Merge branch 'linus' into sched/core

Conflicts:
	fs/proc/array.c

Merge reason: resolve conflict and queue up dependent patch.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

359 files changed, 19222 insertions, 7787 deletions

diff --git a/fs/9p/Kconfig b/fs/9p/Kconfig
index 74e0723e90bc..795233702a4e 100644
--- a/fs/9p/Kconfig
+++ b/fs/9p/Kconfig
@@ -8,3 +8,12 @@ config 9P_FS
 	  See <http://v9fs.sf.net> for more information.
 
 	  If unsure, say N.
+
+config 9P_FSCACHE
+	bool "Enable 9P client caching support (EXPERIMENTAL)"
+	depends on EXPERIMENTAL
+	depends on 9P_FS=m && FSCACHE || 9P_FS=y && FSCACHE=y
+	help
+	  Choose Y here to enable persistent, read-only local
+	  caching support for 9p clients using FS-Cache
+
diff --git a/fs/9p/Makefile b/fs/9p/Makefile
index bc7f0d1551e6..1a940ec7af61 100644
--- a/fs/9p/Makefile
+++ b/fs/9p/Makefile
@@ -8,5 +8,6 @@ obj-$(CONFIG_9P_FS) := 9p.o
 	vfs_dir.o \
 	vfs_dentry.o \
 	v9fs.o \
-	fid.o \
+	fid.o
 
+9p-$(CONFIG_9P_FSCACHE) += cache.o
diff --git a/fs/9p/cache.c b/fs/9p/cache.c
new file mode 100644
index 000000000000..51c94e26a346
--- /dev/null
+++ b/fs/9p/cache.c
@@ -0,0 +1,474 @@
+/*
+ * V9FS cache definitions.
+ *
+ *  Copyright (C) 2009 by Abhishek Kulkarni <adkulkar@umail.iu.edu>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#include <linux/jiffies.h>
+#include <linux/file.h>
+#include <linux/stat.h>
+#include <linux/sched.h>
+#include <linux/fs.h>
+#include <net/9p/9p.h>
+
+#include "v9fs.h"
+#include "cache.h"
+
+#define CACHETAG_LEN  11
+
+struct kmem_cache *vcookie_cache;
+
+struct fscache_netfs v9fs_cache_netfs = {
+	.name 		= "9p",
+	.version 	= 0,
+};
+
+static void init_once(void *foo)
+{
+	struct v9fs_cookie *vcookie = (struct v9fs_cookie *) foo;
+	vcookie->fscache = NULL;
+	vcookie->qid = NULL;
+	inode_init_once(&vcookie->inode);
+}
+
+/**
+ * v9fs_init_vcookiecache - initialize a cache for vcookies to maintain
+ *			    vcookie to inode mapping
+ *
+ * Returns 0 on success.
+ */
+
+static int v9fs_init_vcookiecache(void)
+{
+	vcookie_cache = kmem_cache_create("vcookie_cache",
+					  sizeof(struct v9fs_cookie),
+					  0, (SLAB_RECLAIM_ACCOUNT|
+					      SLAB_MEM_SPREAD),
+					  init_once);
+	if (!vcookie_cache)
+		return -ENOMEM;
+
+	return 0;
+}
+
+/**
+ * v9fs_destroy_vcookiecache - destroy the cache of vcookies
+ *
+ */
+
+static void v9fs_destroy_vcookiecache(void)
+{
+	kmem_cache_destroy(vcookie_cache);
+}
+
+int __v9fs_cache_register(void)
+{
+	int ret;
+	ret = v9fs_init_vcookiecache();
+	if (ret < 0)
+		return ret;
+
+	return fscache_register_netfs(&v9fs_cache_netfs);
+}
+
+void __v9fs_cache_unregister(void)
+{
+	v9fs_destroy_vcookiecache();
+	fscache_unregister_netfs(&v9fs_cache_netfs);
+}
+
+/**
+ * v9fs_random_cachetag - Generate a random tag to be associated
+ *			  with a new cache session.
+ *
+ * The value of jiffies is used for a fairly randomly cache tag.
+ */
+
+static
+int v9fs_random_cachetag(struct v9fs_session_info *v9ses)
+{
+	v9ses->cachetag = kmalloc(CACHETAG_LEN, GFP_KERNEL);
+	if (!v9ses->cachetag)
+		return -ENOMEM;
+
+	return scnprintf(v9ses->cachetag, CACHETAG_LEN, "%lu", jiffies);
+}
+
+static uint16_t v9fs_cache_session_get_key(const void *cookie_netfs_data,
+					   void *buffer, uint16_t bufmax)
+{
+	struct v9fs_session_info *v9ses;
+	uint16_t klen = 0;
+
+	v9ses = (struct v9fs_session_info *)cookie_netfs_data;
+	P9_DPRINTK(P9_DEBUG_FSC, "session %p buf %p size %u", v9ses,
+		   buffer, bufmax);
+
+	if (v9ses->cachetag)
+		klen = strlen(v9ses->cachetag);
+
+	if (klen > bufmax)
+		return 0;
+
+	memcpy(buffer, v9ses->cachetag, klen);
+	P9_DPRINTK(P9_DEBUG_FSC, "cache session tag %s", v9ses->cachetag);
+	return klen;
+}
+
+const struct fscache_cookie_def v9fs_cache_session_index_def = {
+	.name 		= "9P.session",
+	.type 		= FSCACHE_COOKIE_TYPE_INDEX,
+	.get_key 	= v9fs_cache_session_get_key,
+};
+
+void v9fs_cache_session_get_cookie(struct v9fs_session_info *v9ses)
+{
+	/* If no cache session tag was specified, we generate a random one. */
+	if (!v9ses->cachetag)
+		v9fs_random_cachetag(v9ses);
+
+	v9ses->fscache = fscache_acquire_cookie(v9fs_cache_netfs.primary_index,
+						&v9fs_cache_session_index_def,
+						v9ses);
+	P9_DPRINTK(P9_DEBUG_FSC, "session %p get cookie %p", v9ses,
+		   v9ses->fscache);
+}
+
+void v9fs_cache_session_put_cookie(struct v9fs_session_info *v9ses)
+{
+	P9_DPRINTK(P9_DEBUG_FSC, "session %p put cookie %p", v9ses,
+		   v9ses->fscache);
+	fscache_relinquish_cookie(v9ses->fscache, 0);
+	v9ses->fscache = NULL;
+}
+
+
+static uint16_t v9fs_cache_inode_get_key(const void *cookie_netfs_data,
+					 void *buffer, uint16_t bufmax)
+{
+	const struct v9fs_cookie *vcookie = cookie_netfs_data;
+	memcpy(buffer, &vcookie->qid->path, sizeof(vcookie->qid->path));
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p get key %llu", &vcookie->inode,
+		   vcookie->qid->path);
+	return sizeof(vcookie->qid->path);
+}
+
+static void v9fs_cache_inode_get_attr(const void *cookie_netfs_data,
+				      uint64_t *size)
+{
+	const struct v9fs_cookie *vcookie = cookie_netfs_data;
+	*size = i_size_read(&vcookie->inode);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p get attr %llu", &vcookie->inode,
+		   *size);
+}
+
+static uint16_t v9fs_cache_inode_get_aux(const void *cookie_netfs_data,
+					 void *buffer, uint16_t buflen)
+{
+	const struct v9fs_cookie *vcookie = cookie_netfs_data;
+	memcpy(buffer, &vcookie->qid->version, sizeof(vcookie->qid->version));
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p get aux %u", &vcookie->inode,
+		   vcookie->qid->version);
+	return sizeof(vcookie->qid->version);
+}
+
+static enum
+fscache_checkaux v9fs_cache_inode_check_aux(void *cookie_netfs_data,
+					    const void *buffer,
+					    uint16_t buflen)
+{
+	const struct v9fs_cookie *vcookie = cookie_netfs_data;
+
+	if (buflen != sizeof(vcookie->qid->version))
+		return FSCACHE_CHECKAUX_OBSOLETE;
+
+	if (memcmp(buffer, &vcookie->qid->version,
+		   sizeof(vcookie->qid->version)))
+		return FSCACHE_CHECKAUX_OBSOLETE;
+
+	return FSCACHE_CHECKAUX_OKAY;
+}
+
+static void v9fs_cache_inode_now_uncached(void *cookie_netfs_data)
+{
+	struct v9fs_cookie *vcookie = cookie_netfs_data;
+	struct pagevec pvec;
+	pgoff_t first;
+	int loop, nr_pages;
+
+	pagevec_init(&pvec, 0);
+	first = 0;
+
+	for (;;) {
+		nr_pages = pagevec_lookup(&pvec, vcookie->inode.i_mapping,
+					  first,
+					  PAGEVEC_SIZE - pagevec_count(&pvec));
+		if (!nr_pages)
+			break;
+
+		for (loop = 0; loop < nr_pages; loop++)
+			ClearPageFsCache(pvec.pages[loop]);
+
+		first = pvec.pages[nr_pages - 1]->index + 1;
+
+		pvec.nr = nr_pages;
+		pagevec_release(&pvec);
+		cond_resched();
+	}
+}
+
+const struct fscache_cookie_def v9fs_cache_inode_index_def = {
+	.name		= "9p.inode",
+	.type		= FSCACHE_COOKIE_TYPE_DATAFILE,
+	.get_key	= v9fs_cache_inode_get_key,
+	.get_attr	= v9fs_cache_inode_get_attr,
+	.get_aux	= v9fs_cache_inode_get_aux,
+	.check_aux	= v9fs_cache_inode_check_aux,
+	.now_uncached	= v9fs_cache_inode_now_uncached,
+};
+
+void v9fs_cache_inode_get_cookie(struct inode *inode)
+{
+	struct v9fs_cookie *vcookie;
+	struct v9fs_session_info *v9ses;
+
+	if (!S_ISREG(inode->i_mode))
+		return;
+
+	vcookie = v9fs_inode2cookie(inode);
+	if (vcookie->fscache)
+		return;
+
+	v9ses = v9fs_inode2v9ses(inode);
+	vcookie->fscache = fscache_acquire_cookie(v9ses->fscache,
+						  &v9fs_cache_inode_index_def,
+						  vcookie);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p get cookie %p", inode,
+		   vcookie->fscache);
+}
+
+void v9fs_cache_inode_put_cookie(struct inode *inode)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	if (!vcookie->fscache)
+		return;
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p put cookie %p", inode,
+		   vcookie->fscache);
+
+	fscache_relinquish_cookie(vcookie->fscache, 0);
+	vcookie->fscache = NULL;
+}
+
+void v9fs_cache_inode_flush_cookie(struct inode *inode)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	if (!vcookie->fscache)
+		return;
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p flush cookie %p", inode,
+		   vcookie->fscache);
+
+	fscache_relinquish_cookie(vcookie->fscache, 1);
+	vcookie->fscache = NULL;
+}
+
+void v9fs_cache_inode_set_cookie(struct inode *inode, struct file *filp)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+	struct p9_fid *fid;
+
+	if (!vcookie->fscache)
+		return;
+
+	spin_lock(&vcookie->lock);
+	fid = filp->private_data;
+	if ((filp->f_flags & O_ACCMODE) != O_RDONLY)
+		v9fs_cache_inode_flush_cookie(inode);
+	else
+		v9fs_cache_inode_get_cookie(inode);
+
+	spin_unlock(&vcookie->lock);
+}
+
+void v9fs_cache_inode_reset_cookie(struct inode *inode)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+	struct v9fs_session_info *v9ses;
+	struct fscache_cookie *old;
+
+	if (!vcookie->fscache)
+		return;
+
+	old = vcookie->fscache;
+
+	spin_lock(&vcookie->lock);
+	fscache_relinquish_cookie(vcookie->fscache, 1);
+
+	v9ses = v9fs_inode2v9ses(inode);
+	vcookie->fscache = fscache_acquire_cookie(v9ses->fscache,
+						  &v9fs_cache_inode_index_def,
+						  vcookie);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p revalidating cookie old %p new %p",
+		   inode, old, vcookie->fscache);
+
+	spin_unlock(&vcookie->lock);
+}
+
+int __v9fs_fscache_release_page(struct page *page, gfp_t gfp)
+{
+	struct inode *inode = page->mapping->host;
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	BUG_ON(!vcookie->fscache);
+
+	if (PageFsCache(page)) {
+		if (fscache_check_page_write(vcookie->fscache, page)) {
+			if (!(gfp & __GFP_WAIT))
+				return 0;
+			fscache_wait_on_page_write(vcookie->fscache, page);
+		}
+
+		fscache_uncache_page(vcookie->fscache, page);
+		ClearPageFsCache(page);
+	}
+
+	return 1;
+}
+
+void __v9fs_fscache_invalidate_page(struct page *page)
+{
+	struct inode *inode = page->mapping->host;
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	BUG_ON(!vcookie->fscache);
+
+	if (PageFsCache(page)) {
+		fscache_wait_on_page_write(vcookie->fscache, page);
+		BUG_ON(!PageLocked(page));
+		fscache_uncache_page(vcookie->fscache, page);
+		ClearPageFsCache(page);
+	}
+}
+
+static void v9fs_vfs_readpage_complete(struct page *page, void *data,
+				       int error)
+{
+	if (!error)
+		SetPageUptodate(page);
+
+	unlock_page(page);
+}
+
+/**
+ * __v9fs_readpage_from_fscache - read a page from cache
+ *
+ * Returns 0 if the pages are in cache and a BIO is submitted,
+ * 1 if the pages are not in cache and -error otherwise.
+ */
+
+int __v9fs_readpage_from_fscache(struct inode *inode, struct page *page)
+{
+	int ret;
+	const struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p page %p", inode, page);
+	if (!vcookie->fscache)
+		return -ENOBUFS;
+
+	ret = fscache_read_or_alloc_page(vcookie->fscache,
+					 page,
+					 v9fs_vfs_readpage_complete,
+					 NULL,
+					 GFP_KERNEL);
+	switch (ret) {
+	case -ENOBUFS:
+	case -ENODATA:
+		P9_DPRINTK(P9_DEBUG_FSC, "page/inode not in cache %d", ret);
+		return 1;
+	case 0:
+		P9_DPRINTK(P9_DEBUG_FSC, "BIO submitted");
+		return ret;
+	default:
+		P9_DPRINTK(P9_DEBUG_FSC, "ret %d", ret);
+		return ret;
+	}
+}
+
+/**
+ * __v9fs_readpages_from_fscache - read multiple pages from cache
+ *
+ * Returns 0 if the pages are in cache and a BIO is submitted,
+ * 1 if the pages are not in cache and -error otherwise.
+ */
+
+int __v9fs_readpages_from_fscache(struct inode *inode,
+				  struct address_space *mapping,
+				  struct list_head *pages,
+				  unsigned *nr_pages)
+{
+	int ret;
+	const struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p pages %u", inode, *nr_pages);
+	if (!vcookie->fscache)
+		return -ENOBUFS;
+
+	ret = fscache_read_or_alloc_pages(vcookie->fscache,
+					  mapping, pages, nr_pages,
+					  v9fs_vfs_readpage_complete,
+					  NULL,
+					  mapping_gfp_mask(mapping));
+	switch (ret) {
+	case -ENOBUFS:
+	case -ENODATA:
+		P9_DPRINTK(P9_DEBUG_FSC, "pages/inodes not in cache %d", ret);
+		return 1;
+	case 0:
+		BUG_ON(!list_empty(pages));
+		BUG_ON(*nr_pages != 0);
+		P9_DPRINTK(P9_DEBUG_FSC, "BIO submitted");
+		return ret;
+	default:
+		P9_DPRINTK(P9_DEBUG_FSC, "ret %d", ret);
+		return ret;
+	}
+}
+
+/**
+ * __v9fs_readpage_to_fscache - write a page to the cache
+ *
+ */
+
+void __v9fs_readpage_to_fscache(struct inode *inode, struct page *page)
+{
+	int ret;
+	const struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+
+	P9_DPRINTK(P9_DEBUG_FSC, "inode %p page %p", inode, page);
+	ret = fscache_write_page(vcookie->fscache, page, GFP_KERNEL);
+	P9_DPRINTK(P9_DEBUG_FSC, "ret =  %d", ret);
+	if (ret != 0)
+		v9fs_uncache_page(inode, page);
+}
diff --git a/fs/9p/cache.h b/fs/9p/cache.h
new file mode 100644
index 000000000000..a94192bfaee8
--- /dev/null
+++ b/fs/9p/cache.h
@@ -0,0 +1,176 @@
+/*
+ * V9FS cache definitions.
+ *
+ *  Copyright (C) 2009 by Abhishek Kulkarni <adkulkar@umail.iu.edu>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#ifndef _9P_CACHE_H
+#ifdef CONFIG_9P_FSCACHE
+#include <linux/fscache.h>
+#include <linux/spinlock.h>
+
+extern struct kmem_cache *vcookie_cache;
+
+struct v9fs_cookie {
+	spinlock_t lock;
+	struct inode inode;
+	struct fscache_cookie *fscache;
+	struct p9_qid *qid;
+};
+
+static inline struct v9fs_cookie *v9fs_inode2cookie(const struct inode *inode)
+{
+	return container_of(inode, struct v9fs_cookie, inode);
+}
+
+extern struct fscache_netfs v9fs_cache_netfs;
+extern const struct fscache_cookie_def v9fs_cache_session_index_def;
+extern const struct fscache_cookie_def v9fs_cache_inode_index_def;
+
+extern void v9fs_cache_session_get_cookie(struct v9fs_session_info *v9ses);
+extern void v9fs_cache_session_put_cookie(struct v9fs_session_info *v9ses);
+
+extern void v9fs_cache_inode_get_cookie(struct inode *inode);
+extern void v9fs_cache_inode_put_cookie(struct inode *inode);
+extern void v9fs_cache_inode_flush_cookie(struct inode *inode);
+extern void v9fs_cache_inode_set_cookie(struct inode *inode, struct file *filp);
+extern void v9fs_cache_inode_reset_cookie(struct inode *inode);
+
+extern int __v9fs_cache_register(void);
+extern void __v9fs_cache_unregister(void);
+
+extern int __v9fs_fscache_release_page(struct page *page, gfp_t gfp);
+extern void __v9fs_fscache_invalidate_page(struct page *page);
+extern int __v9fs_readpage_from_fscache(struct inode *inode,
+					struct page *page);
+extern int __v9fs_readpages_from_fscache(struct inode *inode,
+					 struct address_space *mapping,
+					 struct list_head *pages,
+					 unsigned *nr_pages);
+extern void __v9fs_readpage_to_fscache(struct inode *inode, struct page *page);
+
+
+/**
+ * v9fs_cache_register - Register v9fs file system with the cache
+ */
+static inline int v9fs_cache_register(void)
+{
+	return __v9fs_cache_register();
+}
+
+/**
+ * v9fs_cache_unregister - Unregister v9fs from the cache
+ */
+static inline void v9fs_cache_unregister(void)
+{
+	__v9fs_cache_unregister();
+}
+
+static inline int v9fs_fscache_release_page(struct page *page,
+					    gfp_t gfp)
+{
+	return __v9fs_fscache_release_page(page, gfp);
+}
+
+static inline void v9fs_fscache_invalidate_page(struct page *page)
+{
+	__v9fs_fscache_invalidate_page(page);
+}
+
+static inline int v9fs_readpage_from_fscache(struct inode *inode,
+					     struct page *page)
+{
+	return __v9fs_readpage_from_fscache(inode, page);
+}
+
+static inline int v9fs_readpages_from_fscache(struct inode *inode,
+					      struct address_space *mapping,
+					      struct list_head *pages,
+					      unsigned *nr_pages)
+{
+	return __v9fs_readpages_from_fscache(inode, mapping, pages,
+					     nr_pages);
+}
+
+static inline void v9fs_readpage_to_fscache(struct inode *inode,
+					    struct page *page)
+{
+	if (PageFsCache(page))
+		__v9fs_readpage_to_fscache(inode, page);
+}
+
+static inline void v9fs_uncache_page(struct inode *inode, struct page *page)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+	fscache_uncache_page(vcookie->fscache, page);
+	BUG_ON(PageFsCache(page));
+}
+
+static inline void v9fs_vcookie_set_qid(struct inode *inode,
+					struct p9_qid *qid)
+{
+	struct v9fs_cookie *vcookie = v9fs_inode2cookie(inode);
+	spin_lock(&vcookie->lock);
+	vcookie->qid = qid;
+	spin_unlock(&vcookie->lock);
+}
+
+#else /* CONFIG_9P_FSCACHE */
+
+static inline int v9fs_cache_register(void)
+{
+	return 1;
+}
+
+static inline void v9fs_cache_unregister(void) {}
+
+static inline int v9fs_fscache_release_page(struct page *page,
+					    gfp_t gfp) {
+	return 1;
+}
+
+static inline void v9fs_fscache_invalidate_page(struct page *page) {}
+
+static inline int v9fs_readpage_from_fscache(struct inode *inode,
+					     struct page *page)
+{
+	return -ENOBUFS;
+}
+
+static inline int v9fs_readpages_from_fscache(struct inode *inode,
+					      struct address_space *mapping,
+					      struct list_head *pages,
+					      unsigned *nr_pages)
+{
+	return -ENOBUFS;
+}
+
+static inline void v9fs_readpage_to_fscache(struct inode *inode,
+					    struct page *page)
+{}
+
+static inline void v9fs_uncache_page(struct inode *inode, struct page *page)
+{}
+
+static inline void v9fs_vcookie_set_qid(struct inode *inode,
+					struct p9_qid *qid)
+{}
+
+#endif /* CONFIG_9P_FSCACHE */
+#endif /* _9P_CACHE_H */
diff --git a/fs/9p/v9fs.c b/fs/9p/v9fs.c
index f7003cfac63d..cf62b05e296a 100644
--- a/fs/9p/v9fs.c
+++ b/fs/9p/v9fs.c
@@ -34,21 +34,25 @@
 #include <net/9p/transport.h>
 #include "v9fs.h"
 #include "v9fs_vfs.h"
+#include "cache.h"
+
+static DEFINE_SPINLOCK(v9fs_sessionlist_lock);
+static LIST_HEAD(v9fs_sessionlist);
 
 /*
-  * Option Parsing (code inspired by NFS code)
-  *  NOTE: each transport will parse its own options
-  */
+ * Option Parsing (code inspired by NFS code)
+ *  NOTE: each transport will parse its own options
+ */
 
 enum {
 	/* Options that take integer arguments */
 	Opt_debug, Opt_dfltuid, Opt_dfltgid, Opt_afid,
 	/* String options */
-	Opt_uname, Opt_remotename, Opt_trans,
+	Opt_uname, Opt_remotename, Opt_trans, Opt_cache, Opt_cachetag,
 	/* Options that take no arguments */
 	Opt_nodevmap,
 	/* Cache options */
-	Opt_cache_loose,
+	Opt_cache_loose, Opt_fscache,
 	/* Access options */
 	Opt_access,
 	/* Error token */
@@ -63,8 +67,10 @@ static const match_table_t tokens = {
 	{Opt_uname, "uname=%s"},
 	{Opt_remotename, "aname=%s"},
 	{Opt_nodevmap, "nodevmap"},
-	{Opt_cache_loose, "cache=loose"},
+	{Opt_cache, "cache=%s"},
 	{Opt_cache_loose, "loose"},
+	{Opt_fscache, "fscache"},
+	{Opt_cachetag, "cachetag=%s"},
 	{Opt_access, "access=%s"},
 	{Opt_err, NULL}
 };
@@ -89,16 +95,16 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts)
 	v9ses->afid = ~0;
 	v9ses->debug = 0;
 	v9ses->cache = 0;
+#ifdef CONFIG_9P_FSCACHE
+	v9ses->cachetag = NULL;
+#endif
 
 	if (!opts)
 		return 0;
 
 	options = kstrdup(opts, GFP_KERNEL);
-	if (!options) {
-		P9_DPRINTK(P9_DEBUG_ERROR,
-			   "failed to allocate copy of option string\n");
-		return -ENOMEM;
-	}
+	if (!options)
+		goto fail_option_alloc;
 
 	while ((p = strsep(&options, ",")) != NULL) {
 		int token;
@@ -143,16 +149,33 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts)
 		case Opt_cache_loose:
 			v9ses->cache = CACHE_LOOSE;
 			break;
+		case Opt_fscache:
+			v9ses->cache = CACHE_FSCACHE;
+			break;
+		case Opt_cachetag:
+#ifdef CONFIG_9P_FSCACHE
+			v9ses->cachetag = match_strdup(&args[0]);
+#endif
+			break;
+		case Opt_cache:
+			s = match_strdup(&args[0]);
+			if (!s)
+				goto fail_option_alloc;
+
+			if (strcmp(s, "loose") == 0)
+				v9ses->cache = CACHE_LOOSE;
+			else if (strcmp(s, "fscache") == 0)
+				v9ses->cache = CACHE_FSCACHE;
+			else
+				v9ses->cache = CACHE_NONE;
+			kfree(s);
+			break;
 
 		case Opt_access:
 			s = match_strdup(&args[0]);
-			if (!s) {
-				P9_DPRINTK(P9_DEBUG_ERROR,
-					   "failed to allocate copy"
-					   " of option argument\n");
-				ret = -ENOMEM;
-				break;
-			}
+			if (!s)
+				goto fail_option_alloc;
+
 			v9ses->flags &= ~V9FS_ACCESS_MASK;
 			if (strcmp(s, "user") == 0)
 				v9ses->flags |= V9FS_ACCESS_USER;
@@ -173,6 +196,11 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts)
 	}
 	kfree(options);
 	return ret;
+
+fail_option_alloc:
+	P9_DPRINTK(P9_DEBUG_ERROR,
+		   "failed to allocate copy of option argument\n");
+	return -ENOMEM;
 }
 
 /**
@@ -200,6 +228,10 @@ struct p9_fid *v9fs_session_init(struct v9fs_session_info *v9ses,
 		return ERR_PTR(-ENOMEM);
 	}
 
+	spin_lock(&v9fs_sessionlist_lock);
+	list_add(&v9ses->slist, &v9fs_sessionlist);
+	spin_unlock(&v9fs_sessionlist_lock);
+
 	v9ses->flags = V9FS_EXTENDED | V9FS_ACCESS_USER;
 	strcpy(v9ses->uname, V9FS_DEFUSER);
 	strcpy(v9ses->aname, V9FS_DEFANAME);
@@ -249,6 +281,11 @@ struct p9_fid *v9fs_session_init(struct v9fs_session_info *v9ses,
 	else
 		fid->uid = ~0;
 
+#ifdef CONFIG_9P_FSCACHE
+	/* register the session for caching */
+	v9fs_cache_session_get_cookie(v9ses);
+#endif
+
 	return fid;
 
 error:
@@ -268,8 +305,18 @@ void v9fs_session_close(struct v9fs_session_info *v9ses)
 		v9ses->clnt = NULL;
 	}
 
+#ifdef CONFIG_9P_FSCACHE
+	if (v9ses->fscache) {
+		v9fs_cache_session_put_cookie(v9ses);
+		kfree(v9ses->cachetag);
+	}
+#endif
 	__putname(v9ses->uname);
 	__putname(v9ses->aname);
+
+	spin_lock(&v9fs_sessionlist_lock);
+	list_del(&v9ses->slist);
+	spin_unlock(&v9fs_sessionlist_lock);
 }
 
 /**
@@ -286,25 +333,132 @@ void v9fs_session_cancel(struct v9fs_session_info *v9ses) {
 
 extern int v9fs_error_init(void);
 
+static struct kobject *v9fs_kobj;
+
+#ifdef CONFIG_9P_FSCACHE
 /**
- * v9fs_init - Initialize module
+ * caches_show - list caches associated with a session
+ *
+ * Returns the size of buffer written.
+ */
+
+static ssize_t caches_show(struct kobject *kobj,
+			   struct kobj_attribute *attr,
+			   char *buf)
+{
+	ssize_t n = 0, count = 0, limit = PAGE_SIZE;
+	struct v9fs_session_info *v9ses;
+
+	spin_lock(&v9fs_sessionlist_lock);
+	list_for_each_entry(v9ses, &v9fs_sessionlist, slist) {
+		if (v9ses->cachetag) {
+			n = snprintf(buf, limit, "%s\n", v9ses->cachetag);
+			if (n < 0) {
+				count = n;
+				break;
+			}
+
+			count += n;
+			limit -= n;
+		}
+	}
+
+	spin_unlock(&v9fs_sessionlist_lock);
+	return count;
+}
+
+static struct kobj_attribute v9fs_attr_cache = __ATTR_RO(caches);
+#endif /* CONFIG_9P_FSCACHE */
+
+static struct attribute *v9fs_attrs[] = {
+#ifdef CONFIG_9P_FSCACHE
+	&v9fs_attr_cache.attr,
+#endif
+	NULL,
+};
+
+static struct attribute_group v9fs_attr_group = {
+	.attrs = v9fs_attrs,
+};
+
+/**
+ * v9fs_sysfs_init - Initialize the v9fs sysfs interface
+ *
+ */
+
+static int v9fs_sysfs_init(void)
+{
+	v9fs_kobj = kobject_create_and_add("9p", fs_kobj);
+	if (!v9fs_kobj)
+		return -ENOMEM;
+
+	if (sysfs_create_group(v9fs_kobj, &v9fs_attr_group)) {
+		kobject_put(v9fs_kobj);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+/**
+ * v9fs_sysfs_cleanup - Unregister the v9fs sysfs interface
+ *
+ */
+
+static void v9fs_sysfs_cleanup(void)
+{
+	sysfs_remove_group(v9fs_kobj, &v9fs_attr_group);
+	kobject_put(v9fs_kobj);
+}
+
+/**
+ * init_v9fs - Initialize module
  *
  */
 
 static int __init init_v9fs(void)
 {
+	int err;
 	printk(KERN_INFO "Installing v9fs 9p2000 file system support\n");
 	/* TODO: Setup list of registered trasnport modules */
-	return register_filesystem(&v9fs_fs_type);
+	err = register_filesystem(&v9fs_fs_type);
+	if (err < 0) {
+		printk(KERN_ERR "Failed to register filesystem\n");
+		return err;
+	}
+
+	err = v9fs_cache_register();
+	if (err < 0) {
+		printk(KERN_ERR "Failed to register v9fs for caching\n");
+		goto out_fs_unreg;
+	}
+
+	err = v9fs_sysfs_init();
+	if (err < 0) {
+		printk(KERN_ERR "Failed to register with sysfs\n");
+		goto out_sysfs_cleanup;
+	}
+
+	return 0;
+
+out_sysfs_cleanup:
+	v9fs_sysfs_cleanup();
+
+out_fs_unreg:
+	unregister_filesystem(&v9fs_fs_type);
+
+	return err;
 }
 
 /**
- * v9fs_init - shutdown module
+ * exit_v9fs - shutdown module
  *
  */
 
 static void __exit exit_v9fs(void)
 {
+	v9fs_sysfs_cleanup();
+	v9fs_cache_unregister();
 	unregister_filesystem(&v9fs_fs_type);
 }
 
diff --git a/fs/9p/v9fs.h b/fs/9p/v9fs.h
index 38762bf102a9..019f4ccb70c1 100644
--- a/fs/9p/v9fs.h
+++ b/fs/9p/v9fs.h
@@ -51,6 +51,7 @@ enum p9_session_flags {
 enum p9_cache_modes {
 	CACHE_NONE,
 	CACHE_LOOSE,
+	CACHE_FSCACHE,
 };
 
 /**
@@ -60,6 +61,8 @@ enum p9_cache_modes {
  * @debug: debug level
  * @afid: authentication handle
  * @cache: cache mode of type &p9_cache_modes
+ * @cachetag: the tag of the cache associated with this session
+ * @fscache: session cookie associated with FS-Cache
  * @options: copy of options string given by user
  * @uname: string user name to mount hierarchy as
  * @aname: mount specifier for remote hierarchy
@@ -68,7 +71,7 @@ enum p9_cache_modes {
  * @dfltgid: default numeric groupid to mount hierarchy as
  * @uid: if %V9FS_ACCESS_SINGLE, the numeric uid which mounted the hierarchy
  * @clnt: reference to 9P network client instantiated for this session
- * @debugfs_dir: reference to debugfs_dir which can be used for add'l debug
+ * @slist: reference to list of registered 9p sessions
  *
  * This structure holds state for each session instance established during
  * a sys_mount() .
@@ -84,6 +87,10 @@ struct v9fs_session_info {
 	unsigned short debug;
 	unsigned int afid;
 	unsigned int cache;
+#ifdef CONFIG_9P_FSCACHE
+	char *cachetag;
+	struct fscache_cookie *fscache;
+#endif
 
 	char *uname;		/* user name to mount as */
 	char *aname;		/* name of remote hierarchy being mounted */
@@ -92,11 +99,9 @@ struct v9fs_session_info {
 	unsigned int dfltgid;	/* default gid for legacy support */
 	u32 uid;		/* if ACCESS_SINGLE, the uid that has access */
 	struct p9_client *clnt;	/* 9p client */
-	struct dentry *debugfs_dir;
+	struct list_head slist; /* list of sessions registered with v9fs */
 };
 
-extern struct dentry *v9fs_debugfs_root;
-
 struct p9_fid *v9fs_session_init(struct v9fs_session_info *, const char *,
 									char *);
 void v9fs_session_close(struct v9fs_session_info *v9ses);
diff --git a/fs/9p/v9fs_vfs.h b/fs/9p/v9fs_vfs.h
index f0c7de78e205..3a7560e35865 100644
--- a/fs/9p/v9fs_vfs.h
+++ b/fs/9p/v9fs_vfs.h
@@ -44,7 +44,13 @@ extern const struct file_operations v9fs_dir_operations;
 extern const struct dentry_operations v9fs_dentry_operations;
 extern const struct dentry_operations v9fs_cached_dentry_operations;
 
+#ifdef CONFIG_9P_FSCACHE
+struct inode *v9fs_alloc_inode(struct super_block *sb);
+void v9fs_destroy_inode(struct inode *inode);
+#endif
+
 struct inode *v9fs_get_inode(struct super_block *sb, int mode);
+void v9fs_clear_inode(struct inode *inode);
 ino_t v9fs_qid2ino(struct p9_qid *qid);
 void v9fs_stat2inode(struct p9_wstat *, struct inode *, struct super_block *);
 int v9fs_dir_release(struct inode *inode, struct file *filp);
diff --git a/fs/9p/vfs_addr.c b/fs/9p/vfs_addr.c
index 92828281a30b..90e38449f4b3 100644
--- a/fs/9p/vfs_addr.c
+++ b/fs/9p/vfs_addr.c
@@ -38,6 +38,7 @@
 
 #include "v9fs.h"
 #include "v9fs_vfs.h"
+#include "cache.h"
 
 /**
  * v9fs_vfs_readpage - read an entire page in from 9P
@@ -52,18 +53,31 @@ static int v9fs_vfs_readpage(struct file *filp, struct page *page)
 	int retval;
 	loff_t offset;
 	char *buffer;
+	struct inode *inode;
 
+	inode = page->mapping->host;
 	P9_DPRINTK(P9_DEBUG_VFS, "\n");
+
+	BUG_ON(!PageLocked(page));
+
+	retval = v9fs_readpage_from_fscache(inode, page);
+	if (retval == 0)
+		return retval;
+
 	buffer = kmap(page);
 	offset = page_offset(page);
 
 	retval = v9fs_file_readn(filp, buffer, NULL, PAGE_CACHE_SIZE, offset);
-	if (retval < 0)
+	if (retval < 0) {
+		v9fs_uncache_page(inode, page);
 		goto done;
+	}
 
 	memset(buffer + retval, 0, PAGE_CACHE_SIZE - retval);
 	flush_dcache_page(page);
 	SetPageUptodate(page);
+
+	v9fs_readpage_to_fscache(inode, page);
 	retval = 0;
 
 done:
@@ -72,6 +86,78 @@ done:
 	return retval;
 }
 
+/**
+ * v9fs_vfs_readpages - read a set of pages from 9P
+ *
+ * @filp: file being read
+ * @mapping: the address space
+ * @pages: list of pages to read
+ * @nr_pages: count of pages to read
+ *
+ */
+
+static int v9fs_vfs_readpages(struct file *filp, struct address_space *mapping,
+			     struct list_head *pages, unsigned nr_pages)
+{
+	int ret = 0;
+	struct inode *inode;
+
+	inode = mapping->host;
+	P9_DPRINTK(P9_DEBUG_VFS, "inode: %p file: %p\n", inode, filp);
+
+	ret = v9fs_readpages_from_fscache(inode, mapping, pages, &nr_pages);
+	if (ret == 0)
+		return ret;
+
+	ret = read_cache_pages(mapping, pages, (void *)v9fs_vfs_readpage, filp);
+	P9_DPRINTK(P9_DEBUG_VFS, "  = %d\n", ret);
+	return ret;
+}
+
+/**
+ * v9fs_release_page - release the private state associated with a page
+ *
+ * Returns 1 if the page can be released, false otherwise.
+ */
+
+static int v9fs_release_page(struct page *page, gfp_t gfp)
+{
+	if (PagePrivate(page))
+		return 0;
+
+	return v9fs_fscache_release_page(page, gfp);
+}
+
+/**
+ * v9fs_invalidate_page - Invalidate a page completely or partially
+ *
+ * @page: structure to page
+ * @offset: offset in the page
+ */
+
+static void v9fs_invalidate_page(struct page *page, unsigned long offset)
+{
+	if (offset == 0)
+		v9fs_fscache_invalidate_page(page);
+}
+
+/**
+ * v9fs_launder_page - Writeback a dirty page
+ * Since the writes go directly to the server, we simply return a 0
+ * here to indicate success.
+ *
+ * Returns 0 on success.
+ */
+
+static int v9fs_launder_page(struct page *page)
+{
+	return 0;
+}
+
 const struct address_space_operations v9fs_addr_operations = {
       .readpage = v9fs_vfs_readpage,
+      .readpages = v9fs_vfs_readpages,
+      .releasepage = v9fs_release_page,
+      .invalidatepage = v9fs_invalidate_page,
+      .launder_page = v9fs_launder_page,
 };
diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
index 68bf2af6c389..3902bf43a088 100644
--- a/fs/9p/vfs_file.c
+++ b/fs/9p/vfs_file.c
@@ -32,6 +32,7 @@
 #include <linux/string.h>
 #include <linux/inet.h>
 #include <linux/list.h>
+#include <linux/pagemap.h>
 #include <asm/uaccess.h>
 #include <linux/idr.h>
 #include <net/9p/9p.h>
@@ -40,6 +41,7 @@
 #include "v9fs.h"
 #include "v9fs_vfs.h"
 #include "fid.h"
+#include "cache.h"
 
 static const struct file_operations v9fs_cached_file_operations;
 
@@ -72,7 +74,7 @@ int v9fs_file_open(struct inode *inode, struct file *file)
 			return err;
 		}
 		if (omode & P9_OTRUNC) {
-			inode->i_size = 0;
+			i_size_write(inode, 0);
 			inode->i_blocks = 0;
 		}
 		if ((file->f_flags & O_APPEND) && (!v9fs_extended(v9ses)))
@@ -85,6 +87,10 @@ int v9fs_file_open(struct inode *inode, struct file *file)
 		/* enable cached file options */
 		if(file->f_op == &v9fs_file_operations)
 			file->f_op = &v9fs_cached_file_operations;
+
+#ifdef CONFIG_9P_FSCACHE
+		v9fs_cache_inode_set_cookie(inode, file);
+#endif
 	}
 
 	return 0;
@@ -210,6 +216,7 @@ v9fs_file_write(struct file *filp, const char __user * data,
 	struct p9_client *clnt;
 	struct inode *inode = filp->f_path.dentry->d_inode;
 	int origin = *offset;
+	unsigned long pg_start, pg_end;
 
 	P9_DPRINTK(P9_DEBUG_VFS, "data %p count %d offset %x\n", data,
 		(int)count, (int)*offset);
@@ -225,7 +232,7 @@ v9fs_file_write(struct file *filp, const char __user * data,
 		if (count < rsize)
 			rsize = count;
 
-		n = p9_client_write(fid, NULL, data+total, *offset+total,
+		n = p9_client_write(fid, NULL, data+total, origin+total,
 									rsize);
 		if (n <= 0)
 			break;
@@ -234,14 +241,14 @@ v9fs_file_write(struct file *filp, const char __user * data,
 	} while (count > 0);
 
 	if (total > 0) {
-		invalidate_inode_pages2_range(inode->i_mapping, origin,
-								origin+total);
+		pg_start = origin >> PAGE_CACHE_SHIFT;
+		pg_end = (origin + total - 1) >> PAGE_CACHE_SHIFT;
+		if (inode->i_mapping && inode->i_mapping->nrpages)
+			invalidate_inode_pages2_range(inode->i_mapping,
+						      pg_start, pg_end);
 		*offset += total;
-	}
-
-	if (*offset > inode->i_size) {
-		inode->i_size = *offset;
-		inode->i_blocks = (inode->i_size + 512 - 1) >> 9;
+		i_size_write(inode, i_size_read(inode) + total);
+		inode->i_blocks = (i_size_read(inode) + 512 - 1) >> 9;
 	}
 
 	if (n < 0)
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 06a223d50a81..5947628aefef 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -40,6 +40,7 @@
 #include "v9fs.h"
 #include "v9fs_vfs.h"
 #include "fid.h"
+#include "cache.h"
 
 static const struct inode_operations v9fs_dir_inode_operations;
 static const struct inode_operations v9fs_dir_inode_operations_ext;
@@ -197,6 +198,39 @@ v9fs_blank_wstat(struct p9_wstat *wstat)
 	wstat->extension = NULL;
 }
 
+#ifdef CONFIG_9P_FSCACHE
+/**
+ * v9fs_alloc_inode - helper function to allocate an inode
+ * This callback is executed before setting up the inode so that we
+ * can associate a vcookie with each inode.
+ *
+ */
+
+struct inode *v9fs_alloc_inode(struct super_block *sb)
+{
+	struct v9fs_cookie *vcookie;
+	vcookie = (struct v9fs_cookie *)kmem_cache_alloc(vcookie_cache,
+							 GFP_KERNEL);
+	if (!vcookie)
+		return NULL;
+
+	vcookie->fscache = NULL;
+	vcookie->qid = NULL;
+	spin_lock_init(&vcookie->lock);
+	return &vcookie->inode;
+}
+
+/**
+ * v9fs_destroy_inode - destroy an inode
+ *
+ */
+
+void v9fs_destroy_inode(struct inode *inode)
+{
+	kmem_cache_free(vcookie_cache, v9fs_inode2cookie(inode));
+}
+#endif
+
 /**
  * v9fs_get_inode - helper function to setup an inode
  * @sb: superblock
@@ -326,6 +360,21 @@ error:
 }
 */
 
+
+/**
+ * v9fs_clear_inode - release an inode
+ * @inode: inode to release
+ *
+ */
+void v9fs_clear_inode(struct inode *inode)
+{
+	filemap_fdatawrite(inode->i_mapping);
+
+#ifdef CONFIG_9P_FSCACHE
+	v9fs_cache_inode_put_cookie(inode);
+#endif
+}
+
 /**
  * v9fs_inode_from_fid - populate an inode by issuing a attribute request
  * @v9ses: session information
@@ -356,8 +405,14 @@ v9fs_inode_from_fid(struct v9fs_session_info *v9ses, struct p9_fid *fid,
 
 	v9fs_stat2inode(st, ret, sb);
 	ret->i_ino = v9fs_qid2ino(&st->qid);
+
+#ifdef CONFIG_9P_FSCACHE
+	v9fs_vcookie_set_qid(ret, &st->qid);
+	v9fs_cache_inode_get_cookie(ret);
+#endif
 	p9stat_free(st);
 	kfree(st);
+
 	return ret;
 
 error:
@@ -751,7 +806,7 @@ v9fs_vfs_getattr(struct vfsmount *mnt, struct dentry *dentry,
 	P9_DPRINTK(P9_DEBUG_VFS, "dentry: %p\n", dentry);
 	err = -EPERM;
 	v9ses = v9fs_inode2v9ses(dentry->d_inode);
-	if (v9ses->cache == CACHE_LOOSE)
+	if (v9ses->cache == CACHE_LOOSE || v9ses->cache == CACHE_FSCACHE)
 		return simple_getattr(mnt, dentry, stat);
 
 	fid = v9fs_fid_lookup(dentry);
@@ -872,10 +927,10 @@ v9fs_stat2inode(struct p9_wstat *stat, struct inode *inode,
 	} else
 		inode->i_rdev = 0;
 
-	inode->i_size = stat->length;
+	i_size_write(inode, stat->length);
 
 	/* not real number of blocks, but 512 byte ones ... */
-	inode->i_blocks = (inode->i_size + 512 - 1) >> 9;
+	inode->i_blocks = (i_size_read(inode) + 512 - 1) >> 9;
 }
 
 /**
diff --git a/fs/9p/vfs_super.c b/fs/9p/vfs_super.c
index 8961f1a8f668..14a86448572c 100644
--- a/fs/9p/vfs_super.c
+++ b/fs/9p/vfs_super.c
@@ -44,21 +44,9 @@
 #include "v9fs_vfs.h"
 #include "fid.h"
 
-static void v9fs_clear_inode(struct inode *);
 static const struct super_operations v9fs_super_ops;
 
 /**
- * v9fs_clear_inode - release an inode
- * @inode: inode to release
- *
- */
-
-static void v9fs_clear_inode(struct inode *inode)
-{
-	filemap_fdatawrite(inode->i_mapping);
-}
-
-/**
  * v9fs_set_super - set the superblock
  * @s: super block
  * @data: file system specific data
@@ -220,6 +208,10 @@ v9fs_umount_begin(struct super_block *sb)
 }
 
 static const struct super_operations v9fs_super_ops = {
+#ifdef CONFIG_9P_FSCACHE
+	.alloc_inode = v9fs_alloc_inode,
+	.destroy_inode = v9fs_destroy_inode,
+#endif
 	.statfs = simple_statfs,
 	.clear_inode = v9fs_clear_inode,
 	.show_options = generic_show_options,
diff --git a/fs/Kconfig b/fs/Kconfig
index 455aa207e67e..d4bf8caad8d0 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -109,6 +109,7 @@ source "fs/sysfs/Kconfig"
 
 config TMPFS
 	bool "Virtual memory file system support (former shm fs)"
+	depends on SHMEM
 	help
 	  Tmpfs is a file system which keeps all files in virtual memory.
 
diff --git a/fs/adfs/inode.c b/fs/adfs/inode.c
index 798cb071d132..3f57ce4bee5d 100644
--- a/fs/adfs/inode.c
+++ b/fs/adfs/inode.c
@@ -19,9 +19,6 @@ static int
 adfs_get_block(struct inode *inode, sector_t block, struct buffer_head *bh,
 	       int create)
 {
-	if (block < 0)
-		goto abort_negative;
-
 	if (!create) {
 		if (block >= inode->i_blocks)
 			goto abort_toobig;
@@ -34,10 +31,6 @@ adfs_get_block(struct inode *inode, sector_t block, struct buffer_head *bh,
 	/* don't support allocation of blocks yet */
 	return -EIO;
 
-abort_negative:
-	adfs_error(inode->i_sb, "block %d < 0", block);
-	return -EIO;
-
 abort_toobig:
 	return 0;
 }
diff --git a/fs/afs/cache.h b/fs/afs/cache.h
deleted file mode 100644
index 5c4f6b499e90..000000000000
--- a/fs/afs/cache.h
+++ /dev/null
@@ -1,12 +0,0 @@
-/* AFS local cache management interface
- *
- * Copyright (C) 2008 Red Hat, Inc. All Rights Reserved.
- * Written by David Howells (dhowells@redhat.com)
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License
- * as published by the Free Software Foundation; either version
- * 2 of the License, or (at your option) any later version.
- */
-
-#include <linux/fscache.h>
diff --git a/fs/afs/flock.c b/fs/afs/flock.c
index 3ff8bdd18fb3..0931bc1325eb 100644
--- a/fs/afs/flock.c
+++ b/fs/afs/flock.c
@@ -21,7 +21,7 @@ static void afs_fl_release_private(struct file_lock *fl);
 static struct workqueue_struct *afs_lock_manager;
 static DEFINE_MUTEX(afs_lock_manager_mutex);
 
-static struct file_lock_operations afs_lock_ops = {
+static const struct file_lock_operations afs_lock_ops = {
 	.fl_copy_lock		= afs_fl_copy_lock,
 	.fl_release_private	= afs_fl_release_private,
 };
diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index 106be66dafd2..6ece2a13bf71 100644
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -18,10 +18,10 @@
 #include <linux/key.h>
 #include <linux/workqueue.h>
 #include <linux/sched.h>
+#include <linux/fscache.h>
 
 #include "afs.h"
 #include "afs_vl.h"
-#include "cache.h"
 
 #define AFS_CELL_MAX_ADDRS 15
 
diff --git a/fs/afs/proc.c b/fs/afs/proc.c
index 8630615e57fe..852739d262a9 100644
--- a/fs/afs/proc.c
+++ b/fs/afs/proc.c
@@ -28,7 +28,7 @@ static int afs_proc_cells_show(struct seq_file *m, void *v);
 static ssize_t afs_proc_cells_write(struct file *file, const char __user *buf,
 				    size_t size, loff_t *_pos);
 
-static struct seq_operations afs_proc_cells_ops = {
+static const struct seq_operations afs_proc_cells_ops = {
 	.start	= afs_proc_cells_start,
 	.next	= afs_proc_cells_next,
 	.stop	= afs_proc_cells_stop,
@@ -70,7 +70,7 @@ static void *afs_proc_cell_volumes_next(struct seq_file *p, void *v,
 static void afs_proc_cell_volumes_stop(struct seq_file *p, void *v);
 static int afs_proc_cell_volumes_show(struct seq_file *m, void *v);
 
-static struct seq_operations afs_proc_cell_volumes_ops = {
+static const struct seq_operations afs_proc_cell_volumes_ops = {
 	.start	= afs_proc_cell_volumes_start,
 	.next	= afs_proc_cell_volumes_next,
 	.stop	= afs_proc_cell_volumes_stop,
@@ -95,7 +95,7 @@ static void *afs_proc_cell_vlservers_next(struct seq_file *p, void *v,
 static void afs_proc_cell_vlservers_stop(struct seq_file *p, void *v);
 static int afs_proc_cell_vlservers_show(struct seq_file *m, void *v);
 
-static struct seq_operations afs_proc_cell_vlservers_ops = {
+static const struct seq_operations afs_proc_cell_vlservers_ops = {
 	.start	= afs_proc_cell_vlservers_start,
 	.next	= afs_proc_cell_vlservers_next,
 	.stop	= afs_proc_cell_vlservers_stop,
@@ -119,7 +119,7 @@ static void *afs_proc_cell_servers_next(struct seq_file *p, void *v,
 static void afs_proc_cell_servers_stop(struct seq_file *p, void *v);
 static int afs_proc_cell_servers_show(struct seq_file *m, void *v);
 
-static struct seq_operations afs_proc_cell_servers_ops = {
+static const struct seq_operations afs_proc_cell_servers_ops = {
 	.start	= afs_proc_cell_servers_start,
 	.next	= afs_proc_cell_servers_next,
 	.stop	= afs_proc_cell_servers_stop,
diff --git a/fs/aio.c b/fs/aio.c
index d065b2c3273e..02a2c9340573 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -24,6 +24,7 @@
 #include <linux/file.h>
 #include <linux/mm.h>
 #include <linux/mman.h>
+#include <linux/mmu_context.h>
 #include <linux/slab.h>
 #include <linux/timer.h>
 #include <linux/aio.h>
@@ -34,7 +35,6 @@
 
 #include <asm/kmap_types.h>
 #include <asm/uaccess.h>
-#include <asm/mmu_context.h>
 
 #if DEBUG > 1
 #define dprintk		printk
@@ -78,6 +78,7 @@ static int __init aio_setup(void)
 
 	return 0;
 }
+__initcall(aio_setup);
 
 static void aio_free_ring(struct kioctx *ctx)
 {
@@ -380,6 +381,7 @@ ssize_t wait_on_sync_kiocb(struct kiocb *iocb)
 	__set_current_state(TASK_RUNNING);
 	return iocb->ki_user_data;
 }
+EXPORT_SYMBOL(wait_on_sync_kiocb);
 
 /* exit_aio: called when the last user of mm goes away.  At this point, 
  * there is no way for any new requests to be submited or any of the 
@@ -573,6 +575,7 @@ int aio_put_req(struct kiocb *req)
 	spin_unlock_irq(&ctx->ctx_lock);
 	return ret;
 }
+EXPORT_SYMBOL(aio_put_req);
 
 static struct kioctx *lookup_ioctx(unsigned long ctx_id)
 {
@@ -595,51 +598,6 @@ static struct kioctx *lookup_ioctx(unsigned long ctx_id)
 }
 
 /*
- * use_mm
- *	Makes the calling kernel thread take on the specified
- *	mm context.
- *	Called by the retry thread execute retries within the
- *	iocb issuer's mm context, so that copy_from/to_user
- *	operations work seamlessly for aio.
- *	(Note: this routine is intended to be called only
- *	from a kernel thread context)
- */
-static void use_mm(struct mm_struct *mm)
-{
-	struct mm_struct *active_mm;
-	struct task_struct *tsk = current;
-
-	task_lock(tsk);
-	active_mm = tsk->active_mm;
-	atomic_inc(&mm->mm_count);
-	tsk->mm = mm;
-	tsk->active_mm = mm;
-	switch_mm(active_mm, mm, tsk);
-	task_unlock(tsk);
-
-	mmdrop(active_mm);
-}
-
-/*
- * unuse_mm
- *	Reverses the effect of use_mm, i.e. releases the
- *	specified mm context which was earlier taken on
- *	by the calling kernel thread
- *	(Note: this routine is intended to be called only
- *	from a kernel thread context)
- */
-static void unuse_mm(struct mm_struct *mm)
-{
-	struct task_struct *tsk = current;
-
-	task_lock(tsk);
-	tsk->mm = NULL;
-	/* active_mm is still 'mm' */
-	enter_lazy_tlb(mm, tsk);
-	task_unlock(tsk);
-}
-
-/*
  * Queue up a kiocb to be retried. Assumes that the kiocb
  * has already been marked as kicked, and places it on
  * the retry run list for the corresponding ioctx, if it
@@ -1037,6 +995,7 @@ put_rq:
 	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
 	return ret;
 }
+EXPORT_SYMBOL(aio_complete);
 
 /* aio_read_evt
  *	Pull an event off of the ioctx's event ring.  Returns the number of 
@@ -1825,9 +1784,3 @@ SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id,
 	asmlinkage_protect(5, ret, ctx_id, min_nr, nr, events, timeout);
 	return ret;
 }
-
-__initcall(aio_setup);
-
-EXPORT_SYMBOL(aio_complete);
-EXPORT_SYMBOL(aio_put_req);
-EXPORT_SYMBOL(wait_on_sync_kiocb);
diff --git a/fs/anon_inodes.c b/fs/anon_inodes.c
index 47d4a01c5393..2ca7a7cafdbf 100644
--- a/fs/anon_inodes.c
+++ b/fs/anon_inodes.c
@@ -8,8 +8,10 @@
  *
  */
 
+#include <linux/cred.h>
 #include <linux/file.h>
 #include <linux/poll.h>
+#include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/init.h>
 #include <linux/fs.h>
@@ -77,28 +79,24 @@ static const struct address_space_operations anon_aops = {
  *
  * Creates a new file by hooking it on a single inode. This is useful for files
  * that do not need to have a full-fledged inode in order to operate correctly.
- * All the files created with anon_inode_getfd() will share a single inode,
+ * All the files created with anon_inode_getfile() will share a single inode,
  * hence saving memory and avoiding code duplication for the file/inode/dentry
- * setup.  Returns new descriptor or -error.
+ * setup.  Returns the newly created file* or an error pointer.
  */
-int anon_inode_getfd(const char *name, const struct file_operations *fops,
-		     void *priv, int flags)
+struct file *anon_inode_getfile(const char *name,
+				const struct file_operations *fops,
+				void *priv, int flags)
 {
 	struct qstr this;
 	struct dentry *dentry;
 	struct file *file;
-	int error, fd;
+	int error;
 
 	if (IS_ERR(anon_inode_inode))
-		return -ENODEV;
+		return ERR_PTR(-ENODEV);
 
 	if (fops->owner && !try_module_get(fops->owner))
-		return -ENOENT;
-
-	error = get_unused_fd_flags(flags);
-	if (error < 0)
-		goto err_module;
-	fd = error;
+		return ERR_PTR(-ENOENT);
 
 	/*
 	 * Link the inode to a directory entry by creating a unique name
@@ -110,7 +108,7 @@ int anon_inode_getfd(const char *name, const struct file_operations *fops,
 	this.hash = 0;
 	dentry = d_alloc(anon_inode_mnt->mnt_sb->s_root, &this);
 	if (!dentry)
-		goto err_put_unused_fd;
+		goto err_module;
 
 	/*
 	 * We know the anon_inode inode count is always greater than zero,
@@ -136,16 +134,54 @@ int anon_inode_getfd(const char *name, const struct file_operations *fops,
 	file->f_version = 0;
 	file->private_data = priv;
 
+	return file;
+
+err_dput:
+	dput(dentry);
+err_module:
+	module_put(fops->owner);
+	return ERR_PTR(error);
+}
+EXPORT_SYMBOL_GPL(anon_inode_getfile);
+
+/**
+ * anon_inode_getfd - creates a new file instance by hooking it up to an
+ *                    anonymous inode, and a dentry that describe the "class"
+ *                    of the file
+ *
+ * @name:    [in]    name of the "class" of the new file
+ * @fops:    [in]    file operations for the new file
+ * @priv:    [in]    private data for the new file (will be file's private_data)
+ * @flags:   [in]    flags
+ *
+ * Creates a new file by hooking it on a single inode. This is useful for files
+ * that do not need to have a full-fledged inode in order to operate correctly.
+ * All the files created with anon_inode_getfd() will share a single inode,
+ * hence saving memory and avoiding code duplication for the file/inode/dentry
+ * setup.  Returns new descriptor or an error code.
+ */
+int anon_inode_getfd(const char *name, const struct file_operations *fops,
+		     void *priv, int flags)
+{
+	int error, fd;
+	struct file *file;
+
+	error = get_unused_fd_flags(flags);
+	if (error < 0)
+		return error;
+	fd = error;
+
+	file = anon_inode_getfile(name, fops, priv, flags);
+	if (IS_ERR(file)) {
+		error = PTR_ERR(file);
+		goto err_put_unused_fd;
+	}
 	fd_install(fd, file);
 
 	return fd;
 
-err_dput:
-	dput(dentry);
 err_put_unused_fd:
 	put_unused_fd(fd);
-err_module:
-	module_put(fops->owner);
 	return error;
 }
 EXPORT_SYMBOL_GPL(anon_inode_getfd);
diff --git a/fs/attr.c b/fs/attr.c
index 9fe1b1bd30a8..96d394bdaddf 100644
--- a/fs/attr.c
+++ b/fs/attr.c
@@ -18,7 +18,7 @@
 /* Taken over from the old code... */
 
 /* POSIX UID/GID verification for setting inode attributes. */
-int inode_change_ok(struct inode *inode, struct iattr *attr)
+int inode_change_ok(const struct inode *inode, struct iattr *attr)
 {
 	int retval = -EPERM;
 	unsigned int ia_valid = attr->ia_valid;
@@ -60,9 +60,51 @@ fine:
 error:
 	return retval;
 }
-
 EXPORT_SYMBOL(inode_change_ok);
 
+/**
+ * inode_newsize_ok - may this inode be truncated to a given size
+ * @inode:	the inode to be truncated
+ * @offset:	the new size to assign to the inode
+ * @Returns:	0 on success, -ve errno on failure
+ *
+ * inode_newsize_ok will check filesystem limits and ulimits to check that the
+ * new inode size is within limits. inode_newsize_ok will also send SIGXFSZ
+ * when necessary. Caller must not proceed with inode size change if failure is
+ * returned. @inode must be a file (not directory), with appropriate
+ * permissions to allow truncate (inode_newsize_ok does NOT check these
+ * conditions).
+ *
+ * inode_newsize_ok must be called with i_mutex held.
+ */
+int inode_newsize_ok(const struct inode *inode, loff_t offset)
+{
+	if (inode->i_size < offset) {
+		unsigned long limit;
+
+		limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
+		if (limit != RLIM_INFINITY && offset > limit)
+			goto out_sig;
+		if (offset > inode->i_sb->s_maxbytes)
+			goto out_big;
+	} else {
+		/*
+		 * truncation of in-use swapfiles is disallowed - it would
+		 * cause subsequent swapout to scribble on the now-freed
+		 * blocks.
+		 */
+		if (IS_SWAPFILE(inode))
+			return -ETXTBSY;
+	}
+
+	return 0;
+out_sig:
+	send_sig(SIGXFSZ, current, 0);
+out_big:
+	return -EFBIG;
+}
+EXPORT_SYMBOL(inode_newsize_ok);
+
 int inode_setattr(struct inode * inode, struct iattr * attr)
 {
 	unsigned int ia_valid = attr->ia_valid;
diff --git a/fs/autofs/dirhash.c b/fs/autofs/dirhash.c
index 2316e944a109..e947915109e5 100644
--- a/fs/autofs/dirhash.c
+++ b/fs/autofs/dirhash.c
@@ -90,7 +90,7 @@ struct autofs_dir_ent *autofs_expire(struct super_block *sb,
 			DPRINTK(("autofs: not expirable (not a mounted directory): %s\n", ent->name));
 			continue;
 		}
-		while (d_mountpoint(path.dentry) && follow_down(&path));
+		while (d_mountpoint(path.dentry) && follow_down(&path))
 			;
 		umount_ok = may_umount(path.mnt);
 		path_put(&path);
diff --git a/fs/befs/linuxvfs.c b/fs/befs/linuxvfs.c
index 615d5496fe0f..33baf27fac78 100644
--- a/fs/befs/linuxvfs.c
+++ b/fs/befs/linuxvfs.c
@@ -737,12 +737,7 @@ befs_put_super(struct super_block *sb)
 {
 	kfree(BEFS_SB(sb)->mount_opts.iocharset);
 	BEFS_SB(sb)->mount_opts.iocharset = NULL;
-
-	if (BEFS_SB(sb)->nls) {
-		unload_nls(BEFS_SB(sb)->nls);
-		BEFS_SB(sb)->nls = NULL;
-	}
-
+	unload_nls(BEFS_SB(sb)->nls);
 	kfree(sb->s_fs_info);
 	sb->s_fs_info = NULL;
 }
@@ -842,7 +837,7 @@ befs_fill_super(struct super_block *sb, void *data, int silent)
 	sb->s_magic = BEFS_SUPER_MAGIC;
 	/* Set real blocksize of fs */
 	sb_set_blocksize(sb, (ulong) befs_sb->block_size);
-	sb->s_op = (struct super_operations *) &befs_sops;
+	sb->s_op = &befs_sops;
 	root = befs_iget(sb, iaddr2blockno(sb, &(befs_sb->root_dir)));
 	if (IS_ERR(root)) {
 		ret = PTR_ERR(root);
diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index 7c1e65d54872..b9b3bb51b1e4 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -1280,9 +1280,6 @@ static int writenote(struct memelfnote *men, struct file *file,
 #define DUMP_WRITE(addr, nr)	\
 	if ((size += (nr)) > limit || !dump_write(file, (addr), (nr))) \
 		goto end_coredump;
-#define DUMP_SEEK(off)	\
-	if (!dump_seek(file, (off))) \
-		goto end_coredump;
 
 static void fill_elf_header(struct elfhdr *elf, int segs,
 			    u16 machine, u32 flags, u8 osabi)
@@ -1714,42 +1711,52 @@ struct elf_note_info {
 	int numnote;
 };
 
-static int fill_note_info(struct elfhdr *elf, int phdrs,
-			  struct elf_note_info *info,
-			  long signr, struct pt_regs *regs)
+static int elf_note_info_init(struct elf_note_info *info)
 {
-#define	NUM_NOTES	6
-	struct list_head *t;
-
-	info->notes = NULL;
-	info->prstatus = NULL;
-	info->psinfo = NULL;
-	info->fpu = NULL;
-#ifdef ELF_CORE_COPY_XFPREGS
-	info->xfpu = NULL;
-#endif
+	memset(info, 0, sizeof(*info));
 	INIT_LIST_HEAD(&info->thread_list);
 
-	info->notes = kmalloc(NUM_NOTES * sizeof(struct memelfnote),
-			      GFP_KERNEL);
+	/* Allocate space for six ELF notes */
+	info->notes = kmalloc(6 * sizeof(struct memelfnote), GFP_KERNEL);
 	if (!info->notes)
 		return 0;
 	info->psinfo = kmalloc(sizeof(*info->psinfo), GFP_KERNEL);
 	if (!info->psinfo)
-		return 0;
+		goto notes_free;
 	info->prstatus = kmalloc(sizeof(*info->prstatus), GFP_KERNEL);
 	if (!info->prstatus)
-		return 0;
+		goto psinfo_free;
 	info->fpu = kmalloc(sizeof(*info->fpu), GFP_KERNEL);
 	if (!info->fpu)
-		return 0;
+		goto prstatus_free;
 #ifdef ELF_CORE_COPY_XFPREGS
 	info->xfpu = kmalloc(sizeof(*info->xfpu), GFP_KERNEL);
 	if (!info->xfpu)
-		return 0;
+		goto fpu_free;
+#endif
+	return 1;
+#ifdef ELF_CORE_COPY_XFPREGS
+ fpu_free:
+	kfree(info->fpu);
 #endif
+ prstatus_free:
+	kfree(info->prstatus);
+ psinfo_free:
+	kfree(info->psinfo);
+ notes_free:
+	kfree(info->notes);
+	return 0;
+}
+
+static int fill_note_info(struct elfhdr *elf, int phdrs,
+			  struct elf_note_info *info,
+			  long signr, struct pt_regs *regs)
+{
+	struct list_head *t;
+
+	if (!elf_note_info_init(info))
+		return 0;
 
-	info->thread_status_size = 0;
 	if (signr) {
 		struct core_thread *ct;
 		struct elf_thread_status *ets;
@@ -1809,8 +1816,6 @@ static int fill_note_info(struct elfhdr *elf, int phdrs,
 #endif
 
 	return 1;
-
-#undef NUM_NOTES
 }
 
 static size_t get_note_info_size(struct elf_note_info *info)
@@ -2016,7 +2021,8 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file, un
 		goto end_coredump;
 
 	/* Align to page */
-	DUMP_SEEK(dataoff - foffset);
+	if (!dump_seek(file, dataoff - foffset))
+		goto end_coredump;
 
 	for (vma = first_vma(current, gate_vma); vma != NULL;
 			vma = next_vma(vma, gate_vma)) {
@@ -2027,33 +2033,19 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file, un
 
 		for (addr = vma->vm_start; addr < end; addr += PAGE_SIZE) {
 			struct page *page;
-			struct vm_area_struct *tmp_vma;
-
-			if (get_user_pages(current, current->mm, addr, 1, 0, 1,
-						&page, &tmp_vma) <= 0) {
-				DUMP_SEEK(PAGE_SIZE);
-			} else {
-				if (page == ZERO_PAGE(0)) {
-					if (!dump_seek(file, PAGE_SIZE)) {
-						page_cache_release(page);
-						goto end_coredump;
-					}
-				} else {
-					void *kaddr;
-					flush_cache_page(tmp_vma, addr,
-							 page_to_pfn(page));
-					kaddr = kmap(page);
-					if ((size += PAGE_SIZE) > limit ||
-					    !dump_write(file, kaddr,
-					    PAGE_SIZE)) {
-						kunmap(page);
-						page_cache_release(page);
-						goto end_coredump;
-					}
-					kunmap(page);
-				}
+			int stop;
+
+			page = get_dump_page(addr);
+			if (page) {
+				void *kaddr = kmap(page);
+				stop = ((size += PAGE_SIZE) > limit) ||
+					!dump_write(file, kaddr, PAGE_SIZE);
+				kunmap(page);
 				page_cache_release(page);
-			}
+			} else
+				stop = !dump_seek(file, PAGE_SIZE);
+			if (stop)
+				goto end_coredump;
 		}
 	}
 
diff --git a/fs/binfmt_elf_fdpic.c b/fs/binfmt_elf_fdpic.c
index 20fbeced472b..38502c67987c 100644
--- a/fs/binfmt_elf_fdpic.c
+++ b/fs/binfmt_elf_fdpic.c
@@ -283,20 +283,23 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
 	}
 
 	stack_size = exec_params.stack_size;
-	if (stack_size < interp_params.stack_size)
-		stack_size = interp_params.stack_size;
-
 	if (exec_params.flags & ELF_FDPIC_FLAG_EXEC_STACK)
 		executable_stack = EXSTACK_ENABLE_X;
 	else if (exec_params.flags & ELF_FDPIC_FLAG_NOEXEC_STACK)
 		executable_stack = EXSTACK_DISABLE_X;
-	else if (interp_params.flags & ELF_FDPIC_FLAG_EXEC_STACK)
-		executable_stack = EXSTACK_ENABLE_X;
-	else if (interp_params.flags & ELF_FDPIC_FLAG_NOEXEC_STACK)
-		executable_stack = EXSTACK_DISABLE_X;
 	else
 		executable_stack = EXSTACK_DEFAULT;
 
+	if (stack_size == 0) {
+		stack_size = interp_params.stack_size;
+		if (interp_params.flags & ELF_FDPIC_FLAG_EXEC_STACK)
+			executable_stack = EXSTACK_ENABLE_X;
+		else if (interp_params.flags & ELF_FDPIC_FLAG_NOEXEC_STACK)
+			executable_stack = EXSTACK_DISABLE_X;
+		else
+			executable_stack = EXSTACK_DEFAULT;
+	}
+
 	retval = -ENOEXEC;
 	if (stack_size == 0)
 		goto error;
@@ -1325,9 +1328,6 @@ static int writenote(struct memelfnote *men, struct file *file)
 #define DUMP_WRITE(addr, nr)	\
 	if ((size += (nr)) > limit || !dump_write(file, (addr), (nr))) \
 		goto end_coredump;
-#define DUMP_SEEK(off)	\
-	if (!dump_seek(file, (off))) \
-		goto end_coredump;
 
 static inline void fill_elf_fdpic_header(struct elfhdr *elf, int segs)
 {
@@ -1518,6 +1518,7 @@ static int elf_fdpic_dump_segments(struct file *file, size_t *size,
 			   unsigned long *limit, unsigned long mm_flags)
 {
 	struct vm_area_struct *vma;
+	int err = 0;
 
 	for (vma = current->mm->mmap; vma; vma = vma->vm_next) {
 		unsigned long addr;
@@ -1525,43 +1526,26 @@ static int elf_fdpic_dump_segments(struct file *file, size_t *size,
 		if (!maydump(vma, mm_flags))
 			continue;
 
-		for (addr = vma->vm_start;
-		     addr < vma->vm_end;
-		     addr += PAGE_SIZE
-		     ) {
-			struct vm_area_struct *vma;
-			struct page *page;
-
-			if (get_user_pages(current, current->mm, addr, 1, 0, 1,
-					   &page, &vma) <= 0) {
-				DUMP_SEEK(file->f_pos + PAGE_SIZE);
-			}
-			else if (page == ZERO_PAGE(0)) {
-				page_cache_release(page);
-				DUMP_SEEK(file->f_pos + PAGE_SIZE);
-			}
-			else {
-				void *kaddr;
-
-				flush_cache_page(vma, addr, page_to_pfn(page));
-				kaddr = kmap(page);
-				if ((*size += PAGE_SIZE) > *limit ||
-				    !dump_write(file, kaddr, PAGE_SIZE)
-				    ) {
-					kunmap(page);
-					page_cache_release(page);
-					return -EIO;
-				}
+		for (addr = vma->vm_start; addr < vma->vm_end;
+							addr += PAGE_SIZE) {
+			struct page *page = get_dump_page(addr);
+			if (page) {
+				void *kaddr = kmap(page);
+				*size += PAGE_SIZE;
+				if (*size > *limit)
+					err = -EFBIG;
+				else if (!dump_write(file, kaddr, PAGE_SIZE))
+					err = -EIO;
 				kunmap(page);
 				page_cache_release(page);
-			}
+			} else if (!dump_seek(file, file->f_pos + PAGE_SIZE))
+				err = -EFBIG;
+			if (err)
+				goto out;
 		}
 	}
-
-	return 0;
-
-end_coredump:
-	return -EFBIG;
+out:
+	return err;
 }
 #endif
 
@@ -1802,7 +1786,8 @@ static int elf_fdpic_core_dump(long signr, struct pt_regs *regs,
 				goto end_coredump;
 	}
 
-	DUMP_SEEK(dataoff);
+	if (!dump_seek(file, dataoff))
+		goto end_coredump;
 
 	if (elf_fdpic_dump_segments(file, &size, &limit, mm_flags) < 0)
 		goto end_coredump;
diff --git a/fs/binfmt_flat.c b/fs/binfmt_flat.c
index e92f229e3c6e..a2796651e756 100644
--- a/fs/binfmt_flat.c
+++ b/fs/binfmt_flat.c
@@ -278,8 +278,6 @@ static int decompress_exec(
 		ret = bprm->file->f_op->read(bprm->file, buf, LBUFSIZE, &fpos);
 		if (ret <= 0)
 			break;
-		if (ret >= (unsigned long) -4096)
-			break;
 		len -= ret;
 
 		strm.next_in = buf;
@@ -335,7 +333,7 @@ calc_reloc(unsigned long r, struct lib_info *p, int curid, int internalp)
 					"(%d != %d)", (unsigned) r, curid, id);
 			goto failed;
 		} else if ( ! p->lib_list[id].loaded &&
-				load_flat_shared_library(id, p) > (unsigned long) -4096) {
+				IS_ERR_VALUE(load_flat_shared_library(id, p))) {
 			printk("BINFMT_FLAT: failed to load library %d", id);
 			goto failed;
 		}
@@ -545,7 +543,7 @@ static int load_flat_file(struct linux_binprm * bprm,
 		textpos = do_mmap(bprm->file, 0, text_len, PROT_READ|PROT_EXEC,
 				  MAP_PRIVATE|MAP_EXECUTABLE, 0);
 		up_write(&current->mm->mmap_sem);
-		if (!textpos  || textpos >= (unsigned long) -4096) {
+		if (!textpos || IS_ERR_VALUE(textpos)) {
 			if (!textpos)
 				textpos = (unsigned long) -ENOMEM;
 			printk("Unable to mmap process text, errno %d\n", (int)-textpos);
@@ -560,7 +558,7 @@ static int load_flat_file(struct linux_binprm * bprm,
 			PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE, 0);
 		up_write(&current->mm->mmap_sem);
 
-		if (realdatastart == 0 || realdatastart >= (unsigned long)-4096) {
+		if (realdatastart == 0 || IS_ERR_VALUE(realdatastart)) {
 			if (!realdatastart)
 				realdatastart = (unsigned long) -ENOMEM;
 			printk("Unable to allocate RAM for process data, errno %d\n",
@@ -587,7 +585,7 @@ static int load_flat_file(struct linux_binprm * bprm,
 			result = bprm->file->f_op->read(bprm->file, (char *) datapos,
 					data_len + (relocs * sizeof(unsigned long)), &fpos);
 		}
-		if (result >= (unsigned long)-4096) {
+		if (IS_ERR_VALUE(result)) {
 			printk("Unable to read data+bss, errno %d\n", (int)-result);
 			do_munmap(current->mm, textpos, text_len);
 			do_munmap(current->mm, realdatastart, data_len + extra);
@@ -607,7 +605,7 @@ static int load_flat_file(struct linux_binprm * bprm,
 			PROT_READ | PROT_EXEC | PROT_WRITE, MAP_PRIVATE, 0);
 		up_write(&current->mm->mmap_sem);
 
-		if (!textpos  || textpos >= (unsigned long) -4096) {
+		if (!textpos || IS_ERR_VALUE(textpos)) {
 			if (!textpos)
 				textpos = (unsigned long) -ENOMEM;
 			printk("Unable to allocate RAM for process text/data, errno %d\n",
@@ -641,7 +639,7 @@ static int load_flat_file(struct linux_binprm * bprm,
 			fpos = 0;
 			result = bprm->file->f_op->read(bprm->file,
 					(char *) textpos, text_len, &fpos);
-			if (result < (unsigned long) -4096)
+			if (!IS_ERR_VALUE(result))
 				result = decompress_exec(bprm, text_len, (char *) datapos,
 						 data_len + (relocs * sizeof(unsigned long)), 0);
 		}
@@ -651,13 +649,13 @@ static int load_flat_file(struct linux_binprm * bprm,
 			fpos = 0;
 			result = bprm->file->f_op->read(bprm->file,
 					(char *) textpos, text_len, &fpos);
-			if (result < (unsigned long) -4096) {
+			if (!IS_ERR_VALUE(result)) {
 				fpos = ntohl(hdr->data_start);
 				result = bprm->file->f_op->read(bprm->file, (char *) datapos,
 					data_len + (relocs * sizeof(unsigned long)), &fpos);
 			}
 		}
-		if (result >= (unsigned long)-4096) {
+		if (IS_ERR_VALUE(result)) {
 			printk("Unable to read code+data+bss, errno %d\n",(int)-result);
 			do_munmap(current->mm, textpos, text_len + data_len + extra +
 				MAX_SHARED_LIBS * sizeof(unsigned long));
@@ -835,7 +833,7 @@ static int load_flat_shared_library(int id, struct lib_info *libs)
 
 	res = prepare_binprm(&bprm);
 
-	if (res <= (unsigned long)-4096)
+	if (!IS_ERR_VALUE(res))
 		res = load_flat_file(&bprm, libs, id, NULL);
 
 	abort_creds(bprm.cred);
@@ -880,7 +878,7 @@ static int load_flat_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 	stack_len += FLAT_DATA_ALIGN - 1;  /* reserve for upcoming alignment */
 	
 	res = load_flat_file(bprm, &libinfo, 0, &stack_len);
-	if (res > (unsigned long)-4096)
+	if (IS_ERR_VALUE(res))
 		return res;
 	
 	/* Update data segment pointers for all libraries */
diff --git a/fs/bio.c b/fs/bio.c
index 76738005c8e8..402cb84a92a1 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -249,6 +249,7 @@ void bio_free(struct bio *bio, struct bio_set *bs)
 
 	mempool_free(p, bs->bio_pool);
 }
+EXPORT_SYMBOL(bio_free);
 
 void bio_init(struct bio *bio)
 {
@@ -257,6 +258,7 @@ void bio_init(struct bio *bio)
 	bio->bi_comp_cpu = -1;
 	atomic_set(&bio->bi_cnt, 1);
 }
+EXPORT_SYMBOL(bio_init);
 
 /**
  * bio_alloc_bioset - allocate a bio for I/O
@@ -311,6 +313,7 @@ err_free:
 	mempool_free(p, bs->bio_pool);
 	return NULL;
 }
+EXPORT_SYMBOL(bio_alloc_bioset);
 
 static void bio_fs_destructor(struct bio *bio)
 {
@@ -337,6 +340,7 @@ struct bio *bio_alloc(gfp_t gfp_mask, int nr_iovecs)
 
 	return bio;
 }
+EXPORT_SYMBOL(bio_alloc);
 
 static void bio_kmalloc_destructor(struct bio *bio)
 {
@@ -380,6 +384,7 @@ struct bio *bio_kmalloc(gfp_t gfp_mask, int nr_iovecs)
 
 	return bio;
 }
+EXPORT_SYMBOL(bio_kmalloc);
 
 void zero_fill_bio(struct bio *bio)
 {
@@ -416,6 +421,7 @@ void bio_put(struct bio *bio)
 		bio->bi_destructor(bio);
 	}
 }
+EXPORT_SYMBOL(bio_put);
 
 inline int bio_phys_segments(struct request_queue *q, struct bio *bio)
 {
@@ -424,6 +430,7 @@ inline int bio_phys_segments(struct request_queue *q, struct bio *bio)
 
 	return bio->bi_phys_segments;
 }
+EXPORT_SYMBOL(bio_phys_segments);
 
 /**
  * 	__bio_clone	-	clone a bio
@@ -451,6 +458,7 @@ void __bio_clone(struct bio *bio, struct bio *bio_src)
 	bio->bi_size = bio_src->bi_size;
 	bio->bi_idx = bio_src->bi_idx;
 }
+EXPORT_SYMBOL(__bio_clone);
 
 /**
  *	bio_clone	-	clone a bio
@@ -482,6 +490,7 @@ struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
 
 	return b;
 }
+EXPORT_SYMBOL(bio_clone);
 
 /**
  *	bio_get_nr_vecs		- return approx number of vecs
@@ -505,6 +514,7 @@ int bio_get_nr_vecs(struct block_device *bdev)
 
 	return nr_pages;
 }
+EXPORT_SYMBOL(bio_get_nr_vecs);
 
 static int __bio_add_page(struct request_queue *q, struct bio *bio, struct page
 			  *page, unsigned int len, unsigned int offset,
@@ -635,6 +645,7 @@ int bio_add_pc_page(struct request_queue *q, struct bio *bio, struct page *page,
 	return __bio_add_page(q, bio, page, len, offset,
 			      queue_max_hw_sectors(q));
 }
+EXPORT_SYMBOL(bio_add_pc_page);
 
 /**
  *	bio_add_page	-	attempt to add page to bio
@@ -655,6 +666,7 @@ int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 	return __bio_add_page(q, bio, page, len, offset, queue_max_sectors(q));
 }
+EXPORT_SYMBOL(bio_add_page);
 
 struct bio_map_data {
 	struct bio_vec *iovecs;
@@ -776,6 +788,7 @@ int bio_uncopy_user(struct bio *bio)
 	bio_put(bio);
 	return ret;
 }
+EXPORT_SYMBOL(bio_uncopy_user);
 
 /**
  *	bio_copy_user_iov	-	copy user data to bio
@@ -920,6 +933,7 @@ struct bio *bio_copy_user(struct request_queue *q, struct rq_map_data *map_data,
 
 	return bio_copy_user_iov(q, map_data, &iov, 1, write_to_vm, gfp_mask);
 }
+EXPORT_SYMBOL(bio_copy_user);
 
 static struct bio *__bio_map_user_iov(struct request_queue *q,
 				      struct block_device *bdev,
@@ -1050,6 +1064,7 @@ struct bio *bio_map_user(struct request_queue *q, struct block_device *bdev,
 
 	return bio_map_user_iov(q, bdev, &iov, 1, write_to_vm, gfp_mask);
 }
+EXPORT_SYMBOL(bio_map_user);
 
 /**
  *	bio_map_user_iov - map user sg_iovec table into bio
@@ -1117,13 +1132,13 @@ void bio_unmap_user(struct bio *bio)
 	__bio_unmap_user(bio);
 	bio_put(bio);
 }
+EXPORT_SYMBOL(bio_unmap_user);
 
 static void bio_map_kern_endio(struct bio *bio, int err)
 {
 	bio_put(bio);
 }
 
-
 static struct bio *__bio_map_kern(struct request_queue *q, void *data,
 				  unsigned int len, gfp_t gfp_mask)
 {
@@ -1189,6 +1204,7 @@ struct bio *bio_map_kern(struct request_queue *q, void *data, unsigned int len,
 	bio_put(bio);
 	return ERR_PTR(-EINVAL);
 }
+EXPORT_SYMBOL(bio_map_kern);
 
 static void bio_copy_kern_endio(struct bio *bio, int err)
 {
@@ -1250,6 +1266,7 @@ struct bio *bio_copy_kern(struct request_queue *q, void *data, unsigned int len,
 
 	return bio;
 }
+EXPORT_SYMBOL(bio_copy_kern);
 
 /*
  * bio_set_pages_dirty() and bio_check_pages_dirty() are support functions
@@ -1400,6 +1417,7 @@ void bio_endio(struct bio *bio, int error)
 	if (bio->bi_end_io)
 		bio->bi_end_io(bio, error);
 }
+EXPORT_SYMBOL(bio_endio);
 
 void bio_pair_release(struct bio_pair *bp)
 {
@@ -1410,6 +1428,7 @@ void bio_pair_release(struct bio_pair *bp)
 		mempool_free(bp, bp->bio2.bi_private);
 	}
 }
+EXPORT_SYMBOL(bio_pair_release);
 
 static void bio_pair_end_1(struct bio *bi, int err)
 {
@@ -1477,6 +1496,7 @@ struct bio_pair *bio_split(struct bio *bi, int first_sectors)
 
 	return bp;
 }
+EXPORT_SYMBOL(bio_split);
 
 /**
  *      bio_sector_offset - Find hardware sector offset in bio
@@ -1547,6 +1567,7 @@ void bioset_free(struct bio_set *bs)
 
 	kfree(bs);
 }
+EXPORT_SYMBOL(bioset_free);
 
 /**
  * bioset_create  - Create a bio_set
@@ -1592,6 +1613,7 @@ bad:
 	bioset_free(bs);
 	return NULL;
 }
+EXPORT_SYMBOL(bioset_create);
 
 static void __init biovec_init_slabs(void)
 {
@@ -1636,29 +1658,4 @@ static int __init init_bio(void)
 
 	return 0;
 }
-
 subsys_initcall(init_bio);
-
-EXPORT_SYMBOL(bio_alloc);
-EXPORT_SYMBOL(bio_kmalloc);
-EXPORT_SYMBOL(bio_put);
-EXPORT_SYMBOL(bio_free);
-EXPORT_SYMBOL(bio_endio);
-EXPORT_SYMBOL(bio_init);
-EXPORT_SYMBOL(__bio_clone);
-EXPORT_SYMBOL(bio_clone);
-EXPORT_SYMBOL(bio_phys_segments);
-EXPORT_SYMBOL(bio_add_page);
-EXPORT_SYMBOL(bio_add_pc_page);
-EXPORT_SYMBOL(bio_get_nr_vecs);
-EXPORT_SYMBOL(bio_map_user);
-EXPORT_SYMBOL(bio_unmap_user);
-EXPORT_SYMBOL(bio_map_kern);
-EXPORT_SYMBOL(bio_copy_kern);
-EXPORT_SYMBOL(bio_pair_release);
-EXPORT_SYMBOL(bio_split);
-EXPORT_SYMBOL(bio_copy_user);
-EXPORT_SYMBOL(bio_uncopy_user);
-EXPORT_SYMBOL(bioset_create);
-EXPORT_SYMBOL(bioset_free);
-EXPORT_SYMBOL(bio_alloc_bioset);
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 71e7e03ac343..9cf4b926f8e4 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -216,8 +216,6 @@ EXPORT_SYMBOL(fsync_bdev);
  * freeze_bdev  --  lock a filesystem and force it into a consistent state
  * @bdev:	blockdevice to lock
  *
- * This takes the block device bd_mount_sem to make sure no new mounts
- * happen on bdev until thaw_bdev() is called.
  * If a superblock is found on this device, we take the s_umount semaphore
  * on it to make sure nobody unmounts until the snapshot creation is done.
  * The reference counter (bd_fsfreeze_count) guarantees that only the last
@@ -232,46 +230,55 @@ struct super_block *freeze_bdev(struct block_device *bdev)
 	int error = 0;
 
 	mutex_lock(&bdev->bd_fsfreeze_mutex);
-	if (bdev->bd_fsfreeze_count > 0) {
-		bdev->bd_fsfreeze_count++;
+	if (++bdev->bd_fsfreeze_count > 1) {
+		/*
+		 * We don't even need to grab a reference - the first call
+		 * to freeze_bdev grab an active reference and only the last
+		 * thaw_bdev drops it.
+		 */
 		sb = get_super(bdev);
+		drop_super(sb);
 		mutex_unlock(&bdev->bd_fsfreeze_mutex);
 		return sb;
 	}
-	bdev->bd_fsfreeze_count++;
-
-	down(&bdev->bd_mount_sem);
-	sb = get_super(bdev);
-	if (sb && !(sb->s_flags & MS_RDONLY)) {
-		sb->s_frozen = SB_FREEZE_WRITE;
-		smp_wmb();
-
-		sync_filesystem(sb);
-
-		sb->s_frozen = SB_FREEZE_TRANS;
-		smp_wmb();
-
-		sync_blockdev(sb->s_bdev);
-
-		if (sb->s_op->freeze_fs) {
-			error = sb->s_op->freeze_fs(sb);
-			if (error) {
-				printk(KERN_ERR
-					"VFS:Filesystem freeze failed\n");
-				sb->s_frozen = SB_UNFROZEN;
-				drop_super(sb);
-				up(&bdev->bd_mount_sem);
-				bdev->bd_fsfreeze_count--;
-				mutex_unlock(&bdev->bd_fsfreeze_mutex);
-				return ERR_PTR(error);
-			}
+
+	sb = get_active_super(bdev);
+	if (!sb)
+		goto out;
+	if (sb->s_flags & MS_RDONLY) {
+		deactivate_locked_super(sb);
+		mutex_unlock(&bdev->bd_fsfreeze_mutex);
+		return sb;
+	}
+
+	sb->s_frozen = SB_FREEZE_WRITE;
+	smp_wmb();
+
+	sync_filesystem(sb);
+
+	sb->s_frozen = SB_FREEZE_TRANS;
+	smp_wmb();
+
+	sync_blockdev(sb->s_bdev);
+
+	if (sb->s_op->freeze_fs) {
+		error = sb->s_op->freeze_fs(sb);
+		if (error) {
+			printk(KERN_ERR
+				"VFS:Filesystem freeze failed\n");
+			sb->s_frozen = SB_UNFROZEN;
+			deactivate_locked_super(sb);
+			bdev->bd_fsfreeze_count--;
+			mutex_unlock(&bdev->bd_fsfreeze_mutex);
+			return ERR_PTR(error);
 		}
 	}
+	up_write(&sb->s_umount);
 
+ out:
 	sync_blockdev(bdev);
 	mutex_unlock(&bdev->bd_fsfreeze_mutex);
-
-	return sb;	/* thaw_bdev releases s->s_umount and bd_mount_sem */
+	return sb;	/* thaw_bdev releases s->s_umount */
 }
 EXPORT_SYMBOL(freeze_bdev);
 
@@ -284,44 +291,44 @@ EXPORT_SYMBOL(freeze_bdev);
  */
 int thaw_bdev(struct block_device *bdev, struct super_block *sb)
 {
-	int error = 0;
+	int error = -EINVAL;
 
 	mutex_lock(&bdev->bd_fsfreeze_mutex);
-	if (!bdev->bd_fsfreeze_count) {
-		mutex_unlock(&bdev->bd_fsfreeze_mutex);
-		return -EINVAL;
-	}
-
-	bdev->bd_fsfreeze_count--;
-	if (bdev->bd_fsfreeze_count > 0) {
-		if (sb)
-			drop_super(sb);
-		mutex_unlock(&bdev->bd_fsfreeze_mutex);
-		return 0;
-	}
-
-	if (sb) {
-		BUG_ON(sb->s_bdev != bdev);
-		if (!(sb->s_flags & MS_RDONLY)) {
-			if (sb->s_op->unfreeze_fs) {
-				error = sb->s_op->unfreeze_fs(sb);
-				if (error) {
-					printk(KERN_ERR
-						"VFS:Filesystem thaw failed\n");
-					sb->s_frozen = SB_FREEZE_TRANS;
-					bdev->bd_fsfreeze_count++;
-					mutex_unlock(&bdev->bd_fsfreeze_mutex);
-					return error;
-				}
-			}
-			sb->s_frozen = SB_UNFROZEN;
-			smp_wmb();
-			wake_up(&sb->s_wait_unfrozen);
+	if (!bdev->bd_fsfreeze_count)
+		goto out_unlock;
+
+	error = 0;
+	if (--bdev->bd_fsfreeze_count > 0)
+		goto out_unlock;
+
+	if (!sb)
+		goto out_unlock;
+
+	BUG_ON(sb->s_bdev != bdev);
+	down_write(&sb->s_umount);
+	if (sb->s_flags & MS_RDONLY)
+		goto out_deactivate;
+
+	if (sb->s_op->unfreeze_fs) {
+		error = sb->s_op->unfreeze_fs(sb);
+		if (error) {
+			printk(KERN_ERR
+				"VFS:Filesystem thaw failed\n");
+			sb->s_frozen = SB_FREEZE_TRANS;
+			bdev->bd_fsfreeze_count++;
+			mutex_unlock(&bdev->bd_fsfreeze_mutex);
+			return error;
 		}
-		drop_super(sb);
 	}
 
-	up(&bdev->bd_mount_sem);
+	sb->s_frozen = SB_UNFROZEN;
+	smp_wmb();
+	wake_up(&sb->s_wait_unfrozen);
+
+out_deactivate:
+	if (sb)
+		deactivate_locked_super(sb);
+out_unlock:
 	mutex_unlock(&bdev->bd_fsfreeze_mutex);
 	return 0;
 }
@@ -430,7 +437,6 @@ static void init_once(void *foo)
 
 	memset(bdev, 0, sizeof(*bdev));
 	mutex_init(&bdev->bd_mutex);
-	sema_init(&bdev->bd_mount_sem, 1);
 	INIT_LIST_HEAD(&bdev->bd_inodes);
 	INIT_LIST_HEAD(&bdev->bd_list);
 #ifdef CONFIG_SYSFS
@@ -1114,7 +1120,7 @@ EXPORT_SYMBOL(revalidate_disk);
 int check_disk_change(struct block_device *bdev)
 {
 	struct gendisk *disk = bdev->bd_disk;
-	struct block_device_operations * bdops = disk->fops;
+	const struct block_device_operations *bdops = disk->fops;
 
 	if (!bdops->media_changed)
 		return 0;
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index f128427b995b..361604244271 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -27,7 +27,7 @@
 #include "btrfs_inode.h"
 #include "xattr.h"
 
-#ifdef CONFIG_FS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 
 static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
 {
@@ -313,7 +313,7 @@ struct xattr_handler btrfs_xattr_acl_access_handler = {
 	.set	= btrfs_xattr_acl_access_set,
 };
 
-#else /* CONFIG_FS_POSIX_ACL */
+#else /* CONFIG_BTRFS_FS_POSIX_ACL */
 
 int btrfs_acl_chmod(struct inode *inode)
 {
@@ -325,4 +325,4 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir)
 	return 0;
 }
 
-#endif /* CONFIG_FS_POSIX_ACL */
+#endif /* CONFIG_BTRFS_FS_POSIX_ACL */
diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index 019e8af449ab..c0861e781cdb 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -48,6 +48,9 @@ struct btrfs_worker_thread {
 	/* number of things on the pending list */
 	atomic_t num_pending;
 
+	/* reference counter for this struct */
+	atomic_t refs;
+
 	unsigned long sequence;
 
 	/* protects the pending list. */
@@ -61,6 +64,51 @@ struct btrfs_worker_thread {
 };
 
 /*
+ * btrfs_start_workers uses kthread_run, which can block waiting for memory
+ * for a very long time.  It will actually throttle on page writeback,
+ * and so it may not make progress until after our btrfs worker threads
+ * process all of the pending work structs in their queue
+ *
+ * This means we can't use btrfs_start_workers from inside a btrfs worker
+ * thread that is used as part of cleaning dirty memory, which pretty much
+ * involves all of the worker threads.
+ *
+ * Instead we have a helper queue who never has more than one thread
+ * where we scheduler thread start operations.  This worker_start struct
+ * is used to contain the work and hold a pointer to the queue that needs
+ * another worker.
+ */
+struct worker_start {
+	struct btrfs_work work;
+	struct btrfs_workers *queue;
+};
+
+static void start_new_worker_func(struct btrfs_work *work)
+{
+	struct worker_start *start;
+	start = container_of(work, struct worker_start, work);
+	btrfs_start_workers(start->queue, 1);
+	kfree(start);
+}
+
+static int start_new_worker(struct btrfs_workers *queue)
+{
+	struct worker_start *start;
+	int ret;
+
+	start = kzalloc(sizeof(*start), GFP_NOFS);
+	if (!start)
+		return -ENOMEM;
+
+	start->work.func = start_new_worker_func;
+	start->queue = queue;
+	ret = btrfs_queue_worker(queue->atomic_worker_start, &start->work);
+	if (ret)
+		kfree(start);
+	return ret;
+}
+
+/*
  * helper function to move a thread onto the idle list after it
  * has finished some requests.
  */
@@ -71,7 +119,12 @@ static void check_idle_worker(struct btrfs_worker_thread *worker)
 		unsigned long flags;
 		spin_lock_irqsave(&worker->workers->lock, flags);
 		worker->idle = 1;
-		list_move(&worker->worker_list, &worker->workers->idle_list);
+
+		/* the list may be empty if the worker is just starting */
+		if (!list_empty(&worker->worker_list)) {
+			list_move(&worker->worker_list,
+				 &worker->workers->idle_list);
+		}
 		spin_unlock_irqrestore(&worker->workers->lock, flags);
 	}
 }
@@ -87,23 +140,51 @@ static void check_busy_worker(struct btrfs_worker_thread *worker)
 		unsigned long flags;
 		spin_lock_irqsave(&worker->workers->lock, flags);
 		worker->idle = 0;
-		list_move_tail(&worker->worker_list,
-			       &worker->workers->worker_list);
+
+		if (!list_empty(&worker->worker_list)) {
+			list_move_tail(&worker->worker_list,
+				      &worker->workers->worker_list);
+		}
 		spin_unlock_irqrestore(&worker->workers->lock, flags);
 	}
 }
 
-static noinline int run_ordered_completions(struct btrfs_workers *workers,
-					    struct btrfs_work *work)
+static void check_pending_worker_creates(struct btrfs_worker_thread *worker)
 {
+	struct btrfs_workers *workers = worker->workers;
 	unsigned long flags;
 
+	rmb();
+	if (!workers->atomic_start_pending)
+		return;
+
+	spin_lock_irqsave(&workers->lock, flags);
+	if (!workers->atomic_start_pending)
+		goto out;
+
+	workers->atomic_start_pending = 0;
+	if (workers->num_workers + workers->num_workers_starting >=
+	    workers->max_workers)
+		goto out;
+
+	workers->num_workers_starting += 1;
+	spin_unlock_irqrestore(&workers->lock, flags);
+	start_new_worker(workers);
+	return;
+
+out:
+	spin_unlock_irqrestore(&workers->lock, flags);
+}
+
+static noinline int run_ordered_completions(struct btrfs_workers *workers,
+					    struct btrfs_work *work)
+{
 	if (!workers->ordered)
 		return 0;
 
 	set_bit(WORK_DONE_BIT, &work->flags);
 
-	spin_lock_irqsave(&workers->lock, flags);
+	spin_lock(&workers->order_lock);
 
 	while (1) {
 		if (!list_empty(&workers->prio_order_list)) {
@@ -126,45 +207,118 @@ static noinline int run_ordered_completions(struct btrfs_workers *workers,
 		if (test_and_set_bit(WORK_ORDER_DONE_BIT, &work->flags))
 			break;
 
-		spin_unlock_irqrestore(&workers->lock, flags);
+		spin_unlock(&workers->order_lock);
 
 		work->ordered_func(work);
 
 		/* now take the lock again and call the freeing code */
-		spin_lock_irqsave(&workers->lock, flags);
+		spin_lock(&workers->order_lock);
 		list_del(&work->order_list);
 		work->ordered_free(work);
 	}
 
-	spin_unlock_irqrestore(&workers->lock, flags);
+	spin_unlock(&workers->order_lock);
 	return 0;
 }
 
+static void put_worker(struct btrfs_worker_thread *worker)
+{
+	if (atomic_dec_and_test(&worker->refs))
+		kfree(worker);
+}
+
+static int try_worker_shutdown(struct btrfs_worker_thread *worker)
+{
+	int freeit = 0;
+
+	spin_lock_irq(&worker->lock);
+	spin_lock(&worker->workers->lock);
+	if (worker->workers->num_workers > 1 &&
+	    worker->idle &&
+	    !worker->working &&
+	    !list_empty(&worker->worker_list) &&
+	    list_empty(&worker->prio_pending) &&
+	    list_empty(&worker->pending) &&
+	    atomic_read(&worker->num_pending) == 0) {
+		freeit = 1;
+		list_del_init(&worker->worker_list);
+		worker->workers->num_workers--;
+	}
+	spin_unlock(&worker->workers->lock);
+	spin_unlock_irq(&worker->lock);
+
+	if (freeit)
+		put_worker(worker);
+	return freeit;
+}
+
+static struct btrfs_work *get_next_work(struct btrfs_worker_thread *worker,
+					struct list_head *prio_head,
+					struct list_head *head)
+{
+	struct btrfs_work *work = NULL;
+	struct list_head *cur = NULL;
+
+	if(!list_empty(prio_head))
+		cur = prio_head->next;
+
+	smp_mb();
+	if (!list_empty(&worker->prio_pending))
+		goto refill;
+
+	if (!list_empty(head))
+		cur = head->next;
+
+	if (cur)
+		goto out;
+
+refill:
+	spin_lock_irq(&worker->lock);
+	list_splice_tail_init(&worker->prio_pending, prio_head);
+	list_splice_tail_init(&worker->pending, head);
+
+	if (!list_empty(prio_head))
+		cur = prio_head->next;
+	else if (!list_empty(head))
+		cur = head->next;
+	spin_unlock_irq(&worker->lock);
+
+	if (!cur)
+		goto out_fail;
+
+out:
+	work = list_entry(cur, struct btrfs_work, list);
+
+out_fail:
+	return work;
+}
+
 /*
  * main loop for servicing work items
  */
 static int worker_loop(void *arg)
 {
 	struct btrfs_worker_thread *worker = arg;
-	struct list_head *cur;
+	struct list_head head;
+	struct list_head prio_head;
 	struct btrfs_work *work;
+
+	INIT_LIST_HEAD(&head);
+	INIT_LIST_HEAD(&prio_head);
+
 	do {
-		spin_lock_irq(&worker->lock);
-again_locked:
+again:
 		while (1) {
-			if (!list_empty(&worker->prio_pending))
-				cur = worker->prio_pending.next;
-			else if (!list_empty(&worker->pending))
-				cur = worker->pending.next;
-			else
+
+
+			work = get_next_work(worker, &prio_head, &head);
+			if (!work)
 				break;
 
-			work = list_entry(cur, struct btrfs_work, list);
 			list_del(&work->list);
 			clear_bit(WORK_QUEUED_BIT, &work->flags);
 
 			work->worker = worker;
-			spin_unlock_irq(&worker->lock);
 
 			work->func(work);
 
@@ -175,9 +329,13 @@ again_locked:
 			 */
 			run_ordered_completions(worker->workers, work);
 
-			spin_lock_irq(&worker->lock);
-			check_idle_worker(worker);
+			check_pending_worker_creates(worker);
+
 		}
+
+		spin_lock_irq(&worker->lock);
+		check_idle_worker(worker);
+
 		if (freezing(current)) {
 			worker->working = 0;
 			spin_unlock_irq(&worker->lock);
@@ -216,8 +374,10 @@ again_locked:
 				spin_lock_irq(&worker->lock);
 				set_current_state(TASK_INTERRUPTIBLE);
 				if (!list_empty(&worker->pending) ||
-				    !list_empty(&worker->prio_pending))
-					goto again_locked;
+				    !list_empty(&worker->prio_pending)) {
+					spin_unlock_irq(&worker->lock);
+					goto again;
+				}
 
 				/*
 				 * this makes sure we get a wakeup when someone
@@ -226,8 +386,13 @@ again_locked:
 				worker->working = 0;
 				spin_unlock_irq(&worker->lock);
 
-				if (!kthread_should_stop())
-					schedule();
+				if (!kthread_should_stop()) {
+					schedule_timeout(HZ * 120);
+					if (!worker->working &&
+					    try_worker_shutdown(worker)) {
+						return 0;
+					}
+				}
 			}
 			__set_current_state(TASK_RUNNING);
 		}
@@ -242,41 +407,61 @@ int btrfs_stop_workers(struct btrfs_workers *workers)
 {
 	struct list_head *cur;
 	struct btrfs_worker_thread *worker;
+	int can_stop;
 
+	spin_lock_irq(&workers->lock);
 	list_splice_init(&workers->idle_list, &workers->worker_list);
 	while (!list_empty(&workers->worker_list)) {
 		cur = workers->worker_list.next;
 		worker = list_entry(cur, struct btrfs_worker_thread,
 				    worker_list);
-		kthread_stop(worker->task);
-		list_del(&worker->worker_list);
-		kfree(worker);
+
+		atomic_inc(&worker->refs);
+		workers->num_workers -= 1;
+		if (!list_empty(&worker->worker_list)) {
+			list_del_init(&worker->worker_list);
+			put_worker(worker);
+			can_stop = 1;
+		} else
+			can_stop = 0;
+		spin_unlock_irq(&workers->lock);
+		if (can_stop)
+			kthread_stop(worker->task);
+		spin_lock_irq(&workers->lock);
+		put_worker(worker);
 	}
+	spin_unlock_irq(&workers->lock);
 	return 0;
 }
 
 /*
  * simple init on struct btrfs_workers
  */
-void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max)
+void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max,
+			struct btrfs_workers *async_helper)
 {
 	workers->num_workers = 0;
+	workers->num_workers_starting = 0;
 	INIT_LIST_HEAD(&workers->worker_list);
 	INIT_LIST_HEAD(&workers->idle_list);
 	INIT_LIST_HEAD(&workers->order_list);
 	INIT_LIST_HEAD(&workers->prio_order_list);
 	spin_lock_init(&workers->lock);
+	spin_lock_init(&workers->order_lock);
 	workers->max_workers = max;
 	workers->idle_thresh = 32;
 	workers->name = name;
 	workers->ordered = 0;
+	workers->atomic_start_pending = 0;
+	workers->atomic_worker_start = async_helper;
 }
 
 /*
  * starts new worker threads.  This does not enforce the max worker
  * count in case you need to temporarily go past it.
  */
-int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
+static int __btrfs_start_workers(struct btrfs_workers *workers,
+				 int num_workers)
 {
 	struct btrfs_worker_thread *worker;
 	int ret = 0;
@@ -293,7 +478,9 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
 		INIT_LIST_HEAD(&worker->prio_pending);
 		INIT_LIST_HEAD(&worker->worker_list);
 		spin_lock_init(&worker->lock);
+
 		atomic_set(&worker->num_pending, 0);
+		atomic_set(&worker->refs, 1);
 		worker->workers = workers;
 		worker->task = kthread_run(worker_loop, worker,
 					   "btrfs-%s-%d", workers->name,
@@ -303,11 +490,12 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
 			kfree(worker);
 			goto fail;
 		}
-
 		spin_lock_irq(&workers->lock);
 		list_add_tail(&worker->worker_list, &workers->idle_list);
 		worker->idle = 1;
 		workers->num_workers++;
+		workers->num_workers_starting--;
+		WARN_ON(workers->num_workers_starting < 0);
 		spin_unlock_irq(&workers->lock);
 	}
 	return 0;
@@ -316,6 +504,14 @@ fail:
 	return ret;
 }
 
+int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
+{
+	spin_lock_irq(&workers->lock);
+	workers->num_workers_starting += num_workers;
+	spin_unlock_irq(&workers->lock);
+	return __btrfs_start_workers(workers, num_workers);
+}
+
 /*
  * run through the list and find a worker thread that doesn't have a lot
  * to do right now.  This can return null if we aren't yet at the thread
@@ -325,7 +521,10 @@ static struct btrfs_worker_thread *next_worker(struct btrfs_workers *workers)
 {
 	struct btrfs_worker_thread *worker;
 	struct list_head *next;
-	int enforce_min = workers->num_workers < workers->max_workers;
+	int enforce_min;
+
+	enforce_min = (workers->num_workers + workers->num_workers_starting) <
+		workers->max_workers;
 
 	/*
 	 * if we find an idle thread, don't move it to the end of the
@@ -350,7 +549,6 @@ static struct btrfs_worker_thread *next_worker(struct btrfs_workers *workers)
 	 */
 	next = workers->worker_list.next;
 	worker = list_entry(next, struct btrfs_worker_thread, worker_list);
-	atomic_inc(&worker->num_pending);
 	worker->sequence++;
 
 	if (worker->sequence % workers->idle_thresh == 0)
@@ -367,35 +565,49 @@ static struct btrfs_worker_thread *find_worker(struct btrfs_workers *workers)
 {
 	struct btrfs_worker_thread *worker;
 	unsigned long flags;
+	struct list_head *fallback;
 
 again:
 	spin_lock_irqsave(&workers->lock, flags);
 	worker = next_worker(workers);
-	spin_unlock_irqrestore(&workers->lock, flags);
 
 	if (!worker) {
-		spin_lock_irqsave(&workers->lock, flags);
-		if (workers->num_workers >= workers->max_workers) {
-			struct list_head *fallback = NULL;
-			/*
-			 * we have failed to find any workers, just
-			 * return the force one
-			 */
-			if (!list_empty(&workers->worker_list))
-				fallback = workers->worker_list.next;
-			if (!list_empty(&workers->idle_list))
-				fallback = workers->idle_list.next;
-			BUG_ON(!fallback);
-			worker = list_entry(fallback,
-				  struct btrfs_worker_thread, worker_list);
-			spin_unlock_irqrestore(&workers->lock, flags);
+		if (workers->num_workers + workers->num_workers_starting >=
+		    workers->max_workers) {
+			goto fallback;
+		} else if (workers->atomic_worker_start) {
+			workers->atomic_start_pending = 1;
+			goto fallback;
 		} else {
+			workers->num_workers_starting++;
 			spin_unlock_irqrestore(&workers->lock, flags);
 			/* we're below the limit, start another worker */
-			btrfs_start_workers(workers, 1);
+			__btrfs_start_workers(workers, 1);
 			goto again;
 		}
 	}
+	goto found;
+
+fallback:
+	fallback = NULL;
+	/*
+	 * we have failed to find any workers, just
+	 * return the first one we can find.
+	 */
+	if (!list_empty(&workers->worker_list))
+		fallback = workers->worker_list.next;
+	if (!list_empty(&workers->idle_list))
+		fallback = workers->idle_list.next;
+	BUG_ON(!fallback);
+	worker = list_entry(fallback,
+		  struct btrfs_worker_thread, worker_list);
+found:
+	/*
+	 * this makes sure the worker doesn't exit before it is placed
+	 * onto a busy/idle list
+	 */
+	atomic_inc(&worker->num_pending);
+	spin_unlock_irqrestore(&workers->lock, flags);
 	return worker;
 }
 
@@ -427,7 +639,7 @@ int btrfs_requeue_work(struct btrfs_work *work)
 		spin_lock(&worker->workers->lock);
 		worker->idle = 0;
 		list_move_tail(&worker->worker_list,
-			       &worker->workers->worker_list);
+			      &worker->workers->worker_list);
 		spin_unlock(&worker->workers->lock);
 	}
 	if (!worker->working) {
@@ -435,9 +647,9 @@ int btrfs_requeue_work(struct btrfs_work *work)
 		worker->working = 1;
 	}
 
-	spin_unlock_irqrestore(&worker->lock, flags);
 	if (wake)
 		wake_up_process(worker->task);
+	spin_unlock_irqrestore(&worker->lock, flags);
 out:
 
 	return 0;
@@ -463,14 +675,18 @@ int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 
 	worker = find_worker(workers);
 	if (workers->ordered) {
-		spin_lock_irqsave(&workers->lock, flags);
+		/*
+		 * you're not allowed to do ordered queues from an
+		 * interrupt handler
+		 */
+		spin_lock(&workers->order_lock);
 		if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags)) {
 			list_add_tail(&work->order_list,
 				      &workers->prio_order_list);
 		} else {
 			list_add_tail(&work->order_list, &workers->order_list);
 		}
-		spin_unlock_irqrestore(&workers->lock, flags);
+		spin_unlock(&workers->order_lock);
 	} else {
 		INIT_LIST_HEAD(&work->order_list);
 	}
@@ -481,7 +697,6 @@ int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 		list_add_tail(&work->list, &worker->prio_pending);
 	else
 		list_add_tail(&work->list, &worker->pending);
-	atomic_inc(&worker->num_pending);
 	check_busy_worker(worker);
 
 	/*
@@ -492,10 +707,10 @@ int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 		wake = 1;
 	worker->working = 1;
 
-	spin_unlock_irqrestore(&worker->lock, flags);
-
 	if (wake)
 		wake_up_process(worker->task);
+	spin_unlock_irqrestore(&worker->lock, flags);
+
 out:
 	return 0;
 }
diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h
index 1b511c109db6..5077746cf85e 100644
--- a/fs/btrfs/async-thread.h
+++ b/fs/btrfs/async-thread.h
@@ -64,6 +64,8 @@ struct btrfs_workers {
 	/* current number of running workers */
 	int num_workers;
 
+	int num_workers_starting;
+
 	/* max number of workers allowed.  changed by btrfs_start_workers */
 	int max_workers;
 
@@ -73,6 +75,16 @@ struct btrfs_workers {
 	/* force completions in the order they were queued */
 	int ordered;
 
+	/* more workers required, but in an interrupt handler */
+	int atomic_start_pending;
+
+	/*
+	 * are we allowed to sleep while starting workers or are we required
+	 * to start them at a later time?  If we can't sleep, this indicates
+	 * which queue we need to use to schedule thread creation.
+	 */
+	struct btrfs_workers *atomic_worker_start;
+
 	/* list with all the work threads.  The workers on the idle thread
 	 * may be actively servicing jobs, but they haven't yet hit the
 	 * idle thresh limit above.
@@ -90,6 +102,9 @@ struct btrfs_workers {
 	/* lock for finding the next worker thread to queue on */
 	spinlock_t lock;
 
+	/* lock for the ordered lists */
+	spinlock_t order_lock;
+
 	/* extra name for this worker, used for current->name */
 	char *name;
 };
@@ -97,7 +112,8 @@ struct btrfs_workers {
 int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work);
 int btrfs_start_workers(struct btrfs_workers *workers, int num_workers);
 int btrfs_stop_workers(struct btrfs_workers *workers);
-void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max);
+void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max,
+			struct btrfs_workers *async_starter);
 int btrfs_requeue_work(struct btrfs_work *work);
 void btrfs_set_work_high_prio(struct btrfs_work *work);
 #endif
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index ea1ea0af8c0e..f6783a42f010 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -86,6 +86,12 @@ struct btrfs_inode {
 	 * transid of the trans_handle that last modified this inode
 	 */
 	u64 last_trans;
+
+	/*
+	 * log transid when this inode was last modified
+	 */
+	u64 last_sub_trans;
+
 	/*
 	 * transid that last logged this inode
 	 */
@@ -128,6 +134,16 @@ struct btrfs_inode {
 	u64 last_unlink_trans;
 
 	/*
+	 * Counters to keep track of the number of extent item's we may use due
+	 * to delalloc and such.  outstanding_extents is the number of extent
+	 * items we think we'll end up using, and reserved_extents is the number
+	 * of extent items we've reserved metadata for.
+	 */
+	spinlock_t accounting_lock;
+	int reserved_extents;
+	int outstanding_extents;
+
+	/*
 	 * ordered_data_close is set by truncate when a file that used
 	 * to have good data has been truncated to zero.  When it is set
 	 * the btrfs file release call will add this inode to the
@@ -138,6 +154,7 @@ struct btrfs_inode {
 	 * of these.
 	 */
 	unsigned ordered_data_close:1;
+	unsigned dummy_inode:1;
 
 	struct inode vfs_inode;
 };
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 9d8ba4d54a37..a11a32058b50 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -506,10 +506,10 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 		 */
 		set_page_extent_mapped(page);
 		lock_extent(tree, last_offset, end, GFP_NOFS);
-		spin_lock(&em_tree->lock);
+		read_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, last_offset,
 					   PAGE_CACHE_SIZE);
-		spin_unlock(&em_tree->lock);
+		read_unlock(&em_tree->lock);
 
 		if (!em || last_offset < em->start ||
 		    (last_offset + PAGE_CACHE_SIZE > extent_map_end(em)) ||
@@ -593,11 +593,11 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	em_tree = &BTRFS_I(inode)->extent_tree;
 
 	/* we need the actual starting offset of this extent in the file */
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree,
 				   page_offset(bio->bi_io_vec->bv_page),
 				   PAGE_CACHE_SIZE);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	compressed_len = em->block_len;
 	cb = kmalloc(compressed_bio_size(root, compressed_len), GFP_NOFS);
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 3fdcc0512d3a..ec96f3a6d536 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -2853,6 +2853,12 @@ static noinline int split_leaf(struct btrfs_trans_handle *trans,
 	int split;
 	int num_doubles = 0;
 
+	l = path->nodes[0];
+	slot = path->slots[0];
+	if (extend && data_size + btrfs_item_size_nr(l, slot) +
+	    sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(root))
+		return -EOVERFLOW;
+
 	/* first try to make some room by pushing left and right */
 	if (data_size && ins_key->type != BTRFS_DIR_ITEM_KEY) {
 		wret = push_leaf_right(trans, root, path, data_size, 0);
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 837435ce84ca..444b3e9b92a4 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -114,6 +114,10 @@ struct btrfs_ordered_sum;
  */
 #define BTRFS_DEV_ITEMS_OBJECTID 1ULL
 
+#define BTRFS_BTREE_INODE_OBJECTID 1
+
+#define BTRFS_EMPTY_SUBVOL_DIR_OBJECTID 2
+
 /*
  * we can actually store much bigger names, but lets not confuse the rest
  * of linux
@@ -670,21 +674,29 @@ struct btrfs_space_info {
 	u64 bytes_reserved;	/* total bytes the allocator has reserved for
 				   current allocations */
 	u64 bytes_readonly;	/* total bytes that are read only */
-
-	/* delalloc accounting */
-	u64 bytes_delalloc;	/* number of bytes reserved for allocation,
-				   this space is not necessarily reserved yet
-				   by the allocator */
+	u64 bytes_super;	/* total bytes reserved for the super blocks */
+	u64 bytes_root;		/* the number of bytes needed to commit a
+				   transaction */
 	u64 bytes_may_use;	/* number of bytes that may be used for
-				   delalloc */
+				   delalloc/allocations */
+	u64 bytes_delalloc;	/* number of bytes currently reserved for
+				   delayed allocation */
 
 	int full;		/* indicates that we cannot allocate any more
 				   chunks for this space */
 	int force_alloc;	/* set if we need to force a chunk alloc for
 				   this space */
+	int force_delalloc;	/* make people start doing filemap_flush until
+				   we're under a threshold */
 
 	struct list_head list;
 
+	/* for controlling how we free up space for allocations */
+	wait_queue_head_t allocate_wait;
+	wait_queue_head_t flush_wait;
+	int allocating_chunk;
+	int flushing;
+
 	/* for block groups in our same type */
 	struct list_head block_groups;
 	spinlock_t lock;
@@ -726,6 +738,15 @@ enum btrfs_caching_type {
 	BTRFS_CACHE_FINISHED	= 2,
 };
 
+struct btrfs_caching_control {
+	struct list_head list;
+	struct mutex mutex;
+	wait_queue_head_t wait;
+	struct btrfs_block_group_cache *block_group;
+	u64 progress;
+	atomic_t count;
+};
+
 struct btrfs_block_group_cache {
 	struct btrfs_key key;
 	struct btrfs_block_group_item item;
@@ -733,6 +754,7 @@ struct btrfs_block_group_cache {
 	spinlock_t lock;
 	u64 pinned;
 	u64 reserved;
+	u64 bytes_super;
 	u64 flags;
 	u64 sectorsize;
 	int extents_thresh;
@@ -742,8 +764,9 @@ struct btrfs_block_group_cache {
 	int dirty;
 
 	/* cache tracking stuff */
-	wait_queue_head_t caching_q;
 	int cached;
+	struct btrfs_caching_control *caching_ctl;
+	u64 last_byte_to_unpin;
 
 	struct btrfs_space_info *space_info;
 
@@ -782,13 +805,16 @@ struct btrfs_fs_info {
 
 	/* the log root tree is a directory of all the other log roots */
 	struct btrfs_root *log_root_tree;
+
+	spinlock_t fs_roots_radix_lock;
 	struct radix_tree_root fs_roots_radix;
 
 	/* block group cache stuff */
 	spinlock_t block_group_cache_lock;
 	struct rb_root block_group_cache_tree;
 
-	struct extent_io_tree pinned_extents;
+	struct extent_io_tree freed_extents[2];
+	struct extent_io_tree *pinned_extents;
 
 	/* logical->physical extent mapping */
 	struct btrfs_mapping_tree mapping_tree;
@@ -822,11 +848,7 @@ struct btrfs_fs_info {
 	struct mutex transaction_kthread_mutex;
 	struct mutex cleaner_mutex;
 	struct mutex chunk_mutex;
-	struct mutex drop_mutex;
 	struct mutex volume_mutex;
-	struct mutex tree_reloc_mutex;
-	struct rw_semaphore extent_commit_sem;
-
 	/*
 	 * this protects the ordered operations list only while we are
 	 * processing all of the entries on it.  This way we make
@@ -835,10 +857,16 @@ struct btrfs_fs_info {
 	 * before jumping into the main commit.
 	 */
 	struct mutex ordered_operations_mutex;
+	struct rw_semaphore extent_commit_sem;
+
+	struct rw_semaphore subvol_sem;
+
+	struct srcu_struct subvol_srcu;
 
 	struct list_head trans_list;
 	struct list_head hashers;
 	struct list_head dead_roots;
+	struct list_head caching_block_groups;
 
 	atomic_t nr_async_submits;
 	atomic_t async_submit_draining;
@@ -882,6 +910,7 @@ struct btrfs_fs_info {
 	 * A third pool does submit_bio to avoid deadlocking with the other
 	 * two
 	 */
+	struct btrfs_workers generic_worker;
 	struct btrfs_workers workers;
 	struct btrfs_workers delalloc_workers;
 	struct btrfs_workers endio_workers;
@@ -889,6 +918,7 @@ struct btrfs_fs_info {
 	struct btrfs_workers endio_meta_write_workers;
 	struct btrfs_workers endio_write_workers;
 	struct btrfs_workers submit_workers;
+	struct btrfs_workers enospc_workers;
 	/*
 	 * fixup workers take dirty pages that didn't properly go through
 	 * the cow mechanism and make them safe to write.  It happens
@@ -979,7 +1009,10 @@ struct btrfs_root {
 	atomic_t log_writers;
 	atomic_t log_commit[2];
 	unsigned long log_transid;
+	unsigned long last_log_commit;
 	unsigned long log_batch;
+	pid_t log_start_pid;
+	bool log_multiple_pids;
 
 	u64 objectid;
 	u64 last_trans;
@@ -996,10 +1029,12 @@ struct btrfs_root {
 	u32 stripesize;
 
 	u32 type;
-	u64 highest_inode;
-	u64 last_inode_alloc;
+
+	u64 highest_objectid;
 	int ref_cows;
 	int track_dirty;
+	int in_radix;
+
 	u64 defrag_trans_start;
 	struct btrfs_key defrag_progress;
 	struct btrfs_key defrag_max;
@@ -1118,6 +1153,7 @@ struct btrfs_root {
 #define BTRFS_MOUNT_FLUSHONCOMMIT       (1 << 7)
 #define BTRFS_MOUNT_SSD_SPREAD		(1 << 8)
 #define BTRFS_MOUNT_NOSSD		(1 << 9)
+#define BTRFS_MOUNT_DISCARD		(1 << 10)
 
 #define btrfs_clear_opt(o, opt)		((o) &= ~BTRFS_MOUNT_##opt)
 #define btrfs_set_opt(o, opt)		((o) |= BTRFS_MOUNT_##opt)
@@ -1920,8 +1956,8 @@ void btrfs_put_block_group(struct btrfs_block_group_cache *cache);
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root, unsigned long count);
 int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len);
-int btrfs_update_pinned_extents(struct btrfs_root *root,
-				u64 bytenr, u64 num, int pin);
+int btrfs_pin_extent(struct btrfs_root *root,
+		     u64 bytenr, u64 num, int reserved);
 int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root, struct extent_buffer *leaf);
 int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans,
@@ -1971,9 +2007,10 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
 		      u64 root_objectid, u64 owner, u64 offset);
 
 int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len);
+int btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root);
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct extent_io_tree *unpin);
+			       struct btrfs_root *root);
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_root *root,
 			 u64 bytenr, u64 num_bytes, u64 parent,
@@ -1984,6 +2021,7 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
 int btrfs_extent_readonly(struct btrfs_root *root, u64 bytenr);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
 int btrfs_read_block_groups(struct btrfs_root *root);
+int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr);
 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root, u64 bytes_used,
 			   u64 type, u64 chunk_objectid, u64 chunk_offset,
@@ -1997,7 +2035,12 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags);
 void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *ionde);
 void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
 
-int btrfs_check_metadata_free_space(struct btrfs_root *root);
+int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items);
+int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items);
+int btrfs_unreserve_metadata_for_delalloc(struct btrfs_root *root,
+					  struct inode *inode, int num_items);
+int btrfs_reserve_metadata_for_delalloc(struct btrfs_root *root,
+					struct inode *inode, int num_items);
 int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode,
 				u64 bytes);
 void btrfs_free_reserved_data_space(struct btrfs_root *root,
@@ -2006,7 +2049,6 @@ void btrfs_delalloc_reserve_space(struct btrfs_root *root, struct inode *inode,
 				 u64 bytes);
 void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
 			      u64 bytes);
-void btrfs_free_pinned_extents(struct btrfs_fs_info *info);
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
 		     int level, int *slot);
@@ -2100,12 +2142,15 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 			struct extent_buffer *parent);
 /* root-item.c */
 int btrfs_find_root_ref(struct btrfs_root *tree_root,
-		   struct btrfs_path *path,
-		   u64 root_id, u64 ref_id);
+			struct btrfs_path *path,
+			u64 root_id, u64 ref_id);
 int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *tree_root,
-		       u64 root_id, u8 type, u64 ref_id,
-		       u64 dirid, u64 sequence,
+		       u64 root_id, u64 ref_id, u64 dirid, u64 sequence,
+		       const char *name, int name_len);
+int btrfs_del_root_ref(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *tree_root,
+		       u64 root_id, u64 ref_id, u64 dirid, u64 *sequence,
 		       const char *name, int name_len);
 int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		   struct btrfs_key *key);
@@ -2120,6 +2165,7 @@ int btrfs_find_last_root(struct btrfs_root *root, u64 objectid, struct
 int btrfs_search_root(struct btrfs_root *root, u64 search_start,
 		      u64 *found_objectid);
 int btrfs_find_dead_roots(struct btrfs_root *root, u64 objectid);
+int btrfs_find_orphan_roots(struct btrfs_root *tree_root);
 int btrfs_set_root_node(struct btrfs_root_item *item,
 			struct extent_buffer *node);
 /* dir-item.c */
@@ -2138,6 +2184,10 @@ btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
 			    struct btrfs_path *path, u64 dir,
 			    u64 objectid, const char *name, int name_len,
 			    int mod);
+struct btrfs_dir_item *
+btrfs_search_dir_index_item(struct btrfs_root *root,
+			    struct btrfs_path *path, u64 dirid,
+			    const char *name, int name_len);
 struct btrfs_dir_item *btrfs_match_dir_item_name(struct btrfs_root *root,
 			      struct btrfs_path *path,
 			      const char *name, int name_len);
@@ -2160,6 +2210,7 @@ int btrfs_insert_orphan_item(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root, u64 offset);
 int btrfs_del_orphan_item(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, u64 offset);
+int btrfs_find_orphan_item(struct btrfs_root *root, u64 offset);
 
 /* inode-map.c */
 int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
@@ -2232,6 +2283,10 @@ int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
 int btrfs_add_link(struct btrfs_trans_handle *trans,
 		   struct inode *parent_inode, struct inode *inode,
 		   const char *name, int name_len, int add_backref, u64 index);
+int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root,
+			struct inode *dir, u64 objectid,
+			const char *name, int name_len);
 int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
 			       struct inode *inode, u64 new_size,
@@ -2242,7 +2297,7 @@ int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end);
 int btrfs_writepages(struct address_space *mapping,
 		     struct writeback_control *wbc);
 int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *new_root, struct dentry *dentry,
+			     struct btrfs_root *new_root,
 			     u64 new_dirid, u64 alloc_hint);
 int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
 			 size_t size, struct bio *bio, unsigned long bio_flags);
@@ -2258,6 +2313,7 @@ int btrfs_write_inode(struct inode *inode, int wait);
 void btrfs_dirty_inode(struct inode *inode);
 struct inode *btrfs_alloc_inode(struct super_block *sb);
 void btrfs_destroy_inode(struct inode *inode);
+void btrfs_drop_inode(struct inode *inode);
 int btrfs_init_cachep(void);
 void btrfs_destroy_cachep(void);
 long btrfs_ioctl_trans_end(struct file *file);
@@ -2275,6 +2331,8 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode);
 int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode);
 void btrfs_orphan_cleanup(struct btrfs_root *root);
 int btrfs_cont_expand(struct inode *inode, loff_t size);
+int btrfs_invalidate_inodes(struct btrfs_root *root);
+extern const struct dentry_operations btrfs_dentry_operations;
 
 /* ioctl.c */
 long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
@@ -2286,11 +2344,11 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync);
 int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			    int skip_pinned);
 int btrfs_check_file(struct btrfs_root *root, struct inode *inode);
-extern struct file_operations btrfs_file_operations;
+extern const struct file_operations btrfs_file_operations;
 int btrfs_drop_extents(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root, struct inode *inode,
 		       u64 start, u64 end, u64 locked_end,
-		       u64 inline_limit, u64 *hint_block);
+		       u64 inline_limit, u64 *hint_block, int drop_cache);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root,
 			      struct inode *inode, u64 start, u64 end);
@@ -2317,7 +2375,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options);
 int btrfs_sync_fs(struct super_block *sb, int wait);
 
 /* acl.c */
-#ifdef CONFIG_FS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 int btrfs_check_acl(struct inode *inode, int mask);
 #else
 #define btrfs_check_acl NULL
diff --git a/fs/btrfs/dir-item.c b/fs/btrfs/dir-item.c
index 1d70236ba00c..f3a6075519cc 100644
--- a/fs/btrfs/dir-item.c
+++ b/fs/btrfs/dir-item.c
@@ -281,6 +281,53 @@ btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
 	return btrfs_match_dir_item_name(root, path, name, name_len);
 }
 
+struct btrfs_dir_item *
+btrfs_search_dir_index_item(struct btrfs_root *root,
+			    struct btrfs_path *path, u64 dirid,
+			    const char *name, int name_len)
+{
+	struct extent_buffer *leaf;
+	struct btrfs_dir_item *di;
+	struct btrfs_key key;
+	u32 nritems;
+	int ret;
+
+	key.objectid = dirid;
+	key.type = BTRFS_DIR_INDEX_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ERR_PTR(ret);
+
+	leaf = path->nodes[0];
+	nritems = btrfs_header_nritems(leaf);
+
+	while (1) {
+		if (path->slots[0] >= nritems) {
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				return ERR_PTR(ret);
+			if (ret > 0)
+				break;
+			leaf = path->nodes[0];
+			nritems = btrfs_header_nritems(leaf);
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		if (key.objectid != dirid || key.type != BTRFS_DIR_INDEX_KEY)
+			break;
+
+		di = btrfs_match_dir_item_name(root, path, name, name_len);
+		if (di)
+			return di;
+
+		path->slots[0]++;
+	}
+	return NULL;
+}
+
 struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
 					  struct btrfs_root *root,
 					  struct btrfs_path *path, u64 dir,
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 8b8192790011..02b6afbd7450 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -41,6 +41,7 @@
 
 static struct extent_io_ops btree_extent_io_ops;
 static void end_workqueue_fn(struct btrfs_work *work);
+static void free_fs_root(struct btrfs_root *root);
 
 static atomic_t btrfs_bdi_num = ATOMIC_INIT(0);
 
@@ -123,15 +124,15 @@ static struct extent_map *btree_get_extent(struct inode *inode,
 	struct extent_map *em;
 	int ret;
 
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, start, len);
 	if (em) {
 		em->bdev =
 			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
-		spin_unlock(&em_tree->lock);
+		read_unlock(&em_tree->lock);
 		goto out;
 	}
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	em = alloc_extent_map(GFP_NOFS);
 	if (!em) {
@@ -144,7 +145,7 @@ static struct extent_map *btree_get_extent(struct inode *inode,
 	em->block_start = 0;
 	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
 
-	spin_lock(&em_tree->lock);
+	write_lock(&em_tree->lock);
 	ret = add_extent_mapping(em_tree, em);
 	if (ret == -EEXIST) {
 		u64 failed_start = em->start;
@@ -163,7 +164,7 @@ static struct extent_map *btree_get_extent(struct inode *inode,
 		free_extent_map(em);
 		em = NULL;
 	}
-	spin_unlock(&em_tree->lock);
+	write_unlock(&em_tree->lock);
 
 	if (ret)
 		em = ERR_PTR(ret);
@@ -772,7 +773,7 @@ static void btree_invalidatepage(struct page *page, unsigned long offset)
 	}
 }
 
-static struct address_space_operations btree_aops = {
+static const struct address_space_operations btree_aops = {
 	.readpage	= btree_readpage,
 	.writepage	= btree_writepage,
 	.writepages	= btree_writepages,
@@ -821,14 +822,14 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
 
 int btrfs_write_tree_block(struct extent_buffer *buf)
 {
-	return btrfs_fdatawrite_range(buf->first_page->mapping, buf->start,
-				      buf->start + buf->len - 1, WB_SYNC_ALL);
+	return filemap_fdatawrite_range(buf->first_page->mapping, buf->start,
+					buf->start + buf->len - 1);
 }
 
 int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
 {
-	return btrfs_wait_on_page_writeback_range(buf->first_page->mapping,
-				  buf->start, buf->start + buf->len - 1);
+	return filemap_fdatawait_range(buf->first_page->mapping,
+				       buf->start, buf->start + buf->len - 1);
 }
 
 struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
@@ -895,8 +896,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	root->fs_info = fs_info;
 	root->objectid = objectid;
 	root->last_trans = 0;
-	root->highest_inode = 0;
-	root->last_inode_alloc = 0;
+	root->highest_objectid = 0;
 	root->name = NULL;
 	root->in_sysfs = 0;
 	root->inode_tree.rb_node = NULL;
@@ -917,6 +917,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	atomic_set(&root->log_writers, 0);
 	root->log_batch = 0;
 	root->log_transid = 0;
+	root->last_log_commit = 0;
 	extent_io_tree_init(&root->dirty_log_pages,
 			     fs_info->btree_inode->i_mapping, GFP_NOFS);
 
@@ -952,14 +953,16 @@ static int find_and_setup_root(struct btrfs_root *tree_root,
 		     root, fs_info, objectid);
 	ret = btrfs_find_last_root(tree_root, objectid,
 				   &root->root_item, &root->root_key);
+	if (ret > 0)
+		return -ENOENT;
 	BUG_ON(ret);
 
 	generation = btrfs_root_generation(&root->root_item);
 	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
 	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
 				     blocksize, generation);
-	root->commit_root = btrfs_root_node(root);
 	BUG_ON(!root->node);
+	root->commit_root = btrfs_root_node(root);
 	return 0;
 }
 
@@ -1085,6 +1088,7 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 	WARN_ON(root->log_root);
 	root->log_root = log_root;
 	root->log_transid = 0;
+	root->last_log_commit = 0;
 	return 0;
 }
 
@@ -1095,7 +1099,6 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
 	struct btrfs_fs_info *fs_info = tree_root->fs_info;
 	struct btrfs_path *path;
 	struct extent_buffer *l;
-	u64 highest_inode;
 	u64 generation;
 	u32 blocksize;
 	int ret = 0;
@@ -1110,7 +1113,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
 			kfree(root);
 			return ERR_PTR(ret);
 		}
-		goto insert;
+		goto out;
 	}
 
 	__setup_root(tree_root->nodesize, tree_root->leafsize,
@@ -1120,39 +1123,30 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
 	ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0);
-	if (ret != 0) {
-		if (ret > 0)
-			ret = -ENOENT;
-		goto out;
+	if (ret == 0) {
+		l = path->nodes[0];
+		read_extent_buffer(l, &root->root_item,
+				btrfs_item_ptr_offset(l, path->slots[0]),
+				sizeof(root->root_item));
+		memcpy(&root->root_key, location, sizeof(*location));
 	}
-	l = path->nodes[0];
-	read_extent_buffer(l, &root->root_item,
-	       btrfs_item_ptr_offset(l, path->slots[0]),
-	       sizeof(root->root_item));
-	memcpy(&root->root_key, location, sizeof(*location));
-	ret = 0;
-out:
-	btrfs_release_path(root, path);
 	btrfs_free_path(path);
 	if (ret) {
-		kfree(root);
+		if (ret > 0)
+			ret = -ENOENT;
 		return ERR_PTR(ret);
 	}
+
 	generation = btrfs_root_generation(&root->root_item);
 	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
 	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
 				     blocksize, generation);
 	root->commit_root = btrfs_root_node(root);
 	BUG_ON(!root->node);
-insert:
-	if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
+out:
+	if (location->objectid != BTRFS_TREE_LOG_OBJECTID)
 		root->ref_cows = 1;
-		ret = btrfs_find_highest_inode(root, &highest_inode);
-		if (ret == 0) {
-			root->highest_inode = highest_inode;
-			root->last_inode_alloc = highest_inode;
-		}
-	}
+
 	return root;
 }
 
@@ -1187,39 +1181,66 @@ struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
 		return fs_info->dev_root;
 	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
 		return fs_info->csum_root;
-
+again:
+	spin_lock(&fs_info->fs_roots_radix_lock);
 	root = radix_tree_lookup(&fs_info->fs_roots_radix,
 				 (unsigned long)location->objectid);
+	spin_unlock(&fs_info->fs_roots_radix_lock);
 	if (root)
 		return root;
 
+	ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
+	if (ret == 0)
+		ret = -ENOENT;
+	if (ret < 0)
+		return ERR_PTR(ret);
+
 	root = btrfs_read_fs_root_no_radix(fs_info->tree_root, location);
 	if (IS_ERR(root))
 		return root;
 
+	WARN_ON(btrfs_root_refs(&root->root_item) == 0);
 	set_anon_super(&root->anon_super, NULL);
 
+	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
+	if (ret)
+		goto fail;
+
+	spin_lock(&fs_info->fs_roots_radix_lock);
 	ret = radix_tree_insert(&fs_info->fs_roots_radix,
 				(unsigned long)root->root_key.objectid,
 				root);
+	if (ret == 0)
+		root->in_radix = 1;
+	spin_unlock(&fs_info->fs_roots_radix_lock);
+	radix_tree_preload_end();
 	if (ret) {
-		free_extent_buffer(root->node);
-		kfree(root);
-		return ERR_PTR(ret);
+		if (ret == -EEXIST) {
+			free_fs_root(root);
+			goto again;
+		}
+		goto fail;
 	}
-	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
-		ret = btrfs_find_dead_roots(fs_info->tree_root,
-					    root->root_key.objectid);
-		BUG_ON(ret);
+
+	ret = btrfs_find_dead_roots(fs_info->tree_root,
+				    root->root_key.objectid);
+	WARN_ON(ret);
+
+	if (!(fs_info->sb->s_flags & MS_RDONLY))
 		btrfs_orphan_cleanup(root);
-	}
+
 	return root;
+fail:
+	free_fs_root(root);
+	return ERR_PTR(ret);
 }
 
 struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
 				      struct btrfs_key *location,
 				      const char *name, int namelen)
 {
+	return btrfs_read_fs_root_no_name(fs_info, location);
+#if 0
 	struct btrfs_root *root;
 	int ret;
 
@@ -1236,7 +1257,7 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
 		kfree(root);
 		return ERR_PTR(ret);
 	}
-#if 0
+
 	ret = btrfs_sysfs_add_root(root);
 	if (ret) {
 		free_extent_buffer(root->node);
@@ -1244,9 +1265,9 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
 		kfree(root);
 		return ERR_PTR(ret);
 	}
-#endif
 	root->in_sysfs = 1;
 	return root;
+#endif
 }
 
 static int btrfs_congested_fn(void *congested_data, int bdi_bits)
@@ -1325,9 +1346,9 @@ static void btrfs_unplug_io_fn(struct backing_dev_info *bdi, struct page *page)
 	offset = page_offset(page);
 
 	em_tree = &BTRFS_I(inode)->extent_tree;
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 	if (!em) {
 		__unplug_io_fn(bdi, page);
 		return;
@@ -1360,8 +1381,10 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
 
 	err = bdi_register(bdi, NULL, "btrfs-%d",
 				atomic_inc_return(&btrfs_bdi_num));
-	if (err)
+	if (err) {
+		bdi_destroy(bdi);
 		return err;
+	}
 
 	bdi->ra_pages	= default_backing_dev_info.ra_pages;
 	bdi->unplug_io_fn	= btrfs_unplug_io_fn;
@@ -1451,9 +1474,12 @@ static int cleaner_kthread(void *arg)
 			break;
 
 		vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
-		mutex_lock(&root->fs_info->cleaner_mutex);
-		btrfs_clean_old_snapshots(root);
-		mutex_unlock(&root->fs_info->cleaner_mutex);
+
+		if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
+		    mutex_trylock(&root->fs_info->cleaner_mutex)) {
+			btrfs_clean_old_snapshots(root);
+			mutex_unlock(&root->fs_info->cleaner_mutex);
+		}
 
 		if (freezing(current)) {
 			refrigerator();
@@ -1558,15 +1584,36 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 		err = -ENOMEM;
 		goto fail;
 	}
-	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_NOFS);
+
+	ret = init_srcu_struct(&fs_info->subvol_srcu);
+	if (ret) {
+		err = ret;
+		goto fail;
+	}
+
+	ret = setup_bdi(fs_info, &fs_info->bdi);
+	if (ret) {
+		err = ret;
+		goto fail_srcu;
+	}
+
+	fs_info->btree_inode = new_inode(sb);
+	if (!fs_info->btree_inode) {
+		err = -ENOMEM;
+		goto fail_bdi;
+	}
+
+	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
 	INIT_LIST_HEAD(&fs_info->trans_list);
 	INIT_LIST_HEAD(&fs_info->dead_roots);
 	INIT_LIST_HEAD(&fs_info->hashers);
 	INIT_LIST_HEAD(&fs_info->delalloc_inodes);
 	INIT_LIST_HEAD(&fs_info->ordered_operations);
+	INIT_LIST_HEAD(&fs_info->caching_block_groups);
 	spin_lock_init(&fs_info->delalloc_lock);
 	spin_lock_init(&fs_info->new_trans_lock);
 	spin_lock_init(&fs_info->ref_cache_lock);
+	spin_lock_init(&fs_info->fs_roots_radix_lock);
 
 	init_completion(&fs_info->kobj_unregister);
 	fs_info->tree_root = tree_root;
@@ -1585,12 +1632,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->sb = sb;
 	fs_info->max_extent = (u64)-1;
 	fs_info->max_inline = 8192 * 1024;
-	if (setup_bdi(fs_info, &fs_info->bdi))
-		goto fail_bdi;
-	fs_info->btree_inode = new_inode(sb);
-	fs_info->btree_inode->i_ino = 1;
-	fs_info->btree_inode->i_nlink = 1;
-	fs_info->metadata_ratio = 8;
+	fs_info->metadata_ratio = 0;
 
 	fs_info->thread_pool_size = min_t(unsigned long,
 					  num_online_cpus() + 2, 8);
@@ -1602,6 +1644,8 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	sb->s_blocksize_bits = blksize_bits(4096);
 	sb->s_bdi = &fs_info->bdi;
 
+	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
+	fs_info->btree_inode->i_nlink = 1;
 	/*
 	 * we set the i_size on the btree inode to the max possible int.
 	 * the real end of the address space is determined by all of
@@ -1620,28 +1664,32 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 
 	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
 
+	BTRFS_I(fs_info->btree_inode)->root = tree_root;
+	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
+	       sizeof(struct btrfs_key));
+	BTRFS_I(fs_info->btree_inode)->dummy_inode = 1;
+	insert_inode_hash(fs_info->btree_inode);
+
 	spin_lock_init(&fs_info->block_group_cache_lock);
 	fs_info->block_group_cache_tree.rb_node = NULL;
 
-	extent_io_tree_init(&fs_info->pinned_extents,
+	extent_io_tree_init(&fs_info->freed_extents[0],
 			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	extent_io_tree_init(&fs_info->freed_extents[1],
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	fs_info->pinned_extents = &fs_info->freed_extents[0];
 	fs_info->do_barriers = 1;
 
-	BTRFS_I(fs_info->btree_inode)->root = tree_root;
-	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
-	       sizeof(struct btrfs_key));
-	insert_inode_hash(fs_info->btree_inode);
 
 	mutex_init(&fs_info->trans_mutex);
 	mutex_init(&fs_info->ordered_operations_mutex);
 	mutex_init(&fs_info->tree_log_mutex);
-	mutex_init(&fs_info->drop_mutex);
 	mutex_init(&fs_info->chunk_mutex);
 	mutex_init(&fs_info->transaction_kthread_mutex);
 	mutex_init(&fs_info->cleaner_mutex);
 	mutex_init(&fs_info->volume_mutex);
-	mutex_init(&fs_info->tree_reloc_mutex);
 	init_rwsem(&fs_info->extent_commit_sem);
+	init_rwsem(&fs_info->subvol_sem);
 
 	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
 	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);
@@ -1701,20 +1749,24 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 		goto fail_iput;
 	}
 
-	/*
-	 * we need to start all the end_io workers up front because the
-	 * queue work function gets called at interrupt time, and so it
-	 * cannot dynamically grow.
-	 */
+	btrfs_init_workers(&fs_info->generic_worker,
+			   "genwork", 1, NULL);
+
 	btrfs_init_workers(&fs_info->workers, "worker",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 
 	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 
 	btrfs_init_workers(&fs_info->submit_workers, "submit",
 			   min_t(u64, fs_devices->num_devices,
-			   fs_info->thread_pool_size));
+			   fs_info->thread_pool_size),
+			   &fs_info->generic_worker);
+	btrfs_init_workers(&fs_info->enospc_workers, "enospc",
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 
 	/* a higher idle thresh on the submit workers makes it much more
 	 * likely that bios will be send down in a sane order to the
@@ -1728,15 +1780,20 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->delalloc_workers.idle_thresh = 2;
 	fs_info->delalloc_workers.ordered = 1;
 
-	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1);
+	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
+			   &fs_info->generic_worker);
 	btrfs_init_workers(&fs_info->endio_workers, "endio",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 	btrfs_init_workers(&fs_info->endio_meta_write_workers,
-			   "endio-meta-write", fs_info->thread_pool_size);
+			   "endio-meta-write", fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size,
+			   &fs_info->generic_worker);
 
 	/*
 	 * endios are largely parallel and should have a very
@@ -1745,20 +1802,19 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->endio_workers.idle_thresh = 4;
 	fs_info->endio_meta_workers.idle_thresh = 4;
 
-	fs_info->endio_write_workers.idle_thresh = 64;
-	fs_info->endio_meta_write_workers.idle_thresh = 64;
+	fs_info->endio_write_workers.idle_thresh = 2;
+	fs_info->endio_meta_write_workers.idle_thresh = 2;
 
 	btrfs_start_workers(&fs_info->workers, 1);
+	btrfs_start_workers(&fs_info->generic_worker, 1);
 	btrfs_start_workers(&fs_info->submit_workers, 1);
 	btrfs_start_workers(&fs_info->delalloc_workers, 1);
 	btrfs_start_workers(&fs_info->fixup_workers, 1);
-	btrfs_start_workers(&fs_info->endio_workers, fs_info->thread_pool_size);
-	btrfs_start_workers(&fs_info->endio_meta_workers,
-			    fs_info->thread_pool_size);
-	btrfs_start_workers(&fs_info->endio_meta_write_workers,
-			    fs_info->thread_pool_size);
-	btrfs_start_workers(&fs_info->endio_write_workers,
-			    fs_info->thread_pool_size);
+	btrfs_start_workers(&fs_info->endio_workers, 1);
+	btrfs_start_workers(&fs_info->endio_meta_workers, 1);
+	btrfs_start_workers(&fs_info->endio_meta_write_workers, 1);
+	btrfs_start_workers(&fs_info->endio_write_workers, 1);
+	btrfs_start_workers(&fs_info->enospc_workers, 1);
 
 	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
 	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
@@ -1918,6 +1974,9 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 		}
 	}
 
+	ret = btrfs_find_orphan_roots(tree_root);
+	BUG_ON(ret);
+
 	if (!(sb->s_flags & MS_RDONLY)) {
 		ret = btrfs_recover_relocation(tree_root);
 		BUG_ON(ret);
@@ -1961,6 +2020,7 @@ fail_chunk_root:
 	free_extent_buffer(chunk_root->node);
 	free_extent_buffer(chunk_root->commit_root);
 fail_sb_buffer:
+	btrfs_stop_workers(&fs_info->generic_worker);
 	btrfs_stop_workers(&fs_info->fixup_workers);
 	btrfs_stop_workers(&fs_info->delalloc_workers);
 	btrfs_stop_workers(&fs_info->workers);
@@ -1969,6 +2029,7 @@ fail_sb_buffer:
 	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
 	btrfs_stop_workers(&fs_info->endio_write_workers);
 	btrfs_stop_workers(&fs_info->submit_workers);
+	btrfs_stop_workers(&fs_info->enospc_workers);
 fail_iput:
 	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
 	iput(fs_info->btree_inode);
@@ -1977,6 +2038,8 @@ fail_iput:
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 fail_bdi:
 	bdi_destroy(&fs_info->bdi);
+fail_srcu:
+	cleanup_srcu_struct(&fs_info->subvol_srcu);
 fail:
 	kfree(extent_root);
 	kfree(tree_root);
@@ -2236,20 +2299,29 @@ int write_ctree_super(struct btrfs_trans_handle *trans,
 
 int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
 {
-	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
+	spin_lock(&fs_info->fs_roots_radix_lock);
 	radix_tree_delete(&fs_info->fs_roots_radix,
 			  (unsigned long)root->root_key.objectid);
+	spin_unlock(&fs_info->fs_roots_radix_lock);
+
+	if (btrfs_root_refs(&root->root_item) == 0)
+		synchronize_srcu(&fs_info->subvol_srcu);
+
+	free_fs_root(root);
+	return 0;
+}
+
+static void free_fs_root(struct btrfs_root *root)
+{
+	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
 	if (root->anon_super.s_dev) {
 		down_write(&root->anon_super.s_umount);
 		kill_anon_super(&root->anon_super);
 	}
-	if (root->node)
-		free_extent_buffer(root->node);
-	if (root->commit_root)
-		free_extent_buffer(root->commit_root);
+	free_extent_buffer(root->node);
+	free_extent_buffer(root->commit_root);
 	kfree(root->name);
 	kfree(root);
-	return 0;
 }
 
 static int del_fs_roots(struct btrfs_fs_info *fs_info)
@@ -2258,6 +2330,20 @@ static int del_fs_roots(struct btrfs_fs_info *fs_info)
 	struct btrfs_root *gang[8];
 	int i;
 
+	while (!list_empty(&fs_info->dead_roots)) {
+		gang[0] = list_entry(fs_info->dead_roots.next,
+				     struct btrfs_root, root_list);
+		list_del(&gang[0]->root_list);
+
+		if (gang[0]->in_radix) {
+			btrfs_free_fs_root(fs_info, gang[0]);
+		} else {
+			free_extent_buffer(gang[0]->node);
+			free_extent_buffer(gang[0]->commit_root);
+			kfree(gang[0]);
+		}
+	}
+
 	while (1) {
 		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
 					     (void **)gang, 0,
@@ -2287,9 +2373,6 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 		root_objectid = gang[ret - 1]->root_key.objectid + 1;
 		for (i = 0; i < ret; i++) {
 			root_objectid = gang[i]->root_key.objectid;
-			ret = btrfs_find_dead_roots(fs_info->tree_root,
-						    root_objectid);
-			BUG_ON(ret);
 			btrfs_orphan_cleanup(gang[i]);
 		}
 		root_objectid++;
@@ -2359,12 +2442,12 @@ int close_ctree(struct btrfs_root *root)
 	free_extent_buffer(root->fs_info->csum_root->commit_root);
 
 	btrfs_free_block_groups(root->fs_info);
-	btrfs_free_pinned_extents(root->fs_info);
 
 	del_fs_roots(fs_info);
 
 	iput(fs_info->btree_inode);
 
+	btrfs_stop_workers(&fs_info->generic_worker);
 	btrfs_stop_workers(&fs_info->fixup_workers);
 	btrfs_stop_workers(&fs_info->delalloc_workers);
 	btrfs_stop_workers(&fs_info->workers);
@@ -2373,11 +2456,13 @@ int close_ctree(struct btrfs_root *root)
 	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
 	btrfs_stop_workers(&fs_info->endio_write_workers);
 	btrfs_stop_workers(&fs_info->submit_workers);
+	btrfs_stop_workers(&fs_info->enospc_workers);
 
 	btrfs_close_devices(fs_info->fs_devices);
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 
 	bdi_destroy(&fs_info->bdi);
+	cleanup_srcu_struct(&fs_info->subvol_srcu);
 
 	kfree(fs_info->extent_root);
 	kfree(fs_info->tree_root);
diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c
index 9596b40caa4e..ba5c3fd5ab8c 100644
--- a/fs/btrfs/export.c
+++ b/fs/btrfs/export.c
@@ -28,7 +28,7 @@ static int btrfs_encode_fh(struct dentry *dentry, u32 *fh, int *max_len,
 	len  = BTRFS_FID_SIZE_NON_CONNECTABLE;
 	type = FILEID_BTRFS_WITHOUT_PARENT;
 
-	fid->objectid = BTRFS_I(inode)->location.objectid;
+	fid->objectid = inode->i_ino;
 	fid->root_objectid = BTRFS_I(inode)->root->objectid;
 	fid->gen = inode->i_generation;
 
@@ -60,34 +60,61 @@ static int btrfs_encode_fh(struct dentry *dentry, u32 *fh, int *max_len,
 }
 
 static struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
-				       u64 root_objectid, u32 generation)
+				       u64 root_objectid, u32 generation,
+				       int check_generation)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(sb)->fs_info;
 	struct btrfs_root *root;
+	struct dentry *dentry;
 	struct inode *inode;
 	struct btrfs_key key;
+	int index;
+	int err = 0;
+
+	if (objectid < BTRFS_FIRST_FREE_OBJECTID)
+		return ERR_PTR(-ESTALE);
 
 	key.objectid = root_objectid;
 	btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
 	key.offset = (u64)-1;
 
-	root = btrfs_read_fs_root_no_name(btrfs_sb(sb)->fs_info, &key);
-	if (IS_ERR(root))
-		return ERR_CAST(root);
+	index = srcu_read_lock(&fs_info->subvol_srcu);
+
+	root = btrfs_read_fs_root_no_name(fs_info, &key);
+	if (IS_ERR(root)) {
+		err = PTR_ERR(root);
+		goto fail;
+	}
+
+	if (btrfs_root_refs(&root->root_item) == 0) {
+		err = -ENOENT;
+		goto fail;
+	}
 
 	key.objectid = objectid;
 	btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
 	key.offset = 0;
 
 	inode = btrfs_iget(sb, &key, root);
-	if (IS_ERR(inode))
-		return (void *)inode;
+	if (IS_ERR(inode)) {
+		err = PTR_ERR(inode);
+		goto fail;
+	}
+
+	srcu_read_unlock(&fs_info->subvol_srcu, index);
 
-	if (generation != inode->i_generation) {
+	if (check_generation && generation != inode->i_generation) {
 		iput(inode);
 		return ERR_PTR(-ESTALE);
 	}
 
-	return d_obtain_alias(inode);
+	dentry = d_obtain_alias(inode);
+	if (!IS_ERR(dentry))
+		dentry->d_op = &btrfs_dentry_operations;
+	return dentry;
+fail:
+	srcu_read_unlock(&fs_info->subvol_srcu, index);
+	return ERR_PTR(err);
 }
 
 static struct dentry *btrfs_fh_to_parent(struct super_block *sb, struct fid *fh,
@@ -111,7 +138,7 @@ static struct dentry *btrfs_fh_to_parent(struct super_block *sb, struct fid *fh,
 	objectid = fid->parent_objectid;
 	generation = fid->parent_gen;
 
-	return btrfs_get_dentry(sb, objectid, root_objectid, generation);
+	return btrfs_get_dentry(sb, objectid, root_objectid, generation, 1);
 }
 
 static struct dentry *btrfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
@@ -133,66 +160,76 @@ static struct dentry *btrfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
 	root_objectid = fid->root_objectid;
 	generation = fid->gen;
 
-	return btrfs_get_dentry(sb, objectid, root_objectid, generation);
+	return btrfs_get_dentry(sb, objectid, root_objectid, generation, 1);
 }
 
 static struct dentry *btrfs_get_parent(struct dentry *child)
 {
 	struct inode *dir = child->d_inode;
+	static struct dentry *dentry;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
-	struct btrfs_key key;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
-	int slot;
-	u64 objectid;
+	struct btrfs_root_ref *ref;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
 	int ret;
 
 	path = btrfs_alloc_path();
 
-	key.objectid = dir->i_ino;
-	btrfs_set_key_type(&key, BTRFS_INODE_REF_KEY);
-	key.offset = (u64)-1;
+	if (dir->i_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		key.objectid = root->root_key.objectid;
+		key.type = BTRFS_ROOT_BACKREF_KEY;
+		key.offset = (u64)-1;
+		root = root->fs_info->tree_root;
+	} else {
+		key.objectid = dir->i_ino;
+		key.type = BTRFS_INODE_REF_KEY;
+		key.offset = (u64)-1;
+	}
 
 	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0) {
-		/* Error */
-		btrfs_free_path(path);
-		return ERR_PTR(ret);
+	if (ret < 0)
+		goto fail;
+
+	BUG_ON(ret == 0);
+	if (path->slots[0] == 0) {
+		ret = -ENOENT;
+		goto fail;
 	}
+
+	path->slots[0]--;
 	leaf = path->nodes[0];
-	slot = path->slots[0];
-	if (ret) {
-		/* btrfs_search_slot() returns the slot where we'd want to
-		   insert a backref for parent inode #0xFFFFFFFFFFFFFFFF.
-		   The _real_ backref, telling us what the parent inode
-		   _actually_ is, will be in the slot _before_ the one
-		   that btrfs_search_slot() returns. */
-		if (!slot) {
-			/* Unless there is _no_ key in the tree before... */
-			btrfs_free_path(path);
-			return ERR_PTR(-EIO);
-		}
-		slot--;
+
+	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+	if (found_key.objectid != key.objectid || found_key.type != key.type) {
+		ret = -ENOENT;
+		goto fail;
 	}
 
-	btrfs_item_key_to_cpu(leaf, &key, slot);
+	if (found_key.type == BTRFS_ROOT_BACKREF_KEY) {
+		ref = btrfs_item_ptr(leaf, path->slots[0],
+				     struct btrfs_root_ref);
+		key.objectid = btrfs_root_ref_dirid(leaf, ref);
+	} else {
+		key.objectid = found_key.offset;
+	}
 	btrfs_free_path(path);
 
-	if (key.objectid != dir->i_ino || key.type != BTRFS_INODE_REF_KEY)
-		return ERR_PTR(-EINVAL);
-
-	objectid = key.offset;
-
-	/* If we are already at the root of a subvol, return the real root */
-	if (objectid == dir->i_ino)
-		return dget(dir->i_sb->s_root);
+	if (found_key.type == BTRFS_ROOT_BACKREF_KEY) {
+		return btrfs_get_dentry(root->fs_info->sb, key.objectid,
+					found_key.offset, 0, 0);
+	}
 
-	/* Build a new key for the inode item */
-	key.objectid = objectid;
-	btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
+	key.type = BTRFS_INODE_ITEM_KEY;
 	key.offset = 0;
-
-	return d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root));
+	dentry = d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root));
+	if (!IS_ERR(dentry))
+		dentry->d_op = &btrfs_dentry_operations;
+	return dentry;
+fail:
+	btrfs_free_path(path);
+	return ERR_PTR(ret);
 }
 
 const struct export_operations btrfs_export_ops = {
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 535f85ba104f..e238a0cdac67 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -32,12 +32,12 @@
 #include "locking.h"
 #include "free-space-cache.h"
 
-static int update_reserved_extents(struct btrfs_root *root,
-				   u64 bytenr, u64 num, int reserve);
 static int update_block_group(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root,
 			      u64 bytenr, u64 num_bytes, int alloc,
 			      int mark_free);
+static int update_reserved_extents(struct btrfs_block_group_cache *cache,
+				   u64 num_bytes, int reserve);
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes, u64 parent,
@@ -57,10 +57,19 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 				     u64 parent, u64 root_objectid,
 				     u64 flags, struct btrfs_disk_key *key,
 				     int level, struct btrfs_key *ins);
-
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *extent_root, u64 alloc_bytes,
 			  u64 flags, int force);
+static int pin_down_bytes(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path,
+			  u64 bytenr, u64 num_bytes,
+			  int is_data, int reserved,
+			  struct extent_buffer **must_clean);
+static int find_next_key(struct btrfs_path *path, int level,
+			 struct btrfs_key *key);
+static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
+			    int dump_block_groups);
 
 static noinline int
 block_group_cache_done(struct btrfs_block_group_cache *cache)
@@ -153,34 +162,34 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
 	return ret;
 }
 
-/*
- * We always set EXTENT_LOCKED for the super mirror extents so we don't
- * overwrite them, so those bits need to be unset.  Also, if we are unmounting
- * with pinned extents still sitting there because we had a block group caching,
- * we need to clear those now, since we are done.
- */
-void btrfs_free_pinned_extents(struct btrfs_fs_info *info)
+static int add_excluded_extent(struct btrfs_root *root,
+			       u64 start, u64 num_bytes)
 {
-	u64 start, end, last = 0;
-	int ret;
+	u64 end = start + num_bytes - 1;
+	set_extent_bits(&root->fs_info->freed_extents[0],
+			start, end, EXTENT_UPTODATE, GFP_NOFS);
+	set_extent_bits(&root->fs_info->freed_extents[1],
+			start, end, EXTENT_UPTODATE, GFP_NOFS);
+	return 0;
+}
 
-	while (1) {
-		ret = find_first_extent_bit(&info->pinned_extents, last,
-					    &start, &end,
-					    EXTENT_LOCKED|EXTENT_DIRTY);
-		if (ret)
-			break;
+static void free_excluded_extents(struct btrfs_root *root,
+				  struct btrfs_block_group_cache *cache)
+{
+	u64 start, end;
 
-		clear_extent_bits(&info->pinned_extents, start, end,
-				  EXTENT_LOCKED|EXTENT_DIRTY, GFP_NOFS);
-		last = end+1;
-	}
+	start = cache->key.objectid;
+	end = start + cache->key.offset - 1;
+
+	clear_extent_bits(&root->fs_info->freed_extents[0],
+			  start, end, EXTENT_UPTODATE, GFP_NOFS);
+	clear_extent_bits(&root->fs_info->freed_extents[1],
+			  start, end, EXTENT_UPTODATE, GFP_NOFS);
 }
 
-static int remove_sb_from_cache(struct btrfs_root *root,
-				struct btrfs_block_group_cache *cache)
+static int exclude_super_stripes(struct btrfs_root *root,
+				 struct btrfs_block_group_cache *cache)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 bytenr;
 	u64 *logical;
 	int stripe_len;
@@ -192,17 +201,42 @@ static int remove_sb_from_cache(struct btrfs_root *root,
 				       cache->key.objectid, bytenr,
 				       0, &logical, &nr, &stripe_len);
 		BUG_ON(ret);
+
 		while (nr--) {
-			try_lock_extent(&fs_info->pinned_extents,
-					logical[nr],
-					logical[nr] + stripe_len - 1, GFP_NOFS);
+			cache->bytes_super += stripe_len;
+			ret = add_excluded_extent(root, logical[nr],
+						  stripe_len);
+			BUG_ON(ret);
 		}
+
 		kfree(logical);
 	}
-
 	return 0;
 }
 
+static struct btrfs_caching_control *
+get_caching_control(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_caching_control *ctl;
+
+	spin_lock(&cache->lock);
+	if (cache->cached != BTRFS_CACHE_STARTED) {
+		spin_unlock(&cache->lock);
+		return NULL;
+	}
+
+	ctl = cache->caching_ctl;
+	atomic_inc(&ctl->count);
+	spin_unlock(&cache->lock);
+	return ctl;
+}
+
+static void put_caching_control(struct btrfs_caching_control *ctl)
+{
+	if (atomic_dec_and_test(&ctl->count))
+		kfree(ctl);
+}
+
 /*
  * this is only called by cache_block_group, since we could have freed extents
  * we need to check the pinned_extents for any extents that can't be used yet
@@ -215,9 +249,9 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 	int ret;
 
 	while (start < end) {
-		ret = find_first_extent_bit(&info->pinned_extents, start,
+		ret = find_first_extent_bit(info->pinned_extents, start,
 					    &extent_start, &extent_end,
-					    EXTENT_DIRTY|EXTENT_LOCKED);
+					    EXTENT_DIRTY | EXTENT_UPTODATE);
 		if (ret)
 			break;
 
@@ -249,22 +283,27 @@ static int caching_kthread(void *data)
 {
 	struct btrfs_block_group_cache *block_group = data;
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	u64 last = 0;
+	struct btrfs_caching_control *caching_ctl = block_group->caching_ctl;
+	struct btrfs_root *extent_root = fs_info->extent_root;
 	struct btrfs_path *path;
-	int ret = 0;
-	struct btrfs_key key;
 	struct extent_buffer *leaf;
-	int slot;
+	struct btrfs_key key;
 	u64 total_found = 0;
-
-	BUG_ON(!fs_info);
+	u64 last = 0;
+	u32 nritems;
+	int ret = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	atomic_inc(&block_group->space_info->caching_threads);
+	exclude_super_stripes(extent_root, block_group);
+	spin_lock(&block_group->space_info->lock);
+	block_group->space_info->bytes_super += block_group->bytes_super;
+	spin_unlock(&block_group->space_info->lock);
+
 	last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
+
 	/*
 	 * We don't want to deadlock with somebody trying to allocate a new
 	 * extent for the extent root while also trying to search the extent
@@ -277,74 +316,64 @@ static int caching_kthread(void *data)
 
 	key.objectid = last;
 	key.offset = 0;
-	btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+	key.type = BTRFS_EXTENT_ITEM_KEY;
 again:
+	mutex_lock(&caching_ctl->mutex);
 	/* need to make sure the commit_root doesn't disappear */
 	down_read(&fs_info->extent_commit_sem);
 
-	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
 	if (ret < 0)
 		goto err;
 
+	leaf = path->nodes[0];
+	nritems = btrfs_header_nritems(leaf);
+
 	while (1) {
 		smp_mb();
-		if (block_group->fs_info->closing > 1) {
+		if (fs_info->closing > 1) {
 			last = (u64)-1;
 			break;
 		}
 
-		leaf = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(fs_info->extent_root, path);
-			if (ret < 0)
-				goto err;
-			else if (ret)
+		if (path->slots[0] < nritems) {
+			btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		} else {
+			ret = find_next_key(path, 0, &key);
+			if (ret)
 				break;
 
-			if (need_resched() ||
-			    btrfs_transaction_in_commit(fs_info)) {
-				leaf = path->nodes[0];
-
-				/* this shouldn't happen, but if the
-				 * leaf is empty just move on.
-				 */
-				if (btrfs_header_nritems(leaf) == 0)
-					break;
-				/*
-				 * we need to copy the key out so that
-				 * we are sure the next search advances
-				 * us forward in the btree.
-				 */
-				btrfs_item_key_to_cpu(leaf, &key, 0);
-				btrfs_release_path(fs_info->extent_root, path);
-				up_read(&fs_info->extent_commit_sem);
+			caching_ctl->progress = last;
+			btrfs_release_path(extent_root, path);
+			up_read(&fs_info->extent_commit_sem);
+			mutex_unlock(&caching_ctl->mutex);
+			if (btrfs_transaction_in_commit(fs_info))
 				schedule_timeout(1);
-				goto again;
-			}
+			else
+				cond_resched();
+			goto again;
+		}
 
+		if (key.objectid < block_group->key.objectid) {
+			path->slots[0]++;
 			continue;
 		}
-		btrfs_item_key_to_cpu(leaf, &key, slot);
-		if (key.objectid < block_group->key.objectid)
-			goto next;
 
 		if (key.objectid >= block_group->key.objectid +
 		    block_group->key.offset)
 			break;
 
-		if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
+		if (key.type == BTRFS_EXTENT_ITEM_KEY) {
 			total_found += add_new_free_space(block_group,
 							  fs_info, last,
 							  key.objectid);
 			last = key.objectid + key.offset;
-		}
 
-		if (total_found > (1024 * 1024 * 2)) {
-			total_found = 0;
-			wake_up(&block_group->caching_q);
+			if (total_found > (1024 * 1024 * 2)) {
+				total_found = 0;
+				wake_up(&caching_ctl->wait);
+			}
 		}
-next:
 		path->slots[0]++;
 	}
 	ret = 0;
@@ -352,33 +381,65 @@ next:
 	total_found += add_new_free_space(block_group, fs_info, last,
 					  block_group->key.objectid +
 					  block_group->key.offset);
+	caching_ctl->progress = (u64)-1;
 
 	spin_lock(&block_group->lock);
+	block_group->caching_ctl = NULL;
 	block_group->cached = BTRFS_CACHE_FINISHED;
 	spin_unlock(&block_group->lock);
 
 err:
 	btrfs_free_path(path);
 	up_read(&fs_info->extent_commit_sem);
-	atomic_dec(&block_group->space_info->caching_threads);
-	wake_up(&block_group->caching_q);
 
+	free_excluded_extents(extent_root, block_group);
+
+	mutex_unlock(&caching_ctl->mutex);
+	wake_up(&caching_ctl->wait);
+
+	put_caching_control(caching_ctl);
+	atomic_dec(&block_group->space_info->caching_threads);
 	return 0;
 }
 
 static int cache_block_group(struct btrfs_block_group_cache *cache)
 {
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_caching_control *caching_ctl;
 	struct task_struct *tsk;
 	int ret = 0;
 
+	smp_mb();
+	if (cache->cached != BTRFS_CACHE_NO)
+		return 0;
+
+	caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_KERNEL);
+	BUG_ON(!caching_ctl);
+
+	INIT_LIST_HEAD(&caching_ctl->list);
+	mutex_init(&caching_ctl->mutex);
+	init_waitqueue_head(&caching_ctl->wait);
+	caching_ctl->block_group = cache;
+	caching_ctl->progress = cache->key.objectid;
+	/* one for caching kthread, one for caching block group list */
+	atomic_set(&caching_ctl->count, 2);
+
 	spin_lock(&cache->lock);
 	if (cache->cached != BTRFS_CACHE_NO) {
 		spin_unlock(&cache->lock);
-		return ret;
+		kfree(caching_ctl);
+		return 0;
 	}
+	cache->caching_ctl = caching_ctl;
 	cache->cached = BTRFS_CACHE_STARTED;
 	spin_unlock(&cache->lock);
 
+	down_write(&fs_info->extent_commit_sem);
+	list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups);
+	up_write(&fs_info->extent_commit_sem);
+
+	atomic_inc(&cache->space_info->caching_threads);
+
 	tsk = kthread_run(caching_kthread, cache, "btrfs-cache-%llu\n",
 			  cache->key.objectid);
 	if (IS_ERR(tsk)) {
@@ -1507,23 +1568,23 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-#ifdef BIO_RW_DISCARD
 static void btrfs_issue_discard(struct block_device *bdev,
 				u64 start, u64 len)
 {
 	blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL,
 			     DISCARD_FL_BARRIER);
 }
-#endif
 
 static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 				u64 num_bytes)
 {
-#ifdef BIO_RW_DISCARD
 	int ret;
 	u64 map_length = num_bytes;
 	struct btrfs_multi_bio *multi = NULL;
 
+	if (!btrfs_test_opt(root, DISCARD))
+		return 0;
+
 	/* Tell the block device(s) that the sectors can be discarded */
 	ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
 			      bytenr, &map_length, &multi, 0);
@@ -1543,9 +1604,6 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 	}
 
 	return ret;
-#else
-	return 0;
-#endif
 }
 
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
@@ -1657,7 +1715,6 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
 						 parent, ref_root, flags,
 						 ref->objectid, ref->offset,
 						 &ins, node->ref_mod);
-		update_reserved_extents(root, ins.objectid, ins.offset, 0);
 	} else if (node->action == BTRFS_ADD_DELAYED_REF) {
 		ret = __btrfs_inc_extent_ref(trans, root, node->bytenr,
 					     node->num_bytes, parent,
@@ -1783,7 +1840,6 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
 						extent_op->flags_to_set,
 						&extent_op->key,
 						ref->level, &ins);
-		update_reserved_extents(root, ins.objectid, ins.offset, 0);
 	} else if (node->action == BTRFS_ADD_DELAYED_REF) {
 		ret = __btrfs_inc_extent_ref(trans, root, node->bytenr,
 					     node->num_bytes, parent, ref_root,
@@ -1818,16 +1874,32 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 		BUG_ON(extent_op);
 		head = btrfs_delayed_node_to_head(node);
 		if (insert_reserved) {
+			int mark_free = 0;
+			struct extent_buffer *must_clean = NULL;
+
+			ret = pin_down_bytes(trans, root, NULL,
+					     node->bytenr, node->num_bytes,
+					     head->is_data, 1, &must_clean);
+			if (ret > 0)
+				mark_free = 1;
+
+			if (must_clean) {
+				clean_tree_block(NULL, root, must_clean);
+				btrfs_tree_unlock(must_clean);
+				free_extent_buffer(must_clean);
+			}
 			if (head->is_data) {
 				ret = btrfs_del_csums(trans, root,
 						      node->bytenr,
 						      node->num_bytes);
 				BUG_ON(ret);
 			}
-			btrfs_update_pinned_extents(root, node->bytenr,
-						    node->num_bytes, 1);
-			update_reserved_extents(root, node->bytenr,
-						node->num_bytes, 0);
+			if (mark_free) {
+				ret = btrfs_free_reserved_extent(root,
+							node->bytenr,
+							node->num_bytes);
+				BUG_ON(ret);
+			}
 		}
 		mutex_unlock(&head->mutex);
 		return 0;
@@ -2692,60 +2764,448 @@ void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *inode)
 						       alloc_target);
 }
 
+static u64 calculate_bytes_needed(struct btrfs_root *root, int num_items)
+{
+	u64 num_bytes;
+	int level;
+
+	level = BTRFS_MAX_LEVEL - 2;
+	/*
+	 * NOTE: these calculations are absolutely the worst possible case.
+	 * This assumes that _every_ item we insert will require a new leaf, and
+	 * that the tree has grown to its maximum level size.
+	 */
+
+	/*
+	 * for every item we insert we could insert both an extent item and a
+	 * extent ref item.  Then for ever item we insert, we will need to cow
+	 * both the original leaf, plus the leaf to the left and right of it.
+	 *
+	 * Unless we are talking about the extent root, then we just want the
+	 * number of items * 2, since we just need the extent item plus its ref.
+	 */
+	if (root == root->fs_info->extent_root)
+		num_bytes = num_items * 2;
+	else
+		num_bytes = (num_items + (2 * num_items)) * 3;
+
+	/*
+	 * num_bytes is total number of leaves we could need times the leaf
+	 * size, and then for every leaf we could end up cow'ing 2 nodes per
+	 * level, down to the leaf level.
+	 */
+	num_bytes = (num_bytes * root->leafsize) +
+		(num_bytes * (level * 2)) * root->nodesize;
+
+	return num_bytes;
+}
+
 /*
- * for now this just makes sure we have at least 5% of our metadata space free
- * for use.
+ * Unreserve metadata space for delalloc.  If we have less reserved credits than
+ * we have extents, this function does nothing.
  */
-int btrfs_check_metadata_free_space(struct btrfs_root *root)
+int btrfs_unreserve_metadata_for_delalloc(struct btrfs_root *root,
+					  struct inode *inode, int num_items)
 {
 	struct btrfs_fs_info *info = root->fs_info;
 	struct btrfs_space_info *meta_sinfo;
-	u64 alloc_target, thresh;
-	int committed = 0, ret;
+	u64 num_bytes;
+	u64 alloc_target;
+	bool bug = false;
 
 	/* get the space info for where the metadata will live */
 	alloc_target = btrfs_get_alloc_profile(root, 0);
 	meta_sinfo = __find_space_info(info, alloc_target);
 
-again:
+	num_bytes = calculate_bytes_needed(root->fs_info->extent_root,
+					   num_items);
+
 	spin_lock(&meta_sinfo->lock);
-	if (!meta_sinfo->full)
-		thresh = meta_sinfo->total_bytes * 80;
-	else
-		thresh = meta_sinfo->total_bytes * 95;
+	spin_lock(&BTRFS_I(inode)->accounting_lock);
+	if (BTRFS_I(inode)->reserved_extents <=
+	    BTRFS_I(inode)->outstanding_extents) {
+		spin_unlock(&BTRFS_I(inode)->accounting_lock);
+		spin_unlock(&meta_sinfo->lock);
+		return 0;
+	}
+	spin_unlock(&BTRFS_I(inode)->accounting_lock);
+
+	BTRFS_I(inode)->reserved_extents--;
+	BUG_ON(BTRFS_I(inode)->reserved_extents < 0);
+
+	if (meta_sinfo->bytes_delalloc < num_bytes) {
+		bug = true;
+		meta_sinfo->bytes_delalloc = 0;
+	} else {
+		meta_sinfo->bytes_delalloc -= num_bytes;
+	}
+	spin_unlock(&meta_sinfo->lock);
+
+	BUG_ON(bug);
+
+	return 0;
+}
+
+static void check_force_delalloc(struct btrfs_space_info *meta_sinfo)
+{
+	u64 thresh;
+
+	thresh = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
+		meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly +
+		meta_sinfo->bytes_super + meta_sinfo->bytes_root +
+		meta_sinfo->bytes_may_use;
 
+	thresh = meta_sinfo->total_bytes - thresh;
+	thresh *= 80;
 	do_div(thresh, 100);
+	if (thresh <= meta_sinfo->bytes_delalloc)
+		meta_sinfo->force_delalloc = 1;
+	else
+		meta_sinfo->force_delalloc = 0;
+}
 
-	if (meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
-	    meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly > thresh) {
-		struct btrfs_trans_handle *trans;
-		if (!meta_sinfo->full) {
-			meta_sinfo->force_alloc = 1;
-			spin_unlock(&meta_sinfo->lock);
+struct async_flush {
+	struct btrfs_root *root;
+	struct btrfs_space_info *info;
+	struct btrfs_work work;
+};
 
-			trans = btrfs_start_transaction(root, 1);
-			if (!trans)
-				return -ENOMEM;
+static noinline void flush_delalloc_async(struct btrfs_work *work)
+{
+	struct async_flush *async;
+	struct btrfs_root *root;
+	struct btrfs_space_info *info;
 
-			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-					     2 * 1024 * 1024, alloc_target, 0);
-			btrfs_end_transaction(trans, root);
+	async = container_of(work, struct async_flush, work);
+	root = async->root;
+	info = async->info;
+
+	btrfs_start_delalloc_inodes(root);
+	wake_up(&info->flush_wait);
+	btrfs_wait_ordered_extents(root, 0);
+
+	spin_lock(&info->lock);
+	info->flushing = 0;
+	spin_unlock(&info->lock);
+	wake_up(&info->flush_wait);
+
+	kfree(async);
+}
+
+static void wait_on_flush(struct btrfs_space_info *info)
+{
+	DEFINE_WAIT(wait);
+	u64 used;
+
+	while (1) {
+		prepare_to_wait(&info->flush_wait, &wait,
+				TASK_UNINTERRUPTIBLE);
+		spin_lock(&info->lock);
+		if (!info->flushing) {
+			spin_unlock(&info->lock);
+			break;
+		}
+
+		used = info->bytes_used + info->bytes_reserved +
+			info->bytes_pinned + info->bytes_readonly +
+			info->bytes_super + info->bytes_root +
+			info->bytes_may_use + info->bytes_delalloc;
+		if (used < info->total_bytes) {
+			spin_unlock(&info->lock);
+			break;
+		}
+		spin_unlock(&info->lock);
+		schedule();
+	}
+	finish_wait(&info->flush_wait, &wait);
+}
+
+static void flush_delalloc(struct btrfs_root *root,
+				 struct btrfs_space_info *info)
+{
+	struct async_flush *async;
+	bool wait = false;
+
+	spin_lock(&info->lock);
+
+	if (!info->flushing) {
+		info->flushing = 1;
+		init_waitqueue_head(&info->flush_wait);
+	} else {
+		wait = true;
+	}
+
+	spin_unlock(&info->lock);
+
+	if (wait) {
+		wait_on_flush(info);
+		return;
+	}
+
+	async = kzalloc(sizeof(*async), GFP_NOFS);
+	if (!async)
+		goto flush;
+
+	async->root = root;
+	async->info = info;
+	async->work.func = flush_delalloc_async;
+
+	btrfs_queue_worker(&root->fs_info->enospc_workers,
+			   &async->work);
+	wait_on_flush(info);
+	return;
+
+flush:
+	btrfs_start_delalloc_inodes(root);
+	btrfs_wait_ordered_extents(root, 0);
+
+	spin_lock(&info->lock);
+	info->flushing = 0;
+	spin_unlock(&info->lock);
+	wake_up(&info->flush_wait);
+}
+
+static int maybe_allocate_chunk(struct btrfs_root *root,
+				 struct btrfs_space_info *info)
+{
+	struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
+	struct btrfs_trans_handle *trans;
+	bool wait = false;
+	int ret = 0;
+	u64 min_metadata;
+	u64 free_space;
+
+	free_space = btrfs_super_total_bytes(disk_super);
+	/*
+	 * we allow the metadata to grow to a max of either 5gb or 5% of the
+	 * space in the volume.
+	 */
+	min_metadata = min((u64)5 * 1024 * 1024 * 1024,
+			     div64_u64(free_space * 5, 100));
+	if (info->total_bytes >= min_metadata) {
+		spin_unlock(&info->lock);
+		return 0;
+	}
+
+	if (info->full) {
+		spin_unlock(&info->lock);
+		return 0;
+	}
+
+	if (!info->allocating_chunk) {
+		info->force_alloc = 1;
+		info->allocating_chunk = 1;
+		init_waitqueue_head(&info->allocate_wait);
+	} else {
+		wait = true;
+	}
+
+	spin_unlock(&info->lock);
+
+	if (wait) {
+		wait_event(info->allocate_wait,
+			   !info->allocating_chunk);
+		return 1;
+	}
+
+	trans = btrfs_start_transaction(root, 1);
+	if (!trans) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+			     4096 + 2 * 1024 * 1024,
+			     info->flags, 0);
+	btrfs_end_transaction(trans, root);
+	if (ret)
+		goto out;
+out:
+	spin_lock(&info->lock);
+	info->allocating_chunk = 0;
+	spin_unlock(&info->lock);
+	wake_up(&info->allocate_wait);
+
+	if (ret)
+		return 0;
+	return 1;
+}
+
+/*
+ * Reserve metadata space for delalloc.
+ */
+int btrfs_reserve_metadata_for_delalloc(struct btrfs_root *root,
+					struct inode *inode, int num_items)
+{
+	struct btrfs_fs_info *info = root->fs_info;
+	struct btrfs_space_info *meta_sinfo;
+	u64 num_bytes;
+	u64 used;
+	u64 alloc_target;
+	int flushed = 0;
+	int force_delalloc;
+
+	/* get the space info for where the metadata will live */
+	alloc_target = btrfs_get_alloc_profile(root, 0);
+	meta_sinfo = __find_space_info(info, alloc_target);
+
+	num_bytes = calculate_bytes_needed(root->fs_info->extent_root,
+					   num_items);
+again:
+	spin_lock(&meta_sinfo->lock);
+
+	force_delalloc = meta_sinfo->force_delalloc;
+
+	if (unlikely(!meta_sinfo->bytes_root))
+		meta_sinfo->bytes_root = calculate_bytes_needed(root, 6);
+
+	if (!flushed)
+		meta_sinfo->bytes_delalloc += num_bytes;
+
+	used = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
+		meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly +
+		meta_sinfo->bytes_super + meta_sinfo->bytes_root +
+		meta_sinfo->bytes_may_use + meta_sinfo->bytes_delalloc;
+
+	if (used > meta_sinfo->total_bytes) {
+		flushed++;
+
+		if (flushed == 1) {
+			if (maybe_allocate_chunk(root, meta_sinfo))
+				goto again;
+			flushed++;
+		} else {
+			spin_unlock(&meta_sinfo->lock);
+		}
+
+		if (flushed == 2) {
+			filemap_flush(inode->i_mapping);
+			goto again;
+		} else if (flushed == 3) {
+			flush_delalloc(root, meta_sinfo);
 			goto again;
 		}
+		spin_lock(&meta_sinfo->lock);
+		meta_sinfo->bytes_delalloc -= num_bytes;
 		spin_unlock(&meta_sinfo->lock);
+		printk(KERN_ERR "enospc, has %d, reserved %d\n",
+		       BTRFS_I(inode)->outstanding_extents,
+		       BTRFS_I(inode)->reserved_extents);
+		dump_space_info(meta_sinfo, 0, 0);
+		return -ENOSPC;
+	}
 
-		if (!committed) {
-			committed = 1;
-			trans = btrfs_join_transaction(root, 1);
-			if (!trans)
-				return -ENOMEM;
-			ret = btrfs_commit_transaction(trans, root);
-			if (ret)
-				return ret;
+	BTRFS_I(inode)->reserved_extents++;
+	check_force_delalloc(meta_sinfo);
+	spin_unlock(&meta_sinfo->lock);
+
+	if (!flushed && force_delalloc)
+		filemap_flush(inode->i_mapping);
+
+	return 0;
+}
+
+/*
+ * unreserve num_items number of items worth of metadata space.  This needs to
+ * be paired with btrfs_reserve_metadata_space.
+ *
+ * NOTE: if you have the option, run this _AFTER_ you do a
+ * btrfs_end_transaction, since btrfs_end_transaction will run delayed ref
+ * oprations which will result in more used metadata, so we want to make sure we
+ * can do that without issue.
+ */
+int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items)
+{
+	struct btrfs_fs_info *info = root->fs_info;
+	struct btrfs_space_info *meta_sinfo;
+	u64 num_bytes;
+	u64 alloc_target;
+	bool bug = false;
+
+	/* get the space info for where the metadata will live */
+	alloc_target = btrfs_get_alloc_profile(root, 0);
+	meta_sinfo = __find_space_info(info, alloc_target);
+
+	num_bytes = calculate_bytes_needed(root, num_items);
+
+	spin_lock(&meta_sinfo->lock);
+	if (meta_sinfo->bytes_may_use < num_bytes) {
+		bug = true;
+		meta_sinfo->bytes_may_use = 0;
+	} else {
+		meta_sinfo->bytes_may_use -= num_bytes;
+	}
+	spin_unlock(&meta_sinfo->lock);
+
+	BUG_ON(bug);
+
+	return 0;
+}
+
+/*
+ * Reserve some metadata space for use.  We'll calculate the worste case number
+ * of bytes that would be needed to modify num_items number of items.  If we
+ * have space, fantastic, if not, you get -ENOSPC.  Please call
+ * btrfs_unreserve_metadata_space when you are done for the _SAME_ number of
+ * items you reserved, since whatever metadata you needed should have already
+ * been allocated.
+ *
+ * This will commit the transaction to make more space if we don't have enough
+ * metadata space.  THe only time we don't do this is if we're reserving space
+ * inside of a transaction, then we will just return -ENOSPC and it is the
+ * callers responsibility to handle it properly.
+ */
+int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items)
+{
+	struct btrfs_fs_info *info = root->fs_info;
+	struct btrfs_space_info *meta_sinfo;
+	u64 num_bytes;
+	u64 used;
+	u64 alloc_target;
+	int retries = 0;
+
+	/* get the space info for where the metadata will live */
+	alloc_target = btrfs_get_alloc_profile(root, 0);
+	meta_sinfo = __find_space_info(info, alloc_target);
+
+	num_bytes = calculate_bytes_needed(root, num_items);
+again:
+	spin_lock(&meta_sinfo->lock);
+
+	if (unlikely(!meta_sinfo->bytes_root))
+		meta_sinfo->bytes_root = calculate_bytes_needed(root, 6);
+
+	if (!retries)
+		meta_sinfo->bytes_may_use += num_bytes;
+
+	used = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
+		meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly +
+		meta_sinfo->bytes_super + meta_sinfo->bytes_root +
+		meta_sinfo->bytes_may_use + meta_sinfo->bytes_delalloc;
+
+	if (used > meta_sinfo->total_bytes) {
+		retries++;
+		if (retries == 1) {
+			if (maybe_allocate_chunk(root, meta_sinfo))
+				goto again;
+			retries++;
+		} else {
+			spin_unlock(&meta_sinfo->lock);
+		}
+
+		if (retries == 2) {
+			flush_delalloc(root, meta_sinfo);
 			goto again;
 		}
+		spin_lock(&meta_sinfo->lock);
+		meta_sinfo->bytes_may_use -= num_bytes;
+		spin_unlock(&meta_sinfo->lock);
+
+		dump_space_info(meta_sinfo, 0, 0);
 		return -ENOSPC;
 	}
+
+	check_force_delalloc(meta_sinfo);
 	spin_unlock(&meta_sinfo->lock);
 
 	return 0;
@@ -2765,13 +3225,16 @@ int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode,
 	bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
 
 	data_sinfo = BTRFS_I(inode)->space_info;
+	if (!data_sinfo)
+		goto alloc;
+
 again:
 	/* make sure we have enough space to handle the data first */
 	spin_lock(&data_sinfo->lock);
 	if (data_sinfo->total_bytes - data_sinfo->bytes_used -
 	    data_sinfo->bytes_delalloc - data_sinfo->bytes_reserved -
 	    data_sinfo->bytes_pinned - data_sinfo->bytes_readonly -
-	    data_sinfo->bytes_may_use < bytes) {
+	    data_sinfo->bytes_may_use - data_sinfo->bytes_super < bytes) {
 		struct btrfs_trans_handle *trans;
 
 		/*
@@ -2783,7 +3246,7 @@ again:
 
 			data_sinfo->force_alloc = 1;
 			spin_unlock(&data_sinfo->lock);
-
+alloc:
 			alloc_target = btrfs_get_alloc_profile(root, 1);
 			trans = btrfs_start_transaction(root, 1);
 			if (!trans)
@@ -2795,12 +3258,17 @@ again:
 			btrfs_end_transaction(trans, root);
 			if (ret)
 				return ret;
+
+			if (!data_sinfo) {
+				btrfs_set_inode_space_info(root, inode);
+				data_sinfo = BTRFS_I(inode)->space_info;
+			}
 			goto again;
 		}
 		spin_unlock(&data_sinfo->lock);
 
 		/* commit the current transaction and try again */
-		if (!committed) {
+		if (!committed && !root->fs_info->open_ioctl_trans) {
 			committed = 1;
 			trans = btrfs_join_transaction(root, 1);
 			if (!trans)
@@ -2828,7 +3296,7 @@ again:
 	BTRFS_I(inode)->reserved_bytes += bytes;
 	spin_unlock(&data_sinfo->lock);
 
-	return btrfs_check_metadata_free_space(root);
+	return 0;
 }
 
 /*
@@ -2927,17 +3395,15 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	BUG_ON(!space_info);
 
 	spin_lock(&space_info->lock);
-	if (space_info->force_alloc) {
+	if (space_info->force_alloc)
 		force = 1;
-		space_info->force_alloc = 0;
-	}
 	if (space_info->full) {
 		spin_unlock(&space_info->lock);
 		goto out;
 	}
 
 	thresh = space_info->total_bytes - space_info->bytes_readonly;
-	thresh = div_factor(thresh, 6);
+	thresh = div_factor(thresh, 8);
 	if (!force &&
 	   (space_info->bytes_used + space_info->bytes_pinned +
 	    space_info->bytes_reserved + alloc_bytes) < thresh) {
@@ -2951,7 +3417,7 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	 * we keep a reasonable number of metadata chunks allocated in the
 	 * FS as well.
 	 */
-	if (flags & BTRFS_BLOCK_GROUP_DATA) {
+	if (flags & BTRFS_BLOCK_GROUP_DATA && fs_info->metadata_ratio) {
 		fs_info->data_chunk_allocations++;
 		if (!(fs_info->data_chunk_allocations %
 		      fs_info->metadata_ratio))
@@ -2959,8 +3425,11 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	}
 
 	ret = btrfs_alloc_chunk(trans, extent_root, flags);
+	spin_lock(&space_info->lock);
 	if (ret)
 		space_info->full = 1;
+	space_info->force_alloc = 0;
+	spin_unlock(&space_info->lock);
 out:
 	mutex_unlock(&extent_root->fs_info->chunk_mutex);
 	return ret;
@@ -3009,10 +3478,12 @@ static int update_block_group(struct btrfs_trans_handle *trans,
 		num_bytes = min(total, cache->key.offset - byte_in_group);
 		if (alloc) {
 			old_val += num_bytes;
+			btrfs_set_block_group_used(&cache->item, old_val);
+			cache->reserved -= num_bytes;
 			cache->space_info->bytes_used += num_bytes;
+			cache->space_info->bytes_reserved -= num_bytes;
 			if (cache->ro)
 				cache->space_info->bytes_readonly -= num_bytes;
-			btrfs_set_block_group_used(&cache->item, old_val);
 			spin_unlock(&cache->lock);
 			spin_unlock(&cache->space_info->lock);
 		} else {
@@ -3057,127 +3528,136 @@ static u64 first_logical_byte(struct btrfs_root *root, u64 search_start)
 	return bytenr;
 }
 
-int btrfs_update_pinned_extents(struct btrfs_root *root,
-				u64 bytenr, u64 num, int pin)
+/*
+ * this function must be called within transaction
+ */
+int btrfs_pin_extent(struct btrfs_root *root,
+		     u64 bytenr, u64 num_bytes, int reserved)
 {
-	u64 len;
-	struct btrfs_block_group_cache *cache;
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_block_group_cache *cache;
 
-	if (pin)
-		set_extent_dirty(&fs_info->pinned_extents,
-				bytenr, bytenr + num - 1, GFP_NOFS);
-
-	while (num > 0) {
-		cache = btrfs_lookup_block_group(fs_info, bytenr);
-		BUG_ON(!cache);
-		len = min(num, cache->key.offset -
-			  (bytenr - cache->key.objectid));
-		if (pin) {
-			spin_lock(&cache->space_info->lock);
-			spin_lock(&cache->lock);
-			cache->pinned += len;
-			cache->space_info->bytes_pinned += len;
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
-			fs_info->total_pinned += len;
-		} else {
-			int unpin = 0;
+	cache = btrfs_lookup_block_group(fs_info, bytenr);
+	BUG_ON(!cache);
 
-			/*
-			 * in order to not race with the block group caching, we
-			 * only want to unpin the extent if we are cached.  If
-			 * we aren't cached, we want to start async caching this
-			 * block group so we can free the extent the next time
-			 * around.
-			 */
-			spin_lock(&cache->space_info->lock);
-			spin_lock(&cache->lock);
-			unpin = (cache->cached == BTRFS_CACHE_FINISHED);
-			if (likely(unpin)) {
-				cache->pinned -= len;
-				cache->space_info->bytes_pinned -= len;
-				fs_info->total_pinned -= len;
-			}
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
+	spin_lock(&cache->space_info->lock);
+	spin_lock(&cache->lock);
+	cache->pinned += num_bytes;
+	cache->space_info->bytes_pinned += num_bytes;
+	if (reserved) {
+		cache->reserved -= num_bytes;
+		cache->space_info->bytes_reserved -= num_bytes;
+	}
+	spin_unlock(&cache->lock);
+	spin_unlock(&cache->space_info->lock);
 
-			if (likely(unpin))
-				clear_extent_dirty(&fs_info->pinned_extents,
-						   bytenr, bytenr + len -1,
-						   GFP_NOFS);
-			else
-				cache_block_group(cache);
+	btrfs_put_block_group(cache);
 
-			if (unpin)
-				btrfs_add_free_space(cache, bytenr, len);
-		}
-		btrfs_put_block_group(cache);
-		bytenr += len;
-		num -= len;
+	set_extent_dirty(fs_info->pinned_extents,
+			 bytenr, bytenr + num_bytes - 1, GFP_NOFS);
+	return 0;
+}
+
+static int update_reserved_extents(struct btrfs_block_group_cache *cache,
+				   u64 num_bytes, int reserve)
+{
+	spin_lock(&cache->space_info->lock);
+	spin_lock(&cache->lock);
+	if (reserve) {
+		cache->reserved += num_bytes;
+		cache->space_info->bytes_reserved += num_bytes;
+	} else {
+		cache->reserved -= num_bytes;
+		cache->space_info->bytes_reserved -= num_bytes;
 	}
+	spin_unlock(&cache->lock);
+	spin_unlock(&cache->space_info->lock);
 	return 0;
 }
 
-static int update_reserved_extents(struct btrfs_root *root,
-				   u64 bytenr, u64 num, int reserve)
+int btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root)
 {
-	u64 len;
-	struct btrfs_block_group_cache *cache;
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_caching_control *next;
+	struct btrfs_caching_control *caching_ctl;
+	struct btrfs_block_group_cache *cache;
 
-	while (num > 0) {
-		cache = btrfs_lookup_block_group(fs_info, bytenr);
-		BUG_ON(!cache);
-		len = min(num, cache->key.offset -
-			  (bytenr - cache->key.objectid));
+	down_write(&fs_info->extent_commit_sem);
 
-		spin_lock(&cache->space_info->lock);
-		spin_lock(&cache->lock);
-		if (reserve) {
-			cache->reserved += len;
-			cache->space_info->bytes_reserved += len;
+	list_for_each_entry_safe(caching_ctl, next,
+				 &fs_info->caching_block_groups, list) {
+		cache = caching_ctl->block_group;
+		if (block_group_cache_done(cache)) {
+			cache->last_byte_to_unpin = (u64)-1;
+			list_del_init(&caching_ctl->list);
+			put_caching_control(caching_ctl);
 		} else {
-			cache->reserved -= len;
-			cache->space_info->bytes_reserved -= len;
+			cache->last_byte_to_unpin = caching_ctl->progress;
 		}
-		spin_unlock(&cache->lock);
-		spin_unlock(&cache->space_info->lock);
-		btrfs_put_block_group(cache);
-		bytenr += len;
-		num -= len;
 	}
+
+	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
+		fs_info->pinned_extents = &fs_info->freed_extents[1];
+	else
+		fs_info->pinned_extents = &fs_info->freed_extents[0];
+
+	up_write(&fs_info->extent_commit_sem);
 	return 0;
 }
 
-int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
+static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 {
-	u64 last = 0;
-	u64 start;
-	u64 end;
-	struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
-	int ret;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_block_group_cache *cache = NULL;
+	u64 len;
 
-	while (1) {
-		ret = find_first_extent_bit(pinned_extents, last,
-					    &start, &end, EXTENT_DIRTY);
-		if (ret)
-			break;
+	while (start <= end) {
+		if (!cache ||
+		    start >= cache->key.objectid + cache->key.offset) {
+			if (cache)
+				btrfs_put_block_group(cache);
+			cache = btrfs_lookup_block_group(fs_info, start);
+			BUG_ON(!cache);
+		}
 
-		set_extent_dirty(copy, start, end, GFP_NOFS);
-		last = end + 1;
+		len = cache->key.objectid + cache->key.offset - start;
+		len = min(len, end + 1 - start);
+
+		if (start < cache->last_byte_to_unpin) {
+			len = min(len, cache->last_byte_to_unpin - start);
+			btrfs_add_free_space(cache, start, len);
+		}
+
+		spin_lock(&cache->space_info->lock);
+		spin_lock(&cache->lock);
+		cache->pinned -= len;
+		cache->space_info->bytes_pinned -= len;
+		spin_unlock(&cache->lock);
+		spin_unlock(&cache->space_info->lock);
+
+		start += len;
 	}
+
+	if (cache)
+		btrfs_put_block_group(cache);
 	return 0;
 }
 
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct extent_io_tree *unpin)
+			       struct btrfs_root *root)
 {
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct extent_io_tree *unpin;
 	u64 start;
 	u64 end;
 	int ret;
 
+	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
+		unpin = &fs_info->freed_extents[1];
+	else
+		unpin = &fs_info->freed_extents[0];
+
 	while (1) {
 		ret = find_first_extent_bit(unpin, 0, &start, &end,
 					    EXTENT_DIRTY);
@@ -3186,10 +3666,8 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_discard_extent(root, start, end + 1 - start);
 
-		/* unlocks the pinned mutex */
-		btrfs_update_pinned_extents(root, start, end + 1 - start, 0);
 		clear_extent_dirty(unpin, start, end, GFP_NOFS);
-
+		unpin_extent_range(root, start, end);
 		cond_resched();
 	}
 
@@ -3199,7 +3677,8 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 static int pin_down_bytes(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root,
 			  struct btrfs_path *path,
-			  u64 bytenr, u64 num_bytes, int is_data,
+			  u64 bytenr, u64 num_bytes,
+			  int is_data, int reserved,
 			  struct extent_buffer **must_clean)
 {
 	int err = 0;
@@ -3208,6 +3687,14 @@ static int pin_down_bytes(struct btrfs_trans_handle *trans,
 	if (is_data)
 		goto pinit;
 
+	/*
+	 * discard is sloooow, and so triggering discards on
+	 * individual btree blocks isn't a good plan.  Just
+	 * pin everything in discard mode.
+	 */
+	if (btrfs_test_opt(root, DISCARD))
+		goto pinit;
+
 	buf = btrfs_find_tree_block(root, bytenr, num_bytes);
 	if (!buf)
 		goto pinit;
@@ -3231,15 +3718,15 @@ static int pin_down_bytes(struct btrfs_trans_handle *trans,
 	}
 	free_extent_buffer(buf);
 pinit:
-	btrfs_set_path_blocking(path);
+	if (path)
+		btrfs_set_path_blocking(path);
 	/* unlocks the pinned mutex */
-	btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
+	btrfs_pin_extent(root, bytenr, num_bytes, reserved);
 
 	BUG_ON(err < 0);
 	return 0;
 }
 
-
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes, u64 parent,
@@ -3413,7 +3900,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		}
 
 		ret = pin_down_bytes(trans, root, path, bytenr,
-				     num_bytes, is_data, &must_clean);
+				     num_bytes, is_data, 0, &must_clean);
 		if (ret > 0)
 			mark_free = 1;
 		BUG_ON(ret < 0);
@@ -3544,8 +4031,7 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
 	if (root_objectid == BTRFS_TREE_LOG_OBJECTID) {
 		WARN_ON(owner >= BTRFS_FIRST_FREE_OBJECTID);
 		/* unlocks the pinned mutex */
-		btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
-		update_reserved_extents(root, bytenr, num_bytes, 0);
+		btrfs_pin_extent(root, bytenr, num_bytes, 1);
 		ret = 0;
 	} else if (owner < BTRFS_FIRST_FREE_OBJECTID) {
 		ret = btrfs_add_delayed_tree_ref(trans, bytenr, num_bytes,
@@ -3585,19 +4071,33 @@ static noinline int
 wait_block_group_cache_progress(struct btrfs_block_group_cache *cache,
 				u64 num_bytes)
 {
+	struct btrfs_caching_control *caching_ctl;
 	DEFINE_WAIT(wait);
 
-	prepare_to_wait(&cache->caching_q, &wait, TASK_UNINTERRUPTIBLE);
-
-	if (block_group_cache_done(cache)) {
-		finish_wait(&cache->caching_q, &wait);
+	caching_ctl = get_caching_control(cache);
+	if (!caching_ctl)
 		return 0;
-	}
-	schedule();
-	finish_wait(&cache->caching_q, &wait);
 
-	wait_event(cache->caching_q, block_group_cache_done(cache) ||
+	wait_event(caching_ctl->wait, block_group_cache_done(cache) ||
 		   (cache->free_space >= num_bytes));
+
+	put_caching_control(caching_ctl);
+	return 0;
+}
+
+static noinline int
+wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_caching_control *caching_ctl;
+	DEFINE_WAIT(wait);
+
+	caching_ctl = get_caching_control(cache);
+	if (!caching_ctl)
+		return 0;
+
+	wait_event(caching_ctl->wait, block_group_cache_done(cache));
+
+	put_caching_control(caching_ctl);
 	return 0;
 }
 
@@ -3635,6 +4135,8 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	int last_ptr_loop = 0;
 	int loop = 0;
 	bool found_uncached_bg = false;
+	bool failed_cluster_refill = false;
+	bool failed_alloc = false;
 
 	WARN_ON(num_bytes < root->sectorsize);
 	btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
@@ -3732,7 +4234,16 @@ have_block_group:
 		if (unlikely(block_group->ro))
 			goto loop;
 
-		if (last_ptr) {
+		/*
+		 * Ok we want to try and use the cluster allocator, so lets look
+		 * there, unless we are on LOOP_NO_EMPTY_SIZE, since we will
+		 * have tried the cluster allocator plenty of times at this
+		 * point and not have found anything, so we are likely way too
+		 * fragmented for the clustering stuff to find anything, so lets
+		 * just skip it and let the allocator find whatever block it can
+		 * find
+		 */
+		if (last_ptr && loop < LOOP_NO_EMPTY_SIZE) {
 			/*
 			 * the refill lock keeps out other
 			 * people trying to start a new cluster
@@ -3807,9 +4318,11 @@ refill_cluster:
 					spin_unlock(&last_ptr->refill_lock);
 					goto checks;
 				}
-			} else if (!cached && loop > LOOP_CACHING_NOWAIT) {
+			} else if (!cached && loop > LOOP_CACHING_NOWAIT
+				   && !failed_cluster_refill) {
 				spin_unlock(&last_ptr->refill_lock);
 
+				failed_cluster_refill = true;
 				wait_block_group_cache_progress(block_group,
 				       num_bytes + empty_cluster + empty_size);
 				goto have_block_group;
@@ -3821,25 +4334,30 @@ refill_cluster:
 			 * cluster.  Free the cluster we've been trying
 			 * to use, and go to the next block group
 			 */
-			if (loop < LOOP_NO_EMPTY_SIZE) {
-				btrfs_return_cluster_to_free_space(NULL,
-								   last_ptr);
-				spin_unlock(&last_ptr->refill_lock);
-				goto loop;
-			}
+			btrfs_return_cluster_to_free_space(NULL, last_ptr);
 			spin_unlock(&last_ptr->refill_lock);
+			goto loop;
 		}
 
 		offset = btrfs_find_space_for_alloc(block_group, search_start,
 						    num_bytes, empty_size);
-		if (!offset && (cached || (!cached &&
-					   loop == LOOP_CACHING_NOWAIT))) {
-			goto loop;
-		} else if (!offset && (!cached &&
-				       loop > LOOP_CACHING_NOWAIT)) {
+		/*
+		 * If we didn't find a chunk, and we haven't failed on this
+		 * block group before, and this block group is in the middle of
+		 * caching and we are ok with waiting, then go ahead and wait
+		 * for progress to be made, and set failed_alloc to true.
+		 *
+		 * If failed_alloc is true then we've already waited on this
+		 * block group once and should move on to the next block group.
+		 */
+		if (!offset && !failed_alloc && !cached &&
+		    loop > LOOP_CACHING_NOWAIT) {
 			wait_block_group_cache_progress(block_group,
-					num_bytes + empty_size);
+						num_bytes + empty_size);
+			failed_alloc = true;
 			goto have_block_group;
+		} else if (!offset) {
+			goto loop;
 		}
 checks:
 		search_start = stripe_align(root, offset);
@@ -3881,9 +4399,13 @@ checks:
 					     search_start - offset);
 		BUG_ON(offset > search_start);
 
+		update_reserved_extents(block_group, num_bytes, 1);
+
 		/* we are all good, lets return */
 		break;
 loop:
+		failed_cluster_refill = false;
+		failed_alloc = false;
 		btrfs_put_block_group(block_group);
 	}
 	up_read(&space_info->groups_sem);
@@ -3941,21 +4463,32 @@ loop:
 	return ret;
 }
 
-static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
+static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
+			    int dump_block_groups)
 {
 	struct btrfs_block_group_cache *cache;
 
+	spin_lock(&info->lock);
 	printk(KERN_INFO "space_info has %llu free, is %sfull\n",
 	       (unsigned long long)(info->total_bytes - info->bytes_used -
-				    info->bytes_pinned - info->bytes_reserved),
+				    info->bytes_pinned - info->bytes_reserved -
+				    info->bytes_super),
 	       (info->full) ? "" : "not ");
 	printk(KERN_INFO "space_info total=%llu, pinned=%llu, delalloc=%llu,"
-	       " may_use=%llu, used=%llu\n",
+	       " may_use=%llu, used=%llu, root=%llu, super=%llu, reserved=%llu"
+	       "\n",
 	       (unsigned long long)info->total_bytes,
 	       (unsigned long long)info->bytes_pinned,
 	       (unsigned long long)info->bytes_delalloc,
 	       (unsigned long long)info->bytes_may_use,
-	       (unsigned long long)info->bytes_used);
+	       (unsigned long long)info->bytes_used,
+	       (unsigned long long)info->bytes_root,
+	       (unsigned long long)info->bytes_super,
+	       (unsigned long long)info->bytes_reserved);
+	spin_unlock(&info->lock);
+
+	if (!dump_block_groups)
+		return;
 
 	down_read(&info->groups_sem);
 	list_for_each_entry(cache, &info->block_groups, list) {
@@ -3973,12 +4506,12 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
 	up_read(&info->groups_sem);
 }
 
-static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
-				  struct btrfs_root *root,
-				  u64 num_bytes, u64 min_alloc_size,
-				  u64 empty_size, u64 hint_byte,
-				  u64 search_end, struct btrfs_key *ins,
-				  u64 data)
+int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root,
+			 u64 num_bytes, u64 min_alloc_size,
+			 u64 empty_size, u64 hint_byte,
+			 u64 search_end, struct btrfs_key *ins,
+			 u64 data)
 {
 	int ret;
 	u64 search_start = 0;
@@ -4023,7 +4556,7 @@ again:
 		printk(KERN_ERR "btrfs allocation failed flags %llu, "
 		       "wanted %llu\n", (unsigned long long)data,
 		       (unsigned long long)num_bytes);
-		dump_space_info(sinfo, num_bytes);
+		dump_space_info(sinfo, num_bytes, 1);
 	}
 
 	return ret;
@@ -4044,25 +4577,8 @@ int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
 	ret = btrfs_discard_extent(root, start, len);
 
 	btrfs_add_free_space(cache, start, len);
+	update_reserved_extents(cache, len, 0);
 	btrfs_put_block_group(cache);
-	update_reserved_extents(root, start, len, 0);
-
-	return ret;
-}
-
-int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
-				  struct btrfs_root *root,
-				  u64 num_bytes, u64 min_alloc_size,
-				  u64 empty_size, u64 hint_byte,
-				  u64 search_end, struct btrfs_key *ins,
-				  u64 data)
-{
-	int ret;
-	ret = __btrfs_reserve_extent(trans, root, num_bytes, min_alloc_size,
-				     empty_size, hint_byte, search_end, ins,
-				     data);
-	if (!ret)
-		update_reserved_extents(root, ins->objectid, ins->offset, 1);
 
 	return ret;
 }
@@ -4223,15 +4739,46 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 {
 	int ret;
 	struct btrfs_block_group_cache *block_group;
+	struct btrfs_caching_control *caching_ctl;
+	u64 start = ins->objectid;
+	u64 num_bytes = ins->offset;
 
 	block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
 	cache_block_group(block_group);
-	wait_event(block_group->caching_q,
-		   block_group_cache_done(block_group));
+	caching_ctl = get_caching_control(block_group);
 
-	ret = btrfs_remove_free_space(block_group, ins->objectid,
-				      ins->offset);
-	BUG_ON(ret);
+	if (!caching_ctl) {
+		BUG_ON(!block_group_cache_done(block_group));
+		ret = btrfs_remove_free_space(block_group, start, num_bytes);
+		BUG_ON(ret);
+	} else {
+		mutex_lock(&caching_ctl->mutex);
+
+		if (start >= caching_ctl->progress) {
+			ret = add_excluded_extent(root, start, num_bytes);
+			BUG_ON(ret);
+		} else if (start + num_bytes <= caching_ctl->progress) {
+			ret = btrfs_remove_free_space(block_group,
+						      start, num_bytes);
+			BUG_ON(ret);
+		} else {
+			num_bytes = caching_ctl->progress - start;
+			ret = btrfs_remove_free_space(block_group,
+						      start, num_bytes);
+			BUG_ON(ret);
+
+			start = caching_ctl->progress;
+			num_bytes = ins->objectid + ins->offset -
+				    caching_ctl->progress;
+			ret = add_excluded_extent(root, start, num_bytes);
+			BUG_ON(ret);
+		}
+
+		mutex_unlock(&caching_ctl->mutex);
+		put_caching_control(caching_ctl);
+	}
+
+	update_reserved_extents(block_group, ins->offset, 1);
 	btrfs_put_block_group(block_group);
 	ret = alloc_reserved_file_extent(trans, root, 0, root_objectid,
 					 0, owner, offset, ins, 1);
@@ -4255,9 +4802,9 @@ static int alloc_tree_block(struct btrfs_trans_handle *trans,
 	int ret;
 	u64 flags = 0;
 
-	ret = __btrfs_reserve_extent(trans, root, num_bytes, num_bytes,
-				     empty_size, hint_byte, search_end,
-				     ins, 0);
+	ret = btrfs_reserve_extent(trans, root, num_bytes, num_bytes,
+				   empty_size, hint_byte, search_end,
+				   ins, 0);
 	if (ret)
 		return ret;
 
@@ -4268,7 +4815,6 @@ static int alloc_tree_block(struct btrfs_trans_handle *trans,
 	} else
 		BUG_ON(parent > 0);
 
-	update_reserved_extents(root, ins->objectid, ins->offset, 1);
 	if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
 		struct btrfs_delayed_extent_op *extent_op;
 		extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
@@ -4347,452 +4893,108 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 	return buf;
 }
 
-#if 0
-int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
-			struct btrfs_root *root, struct extent_buffer *leaf)
-{
-	u64 disk_bytenr;
-	u64 num_bytes;
-	struct btrfs_key key;
-	struct btrfs_file_extent_item *fi;
-	u32 nritems;
-	int i;
-	int ret;
-
-	BUG_ON(!btrfs_is_leaf(leaf));
-	nritems = btrfs_header_nritems(leaf);
-
-	for (i = 0; i < nritems; i++) {
-		cond_resched();
-		btrfs_item_key_to_cpu(leaf, &key, i);
-
-		/* only extents have references, skip everything else */
-		if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
-			continue;
-
-		fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
-
-		/* inline extents live in the btree, they don't have refs */
-		if (btrfs_file_extent_type(leaf, fi) ==
-		    BTRFS_FILE_EXTENT_INLINE)
-			continue;
-
-		disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
-
-		/* holes don't have refs */
-		if (disk_bytenr == 0)
-			continue;
-
-		num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
-		ret = btrfs_free_extent(trans, root, disk_bytenr, num_bytes,
-					leaf->start, 0, key.objectid, 0);
-		BUG_ON(ret);
-	}
-	return 0;
-}
-
-static noinline int cache_drop_leaf_ref(struct btrfs_trans_handle *trans,
-					struct btrfs_root *root,
-					struct btrfs_leaf_ref *ref)
-{
-	int i;
-	int ret;
-	struct btrfs_extent_info *info;
-	struct refsort *sorted;
-
-	if (ref->nritems == 0)
-		return 0;
-
-	sorted = kmalloc(sizeof(*sorted) * ref->nritems, GFP_NOFS);
-	for (i = 0; i < ref->nritems; i++) {
-		sorted[i].bytenr = ref->extents[i].bytenr;
-		sorted[i].slot = i;
-	}
-	sort(sorted, ref->nritems, sizeof(struct refsort), refsort_cmp, NULL);
-
-	/*
-	 * the items in the ref were sorted when the ref was inserted
-	 * into the ref cache, so this is already in order
-	 */
-	for (i = 0; i < ref->nritems; i++) {
-		info = ref->extents + sorted[i].slot;
-		ret = btrfs_free_extent(trans, root, info->bytenr,
-					  info->num_bytes, ref->bytenr,
-					  ref->owner, ref->generation,
-					  info->objectid, 0);
-
-		atomic_inc(&root->fs_info->throttle_gen);
-		wake_up(&root->fs_info->transaction_throttle);
-		cond_resched();
-
-		BUG_ON(ret);
-		info++;
-	}
-
-	kfree(sorted);
-	return 0;
-}
-
-
-static int drop_snap_lookup_refcount(struct btrfs_trans_handle *trans,
-				     struct btrfs_root *root, u64 start,
-				     u64 len, u32 *refs)
-{
-	int ret;
-
-	ret = btrfs_lookup_extent_refs(trans, root, start, len, refs);
-	BUG_ON(ret);
-
-#if 0 /* some debugging code in case we see problems here */
-	/* if the refs count is one, it won't get increased again.  But
-	 * if the ref count is > 1, someone may be decreasing it at
-	 * the same time we are.
-	 */
-	if (*refs != 1) {
-		struct extent_buffer *eb = NULL;
-		eb = btrfs_find_create_tree_block(root, start, len);
-		if (eb)
-			btrfs_tree_lock(eb);
-
-		mutex_lock(&root->fs_info->alloc_mutex);
-		ret = lookup_extent_ref(NULL, root, start, len, refs);
-		BUG_ON(ret);
-		mutex_unlock(&root->fs_info->alloc_mutex);
-
-		if (eb) {
-			btrfs_tree_unlock(eb);
-			free_extent_buffer(eb);
-		}
-		if (*refs == 1) {
-			printk(KERN_ERR "btrfs block %llu went down to one "
-			       "during drop_snap\n", (unsigned long long)start);
-		}
-
-	}
-#endif
-
-	cond_resched();
-	return ret;
-}
+struct walk_control {
+	u64 refs[BTRFS_MAX_LEVEL];
+	u64 flags[BTRFS_MAX_LEVEL];
+	struct btrfs_key update_progress;
+	int stage;
+	int level;
+	int shared_level;
+	int update_ref;
+	int keep_locks;
+	int reada_slot;
+	int reada_count;
+};
 
+#define DROP_REFERENCE	1
+#define UPDATE_BACKREF	2
 
-/*
- * this is used while deleting old snapshots, and it drops the refs
- * on a whole subtree starting from a level 1 node.
- *
- * The idea is to sort all the leaf pointers, and then drop the
- * ref on all the leaves in order.  Most of the time the leaves
- * will have ref cache entries, so no leaf IOs will be required to
- * find the extents they have references on.
- *
- * For each leaf, any references it has are also dropped in order
- *
- * This ends up dropping the references in something close to optimal
- * order for reading and modifying the extent allocation tree.
- */
-static noinline int drop_level_one_refs(struct btrfs_trans_handle *trans,
-					struct btrfs_root *root,
-					struct btrfs_path *path)
+static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
+				     struct btrfs_root *root,
+				     struct walk_control *wc,
+				     struct btrfs_path *path)
 {
 	u64 bytenr;
-	u64 root_owner;
-	u64 root_gen;
-	struct extent_buffer *eb = path->nodes[1];
-	struct extent_buffer *leaf;
-	struct btrfs_leaf_ref *ref;
-	struct refsort *sorted = NULL;
-	int nritems = btrfs_header_nritems(eb);
+	u64 generation;
+	u64 refs;
+	u64 flags;
+	u64 last = 0;
+	u32 nritems;
+	u32 blocksize;
+	struct btrfs_key key;
+	struct extent_buffer *eb;
 	int ret;
-	int i;
-	int refi = 0;
-	int slot = path->slots[1];
-	u32 blocksize = btrfs_level_size(root, 0);
-	u32 refs;
-
-	if (nritems == 0)
-		goto out;
-
-	root_owner = btrfs_header_owner(eb);
-	root_gen = btrfs_header_generation(eb);
-	sorted = kmalloc(sizeof(*sorted) * nritems, GFP_NOFS);
+	int slot;
+	int nread = 0;
 
-	/*
-	 * step one, sort all the leaf pointers so we don't scribble
-	 * randomly into the extent allocation tree
-	 */
-	for (i = slot; i < nritems; i++) {
-		sorted[refi].bytenr = btrfs_node_blockptr(eb, i);
-		sorted[refi].slot = i;
-		refi++;
+	if (path->slots[wc->level] < wc->reada_slot) {
+		wc->reada_count = wc->reada_count * 2 / 3;
+		wc->reada_count = max(wc->reada_count, 2);
+	} else {
+		wc->reada_count = wc->reada_count * 3 / 2;
+		wc->reada_count = min_t(int, wc->reada_count,
+					BTRFS_NODEPTRS_PER_BLOCK(root));
 	}
 
-	/*
-	 * nritems won't be zero, but if we're picking up drop_snapshot
-	 * after a crash, slot might be > 0, so double check things
-	 * just in case.
-	 */
-	if (refi == 0)
-		goto out;
+	eb = path->nodes[wc->level];
+	nritems = btrfs_header_nritems(eb);
+	blocksize = btrfs_level_size(root, wc->level - 1);
 
-	sort(sorted, refi, sizeof(struct refsort), refsort_cmp, NULL);
+	for (slot = path->slots[wc->level]; slot < nritems; slot++) {
+		if (nread >= wc->reada_count)
+			break;
 
-	/*
-	 * the first loop frees everything the leaves point to
-	 */
-	for (i = 0; i < refi; i++) {
-		u64 ptr_gen;
+		cond_resched();
+		bytenr = btrfs_node_blockptr(eb, slot);
+		generation = btrfs_node_ptr_generation(eb, slot);
 
-		bytenr = sorted[i].bytenr;
+		if (slot == path->slots[wc->level])
+			goto reada;
 
-		/*
-		 * check the reference count on this leaf.  If it is > 1
-		 * we just decrement it below and don't update any
-		 * of the refs the leaf points to.
-		 */
-		ret = drop_snap_lookup_refcount(trans, root, bytenr,
-						blocksize, &refs);
-		BUG_ON(ret);
-		if (refs != 1)
+		if (wc->stage == UPDATE_BACKREF &&
+		    generation <= root->root_key.offset)
 			continue;
 
-		ptr_gen = btrfs_node_ptr_generation(eb, sorted[i].slot);
-
-		/*
-		 * the leaf only had one reference, which means the
-		 * only thing pointing to this leaf is the snapshot
-		 * we're deleting.  It isn't possible for the reference
-		 * count to increase again later
-		 *
-		 * The reference cache is checked for the leaf,
-		 * and if found we'll be able to drop any refs held by
-		 * the leaf without needing to read it in.
-		 */
-		ref = btrfs_lookup_leaf_ref(root, bytenr);
-		if (ref && ref->generation != ptr_gen) {
-			btrfs_free_leaf_ref(root, ref);
-			ref = NULL;
-		}
-		if (ref) {
-			ret = cache_drop_leaf_ref(trans, root, ref);
-			BUG_ON(ret);
-			btrfs_remove_leaf_ref(root, ref);
-			btrfs_free_leaf_ref(root, ref);
-		} else {
-			/*
-			 * the leaf wasn't in the reference cache, so
-			 * we have to read it.
-			 */
-			leaf = read_tree_block(root, bytenr, blocksize,
-					       ptr_gen);
-			ret = btrfs_drop_leaf_ref(trans, root, leaf);
-			BUG_ON(ret);
-			free_extent_buffer(leaf);
-		}
-		atomic_inc(&root->fs_info->throttle_gen);
-		wake_up(&root->fs_info->transaction_throttle);
-		cond_resched();
-	}
-
-	/*
-	 * run through the loop again to free the refs on the leaves.
-	 * This is faster than doing it in the loop above because
-	 * the leaves are likely to be clustered together.  We end up
-	 * working in nice chunks on the extent allocation tree.
-	 */
-	for (i = 0; i < refi; i++) {
-		bytenr = sorted[i].bytenr;
-		ret = btrfs_free_extent(trans, root, bytenr,
-					blocksize, eb->start,
-					root_owner, root_gen, 0, 1);
+		/* We don't lock the tree block, it's OK to be racy here */
+		ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
+					       &refs, &flags);
 		BUG_ON(ret);
+		BUG_ON(refs == 0);
 
-		atomic_inc(&root->fs_info->throttle_gen);
-		wake_up(&root->fs_info->transaction_throttle);
-		cond_resched();
-	}
-out:
-	kfree(sorted);
-
-	/*
-	 * update the path to show we've processed the entire level 1
-	 * node.  This will get saved into the root's drop_snapshot_progress
-	 * field so these drops are not repeated again if this transaction
-	 * commits.
-	 */
-	path->slots[1] = nritems;
-	return 0;
-}
-
-/*
- * helper function for drop_snapshot, this walks down the tree dropping ref
- * counts as it goes.
- */
-static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
-				   struct btrfs_root *root,
-				   struct btrfs_path *path, int *level)
-{
-	u64 root_owner;
-	u64 root_gen;
-	u64 bytenr;
-	u64 ptr_gen;
-	struct extent_buffer *next;
-	struct extent_buffer *cur;
-	struct extent_buffer *parent;
-	u32 blocksize;
-	int ret;
-	u32 refs;
-
-	WARN_ON(*level < 0);
-	WARN_ON(*level >= BTRFS_MAX_LEVEL);
-	ret = drop_snap_lookup_refcount(trans, root, path->nodes[*level]->start,
-				path->nodes[*level]->len, &refs);
-	BUG_ON(ret);
-	if (refs > 1)
-		goto out;
-
-	/*
-	 * walk down to the last node level and free all the leaves
-	 */
-	while (*level >= 0) {
-		WARN_ON(*level < 0);
-		WARN_ON(*level >= BTRFS_MAX_LEVEL);
-		cur = path->nodes[*level];
-
-		if (btrfs_header_level(cur) != *level)
-			WARN_ON(1);
-
-		if (path->slots[*level] >=
-		    btrfs_header_nritems(cur))
-			break;
+		if (wc->stage == DROP_REFERENCE) {
+			if (refs == 1)
+				goto reada;
 
-		/* the new code goes down to level 1 and does all the
-		 * leaves pointed to that node in bulk.  So, this check
-		 * for level 0 will always be false.
-		 *
-		 * But, the disk format allows the drop_snapshot_progress
-		 * field in the root to leave things in a state where
-		 * a leaf will need cleaning up here.  If someone crashes
-		 * with the old code and then boots with the new code,
-		 * we might find a leaf here.
-		 */
-		if (*level == 0) {
-			ret = btrfs_drop_leaf_ref(trans, root, cur);
-			BUG_ON(ret);
-			break;
+			if (wc->level == 1 &&
+			    (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
+				continue;
+			if (!wc->update_ref ||
+			    generation <= root->root_key.offset)
+				continue;
+			btrfs_node_key_to_cpu(eb, &key, slot);
+			ret = btrfs_comp_cpu_keys(&key,
+						  &wc->update_progress);
+			if (ret < 0)
+				continue;
+		} else {
+			if (wc->level == 1 &&
+			    (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
+				continue;
 		}
-
-		/*
-		 * once we get to level one, process the whole node
-		 * at once, including everything below it.
-		 */
-		if (*level == 1) {
-			ret = drop_level_one_refs(trans, root, path);
-			BUG_ON(ret);
+reada:
+		ret = readahead_tree_block(root, bytenr, blocksize,
+					   generation);
+		if (ret)
 			break;
-		}
-
-		bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
-		ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
-		blocksize = btrfs_level_size(root, *level - 1);
-
-		ret = drop_snap_lookup_refcount(trans, root, bytenr,
-						blocksize, &refs);
-		BUG_ON(ret);
-
-		/*
-		 * if there is more than one reference, we don't need
-		 * to read that node to drop any references it has.  We
-		 * just drop the ref we hold on that node and move on to the
-		 * next slot in this level.
-		 */
-		if (refs != 1) {
-			parent = path->nodes[*level];
-			root_owner = btrfs_header_owner(parent);
-			root_gen = btrfs_header_generation(parent);
-			path->slots[*level]++;
-
-			ret = btrfs_free_extent(trans, root, bytenr,
-						blocksize, parent->start,
-						root_owner, root_gen,
-						*level - 1, 1);
-			BUG_ON(ret);
-
-			atomic_inc(&root->fs_info->throttle_gen);
-			wake_up(&root->fs_info->transaction_throttle);
-			cond_resched();
-
-			continue;
-		}
-
-		/*
-		 * we need to keep freeing things in the next level down.
-		 * read the block and loop around to process it
-		 */
-		next = read_tree_block(root, bytenr, blocksize, ptr_gen);
-		WARN_ON(*level <= 0);
-		if (path->nodes[*level-1])
-			free_extent_buffer(path->nodes[*level-1]);
-		path->nodes[*level-1] = next;
-		*level = btrfs_header_level(next);
-		path->slots[*level] = 0;
-		cond_resched();
+		last = bytenr + blocksize;
+		nread++;
 	}
-out:
-	WARN_ON(*level < 0);
-	WARN_ON(*level >= BTRFS_MAX_LEVEL);
-
-	if (path->nodes[*level] == root->node) {
-		parent = path->nodes[*level];
-		bytenr = path->nodes[*level]->start;
-	} else {
-		parent = path->nodes[*level + 1];
-		bytenr = btrfs_node_blockptr(parent, path->slots[*level + 1]);
-	}
-
-	blocksize = btrfs_level_size(root, *level);
-	root_owner = btrfs_header_owner(parent);
-	root_gen = btrfs_header_generation(parent);
-
-	/*
-	 * cleanup and free the reference on the last node
-	 * we processed
-	 */
-	ret = btrfs_free_extent(trans, root, bytenr, blocksize,
-				  parent->start, root_owner, root_gen,
-				  *level, 1);
-	free_extent_buffer(path->nodes[*level]);
-	path->nodes[*level] = NULL;
-
-	*level += 1;
-	BUG_ON(ret);
-
-	cond_resched();
-	return 0;
+	wc->reada_slot = slot;
 }
-#endif
-
-struct walk_control {
-	u64 refs[BTRFS_MAX_LEVEL];
-	u64 flags[BTRFS_MAX_LEVEL];
-	struct btrfs_key update_progress;
-	int stage;
-	int level;
-	int shared_level;
-	int update_ref;
-	int keep_locks;
-};
-
-#define DROP_REFERENCE	1
-#define UPDATE_BACKREF	2
 
 /*
  * hepler to process tree block while walking down the tree.
  *
- * when wc->stage == DROP_REFERENCE, this function checks
- * reference count of the block. if the block is shared and
- * we need update back refs for the subtree rooted at the
- * block, this function changes wc->stage to UPDATE_BACKREF
- *
  * when wc->stage == UPDATE_BACKREF, this function updates
  * back refs for pointers in the block.
  *
@@ -4801,11 +5003,10 @@ struct walk_control {
 static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root,
 				   struct btrfs_path *path,
-				   struct walk_control *wc)
+				   struct walk_control *wc, int lookup_info)
 {
 	int level = wc->level;
 	struct extent_buffer *eb = path->nodes[level];
-	struct btrfs_key key;
 	u64 flag = BTRFS_BLOCK_FLAG_FULL_BACKREF;
 	int ret;
 
@@ -4817,8 +5018,9 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 	 * when reference count of tree block is 1, it won't increase
 	 * again. once full backref flag is set, we never clear it.
 	 */
-	if ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
-	    (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag))) {
+	if (lookup_info &&
+	    ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
+	     (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) {
 		BUG_ON(!path->locks[level]);
 		ret = btrfs_lookup_extent_info(trans, root,
 					       eb->start, eb->len,
@@ -4828,21 +5030,6 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 		BUG_ON(wc->refs[level] == 0);
 	}
 
-	if (wc->stage == DROP_REFERENCE &&
-	    wc->update_ref && wc->refs[level] > 1) {
-		BUG_ON(eb == root->node);
-		BUG_ON(path->slots[level] > 0);
-		if (level == 0)
-			btrfs_item_key_to_cpu(eb, &key, path->slots[level]);
-		else
-			btrfs_node_key_to_cpu(eb, &key, path->slots[level]);
-		if (btrfs_header_owner(eb) == root->root_key.objectid &&
-		    btrfs_comp_cpu_keys(&key, &wc->update_progress) >= 0) {
-			wc->stage = UPDATE_BACKREF;
-			wc->shared_level = level;
-		}
-	}
-
 	if (wc->stage == DROP_REFERENCE) {
 		if (wc->refs[level] > 1)
 			return 1;
@@ -4879,6 +5066,136 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 }
 
 /*
+ * hepler to process tree block pointer.
+ *
+ * when wc->stage == DROP_REFERENCE, this function checks
+ * reference count of the block pointed to. if the block
+ * is shared and we need update back refs for the subtree
+ * rooted at the block, this function changes wc->stage to
+ * UPDATE_BACKREF. if the block is shared and there is no
+ * need to update back, this function drops the reference
+ * to the block.
+ *
+ * NOTE: return value 1 means we should stop walking down.
+ */
+static noinline int do_walk_down(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 struct walk_control *wc, int *lookup_info)
+{
+	u64 bytenr;
+	u64 generation;
+	u64 parent;
+	u32 blocksize;
+	struct btrfs_key key;
+	struct extent_buffer *next;
+	int level = wc->level;
+	int reada = 0;
+	int ret = 0;
+
+	generation = btrfs_node_ptr_generation(path->nodes[level],
+					       path->slots[level]);
+	/*
+	 * if the lower level block was created before the snapshot
+	 * was created, we know there is no need to update back refs
+	 * for the subtree
+	 */
+	if (wc->stage == UPDATE_BACKREF &&
+	    generation <= root->root_key.offset) {
+		*lookup_info = 1;
+		return 1;
+	}
+
+	bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]);
+	blocksize = btrfs_level_size(root, level - 1);
+
+	next = btrfs_find_tree_block(root, bytenr, blocksize);
+	if (!next) {
+		next = btrfs_find_create_tree_block(root, bytenr, blocksize);
+		reada = 1;
+	}
+	btrfs_tree_lock(next);
+	btrfs_set_lock_blocking(next);
+
+	ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
+				       &wc->refs[level - 1],
+				       &wc->flags[level - 1]);
+	BUG_ON(ret);
+	BUG_ON(wc->refs[level - 1] == 0);
+	*lookup_info = 0;
+
+	if (wc->stage == DROP_REFERENCE) {
+		if (wc->refs[level - 1] > 1) {
+			if (level == 1 &&
+			    (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
+				goto skip;
+
+			if (!wc->update_ref ||
+			    generation <= root->root_key.offset)
+				goto skip;
+
+			btrfs_node_key_to_cpu(path->nodes[level], &key,
+					      path->slots[level]);
+			ret = btrfs_comp_cpu_keys(&key, &wc->update_progress);
+			if (ret < 0)
+				goto skip;
+
+			wc->stage = UPDATE_BACKREF;
+			wc->shared_level = level - 1;
+		}
+	} else {
+		if (level == 1 &&
+		    (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
+			goto skip;
+	}
+
+	if (!btrfs_buffer_uptodate(next, generation)) {
+		btrfs_tree_unlock(next);
+		free_extent_buffer(next);
+		next = NULL;
+		*lookup_info = 1;
+	}
+
+	if (!next) {
+		if (reada && level == 1)
+			reada_walk_down(trans, root, wc, path);
+		next = read_tree_block(root, bytenr, blocksize, generation);
+		btrfs_tree_lock(next);
+		btrfs_set_lock_blocking(next);
+	}
+
+	level--;
+	BUG_ON(level != btrfs_header_level(next));
+	path->nodes[level] = next;
+	path->slots[level] = 0;
+	path->locks[level] = 1;
+	wc->level = level;
+	if (wc->level == 1)
+		wc->reada_slot = 0;
+	return 0;
+skip:
+	wc->refs[level - 1] = 0;
+	wc->flags[level - 1] = 0;
+	if (wc->stage == DROP_REFERENCE) {
+		if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
+			parent = path->nodes[level]->start;
+		} else {
+			BUG_ON(root->root_key.objectid !=
+			       btrfs_header_owner(path->nodes[level]));
+			parent = 0;
+		}
+
+		ret = btrfs_free_extent(trans, root, bytenr, blocksize, parent,
+					root->root_key.objectid, level - 1, 0);
+		BUG_ON(ret);
+	}
+	btrfs_tree_unlock(next);
+	free_extent_buffer(next);
+	*lookup_info = 1;
+	return 1;
+}
+
+/*
  * hepler to process tree block while walking up the tree.
  *
  * when wc->stage == DROP_REFERENCE, this function drops
@@ -4905,7 +5222,6 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 		if (level < wc->shared_level)
 			goto out;
 
-		BUG_ON(wc->refs[level] <= 1);
 		ret = find_next_key(path, level + 1, &wc->update_progress);
 		if (ret > 0)
 			wc->update_ref = 0;
@@ -4936,8 +5252,6 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 				path->locks[level] = 0;
 				return 1;
 			}
-		} else {
-			BUG_ON(level != 0);
 		}
 	}
 
@@ -4990,39 +5304,28 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
 				   struct btrfs_path *path,
 				   struct walk_control *wc)
 {
-	struct extent_buffer *next;
-	struct extent_buffer *cur;
-	u64 bytenr;
-	u64 ptr_gen;
-	u32 blocksize;
 	int level = wc->level;
+	int lookup_info = 1;
 	int ret;
 
 	while (level >= 0) {
-		cur = path->nodes[level];
-		BUG_ON(path->slots[level] >= btrfs_header_nritems(cur));
+		if (path->slots[level] >=
+		    btrfs_header_nritems(path->nodes[level]))
+			break;
 
-		ret = walk_down_proc(trans, root, path, wc);
+		ret = walk_down_proc(trans, root, path, wc, lookup_info);
 		if (ret > 0)
 			break;
 
 		if (level == 0)
 			break;
 
-		bytenr = btrfs_node_blockptr(cur, path->slots[level]);
-		blocksize = btrfs_level_size(root, level - 1);
-		ptr_gen = btrfs_node_ptr_generation(cur, path->slots[level]);
-
-		next = read_tree_block(root, bytenr, blocksize, ptr_gen);
-		btrfs_tree_lock(next);
-		btrfs_set_lock_blocking(next);
-
-		level--;
-		BUG_ON(level != btrfs_header_level(next));
-		path->nodes[level] = next;
-		path->slots[level] = 0;
-		path->locks[level] = 1;
-		wc->level = level;
+		ret = do_walk_down(trans, root, path, wc, &lookup_info);
+		if (ret > 0) {
+			path->slots[level]++;
+			continue;
+		}
+		level = wc->level;
 	}
 	return 0;
 }
@@ -5112,9 +5415,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref)
 			err = ret;
 			goto out;
 		}
-		btrfs_node_key_to_cpu(path->nodes[level], &key,
-				      path->slots[level]);
-		WARN_ON(memcmp(&key, &wc->update_progress, sizeof(key)));
+		WARN_ON(ret > 0);
 
 		/*
 		 * unlock our path, this is safe because only this
@@ -5149,6 +5450,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref)
 	wc->stage = DROP_REFERENCE;
 	wc->update_ref = update_ref;
 	wc->keep_locks = 0;
+	wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(root);
 
 	while (1) {
 		ret = walk_down_tree(trans, root, path, wc);
@@ -5201,9 +5503,24 @@ int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref)
 	ret = btrfs_del_root(trans, tree_root, &root->root_key);
 	BUG_ON(ret);
 
-	free_extent_buffer(root->node);
-	free_extent_buffer(root->commit_root);
-	kfree(root);
+	if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
+		ret = btrfs_find_last_root(tree_root, root->root_key.objectid,
+					   NULL, NULL);
+		BUG_ON(ret < 0);
+		if (ret > 0) {
+			ret = btrfs_del_orphan_item(trans, tree_root,
+						    root->root_key.objectid);
+			BUG_ON(ret);
+		}
+	}
+
+	if (root->in_radix) {
+		btrfs_free_fs_root(tree_root->fs_info, root);
+	} else {
+		free_extent_buffer(root->node);
+		free_extent_buffer(root->commit_root);
+		kfree(root);
+	}
 out:
 	btrfs_end_transaction(trans, tree_root);
 	kfree(wc);
@@ -5255,6 +5572,7 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 	wc->stage = DROP_REFERENCE;
 	wc->update_ref = 0;
 	wc->keep_locks = 1;
+	wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(root);
 
 	while (1) {
 		wret = walk_down_tree(trans, root, path, wc);
@@ -5397,9 +5715,9 @@ static noinline int relocate_data_extent(struct inode *reloc_inode,
 	lock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS);
 	while (1) {
 		int ret;
-		spin_lock(&em_tree->lock);
+		write_lock(&em_tree->lock);
 		ret = add_extent_mapping(em_tree, em);
-		spin_unlock(&em_tree->lock);
+		write_unlock(&em_tree->lock);
 		if (ret != -EEXIST) {
 			free_extent_map(em);
 			break;
@@ -6842,287 +7160,86 @@ int btrfs_prepare_block_group_relocation(struct btrfs_root *root,
 	return 0;
 }
 
-#if 0
-static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root,
-				 u64 objectid, u64 size)
-{
-	struct btrfs_path *path;
-	struct btrfs_inode_item *item;
-	struct extent_buffer *leaf;
-	int ret;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	path->leave_spinning = 1;
-	ret = btrfs_insert_empty_inode(trans, root, path, objectid);
-	if (ret)
-		goto out;
-
-	leaf = path->nodes[0];
-	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
-	memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
-	btrfs_set_inode_generation(leaf, item, 1);
-	btrfs_set_inode_size(leaf, item, size);
-	btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
-	btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NOCOMPRESS);
-	btrfs_mark_buffer_dirty(leaf);
-	btrfs_release_path(root, path);
-out:
-	btrfs_free_path(path);
-	return ret;
-}
-
-static noinline struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
-					struct btrfs_block_group_cache *group)
+/*
+ * checks to see if its even possible to relocate this block group.
+ *
+ * @return - -1 if it's not a good idea to relocate this block group, 0 if its
+ * ok to go ahead and try.
+ */
+int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr)
 {
-	struct inode *inode = NULL;
-	struct btrfs_trans_handle *trans;
-	struct btrfs_root *root;
-	struct btrfs_key root_key;
-	u64 objectid = BTRFS_FIRST_FREE_OBJECTID;
-	int err = 0;
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_space_info *space_info;
+	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
+	struct btrfs_device *device;
+	int full = 0;
+	int ret = 0;
 
-	root_key.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
-	root_key.type = BTRFS_ROOT_ITEM_KEY;
-	root_key.offset = (u64)-1;
-	root = btrfs_read_fs_root_no_name(fs_info, &root_key);
-	if (IS_ERR(root))
-		return ERR_CAST(root);
+	block_group = btrfs_lookup_block_group(root->fs_info, bytenr);
 
-	trans = btrfs_start_transaction(root, 1);
-	BUG_ON(!trans);
+	/* odd, couldn't find the block group, leave it alone */
+	if (!block_group)
+		return -1;
 
-	err = btrfs_find_free_objectid(trans, root, objectid, &objectid);
-	if (err)
+	/* no bytes used, we're good */
+	if (!btrfs_block_group_used(&block_group->item))
 		goto out;
 
-	err = __insert_orphan_inode(trans, root, objectid, group->key.offset);
-	BUG_ON(err);
-
-	err = btrfs_insert_file_extent(trans, root, objectid, 0, 0, 0,
-				       group->key.offset, 0, group->key.offset,
-				       0, 0, 0);
-	BUG_ON(err);
-
-	inode = btrfs_iget_locked(root->fs_info->sb, objectid, root);
-	if (inode->i_state & I_NEW) {
-		BTRFS_I(inode)->root = root;
-		BTRFS_I(inode)->location.objectid = objectid;
-		BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
-		BTRFS_I(inode)->location.offset = 0;
-		btrfs_read_locked_inode(inode);
-		unlock_new_inode(inode);
-		BUG_ON(is_bad_inode(inode));
-	} else {
-		BUG_ON(1);
-	}
-	BTRFS_I(inode)->index_cnt = group->key.objectid;
-
-	err = btrfs_orphan_add(trans, inode);
-out:
-	btrfs_end_transaction(trans, root);
-	if (err) {
-		if (inode)
-			iput(inode);
-		inode = ERR_PTR(err);
-	}
-	return inode;
-}
-
-int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
-{
-
-	struct btrfs_ordered_sum *sums;
-	struct btrfs_sector_sum *sector_sum;
-	struct btrfs_ordered_extent *ordered;
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct list_head list;
-	size_t offset;
-	int ret;
-	u64 disk_bytenr;
-
-	INIT_LIST_HEAD(&list);
-
-	ordered = btrfs_lookup_ordered_extent(inode, file_pos);
-	BUG_ON(ordered->file_offset != file_pos || ordered->len != len);
-
-	disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt;
-	ret = btrfs_lookup_csums_range(root->fs_info->csum_root, disk_bytenr,
-				       disk_bytenr + len - 1, &list);
-
-	while (!list_empty(&list)) {
-		sums = list_entry(list.next, struct btrfs_ordered_sum, list);
-		list_del_init(&sums->list);
-
-		sector_sum = sums->sums;
-		sums->bytenr = ordered->start;
+	space_info = block_group->space_info;
+	spin_lock(&space_info->lock);
 
-		offset = 0;
-		while (offset < sums->len) {
-			sector_sum->bytenr += ordered->start - disk_bytenr;
-			sector_sum++;
-			offset += root->sectorsize;
-		}
+	full = space_info->full;
 
-		btrfs_add_ordered_sum(inode, ordered, sums);
+	/*
+	 * if this is the last block group we have in this space, we can't
+	 * relocate it unless we're able to allocate a new chunk below.
+	 *
+	 * Otherwise, we need to make sure we have room in the space to handle
+	 * all of the extents from this block group.  If we can, we're good
+	 */
+	if ((space_info->total_bytes != block_group->key.offset) &&
+	   (space_info->bytes_used + space_info->bytes_reserved +
+	    space_info->bytes_pinned + space_info->bytes_readonly +
+	    btrfs_block_group_used(&block_group->item) <
+	    space_info->total_bytes)) {
+		spin_unlock(&space_info->lock);
+		goto out;
 	}
-	btrfs_put_ordered_extent(ordered);
-	return 0;
-}
-
-int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start)
-{
-	struct btrfs_trans_handle *trans;
-	struct btrfs_path *path;
-	struct btrfs_fs_info *info = root->fs_info;
-	struct extent_buffer *leaf;
-	struct inode *reloc_inode;
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_key key;
-	u64 skipped;
-	u64 cur_byte;
-	u64 total_found;
-	u32 nritems;
-	int ret;
-	int progress;
-	int pass = 0;
-
-	root = root->fs_info->extent_root;
-
-	block_group = btrfs_lookup_block_group(info, group_start);
-	BUG_ON(!block_group);
-
-	printk(KERN_INFO "btrfs relocating block group %llu flags %llu\n",
-	       (unsigned long long)block_group->key.objectid,
-	       (unsigned long long)block_group->flags);
-
-	path = btrfs_alloc_path();
-	BUG_ON(!path);
-
-	reloc_inode = create_reloc_inode(info, block_group);
-	BUG_ON(IS_ERR(reloc_inode));
-
-	__alloc_chunk_for_shrink(root, block_group, 1);
-	set_block_group_readonly(block_group);
-
-	btrfs_start_delalloc_inodes(info->tree_root);
-	btrfs_wait_ordered_extents(info->tree_root, 0);
-again:
-	skipped = 0;
-	total_found = 0;
-	progress = 0;
-	key.objectid = block_group->key.objectid;
-	key.offset = 0;
-	key.type = 0;
-	cur_byte = key.objectid;
-
-	trans = btrfs_start_transaction(info->tree_root, 1);
-	btrfs_commit_transaction(trans, info->tree_root);
+	spin_unlock(&space_info->lock);
 
-	mutex_lock(&root->fs_info->cleaner_mutex);
-	btrfs_clean_old_snapshots(info->tree_root);
-	btrfs_remove_leaf_refs(info->tree_root, (u64)-1, 1);
-	mutex_unlock(&root->fs_info->cleaner_mutex);
+	/*
+	 * ok we don't have enough space, but maybe we have free space on our
+	 * devices to allocate new chunks for relocation, so loop through our
+	 * alloc devices and guess if we have enough space.  However, if we
+	 * were marked as full, then we know there aren't enough chunks, and we
+	 * can just return.
+	 */
+	ret = -1;
+	if (full)
+		goto out;
 
-	trans = btrfs_start_transaction(info->tree_root, 1);
-	btrfs_commit_transaction(trans, info->tree_root);
+	mutex_lock(&root->fs_info->chunk_mutex);
+	list_for_each_entry(device, &fs_devices->alloc_list, dev_alloc_list) {
+		u64 min_free = btrfs_block_group_used(&block_group->item);
+		u64 dev_offset, max_avail;
 
-	while (1) {
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0)
-			goto out;
-next:
-		leaf = path->nodes[0];
-		nritems = btrfs_header_nritems(leaf);
-		if (path->slots[0] >= nritems) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto out;
-			if (ret == 1) {
-				ret = 0;
+		/*
+		 * check to make sure we can actually find a chunk with enough
+		 * space to fit our block group in.
+		 */
+		if (device->total_bytes > device->bytes_used + min_free) {
+			ret = find_free_dev_extent(NULL, device, min_free,
+						   &dev_offset, &max_avail);
+			if (!ret)
 				break;
-			}
-			leaf = path->nodes[0];
-			nritems = btrfs_header_nritems(leaf);
-		}
-
-		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
-
-		if (key.objectid >= block_group->key.objectid +
-		    block_group->key.offset)
-			break;
-
-		if (progress && need_resched()) {
-			btrfs_release_path(root, path);
-			cond_resched();
-			progress = 0;
-			continue;
-		}
-		progress = 1;
-
-		if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY ||
-		    key.objectid + key.offset <= cur_byte) {
-			path->slots[0]++;
-			goto next;
+			ret = -1;
 		}
-
-		total_found++;
-		cur_byte = key.objectid + key.offset;
-		btrfs_release_path(root, path);
-
-		__alloc_chunk_for_shrink(root, block_group, 0);
-		ret = relocate_one_extent(root, path, &key, block_group,
-					  reloc_inode, pass);
-		BUG_ON(ret < 0);
-		if (ret > 0)
-			skipped++;
-
-		key.objectid = cur_byte;
-		key.type = 0;
-		key.offset = 0;
 	}
-
-	btrfs_release_path(root, path);
-
-	if (pass == 0) {
-		btrfs_wait_ordered_range(reloc_inode, 0, (u64)-1);
-		invalidate_mapping_pages(reloc_inode->i_mapping, 0, -1);
-	}
-
-	if (total_found > 0) {
-		printk(KERN_INFO "btrfs found %llu extents in pass %d\n",
-		       (unsigned long long)total_found, pass);
-		pass++;
-		if (total_found == skipped && pass > 2) {
-			iput(reloc_inode);
-			reloc_inode = create_reloc_inode(info, block_group);
-			pass = 0;
-		}
-		goto again;
-	}
-
-	/* delete reloc_inode */
-	iput(reloc_inode);
-
-	/* unpin extents in this range */
-	trans = btrfs_start_transaction(info->tree_root, 1);
-	btrfs_commit_transaction(trans, info->tree_root);
-
-	spin_lock(&block_group->lock);
-	WARN_ON(block_group->pinned > 0);
-	WARN_ON(block_group->reserved > 0);
-	WARN_ON(btrfs_block_group_used(&block_group->item) > 0);
-	spin_unlock(&block_group->lock);
-	btrfs_put_block_group(block_group);
-	ret = 0;
+	mutex_unlock(&root->fs_info->chunk_mutex);
 out:
-	btrfs_free_path(path);
+	btrfs_put_block_group(block_group);
 	return ret;
 }
-#endif
 
 static int find_first_block_group(struct btrfs_root *root,
 		struct btrfs_path *path, struct btrfs_key *key)
@@ -7165,8 +7282,18 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 {
 	struct btrfs_block_group_cache *block_group;
 	struct btrfs_space_info *space_info;
+	struct btrfs_caching_control *caching_ctl;
 	struct rb_node *n;
 
+	down_write(&info->extent_commit_sem);
+	while (!list_empty(&info->caching_block_groups)) {
+		caching_ctl = list_entry(info->caching_block_groups.next,
+					 struct btrfs_caching_control, list);
+		list_del(&caching_ctl->list);
+		put_caching_control(caching_ctl);
+	}
+	up_write(&info->extent_commit_sem);
+
 	spin_lock(&info->block_group_cache_lock);
 	while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
 		block_group = rb_entry(n, struct btrfs_block_group_cache,
@@ -7180,8 +7307,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		up_write(&block_group->space_info->groups_sem);
 
 		if (block_group->cached == BTRFS_CACHE_STARTED)
-			wait_event(block_group->caching_q,
-				   block_group_cache_done(block_group));
+			wait_block_group_cache_done(block_group);
 
 		btrfs_remove_free_space_cache(block_group);
 
@@ -7251,7 +7377,6 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		spin_lock_init(&cache->lock);
 		spin_lock_init(&cache->tree_lock);
 		cache->fs_info = info;
-		init_waitqueue_head(&cache->caching_q);
 		INIT_LIST_HEAD(&cache->list);
 		INIT_LIST_HEAD(&cache->cluster_list);
 
@@ -7273,8 +7398,6 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		cache->flags = btrfs_block_group_flags(&cache->item);
 		cache->sectorsize = root->sectorsize;
 
-		remove_sb_from_cache(root, cache);
-
 		/*
 		 * check for two cases, either we are full, and therefore
 		 * don't need to bother with the caching work since we won't
@@ -7283,13 +7406,19 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		 * time, particularly in the full case.
 		 */
 		if (found_key.offset == btrfs_block_group_used(&cache->item)) {
+			exclude_super_stripes(root, cache);
+			cache->last_byte_to_unpin = (u64)-1;
 			cache->cached = BTRFS_CACHE_FINISHED;
+			free_excluded_extents(root, cache);
 		} else if (btrfs_block_group_used(&cache->item) == 0) {
+			exclude_super_stripes(root, cache);
+			cache->last_byte_to_unpin = (u64)-1;
 			cache->cached = BTRFS_CACHE_FINISHED;
 			add_new_free_space(cache, root->fs_info,
 					   found_key.objectid,
 					   found_key.objectid +
 					   found_key.offset);
+			free_excluded_extents(root, cache);
 		}
 
 		ret = update_space_info(info, cache->flags, found_key.offset,
@@ -7297,6 +7426,10 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 					&space_info);
 		BUG_ON(ret);
 		cache->space_info = space_info;
+		spin_lock(&cache->space_info->lock);
+		cache->space_info->bytes_super += cache->bytes_super;
+		spin_unlock(&cache->space_info->lock);
+
 		down_write(&space_info->groups_sem);
 		list_add_tail(&cache->list, &space_info->block_groups);
 		up_write(&space_info->groups_sem);
@@ -7346,7 +7479,6 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	atomic_set(&cache->count, 1);
 	spin_lock_init(&cache->lock);
 	spin_lock_init(&cache->tree_lock);
-	init_waitqueue_head(&cache->caching_q);
 	INIT_LIST_HEAD(&cache->list);
 	INIT_LIST_HEAD(&cache->cluster_list);
 
@@ -7355,15 +7487,23 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	cache->flags = type;
 	btrfs_set_block_group_flags(&cache->item, type);
 
+	cache->last_byte_to_unpin = (u64)-1;
 	cache->cached = BTRFS_CACHE_FINISHED;
-	remove_sb_from_cache(root, cache);
+	exclude_super_stripes(root, cache);
 
 	add_new_free_space(cache, root->fs_info, chunk_offset,
 			   chunk_offset + size);
 
+	free_excluded_extents(root, cache);
+
 	ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
 				&cache->space_info);
 	BUG_ON(ret);
+
+	spin_lock(&cache->space_info->lock);
+	cache->space_info->bytes_super += cache->bytes_super;
+	spin_unlock(&cache->space_info->lock);
+
 	down_write(&cache->space_info->groups_sem);
 	list_add_tail(&cache->list, &cache->space_info->block_groups);
 	up_write(&cache->space_info->groups_sem);
@@ -7429,8 +7569,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	up_write(&block_group->space_info->groups_sem);
 
 	if (block_group->cached == BTRFS_CACHE_STARTED)
-		wait_event(block_group->caching_q,
-			   block_group_cache_done(block_group));
+		wait_block_group_cache_done(block_group);
 
 	btrfs_remove_free_space_cache(block_group);
 
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 68260180f587..96577e8bf9fd 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -280,6 +280,14 @@ static struct extent_buffer *buffer_search(struct extent_io_tree *tree,
 	return NULL;
 }
 
+static void merge_cb(struct extent_io_tree *tree, struct extent_state *new,
+		     struct extent_state *other)
+{
+	if (tree->ops && tree->ops->merge_extent_hook)
+		tree->ops->merge_extent_hook(tree->mapping->host, new,
+					     other);
+}
+
 /*
  * utility function to look for merge candidates inside a given range.
  * Any extents with matching state are merged together into a single
@@ -303,6 +311,7 @@ static int merge_state(struct extent_io_tree *tree,
 		other = rb_entry(other_node, struct extent_state, rb_node);
 		if (other->end == state->start - 1 &&
 		    other->state == state->state) {
+			merge_cb(tree, state, other);
 			state->start = other->start;
 			other->tree = NULL;
 			rb_erase(&other->rb_node, &tree->state);
@@ -314,33 +323,37 @@ static int merge_state(struct extent_io_tree *tree,
 		other = rb_entry(other_node, struct extent_state, rb_node);
 		if (other->start == state->end + 1 &&
 		    other->state == state->state) {
+			merge_cb(tree, state, other);
 			other->start = state->start;
 			state->tree = NULL;
 			rb_erase(&state->rb_node, &tree->state);
 			free_extent_state(state);
+			state = NULL;
 		}
 	}
+
 	return 0;
 }
 
-static void set_state_cb(struct extent_io_tree *tree,
+static int set_state_cb(struct extent_io_tree *tree,
 			 struct extent_state *state,
 			 unsigned long bits)
 {
 	if (tree->ops && tree->ops->set_bit_hook) {
-		tree->ops->set_bit_hook(tree->mapping->host, state->start,
-					state->end, state->state, bits);
+		return tree->ops->set_bit_hook(tree->mapping->host,
+					       state->start, state->end,
+					       state->state, bits);
 	}
+
+	return 0;
 }
 
 static void clear_state_cb(struct extent_io_tree *tree,
 			   struct extent_state *state,
 			   unsigned long bits)
 {
-	if (tree->ops && tree->ops->clear_bit_hook) {
-		tree->ops->clear_bit_hook(tree->mapping->host, state->start,
-					  state->end, state->state, bits);
-	}
+	if (tree->ops && tree->ops->clear_bit_hook)
+		tree->ops->clear_bit_hook(tree->mapping->host, state, bits);
 }
 
 /*
@@ -358,6 +371,7 @@ static int insert_state(struct extent_io_tree *tree,
 			int bits)
 {
 	struct rb_node *node;
+	int ret;
 
 	if (end < start) {
 		printk(KERN_ERR "btrfs end < start %llu %llu\n",
@@ -365,12 +379,15 @@ static int insert_state(struct extent_io_tree *tree,
 		       (unsigned long long)start);
 		WARN_ON(1);
 	}
+	state->start = start;
+	state->end = end;
+	ret = set_state_cb(tree, state, bits);
+	if (ret)
+		return ret;
+
 	if (bits & EXTENT_DIRTY)
 		tree->dirty_bytes += end - start + 1;
-	set_state_cb(tree, state, bits);
 	state->state |= bits;
-	state->start = start;
-	state->end = end;
 	node = tree_insert(&tree->state, end, &state->rb_node);
 	if (node) {
 		struct extent_state *found;
@@ -387,6 +404,15 @@ static int insert_state(struct extent_io_tree *tree,
 	return 0;
 }
 
+static int split_cb(struct extent_io_tree *tree, struct extent_state *orig,
+		     u64 split)
+{
+	if (tree->ops && tree->ops->split_extent_hook)
+		return tree->ops->split_extent_hook(tree->mapping->host,
+						    orig, split);
+	return 0;
+}
+
 /*
  * split a given extent state struct in two, inserting the preallocated
  * struct 'prealloc' as the newly created second half.  'split' indicates an
@@ -405,6 +431,9 @@ static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
 		       struct extent_state *prealloc, u64 split)
 {
 	struct rb_node *node;
+
+	split_cb(tree, orig, split);
+
 	prealloc->start = orig->start;
 	prealloc->end = split - 1;
 	prealloc->state = orig->state;
@@ -431,7 +460,8 @@ static int clear_state_bit(struct extent_io_tree *tree,
 			    struct extent_state *state, int bits, int wake,
 			    int delete)
 {
-	int ret = state->state & bits;
+	int bits_to_clear = bits & ~EXTENT_DO_ACCOUNTING;
+	int ret = state->state & bits_to_clear;
 
 	if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
 		u64 range = state->end - state->start + 1;
@@ -439,7 +469,7 @@ static int clear_state_bit(struct extent_io_tree *tree,
 		tree->dirty_bytes -= range;
 	}
 	clear_state_cb(tree, state, bits);
-	state->state &= ~bits;
+	state->state &= ~bits_to_clear;
 	if (wake)
 		wake_up(&state->wq);
 	if (delete || state->state == 0) {
@@ -471,10 +501,14 @@ static int clear_state_bit(struct extent_io_tree *tree,
  * bits were already set, or zero if none of the bits were already set.
  */
 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		     int bits, int wake, int delete, gfp_t mask)
+		     int bits, int wake, int delete,
+		     struct extent_state **cached_state,
+		     gfp_t mask)
 {
 	struct extent_state *state;
+	struct extent_state *cached;
 	struct extent_state *prealloc = NULL;
+	struct rb_node *next_node;
 	struct rb_node *node;
 	u64 last_end;
 	int err;
@@ -488,6 +522,17 @@ again:
 	}
 
 	spin_lock(&tree->lock);
+	if (cached_state) {
+		cached = *cached_state;
+		*cached_state = NULL;
+		cached_state = NULL;
+		if (cached && cached->tree && cached->start == start) {
+			atomic_dec(&cached->refs);
+			state = cached;
+			goto hit_next;
+		}
+		free_extent_state(cached);
+	}
 	/*
 	 * this search will find the extents that end after
 	 * our range starts
@@ -496,6 +541,7 @@ again:
 	if (!node)
 		goto out;
 	state = rb_entry(node, struct extent_state, rb_node);
+hit_next:
 	if (state->start > end)
 		goto out;
 	WARN_ON(state->end < start);
@@ -526,13 +572,11 @@ again:
 		if (err)
 			goto out;
 		if (state->end <= end) {
-			set |= clear_state_bit(tree, state, bits,
-					wake, delete);
+			set |= clear_state_bit(tree, state, bits, wake,
+					       delete);
 			if (last_end == (u64)-1)
 				goto out;
 			start = last_end + 1;
-		} else {
-			start = state->start;
 		}
 		goto search_again;
 	}
@@ -547,19 +591,30 @@ again:
 			prealloc = alloc_extent_state(GFP_ATOMIC);
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
-
 		if (wake)
 			wake_up(&state->wq);
-		set |= clear_state_bit(tree, prealloc, bits,
-				       wake, delete);
+
+		set |= clear_state_bit(tree, prealloc, bits, wake, delete);
+
 		prealloc = NULL;
 		goto out;
 	}
 
+	if (state->end < end && prealloc && !need_resched())
+		next_node = rb_next(&state->rb_node);
+	else
+		next_node = NULL;
+
 	set |= clear_state_bit(tree, state, bits, wake, delete);
 	if (last_end == (u64)-1)
 		goto out;
 	start = last_end + 1;
+	if (start <= end && next_node) {
+		state = rb_entry(next_node, struct extent_state,
+				 rb_node);
+		if (state->start == start)
+			goto hit_next;
+	}
 	goto search_again;
 
 out:
@@ -641,40 +696,59 @@ out:
 	return 0;
 }
 
-static void set_state_bits(struct extent_io_tree *tree,
+static int set_state_bits(struct extent_io_tree *tree,
 			   struct extent_state *state,
 			   int bits)
 {
+	int ret;
+
+	ret = set_state_cb(tree, state, bits);
+	if (ret)
+		return ret;
+
 	if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
 		u64 range = state->end - state->start + 1;
 		tree->dirty_bytes += range;
 	}
-	set_state_cb(tree, state, bits);
 	state->state |= bits;
+
+	return 0;
+}
+
+static void cache_state(struct extent_state *state,
+			struct extent_state **cached_ptr)
+{
+	if (cached_ptr && !(*cached_ptr)) {
+		if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY)) {
+			*cached_ptr = state;
+			atomic_inc(&state->refs);
+		}
+	}
 }
 
 /*
- * set some bits on a range in the tree.  This may require allocations
- * or sleeping, so the gfp mask is used to indicate what is allowed.
+ * set some bits on a range in the tree.  This may require allocations or
+ * sleeping, so the gfp mask is used to indicate what is allowed.
  *
- * If 'exclusive' == 1, this will fail with -EEXIST if some part of the
- * range already has the desired bits set.  The start of the existing
- * range is returned in failed_start in this case.
+ * If any of the exclusive bits are set, this will fail with -EEXIST if some
+ * part of the range already has the desired bits set.  The start of the
+ * existing range is returned in failed_start in this case.
  *
- * [start, end] is inclusive
- * This takes the tree lock.
+ * [start, end] is inclusive This takes the tree lock.
  */
+
 static int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-			  int bits, int exclusive, u64 *failed_start,
+			  int bits, int exclusive_bits, u64 *failed_start,
+			  struct extent_state **cached_state,
 			  gfp_t mask)
 {
 	struct extent_state *state;
 	struct extent_state *prealloc = NULL;
 	struct rb_node *node;
 	int err = 0;
-	int set;
 	u64 last_start;
 	u64 last_end;
+
 again:
 	if (!prealloc && (mask & __GFP_WAIT)) {
 		prealloc = alloc_extent_state(mask);
@@ -683,6 +757,13 @@ again:
 	}
 
 	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->start == start && state->tree) {
+			node = &state->rb_node;
+			goto hit_next;
+		}
+	}
 	/*
 	 * this search will find all the extents that end after
 	 * our range starts.
@@ -694,8 +775,8 @@ again:
 		BUG_ON(err == -EEXIST);
 		goto out;
 	}
-
 	state = rb_entry(node, struct extent_state, rb_node);
+hit_next:
 	last_start = state->start;
 	last_end = state->end;
 
@@ -706,17 +787,32 @@ again:
 	 * Just lock what we found and keep going
 	 */
 	if (state->start == start && state->end <= end) {
-		set = state->state & bits;
-		if (set && exclusive) {
+		struct rb_node *next_node;
+		if (state->state & exclusive_bits) {
 			*failed_start = state->start;
 			err = -EEXIST;
 			goto out;
 		}
-		set_state_bits(tree, state, bits);
+
+		err = set_state_bits(tree, state, bits);
+		if (err)
+			goto out;
+
+		cache_state(state, cached_state);
 		merge_state(tree, state);
 		if (last_end == (u64)-1)
 			goto out;
+
 		start = last_end + 1;
+		if (start < end && prealloc && !need_resched()) {
+			next_node = rb_next(node);
+			if (next_node) {
+				state = rb_entry(next_node, struct extent_state,
+						 rb_node);
+				if (state->start == start)
+					goto hit_next;
+			}
+		}
 		goto search_again;
 	}
 
@@ -737,8 +833,7 @@ again:
 	 * desired bit on it.
 	 */
 	if (state->start < start) {
-		set = state->state & bits;
-		if (exclusive && set) {
+		if (state->state & exclusive_bits) {
 			*failed_start = start;
 			err = -EEXIST;
 			goto out;
@@ -749,13 +844,14 @@ again:
 		if (err)
 			goto out;
 		if (state->end <= end) {
-			set_state_bits(tree, state, bits);
+			err = set_state_bits(tree, state, bits);
+			if (err)
+				goto out;
+			cache_state(state, cached_state);
 			merge_state(tree, state);
 			if (last_end == (u64)-1)
 				goto out;
 			start = last_end + 1;
-		} else {
-			start = state->start;
 		}
 		goto search_again;
 	}
@@ -774,10 +870,13 @@ again:
 			this_end = last_start - 1;
 		err = insert_state(tree, prealloc, start, this_end,
 				   bits);
-		prealloc = NULL;
 		BUG_ON(err == -EEXIST);
-		if (err)
+		if (err) {
+			prealloc = NULL;
 			goto out;
+		}
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
 		start = this_end + 1;
 		goto search_again;
 	}
@@ -788,8 +887,7 @@ again:
 	 * on the first half
 	 */
 	if (state->start <= end && state->end > end) {
-		set = state->state & bits;
-		if (exclusive && set) {
+		if (state->state & exclusive_bits) {
 			*failed_start = start;
 			err = -EEXIST;
 			goto out;
@@ -797,7 +895,12 @@ again:
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
 
-		set_state_bits(tree, prealloc, bits);
+		err = set_state_bits(tree, prealloc, bits);
+		if (err) {
+			prealloc = NULL;
+			goto out;
+		}
+		cache_state(prealloc, cached_state);
 		merge_state(tree, prealloc);
 		prealloc = NULL;
 		goto out;
@@ -826,86 +929,65 @@ int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 		     gfp_t mask)
 {
 	return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
-			      mask);
-}
-
-int set_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
-		       gfp_t mask)
-{
-	return set_extent_bit(tree, start, end, EXTENT_ORDERED, 0, NULL, mask);
+			      NULL, mask);
 }
 
 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		    int bits, gfp_t mask)
 {
 	return set_extent_bit(tree, start, end, bits, 0, NULL,
-			      mask);
+			      NULL, mask);
 }
 
 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		      int bits, gfp_t mask)
 {
-	return clear_extent_bit(tree, start, end, bits, 0, 0, mask);
+	return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask);
 }
 
 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 		     gfp_t mask)
 {
 	return set_extent_bit(tree, start, end,
-			      EXTENT_DELALLOC | EXTENT_DIRTY,
-			      0, NULL, mask);
+			      EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE,
+			      0, NULL, NULL, mask);
 }
 
 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 		       gfp_t mask)
 {
 	return clear_extent_bit(tree, start, end,
-				EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, mask);
-}
-
-int clear_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
-			 gfp_t mask)
-{
-	return clear_extent_bit(tree, start, end, EXTENT_ORDERED, 1, 0, mask);
+				EXTENT_DIRTY | EXTENT_DELALLOC |
+				EXTENT_DO_ACCOUNTING, 0, 0,
+				NULL, mask);
 }
 
 int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 		     gfp_t mask)
 {
 	return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
-			      mask);
+			      NULL, mask);
 }
 
 static int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 		       gfp_t mask)
 {
-	return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0, mask);
+	return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0,
+				NULL, mask);
 }
 
 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 			gfp_t mask)
 {
 	return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL,
-			      mask);
+			      NULL, mask);
 }
 
 static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 				 u64 end, gfp_t mask)
 {
-	return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, mask);
-}
-
-static int set_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end,
-			 gfp_t mask)
-{
-	return set_extent_bit(tree, start, end, EXTENT_WRITEBACK,
-			      0, NULL, mask);
-}
-
-static int clear_extent_writeback(struct extent_io_tree *tree, u64 start,
-				  u64 end, gfp_t mask)
-{
-	return clear_extent_bit(tree, start, end, EXTENT_WRITEBACK, 1, 0, mask);
+	return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
+				NULL, mask);
 }
 
 int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
@@ -917,13 +999,15 @@ int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
  * either insert or lock state struct between start and end use mask to tell
  * us if waiting is desired.
  */
-int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+		     int bits, struct extent_state **cached_state, gfp_t mask)
 {
 	int err;
 	u64 failed_start;
 	while (1) {
-		err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1,
-				     &failed_start, mask);
+		err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
+				     EXTENT_LOCKED, &failed_start,
+				     cached_state, mask);
 		if (err == -EEXIST && (mask & __GFP_WAIT)) {
 			wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
 			start = failed_start;
@@ -935,27 +1019,40 @@ int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
 	return err;
 }
 
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+{
+	return lock_extent_bits(tree, start, end, 0, NULL, mask);
+}
+
 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 		    gfp_t mask)
 {
 	int err;
 	u64 failed_start;
 
-	err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1,
-			     &failed_start, mask);
+	err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
+			     &failed_start, NULL, mask);
 	if (err == -EEXIST) {
 		if (failed_start > start)
 			clear_extent_bit(tree, start, failed_start - 1,
-					 EXTENT_LOCKED, 1, 0, mask);
+					 EXTENT_LOCKED, 1, 0, NULL, mask);
 		return 0;
 	}
 	return 1;
 }
 
+int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
+			 struct extent_state **cached, gfp_t mask)
+{
+	return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
+				mask);
+}
+
 int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 		  gfp_t mask)
 {
-	return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, mask);
+	return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL,
+				mask);
 }
 
 /*
@@ -974,7 +1071,6 @@ int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end)
 		page_cache_release(page);
 		index++;
 	}
-	set_extent_dirty(tree, start, end, GFP_NOFS);
 	return 0;
 }
 
@@ -994,7 +1090,6 @@ static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
 		page_cache_release(page);
 		index++;
 	}
-	set_extent_writeback(tree, start, end, GFP_NOFS);
 	return 0;
 }
 
@@ -1232,6 +1327,7 @@ static noinline u64 find_lock_delalloc_range(struct inode *inode,
 	u64 delalloc_start;
 	u64 delalloc_end;
 	u64 found;
+	struct extent_state *cached_state = NULL;
 	int ret;
 	int loops = 0;
 
@@ -1269,6 +1365,7 @@ again:
 		/* some of the pages are gone, lets avoid looping by
 		 * shortening the size of the delalloc range we're searching
 		 */
+		free_extent_state(cached_state);
 		if (!loops) {
 			unsigned long offset = (*start) & (PAGE_CACHE_SIZE - 1);
 			max_bytes = PAGE_CACHE_SIZE - offset;
@@ -1282,18 +1379,21 @@ again:
 	BUG_ON(ret);
 
 	/* step three, lock the state bits for the whole range */
-	lock_extent(tree, delalloc_start, delalloc_end, GFP_NOFS);
+	lock_extent_bits(tree, delalloc_start, delalloc_end,
+			 0, &cached_state, GFP_NOFS);
 
 	/* then test to make sure it is all still delalloc */
 	ret = test_range_bit(tree, delalloc_start, delalloc_end,
-			     EXTENT_DELALLOC, 1);
+			     EXTENT_DELALLOC, 1, cached_state);
 	if (!ret) {
-		unlock_extent(tree, delalloc_start, delalloc_end, GFP_NOFS);
+		unlock_extent_cached(tree, delalloc_start, delalloc_end,
+				     &cached_state, GFP_NOFS);
 		__unlock_for_delalloc(inode, locked_page,
 			      delalloc_start, delalloc_end);
 		cond_resched();
 		goto again;
 	}
+	free_extent_state(cached_state);
 	*start = delalloc_start;
 	*end = delalloc_end;
 out_failed:
@@ -1303,11 +1403,7 @@ out_failed:
 int extent_clear_unlock_delalloc(struct inode *inode,
 				struct extent_io_tree *tree,
 				u64 start, u64 end, struct page *locked_page,
-				int unlock_pages,
-				int clear_unlock,
-				int clear_delalloc, int clear_dirty,
-				int set_writeback,
-				int end_writeback)
+				unsigned long op)
 {
 	int ret;
 	struct page *pages[16];
@@ -1317,16 +1413,21 @@ int extent_clear_unlock_delalloc(struct inode *inode,
 	int i;
 	int clear_bits = 0;
 
-	if (clear_unlock)
+	if (op & EXTENT_CLEAR_UNLOCK)
 		clear_bits |= EXTENT_LOCKED;
-	if (clear_dirty)
+	if (op & EXTENT_CLEAR_DIRTY)
 		clear_bits |= EXTENT_DIRTY;
 
-	if (clear_delalloc)
+	if (op & EXTENT_CLEAR_DELALLOC)
 		clear_bits |= EXTENT_DELALLOC;
 
-	clear_extent_bit(tree, start, end, clear_bits, 1, 0, GFP_NOFS);
-	if (!(unlock_pages || clear_dirty || set_writeback || end_writeback))
+	if (op & EXTENT_CLEAR_ACCOUNTING)
+		clear_bits |= EXTENT_DO_ACCOUNTING;
+
+	clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS);
+	if (!(op & (EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
+		    EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK |
+		    EXTENT_SET_PRIVATE2)))
 		return 0;
 
 	while (nr_pages > 0) {
@@ -1334,17 +1435,21 @@ int extent_clear_unlock_delalloc(struct inode *inode,
 				     min_t(unsigned long,
 				     nr_pages, ARRAY_SIZE(pages)), pages);
 		for (i = 0; i < ret; i++) {
+
+			if (op & EXTENT_SET_PRIVATE2)
+				SetPagePrivate2(pages[i]);
+
 			if (pages[i] == locked_page) {
 				page_cache_release(pages[i]);
 				continue;
 			}
-			if (clear_dirty)
+			if (op & EXTENT_CLEAR_DIRTY)
 				clear_page_dirty_for_io(pages[i]);
-			if (set_writeback)
+			if (op & EXTENT_SET_WRITEBACK)
 				set_page_writeback(pages[i]);
-			if (end_writeback)
+			if (op & EXTENT_END_WRITEBACK)
 				end_page_writeback(pages[i]);
-			if (unlock_pages)
+			if (op & EXTENT_CLEAR_UNLOCK_PAGE)
 				unlock_page(pages[i]);
 			page_cache_release(pages[i]);
 		}
@@ -1476,14 +1581,17 @@ out:
  * range is found set.
  */
 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   int bits, int filled)
+		   int bits, int filled, struct extent_state *cached)
 {
 	struct extent_state *state = NULL;
 	struct rb_node *node;
 	int bitset = 0;
 
 	spin_lock(&tree->lock);
-	node = tree_search(tree, start);
+	if (cached && cached->tree && cached->start == start)
+		node = &cached->rb_node;
+	else
+		node = tree_search(tree, start);
 	while (node && start <= end) {
 		state = rb_entry(node, struct extent_state, rb_node);
 
@@ -1503,6 +1611,10 @@ int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
 			bitset = 0;
 			break;
 		}
+
+		if (state->end == (u64)-1)
+			break;
+
 		start = state->end + 1;
 		if (start > end)
 			break;
@@ -1526,7 +1638,7 @@ static int check_page_uptodate(struct extent_io_tree *tree,
 {
 	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
 	u64 end = start + PAGE_CACHE_SIZE - 1;
-	if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1))
+	if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
 		SetPageUptodate(page);
 	return 0;
 }
@@ -1540,7 +1652,7 @@ static int check_page_locked(struct extent_io_tree *tree,
 {
 	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
 	u64 end = start + PAGE_CACHE_SIZE - 1;
-	if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0))
+	if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
 		unlock_page(page);
 	return 0;
 }
@@ -1552,10 +1664,7 @@ static int check_page_locked(struct extent_io_tree *tree,
 static int check_page_writeback(struct extent_io_tree *tree,
 			     struct page *page)
 {
-	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
-	u64 end = start + PAGE_CACHE_SIZE - 1;
-	if (!test_range_bit(tree, start, end, EXTENT_WRITEBACK, 0))
-		end_page_writeback(page);
+	end_page_writeback(page);
 	return 0;
 }
 
@@ -1613,13 +1722,11 @@ static void end_bio_extent_writepage(struct bio *bio, int err)
 		}
 
 		if (!uptodate) {
-			clear_extent_uptodate(tree, start, end, GFP_ATOMIC);
+			clear_extent_uptodate(tree, start, end, GFP_NOFS);
 			ClearPageUptodate(page);
 			SetPageError(page);
 		}
 
-		clear_extent_writeback(tree, start, end, GFP_ATOMIC);
-
 		if (whole_page)
 			end_page_writeback(page);
 		else
@@ -1983,7 +2090,8 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			continue;
 		}
 		/* the get_extent function already copied into the page */
-		if (test_range_bit(tree, cur, cur_end, EXTENT_UPTODATE, 1)) {
+		if (test_range_bit(tree, cur, cur_end,
+				   EXTENT_UPTODATE, 1, NULL)) {
 			check_page_uptodate(tree, page);
 			unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
 			cur = cur + iosize;
@@ -2078,6 +2186,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	u64 iosize;
 	u64 unlock_start;
 	sector_t sector;
+	struct extent_state *cached_state = NULL;
 	struct extent_map *em;
 	struct block_device *bdev;
 	int ret;
@@ -2124,6 +2233,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	delalloc_end = 0;
 	page_started = 0;
 	if (!epd->extent_locked) {
+		u64 delalloc_to_write = 0;
 		/*
 		 * make sure the wbc mapping index is at least updated
 		 * to this page.
@@ -2143,8 +2253,24 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 			tree->ops->fill_delalloc(inode, page, delalloc_start,
 						 delalloc_end, &page_started,
 						 &nr_written);
+			/*
+			 * delalloc_end is already one less than the total
+			 * length, so we don't subtract one from
+			 * PAGE_CACHE_SIZE
+			 */
+			delalloc_to_write += (delalloc_end - delalloc_start +
+					      PAGE_CACHE_SIZE) >>
+					      PAGE_CACHE_SHIFT;
 			delalloc_start = delalloc_end + 1;
 		}
+		if (wbc->nr_to_write < delalloc_to_write) {
+			int thresh = 8192;
+
+			if (delalloc_to_write < thresh * 2)
+				thresh = delalloc_to_write;
+			wbc->nr_to_write = min_t(u64, delalloc_to_write,
+						 thresh);
+		}
 
 		/* did the fill delalloc function already unlock and start
 		 * the IO?
@@ -2160,15 +2286,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 			goto done_unlocked;
 		}
 	}
-	lock_extent(tree, start, page_end, GFP_NOFS);
-
-	unlock_start = start;
-
 	if (tree->ops && tree->ops->writepage_start_hook) {
 		ret = tree->ops->writepage_start_hook(page, start,
 						      page_end);
 		if (ret == -EAGAIN) {
-			unlock_extent(tree, start, page_end, GFP_NOFS);
 			redirty_page_for_writepage(wbc, page);
 			update_nr_written(page, wbc, nr_written);
 			unlock_page(page);
@@ -2184,12 +2305,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	update_nr_written(page, wbc, nr_written + 1);
 
 	end = page_end;
-	if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0))
-		printk(KERN_ERR "btrfs delalloc bits after lock_extent\n");
-
 	if (last_byte <= start) {
-		clear_extent_dirty(tree, start, page_end, GFP_NOFS);
-		unlock_extent(tree, start, page_end, GFP_NOFS);
 		if (tree->ops && tree->ops->writepage_end_io_hook)
 			tree->ops->writepage_end_io_hook(page, start,
 							 page_end, NULL, 1);
@@ -2197,13 +2313,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		goto done;
 	}
 
-	set_extent_uptodate(tree, start, page_end, GFP_NOFS);
 	blocksize = inode->i_sb->s_blocksize;
 
 	while (cur <= end) {
 		if (cur >= last_byte) {
-			clear_extent_dirty(tree, cur, page_end, GFP_NOFS);
-			unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
 			if (tree->ops && tree->ops->writepage_end_io_hook)
 				tree->ops->writepage_end_io_hook(page, cur,
 							 page_end, NULL, 1);
@@ -2235,12 +2348,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		 */
 		if (compressed || block_start == EXTENT_MAP_HOLE ||
 		    block_start == EXTENT_MAP_INLINE) {
-			clear_extent_dirty(tree, cur,
-					   cur + iosize - 1, GFP_NOFS);
-
-			unlock_extent(tree, unlock_start, cur + iosize - 1,
-				      GFP_NOFS);
-
 			/*
 			 * end_io notification does not happen here for
 			 * compressed extents
@@ -2265,13 +2372,12 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		}
 		/* leave this out until we have a page_mkwrite call */
 		if (0 && !test_range_bit(tree, cur, cur + iosize - 1,
-				   EXTENT_DIRTY, 0)) {
+				   EXTENT_DIRTY, 0, NULL)) {
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
 		}
 
-		clear_extent_dirty(tree, cur, cur + iosize - 1, GFP_NOFS);
 		if (tree->ops && tree->ops->writepage_io_hook) {
 			ret = tree->ops->writepage_io_hook(page, cur,
 						cur + iosize - 1);
@@ -2309,12 +2415,12 @@ done:
 		set_page_writeback(page);
 		end_page_writeback(page);
 	}
-	if (unlock_start <= page_end)
-		unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
 	unlock_page(page);
 
 done_unlocked:
 
+	/* drop our reference on any cached states */
+	free_extent_state(cached_state);
 	return 0;
 }
 
@@ -2339,9 +2445,9 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
 			     writepage_t writepage, void *data,
 			     void (*flush_fn)(void *))
 {
-	struct backing_dev_info *bdi = mapping->backing_dev_info;
 	int ret = 0;
 	int done = 0;
+	int nr_to_write_done = 0;
 	struct pagevec pvec;
 	int nr_pages;
 	pgoff_t index;
@@ -2361,7 +2467,7 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
 		scanned = 1;
 	}
 retry:
-	while (!done && (index <= end) &&
+	while (!done && !nr_to_write_done && (index <= end) &&
 	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
 			      PAGECACHE_TAG_DIRTY, min(end - index,
 				  (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
@@ -2412,12 +2518,15 @@ retry:
 				unlock_page(page);
 				ret = 0;
 			}
-			if (ret || wbc->nr_to_write <= 0)
-				done = 1;
-			if (wbc->nonblocking && bdi_write_congested(bdi)) {
-				wbc->encountered_congestion = 1;
+			if (ret)
 				done = 1;
-			}
+
+			/*
+			 * the filesystem may choose to bump up nr_to_write.
+			 * We have to make sure to honor the new nr_to_write
+			 * at any time
+			 */
+			nr_to_write_done = wbc->nr_to_write <= 0;
 		}
 		pagevec_release(&pvec);
 		cond_resched();
@@ -2604,10 +2713,11 @@ int extent_invalidatepage(struct extent_io_tree *tree,
 		return 0;
 
 	lock_extent(tree, start, end, GFP_NOFS);
-	wait_on_extent_writeback(tree, start, end);
+	wait_on_page_writeback(page);
 	clear_extent_bit(tree, start, end,
-			 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC,
-			 1, 1, GFP_NOFS);
+			 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
+			 EXTENT_DO_ACCOUNTING,
+			 1, 1, NULL, GFP_NOFS);
 	return 0;
 }
 
@@ -2687,7 +2797,7 @@ int extent_prepare_write(struct extent_io_tree *tree,
 		    !isnew && !PageUptodate(page) &&
 		    (block_off_end > to || block_off_start < from) &&
 		    !test_range_bit(tree, block_start, cur_end,
-				    EXTENT_UPTODATE, 1)) {
+				    EXTENT_UPTODATE, 1, NULL)) {
 			u64 sector;
 			u64 extent_offset = block_start - em->start;
 			size_t iosize;
@@ -2701,7 +2811,7 @@ int extent_prepare_write(struct extent_io_tree *tree,
 			 */
 			set_extent_bit(tree, block_start,
 				       block_start + iosize - 1,
-				       EXTENT_LOCKED, 0, NULL, GFP_NOFS);
+				       EXTENT_LOCKED, 0, NULL, NULL, GFP_NOFS);
 			ret = submit_extent_page(READ, tree, page,
 					 sector, iosize, page_offset, em->bdev,
 					 NULL, 1,
@@ -2742,13 +2852,18 @@ int try_release_extent_state(struct extent_map_tree *map,
 	int ret = 1;
 
 	if (test_range_bit(tree, start, end,
-			   EXTENT_IOBITS | EXTENT_ORDERED, 0))
+			   EXTENT_IOBITS, 0, NULL))
 		ret = 0;
 	else {
 		if ((mask & GFP_NOFS) == GFP_NOFS)
 			mask = GFP_NOFS;
-		clear_extent_bit(tree, start, end, EXTENT_UPTODATE,
-				 1, 1, mask);
+		/*
+		 * at this point we can safely clear everything except the
+		 * locked bit and the nodatasum bit
+		 */
+		clear_extent_bit(tree, start, end,
+				 ~(EXTENT_LOCKED | EXTENT_NODATASUM),
+				 0, 0, NULL, mask);
 	}
 	return ret;
 }
@@ -2771,29 +2886,28 @@ int try_release_extent_mapping(struct extent_map_tree *map,
 		u64 len;
 		while (start <= end) {
 			len = end - start + 1;
-			spin_lock(&map->lock);
+			write_lock(&map->lock);
 			em = lookup_extent_mapping(map, start, len);
 			if (!em || IS_ERR(em)) {
-				spin_unlock(&map->lock);
+				write_unlock(&map->lock);
 				break;
 			}
 			if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
 			    em->start != start) {
-				spin_unlock(&map->lock);
+				write_unlock(&map->lock);
 				free_extent_map(em);
 				break;
 			}
 			if (!test_range_bit(tree, em->start,
 					    extent_map_end(em) - 1,
-					    EXTENT_LOCKED | EXTENT_WRITEBACK |
-					    EXTENT_ORDERED,
-					    0)) {
+					    EXTENT_LOCKED | EXTENT_WRITEBACK,
+					    0, NULL)) {
 				remove_extent_mapping(map, em);
 				/* once for the rb tree */
 				free_extent_map(em);
 			}
 			start = extent_map_end(em);
-			spin_unlock(&map->lock);
+			write_unlock(&map->lock);
 
 			/* once for us */
 			free_extent_map(em);
@@ -3203,7 +3317,7 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	int uptodate;
 	unsigned long index;
 
-	ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1);
+	ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL);
 	if (ret)
 		return 1;
 	while (start <= end) {
@@ -3233,7 +3347,7 @@ int extent_buffer_uptodate(struct extent_io_tree *tree,
 		return 1;
 
 	ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-			   EXTENT_UPTODATE, 1);
+			   EXTENT_UPTODATE, 1, NULL);
 	if (ret)
 		return ret;
 
@@ -3269,7 +3383,7 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 		return 0;
 
 	if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-			   EXTENT_UPTODATE, 1)) {
+			   EXTENT_UPTODATE, 1, NULL)) {
 		return 0;
 	}
 
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 5bc20abf3f3d..36de250a7b2b 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -13,10 +13,9 @@
 #define EXTENT_DEFRAG (1 << 6)
 #define EXTENT_DEFRAG_DONE (1 << 7)
 #define EXTENT_BUFFER_FILLED (1 << 8)
-#define EXTENT_ORDERED (1 << 9)
-#define EXTENT_ORDERED_METADATA (1 << 10)
-#define EXTENT_BOUNDARY (1 << 11)
-#define EXTENT_NODATASUM (1 << 12)
+#define EXTENT_BOUNDARY (1 << 9)
+#define EXTENT_NODATASUM (1 << 10)
+#define EXTENT_DO_ACCOUNTING (1 << 11)
 #define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
 
 /* flags for bio submission */
@@ -27,6 +26,16 @@
 #define EXTENT_BUFFER_BLOCKING 1
 #define EXTENT_BUFFER_DIRTY 2
 
+/* these are flags for extent_clear_unlock_delalloc */
+#define EXTENT_CLEAR_UNLOCK_PAGE 0x1
+#define EXTENT_CLEAR_UNLOCK	 0x2
+#define EXTENT_CLEAR_DELALLOC	 0x4
+#define EXTENT_CLEAR_DIRTY	 0x8
+#define EXTENT_SET_WRITEBACK	 0x10
+#define EXTENT_END_WRITEBACK	 0x20
+#define EXTENT_SET_PRIVATE2	 0x40
+#define EXTENT_CLEAR_ACCOUNTING  0x80
+
 /*
  * page->private values.  Every page that is controlled by the extent
  * map has page->private set to one.
@@ -62,8 +71,13 @@ struct extent_io_ops {
 				      struct extent_state *state, int uptodate);
 	int (*set_bit_hook)(struct inode *inode, u64 start, u64 end,
 			    unsigned long old, unsigned long bits);
-	int (*clear_bit_hook)(struct inode *inode, u64 start, u64 end,
-			    unsigned long old, unsigned long bits);
+	int (*clear_bit_hook)(struct inode *inode, struct extent_state *state,
+			      unsigned long bits);
+	int (*merge_extent_hook)(struct inode *inode,
+				 struct extent_state *new,
+				 struct extent_state *other);
+	int (*split_extent_hook)(struct inode *inode,
+				 struct extent_state *orig, u64 split);
 	int (*write_cache_pages_lock_hook)(struct page *page);
 };
 
@@ -81,10 +95,14 @@ struct extent_state {
 	u64 start;
 	u64 end; /* inclusive */
 	struct rb_node rb_node;
+
+	/* ADD NEW ELEMENTS AFTER THIS */
 	struct extent_io_tree *tree;
 	wait_queue_head_t wq;
 	atomic_t refs;
 	unsigned long state;
+	u64 split_start;
+	u64 split_end;
 
 	/* for use by the FS */
 	u64 private;
@@ -142,6 +160,8 @@ int try_release_extent_state(struct extent_map_tree *map,
 			     struct extent_io_tree *tree, struct page *page,
 			     gfp_t mask);
 int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
+int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+		     int bits, struct extent_state **cached, gfp_t mask);
 int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 		    gfp_t mask);
@@ -155,11 +175,12 @@ u64 count_range_bits(struct extent_io_tree *tree,
 		     u64 max_bytes, unsigned long bits);
 
 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   int bits, int filled);
+		   int bits, int filled, struct extent_state *cached_state);
 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		      int bits, gfp_t mask);
 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		     int bits, int wake, int delete, gfp_t mask);
+		     int bits, int wake, int delete, struct extent_state **cached,
+		     gfp_t mask);
 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		    int bits, gfp_t mask);
 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
@@ -278,9 +299,5 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 int extent_clear_unlock_delalloc(struct inode *inode,
 				struct extent_io_tree *tree,
 				u64 start, u64 end, struct page *locked_page,
-				int unlock_page,
-				int clear_unlock,
-				int clear_delalloc, int clear_dirty,
-				int set_writeback,
-				int end_writeback);
+				unsigned long op);
 #endif
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index 30c9365861e6..2c726b7b9faa 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -36,7 +36,7 @@ void extent_map_exit(void)
 void extent_map_tree_init(struct extent_map_tree *tree, gfp_t mask)
 {
 	tree->map.rb_node = NULL;
-	spin_lock_init(&tree->lock);
+	rwlock_init(&tree->lock);
 }
 
 /**
@@ -198,6 +198,56 @@ static int mergable_maps(struct extent_map *prev, struct extent_map *next)
 	return 0;
 }
 
+int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
+{
+	int ret = 0;
+	struct extent_map *merge = NULL;
+	struct rb_node *rb;
+	struct extent_map *em;
+
+	write_lock(&tree->lock);
+	em = lookup_extent_mapping(tree, start, len);
+
+	WARN_ON(em->start != start || !em);
+
+	if (!em)
+		goto out;
+
+	clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+
+	if (em->start != 0) {
+		rb = rb_prev(&em->rb_node);
+		if (rb)
+			merge = rb_entry(rb, struct extent_map, rb_node);
+		if (rb && mergable_maps(merge, em)) {
+			em->start = merge->start;
+			em->len += merge->len;
+			em->block_len += merge->block_len;
+			em->block_start = merge->block_start;
+			merge->in_tree = 0;
+			rb_erase(&merge->rb_node, &tree->map);
+			free_extent_map(merge);
+		}
+	}
+
+	rb = rb_next(&em->rb_node);
+	if (rb)
+		merge = rb_entry(rb, struct extent_map, rb_node);
+	if (rb && mergable_maps(em, merge)) {
+		em->len += merge->len;
+		em->block_len += merge->len;
+		rb_erase(&merge->rb_node, &tree->map);
+		merge->in_tree = 0;
+		free_extent_map(merge);
+	}
+
+	free_extent_map(em);
+out:
+	write_unlock(&tree->lock);
+	return ret;
+
+}
+
 /**
  * add_extent_mapping - add new extent map to the extent tree
  * @tree:	tree to insert new map in
@@ -222,7 +272,6 @@ int add_extent_mapping(struct extent_map_tree *tree,
 		ret = -EEXIST;
 		goto out;
 	}
-	assert_spin_locked(&tree->lock);
 	rb = tree_insert(&tree->map, em->start, &em->rb_node);
 	if (rb) {
 		ret = -EEXIST;
@@ -285,7 +334,6 @@ struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree,
 	struct rb_node *next = NULL;
 	u64 end = range_end(start, len);
 
-	assert_spin_locked(&tree->lock);
 	rb_node = __tree_search(&tree->map, start, &prev, &next);
 	if (!rb_node && prev) {
 		em = rb_entry(prev, struct extent_map, rb_node);
@@ -319,6 +367,54 @@ out:
 }
 
 /**
+ * search_extent_mapping - find a nearby extent map
+ * @tree:	tree to lookup in
+ * @start:	byte offset to start the search
+ * @len:	length of the lookup range
+ *
+ * Find and return the first extent_map struct in @tree that intersects the
+ * [start, len] range.
+ *
+ * If one can't be found, any nearby extent may be returned
+ */
+struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
+					 u64 start, u64 len)
+{
+	struct extent_map *em;
+	struct rb_node *rb_node;
+	struct rb_node *prev = NULL;
+	struct rb_node *next = NULL;
+
+	rb_node = __tree_search(&tree->map, start, &prev, &next);
+	if (!rb_node && prev) {
+		em = rb_entry(prev, struct extent_map, rb_node);
+		goto found;
+	}
+	if (!rb_node && next) {
+		em = rb_entry(next, struct extent_map, rb_node);
+		goto found;
+	}
+	if (!rb_node) {
+		em = NULL;
+		goto out;
+	}
+	if (IS_ERR(rb_node)) {
+		em = ERR_PTR(PTR_ERR(rb_node));
+		goto out;
+	}
+	em = rb_entry(rb_node, struct extent_map, rb_node);
+	goto found;
+
+	em = NULL;
+	goto out;
+
+found:
+	atomic_inc(&em->refs);
+out:
+	return em;
+}
+
+/**
  * remove_extent_mapping - removes an extent_map from the extent tree
  * @tree:	extent tree to remove from
  * @em:		extent map beeing removed
@@ -331,7 +427,6 @@ int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em)
 	int ret = 0;
 
 	WARN_ON(test_bit(EXTENT_FLAG_PINNED, &em->flags));
-	assert_spin_locked(&tree->lock);
 	rb_erase(&em->rb_node, &tree->map);
 	em->in_tree = 0;
 	return ret;
diff --git a/fs/btrfs/extent_map.h b/fs/btrfs/extent_map.h
index fb6eeef06bb0..ab6d74b6e647 100644
--- a/fs/btrfs/extent_map.h
+++ b/fs/btrfs/extent_map.h
@@ -31,7 +31,7 @@ struct extent_map {
 
 struct extent_map_tree {
 	struct rb_root map;
-	spinlock_t lock;
+	rwlock_t lock;
 };
 
 static inline u64 extent_map_end(struct extent_map *em)
@@ -59,4 +59,7 @@ struct extent_map *alloc_extent_map(gfp_t mask);
 void free_extent_map(struct extent_map *em);
 int __init extent_map_init(void);
 void extent_map_exit(void);
+int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len);
+struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
+					 u64 start, u64 len);
 #endif
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 4b833972273a..06550affbd27 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -112,8 +112,6 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans,
 	int err = 0;
 	int i;
 	struct inode *inode = fdentry(file)->d_inode;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-	u64 hint_byte;
 	u64 num_bytes;
 	u64 start_pos;
 	u64 end_of_last_block;
@@ -125,23 +123,10 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans,
 		    root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
 
 	end_of_last_block = start_pos + num_bytes - 1;
+	err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block);
+	if (err)
+		return err;
 
-	lock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS);
-	trans = btrfs_join_transaction(root, 1);
-	if (!trans) {
-		err = -ENOMEM;
-		goto out_unlock;
-	}
-	btrfs_set_trans_block_group(trans, inode);
-	hint_byte = 0;
-
-	set_extent_uptodate(io_tree, start_pos, end_of_last_block, GFP_NOFS);
-
-	/* check for reserved extents on each page, we don't want
-	 * to reset the delalloc bit on things that already have
-	 * extents reserved.
-	 */
-	btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block);
 	for (i = 0; i < num_pages; i++) {
 		struct page *p = pages[i];
 		SetPageUptodate(p);
@@ -155,9 +140,6 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans,
 		 * at this time.
 		 */
 	}
-	err = btrfs_end_transaction(trans, root);
-out_unlock:
-	unlock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS);
 	return err;
 }
 
@@ -189,18 +171,18 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		if (!split2)
 			split2 = alloc_extent_map(GFP_NOFS);
 
-		spin_lock(&em_tree->lock);
+		write_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, len);
 		if (!em) {
-			spin_unlock(&em_tree->lock);
+			write_unlock(&em_tree->lock);
 			break;
 		}
 		flags = em->flags;
 		if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
-			spin_unlock(&em_tree->lock);
 			if (em->start <= start &&
 			    (!testend || em->start + em->len >= start + len)) {
 				free_extent_map(em);
+				write_unlock(&em_tree->lock);
 				break;
 			}
 			if (start < em->start) {
@@ -210,6 +192,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 				start = em->start + em->len;
 			}
 			free_extent_map(em);
+			write_unlock(&em_tree->lock);
 			continue;
 		}
 		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
@@ -260,7 +243,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			free_extent_map(split);
 			split = NULL;
 		}
-		spin_unlock(&em_tree->lock);
+		write_unlock(&em_tree->lock);
 
 		/* once for us */
 		free_extent_map(em);
@@ -289,7 +272,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 noinline int btrfs_drop_extents(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root, struct inode *inode,
 		       u64 start, u64 end, u64 locked_end,
-		       u64 inline_limit, u64 *hint_byte)
+		       u64 inline_limit, u64 *hint_byte, int drop_cache)
 {
 	u64 extent_end = 0;
 	u64 search_start = start;
@@ -314,7 +297,8 @@ noinline int btrfs_drop_extents(struct btrfs_trans_handle *trans,
 	int ret;
 
 	inline_limit = 0;
-	btrfs_drop_extent_cache(inode, start, end - 1, 0);
+	if (drop_cache)
+		btrfs_drop_extent_cache(inode, start, end - 1, 0);
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -894,7 +878,8 @@ again:
 			btrfs_put_ordered_extent(ordered);
 
 		clear_extent_bits(&BTRFS_I(inode)->io_tree, start_pos,
-				  last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC,
+				  last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
+				  EXTENT_DO_ACCOUNTING,
 				  GFP_NOFS);
 		unlock_extent(&BTRFS_I(inode)->io_tree,
 			      start_pos, last_pos - 1, GFP_NOFS);
@@ -936,21 +921,35 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
 	start_pos = pos;
 
 	vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);
+
+	/* do the reserve before the mutex lock in case we have to do some
+	 * flushing.  We wouldn't deadlock, but this is more polite.
+	 */
+	err = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+	if (err)
+		goto out_nolock;
+
+	mutex_lock(&inode->i_mutex);
+
 	current->backing_dev_info = inode->i_mapping->backing_dev_info;
 	err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode));
 	if (err)
-		goto out_nolock;
+		goto out;
+
 	if (count == 0)
-		goto out_nolock;
+		goto out;
 
 	err = file_remove_suid(file);
 	if (err)
-		goto out_nolock;
+		goto out;
+
 	file_update_time(file);
 
 	pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL);
 
-	mutex_lock(&inode->i_mutex);
+	/* generic_write_checks can change our pos */
+	start_pos = pos;
+
 	BTRFS_I(inode)->sequence++;
 	first_index = pos >> PAGE_CACHE_SHIFT;
 	last_index = (pos + count) >> PAGE_CACHE_SHIFT;
@@ -1024,9 +1023,8 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
 		}
 
 		if (will_write) {
-			btrfs_fdatawrite_range(inode->i_mapping, pos,
-					       pos + write_bytes - 1,
-					       WB_SYNC_ALL);
+			filemap_fdatawrite_range(inode->i_mapping, pos,
+						 pos + write_bytes - 1);
 		} else {
 			balance_dirty_pages_ratelimited_nr(inode->i_mapping,
 							   num_pages);
@@ -1047,6 +1045,7 @@ out:
 	mutex_unlock(&inode->i_mutex);
 	if (ret)
 		err = ret;
+	btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
 
 out_nolock:
 	kfree(pages);
@@ -1087,8 +1086,10 @@ out_nolock:
 					btrfs_end_transaction(trans, root);
 				else
 					btrfs_commit_transaction(trans, root);
-			} else {
+			} else if (ret != BTRFS_NO_LOG_SYNC) {
 				btrfs_commit_transaction(trans, root);
+			} else {
+				btrfs_end_transaction(trans, root);
 			}
 		}
 		if (file->f_flags & O_DIRECT) {
@@ -1138,6 +1139,13 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
 	int ret = 0;
 	struct btrfs_trans_handle *trans;
 
+
+	/* we wait first, since the writeback may change the inode */
+	root->log_batch++;
+	/* the VFS called filemap_fdatawrite for us */
+	btrfs_wait_ordered_range(inode, 0, (u64)-1);
+	root->log_batch++;
+
 	/*
 	 * check the transaction that last modified this inode
 	 * and see if its already been committed
@@ -1145,6 +1153,11 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
 	if (!BTRFS_I(inode)->last_trans)
 		goto out;
 
+	/*
+	 * if the last transaction that changed this file was before
+	 * the current transaction, we can bail out now without any
+	 * syncing
+	 */
 	mutex_lock(&root->fs_info->trans_mutex);
 	if (BTRFS_I(inode)->last_trans <=
 	    root->fs_info->last_trans_committed) {
@@ -1154,13 +1167,6 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
 	}
 	mutex_unlock(&root->fs_info->trans_mutex);
 
-	root->log_batch++;
-	filemap_fdatawrite(inode->i_mapping);
-	btrfs_wait_ordered_range(inode, 0, (u64)-1);
-	root->log_batch++;
-
-	if (datasync && !(inode->i_state & I_DIRTY_PAGES))
-		goto out;
 	/*
 	 * ok we haven't committed the transaction yet, lets do a commit
 	 */
@@ -1189,21 +1195,25 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
 	 */
 	mutex_unlock(&dentry->d_inode->i_mutex);
 
-	if (ret > 0) {
-		ret = btrfs_commit_transaction(trans, root);
-	} else {
-		ret = btrfs_sync_log(trans, root);
-		if (ret == 0)
-			ret = btrfs_end_transaction(trans, root);
-		else
+	if (ret != BTRFS_NO_LOG_SYNC) {
+		if (ret > 0) {
 			ret = btrfs_commit_transaction(trans, root);
+		} else {
+			ret = btrfs_sync_log(trans, root);
+			if (ret == 0)
+				ret = btrfs_end_transaction(trans, root);
+			else
+				ret = btrfs_commit_transaction(trans, root);
+		}
+	} else {
+		ret = btrfs_end_transaction(trans, root);
 	}
 	mutex_lock(&dentry->d_inode->i_mutex);
 out:
 	return ret > 0 ? EIO : ret;
 }
 
-static struct vm_operations_struct btrfs_file_vm_ops = {
+static const struct vm_operations_struct btrfs_file_vm_ops = {
 	.fault		= filemap_fault,
 	.page_mkwrite	= btrfs_page_mkwrite,
 };
@@ -1215,7 +1225,7 @@ static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
 	return 0;
 }
 
-struct file_operations btrfs_file_operations = {
+const struct file_operations btrfs_file_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= do_sync_read,
 	.aio_read       = generic_file_aio_read,
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 5edcee3a617f..5c2caad76212 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -259,7 +259,9 @@ static int link_free_space(struct btrfs_block_group_cache *block_group,
 
 static void recalculate_thresholds(struct btrfs_block_group_cache *block_group)
 {
-	u64 max_bytes, possible_bytes;
+	u64 max_bytes;
+	u64 bitmap_bytes;
+	u64 extent_bytes;
 
 	/*
 	 * The goal is to keep the total amount of memory used per 1gb of space
@@ -269,22 +271,27 @@ static void recalculate_thresholds(struct btrfs_block_group_cache *block_group)
 	max_bytes = MAX_CACHE_BYTES_PER_GIG *
 		(div64_u64(block_group->key.offset, 1024 * 1024 * 1024));
 
-	possible_bytes = (block_group->total_bitmaps * PAGE_CACHE_SIZE) +
-		(sizeof(struct btrfs_free_space) *
-		 block_group->extents_thresh);
+	/*
+	 * we want to account for 1 more bitmap than what we have so we can make
+	 * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
+	 * we add more bitmaps.
+	 */
+	bitmap_bytes = (block_group->total_bitmaps + 1) * PAGE_CACHE_SIZE;
 
-	if (possible_bytes > max_bytes) {
-		int extent_bytes = max_bytes -
-			(block_group->total_bitmaps * PAGE_CACHE_SIZE);
+	if (bitmap_bytes >= max_bytes) {
+		block_group->extents_thresh = 0;
+		return;
+	}
 
-		if (extent_bytes <= 0) {
-			block_group->extents_thresh = 0;
-			return;
-		}
+	/*
+	 * we want the extent entry threshold to always be at most 1/2 the maxw
+	 * bytes we can have, or whatever is less than that.
+	 */
+	extent_bytes = max_bytes - bitmap_bytes;
+	extent_bytes = min_t(u64, extent_bytes, div64_u64(max_bytes, 2));
 
-		block_group->extents_thresh = extent_bytes /
-			(sizeof(struct btrfs_free_space));
-	}
+	block_group->extents_thresh =
+		div64_u64(extent_bytes, (sizeof(struct btrfs_free_space)));
 }
 
 static void bitmap_clear_bits(struct btrfs_block_group_cache *block_group,
@@ -403,6 +410,7 @@ static void add_new_bitmap(struct btrfs_block_group_cache *block_group,
 	BUG_ON(block_group->total_bitmaps >= max_bitmaps);
 
 	info->offset = offset_to_bitmap(block_group, offset);
+	info->bytes = 0;
 	link_free_space(block_group, info);
 	block_group->total_bitmaps++;
 
diff --git a/fs/btrfs/inode-item.c b/fs/btrfs/inode-item.c
index 6b627c611808..72ce3c173d6a 100644
--- a/fs/btrfs/inode-item.c
+++ b/fs/btrfs/inode-item.c
@@ -149,6 +149,8 @@ int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 		ptr = (unsigned long)(ref + 1);
 		ret = 0;
 	} else if (ret < 0) {
+		if (ret == -EOVERFLOW)
+			ret = -EMLINK;
 		goto out;
 	} else {
 		ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
@@ -177,8 +179,6 @@ int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
 				      sizeof(struct btrfs_inode_item));
-	if (ret == 0 && objectid > root->highest_inode)
-		root->highest_inode = objectid;
 	return ret;
 }
 
diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
index 9abbced1123d..c56eb5909172 100644
--- a/fs/btrfs/inode-map.c
+++ b/fs/btrfs/inode-map.c
@@ -43,9 +43,10 @@ int btrfs_find_highest_inode(struct btrfs_root *root, u64 *objectid)
 		slot = path->slots[0] - 1;
 		l = path->nodes[0];
 		btrfs_item_key_to_cpu(l, &found_key, slot);
-		*objectid = found_key.objectid;
+		*objectid = max_t(u64, found_key.objectid,
+				  BTRFS_FIRST_FREE_OBJECTID - 1);
 	} else {
-		*objectid = BTRFS_FIRST_FREE_OBJECTID;
+		*objectid = BTRFS_FIRST_FREE_OBJECTID - 1;
 	}
 	ret = 0;
 error:
@@ -53,91 +54,27 @@ error:
 	return ret;
 }
 
-/*
- * walks the btree of allocated inodes and find a hole.
- */
 int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
 			     u64 dirid, u64 *objectid)
 {
-	struct btrfs_path *path;
-	struct btrfs_key key;
 	int ret;
-	int slot = 0;
-	u64 last_ino = 0;
-	int start_found;
-	struct extent_buffer *l;
-	struct btrfs_key search_key;
-	u64 search_start = dirid;
-
 	mutex_lock(&root->objectid_mutex);
-	if (root->last_inode_alloc >= BTRFS_FIRST_FREE_OBJECTID &&
-	    root->last_inode_alloc < BTRFS_LAST_FREE_OBJECTID) {
-		*objectid = ++root->last_inode_alloc;
-		mutex_unlock(&root->objectid_mutex);
-		return 0;
-	}
-	path = btrfs_alloc_path();
-	BUG_ON(!path);
-	search_start = max(search_start, (u64)BTRFS_FIRST_FREE_OBJECTID);
-	search_key.objectid = search_start;
-	search_key.type = 0;
-	search_key.offset = 0;
-
-	start_found = 0;
-	ret = btrfs_search_slot(trans, root, &search_key, path, 0, 0);
-	if (ret < 0)
-		goto error;
 
-	while (1) {
-		l = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(l)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret == 0)
-				continue;
-			if (ret < 0)
-				goto error;
-			if (!start_found) {
-				*objectid = search_start;
-				start_found = 1;
-				goto found;
-			}
-			*objectid = last_ino > search_start ?
-				last_ino : search_start;
-			goto found;
-		}
-		btrfs_item_key_to_cpu(l, &key, slot);
-		if (key.objectid >= search_start) {
-			if (start_found) {
-				if (last_ino < search_start)
-					last_ino = search_start;
-				if (key.objectid > last_ino) {
-					*objectid = last_ino;
-					goto found;
-				}
-			} else if (key.objectid > search_start) {
-				*objectid = search_start;
-				goto found;
-			}
-		}
-		if (key.objectid >= BTRFS_LAST_FREE_OBJECTID)
-			break;
+	if (unlikely(root->highest_objectid < BTRFS_FIRST_FREE_OBJECTID)) {
+		ret = btrfs_find_highest_inode(root, &root->highest_objectid);
+		if (ret)
+			goto out;
+	}
 
-		start_found = 1;
-		last_ino = key.objectid + 1;
-		path->slots[0]++;
+	if (unlikely(root->highest_objectid >= BTRFS_LAST_FREE_OBJECTID)) {
+		ret = -ENOSPC;
+		goto out;
 	}
-	BUG_ON(1);
-found:
-	btrfs_release_path(root, path);
-	btrfs_free_path(path);
-	BUG_ON(*objectid < search_start);
-	mutex_unlock(&root->objectid_mutex);
-	return 0;
-error:
-	btrfs_release_path(root, path);
-	btrfs_free_path(path);
+
+	*objectid = ++root->highest_objectid;
+	ret = 0;
+out:
 	mutex_unlock(&root->objectid_mutex);
 	return ret;
 }
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 59cba180fe83..dae12dc7e159 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -55,14 +55,14 @@ struct btrfs_iget_args {
 	struct btrfs_root *root;
 };
 
-static struct inode_operations btrfs_dir_inode_operations;
-static struct inode_operations btrfs_symlink_inode_operations;
-static struct inode_operations btrfs_dir_ro_inode_operations;
-static struct inode_operations btrfs_special_inode_operations;
-static struct inode_operations btrfs_file_inode_operations;
-static struct address_space_operations btrfs_aops;
-static struct address_space_operations btrfs_symlink_aops;
-static struct file_operations btrfs_dir_file_operations;
+static const struct inode_operations btrfs_dir_inode_operations;
+static const struct inode_operations btrfs_symlink_inode_operations;
+static const struct inode_operations btrfs_dir_ro_inode_operations;
+static const struct inode_operations btrfs_special_inode_operations;
+static const struct inode_operations btrfs_file_inode_operations;
+static const struct address_space_operations btrfs_aops;
+static const struct address_space_operations btrfs_symlink_aops;
+static const struct file_operations btrfs_dir_file_operations;
 static struct extent_io_ops btrfs_extent_io_ops;
 
 static struct kmem_cache *btrfs_inode_cachep;
@@ -231,7 +231,8 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
 	}
 
 	ret = btrfs_drop_extents(trans, root, inode, start,
-				 aligned_end, aligned_end, start, &hint_byte);
+				 aligned_end, aligned_end, start,
+				 &hint_byte, 1);
 	BUG_ON(ret);
 
 	if (isize > actual_end)
@@ -240,7 +241,7 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
 				   inline_len, compressed_size,
 				   compressed_pages);
 	BUG_ON(ret);
-	btrfs_drop_extent_cache(inode, start, aligned_end, 0);
+	btrfs_drop_extent_cache(inode, start, aligned_end - 1, 0);
 	return 0;
 }
 
@@ -423,9 +424,12 @@ again:
 			 * and free up our temp pages.
 			 */
 			extent_clear_unlock_delalloc(inode,
-						     &BTRFS_I(inode)->io_tree,
-						     start, end, NULL, 1, 0,
-						     0, 1, 1, 1);
+			     &BTRFS_I(inode)->io_tree,
+			     start, end, NULL,
+			     EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
+			     EXTENT_CLEAR_DELALLOC |
+			     EXTENT_CLEAR_ACCOUNTING |
+			     EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK);
 			ret = 0;
 			goto free_pages_out;
 		}
@@ -611,9 +615,9 @@ static noinline int submit_compressed_extents(struct inode *inode,
 		set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
 
 		while (1) {
-			spin_lock(&em_tree->lock);
+			write_lock(&em_tree->lock);
 			ret = add_extent_mapping(em_tree, em);
-			spin_unlock(&em_tree->lock);
+			write_unlock(&em_tree->lock);
 			if (ret != -EEXIST) {
 				free_extent_map(em);
 				break;
@@ -636,11 +640,14 @@ static noinline int submit_compressed_extents(struct inode *inode,
 		 * clear dirty, set writeback and unlock the pages.
 		 */
 		extent_clear_unlock_delalloc(inode,
-					     &BTRFS_I(inode)->io_tree,
-					     async_extent->start,
-					     async_extent->start +
-					     async_extent->ram_size - 1,
-					     NULL, 1, 1, 0, 1, 1, 0);
+				&BTRFS_I(inode)->io_tree,
+				async_extent->start,
+				async_extent->start +
+				async_extent->ram_size - 1,
+				NULL, EXTENT_CLEAR_UNLOCK_PAGE |
+				EXTENT_CLEAR_UNLOCK |
+				EXTENT_CLEAR_DELALLOC |
+				EXTENT_CLEAR_DIRTY | EXTENT_SET_WRITEBACK);
 
 		ret = btrfs_submit_compressed_write(inode,
 				    async_extent->start,
@@ -711,9 +718,15 @@ static noinline int cow_file_range(struct inode *inode,
 					    start, end, 0, NULL);
 		if (ret == 0) {
 			extent_clear_unlock_delalloc(inode,
-						     &BTRFS_I(inode)->io_tree,
-						     start, end, NULL, 1, 1,
-						     1, 1, 1, 1);
+				     &BTRFS_I(inode)->io_tree,
+				     start, end, NULL,
+				     EXTENT_CLEAR_UNLOCK_PAGE |
+				     EXTENT_CLEAR_UNLOCK |
+				     EXTENT_CLEAR_DELALLOC |
+				     EXTENT_CLEAR_ACCOUNTING |
+				     EXTENT_CLEAR_DIRTY |
+				     EXTENT_SET_WRITEBACK |
+				     EXTENT_END_WRITEBACK);
 			*nr_written = *nr_written +
 			     (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
 			*page_started = 1;
@@ -725,9 +738,20 @@ static noinline int cow_file_range(struct inode *inode,
 	BUG_ON(disk_num_bytes >
 	       btrfs_super_total_bytes(&root->fs_info->super_copy));
 
+
+	read_lock(&BTRFS_I(inode)->extent_tree.lock);
+	em = search_extent_mapping(&BTRFS_I(inode)->extent_tree,
+				   start, num_bytes);
+	if (em) {
+		alloc_hint = em->block_start;
+		free_extent_map(em);
+	}
+	read_unlock(&BTRFS_I(inode)->extent_tree.lock);
 	btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
 
 	while (disk_num_bytes > 0) {
+		unsigned long op;
+
 		cur_alloc_size = min(disk_num_bytes, root->fs_info->max_extent);
 		ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
 					   root->sectorsize, 0, alloc_hint,
@@ -737,7 +761,6 @@ static noinline int cow_file_range(struct inode *inode,
 		em = alloc_extent_map(GFP_NOFS);
 		em->start = start;
 		em->orig_start = em->start;
-
 		ram_size = ins.offset;
 		em->len = ins.offset;
 
@@ -747,9 +770,9 @@ static noinline int cow_file_range(struct inode *inode,
 		set_bit(EXTENT_FLAG_PINNED, &em->flags);
 
 		while (1) {
-			spin_lock(&em_tree->lock);
+			write_lock(&em_tree->lock);
 			ret = add_extent_mapping(em_tree, em);
-			spin_unlock(&em_tree->lock);
+			write_unlock(&em_tree->lock);
 			if (ret != -EEXIST) {
 				free_extent_map(em);
 				break;
@@ -776,11 +799,17 @@ static noinline int cow_file_range(struct inode *inode,
 		/* we're not doing compressed IO, don't unlock the first
 		 * page (which the caller expects to stay locked), don't
 		 * clear any dirty bits and don't set any writeback bits
+		 *
+		 * Do set the Private2 bit so we know this page was properly
+		 * setup for writepage
 		 */
+		op = unlock ? EXTENT_CLEAR_UNLOCK_PAGE : 0;
+		op |= EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
+			EXTENT_SET_PRIVATE2;
+
 		extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
 					     start, start + ram_size - 1,
-					     locked_page, unlock, 1,
-					     1, 0, 0, 0);
+					     locked_page, op);
 		disk_num_bytes -= cur_alloc_size;
 		num_bytes -= cur_alloc_size;
 		alloc_hint = ins.objectid + ins.offset;
@@ -852,8 +881,8 @@ static int cow_file_range_async(struct inode *inode, struct page *locked_page,
 	u64 cur_end;
 	int limit = 10 * 1024 * 1042;
 
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED |
-			 EXTENT_DELALLOC, 1, 0, GFP_NOFS);
+	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED,
+			 1, 0, NULL, GFP_NOFS);
 	while (start < end) {
 		async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS);
 		async_cow->inode = inode;
@@ -994,6 +1023,7 @@ next_slot:
 
 		if (found_key.offset > cur_offset) {
 			extent_end = found_key.offset;
+			extent_type = 0;
 			goto out_check;
 		}
 
@@ -1080,9 +1110,9 @@ out_check:
 			em->bdev = root->fs_info->fs_devices->latest_bdev;
 			set_bit(EXTENT_FLAG_PINNED, &em->flags);
 			while (1) {
-				spin_lock(&em_tree->lock);
+				write_lock(&em_tree->lock);
 				ret = add_extent_mapping(em_tree, em);
-				spin_unlock(&em_tree->lock);
+				write_unlock(&em_tree->lock);
 				if (ret != -EEXIST) {
 					free_extent_map(em);
 					break;
@@ -1100,8 +1130,10 @@ out_check:
 		BUG_ON(ret);
 
 		extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
-					cur_offset, cur_offset + num_bytes - 1,
-					locked_page, 1, 1, 1, 0, 0, 0);
+				cur_offset, cur_offset + num_bytes - 1,
+				locked_page, EXTENT_CLEAR_UNLOCK_PAGE |
+				EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
+				EXTENT_SET_PRIVATE2);
 		cur_offset = extent_end;
 		if (cur_offset > end)
 			break;
@@ -1147,6 +1179,89 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page,
 	return ret;
 }
 
+static int btrfs_split_extent_hook(struct inode *inode,
+				    struct extent_state *orig, u64 split)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	u64 size;
+
+	if (!(orig->state & EXTENT_DELALLOC))
+		return 0;
+
+	size = orig->end - orig->start + 1;
+	if (size > root->fs_info->max_extent) {
+		u64 num_extents;
+		u64 new_size;
+
+		new_size = orig->end - split + 1;
+		num_extents = div64_u64(size + root->fs_info->max_extent - 1,
+					root->fs_info->max_extent);
+
+		/*
+		 * if we break a large extent up then leave oustanding_extents
+		 * be, since we've already accounted for the large extent.
+		 */
+		if (div64_u64(new_size + root->fs_info->max_extent - 1,
+			      root->fs_info->max_extent) < num_extents)
+			return 0;
+	}
+
+	spin_lock(&BTRFS_I(inode)->accounting_lock);
+	BTRFS_I(inode)->outstanding_extents++;
+	spin_unlock(&BTRFS_I(inode)->accounting_lock);
+
+	return 0;
+}
+
+/*
+ * extent_io.c merge_extent_hook, used to track merged delayed allocation
+ * extents so we can keep track of new extents that are just merged onto old
+ * extents, such as when we are doing sequential writes, so we can properly
+ * account for the metadata space we'll need.
+ */
+static int btrfs_merge_extent_hook(struct inode *inode,
+				   struct extent_state *new,
+				   struct extent_state *other)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	u64 new_size, old_size;
+	u64 num_extents;
+
+	/* not delalloc, ignore it */
+	if (!(other->state & EXTENT_DELALLOC))
+		return 0;
+
+	old_size = other->end - other->start + 1;
+	if (new->start < other->start)
+		new_size = other->end - new->start + 1;
+	else
+		new_size = new->end - other->start + 1;
+
+	/* we're not bigger than the max, unreserve the space and go */
+	if (new_size <= root->fs_info->max_extent) {
+		spin_lock(&BTRFS_I(inode)->accounting_lock);
+		BTRFS_I(inode)->outstanding_extents--;
+		spin_unlock(&BTRFS_I(inode)->accounting_lock);
+		return 0;
+	}
+
+	/*
+	 * If we grew by another max_extent, just return, we want to keep that
+	 * reserved amount.
+	 */
+	num_extents = div64_u64(old_size + root->fs_info->max_extent - 1,
+				root->fs_info->max_extent);
+	if (div64_u64(new_size + root->fs_info->max_extent - 1,
+		      root->fs_info->max_extent) > num_extents)
+		return 0;
+
+	spin_lock(&BTRFS_I(inode)->accounting_lock);
+	BTRFS_I(inode)->outstanding_extents--;
+	spin_unlock(&BTRFS_I(inode)->accounting_lock);
+
+	return 0;
+}
+
 /*
  * extent_io.c set_bit_hook, used to track delayed allocation
  * bytes in this file, and to maintain the list of inodes that
@@ -1155,6 +1270,7 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page,
 static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
 		       unsigned long old, unsigned long bits)
 {
+
 	/*
 	 * set_bit and clear bit hooks normally require _irqsave/restore
 	 * but in this case, we are only testeing for the DELALLOC
@@ -1162,6 +1278,10 @@ static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
 	 */
 	if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
+
+		spin_lock(&BTRFS_I(inode)->accounting_lock);
+		BTRFS_I(inode)->outstanding_extents++;
+		spin_unlock(&BTRFS_I(inode)->accounting_lock);
 		btrfs_delalloc_reserve_space(root, inode, end - start + 1);
 		spin_lock(&root->fs_info->delalloc_lock);
 		BTRFS_I(inode)->delalloc_bytes += end - start + 1;
@@ -1178,22 +1298,31 @@ static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
 /*
  * extent_io.c clear_bit_hook, see set_bit_hook for why
  */
-static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
-			 unsigned long old, unsigned long bits)
+static int btrfs_clear_bit_hook(struct inode *inode,
+				struct extent_state *state, unsigned long bits)
 {
 	/*
 	 * set_bit and clear bit hooks normally require _irqsave/restore
 	 * but in this case, we are only testeing for the DELALLOC
 	 * bit, which is only set or cleared with irqs on
 	 */
-	if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+	if ((state->state & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
 
+		if (bits & EXTENT_DO_ACCOUNTING) {
+			spin_lock(&BTRFS_I(inode)->accounting_lock);
+			BTRFS_I(inode)->outstanding_extents--;
+			spin_unlock(&BTRFS_I(inode)->accounting_lock);
+			btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
+		}
+
 		spin_lock(&root->fs_info->delalloc_lock);
-		if (end - start + 1 > root->fs_info->delalloc_bytes) {
+		if (state->end - state->start + 1 >
+		    root->fs_info->delalloc_bytes) {
 			printk(KERN_INFO "btrfs warning: delalloc account "
 			       "%llu %llu\n",
-			       (unsigned long long)end - start + 1,
+			       (unsigned long long)
+			       state->end - state->start + 1,
 			       (unsigned long long)
 			       root->fs_info->delalloc_bytes);
 			btrfs_delalloc_free_space(root, inode, (u64)-1);
@@ -1201,9 +1330,12 @@ static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
 			BTRFS_I(inode)->delalloc_bytes = 0;
 		} else {
 			btrfs_delalloc_free_space(root, inode,
-						  end - start + 1);
-			root->fs_info->delalloc_bytes -= end - start + 1;
-			BTRFS_I(inode)->delalloc_bytes -= end - start + 1;
+						  state->end -
+						  state->start + 1);
+			root->fs_info->delalloc_bytes -= state->end -
+				state->start + 1;
+			BTRFS_I(inode)->delalloc_bytes -= state->end -
+				state->start + 1;
 		}
 		if (BTRFS_I(inode)->delalloc_bytes == 0 &&
 		    !list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
@@ -1374,10 +1506,8 @@ again:
 	lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
 
 	/* already ordered? We're done */
-	if (test_range_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
-			     EXTENT_ORDERED, 0)) {
+	if (PagePrivate2(page))
 		goto out;
-	}
 
 	ordered = btrfs_lookup_ordered_extent(inode, page_start);
 	if (ordered) {
@@ -1413,11 +1543,9 @@ static int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end)
 	struct inode *inode = page->mapping->host;
 	struct btrfs_writepage_fixup *fixup;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	int ret;
 
-	ret = test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
-			     EXTENT_ORDERED, 0);
-	if (ret)
+	/* this page is properly in the ordered list */
+	if (TestClearPagePrivate2(page))
 		return 0;
 
 	if (PageChecked(page))
@@ -1455,9 +1583,19 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	BUG_ON(!path);
 
 	path->leave_spinning = 1;
+
+	/*
+	 * we may be replacing one extent in the tree with another.
+	 * The new extent is pinned in the extent map, and we don't want
+	 * to drop it from the cache until it is completely in the btree.
+	 *
+	 * So, tell btrfs_drop_extents to leave this extent in the cache.
+	 * the caller is expected to unpin it and allow it to be merged
+	 * with the others.
+	 */
 	ret = btrfs_drop_extents(trans, root, inode, file_pos,
 				 file_pos + num_bytes, locked_end,
-				 file_pos, &hint);
+				 file_pos, &hint, 0);
 	BUG_ON(ret);
 
 	ins.objectid = inode->i_ino;
@@ -1485,7 +1623,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 
 	inode_add_bytes(inode, num_bytes);
-	btrfs_drop_extent_cache(inode, file_pos, file_pos + num_bytes - 1, 0);
 
 	ins.objectid = disk_bytenr;
 	ins.offset = disk_num_bytes;
@@ -1596,6 +1733,9 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 						ordered_extent->len,
 						compressed, 0, 0,
 						BTRFS_FILE_EXTENT_REG);
+		unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
+				   ordered_extent->file_offset,
+				   ordered_extent->len);
 		BUG_ON(ret);
 	}
 	unlock_extent(io_tree, ordered_extent->file_offset,
@@ -1623,6 +1763,7 @@ nocow:
 static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
 				struct extent_state *state, int uptodate)
 {
+	ClearPagePrivate2(page);
 	return btrfs_finish_ordered_io(page->mapping->host, start, end);
 }
 
@@ -1669,13 +1810,13 @@ static int btrfs_io_failed_hook(struct bio *failed_bio,
 		failrec->last_mirror = 0;
 		failrec->bio_flags = 0;
 
-		spin_lock(&em_tree->lock);
+		read_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, failrec->len);
 		if (em->start > start || em->start + em->len < start) {
 			free_extent_map(em);
 			em = NULL;
 		}
-		spin_unlock(&em_tree->lock);
+		read_unlock(&em_tree->lock);
 
 		if (!em || IS_ERR(em)) {
 			kfree(failrec);
@@ -1794,7 +1935,7 @@ static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		return 0;
 
 	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID &&
-	    test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1)) {
+	    test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) {
 		clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM,
 				  GFP_NOFS);
 		return 0;
@@ -2352,6 +2493,69 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 	return ret;
 }
 
+int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root,
+			struct inode *dir, u64 objectid,
+			const char *name, int name_len)
+{
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_dir_item *di;
+	struct btrfs_key key;
+	u64 index;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
+				   name, name_len, -1);
+	BUG_ON(!di || IS_ERR(di));
+
+	leaf = path->nodes[0];
+	btrfs_dir_item_key_to_cpu(leaf, di, &key);
+	WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
+	ret = btrfs_delete_one_dir_name(trans, root, path, di);
+	BUG_ON(ret);
+	btrfs_release_path(root, path);
+
+	ret = btrfs_del_root_ref(trans, root->fs_info->tree_root,
+				 objectid, root->root_key.objectid,
+				 dir->i_ino, &index, name, name_len);
+	if (ret < 0) {
+		BUG_ON(ret != -ENOENT);
+		di = btrfs_search_dir_index_item(root, path, dir->i_ino,
+						 name, name_len);
+		BUG_ON(!di || IS_ERR(di));
+
+		leaf = path->nodes[0];
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		btrfs_release_path(root, path);
+		index = key.offset;
+	}
+
+	di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino,
+					 index, name, name_len, -1);
+	BUG_ON(!di || IS_ERR(di));
+
+	leaf = path->nodes[0];
+	btrfs_dir_item_key_to_cpu(leaf, di, &key);
+	WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
+	ret = btrfs_delete_one_dir_name(trans, root, path, di);
+	BUG_ON(ret);
+	btrfs_release_path(root, path);
+
+	btrfs_i_size_write(dir, dir->i_size - name_len * 2);
+	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
+	ret = btrfs_update_inode(trans, root, dir);
+	BUG_ON(ret);
+	dir->i_sb->s_dirt = 1;
+
+	btrfs_free_path(path);
+	return 0;
+}
+
 static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 {
 	struct inode *inode = dentry->d_inode;
@@ -2361,29 +2565,31 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	struct btrfs_trans_handle *trans;
 	unsigned long nr = 0;
 
-	/*
-	 * the FIRST_FREE_OBJECTID check makes sure we don't try to rmdir
-	 * the root of a subvolume or snapshot
-	 */
 	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE ||
-	    inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) {
+	    inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
 		return -ENOTEMPTY;
-	}
 
 	trans = btrfs_start_transaction(root, 1);
 	btrfs_set_trans_block_group(trans, dir);
 
+	if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
+		err = btrfs_unlink_subvol(trans, root, dir,
+					  BTRFS_I(inode)->location.objectid,
+					  dentry->d_name.name,
+					  dentry->d_name.len);
+		goto out;
+	}
+
 	err = btrfs_orphan_add(trans, inode);
 	if (err)
-		goto fail_trans;
+		goto out;
 
 	/* now the directory is empty */
 	err = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
 				 dentry->d_name.name, dentry->d_name.len);
 	if (!err)
 		btrfs_i_size_write(inode, 0);
-
-fail_trans:
+out:
 	nr = trans->blocks_used;
 	ret = btrfs_end_transaction_throttle(trans, root);
 	btrfs_btree_balance_dirty(root, nr);
@@ -2826,12 +3032,22 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
 
 	if ((offset & (blocksize - 1)) == 0)
 		goto out;
+	ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE);
+	if (ret)
+		goto out;
+
+	ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+	if (ret)
+		goto out;
 
 	ret = -ENOMEM;
 again:
 	page = grab_cache_page(mapping, index);
-	if (!page)
+	if (!page) {
+		btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+		btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
 		goto out;
+	}
 
 	page_start = page_offset(page);
 	page_end = page_start + PAGE_CACHE_SIZE - 1;
@@ -2864,7 +3080,16 @@ again:
 		goto again;
 	}
 
-	btrfs_set_extent_delalloc(inode, page_start, page_end);
+	clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
+			  EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+			  GFP_NOFS);
+
+	ret = btrfs_set_extent_delalloc(inode, page_start, page_end);
+	if (ret) {
+		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+		goto out_unlock;
+	}
+
 	ret = 0;
 	if (offset != PAGE_CACHE_SIZE) {
 		kaddr = kmap(page);
@@ -2877,6 +3102,9 @@ again:
 	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 out_unlock:
+	if (ret)
+		btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+	btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
 	unlock_page(page);
 	page_cache_release(page);
 out:
@@ -2895,17 +3123,15 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
 	u64 last_byte;
 	u64 cur_offset;
 	u64 hole_size;
-	int err;
+	int err = 0;
 
 	if (size <= hole_start)
 		return 0;
 
-	err = btrfs_check_metadata_free_space(root);
+	err = btrfs_truncate_page(inode->i_mapping, inode->i_size);
 	if (err)
 		return err;
 
-	btrfs_truncate_page(inode->i_mapping, inode->i_size);
-
 	while (1) {
 		struct btrfs_ordered_extent *ordered;
 		btrfs_wait_ordered_range(inode, hole_start,
@@ -2935,15 +3161,21 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
 						 cur_offset,
 						 cur_offset + hole_size,
 						 block_end,
-						 cur_offset, &hint_byte);
+						 cur_offset, &hint_byte, 1);
+			if (err)
+				break;
+
+			err = btrfs_reserve_metadata_space(root, 1);
 			if (err)
 				break;
+
 			err = btrfs_insert_file_extent(trans, root,
 					inode->i_ino, cur_offset, 0,
 					0, hole_size, 0, hole_size,
 					0, 0, 0);
 			btrfs_drop_extent_cache(inode, hole_start,
 					last_byte - 1, 0);
+			btrfs_unreserve_metadata_space(root, 1);
 		}
 		free_extent_map(em);
 		cur_offset = last_byte;
@@ -3003,6 +3235,11 @@ void btrfs_delete_inode(struct inode *inode)
 	}
 	btrfs_wait_ordered_range(inode, 0, (u64)-1);
 
+	if (inode->i_nlink > 0) {
+		BUG_ON(btrfs_root_refs(&root->root_item) != 0);
+		goto no_delete;
+	}
+
 	btrfs_i_size_write(inode, 0);
 	trans = btrfs_join_transaction(root, 1);
 
@@ -3070,29 +3307,67 @@ out_err:
  * is kind of like crossing a mount point.
  */
 static int fixup_tree_root_location(struct btrfs_root *root,
-			     struct btrfs_key *location,
-			     struct btrfs_root **sub_root,
-			     struct dentry *dentry)
+				    struct inode *dir,
+				    struct dentry *dentry,
+				    struct btrfs_key *location,
+				    struct btrfs_root **sub_root)
 {
-	struct btrfs_root_item *ri;
+	struct btrfs_path *path;
+	struct btrfs_root *new_root;
+	struct btrfs_root_ref *ref;
+	struct extent_buffer *leaf;
+	int ret;
+	int err = 0;
 
-	if (btrfs_key_type(location) != BTRFS_ROOT_ITEM_KEY)
-		return 0;
-	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
-		return 0;
+	path = btrfs_alloc_path();
+	if (!path) {
+		err = -ENOMEM;
+		goto out;
+	}
 
-	*sub_root = btrfs_read_fs_root(root->fs_info, location,
-					dentry->d_name.name,
-					dentry->d_name.len);
-	if (IS_ERR(*sub_root))
-		return PTR_ERR(*sub_root);
+	err = -ENOENT;
+	ret = btrfs_find_root_ref(root->fs_info->tree_root, path,
+				  BTRFS_I(dir)->root->root_key.objectid,
+				  location->objectid);
+	if (ret) {
+		if (ret < 0)
+			err = ret;
+		goto out;
+	}
 
-	ri = &(*sub_root)->root_item;
-	location->objectid = btrfs_root_dirid(ri);
-	btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY);
-	location->offset = 0;
+	leaf = path->nodes[0];
+	ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_root_ref);
+	if (btrfs_root_ref_dirid(leaf, ref) != dir->i_ino ||
+	    btrfs_root_ref_name_len(leaf, ref) != dentry->d_name.len)
+		goto out;
 
-	return 0;
+	ret = memcmp_extent_buffer(leaf, dentry->d_name.name,
+				   (unsigned long)(ref + 1),
+				   dentry->d_name.len);
+	if (ret)
+		goto out;
+
+	btrfs_release_path(root->fs_info->tree_root, path);
+
+	new_root = btrfs_read_fs_root_no_name(root->fs_info, location);
+	if (IS_ERR(new_root)) {
+		err = PTR_ERR(new_root);
+		goto out;
+	}
+
+	if (btrfs_root_refs(&new_root->root_item) == 0) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	*sub_root = new_root;
+	location->objectid = btrfs_root_dirid(&new_root->root_item);
+	location->type = BTRFS_INODE_ITEM_KEY;
+	location->offset = 0;
+	err = 0;
+out:
+	btrfs_free_path(path);
+	return err;
 }
 
 static void inode_tree_add(struct inode *inode)
@@ -3101,11 +3376,13 @@ static void inode_tree_add(struct inode *inode)
 	struct btrfs_inode *entry;
 	struct rb_node **p;
 	struct rb_node *parent;
-
 again:
 	p = &root->inode_tree.rb_node;
 	parent = NULL;
 
+	if (hlist_unhashed(&inode->i_hash))
+		return;
+
 	spin_lock(&root->inode_lock);
 	while (*p) {
 		parent = *p;
@@ -3132,13 +3409,87 @@ again:
 static void inode_tree_del(struct inode *inode)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	int empty = 0;
 
 	spin_lock(&root->inode_lock);
 	if (!RB_EMPTY_NODE(&BTRFS_I(inode)->rb_node)) {
 		rb_erase(&BTRFS_I(inode)->rb_node, &root->inode_tree);
 		RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
+		empty = RB_EMPTY_ROOT(&root->inode_tree);
 	}
 	spin_unlock(&root->inode_lock);
+
+	if (empty && btrfs_root_refs(&root->root_item) == 0) {
+		synchronize_srcu(&root->fs_info->subvol_srcu);
+		spin_lock(&root->inode_lock);
+		empty = RB_EMPTY_ROOT(&root->inode_tree);
+		spin_unlock(&root->inode_lock);
+		if (empty)
+			btrfs_add_dead_root(root);
+	}
+}
+
+int btrfs_invalidate_inodes(struct btrfs_root *root)
+{
+	struct rb_node *node;
+	struct rb_node *prev;
+	struct btrfs_inode *entry;
+	struct inode *inode;
+	u64 objectid = 0;
+
+	WARN_ON(btrfs_root_refs(&root->root_item) != 0);
+
+	spin_lock(&root->inode_lock);
+again:
+	node = root->inode_tree.rb_node;
+	prev = NULL;
+	while (node) {
+		prev = node;
+		entry = rb_entry(node, struct btrfs_inode, rb_node);
+
+		if (objectid < entry->vfs_inode.i_ino)
+			node = node->rb_left;
+		else if (objectid > entry->vfs_inode.i_ino)
+			node = node->rb_right;
+		else
+			break;
+	}
+	if (!node) {
+		while (prev) {
+			entry = rb_entry(prev, struct btrfs_inode, rb_node);
+			if (objectid <= entry->vfs_inode.i_ino) {
+				node = prev;
+				break;
+			}
+			prev = rb_next(prev);
+		}
+	}
+	while (node) {
+		entry = rb_entry(node, struct btrfs_inode, rb_node);
+		objectid = entry->vfs_inode.i_ino + 1;
+		inode = igrab(&entry->vfs_inode);
+		if (inode) {
+			spin_unlock(&root->inode_lock);
+			if (atomic_read(&inode->i_count) > 1)
+				d_prune_aliases(inode);
+			/*
+			 * btrfs_drop_inode will remove it from
+			 * the inode cache when its usage count
+			 * hits zero.
+			 */
+			iput(inode);
+			cond_resched();
+			spin_lock(&root->inode_lock);
+			goto again;
+		}
+
+		if (cond_resched_lock(&root->inode_lock))
+			goto again;
+
+		node = rb_next(node);
+	}
+	spin_unlock(&root->inode_lock);
+	return 0;
 }
 
 static noinline void init_btrfs_i(struct inode *inode)
@@ -3148,6 +3499,7 @@ static noinline void init_btrfs_i(struct inode *inode)
 	bi->generation = 0;
 	bi->sequence = 0;
 	bi->last_trans = 0;
+	bi->last_sub_trans = 0;
 	bi->logged_trans = 0;
 	bi->delalloc_bytes = 0;
 	bi->reserved_bytes = 0;
@@ -3225,15 +3577,41 @@ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
 	return inode;
 }
 
+static struct inode *new_simple_dir(struct super_block *s,
+				    struct btrfs_key *key,
+				    struct btrfs_root *root)
+{
+	struct inode *inode = new_inode(s);
+
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
+
+	init_btrfs_i(inode);
+
+	BTRFS_I(inode)->root = root;
+	memcpy(&BTRFS_I(inode)->location, key, sizeof(*key));
+	BTRFS_I(inode)->dummy_inode = 1;
+
+	inode->i_ino = BTRFS_EMPTY_SUBVOL_DIR_OBJECTID;
+	inode->i_op = &simple_dir_inode_operations;
+	inode->i_fop = &simple_dir_operations;
+	inode->i_mode = S_IFDIR | S_IRUGO | S_IWUSR | S_IXUGO;
+	inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
+
+	return inode;
+}
+
 struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 {
 	struct inode *inode;
-	struct btrfs_inode *bi = BTRFS_I(dir);
-	struct btrfs_root *root = bi->root;
+	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_root *sub_root = root;
 	struct btrfs_key location;
+	int index;
 	int ret;
 
+	dentry->d_op = &btrfs_dentry_operations;
+
 	if (dentry->d_name.len > BTRFS_NAME_LEN)
 		return ERR_PTR(-ENAMETOOLONG);
 
@@ -3242,29 +3620,52 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 	if (ret < 0)
 		return ERR_PTR(ret);
 
-	inode = NULL;
-	if (location.objectid) {
-		ret = fixup_tree_root_location(root, &location, &sub_root,
-						dentry);
-		if (ret < 0)
-			return ERR_PTR(ret);
-		if (ret > 0)
-			return ERR_PTR(-ENOENT);
+	if (location.objectid == 0)
+		return NULL;
+
+	if (location.type == BTRFS_INODE_ITEM_KEY) {
+		inode = btrfs_iget(dir->i_sb, &location, root);
+		return inode;
+	}
+
+	BUG_ON(location.type != BTRFS_ROOT_ITEM_KEY);
+
+	index = srcu_read_lock(&root->fs_info->subvol_srcu);
+	ret = fixup_tree_root_location(root, dir, dentry,
+				       &location, &sub_root);
+	if (ret < 0) {
+		if (ret != -ENOENT)
+			inode = ERR_PTR(ret);
+		else
+			inode = new_simple_dir(dir->i_sb, &location, sub_root);
+	} else {
 		inode = btrfs_iget(dir->i_sb, &location, sub_root);
-		if (IS_ERR(inode))
-			return ERR_CAST(inode);
 	}
+	srcu_read_unlock(&root->fs_info->subvol_srcu, index);
+
 	return inode;
 }
 
+static int btrfs_dentry_delete(struct dentry *dentry)
+{
+	struct btrfs_root *root;
+
+	if (!dentry->d_inode && !IS_ROOT(dentry))
+		dentry = dentry->d_parent;
+
+	if (dentry->d_inode) {
+		root = BTRFS_I(dentry->d_inode)->root;
+		if (btrfs_root_refs(&root->root_item) == 0)
+			return 1;
+	}
+	return 0;
+}
+
 static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry,
 				   struct nameidata *nd)
 {
 	struct inode *inode;
 
-	if (dentry->d_name.len > BTRFS_NAME_LEN)
-		return ERR_PTR(-ENAMETOOLONG);
-
 	inode = btrfs_lookup_dentry(dir, dentry);
 	if (IS_ERR(inode))
 		return ERR_CAST(inode);
@@ -3603,9 +4004,6 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 	if (ret != 0)
 		goto fail;
 
-	if (objectid > root->highest_inode)
-		root->highest_inode = objectid;
-
 	inode->i_uid = current_fsuid();
 
 	if (dir && (dir->i_mode & S_ISGID)) {
@@ -3673,26 +4071,35 @@ int btrfs_add_link(struct btrfs_trans_handle *trans,
 		   struct inode *parent_inode, struct inode *inode,
 		   const char *name, int name_len, int add_backref, u64 index)
 {
-	int ret;
+	int ret = 0;
 	struct btrfs_key key;
 	struct btrfs_root *root = BTRFS_I(parent_inode)->root;
 
-	key.objectid = inode->i_ino;
-	btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
-	key.offset = 0;
+	if (unlikely(inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) {
+		memcpy(&key, &BTRFS_I(inode)->root->root_key, sizeof(key));
+	} else {
+		key.objectid = inode->i_ino;
+		btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
+		key.offset = 0;
+	}
+
+	if (unlikely(inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) {
+		ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
+					 key.objectid, root->root_key.objectid,
+					 parent_inode->i_ino,
+					 index, name, name_len);
+	} else if (add_backref) {
+		ret = btrfs_insert_inode_ref(trans, root,
+					     name, name_len, inode->i_ino,
+					     parent_inode->i_ino, index);
+	}
 
-	ret = btrfs_insert_dir_item(trans, root, name, name_len,
-				    parent_inode->i_ino,
-				    &key, btrfs_inode_type(inode),
-				    index);
 	if (ret == 0) {
-		if (add_backref) {
-			ret = btrfs_insert_inode_ref(trans, root,
-						     name, name_len,
-						     inode->i_ino,
-						     parent_inode->i_ino,
-						     index);
-		}
+		ret = btrfs_insert_dir_item(trans, root, name, name_len,
+					    parent_inode->i_ino, &key,
+					    btrfs_inode_type(inode), index);
+		BUG_ON(ret);
+
 		btrfs_i_size_write(parent_inode, parent_inode->i_size +
 				   name_len * 2);
 		parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
@@ -3732,11 +4139,18 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
 	if (!new_valid_dev(rdev))
 		return -EINVAL;
 
-	err = btrfs_check_metadata_free_space(root);
+	/*
+	 * 2 for inode item and ref
+	 * 2 for dir items
+	 * 1 for xattr if selinux is on
+	 */
+	err = btrfs_reserve_metadata_space(root, 5);
 	if (err)
-		goto fail;
+		return err;
 
 	trans = btrfs_start_transaction(root, 1);
+	if (!trans)
+		goto fail;
 	btrfs_set_trans_block_group(trans, dir);
 
 	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
@@ -3774,6 +4188,7 @@ out_unlock:
 	nr = trans->blocks_used;
 	btrfs_end_transaction_throttle(trans, root);
 fail:
+	btrfs_unreserve_metadata_space(root, 5);
 	if (drop_inode) {
 		inode_dec_link_count(inode);
 		iput(inode);
@@ -3794,10 +4209,18 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
 	u64 objectid;
 	u64 index = 0;
 
-	err = btrfs_check_metadata_free_space(root);
+	/*
+	 * 2 for inode item and ref
+	 * 2 for dir items
+	 * 1 for xattr if selinux is on
+	 */
+	err = btrfs_reserve_metadata_space(root, 5);
 	if (err)
-		goto fail;
+		return err;
+
 	trans = btrfs_start_transaction(root, 1);
+	if (!trans)
+		goto fail;
 	btrfs_set_trans_block_group(trans, dir);
 
 	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
@@ -3838,6 +4261,7 @@ out_unlock:
 	nr = trans->blocks_used;
 	btrfs_end_transaction_throttle(trans, root);
 fail:
+	btrfs_unreserve_metadata_space(root, 5);
 	if (drop_inode) {
 		inode_dec_link_count(inode);
 		iput(inode);
@@ -3860,10 +4284,16 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 	if (inode->i_nlink == 0)
 		return -ENOENT;
 
-	btrfs_inc_nlink(inode);
-	err = btrfs_check_metadata_free_space(root);
+	/*
+	 * 1 item for inode ref
+	 * 2 items for dir items
+	 */
+	err = btrfs_reserve_metadata_space(root, 3);
 	if (err)
-		goto fail;
+		return err;
+
+	btrfs_inc_nlink(inode);
+
 	err = btrfs_set_inode_index(dir, &index);
 	if (err)
 		goto fail;
@@ -3875,20 +4305,19 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 
 	err = btrfs_add_nondir(trans, dentry, inode, 1, index);
 
-	if (err)
-		drop_inode = 1;
-
-	btrfs_update_inode_block_group(trans, dir);
-	err = btrfs_update_inode(trans, root, inode);
-
-	if (err)
+	if (err) {
 		drop_inode = 1;
+	} else {
+		btrfs_update_inode_block_group(trans, dir);
+		err = btrfs_update_inode(trans, root, inode);
+		BUG_ON(err);
+		btrfs_log_new_name(trans, inode, NULL, dentry->d_parent);
+	}
 
 	nr = trans->blocks_used;
-
-	btrfs_log_new_name(trans, inode, NULL, dentry->d_parent);
 	btrfs_end_transaction_throttle(trans, root);
 fail:
+	btrfs_unreserve_metadata_space(root, 3);
 	if (drop_inode) {
 		inode_dec_link_count(inode);
 		iput(inode);
@@ -3908,17 +4337,21 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 	u64 index = 0;
 	unsigned long nr = 1;
 
-	err = btrfs_check_metadata_free_space(root);
+	/*
+	 * 2 items for inode and ref
+	 * 2 items for dir items
+	 * 1 for xattr if selinux is on
+	 */
+	err = btrfs_reserve_metadata_space(root, 5);
 	if (err)
-		goto out_unlock;
+		return err;
 
 	trans = btrfs_start_transaction(root, 1);
-	btrfs_set_trans_block_group(trans, dir);
-
-	if (IS_ERR(trans)) {
-		err = PTR_ERR(trans);
+	if (!trans) {
+		err = -ENOMEM;
 		goto out_unlock;
 	}
+	btrfs_set_trans_block_group(trans, dir);
 
 	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
 	if (err) {
@@ -3967,6 +4400,7 @@ out_fail:
 	btrfs_end_transaction_throttle(trans, root);
 
 out_unlock:
+	btrfs_unreserve_metadata_space(root, 5);
 	if (drop_on_err)
 		iput(inode);
 	btrfs_btree_balance_dirty(root, nr);
@@ -4064,11 +4498,11 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
 	int compressed;
 
 again:
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, start, len);
 	if (em)
 		em->bdev = root->fs_info->fs_devices->latest_bdev;
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	if (em) {
 		if (em->start > start || em->start + em->len <= start)
@@ -4215,6 +4649,11 @@ again:
 				map = kmap(page);
 				read_extent_buffer(leaf, map + pg_offset, ptr,
 						   copy_size);
+				if (pg_offset + copy_size < PAGE_CACHE_SIZE) {
+					memset(map + pg_offset + copy_size, 0,
+					       PAGE_CACHE_SIZE - pg_offset -
+					       copy_size);
+				}
 				kunmap(page);
 			}
 			flush_dcache_page(page);
@@ -4259,7 +4698,7 @@ insert:
 	}
 
 	err = 0;
-	spin_lock(&em_tree->lock);
+	write_lock(&em_tree->lock);
 	ret = add_extent_mapping(em_tree, em);
 	/* it is possible that someone inserted the extent into the tree
 	 * while we had the lock dropped.  It is also possible that
@@ -4299,7 +4738,7 @@ insert:
 			err = 0;
 		}
 	}
-	spin_unlock(&em_tree->lock);
+	write_unlock(&em_tree->lock);
 out:
 	if (path)
 		btrfs_free_path(path);
@@ -4398,13 +4837,21 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 	u64 page_start = page_offset(page);
 	u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
 
+
+	/*
+	 * we have the page locked, so new writeback can't start,
+	 * and the dirty bit won't be cleared while we are here.
+	 *
+	 * Wait for IO on this page so that we can safely clear
+	 * the PagePrivate2 bit and do ordered accounting
+	 */
 	wait_on_page_writeback(page);
+
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
 	if (offset) {
 		btrfs_releasepage(page, GFP_NOFS);
 		return;
 	}
-
 	lock_extent(tree, page_start, page_end, GFP_NOFS);
 	ordered = btrfs_lookup_ordered_extent(page->mapping->host,
 					   page_offset(page));
@@ -4415,16 +4862,22 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 		 */
 		clear_extent_bit(tree, page_start, page_end,
 				 EXTENT_DIRTY | EXTENT_DELALLOC |
-				 EXTENT_LOCKED, 1, 0, GFP_NOFS);
-		btrfs_finish_ordered_io(page->mapping->host,
-					page_start, page_end);
+				 EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0,
+				 NULL, GFP_NOFS);
+		/*
+		 * whoever cleared the private bit is responsible
+		 * for the finish_ordered_io
+		 */
+		if (TestClearPagePrivate2(page)) {
+			btrfs_finish_ordered_io(page->mapping->host,
+						page_start, page_end);
+		}
 		btrfs_put_ordered_extent(ordered);
 		lock_extent(tree, page_start, page_end, GFP_NOFS);
 	}
 	clear_extent_bit(tree, page_start, page_end,
 		 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
-		 EXTENT_ORDERED,
-		 1, 1, GFP_NOFS);
+		 EXTENT_DO_ACCOUNTING, 1, 1, NULL, GFP_NOFS);
 	__btrfs_releasepage(page, GFP_NOFS);
 
 	ClearPageChecked(page);
@@ -4473,6 +4926,13 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 		goto out;
 	}
 
+	ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+	if (ret) {
+		btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+		ret = VM_FAULT_SIGBUS;
+		goto out;
+	}
+
 	ret = VM_FAULT_NOPAGE; /* make the VM retry the fault */
 again:
 	lock_page(page);
@@ -4504,7 +4964,24 @@ again:
 		goto again;
 	}
 
-	btrfs_set_extent_delalloc(inode, page_start, page_end);
+	/*
+	 * XXX - page_mkwrite gets called every time the page is dirtied, even
+	 * if it was already dirty, so for space accounting reasons we need to
+	 * clear any delalloc bits for the range we are fixing to save.  There
+	 * is probably a better way to do this, but for now keep consistent with
+	 * prepare_pages in the normal write path.
+	 */
+	clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
+			  EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+			  GFP_NOFS);
+
+	ret = btrfs_set_extent_delalloc(inode, page_start, page_end);
+	if (ret) {
+		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+		ret = VM_FAULT_SIGBUS;
+		btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+		goto out_unlock;
+	}
 	ret = 0;
 
 	/* page is wholly or partially inside EOF */
@@ -4521,11 +4998,17 @@ again:
 	}
 	ClearPageChecked(page);
 	set_page_dirty(page);
+	SetPageUptodate(page);
+
+	BTRFS_I(inode)->last_trans = root->fs_info->generation;
+	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
 
-	BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
 	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 out_unlock:
+	btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
+	if (!ret)
+		return VM_FAULT_LOCKED;
 	unlock_page(page);
 out:
 	return ret;
@@ -4544,7 +5027,9 @@ static void btrfs_truncate(struct inode *inode)
 	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
 		return;
 
-	btrfs_truncate_page(inode->i_mapping, inode->i_size);
+	ret = btrfs_truncate_page(inode->i_mapping, inode->i_size);
+	if (ret)
+		return;
 	btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
 
 	trans = btrfs_start_transaction(root, 1);
@@ -4594,11 +5079,11 @@ out:
  * create a new subvolume directory/inode (helper for the ioctl).
  */
 int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *new_root, struct dentry *dentry,
+			     struct btrfs_root *new_root,
 			     u64 new_dirid, u64 alloc_hint)
 {
 	struct inode *inode;
-	int error;
+	int err;
 	u64 index = 0;
 
 	inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
@@ -4611,11 +5096,10 @@ int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
 	inode->i_nlink = 1;
 	btrfs_i_size_write(inode, 0);
 
-	error = btrfs_update_inode(trans, new_root, inode);
-	if (error)
-		return error;
+	err = btrfs_update_inode(trans, new_root, inode);
+	BUG_ON(err);
 
-	d_instantiate(dentry, inode);
+	iput(inode);
 	return 0;
 }
 
@@ -4640,7 +5124,11 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	if (!ei)
 		return NULL;
 	ei->last_trans = 0;
+	ei->last_sub_trans = 0;
 	ei->logged_trans = 0;
+	ei->outstanding_extents = 0;
+	ei->reserved_extents = 0;
+	spin_lock_init(&ei->accounting_lock);
 	btrfs_ordered_inode_tree_init(&ei->ordered_tree);
 	INIT_LIST_HEAD(&ei->i_orphan);
 	INIT_LIST_HEAD(&ei->ordered_operations);
@@ -4693,6 +5181,16 @@ void btrfs_destroy_inode(struct inode *inode)
 	kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
 }
 
+void btrfs_drop_inode(struct inode *inode)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+
+	if (inode->i_nlink > 0 && btrfs_root_refs(&root->root_item) == 0)
+		generic_delete_inode(inode);
+	else
+		generic_drop_inode(inode);
+}
+
 static void init_once(void *foo)
 {
 	struct btrfs_inode *ei = (struct btrfs_inode *) foo;
@@ -4761,31 +5259,37 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 {
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root = BTRFS_I(old_dir)->root;
+	struct btrfs_root *dest = BTRFS_I(new_dir)->root;
 	struct inode *new_inode = new_dentry->d_inode;
 	struct inode *old_inode = old_dentry->d_inode;
 	struct timespec ctime = CURRENT_TIME;
 	u64 index = 0;
+	u64 root_objectid;
 	int ret;
 
-	/* we're not allowed to rename between subvolumes */
-	if (BTRFS_I(old_inode)->root->root_key.objectid !=
-	    BTRFS_I(new_dir)->root->root_key.objectid)
+	if (new_dir->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
+		return -EPERM;
+
+	/* we only allow rename subvolume link between subvolumes */
+	if (old_inode->i_ino != BTRFS_FIRST_FREE_OBJECTID && root != dest)
 		return -EXDEV;
 
+	if (old_inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID ||
+	    (new_inode && new_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID))
+		return -ENOTEMPTY;
+
 	if (S_ISDIR(old_inode->i_mode) && new_inode &&
-	    new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) {
+	    new_inode->i_size > BTRFS_EMPTY_DIR_SIZE)
 		return -ENOTEMPTY;
-	}
 
-	/* to rename a snapshot or subvolume, we need to juggle the
-	 * backrefs.  This isn't coded yet
+	/*
+	 * 2 items for dir items
+	 * 1 item for orphan entry
+	 * 1 item for ref
 	 */
-	if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
-		return -EXDEV;
-
-	ret = btrfs_check_metadata_free_space(root);
+	ret = btrfs_reserve_metadata_space(root, 4);
 	if (ret)
-		goto out_unlock;
+		return ret;
 
 	/*
 	 * we're using rename to replace one file with another.
@@ -4796,8 +5300,40 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	    old_inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT)
 		filemap_flush(old_inode->i_mapping);
 
+	/* close the racy window with snapshot create/destroy ioctl */
+	if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
+		down_read(&root->fs_info->subvol_sem);
+
 	trans = btrfs_start_transaction(root, 1);
+	btrfs_set_trans_block_group(trans, new_dir);
+
+	if (dest != root)
+		btrfs_record_root_in_trans(trans, dest);
 
+	ret = btrfs_set_inode_index(new_dir, &index);
+	if (ret)
+		goto out_fail;
+
+	if (unlikely(old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) {
+		/* force full log commit if subvolume involved. */
+		root->fs_info->last_trans_log_full_commit = trans->transid;
+	} else {
+		ret = btrfs_insert_inode_ref(trans, dest,
+					     new_dentry->d_name.name,
+					     new_dentry->d_name.len,
+					     old_inode->i_ino,
+					     new_dir->i_ino, index);
+		if (ret)
+			goto out_fail;
+		/*
+		 * this is an ugly little race, but the rename is required
+		 * to make sure that if we crash, the inode is either at the
+		 * old name or the new one.  pinning the log transaction lets
+		 * us make sure we don't allow a log commit to come in after
+		 * we unlink the name but before we add the new name back in.
+		 */
+		btrfs_pin_log_trans(root);
+	}
 	/*
 	 * make sure the inode gets flushed if it is replacing
 	 * something.
@@ -4807,18 +5343,6 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		btrfs_add_ordered_operation(trans, root, old_inode);
 	}
 
-	/*
-	 * this is an ugly little race, but the rename is required to make
-	 * sure that if we crash, the inode is either at the old name
-	 * or the new one.  pinning the log transaction lets us make sure
-	 * we don't allow a log commit to come in after we unlink the
-	 * name but before we add the new name back in.
-	 */
-	btrfs_pin_log_trans(root);
-
-	btrfs_set_trans_block_group(trans, new_dir);
-
-	btrfs_inc_nlink(old_dentry->d_inode);
 	old_dir->i_ctime = old_dir->i_mtime = ctime;
 	new_dir->i_ctime = new_dir->i_mtime = ctime;
 	old_inode->i_ctime = ctime;
@@ -4826,47 +5350,60 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (old_dentry->d_parent != new_dentry->d_parent)
 		btrfs_record_unlink_dir(trans, old_dir, old_inode, 1);
 
-	ret = btrfs_unlink_inode(trans, root, old_dir, old_dentry->d_inode,
-				 old_dentry->d_name.name,
-				 old_dentry->d_name.len);
-	if (ret)
-		goto out_fail;
+	if (unlikely(old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) {
+		root_objectid = BTRFS_I(old_inode)->root->root_key.objectid;
+		ret = btrfs_unlink_subvol(trans, root, old_dir, root_objectid,
+					old_dentry->d_name.name,
+					old_dentry->d_name.len);
+	} else {
+		btrfs_inc_nlink(old_dentry->d_inode);
+		ret = btrfs_unlink_inode(trans, root, old_dir,
+					 old_dentry->d_inode,
+					 old_dentry->d_name.name,
+					 old_dentry->d_name.len);
+	}
+	BUG_ON(ret);
 
 	if (new_inode) {
 		new_inode->i_ctime = CURRENT_TIME;
-		ret = btrfs_unlink_inode(trans, root, new_dir,
-					 new_dentry->d_inode,
-					 new_dentry->d_name.name,
-					 new_dentry->d_name.len);
-		if (ret)
-			goto out_fail;
+		if (unlikely(new_inode->i_ino ==
+			     BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
+			root_objectid = BTRFS_I(new_inode)->location.objectid;
+			ret = btrfs_unlink_subvol(trans, dest, new_dir,
+						root_objectid,
+						new_dentry->d_name.name,
+						new_dentry->d_name.len);
+			BUG_ON(new_inode->i_nlink == 0);
+		} else {
+			ret = btrfs_unlink_inode(trans, dest, new_dir,
+						 new_dentry->d_inode,
+						 new_dentry->d_name.name,
+						 new_dentry->d_name.len);
+		}
+		BUG_ON(ret);
 		if (new_inode->i_nlink == 0) {
 			ret = btrfs_orphan_add(trans, new_dentry->d_inode);
-			if (ret)
-				goto out_fail;
+			BUG_ON(ret);
 		}
-
 	}
-	ret = btrfs_set_inode_index(new_dir, &index);
-	if (ret)
-		goto out_fail;
 
-	ret = btrfs_add_link(trans, new_dentry->d_parent->d_inode,
-			     old_inode, new_dentry->d_name.name,
-			     new_dentry->d_name.len, 1, index);
-	if (ret)
-		goto out_fail;
+	ret = btrfs_add_link(trans, new_dir, old_inode,
+			     new_dentry->d_name.name,
+			     new_dentry->d_name.len, 0, index);
+	BUG_ON(ret);
 
-	btrfs_log_new_name(trans, old_inode, old_dir,
-				       new_dentry->d_parent);
+	if (old_inode->i_ino != BTRFS_FIRST_FREE_OBJECTID) {
+		btrfs_log_new_name(trans, old_inode, old_dir,
+				   new_dentry->d_parent);
+		btrfs_end_log_trans(root);
+	}
 out_fail:
-
-	/* this btrfs_end_log_trans just allows the current
-	 * log-sub transaction to complete
-	 */
-	btrfs_end_log_trans(root);
 	btrfs_end_transaction_throttle(trans, root);
-out_unlock:
+
+	if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
+		up_read(&root->fs_info->subvol_sem);
+
+	btrfs_unreserve_metadata_space(root, 4);
 	return ret;
 }
 
@@ -4938,11 +5475,18 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
 		return -ENAMETOOLONG;
 
-	err = btrfs_check_metadata_free_space(root);
+	/*
+	 * 2 items for inode item and ref
+	 * 2 items for dir items
+	 * 1 item for xattr if selinux is on
+	 */
+	err = btrfs_reserve_metadata_space(root, 5);
 	if (err)
-		goto out_fail;
+		return err;
 
 	trans = btrfs_start_transaction(root, 1);
+	if (!trans)
+		goto out_fail;
 	btrfs_set_trans_block_group(trans, dir);
 
 	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
@@ -5023,6 +5567,7 @@ out_unlock:
 	nr = trans->blocks_used;
 	btrfs_end_transaction_throttle(trans, root);
 out_fail:
+	btrfs_unreserve_metadata_space(root, 5);
 	if (drop_inode) {
 		inode_dec_link_count(inode);
 		iput(inode);
@@ -5044,6 +5589,11 @@ static int prealloc_file_range(struct btrfs_trans_handle *trans,
 
 	while (num_bytes > 0) {
 		alloc_size = min(num_bytes, root->fs_info->max_extent);
+
+		ret = btrfs_reserve_metadata_space(root, 1);
+		if (ret)
+			goto out;
+
 		ret = btrfs_reserve_extent(trans, root, alloc_size,
 					   root->sectorsize, 0, alloc_hint,
 					   (u64)-1, &ins, 1);
@@ -5058,9 +5608,12 @@ static int prealloc_file_range(struct btrfs_trans_handle *trans,
 						  0, 0, 0,
 						  BTRFS_FILE_EXTENT_PREALLOC);
 		BUG_ON(ret);
+		btrfs_drop_extent_cache(inode, cur_offset,
+					cur_offset + ins.offset -1, 0);
 		num_bytes -= ins.offset;
 		cur_offset += ins.offset;
 		alloc_hint = ins.objectid + ins.offset;
+		btrfs_unreserve_metadata_space(root, 1);
 	}
 out:
 	if (cur_offset > start) {
@@ -5201,7 +5754,7 @@ static int btrfs_permission(struct inode *inode, int mask)
 	return generic_permission(inode, mask, btrfs_check_acl);
 }
 
-static struct inode_operations btrfs_dir_inode_operations = {
+static const struct inode_operations btrfs_dir_inode_operations = {
 	.getattr	= btrfs_getattr,
 	.lookup		= btrfs_lookup,
 	.create		= btrfs_create,
@@ -5219,11 +5772,12 @@ static struct inode_operations btrfs_dir_inode_operations = {
 	.removexattr	= btrfs_removexattr,
 	.permission	= btrfs_permission,
 };
-static struct inode_operations btrfs_dir_ro_inode_operations = {
+static const struct inode_operations btrfs_dir_ro_inode_operations = {
 	.lookup		= btrfs_lookup,
 	.permission	= btrfs_permission,
 };
-static struct file_operations btrfs_dir_file_operations = {
+
+static const struct file_operations btrfs_dir_file_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
 	.readdir	= btrfs_real_readdir,
@@ -5245,6 +5799,8 @@ static struct extent_io_ops btrfs_extent_io_ops = {
 	.readpage_io_failed_hook = btrfs_io_failed_hook,
 	.set_bit_hook = btrfs_set_bit_hook,
 	.clear_bit_hook = btrfs_clear_bit_hook,
+	.merge_extent_hook = btrfs_merge_extent_hook,
+	.split_extent_hook = btrfs_split_extent_hook,
 };
 
 /*
@@ -5259,7 +5815,7 @@ static struct extent_io_ops btrfs_extent_io_ops = {
  *
  * For now we're avoiding this by dropping bmap.
  */
-static struct address_space_operations btrfs_aops = {
+static const struct address_space_operations btrfs_aops = {
 	.readpage	= btrfs_readpage,
 	.writepage	= btrfs_writepage,
 	.writepages	= btrfs_writepages,
@@ -5269,16 +5825,17 @@ static struct address_space_operations btrfs_aops = {
 	.invalidatepage = btrfs_invalidatepage,
 	.releasepage	= btrfs_releasepage,
 	.set_page_dirty	= btrfs_set_page_dirty,
+	.error_remove_page = generic_error_remove_page,
 };
 
-static struct address_space_operations btrfs_symlink_aops = {
+static const struct address_space_operations btrfs_symlink_aops = {
 	.readpage	= btrfs_readpage,
 	.writepage	= btrfs_writepage,
 	.invalidatepage = btrfs_invalidatepage,
 	.releasepage	= btrfs_releasepage,
 };
 
-static struct inode_operations btrfs_file_inode_operations = {
+static const struct inode_operations btrfs_file_inode_operations = {
 	.truncate	= btrfs_truncate,
 	.getattr	= btrfs_getattr,
 	.setattr	= btrfs_setattr,
@@ -5290,7 +5847,7 @@ static struct inode_operations btrfs_file_inode_operations = {
 	.fallocate	= btrfs_fallocate,
 	.fiemap		= btrfs_fiemap,
 };
-static struct inode_operations btrfs_special_inode_operations = {
+static const struct inode_operations btrfs_special_inode_operations = {
 	.getattr	= btrfs_getattr,
 	.setattr	= btrfs_setattr,
 	.permission	= btrfs_permission,
@@ -5299,7 +5856,7 @@ static struct inode_operations btrfs_special_inode_operations = {
 	.listxattr	= btrfs_listxattr,
 	.removexattr	= btrfs_removexattr,
 };
-static struct inode_operations btrfs_symlink_inode_operations = {
+static const struct inode_operations btrfs_symlink_inode_operations = {
 	.readlink	= generic_readlink,
 	.follow_link	= page_follow_link_light,
 	.put_link	= page_put_link,
@@ -5309,3 +5866,7 @@ static struct inode_operations btrfs_symlink_inode_operations = {
 	.listxattr	= btrfs_listxattr,
 	.removexattr	= btrfs_removexattr,
 };
+
+const struct dentry_operations btrfs_dentry_operations = {
+	.d_delete	= btrfs_dentry_delete,
+};
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index bd88f25889f7..cdbb054102b9 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -230,8 +230,8 @@ static noinline int create_subvol(struct btrfs_root *root,
 	struct btrfs_root_item root_item;
 	struct btrfs_inode_item *inode_item;
 	struct extent_buffer *leaf;
-	struct btrfs_root *new_root = root;
-	struct inode *dir;
+	struct btrfs_root *new_root;
+	struct inode *dir = dentry->d_parent->d_inode;
 	int ret;
 	int err;
 	u64 objectid;
@@ -239,9 +239,15 @@ static noinline int create_subvol(struct btrfs_root *root,
 	u64 index = 0;
 	unsigned long nr = 1;
 
-	ret = btrfs_check_metadata_free_space(root);
+	/*
+	 * 1 - inode item
+	 * 2 - refs
+	 * 1 - root item
+	 * 2 - dir items
+	 */
+	ret = btrfs_reserve_metadata_space(root, 6);
 	if (ret)
-		goto fail_commit;
+		return ret;
 
 	trans = btrfs_start_transaction(root, 1);
 	BUG_ON(!trans);
@@ -304,11 +310,17 @@ static noinline int create_subvol(struct btrfs_root *root,
 	if (ret)
 		goto fail;
 
+	key.offset = (u64)-1;
+	new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
+	BUG_ON(IS_ERR(new_root));
+
+	btrfs_record_root_in_trans(trans, new_root);
+
+	ret = btrfs_create_subvol_root(trans, new_root, new_dirid,
+				       BTRFS_I(dir)->block_group);
 	/*
 	 * insert the directory item
 	 */
-	key.offset = (u64)-1;
-	dir = dentry->d_parent->d_inode;
 	ret = btrfs_set_inode_index(dir, &index);
 	BUG_ON(ret);
 
@@ -322,43 +334,20 @@ static noinline int create_subvol(struct btrfs_root *root,
 	ret = btrfs_update_inode(trans, root, dir);
 	BUG_ON(ret);
 
-	/* add the backref first */
 	ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
-				 objectid, BTRFS_ROOT_BACKREF_KEY,
-				 root->root_key.objectid,
+				 objectid, root->root_key.objectid,
 				 dir->i_ino, index, name, namelen);
 
 	BUG_ON(ret);
 
-	/* now add the forward ref */
-	ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
-				 root->root_key.objectid, BTRFS_ROOT_REF_KEY,
-				 objectid,
-				 dir->i_ino, index, name, namelen);
-
-	BUG_ON(ret);
-
-	ret = btrfs_commit_transaction(trans, root);
-	if (ret)
-		goto fail_commit;
-
-	new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
-	BUG_ON(!new_root);
-
-	trans = btrfs_start_transaction(new_root, 1);
-	BUG_ON(!trans);
-
-	ret = btrfs_create_subvol_root(trans, new_root, dentry, new_dirid,
-				       BTRFS_I(dir)->block_group);
-	if (ret)
-		goto fail;
-
+	d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
 fail:
 	nr = trans->blocks_used;
-	err = btrfs_commit_transaction(trans, new_root);
+	err = btrfs_commit_transaction(trans, root);
 	if (err && !ret)
 		ret = err;
-fail_commit:
+
+	btrfs_unreserve_metadata_space(root, 6);
 	btrfs_btree_balance_dirty(root, nr);
 	return ret;
 }
@@ -375,19 +364,27 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 	if (!root->ref_cows)
 		return -EINVAL;
 
-	ret = btrfs_check_metadata_free_space(root);
+	/*
+	 * 1 - inode item
+	 * 2 - refs
+	 * 1 - root item
+	 * 2 - dir items
+	 */
+	ret = btrfs_reserve_metadata_space(root, 6);
 	if (ret)
 		goto fail_unlock;
 
 	pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
 	if (!pending_snapshot) {
 		ret = -ENOMEM;
+		btrfs_unreserve_metadata_space(root, 6);
 		goto fail_unlock;
 	}
 	pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
 	if (!pending_snapshot->name) {
 		ret = -ENOMEM;
 		kfree(pending_snapshot);
+		btrfs_unreserve_metadata_space(root, 6);
 		goto fail_unlock;
 	}
 	memcpy(pending_snapshot->name, name, namelen);
@@ -420,14 +417,15 @@ static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
  * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
  * inside this filesystem so it's quite a bit simpler.
  */
-static noinline int btrfs_mksubvol(struct path *parent, char *name,
-				   int mode, int namelen,
+static noinline int btrfs_mksubvol(struct path *parent,
+				   char *name, int namelen,
 				   struct btrfs_root *snap_src)
 {
+	struct inode *dir  = parent->dentry->d_inode;
 	struct dentry *dentry;
 	int error;
 
-	mutex_lock_nested(&parent->dentry->d_inode->i_mutex, I_MUTEX_PARENT);
+	mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
 
 	dentry = lookup_one_len(name, parent->dentry, namelen);
 	error = PTR_ERR(dentry);
@@ -438,99 +436,39 @@ static noinline int btrfs_mksubvol(struct path *parent, char *name,
 	if (dentry->d_inode)
 		goto out_dput;
 
-	if (!IS_POSIXACL(parent->dentry->d_inode))
-		mode &= ~current_umask();
-
 	error = mnt_want_write(parent->mnt);
 	if (error)
 		goto out_dput;
 
-	error = btrfs_may_create(parent->dentry->d_inode, dentry);
+	error = btrfs_may_create(dir, dentry);
 	if (error)
 		goto out_drop_write;
 
-	/*
-	 * Actually perform the low-level subvolume creation after all
-	 * this VFS fuzz.
-	 *
-	 * Eventually we want to pass in an inode under which we create this
-	 * subvolume, but for now all are under the filesystem root.
-	 *
-	 * Also we should pass on the mode eventually to allow creating new
-	 * subvolume with specific mode bits.
-	 */
+	down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
+
+	if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
+		goto out_up_read;
+
 	if (snap_src) {
-		struct dentry *dir = dentry->d_parent;
-		struct dentry *test = dir->d_parent;
-		struct btrfs_path *path = btrfs_alloc_path();
-		int ret;
-		u64 test_oid;
-		u64 parent_oid = BTRFS_I(dir->d_inode)->root->root_key.objectid;
-
-		test_oid = snap_src->root_key.objectid;
-
-		ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
-					  path, parent_oid, test_oid);
-		if (ret == 0)
-			goto create;
-		btrfs_release_path(snap_src->fs_info->tree_root, path);
-
-		/* we need to make sure we aren't creating a directory loop
-		 * by taking a snapshot of something that has our current
-		 * subvol in its directory tree.  So, this loops through
-		 * the dentries and checks the forward refs for each subvolume
-		 * to see if is references the subvolume where we are
-		 * placing this new snapshot.
-		 */
-		while (1) {
-			if (!test ||
-			    dir == snap_src->fs_info->sb->s_root ||
-			    test == snap_src->fs_info->sb->s_root ||
-			    test->d_inode->i_sb != snap_src->fs_info->sb) {
-				break;
-			}
-			if (S_ISLNK(test->d_inode->i_mode)) {
-				printk(KERN_INFO "Btrfs symlink in snapshot "
-				       "path, failed\n");
-				error = -EMLINK;
-				btrfs_free_path(path);
-				goto out_drop_write;
-			}
-			test_oid =
-				BTRFS_I(test->d_inode)->root->root_key.objectid;
-			ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
-				  path, test_oid, parent_oid);
-			if (ret == 0) {
-				printk(KERN_INFO "Btrfs snapshot creation "
-				       "failed, looping\n");
-				error = -EMLINK;
-				btrfs_free_path(path);
-				goto out_drop_write;
-			}
-			btrfs_release_path(snap_src->fs_info->tree_root, path);
-			test = test->d_parent;
-		}
-create:
-		btrfs_free_path(path);
-		error = create_snapshot(snap_src, dentry, name, namelen);
+		error = create_snapshot(snap_src, dentry,
+					name, namelen);
 	} else {
-		error = create_subvol(BTRFS_I(parent->dentry->d_inode)->root,
-				      dentry, name, namelen);
+		error = create_subvol(BTRFS_I(dir)->root, dentry,
+				      name, namelen);
 	}
-	if (error)
-		goto out_drop_write;
-
-	fsnotify_mkdir(parent->dentry->d_inode, dentry);
+	if (!error)
+		fsnotify_mkdir(dir, dentry);
+out_up_read:
+	up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
 out_drop_write:
 	mnt_drop_write(parent->mnt);
 out_dput:
 	dput(dentry);
 out_unlock:
-	mutex_unlock(&parent->dentry->d_inode->i_mutex);
+	mutex_unlock(&dir->i_mutex);
 	return error;
 }
 
-
 static int btrfs_defrag_file(struct file *file)
 {
 	struct inode *inode = fdentry(file)->d_inode;
@@ -596,9 +534,8 @@ again:
 		clear_page_dirty_for_io(page);
 
 		btrfs_set_extent_delalloc(inode, page_start, page_end);
-
-		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 		set_page_dirty(page);
+		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 		unlock_page(page);
 		page_cache_release(page);
 		balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
@@ -609,7 +546,8 @@ out_unlock:
 	return 0;
 }
 
-static int btrfs_ioctl_resize(struct btrfs_root *root, void __user *arg)
+static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
+					void __user *arg)
 {
 	u64 new_size;
 	u64 old_size;
@@ -718,10 +656,7 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
 {
 	struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
 	struct btrfs_ioctl_vol_args *vol_args;
-	struct btrfs_dir_item *di;
-	struct btrfs_path *path;
 	struct file *src_file;
-	u64 root_dirid;
 	int namelen;
 	int ret = 0;
 
@@ -739,32 +674,9 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
 		goto out;
 	}
 
-	path = btrfs_alloc_path();
-	if (!path) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
-	di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
-			    path, root_dirid,
-			    vol_args->name, namelen, 0);
-	btrfs_free_path(path);
-
-	if (di && !IS_ERR(di)) {
-		ret = -EEXIST;
-		goto out;
-	}
-
-	if (IS_ERR(di)) {
-		ret = PTR_ERR(di);
-		goto out;
-	}
-
 	if (subvol) {
-		ret = btrfs_mksubvol(&file->f_path, vol_args->name,
-				     file->f_path.dentry->d_inode->i_mode,
-				     namelen, NULL);
+		ret = btrfs_mksubvol(&file->f_path, vol_args->name, namelen,
+				     NULL);
 	} else {
 		struct inode *src_inode;
 		src_file = fget(vol_args->fd);
@@ -781,17 +693,157 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
 			fput(src_file);
 			goto out;
 		}
-		ret = btrfs_mksubvol(&file->f_path, vol_args->name,
-			     file->f_path.dentry->d_inode->i_mode,
-			     namelen, BTRFS_I(src_inode)->root);
+		ret = btrfs_mksubvol(&file->f_path, vol_args->name, namelen,
+				     BTRFS_I(src_inode)->root);
 		fput(src_file);
 	}
-
 out:
 	kfree(vol_args);
 	return ret;
 }
 
+/*
+ * helper to check if the subvolume references other subvolumes
+ */
+static noinline int may_destroy_subvol(struct btrfs_root *root)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	key.objectid = root->root_key.objectid;
+	key.type = BTRFS_ROOT_REF_KEY;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
+				&key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+	BUG_ON(ret == 0);
+
+	ret = 0;
+	if (path->slots[0] > 0) {
+		path->slots[0]--;
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		if (key.objectid == root->root_key.objectid &&
+		    key.type == BTRFS_ROOT_REF_KEY)
+			ret = -ENOTEMPTY;
+	}
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+static noinline int btrfs_ioctl_snap_destroy(struct file *file,
+					     void __user *arg)
+{
+	struct dentry *parent = fdentry(file);
+	struct dentry *dentry;
+	struct inode *dir = parent->d_inode;
+	struct inode *inode;
+	struct btrfs_root *root = BTRFS_I(dir)->root;
+	struct btrfs_root *dest = NULL;
+	struct btrfs_ioctl_vol_args *vol_args;
+	struct btrfs_trans_handle *trans;
+	int namelen;
+	int ret;
+	int err = 0;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	vol_args = memdup_user(arg, sizeof(*vol_args));
+	if (IS_ERR(vol_args))
+		return PTR_ERR(vol_args);
+
+	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
+	namelen = strlen(vol_args->name);
+	if (strchr(vol_args->name, '/') ||
+	    strncmp(vol_args->name, "..", namelen) == 0) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = mnt_want_write(file->f_path.mnt);
+	if (err)
+		goto out;
+
+	mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
+	dentry = lookup_one_len(vol_args->name, parent, namelen);
+	if (IS_ERR(dentry)) {
+		err = PTR_ERR(dentry);
+		goto out_unlock_dir;
+	}
+
+	if (!dentry->d_inode) {
+		err = -ENOENT;
+		goto out_dput;
+	}
+
+	inode = dentry->d_inode;
+	if (inode->i_ino != BTRFS_FIRST_FREE_OBJECTID) {
+		err = -EINVAL;
+		goto out_dput;
+	}
+
+	dest = BTRFS_I(inode)->root;
+
+	mutex_lock(&inode->i_mutex);
+	err = d_invalidate(dentry);
+	if (err)
+		goto out_unlock;
+
+	down_write(&root->fs_info->subvol_sem);
+
+	err = may_destroy_subvol(dest);
+	if (err)
+		goto out_up_write;
+
+	trans = btrfs_start_transaction(root, 1);
+	ret = btrfs_unlink_subvol(trans, root, dir,
+				dest->root_key.objectid,
+				dentry->d_name.name,
+				dentry->d_name.len);
+	BUG_ON(ret);
+
+	btrfs_record_root_in_trans(trans, dest);
+
+	memset(&dest->root_item.drop_progress, 0,
+		sizeof(dest->root_item.drop_progress));
+	dest->root_item.drop_level = 0;
+	btrfs_set_root_refs(&dest->root_item, 0);
+
+	ret = btrfs_insert_orphan_item(trans,
+				root->fs_info->tree_root,
+				dest->root_key.objectid);
+	BUG_ON(ret);
+
+	ret = btrfs_commit_transaction(trans, root);
+	BUG_ON(ret);
+	inode->i_flags |= S_DEAD;
+out_up_write:
+	up_write(&root->fs_info->subvol_sem);
+out_unlock:
+	mutex_unlock(&inode->i_mutex);
+	if (!err) {
+		shrink_dcache_sb(root->fs_info->sb);
+		btrfs_invalidate_inodes(dest);
+		d_delete(dentry);
+	}
+out_dput:
+	dput(dentry);
+out_unlock_dir:
+	mutex_unlock(&dir->i_mutex);
+	mnt_drop_write(file->f_path.mnt);
+out:
+	kfree(vol_args);
+	return err;
+}
+
 static int btrfs_ioctl_defrag(struct file *file)
 {
 	struct inode *inode = fdentry(file)->d_inode;
@@ -865,8 +917,8 @@ static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
 	return ret;
 }
 
-static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
-		u64 off, u64 olen, u64 destoff)
+static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
+				       u64 off, u64 olen, u64 destoff)
 {
 	struct inode *inode = fdentry(file)->d_inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -976,7 +1028,7 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 
 	/* punch hole in destination first */
 	btrfs_drop_extents(trans, root, inode, off, off + len,
-			   off + len, 0, &hint_byte);
+			   off + len, 0, &hint_byte, 1);
 
 	/* clone data */
 	key.objectid = src->i_ino;
@@ -1071,9 +1123,10 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 					datao += off - key.offset;
 					datal -= off - key.offset;
 				}
-				if (key.offset + datao + datal + key.offset >
-				    off + len)
-					datal = off + len - key.offset - datao;
+
+				if (key.offset + datal > off + len)
+					datal = off + len - key.offset;
+
 				/* disko == 0 means it's a hole */
 				if (!disko)
 					datao = 0;
@@ -1182,15 +1235,15 @@ static long btrfs_ioctl_trans_start(struct file *file)
 	struct inode *inode = fdentry(file)->d_inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_trans_handle *trans;
-	int ret = 0;
+	int ret;
 
+	ret = -EPERM;
 	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
+		goto out;
 
-	if (file->private_data) {
-		ret = -EINPROGRESS;
+	ret = -EINPROGRESS;
+	if (file->private_data)
 		goto out;
-	}
 
 	ret = mnt_want_write(file->f_path.mnt);
 	if (ret)
@@ -1200,12 +1253,19 @@ static long btrfs_ioctl_trans_start(struct file *file)
 	root->fs_info->open_ioctl_trans++;
 	mutex_unlock(&root->fs_info->trans_mutex);
 
+	ret = -ENOMEM;
 	trans = btrfs_start_ioctl_transaction(root, 0);
-	if (trans)
-		file->private_data = trans;
-	else
-		ret = -ENOMEM;
-	/*printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);*/
+	if (!trans)
+		goto out_drop;
+
+	file->private_data = trans;
+	return 0;
+
+out_drop:
+	mutex_lock(&root->fs_info->trans_mutex);
+	root->fs_info->open_ioctl_trans--;
+	mutex_unlock(&root->fs_info->trans_mutex);
+	mnt_drop_write(file->f_path.mnt);
 out:
 	return ret;
 }
@@ -1221,24 +1281,20 @@ long btrfs_ioctl_trans_end(struct file *file)
 	struct inode *inode = fdentry(file)->d_inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_trans_handle *trans;
-	int ret = 0;
 
 	trans = file->private_data;
-	if (!trans) {
-		ret = -EINVAL;
-		goto out;
-	}
-	btrfs_end_transaction(trans, root);
+	if (!trans)
+		return -EINVAL;
 	file->private_data = NULL;
 
+	btrfs_end_transaction(trans, root);
+
 	mutex_lock(&root->fs_info->trans_mutex);
 	root->fs_info->open_ioctl_trans--;
 	mutex_unlock(&root->fs_info->trans_mutex);
 
 	mnt_drop_write(file->f_path.mnt);
-
-out:
-	return ret;
+	return 0;
 }
 
 long btrfs_ioctl(struct file *file, unsigned int
@@ -1258,6 +1314,8 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_snap_create(file, argp, 0);
 	case BTRFS_IOC_SUBVOL_CREATE:
 		return btrfs_ioctl_snap_create(file, argp, 1);
+	case BTRFS_IOC_SNAP_DESTROY:
+		return btrfs_ioctl_snap_destroy(file, argp);
 	case BTRFS_IOC_DEFRAG:
 		return btrfs_ioctl_defrag(file);
 	case BTRFS_IOC_RESIZE:
diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h
index b320b103fa13..bc49914475eb 100644
--- a/fs/btrfs/ioctl.h
+++ b/fs/btrfs/ioctl.h
@@ -65,5 +65,6 @@ struct btrfs_ioctl_clone_range_args {
 
 #define BTRFS_IOC_SUBVOL_CREATE _IOW(BTRFS_IOCTL_MAGIC, 14, \
 				   struct btrfs_ioctl_vol_args)
-
+#define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \
+				struct btrfs_ioctl_vol_args)
 #endif
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 7b2f401e604e..5799bc46a309 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -159,8 +159,6 @@ static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
  *
  * len is the length of the extent
  *
- * This also sets the EXTENT_ORDERED bit on the range in the inode.
- *
  * The tree is given a single reference on the ordered extent that was
  * inserted.
  */
@@ -181,6 +179,7 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	entry->start = start;
 	entry->len = len;
 	entry->disk_len = disk_len;
+	entry->bytes_left = len;
 	entry->inode = inode;
 	if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE)
 		set_bit(type, &entry->flags);
@@ -195,9 +194,6 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 			   &entry->rb_node);
 	BUG_ON(node);
 
-	set_extent_ordered(&BTRFS_I(inode)->io_tree, file_offset,
-			   entry_end(entry) - 1, GFP_NOFS);
-
 	spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 	list_add_tail(&entry->root_extent_list,
 		      &BTRFS_I(inode)->root->fs_info->ordered_extents);
@@ -241,13 +237,10 @@ int btrfs_dec_test_ordered_pending(struct inode *inode,
 	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	int ret;
 
 	tree = &BTRFS_I(inode)->ordered_tree;
 	mutex_lock(&tree->mutex);
-	clear_extent_ordered(io_tree, file_offset, file_offset + io_size - 1,
-			     GFP_NOFS);
 	node = tree_search(tree, file_offset);
 	if (!node) {
 		ret = 1;
@@ -260,11 +253,16 @@ int btrfs_dec_test_ordered_pending(struct inode *inode,
 		goto out;
 	}
 
-	ret = test_range_bit(io_tree, entry->file_offset,
-			     entry->file_offset + entry->len - 1,
-			     EXTENT_ORDERED, 0);
-	if (ret == 0)
+	if (io_size > entry->bytes_left) {
+		printk(KERN_CRIT "bad ordered accounting left %llu size %llu\n",
+		       (unsigned long long)entry->bytes_left,
+		       (unsigned long long)io_size);
+	}
+	entry->bytes_left -= io_size;
+	if (entry->bytes_left == 0)
 		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+	else
+		ret = 1;
 out:
 	mutex_unlock(&tree->mutex);
 	return ret == 0;
@@ -308,6 +306,12 @@ int btrfs_remove_ordered_extent(struct inode *inode,
 	tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
 
+	spin_lock(&BTRFS_I(inode)->accounting_lock);
+	BTRFS_I(inode)->outstanding_extents--;
+	spin_unlock(&BTRFS_I(inode)->accounting_lock);
+	btrfs_unreserve_metadata_for_delalloc(BTRFS_I(inode)->root,
+					      inode, 1);
+
 	spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 	list_del_init(&entry->root_extent_list);
 
@@ -460,7 +464,7 @@ void btrfs_start_ordered_extent(struct inode *inode,
 	 * start IO on any dirty ones so the wait doesn't stall waiting
 	 * for pdflush to find them
 	 */
-	btrfs_fdatawrite_range(inode->i_mapping, start, end, WB_SYNC_ALL);
+	filemap_fdatawrite_range(inode->i_mapping, start, end);
 	if (wait) {
 		wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE,
 						 &entry->flags));
@@ -476,6 +480,7 @@ int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
 	u64 orig_end;
 	u64 wait_end;
 	struct btrfs_ordered_extent *ordered;
+	int found;
 
 	if (start + len < start) {
 		orig_end = INT_LIMIT(loff_t);
@@ -489,19 +494,18 @@ again:
 	/* start IO across the range first to instantiate any delalloc
 	 * extents
 	 */
-	btrfs_fdatawrite_range(inode->i_mapping, start, orig_end, WB_SYNC_ALL);
+	filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
 
 	/* The compression code will leave pages locked but return from
 	 * writepage without setting the page writeback.  Starting again
 	 * with WB_SYNC_ALL will end up waiting for the IO to actually start.
 	 */
-	btrfs_fdatawrite_range(inode->i_mapping, start, orig_end, WB_SYNC_ALL);
+	filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
 
-	btrfs_wait_on_page_writeback_range(inode->i_mapping,
-					   start >> PAGE_CACHE_SHIFT,
-					   orig_end >> PAGE_CACHE_SHIFT);
+	filemap_fdatawait_range(inode->i_mapping, start, orig_end);
 
 	end = orig_end;
+	found = 0;
 	while (1) {
 		ordered = btrfs_lookup_first_ordered_extent(inode, end);
 		if (!ordered)
@@ -514,6 +518,7 @@ again:
 			btrfs_put_ordered_extent(ordered);
 			break;
 		}
+		found++;
 		btrfs_start_ordered_extent(inode, ordered, 1);
 		end = ordered->file_offset;
 		btrfs_put_ordered_extent(ordered);
@@ -521,8 +526,8 @@ again:
 			break;
 		end--;
 	}
-	if (test_range_bit(&BTRFS_I(inode)->io_tree, start, orig_end,
-			   EXTENT_ORDERED | EXTENT_DELALLOC, 0)) {
+	if (found || test_range_bit(&BTRFS_I(inode)->io_tree, start, orig_end,
+			   EXTENT_DELALLOC, 0, NULL)) {
 		schedule_timeout(1);
 		goto again;
 	}
@@ -613,7 +618,7 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 */
 	if (test_range_bit(io_tree, disk_i_size,
 			   ordered->file_offset + ordered->len - 1,
-			   EXTENT_DELALLOC, 0)) {
+			   EXTENT_DELALLOC, 0, NULL)) {
 		goto out;
 	}
 	/*
@@ -664,7 +669,7 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 */
 	if (i_size_test > entry_end(ordered) &&
 	    !test_range_bit(io_tree, entry_end(ordered), i_size_test - 1,
-			   EXTENT_DELALLOC, 0)) {
+			   EXTENT_DELALLOC, 0, NULL)) {
 		new_i_size = min_t(u64, i_size_test, i_size_read(inode));
 	}
 	BTRFS_I(inode)->disk_i_size = new_i_size;
@@ -715,89 +720,6 @@ out:
 }
 
 
-/**
- * taken from mm/filemap.c because it isn't exported
- *
- * __filemap_fdatawrite_range - start writeback on mapping dirty pages in range
- * @mapping:	address space structure to write
- * @start:	offset in bytes where the range starts
- * @end:	offset in bytes where the range ends (inclusive)
- * @sync_mode:	enable synchronous operation
- *
- * Start writeback against all of a mapping's dirty pages that lie
- * within the byte offsets <start, end> inclusive.
- *
- * If sync_mode is WB_SYNC_ALL then this is a "data integrity" operation, as
- * opposed to a regular memory cleansing writeback.  The difference between
- * these two operations is that if a dirty page/buffer is encountered, it must
- * be waited upon, and not just skipped over.
- */
-int btrfs_fdatawrite_range(struct address_space *mapping, loff_t start,
-			   loff_t end, int sync_mode)
-{
-	struct writeback_control wbc = {
-		.sync_mode = sync_mode,
-		.nr_to_write = mapping->nrpages * 2,
-		.range_start = start,
-		.range_end = end,
-	};
-	return btrfs_writepages(mapping, &wbc);
-}
-
-/**
- * taken from mm/filemap.c because it isn't exported
- *
- * wait_on_page_writeback_range - wait for writeback to complete
- * @mapping:	target address_space
- * @start:	beginning page index
- * @end:	ending page index
- *
- * Wait for writeback to complete against pages indexed by start->end
- * inclusive
- */
-int btrfs_wait_on_page_writeback_range(struct address_space *mapping,
-				       pgoff_t start, pgoff_t end)
-{
-	struct pagevec pvec;
-	int nr_pages;
-	int ret = 0;
-	pgoff_t index;
-
-	if (end < start)
-		return 0;
-
-	pagevec_init(&pvec, 0);
-	index = start;
-	while ((index <= end) &&
-			(nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
-			PAGECACHE_TAG_WRITEBACK,
-			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1)) != 0) {
-		unsigned i;
-
-		for (i = 0; i < nr_pages; i++) {
-			struct page *page = pvec.pages[i];
-
-			/* until radix tree lookup accepts end_index */
-			if (page->index > end)
-				continue;
-
-			wait_on_page_writeback(page);
-			if (PageError(page))
-				ret = -EIO;
-		}
-		pagevec_release(&pvec);
-		cond_resched();
-	}
-
-	/* Check for outstanding write errors */
-	if (test_and_clear_bit(AS_ENOSPC, &mapping->flags))
-		ret = -ENOSPC;
-	if (test_and_clear_bit(AS_EIO, &mapping->flags))
-		ret = -EIO;
-
-	return ret;
-}
-
 /*
  * add a given inode to the list of inodes that must be fully on
  * disk before a transaction commit finishes.
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index 3d31c8827b01..f82e87488ca8 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -85,6 +85,9 @@ struct btrfs_ordered_extent {
 	/* extent length on disk */
 	u64 disk_len;
 
+	/* number of bytes that still need writing */
+	u64 bytes_left;
+
 	/* flags (described above) */
 	unsigned long flags;
 
@@ -150,10 +153,6 @@ btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
 int btrfs_ordered_update_i_size(struct inode *inode,
 				struct btrfs_ordered_extent *ordered);
 int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
-int btrfs_wait_on_page_writeback_range(struct address_space *mapping,
-				       pgoff_t start, pgoff_t end);
-int btrfs_fdatawrite_range(struct address_space *mapping, loff_t start,
-			   loff_t end, int sync_mode);
 int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only);
 int btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
 int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
diff --git a/fs/btrfs/orphan.c b/fs/btrfs/orphan.c
index 3c0d52af4f80..79cba5fbc28e 100644
--- a/fs/btrfs/orphan.c
+++ b/fs/btrfs/orphan.c
@@ -65,3 +65,23 @@ out:
 	btrfs_free_path(path);
 	return ret;
 }
+
+int btrfs_find_orphan_item(struct btrfs_root *root, u64 offset)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int ret;
+
+	key.objectid = BTRFS_ORPHAN_OBJECTID;
+	key.type = BTRFS_ORPHAN_ITEM_KEY;
+	key.offset = offset;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+
+	btrfs_free_path(path);
+	return ret;
+}
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index c04f7f212602..cfcc93c93a7b 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -121,6 +121,15 @@ struct inodevec {
 	int nr;
 };
 
+#define MAX_EXTENTS 128
+
+struct file_extent_cluster {
+	u64 start;
+	u64 end;
+	u64 boundary[MAX_EXTENTS];
+	unsigned int nr;
+};
+
 struct reloc_control {
 	/* block group to relocate */
 	struct btrfs_block_group_cache *block_group;
@@ -2180,7 +2189,7 @@ static int tree_block_processed(u64 bytenr, u32 blocksize,
 				struct reloc_control *rc)
 {
 	if (test_range_bit(&rc->processed_blocks, bytenr,
-			   bytenr + blocksize - 1, EXTENT_DIRTY, 1))
+			   bytenr + blocksize - 1, EXTENT_DIRTY, 1, NULL))
 		return 1;
 	return 0;
 }
@@ -2529,56 +2538,94 @@ out:
 }
 
 static noinline_for_stack
-int relocate_inode_pages(struct inode *inode, u64 start, u64 len)
+int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
+			 u64 block_start)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map *em;
+	int ret = 0;
+
+	em = alloc_extent_map(GFP_NOFS);
+	if (!em)
+		return -ENOMEM;
+
+	em->start = start;
+	em->len = end + 1 - start;
+	em->block_len = em->len;
+	em->block_start = block_start;
+	em->bdev = root->fs_info->fs_devices->latest_bdev;
+	set_bit(EXTENT_FLAG_PINNED, &em->flags);
+
+	lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+	while (1) {
+		write_lock(&em_tree->lock);
+		ret = add_extent_mapping(em_tree, em);
+		write_unlock(&em_tree->lock);
+		if (ret != -EEXIST) {
+			free_extent_map(em);
+			break;
+		}
+		btrfs_drop_extent_cache(inode, start, end, 0);
+	}
+	unlock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+	return ret;
+}
+
+static int relocate_file_extent_cluster(struct inode *inode,
+					struct file_extent_cluster *cluster)
 {
 	u64 page_start;
 	u64 page_end;
-	unsigned long i;
-	unsigned long first_index;
+	u64 offset = BTRFS_I(inode)->index_cnt;
+	unsigned long index;
 	unsigned long last_index;
-	unsigned int total_read = 0;
-	unsigned int total_dirty = 0;
+	unsigned int dirty_page = 0;
 	struct page *page;
 	struct file_ra_state *ra;
-	struct btrfs_ordered_extent *ordered;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	int nr = 0;
 	int ret = 0;
 
+	if (!cluster->nr)
+		return 0;
+
 	ra = kzalloc(sizeof(*ra), GFP_NOFS);
 	if (!ra)
 		return -ENOMEM;
 
+	index = (cluster->start - offset) >> PAGE_CACHE_SHIFT;
+	last_index = (cluster->end - offset) >> PAGE_CACHE_SHIFT;
+
 	mutex_lock(&inode->i_mutex);
-	first_index = start >> PAGE_CACHE_SHIFT;
-	last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
 
-	/* make sure the dirty trick played by the caller work */
-	while (1) {
-		ret = invalidate_inode_pages2_range(inode->i_mapping,
-						    first_index, last_index);
-		if (ret != -EBUSY)
-			break;
-		schedule_timeout(HZ/10);
-	}
+	i_size_write(inode, cluster->end + 1 - offset);
+	ret = setup_extent_mapping(inode, cluster->start - offset,
+				   cluster->end - offset, cluster->start);
 	if (ret)
 		goto out_unlock;
 
 	file_ra_state_init(ra, inode->i_mapping);
 
-	for (i = first_index ; i <= last_index; i++) {
-		if (total_read % ra->ra_pages == 0) {
-			btrfs_force_ra(inode->i_mapping, ra, NULL, i,
-				min(last_index, ra->ra_pages + i - 1));
-		}
-		total_read++;
-again:
-		if (((u64)i << PAGE_CACHE_SHIFT) > i_size_read(inode))
-			BUG_ON(1);
-		page = grab_cache_page(inode->i_mapping, i);
+	WARN_ON(cluster->start != cluster->boundary[0]);
+	while (index <= last_index) {
+		page = find_lock_page(inode->i_mapping, index);
 		if (!page) {
-			ret = -ENOMEM;
-			goto out_unlock;
+			page_cache_sync_readahead(inode->i_mapping,
+						  ra, NULL, index,
+						  last_index + 1 - index);
+			page = grab_cache_page(inode->i_mapping, index);
+			if (!page) {
+				ret = -ENOMEM;
+				goto out_unlock;
+			}
+		}
+
+		if (PageReadahead(page)) {
+			page_cache_async_readahead(inode->i_mapping,
+						   ra, NULL, page, index,
+						   last_index + 1 - index);
 		}
+
 		if (!PageUptodate(page)) {
 			btrfs_readpage(NULL, page);
 			lock_page(page);
@@ -2589,75 +2636,79 @@ again:
 				goto out_unlock;
 			}
 		}
-		wait_on_page_writeback(page);
 
 		page_start = (u64)page->index << PAGE_CACHE_SHIFT;
 		page_end = page_start + PAGE_CACHE_SIZE - 1;
-		lock_extent(io_tree, page_start, page_end, GFP_NOFS);
-
-		ordered = btrfs_lookup_ordered_extent(inode, page_start);
-		if (ordered) {
-			unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
-			unlock_page(page);
-			page_cache_release(page);
-			btrfs_start_ordered_extent(inode, ordered, 1);
-			btrfs_put_ordered_extent(ordered);
-			goto again;
-		}
+
+		lock_extent(&BTRFS_I(inode)->io_tree,
+			    page_start, page_end, GFP_NOFS);
+
 		set_page_extent_mapped(page);
 
-		if (i == first_index)
-			set_extent_bits(io_tree, page_start, page_end,
+		if (nr < cluster->nr &&
+		    page_start + offset == cluster->boundary[nr]) {
+			set_extent_bits(&BTRFS_I(inode)->io_tree,
+					page_start, page_end,
 					EXTENT_BOUNDARY, GFP_NOFS);
+			nr++;
+		}
 		btrfs_set_extent_delalloc(inode, page_start, page_end);
 
 		set_page_dirty(page);
-		total_dirty++;
+		dirty_page++;
 
-		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+		unlock_extent(&BTRFS_I(inode)->io_tree,
+			      page_start, page_end, GFP_NOFS);
 		unlock_page(page);
 		page_cache_release(page);
+
+		index++;
+		if (nr < cluster->nr &&
+		    page_end + 1 + offset == cluster->boundary[nr]) {
+			balance_dirty_pages_ratelimited_nr(inode->i_mapping,
+							   dirty_page);
+			dirty_page = 0;
+		}
+	}
+	if (dirty_page) {
+		balance_dirty_pages_ratelimited_nr(inode->i_mapping,
+						   dirty_page);
 	}
+	WARN_ON(nr != cluster->nr);
 out_unlock:
 	mutex_unlock(&inode->i_mutex);
 	kfree(ra);
-	balance_dirty_pages_ratelimited_nr(inode->i_mapping, total_dirty);
 	return ret;
 }
 
 static noinline_for_stack
-int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key)
+int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key,
+			 struct file_extent_cluster *cluster)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
-	struct extent_map *em;
-	u64 start = extent_key->objectid - BTRFS_I(inode)->index_cnt;
-	u64 end = start + extent_key->offset - 1;
-
-	em = alloc_extent_map(GFP_NOFS);
-	em->start = start;
-	em->len = extent_key->offset;
-	em->block_len = extent_key->offset;
-	em->block_start = extent_key->objectid;
-	em->bdev = root->fs_info->fs_devices->latest_bdev;
-	set_bit(EXTENT_FLAG_PINNED, &em->flags);
+	int ret;
 
-	/* setup extent map to cheat btrfs_readpage */
-	lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
-	while (1) {
-		int ret;
-		spin_lock(&em_tree->lock);
-		ret = add_extent_mapping(em_tree, em);
-		spin_unlock(&em_tree->lock);
-		if (ret != -EEXIST) {
-			free_extent_map(em);
-			break;
-		}
-		btrfs_drop_extent_cache(inode, start, end, 0);
+	if (cluster->nr > 0 && extent_key->objectid != cluster->end + 1) {
+		ret = relocate_file_extent_cluster(inode, cluster);
+		if (ret)
+			return ret;
+		cluster->nr = 0;
 	}
-	unlock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
 
-	return relocate_inode_pages(inode, start, extent_key->offset);
+	if (!cluster->nr)
+		cluster->start = extent_key->objectid;
+	else
+		BUG_ON(cluster->nr >= MAX_EXTENTS);
+	cluster->end = extent_key->objectid + extent_key->offset - 1;
+	cluster->boundary[cluster->nr] = extent_key->objectid;
+	cluster->nr++;
+
+	if (cluster->nr >= MAX_EXTENTS) {
+		ret = relocate_file_extent_cluster(inode, cluster);
+		if (ret)
+			return ret;
+		cluster->nr = 0;
+	}
+	return 0;
 }
 
 #ifdef BTRFS_COMPAT_EXTENT_TREE_V0
@@ -3203,10 +3254,12 @@ static int check_extent_flags(u64 flags)
 	return 0;
 }
 
+
 static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 {
 	struct rb_root blocks = RB_ROOT;
 	struct btrfs_key key;
+	struct file_extent_cluster *cluster;
 	struct btrfs_trans_handle *trans = NULL;
 	struct btrfs_path *path;
 	struct btrfs_extent_item *ei;
@@ -3216,10 +3269,17 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 	int ret;
 	int err = 0;
 
+	cluster = kzalloc(sizeof(*cluster), GFP_NOFS);
+	if (!cluster)
+		return -ENOMEM;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
+	rc->extents_found = 0;
+	rc->extents_skipped = 0;
+
 	rc->search_start = rc->block_group->key.objectid;
 	clear_extent_bits(&rc->processed_blocks, 0, (u64)-1, EXTENT_DIRTY,
 			  GFP_NOFS);
@@ -3306,14 +3366,15 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 		}
 
 		nr = trans->blocks_used;
-		btrfs_end_transaction_throttle(trans, rc->extent_root);
+		btrfs_end_transaction(trans, rc->extent_root);
 		trans = NULL;
 		btrfs_btree_balance_dirty(rc->extent_root, nr);
 
 		if (rc->stage == MOVE_DATA_EXTENTS &&
 		    (flags & BTRFS_EXTENT_FLAG_DATA)) {
 			rc->found_file_extent = 1;
-			ret = relocate_data_extent(rc->data_inode, &key);
+			ret = relocate_data_extent(rc->data_inode,
+						   &key, cluster);
 			if (ret < 0) {
 				err = ret;
 				break;
@@ -3328,6 +3389,14 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 		btrfs_btree_balance_dirty(rc->extent_root, nr);
 	}
 
+	if (!err) {
+		ret = relocate_file_extent_cluster(rc->data_inode, cluster);
+		if (ret < 0)
+			err = ret;
+	}
+
+	kfree(cluster);
+
 	rc->create_reloc_root = 0;
 	smp_mb();
 
@@ -3348,8 +3417,7 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 }
 
 static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root,
-				 u64 objectid, u64 size)
+				 struct btrfs_root *root, u64 objectid)
 {
 	struct btrfs_path *path;
 	struct btrfs_inode_item *item;
@@ -3368,7 +3436,7 @@ static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
 	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
 	memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
 	btrfs_set_inode_generation(leaf, item, 1);
-	btrfs_set_inode_size(leaf, item, size);
+	btrfs_set_inode_size(leaf, item, 0);
 	btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
 	btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NOCOMPRESS);
 	btrfs_mark_buffer_dirty(leaf);
@@ -3404,12 +3472,7 @@ static struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
 	if (err)
 		goto out;
 
-	err = __insert_orphan_inode(trans, root, objectid, group->key.offset);
-	BUG_ON(err);
-
-	err = btrfs_insert_file_extent(trans, root, objectid, 0, 0, 0,
-				       group->key.offset, 0, group->key.offset,
-				       0, 0, 0);
+	err = __insert_orphan_inode(trans, root, objectid);
 	BUG_ON(err);
 
 	key.objectid = objectid;
@@ -3455,7 +3518,7 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	BUG_ON(!rc->block_group);
 
 	btrfs_init_workers(&rc->workers, "relocate",
-			   fs_info->thread_pool_size);
+			   fs_info->thread_pool_size, NULL);
 
 	rc->extent_root = extent_root;
 	btrfs_prepare_block_group_relocation(extent_root, rc->block_group);
@@ -3475,14 +3538,15 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	btrfs_wait_ordered_extents(fs_info->tree_root, 0);
 
 	while (1) {
-		mutex_lock(&fs_info->cleaner_mutex);
-		btrfs_clean_old_snapshots(fs_info->tree_root);
-		mutex_unlock(&fs_info->cleaner_mutex);
-
 		rc->extents_found = 0;
 		rc->extents_skipped = 0;
 
+		mutex_lock(&fs_info->cleaner_mutex);
+
+		btrfs_clean_old_snapshots(fs_info->tree_root);
 		ret = relocate_block_group(rc);
+
+		mutex_unlock(&fs_info->cleaner_mutex);
 		if (ret < 0) {
 			err = ret;
 			break;
@@ -3514,10 +3578,10 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 		}
 	}
 
-	filemap_fdatawrite_range(fs_info->btree_inode->i_mapping,
-				 rc->block_group->key.objectid,
-				 rc->block_group->key.objectid +
-				 rc->block_group->key.offset - 1);
+	filemap_write_and_wait_range(fs_info->btree_inode->i_mapping,
+				     rc->block_group->key.objectid,
+				     rc->block_group->key.objectid +
+				     rc->block_group->key.offset - 1);
 
 	WARN_ON(rc->block_group->pinned > 0);
 	WARN_ON(rc->block_group->reserved > 0);
@@ -3530,6 +3594,26 @@ out:
 	return err;
 }
 
+static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
+{
+	struct btrfs_trans_handle *trans;
+	int ret;
+
+	trans = btrfs_start_transaction(root->fs_info->tree_root, 1);
+
+	memset(&root->root_item.drop_progress, 0,
+		sizeof(root->root_item.drop_progress));
+	root->root_item.drop_level = 0;
+	btrfs_set_root_refs(&root->root_item, 0);
+	ret = btrfs_update_root(trans, root->fs_info->tree_root,
+				&root->root_key, &root->root_item);
+	BUG_ON(ret);
+
+	ret = btrfs_end_transaction(trans, root->fs_info->tree_root);
+	BUG_ON(ret);
+	return 0;
+}
+
 /*
  * recover relocation interrupted by system crash.
  *
@@ -3589,8 +3673,12 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 			fs_root = read_fs_root(root->fs_info,
 					       reloc_root->root_key.offset);
 			if (IS_ERR(fs_root)) {
-				err = PTR_ERR(fs_root);
-				goto out;
+				ret = PTR_ERR(fs_root);
+				if (ret != -ENOENT) {
+					err = ret;
+					goto out;
+				}
+				mark_garbage_root(reloc_root);
 			}
 		}
 
@@ -3613,7 +3701,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 	mapping_tree_init(&rc->reloc_root_tree);
 	INIT_LIST_HEAD(&rc->reloc_roots);
 	btrfs_init_workers(&rc->workers, "relocate",
-			   root->fs_info->thread_pool_size);
+			   root->fs_info->thread_pool_size, NULL);
 	rc->extent_root = root->fs_info->extent_root;
 
 	set_reloc_control(rc);
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 0ddc6d61c55a..9351428f30e2 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -94,17 +94,23 @@ int btrfs_find_last_root(struct btrfs_root *root, u64 objectid,
 		goto out;
 
 	BUG_ON(ret == 0);
+	if (path->slots[0] == 0) {
+		ret = 1;
+		goto out;
+	}
 	l = path->nodes[0];
-	BUG_ON(path->slots[0] == 0);
 	slot = path->slots[0] - 1;
 	btrfs_item_key_to_cpu(l, &found_key, slot);
-	if (found_key.objectid != objectid) {
+	if (found_key.objectid != objectid ||
+	    found_key.type != BTRFS_ROOT_ITEM_KEY) {
 		ret = 1;
 		goto out;
 	}
-	read_extent_buffer(l, item, btrfs_item_ptr_offset(l, slot),
-			   sizeof(*item));
-	memcpy(key, &found_key, sizeof(found_key));
+	if (item)
+		read_extent_buffer(l, item, btrfs_item_ptr_offset(l, slot),
+				   sizeof(*item));
+	if (key)
+		memcpy(key, &found_key, sizeof(found_key));
 	ret = 0;
 out:
 	btrfs_free_path(path);
@@ -249,6 +255,59 @@ err:
 	return ret;
 }
 
+int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
+{
+	struct extent_buffer *leaf;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int err = 0;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	key.objectid = BTRFS_ORPHAN_OBJECTID;
+	key.type = BTRFS_ORPHAN_ITEM_KEY;
+	key.offset = 0;
+
+	while (1) {
+		ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
+		if (ret < 0) {
+			err = ret;
+			break;
+		}
+
+		leaf = path->nodes[0];
+		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
+			ret = btrfs_next_leaf(tree_root, path);
+			if (ret < 0)
+				err = ret;
+			if (ret != 0)
+				break;
+			leaf = path->nodes[0];
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		btrfs_release_path(tree_root, path);
+
+		if (key.objectid != BTRFS_ORPHAN_OBJECTID ||
+		    key.type != BTRFS_ORPHAN_ITEM_KEY)
+			break;
+
+		ret = btrfs_find_dead_roots(tree_root, key.offset);
+		if (ret) {
+			err = ret;
+			break;
+		}
+
+		key.offset++;
+	}
+
+	btrfs_free_path(path);
+	return err;
+}
+
 /* drop the root item for 'key' from 'root' */
 int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		   struct btrfs_key *key)
@@ -278,31 +337,57 @@ out:
 	return ret;
 }
 
-#if 0 /* this will get used when snapshot deletion is implemented */
 int btrfs_del_root_ref(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *tree_root,
-		       u64 root_id, u8 type, u64 ref_id)
+		       u64 root_id, u64 ref_id, u64 dirid, u64 *sequence,
+		       const char *name, int name_len)
+
 {
+	struct btrfs_path *path;
+	struct btrfs_root_ref *ref;
+	struct extent_buffer *leaf;
 	struct btrfs_key key;
+	unsigned long ptr;
+	int err = 0;
 	int ret;
-	struct btrfs_path *path;
 
 	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
 
 	key.objectid = root_id;
-	key.type = type;
+	key.type = BTRFS_ROOT_BACKREF_KEY;
 	key.offset = ref_id;
-
+again:
 	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
-	BUG_ON(ret);
-
-	ret = btrfs_del_item(trans, tree_root, path);
-	BUG_ON(ret);
+	BUG_ON(ret < 0);
+	if (ret == 0) {
+		leaf = path->nodes[0];
+		ref = btrfs_item_ptr(leaf, path->slots[0],
+				     struct btrfs_root_ref);
+
+		WARN_ON(btrfs_root_ref_dirid(leaf, ref) != dirid);
+		WARN_ON(btrfs_root_ref_name_len(leaf, ref) != name_len);
+		ptr = (unsigned long)(ref + 1);
+		WARN_ON(memcmp_extent_buffer(leaf, name, ptr, name_len));
+		*sequence = btrfs_root_ref_sequence(leaf, ref);
+
+		ret = btrfs_del_item(trans, tree_root, path);
+		BUG_ON(ret);
+	} else
+		err = -ENOENT;
+
+	if (key.type == BTRFS_ROOT_BACKREF_KEY) {
+		btrfs_release_path(tree_root, path);
+		key.objectid = ref_id;
+		key.type = BTRFS_ROOT_REF_KEY;
+		key.offset = root_id;
+		goto again;
+	}
 
 	btrfs_free_path(path);
-	return ret;
+	return err;
 }
-#endif
 
 int btrfs_find_root_ref(struct btrfs_root *tree_root,
 		   struct btrfs_path *path,
@@ -319,7 +404,6 @@ int btrfs_find_root_ref(struct btrfs_root *tree_root,
 	return ret;
 }
 
-
 /*
  * add a btrfs_root_ref item.  type is either BTRFS_ROOT_REF_KEY
  * or BTRFS_ROOT_BACKREF_KEY.
@@ -335,8 +419,7 @@ int btrfs_find_root_ref(struct btrfs_root *tree_root,
  */
 int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *tree_root,
-		       u64 root_id, u8 type, u64 ref_id,
-		       u64 dirid, u64 sequence,
+		       u64 root_id, u64 ref_id, u64 dirid, u64 sequence,
 		       const char *name, int name_len)
 {
 	struct btrfs_key key;
@@ -346,13 +429,14 @@ int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	unsigned long ptr;
 
-
 	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
 
 	key.objectid = root_id;
-	key.type = type;
+	key.type = BTRFS_ROOT_BACKREF_KEY;
 	key.offset = ref_id;
-
+again:
 	ret = btrfs_insert_empty_item(trans, tree_root, path, &key,
 				      sizeof(*ref) + name_len);
 	BUG_ON(ret);
@@ -366,6 +450,14 @@ int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 	write_extent_buffer(leaf, name, ptr, name_len);
 	btrfs_mark_buffer_dirty(leaf);
 
+	if (key.type == BTRFS_ROOT_BACKREF_KEY) {
+		btrfs_release_path(tree_root, path);
+		key.objectid = ref_id;
+		key.type = BTRFS_ROOT_REF_KEY;
+		key.offset = root_id;
+		goto again;
+	}
+
 	btrfs_free_path(path);
-	return ret;
+	return 0;
 }
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 6d6d06cb6dfc..752a5463bf53 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -51,7 +51,7 @@
 #include "export.h"
 #include "compression.h"
 
-static struct super_operations btrfs_super_ops;
+static const struct super_operations btrfs_super_ops;
 
 static void btrfs_put_super(struct super_block *sb)
 {
@@ -66,7 +66,8 @@ enum {
 	Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
 	Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
 	Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
-	Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_err,
+	Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
+	Opt_discard, Opt_err,
 };
 
 static match_table_t tokens = {
@@ -88,6 +89,7 @@ static match_table_t tokens = {
 	{Opt_notreelog, "notreelog"},
 	{Opt_flushoncommit, "flushoncommit"},
 	{Opt_ratio, "metadata_ratio=%d"},
+	{Opt_discard, "discard"},
 	{Opt_err, NULL},
 };
 
@@ -257,6 +259,9 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 				       info->metadata_ratio);
 			}
 			break;
+		case Opt_discard:
+			btrfs_set_opt(info->mount_opt, DISCARD);
+			break;
 		default:
 			break;
 		}
@@ -344,7 +349,9 @@ static int btrfs_fill_super(struct super_block *sb,
 	sb->s_export_op = &btrfs_export_ops;
 	sb->s_xattr = btrfs_xattr_handlers;
 	sb->s_time_gran = 1;
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 	sb->s_flags |= MS_POSIXACL;
+#endif
 
 	tree_root = open_ctree(sb, fs_devices, (char *)data);
 
@@ -675,7 +682,8 @@ static int btrfs_unfreeze(struct super_block *sb)
 	return 0;
 }
 
-static struct super_operations btrfs_super_ops = {
+static const struct super_operations btrfs_super_ops = {
+	.drop_inode	= btrfs_drop_inode,
 	.delete_inode	= btrfs_delete_inode,
 	.put_super	= btrfs_put_super,
 	.sync_fs	= btrfs_sync_fs,
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index cdbb5022da52..bca82a4ca8e6 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -104,7 +104,6 @@ static noinline int record_root_in_trans(struct btrfs_trans_handle *trans,
 {
 	if (root->ref_cows && root->last_trans < trans->transid) {
 		WARN_ON(root == root->fs_info->extent_root);
-		WARN_ON(root->root_item.refs == 0);
 		WARN_ON(root->commit_root != root->node);
 
 		radix_tree_tag_set(&root->fs_info->fs_roots_radix,
@@ -187,6 +186,9 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 	h->alloc_exclude_start = 0;
 	h->delayed_ref_updates = 0;
 
+	if (!current->journal_info)
+		current->journal_info = h;
+
 	root->fs_info->running_transaction->use_count++;
 	record_root_in_trans(h, root);
 	mutex_unlock(&root->fs_info->trans_mutex);
@@ -318,6 +320,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 		wake_up(&cur_trans->writer_wait);
 	put_transaction(cur_trans);
 	mutex_unlock(&info->trans_mutex);
+
+	if (current->journal_info == trans)
+		current->journal_info = NULL;
 	memset(trans, 0, sizeof(*trans));
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
@@ -339,10 +344,10 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
 /*
  * when btree blocks are allocated, they have some corresponding bits set for
  * them in one of two extent_io trees.  This is used to make sure all of
- * those extents are on disk for transaction or log commit
+ * those extents are sent to disk but does not wait on them
  */
-int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
-					struct extent_io_tree *dirty_pages)
+int btrfs_write_marked_extents(struct btrfs_root *root,
+			       struct extent_io_tree *dirty_pages)
 {
 	int ret;
 	int err = 0;
@@ -389,6 +394,29 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
 			page_cache_release(page);
 		}
 	}
+	if (err)
+		werr = err;
+	return werr;
+}
+
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees.  This is used to make sure all of
+ * those extents are on disk for transaction or log commit.  We wait
+ * on all the pages and clear them from the dirty pages state tree
+ */
+int btrfs_wait_marked_extents(struct btrfs_root *root,
+			      struct extent_io_tree *dirty_pages)
+{
+	int ret;
+	int err = 0;
+	int werr = 0;
+	struct page *page;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	u64 start = 0;
+	u64 end;
+	unsigned long index;
+
 	while (1) {
 		ret = find_first_extent_bit(dirty_pages, 0, &start, &end,
 					    EXTENT_DIRTY);
@@ -419,6 +447,22 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
 	return werr;
 }
 
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees.  This is used to make sure all of
+ * those extents are on disk for transaction or log commit
+ */
+int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
+					struct extent_io_tree *dirty_pages)
+{
+	int ret;
+	int ret2;
+
+	ret = btrfs_write_marked_extents(root, dirty_pages);
+	ret2 = btrfs_wait_marked_extents(root, dirty_pages);
+	return ret || ret2;
+}
+
 int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
 				     struct btrfs_root *root)
 {
@@ -720,7 +764,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
 
 	key.objectid = objectid;
-	key.offset = 0;
+	/* record when the snapshot was created in key.offset */
+	key.offset = trans->transid;
 	btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
 
 	old = btrfs_lock_root_node(root);
@@ -743,6 +788,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	memcpy(&pending->root_key, &key, sizeof(key));
 fail:
 	kfree(new_root_item);
+	btrfs_unreserve_metadata_space(root, 6);
 	return ret;
 }
 
@@ -778,24 +824,14 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
 	ret = btrfs_update_inode(trans, parent_root, parent_inode);
 	BUG_ON(ret);
 
-	/* add the backref first */
 	ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
 				 pending->root_key.objectid,
-				 BTRFS_ROOT_BACKREF_KEY,
 				 parent_root->root_key.objectid,
 				 parent_inode->i_ino, index, pending->name,
 				 namelen);
 
 	BUG_ON(ret);
 
-	/* now add the forward ref */
-	ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
-				 parent_root->root_key.objectid,
-				 BTRFS_ROOT_REF_KEY,
-				 pending->root_key.objectid,
-				 parent_inode->i_ino, index, pending->name,
-				 namelen);
-
 	inode = btrfs_lookup_dentry(parent_inode, pending->dentry);
 	d_instantiate(pending->dentry, inode);
 fail:
@@ -874,7 +910,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	unsigned long timeout = 1;
 	struct btrfs_transaction *cur_trans;
 	struct btrfs_transaction *prev_trans = NULL;
-	struct extent_io_tree *pinned_copy;
 	DEFINE_WAIT(wait);
 	int ret;
 	int should_grow = 0;
@@ -915,13 +950,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		return 0;
 	}
 
-	pinned_copy = kmalloc(sizeof(*pinned_copy), GFP_NOFS);
-	if (!pinned_copy)
-		return -ENOMEM;
-
-	extent_io_tree_init(pinned_copy,
-			     root->fs_info->btree_inode->i_mapping, GFP_NOFS);
-
 	trans->transaction->in_commit = 1;
 	trans->transaction->blocked = 1;
 	if (cur_trans->list.prev != &root->fs_info->trans_list) {
@@ -1019,6 +1047,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	ret = commit_cowonly_roots(trans, root);
 	BUG_ON(ret);
 
+	btrfs_prepare_extent_commit(trans, root);
+
 	cur_trans = root->fs_info->running_transaction;
 	spin_lock(&root->fs_info->new_trans_lock);
 	root->fs_info->running_transaction = NULL;
@@ -1042,8 +1072,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	memcpy(&root->fs_info->super_for_commit, &root->fs_info->super_copy,
 	       sizeof(root->fs_info->super_copy));
 
-	btrfs_copy_pinned(root, pinned_copy);
-
 	trans->transaction->blocked = 0;
 
 	wake_up(&root->fs_info->transaction_wait);
@@ -1059,8 +1087,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	 */
 	mutex_unlock(&root->fs_info->tree_log_mutex);
 
-	btrfs_finish_extent_commit(trans, root, pinned_copy);
-	kfree(pinned_copy);
+	btrfs_finish_extent_commit(trans, root);
 
 	/* do the directory inserts of any pending snapshot creations */
 	finish_pending_snapshots(trans, root->fs_info);
@@ -1078,6 +1105,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	mutex_unlock(&root->fs_info->trans_mutex);
 
+	if (current->journal_info == trans)
+		current->journal_info = NULL;
+
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 	return ret;
 }
@@ -1096,8 +1126,13 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
 
 	while (!list_empty(&list)) {
 		root = list_entry(list.next, struct btrfs_root, root_list);
-		list_del_init(&root->root_list);
-		btrfs_drop_snapshot(root, 0);
+		list_del(&root->root_list);
+
+		if (btrfs_header_backref_rev(root->node) <
+		    BTRFS_MIXED_BACKREF_REV)
+			btrfs_drop_snapshot(root, 0);
+		else
+			btrfs_drop_snapshot(root, 1);
 	}
 	return 0;
 }
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 663c67404918..d4e3e7a6938c 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -79,6 +79,7 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
 					      struct inode *inode)
 {
 	BTRFS_I(inode)->last_trans = trans->transaction->transid;
+	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
 }
 
 int btrfs_end_transaction(struct btrfs_trans_handle *trans,
@@ -107,5 +108,9 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root);
 int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
 					struct extent_io_tree *dirty_pages);
+int btrfs_write_marked_extents(struct btrfs_root *root,
+					struct extent_io_tree *dirty_pages);
+int btrfs_wait_marked_extents(struct btrfs_root *root,
+					struct extent_io_tree *dirty_pages);
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info);
 #endif
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index d91b0de7c502..741666a7676a 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -137,11 +137,20 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 
 	mutex_lock(&root->log_mutex);
 	if (root->log_root) {
+		if (!root->log_start_pid) {
+			root->log_start_pid = current->pid;
+			root->log_multiple_pids = false;
+		} else if (root->log_start_pid != current->pid) {
+			root->log_multiple_pids = true;
+		}
+
 		root->log_batch++;
 		atomic_inc(&root->log_writers);
 		mutex_unlock(&root->log_mutex);
 		return 0;
 	}
+	root->log_multiple_pids = false;
+	root->log_start_pid = current->pid;
 	mutex_lock(&root->fs_info->tree_log_mutex);
 	if (!root->fs_info->log_root_tree) {
 		ret = btrfs_init_log_root_tree(trans, root->fs_info);
@@ -263,8 +272,8 @@ static int process_one_buffer(struct btrfs_root *log,
 			      struct walk_control *wc, u64 gen)
 {
 	if (wc->pin)
-		btrfs_update_pinned_extents(log->fs_info->extent_root,
-					    eb->start, eb->len, 1);
+		btrfs_pin_extent(log->fs_info->extent_root,
+				 eb->start, eb->len, 0);
 
 	if (btrfs_buffer_uptodate(eb, gen)) {
 		if (wc->write)
@@ -534,7 +543,7 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 	saved_nbytes = inode_get_bytes(inode);
 	/* drop any overlapping extents */
 	ret = btrfs_drop_extents(trans, root, inode,
-			 start, extent_end, extent_end, start, &alloc_hint);
+			 start, extent_end, extent_end, start, &alloc_hint, 1);
 	BUG_ON(ret);
 
 	if (found_type == BTRFS_FILE_EXTENT_REG ||
@@ -1971,6 +1980,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	int ret;
 	struct btrfs_root *log = root->log_root;
 	struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
+	u64 log_transid = 0;
 
 	mutex_lock(&root->log_mutex);
 	index1 = root->log_transid % 2;
@@ -1987,10 +1997,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	while (1) {
 		unsigned long batch = root->log_batch;
-		mutex_unlock(&root->log_mutex);
-		schedule_timeout_uninterruptible(1);
-		mutex_lock(&root->log_mutex);
-
+		if (root->log_multiple_pids) {
+			mutex_unlock(&root->log_mutex);
+			schedule_timeout_uninterruptible(1);
+			mutex_lock(&root->log_mutex);
+		}
 		wait_for_writer(trans, root);
 		if (batch == root->log_batch)
 			break;
@@ -2003,14 +2014,19 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 
-	ret = btrfs_write_and_wait_marked_extents(log, &log->dirty_log_pages);
+	/* we start IO on  all the marked extents here, but we don't actually
+	 * wait for them until later.
+	 */
+	ret = btrfs_write_marked_extents(log, &log->dirty_log_pages);
 	BUG_ON(ret);
 
 	btrfs_set_root_node(&log->root_item, log->node);
 
 	root->log_batch = 0;
+	log_transid = root->log_transid;
 	root->log_transid++;
 	log->log_transid = root->log_transid;
+	root->log_start_pid = 0;
 	smp_mb();
 	/*
 	 * log tree has been flushed to disk, new modifications of
@@ -2036,6 +2052,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	index2 = log_root_tree->log_transid % 2;
 	if (atomic_read(&log_root_tree->log_commit[index2])) {
+		btrfs_wait_marked_extents(log, &log->dirty_log_pages);
 		wait_log_commit(trans, log_root_tree,
 				log_root_tree->log_transid);
 		mutex_unlock(&log_root_tree->log_mutex);
@@ -2055,6 +2072,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 * check the full commit flag again
 	 */
 	if (root->fs_info->last_trans_log_full_commit == trans->transid) {
+		btrfs_wait_marked_extents(log, &log->dirty_log_pages);
 		mutex_unlock(&log_root_tree->log_mutex);
 		ret = -EAGAIN;
 		goto out_wake_log_root;
@@ -2063,6 +2081,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	ret = btrfs_write_and_wait_marked_extents(log_root_tree,
 				&log_root_tree->dirty_log_pages);
 	BUG_ON(ret);
+	btrfs_wait_marked_extents(log, &log->dirty_log_pages);
 
 	btrfs_set_super_log_root(&root->fs_info->super_for_commit,
 				log_root_tree->node->start);
@@ -2082,9 +2101,14 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 * the running transaction open, so a full commit can't hop
 	 * in and cause problems either.
 	 */
-	write_ctree_super(trans, root->fs_info->tree_root, 2);
+	write_ctree_super(trans, root->fs_info->tree_root, 1);
 	ret = 0;
 
+	mutex_lock(&root->log_mutex);
+	if (root->last_log_commit < log_transid)
+		root->last_log_commit = log_transid;
+	mutex_unlock(&root->log_mutex);
+
 out_wake_log_root:
 	atomic_set(&log_root_tree->log_commit[index2], 0);
 	smp_mb();
@@ -2605,7 +2629,7 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 								extent);
 				cs = btrfs_file_extent_offset(src, extent);
 				cl = btrfs_file_extent_num_bytes(src,
-								extent);;
+								extent);
 				if (btrfs_file_extent_compression(src,
 								  extent)) {
 					cs = 0;
@@ -2841,7 +2865,7 @@ static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans,
 		if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
 			break;
 
-		if (parent == sb->s_root)
+		if (IS_ROOT(parent))
 			break;
 
 		parent = parent->d_parent;
@@ -2852,6 +2876,21 @@ out:
 	return ret;
 }
 
+static int inode_in_log(struct btrfs_trans_handle *trans,
+		 struct inode *inode)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	int ret = 0;
+
+	mutex_lock(&root->log_mutex);
+	if (BTRFS_I(inode)->logged_trans == trans->transid &&
+	    BTRFS_I(inode)->last_sub_trans <= root->last_log_commit)
+		ret = 1;
+	mutex_unlock(&root->log_mutex);
+	return ret;
+}
+
+
 /*
  * helper function around btrfs_log_inode to make sure newly created
  * parent directories also end up in the log.  A minimal inode and backref
@@ -2880,11 +2919,22 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		goto end_no_trans;
 	}
 
+	if (root != BTRFS_I(inode)->root ||
+	    btrfs_root_refs(&root->root_item) == 0) {
+		ret = 1;
+		goto end_no_trans;
+	}
+
 	ret = check_parent_dirs_for_sync(trans, inode, parent,
 					 sb, last_committed);
 	if (ret)
 		goto end_no_trans;
 
+	if (inode_in_log(trans, inode)) {
+		ret = BTRFS_NO_LOG_SYNC;
+		goto end_no_trans;
+	}
+
 	start_log_trans(trans, root);
 
 	ret = btrfs_log_inode(trans, root, inode, inode_only);
@@ -2907,12 +2957,15 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 			break;
 
 		inode = parent->d_inode;
+		if (root != BTRFS_I(inode)->root)
+			break;
+
 		if (BTRFS_I(inode)->generation >
 		    root->fs_info->last_trans_committed) {
 			ret = btrfs_log_inode(trans, root, inode, inode_only);
 			BUG_ON(ret);
 		}
-		if (parent == sb->s_root)
+		if (IS_ROOT(parent))
 			break;
 
 		parent = parent->d_parent;
@@ -2951,7 +3004,6 @@ int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
 	struct btrfs_key tmp_key;
 	struct btrfs_root *log;
 	struct btrfs_fs_info *fs_info = log_root_tree->fs_info;
-	u64 highest_inode;
 	struct walk_control wc = {
 		.process_func = process_one_buffer,
 		.stage = 0,
@@ -3010,11 +3062,6 @@ again:
 						      path);
 			BUG_ON(ret);
 		}
-		ret = btrfs_find_highest_inode(wc.replay_dest, &highest_inode);
-		if (ret == 0) {
-			wc.replay_dest->highest_inode = highest_inode;
-			wc.replay_dest->last_inode_alloc = highest_inode;
-		}
 
 		key.offset = found_key.offset - 1;
 		wc.replay_dest->log_root = NULL;
diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h
index d09c7609e16b..0776eacb5083 100644
--- a/fs/btrfs/tree-log.h
+++ b/fs/btrfs/tree-log.h
@@ -19,6 +19,9 @@
 #ifndef __TREE_LOG_
 #define __TREE_LOG_
 
+/* return value for btrfs_log_dentry_safe that means we don't need to log it at all */
+#define BTRFS_NO_LOG_SYNC 256
+
 int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		   struct btrfs_root *root);
 int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 5cf405b0828d..7eda483d7b5a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -276,7 +276,7 @@ loop_lock:
 		 * is now congested.  Back off and let other work structs
 		 * run instead
 		 */
-		if (pending && bdi_write_congested(bdi) && batch_run > 32 &&
+		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
 		    fs_info->fs_devices->open_devices > 1) {
 			struct io_context *ioc;
 
@@ -446,8 +446,10 @@ static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig)
 			goto error;
 
 		device->name = kstrdup(orig_dev->name, GFP_NOFS);
-		if (!device->name)
+		if (!device->name) {
+			kfree(device);
 			goto error;
+		}
 
 		device->devid = orig_dev->devid;
 		device->work.func = pending_bios_fn;
@@ -719,10 +721,9 @@ error:
  * called very infrequently and that a given device has a small number
  * of extents
  */
-static noinline int find_free_dev_extent(struct btrfs_trans_handle *trans,
-					 struct btrfs_device *device,
-					 u64 num_bytes, u64 *start,
-					 u64 *max_avail)
+int find_free_dev_extent(struct btrfs_trans_handle *trans,
+			 struct btrfs_device *device, u64 num_bytes,
+			 u64 *start, u64 *max_avail)
 {
 	struct btrfs_key key;
 	struct btrfs_root *root = device->dev_root;
@@ -1736,6 +1737,10 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	extent_root = root->fs_info->extent_root;
 	em_tree = &root->fs_info->mapping_tree.map_tree;
 
+	ret = btrfs_can_relocate(extent_root, chunk_offset);
+	if (ret)
+		return -ENOSPC;
+
 	/* step one, relocate all the extents inside this chunk */
 	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
 	BUG_ON(ret);
@@ -1749,9 +1754,9 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	 * step two, delete the device extents and the
 	 * chunk tree entries
 	 */
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	BUG_ON(em->start > chunk_offset ||
 	       em->start + em->len < chunk_offset);
@@ -1780,9 +1785,9 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
 	BUG_ON(ret);
 
-	spin_lock(&em_tree->lock);
+	write_lock(&em_tree->lock);
 	remove_extent_mapping(em_tree, em);
-	spin_unlock(&em_tree->lock);
+	write_unlock(&em_tree->lock);
 
 	kfree(map);
 	em->bdev = NULL;
@@ -1807,12 +1812,15 @@ static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
 	struct btrfs_key found_key;
 	u64 chunk_tree = chunk_root->root_key.objectid;
 	u64 chunk_type;
+	bool retried = false;
+	int failed = 0;
 	int ret;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
+again:
 	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
 	key.offset = (u64)-1;
 	key.type = BTRFS_CHUNK_ITEM_KEY;
@@ -1842,7 +1850,10 @@ static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
 			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
 						   found_key.objectid,
 						   found_key.offset);
-			BUG_ON(ret);
+			if (ret == -ENOSPC)
+				failed++;
+			else if (ret)
+				BUG();
 		}
 
 		if (found_key.offset == 0)
@@ -1850,6 +1861,14 @@ static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
 		key.offset = found_key.offset - 1;
 	}
 	ret = 0;
+	if (failed && !retried) {
+		failed = 0;
+		retried = true;
+		goto again;
+	} else if (failed && retried) {
+		WARN_ON(1);
+		ret = -ENOSPC;
+	}
 error:
 	btrfs_free_path(path);
 	return ret;
@@ -1894,6 +1913,8 @@ int btrfs_balance(struct btrfs_root *dev_root)
 			continue;
 
 		ret = btrfs_shrink_device(device, old_size - size_to_free);
+		if (ret == -ENOSPC)
+			break;
 		BUG_ON(ret);
 
 		trans = btrfs_start_transaction(dev_root, 1);
@@ -1938,9 +1959,8 @@ int btrfs_balance(struct btrfs_root *dev_root)
 		chunk = btrfs_item_ptr(path->nodes[0],
 				       path->slots[0],
 				       struct btrfs_chunk);
-		key.offset = found_key.offset;
 		/* chunk zero is special */
-		if (key.offset == 0)
+		if (found_key.offset == 0)
 			break;
 
 		btrfs_release_path(chunk_root, path);
@@ -1948,7 +1968,8 @@ int btrfs_balance(struct btrfs_root *dev_root)
 					   chunk_root->root_key.objectid,
 					   found_key.objectid,
 					   found_key.offset);
-		BUG_ON(ret);
+		BUG_ON(ret && ret != -ENOSPC);
+		key.offset = found_key.offset - 1;
 	}
 	ret = 0;
 error:
@@ -1974,10 +1995,13 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	u64 chunk_offset;
 	int ret;
 	int slot;
+	int failed = 0;
+	bool retried = false;
 	struct extent_buffer *l;
 	struct btrfs_key key;
 	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
 	u64 old_total = btrfs_super_total_bytes(super_copy);
+	u64 old_size = device->total_bytes;
 	u64 diff = device->total_bytes - new_size;
 
 	if (new_size >= device->total_bytes)
@@ -1987,12 +2011,6 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	if (!path)
 		return -ENOMEM;
 
-	trans = btrfs_start_transaction(root, 1);
-	if (!trans) {
-		ret = -ENOMEM;
-		goto done;
-	}
-
 	path->reada = 2;
 
 	lock_chunks(root);
@@ -2001,8 +2019,8 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	if (device->writeable)
 		device->fs_devices->total_rw_bytes -= diff;
 	unlock_chunks(root);
-	btrfs_end_transaction(trans, root);
 
+again:
 	key.objectid = device->devid;
 	key.offset = (u64)-1;
 	key.type = BTRFS_DEV_EXTENT_KEY;
@@ -2017,6 +2035,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 			goto done;
 		if (ret) {
 			ret = 0;
+			btrfs_release_path(root, path);
 			break;
 		}
 
@@ -2024,14 +2043,18 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(l, &key, path->slots[0]);
 
-		if (key.objectid != device->devid)
+		if (key.objectid != device->devid) {
+			btrfs_release_path(root, path);
 			break;
+		}
 
 		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
 		length = btrfs_dev_extent_length(l, dev_extent);
 
-		if (key.offset + length <= new_size)
+		if (key.offset + length <= new_size) {
+			btrfs_release_path(root, path);
 			break;
+		}
 
 		chunk_tree = btrfs_dev_extent_chunk_tree(l, dev_extent);
 		chunk_objectid = btrfs_dev_extent_chunk_objectid(l, dev_extent);
@@ -2040,8 +2063,26 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 
 		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
 					   chunk_offset);
-		if (ret)
+		if (ret && ret != -ENOSPC)
 			goto done;
+		if (ret == -ENOSPC)
+			failed++;
+		key.offset -= 1;
+	}
+
+	if (failed && !retried) {
+		failed = 0;
+		retried = true;
+		goto again;
+	} else if (failed && retried) {
+		ret = -ENOSPC;
+		lock_chunks(root);
+
+		device->total_bytes = old_size;
+		if (device->writeable)
+			device->fs_devices->total_rw_bytes += diff;
+		unlock_chunks(root);
+		goto done;
 	}
 
 	/* Shrinking succeeded, else we would be at "done". */
@@ -2294,9 +2335,9 @@ again:
 	em->block_len = em->len;
 
 	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
-	spin_lock(&em_tree->lock);
+	write_lock(&em_tree->lock);
 	ret = add_extent_mapping(em_tree, em);
-	spin_unlock(&em_tree->lock);
+	write_unlock(&em_tree->lock);
 	BUG_ON(ret);
 	free_extent_map(em);
 
@@ -2491,9 +2532,9 @@ int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
 	int readonly = 0;
 	int i;
 
-	spin_lock(&map_tree->map_tree.lock);
+	read_lock(&map_tree->map_tree.lock);
 	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
-	spin_unlock(&map_tree->map_tree.lock);
+	read_unlock(&map_tree->map_tree.lock);
 	if (!em)
 		return 1;
 
@@ -2518,11 +2559,11 @@ void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree)
 	struct extent_map *em;
 
 	while (1) {
-		spin_lock(&tree->map_tree.lock);
+		write_lock(&tree->map_tree.lock);
 		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
 		if (em)
 			remove_extent_mapping(&tree->map_tree, em);
-		spin_unlock(&tree->map_tree.lock);
+		write_unlock(&tree->map_tree.lock);
 		if (!em)
 			break;
 		kfree(em->bdev);
@@ -2540,9 +2581,9 @@ int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len)
 	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	int ret;
 
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, logical, len);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 	BUG_ON(!em);
 
 	BUG_ON(em->start > logical || em->start + em->len < logical);
@@ -2604,9 +2645,9 @@ again:
 		atomic_set(&multi->error, 0);
 	}
 
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, logical, *length);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	if (!em && unplug_page)
 		return 0;
@@ -2763,9 +2804,9 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 	u64 stripe_nr;
 	int i, j, nr = 0;
 
-	spin_lock(&em_tree->lock);
+	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, chunk_start, 1);
-	spin_unlock(&em_tree->lock);
+	read_unlock(&em_tree->lock);
 
 	BUG_ON(!em || em->start != chunk_start);
 	map = (struct map_lookup *)em->bdev;
@@ -3053,9 +3094,9 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 	logical = key->offset;
 	length = btrfs_chunk_length(leaf, chunk);
 
-	spin_lock(&map_tree->map_tree.lock);
+	read_lock(&map_tree->map_tree.lock);
 	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
-	spin_unlock(&map_tree->map_tree.lock);
+	read_unlock(&map_tree->map_tree.lock);
 
 	/* already mapped? */
 	if (em && em->start <= logical && em->start + em->len > logical) {
@@ -3114,9 +3155,9 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 		map->stripes[i].dev->in_fs_metadata = 1;
 	}
 
-	spin_lock(&map_tree->map_tree.lock);
+	write_lock(&map_tree->map_tree.lock);
 	ret = add_extent_mapping(&map_tree->map_tree, em);
-	spin_unlock(&map_tree->map_tree.lock);
+	write_unlock(&map_tree->map_tree.lock);
 	BUG_ON(ret);
 	free_extent_map(em);
 
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 5139a833f721..31b0fabdd2ea 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -181,4 +181,7 @@ int btrfs_balance(struct btrfs_root *dev_root);
 void btrfs_unlock_volumes(void);
 void btrfs_lock_volumes(void);
 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
+int find_free_dev_extent(struct btrfs_trans_handle *trans,
+			 struct btrfs_device *device, u64 num_bytes,
+			 u64 *start, u64 *max_avail);
 #endif
diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c
index a9d3bf4d2689..b6dd5967c48a 100644
--- a/fs/btrfs/xattr.c
+++ b/fs/btrfs/xattr.c
@@ -260,7 +260,7 @@ err:
  * attributes are handled directly.
  */
 struct xattr_handler *btrfs_xattr_handlers[] = {
-#ifdef CONFIG_FS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 	&btrfs_xattr_acl_access_handler,
 	&btrfs_xattr_acl_default_handler,
 #endif
diff --git a/fs/buffer.c b/fs/buffer.c
index 90a98865b0cc..6fa530256bfd 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -52,6 +52,7 @@ init_buffer(struct buffer_head *bh, bh_end_io_t *handler, void *private)
 	bh->b_end_io = handler;
 	bh->b_private = private;
 }
+EXPORT_SYMBOL(init_buffer);
 
 static int sync_buffer(void *word)
 {
@@ -80,6 +81,7 @@ void unlock_buffer(struct buffer_head *bh)
 	smp_mb__after_clear_bit();
 	wake_up_bit(&bh->b_state, BH_Lock);
 }
+EXPORT_SYMBOL(unlock_buffer);
 
 /*
  * Block until a buffer comes unlocked.  This doesn't stop it
@@ -90,6 +92,7 @@ void __wait_on_buffer(struct buffer_head * bh)
 {
 	wait_on_bit(&bh->b_state, BH_Lock, sync_buffer, TASK_UNINTERRUPTIBLE);
 }
+EXPORT_SYMBOL(__wait_on_buffer);
 
 static void
 __clear_page_buffers(struct page *page)
@@ -144,6 +147,7 @@ void end_buffer_read_sync(struct buffer_head *bh, int uptodate)
 	__end_buffer_read_notouch(bh, uptodate);
 	put_bh(bh);
 }
+EXPORT_SYMBOL(end_buffer_read_sync);
 
 void end_buffer_write_sync(struct buffer_head *bh, int uptodate)
 {
@@ -164,6 +168,7 @@ void end_buffer_write_sync(struct buffer_head *bh, int uptodate)
 	unlock_buffer(bh);
 	put_bh(bh);
 }
+EXPORT_SYMBOL(end_buffer_write_sync);
 
 /*
  * Various filesystems appear to want __find_get_block to be non-blocking.
@@ -272,9 +277,10 @@ void invalidate_bdev(struct block_device *bdev)
 	invalidate_bh_lrus();
 	invalidate_mapping_pages(mapping, 0, -1);
 }
+EXPORT_SYMBOL(invalidate_bdev);
 
 /*
- * Kick pdflush then try to free up some ZONE_NORMAL memory.
+ * Kick the writeback threads then try to free up some ZONE_NORMAL memory.
  */
 static void free_more_memory(void)
 {
@@ -410,6 +416,7 @@ still_busy:
 	local_irq_restore(flags);
 	return;
 }
+EXPORT_SYMBOL(end_buffer_async_write);
 
 /*
  * If a page's buffers are under async readin (end_buffer_async_read
@@ -438,8 +445,8 @@ static void mark_buffer_async_read(struct buffer_head *bh)
 	set_buffer_async_read(bh);
 }
 
-void mark_buffer_async_write_endio(struct buffer_head *bh,
-				   bh_end_io_t *handler)
+static void mark_buffer_async_write_endio(struct buffer_head *bh,
+					  bh_end_io_t *handler)
 {
 	bh->b_end_io = handler;
 	set_buffer_async_write(bh);
@@ -553,7 +560,7 @@ repeat:
 	return err;
 }
 
-void do_thaw_all(struct work_struct *work)
+static void do_thaw_all(struct work_struct *work)
 {
 	struct super_block *sb;
 	char b[BDEVNAME_SIZE];
@@ -1172,6 +1179,7 @@ void mark_buffer_dirty(struct buffer_head *bh)
 		}
 	}
 }
+EXPORT_SYMBOL(mark_buffer_dirty);
 
 /*
  * Decrement a buffer_head's reference count.  If all buffers against a page
@@ -1188,6 +1196,7 @@ void __brelse(struct buffer_head * buf)
 	}
 	WARN(1, KERN_ERR "VFS: brelse: Trying to free free buffer\n");
 }
+EXPORT_SYMBOL(__brelse);
 
 /*
  * bforget() is like brelse(), except it discards any
@@ -1206,6 +1215,7 @@ void __bforget(struct buffer_head *bh)
 	}
 	__brelse(bh);
 }
+EXPORT_SYMBOL(__bforget);
 
 static struct buffer_head *__bread_slow(struct buffer_head *bh)
 {
@@ -1699,9 +1709,9 @@ static int __block_write_full_page(struct inode *inode, struct page *page,
 		/*
 		 * If it's a fully non-blocking write attempt and we cannot
 		 * lock the buffer then redirty the page.  Note that this can
-		 * potentially cause a busy-wait loop from pdflush and kswapd
-		 * activity, but those code paths have their own higher-level
-		 * throttling.
+		 * potentially cause a busy-wait loop from writeback threads
+		 * and kswapd activity, but those code paths have their own
+		 * higher-level throttling.
 		 */
 		if (wbc->sync_mode != WB_SYNC_NONE || !wbc->nonblocking) {
 			lock_buffer(bh);
@@ -2218,6 +2228,7 @@ int block_read_full_page(struct page *page, get_block_t *get_block)
 	}
 	return 0;
 }
+EXPORT_SYMBOL(block_read_full_page);
 
 /* utility function for filesystems that need to do work on expanding
  * truncates.  Uses filesystem pagecache writes to allow the filesystem to
@@ -2228,16 +2239,10 @@ int generic_cont_expand_simple(struct inode *inode, loff_t size)
 	struct address_space *mapping = inode->i_mapping;
 	struct page *page;
 	void *fsdata;
-	unsigned long limit;
 	int err;
 
-	err = -EFBIG;
-        limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
-	if (limit != RLIM_INFINITY && size > (loff_t)limit) {
-		send_sig(SIGXFSZ, current, 0);
-		goto out;
-	}
-	if (size > inode->i_sb->s_maxbytes)
+	err = inode_newsize_ok(inode, size);
+	if (err)
 		goto out;
 
 	err = pagecache_write_begin(NULL, mapping, size, 0,
@@ -2252,6 +2257,7 @@ int generic_cont_expand_simple(struct inode *inode, loff_t size)
 out:
 	return err;
 }
+EXPORT_SYMBOL(generic_cont_expand_simple);
 
 static int cont_expand_zero(struct file *file, struct address_space *mapping,
 			    loff_t pos, loff_t *bytes)
@@ -2352,6 +2358,7 @@ int cont_write_begin(struct file *file, struct address_space *mapping,
 out:
 	return err;
 }
+EXPORT_SYMBOL(cont_write_begin);
 
 int block_prepare_write(struct page *page, unsigned from, unsigned to,
 			get_block_t *get_block)
@@ -2362,6 +2369,7 @@ int block_prepare_write(struct page *page, unsigned from, unsigned to,
 		ClearPageUptodate(page);
 	return err;
 }
+EXPORT_SYMBOL(block_prepare_write);
 
 int block_commit_write(struct page *page, unsigned from, unsigned to)
 {
@@ -2369,6 +2377,7 @@ int block_commit_write(struct page *page, unsigned from, unsigned to)
 	__block_commit_write(inode,page,from,to);
 	return 0;
 }
+EXPORT_SYMBOL(block_commit_write);
 
 /*
  * block_page_mkwrite() is not allowed to change the file size as it gets
@@ -2426,6 +2435,7 @@ block_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf,
 out:
 	return ret;
 }
+EXPORT_SYMBOL(block_page_mkwrite);
 
 /*
  * nobh_write_begin()'s prereads are special: the buffer_heads are freed
@@ -2849,6 +2859,7 @@ unlock:
 out:
 	return err;
 }
+EXPORT_SYMBOL(block_truncate_page);
 
 /*
  * The generic ->writepage function for buffer-backed address_spaces
@@ -2890,6 +2901,7 @@ int block_write_full_page_endio(struct page *page, get_block_t *get_block,
 	zero_user_segment(page, offset, PAGE_CACHE_SIZE);
 	return __block_write_full_page(inode, page, get_block, wbc, handler);
 }
+EXPORT_SYMBOL(block_write_full_page_endio);
 
 /*
  * The generic ->writepage function for buffer-backed address_spaces
@@ -2900,7 +2912,7 @@ int block_write_full_page(struct page *page, get_block_t *get_block,
 	return block_write_full_page_endio(page, get_block, wbc,
 					   end_buffer_async_write);
 }
-
+EXPORT_SYMBOL(block_write_full_page);
 
 sector_t generic_block_bmap(struct address_space *mapping, sector_t block,
 			    get_block_t *get_block)
@@ -2913,6 +2925,7 @@ sector_t generic_block_bmap(struct address_space *mapping, sector_t block,
 	get_block(inode, block, &tmp, 0);
 	return tmp.b_blocknr;
 }
+EXPORT_SYMBOL(generic_block_bmap);
 
 static void end_bio_bh_io_sync(struct bio *bio, int err)
 {
@@ -2982,6 +2995,7 @@ int submit_bh(int rw, struct buffer_head * bh)
 	bio_put(bio);
 	return ret;
 }
+EXPORT_SYMBOL(submit_bh);
 
 /**
  * ll_rw_block: low-level access to block devices (DEPRECATED)
@@ -3043,6 +3057,7 @@ void ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
 		unlock_buffer(bh);
 	}
 }
+EXPORT_SYMBOL(ll_rw_block);
 
 /*
  * For a data-integrity writeout, we need to wait upon any in-progress I/O
@@ -3071,6 +3086,7 @@ int sync_dirty_buffer(struct buffer_head *bh)
 	}
 	return ret;
 }
+EXPORT_SYMBOL(sync_dirty_buffer);
 
 /*
  * try_to_free_buffers() checks if all the buffers on this particular page
@@ -3185,13 +3201,14 @@ void block_sync_page(struct page *page)
 	if (mapping)
 		blk_run_backing_dev(mapping->backing_dev_info, page);
 }
+EXPORT_SYMBOL(block_sync_page);
 
 /*
  * There are no bdflush tunables left.  But distributions are
  * still running obsolete flush daemons, so we terminate them here.
  *
  * Use of bdflush() is deprecated and will be removed in a future kernel.
- * The `pdflush' kernel threads fully replace bdflush daemons and this call.
+ * The `flush-X' kernel threads fully replace bdflush daemons and this call.
  */
 SYSCALL_DEFINE2(bdflush, int, func, long, data)
 {
@@ -3361,29 +3378,3 @@ void __init buffer_init(void)
 	max_buffer_heads = nrpages * (PAGE_SIZE / sizeof(struct buffer_head));
 	hotcpu_notifier(buffer_cpu_notify, 0);
 }
-
-EXPORT_SYMBOL(__bforget);
-EXPORT_SYMBOL(__brelse);
-EXPORT_SYMBOL(__wait_on_buffer);
-EXPORT_SYMBOL(block_commit_write);
-EXPORT_SYMBOL(block_prepare_write);
-EXPORT_SYMBOL(block_page_mkwrite);
-EXPORT_SYMBOL(block_read_full_page);
-EXPORT_SYMBOL(block_sync_page);
-EXPORT_SYMBOL(block_truncate_page);
-EXPORT_SYMBOL(block_write_full_page);
-EXPORT_SYMBOL(block_write_full_page_endio);
-EXPORT_SYMBOL(cont_write_begin);
-EXPORT_SYMBOL(end_buffer_read_sync);
-EXPORT_SYMBOL(end_buffer_write_sync);
-EXPORT_SYMBOL(end_buffer_async_write);
-EXPORT_SYMBOL(file_fsync);
-EXPORT_SYMBOL(generic_block_bmap);
-EXPORT_SYMBOL(generic_cont_expand_simple);
-EXPORT_SYMBOL(init_buffer);
-EXPORT_SYMBOL(invalidate_bdev);
-EXPORT_SYMBOL(ll_rw_block);
-EXPORT_SYMBOL(mark_buffer_dirty);
-EXPORT_SYMBOL(submit_bh);
-EXPORT_SYMBOL(sync_dirty_buffer);
-EXPORT_SYMBOL(unlock_buffer);
diff --git a/fs/char_dev.c b/fs/char_dev.c
index 3cbc57f932d2..d6db933df2b2 100644
--- a/fs/char_dev.c
+++ b/fs/char_dev.c
@@ -264,7 +264,6 @@ int __register_chrdev(unsigned int major, unsigned int baseminor,
 {
 	struct char_device_struct *cd;
 	struct cdev *cdev;
-	char *s;
 	int err = -ENOMEM;
 
 	cd = __register_chrdev_region(major, baseminor, count, name);
@@ -278,8 +277,6 @@ int __register_chrdev(unsigned int major, unsigned int baseminor,
 	cdev->owner = fops->owner;
 	cdev->ops = fops;
 	kobject_set_name(&cdev->kobj, "%s", name);
-	for (s = strchr(kobject_name(&cdev->kobj),'/'); s; s = strchr(s, '/'))
-		*s = '!';
 		
 	err = cdev_add(cdev, MKDEV(cd->major, baseminor), count);
 	if (err)
diff --git a/fs/cifs/Kconfig b/fs/cifs/Kconfig
index 6994a0f54f02..80f352596807 100644
--- a/fs/cifs/Kconfig
+++ b/fs/cifs/Kconfig
@@ -2,6 +2,7 @@ config CIFS
 	tristate "CIFS support (advanced network filesystem, SMBFS successor)"
 	depends on INET
 	select NLS
+	select SLOW_WORK
 	help
 	  This is the client VFS module for the Common Internet File System
 	  (CIFS) protocol which is the successor to the Server Message Block
diff --git a/fs/cifs/cifs_dfs_ref.c b/fs/cifs/cifs_dfs_ref.c
index 606912d8f2a8..fea9e898c4ba 100644
--- a/fs/cifs/cifs_dfs_ref.c
+++ b/fs/cifs/cifs_dfs_ref.c
@@ -142,7 +142,7 @@ char *cifs_compose_mount_options(const char *sb_mountdata,
 	rc = dns_resolve_server_name_to_ip(*devname, &srvIP);
 	if (rc != 0) {
 		cERROR(1, ("%s: Failed to resolve server part of %s to IP: %d",
-			  __func__, *devname, rc));;
+			  __func__, *devname, rc));
 		goto compose_mount_options_err;
 	}
 	/* md_len = strlen(...) + 12 for 'sep+prefixpath='
@@ -385,7 +385,7 @@ out_err:
 	goto out;
 }
 
-struct inode_operations cifs_dfs_referral_inode_operations = {
+const struct inode_operations cifs_dfs_referral_inode_operations = {
 	.follow_link = cifs_dfs_follow_mountpoint,
 };
 
diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c
index 3610e9958b4c..9a5e4f5f3122 100644
--- a/fs/cifs/cifsfs.c
+++ b/fs/cifs/cifsfs.c
@@ -50,7 +50,7 @@
 #define CIFS_MAGIC_NUMBER 0xFF534D42	/* the first four bytes of SMB PDUs */
 
 #ifdef CONFIG_CIFS_QUOTA
-static struct quotactl_ops cifs_quotactl_ops;
+static const struct quotactl_ops cifs_quotactl_ops;
 #endif /* QUOTA */
 
 int cifsFYI = 0;
@@ -64,9 +64,6 @@ unsigned int multiuser_mount = 0;
 unsigned int extended_security = CIFSSEC_DEF;
 /* unsigned int ntlmv2_support = 0; */
 unsigned int sign_CIFS_PDUs = 1;
-extern struct task_struct *oplockThread; /* remove sparse warning */
-struct task_struct *oplockThread = NULL;
-/* extern struct task_struct * dnotifyThread; remove sparse warning */
 static const struct super_operations cifs_super_ops;
 unsigned int CIFSMaxBufSize = CIFS_MAX_MSGSIZE;
 module_param(CIFSMaxBufSize, int, 0);
@@ -185,8 +182,7 @@ out_mount_failed:
 			cifs_sb->mountdata = NULL;
 		}
 #endif
-		if (cifs_sb->local_nls)
-			unload_nls(cifs_sb->local_nls);
+		unload_nls(cifs_sb->local_nls);
 		kfree(cifs_sb);
 	}
 	return rc;
@@ -517,7 +513,7 @@ int cifs_xstate_get(struct super_block *sb, struct fs_quota_stat *qstats)
 	return rc;
 }
 
-static struct quotactl_ops cifs_quotactl_ops = {
+static const struct quotactl_ops cifs_quotactl_ops = {
 	.set_xquota	= cifs_xquota_set,
 	.get_xquota	= cifs_xquota_get,
 	.set_xstate	= cifs_xstate_set,
@@ -973,89 +969,12 @@ cifs_destroy_mids(void)
 	kmem_cache_destroy(cifs_oplock_cachep);
 }
 
-static int cifs_oplock_thread(void *dummyarg)
-{
-	struct oplock_q_entry *oplock_item;
-	struct cifsTconInfo *pTcon;
-	struct inode *inode;
-	__u16  netfid;
-	int rc, waitrc = 0;
-
-	set_freezable();
-	do {
-		if (try_to_freeze())
-			continue;
-
-		spin_lock(&cifs_oplock_lock);
-		if (list_empty(&cifs_oplock_list)) {
-			spin_unlock(&cifs_oplock_lock);
-			set_current_state(TASK_INTERRUPTIBLE);
-			schedule_timeout(39*HZ);
-		} else {
-			oplock_item = list_entry(cifs_oplock_list.next,
-						struct oplock_q_entry, qhead);
-			cFYI(1, ("found oplock item to write out"));
-			pTcon = oplock_item->tcon;
-			inode = oplock_item->pinode;
-			netfid = oplock_item->netfid;
-			spin_unlock(&cifs_oplock_lock);
-			DeleteOplockQEntry(oplock_item);
-			/* can not grab inode sem here since it would
-				deadlock when oplock received on delete
-				since vfs_unlink holds the i_mutex across
-				the call */
-			/* mutex_lock(&inode->i_mutex);*/
-			if (S_ISREG(inode->i_mode)) {
-#ifdef CONFIG_CIFS_EXPERIMENTAL
-				if (CIFS_I(inode)->clientCanCacheAll == 0)
-					break_lease(inode, FMODE_READ);
-				else if (CIFS_I(inode)->clientCanCacheRead == 0)
-					break_lease(inode, FMODE_WRITE);
-#endif
-				rc = filemap_fdatawrite(inode->i_mapping);
-				if (CIFS_I(inode)->clientCanCacheRead == 0) {
-					waitrc = filemap_fdatawait(
-							      inode->i_mapping);
-					invalidate_remote_inode(inode);
-				}
-				if (rc == 0)
-					rc = waitrc;
-			} else
-				rc = 0;
-			/* mutex_unlock(&inode->i_mutex);*/
-			if (rc)
-				CIFS_I(inode)->write_behind_rc = rc;
-			cFYI(1, ("Oplock flush inode %p rc %d",
-				inode, rc));
-
-				/* releasing stale oplock after recent reconnect
-				of smb session using a now incorrect file
-				handle is not a data integrity issue but do
-				not bother sending an oplock release if session
-				to server still is disconnected since oplock
-				already released by the server in that case */
-			if (!pTcon->need_reconnect) {
-				rc = CIFSSMBLock(0, pTcon, netfid,
-						0 /* len */ , 0 /* offset */, 0,
-						0, LOCKING_ANDX_OPLOCK_RELEASE,
-						false /* wait flag */);
-				cFYI(1, ("Oplock release rc = %d", rc));
-			}
-			set_current_state(TASK_INTERRUPTIBLE);
-			schedule_timeout(1);  /* yield in case q were corrupt */
-		}
-	} while (!kthread_should_stop());
-
-	return 0;
-}
-
 static int __init
 init_cifs(void)
 {
 	int rc = 0;
 	cifs_proc_init();
 	INIT_LIST_HEAD(&cifs_tcp_ses_list);
-	INIT_LIST_HEAD(&cifs_oplock_list);
 #ifdef CONFIG_CIFS_EXPERIMENTAL
 	INIT_LIST_HEAD(&GlobalDnotifyReqList);
 	INIT_LIST_HEAD(&GlobalDnotifyRsp_Q);
@@ -1084,7 +1003,6 @@ init_cifs(void)
 	rwlock_init(&GlobalSMBSeslock);
 	rwlock_init(&cifs_tcp_ses_lock);
 	spin_lock_init(&GlobalMid_Lock);
-	spin_lock_init(&cifs_oplock_lock);
 
 	if (cifs_max_pending < 2) {
 		cifs_max_pending = 2;
@@ -1119,16 +1037,13 @@ init_cifs(void)
 	if (rc)
 		goto out_unregister_key_type;
 #endif
-	oplockThread = kthread_run(cifs_oplock_thread, NULL, "cifsoplockd");
-	if (IS_ERR(oplockThread)) {
-		rc = PTR_ERR(oplockThread);
-		cERROR(1, ("error %d create oplock thread", rc));
-		goto out_unregister_dfs_key_type;
-	}
+	rc = slow_work_register_user();
+	if (rc)
+		goto out_unregister_resolver_key;
 
 	return 0;
 
- out_unregister_dfs_key_type:
+ out_unregister_resolver_key:
 #ifdef CONFIG_CIFS_DFS_UPCALL
 	unregister_key_type(&key_type_dns_resolver);
  out_unregister_key_type:
@@ -1165,7 +1080,6 @@ exit_cifs(void)
 	cifs_destroy_inodecache();
 	cifs_destroy_mids();
 	cifs_destroy_request_bufs();
-	kthread_stop(oplockThread);
 }
 
 MODULE_AUTHOR("Steve French <sfrench@us.ibm.com>");
diff --git a/fs/cifs/cifsfs.h b/fs/cifs/cifsfs.h
index 094325e3f714..ac2b24c192f8 100644
--- a/fs/cifs/cifsfs.h
+++ b/fs/cifs/cifsfs.h
@@ -67,7 +67,7 @@ extern int cifs_setattr(struct dentry *, struct iattr *);
 
 extern const struct inode_operations cifs_file_inode_ops;
 extern const struct inode_operations cifs_symlink_inode_ops;
-extern struct inode_operations cifs_dfs_referral_inode_operations;
+extern const struct inode_operations cifs_dfs_referral_inode_operations;
 
 
 /* Functions related to files and directories */
diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h
index 6cfc81a32703..5d0fde18039c 100644
--- a/fs/cifs/cifsglob.h
+++ b/fs/cifs/cifsglob.h
@@ -18,6 +18,7 @@
  */
 #include <linux/in.h>
 #include <linux/in6.h>
+#include <linux/slow-work.h>
 #include "cifs_fs_sb.h"
 #include "cifsacl.h"
 /*
@@ -346,14 +347,16 @@ struct cifsFileInfo {
 	/* lock scope id (0 if none) */
 	struct file *pfile; /* needed for writepage */
 	struct inode *pInode; /* needed for oplock break */
+	struct vfsmount *mnt;
 	struct mutex lock_mutex;
 	struct list_head llist; /* list of byte range locks we have. */
 	bool closePend:1;	/* file is marked to close */
 	bool invalidHandle:1;	/* file closed via session abend */
-	bool messageMode:1;	/* for pipes: message vs byte mode */
+	bool oplock_break_cancelled:1;
 	atomic_t count;		/* reference count */
 	struct mutex fh_mutex; /* prevents reopen race after dead ses*/
 	struct cifs_search_info srch_inf;
+	struct slow_work oplock_break; /* slow_work job for oplock breaks */
 };
 
 /* Take a reference on the file private data */
@@ -365,8 +368,10 @@ static inline void cifsFileInfo_get(struct cifsFileInfo *cifs_file)
 /* Release a reference on the file private data */
 static inline void cifsFileInfo_put(struct cifsFileInfo *cifs_file)
 {
-	if (atomic_dec_and_test(&cifs_file->count))
+	if (atomic_dec_and_test(&cifs_file->count)) {
+		iput(cifs_file->pInode);
 		kfree(cifs_file);
+	}
 }
 
 /*
@@ -382,7 +387,6 @@ struct cifsInodeInfo {
 	unsigned long time;	/* jiffies of last update/check of inode */
 	bool clientCanCacheRead:1;	/* read oplock */
 	bool clientCanCacheAll:1;	/* read and writebehind oplock */
-	bool oplockPending:1;
 	bool delete_pending:1;		/* DELETE_ON_CLOSE is set */
 	u64  server_eof;		/* current file size on server */
 	u64  uniqueid;			/* server inode number */
@@ -585,9 +589,9 @@ require use of the stronger protocol */
 #define   CIFSSEC_MUST_LANMAN	0x10010
 #define   CIFSSEC_MUST_PLNTXT	0x20020
 #ifdef CONFIG_CIFS_UPCALL
-#define   CIFSSEC_MASK          0xAF0AF /* allows weak security but also krb5 */
+#define   CIFSSEC_MASK          0xBF0BF /* allows weak security but also krb5 */
 #else
-#define   CIFSSEC_MASK          0xA70A7 /* current flags supported if weak */
+#define   CIFSSEC_MASK          0xB70B7 /* current flags supported if weak */
 #endif /* UPCALL */
 #else /* do not allow weak pw hash */
 #ifdef CONFIG_CIFS_UPCALL
@@ -669,12 +673,6 @@ GLOBAL_EXTERN rwlock_t		cifs_tcp_ses_lock;
  */
 GLOBAL_EXTERN rwlock_t GlobalSMBSeslock;
 
-/* Global list of oplocks */
-GLOBAL_EXTERN struct list_head cifs_oplock_list;
-
-/* Protects the cifs_oplock_list */
-GLOBAL_EXTERN spinlock_t cifs_oplock_lock;
-
 /* Outstanding dir notify requests */
 GLOBAL_EXTERN struct list_head GlobalDnotifyReqList;
 /* DirNotify response queue */
@@ -725,3 +723,4 @@ GLOBAL_EXTERN unsigned int cifs_min_rcv;    /* min size of big ntwrk buf pool */
 GLOBAL_EXTERN unsigned int cifs_min_small;  /* min size of small buf pool */
 GLOBAL_EXTERN unsigned int cifs_max_pending; /* MAX requests at once to server*/
 
+extern const struct slow_work_ops cifs_oplock_break_ops;
diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
index da8fbf565991..6928c24d1d42 100644
--- a/fs/cifs/cifsproto.h
+++ b/fs/cifs/cifsproto.h
@@ -86,18 +86,17 @@ extern int CIFS_SessSetup(unsigned int xid, struct cifsSesInfo *ses,
 			     const int stage,
 			     const struct nls_table *nls_cp);
 extern __u16 GetNextMid(struct TCP_Server_Info *server);
-extern struct oplock_q_entry *AllocOplockQEntry(struct inode *, u16,
-						 struct cifsTconInfo *);
-extern void DeleteOplockQEntry(struct oplock_q_entry *);
-extern void DeleteTconOplockQEntries(struct cifsTconInfo *);
 extern struct timespec cifs_NTtimeToUnix(__le64 utc_nanoseconds_since_1601);
 extern u64 cifs_UnixTimeToNT(struct timespec);
 extern struct timespec cnvrtDosUnixTm(__le16 le_date, __le16 le_time,
 				      int offset);
 
+extern struct cifsFileInfo *cifs_new_fileinfo(struct inode *newinode,
+				__u16 fileHandle, struct file *file,
+				struct vfsmount *mnt, unsigned int oflags);
 extern int cifs_posix_open(char *full_path, struct inode **pinode,
-			   struct super_block *sb, int mode, int oflags,
-			   int *poplock, __u16 *pnetfid, int xid);
+			   struct vfsmount *mnt, int mode, int oflags,
+			   __u32 *poplock, __u16 *pnetfid, int xid);
 extern void cifs_unix_basic_to_fattr(struct cifs_fattr *fattr,
 				     FILE_UNIX_BASIC_INFO *info,
 				     struct cifs_sb_info *cifs_sb);
diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c
index 301e307e1279..941441d3e386 100644
--- a/fs/cifs/cifssmb.c
+++ b/fs/cifs/cifssmb.c
@@ -94,6 +94,7 @@ static void mark_open_files_invalid(struct cifsTconInfo *pTcon)
 	list_for_each_safe(tmp, tmp1, &pTcon->openFileList) {
 		open_file = list_entry(tmp, struct cifsFileInfo, tlist);
 		open_file->invalidHandle = true;
+		open_file->oplock_break_cancelled = true;
 	}
 	write_unlock(&GlobalSMBSeslock);
 	/* BB Add call to invalidate_inodes(sb) for all superblocks mounted
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index d49682433c20..43003e0bef18 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -1670,7 +1670,6 @@ cifs_put_tcon(struct cifsTconInfo *tcon)
 	CIFSSMBTDis(xid, tcon);
 	_FreeXid(xid);
 
-	DeleteTconOplockQEntries(tcon);
 	tconInfoFree(tcon);
 	cifs_put_smb_ses(ses);
 }
diff --git a/fs/cifs/dir.c b/fs/cifs/dir.c
index a6424cfc0121..627a60a6c1b1 100644
--- a/fs/cifs/dir.c
+++ b/fs/cifs/dir.c
@@ -24,6 +24,7 @@
 #include <linux/stat.h>
 #include <linux/slab.h>
 #include <linux/namei.h>
+#include <linux/mount.h>
 #include "cifsfs.h"
 #include "cifspdu.h"
 #include "cifsglob.h"
@@ -129,44 +130,45 @@ cifs_bp_rename_retry:
 	return full_path;
 }
 
-static void
-cifs_fill_fileinfo(struct inode *newinode, __u16 fileHandle,
-			struct cifsTconInfo *tcon, bool write_only)
+struct cifsFileInfo *
+cifs_new_fileinfo(struct inode *newinode, __u16 fileHandle,
+		  struct file *file, struct vfsmount *mnt, unsigned int oflags)
 {
 	int oplock = 0;
 	struct cifsFileInfo *pCifsFile;
 	struct cifsInodeInfo *pCifsInode;
+	struct cifs_sb_info *cifs_sb = CIFS_SB(mnt->mnt_sb);
 
 	pCifsFile = kzalloc(sizeof(struct cifsFileInfo), GFP_KERNEL);
-
 	if (pCifsFile == NULL)
-		return;
+		return pCifsFile;
 
 	if (oplockEnabled)
 		oplock = REQ_OPLOCK;
 
 	pCifsFile->netfid = fileHandle;
 	pCifsFile->pid = current->tgid;
-	pCifsFile->pInode = newinode;
+	pCifsFile->pInode = igrab(newinode);
+	pCifsFile->mnt = mnt;
+	pCifsFile->pfile = file;
 	pCifsFile->invalidHandle = false;
 	pCifsFile->closePend = false;
 	mutex_init(&pCifsFile->fh_mutex);
 	mutex_init(&pCifsFile->lock_mutex);
 	INIT_LIST_HEAD(&pCifsFile->llist);
 	atomic_set(&pCifsFile->count, 1);
+	slow_work_init(&pCifsFile->oplock_break, &cifs_oplock_break_ops);
 
-	/* set the following in open now
-			pCifsFile->pfile = file; */
 	write_lock(&GlobalSMBSeslock);
-	list_add(&pCifsFile->tlist, &tcon->openFileList);
+	list_add(&pCifsFile->tlist, &cifs_sb->tcon->openFileList);
 	pCifsInode = CIFS_I(newinode);
 	if (pCifsInode) {
 		/* if readable file instance put first in list*/
-		if (write_only)
+		if (oflags & FMODE_READ)
+			list_add(&pCifsFile->flist, &pCifsInode->openFileList);
+		else
 			list_add_tail(&pCifsFile->flist,
 				      &pCifsInode->openFileList);
-		else
-			list_add(&pCifsFile->flist, &pCifsInode->openFileList);
 
 		if ((oplock & 0xF) == OPLOCK_EXCLUSIVE) {
 			pCifsInode->clientCanCacheAll = true;
@@ -176,18 +178,18 @@ cifs_fill_fileinfo(struct inode *newinode, __u16 fileHandle,
 				pCifsInode->clientCanCacheRead = true;
 	}
 	write_unlock(&GlobalSMBSeslock);
+
+	return pCifsFile;
 }
 
 int cifs_posix_open(char *full_path, struct inode **pinode,
-		    struct super_block *sb, int mode, int oflags,
-		    int *poplock, __u16 *pnetfid, int xid)
+		    struct vfsmount *mnt, int mode, int oflags,
+		    __u32 *poplock, __u16 *pnetfid, int xid)
 {
 	int rc;
-	__u32 oplock;
-	bool write_only = false;
 	FILE_UNIX_BASIC_INFO *presp_data;
 	__u32 posix_flags = 0;
-	struct cifs_sb_info *cifs_sb = CIFS_SB(sb);
+	struct cifs_sb_info *cifs_sb = CIFS_SB(mnt->mnt_sb);
 	struct cifs_fattr fattr;
 
 	cFYI(1, ("posix open %s", full_path));
@@ -223,12 +225,9 @@ int cifs_posix_open(char *full_path, struct inode **pinode,
 	if (oflags & O_DIRECT)
 		posix_flags |= SMB_O_DIRECT;
 
-	if (!(oflags & FMODE_READ))
-		write_only = true;
-
 	mode &= ~current_umask();
 	rc = CIFSPOSIXCreate(xid, cifs_sb->tcon, posix_flags, mode,
-			pnetfid, presp_data, &oplock, full_path,
+			pnetfid, presp_data, poplock, full_path,
 			cifs_sb->local_nls, cifs_sb->mnt_cifs_flags &
 					CIFS_MOUNT_MAP_SPECIAL_CHR);
 	if (rc)
@@ -244,7 +243,7 @@ int cifs_posix_open(char *full_path, struct inode **pinode,
 
 	/* get new inode and set it up */
 	if (*pinode == NULL) {
-		*pinode = cifs_iget(sb, &fattr);
+		*pinode = cifs_iget(mnt->mnt_sb, &fattr);
 		if (!*pinode) {
 			rc = -ENOMEM;
 			goto posix_open_ret;
@@ -253,7 +252,7 @@ int cifs_posix_open(char *full_path, struct inode **pinode,
 		cifs_fattr_to_inode(*pinode, &fattr);
 	}
 
-	cifs_fill_fileinfo(*pinode, *pnetfid, cifs_sb->tcon, write_only);
+	cifs_new_fileinfo(*pinode, *pnetfid, NULL, mnt, oflags);
 
 posix_open_ret:
 	kfree(presp_data);
@@ -280,7 +279,7 @@ cifs_create(struct inode *inode, struct dentry *direntry, int mode,
 	int rc = -ENOENT;
 	int xid;
 	int create_options = CREATE_NOT_DIR;
-	int oplock = 0;
+	__u32 oplock = 0;
 	int oflags;
 	bool posix_create = false;
 	/*
@@ -298,7 +297,6 @@ cifs_create(struct inode *inode, struct dentry *direntry, int mode,
 	FILE_ALL_INFO *buf = NULL;
 	struct inode *newinode = NULL;
 	int disposition = FILE_OVERWRITE_IF;
-	bool write_only = false;
 
 	xid = GetXid();
 
@@ -323,7 +321,7 @@ cifs_create(struct inode *inode, struct dentry *direntry, int mode,
 	if (tcon->unix_ext && (tcon->ses->capabilities & CAP_UNIX) &&
 	    (CIFS_UNIX_POSIX_PATH_OPS_CAP &
 			le64_to_cpu(tcon->fsUnixInfo.Capability))) {
-		rc = cifs_posix_open(full_path, &newinode, inode->i_sb,
+		rc = cifs_posix_open(full_path, &newinode, nd->path.mnt,
 				     mode, oflags, &oplock, &fileHandle, xid);
 		/* EIO could indicate that (posix open) operation is not
 		   supported, despite what server claimed in capability
@@ -351,11 +349,8 @@ cifs_create(struct inode *inode, struct dentry *direntry, int mode,
 		desiredAccess = 0;
 		if (oflags & FMODE_READ)
 			desiredAccess |= GENERIC_READ; /* is this too little? */
-		if (oflags & FMODE_WRITE) {
+		if (oflags & FMODE_WRITE)
 			desiredAccess |= GENERIC_WRITE;
-			if (!(oflags & FMODE_READ))
-				write_only = true;
-		}
 
 		if ((oflags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))
 			disposition = FILE_CREATE;
@@ -470,8 +465,8 @@ cifs_create_set_dentry:
 		/* mknod case - do not leave file open */
 		CIFSSMBClose(xid, tcon, fileHandle);
 	} else if (!(posix_create) && (newinode)) {
-			cifs_fill_fileinfo(newinode, fileHandle,
-					cifs_sb->tcon, write_only);
+			cifs_new_fileinfo(newinode, fileHandle, NULL,
+						nd->path.mnt, oflags);
 	}
 cifs_create_out:
 	kfree(buf);
@@ -611,7 +606,7 @@ cifs_lookup(struct inode *parent_dir_inode, struct dentry *direntry,
 {
 	int xid;
 	int rc = 0; /* to get around spurious gcc warning, set to zero here */
-	int oplock = 0;
+	__u32 oplock = 0;
 	__u16 fileHandle = 0;
 	bool posix_open = false;
 	struct cifs_sb_info *cifs_sb;
@@ -683,8 +678,7 @@ cifs_lookup(struct inode *parent_dir_inode, struct dentry *direntry,
 		if (!(nd->flags & (LOOKUP_PARENT | LOOKUP_DIRECTORY)) &&
 		     (nd->flags & LOOKUP_OPEN) && !pTcon->broken_posix_open &&
 		     (nd->intent.open.flags & O_CREAT)) {
-			rc = cifs_posix_open(full_path, &newInode,
-					parent_dir_inode->i_sb,
+			rc = cifs_posix_open(full_path, &newInode, nd->path.mnt,
 					nd->intent.open.create_mode,
 					nd->intent.open.flags, &oplock,
 					&fileHandle, xid);
diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index fa7beac8b80e..429337eb7afe 100644
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -30,6 +30,7 @@
 #include <linux/writeback.h>
 #include <linux/task_io_accounting_ops.h>
 #include <linux/delay.h>
+#include <linux/mount.h>
 #include <asm/div64.h>
 #include "cifsfs.h"
 #include "cifspdu.h"
@@ -39,27 +40,6 @@
 #include "cifs_debug.h"
 #include "cifs_fs_sb.h"
 
-static inline struct cifsFileInfo *cifs_init_private(
-	struct cifsFileInfo *private_data, struct inode *inode,
-	struct file *file, __u16 netfid)
-{
-	memset(private_data, 0, sizeof(struct cifsFileInfo));
-	private_data->netfid = netfid;
-	private_data->pid = current->tgid;
-	mutex_init(&private_data->fh_mutex);
-	mutex_init(&private_data->lock_mutex);
-	INIT_LIST_HEAD(&private_data->llist);
-	private_data->pfile = file; /* needed for writepage */
-	private_data->pInode = inode;
-	private_data->invalidHandle = false;
-	private_data->closePend = false;
-	/* Initialize reference count to one.  The private data is
-	freed on the release of the last reference */
-	atomic_set(&private_data->count, 1);
-
-	return private_data;
-}
-
 static inline int cifs_convert_flags(unsigned int flags)
 {
 	if ((flags & O_ACCMODE) == O_RDONLY)
@@ -123,9 +103,11 @@ static inline int cifs_get_disposition(unsigned int flags)
 }
 
 /* all arguments to this function must be checked for validity in caller */
-static inline int cifs_posix_open_inode_helper(struct inode *inode,
-			struct file *file, struct cifsInodeInfo *pCifsInode,
-			struct cifsFileInfo *pCifsFile, int oplock, u16 netfid)
+static inline int
+cifs_posix_open_inode_helper(struct inode *inode, struct file *file,
+			     struct cifsInodeInfo *pCifsInode,
+			     struct cifsFileInfo *pCifsFile, __u32 oplock,
+			     u16 netfid)
 {
 
 	write_lock(&GlobalSMBSeslock);
@@ -219,17 +201,6 @@ static inline int cifs_open_inode_helper(struct inode *inode, struct file *file,
 	struct timespec temp;
 	int rc;
 
-	/* want handles we can use to read with first
-	   in the list so we do not have to walk the
-	   list to search for one in write_begin */
-	if ((file->f_flags & O_ACCMODE) == O_WRONLY) {
-		list_add_tail(&pCifsFile->flist,
-			      &pCifsInode->openFileList);
-	} else {
-		list_add(&pCifsFile->flist,
-			 &pCifsInode->openFileList);
-	}
-	write_unlock(&GlobalSMBSeslock);
 	if (pCifsInode->clientCanCacheRead) {
 		/* we have the inode open somewhere else
 		   no need to discard cache data */
@@ -279,7 +250,8 @@ client_can_cache:
 int cifs_open(struct inode *inode, struct file *file)
 {
 	int rc = -EACCES;
-	int xid, oplock;
+	int xid;
+	__u32 oplock;
 	struct cifs_sb_info *cifs_sb;
 	struct cifsTconInfo *tcon;
 	struct cifsFileInfo *pCifsFile;
@@ -324,7 +296,7 @@ int cifs_open(struct inode *inode, struct file *file)
 			le64_to_cpu(tcon->fsUnixInfo.Capability))) {
 		int oflags = (int) cifs_posix_convert_flags(file->f_flags);
 		/* can not refresh inode info since size could be stale */
-		rc = cifs_posix_open(full_path, &inode, inode->i_sb,
+		rc = cifs_posix_open(full_path, &inode, file->f_path.mnt,
 				     cifs_sb->mnt_file_mode /* ignored */,
 				     oflags, &oplock, &netfid, xid);
 		if (rc == 0) {
@@ -414,24 +386,17 @@ int cifs_open(struct inode *inode, struct file *file)
 		cFYI(1, ("cifs_open returned 0x%x", rc));
 		goto out;
 	}
-	file->private_data =
-		kmalloc(sizeof(struct cifsFileInfo), GFP_KERNEL);
+
+	pCifsFile = cifs_new_fileinfo(inode, netfid, file, file->f_path.mnt,
+					file->f_flags);
+	file->private_data = pCifsFile;
 	if (file->private_data == NULL) {
 		rc = -ENOMEM;
 		goto out;
 	}
-	pCifsFile = cifs_init_private(file->private_data, inode, file, netfid);
-	write_lock(&GlobalSMBSeslock);
-	list_add(&pCifsFile->tlist, &tcon->openFileList);
 
-	pCifsInode = CIFS_I(file->f_path.dentry->d_inode);
-	if (pCifsInode) {
-		rc = cifs_open_inode_helper(inode, file, pCifsInode,
-					    pCifsFile, tcon,
-					    &oplock, buf, full_path, xid);
-	} else {
-		write_unlock(&GlobalSMBSeslock);
-	}
+	rc = cifs_open_inode_helper(inode, file, pCifsInode, pCifsFile, tcon,
+				    &oplock, buf, full_path, xid);
 
 	if (oplock & CIFS_CREATE_ACTION) {
 		/* time to set mode which we can not set earlier due to
@@ -474,7 +439,8 @@ static int cifs_relock_file(struct cifsFileInfo *cifsFile)
 static int cifs_reopen_file(struct file *file, bool can_flush)
 {
 	int rc = -EACCES;
-	int xid, oplock;
+	int xid;
+	__u32 oplock;
 	struct cifs_sb_info *cifs_sb;
 	struct cifsTconInfo *tcon;
 	struct cifsFileInfo *pCifsFile;
@@ -543,7 +509,7 @@ reopen_error_exit:
 			le64_to_cpu(tcon->fsUnixInfo.Capability))) {
 		int oflags = (int) cifs_posix_convert_flags(file->f_flags);
 		/* can not refresh inode info since size could be stale */
-		rc = cifs_posix_open(full_path, NULL, inode->i_sb,
+		rc = cifs_posix_open(full_path, NULL, file->f_path.mnt,
 				     cifs_sb->mnt_file_mode /* ignored */,
 				     oflags, &oplock, &netfid, xid);
 		if (rc == 0) {
@@ -2308,6 +2274,73 @@ out:
 	return rc;
 }
 
+static void
+cifs_oplock_break(struct slow_work *work)
+{
+	struct cifsFileInfo *cfile = container_of(work, struct cifsFileInfo,
+						  oplock_break);
+	struct inode *inode = cfile->pInode;
+	struct cifsInodeInfo *cinode = CIFS_I(inode);
+	struct cifs_sb_info *cifs_sb = CIFS_SB(cfile->mnt->mnt_sb);
+	int rc, waitrc = 0;
+
+	if (inode && S_ISREG(inode->i_mode)) {
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+		if (cinode->clientCanCacheAll == 0)
+			break_lease(inode, FMODE_READ);
+		else if (cinode->clientCanCacheRead == 0)
+			break_lease(inode, FMODE_WRITE);
+#endif
+		rc = filemap_fdatawrite(inode->i_mapping);
+		if (cinode->clientCanCacheRead == 0) {
+			waitrc = filemap_fdatawait(inode->i_mapping);
+			invalidate_remote_inode(inode);
+		}
+		if (!rc)
+			rc = waitrc;
+		if (rc)
+			cinode->write_behind_rc = rc;
+		cFYI(1, ("Oplock flush inode %p rc %d", inode, rc));
+	}
+
+	/*
+	 * releasing stale oplock after recent reconnect of smb session using
+	 * a now incorrect file handle is not a data integrity issue but do
+	 * not bother sending an oplock release if session to server still is
+	 * disconnected since oplock already released by the server
+	 */
+	if (!cfile->closePend && !cfile->oplock_break_cancelled) {
+		rc = CIFSSMBLock(0, cifs_sb->tcon, cfile->netfid, 0, 0, 0, 0,
+				 LOCKING_ANDX_OPLOCK_RELEASE, false);
+		cFYI(1, ("Oplock release rc = %d", rc));
+	}
+}
+
+static int
+cifs_oplock_break_get(struct slow_work *work)
+{
+	struct cifsFileInfo *cfile = container_of(work, struct cifsFileInfo,
+						  oplock_break);
+	mntget(cfile->mnt);
+	cifsFileInfo_get(cfile);
+	return 0;
+}
+
+static void
+cifs_oplock_break_put(struct slow_work *work)
+{
+	struct cifsFileInfo *cfile = container_of(work, struct cifsFileInfo,
+						  oplock_break);
+	mntput(cfile->mnt);
+	cifsFileInfo_put(cfile);
+}
+
+const struct slow_work_ops cifs_oplock_break_ops = {
+	.get_ref	= cifs_oplock_break_get,
+	.put_ref	= cifs_oplock_break_put,
+	.execute	= cifs_oplock_break,
+};
+
 const struct address_space_operations cifs_addr_ops = {
 	.readpage = cifs_readpage,
 	.readpages = cifs_readpages,
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 1f09c7619319..5e2492535daa 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -1557,57 +1557,24 @@ static int cifs_truncate_page(struct address_space *mapping, loff_t from)
 
 static int cifs_vmtruncate(struct inode *inode, loff_t offset)
 {
-	struct address_space *mapping = inode->i_mapping;
-	unsigned long limit;
+	loff_t oldsize;
+	int err;
 
 	spin_lock(&inode->i_lock);
-	if (inode->i_size < offset)
-		goto do_expand;
-	/*
-	 * truncation of in-use swapfiles is disallowed - it would cause
-	 * subsequent swapout to scribble on the now-freed blocks.
-	 */
-	if (IS_SWAPFILE(inode)) {
-		spin_unlock(&inode->i_lock);
-		goto out_busy;
-	}
-	i_size_write(inode, offset);
-	spin_unlock(&inode->i_lock);
-	/*
-	 * unmap_mapping_range is called twice, first simply for efficiency
-	 * so that truncate_inode_pages does fewer single-page unmaps. However
-	 * after this first call, and before truncate_inode_pages finishes,
-	 * it is possible for private pages to be COWed, which remain after
-	 * truncate_inode_pages finishes, hence the second unmap_mapping_range
-	 * call must be made for correctness.
-	 */
-	unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-	truncate_inode_pages(mapping, offset);
-	unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-	goto out_truncate;
-
-do_expand:
-	limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
-	if (limit != RLIM_INFINITY && offset > limit) {
+	err = inode_newsize_ok(inode, offset);
+	if (err) {
 		spin_unlock(&inode->i_lock);
-		goto out_sig;
-	}
-	if (offset > inode->i_sb->s_maxbytes) {
-		spin_unlock(&inode->i_lock);
-		goto out_big;
+		goto out;
 	}
+
+	oldsize = inode->i_size;
 	i_size_write(inode, offset);
 	spin_unlock(&inode->i_lock);
-out_truncate:
+	truncate_pagecache(inode, oldsize, offset);
 	if (inode->i_op->truncate)
 		inode->i_op->truncate(inode);
-	return 0;
-out_sig:
-	send_sig(SIGXFSZ, current, 0);
-out_big:
-	return -EFBIG;
-out_busy:
-	return -ETXTBSY;
+out:
+	return err;
 }
 
 static int
diff --git a/fs/cifs/misc.c b/fs/cifs/misc.c
index e079a9190ec4..0241b25ac33f 100644
--- a/fs/cifs/misc.c
+++ b/fs/cifs/misc.c
@@ -32,7 +32,6 @@
 
 extern mempool_t *cifs_sm_req_poolp;
 extern mempool_t *cifs_req_poolp;
-extern struct task_struct *oplockThread;
 
 /* The xid serves as a useful identifier for each incoming vfs request,
    in a similar way to the mid which is useful to track each sent smb,
@@ -500,6 +499,7 @@ is_valid_oplock_break(struct smb_hdr *buf, struct TCP_Server_Info *srv)
 	struct cifsTconInfo *tcon;
 	struct cifsInodeInfo *pCifsInode;
 	struct cifsFileInfo *netfile;
+	int rc;
 
 	cFYI(1, ("Checking for oplock break or dnotify response"));
 	if ((pSMB->hdr.Command == SMB_COM_NT_TRANSACT) &&
@@ -562,30 +562,40 @@ is_valid_oplock_break(struct smb_hdr *buf, struct TCP_Server_Info *srv)
 				continue;
 
 			cifs_stats_inc(&tcon->num_oplock_brks);
-			write_lock(&GlobalSMBSeslock);
+			read_lock(&GlobalSMBSeslock);
 			list_for_each(tmp2, &tcon->openFileList) {
 				netfile = list_entry(tmp2, struct cifsFileInfo,
 						     tlist);
 				if (pSMB->Fid != netfile->netfid)
 					continue;
 
-				write_unlock(&GlobalSMBSeslock);
-				read_unlock(&cifs_tcp_ses_lock);
+				/*
+				 * don't do anything if file is about to be
+				 * closed anyway.
+				 */
+				if (netfile->closePend) {
+					read_unlock(&GlobalSMBSeslock);
+					read_unlock(&cifs_tcp_ses_lock);
+					return true;
+				}
+
 				cFYI(1, ("file id match, oplock break"));
 				pCifsInode = CIFS_I(netfile->pInode);
 				pCifsInode->clientCanCacheAll = false;
 				if (pSMB->OplockLevel == 0)
 					pCifsInode->clientCanCacheRead = false;
-				pCifsInode->oplockPending = true;
-				AllocOplockQEntry(netfile->pInode,
-						  netfile->netfid, tcon);
-				cFYI(1, ("about to wake up oplock thread"));
-				if (oplockThread)
-					wake_up_process(oplockThread);
-
+				rc = slow_work_enqueue(&netfile->oplock_break);
+				if (rc) {
+					cERROR(1, ("failed to enqueue oplock "
+						   "break: %d\n", rc));
+				} else {
+					netfile->oplock_break_cancelled = false;
+				}
+				read_unlock(&GlobalSMBSeslock);
+				read_unlock(&cifs_tcp_ses_lock);
 				return true;
 			}
-			write_unlock(&GlobalSMBSeslock);
+			read_unlock(&GlobalSMBSeslock);
 			read_unlock(&cifs_tcp_ses_lock);
 			cFYI(1, ("No matching file for oplock break"));
 			return true;
diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
index f823a4a208a7..1f098ca71636 100644
--- a/fs/cifs/readdir.c
+++ b/fs/cifs/readdir.c
@@ -146,7 +146,7 @@ cifs_fill_common_info(struct cifs_fattr *fattr, struct cifs_sb_info *cifs_sb)
 	}
 }
 
-void
+static void
 cifs_dir_info_to_fattr(struct cifs_fattr *fattr, FILE_DIRECTORY_INFO *info,
 		       struct cifs_sb_info *cifs_sb)
 {
@@ -161,7 +161,7 @@ cifs_dir_info_to_fattr(struct cifs_fattr *fattr, FILE_DIRECTORY_INFO *info,
 	cifs_fill_common_info(fattr, cifs_sb);
 }
 
-void
+static void
 cifs_std_info_to_fattr(struct cifs_fattr *fattr, FIND_FILE_STANDARD_INFO *info,
 		       struct cifs_sb_info *cifs_sb)
 {
diff --git a/fs/cifs/transport.c b/fs/cifs/transport.c
index 1da4ab250eae..07b8e71544ee 100644
--- a/fs/cifs/transport.c
+++ b/fs/cifs/transport.c
@@ -103,56 +103,6 @@ DeleteMidQEntry(struct mid_q_entry *midEntry)
 	mempool_free(midEntry, cifs_mid_poolp);
 }
 
-struct oplock_q_entry *
-AllocOplockQEntry(struct inode *pinode, __u16 fid, struct cifsTconInfo *tcon)
-{
-	struct oplock_q_entry *temp;
-	if ((pinode == NULL) || (tcon == NULL)) {
-		cERROR(1, ("Null parms passed to AllocOplockQEntry"));
-		return NULL;
-	}
-	temp = (struct oplock_q_entry *) kmem_cache_alloc(cifs_oplock_cachep,
-						       GFP_KERNEL);
-	if (temp == NULL)
-		return temp;
-	else {
-		temp->pinode = pinode;
-		temp->tcon = tcon;
-		temp->netfid = fid;
-		spin_lock(&cifs_oplock_lock);
-		list_add_tail(&temp->qhead, &cifs_oplock_list);
-		spin_unlock(&cifs_oplock_lock);
-	}
-	return temp;
-}
-
-void DeleteOplockQEntry(struct oplock_q_entry *oplockEntry)
-{
-	spin_lock(&cifs_oplock_lock);
-    /* should we check if list empty first? */
-	list_del(&oplockEntry->qhead);
-	spin_unlock(&cifs_oplock_lock);
-	kmem_cache_free(cifs_oplock_cachep, oplockEntry);
-}
-
-
-void DeleteTconOplockQEntries(struct cifsTconInfo *tcon)
-{
-	struct oplock_q_entry *temp;
-
-	if (tcon == NULL)
-		return;
-
-	spin_lock(&cifs_oplock_lock);
-	list_for_each_entry(temp, &cifs_oplock_list, qhead) {
-		if ((temp->tcon) && (temp->tcon == tcon)) {
-			list_del(&temp->qhead);
-			kmem_cache_free(cifs_oplock_cachep, temp);
-		}
-	}
-	spin_unlock(&cifs_oplock_lock);
-}
-
 static int
 smb_sendv(struct TCP_Server_Info *server, struct kvec *iov, int n_vec)
 {
diff --git a/fs/coda/coda_int.h b/fs/coda/coda_int.h
index 8ccd5ed81d9c..d99860a33890 100644
--- a/fs/coda/coda_int.h
+++ b/fs/coda/coda_int.h
@@ -2,6 +2,7 @@
 #define _CODA_INT_
 
 struct dentry;
+struct file;
 
 extern struct file_system_type coda_fs_type;
 extern unsigned long coda_timeout;
diff --git a/fs/coda/psdev.c b/fs/coda/psdev.c
index 0376ac66c44a..be4392ca2098 100644
--- a/fs/coda/psdev.c
+++ b/fs/coda/psdev.c
@@ -22,6 +22,7 @@
 #include <linux/kernel.h>
 #include <linux/major.h>
 #include <linux/time.h>
+#include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/ioport.h>
 #include <linux/fcntl.h>
diff --git a/fs/compat.c b/fs/compat.c
index 6d6f98fe64a0..d576b552e8e2 100644
--- a/fs/compat.c
+++ b/fs/compat.c
@@ -100,13 +100,6 @@ asmlinkage long compat_sys_utimensat(unsigned int dfd, char __user *filename, st
 		    get_compat_timespec(&tv[1], &t[1]))
 			return -EFAULT;
 
-		if ((tv[0].tv_nsec == UTIME_OMIT || tv[0].tv_nsec == UTIME_NOW)
-		    && tv[0].tv_sec != 0)
-			return -EINVAL;
-		if ((tv[1].tv_nsec == UTIME_OMIT || tv[1].tv_nsec == UTIME_NOW)
-		    && tv[1].tv_sec != 0)
-			return -EINVAL;
-
 		if (tv[0].tv_nsec == UTIME_OMIT && tv[1].tv_nsec == UTIME_OMIT)
 			return 0;
 	}
@@ -775,13 +768,13 @@ asmlinkage long compat_sys_mount(char __user * dev_name, char __user * dir_name,
 				 char __user * type, unsigned long flags,
 				 void __user * data)
 {
-	unsigned long type_page;
+	char *kernel_type;
 	unsigned long data_page;
-	unsigned long dev_page;
+	char *kernel_dev;
 	char *dir_page;
 	int retval;
 
-	retval = copy_mount_options (type, &type_page);
+	retval = copy_mount_string(type, &kernel_type);
 	if (retval < 0)
 		goto out;
 
@@ -790,38 +783,38 @@ asmlinkage long compat_sys_mount(char __user * dev_name, char __user * dir_name,
 	if (IS_ERR(dir_page))
 		goto out1;
 
-	retval = copy_mount_options (dev_name, &dev_page);
+	retval = copy_mount_string(dev_name, &kernel_dev);
 	if (retval < 0)
 		goto out2;
 
-	retval = copy_mount_options (data, &data_page);
+	retval = copy_mount_options(data, &data_page);
 	if (retval < 0)
 		goto out3;
 
 	retval = -EINVAL;
 
-	if (type_page && data_page) {
-		if (!strcmp((char *)type_page, SMBFS_NAME)) {
+	if (kernel_type && data_page) {
+		if (!strcmp(kernel_type, SMBFS_NAME)) {
 			do_smb_super_data_conv((void *)data_page);
-		} else if (!strcmp((char *)type_page, NCPFS_NAME)) {
+		} else if (!strcmp(kernel_type, NCPFS_NAME)) {
 			do_ncp_super_data_conv((void *)data_page);
-		} else if (!strcmp((char *)type_page, NFS4_NAME)) {
+		} else if (!strcmp(kernel_type, NFS4_NAME)) {
 			if (do_nfs4_super_data_conv((void *) data_page))
 				goto out4;
 		}
 	}
 
-	retval = do_mount((char*)dev_page, dir_page, (char*)type_page,
+	retval = do_mount(kernel_dev, dir_page, kernel_type,
 			flags, (void*)data_page);
 
  out4:
 	free_page(data_page);
  out3:
-	free_page(dev_page);
+	kfree(kernel_dev);
  out2:
 	putname(dir_page);
  out1:
-	free_page(type_page);
+	kfree(kernel_type);
  out:
 	return retval;
 }
diff --git a/fs/devpts/inode.c b/fs/devpts/inode.c
index 75efb028974b..d5f8c96964be 100644
--- a/fs/devpts/inode.c
+++ b/fs/devpts/inode.c
@@ -18,14 +18,13 @@
 #include <linux/mount.h>
 #include <linux/tty.h>
 #include <linux/mutex.h>
+#include <linux/magic.h>
 #include <linux/idr.h>
 #include <linux/devpts_fs.h>
 #include <linux/parser.h>
 #include <linux/fsnotify.h>
 #include <linux/seq_file.h>
 
-#define DEVPTS_SUPER_MAGIC 0x1cd1
-
 #define DEVPTS_DEFAULT_MODE 0600
 /*
  * ptmx is a new node in /dev/pts and will be unused in legacy (single-
diff --git a/fs/dlm/debug_fs.c b/fs/dlm/debug_fs.c
index 1d1d27442235..1c8bb8c3a82e 100644
--- a/fs/dlm/debug_fs.c
+++ b/fs/dlm/debug_fs.c
@@ -386,9 +386,9 @@ static int table_seq_show(struct seq_file *seq, void *iter_ptr)
 	return rv;
 }
 
-static struct seq_operations format1_seq_ops;
-static struct seq_operations format2_seq_ops;
-static struct seq_operations format3_seq_ops;
+static const struct seq_operations format1_seq_ops;
+static const struct seq_operations format2_seq_ops;
+static const struct seq_operations format3_seq_ops;
 
 static void *table_seq_start(struct seq_file *seq, loff_t *pos)
 {
@@ -534,21 +534,21 @@ static void table_seq_stop(struct seq_file *seq, void *iter_ptr)
 	}
 }
 
-static struct seq_operations format1_seq_ops = {
+static const struct seq_operations format1_seq_ops = {
 	.start = table_seq_start,
 	.next  = table_seq_next,
 	.stop  = table_seq_stop,
 	.show  = table_seq_show,
 };
 
-static struct seq_operations format2_seq_ops = {
+static const struct seq_operations format2_seq_ops = {
 	.start = table_seq_start,
 	.next  = table_seq_next,
 	.stop  = table_seq_stop,
 	.show  = table_seq_show,
 };
 
-static struct seq_operations format3_seq_ops = {
+static const struct seq_operations format3_seq_ops = {
 	.start = table_seq_start,
 	.next  = table_seq_next,
 	.stop  = table_seq_stop,
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 618a60f03886..70736eb4b516 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -106,6 +106,7 @@ struct connection {
 #define CF_CONNECT_PENDING 3
 #define CF_INIT_PENDING 4
 #define CF_IS_OTHERCON 5
+#define CF_CLOSE 6
 	struct list_head writequeue;  /* List of outgoing writequeue_entries */
 	spinlock_t writequeue_lock;
 	int (*rx_action) (struct connection *);	/* What to do when active */
@@ -299,6 +300,8 @@ static void lowcomms_write_space(struct sock *sk)
 
 static inline void lowcomms_connect_sock(struct connection *con)
 {
+	if (test_bit(CF_CLOSE, &con->flags))
+		return;
 	if (!test_and_set_bit(CF_CONNECT_PENDING, &con->flags))
 		queue_work(send_workqueue, &con->swork);
 }
@@ -313,6 +316,10 @@ int dlm_lowcomms_connect_node(int nodeid)
 {
 	struct connection *con;
 
+	/* with sctp there's no connecting without sending */
+	if (dlm_config.ci_protocol != 0)
+		return 0;
+
 	if (nodeid == dlm_our_nodeid())
 		return 0;
 
@@ -452,9 +459,9 @@ static void process_sctp_notification(struct connection *con,
 			int prim_len, ret;
 			int addr_len;
 			struct connection *new_con;
-			struct file *file;
 			sctp_peeloff_arg_t parg;
 			int parglen = sizeof(parg);
+			int err;
 
 			/*
 			 * We get this before any data for an association.
@@ -509,19 +516,22 @@ static void process_sctp_notification(struct connection *con,
 			ret = kernel_getsockopt(con->sock, IPPROTO_SCTP,
 						SCTP_SOCKOPT_PEELOFF,
 						(void *)&parg, &parglen);
-			if (ret) {
+			if (ret < 0) {
 				log_print("Can't peel off a socket for "
-					  "connection %d to node %d: err=%d\n",
+					  "connection %d to node %d: err=%d",
 					  parg.associd, nodeid, ret);
+				return;
+			}
+			new_con->sock = sockfd_lookup(parg.sd, &err);
+			if (!new_con->sock) {
+				log_print("sockfd_lookup error %d", err);
+				return;
 			}
-			file = fget(parg.sd);
-			new_con->sock = SOCKET_I(file->f_dentry->d_inode);
 			add_sock(new_con->sock, new_con);
-			fput(file);
-			put_unused_fd(parg.sd);
+			sockfd_put(new_con->sock);
 
-			log_print("got new/restarted association %d nodeid %d",
-				 (int)sn->sn_assoc_change.sac_assoc_id, nodeid);
+			log_print("connecting to %d sctp association %d",
+				 nodeid, (int)sn->sn_assoc_change.sac_assoc_id);
 
 			/* Send any pending writes */
 			clear_bit(CF_CONNECT_PENDING, &new_con->flags);
@@ -834,8 +844,6 @@ static void sctp_init_assoc(struct connection *con)
 	if (con->retries++ > MAX_CONNECT_RETRIES)
 		return;
 
-	log_print("Initiating association with node %d", con->nodeid);
-
 	if (nodeid_to_addr(con->nodeid, (struct sockaddr *)&rem_addr)) {
 		log_print("no address for nodeid %d", con->nodeid);
 		return;
@@ -852,11 +860,14 @@ static void sctp_init_assoc(struct connection *con)
 	outmessage.msg_flags = MSG_EOR;
 
 	spin_lock(&con->writequeue_lock);
-	e = list_entry(con->writequeue.next, struct writequeue_entry,
-		       list);
 
-	BUG_ON((struct list_head *) e == &con->writequeue);
+	if (list_empty(&con->writequeue)) {
+		spin_unlock(&con->writequeue_lock);
+		log_print("writequeue empty for nodeid %d", con->nodeid);
+		return;
+	}
 
+	e = list_first_entry(&con->writequeue, struct writequeue_entry, list);
 	len = e->len;
 	offset = e->offset;
 	spin_unlock(&con->writequeue_lock);
@@ -926,10 +937,8 @@ static void tcp_connect_to_sock(struct connection *con)
 		goto out_err;
 
 	memset(&saddr, 0, sizeof(saddr));
-	if (dlm_nodeid_to_addr(con->nodeid, &saddr)) {
-		sock_release(sock);
+	if (dlm_nodeid_to_addr(con->nodeid, &saddr))
 		goto out_err;
-	}
 
 	sock->sk->sk_user_data = con;
 	con->rx_action = receive_from_sock;
@@ -1284,7 +1293,6 @@ out:
 static void send_to_sock(struct connection *con)
 {
 	int ret = 0;
-	ssize_t(*sendpage) (struct socket *, struct page *, int, size_t, int);
 	const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
 	struct writequeue_entry *e;
 	int len, offset;
@@ -1293,8 +1301,6 @@ static void send_to_sock(struct connection *con)
 	if (con->sock == NULL)
 		goto out_connect;
 
-	sendpage = con->sock->ops->sendpage;
-
 	spin_lock(&con->writequeue_lock);
 	for (;;) {
 		e = list_entry(con->writequeue.next, struct writequeue_entry,
@@ -1309,8 +1315,8 @@ static void send_to_sock(struct connection *con)
 
 		ret = 0;
 		if (len) {
-			ret = sendpage(con->sock, e->page, offset, len,
-				       msg_flags);
+			ret = kernel_sendpage(con->sock, e->page, offset, len,
+					      msg_flags);
 			if (ret == -EAGAIN || ret == 0) {
 				cond_resched();
 				goto out;
@@ -1370,6 +1376,13 @@ int dlm_lowcomms_close(int nodeid)
 	log_print("closing connection to node %d", nodeid);
 	con = nodeid2con(nodeid, 0);
 	if (con) {
+		clear_bit(CF_CONNECT_PENDING, &con->flags);
+		clear_bit(CF_WRITE_PENDING, &con->flags);
+		set_bit(CF_CLOSE, &con->flags);
+		if (cancel_work_sync(&con->swork))
+			log_print("canceled swork for node %d", nodeid);
+		if (cancel_work_sync(&con->rwork))
+			log_print("canceled rwork for node %d", nodeid);
 		clean_one_writequeue(con);
 		close_connection(con, true);
 	}
@@ -1395,9 +1408,10 @@ static void process_send_sockets(struct work_struct *work)
 
 	if (test_and_clear_bit(CF_CONNECT_PENDING, &con->flags)) {
 		con->connect_action(con);
+		set_bit(CF_WRITE_PENDING, &con->flags);
 	}
-	clear_bit(CF_WRITE_PENDING, &con->flags);
-	send_to_sock(con);
+	if (test_and_clear_bit(CF_WRITE_PENDING, &con->flags))
+		send_to_sock(con);
 }
 
 
diff --git a/fs/drop_caches.c b/fs/drop_caches.c
index a2edb7913447..31f4b0e6d72c 100644
--- a/fs/drop_caches.c
+++ b/fs/drop_caches.c
@@ -63,9 +63,9 @@ static void drop_slab(void)
 }
 
 int drop_caches_sysctl_handler(ctl_table *table, int write,
-	struct file *file, void __user *buffer, size_t *length, loff_t *ppos)
+	void __user *buffer, size_t *length, loff_t *ppos)
 {
-	proc_dointvec_minmax(table, write, file, buffer, length, ppos);
+	proc_dointvec_minmax(table, write, buffer, length, ppos);
 	if (write) {
 		if (sysctl_drop_caches & 1)
 			drop_pagecache();
diff --git a/fs/ecryptfs/Kconfig b/fs/ecryptfs/Kconfig
index 0c754e64232b..1cd6d9d3e29a 100644
--- a/fs/ecryptfs/Kconfig
+++ b/fs/ecryptfs/Kconfig
@@ -1,6 +1,9 @@
 config ECRYPT_FS
 	tristate "eCrypt filesystem layer support (EXPERIMENTAL)"
-	depends on EXPERIMENTAL && KEYS && CRYPTO && NET
+	depends on EXPERIMENTAL && KEYS && CRYPTO
+	select CRYPTO_ECB
+	select CRYPTO_CBC
+	select CRYPTO_MD5
 	help
 	  Encrypted filesystem that operates on the VFS layer.  See
 	  <file:Documentation/filesystems/ecryptfs.txt> to learn more about
diff --git a/fs/ecryptfs/crypto.c b/fs/ecryptfs/crypto.c
index b91851f1cda3..fbb6e5eed697 100644
--- a/fs/ecryptfs/crypto.c
+++ b/fs/ecryptfs/crypto.c
@@ -245,13 +245,11 @@ void ecryptfs_destroy_crypt_stat(struct ecryptfs_crypt_stat *crypt_stat)
 		crypto_free_blkcipher(crypt_stat->tfm);
 	if (crypt_stat->hash_tfm)
 		crypto_free_hash(crypt_stat->hash_tfm);
-	mutex_lock(&crypt_stat->keysig_list_mutex);
 	list_for_each_entry_safe(key_sig, key_sig_tmp,
 				 &crypt_stat->keysig_list, crypt_stat_list) {
 		list_del(&key_sig->crypt_stat_list);
 		kmem_cache_free(ecryptfs_key_sig_cache, key_sig);
 	}
-	mutex_unlock(&crypt_stat->keysig_list_mutex);
 	memset(crypt_stat, 0, sizeof(struct ecryptfs_crypt_stat));
 }
 
@@ -511,13 +509,14 @@ int ecryptfs_encrypt_page(struct page *page)
 				  + extent_offset), crypt_stat);
 		rc = ecryptfs_write_lower(ecryptfs_inode, enc_extent_virt,
 					  offset, crypt_stat->extent_size);
-		if (rc) {
+		if (rc < 0) {
 			ecryptfs_printk(KERN_ERR, "Error attempting "
 					"to write lower page; rc = [%d]"
 					"\n", rc);
 			goto out;
 		}
 	}
+	rc = 0;
 out:
 	if (enc_extent_page) {
 		kunmap(enc_extent_page);
@@ -633,7 +632,7 @@ int ecryptfs_decrypt_page(struct page *page)
 		rc = ecryptfs_read_lower(enc_extent_virt, offset,
 					 crypt_stat->extent_size,
 					 ecryptfs_inode);
-		if (rc) {
+		if (rc < 0) {
 			ecryptfs_printk(KERN_ERR, "Error attempting "
 					"to read lower page; rc = [%d]"
 					"\n", rc);
@@ -797,6 +796,7 @@ int ecryptfs_init_crypt_ctx(struct ecryptfs_crypt_stat *crypt_stat)
 	kfree(full_alg_name);
 	if (IS_ERR(crypt_stat->tfm)) {
 		rc = PTR_ERR(crypt_stat->tfm);
+		crypt_stat->tfm = NULL;
 		ecryptfs_printk(KERN_ERR, "cryptfs: init_crypt_ctx(): "
 				"Error initializing cipher [%s]\n",
 				crypt_stat->cipher);
@@ -925,7 +925,9 @@ static int ecryptfs_copy_mount_wide_sigs_to_inode_sigs(
 	struct ecryptfs_global_auth_tok *global_auth_tok;
 	int rc = 0;
 
+	mutex_lock(&crypt_stat->keysig_list_mutex);
 	mutex_lock(&mount_crypt_stat->global_auth_tok_list_mutex);
+
 	list_for_each_entry(global_auth_tok,
 			    &mount_crypt_stat->global_auth_tok_list,
 			    mount_crypt_stat_list) {
@@ -934,13 +936,13 @@ static int ecryptfs_copy_mount_wide_sigs_to_inode_sigs(
 		rc = ecryptfs_add_keysig(crypt_stat, global_auth_tok->sig);
 		if (rc) {
 			printk(KERN_ERR "Error adding keysig; rc = [%d]\n", rc);
-			mutex_unlock(
-				&mount_crypt_stat->global_auth_tok_list_mutex);
 			goto out;
 		}
 	}
-	mutex_unlock(&mount_crypt_stat->global_auth_tok_list_mutex);
+
 out:
+	mutex_unlock(&mount_crypt_stat->global_auth_tok_list_mutex);
+	mutex_unlock(&crypt_stat->keysig_list_mutex);
 	return rc;
 }
 
@@ -1212,14 +1214,15 @@ int ecryptfs_read_and_validate_header_region(char *data,
 		crypt_stat->extent_size = ECRYPTFS_DEFAULT_EXTENT_SIZE;
 	rc = ecryptfs_read_lower(data, 0, crypt_stat->extent_size,
 				 ecryptfs_inode);
-	if (rc) {
+	if (rc < 0) {
 		printk(KERN_ERR "%s: Error reading header region; rc = [%d]\n",
 		       __func__, rc);
 		goto out;
 	}
 	if (!contains_ecryptfs_marker(data + ECRYPTFS_FILE_SIZE_BYTES)) {
 		rc = -EINVAL;
-	}
+	} else
+		rc = 0;
 out:
 	return rc;
 }
@@ -1314,10 +1317,11 @@ ecryptfs_write_metadata_to_contents(struct dentry *ecryptfs_dentry,
 
 	rc = ecryptfs_write_lower(ecryptfs_dentry->d_inode, virt,
 				  0, virt_len);
-	if (rc)
+	if (rc < 0)
 		printk(KERN_ERR "%s: Error attempting to write header "
-		       "information to lower file; rc = [%d]\n", __func__,
-		       rc);
+		       "information to lower file; rc = [%d]\n", __func__, rc);
+	else
+		rc = 0;
 	return rc;
 }
 
@@ -1597,7 +1601,7 @@ int ecryptfs_read_metadata(struct dentry *ecryptfs_dentry)
 	}
 	rc = ecryptfs_read_lower(page_virt, 0, crypt_stat->extent_size,
 				 ecryptfs_inode);
-	if (!rc)
+	if (rc >= 0)
 		rc = ecryptfs_read_headers_virt(page_virt, crypt_stat,
 						ecryptfs_dentry,
 						ECRYPTFS_VALIDATE_HEADER_SIZE);
@@ -1702,7 +1706,7 @@ ecryptfs_encrypt_filename(struct ecryptfs_filename *filename,
 	} else {
 		printk(KERN_ERR "%s: No support for requested filename "
 		       "encryption method in this release\n", __func__);
-		rc = -ENOTSUPP;
+		rc = -EOPNOTSUPP;
 		goto out;
 	}
 out:
@@ -1763,7 +1767,7 @@ ecryptfs_process_key_cipher(struct crypto_blkcipher **key_tfm,
 	if (IS_ERR(*key_tfm)) {
 		rc = PTR_ERR(*key_tfm);
 		printk(KERN_ERR "Unable to allocate crypto cipher with name "
-		       "[%s]; rc = [%d]\n", cipher_name, rc);
+		       "[%s]; rc = [%d]\n", full_alg_name, rc);
 		goto out;
 	}
 	crypto_blkcipher_set_flags(*key_tfm, CRYPTO_TFM_REQ_WEAK_KEY);
@@ -1776,7 +1780,8 @@ ecryptfs_process_key_cipher(struct crypto_blkcipher **key_tfm,
 	rc = crypto_blkcipher_setkey(*key_tfm, dummy_key, *key_size);
 	if (rc) {
 		printk(KERN_ERR "Error attempting to set key of size [%zd] for "
-		       "cipher [%s]; rc = [%d]\n", *key_size, cipher_name, rc);
+		       "cipher [%s]; rc = [%d]\n", *key_size, full_alg_name,
+		       rc);
 		rc = -EINVAL;
 		goto out;
 	}
@@ -2166,7 +2171,7 @@ int ecryptfs_encrypt_and_encode_filename(
 			(*encoded_name)[(*encoded_name_size)] = '\0';
 			(*encoded_name_size)++;
 		} else {
-			rc = -ENOTSUPP;
+			rc = -EOPNOTSUPP;
 		}
 		if (rc) {
 			printk(KERN_ERR "%s: Error attempting to encode "
diff --git a/fs/ecryptfs/ecryptfs_kernel.h b/fs/ecryptfs/ecryptfs_kernel.h
index 00b30a2d5466..542f625312f3 100644
--- a/fs/ecryptfs/ecryptfs_kernel.h
+++ b/fs/ecryptfs/ecryptfs_kernel.h
@@ -582,7 +582,7 @@ extern const struct inode_operations ecryptfs_dir_iops;
 extern const struct inode_operations ecryptfs_symlink_iops;
 extern const struct super_operations ecryptfs_sops;
 extern const struct dentry_operations ecryptfs_dops;
-extern struct address_space_operations ecryptfs_aops;
+extern const struct address_space_operations ecryptfs_aops;
 extern int ecryptfs_verbosity;
 extern unsigned int ecryptfs_message_buf_len;
 extern signed long ecryptfs_message_wait_timeout;
diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
index 2f0945d63297..056fed62d0de 100644
--- a/fs/ecryptfs/inode.c
+++ b/fs/ecryptfs/inode.c
@@ -476,6 +476,7 @@ static int ecryptfs_unlink(struct inode *dir, struct dentry *dentry)
 	struct inode *lower_dir_inode = ecryptfs_inode_to_lower(dir);
 	struct dentry *lower_dir_dentry;
 
+	dget(lower_dentry);
 	lower_dir_dentry = lock_parent(lower_dentry);
 	rc = vfs_unlink(lower_dir_inode, lower_dentry);
 	if (rc) {
@@ -489,6 +490,7 @@ static int ecryptfs_unlink(struct inode *dir, struct dentry *dentry)
 	d_drop(dentry);
 out_unlock:
 	unlock_dir(lower_dir_dentry);
+	dput(lower_dentry);
 	return rc;
 }
 
diff --git a/fs/ecryptfs/keystore.c b/fs/ecryptfs/keystore.c
index 259525c9abb8..a0a7847567e9 100644
--- a/fs/ecryptfs/keystore.c
+++ b/fs/ecryptfs/keystore.c
@@ -416,7 +416,9 @@ ecryptfs_find_global_auth_tok_for_sig(
 			    &mount_crypt_stat->global_auth_tok_list,
 			    mount_crypt_stat_list) {
 		if (memcmp(walker->sig, sig, ECRYPTFS_SIG_SIZE_HEX) == 0) {
-			(*global_auth_tok) = walker;
+			rc = key_validate(walker->global_auth_tok_key);
+			if (!rc)
+				(*global_auth_tok) = walker;
 			goto out;
 		}
 	}
@@ -612,7 +614,12 @@ ecryptfs_write_tag_70_packet(char *dest, size_t *remaining_bytes,
 	}
 	/* TODO: Support other key modules than passphrase for
 	 * filename encryption */
-	BUG_ON(s->auth_tok->token_type != ECRYPTFS_PASSWORD);
+	if (s->auth_tok->token_type != ECRYPTFS_PASSWORD) {
+		rc = -EOPNOTSUPP;
+		printk(KERN_INFO "%s: Filename encryption only supports "
+		       "password tokens\n", __func__);
+		goto out_free_unlock;
+	}
 	sg_init_one(
 		&s->hash_sg,
 		(u8 *)s->auth_tok->token.password.session_key_encryption_key,
@@ -910,7 +917,12 @@ ecryptfs_parse_tag_70_packet(char **filename, size_t *filename_size,
 	}
 	/* TODO: Support other key modules than passphrase for
 	 * filename encryption */
-	BUG_ON(s->auth_tok->token_type != ECRYPTFS_PASSWORD);
+	if (s->auth_tok->token_type != ECRYPTFS_PASSWORD) {
+		rc = -EOPNOTSUPP;
+		printk(KERN_INFO "%s: Filename encryption only supports "
+		       "password tokens\n", __func__);
+		goto out_free_unlock;
+	}
 	rc = crypto_blkcipher_setkey(
 		s->desc.tfm,
 		s->auth_tok->token.password.session_key_encryption_key,
@@ -1316,8 +1328,10 @@ parse_tag_3_packet(struct ecryptfs_crypt_stat *crypt_stat,
 		rc = -EINVAL;
 		goto out_free;
 	}
-	ecryptfs_cipher_code_to_string(crypt_stat->cipher,
-				       (u16)data[(*packet_size)]);
+	rc = ecryptfs_cipher_code_to_string(crypt_stat->cipher,
+					    (u16)data[(*packet_size)]);
+	if (rc)
+		goto out_free;
 	/* A little extra work to differentiate among the AES key
 	 * sizes; see RFC2440 */
 	switch(data[(*packet_size)++]) {
@@ -1328,7 +1342,9 @@ parse_tag_3_packet(struct ecryptfs_crypt_stat *crypt_stat,
 		crypt_stat->key_size =
 			(*new_auth_tok)->session_key.encrypted_key_size;
 	}
-	ecryptfs_init_crypt_ctx(crypt_stat);
+	rc = ecryptfs_init_crypt_ctx(crypt_stat);
+	if (rc)
+		goto out_free;
 	if (unlikely(data[(*packet_size)++] != 0x03)) {
 		printk(KERN_WARNING "Only S2K ID 3 is currently supported\n");
 		rc = -ENOSYS;
@@ -2366,21 +2382,18 @@ struct kmem_cache *ecryptfs_key_sig_cache;
 int ecryptfs_add_keysig(struct ecryptfs_crypt_stat *crypt_stat, char *sig)
 {
 	struct ecryptfs_key_sig *new_key_sig;
-	int rc = 0;
 
 	new_key_sig = kmem_cache_alloc(ecryptfs_key_sig_cache, GFP_KERNEL);
 	if (!new_key_sig) {
-		rc = -ENOMEM;
 		printk(KERN_ERR
 		       "Error allocating from ecryptfs_key_sig_cache\n");
-		goto out;
+		return -ENOMEM;
 	}
 	memcpy(new_key_sig->keysig, sig, ECRYPTFS_SIG_SIZE_HEX);
-	mutex_lock(&crypt_stat->keysig_list_mutex);
+	/* Caller must hold keysig_list_mutex */
 	list_add(&new_key_sig->crypt_stat_list, &crypt_stat->keysig_list);
-	mutex_unlock(&crypt_stat->keysig_list_mutex);
-out:
-	return rc;
+
+	return 0;
 }
 
 struct kmem_cache *ecryptfs_global_auth_tok_cache;
diff --git a/fs/ecryptfs/kthread.c b/fs/ecryptfs/kthread.c
index c6d7a4d748a0..e14cf7e588db 100644
--- a/fs/ecryptfs/kthread.c
+++ b/fs/ecryptfs/kthread.c
@@ -136,6 +136,7 @@ int ecryptfs_privileged_open(struct file **lower_file,
 			     const struct cred *cred)
 {
 	struct ecryptfs_open_req *req;
+	int flags = O_LARGEFILE;
 	int rc = 0;
 
 	/* Corresponding dput() and mntput() are done when the
@@ -143,10 +144,14 @@ int ecryptfs_privileged_open(struct file **lower_file,
 	 * destroyed. */
 	dget(lower_dentry);
 	mntget(lower_mnt);
-	(*lower_file) = dentry_open(lower_dentry, lower_mnt,
-				    (O_RDWR | O_LARGEFILE), cred);
+	flags |= IS_RDONLY(lower_dentry->d_inode) ? O_RDONLY : O_RDWR;
+	(*lower_file) = dentry_open(lower_dentry, lower_mnt, flags, cred);
 	if (!IS_ERR(*lower_file))
 		goto out;
+	if (flags & O_RDONLY) {
+		rc = PTR_ERR((*lower_file));
+		goto out;
+	}
 	req = kmem_cache_alloc(ecryptfs_open_req_cache, GFP_KERNEL);
 	if (!req) {
 		rc = -ENOMEM;
@@ -180,21 +185,8 @@ int ecryptfs_privileged_open(struct file **lower_file,
 		       __func__);
 		goto out_unlock;
 	}
-	if (IS_ERR(*req->lower_file)) {
+	if (IS_ERR(*req->lower_file))
 		rc = PTR_ERR(*req->lower_file);
-		dget(lower_dentry);
-		mntget(lower_mnt);
-		(*lower_file) = dentry_open(lower_dentry, lower_mnt,
-					    (O_RDONLY | O_LARGEFILE), cred);
-		if (IS_ERR(*lower_file)) {
-			rc = PTR_ERR(*req->lower_file);
-			(*lower_file) = NULL;
-			printk(KERN_WARNING "%s: Error attempting privileged "
-			       "open of lower file with either RW or RO "
-			       "perms; rc = [%d]. Giving up.\n",
-			       __func__, rc);
-		}
-	}
 out_unlock:
 	mutex_unlock(&req->mux);
 out_free:
diff --git a/fs/ecryptfs/main.c b/fs/ecryptfs/main.c
index 9f0aa9883c28..c6ac85d6c701 100644
--- a/fs/ecryptfs/main.c
+++ b/fs/ecryptfs/main.c
@@ -35,6 +35,7 @@
 #include <linux/key.h>
 #include <linux/parser.h>
 #include <linux/fs_stack.h>
+#include <linux/ima.h>
 #include "ecryptfs_kernel.h"
 
 /**
@@ -118,6 +119,7 @@ int ecryptfs_init_persistent_file(struct dentry *ecryptfs_dentry)
 	const struct cred *cred = current_cred();
 	struct ecryptfs_inode_info *inode_info =
 		ecryptfs_inode_to_private(ecryptfs_dentry->d_inode);
+	int opened_lower_file = 0;
 	int rc = 0;
 
 	mutex_lock(&inode_info->lower_file_mutex);
@@ -129,15 +131,17 @@ int ecryptfs_init_persistent_file(struct dentry *ecryptfs_dentry)
 		lower_dentry = ecryptfs_dentry_to_lower(ecryptfs_dentry);
 		rc = ecryptfs_privileged_open(&inode_info->lower_file,
 					      lower_dentry, lower_mnt, cred);
-		if (rc || IS_ERR(inode_info->lower_file)) {
+		if (rc) {
 			printk(KERN_ERR "Error opening lower persistent file "
 			       "for lower_dentry [0x%p] and lower_mnt [0x%p]; "
 			       "rc = [%d]\n", lower_dentry, lower_mnt, rc);
-			rc = PTR_ERR(inode_info->lower_file);
 			inode_info->lower_file = NULL;
-		}
+		} else
+			opened_lower_file = 1;
 	}
 	mutex_unlock(&inode_info->lower_file_mutex);
+	if (opened_lower_file)
+		ima_counts_get(inode_info->lower_file);
 	return rc;
 }
 
diff --git a/fs/ecryptfs/mmap.c b/fs/ecryptfs/mmap.c
index 5c6bab9786e3..df4ce99d0597 100644
--- a/fs/ecryptfs/mmap.c
+++ b/fs/ecryptfs/mmap.c
@@ -396,9 +396,11 @@ static int ecryptfs_write_inode_size_to_header(struct inode *ecryptfs_inode)
 	rc = ecryptfs_write_lower(ecryptfs_inode, file_size_virt, 0,
 				  sizeof(u64));
 	kfree(file_size_virt);
-	if (rc)
+	if (rc < 0)
 		printk(KERN_ERR "%s: Error writing file size to header; "
 		       "rc = [%d]\n", __func__, rc);
+	else
+		rc = 0;
 out:
 	return rc;
 }
@@ -545,7 +547,7 @@ static sector_t ecryptfs_bmap(struct address_space *mapping, sector_t block)
 	return rc;
 }
 
-struct address_space_operations ecryptfs_aops = {
+const struct address_space_operations ecryptfs_aops = {
 	.writepage = ecryptfs_writepage,
 	.readpage = ecryptfs_readpage,
 	.write_begin = ecryptfs_write_begin,
diff --git a/fs/ecryptfs/read_write.c b/fs/ecryptfs/read_write.c
index a137c6ea2fee..0cc4fafd6552 100644
--- a/fs/ecryptfs/read_write.c
+++ b/fs/ecryptfs/read_write.c
@@ -34,15 +34,14 @@
  *
  * Write data to the lower file.
  *
- * Returns zero on success; non-zero on error
+ * Returns bytes written on success; less than zero on error
  */
 int ecryptfs_write_lower(struct inode *ecryptfs_inode, char *data,
 			 loff_t offset, size_t size)
 {
 	struct ecryptfs_inode_info *inode_info;
-	ssize_t octets_written;
 	mm_segment_t fs_save;
-	int rc = 0;
+	ssize_t rc;
 
 	inode_info = ecryptfs_inode_to_private(ecryptfs_inode);
 	mutex_lock(&inode_info->lower_file_mutex);
@@ -50,14 +49,9 @@ int ecryptfs_write_lower(struct inode *ecryptfs_inode, char *data,
 	inode_info->lower_file->f_pos = offset;
 	fs_save = get_fs();
 	set_fs(get_ds());
-	octets_written = vfs_write(inode_info->lower_file, data, size,
-				   &inode_info->lower_file->f_pos);
+	rc = vfs_write(inode_info->lower_file, data, size,
+		       &inode_info->lower_file->f_pos);
 	set_fs(fs_save);
-	if (octets_written < 0) {
-		printk(KERN_ERR "%s: octets_written = [%td]; "
-		       "expected [%td]\n", __func__, octets_written, size);
-		rc = -EINVAL;
-	}
 	mutex_unlock(&inode_info->lower_file_mutex);
 	mark_inode_dirty_sync(ecryptfs_inode);
 	return rc;
@@ -91,6 +85,8 @@ int ecryptfs_write_lower_page_segment(struct inode *ecryptfs_inode,
 		  + offset_in_page);
 	virt = kmap(page_for_lower);
 	rc = ecryptfs_write_lower(ecryptfs_inode, virt, offset, size);
+	if (rc > 0)
+		rc = 0;
 	kunmap(page_for_lower);
 	return rc;
 }
@@ -229,30 +225,24 @@ out:
  * Read @size bytes of data at byte offset @offset from the lower
  * inode into memory location @data.
  *
- * Returns zero on success; non-zero on error
+ * Returns bytes read on success; 0 on EOF; less than zero on error
  */
 int ecryptfs_read_lower(char *data, loff_t offset, size_t size,
 			struct inode *ecryptfs_inode)
 {
 	struct ecryptfs_inode_info *inode_info =
 		ecryptfs_inode_to_private(ecryptfs_inode);
-	ssize_t octets_read;
 	mm_segment_t fs_save;
-	int rc = 0;
+	ssize_t rc;
 
 	mutex_lock(&inode_info->lower_file_mutex);
 	BUG_ON(!inode_info->lower_file);
 	inode_info->lower_file->f_pos = offset;
 	fs_save = get_fs();
 	set_fs(get_ds());
-	octets_read = vfs_read(inode_info->lower_file, data, size,
-			       &inode_info->lower_file->f_pos);
+	rc = vfs_read(inode_info->lower_file, data, size,
+		      &inode_info->lower_file->f_pos);
 	set_fs(fs_save);
-	if (octets_read < 0) {
-		printk(KERN_ERR "%s: octets_read = [%td]; "
-		       "expected [%td]\n", __func__, octets_read, size);
-		rc = -EINVAL;
-	}
 	mutex_unlock(&inode_info->lower_file_mutex);
 	return rc;
 }
@@ -284,6 +274,8 @@ int ecryptfs_read_lower_page_segment(struct page *page_for_ecryptfs,
 	offset = ((((loff_t)page_index) << PAGE_CACHE_SHIFT) + offset_in_page);
 	virt = kmap(page_for_ecryptfs);
 	rc = ecryptfs_read_lower(virt, offset, size, ecryptfs_inode);
+	if (rc > 0)
+		rc = 0;
 	kunmap(page_for_ecryptfs);
 	flush_dcache_page(page_for_ecryptfs);
 	return rc;
diff --git a/fs/ecryptfs/super.c b/fs/ecryptfs/super.c
index 12d649602d3a..b15a43a80ab7 100644
--- a/fs/ecryptfs/super.c
+++ b/fs/ecryptfs/super.c
@@ -77,7 +77,6 @@ static void ecryptfs_destroy_inode(struct inode *inode)
 	struct ecryptfs_inode_info *inode_info;
 
 	inode_info = ecryptfs_inode_to_private(inode);
-	mutex_lock(&inode_info->lower_file_mutex);
 	if (inode_info->lower_file) {
 		struct dentry *lower_dentry =
 			inode_info->lower_file->f_dentry;
@@ -89,7 +88,6 @@ static void ecryptfs_destroy_inode(struct inode *inode)
 			d_drop(lower_dentry);
 		}
 	}
-	mutex_unlock(&inode_info->lower_file_mutex);
 	ecryptfs_destroy_crypt_stat(&inode_info->crypt_stat);
 	kmem_cache_free(ecryptfs_inode_info_cache, inode_info);
 }
diff --git a/fs/eventfd.c b/fs/eventfd.c
index 31d12de83a2a..8b47e4200e65 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -68,11 +68,16 @@ int eventfd_signal(struct eventfd_ctx *ctx, int n)
 }
 EXPORT_SYMBOL_GPL(eventfd_signal);
 
+static void eventfd_free_ctx(struct eventfd_ctx *ctx)
+{
+	kfree(ctx);
+}
+
 static void eventfd_free(struct kref *kref)
 {
 	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
 
-	kfree(ctx);
+	eventfd_free_ctx(ctx);
 }
 
 /**
@@ -298,9 +303,23 @@ struct eventfd_ctx *eventfd_ctx_fileget(struct file *file)
 }
 EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
 
-SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
+/**
+ * eventfd_file_create - Creates an eventfd file pointer.
+ * @count: Initial eventfd counter value.
+ * @flags: Flags for the eventfd file.
+ *
+ * This function creates an eventfd file pointer, w/out installing it into
+ * the fd table. This is useful when the eventfd file is used during the
+ * initialization of data structures that require extra setup after the eventfd
+ * creation. So the eventfd creation is split into the file pointer creation
+ * phase, and the file descriptor installation phase.
+ * In this way races with userspace closing the newly installed file descriptor
+ * can be avoided.
+ * Returns an eventfd file pointer, or a proper error pointer.
+ */
+struct file *eventfd_file_create(unsigned int count, int flags)
 {
-	int fd;
+	struct file *file;
 	struct eventfd_ctx *ctx;
 
 	/* Check the EFD_* constants for consistency.  */
@@ -308,26 +327,48 @@ SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
 	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
 
 	if (flags & ~EFD_FLAGS_SET)
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 
 	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
 	if (!ctx)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	kref_init(&ctx->kref);
 	init_waitqueue_head(&ctx->wqh);
 	ctx->count = count;
 	ctx->flags = flags;
 
-	/*
-	 * When we call this, the initialization must be complete, since
-	 * anon_inode_getfd() will install the fd.
-	 */
-	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
-			      flags & EFD_SHARED_FCNTL_FLAGS);
-	if (fd < 0)
-		kfree(ctx);
+	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx,
+				  flags & EFD_SHARED_FCNTL_FLAGS);
+	if (IS_ERR(file))
+		eventfd_free_ctx(ctx);
+
+	return file;
+}
+
+SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
+{
+	int fd, error;
+	struct file *file;
+
+	error = get_unused_fd_flags(flags & EFD_SHARED_FCNTL_FLAGS);
+	if (error < 0)
+		return error;
+	fd = error;
+
+	file = eventfd_file_create(count, flags);
+	if (IS_ERR(file)) {
+		error = PTR_ERR(file);
+		goto err_put_unused_fd;
+	}
+	fd_install(fd, file);
+
 	return fd;
+
+err_put_unused_fd:
+	put_unused_fd(fd);
+
+	return error;
 }
 
 SYSCALL_DEFINE1(eventfd, unsigned int, count)
diff --git a/fs/exec.c b/fs/exec.c
index 172ceb6edde4..d49be6bc1793 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -33,7 +33,7 @@
 #include <linux/string.h>
 #include <linux/init.h>
 #include <linux/pagemap.h>
-#include <linux/perf_counter.h>
+#include <linux/perf_event.h>
 #include <linux/highmem.h>
 #include <linux/spinlock.h>
 #include <linux/key.h>
@@ -55,6 +55,7 @@
 #include <linux/kmod.h>
 #include <linux/fsnotify.h>
 #include <linux/fs_struct.h>
+#include <linux/pipe_fs_i.h>
 
 #include <asm/uaccess.h>
 #include <asm/mmu_context.h>
@@ -63,6 +64,7 @@
 
 int core_uses_pid;
 char core_pattern[CORENAME_MAX_SIZE] = "core";
+unsigned int core_pipe_limit;
 int suid_dumpable = 0;
 
 /* The maximal length of core_pattern is also specified in sysctl.c */
@@ -845,6 +847,9 @@ static int de_thread(struct task_struct *tsk)
 	sig->notify_count = 0;
 
 no_thread_group:
+	if (current->mm)
+		setmax_mm_hiwater_rss(&sig->maxrss, current->mm);
+
 	exit_itimers(sig);
 	flush_itimer_signals();
 
@@ -923,7 +928,7 @@ void set_task_comm(struct task_struct *tsk, char *buf)
 	task_lock(tsk);
 	strlcpy(tsk->comm, buf, sizeof(tsk->comm));
 	task_unlock(tsk);
-	perf_counter_comm(tsk);
+	perf_event_comm(tsk);
 }
 
 int flush_old_exec(struct linux_binprm * bprm)
@@ -997,7 +1002,7 @@ int flush_old_exec(struct linux_binprm * bprm)
 	 * security domain:
 	 */
 	if (!get_dumpable(current->mm))
-		perf_counter_exit_task(current);
+		perf_event_exit_task(current);
 
 	/* An exec changes our domain. We are no longer part of the thread
 	   group */
@@ -1354,6 +1359,8 @@ int do_execve(char * filename,
 	if (retval < 0)
 		goto out;
 
+	current->stack_start = current->mm->start_stack;
+
 	/* execve succeeded */
 	current->fs->in_exec = 0;
 	current->in_execve = 0;
@@ -1388,18 +1395,16 @@ out_ret:
 	return retval;
 }
 
-int set_binfmt(struct linux_binfmt *new)
+void set_binfmt(struct linux_binfmt *new)
 {
-	struct linux_binfmt *old = current->binfmt;
+	struct mm_struct *mm = current->mm;
 
-	if (new) {
-		if (!try_module_get(new->module))
-			return -1;
-	}
-	current->binfmt = new;
-	if (old)
-		module_put(old->module);
-	return 0;
+	if (mm->binfmt)
+		module_put(mm->binfmt->module);
+
+	mm->binfmt = new;
+	if (new)
+		__module_get(new->module);
 }
 
 EXPORT_SYMBOL(set_binfmt);
@@ -1723,6 +1728,29 @@ int get_dumpable(struct mm_struct *mm)
 	return (ret >= 2) ? 2 : ret;
 }
 
+static void wait_for_dump_helpers(struct file *file)
+{
+	struct pipe_inode_info *pipe;
+
+	pipe = file->f_path.dentry->d_inode->i_pipe;
+
+	pipe_lock(pipe);
+	pipe->readers++;
+	pipe->writers--;
+
+	while ((pipe->readers > 1) && (!signal_pending(current))) {
+		wake_up_interruptible_sync(&pipe->wait);
+		kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
+		pipe_wait(pipe);
+	}
+
+	pipe->readers--;
+	pipe->writers++;
+	pipe_unlock(pipe);
+
+}
+
+
 void do_coredump(long signr, int exit_code, struct pt_regs *regs)
 {
 	struct core_state core_state;
@@ -1739,11 +1767,12 @@ void do_coredump(long signr, int exit_code, struct pt_regs *regs)
 	unsigned long core_limit = current->signal->rlim[RLIMIT_CORE].rlim_cur;
 	char **helper_argv = NULL;
 	int helper_argc = 0;
-	char *delimit;
+	int dump_count = 0;
+	static atomic_t core_dump_count = ATOMIC_INIT(0);
 
 	audit_core_dumps(signr);
 
-	binfmt = current->binfmt;
+	binfmt = mm->binfmt;
 	if (!binfmt || !binfmt->core_dump)
 		goto fail;
 
@@ -1794,54 +1823,63 @@ void do_coredump(long signr, int exit_code, struct pt_regs *regs)
  	lock_kernel();
 	ispipe = format_corename(corename, signr);
 	unlock_kernel();
-	/*
-	 * Don't bother to check the RLIMIT_CORE value if core_pattern points
-	 * to a pipe.  Since we're not writing directly to the filesystem
-	 * RLIMIT_CORE doesn't really apply, as no actual core file will be
-	 * created unless the pipe reader choses to write out the core file
-	 * at which point file size limits and permissions will be imposed
-	 * as it does with any other process
-	 */
+
 	if ((!ispipe) && (core_limit < binfmt->min_coredump))
 		goto fail_unlock;
 
  	if (ispipe) {
+		if (core_limit == 0) {
+			/*
+			 * Normally core limits are irrelevant to pipes, since
+			 * we're not writing to the file system, but we use
+			 * core_limit of 0 here as a speacial value. Any
+			 * non-zero limit gets set to RLIM_INFINITY below, but
+			 * a limit of 0 skips the dump.  This is a consistent
+			 * way to catch recursive crashes.  We can still crash
+			 * if the core_pattern binary sets RLIM_CORE =  !0
+			 * but it runs as root, and can do lots of stupid things
+			 * Note that we use task_tgid_vnr here to grab the pid
+			 * of the process group leader.  That way we get the
+			 * right pid if a thread in a multi-threaded
+			 * core_pattern process dies.
+			 */
+			printk(KERN_WARNING
+				"Process %d(%s) has RLIMIT_CORE set to 0\n",
+				task_tgid_vnr(current), current->comm);
+			printk(KERN_WARNING "Aborting core\n");
+			goto fail_unlock;
+		}
+
+		dump_count = atomic_inc_return(&core_dump_count);
+		if (core_pipe_limit && (core_pipe_limit < dump_count)) {
+			printk(KERN_WARNING "Pid %d(%s) over core_pipe_limit\n",
+			       task_tgid_vnr(current), current->comm);
+			printk(KERN_WARNING "Skipping core dump\n");
+			goto fail_dropcount;
+		}
+
 		helper_argv = argv_split(GFP_KERNEL, corename+1, &helper_argc);
 		if (!helper_argv) {
 			printk(KERN_WARNING "%s failed to allocate memory\n",
 			       __func__);
-			goto fail_unlock;
-		}
-		/* Terminate the string before the first option */
-		delimit = strchr(corename, ' ');
-		if (delimit)
-			*delimit = '\0';
-		delimit = strrchr(helper_argv[0], '/');
-		if (delimit)
-			delimit++;
-		else
-			delimit = helper_argv[0];
-		if (!strcmp(delimit, current->comm)) {
-			printk(KERN_NOTICE "Recursive core dump detected, "
-					"aborting\n");
-			goto fail_unlock;
+			goto fail_dropcount;
 		}
 
 		core_limit = RLIM_INFINITY;
 
 		/* SIGPIPE can happen, but it's just never processed */
- 		if (call_usermodehelper_pipe(corename+1, helper_argv, NULL,
+		if (call_usermodehelper_pipe(helper_argv[0], helper_argv, NULL,
 				&file)) {
  			printk(KERN_INFO "Core dump to %s pipe failed\n",
 			       corename);
- 			goto fail_unlock;
+			goto fail_dropcount;
  		}
  	} else
  		file = filp_open(corename,
 				 O_CREAT | 2 | O_NOFOLLOW | O_LARGEFILE | flag,
 				 0600);
 	if (IS_ERR(file))
-		goto fail_unlock;
+		goto fail_dropcount;
 	inode = file->f_path.dentry->d_inode;
 	if (inode->i_nlink > 1)
 		goto close_fail;	/* multiple links - don't dump */
@@ -1870,7 +1908,12 @@ void do_coredump(long signr, int exit_code, struct pt_regs *regs)
 	if (retval)
 		current->signal->group_exit_code |= 0x80;
 close_fail:
+	if (ispipe && core_pipe_limit)
+		wait_for_dump_helpers(file);
 	filp_close(file, NULL);
+fail_dropcount:
+	if (dump_count)
+		atomic_dec(&core_dump_count);
 fail_unlock:
 	if (helper_argv)
 		argv_free(helper_argv);
diff --git a/fs/exofs/super.c b/fs/exofs/super.c
index 5ab10c3bbebe..9f500dec3b59 100644
--- a/fs/exofs/super.c
+++ b/fs/exofs/super.c
@@ -214,7 +214,6 @@ int exofs_sync_fs(struct super_block *sb, int wait)
 	}
 
 	lock_super(sb);
-	lock_kernel();
 	sbi = sb->s_fs_info;
 	fscb->s_nextid = cpu_to_le64(sbi->s_nextid);
 	fscb->s_numfiles = cpu_to_le32(sbi->s_numfiles);
@@ -245,7 +244,6 @@ int exofs_sync_fs(struct super_block *sb, int wait)
 out:
 	if (or)
 		osd_end_request(or);
-	unlock_kernel();
 	unlock_super(sb);
 	kfree(fscb);
 	return ret;
@@ -268,8 +266,6 @@ static void exofs_put_super(struct super_block *sb)
 	int num_pend;
 	struct exofs_sb_info *sbi = sb->s_fs_info;
 
-	lock_kernel();
-
 	if (sb->s_dirt)
 		exofs_write_super(sb);
 
@@ -286,8 +282,6 @@ static void exofs_put_super(struct super_block *sb)
 	osduld_put_device(sbi->s_dev);
 	kfree(sb->s_fs_info);
 	sb->s_fs_info = NULL;
-
-	unlock_kernel();
 }
 
 /*
diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c
index 1c1638f873a4..ade634076d0a 100644
--- a/fs/ext2/inode.c
+++ b/fs/ext2/inode.c
@@ -819,6 +819,7 @@ const struct address_space_operations ext2_aops = {
 	.writepages		= ext2_writepages,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate	= block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 const struct address_space_operations ext2_aops_xip = {
@@ -837,6 +838,7 @@ const struct address_space_operations ext2_nobh_aops = {
 	.direct_IO		= ext2_direct_IO,
 	.writepages		= ext2_writepages,
 	.migratepage		= buffer_migrate_page,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 /*
diff --git a/fs/ext2/namei.c b/fs/ext2/namei.c
index 23701f289e98..dd7175ce5606 100644
--- a/fs/ext2/namei.c
+++ b/fs/ext2/namei.c
@@ -70,7 +70,7 @@ static struct dentry *ext2_lookup(struct inode * dir, struct dentry *dentry, str
 			if (PTR_ERR(inode) == -ESTALE) {
 				ext2_error(dir->i_sb, __func__,
 						"deleted inode referenced: %lu",
-						ino);
+						(unsigned long) ino);
 				return ERR_PTR(-EIO);
 			} else {
 				return ERR_CAST(inode);
diff --git a/fs/ext2/xip.c b/fs/ext2/xip.c
index b72b85884223..c18fbf3e4068 100644
--- a/fs/ext2/xip.c
+++ b/fs/ext2/xip.c
@@ -20,7 +20,7 @@ __inode_direct_access(struct inode *inode, sector_t block,
 		      void **kaddr, unsigned long *pfn)
 {
 	struct block_device *bdev = inode->i_sb->s_bdev;
-	struct block_device_operations *ops = bdev->bd_disk->fops;
+	const struct block_device_operations *ops = bdev->bd_disk->fops;
 	sector_t sector;
 
 	sector = block * (PAGE_SIZE / 512); /* ext2 block to bdev sector */
diff --git a/fs/ext3/fsync.c b/fs/ext3/fsync.c
index d33634119e17..451d166bbe93 100644
--- a/fs/ext3/fsync.c
+++ b/fs/ext3/fsync.c
@@ -23,6 +23,7 @@
  */
 
 #include <linux/time.h>
+#include <linux/blkdev.h>
 #include <linux/fs.h>
 #include <linux/sched.h>
 #include <linux/writeback.h>
@@ -73,7 +74,7 @@ int ext3_sync_file(struct file * file, struct dentry *dentry, int datasync)
 	}
 
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
-		goto out;
+		goto flush;
 
 	/*
 	 * The VFS has written the file data.  If the inode is unaltered
@@ -85,7 +86,16 @@ int ext3_sync_file(struct file * file, struct dentry *dentry, int datasync)
 			.nr_to_write = 0, /* sys_fsync did this */
 		};
 		ret = sync_inode(inode, &wbc);
+		goto out;
 	}
+flush:
+	/*
+	 * In case we didn't commit a transaction, we have to flush
+	 * disk caches manually so that data really is on persistent
+	 * storage
+	 */
+	if (test_opt(inode->i_sb, BARRIER))
+		blkdev_issue_flush(inode->i_sb->s_bdev, NULL);
 out:
 	return ret;
 }
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index b49908a167ae..acf1b1423327 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -172,10 +172,21 @@ static int try_to_extend_transaction(handle_t *handle, struct inode *inode)
  * so before we call here everything must be consistently dirtied against
  * this transaction.
  */
-static int ext3_journal_test_restart(handle_t *handle, struct inode *inode)
+static int truncate_restart_transaction(handle_t *handle, struct inode *inode)
 {
+	int ret;
+
 	jbd_debug(2, "restarting handle %p\n", handle);
-	return ext3_journal_restart(handle, blocks_for_truncate(inode));
+	/*
+	 * Drop truncate_mutex to avoid deadlock with ext3_get_blocks_handle
+	 * At this moment, get_block can be called only for blocks inside
+	 * i_size since page cache has been already dropped and writes are
+	 * blocked by i_mutex. So we can safely drop the truncate_mutex.
+	 */
+	mutex_unlock(&EXT3_I(inode)->truncate_mutex);
+	ret = ext3_journal_restart(handle, blocks_for_truncate(inode));
+	mutex_lock(&EXT3_I(inode)->truncate_mutex);
+	return ret;
 }
 
 /*
@@ -1819,6 +1830,7 @@ static const struct address_space_operations ext3_ordered_aops = {
 	.direct_IO		= ext3_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 static const struct address_space_operations ext3_writeback_aops = {
@@ -1834,6 +1846,7 @@ static const struct address_space_operations ext3_writeback_aops = {
 	.direct_IO		= ext3_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 static const struct address_space_operations ext3_journalled_aops = {
@@ -1848,6 +1861,7 @@ static const struct address_space_operations ext3_journalled_aops = {
 	.invalidatepage		= ext3_invalidatepage,
 	.releasepage		= ext3_releasepage,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 void ext3_set_aops(struct inode *inode)
@@ -2072,7 +2086,7 @@ static void ext3_clear_blocks(handle_t *handle, struct inode *inode,
 			ext3_journal_dirty_metadata(handle, bh);
 		}
 		ext3_mark_inode_dirty(handle, inode);
-		ext3_journal_test_restart(handle, inode);
+		truncate_restart_transaction(handle, inode);
 		if (bh) {
 			BUFFER_TRACE(bh, "retaking write access");
 			ext3_journal_get_write_access(handle, bh);
@@ -2282,7 +2296,7 @@ static void ext3_free_branches(handle_t *handle, struct inode *inode,
 				return;
 			if (try_to_extend_transaction(handle, inode)) {
 				ext3_mark_inode_dirty(handle, inode);
-				ext3_journal_test_restart(handle, inode);
+				truncate_restart_transaction(handle, inode);
 			}
 
 			ext3_free_blocks(handle, inode, nr, 1);
@@ -2892,6 +2906,10 @@ static int ext3_do_update_inode(handle_t *handle,
 	struct buffer_head *bh = iloc->bh;
 	int err = 0, rc, block;
 
+again:
+	/* we can't allow multiple procs in here at once, its a bit racey */
+	lock_buffer(bh);
+
 	/* For fields not not tracking in the in-memory inode,
 	 * initialise them to zero for new inodes. */
 	if (ei->i_state & EXT3_STATE_NEW)
@@ -2951,16 +2969,20 @@ static int ext3_do_update_inode(handle_t *handle,
 			       /* If this is the first large file
 				* created, add a flag to the superblock.
 				*/
+				unlock_buffer(bh);
 				err = ext3_journal_get_write_access(handle,
 						EXT3_SB(sb)->s_sbh);
 				if (err)
 					goto out_brelse;
+
 				ext3_update_dynamic_rev(sb);
 				EXT3_SET_RO_COMPAT_FEATURE(sb,
 					EXT3_FEATURE_RO_COMPAT_LARGE_FILE);
 				handle->h_sync = 1;
 				err = ext3_journal_dirty_metadata(handle,
 						EXT3_SB(sb)->s_sbh);
+				/* get our lock and start over */
+				goto again;
 			}
 		}
 	}
@@ -2983,6 +3005,7 @@ static int ext3_do_update_inode(handle_t *handle,
 		raw_inode->i_extra_isize = cpu_to_le16(ei->i_extra_isize);
 
 	BUFFER_TRACE(bh, "call ext3_journal_dirty_metadata");
+	unlock_buffer(bh);
 	rc = ext3_journal_dirty_metadata(handle, bh);
 	if (!err)
 		err = rc;
diff --git a/fs/ext3/super.c b/fs/ext3/super.c
index a8d80a7f1105..7a520a862f49 100644
--- a/fs/ext3/super.c
+++ b/fs/ext3/super.c
@@ -720,7 +720,7 @@ static ssize_t ext3_quota_read(struct super_block *sb, int type, char *data,
 static ssize_t ext3_quota_write(struct super_block *sb, int type,
 				const char *data, size_t len, loff_t off);
 
-static struct dquot_operations ext3_quota_operations = {
+static const struct dquot_operations ext3_quota_operations = {
 	.initialize	= dquot_initialize,
 	.drop		= dquot_drop,
 	.alloc_space	= dquot_alloc_space,
@@ -737,7 +737,7 @@ static struct dquot_operations ext3_quota_operations = {
 	.destroy_dquot	= dquot_destroy,
 };
 
-static struct quotactl_ops ext3_qctl_operations = {
+static const struct quotactl_ops ext3_qctl_operations = {
 	.quota_on	= ext3_quota_on,
 	.quota_off	= vfs_quota_off,
 	.quota_sync	= vfs_quota_sync,
@@ -2321,7 +2321,18 @@ static int ext3_commit_super(struct super_block *sb,
 
 	if (!sbh)
 		return error;
-	es->s_wtime = cpu_to_le32(get_seconds());
+	/*
+	 * If the file system is mounted read-only, don't update the
+	 * superblock write time.  This avoids updating the superblock
+	 * write time when we are mounting the root file system
+	 * read/only but we need to replay the journal; at that point,
+	 * for people who are east of GMT and who make their clock
+	 * tick in localtime for Windows bug-for-bug compatibility,
+	 * the clock is set in the future, and this will cause e2fsck
+	 * to complain and force a full file system check.
+	 */
+	if (!(sb->s_flags & MS_RDONLY))
+		es->s_wtime = cpu_to_le32(get_seconds());
 	es->s_free_blocks_count = cpu_to_le32(ext3_count_free_blocks(sb));
 	es->s_free_inodes_count = cpu_to_le32(ext3_count_free_inodes(sb));
 	BUFFER_TRACE(sbh, "marking dirty");
diff --git a/fs/ext4/Kconfig b/fs/ext4/Kconfig
index 418b6f3b0ae8..9f2d45d75b1a 100644
--- a/fs/ext4/Kconfig
+++ b/fs/ext4/Kconfig
@@ -26,20 +26,6 @@ config EXT4_FS
 
 	  If unsure, say N.
 
-config EXT4DEV_COMPAT
-	bool "Enable ext4dev compatibility"
-	depends on EXT4_FS
-	help
-	  Starting with 2.6.28, the name of the ext4 filesystem was
-	  renamed from ext4dev to ext4.  Unfortunately there are some
-	  legacy userspace programs (such as klibc's fstype) have
-	  "ext4dev" hardcoded.
-
-	  To enable backwards compatibility so that systems that are
-	  still expecting to mount ext4 filesystems using ext4dev,
-	  chose Y here.   This feature will go away by 2.6.31, so
-	  please arrange to get your userspace programs fixed!
-
 config EXT4_FS_XATTR
 	bool "Ext4 extended attributes"
 	depends on EXT4_FS
@@ -77,3 +63,12 @@ config EXT4_FS_SECURITY
 
 	  If you are not using a security module that requires using
 	  extended attributes for file security labels, say N.
+
+config EXT4_DEBUG
+	bool "EXT4 debugging support"
+	depends on EXT4_FS
+	help
+	  Enables run-time debugging support for the ext4 filesystem.
+
+	  If you select Y here, then you will be able to turn on debugging
+	  with a command such as "echo 1 > /sys/kernel/debug/ext4/mballoc-debug"
diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index e2126d70dff5..1d0418980f8d 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -478,7 +478,7 @@ void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
 	 * new bitmap information
 	 */
 	set_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &(grp->bb_state));
-	ext4_mb_update_group_info(grp, blocks_freed);
+	grp->bb_free += blocks_freed;
 	up_write(&grp->alloc_sem);
 
 	/* We dirtied the bitmap block */
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 9714db393efe..984ca0cb38c3 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -65,29 +65,37 @@ typedef __u32 ext4_lblk_t;
 /* data type for block group number */
 typedef unsigned int ext4_group_t;
 
+/*
+ * Flags used in mballoc's allocation_context flags field.  
+ *
+ * Also used to show what's going on for debugging purposes when the
+ * flag field is exported via the traceport interface
+ */
 
 /* prefer goal again. length */
-#define EXT4_MB_HINT_MERGE		1
+#define EXT4_MB_HINT_MERGE		0x0001
 /* blocks already reserved */
-#define EXT4_MB_HINT_RESERVED		2
+#define EXT4_MB_HINT_RESERVED		0x0002
 /* metadata is being allocated */
-#define EXT4_MB_HINT_METADATA		4
+#define EXT4_MB_HINT_METADATA		0x0004
 /* first blocks in the file */
-#define EXT4_MB_HINT_FIRST		8
+#define EXT4_MB_HINT_FIRST		0x0008
 /* search for the best chunk */
-#define EXT4_MB_HINT_BEST		16
+#define EXT4_MB_HINT_BEST		0x0010
 /* data is being allocated */
-#define EXT4_MB_HINT_DATA		32
+#define EXT4_MB_HINT_DATA		0x0020
 /* don't preallocate (for tails) */
-#define EXT4_MB_HINT_NOPREALLOC		64
+#define EXT4_MB_HINT_NOPREALLOC		0x0040
 /* allocate for locality group */
-#define EXT4_MB_HINT_GROUP_ALLOC	128
+#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
 /* allocate goal blocks or none */
-#define EXT4_MB_HINT_GOAL_ONLY		256
+#define EXT4_MB_HINT_GOAL_ONLY		0x0100
 /* goal is meaningful */
-#define EXT4_MB_HINT_TRY_GOAL		512
+#define EXT4_MB_HINT_TRY_GOAL		0x0200
 /* blocks already pre-reserved by delayed allocation */
-#define EXT4_MB_DELALLOC_RESERVED      1024
+#define EXT4_MB_DELALLOC_RESERVED	0x0400
+/* We are doing stream allocation */
+#define EXT4_MB_STREAM_ALLOC		0x0800
 
 
 struct ext4_allocation_request {
@@ -112,6 +120,31 @@ struct ext4_allocation_request {
 };
 
 /*
+ * For delayed allocation tracking
+ */
+struct mpage_da_data {
+	struct inode *inode;
+	sector_t b_blocknr;		/* start block number of extent */
+	size_t b_size;			/* size of extent */
+	unsigned long b_state;		/* state of the extent */
+	unsigned long first_page, next_page;	/* extent of pages */
+	struct writeback_control *wbc;
+	int io_done;
+	int pages_written;
+	int retval;
+};
+#define	DIO_AIO_UNWRITTEN	0x1
+typedef struct ext4_io_end {
+	struct list_head	list;		/* per-file finished AIO list */
+	struct inode		*inode;		/* file being written to */
+	unsigned int		flag;		/* unwritten or not */
+	int			error;		/* I/O error code */
+	ext4_lblk_t		offset;		/* offset in the file */
+	size_t			size;		/* size of the extent */
+	struct work_struct	work;		/* data work queue */
+} ext4_io_end_t;
+
+/*
  * Special inodes numbers
  */
 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
@@ -251,7 +284,6 @@ struct flex_groups {
 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
-#define EXT4_EXT_MIGRATE		0x00100000 /* Inode is migrating */
 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
 
 #define EXT4_FL_USER_VISIBLE		0x000BDFFF /* User visible flags */
@@ -289,6 +321,7 @@ static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 #define EXT4_STATE_XATTR		0x00000004 /* has in-inode xattrs */
 #define EXT4_STATE_NO_EXPAND		0x00000008 /* No space for expansion */
 #define EXT4_STATE_DA_ALLOC_CLOSE	0x00000010 /* Alloc DA blks on close */
+#define EXT4_STATE_EXT_MIGRATE		0x00000020 /* Inode is migrating */
 
 /* Used to pass group descriptor data when online resize is done */
 struct ext4_new_group_input {
@@ -330,7 +363,16 @@ struct ext4_new_group_data {
 	/* Call ext4_da_update_reserve_space() after successfully 
 	   allocating the blocks */
 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE	0x0008
-
+	/* caller is from the direct IO path, request to creation of an
+	unitialized extents if not allocated, split the uninitialized
+	extent if blocks has been preallocated already*/
+#define EXT4_GET_BLOCKS_DIO			0x0010
+#define EXT4_GET_BLOCKS_CONVERT			0x0020
+#define EXT4_GET_BLOCKS_DIO_CREATE_EXT		(EXT4_GET_BLOCKS_DIO|\
+					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
+	/* Convert extent to initialized after direct IO complete */
+#define EXT4_GET_BLOCKS_DIO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
+					 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
 
 /*
  * ioctl commands
@@ -386,6 +428,9 @@ struct ext4_mount_options {
 #endif
 };
 
+/* Max physical block we can addres w/o extents */
+#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
+
 /*
  * Structure of an inode on the disk
  */
@@ -456,7 +501,6 @@ struct move_extent {
 	__u64 len;		/* block length to be moved */
 	__u64 moved_len;	/* moved block length */
 };
-#define MAX_DEFRAG_SIZE         ((1UL<<31) - 1)
 
 #define EXT4_EPOCH_BITS 2
 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
@@ -481,8 +525,8 @@ struct move_extent {
 static inline __le32 ext4_encode_extra_time(struct timespec *time)
 {
        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
-			   time->tv_sec >> 32 : 0) |
-			   ((time->tv_nsec << 2) & EXT4_NSEC_MASK));
+			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
+                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
 }
 
 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
@@ -490,7 +534,7 @@ static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
        if (sizeof(time->tv_sec) > 4)
 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
 			       << 32;
-       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> 2;
+       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 }
 
 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
@@ -653,6 +697,11 @@ struct ext4_inode_info {
 	__u16 i_extra_isize;
 
 	spinlock_t i_block_reservation_lock;
+
+	/* completed async DIOs that might need unwritten extents handling */
+	struct list_head i_aio_dio_complete_list;
+	/* current io_end structure for async DIO write*/
+	ext4_io_end_t *cur_aio_dio;
 };
 
 /*
@@ -694,7 +743,6 @@ struct ext4_inode_info {
 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
-#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
@@ -841,6 +889,7 @@ struct ext4_sb_info {
 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
+	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
 	unsigned long s_overhead_last;  /* Last calculated overhead */
 	unsigned long s_blocks_last;    /* Last seen block count */
 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
@@ -923,18 +972,11 @@ struct ext4_sb_info {
 	unsigned int s_mb_stats;
 	unsigned int s_mb_order2_reqs;
 	unsigned int s_mb_group_prealloc;
+	unsigned int s_max_writeback_mb_bump;
 	/* where last allocation was done - for stream allocation */
 	unsigned long s_mb_last_group;
 	unsigned long s_mb_last_start;
 
-	/* history to debug policy */
-	struct ext4_mb_history *s_mb_history;
-	int s_mb_history_cur;
-	int s_mb_history_max;
-	int s_mb_history_num;
-	spinlock_t s_mb_history_lock;
-	int s_mb_history_filter;
-
 	/* stats for buddy allocator */
 	spinlock_t s_mb_pa_lock;
 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
@@ -950,6 +992,7 @@ struct ext4_sb_info {
 	atomic_t s_mb_lost_chunks;
 	atomic_t s_mb_preallocated;
 	atomic_t s_mb_discarded;
+	atomic_t s_lock_busy;
 
 	/* locality groups */
 	struct ext4_locality_group *s_locality_groups;
@@ -960,6 +1003,9 @@ struct ext4_sb_info {
 
 	unsigned int s_log_groups_per_flex;
 	struct flex_groups *s_flex_groups;
+
+	/* workqueue for dio unwritten */
+	struct workqueue_struct *dio_unwritten_wq;
 };
 
 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
@@ -1340,8 +1386,6 @@ extern void ext4_mb_free_blocks(handle_t *, struct inode *,
 		ext4_fsblk_t, unsigned long, int, unsigned long *);
 extern int ext4_mb_add_groupinfo(struct super_block *sb,
 		ext4_group_t i, struct ext4_group_desc *desc);
-extern void ext4_mb_update_group_info(struct ext4_group_info *grp,
-		ext4_grpblk_t add);
 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
 						ext4_group_t, int);
@@ -1367,6 +1411,7 @@ extern int ext4_change_inode_journal_flag(struct inode *, int);
 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
 extern int ext4_can_truncate(struct inode *inode);
 extern void ext4_truncate(struct inode *);
+extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
 extern void ext4_set_inode_flags(struct inode *);
 extern void ext4_get_inode_flags(struct ext4_inode_info *);
 extern int ext4_alloc_da_blocks(struct inode *inode);
@@ -1378,7 +1423,7 @@ extern int ext4_block_truncate_page(handle_t *handle,
 		struct address_space *mapping, loff_t from);
 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
 extern qsize_t ext4_get_reserved_space(struct inode *inode);
-
+extern int flush_aio_dio_completed_IO(struct inode *inode);
 /* ioctl.c */
 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
@@ -1575,15 +1620,18 @@ static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
 struct ext4_group_info {
 	unsigned long   bb_state;
 	struct rb_root  bb_free_root;
-	unsigned short  bb_first_free;
-	unsigned short  bb_free;
-	unsigned short  bb_fragments;
+	ext4_grpblk_t	bb_first_free;	/* first free block */
+	ext4_grpblk_t	bb_free;	/* total free blocks */
+	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
 	struct          list_head bb_prealloc_list;
 #ifdef DOUBLE_CHECK
 	void            *bb_bitmap;
 #endif
 	struct rw_semaphore alloc_sem;
-	unsigned short  bb_counters[];
+	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
+					 * regions, index is order.
+					 * bb_counters[3] = 5 means
+					 * 5 free 8-block regions. */
 };
 
 #define EXT4_GROUP_INFO_NEED_INIT_BIT	0
@@ -1591,15 +1639,42 @@ struct ext4_group_info {
 #define EXT4_MB_GRP_NEED_INIT(grp)	\
 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
 
+#define EXT4_MAX_CONTENTION		8
+#define EXT4_CONTENTION_THRESHOLD	2
+
 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
 					      ext4_group_t group)
 {
 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
 }
 
+/*
+ * Returns true if the filesystem is busy enough that attempts to
+ * access the block group locks has run into contention.
+ */
+static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
+{
+	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
+}
+
 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
 {
-	spin_lock(ext4_group_lock_ptr(sb, group));
+	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
+	if (spin_trylock(lock))
+		/*
+		 * We're able to grab the lock right away, so drop the
+		 * lock contention counter.
+		 */
+		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
+	else {
+		/*
+		 * The lock is busy, so bump the contention counter,
+		 * and then wait on the spin lock.
+		 */
+		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
+				  EXT4_MAX_CONTENTION);
+		spin_lock(lock);
+	}
 }
 
 static inline void ext4_unlock_group(struct super_block *sb,
@@ -1650,6 +1725,8 @@ extern void ext4_ext_init(struct super_block *);
 extern void ext4_ext_release(struct super_block *);
 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
 			  loff_t len);
+extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
+			  loff_t len);
 extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
 			   sector_t block, unsigned int max_blocks,
 			   struct buffer_head *bh, int flags);
diff --git a/fs/ext4/ext4_extents.h b/fs/ext4/ext4_extents.h
index 20a84105a10b..2ca686454e87 100644
--- a/fs/ext4/ext4_extents.h
+++ b/fs/ext4/ext4_extents.h
@@ -43,8 +43,7 @@
 #define CHECK_BINSEARCH__
 
 /*
- * If EXT_DEBUG is defined you can use the 'extdebug' mount option
- * to get lots of info about what's going on.
+ * Turn on EXT_DEBUG to get lots of info about extents operations.
  */
 #define EXT_DEBUG__
 #ifdef EXT_DEBUG
@@ -138,6 +137,7 @@ typedef int (*ext_prepare_callback)(struct inode *, struct ext4_ext_path *,
 #define EXT_BREAK      1
 #define EXT_REPEAT     2
 
+/* Maximum logical block in a file; ext4_extent's ee_block is __le32 */
 #define EXT_MAX_BLOCK	0xffffffff
 
 /*
@@ -220,6 +220,11 @@ static inline int ext4_ext_get_actual_len(struct ext4_extent *ext)
 		(le16_to_cpu(ext->ee_len) - EXT_INIT_MAX_LEN));
 }
 
+static inline void ext4_ext_mark_initialized(struct ext4_extent *ext)
+{
+	ext->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ext));
+}
+
 extern int ext4_ext_calc_metadata_amount(struct inode *inode, int blocks);
 extern ext4_fsblk_t ext_pblock(struct ext4_extent *ex);
 extern ext4_fsblk_t idx_pblock(struct ext4_extent_idx *);
@@ -235,7 +240,7 @@ extern int ext4_ext_try_to_merge(struct inode *inode,
 				 struct ext4_ext_path *path,
 				 struct ext4_extent *);
 extern unsigned int ext4_ext_check_overlap(struct inode *, struct ext4_extent *, struct ext4_ext_path *);
-extern int ext4_ext_insert_extent(handle_t *, struct inode *, struct ext4_ext_path *, struct ext4_extent *);
+extern int ext4_ext_insert_extent(handle_t *, struct inode *, struct ext4_ext_path *, struct ext4_extent *, int);
 extern int ext4_ext_walk_space(struct inode *, ext4_lblk_t, ext4_lblk_t,
 							ext_prepare_callback, void *);
 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
diff --git a/fs/ext4/ext4_jbd2.c b/fs/ext4/ext4_jbd2.c
index eb27fd0f2ee8..6a9409920dee 100644
--- a/fs/ext4/ext4_jbd2.c
+++ b/fs/ext4/ext4_jbd2.c
@@ -44,7 +44,7 @@ int __ext4_journal_forget(const char *where, handle_t *handle,
 						  handle, err);
 	}
 	else
-		brelse(bh);
+		bforget(bh);
 	return err;
 }
 
@@ -60,7 +60,7 @@ int __ext4_journal_revoke(const char *where, handle_t *handle,
 						  handle, err);
 	}
 	else
-		brelse(bh);
+		bforget(bh);
 	return err;
 }
 
@@ -89,7 +89,10 @@ int __ext4_handle_dirty_metadata(const char *where, handle_t *handle,
 			ext4_journal_abort_handle(where, __func__, bh,
 						  handle, err);
 	} else {
-		mark_buffer_dirty(bh);
+		if (inode && bh)
+			mark_buffer_dirty_inode(bh, inode);
+		else
+			mark_buffer_dirty(bh);
 		if (inode && inode_needs_sync(inode)) {
 			sync_dirty_buffer(bh);
 			if (buffer_req(bh) && !buffer_uptodate(bh)) {
diff --git a/fs/ext4/ext4_jbd2.h b/fs/ext4/ext4_jbd2.h
index 139fb8cb87e4..a2865980342f 100644
--- a/fs/ext4/ext4_jbd2.h
+++ b/fs/ext4/ext4_jbd2.h
@@ -161,11 +161,13 @@ int __ext4_handle_dirty_metadata(const char *where, handle_t *handle,
 handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks);
 int __ext4_journal_stop(const char *where, handle_t *handle);
 
-#define EXT4_NOJOURNAL_HANDLE	((handle_t *) 0x1)
+#define EXT4_NOJOURNAL_MAX_REF_COUNT ((unsigned long) 4096)
 
+/* Note:  Do not use this for NULL handles.  This is only to determine if
+ * a properly allocated handle is using a journal or not. */
 static inline int ext4_handle_valid(handle_t *handle)
 {
-	if (handle == EXT4_NOJOURNAL_HANDLE)
+	if ((unsigned long)handle < EXT4_NOJOURNAL_MAX_REF_COUNT)
 		return 0;
 	return 1;
 }
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 73ebfb44ad75..10539e364283 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -93,7 +93,9 @@ static void ext4_idx_store_pblock(struct ext4_extent_idx *ix, ext4_fsblk_t pb)
 	ix->ei_leaf_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) & 0xffff);
 }
 
-static int ext4_ext_journal_restart(handle_t *handle, int needed)
+static int ext4_ext_truncate_extend_restart(handle_t *handle,
+					    struct inode *inode,
+					    int needed)
 {
 	int err;
 
@@ -104,7 +106,14 @@ static int ext4_ext_journal_restart(handle_t *handle, int needed)
 	err = ext4_journal_extend(handle, needed);
 	if (err <= 0)
 		return err;
-	return ext4_journal_restart(handle, needed);
+	err = ext4_truncate_restart_trans(handle, inode, needed);
+	/*
+	 * We have dropped i_data_sem so someone might have cached again
+	 * an extent we are going to truncate.
+	 */
+	ext4_ext_invalidate_cache(inode);
+
+	return err;
 }
 
 /*
@@ -220,57 +229,65 @@ ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
 	return newblock;
 }
 
-static int ext4_ext_space_block(struct inode *inode)
+static inline int ext4_ext_space_block(struct inode *inode, int check)
 {
 	int size;
 
 	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
 			/ sizeof(struct ext4_extent);
+	if (!check) {
 #ifdef AGGRESSIVE_TEST
-	if (size > 6)
-		size = 6;
+		if (size > 6)
+			size = 6;
 #endif
+	}
 	return size;
 }
 
-static int ext4_ext_space_block_idx(struct inode *inode)
+static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
 {
 	int size;
 
 	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
 			/ sizeof(struct ext4_extent_idx);
+	if (!check) {
 #ifdef AGGRESSIVE_TEST
-	if (size > 5)
-		size = 5;
+		if (size > 5)
+			size = 5;
 #endif
+	}
 	return size;
 }
 
-static int ext4_ext_space_root(struct inode *inode)
+static inline int ext4_ext_space_root(struct inode *inode, int check)
 {
 	int size;
 
 	size = sizeof(EXT4_I(inode)->i_data);
 	size -= sizeof(struct ext4_extent_header);
 	size /= sizeof(struct ext4_extent);
+	if (!check) {
 #ifdef AGGRESSIVE_TEST
-	if (size > 3)
-		size = 3;
+		if (size > 3)
+			size = 3;
 #endif
+	}
 	return size;
 }
 
-static int ext4_ext_space_root_idx(struct inode *inode)
+static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
 {
 	int size;
 
 	size = sizeof(EXT4_I(inode)->i_data);
 	size -= sizeof(struct ext4_extent_header);
 	size /= sizeof(struct ext4_extent_idx);
+	if (!check) {
 #ifdef AGGRESSIVE_TEST
-	if (size > 4)
-		size = 4;
+		if (size > 4)
+			size = 4;
 #endif
+	}
 	return size;
 }
 
@@ -284,9 +301,9 @@ int ext4_ext_calc_metadata_amount(struct inode *inode, int blocks)
 	int lcap, icap, rcap, leafs, idxs, num;
 	int newextents = blocks;
 
-	rcap = ext4_ext_space_root_idx(inode);
-	lcap = ext4_ext_space_block(inode);
-	icap = ext4_ext_space_block_idx(inode);
+	rcap = ext4_ext_space_root_idx(inode, 0);
+	lcap = ext4_ext_space_block(inode, 0);
+	icap = ext4_ext_space_block_idx(inode, 0);
 
 	/* number of new leaf blocks needed */
 	num = leafs = (newextents + lcap - 1) / lcap;
@@ -311,14 +328,14 @@ ext4_ext_max_entries(struct inode *inode, int depth)
 
 	if (depth == ext_depth(inode)) {
 		if (depth == 0)
-			max = ext4_ext_space_root(inode);
+			max = ext4_ext_space_root(inode, 1);
 		else
-			max = ext4_ext_space_root_idx(inode);
+			max = ext4_ext_space_root_idx(inode, 1);
 	} else {
 		if (depth == 0)
-			max = ext4_ext_space_block(inode);
+			max = ext4_ext_space_block(inode, 1);
 		else
-			max = ext4_ext_space_block_idx(inode);
+			max = ext4_ext_space_block_idx(inode, 1);
 	}
 
 	return max;
@@ -437,8 +454,9 @@ static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
 		  ext_debug("  %d->%llu", le32_to_cpu(path->p_idx->ei_block),
 			    idx_pblock(path->p_idx));
 		} else if (path->p_ext) {
-			ext_debug("  %d:%d:%llu ",
+			ext_debug("  %d:[%d]%d:%llu ",
 				  le32_to_cpu(path->p_ext->ee_block),
+				  ext4_ext_is_uninitialized(path->p_ext),
 				  ext4_ext_get_actual_len(path->p_ext),
 				  ext_pblock(path->p_ext));
 		} else
@@ -460,8 +478,11 @@ static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
 	eh = path[depth].p_hdr;
 	ex = EXT_FIRST_EXTENT(eh);
 
+	ext_debug("Displaying leaf extents for inode %lu\n", inode->i_ino);
+
 	for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
-		ext_debug("%d:%d:%llu ", le32_to_cpu(ex->ee_block),
+		ext_debug("%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
+			  ext4_ext_is_uninitialized(ex),
 			  ext4_ext_get_actual_len(ex), ext_pblock(ex));
 	}
 	ext_debug("\n");
@@ -580,9 +601,10 @@ ext4_ext_binsearch(struct inode *inode,
 	}
 
 	path->p_ext = l - 1;
-	ext_debug("  -> %d:%llu:%d ",
+	ext_debug("  -> %d:%llu:[%d]%d ",
 			le32_to_cpu(path->p_ext->ee_block),
 			ext_pblock(path->p_ext),
+			ext4_ext_is_uninitialized(path->p_ext),
 			ext4_ext_get_actual_len(path->p_ext));
 
 #ifdef CHECK_BINSEARCH
@@ -612,7 +634,7 @@ int ext4_ext_tree_init(handle_t *handle, struct inode *inode)
 	eh->eh_depth = 0;
 	eh->eh_entries = 0;
 	eh->eh_magic = EXT4_EXT_MAGIC;
-	eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode));
+	eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
 	ext4_mark_inode_dirty(handle, inode);
 	ext4_ext_invalidate_cache(inode);
 	return 0;
@@ -701,7 +723,7 @@ err:
  * insert new index [@logical;@ptr] into the block at @curp;
  * check where to insert: before @curp or after @curp
  */
-static int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
+int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
 				struct ext4_ext_path *curp,
 				int logical, ext4_fsblk_t ptr)
 {
@@ -837,7 +859,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 
 	neh = ext_block_hdr(bh);
 	neh->eh_entries = 0;
-	neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode));
+	neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
 	neh->eh_magic = EXT4_EXT_MAGIC;
 	neh->eh_depth = 0;
 	ex = EXT_FIRST_EXTENT(neh);
@@ -850,9 +872,10 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 	path[depth].p_ext++;
 	while (path[depth].p_ext <=
 			EXT_MAX_EXTENT(path[depth].p_hdr)) {
-		ext_debug("move %d:%llu:%d in new leaf %llu\n",
+		ext_debug("move %d:%llu:[%d]%d in new leaf %llu\n",
 				le32_to_cpu(path[depth].p_ext->ee_block),
 				ext_pblock(path[depth].p_ext),
+				ext4_ext_is_uninitialized(path[depth].p_ext),
 				ext4_ext_get_actual_len(path[depth].p_ext),
 				newblock);
 		/*memmove(ex++, path[depth].p_ext++,
@@ -912,7 +935,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 		neh = ext_block_hdr(bh);
 		neh->eh_entries = cpu_to_le16(1);
 		neh->eh_magic = EXT4_EXT_MAGIC;
-		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode));
+		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
 		neh->eh_depth = cpu_to_le16(depth - i);
 		fidx = EXT_FIRST_INDEX(neh);
 		fidx->ei_block = border;
@@ -1037,9 +1060,9 @@ static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
 	/* old root could have indexes or leaves
 	 * so calculate e_max right way */
 	if (ext_depth(inode))
-	  neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode));
+		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
 	else
-	  neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode));
+		neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
 	neh->eh_magic = EXT4_EXT_MAGIC;
 	set_buffer_uptodate(bh);
 	unlock_buffer(bh);
@@ -1054,7 +1077,7 @@ static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
 		goto out;
 
 	curp->p_hdr->eh_magic = EXT4_EXT_MAGIC;
-	curp->p_hdr->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode));
+	curp->p_hdr->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
 	curp->p_hdr->eh_entries = cpu_to_le16(1);
 	curp->p_idx = EXT_FIRST_INDEX(curp->p_hdr);
 
@@ -1563,7 +1586,7 @@ out:
  */
 int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
 				struct ext4_ext_path *path,
-				struct ext4_extent *newext)
+				struct ext4_extent *newext, int flag)
 {
 	struct ext4_extent_header *eh;
 	struct ext4_extent *ex, *fex;
@@ -1579,10 +1602,13 @@ int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
 	BUG_ON(path[depth].p_hdr == NULL);
 
 	/* try to insert block into found extent and return */
-	if (ex && ext4_can_extents_be_merged(inode, ex, newext)) {
-		ext_debug("append %d block to %d:%d (from %llu)\n",
+	if (ex && (flag != EXT4_GET_BLOCKS_DIO_CREATE_EXT)
+		&& ext4_can_extents_be_merged(inode, ex, newext)) {
+		ext_debug("append [%d]%d block to %d:[%d]%d (from %llu)\n",
+				ext4_ext_is_uninitialized(newext),
 				ext4_ext_get_actual_len(newext),
 				le32_to_cpu(ex->ee_block),
+				ext4_ext_is_uninitialized(ex),
 				ext4_ext_get_actual_len(ex), ext_pblock(ex));
 		err = ext4_ext_get_access(handle, inode, path + depth);
 		if (err)
@@ -1651,9 +1677,10 @@ has_space:
 
 	if (!nearex) {
 		/* there is no extent in this leaf, create first one */
-		ext_debug("first extent in the leaf: %d:%llu:%d\n",
+		ext_debug("first extent in the leaf: %d:%llu:[%d]%d\n",
 				le32_to_cpu(newext->ee_block),
 				ext_pblock(newext),
+				ext4_ext_is_uninitialized(newext),
 				ext4_ext_get_actual_len(newext));
 		path[depth].p_ext = EXT_FIRST_EXTENT(eh);
 	} else if (le32_to_cpu(newext->ee_block)
@@ -1663,10 +1690,11 @@ has_space:
 			len = EXT_MAX_EXTENT(eh) - nearex;
 			len = (len - 1) * sizeof(struct ext4_extent);
 			len = len < 0 ? 0 : len;
-			ext_debug("insert %d:%llu:%d after: nearest 0x%p, "
+			ext_debug("insert %d:%llu:[%d]%d after: nearest 0x%p, "
 					"move %d from 0x%p to 0x%p\n",
 					le32_to_cpu(newext->ee_block),
 					ext_pblock(newext),
+					ext4_ext_is_uninitialized(newext),
 					ext4_ext_get_actual_len(newext),
 					nearex, len, nearex + 1, nearex + 2);
 			memmove(nearex + 2, nearex + 1, len);
@@ -1676,10 +1704,11 @@ has_space:
 		BUG_ON(newext->ee_block == nearex->ee_block);
 		len = (EXT_MAX_EXTENT(eh) - nearex) * sizeof(struct ext4_extent);
 		len = len < 0 ? 0 : len;
-		ext_debug("insert %d:%llu:%d before: nearest 0x%p, "
+		ext_debug("insert %d:%llu:[%d]%d before: nearest 0x%p, "
 				"move %d from 0x%p to 0x%p\n",
 				le32_to_cpu(newext->ee_block),
 				ext_pblock(newext),
+				ext4_ext_is_uninitialized(newext),
 				ext4_ext_get_actual_len(newext),
 				nearex, len, nearex + 1, nearex + 2);
 		memmove(nearex + 1, nearex, len);
@@ -1694,7 +1723,8 @@ has_space:
 
 merge:
 	/* try to merge extents to the right */
-	ext4_ext_try_to_merge(inode, path, nearex);
+	if (flag != EXT4_GET_BLOCKS_DIO_CREATE_EXT)
+		ext4_ext_try_to_merge(inode, path, nearex);
 
 	/* try to merge extents to the left */
 
@@ -2094,7 +2124,8 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 		else
 			uninitialized = 0;
 
-		ext_debug("remove ext %lu:%u\n", ex_ee_block, ex_ee_len);
+		ext_debug("remove ext %u:[%d]%d\n", ex_ee_block,
+			 uninitialized, ex_ee_len);
 		path[depth].p_ext = ex;
 
 		a = ex_ee_block > start ? ex_ee_block : start;
@@ -2138,7 +2169,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 		}
 		credits += 2 * EXT4_QUOTA_TRANS_BLOCKS(inode->i_sb);
 
-		err = ext4_ext_journal_restart(handle, credits);
+		err = ext4_ext_truncate_extend_restart(handle, inode, credits);
 		if (err)
 			goto out;
 
@@ -2327,7 +2358,7 @@ static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start)
 		if (err == 0) {
 			ext_inode_hdr(inode)->eh_depth = 0;
 			ext_inode_hdr(inode)->eh_max =
-				cpu_to_le16(ext4_ext_space_root(inode));
+				cpu_to_le16(ext4_ext_space_root(inode, 0));
 			err = ext4_ext_dirty(handle, inode, path);
 		}
 	}
@@ -2349,6 +2380,7 @@ void ext4_ext_init(struct super_block *sb)
 	 */
 
 	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS)) {
+#if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
 		printk(KERN_INFO "EXT4-fs: file extents enabled");
 #ifdef AGGRESSIVE_TEST
 		printk(", aggressive tests");
@@ -2360,6 +2392,7 @@ void ext4_ext_init(struct super_block *sb)
 		printk(", stats");
 #endif
 		printk("\n");
+#endif
 #ifdef EXTENTS_STATS
 		spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
 		EXT4_SB(sb)->s_ext_min = 1 << 30;
@@ -2461,7 +2494,6 @@ static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
 }
 
 #define EXT4_EXT_ZERO_LEN 7
-
 /*
  * This function is called by ext4_ext_get_blocks() if someone tries to write
  * to an uninitialized extent. It may result in splitting the uninitialized
@@ -2554,7 +2586,8 @@ static int ext4_ext_convert_to_initialized(handle_t *handle,
 			ex3->ee_block = cpu_to_le32(iblock);
 			ext4_ext_store_pblock(ex3, newblock);
 			ex3->ee_len = cpu_to_le16(allocated);
-			err = ext4_ext_insert_extent(handle, inode, path, ex3);
+			err = ext4_ext_insert_extent(handle, inode, path,
+							ex3, 0);
 			if (err == -ENOSPC) {
 				err =  ext4_ext_zeroout(inode, &orig_ex);
 				if (err)
@@ -2610,7 +2643,7 @@ static int ext4_ext_convert_to_initialized(handle_t *handle,
 		ext4_ext_store_pblock(ex3, newblock + max_blocks);
 		ex3->ee_len = cpu_to_le16(allocated - max_blocks);
 		ext4_ext_mark_uninitialized(ex3);
-		err = ext4_ext_insert_extent(handle, inode, path, ex3);
+		err = ext4_ext_insert_extent(handle, inode, path, ex3, 0);
 		if (err == -ENOSPC) {
 			err =  ext4_ext_zeroout(inode, &orig_ex);
 			if (err)
@@ -2728,7 +2761,191 @@ static int ext4_ext_convert_to_initialized(handle_t *handle,
 	err = ext4_ext_dirty(handle, inode, path + depth);
 	goto out;
 insert:
-	err = ext4_ext_insert_extent(handle, inode, path, &newex);
+	err = ext4_ext_insert_extent(handle, inode, path, &newex, 0);
+	if (err == -ENOSPC) {
+		err =  ext4_ext_zeroout(inode, &orig_ex);
+		if (err)
+			goto fix_extent_len;
+		/* update the extent length and mark as initialized */
+		ex->ee_block = orig_ex.ee_block;
+		ex->ee_len   = orig_ex.ee_len;
+		ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
+		ext4_ext_dirty(handle, inode, path + depth);
+		/* zero out the first half */
+		return allocated;
+	} else if (err)
+		goto fix_extent_len;
+out:
+	ext4_ext_show_leaf(inode, path);
+	return err ? err : allocated;
+
+fix_extent_len:
+	ex->ee_block = orig_ex.ee_block;
+	ex->ee_len   = orig_ex.ee_len;
+	ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
+	ext4_ext_mark_uninitialized(ex);
+	ext4_ext_dirty(handle, inode, path + depth);
+	return err;
+}
+
+/*
+ * This function is called by ext4_ext_get_blocks() from
+ * ext4_get_blocks_dio_write() when DIO to write
+ * to an uninitialized extent.
+ *
+ * Writing to an uninitized extent may result in splitting the uninitialized
+ * extent into multiple /intialized unintialized extents (up to three)
+ * There are three possibilities:
+ *   a> There is no split required: Entire extent should be uninitialized
+ *   b> Splits in two extents: Write is happening at either end of the extent
+ *   c> Splits in three extents: Somone is writing in middle of the extent
+ *
+ * One of more index blocks maybe needed if the extent tree grow after
+ * the unintialized extent split. To prevent ENOSPC occur at the IO
+ * complete, we need to split the uninitialized extent before DIO submit
+ * the IO. The uninitilized extent called at this time will be split
+ * into three uninitialized extent(at most). After IO complete, the part
+ * being filled will be convert to initialized by the end_io callback function
+ * via ext4_convert_unwritten_extents().
+ */
+static int ext4_split_unwritten_extents(handle_t *handle,
+					struct inode *inode,
+					struct ext4_ext_path *path,
+					ext4_lblk_t iblock,
+					unsigned int max_blocks,
+					int flags)
+{
+	struct ext4_extent *ex, newex, orig_ex;
+	struct ext4_extent *ex1 = NULL;
+	struct ext4_extent *ex2 = NULL;
+	struct ext4_extent *ex3 = NULL;
+	struct ext4_extent_header *eh;
+	ext4_lblk_t ee_block;
+	unsigned int allocated, ee_len, depth;
+	ext4_fsblk_t newblock;
+	int err = 0;
+	int ret = 0;
+
+	ext_debug("ext4_split_unwritten_extents: inode %lu,"
+		  "iblock %llu, max_blocks %u\n", inode->i_ino,
+		  (unsigned long long)iblock, max_blocks);
+	depth = ext_depth(inode);
+	eh = path[depth].p_hdr;
+	ex = path[depth].p_ext;
+	ee_block = le32_to_cpu(ex->ee_block);
+	ee_len = ext4_ext_get_actual_len(ex);
+	allocated = ee_len - (iblock - ee_block);
+	newblock = iblock - ee_block + ext_pblock(ex);
+	ex2 = ex;
+	orig_ex.ee_block = ex->ee_block;
+	orig_ex.ee_len   = cpu_to_le16(ee_len);
+	ext4_ext_store_pblock(&orig_ex, ext_pblock(ex));
+
+	/*
+ 	 * if the entire unintialized extent length less than
+ 	 * the size of extent to write, there is no need to split
+ 	 * uninitialized extent
+ 	 */
+ 	if (allocated <= max_blocks)
+		return ret;
+
+	err = ext4_ext_get_access(handle, inode, path + depth);
+	if (err)
+		goto out;
+	/* ex1: ee_block to iblock - 1 : uninitialized */
+	if (iblock > ee_block) {
+		ex1 = ex;
+		ex1->ee_len = cpu_to_le16(iblock - ee_block);
+		ext4_ext_mark_uninitialized(ex1);
+		ex2 = &newex;
+	}
+	/*
+	 * for sanity, update the length of the ex2 extent before
+	 * we insert ex3, if ex1 is NULL. This is to avoid temporary
+	 * overlap of blocks.
+	 */
+	if (!ex1 && allocated > max_blocks)
+		ex2->ee_len = cpu_to_le16(max_blocks);
+	/* ex3: to ee_block + ee_len : uninitialised */
+	if (allocated > max_blocks) {
+		unsigned int newdepth;
+		ex3 = &newex;
+		ex3->ee_block = cpu_to_le32(iblock + max_blocks);
+		ext4_ext_store_pblock(ex3, newblock + max_blocks);
+		ex3->ee_len = cpu_to_le16(allocated - max_blocks);
+		ext4_ext_mark_uninitialized(ex3);
+		err = ext4_ext_insert_extent(handle, inode, path, ex3, flags);
+		if (err == -ENOSPC) {
+			err =  ext4_ext_zeroout(inode, &orig_ex);
+			if (err)
+				goto fix_extent_len;
+			/* update the extent length and mark as initialized */
+			ex->ee_block = orig_ex.ee_block;
+			ex->ee_len   = orig_ex.ee_len;
+			ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
+			ext4_ext_dirty(handle, inode, path + depth);
+			/* zeroed the full extent */
+			/* blocks available from iblock */
+			return allocated;
+
+		} else if (err)
+			goto fix_extent_len;
+		/*
+		 * The depth, and hence eh & ex might change
+		 * as part of the insert above.
+		 */
+		newdepth = ext_depth(inode);
+		/*
+		 * update the extent length after successful insert of the
+		 * split extent
+		 */
+		orig_ex.ee_len = cpu_to_le16(ee_len -
+						ext4_ext_get_actual_len(ex3));
+		depth = newdepth;
+		ext4_ext_drop_refs(path);
+		path = ext4_ext_find_extent(inode, iblock, path);
+		if (IS_ERR(path)) {
+			err = PTR_ERR(path);
+			goto out;
+		}
+		eh = path[depth].p_hdr;
+		ex = path[depth].p_ext;
+		if (ex2 != &newex)
+			ex2 = ex;
+
+		err = ext4_ext_get_access(handle, inode, path + depth);
+		if (err)
+			goto out;
+
+		allocated = max_blocks;
+	}
+	/*
+	 * If there was a change of depth as part of the
+	 * insertion of ex3 above, we need to update the length
+	 * of the ex1 extent again here
+	 */
+	if (ex1 && ex1 != ex) {
+		ex1 = ex;
+		ex1->ee_len = cpu_to_le16(iblock - ee_block);
+		ext4_ext_mark_uninitialized(ex1);
+		ex2 = &newex;
+	}
+	/*
+	 * ex2: iblock to iblock + maxblocks-1 : to be direct IO written,
+	 * uninitialised still.
+	 */
+	ex2->ee_block = cpu_to_le32(iblock);
+	ext4_ext_store_pblock(ex2, newblock);
+	ex2->ee_len = cpu_to_le16(allocated);
+	ext4_ext_mark_uninitialized(ex2);
+	if (ex2 != ex)
+		goto insert;
+	/* Mark modified extent as dirty */
+	err = ext4_ext_dirty(handle, inode, path + depth);
+	ext_debug("out here\n");
+	goto out;
+insert:
+	err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
 	if (err == -ENOSPC) {
 		err =  ext4_ext_zeroout(inode, &orig_ex);
 		if (err)
@@ -2743,6 +2960,7 @@ insert:
 	} else if (err)
 		goto fix_extent_len;
 out:
+	ext4_ext_show_leaf(inode, path);
 	return err ? err : allocated;
 
 fix_extent_len:
@@ -2753,7 +2971,141 @@ fix_extent_len:
 	ext4_ext_dirty(handle, inode, path + depth);
 	return err;
 }
+static int ext4_convert_unwritten_extents_dio(handle_t *handle,
+					      struct inode *inode,
+					      struct ext4_ext_path *path)
+{
+	struct ext4_extent *ex;
+	struct ext4_extent_header *eh;
+	int depth;
+	int err = 0;
+	int ret = 0;
+
+	depth = ext_depth(inode);
+	eh = path[depth].p_hdr;
+	ex = path[depth].p_ext;
+
+	err = ext4_ext_get_access(handle, inode, path + depth);
+	if (err)
+		goto out;
+	/* first mark the extent as initialized */
+	ext4_ext_mark_initialized(ex);
+
+	/*
+	 * We have to see if it can be merged with the extent
+	 * on the left.
+	 */
+	if (ex > EXT_FIRST_EXTENT(eh)) {
+		/*
+		 * To merge left, pass "ex - 1" to try_to_merge(),
+		 * since it merges towards right _only_.
+		 */
+		ret = ext4_ext_try_to_merge(inode, path, ex - 1);
+		if (ret) {
+			err = ext4_ext_correct_indexes(handle, inode, path);
+			if (err)
+				goto out;
+			depth = ext_depth(inode);
+			ex--;
+		}
+	}
+	/*
+	 * Try to Merge towards right.
+	 */
+	ret = ext4_ext_try_to_merge(inode, path, ex);
+	if (ret) {
+		err = ext4_ext_correct_indexes(handle, inode, path);
+		if (err)
+			goto out;
+		depth = ext_depth(inode);
+	}
+	/* Mark modified extent as dirty */
+	err = ext4_ext_dirty(handle, inode, path + depth);
+out:
+	ext4_ext_show_leaf(inode, path);
+	return err;
+}
+
+static int
+ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
+			ext4_lblk_t iblock, unsigned int max_blocks,
+			struct ext4_ext_path *path, int flags,
+			unsigned int allocated, struct buffer_head *bh_result,
+			ext4_fsblk_t newblock)
+{
+	int ret = 0;
+	int err = 0;
+	ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
+
+	ext_debug("ext4_ext_handle_uninitialized_extents: inode %lu, logical"
+		  "block %llu, max_blocks %u, flags %d, allocated %u",
+		  inode->i_ino, (unsigned long long)iblock, max_blocks,
+		  flags, allocated);
+	ext4_ext_show_leaf(inode, path);
+
+	/* DIO get_block() before submit the IO, split the extent */
+	if (flags == EXT4_GET_BLOCKS_DIO_CREATE_EXT) {
+		ret = ext4_split_unwritten_extents(handle,
+						inode, path, iblock,
+						max_blocks, flags);
+		/* flag the io_end struct that we need convert when IO done */
+		if (io)
+			io->flag = DIO_AIO_UNWRITTEN;
+		goto out;
+	}
+	/* DIO end_io complete, convert the filled extent to written */
+	if (flags == EXT4_GET_BLOCKS_DIO_CONVERT_EXT) {
+		ret = ext4_convert_unwritten_extents_dio(handle, inode,
+							path);
+		goto out2;
+	}
+	/* buffered IO case */
+	/*
+	 * repeat fallocate creation request
+	 * we already have an unwritten extent
+	 */
+	if (flags & EXT4_GET_BLOCKS_UNINIT_EXT)
+		goto map_out;
+
+	/* buffered READ or buffered write_begin() lookup */
+	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
+		/*
+		 * We have blocks reserved already.  We
+		 * return allocated blocks so that delalloc
+		 * won't do block reservation for us.  But
+		 * the buffer head will be unmapped so that
+		 * a read from the block returns 0s.
+		 */
+		set_buffer_unwritten(bh_result);
+		goto out1;
+	}
 
+	/* buffered write, writepage time, convert*/
+	ret = ext4_ext_convert_to_initialized(handle, inode,
+						path, iblock,
+						max_blocks);
+out:
+	if (ret <= 0) {
+		err = ret;
+		goto out2;
+	} else
+		allocated = ret;
+	set_buffer_new(bh_result);
+map_out:
+	set_buffer_mapped(bh_result);
+out1:
+	if (allocated > max_blocks)
+		allocated = max_blocks;
+	ext4_ext_show_leaf(inode, path);
+	bh_result->b_bdev = inode->i_sb->s_bdev;
+	bh_result->b_blocknr = newblock;
+out2:
+	if (path) {
+		ext4_ext_drop_refs(path);
+		kfree(path);
+	}
+	return err ? err : allocated;
+}
 /*
  * Block allocation/map/preallocation routine for extents based files
  *
@@ -2784,9 +3136,10 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 	int err = 0, depth, ret, cache_type;
 	unsigned int allocated = 0;
 	struct ext4_allocation_request ar;
+	ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
 
 	__clear_bit(BH_New, &bh_result->b_state);
-	ext_debug("blocks %u/%u requested for inode %u\n",
+	ext_debug("blocks %u/%u requested for inode %lu\n",
 			iblock, max_blocks, inode->i_ino);
 
 	/* check in cache */
@@ -2849,7 +3202,7 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 			newblock = iblock - ee_block + ee_start;
 			/* number of remaining blocks in the extent */
 			allocated = ee_len - (iblock - ee_block);
-			ext_debug("%u fit into %lu:%d -> %llu\n", iblock,
+			ext_debug("%u fit into %u:%d -> %llu\n", iblock,
 					ee_block, ee_len, newblock);
 
 			/* Do not put uninitialized extent in the cache */
@@ -2859,33 +3212,10 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 							EXT4_EXT_CACHE_EXTENT);
 				goto out;
 			}
-			if (flags & EXT4_GET_BLOCKS_UNINIT_EXT)
-				goto out;
-			if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
-				if (allocated > max_blocks)
-					allocated = max_blocks;
-				/*
-				 * We have blocks reserved already.  We
-				 * return allocated blocks so that delalloc
-				 * won't do block reservation for us.  But
-				 * the buffer head will be unmapped so that
-				 * a read from the block returns 0s.
-				 */
-				set_buffer_unwritten(bh_result);
-				bh_result->b_bdev = inode->i_sb->s_bdev;
-				bh_result->b_blocknr = newblock;
-				goto out2;
-			}
-
-			ret = ext4_ext_convert_to_initialized(handle, inode,
-								path, iblock,
-								max_blocks);
-			if (ret <= 0) {
-				err = ret;
-				goto out2;
-			} else
-				allocated = ret;
-			goto outnew;
+			ret = ext4_ext_handle_uninitialized_extents(handle,
+					inode, iblock, max_blocks, path,
+					flags, allocated, bh_result, newblock);
+			return ret;
 		}
 	}
 
@@ -2950,15 +3280,27 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 	newblock = ext4_mb_new_blocks(handle, &ar, &err);
 	if (!newblock)
 		goto out2;
-	ext_debug("allocate new block: goal %llu, found %llu/%lu\n",
+	ext_debug("allocate new block: goal %llu, found %llu/%u\n",
 		  ar.goal, newblock, allocated);
 
 	/* try to insert new extent into found leaf and return */
 	ext4_ext_store_pblock(&newex, newblock);
 	newex.ee_len = cpu_to_le16(ar.len);
-	if (flags & EXT4_GET_BLOCKS_UNINIT_EXT)  /* Mark uninitialized */
+	/* Mark uninitialized */
+	if (flags & EXT4_GET_BLOCKS_UNINIT_EXT){
 		ext4_ext_mark_uninitialized(&newex);
-	err = ext4_ext_insert_extent(handle, inode, path, &newex);
+		/*
+		 * io_end structure was created for every async
+		 * direct IO write to the middle of the file.
+		 * To avoid unecessary convertion for every aio dio rewrite
+		 * to the mid of file, here we flag the IO that is really
+		 * need the convertion.
+		 *
+		 */
+		if (io && flags == EXT4_GET_BLOCKS_DIO_CREATE_EXT)
+			io->flag = DIO_AIO_UNWRITTEN;
+	}
+	err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
 	if (err) {
 		/* free data blocks we just allocated */
 		/* not a good idea to call discard here directly,
@@ -2972,7 +3314,6 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 	/* previous routine could use block we allocated */
 	newblock = ext_pblock(&newex);
 	allocated = ext4_ext_get_actual_len(&newex);
-outnew:
 	set_buffer_new(bh_result);
 
 	/* Cache only when it is _not_ an uninitialized extent */
@@ -3171,6 +3512,63 @@ retry:
 }
 
 /*
+ * This function convert a range of blocks to written extents
+ * The caller of this function will pass the start offset and the size.
+ * all unwritten extents within this range will be converted to
+ * written extents.
+ *
+ * This function is called from the direct IO end io call back
+ * function, to convert the fallocated extents after IO is completed.
+ */
+int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
+				    loff_t len)
+{
+	handle_t *handle;
+	ext4_lblk_t block;
+	unsigned int max_blocks;
+	int ret = 0;
+	int ret2 = 0;
+	struct buffer_head map_bh;
+	unsigned int credits, blkbits = inode->i_blkbits;
+
+	block = offset >> blkbits;
+	/*
+	 * We can't just convert len to max_blocks because
+	 * If blocksize = 4096 offset = 3072 and len = 2048
+	 */
+	max_blocks = (EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits)
+							- block;
+	/*
+	 * credits to insert 1 extent into extent tree
+	 */
+	credits = ext4_chunk_trans_blocks(inode, max_blocks);
+	while (ret >= 0 && ret < max_blocks) {
+		block = block + ret;
+		max_blocks = max_blocks - ret;
+		handle = ext4_journal_start(inode, credits);
+		if (IS_ERR(handle)) {
+			ret = PTR_ERR(handle);
+			break;
+		}
+		map_bh.b_state = 0;
+		ret = ext4_get_blocks(handle, inode, block,
+				      max_blocks, &map_bh,
+				      EXT4_GET_BLOCKS_DIO_CONVERT_EXT);
+		if (ret <= 0) {
+			WARN_ON(ret <= 0);
+			printk(KERN_ERR "%s: ext4_ext_get_blocks "
+				    "returned error inode#%lu, block=%u, "
+				    "max_blocks=%u", __func__,
+				    inode->i_ino, block, max_blocks);
+		}
+		ext4_mark_inode_dirty(handle, inode);
+		ret2 = ext4_journal_stop(handle);
+		if (ret <= 0 || ret2 )
+			break;
+	}
+	return ret > 0 ? ret2 : ret;
+}
+/*
  * Callback function called for each extent to gather FIEMAP information.
  */
 static int ext4_ext_fiemap_cb(struct inode *inode, struct ext4_ext_path *path,
diff --git a/fs/ext4/file.c b/fs/ext4/file.c
index 5ca3eca70a1e..9630583cef28 100644
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -81,7 +81,7 @@ ext4_file_write(struct kiocb *iocb, const struct iovec *iov,
 	return generic_file_aio_write(iocb, iov, nr_segs, pos);
 }
 
-static struct vm_operations_struct ext4_file_vm_ops = {
+static const struct vm_operations_struct ext4_file_vm_ops = {
 	.fault		= filemap_fault,
 	.page_mkwrite   = ext4_page_mkwrite,
 };
diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
index 83cf6415f599..2b1531266ee2 100644
--- a/fs/ext4/fsync.c
+++ b/fs/ext4/fsync.c
@@ -44,18 +44,23 @@
  *
  * What we do is just kick off a commit and wait on it.  This will snapshot the
  * inode to disk.
+ *
+ * i_mutex lock is held when entering and exiting this function
  */
 
 int ext4_sync_file(struct file *file, struct dentry *dentry, int datasync)
 {
 	struct inode *inode = dentry->d_inode;
 	journal_t *journal = EXT4_SB(inode->i_sb)->s_journal;
-	int ret = 0;
+	int err, ret = 0;
 
 	J_ASSERT(ext4_journal_current_handle() == NULL);
 
 	trace_ext4_sync_file(file, dentry, datasync);
 
+	ret = flush_aio_dio_completed_IO(inode);
+	if (ret < 0)
+		goto out;
 	/*
 	 * data=writeback:
 	 *  The caller's filemap_fdatawrite()/wait will sync the data.
@@ -79,6 +84,9 @@ int ext4_sync_file(struct file *file, struct dentry *dentry, int datasync)
 		goto out;
 	}
 
+	if (!journal)
+		ret = sync_mapping_buffers(inode->i_mapping);
+
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
 		goto out;
 
@@ -91,10 +99,12 @@ int ext4_sync_file(struct file *file, struct dentry *dentry, int datasync)
 			.sync_mode = WB_SYNC_ALL,
 			.nr_to_write = 0, /* sys_fsync did this */
 		};
-		ret = sync_inode(inode, &wbc);
-		if (journal && (journal->j_flags & JBD2_BARRIER))
-			blkdev_issue_flush(inode->i_sb->s_bdev, NULL);
+		err = sync_inode(inode, &wbc);
+		if (ret == 0)
+			ret = err;
 	}
 out:
+	if (journal && (journal->j_flags & JBD2_BARRIER))
+		blkdev_issue_flush(inode->i_sb->s_bdev, NULL);
 	return ret;
 }
diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c
index 29e6dc7299b8..f3624ead4f6c 100644
--- a/fs/ext4/ialloc.c
+++ b/fs/ext4/ialloc.c
@@ -1189,7 +1189,7 @@ unsigned long ext4_count_free_inodes(struct super_block *sb)
 
 		x = ext4_count_free(bitmap_bh, EXT4_INODES_PER_GROUP(sb) / 8);
 		printk(KERN_DEBUG "group %lu: stored = %d, counted = %lu\n",
-			i, ext4_free_inodes_count(sb, gdp), x);
+			(unsigned long) i, ext4_free_inodes_count(sb, gdp), x);
 		bitmap_count += x;
 	}
 	brelse(bitmap_bh);
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index f9c642b22efa..5c5bc5dafff8 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -37,6 +37,7 @@
 #include <linux/namei.h>
 #include <linux/uio.h>
 #include <linux/bio.h>
+#include <linux/workqueue.h>
 
 #include "ext4_jbd2.h"
 #include "xattr.h"
@@ -192,11 +193,24 @@ static int try_to_extend_transaction(handle_t *handle, struct inode *inode)
  * so before we call here everything must be consistently dirtied against
  * this transaction.
  */
-static int ext4_journal_test_restart(handle_t *handle, struct inode *inode)
+ int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
+				 int nblocks)
 {
+	int ret;
+
+	/*
+	 * Drop i_data_sem to avoid deadlock with ext4_get_blocks At this
+	 * moment, get_block can be called only for blocks inside i_size since
+	 * page cache has been already dropped and writes are blocked by
+	 * i_mutex. So we can safely drop the i_data_sem here.
+	 */
 	BUG_ON(EXT4_JOURNAL(inode) == NULL);
 	jbd_debug(2, "restarting handle %p\n", handle);
-	return ext4_journal_restart(handle, blocks_for_truncate(inode));
+	up_write(&EXT4_I(inode)->i_data_sem);
+	ret = ext4_journal_restart(handle, blocks_for_truncate(inode));
+	down_write(&EXT4_I(inode)->i_data_sem);
+
+	return ret;
 }
 
 /*
@@ -341,9 +355,7 @@ static int ext4_block_to_path(struct inode *inode,
 	int n = 0;
 	int final = 0;
 
-	if (i_block < 0) {
-		ext4_warning(inode->i_sb, "ext4_block_to_path", "block < 0");
-	} else if (i_block < direct_blocks) {
+	if (i_block < direct_blocks) {
 		offsets[n++] = i_block;
 		final = direct_blocks;
 	} else if ((i_block -= direct_blocks) < indirect_blocks) {
@@ -551,15 +563,21 @@ static ext4_fsblk_t ext4_find_near(struct inode *inode, Indirect *ind)
  *
  *	Normally this function find the preferred place for block allocation,
  *	returns it.
+ *	Because this is only used for non-extent files, we limit the block nr
+ *	to 32 bits.
  */
 static ext4_fsblk_t ext4_find_goal(struct inode *inode, ext4_lblk_t block,
 				   Indirect *partial)
 {
+	ext4_fsblk_t goal;
+
 	/*
 	 * XXX need to get goal block from mballoc's data structures
 	 */
 
-	return ext4_find_near(inode, partial);
+	goal = ext4_find_near(inode, partial);
+	goal = goal & EXT4_MAX_BLOCK_FILE_PHYS;
+	return goal;
 }
 
 /**
@@ -640,6 +658,8 @@ static int ext4_alloc_blocks(handle_t *handle, struct inode *inode,
 		if (*err)
 			goto failed_out;
 
+		BUG_ON(current_block + count > EXT4_MAX_BLOCK_FILE_PHYS);
+
 		target -= count;
 		/* allocate blocks for indirect blocks */
 		while (index < indirect_blks && count) {
@@ -674,6 +694,7 @@ static int ext4_alloc_blocks(handle_t *handle, struct inode *inode,
 		ar.flags = EXT4_MB_HINT_DATA;
 
 	current_block = ext4_mb_new_blocks(handle, &ar, err);
+	BUG_ON(current_block + ar.len > EXT4_MAX_BLOCK_FILE_PHYS);
 
 	if (*err && (target == blks)) {
 		/*
@@ -762,8 +783,9 @@ static int ext4_alloc_branch(handle_t *handle, struct inode *inode,
 		BUFFER_TRACE(bh, "call get_create_access");
 		err = ext4_journal_get_create_access(handle, bh);
 		if (err) {
+			/* Don't brelse(bh) here; it's done in
+			 * ext4_journal_forget() below */
 			unlock_buffer(bh);
-			brelse(bh);
 			goto failed;
 		}
 
@@ -1109,22 +1131,79 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used)
 		ext4_discard_preallocations(inode);
 }
 
-static int check_block_validity(struct inode *inode, sector_t logical,
-				sector_t phys, int len)
+static int check_block_validity(struct inode *inode, const char *msg,
+				sector_t logical, sector_t phys, int len)
 {
 	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), phys, len)) {
-		ext4_error(inode->i_sb, "check_block_validity",
+		ext4_error(inode->i_sb, msg,
 			   "inode #%lu logical block %llu mapped to %llu "
 			   "(size %d)", inode->i_ino,
 			   (unsigned long long) logical,
 			   (unsigned long long) phys, len);
-		WARN_ON(1);
 		return -EIO;
 	}
 	return 0;
 }
 
 /*
+ * Return the number of contiguous dirty pages in a given inode
+ * starting at page frame idx.
+ */
+static pgoff_t ext4_num_dirty_pages(struct inode *inode, pgoff_t idx,
+				    unsigned int max_pages)
+{
+	struct address_space *mapping = inode->i_mapping;
+	pgoff_t	index;
+	struct pagevec pvec;
+	pgoff_t num = 0;
+	int i, nr_pages, done = 0;
+
+	if (max_pages == 0)
+		return 0;
+	pagevec_init(&pvec, 0);
+	while (!done) {
+		index = idx;
+		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
+					      PAGECACHE_TAG_DIRTY,
+					      (pgoff_t)PAGEVEC_SIZE);
+		if (nr_pages == 0)
+			break;
+		for (i = 0; i < nr_pages; i++) {
+			struct page *page = pvec.pages[i];
+			struct buffer_head *bh, *head;
+
+			lock_page(page);
+			if (unlikely(page->mapping != mapping) ||
+			    !PageDirty(page) ||
+			    PageWriteback(page) ||
+			    page->index != idx) {
+				done = 1;
+				unlock_page(page);
+				break;
+			}
+			if (page_has_buffers(page)) {
+				bh = head = page_buffers(page);
+				do {
+					if (!buffer_delay(bh) &&
+					    !buffer_unwritten(bh))
+						done = 1;
+					bh = bh->b_this_page;
+				} while (!done && (bh != head));
+			}
+			unlock_page(page);
+			if (done)
+				break;
+			idx++;
+			num++;
+			if (num >= max_pages)
+				break;
+		}
+		pagevec_release(&pvec);
+	}
+	return num;
+}
+
+/*
  * The ext4_get_blocks() function tries to look up the requested blocks,
  * and returns if the blocks are already mapped.
  *
@@ -1155,6 +1234,9 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block,
 	clear_buffer_mapped(bh);
 	clear_buffer_unwritten(bh);
 
+	ext_debug("ext4_get_blocks(): inode %lu, flag %d, max_blocks %u,"
+		  "logical block %lu\n", inode->i_ino, flags, max_blocks,
+		  (unsigned long)block);
 	/*
 	 * Try to see if we can get the block without requesting a new
 	 * file system block.
@@ -1170,8 +1252,8 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block,
 	up_read((&EXT4_I(inode)->i_data_sem));
 
 	if (retval > 0 && buffer_mapped(bh)) {
-		int ret = check_block_validity(inode, block,
-					       bh->b_blocknr, retval);
+		int ret = check_block_validity(inode, "file system corruption",
+					       block, bh->b_blocknr, retval);
 		if (ret != 0)
 			return ret;
 	}
@@ -1235,8 +1317,7 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block,
 			 * i_data's format changing.  Force the migrate
 			 * to fail by clearing migrate flags
 			 */
-			EXT4_I(inode)->i_flags = EXT4_I(inode)->i_flags &
-							~EXT4_EXT_MIGRATE;
+			EXT4_I(inode)->i_state &= ~EXT4_STATE_EXT_MIGRATE;
 		}
 	}
 
@@ -1252,8 +1333,9 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block,
 
 	up_write((&EXT4_I(inode)->i_data_sem));
 	if (retval > 0 && buffer_mapped(bh)) {
-		int ret = check_block_validity(inode, block,
-					       bh->b_blocknr, retval);
+		int ret = check_block_validity(inode, "file system "
+					       "corruption after allocation",
+					       block, bh->b_blocknr, retval);
 		if (ret != 0)
 			return ret;
 	}
@@ -1776,11 +1858,11 @@ repeat:
 
 	if (ext4_claim_free_blocks(sbi, total)) {
 		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
+		vfs_dq_release_reservation_block(inode, total);
 		if (ext4_should_retry_alloc(inode->i_sb, &retries)) {
 			yield();
 			goto repeat;
 		}
-		vfs_dq_release_reservation_block(inode, total);
 		return -ENOSPC;
 	}
 	EXT4_I(inode)->i_reserved_data_blocks += nrblocks;
@@ -1863,18 +1945,6 @@ static void ext4_da_page_release_reservation(struct page *page,
  * Delayed allocation stuff
  */
 
-struct mpage_da_data {
-	struct inode *inode;
-	sector_t b_blocknr;		/* start block number of extent */
-	size_t b_size;			/* size of extent */
-	unsigned long b_state;		/* state of the extent */
-	unsigned long first_page, next_page;	/* extent of pages */
-	struct writeback_control *wbc;
-	int io_done;
-	int pages_written;
-	int retval;
-};
-
 /*
  * mpage_da_submit_io - walks through extent of pages and try to write
  * them with writepage() call back
@@ -2084,18 +2154,18 @@ static void ext4_da_block_invalidatepages(struct mpage_da_data *mpd,
 static void ext4_print_free_blocks(struct inode *inode)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
-	printk(KERN_EMERG "Total free blocks count %lld\n",
-			ext4_count_free_blocks(inode->i_sb));
-	printk(KERN_EMERG "Free/Dirty block details\n");
-	printk(KERN_EMERG "free_blocks=%lld\n",
-			(long long)percpu_counter_sum(&sbi->s_freeblocks_counter));
-	printk(KERN_EMERG "dirty_blocks=%lld\n",
-			(long long)percpu_counter_sum(&sbi->s_dirtyblocks_counter));
-	printk(KERN_EMERG "Block reservation details\n");
-	printk(KERN_EMERG "i_reserved_data_blocks=%u\n",
-			EXT4_I(inode)->i_reserved_data_blocks);
-	printk(KERN_EMERG "i_reserved_meta_blocks=%u\n",
-			EXT4_I(inode)->i_reserved_meta_blocks);
+	printk(KERN_CRIT "Total free blocks count %lld\n",
+	       ext4_count_free_blocks(inode->i_sb));
+	printk(KERN_CRIT "Free/Dirty block details\n");
+	printk(KERN_CRIT "free_blocks=%lld\n",
+	       (long long) percpu_counter_sum(&sbi->s_freeblocks_counter));
+	printk(KERN_CRIT "dirty_blocks=%lld\n",
+	       (long long) percpu_counter_sum(&sbi->s_dirtyblocks_counter));
+	printk(KERN_CRIT "Block reservation details\n");
+	printk(KERN_CRIT "i_reserved_data_blocks=%u\n",
+	       EXT4_I(inode)->i_reserved_data_blocks);
+	printk(KERN_CRIT "i_reserved_meta_blocks=%u\n",
+	       EXT4_I(inode)->i_reserved_meta_blocks);
 	return;
 }
 
@@ -2181,14 +2251,14 @@ static int mpage_da_map_blocks(struct mpage_da_data *mpd)
 		 * writepage and writepages will again try to write
 		 * the same.
 		 */
-		printk(KERN_EMERG "%s block allocation failed for inode %lu "
-				  "at logical offset %llu with max blocks "
-				  "%zd with error %d\n",
-				  __func__, mpd->inode->i_ino,
-				  (unsigned long long)next,
-				  mpd->b_size >> mpd->inode->i_blkbits, err);
-		printk(KERN_EMERG "This should not happen.!! "
-					"Data will be lost\n");
+		ext4_msg(mpd->inode->i_sb, KERN_CRIT,
+			 "delayed block allocation failed for inode %lu at "
+			 "logical offset %llu with max blocks %zd with "
+			 "error %d\n", mpd->inode->i_ino,
+			 (unsigned long long) next,
+			 mpd->b_size >> mpd->inode->i_blkbits, err);
+		printk(KERN_CRIT "This should not happen!!  "
+		       "Data will be lost\n");
 		if (err == -ENOSPC) {
 			ext4_print_free_blocks(mpd->inode);
 		}
@@ -2329,7 +2399,7 @@ static int __mpage_da_writepage(struct page *page,
 		/*
 		 * Rest of the page in the page_vec
 		 * redirty then and skip then. We will
-		 * try to to write them again after
+		 * try to write them again after
 		 * starting a new transaction
 		 */
 		redirty_page_for_writepage(wbc, page);
@@ -2735,8 +2805,11 @@ static int ext4_da_writepages(struct address_space *mapping,
 	int no_nrwrite_index_update;
 	int pages_written = 0;
 	long pages_skipped;
+	unsigned int max_pages;
 	int range_cyclic, cycled = 1, io_done = 0;
-	int needed_blocks, ret = 0, nr_to_writebump = 0;
+	int needed_blocks, ret = 0;
+	long desired_nr_to_write, nr_to_writebump = 0;
+	loff_t range_start = wbc->range_start;
 	struct ext4_sb_info *sbi = EXT4_SB(mapping->host->i_sb);
 
 	trace_ext4_da_writepages(inode, wbc);
@@ -2762,16 +2835,6 @@ static int ext4_da_writepages(struct address_space *mapping,
 	if (unlikely(sbi->s_mount_flags & EXT4_MF_FS_ABORTED))
 		return -EROFS;
 
-	/*
-	 * Make sure nr_to_write is >= sbi->s_mb_stream_request
-	 * This make sure small files blocks are allocated in
-	 * single attempt. This ensure that small files
-	 * get less fragmented.
-	 */
-	if (wbc->nr_to_write < sbi->s_mb_stream_request) {
-		nr_to_writebump = sbi->s_mb_stream_request - wbc->nr_to_write;
-		wbc->nr_to_write = sbi->s_mb_stream_request;
-	}
 	if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
 		range_whole = 1;
 
@@ -2786,6 +2849,36 @@ static int ext4_da_writepages(struct address_space *mapping,
 	} else
 		index = wbc->range_start >> PAGE_CACHE_SHIFT;
 
+	/*
+	 * This works around two forms of stupidity.  The first is in
+	 * the writeback code, which caps the maximum number of pages
+	 * written to be 1024 pages.  This is wrong on multiple
+	 * levels; different architectues have a different page size,
+	 * which changes the maximum amount of data which gets
+	 * written.  Secondly, 4 megabytes is way too small.  XFS
+	 * forces this value to be 16 megabytes by multiplying
+	 * nr_to_write parameter by four, and then relies on its
+	 * allocator to allocate larger extents to make them
+	 * contiguous.  Unfortunately this brings us to the second
+	 * stupidity, which is that ext4's mballoc code only allocates
+	 * at most 2048 blocks.  So we force contiguous writes up to
+	 * the number of dirty blocks in the inode, or
+	 * sbi->max_writeback_mb_bump whichever is smaller.
+	 */
+	max_pages = sbi->s_max_writeback_mb_bump << (20 - PAGE_CACHE_SHIFT);
+	if (!range_cyclic && range_whole)
+		desired_nr_to_write = wbc->nr_to_write * 8;
+	else
+		desired_nr_to_write = ext4_num_dirty_pages(inode, index,
+							   max_pages);
+	if (desired_nr_to_write > max_pages)
+		desired_nr_to_write = max_pages;
+
+	if (wbc->nr_to_write < desired_nr_to_write) {
+		nr_to_writebump = desired_nr_to_write - wbc->nr_to_write;
+		wbc->nr_to_write = desired_nr_to_write;
+	}
+
 	mpd.wbc = wbc;
 	mpd.inode = mapping->host;
 
@@ -2813,10 +2906,9 @@ retry:
 		handle = ext4_journal_start(inode, needed_blocks);
 		if (IS_ERR(handle)) {
 			ret = PTR_ERR(handle);
-			printk(KERN_CRIT "%s: jbd2_start: "
+			ext4_msg(inode->i_sb, KERN_CRIT, "%s: jbd2_start: "
 			       "%ld pages, ino %lu; err %d\n", __func__,
 				wbc->nr_to_write, inode->i_ino, ret);
-			dump_stack();
 			goto out_writepages;
 		}
 
@@ -2850,6 +2942,7 @@ retry:
 			mpd.io_done = 1;
 			ret = MPAGE_DA_EXTENT_TAIL;
 		}
+		trace_ext4_da_write_pages(inode, &mpd);
 		wbc->nr_to_write -= mpd.pages_written;
 
 		ext4_journal_stop(handle);
@@ -2887,9 +2980,10 @@ retry:
 		goto retry;
 	}
 	if (pages_skipped != wbc->pages_skipped)
-		printk(KERN_EMERG "This should not happen leaving %s "
-				"with nr_to_write = %ld ret = %d\n",
-				__func__, wbc->nr_to_write, ret);
+		ext4_msg(inode->i_sb, KERN_CRIT,
+			 "This should not happen leaving %s "
+			 "with nr_to_write = %ld ret = %d\n",
+			 __func__, wbc->nr_to_write, ret);
 
 	/* Update index */
 	index += pages_written;
@@ -2904,7 +2998,9 @@ retry:
 out_writepages:
 	if (!no_nrwrite_index_update)
 		wbc->no_nrwrite_index_update = 0;
-	wbc->nr_to_write -= nr_to_writebump;
+	if (wbc->nr_to_write > nr_to_writebump)
+		wbc->nr_to_write -= nr_to_writebump;
+	wbc->range_start = range_start;
 	trace_ext4_da_writepages_result(inode, wbc, ret, pages_written);
 	return ret;
 }
@@ -3117,6 +3213,8 @@ out:
  */
 int ext4_alloc_da_blocks(struct inode *inode)
 {
+	trace_ext4_alloc_da_blocks(inode);
+
 	if (!EXT4_I(inode)->i_reserved_data_blocks &&
 	    !EXT4_I(inode)->i_reserved_meta_blocks)
 		return 0;
@@ -3259,6 +3357,8 @@ static int ext4_releasepage(struct page *page, gfp_t wait)
 }
 
 /*
+ * O_DIRECT for ext3 (or indirect map) based files
+ *
  * If the O_DIRECT write will extend the file then add this inode to the
  * orphan list.  So recovery will truncate it back to the original size
  * if the machine crashes during the write.
@@ -3267,7 +3367,7 @@ static int ext4_releasepage(struct page *page, gfp_t wait)
  * crashes then stale disk data _may_ be exposed inside the file. But current
  * VFS code falls back into buffered path in that case so we are safe.
  */
-static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
+static ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
 			      const struct iovec *iov, loff_t offset,
 			      unsigned long nr_segs)
 {
@@ -3278,6 +3378,7 @@ static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
 	ssize_t ret;
 	int orphan = 0;
 	size_t count = iov_length(iov, nr_segs);
+	int retries = 0;
 
 	if (rw == WRITE) {
 		loff_t final_size = offset + count;
@@ -3300,9 +3401,12 @@ static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
 		}
 	}
 
+retry:
 	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
 				 offset, nr_segs,
 				 ext4_get_block, NULL);
+	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
+		goto retry;
 
 	if (orphan) {
 		int err;
@@ -3341,6 +3445,359 @@ out:
 	return ret;
 }
 
+/* Maximum number of blocks we map for direct IO at once. */
+
+static int ext4_get_block_dio_write(struct inode *inode, sector_t iblock,
+		   struct buffer_head *bh_result, int create)
+{
+	handle_t *handle = NULL;
+	int ret = 0;
+	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;
+	int dio_credits;
+
+	ext4_debug("ext4_get_block_dio_write: inode %lu, create flag %d\n",
+		   inode->i_ino, create);
+	/*
+	 * DIO VFS code passes create = 0 flag for write to
+	 * the middle of file. It does this to avoid block
+	 * allocation for holes, to prevent expose stale data
+	 * out when there is parallel buffered read (which does
+	 * not hold the i_mutex lock) while direct IO write has
+	 * not completed. DIO request on holes finally falls back
+	 * to buffered IO for this reason.
+	 *
+	 * For ext4 extent based file, since we support fallocate,
+	 * new allocated extent as uninitialized, for holes, we
+	 * could fallocate blocks for holes, thus parallel
+	 * buffered IO read will zero out the page when read on
+	 * a hole while parallel DIO write to the hole has not completed.
+	 *
+	 * when we come here, we know it's a direct IO write to
+	 * to the middle of file (<i_size)
+	 * so it's safe to override the create flag from VFS.
+	 */
+	create = EXT4_GET_BLOCKS_DIO_CREATE_EXT;
+
+	if (max_blocks > DIO_MAX_BLOCKS)
+		max_blocks = DIO_MAX_BLOCKS;
+	dio_credits = ext4_chunk_trans_blocks(inode, max_blocks);
+	handle = ext4_journal_start(inode, dio_credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		goto out;
+	}
+	ret = ext4_get_blocks(handle, inode, iblock, max_blocks, bh_result,
+			      create);
+	if (ret > 0) {
+		bh_result->b_size = (ret << inode->i_blkbits);
+		ret = 0;
+	}
+	ext4_journal_stop(handle);
+out:
+	return ret;
+}
+
+static void ext4_free_io_end(ext4_io_end_t *io)
+{
+	BUG_ON(!io);
+	iput(io->inode);
+	kfree(io);
+}
+static void dump_aio_dio_list(struct inode * inode)
+{
+#ifdef	EXT4_DEBUG
+	struct list_head *cur, *before, *after;
+	ext4_io_end_t *io, *io0, *io1;
+
+	if (list_empty(&EXT4_I(inode)->i_aio_dio_complete_list)){
+		ext4_debug("inode %lu aio dio list is empty\n", inode->i_ino);
+		return;
+	}
+
+	ext4_debug("Dump inode %lu aio_dio_completed_IO list \n", inode->i_ino);
+	list_for_each_entry(io, &EXT4_I(inode)->i_aio_dio_complete_list, list){
+		cur = &io->list;
+		before = cur->prev;
+		io0 = container_of(before, ext4_io_end_t, list);
+		after = cur->next;
+		io1 = container_of(after, ext4_io_end_t, list);
+
+		ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
+			    io, inode->i_ino, io0, io1);
+	}
+#endif
+}
+
+/*
+ * check a range of space and convert unwritten extents to written.
+ */
+static int ext4_end_aio_dio_nolock(ext4_io_end_t *io)
+{
+	struct inode *inode = io->inode;
+	loff_t offset = io->offset;
+	size_t size = io->size;
+	int ret = 0;
+
+	ext4_debug("end_aio_dio_onlock: io 0x%p from inode %lu,list->next 0x%p,"
+		   "list->prev 0x%p\n",
+	           io, inode->i_ino, io->list.next, io->list.prev);
+
+	if (list_empty(&io->list))
+		return ret;
+
+	if (io->flag != DIO_AIO_UNWRITTEN)
+		return ret;
+
+	if (offset + size <= i_size_read(inode))
+		ret = ext4_convert_unwritten_extents(inode, offset, size);
+
+	if (ret < 0) {
+		printk(KERN_EMERG "%s: failed to convert unwritten"
+			"extents to written extents, error is %d"
+			" io is still on inode %lu aio dio list\n",
+                       __func__, ret, inode->i_ino);
+		return ret;
+	}
+
+	/* clear the DIO AIO unwritten flag */
+	io->flag = 0;
+	return ret;
+}
+/*
+ * work on completed aio dio IO, to convert unwritten extents to extents
+ */
+static void ext4_end_aio_dio_work(struct work_struct *work)
+{
+	ext4_io_end_t *io  = container_of(work, ext4_io_end_t, work);
+	struct inode *inode = io->inode;
+	int ret = 0;
+
+	mutex_lock(&inode->i_mutex);
+	ret = ext4_end_aio_dio_nolock(io);
+	if (ret >= 0) {
+		if (!list_empty(&io->list))
+			list_del_init(&io->list);
+		ext4_free_io_end(io);
+	}
+	mutex_unlock(&inode->i_mutex);
+}
+/*
+ * This function is called from ext4_sync_file().
+ *
+ * When AIO DIO IO is completed, the work to convert unwritten
+ * extents to written is queued on workqueue but may not get immediately
+ * scheduled. When fsync is called, we need to ensure the
+ * conversion is complete before fsync returns.
+ * The inode keeps track of a list of completed AIO from DIO path
+ * that might needs to do the conversion. This function walks through
+ * the list and convert the related unwritten extents to written.
+ */
+int flush_aio_dio_completed_IO(struct inode *inode)
+{
+	ext4_io_end_t *io;
+	int ret = 0;
+	int ret2 = 0;
+
+	if (list_empty(&EXT4_I(inode)->i_aio_dio_complete_list))
+		return ret;
+
+	dump_aio_dio_list(inode);
+	while (!list_empty(&EXT4_I(inode)->i_aio_dio_complete_list)){
+		io = list_entry(EXT4_I(inode)->i_aio_dio_complete_list.next,
+				ext4_io_end_t, list);
+		/*
+		 * Calling ext4_end_aio_dio_nolock() to convert completed
+		 * IO to written.
+		 *
+		 * When ext4_sync_file() is called, run_queue() may already
+		 * about to flush the work corresponding to this io structure.
+		 * It will be upset if it founds the io structure related
+		 * to the work-to-be schedule is freed.
+		 *
+		 * Thus we need to keep the io structure still valid here after
+		 * convertion finished. The io structure has a flag to
+		 * avoid double converting from both fsync and background work
+		 * queue work.
+		 */
+		ret = ext4_end_aio_dio_nolock(io);
+		if (ret < 0)
+			ret2 = ret;
+		else
+			list_del_init(&io->list);
+	}
+	return (ret2 < 0) ? ret2 : 0;
+}
+
+static ext4_io_end_t *ext4_init_io_end (struct inode *inode)
+{
+	ext4_io_end_t *io = NULL;
+
+	io = kmalloc(sizeof(*io), GFP_NOFS);
+
+	if (io) {
+		igrab(inode);
+		io->inode = inode;
+		io->flag = 0;
+		io->offset = 0;
+		io->size = 0;
+		io->error = 0;
+		INIT_WORK(&io->work, ext4_end_aio_dio_work);
+		INIT_LIST_HEAD(&io->list);
+	}
+
+	return io;
+}
+
+static void ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
+			    ssize_t size, void *private)
+{
+        ext4_io_end_t *io_end = iocb->private;
+	struct workqueue_struct *wq;
+
+	ext_debug("ext4_end_io_dio(): io_end 0x%p"
+		  "for inode %lu, iocb 0x%p, offset %llu, size %llu\n",
+ 		  iocb->private, io_end->inode->i_ino, iocb, offset,
+		  size);
+	/* if not async direct IO or dio with 0 bytes write, just return */
+	if (!io_end || !size)
+		return;
+
+	/* if not aio dio with unwritten extents, just free io and return */
+	if (io_end->flag != DIO_AIO_UNWRITTEN){
+		ext4_free_io_end(io_end);
+		iocb->private = NULL;
+		return;
+	}
+
+	io_end->offset = offset;
+	io_end->size = size;
+	wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;
+
+	/* queue the work to convert unwritten extents to written */
+	queue_work(wq, &io_end->work);
+
+	/* Add the io_end to per-inode completed aio dio list*/
+	list_add_tail(&io_end->list,
+		 &EXT4_I(io_end->inode)->i_aio_dio_complete_list);
+	iocb->private = NULL;
+}
+/*
+ * For ext4 extent files, ext4 will do direct-io write to holes,
+ * preallocated extents, and those write extend the file, no need to
+ * fall back to buffered IO.
+ *
+ * For holes, we fallocate those blocks, mark them as unintialized
+ * If those blocks were preallocated, we mark sure they are splited, but
+ * still keep the range to write as unintialized.
+ *
+ * The unwrritten extents will be converted to written when DIO is completed.
+ * For async direct IO, since the IO may still pending when return, we
+ * set up an end_io call back function, which will do the convertion
+ * when async direct IO completed.
+ *
+ * If the O_DIRECT write will extend the file then add this inode to the
+ * orphan list.  So recovery will truncate it back to the original size
+ * if the machine crashes during the write.
+ *
+ */
+static ssize_t ext4_ext_direct_IO(int rw, struct kiocb *iocb,
+			      const struct iovec *iov, loff_t offset,
+			      unsigned long nr_segs)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file->f_mapping->host;
+	ssize_t ret;
+	size_t count = iov_length(iov, nr_segs);
+
+	loff_t final_size = offset + count;
+	if (rw == WRITE && final_size <= inode->i_size) {
+		/*
+ 		 * We could direct write to holes and fallocate.
+		 *
+ 		 * Allocated blocks to fill the hole are marked as uninitialized
+ 		 * to prevent paralel buffered read to expose the stale data
+ 		 * before DIO complete the data IO.
+		 *
+ 		 * As to previously fallocated extents, ext4 get_block
+ 		 * will just simply mark the buffer mapped but still
+ 		 * keep the extents uninitialized.
+ 		 *
+		 * for non AIO case, we will convert those unwritten extents
+		 * to written after return back from blockdev_direct_IO.
+		 *
+		 * for async DIO, the conversion needs to be defered when
+		 * the IO is completed. The ext4 end_io callback function
+		 * will be called to take care of the conversion work.
+		 * Here for async case, we allocate an io_end structure to
+		 * hook to the iocb.
+ 		 */
+		iocb->private = NULL;
+		EXT4_I(inode)->cur_aio_dio = NULL;
+		if (!is_sync_kiocb(iocb)) {
+			iocb->private = ext4_init_io_end(inode);
+			if (!iocb->private)
+				return -ENOMEM;
+			/*
+			 * we save the io structure for current async
+			 * direct IO, so that later ext4_get_blocks()
+			 * could flag the io structure whether there
+			 * is a unwritten extents needs to be converted
+			 * when IO is completed.
+			 */
+			EXT4_I(inode)->cur_aio_dio = iocb->private;
+		}
+
+		ret = blockdev_direct_IO(rw, iocb, inode,
+					 inode->i_sb->s_bdev, iov,
+					 offset, nr_segs,
+					 ext4_get_block_dio_write,
+					 ext4_end_io_dio);
+		if (iocb->private)
+			EXT4_I(inode)->cur_aio_dio = NULL;
+		/*
+		 * The io_end structure takes a reference to the inode,
+		 * that structure needs to be destroyed and the
+		 * reference to the inode need to be dropped, when IO is
+		 * complete, even with 0 byte write, or failed.
+		 *
+		 * In the successful AIO DIO case, the io_end structure will be
+		 * desctroyed and the reference to the inode will be dropped
+		 * after the end_io call back function is called.
+		 *
+		 * In the case there is 0 byte write, or error case, since
+		 * VFS direct IO won't invoke the end_io call back function,
+		 * we need to free the end_io structure here.
+		 */
+		if (ret != -EIOCBQUEUED && ret <= 0 && iocb->private) {
+			ext4_free_io_end(iocb->private);
+			iocb->private = NULL;
+		} else if (ret > 0)
+			/*
+			 * for non AIO case, since the IO is already
+			 * completed, we could do the convertion right here
+			 */
+			ret = ext4_convert_unwritten_extents(inode,
+								offset, ret);
+		return ret;
+	}
+
+	/* for write the the end of file case, we fall back to old way */
+	return ext4_ind_direct_IO(rw, iocb, iov, offset, nr_segs);
+}
+
+static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
+			      const struct iovec *iov, loff_t offset,
+			      unsigned long nr_segs)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file->f_mapping->host;
+
+	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)
+		return ext4_ext_direct_IO(rw, iocb, iov, offset, nr_segs);
+
+	return ext4_ind_direct_IO(rw, iocb, iov, offset, nr_segs);
+}
+
 /*
  * Pages can be marked dirty completely asynchronously from ext4's journalling
  * activity.  By filemap_sync_pte(), try_to_unmap_one(), etc.  We cannot do
@@ -3373,6 +3830,7 @@ static const struct address_space_operations ext4_ordered_aops = {
 	.direct_IO		= ext4_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 static const struct address_space_operations ext4_writeback_aops = {
@@ -3388,6 +3846,7 @@ static const struct address_space_operations ext4_writeback_aops = {
 	.direct_IO		= ext4_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 static const struct address_space_operations ext4_journalled_aops = {
@@ -3402,6 +3861,7 @@ static const struct address_space_operations ext4_journalled_aops = {
 	.invalidatepage		= ext4_invalidatepage,
 	.releasepage		= ext4_releasepage,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 static const struct address_space_operations ext4_da_aops = {
@@ -3418,6 +3878,7 @@ static const struct address_space_operations ext4_da_aops = {
 	.direct_IO		= ext4_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
 
 void ext4_set_aops(struct inode *inode)
@@ -3659,7 +4120,8 @@ static void ext4_clear_blocks(handle_t *handle, struct inode *inode,
 			ext4_handle_dirty_metadata(handle, inode, bh);
 		}
 		ext4_mark_inode_dirty(handle, inode);
-		ext4_journal_test_restart(handle, inode);
+		ext4_truncate_restart_trans(handle, inode,
+					    blocks_for_truncate(inode));
 		if (bh) {
 			BUFFER_TRACE(bh, "retaking write access");
 			ext4_journal_get_write_access(handle, bh);
@@ -3870,7 +4332,8 @@ static void ext4_free_branches(handle_t *handle, struct inode *inode,
 				return;
 			if (try_to_extend_transaction(handle, inode)) {
 				ext4_mark_inode_dirty(handle, inode);
-				ext4_journal_test_restart(handle, inode);
+				ext4_truncate_restart_trans(handle, inode,
+					    blocks_for_truncate(inode));
 			}
 
 			ext4_free_blocks(handle, inode, nr, 1, 1);
@@ -3958,8 +4421,7 @@ void ext4_truncate(struct inode *inode)
 	if (!ext4_can_truncate(inode))
 		return;
 
-	if (ei->i_disksize && inode->i_size == 0 &&
-	    !test_opt(inode->i_sb, NO_AUTO_DA_ALLOC))
+	if (inode->i_size == 0 && !test_opt(inode->i_sb, NO_AUTO_DA_ALLOC))
 		ei->i_state |= EXT4_STATE_DA_ALLOC_CLOSE;
 
 	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
@@ -4581,8 +5043,7 @@ static int ext4_do_update_inode(handle_t *handle,
 	if (ext4_inode_blocks_set(handle, raw_inode, ei))
 		goto out_brelse;
 	raw_inode->i_dtime = cpu_to_le32(ei->i_dtime);
-	/* clear the migrate flag in the raw_inode */
-	raw_inode->i_flags = cpu_to_le32(ei->i_flags & ~EXT4_EXT_MIGRATE);
+	raw_inode->i_flags = cpu_to_le32(ei->i_flags);
 	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
 	    cpu_to_le32(EXT4_OS_HURD))
 		raw_inode->i_file_acl_high =
@@ -4684,19 +5145,40 @@ out_brelse:
  */
 int ext4_write_inode(struct inode *inode, int wait)
 {
+	int err;
+
 	if (current->flags & PF_MEMALLOC)
 		return 0;
 
-	if (ext4_journal_current_handle()) {
-		jbd_debug(1, "called recursively, non-PF_MEMALLOC!\n");
-		dump_stack();
-		return -EIO;
-	}
+	if (EXT4_SB(inode->i_sb)->s_journal) {
+		if (ext4_journal_current_handle()) {
+			jbd_debug(1, "called recursively, non-PF_MEMALLOC!\n");
+			dump_stack();
+			return -EIO;
+		}
 
-	if (!wait)
-		return 0;
+		if (!wait)
+			return 0;
 
-	return ext4_force_commit(inode->i_sb);
+		err = ext4_force_commit(inode->i_sb);
+	} else {
+		struct ext4_iloc iloc;
+
+		err = ext4_get_inode_loc(inode, &iloc);
+		if (err)
+			return err;
+		if (wait)
+			sync_dirty_buffer(iloc.bh);
+		if (buffer_req(iloc.bh) && !buffer_uptodate(iloc.bh)) {
+			ext4_error(inode->i_sb, __func__,
+				   "IO error syncing inode, "
+				   "inode=%lu, block=%llu",
+				   inode->i_ino,
+				   (unsigned long long)iloc.bh->b_blocknr);
+			err = -EIO;
+		}
+	}
+	return err;
 }
 
 /*
@@ -5134,27 +5616,14 @@ int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
  */
 void ext4_dirty_inode(struct inode *inode)
 {
-	handle_t *current_handle = ext4_journal_current_handle();
 	handle_t *handle;
 
-	if (!ext4_handle_valid(current_handle)) {
-		ext4_mark_inode_dirty(current_handle, inode);
-		return;
-	}
-
 	handle = ext4_journal_start(inode, 2);
 	if (IS_ERR(handle))
 		goto out;
-	if (current_handle &&
-		current_handle->h_transaction != handle->h_transaction) {
-		/* This task has a transaction open against a different fs */
-		printk(KERN_EMERG "%s: transactions do not match!\n",
-		       __func__);
-	} else {
-		jbd_debug(5, "marking dirty.  outer handle=%p\n",
-				current_handle);
-		ext4_mark_inode_dirty(handle, inode);
-	}
+
+	ext4_mark_inode_dirty(handle, inode);
+
 	ext4_journal_stop(handle);
 out:
 	return;
@@ -5281,12 +5750,21 @@ int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	else
 		len = PAGE_CACHE_SIZE;
 
+	lock_page(page);
+	/*
+	 * return if we have all the buffers mapped. This avoid
+	 * the need to call write_begin/write_end which does a
+	 * journal_start/journal_stop which can block and take
+	 * long time
+	 */
 	if (page_has_buffers(page)) {
-		/* return if we have all the buffers mapped */
 		if (!walk_page_buffers(NULL, page_buffers(page), 0, len, NULL,
-				       ext4_bh_unmapped))
+					ext4_bh_unmapped)) {
+			unlock_page(page);
 			goto out_unlock;
+		}
 	}
+	unlock_page(page);
 	/*
 	 * OK, we need to fill the hole... Do write_begin write_end
 	 * to do block allocation/reservation.We are not holding
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index 7050a9cd04a4..c1cdf613e725 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -243,10 +243,9 @@ setversion_out:
 					me.donor_start, me.len, &me.moved_len);
 		fput(donor_filp);
 
-		if (!err)
-			if (copy_to_user((struct move_extent *)arg,
-				&me, sizeof(me)))
-				return -EFAULT;
+		if (copy_to_user((struct move_extent *)arg, &me, sizeof(me)))
+			return -EFAULT;
+
 		return err;
 	}
 
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index cd258463e2a9..bba12824defa 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -22,6 +22,7 @@
  */
 
 #include "mballoc.h"
+#include <linux/debugfs.h>
 #include <trace/events/ext4.h>
 
 /*
@@ -622,13 +623,13 @@ static int __mb_check_buddy(struct ext4_buddy *e4b, char *file,
 
 /* FIXME!! need more doc */
 static void ext4_mb_mark_free_simple(struct super_block *sb,
-				void *buddy, unsigned first, int len,
+				void *buddy, ext4_grpblk_t first, ext4_grpblk_t len,
 					struct ext4_group_info *grp)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
-	unsigned short min;
-	unsigned short max;
-	unsigned short chunk;
+	ext4_grpblk_t min;
+	ext4_grpblk_t max;
+	ext4_grpblk_t chunk;
 	unsigned short border;
 
 	BUG_ON(len > EXT4_BLOCKS_PER_GROUP(sb));
@@ -662,10 +663,10 @@ void ext4_mb_generate_buddy(struct super_block *sb,
 				void *buddy, void *bitmap, ext4_group_t group)
 {
 	struct ext4_group_info *grp = ext4_get_group_info(sb, group);
-	unsigned short max = EXT4_BLOCKS_PER_GROUP(sb);
-	unsigned short i = 0;
-	unsigned short first;
-	unsigned short len;
+	ext4_grpblk_t max = EXT4_BLOCKS_PER_GROUP(sb);
+	ext4_grpblk_t i = 0;
+	ext4_grpblk_t first;
+	ext4_grpblk_t len;
 	unsigned free = 0;
 	unsigned fragments = 0;
 	unsigned long long period = get_cycles();
@@ -743,7 +744,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore)
 	char *data;
 	char *bitmap;
 
-	mb_debug("init page %lu\n", page->index);
+	mb_debug(1, "init page %lu\n", page->index);
 
 	inode = page->mapping->host;
 	sb = inode->i_sb;
@@ -822,7 +823,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore)
 		set_bitmap_uptodate(bh[i]);
 		bh[i]->b_end_io = end_buffer_read_sync;
 		submit_bh(READ, bh[i]);
-		mb_debug("read bitmap for group %u\n", first_group + i);
+		mb_debug(1, "read bitmap for group %u\n", first_group + i);
 	}
 
 	/* wait for I/O completion */
@@ -862,12 +863,13 @@ static int ext4_mb_init_cache(struct page *page, char *incore)
 		if ((first_block + i) & 1) {
 			/* this is block of buddy */
 			BUG_ON(incore == NULL);
-			mb_debug("put buddy for group %u in page %lu/%x\n",
+			mb_debug(1, "put buddy for group %u in page %lu/%x\n",
 				group, page->index, i * blocksize);
 			grinfo = ext4_get_group_info(sb, group);
 			grinfo->bb_fragments = 0;
 			memset(grinfo->bb_counters, 0,
-			       sizeof(unsigned short)*(sb->s_blocksize_bits+2));
+			       sizeof(*grinfo->bb_counters) *
+				(sb->s_blocksize_bits+2));
 			/*
 			 * incore got set to the group block bitmap below
 			 */
@@ -878,7 +880,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore)
 		} else {
 			/* this is block of bitmap */
 			BUG_ON(incore != NULL);
-			mb_debug("put bitmap for group %u in page %lu/%x\n",
+			mb_debug(1, "put bitmap for group %u in page %lu/%x\n",
 				group, page->index, i * blocksize);
 
 			/* see comments in ext4_mb_put_pa() */
@@ -908,6 +910,100 @@ out:
 	return err;
 }
 
+static noinline_for_stack
+int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
+{
+
+	int ret = 0;
+	void *bitmap;
+	int blocks_per_page;
+	int block, pnum, poff;
+	int num_grp_locked = 0;
+	struct ext4_group_info *this_grp;
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+	struct inode *inode = sbi->s_buddy_cache;
+	struct page *page = NULL, *bitmap_page = NULL;
+
+	mb_debug(1, "init group %u\n", group);
+	blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
+	this_grp = ext4_get_group_info(sb, group);
+	/*
+	 * This ensures that we don't reinit the buddy cache
+	 * page which map to the group from which we are already
+	 * allocating. If we are looking at the buddy cache we would
+	 * have taken a reference using ext4_mb_load_buddy and that
+	 * would have taken the alloc_sem lock.
+	 */
+	num_grp_locked =  ext4_mb_get_buddy_cache_lock(sb, group);
+	if (!EXT4_MB_GRP_NEED_INIT(this_grp)) {
+		/*
+		 * somebody initialized the group
+		 * return without doing anything
+		 */
+		ret = 0;
+		goto err;
+	}
+	/*
+	 * the buddy cache inode stores the block bitmap
+	 * and buddy information in consecutive blocks.
+	 * So for each group we need two blocks.
+	 */
+	block = group * 2;
+	pnum = block / blocks_per_page;
+	poff = block % blocks_per_page;
+	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
+	if (page) {
+		BUG_ON(page->mapping != inode->i_mapping);
+		ret = ext4_mb_init_cache(page, NULL);
+		if (ret) {
+			unlock_page(page);
+			goto err;
+		}
+		unlock_page(page);
+	}
+	if (page == NULL || !PageUptodate(page)) {
+		ret = -EIO;
+		goto err;
+	}
+	mark_page_accessed(page);
+	bitmap_page = page;
+	bitmap = page_address(page) + (poff * sb->s_blocksize);
+
+	/* init buddy cache */
+	block++;
+	pnum = block / blocks_per_page;
+	poff = block % blocks_per_page;
+	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
+	if (page == bitmap_page) {
+		/*
+		 * If both the bitmap and buddy are in
+		 * the same page we don't need to force
+		 * init the buddy
+		 */
+		unlock_page(page);
+	} else if (page) {
+		BUG_ON(page->mapping != inode->i_mapping);
+		ret = ext4_mb_init_cache(page, bitmap);
+		if (ret) {
+			unlock_page(page);
+			goto err;
+		}
+		unlock_page(page);
+	}
+	if (page == NULL || !PageUptodate(page)) {
+		ret = -EIO;
+		goto err;
+	}
+	mark_page_accessed(page);
+err:
+	ext4_mb_put_buddy_cache_lock(sb, group, num_grp_locked);
+	if (bitmap_page)
+		page_cache_release(bitmap_page);
+	if (page)
+		page_cache_release(page);
+	return ret;
+}
+
 static noinline_for_stack int
 ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 					struct ext4_buddy *e4b)
@@ -922,7 +1018,7 @@ ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct inode *inode = sbi->s_buddy_cache;
 
-	mb_debug("load group %u\n", group);
+	mb_debug(1, "load group %u\n", group);
 
 	blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
 	grp = ext4_get_group_info(sb, group);
@@ -941,8 +1037,26 @@ ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 	 * groups mapped by the page is blocked
 	 * till we are done with allocation
 	 */
+repeat_load_buddy:
 	down_read(e4b->alloc_semp);
 
+	if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
+		/* we need to check for group need init flag
+		 * with alloc_semp held so that we can be sure
+		 * that new blocks didn't get added to the group
+		 * when we are loading the buddy cache
+		 */
+		up_read(e4b->alloc_semp);
+		/*
+		 * we need full data about the group
+		 * to make a good selection
+		 */
+		ret = ext4_mb_init_group(sb, group);
+		if (ret)
+			return ret;
+		goto repeat_load_buddy;
+	}
+
 	/*
 	 * the buddy cache inode stores the block bitmap
 	 * and buddy information in consecutive blocks.
@@ -1360,7 +1474,7 @@ static void ext4_mb_use_best_found(struct ext4_allocation_context *ac,
 	ac->alloc_semp =  e4b->alloc_semp;
 	e4b->alloc_semp = NULL;
 	/* store last allocated for subsequent stream allocation */
-	if ((ac->ac_flags & EXT4_MB_HINT_DATA)) {
+	if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) {
 		spin_lock(&sbi->s_md_lock);
 		sbi->s_mb_last_group = ac->ac_f_ex.fe_group;
 		sbi->s_mb_last_start = ac->ac_f_ex.fe_start;
@@ -1837,97 +1951,6 @@ void ext4_mb_put_buddy_cache_lock(struct super_block *sb,
 
 }
 
-static noinline_for_stack
-int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
-{
-
-	int ret;
-	void *bitmap;
-	int blocks_per_page;
-	int block, pnum, poff;
-	int num_grp_locked = 0;
-	struct ext4_group_info *this_grp;
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-	struct inode *inode = sbi->s_buddy_cache;
-	struct page *page = NULL, *bitmap_page = NULL;
-
-	mb_debug("init group %lu\n", group);
-	blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
-	this_grp = ext4_get_group_info(sb, group);
-	/*
-	 * This ensures we don't add group
-	 * to this buddy cache via resize
-	 */
-	num_grp_locked =  ext4_mb_get_buddy_cache_lock(sb, group);
-	if (!EXT4_MB_GRP_NEED_INIT(this_grp)) {
-		/*
-		 * somebody initialized the group
-		 * return without doing anything
-		 */
-		ret = 0;
-		goto err;
-	}
-	/*
-	 * the buddy cache inode stores the block bitmap
-	 * and buddy information in consecutive blocks.
-	 * So for each group we need two blocks.
-	 */
-	block = group * 2;
-	pnum = block / blocks_per_page;
-	poff = block % blocks_per_page;
-	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
-	if (page) {
-		BUG_ON(page->mapping != inode->i_mapping);
-		ret = ext4_mb_init_cache(page, NULL);
-		if (ret) {
-			unlock_page(page);
-			goto err;
-		}
-		unlock_page(page);
-	}
-	if (page == NULL || !PageUptodate(page)) {
-		ret = -EIO;
-		goto err;
-	}
-	mark_page_accessed(page);
-	bitmap_page = page;
-	bitmap = page_address(page) + (poff * sb->s_blocksize);
-
-	/* init buddy cache */
-	block++;
-	pnum = block / blocks_per_page;
-	poff = block % blocks_per_page;
-	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
-	if (page == bitmap_page) {
-		/*
-		 * If both the bitmap and buddy are in
-		 * the same page we don't need to force
-		 * init the buddy
-		 */
-		unlock_page(page);
-	} else if (page) {
-		BUG_ON(page->mapping != inode->i_mapping);
-		ret = ext4_mb_init_cache(page, bitmap);
-		if (ret) {
-			unlock_page(page);
-			goto err;
-		}
-		unlock_page(page);
-	}
-	if (page == NULL || !PageUptodate(page)) {
-		ret = -EIO;
-		goto err;
-	}
-	mark_page_accessed(page);
-err:
-	ext4_mb_put_buddy_cache_lock(sb, group, num_grp_locked);
-	if (bitmap_page)
-		page_cache_release(bitmap_page);
-	if (page)
-		page_cache_release(page);
-	return ret;
-}
-
 static noinline_for_stack int
 ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 {
@@ -1938,11 +1961,14 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 	struct ext4_sb_info *sbi;
 	struct super_block *sb;
 	struct ext4_buddy e4b;
-	loff_t size, isize;
 
 	sb = ac->ac_sb;
 	sbi = EXT4_SB(sb);
 	ngroups = ext4_get_groups_count(sb);
+	/* non-extent files are limited to low blocks/groups */
+	if (!(EXT4_I(ac->ac_inode)->i_flags & EXT4_EXTENTS_FL))
+		ngroups = sbi->s_blockfile_groups;
+
 	BUG_ON(ac->ac_status == AC_STATUS_FOUND);
 
 	/* first, try the goal */
@@ -1974,20 +2000,16 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 	}
 
 	bsbits = ac->ac_sb->s_blocksize_bits;
-	/* if stream allocation is enabled, use global goal */
-	size = ac->ac_o_ex.fe_logical + ac->ac_o_ex.fe_len;
-	isize = i_size_read(ac->ac_inode) >> bsbits;
-	if (size < isize)
-		size = isize;
 
-	if (size < sbi->s_mb_stream_request &&
-			(ac->ac_flags & EXT4_MB_HINT_DATA)) {
+	/* if stream allocation is enabled, use global goal */
+	if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) {
 		/* TBD: may be hot point */
 		spin_lock(&sbi->s_md_lock);
 		ac->ac_g_ex.fe_group = sbi->s_mb_last_group;
 		ac->ac_g_ex.fe_start = sbi->s_mb_last_start;
 		spin_unlock(&sbi->s_md_lock);
 	}
+
 	/* Let's just scan groups to find more-less suitable blocks */
 	cr = ac->ac_2order ? 0 : 1;
 	/*
@@ -2015,27 +2037,6 @@ repeat:
 			if (grp->bb_free == 0)
 				continue;
 
-			/*
-			 * if the group is already init we check whether it is
-			 * a good group and if not we don't load the buddy
-			 */
-			if (EXT4_MB_GRP_NEED_INIT(grp)) {
-				/*
-				 * we need full data about the group
-				 * to make a good selection
-				 */
-				err = ext4_mb_init_group(sb, group);
-				if (err)
-					goto out;
-			}
-
-			/*
-			 * If the particular group doesn't satisfy our
-			 * criteria we continue with the next group
-			 */
-			if (!ext4_mb_good_group(ac, group, cr))
-				continue;
-
 			err = ext4_mb_load_buddy(sb, group, &e4b);
 			if (err)
 				goto out;
@@ -2095,207 +2096,6 @@ out:
 	return err;
 }
 
-#ifdef EXT4_MB_HISTORY
-struct ext4_mb_proc_session {
-	struct ext4_mb_history *history;
-	struct super_block *sb;
-	int start;
-	int max;
-};
-
-static void *ext4_mb_history_skip_empty(struct ext4_mb_proc_session *s,
-					struct ext4_mb_history *hs,
-					int first)
-{
-	if (hs == s->history + s->max)
-		hs = s->history;
-	if (!first && hs == s->history + s->start)
-		return NULL;
-	while (hs->orig.fe_len == 0) {
-		hs++;
-		if (hs == s->history + s->max)
-			hs = s->history;
-		if (hs == s->history + s->start)
-			return NULL;
-	}
-	return hs;
-}
-
-static void *ext4_mb_seq_history_start(struct seq_file *seq, loff_t *pos)
-{
-	struct ext4_mb_proc_session *s = seq->private;
-	struct ext4_mb_history *hs;
-	int l = *pos;
-
-	if (l == 0)
-		return SEQ_START_TOKEN;
-	hs = ext4_mb_history_skip_empty(s, s->history + s->start, 1);
-	if (!hs)
-		return NULL;
-	while (--l && (hs = ext4_mb_history_skip_empty(s, ++hs, 0)) != NULL);
-	return hs;
-}
-
-static void *ext4_mb_seq_history_next(struct seq_file *seq, void *v,
-				      loff_t *pos)
-{
-	struct ext4_mb_proc_session *s = seq->private;
-	struct ext4_mb_history *hs = v;
-
-	++*pos;
-	if (v == SEQ_START_TOKEN)
-		return ext4_mb_history_skip_empty(s, s->history + s->start, 1);
-	else
-		return ext4_mb_history_skip_empty(s, ++hs, 0);
-}
-
-static int ext4_mb_seq_history_show(struct seq_file *seq, void *v)
-{
-	char buf[25], buf2[25], buf3[25], *fmt;
-	struct ext4_mb_history *hs = v;
-
-	if (v == SEQ_START_TOKEN) {
-		seq_printf(seq, "%-5s %-8s %-23s %-23s %-23s %-5s "
-				"%-5s %-2s %-5s %-5s %-5s %-6s\n",
-			  "pid", "inode", "original", "goal", "result", "found",
-			   "grps", "cr", "flags", "merge", "tail", "broken");
-		return 0;
-	}
-
-	if (hs->op == EXT4_MB_HISTORY_ALLOC) {
-		fmt = "%-5u %-8u %-23s %-23s %-23s %-5u %-5u %-2u "
-			"%-5u %-5s %-5u %-6u\n";
-		sprintf(buf2, "%u/%d/%u@%u", hs->result.fe_group,
-			hs->result.fe_start, hs->result.fe_len,
-			hs->result.fe_logical);
-		sprintf(buf, "%u/%d/%u@%u", hs->orig.fe_group,
-			hs->orig.fe_start, hs->orig.fe_len,
-			hs->orig.fe_logical);
-		sprintf(buf3, "%u/%d/%u@%u", hs->goal.fe_group,
-			hs->goal.fe_start, hs->goal.fe_len,
-			hs->goal.fe_logical);
-		seq_printf(seq, fmt, hs->pid, hs->ino, buf, buf3, buf2,
-				hs->found, hs->groups, hs->cr, hs->flags,
-				hs->merged ? "M" : "", hs->tail,
-				hs->buddy ? 1 << hs->buddy : 0);
-	} else if (hs->op == EXT4_MB_HISTORY_PREALLOC) {
-		fmt = "%-5u %-8u %-23s %-23s %-23s\n";
-		sprintf(buf2, "%u/%d/%u@%u", hs->result.fe_group,
-			hs->result.fe_start, hs->result.fe_len,
-			hs->result.fe_logical);
-		sprintf(buf, "%u/%d/%u@%u", hs->orig.fe_group,
-			hs->orig.fe_start, hs->orig.fe_len,
-			hs->orig.fe_logical);
-		seq_printf(seq, fmt, hs->pid, hs->ino, buf, "", buf2);
-	} else if (hs->op == EXT4_MB_HISTORY_DISCARD) {
-		sprintf(buf2, "%u/%d/%u", hs->result.fe_group,
-			hs->result.fe_start, hs->result.fe_len);
-		seq_printf(seq, "%-5u %-8u %-23s discard\n",
-				hs->pid, hs->ino, buf2);
-	} else if (hs->op == EXT4_MB_HISTORY_FREE) {
-		sprintf(buf2, "%u/%d/%u", hs->result.fe_group,
-			hs->result.fe_start, hs->result.fe_len);
-		seq_printf(seq, "%-5u %-8u %-23s free\n",
-				hs->pid, hs->ino, buf2);
-	}
-	return 0;
-}
-
-static void ext4_mb_seq_history_stop(struct seq_file *seq, void *v)
-{
-}
-
-static struct seq_operations ext4_mb_seq_history_ops = {
-	.start  = ext4_mb_seq_history_start,
-	.next   = ext4_mb_seq_history_next,
-	.stop   = ext4_mb_seq_history_stop,
-	.show   = ext4_mb_seq_history_show,
-};
-
-static int ext4_mb_seq_history_open(struct inode *inode, struct file *file)
-{
-	struct super_block *sb = PDE(inode)->data;
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-	struct ext4_mb_proc_session *s;
-	int rc;
-	int size;
-
-	if (unlikely(sbi->s_mb_history == NULL))
-		return -ENOMEM;
-	s = kmalloc(sizeof(*s), GFP_KERNEL);
-	if (s == NULL)
-		return -ENOMEM;
-	s->sb = sb;
-	size = sizeof(struct ext4_mb_history) * sbi->s_mb_history_max;
-	s->history = kmalloc(size, GFP_KERNEL);
-	if (s->history == NULL) {
-		kfree(s);
-		return -ENOMEM;
-	}
-
-	spin_lock(&sbi->s_mb_history_lock);
-	memcpy(s->history, sbi->s_mb_history, size);
-	s->max = sbi->s_mb_history_max;
-	s->start = sbi->s_mb_history_cur % s->max;
-	spin_unlock(&sbi->s_mb_history_lock);
-
-	rc = seq_open(file, &ext4_mb_seq_history_ops);
-	if (rc == 0) {
-		struct seq_file *m = (struct seq_file *)file->private_data;
-		m->private = s;
-	} else {
-		kfree(s->history);
-		kfree(s);
-	}
-	return rc;
-
-}
-
-static int ext4_mb_seq_history_release(struct inode *inode, struct file *file)
-{
-	struct seq_file *seq = (struct seq_file *)file->private_data;
-	struct ext4_mb_proc_session *s = seq->private;
-	kfree(s->history);
-	kfree(s);
-	return seq_release(inode, file);
-}
-
-static ssize_t ext4_mb_seq_history_write(struct file *file,
-				const char __user *buffer,
-				size_t count, loff_t *ppos)
-{
-	struct seq_file *seq = (struct seq_file *)file->private_data;
-	struct ext4_mb_proc_session *s = seq->private;
-	struct super_block *sb = s->sb;
-	char str[32];
-	int value;
-
-	if (count >= sizeof(str)) {
-		printk(KERN_ERR "EXT4-fs: %s string too long, max %u bytes\n",
-				"mb_history", (int)sizeof(str));
-		return -EOVERFLOW;
-	}
-
-	if (copy_from_user(str, buffer, count))
-		return -EFAULT;
-
-	value = simple_strtol(str, NULL, 0);
-	if (value < 0)
-		return -ERANGE;
-	EXT4_SB(sb)->s_mb_history_filter = value;
-
-	return count;
-}
-
-static struct file_operations ext4_mb_seq_history_fops = {
-	.owner		= THIS_MODULE,
-	.open		= ext4_mb_seq_history_open,
-	.read		= seq_read,
-	.write		= ext4_mb_seq_history_write,
-	.llseek		= seq_lseek,
-	.release	= ext4_mb_seq_history_release,
-};
-
 static void *ext4_mb_seq_groups_start(struct seq_file *seq, loff_t *pos)
 {
 	struct super_block *sb = seq->private;
@@ -2328,7 +2128,7 @@ static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v)
 	struct ext4_buddy e4b;
 	struct sg {
 		struct ext4_group_info info;
-		unsigned short counters[16];
+		ext4_grpblk_t counters[16];
 	} sg;
 
 	group--;
@@ -2366,7 +2166,7 @@ static void ext4_mb_seq_groups_stop(struct seq_file *seq, void *v)
 {
 }
 
-static struct seq_operations ext4_mb_seq_groups_ops = {
+static const struct seq_operations ext4_mb_seq_groups_ops = {
 	.start  = ext4_mb_seq_groups_start,
 	.next   = ext4_mb_seq_groups_next,
 	.stop   = ext4_mb_seq_groups_stop,
@@ -2387,7 +2187,7 @@ static int ext4_mb_seq_groups_open(struct inode *inode, struct file *file)
 
 }
 
-static struct file_operations ext4_mb_seq_groups_fops = {
+static const struct file_operations ext4_mb_seq_groups_fops = {
 	.owner		= THIS_MODULE,
 	.open		= ext4_mb_seq_groups_open,
 	.read		= seq_read,
@@ -2395,82 +2195,6 @@ static struct file_operations ext4_mb_seq_groups_fops = {
 	.release	= seq_release,
 };
 
-static void ext4_mb_history_release(struct super_block *sb)
-{
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-
-	if (sbi->s_proc != NULL) {
-		remove_proc_entry("mb_groups", sbi->s_proc);
-		if (sbi->s_mb_history_max)
-			remove_proc_entry("mb_history", sbi->s_proc);
-	}
-	kfree(sbi->s_mb_history);
-}
-
-static void ext4_mb_history_init(struct super_block *sb)
-{
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-	int i;
-
-	if (sbi->s_proc != NULL) {
-		if (sbi->s_mb_history_max)
-			proc_create_data("mb_history", S_IRUGO, sbi->s_proc,
-					 &ext4_mb_seq_history_fops, sb);
-		proc_create_data("mb_groups", S_IRUGO, sbi->s_proc,
-				 &ext4_mb_seq_groups_fops, sb);
-	}
-
-	sbi->s_mb_history_cur = 0;
-	spin_lock_init(&sbi->s_mb_history_lock);
-	i = sbi->s_mb_history_max * sizeof(struct ext4_mb_history);
-	sbi->s_mb_history = i ? kzalloc(i, GFP_KERNEL) : NULL;
-	/* if we can't allocate history, then we simple won't use it */
-}
-
-static noinline_for_stack void
-ext4_mb_store_history(struct ext4_allocation_context *ac)
-{
-	struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
-	struct ext4_mb_history h;
-
-	if (sbi->s_mb_history == NULL)
-		return;
-
-	if (!(ac->ac_op & sbi->s_mb_history_filter))
-		return;
-
-	h.op = ac->ac_op;
-	h.pid = current->pid;
-	h.ino = ac->ac_inode ? ac->ac_inode->i_ino : 0;
-	h.orig = ac->ac_o_ex;
-	h.result = ac->ac_b_ex;
-	h.flags = ac->ac_flags;
-	h.found = ac->ac_found;
-	h.groups = ac->ac_groups_scanned;
-	h.cr = ac->ac_criteria;
-	h.tail = ac->ac_tail;
-	h.buddy = ac->ac_buddy;
-	h.merged = 0;
-	if (ac->ac_op == EXT4_MB_HISTORY_ALLOC) {
-		if (ac->ac_g_ex.fe_start == ac->ac_b_ex.fe_start &&
-				ac->ac_g_ex.fe_group == ac->ac_b_ex.fe_group)
-			h.merged = 1;
-		h.goal = ac->ac_g_ex;
-		h.result = ac->ac_f_ex;
-	}
-
-	spin_lock(&sbi->s_mb_history_lock);
-	memcpy(sbi->s_mb_history + sbi->s_mb_history_cur, &h, sizeof(h));
-	if (++sbi->s_mb_history_cur >= sbi->s_mb_history_max)
-		sbi->s_mb_history_cur = 0;
-	spin_unlock(&sbi->s_mb_history_lock);
-}
-
-#else
-#define ext4_mb_history_release(sb)
-#define ext4_mb_history_init(sb)
-#endif
-
 
 /* Create and initialize ext4_group_info data for the given group. */
 int ext4_mb_add_groupinfo(struct super_block *sb, ext4_group_t group,
@@ -2532,7 +2256,7 @@ int ext4_mb_add_groupinfo(struct super_block *sb, ext4_group_t group,
 
 	INIT_LIST_HEAD(&meta_group_info[i]->bb_prealloc_list);
 	init_rwsem(&meta_group_info[i]->alloc_sem);
-	meta_group_info[i]->bb_free_root.rb_node = NULL;;
+	meta_group_info[i]->bb_free_root.rb_node = NULL;
 
 #ifdef DOUBLE_CHECK
 	{
@@ -2558,26 +2282,15 @@ exit_meta_group_info:
 	return -ENOMEM;
 } /* ext4_mb_add_groupinfo */
 
-/*
- * Update an existing group.
- * This function is used for online resize
- */
-void ext4_mb_update_group_info(struct ext4_group_info *grp, ext4_grpblk_t add)
-{
-	grp->bb_free += add;
-}
-
 static int ext4_mb_init_backend(struct super_block *sb)
 {
 	ext4_group_t ngroups = ext4_get_groups_count(sb);
 	ext4_group_t i;
-	int metalen;
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct ext4_super_block *es = sbi->s_es;
 	int num_meta_group_infos;
 	int num_meta_group_infos_max;
 	int array_size;
-	struct ext4_group_info **meta_group_info;
 	struct ext4_group_desc *desc;
 
 	/* This is the number of blocks used by GDT */
@@ -2622,22 +2335,6 @@ static int ext4_mb_init_backend(struct super_block *sb)
 		goto err_freesgi;
 	}
 	EXT4_I(sbi->s_buddy_cache)->i_disksize = 0;
-
-	metalen = sizeof(*meta_group_info) << EXT4_DESC_PER_BLOCK_BITS(sb);
-	for (i = 0; i < num_meta_group_infos; i++) {
-		if ((i + 1) == num_meta_group_infos)
-			metalen = sizeof(*meta_group_info) *
-				(ngroups -
-					(i << EXT4_DESC_PER_BLOCK_BITS(sb)));
-		meta_group_info = kmalloc(metalen, GFP_KERNEL);
-		if (meta_group_info == NULL) {
-			printk(KERN_ERR "EXT4-fs: can't allocate mem for a "
-			       "buddy group\n");
-			goto err_freemeta;
-		}
-		sbi->s_group_info[i] = meta_group_info;
-	}
-
 	for (i = 0; i < ngroups; i++) {
 		desc = ext4_get_group_desc(sb, i, NULL);
 		if (desc == NULL) {
@@ -2655,7 +2352,6 @@ err_freebuddy:
 	while (i-- > 0)
 		kfree(ext4_get_group_info(sb, i));
 	i = num_meta_group_infos;
-err_freemeta:
 	while (i-- > 0)
 		kfree(sbi->s_group_info[i]);
 	iput(sbi->s_buddy_cache);
@@ -2672,14 +2368,14 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 	unsigned max;
 	int ret;
 
-	i = (sb->s_blocksize_bits + 2) * sizeof(unsigned short);
+	i = (sb->s_blocksize_bits + 2) * sizeof(*sbi->s_mb_offsets);
 
 	sbi->s_mb_offsets = kmalloc(i, GFP_KERNEL);
 	if (sbi->s_mb_offsets == NULL) {
 		return -ENOMEM;
 	}
 
-	i = (sb->s_blocksize_bits + 2) * sizeof(unsigned int);
+	i = (sb->s_blocksize_bits + 2) * sizeof(*sbi->s_mb_maxs);
 	sbi->s_mb_maxs = kmalloc(i, GFP_KERNEL);
 	if (sbi->s_mb_maxs == NULL) {
 		kfree(sbi->s_mb_offsets);
@@ -2717,7 +2413,6 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 	sbi->s_mb_stats = MB_DEFAULT_STATS;
 	sbi->s_mb_stream_request = MB_DEFAULT_STREAM_THRESHOLD;
 	sbi->s_mb_order2_reqs = MB_DEFAULT_ORDER2_REQS;
-	sbi->s_mb_history_filter = EXT4_MB_HISTORY_DEFAULT;
 	sbi->s_mb_group_prealloc = MB_DEFAULT_GROUP_PREALLOC;
 
 	sbi->s_locality_groups = alloc_percpu(struct ext4_locality_group);
@@ -2735,12 +2430,12 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 		spin_lock_init(&lg->lg_prealloc_lock);
 	}
 
-	ext4_mb_history_init(sb);
+	if (sbi->s_proc)
+		proc_create_data("mb_groups", S_IRUGO, sbi->s_proc,
+				 &ext4_mb_seq_groups_fops, sb);
 
 	if (sbi->s_journal)
 		sbi->s_journal->j_commit_callback = release_blocks_on_commit;
-
-	printk(KERN_INFO "EXT4-fs: mballoc enabled\n");
 	return 0;
 }
 
@@ -2758,7 +2453,7 @@ static void ext4_mb_cleanup_pa(struct ext4_group_info *grp)
 		kmem_cache_free(ext4_pspace_cachep, pa);
 	}
 	if (count)
-		mb_debug("mballoc: %u PAs left\n", count);
+		mb_debug(1, "mballoc: %u PAs left\n", count);
 
 }
 
@@ -2817,7 +2512,8 @@ int ext4_mb_release(struct super_block *sb)
 	}
 
 	free_percpu(sbi->s_locality_groups);
-	ext4_mb_history_release(sb);
+	if (sbi->s_proc)
+		remove_proc_entry("mb_groups", sbi->s_proc);
 
 	return 0;
 }
@@ -2839,7 +2535,7 @@ static void release_blocks_on_commit(journal_t *journal, transaction_t *txn)
 	list_for_each_safe(l, ltmp, &txn->t_private_list) {
 		entry = list_entry(l, struct ext4_free_data, list);
 
-		mb_debug("gonna free %u blocks in group %u (0x%p):",
+		mb_debug(1, "gonna free %u blocks in group %u (0x%p):",
 			 entry->count, entry->group, entry);
 
 		err = ext4_mb_load_buddy(sb, entry->group, &e4b);
@@ -2874,9 +2570,43 @@ static void release_blocks_on_commit(journal_t *journal, transaction_t *txn)
 		ext4_mb_release_desc(&e4b);
 	}
 
-	mb_debug("freed %u blocks in %u structures\n", count, count2);
+	mb_debug(1, "freed %u blocks in %u structures\n", count, count2);
+}
+
+#ifdef CONFIG_EXT4_DEBUG
+u8 mb_enable_debug __read_mostly;
+
+static struct dentry *debugfs_dir;
+static struct dentry *debugfs_debug;
+
+static void __init ext4_create_debugfs_entry(void)
+{
+	debugfs_dir = debugfs_create_dir("ext4", NULL);
+	if (debugfs_dir)
+		debugfs_debug = debugfs_create_u8("mballoc-debug",
+						  S_IRUGO | S_IWUSR,
+						  debugfs_dir,
+						  &mb_enable_debug);
+}
+
+static void ext4_remove_debugfs_entry(void)
+{
+	debugfs_remove(debugfs_debug);
+	debugfs_remove(debugfs_dir);
 }
 
+#else
+
+static void __init ext4_create_debugfs_entry(void)
+{
+}
+
+static void ext4_remove_debugfs_entry(void)
+{
+}
+
+#endif
+
 int __init init_ext4_mballoc(void)
 {
 	ext4_pspace_cachep =
@@ -2904,6 +2634,7 @@ int __init init_ext4_mballoc(void)
 		kmem_cache_destroy(ext4_ac_cachep);
 		return -ENOMEM;
 	}
+	ext4_create_debugfs_entry();
 	return 0;
 }
 
@@ -2917,6 +2648,7 @@ void exit_ext4_mballoc(void)
 	kmem_cache_destroy(ext4_pspace_cachep);
 	kmem_cache_destroy(ext4_ac_cachep);
 	kmem_cache_destroy(ext4_free_ext_cachep);
+	ext4_remove_debugfs_entry();
 }
 
 
@@ -3061,7 +2793,7 @@ static void ext4_mb_normalize_group_request(struct ext4_allocation_context *ac)
 		ac->ac_g_ex.fe_len = EXT4_SB(sb)->s_stripe;
 	else
 		ac->ac_g_ex.fe_len = EXT4_SB(sb)->s_mb_group_prealloc;
-	mb_debug("#%u: goal %u blocks for locality group\n",
+	mb_debug(1, "#%u: goal %u blocks for locality group\n",
 		current->pid, ac->ac_g_ex.fe_len);
 }
 
@@ -3180,23 +2912,18 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 		BUG_ON(!(ac->ac_o_ex.fe_logical >= pa_end ||
 			ac->ac_o_ex.fe_logical < pa->pa_lstart));
 
-		/* skip PA normalized request doesn't overlap with */
-		if (pa->pa_lstart >= end) {
-			spin_unlock(&pa->pa_lock);
-			continue;
-		}
-		if (pa_end <= start) {
+		/* skip PAs this normalized request doesn't overlap with */
+		if (pa->pa_lstart >= end || pa_end <= start) {
 			spin_unlock(&pa->pa_lock);
 			continue;
 		}
 		BUG_ON(pa->pa_lstart <= start && pa_end >= end);
 
+		/* adjust start or end to be adjacent to this pa */
 		if (pa_end <= ac->ac_o_ex.fe_logical) {
 			BUG_ON(pa_end < start);
 			start = pa_end;
-		}
-
-		if (pa->pa_lstart > ac->ac_o_ex.fe_logical) {
+		} else if (pa->pa_lstart > ac->ac_o_ex.fe_logical) {
 			BUG_ON(pa->pa_lstart > end);
 			end = pa->pa_lstart;
 		}
@@ -3251,7 +2978,7 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 		ac->ac_flags |= EXT4_MB_HINT_TRY_GOAL;
 	}
 
-	mb_debug("goal: %u(was %u) blocks at %u\n", (unsigned) size,
+	mb_debug(1, "goal: %u(was %u) blocks at %u\n", (unsigned) size,
 		(unsigned) orig_size, (unsigned) start);
 }
 
@@ -3272,7 +2999,10 @@ static void ext4_mb_collect_stats(struct ext4_allocation_context *ac)
 			atomic_inc(&sbi->s_bal_breaks);
 	}
 
-	ext4_mb_store_history(ac);
+	if (ac->ac_op == EXT4_MB_HISTORY_ALLOC)
+		trace_ext4_mballoc_alloc(ac);
+	else
+		trace_ext4_mballoc_prealloc(ac);
 }
 
 /*
@@ -3300,7 +3030,7 @@ static void ext4_mb_use_inode_pa(struct ext4_allocation_context *ac,
 	BUG_ON(pa->pa_free < len);
 	pa->pa_free -= len;
 
-	mb_debug("use %llu/%u from inode pa %p\n", start, len, pa);
+	mb_debug(1, "use %llu/%u from inode pa %p\n", start, len, pa);
 }
 
 /*
@@ -3324,7 +3054,7 @@ static void ext4_mb_use_group_pa(struct ext4_allocation_context *ac,
 	 * in on-disk bitmap -- see ext4_mb_release_context()
 	 * Other CPUs are prevented from allocating from this pa by lg_mutex
 	 */
-	mb_debug("use %u/%u from group pa %p\n", pa->pa_lstart-len, len, pa);
+	mb_debug(1, "use %u/%u from group pa %p\n", pa->pa_lstart-len, len, pa);
 }
 
 /*
@@ -3382,6 +3112,11 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
 			ac->ac_o_ex.fe_logical >= pa->pa_lstart + pa->pa_len)
 			continue;
 
+		/* non-extent files can't have physical blocks past 2^32 */
+		if (!(EXT4_I(ac->ac_inode)->i_flags & EXT4_EXTENTS_FL) &&
+			pa->pa_pstart + pa->pa_len > EXT4_MAX_BLOCK_FILE_PHYS)
+			continue;
+
 		/* found preallocated blocks, use them */
 		spin_lock(&pa->pa_lock);
 		if (pa->pa_deleted == 0 && pa->pa_free) {
@@ -3503,7 +3238,7 @@ void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap,
 		preallocated += len;
 		count++;
 	}
-	mb_debug("prellocated %u for group %u\n", preallocated, group);
+	mb_debug(1, "prellocated %u for group %u\n", preallocated, group);
 }
 
 static void ext4_mb_pa_callback(struct rcu_head *head)
@@ -3638,7 +3373,7 @@ ext4_mb_new_inode_pa(struct ext4_allocation_context *ac)
 	pa->pa_deleted = 0;
 	pa->pa_type = MB_INODE_PA;
 
-	mb_debug("new inode pa %p: %llu/%u for %u\n", pa,
+	mb_debug(1, "new inode pa %p: %llu/%u for %u\n", pa,
 			pa->pa_pstart, pa->pa_len, pa->pa_lstart);
 	trace_ext4_mb_new_inode_pa(ac, pa);
 
@@ -3698,7 +3433,7 @@ ext4_mb_new_group_pa(struct ext4_allocation_context *ac)
 	pa->pa_deleted = 0;
 	pa->pa_type = MB_GROUP_PA;
 
-	mb_debug("new group pa %p: %llu/%u for %u\n", pa,
+	mb_debug(1, "new group pa %p: %llu/%u for %u\n", pa,
 			pa->pa_pstart, pa->pa_len, pa->pa_lstart);
 	trace_ext4_mb_new_group_pa(ac, pa);
 
@@ -3767,7 +3502,6 @@ ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh,
 	if (ac) {
 		ac->ac_sb = sb;
 		ac->ac_inode = pa->pa_inode;
-		ac->ac_op = EXT4_MB_HISTORY_DISCARD;
 	}
 
 	while (bit < end) {
@@ -3777,7 +3511,7 @@ ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh,
 		next = mb_find_next_bit(bitmap_bh->b_data, end, bit);
 		start = group * EXT4_BLOCKS_PER_GROUP(sb) + bit +
 				le32_to_cpu(sbi->s_es->s_first_data_block);
-		mb_debug("    free preallocated %u/%u in group %u\n",
+		mb_debug(1, "    free preallocated %u/%u in group %u\n",
 				(unsigned) start, (unsigned) next - bit,
 				(unsigned) group);
 		free += next - bit;
@@ -3787,7 +3521,7 @@ ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh,
 			ac->ac_b_ex.fe_start = bit;
 			ac->ac_b_ex.fe_len = next - bit;
 			ac->ac_b_ex.fe_logical = 0;
-			ext4_mb_store_history(ac);
+			trace_ext4_mballoc_discard(ac);
 		}
 
 		trace_ext4_mb_release_inode_pa(ac, pa, grp_blk_start + bit,
@@ -3822,9 +3556,6 @@ ext4_mb_release_group_pa(struct ext4_buddy *e4b,
 	ext4_group_t group;
 	ext4_grpblk_t bit;
 
-	if (ac)
-		ac->ac_op = EXT4_MB_HISTORY_DISCARD;
-
 	trace_ext4_mb_release_group_pa(ac, pa);
 	BUG_ON(pa->pa_deleted == 0);
 	ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
@@ -3839,7 +3570,7 @@ ext4_mb_release_group_pa(struct ext4_buddy *e4b,
 		ac->ac_b_ex.fe_start = bit;
 		ac->ac_b_ex.fe_len = pa->pa_len;
 		ac->ac_b_ex.fe_logical = 0;
-		ext4_mb_store_history(ac);
+		trace_ext4_mballoc_discard(ac);
 	}
 
 	return 0;
@@ -3868,7 +3599,7 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
 	int busy = 0;
 	int free = 0;
 
-	mb_debug("discard preallocation for group %u\n", group);
+	mb_debug(1, "discard preallocation for group %u\n", group);
 
 	if (list_empty(&grp->bb_prealloc_list))
 		return 0;
@@ -3992,7 +3723,7 @@ void ext4_discard_preallocations(struct inode *inode)
 		return;
 	}
 
-	mb_debug("discard preallocation for inode %lu\n", inode->i_ino);
+	mb_debug(1, "discard preallocation for inode %lu\n", inode->i_ino);
 	trace_ext4_discard_preallocations(inode);
 
 	INIT_LIST_HEAD(&list);
@@ -4097,7 +3828,7 @@ static void ext4_mb_return_to_preallocation(struct inode *inode,
 {
 	BUG_ON(!list_empty(&EXT4_I(inode)->i_prealloc_list));
 }
-#ifdef MB_DEBUG
+#ifdef CONFIG_EXT4_DEBUG
 static void ext4_mb_show_ac(struct ext4_allocation_context *ac)
 {
 	struct super_block *sb = ac->ac_sb;
@@ -4139,14 +3870,14 @@ static void ext4_mb_show_ac(struct ext4_allocation_context *ac)
 			ext4_get_group_no_and_offset(sb, pa->pa_pstart,
 						     NULL, &start);
 			spin_unlock(&pa->pa_lock);
-			printk(KERN_ERR "PA:%lu:%d:%u \n", i,
-							start, pa->pa_len);
+			printk(KERN_ERR "PA:%u:%d:%u \n", i,
+			       start, pa->pa_len);
 		}
 		ext4_unlock_group(sb, i);
 
 		if (grp->bb_free == 0)
 			continue;
-		printk(KERN_ERR "%lu: %d/%d \n",
+		printk(KERN_ERR "%u: %d/%d \n",
 		       i, grp->bb_free, grp->bb_fragments);
 	}
 	printk(KERN_ERR "\n");
@@ -4174,16 +3905,26 @@ static void ext4_mb_group_or_file(struct ext4_allocation_context *ac)
 	if (!(ac->ac_flags & EXT4_MB_HINT_DATA))
 		return;
 
+	if (unlikely(ac->ac_flags & EXT4_MB_HINT_GOAL_ONLY))
+		return;
+
 	size = ac->ac_o_ex.fe_logical + ac->ac_o_ex.fe_len;
-	isize = i_size_read(ac->ac_inode) >> bsbits;
-	size = max(size, isize);
+	isize = (i_size_read(ac->ac_inode) + ac->ac_sb->s_blocksize - 1)
+		>> bsbits;
 
-	/* don't use group allocation for large files */
-	if (size >= sbi->s_mb_stream_request)
+	if ((size == isize) &&
+	    !ext4_fs_is_busy(sbi) &&
+	    (atomic_read(&ac->ac_inode->i_writecount) == 0)) {
+		ac->ac_flags |= EXT4_MB_HINT_NOPREALLOC;
 		return;
+	}
 
-	if (unlikely(ac->ac_flags & EXT4_MB_HINT_GOAL_ONLY))
+	/* don't use group allocation for large files */
+	size = max(size, isize);
+	if (size >= sbi->s_mb_stream_request) {
+		ac->ac_flags |= EXT4_MB_STREAM_ALLOC;
 		return;
+	}
 
 	BUG_ON(ac->ac_lg != NULL);
 	/*
@@ -4246,7 +3987,7 @@ ext4_mb_initialize_context(struct ext4_allocation_context *ac,
 	 * locality group. this is a policy, actually */
 	ext4_mb_group_or_file(ac);
 
-	mb_debug("init ac: %u blocks @ %u, goal %u, flags %x, 2^%d, "
+	mb_debug(1, "init ac: %u blocks @ %u, goal %u, flags %x, 2^%d, "
 			"left: %u/%u, right %u/%u to %swritable\n",
 			(unsigned) ar->len, (unsigned) ar->logical,
 			(unsigned) ar->goal, ac->ac_flags, ac->ac_2order,
@@ -4268,7 +4009,7 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb,
 	struct ext4_prealloc_space *pa, *tmp;
 	struct ext4_allocation_context *ac;
 
-	mb_debug("discard locality group preallocation\n");
+	mb_debug(1, "discard locality group preallocation\n");
 
 	INIT_LIST_HEAD(&discard_list);
 	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
@@ -4720,7 +4461,6 @@ void ext4_mb_free_blocks(handle_t *handle, struct inode *inode,
 
 	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
 	if (ac) {
-		ac->ac_op = EXT4_MB_HISTORY_FREE;
 		ac->ac_inode = inode;
 		ac->ac_sb = sb;
 	}
@@ -4787,7 +4527,7 @@ do_more:
 		ac->ac_b_ex.fe_group = block_group;
 		ac->ac_b_ex.fe_start = bit;
 		ac->ac_b_ex.fe_len = count;
-		ext4_mb_store_history(ac);
+		trace_ext4_mballoc_free(ac);
 	}
 
 	err = ext4_mb_load_buddy(sb, block_group, &e4b);
diff --git a/fs/ext4/mballoc.h b/fs/ext4/mballoc.h
index c96bb19f58f9..0ca811061bc7 100644
--- a/fs/ext4/mballoc.h
+++ b/fs/ext4/mballoc.h
@@ -37,25 +37,23 @@
 
 /*
  */
-#define MB_DEBUG__
-#ifdef MB_DEBUG
-#define mb_debug(fmt, a...)	printk(fmt, ##a)
+#ifdef CONFIG_EXT4_DEBUG
+extern u8 mb_enable_debug;
+
+#define mb_debug(n, fmt, a...)	                                        \
+	do {								\
+		if ((n) <= mb_enable_debug) {		        	\
+			printk(KERN_DEBUG "(%s, %d): %s: ",		\
+			       __FILE__, __LINE__, __func__);		\
+			printk(fmt, ## a);				\
+		}							\
+	} while (0)
 #else
-#define mb_debug(fmt, a...)
+#define mb_debug(n, fmt, a...)
 #endif
 
-/*
- * with EXT4_MB_HISTORY mballoc stores last N allocations in memory
- * and you can monitor it in /proc/fs/ext4/<dev>/mb_history
- */
-#define EXT4_MB_HISTORY
 #define EXT4_MB_HISTORY_ALLOC		1	/* allocation */
 #define EXT4_MB_HISTORY_PREALLOC	2	/* preallocated blocks used */
-#define EXT4_MB_HISTORY_DISCARD		4	/* preallocation discarded */
-#define EXT4_MB_HISTORY_FREE		8	/* free */
-
-#define EXT4_MB_HISTORY_DEFAULT		(EXT4_MB_HISTORY_ALLOC | \
-					 EXT4_MB_HISTORY_PREALLOC)
 
 /*
  * How long mballoc can look for a best extent (in found extents)
@@ -76,7 +74,7 @@
  * with 'ext4_mb_stats' allocator will collect stats that will be
  * shown at umount. The collecting costs though!
  */
-#define MB_DEFAULT_STATS		1
+#define MB_DEFAULT_STATS		0
 
 /*
  * files smaller than MB_DEFAULT_STREAM_THRESHOLD are served
@@ -128,8 +126,8 @@ struct ext4_prealloc_space {
 	unsigned		pa_deleted;
 	ext4_fsblk_t		pa_pstart;	/* phys. block */
 	ext4_lblk_t		pa_lstart;	/* log. block */
-	unsigned short		pa_len;		/* len of preallocated chunk */
-	unsigned short		pa_free;	/* how many blocks are free */
+	ext4_grpblk_t		pa_len;		/* len of preallocated chunk */
+	ext4_grpblk_t		pa_free;	/* how many blocks are free */
 	unsigned short		pa_type;	/* pa type. inode or group */
 	spinlock_t		*pa_obj_lock;
 	struct inode		*pa_inode;	/* hack, for history only */
@@ -144,7 +142,7 @@ struct ext4_free_extent {
 	ext4_lblk_t fe_logical;
 	ext4_grpblk_t fe_start;
 	ext4_group_t fe_group;
-	int fe_len;
+	ext4_grpblk_t fe_len;
 };
 
 /*
@@ -209,22 +207,6 @@ struct ext4_allocation_context {
 #define AC_STATUS_FOUND		2
 #define AC_STATUS_BREAK		3
 
-struct ext4_mb_history {
-	struct ext4_free_extent orig;	/* orig allocation */
-	struct ext4_free_extent goal;	/* goal allocation */
-	struct ext4_free_extent result;	/* result allocation */
-	unsigned pid;
-	unsigned ino;
-	__u16 found;	/* how many extents have been found */
-	__u16 groups;	/* how many groups have been scanned */
-	__u16 tail;	/* what tail broke some buddy */
-	__u16 buddy;	/* buddy the tail ^^^ broke */
-	__u16 flags;
-	__u8 cr:3;	/* which phase the result extent was found at */
-	__u8 op:4;
-	__u8 merged:1;
-};
-
 struct ext4_buddy {
 	struct page *bd_buddy_page;
 	void *bd_buddy;
@@ -239,13 +221,6 @@ struct ext4_buddy {
 #define EXT4_MB_BITMAP(e4b)	((e4b)->bd_bitmap)
 #define EXT4_MB_BUDDY(e4b)	((e4b)->bd_buddy)
 
-#ifndef EXT4_MB_HISTORY
-static inline void ext4_mb_store_history(struct ext4_allocation_context *ac)
-{
-	return;
-}
-#endif
-
 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
 
 static inline ext4_fsblk_t ext4_grp_offs_to_block(struct super_block *sb,
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
index 313a50b39741..a93d5b80f3e2 100644
--- a/fs/ext4/migrate.c
+++ b/fs/ext4/migrate.c
@@ -75,7 +75,7 @@ static int finish_range(handle_t *handle, struct inode *inode,
 				goto err_out;
 		}
 	}
-	retval = ext4_ext_insert_extent(handle, inode, path, &newext);
+	retval = ext4_ext_insert_extent(handle, inode, path, &newext, 0);
 err_out:
 	if (path) {
 		ext4_ext_drop_refs(path);
@@ -353,17 +353,16 @@ static int ext4_ext_swap_inode_data(handle_t *handle, struct inode *inode,
 
 	down_write(&EXT4_I(inode)->i_data_sem);
 	/*
-	 * if EXT4_EXT_MIGRATE is cleared a block allocation
+	 * if EXT4_STATE_EXT_MIGRATE is cleared a block allocation
 	 * happened after we started the migrate. We need to
 	 * fail the migrate
 	 */
-	if (!(EXT4_I(inode)->i_flags & EXT4_EXT_MIGRATE)) {
+	if (!(EXT4_I(inode)->i_state & EXT4_STATE_EXT_MIGRATE)) {
 		retval = -EAGAIN;
 		up_write(&EXT4_I(inode)->i_data_sem);
 		goto err_out;
 	} else
-		EXT4_I(inode)->i_flags = EXT4_I(inode)->i_flags &
-							~EXT4_EXT_MIGRATE;
+		EXT4_I(inode)->i_state &= ~EXT4_STATE_EXT_MIGRATE;
 	/*
 	 * We have the extent map build with the tmp inode.
 	 * Now copy the i_data across
@@ -517,14 +516,15 @@ int ext4_ext_migrate(struct inode *inode)
 	 * when we add extents we extent the journal
 	 */
 	/*
-	 * Even though we take i_mutex we can still cause block allocation
-	 * via mmap write to holes. If we have allocated new blocks we fail
-	 * migrate.  New block allocation will clear EXT4_EXT_MIGRATE flag.
-	 * The flag is updated with i_data_sem held to prevent racing with
-	 * block allocation.
+	 * Even though we take i_mutex we can still cause block
+	 * allocation via mmap write to holes. If we have allocated
+	 * new blocks we fail migrate.  New block allocation will
+	 * clear EXT4_STATE_EXT_MIGRATE flag.  The flag is updated
+	 * with i_data_sem held to prevent racing with block
+	 * allocation.
 	 */
 	down_read((&EXT4_I(inode)->i_data_sem));
-	EXT4_I(inode)->i_flags = EXT4_I(inode)->i_flags | EXT4_EXT_MIGRATE;
+	EXT4_I(inode)->i_state |= EXT4_STATE_EXT_MIGRATE;
 	up_read((&EXT4_I(inode)->i_data_sem));
 
 	handle = ext4_journal_start(inode, 1);
@@ -618,7 +618,7 @@ err_out:
 	tmp_inode->i_nlink = 0;
 
 	ext4_journal_stop(handle);
-
+	unlock_new_inode(tmp_inode);
 	iput(tmp_inode);
 
 	return retval;
diff --git a/fs/ext4/move_extent.c b/fs/ext4/move_extent.c
index bbf2dd9404dc..25b6b1457360 100644
--- a/fs/ext4/move_extent.c
+++ b/fs/ext4/move_extent.c
@@ -19,14 +19,31 @@
 #include "ext4_extents.h"
 #include "ext4.h"
 
-#define get_ext_path(path, inode, block, ret)		\
-	do {								\
-		path = ext4_ext_find_extent(inode, block, path);	\
-		if (IS_ERR(path)) {					\
-			ret = PTR_ERR(path);				\
-			path = NULL;					\
-		}							\
-	} while (0)
+/**
+ * get_ext_path - Find an extent path for designated logical block number.
+ *
+ * @inode:	an inode which is searched
+ * @lblock:	logical block number to find an extent path
+ * @path:	pointer to an extent path pointer (for output)
+ *
+ * ext4_ext_find_extent wrapper. Return 0 on success, or a negative error value
+ * on failure.
+ */
+static inline int
+get_ext_path(struct inode *inode, ext4_lblk_t lblock,
+		struct ext4_ext_path **path)
+{
+	int ret = 0;
+
+	*path = ext4_ext_find_extent(inode, lblock, *path);
+	if (IS_ERR(*path)) {
+		ret = PTR_ERR(*path);
+		*path = NULL;
+	} else if ((*path)[ext_depth(inode)].p_ext == NULL)
+		ret = -ENODATA;
+
+	return ret;
+}
 
 /**
  * copy_extent_status - Copy the extent's initialization status
@@ -113,6 +130,31 @@ mext_next_extent(struct inode *inode, struct ext4_ext_path *path,
 }
 
 /**
+ * mext_check_null_inode - NULL check for two inodes
+ *
+ * If inode1 or inode2 is NULL, return -EIO. Otherwise, return 0.
+ */
+static int
+mext_check_null_inode(struct inode *inode1, struct inode *inode2,
+		const char *function)
+{
+	int ret = 0;
+
+	if (inode1 == NULL) {
+		ext4_error(inode2->i_sb, function,
+			"Both inodes should not be NULL: "
+			"inode1 NULL inode2 %lu", inode2->i_ino);
+		ret = -EIO;
+	} else if (inode2 == NULL) {
+		ext4_error(inode1->i_sb, function,
+			"Both inodes should not be NULL: "
+			"inode1 %lu inode2 NULL", inode1->i_ino);
+		ret = -EIO;
+	}
+	return ret;
+}
+
+/**
  * mext_double_down_read - Acquire two inodes' read semaphore
  *
  * @orig_inode:		original inode structure
@@ -124,8 +166,6 @@ mext_double_down_read(struct inode *orig_inode, struct inode *donor_inode)
 {
 	struct inode *first = orig_inode, *second = donor_inode;
 
-	BUG_ON(orig_inode == NULL || donor_inode == NULL);
-
 	/*
 	 * Use the inode number to provide the stable locking order instead
 	 * of its address, because the C language doesn't guarantee you can
@@ -152,8 +192,6 @@ mext_double_down_write(struct inode *orig_inode, struct inode *donor_inode)
 {
 	struct inode *first = orig_inode, *second = donor_inode;
 
-	BUG_ON(orig_inode == NULL || donor_inode == NULL);
-
 	/*
 	 * Use the inode number to provide the stable locking order instead
 	 * of its address, because the C language doesn't guarantee you can
@@ -178,8 +216,6 @@ mext_double_down_write(struct inode *orig_inode, struct inode *donor_inode)
 static void
 mext_double_up_read(struct inode *orig_inode, struct inode *donor_inode)
 {
-	BUG_ON(orig_inode == NULL || donor_inode == NULL);
-
 	up_read(&EXT4_I(orig_inode)->i_data_sem);
 	up_read(&EXT4_I(donor_inode)->i_data_sem);
 }
@@ -194,8 +230,6 @@ mext_double_up_read(struct inode *orig_inode, struct inode *donor_inode)
 static void
 mext_double_up_write(struct inode *orig_inode, struct inode *donor_inode)
 {
-	BUG_ON(orig_inode == NULL || donor_inode == NULL);
-
 	up_write(&EXT4_I(orig_inode)->i_data_sem);
 	up_write(&EXT4_I(donor_inode)->i_data_sem);
 }
@@ -283,23 +317,23 @@ mext_insert_across_blocks(handle_t *handle, struct inode *orig_inode,
 	}
 
 	if (new_flag) {
-		get_ext_path(orig_path, orig_inode, eblock, err);
-		if (orig_path == NULL)
+		err = get_ext_path(orig_inode, eblock, &orig_path);
+		if (err)
 			goto out;
 
 		if (ext4_ext_insert_extent(handle, orig_inode,
-					orig_path, new_ext))
+					orig_path, new_ext, 0))
 			goto out;
 	}
 
 	if (end_flag) {
-		get_ext_path(orig_path, orig_inode,
-				      le32_to_cpu(end_ext->ee_block) - 1, err);
-		if (orig_path == NULL)
+		err = get_ext_path(orig_inode,
+				le32_to_cpu(end_ext->ee_block) - 1, &orig_path);
+		if (err)
 			goto out;
 
 		if (ext4_ext_insert_extent(handle, orig_inode,
-					   orig_path, end_ext))
+					   orig_path, end_ext, 0))
 			goto out;
 	}
 out:
@@ -519,7 +553,15 @@ mext_leaf_block(handle_t *handle, struct inode *orig_inode,
 	 * oext      |-----------|
 	 * new_ext       |-------|
 	 */
-	BUG_ON(le32_to_cpu(oext->ee_block) + oext_alen - 1 < new_ext_end);
+	if (le32_to_cpu(oext->ee_block) + oext_alen - 1 < new_ext_end) {
+		ext4_error(orig_inode->i_sb, __func__,
+			"new_ext_end(%u) should be less than or equal to "
+			"oext->ee_block(%u) + oext_alen(%d) - 1",
+			new_ext_end, le32_to_cpu(oext->ee_block),
+			oext_alen);
+		ret = -EIO;
+		goto out;
+	}
 
 	/*
 	 * Case: new_ext is smaller than original extent
@@ -543,6 +585,7 @@ mext_leaf_block(handle_t *handle, struct inode *orig_inode,
 
 	ret = mext_insert_extents(handle, orig_inode, orig_path, o_start,
 				o_end, &start_ext, &new_ext, &end_ext);
+out:
 	return ret;
 }
 
@@ -554,8 +597,10 @@ mext_leaf_block(handle_t *handle, struct inode *orig_inode,
  * @orig_off:		block offset of original inode
  * @donor_off:		block offset of donor inode
  * @max_count:		the maximun length of extents
+ *
+ * Return 0 on success, or a negative error value on failure.
  */
-static void
+static int
 mext_calc_swap_extents(struct ext4_extent *tmp_dext,
 			      struct ext4_extent *tmp_oext,
 			      ext4_lblk_t orig_off, ext4_lblk_t donor_off,
@@ -564,6 +609,19 @@ mext_calc_swap_extents(struct ext4_extent *tmp_dext,
 	ext4_lblk_t diff, orig_diff;
 	struct ext4_extent dext_old, oext_old;
 
+	BUG_ON(orig_off != donor_off);
+
+	/* original and donor extents have to cover the same block offset */
+	if (orig_off < le32_to_cpu(tmp_oext->ee_block) ||
+	    le32_to_cpu(tmp_oext->ee_block) +
+			ext4_ext_get_actual_len(tmp_oext) - 1 < orig_off)
+		return -ENODATA;
+
+	if (orig_off < le32_to_cpu(tmp_dext->ee_block) ||
+	    le32_to_cpu(tmp_dext->ee_block) +
+			ext4_ext_get_actual_len(tmp_dext) - 1 < orig_off)
+		return -ENODATA;
+
 	dext_old = *tmp_dext;
 	oext_old = *tmp_oext;
 
@@ -591,6 +649,8 @@ mext_calc_swap_extents(struct ext4_extent *tmp_dext,
 
 	copy_extent_status(&oext_old, tmp_dext);
 	copy_extent_status(&dext_old, tmp_oext);
+
+	return 0;
 }
 
 /**
@@ -631,13 +691,13 @@ mext_replace_branches(handle_t *handle, struct inode *orig_inode,
 	mext_double_down_write(orig_inode, donor_inode);
 
 	/* Get the original extent for the block "orig_off" */
-	get_ext_path(orig_path, orig_inode, orig_off, err);
-	if (orig_path == NULL)
+	err = get_ext_path(orig_inode, orig_off, &orig_path);
+	if (err)
 		goto out;
 
 	/* Get the donor extent for the head */
-	get_ext_path(donor_path, donor_inode, donor_off, err);
-	if (donor_path == NULL)
+	err = get_ext_path(donor_inode, donor_off, &donor_path);
+	if (err)
 		goto out;
 	depth = ext_depth(orig_inode);
 	oext = orig_path[depth].p_ext;
@@ -647,13 +707,28 @@ mext_replace_branches(handle_t *handle, struct inode *orig_inode,
 	dext = donor_path[depth].p_ext;
 	tmp_dext = *dext;
 
-	mext_calc_swap_extents(&tmp_dext, &tmp_oext, orig_off,
+	err = mext_calc_swap_extents(&tmp_dext, &tmp_oext, orig_off,
 				      donor_off, count);
+	if (err)
+		goto out;
 
 	/* Loop for the donor extents */
 	while (1) {
 		/* The extent for donor must be found. */
-		BUG_ON(!dext || donor_off != le32_to_cpu(tmp_dext.ee_block));
+		if (!dext) {
+			ext4_error(donor_inode->i_sb, __func__,
+				   "The extent for donor must be found");
+			err = -EIO;
+			goto out;
+		} else if (donor_off != le32_to_cpu(tmp_dext.ee_block)) {
+			ext4_error(donor_inode->i_sb, __func__,
+				"Donor offset(%u) and the first block of donor "
+				"extent(%u) should be equal",
+				donor_off,
+				le32_to_cpu(tmp_dext.ee_block));
+			err = -EIO;
+			goto out;
+		}
 
 		/* Set donor extent to orig extent */
 		err = mext_leaf_block(handle, orig_inode,
@@ -678,8 +753,8 @@ mext_replace_branches(handle_t *handle, struct inode *orig_inode,
 
 		if (orig_path)
 			ext4_ext_drop_refs(orig_path);
-		get_ext_path(orig_path, orig_inode, orig_off, err);
-		if (orig_path == NULL)
+		err = get_ext_path(orig_inode, orig_off, &orig_path);
+		if (err)
 			goto out;
 		depth = ext_depth(orig_inode);
 		oext = orig_path[depth].p_ext;
@@ -692,9 +767,8 @@ mext_replace_branches(handle_t *handle, struct inode *orig_inode,
 
 		if (donor_path)
 			ext4_ext_drop_refs(donor_path);
-		get_ext_path(donor_path, donor_inode,
-				      donor_off, err);
-		if (donor_path == NULL)
+		err = get_ext_path(donor_inode, donor_off, &donor_path);
+		if (err)
 			goto out;
 		depth = ext_depth(donor_inode);
 		dext = donor_path[depth].p_ext;
@@ -705,9 +779,10 @@ mext_replace_branches(handle_t *handle, struct inode *orig_inode,
 		}
 		tmp_dext = *dext;
 
-		mext_calc_swap_extents(&tmp_dext, &tmp_oext, orig_off,
-					      donor_off,
-					      count - replaced_count);
+		err = mext_calc_swap_extents(&tmp_dext, &tmp_oext, orig_off,
+					   donor_off, count - replaced_count);
+		if (err)
+			goto out;
 	}
 
 out:
@@ -740,7 +815,7 @@ out:
  * on success, or a negative error value on failure.
  */
 static int
-move_extent_par_page(struct file *o_filp, struct inode *donor_inode,
+move_extent_per_page(struct file *o_filp, struct inode *donor_inode,
 		  pgoff_t orig_page_offset, int data_offset_in_page,
 		  int block_len_in_page, int uninit)
 {
@@ -871,6 +946,7 @@ out:
 		if (PageLocked(page))
 			unlock_page(page);
 		page_cache_release(page);
+		ext4_journal_stop(handle);
 	}
 out2:
 	ext4_journal_stop(handle);
@@ -897,6 +973,10 @@ mext_check_arguments(struct inode *orig_inode,
 			  struct inode *donor_inode, __u64 orig_start,
 			  __u64 donor_start, __u64 *len, __u64 moved_len)
 {
+	ext4_lblk_t orig_blocks, donor_blocks;
+	unsigned int blkbits = orig_inode->i_blkbits;
+	unsigned int blocksize = 1 << blkbits;
+
 	/* Regular file check */
 	if (!S_ISREG(orig_inode->i_mode) || !S_ISREG(donor_inode->i_mode)) {
 		ext4_debug("ext4 move extent: The argument files should be "
@@ -921,14 +1001,6 @@ mext_check_arguments(struct inode *orig_inode,
 		return -EINVAL;
 	}
 
-	/* orig and donor should be different file */
-	if (orig_inode->i_ino == donor_inode->i_ino) {
-		ext4_debug("ext4 move extent: The argument files should not "
-			"be same file [ino:orig %lu, donor %lu]\n",
-			orig_inode->i_ino, donor_inode->i_ino);
-		return -EINVAL;
-	}
-
 	/* Ext4 move extent supports only extent based file */
 	if (!(EXT4_I(orig_inode)->i_flags & EXT4_EXTENTS_FL)) {
 		ext4_debug("ext4 move extent: orig file is not extents "
@@ -960,54 +1032,58 @@ mext_check_arguments(struct inode *orig_inode,
 		return -EINVAL;
 	}
 
-	if ((orig_start > MAX_DEFRAG_SIZE) ||
-	    (donor_start > MAX_DEFRAG_SIZE) ||
-	    (*len > MAX_DEFRAG_SIZE) ||
-	    (orig_start + *len > MAX_DEFRAG_SIZE))  {
-		ext4_debug("ext4 move extent: Can't handle over [%lu] blocks "
-			"[ino:orig %lu, donor %lu]\n", MAX_DEFRAG_SIZE,
+	if ((orig_start > EXT_MAX_BLOCK) ||
+	    (donor_start > EXT_MAX_BLOCK) ||
+	    (*len > EXT_MAX_BLOCK) ||
+	    (orig_start + *len > EXT_MAX_BLOCK))  {
+		ext4_debug("ext4 move extent: Can't handle over [%u] blocks "
+			"[ino:orig %lu, donor %lu]\n", EXT_MAX_BLOCK,
 			orig_inode->i_ino, donor_inode->i_ino);
 		return -EINVAL;
 	}
 
 	if (orig_inode->i_size > donor_inode->i_size) {
-		if (orig_start >= donor_inode->i_size) {
+		donor_blocks = (donor_inode->i_size + blocksize - 1) >> blkbits;
+		/* TODO: eliminate this artificial restriction */
+		if (orig_start >= donor_blocks) {
 			ext4_debug("ext4 move extent: orig start offset "
-			"[%llu] should be less than donor file size "
-			"[%lld] [ino:orig %lu, donor_inode %lu]\n",
-			orig_start, donor_inode->i_size,
+			"[%llu] should be less than donor file blocks "
+			"[%u] [ino:orig %lu, donor %lu]\n",
+			orig_start, donor_blocks,
 			orig_inode->i_ino, donor_inode->i_ino);
 			return -EINVAL;
 		}
 
-		if (orig_start + *len > donor_inode->i_size) {
+		/* TODO: eliminate this artificial restriction */
+		if (orig_start + *len > donor_blocks) {
 			ext4_debug("ext4 move extent: End offset [%llu] should "
-				"be less than donor file size [%lld]."
-				"So adjust length from %llu to %lld "
+				"be less than donor file blocks [%u]."
+				"So adjust length from %llu to %llu "
 				"[ino:orig %lu, donor %lu]\n",
-				orig_start + *len, donor_inode->i_size,
-				*len, donor_inode->i_size - orig_start,
+				orig_start + *len, donor_blocks,
+				*len, donor_blocks - orig_start,
 				orig_inode->i_ino, donor_inode->i_ino);
-			*len = donor_inode->i_size - orig_start;
+			*len = donor_blocks - orig_start;
 		}
 	} else {
-		if (orig_start >= orig_inode->i_size) {
+		orig_blocks = (orig_inode->i_size + blocksize - 1) >> blkbits;
+		if (orig_start >= orig_blocks) {
 			ext4_debug("ext4 move extent: start offset [%llu] "
-				"should be less than original file size "
-				"[%lld] [inode:orig %lu, donor %lu]\n",
-				 orig_start, orig_inode->i_size,
+				"should be less than original file blocks "
+				"[%u] [ino:orig %lu, donor %lu]\n",
+				 orig_start, orig_blocks,
 				orig_inode->i_ino, donor_inode->i_ino);
 			return -EINVAL;
 		}
 
-		if (orig_start + *len > orig_inode->i_size) {
+		if (orig_start + *len > orig_blocks) {
 			ext4_debug("ext4 move extent: Adjust length "
-				"from %llu to %lld. Because it should be "
-				"less than original file size "
+				"from %llu to %llu. Because it should be "
+				"less than original file blocks "
 				"[ino:orig %lu, donor %lu]\n",
-				*len, orig_inode->i_size - orig_start,
+				*len, orig_blocks - orig_start,
 				orig_inode->i_ino, donor_inode->i_ino);
-			*len = orig_inode->i_size - orig_start;
+			*len = orig_blocks - orig_start;
 		}
 	}
 
@@ -1027,18 +1103,23 @@ mext_check_arguments(struct inode *orig_inode,
  * @inode1:	the inode structure
  * @inode2:	the inode structure
  *
- * Lock two inodes' i_mutex by i_ino order. This function is moved from
- * fs/inode.c.
+ * Lock two inodes' i_mutex by i_ino order.
+ * If inode1 or inode2 is NULL, return -EIO. Otherwise, return 0.
  */
-static void
+static int
 mext_inode_double_lock(struct inode *inode1, struct inode *inode2)
 {
-	if (inode1 == NULL || inode2 == NULL || inode1 == inode2) {
-		if (inode1)
-			mutex_lock(&inode1->i_mutex);
-		else if (inode2)
-			mutex_lock(&inode2->i_mutex);
-		return;
+	int ret = 0;
+
+	BUG_ON(inode1 == NULL && inode2 == NULL);
+
+	ret = mext_check_null_inode(inode1, inode2, __func__);
+	if (ret < 0)
+		goto out;
+
+	if (inode1 == inode2) {
+		mutex_lock(&inode1->i_mutex);
+		goto out;
 	}
 
 	if (inode1->i_ino < inode2->i_ino) {
@@ -1048,6 +1129,9 @@ mext_inode_double_lock(struct inode *inode1, struct inode *inode2)
 		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_PARENT);
 		mutex_lock_nested(&inode1->i_mutex, I_MUTEX_CHILD);
 	}
+
+out:
+	return ret;
 }
 
 /**
@@ -1056,17 +1140,28 @@ mext_inode_double_lock(struct inode *inode1, struct inode *inode2)
  * @inode1:     the inode that is released first
  * @inode2:     the inode that is released second
  *
- * This function is moved from fs/inode.c.
+ * If inode1 or inode2 is NULL, return -EIO. Otherwise, return 0.
  */
 
-static void
+static int
 mext_inode_double_unlock(struct inode *inode1, struct inode *inode2)
 {
+	int ret = 0;
+
+	BUG_ON(inode1 == NULL && inode2 == NULL);
+
+	ret = mext_check_null_inode(inode1, inode2, __func__);
+	if (ret < 0)
+		goto out;
+
 	if (inode1)
 		mutex_unlock(&inode1->i_mutex);
 
 	if (inode2 && inode2 != inode1)
 		mutex_unlock(&inode2->i_mutex);
+
+out:
+	return ret;
 }
 
 /**
@@ -1123,70 +1218,84 @@ ext4_move_extents(struct file *o_filp, struct file *d_filp,
 	ext4_lblk_t block_end, seq_start, add_blocks, file_end, seq_blocks = 0;
 	ext4_lblk_t rest_blocks;
 	pgoff_t orig_page_offset = 0, seq_end_page;
-	int ret, depth, last_extent = 0;
+	int ret1, ret2, depth, last_extent = 0;
 	int blocks_per_page = PAGE_CACHE_SIZE >> orig_inode->i_blkbits;
 	int data_offset_in_page;
 	int block_len_in_page;
 	int uninit;
 
+	/* orig and donor should be different file */
+	if (orig_inode->i_ino == donor_inode->i_ino) {
+		ext4_debug("ext4 move extent: The argument files should not "
+			"be same file [ino:orig %lu, donor %lu]\n",
+			orig_inode->i_ino, donor_inode->i_ino);
+		return -EINVAL;
+	}
+
 	/* protect orig and donor against a truncate */
-	mext_inode_double_lock(orig_inode, donor_inode);
+	ret1 = mext_inode_double_lock(orig_inode, donor_inode);
+	if (ret1 < 0)
+		return ret1;
 
 	mext_double_down_read(orig_inode, donor_inode);
 	/* Check the filesystem environment whether move_extent can be done */
-	ret = mext_check_arguments(orig_inode, donor_inode, orig_start,
+	ret1 = mext_check_arguments(orig_inode, donor_inode, orig_start,
 					donor_start, &len, *moved_len);
 	mext_double_up_read(orig_inode, donor_inode);
-	if (ret)
-		goto out2;
+	if (ret1)
+		goto out;
 
 	file_end = (i_size_read(orig_inode) - 1) >> orig_inode->i_blkbits;
 	block_end = block_start + len - 1;
 	if (file_end < block_end)
 		len -= block_end - file_end;
 
-	get_ext_path(orig_path, orig_inode, block_start, ret);
-	if (orig_path == NULL)
-		goto out2;
+	ret1 = get_ext_path(orig_inode, block_start, &orig_path);
+	if (ret1)
+		goto out;
 
 	/* Get path structure to check the hole */
-	get_ext_path(holecheck_path, orig_inode, block_start, ret);
-	if (holecheck_path == NULL)
+	ret1 = get_ext_path(orig_inode, block_start, &holecheck_path);
+	if (ret1)
 		goto out;
 
 	depth = ext_depth(orig_inode);
 	ext_cur = holecheck_path[depth].p_ext;
-	if (ext_cur == NULL) {
-		ret = -EINVAL;
-		goto out;
-	}
 
 	/*
-	 * Get proper extent whose ee_block is beyond block_start
-	 * if block_start was within the hole.
+	 * Get proper starting location of block replacement if block_start was
+	 * within the hole.
 	 */
 	if (le32_to_cpu(ext_cur->ee_block) +
 		ext4_ext_get_actual_len(ext_cur) - 1 < block_start) {
+		/*
+		 * The hole exists between extents or the tail of
+		 * original file.
+		 */
 		last_extent = mext_next_extent(orig_inode,
 					holecheck_path, &ext_cur);
 		if (last_extent < 0) {
-			ret = last_extent;
+			ret1 = last_extent;
 			goto out;
 		}
 		last_extent = mext_next_extent(orig_inode, orig_path,
 							&ext_dummy);
 		if (last_extent < 0) {
-			ret = last_extent;
+			ret1 = last_extent;
 			goto out;
 		}
-	}
-	seq_start = block_start;
+		seq_start = le32_to_cpu(ext_cur->ee_block);
+	} else if (le32_to_cpu(ext_cur->ee_block) > block_start)
+		/* The hole exists at the beginning of original file. */
+		seq_start = le32_to_cpu(ext_cur->ee_block);
+	else
+		seq_start = block_start;
 
 	/* No blocks within the specified range. */
 	if (le32_to_cpu(ext_cur->ee_block) > block_end) {
 		ext4_debug("ext4 move extent: The specified range of file "
 							"may be the hole\n");
-		ret = -EINVAL;
+		ret1 = -EINVAL;
 		goto out;
 	}
 
@@ -1206,7 +1315,7 @@ ext4_move_extents(struct file *o_filp, struct file *d_filp,
 		last_extent = mext_next_extent(orig_inode, holecheck_path,
 						&ext_cur);
 		if (last_extent < 0) {
-			ret = last_extent;
+			ret1 = last_extent;
 			break;
 		}
 		add_blocks = ext4_ext_get_actual_len(ext_cur);
@@ -1258,16 +1367,23 @@ ext4_move_extents(struct file *o_filp, struct file *d_filp,
 		while (orig_page_offset <= seq_end_page) {
 
 			/* Swap original branches with new branches */
-			ret = move_extent_par_page(o_filp, donor_inode,
+			ret1 = move_extent_per_page(o_filp, donor_inode,
 						orig_page_offset,
 						data_offset_in_page,
 						block_len_in_page, uninit);
-			if (ret < 0)
+			if (ret1 < 0)
 				goto out;
 			orig_page_offset++;
 			/* Count how many blocks we have exchanged */
 			*moved_len += block_len_in_page;
-			BUG_ON(*moved_len > len);
+			if (*moved_len > len) {
+				ext4_error(orig_inode->i_sb, __func__,
+					"We replaced blocks too much! "
+					"sum of replaced: %llu requested: %llu",
+					*moved_len, len);
+				ret1 = -EIO;
+				goto out;
+			}
 
 			data_offset_in_page = 0;
 			rest_blocks -= block_len_in_page;
@@ -1280,17 +1396,16 @@ ext4_move_extents(struct file *o_filp, struct file *d_filp,
 		/* Decrease buffer counter */
 		if (holecheck_path)
 			ext4_ext_drop_refs(holecheck_path);
-		get_ext_path(holecheck_path, orig_inode,
-				      seq_start, ret);
-		if (holecheck_path == NULL)
+		ret1 = get_ext_path(orig_inode, seq_start, &holecheck_path);
+		if (ret1)
 			break;
 		depth = holecheck_path->p_depth;
 
 		/* Decrease buffer counter */
 		if (orig_path)
 			ext4_ext_drop_refs(orig_path);
-		get_ext_path(orig_path, orig_inode, seq_start, ret);
-		if (orig_path == NULL)
+		ret1 = get_ext_path(orig_inode, seq_start, &orig_path);
+		if (ret1)
 			break;
 
 		ext_cur = holecheck_path[depth].p_ext;
@@ -1307,14 +1422,13 @@ out:
 		ext4_ext_drop_refs(holecheck_path);
 		kfree(holecheck_path);
 	}
-out2:
-	mext_inode_double_unlock(orig_inode, donor_inode);
 
-	if (ret)
-		return ret;
+	ret2 = mext_inode_double_unlock(orig_inode, donor_inode);
 
-	/* All of the specified blocks must be exchanged in succeed */
-	BUG_ON(*moved_len != len);
+	if (ret1)
+		return ret1;
+	else if (ret2)
+		return ret2;
 
 	return 0;
 }
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
index 114abe5d2c1d..7c8fe80bacdd 100644
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -1518,8 +1518,12 @@ static int ext4_add_entry(handle_t *handle, struct dentry *dentry,
 			return retval;
 
 		if (blocks == 1 && !dx_fallback &&
-		    EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_DIR_INDEX))
-			return make_indexed_dir(handle, dentry, inode, bh);
+		    EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_DIR_INDEX)) {
+			retval = make_indexed_dir(handle, dentry, inode, bh);
+			if (retval == -ENOSPC)
+				brelse(bh);
+			return retval;
+		}
 		brelse(bh);
 	}
 	bh = ext4_append(handle, dir, &block, &retval);
@@ -1528,7 +1532,10 @@ static int ext4_add_entry(handle_t *handle, struct dentry *dentry,
 	de = (struct ext4_dir_entry_2 *) bh->b_data;
 	de->inode = 0;
 	de->rec_len = ext4_rec_len_to_disk(blocksize, blocksize);
-	return add_dirent_to_buf(handle, dentry, inode, de, bh);
+	retval = add_dirent_to_buf(handle, dentry, inode, de, bh);
+	if (retval == -ENOSPC)
+		brelse(bh);
+	return retval;
 }
 
 /*
@@ -1590,9 +1597,9 @@ static int ext4_dx_add_entry(handle_t *handle, struct dentry *dentry,
 			goto cleanup;
 		node2 = (struct dx_node *)(bh2->b_data);
 		entries2 = node2->entries;
+		memset(&node2->fake, 0, sizeof(struct fake_dirent));
 		node2->fake.rec_len = ext4_rec_len_to_disk(sb->s_blocksize,
 							   sb->s_blocksize);
-		node2->fake.inode = 0;
 		BUFFER_TRACE(frame->bh, "get_write_access");
 		err = ext4_journal_get_write_access(handle, frame->bh);
 		if (err)
@@ -1657,7 +1664,8 @@ static int ext4_dx_add_entry(handle_t *handle, struct dentry *dentry,
 	if (!de)
 		goto cleanup;
 	err = add_dirent_to_buf(handle, dentry, inode, de, bh);
-	bh = NULL;
+	if (err != -ENOSPC)
+		bh = NULL;
 	goto cleanup;
 
 journal_error:
@@ -2068,7 +2076,8 @@ int ext4_orphan_del(handle_t *handle, struct inode *inode)
 	struct ext4_iloc iloc;
 	int err = 0;
 
-	if (!ext4_handle_valid(handle))
+	/* ext4_handle_valid() assumes a valid handle_t pointer */
+	if (handle && !ext4_handle_valid(handle))
 		return 0;
 
 	mutex_lock(&EXT4_SB(inode->i_sb)->s_orphan_lock);
@@ -2310,7 +2319,7 @@ static int ext4_link(struct dentry *old_dentry,
 	struct inode *inode = old_dentry->d_inode;
 	int err, retries = 0;
 
-	if (EXT4_DIR_LINK_MAX(inode))
+	if (inode->i_nlink >= EXT4_LINK_MAX)
 		return -EMLINK;
 
 	/*
@@ -2413,7 +2422,7 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 			goto end_rename;
 		retval = -EMLINK;
 		if (!new_inode && new_dir != old_dir &&
-				new_dir->i_nlink >= EXT4_LINK_MAX)
+		    EXT4_DIR_LINK_MAX(new_dir))
 			goto end_rename;
 	}
 	if (!new_bh) {
diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c
index 68b0351fc647..3cfc343c41b5 100644
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -746,7 +746,6 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
 	struct inode *inode = NULL;
 	handle_t *handle;
 	int gdb_off, gdb_num;
-	int num_grp_locked = 0;
 	int err, err2;
 
 	gdb_num = input->group / EXT4_DESC_PER_BLOCK(sb);
@@ -856,7 +855,6 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
          * using the new disk blocks.
          */
 
-	num_grp_locked = ext4_mb_get_buddy_cache_lock(sb, input->group);
 	/* Update group descriptor block for new group */
 	gdp = (struct ext4_group_desc *)((char *)primary->b_data +
 					 gdb_off * EXT4_DESC_SIZE(sb));
@@ -875,10 +873,8 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
 	 * descriptor
 	 */
 	err = ext4_mb_add_groupinfo(sb, input->group, gdp);
-	if (err) {
-		ext4_mb_put_buddy_cache_lock(sb, input->group, num_grp_locked);
+	if (err)
 		goto exit_journal;
-	}
 
 	/*
 	 * Make the new blocks and inodes valid next.  We do this before
@@ -920,7 +916,6 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
 
 	/* Update the global fs size fields */
 	sbi->s_groups_count++;
-	ext4_mb_put_buddy_cache_lock(sb, input->group, num_grp_locked);
 
 	ext4_handle_dirty_metadata(handle, NULL, primary);
 
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 8f4f079e6b9a..312211ee05af 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -45,17 +45,11 @@
 #include "ext4_jbd2.h"
 #include "xattr.h"
 #include "acl.h"
+#include "mballoc.h"
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/ext4.h>
 
-static int default_mb_history_length = 1000;
-
-module_param_named(default_mb_history_length, default_mb_history_length,
-		   int, 0644);
-MODULE_PARM_DESC(default_mb_history_length,
-		 "Default number of entries saved for mb_history");
-
 struct proc_dir_entry *ext4_proc_root;
 static struct kset *ext4_kset;
 
@@ -188,6 +182,36 @@ void ext4_itable_unused_set(struct super_block *sb,
 		bg->bg_itable_unused_hi = cpu_to_le16(count >> 16);
 }
 
+
+/* Just increment the non-pointer handle value */
+static handle_t *ext4_get_nojournal(void)
+{
+	handle_t *handle = current->journal_info;
+	unsigned long ref_cnt = (unsigned long)handle;
+
+	BUG_ON(ref_cnt >= EXT4_NOJOURNAL_MAX_REF_COUNT);
+
+	ref_cnt++;
+	handle = (handle_t *)ref_cnt;
+
+	current->journal_info = handle;
+	return handle;
+}
+
+
+/* Decrement the non-pointer handle value */
+static void ext4_put_nojournal(handle_t *handle)
+{
+	unsigned long ref_cnt = (unsigned long)handle;
+
+	BUG_ON(ref_cnt == 0);
+
+	ref_cnt--;
+	handle = (handle_t *)ref_cnt;
+
+	current->journal_info = handle;
+}
+
 /*
  * Wrappers for jbd2_journal_start/end.
  *
@@ -214,11 +238,7 @@ handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks)
 		}
 		return jbd2_journal_start(journal, nblocks);
 	}
-	/*
-	 * We're not journaling, return the appropriate indication.
-	 */
-	current->journal_info = EXT4_NOJOURNAL_HANDLE;
-	return current->journal_info;
+	return ext4_get_nojournal();
 }
 
 /*
@@ -234,11 +254,7 @@ int __ext4_journal_stop(const char *where, handle_t *handle)
 	int rc;
 
 	if (!ext4_handle_valid(handle)) {
-		/*
-		 * Do this here since we don't call jbd2_journal_stop() in
-		 * no-journal mode.
-		 */
-		current->journal_info = NULL;
+		ext4_put_nojournal(handle);
 		return 0;
 	}
 	sb = handle->h_transaction->t_journal->j_private;
@@ -344,7 +360,8 @@ static const char *ext4_decode_error(struct super_block *sb, int errno,
 		errstr = "Out of memory";
 		break;
 	case -EROFS:
-		if (!sb || EXT4_SB(sb)->s_journal->j_flags & JBD2_ABORT)
+		if (!sb || (EXT4_SB(sb)->s_journal &&
+			    EXT4_SB(sb)->s_journal->j_flags & JBD2_ABORT))
 			errstr = "Journal has aborted";
 		else
 			errstr = "Readonly filesystem";
@@ -578,6 +595,9 @@ static void ext4_put_super(struct super_block *sb)
 	struct ext4_super_block *es = sbi->s_es;
 	int i, err;
 
+	flush_workqueue(sbi->dio_unwritten_wq);
+	destroy_workqueue(sbi->dio_unwritten_wq);
+
 	lock_super(sb);
 	lock_kernel();
 	if (sb->s_dirt)
@@ -682,6 +702,8 @@ static struct inode *ext4_alloc_inode(struct super_block *sb)
 	ei->i_allocated_meta_blocks = 0;
 	ei->i_delalloc_reserved_flag = 0;
 	spin_lock_init(&(ei->i_block_reservation_lock));
+	INIT_LIST_HEAD(&ei->i_aio_dio_complete_list);
+	ei->cur_aio_dio = NULL;
 
 	return &ei->vfs_inode;
 }
@@ -962,7 +984,7 @@ static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data,
 static ssize_t ext4_quota_write(struct super_block *sb, int type,
 				const char *data, size_t len, loff_t off);
 
-static struct dquot_operations ext4_quota_operations = {
+static const struct dquot_operations ext4_quota_operations = {
 	.initialize	= dquot_initialize,
 	.drop		= dquot_drop,
 	.alloc_space	= dquot_alloc_space,
@@ -983,7 +1005,7 @@ static struct dquot_operations ext4_quota_operations = {
 	.destroy_dquot	= dquot_destroy,
 };
 
-static struct quotactl_ops ext4_qctl_operations = {
+static const struct quotactl_ops ext4_qctl_operations = {
 	.quota_on	= ext4_quota_on,
 	.quota_off	= vfs_quota_off,
 	.quota_sync	= vfs_quota_sync,
@@ -1050,7 +1072,7 @@ enum {
 	Opt_journal_update, Opt_journal_dev,
 	Opt_journal_checksum, Opt_journal_async_commit,
 	Opt_abort, Opt_data_journal, Opt_data_ordered, Opt_data_writeback,
-	Opt_data_err_abort, Opt_data_err_ignore, Opt_mb_history_length,
+	Opt_data_err_abort, Opt_data_err_ignore,
 	Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
 	Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
 	Opt_ignore, Opt_barrier, Opt_nobarrier, Opt_err, Opt_resize,
@@ -1097,7 +1119,6 @@ static const match_table_t tokens = {
 	{Opt_data_writeback, "data=writeback"},
 	{Opt_data_err_abort, "data_err=abort"},
 	{Opt_data_err_ignore, "data_err=ignore"},
-	{Opt_mb_history_length, "mb_history_length=%u"},
 	{Opt_offusrjquota, "usrjquota="},
 	{Opt_usrjquota, "usrjquota=%s"},
 	{Opt_offgrpjquota, "grpjquota="},
@@ -1279,11 +1300,9 @@ static int parse_options(char *options, struct super_block *sb,
 			*journal_devnum = option;
 			break;
 		case Opt_journal_checksum:
-			set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM);
-			break;
+			break;	/* Kept for backwards compatibility */
 		case Opt_journal_async_commit:
 			set_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT);
-			set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM);
 			break;
 		case Opt_noload:
 			set_opt(sbi->s_mount_opt, NOLOAD);
@@ -1340,13 +1359,6 @@ static int parse_options(char *options, struct super_block *sb,
 		case Opt_data_err_ignore:
 			clear_opt(sbi->s_mount_opt, DATA_ERR_ABORT);
 			break;
-		case Opt_mb_history_length:
-			if (match_int(&args[0], &option))
-				return 0;
-			if (option < 0)
-				return 0;
-			sbi->s_mb_history_max = option;
-			break;
 #ifdef CONFIG_QUOTA
 		case Opt_usrjquota:
 			qtype = USRQUOTA;
@@ -1646,13 +1658,6 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
 			EXT4_INODES_PER_GROUP(sb),
 			sbi->s_mount_opt);
 
-	if (EXT4_SB(sb)->s_journal) {
-		ext4_msg(sb, KERN_INFO, "%s journal on %s",
-		       EXT4_SB(sb)->s_journal->j_inode ? "internal" :
-		       "external", EXT4_SB(sb)->s_journal->j_devname);
-	} else {
-		ext4_msg(sb, KERN_INFO, "no journal");
-	}
 	return res;
 }
 
@@ -1695,12 +1700,12 @@ static int ext4_fill_flex_info(struct super_block *sb)
 		gdp = ext4_get_group_desc(sb, i, NULL);
 
 		flex_group = ext4_flex_group(sbi, i);
-		atomic_set(&sbi->s_flex_groups[flex_group].free_inodes,
-			   ext4_free_inodes_count(sb, gdp));
-		atomic_set(&sbi->s_flex_groups[flex_group].free_blocks,
-			   ext4_free_blks_count(sb, gdp));
-		atomic_set(&sbi->s_flex_groups[flex_group].used_dirs,
-			   ext4_used_dirs_count(sb, gdp));
+		atomic_add(ext4_free_inodes_count(sb, gdp),
+			   &sbi->s_flex_groups[flex_group].free_inodes);
+		atomic_add(ext4_free_blks_count(sb, gdp),
+			   &sbi->s_flex_groups[flex_group].free_blocks);
+		atomic_add(ext4_used_dirs_count(sb, gdp),
+			   &sbi->s_flex_groups[flex_group].used_dirs);
 	}
 
 	return 1;
@@ -2197,6 +2202,7 @@ EXT4_RW_ATTR_SBI_UI(mb_min_to_scan, s_mb_min_to_scan);
 EXT4_RW_ATTR_SBI_UI(mb_order2_req, s_mb_order2_reqs);
 EXT4_RW_ATTR_SBI_UI(mb_stream_req, s_mb_stream_request);
 EXT4_RW_ATTR_SBI_UI(mb_group_prealloc, s_mb_group_prealloc);
+EXT4_RW_ATTR_SBI_UI(max_writeback_mb_bump, s_max_writeback_mb_bump);
 
 static struct attribute *ext4_attrs[] = {
 	ATTR_LIST(delayed_allocation_blocks),
@@ -2210,6 +2216,7 @@ static struct attribute *ext4_attrs[] = {
 	ATTR_LIST(mb_order2_req),
 	ATTR_LIST(mb_stream_req),
 	ATTR_LIST(mb_group_prealloc),
+	ATTR_LIST(max_writeback_mb_bump),
 	NULL,
 };
 
@@ -2253,6 +2260,49 @@ static struct kobj_type ext4_ktype = {
 	.release	= ext4_sb_release,
 };
 
+/*
+ * Check whether this filesystem can be mounted based on
+ * the features present and the RDONLY/RDWR mount requested.
+ * Returns 1 if this filesystem can be mounted as requested,
+ * 0 if it cannot be.
+ */
+static int ext4_feature_set_ok(struct super_block *sb, int readonly)
+{
+	if (EXT4_HAS_INCOMPAT_FEATURE(sb, ~EXT4_FEATURE_INCOMPAT_SUPP)) {
+		ext4_msg(sb, KERN_ERR,
+			"Couldn't mount because of "
+			"unsupported optional features (%x)",
+			(le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_incompat) &
+			~EXT4_FEATURE_INCOMPAT_SUPP));
+		return 0;
+	}
+
+	if (readonly)
+		return 1;
+
+	/* Check that feature set is OK for a read-write mount */
+	if (EXT4_HAS_RO_COMPAT_FEATURE(sb, ~EXT4_FEATURE_RO_COMPAT_SUPP)) {
+		ext4_msg(sb, KERN_ERR, "couldn't mount RDWR because of "
+			 "unsupported optional features (%x)",
+			 (le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_ro_compat) &
+				~EXT4_FEATURE_RO_COMPAT_SUPP));
+		return 0;
+	}
+	/*
+	 * Large file size enabled file system can only be mounted
+	 * read-write on 32-bit systems if kernel is built with CONFIG_LBDAF
+	 */
+	if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_HUGE_FILE)) {
+		if (sizeof(blkcnt_t) < sizeof(u64)) {
+			ext4_msg(sb, KERN_ERR, "Filesystem with huge files "
+				 "cannot be mounted RDWR without "
+				 "CONFIG_LBDAF");
+			return 0;
+		}
+	}
+	return 1;
+}
+
 static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 				__releases(kernel_lock)
 				__acquires(kernel_lock)
@@ -2274,7 +2324,6 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	unsigned int db_count;
 	unsigned int i;
 	int needs_recovery, has_huge_files;
-	int features;
 	__u64 blocks_count;
 	int err;
 	unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO;
@@ -2371,7 +2420,6 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	sbi->s_commit_interval = JBD2_DEFAULT_MAX_COMMIT_AGE * HZ;
 	sbi->s_min_batch_time = EXT4_DEF_MIN_BATCH_TIME;
 	sbi->s_max_batch_time = EXT4_DEF_MAX_BATCH_TIME;
-	sbi->s_mb_history_max = default_mb_history_length;
 
 	set_opt(sbi->s_mount_opt, BARRIER);
 
@@ -2401,39 +2449,9 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	 * previously didn't change the revision level when setting the flags,
 	 * so there is a chance incompat flags are set on a rev 0 filesystem.
 	 */
-	features = EXT4_HAS_INCOMPAT_FEATURE(sb, ~EXT4_FEATURE_INCOMPAT_SUPP);
-	if (features) {
-		ext4_msg(sb, KERN_ERR,
-			"Couldn't mount because of "
-			"unsupported optional features (%x)",
-			(le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_incompat) &
-			~EXT4_FEATURE_INCOMPAT_SUPP));
+	if (!ext4_feature_set_ok(sb, (sb->s_flags & MS_RDONLY)))
 		goto failed_mount;
-	}
-	features = EXT4_HAS_RO_COMPAT_FEATURE(sb, ~EXT4_FEATURE_RO_COMPAT_SUPP);
-	if (!(sb->s_flags & MS_RDONLY) && features) {
-		ext4_msg(sb, KERN_ERR,
-			"Couldn't mount RDWR because of "
-			"unsupported optional features (%x)",
-			(le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_ro_compat) &
-			~EXT4_FEATURE_RO_COMPAT_SUPP));
-		goto failed_mount;
-	}
-	has_huge_files = EXT4_HAS_RO_COMPAT_FEATURE(sb,
-				    EXT4_FEATURE_RO_COMPAT_HUGE_FILE);
-	if (has_huge_files) {
-		/*
-		 * Large file size enabled file system can only be
-		 * mount if kernel is build with CONFIG_LBDAF
-		 */
-		if (sizeof(root->i_blocks) < sizeof(u64) &&
-				!(sb->s_flags & MS_RDONLY)) {
-			ext4_msg(sb, KERN_ERR, "Filesystem with huge "
-					"files cannot be mounted read-write "
-					"without CONFIG_LBDAF");
-			goto failed_mount;
-		}
-	}
+
 	blocksize = BLOCK_SIZE << le32_to_cpu(es->s_log_block_size);
 
 	if (blocksize < EXT4_MIN_BLOCK_SIZE ||
@@ -2469,6 +2487,8 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 		}
 	}
 
+	has_huge_files = EXT4_HAS_RO_COMPAT_FEATURE(sb,
+				EXT4_FEATURE_RO_COMPAT_HUGE_FILE);
 	sbi->s_bitmap_maxbytes = ext4_max_bitmap_size(sb->s_blocksize_bits,
 						      has_huge_files);
 	sb->s_maxbytes = ext4_max_size(sb->s_blocksize_bits, has_huge_files);
@@ -2549,12 +2569,19 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 		goto failed_mount;
 	}
 
-	if (ext4_blocks_count(es) >
-		    (sector_t)(~0ULL) >> (sb->s_blocksize_bits - 9)) {
+	/*
+	 * Test whether we have more sectors than will fit in sector_t,
+	 * and whether the max offset is addressable by the page cache.
+	 */
+	if ((ext4_blocks_count(es) >
+	     (sector_t)(~0ULL) >> (sb->s_blocksize_bits - 9)) ||
+	    (ext4_blocks_count(es) >
+	     (pgoff_t)(~0ULL) >> (PAGE_CACHE_SHIFT - sb->s_blocksize_bits))) {
 		ext4_msg(sb, KERN_ERR, "filesystem"
-			" too large to mount safely");
+			 " too large to mount safely on this system");
 		if (sizeof(sector_t) < 8)
 			ext4_msg(sb, KERN_WARNING, "CONFIG_LBDAF not enabled");
+		ret = -EFBIG;
 		goto failed_mount;
 	}
 
@@ -2595,6 +2622,8 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 		goto failed_mount;
 	}
 	sbi->s_groups_count = blocks_count;
+	sbi->s_blockfile_groups = min_t(ext4_group_t, sbi->s_groups_count,
+			(EXT4_MAX_BLOCK_FILE_PHYS / EXT4_BLOCKS_PER_GROUP(sb)));
 	db_count = (sbi->s_groups_count + EXT4_DESC_PER_BLOCK(sb) - 1) /
 		   EXT4_DESC_PER_BLOCK(sb);
 	sbi->s_group_desc = kmalloc(db_count * sizeof(struct buffer_head *),
@@ -2656,6 +2685,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	}
 
 	sbi->s_stripe = ext4_get_stripe_size(sbi);
+	sbi->s_max_writeback_mb_bump = 128;
 
 	/*
 	 * set up enough so that it can read an inode
@@ -2729,20 +2759,14 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 		goto failed_mount4;
 	}
 
-	if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) {
-		jbd2_journal_set_features(sbi->s_journal,
-				JBD2_FEATURE_COMPAT_CHECKSUM, 0,
+	jbd2_journal_set_features(sbi->s_journal,
+				  JBD2_FEATURE_COMPAT_CHECKSUM, 0, 0);
+	if (test_opt(sb, JOURNAL_ASYNC_COMMIT))
+		jbd2_journal_set_features(sbi->s_journal, 0, 0,
 				JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
-	} else if (test_opt(sb, JOURNAL_CHECKSUM)) {
-		jbd2_journal_set_features(sbi->s_journal,
-				JBD2_FEATURE_COMPAT_CHECKSUM, 0, 0);
+	else
 		jbd2_journal_clear_features(sbi->s_journal, 0, 0,
 				JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
-	} else {
-		jbd2_journal_clear_features(sbi->s_journal,
-				JBD2_FEATURE_COMPAT_CHECKSUM, 0,
-				JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
-	}
 
 	/* We have now updated the journal if required, so we can
 	 * validate the data journaling mode. */
@@ -2781,6 +2805,12 @@ no_journal:
 			clear_opt(sbi->s_mount_opt, NOBH);
 		}
 	}
+	EXT4_SB(sb)->dio_unwritten_wq = create_workqueue("ext4-dio-unwritten");
+	if (!EXT4_SB(sb)->dio_unwritten_wq) {
+		printk(KERN_ERR "EXT4-fs: failed to create DIO workqueue\n");
+		goto failed_mount_wq;
+	}
+
 	/*
 	 * The jbd2_journal_load will have done any necessary log recovery,
 	 * so we can safely mount the rest of the filesystem now.
@@ -2832,12 +2862,12 @@ no_journal:
 			 "available");
 	}
 
-	if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA) {
+	if (test_opt(sb, DELALLOC) &&
+	    (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA)) {
 		ext4_msg(sb, KERN_WARNING, "Ignoring delalloc option - "
 			 "requested data journaling mode");
 		clear_opt(sbi->s_mount_opt, DELALLOC);
-	} else if (test_opt(sb, DELALLOC))
-		ext4_msg(sb, KERN_INFO, "delayed allocation enabled");
+	}
 
 	err = ext4_setup_system_zone(sb);
 	if (err) {
@@ -2893,6 +2923,8 @@ cantfind_ext4:
 
 failed_mount4:
 	ext4_msg(sb, KERN_ERR, "mount failed");
+	destroy_workqueue(EXT4_SB(sb)->dio_unwritten_wq);
+failed_mount_wq:
 	ext4_release_system_zone(sb);
 	if (sbi->s_journal) {
 		jbd2_journal_destroy(sbi->s_journal);
@@ -3147,9 +3179,7 @@ static int ext4_load_journal(struct super_block *sb,
 			return -EINVAL;
 	}
 
-	if (journal->j_flags & JBD2_BARRIER)
-		ext4_msg(sb, KERN_INFO, "barriers enabled");
-	else
+	if (!(journal->j_flags & JBD2_BARRIER))
 		ext4_msg(sb, KERN_INFO, "barriers disabled");
 
 	if (!really_read_only && test_opt(sb, UPDATE_JOURNAL)) {
@@ -3208,7 +3238,18 @@ static int ext4_commit_super(struct super_block *sb, int sync)
 		clear_buffer_write_io_error(sbh);
 		set_buffer_uptodate(sbh);
 	}
-	es->s_wtime = cpu_to_le32(get_seconds());
+	/*
+	 * If the file system is mounted read-only, don't update the
+	 * superblock write time.  This avoids updating the superblock
+	 * write time when we are mounting the root file system
+	 * read/only but we need to replay the journal; at that point,
+	 * for people who are east of GMT and who make their clock
+	 * tick in localtime for Windows bug-for-bug compatibility,
+	 * the clock is set in the future, and this will cause e2fsck
+	 * to complain and force a full file system check.
+	 */
+	if (!(sb->s_flags & MS_RDONLY))
+		es->s_wtime = cpu_to_le32(get_seconds());
 	es->s_kbytes_written =
 		cpu_to_le64(EXT4_SB(sb)->s_kbytes_written + 
 			    ((part_stat_read(sb->s_bdev->bd_part, sectors[1]) -
@@ -3333,11 +3374,13 @@ static int ext4_sync_fs(struct super_block *sb, int wait)
 {
 	int ret = 0;
 	tid_t target;
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
 
 	trace_ext4_sync_fs(sb, wait);
-	if (jbd2_journal_start_commit(EXT4_SB(sb)->s_journal, &target)) {
+	flush_workqueue(sbi->dio_unwritten_wq);
+	if (jbd2_journal_start_commit(sbi->s_journal, &target)) {
 		if (wait)
-			jbd2_log_wait_commit(EXT4_SB(sb)->s_journal, target);
+			jbd2_log_wait_commit(sbi->s_journal, target);
 	}
 	return ret;
 }
@@ -3477,18 +3520,11 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 			if (sbi->s_journal)
 				ext4_mark_recovery_complete(sb, es);
 		} else {
-			int ret;
-			if ((ret = EXT4_HAS_RO_COMPAT_FEATURE(sb,
-					~EXT4_FEATURE_RO_COMPAT_SUPP))) {
-				ext4_msg(sb, KERN_WARNING, "couldn't "
-				       "remount RDWR because of unsupported "
-				       "optional features (%x)",
-				(le32_to_cpu(sbi->s_es->s_feature_ro_compat) &
-					~EXT4_FEATURE_RO_COMPAT_SUPP));
+			/* Make sure we can mount this feature set readwrite */
+			if (!ext4_feature_set_ok(sb, 0)) {
 				err = -EROFS;
 				goto restore_opts;
 			}
-
 			/*
 			 * Make sure the group descriptor checksums
 			 * are sane.  If they aren't, refuse to remount r/w.
@@ -3930,27 +3966,6 @@ static struct file_system_type ext4_fs_type = {
 	.fs_flags	= FS_REQUIRES_DEV,
 };
 
-#ifdef CONFIG_EXT4DEV_COMPAT
-static int ext4dev_get_sb(struct file_system_type *fs_type, int flags,
-			  const char *dev_name, void *data,struct vfsmount *mnt)
-{
-	printk(KERN_WARNING "EXT4-fs (%s): Update your userspace programs "
-	       "to mount using ext4\n", dev_name);
-	printk(KERN_WARNING "EXT4-fs (%s): ext4dev backwards compatibility "
-	       "will go away by 2.6.31\n", dev_name);
-	return get_sb_bdev(fs_type, flags, dev_name, data, ext4_fill_super,mnt);
-}
-
-static struct file_system_type ext4dev_fs_type = {
-	.owner		= THIS_MODULE,
-	.name		= "ext4dev",
-	.get_sb		= ext4dev_get_sb,
-	.kill_sb	= kill_block_super,
-	.fs_flags	= FS_REQUIRES_DEV,
-};
-MODULE_ALIAS("ext4dev");
-#endif
-
 static int __init init_ext4_fs(void)
 {
 	int err;
@@ -3975,13 +3990,6 @@ static int __init init_ext4_fs(void)
 	err = register_filesystem(&ext4_fs_type);
 	if (err)
 		goto out;
-#ifdef CONFIG_EXT4DEV_COMPAT
-	err = register_filesystem(&ext4dev_fs_type);
-	if (err) {
-		unregister_filesystem(&ext4_fs_type);
-		goto out;
-	}
-#endif
 	return 0;
 out:
 	destroy_inodecache();
@@ -4000,9 +4008,6 @@ out4:
 static void __exit exit_ext4_fs(void)
 {
 	unregister_filesystem(&ext4_fs_type);
-#ifdef CONFIG_EXT4DEV_COMPAT
-	unregister_filesystem(&ext4dev_fs_type);
-#endif
 	destroy_inodecache();
 	exit_ext4_xattr();
 	exit_ext4_mballoc();
diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c
index 62b31c246994..fed5b01d7a8d 100644
--- a/fs/ext4/xattr.c
+++ b/fs/ext4/xattr.c
@@ -810,12 +810,23 @@ inserted:
 			get_bh(new_bh);
 		} else {
 			/* We need to allocate a new block */
-			ext4_fsblk_t goal = ext4_group_first_block_no(sb,
+			ext4_fsblk_t goal, block;
+
+			goal = ext4_group_first_block_no(sb,
 						EXT4_I(inode)->i_block_group);
-			ext4_fsblk_t block = ext4_new_meta_blocks(handle, inode,
+
+			/* non-extent files can't have physical blocks past 2^32 */
+			if (!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
+				goal = goal & EXT4_MAX_BLOCK_FILE_PHYS;
+
+			block = ext4_new_meta_blocks(handle, inode,
 						  goal, NULL, &error);
 			if (error)
 				goto cleanup;
+
+			if (!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
+				BUG_ON(block > EXT4_MAX_BLOCK_FILE_PHYS);
+
 			ea_idebug(inode, "creating block %d", block);
 
 			new_bh = sb_getblk(sb, block);
diff --git a/fs/fat/fat.h b/fs/fat/fat.h
index adb0e72a176d..7db0979c6b72 100644
--- a/fs/fat/fat.h
+++ b/fs/fat/fat.h
@@ -323,7 +323,7 @@ extern int fat_flush_inodes(struct super_block *sb, struct inode *i1,
 /* fat/misc.c */
 extern void fat_fs_error(struct super_block *s, const char *fmt, ...)
 	__attribute__ ((format (printf, 2, 3))) __cold;
-extern void fat_clusters_flush(struct super_block *sb);
+extern int fat_clusters_flush(struct super_block *sb);
 extern int fat_chain_add(struct inode *inode, int new_dclus, int nr_cluster);
 extern void fat_time_fat2unix(struct msdos_sb_info *sbi, struct timespec *ts,
 			      __le16 __time, __le16 __date, u8 time_cs);
diff --git a/fs/fat/inode.c b/fs/fat/inode.c
index 8970d8c49bb0..76b7961ab663 100644
--- a/fs/fat/inode.c
+++ b/fs/fat/inode.c
@@ -451,12 +451,16 @@ static void fat_write_super(struct super_block *sb)
 
 static int fat_sync_fs(struct super_block *sb, int wait)
 {
-	lock_super(sb);
-	fat_clusters_flush(sb);
-	sb->s_dirt = 0;
-	unlock_super(sb);
+	int err = 0;
 
-	return 0;
+	if (sb->s_dirt) {
+		lock_super(sb);
+		sb->s_dirt = 0;
+		err = fat_clusters_flush(sb);
+		unlock_super(sb);
+	}
+
+	return err;
 }
 
 static void fat_put_super(struct super_block *sb)
@@ -470,19 +474,11 @@ static void fat_put_super(struct super_block *sb)
 
 	iput(sbi->fat_inode);
 
-	if (sbi->nls_disk) {
-		unload_nls(sbi->nls_disk);
-		sbi->nls_disk = NULL;
-		sbi->options.codepage = fat_default_codepage;
-	}
-	if (sbi->nls_io) {
-		unload_nls(sbi->nls_io);
-		sbi->nls_io = NULL;
-	}
-	if (sbi->options.iocharset != fat_default_iocharset) {
+	unload_nls(sbi->nls_disk);
+	unload_nls(sbi->nls_io);
+
+	if (sbi->options.iocharset != fat_default_iocharset)
 		kfree(sbi->options.iocharset);
-		sbi->options.iocharset = fat_default_iocharset;
-	}
 
 	sb->s_fs_info = NULL;
 	kfree(sbi);
@@ -820,7 +816,7 @@ static int fat_show_options(struct seq_file *m, struct vfsmount *mnt)
 			seq_puts(m, ",shortname=mixed");
 			break;
 		case VFAT_SFN_DISPLAY_LOWER | VFAT_SFN_CREATE_WIN95:
-			/* seq_puts(m, ",shortname=lower"); */
+			seq_puts(m, ",shortname=lower");
 			break;
 		default:
 			seq_puts(m, ",shortname=unknown");
@@ -971,7 +967,7 @@ static int parse_options(char *options, int is_vfat, int silent, int *debug,
 	opts->codepage = fat_default_codepage;
 	opts->iocharset = fat_default_iocharset;
 	if (is_vfat) {
-		opts->shortname = VFAT_SFN_DISPLAY_LOWER|VFAT_SFN_CREATE_WIN95;
+		opts->shortname = VFAT_SFN_DISPLAY_WINNT|VFAT_SFN_CREATE_WIN95;
 		opts->rodir = 0;
 	} else {
 		opts->shortname = 0;
diff --git a/fs/fat/misc.c b/fs/fat/misc.c
index 4e35be873e09..0f55f5cb732f 100644
--- a/fs/fat/misc.c
+++ b/fs/fat/misc.c
@@ -43,19 +43,19 @@ EXPORT_SYMBOL_GPL(fat_fs_error);
 
 /* Flushes the number of free clusters on FAT32 */
 /* XXX: Need to write one per FSINFO block.  Currently only writes 1 */
-void fat_clusters_flush(struct super_block *sb)
+int fat_clusters_flush(struct super_block *sb)
 {
 	struct msdos_sb_info *sbi = MSDOS_SB(sb);
 	struct buffer_head *bh;
 	struct fat_boot_fsinfo *fsinfo;
 
 	if (sbi->fat_bits != 32)
-		return;
+		return 0;
 
 	bh = sb_bread(sb, sbi->fsinfo_sector);
 	if (bh == NULL) {
 		printk(KERN_ERR "FAT: bread failed in fat_clusters_flush\n");
-		return;
+		return -EIO;
 	}
 
 	fsinfo = (struct fat_boot_fsinfo *)bh->b_data;
@@ -74,6 +74,8 @@ void fat_clusters_flush(struct super_block *sb)
 		mark_buffer_dirty(bh);
 	}
 	brelse(bh);
+
+	return 0;
 }
 
 /*
diff --git a/fs/fat/namei_vfat.c b/fs/fat/namei_vfat.c
index cb6e83557112..f565f24019b5 100644
--- a/fs/fat/namei_vfat.c
+++ b/fs/fat/namei_vfat.c
@@ -499,17 +499,10 @@ xlate_to_uni(const unsigned char *name, int len, unsigned char *outname,
 	int charlen;
 
 	if (utf8) {
-		int name_len = strlen(name);
-
-		*outlen = utf8s_to_utf16s(name, PATH_MAX, (wchar_t *) outname);
-
-		/*
-		 * We stripped '.'s before and set len appropriately,
-		 * but utf8s_to_utf16s doesn't care about len
-		 */
-		*outlen -= (name_len - len);
-
-		if (*outlen > 255)
+		*outlen = utf8s_to_utf16s(name, len, (wchar_t *)outname);
+		if (*outlen < 0)
+			return *outlen;
+		else if (*outlen > 255)
 			return -ENAMETOOLONG;
 
 		op = &outname[*outlen * sizeof(wchar_t)];
diff --git a/fs/fcntl.c b/fs/fcntl.c
index ae413086db97..fc089f2f7f56 100644
--- a/fs/fcntl.c
+++ b/fs/fcntl.c
@@ -263,6 +263,79 @@ pid_t f_getown(struct file *filp)
 	return pid;
 }
 
+static int f_setown_ex(struct file *filp, unsigned long arg)
+{
+	struct f_owner_ex * __user owner_p = (void * __user)arg;
+	struct f_owner_ex owner;
+	struct pid *pid;
+	int type;
+	int ret;
+
+	ret = copy_from_user(&owner, owner_p, sizeof(owner));
+	if (ret)
+		return ret;
+
+	switch (owner.type) {
+	case F_OWNER_TID:
+		type = PIDTYPE_MAX;
+		break;
+
+	case F_OWNER_PID:
+		type = PIDTYPE_PID;
+		break;
+
+	case F_OWNER_GID:
+		type = PIDTYPE_PGID;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	rcu_read_lock();
+	pid = find_vpid(owner.pid);
+	if (owner.pid && !pid)
+		ret = -ESRCH;
+	else
+		ret = __f_setown(filp, pid, type, 1);
+	rcu_read_unlock();
+
+	return ret;
+}
+
+static int f_getown_ex(struct file *filp, unsigned long arg)
+{
+	struct f_owner_ex * __user owner_p = (void * __user)arg;
+	struct f_owner_ex owner;
+	int ret = 0;
+
+	read_lock(&filp->f_owner.lock);
+	owner.pid = pid_vnr(filp->f_owner.pid);
+	switch (filp->f_owner.pid_type) {
+	case PIDTYPE_MAX:
+		owner.type = F_OWNER_TID;
+		break;
+
+	case PIDTYPE_PID:
+		owner.type = F_OWNER_PID;
+		break;
+
+	case PIDTYPE_PGID:
+		owner.type = F_OWNER_GID;
+		break;
+
+	default:
+		WARN_ON(1);
+		ret = -EINVAL;
+		break;
+	}
+	read_unlock(&filp->f_owner.lock);
+
+	if (!ret)
+		ret = copy_to_user(owner_p, &owner, sizeof(owner));
+	return ret;
+}
+
 static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
 		struct file *filp)
 {
@@ -313,6 +386,12 @@ static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
 	case F_SETOWN:
 		err = f_setown(filp, arg, 1);
 		break;
+	case F_GETOWN_EX:
+		err = f_getown_ex(filp, arg);
+		break;
+	case F_SETOWN_EX:
+		err = f_setown_ex(filp, arg);
+		break;
 	case F_GETSIG:
 		err = filp->f_owner.signum;
 		break;
@@ -428,8 +507,7 @@ static inline int sigio_perm(struct task_struct *p,
 
 static void send_sigio_to_task(struct task_struct *p,
 			       struct fown_struct *fown,
-			       int fd,
-			       int reason)
+			       int fd, int reason, int group)
 {
 	/*
 	 * F_SETSIG can change ->signum lockless in parallel, make
@@ -461,11 +539,11 @@ static void send_sigio_to_task(struct task_struct *p,
 			else
 				si.si_band = band_table[reason - POLL_IN];
 			si.si_fd    = fd;
-			if (!group_send_sig_info(signum, &si, p))
+			if (!do_send_sig_info(signum, &si, p, group))
 				break;
 		/* fall-through: fall back on the old plain SIGIO signal */
 		case 0:
-			group_send_sig_info(SIGIO, SEND_SIG_PRIV, p);
+			do_send_sig_info(SIGIO, SEND_SIG_PRIV, p, group);
 	}
 }
 
@@ -474,16 +552,23 @@ void send_sigio(struct fown_struct *fown, int fd, int band)
 	struct task_struct *p;
 	enum pid_type type;
 	struct pid *pid;
+	int group = 1;
 	
 	read_lock(&fown->lock);
+
 	type = fown->pid_type;
+	if (type == PIDTYPE_MAX) {
+		group = 0;
+		type = PIDTYPE_PID;
+	}
+
 	pid = fown->pid;
 	if (!pid)
 		goto out_unlock_fown;
 	
 	read_lock(&tasklist_lock);
 	do_each_pid_task(pid, type, p) {
-		send_sigio_to_task(p, fown, fd, band);
+		send_sigio_to_task(p, fown, fd, band, group);
 	} while_each_pid_task(pid, type, p);
 	read_unlock(&tasklist_lock);
  out_unlock_fown:
@@ -491,10 +576,10 @@ void send_sigio(struct fown_struct *fown, int fd, int band)
 }
 
 static void send_sigurg_to_task(struct task_struct *p,
-                                struct fown_struct *fown)
+				struct fown_struct *fown, int group)
 {
 	if (sigio_perm(p, fown, SIGURG))
-		group_send_sig_info(SIGURG, SEND_SIG_PRIV, p);
+		do_send_sig_info(SIGURG, SEND_SIG_PRIV, p, group);
 }
 
 int send_sigurg(struct fown_struct *fown)
@@ -502,10 +587,17 @@ int send_sigurg(struct fown_struct *fown)
 	struct task_struct *p;
 	enum pid_type type;
 	struct pid *pid;
+	int group = 1;
 	int ret = 0;
 	
 	read_lock(&fown->lock);
+
 	type = fown->pid_type;
+	if (type == PIDTYPE_MAX) {
+		group = 0;
+		type = PIDTYPE_PID;
+	}
+
 	pid = fown->pid;
 	if (!pid)
 		goto out_unlock_fown;
@@ -514,7 +606,7 @@ int send_sigurg(struct fown_struct *fown)
 	
 	read_lock(&tasklist_lock);
 	do_each_pid_task(pid, type, p) {
-		send_sigurg_to_task(p, fown);
+		send_sigurg_to_task(p, fown, group);
 	} while_each_pid_task(pid, type, p);
 	read_unlock(&tasklist_lock);
  out_unlock_fown:
diff --git a/fs/file.c b/fs/file.c
index f313314f996f..87e129030ab1 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -10,6 +10,7 @@
 #include <linux/fs.h>
 #include <linux/mm.h>
 #include <linux/time.h>
+#include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/vmalloc.h>
 #include <linux/file.h>
diff --git a/fs/file_table.c b/fs/file_table.c
index 334ce39881f8..8eb44042e009 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -74,14 +74,14 @@ EXPORT_SYMBOL_GPL(get_max_files);
  * Handle nr_files sysctl
  */
 #if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
-int proc_nr_files(ctl_table *table, int write, struct file *filp,
+int proc_nr_files(ctl_table *table, int write,
                      void __user *buffer, size_t *lenp, loff_t *ppos)
 {
 	files_stat.nr_files = get_nr_files();
-	return proc_dointvec(table, write, filp, buffer, lenp, ppos);
+	return proc_dointvec(table, write, buffer, lenp, ppos);
 }
 #else
-int proc_nr_files(ctl_table *table, int write, struct file *filp,
+int proc_nr_files(ctl_table *table, int write,
                      void __user *buffer, size_t *lenp, loff_t *ppos)
 {
 	return -ENOSYS;
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 8e1e5e19d21e..9d5360c4c2af 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -41,8 +41,9 @@ struct wb_writeback_args {
 	long nr_pages;
 	struct super_block *sb;
 	enum writeback_sync_modes sync_mode;
-	int for_kupdate;
-	int range_cyclic;
+	int for_kupdate:1;
+	int range_cyclic:1;
+	int for_background:1;
 };
 
 /*
@@ -249,14 +250,25 @@ static void bdi_sync_writeback(struct backing_dev_info *bdi,
  *   completion. Caller need not hold sb s_umount semaphore.
  *
  */
-void bdi_start_writeback(struct backing_dev_info *bdi, long nr_pages)
+void bdi_start_writeback(struct backing_dev_info *bdi, struct super_block *sb,
+			 long nr_pages)
 {
 	struct wb_writeback_args args = {
+		.sb		= sb,
 		.sync_mode	= WB_SYNC_NONE,
 		.nr_pages	= nr_pages,
 		.range_cyclic	= 1,
 	};
 
+	/*
+	 * We treat @nr_pages=0 as the special case to do background writeback,
+	 * ie. to sync pages until the background dirty threshold is reached.
+	 */
+	if (!nr_pages) {
+		args.nr_pages = LONG_MAX;
+		args.for_background = 1;
+	}
+
 	bdi_alloc_queue_work(bdi, &args);
 }
 
@@ -310,7 +322,7 @@ static bool inode_dirtied_after(struct inode *inode, unsigned long t)
 	 * For inodes being constantly redirtied, dirtied_when can get stuck.
 	 * It _appears_ to be in the future, but is actually in distant past.
 	 * This test is necessary to prevent such wrapped-around relative times
-	 * from permanently stopping the whole pdflush writeback.
+	 * from permanently stopping the whole bdi writeback.
 	 */
 	ret = ret && time_before_eq(inode->dirtied_when, jiffies);
 #endif
@@ -324,13 +336,38 @@ static void move_expired_inodes(struct list_head *delaying_queue,
 			       struct list_head *dispatch_queue,
 				unsigned long *older_than_this)
 {
+	LIST_HEAD(tmp);
+	struct list_head *pos, *node;
+	struct super_block *sb = NULL;
+	struct inode *inode;
+	int do_sb_sort = 0;
+
 	while (!list_empty(delaying_queue)) {
-		struct inode *inode = list_entry(delaying_queue->prev,
-						struct inode, i_list);
+		inode = list_entry(delaying_queue->prev, struct inode, i_list);
 		if (older_than_this &&
 		    inode_dirtied_after(inode, *older_than_this))
 			break;
-		list_move(&inode->i_list, dispatch_queue);
+		if (sb && sb != inode->i_sb)
+			do_sb_sort = 1;
+		sb = inode->i_sb;
+		list_move(&inode->i_list, &tmp);
+	}
+
+	/* just one sb in list, splice to dispatch_queue and we're done */
+	if (!do_sb_sort) {
+		list_splice(&tmp, dispatch_queue);
+		return;
+	}
+
+	/* Move inodes from one superblock together */
+	while (!list_empty(&tmp)) {
+		inode = list_entry(tmp.prev, struct inode, i_list);
+		sb = inode->i_sb;
+		list_for_each_prev_safe(pos, node, &tmp) {
+			inode = list_entry(pos, struct inode, i_list);
+			if (inode->i_sb == sb)
+				list_move(&inode->i_list, dispatch_queue);
+		}
 	}
 }
 
@@ -439,8 +476,18 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
 	spin_lock(&inode_lock);
 	inode->i_state &= ~I_SYNC;
 	if (!(inode->i_state & (I_FREEING | I_CLEAR))) {
-		if (!(inode->i_state & I_DIRTY) &&
-		    mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
+		if ((inode->i_state & I_DIRTY_PAGES) && wbc->for_kupdate) {
+			/*
+			 * More pages get dirtied by a fast dirtier.
+			 */
+			goto select_queue;
+		} else if (inode->i_state & I_DIRTY) {
+			/*
+			 * At least XFS will redirty the inode during the
+			 * writeback (delalloc) and on io completion (isize).
+			 */
+			redirty_tail(inode);
+		} else if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
 			/*
 			 * We didn't write back all the pages.  nfs_writepages()
 			 * sometimes bales out without doing anything. Redirty
@@ -462,6 +509,7 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
 				 * soon as the queue becomes uncongested.
 				 */
 				inode->i_state |= I_DIRTY_PAGES;
+select_queue:
 				if (wbc->nr_to_write <= 0) {
 					/*
 					 * slice used up: queue for next turn
@@ -484,12 +532,6 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
 				inode->i_state |= I_DIRTY_PAGES;
 				redirty_tail(inode);
 			}
-		} else if (inode->i_state & I_DIRTY) {
-			/*
-			 * Someone redirtied the inode while were writing back
-			 * the pages.
-			 */
-			redirty_tail(inode);
 		} else if (atomic_read(&inode->i_count)) {
 			/*
 			 * The inode is clean, inuse
@@ -506,6 +548,17 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
 	return ret;
 }
 
+static void unpin_sb_for_writeback(struct super_block **psb)
+{
+	struct super_block *sb = *psb;
+
+	if (sb) {
+		up_read(&sb->s_umount);
+		put_super(sb);
+		*psb = NULL;
+	}
+}
+
 /*
  * For WB_SYNC_NONE writeback, the caller does not have the sb pinned
  * before calling writeback. So make sure that we do pin it, so it doesn't
@@ -515,11 +568,20 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
  * 1 if we failed.
  */
 static int pin_sb_for_writeback(struct writeback_control *wbc,
-				   struct inode *inode)
+				struct inode *inode, struct super_block **psb)
 {
 	struct super_block *sb = inode->i_sb;
 
 	/*
+	 * If this sb is already pinned, nothing more to do. If not and
+	 * *psb is non-NULL, unpin the old one first
+	 */
+	if (sb == *psb)
+		return 0;
+	else if (*psb)
+		unpin_sb_for_writeback(psb);
+
+	/*
 	 * Caller must already hold the ref for this
 	 */
 	if (wbc->sync_mode == WB_SYNC_ALL) {
@@ -532,7 +594,7 @@ static int pin_sb_for_writeback(struct writeback_control *wbc,
 	if (down_read_trylock(&sb->s_umount)) {
 		if (sb->s_root) {
 			spin_unlock(&sb_lock);
-			return 0;
+			goto pinned;
 		}
 		/*
 		 * umounted, drop rwsem again and fall through to failure
@@ -543,24 +605,15 @@ static int pin_sb_for_writeback(struct writeback_control *wbc,
 	sb->s_count--;
 	spin_unlock(&sb_lock);
 	return 1;
-}
-
-static void unpin_sb_for_writeback(struct writeback_control *wbc,
-				   struct inode *inode)
-{
-	struct super_block *sb = inode->i_sb;
-
-	if (wbc->sync_mode == WB_SYNC_ALL)
-		return;
-
-	up_read(&sb->s_umount);
-	put_super(sb);
+pinned:
+	*psb = sb;
+	return 0;
 }
 
 static void writeback_inodes_wb(struct bdi_writeback *wb,
 				struct writeback_control *wbc)
 {
-	struct super_block *sb = wbc->sb;
+	struct super_block *sb = wbc->sb, *pin_sb = NULL;
 	const int is_blkdev_sb = sb_is_blkdev_sb(sb);
 	const unsigned long start = jiffies;	/* livelock avoidance */
 
@@ -619,7 +672,7 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
 		if (inode_dirtied_after(inode, start))
 			break;
 
-		if (pin_sb_for_writeback(wbc, inode)) {
+		if (pin_sb_for_writeback(wbc, inode, &pin_sb)) {
 			requeue_io(inode);
 			continue;
 		}
@@ -628,7 +681,6 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
 		__iget(inode);
 		pages_skipped = wbc->pages_skipped;
 		writeback_single_inode(inode, wbc);
-		unpin_sb_for_writeback(wbc, inode);
 		if (wbc->pages_skipped != pages_skipped) {
 			/*
 			 * writeback is not making progress due to locked
@@ -648,6 +700,8 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
 			wbc->more_io = 1;
 	}
 
+	unpin_sb_for_writeback(&pin_sb);
+
 	spin_unlock(&inode_lock);
 	/* Leave any unwritten inodes on b_io */
 }
@@ -706,6 +760,7 @@ static long wb_writeback(struct bdi_writeback *wb,
 	};
 	unsigned long oldest_jif;
 	long wrote = 0;
+	struct inode *inode;
 
 	if (wbc.for_kupdate) {
 		wbc.older_than_this = &oldest_jif;
@@ -719,20 +774,16 @@ static long wb_writeback(struct bdi_writeback *wb,
 
 	for (;;) {
 		/*
-		 * Don't flush anything for non-integrity writeback where
-		 * no nr_pages was given
+		 * Stop writeback when nr_pages has been consumed
 		 */
-		if (!args->for_kupdate && args->nr_pages <= 0 &&
-		     args->sync_mode == WB_SYNC_NONE)
+		if (args->nr_pages <= 0)
 			break;
 
 		/*
-		 * If no specific pages were given and this is just a
-		 * periodic background writeout and we are below the
-		 * background dirty threshold, don't do anything
+		 * For background writeout, stop when we are below the
+		 * background dirty threshold
 		 */
-		if (args->for_kupdate && args->nr_pages <= 0 &&
-		    !over_bground_thresh())
+		if (args->for_background && !over_bground_thresh())
 			break;
 
 		wbc.more_io = 0;
@@ -744,13 +795,32 @@ static long wb_writeback(struct bdi_writeback *wb,
 		wrote += MAX_WRITEBACK_PAGES - wbc.nr_to_write;
 
 		/*
-		 * If we ran out of stuff to write, bail unless more_io got set
+		 * If we consumed everything, see if we have more
 		 */
-		if (wbc.nr_to_write > 0 || wbc.pages_skipped > 0) {
-			if (wbc.more_io && !wbc.for_kupdate)
-				continue;
+		if (wbc.nr_to_write <= 0)
+			continue;
+		/*
+		 * Didn't write everything and we don't have more IO, bail
+		 */
+		if (!wbc.more_io)
 			break;
+		/*
+		 * Did we write something? Try for more
+		 */
+		if (wbc.nr_to_write < MAX_WRITEBACK_PAGES)
+			continue;
+		/*
+		 * Nothing written. Wait for some inode to
+		 * become available for writeback. Otherwise
+		 * we'll just busyloop.
+		 */
+		spin_lock(&inode_lock);
+		if (!list_empty(&wb->b_more_io))  {
+			inode = list_entry(wb->b_more_io.prev,
+						struct inode, i_list);
+			inode_wait_for_writeback(inode);
 		}
+		spin_unlock(&inode_lock);
 	}
 
 	return wrote;
@@ -1060,9 +1130,6 @@ EXPORT_SYMBOL(__mark_inode_dirty);
  * If older_than_this is non-NULL, then only write out inodes which
  * had their first dirtying at a time earlier than *older_than_this.
  *
- * If we're a pdlfush thread, then implement pdflush collision avoidance
- * against the entire list.
- *
  * If `bdi' is non-zero then we're being asked to writeback a specific queue.
  * This function assumes that the blockdev superblock's inodes are backed by
  * a variety of queues, so all inodes are searched.  For other superblocks,
@@ -1141,7 +1208,7 @@ void writeback_inodes_sb(struct super_block *sb)
 	nr_to_write = nr_dirty + nr_unstable +
 			(inodes_stat.nr_inodes - inodes_stat.nr_unused);
 
-	bdi_writeback_all(sb, nr_to_write);
+	bdi_start_writeback(sb->s_bdi, sb, nr_to_write);
 }
 EXPORT_SYMBOL(writeback_inodes_sb);
 
diff --git a/fs/fuse/control.c b/fs/fuse/control.c
index 99c99dfb0373..3773fd63d2f9 100644
--- a/fs/fuse/control.c
+++ b/fs/fuse/control.c
@@ -61,6 +61,121 @@ static ssize_t fuse_conn_waiting_read(struct file *file, char __user *buf,
 	return simple_read_from_buffer(buf, len, ppos, tmp, size);
 }
 
+static ssize_t fuse_conn_limit_read(struct file *file, char __user *buf,
+				    size_t len, loff_t *ppos, unsigned val)
+{
+	char tmp[32];
+	size_t size = sprintf(tmp, "%u\n", val);
+
+	return simple_read_from_buffer(buf, len, ppos, tmp, size);
+}
+
+static ssize_t fuse_conn_limit_write(struct file *file, const char __user *buf,
+				     size_t count, loff_t *ppos, unsigned *val,
+				     unsigned global_limit)
+{
+	unsigned long t;
+	char tmp[32];
+	unsigned limit = (1 << 16) - 1;
+	int err;
+
+	if (*ppos || count >= sizeof(tmp) - 1)
+		return -EINVAL;
+
+	if (copy_from_user(tmp, buf, count))
+		return -EINVAL;
+
+	tmp[count] = '\0';
+
+	err = strict_strtoul(tmp, 0, &t);
+	if (err)
+		return err;
+
+	if (!capable(CAP_SYS_ADMIN))
+		limit = min(limit, global_limit);
+
+	if (t > limit)
+		return -EINVAL;
+
+	*val = t;
+
+	return count;
+}
+
+static ssize_t fuse_conn_max_background_read(struct file *file,
+					     char __user *buf, size_t len,
+					     loff_t *ppos)
+{
+	struct fuse_conn *fc;
+	unsigned val;
+
+	fc = fuse_ctl_file_conn_get(file);
+	if (!fc)
+		return 0;
+
+	val = fc->max_background;
+	fuse_conn_put(fc);
+
+	return fuse_conn_limit_read(file, buf, len, ppos, val);
+}
+
+static ssize_t fuse_conn_max_background_write(struct file *file,
+					      const char __user *buf,
+					      size_t count, loff_t *ppos)
+{
+	unsigned val;
+	ssize_t ret;
+
+	ret = fuse_conn_limit_write(file, buf, count, ppos, &val,
+				    max_user_bgreq);
+	if (ret > 0) {
+		struct fuse_conn *fc = fuse_ctl_file_conn_get(file);
+		if (fc) {
+			fc->max_background = val;
+			fuse_conn_put(fc);
+		}
+	}
+
+	return ret;
+}
+
+static ssize_t fuse_conn_congestion_threshold_read(struct file *file,
+						   char __user *buf, size_t len,
+						   loff_t *ppos)
+{
+	struct fuse_conn *fc;
+	unsigned val;
+
+	fc = fuse_ctl_file_conn_get(file);
+	if (!fc)
+		return 0;
+
+	val = fc->congestion_threshold;
+	fuse_conn_put(fc);
+
+	return fuse_conn_limit_read(file, buf, len, ppos, val);
+}
+
+static ssize_t fuse_conn_congestion_threshold_write(struct file *file,
+						    const char __user *buf,
+						    size_t count, loff_t *ppos)
+{
+	unsigned val;
+	ssize_t ret;
+
+	ret = fuse_conn_limit_write(file, buf, count, ppos, &val,
+				    max_user_congthresh);
+	if (ret > 0) {
+		struct fuse_conn *fc = fuse_ctl_file_conn_get(file);
+		if (fc) {
+			fc->congestion_threshold = val;
+			fuse_conn_put(fc);
+		}
+	}
+
+	return ret;
+}
+
 static const struct file_operations fuse_ctl_abort_ops = {
 	.open = nonseekable_open,
 	.write = fuse_conn_abort_write,
@@ -71,6 +186,18 @@ static const struct file_operations fuse_ctl_waiting_ops = {
 	.read = fuse_conn_waiting_read,
 };
 
+static const struct file_operations fuse_conn_max_background_ops = {
+	.open = nonseekable_open,
+	.read = fuse_conn_max_background_read,
+	.write = fuse_conn_max_background_write,
+};
+
+static const struct file_operations fuse_conn_congestion_threshold_ops = {
+	.open = nonseekable_open,
+	.read = fuse_conn_congestion_threshold_read,
+	.write = fuse_conn_congestion_threshold_write,
+};
+
 static struct dentry *fuse_ctl_add_dentry(struct dentry *parent,
 					  struct fuse_conn *fc,
 					  const char *name,
@@ -127,9 +254,14 @@ int fuse_ctl_add_conn(struct fuse_conn *fc)
 		goto err;
 
 	if (!fuse_ctl_add_dentry(parent, fc, "waiting", S_IFREG | 0400, 1,
-				NULL, &fuse_ctl_waiting_ops) ||
+				 NULL, &fuse_ctl_waiting_ops) ||
 	    !fuse_ctl_add_dentry(parent, fc, "abort", S_IFREG | 0200, 1,
-				 NULL, &fuse_ctl_abort_ops))
+				 NULL, &fuse_ctl_abort_ops) ||
+	    !fuse_ctl_add_dentry(parent, fc, "max_background", S_IFREG | 0600,
+				 1, NULL, &fuse_conn_max_background_ops) ||
+	    !fuse_ctl_add_dentry(parent, fc, "congestion_threshold",
+				 S_IFREG | 0600, 1, NULL,
+				 &fuse_conn_congestion_threshold_ops))
 		goto err;
 
 	return 0;
@@ -156,7 +288,7 @@ void fuse_ctl_remove_conn(struct fuse_conn *fc)
 		d_drop(dentry);
 		dput(dentry);
 	}
-	fuse_control_sb->s_root->d_inode->i_nlink--;
+	drop_nlink(fuse_control_sb->s_root->d_inode);
 }
 
 static int fuse_ctl_fill_super(struct super_block *sb, void *data, int silent)
diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index 6484eb75acd6..51d9e33d634f 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -250,7 +250,7 @@ static void queue_request(struct fuse_conn *fc, struct fuse_req *req)
 
 static void flush_bg_queue(struct fuse_conn *fc)
 {
-	while (fc->active_background < FUSE_MAX_BACKGROUND &&
+	while (fc->active_background < fc->max_background &&
 	       !list_empty(&fc->bg_queue)) {
 		struct fuse_req *req;
 
@@ -280,11 +280,11 @@ __releases(&fc->lock)
 	list_del(&req->intr_entry);
 	req->state = FUSE_REQ_FINISHED;
 	if (req->background) {
-		if (fc->num_background == FUSE_MAX_BACKGROUND) {
+		if (fc->num_background == fc->max_background) {
 			fc->blocked = 0;
 			wake_up_all(&fc->blocked_waitq);
 		}
-		if (fc->num_background == FUSE_CONGESTION_THRESHOLD &&
+		if (fc->num_background == fc->congestion_threshold &&
 		    fc->connected && fc->bdi_initialized) {
 			clear_bdi_congested(&fc->bdi, BLK_RW_SYNC);
 			clear_bdi_congested(&fc->bdi, BLK_RW_ASYNC);
@@ -410,9 +410,9 @@ static void fuse_request_send_nowait_locked(struct fuse_conn *fc,
 {
 	req->background = 1;
 	fc->num_background++;
-	if (fc->num_background == FUSE_MAX_BACKGROUND)
+	if (fc->num_background == fc->max_background)
 		fc->blocked = 1;
-	if (fc->num_background == FUSE_CONGESTION_THRESHOLD &&
+	if (fc->num_background == fc->congestion_threshold &&
 	    fc->bdi_initialized) {
 		set_bdi_congested(&fc->bdi, BLK_RW_SYNC);
 		set_bdi_congested(&fc->bdi, BLK_RW_ASYNC);
diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index e703654e7f40..992f6c9410bb 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -1276,14 +1276,9 @@ static int fuse_do_setattr(struct dentry *entry, struct iattr *attr,
 		return 0;
 
 	if (attr->ia_valid & ATTR_SIZE) {
-		unsigned long limit;
-		if (IS_SWAPFILE(inode))
-			return -ETXTBSY;
-		limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
-		if (limit != RLIM_INFINITY && attr->ia_size > (loff_t) limit) {
-			send_sig(SIGXFSZ, current, 0);
-			return -EFBIG;
-		}
+		err = inode_newsize_ok(inode, attr->ia_size);
+		if (err)
+			return err;
 		is_truncate = true;
 	}
 
@@ -1350,8 +1345,7 @@ static int fuse_do_setattr(struct dentry *entry, struct iattr *attr,
 	 * FUSE_NOWRITE, otherwise fuse_launder_page() would deadlock.
 	 */
 	if (S_ISREG(inode->i_mode) && oldsize != outarg.attr.size) {
-		if (outarg.attr.size < oldsize)
-			fuse_truncate(inode->i_mapping, outarg.attr.size);
+		truncate_pagecache(inode, oldsize, outarg.attr.size);
 		invalidate_inode_pages2(inode->i_mapping);
 	}
 
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index cbc464043b6f..a3492f7d207c 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1313,7 +1313,7 @@ static int fuse_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	return 0;
 }
 
-static struct vm_operations_struct fuse_file_vm_ops = {
+static const struct vm_operations_struct fuse_file_vm_ops = {
 	.close		= fuse_vma_close,
 	.fault		= filemap_fault,
 	.page_mkwrite	= fuse_page_mkwrite,
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 52b641fc0faf..01cc462ff45d 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -25,12 +25,6 @@
 /** Max number of pages that can be used in a single read request */
 #define FUSE_MAX_PAGES_PER_REQ 32
 
-/** Maximum number of outstanding background requests */
-#define FUSE_MAX_BACKGROUND 12
-
-/** Congestion starts at 75% of maximum */
-#define FUSE_CONGESTION_THRESHOLD (FUSE_MAX_BACKGROUND * 75 / 100)
-
 /** Bias for fi->writectr, meaning new writepages must not be sent */
 #define FUSE_NOWRITE INT_MIN
 
@@ -38,7 +32,7 @@
 #define FUSE_NAME_MAX 1024
 
 /** Number of dentries for each connection in the control filesystem */
-#define FUSE_CTL_NUM_DENTRIES 3
+#define FUSE_CTL_NUM_DENTRIES 5
 
 /** If the FUSE_DEFAULT_PERMISSIONS flag is given, the filesystem
     module will check permissions based on the file mode.  Otherwise no
@@ -55,6 +49,10 @@ extern struct list_head fuse_conn_list;
 /** Global mutex protecting fuse_conn_list and the control filesystem */
 extern struct mutex fuse_mutex;
 
+/** Module parameters */
+extern unsigned max_user_bgreq;
+extern unsigned max_user_congthresh;
+
 /** FUSE inode */
 struct fuse_inode {
 	/** Inode data */
@@ -349,6 +347,12 @@ struct fuse_conn {
 	/** rbtree of fuse_files waiting for poll events indexed by ph */
 	struct rb_root polled_files;
 
+	/** Maximum number of outstanding background requests */
+	unsigned max_background;
+
+	/** Number of background requests at which congestion starts */
+	unsigned congestion_threshold;
+
 	/** Number of requests currently in the background */
 	unsigned num_background;
 
@@ -602,8 +606,6 @@ void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr,
 void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr,
 				   u64 attr_valid);
 
-void fuse_truncate(struct address_space *mapping, loff_t offset);
-
 /**
  * Initialize the client device
  */
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index e5dbecd87b0f..1a822ce2b24b 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -14,6 +14,7 @@
 #include <linux/seq_file.h>
 #include <linux/init.h>
 #include <linux/module.h>
+#include <linux/moduleparam.h>
 #include <linux/parser.h>
 #include <linux/statfs.h>
 #include <linux/random.h>
@@ -28,10 +29,34 @@ static struct kmem_cache *fuse_inode_cachep;
 struct list_head fuse_conn_list;
 DEFINE_MUTEX(fuse_mutex);
 
+static int set_global_limit(const char *val, struct kernel_param *kp);
+
+unsigned max_user_bgreq;
+module_param_call(max_user_bgreq, set_global_limit, param_get_uint,
+		  &max_user_bgreq, 0644);
+__MODULE_PARM_TYPE(max_user_bgreq, "uint");
+MODULE_PARM_DESC(max_user_bgreq,
+ "Global limit for the maximum number of backgrounded requests an "
+ "unprivileged user can set");
+
+unsigned max_user_congthresh;
+module_param_call(max_user_congthresh, set_global_limit, param_get_uint,
+		  &max_user_congthresh, 0644);
+__MODULE_PARM_TYPE(max_user_congthresh, "uint");
+MODULE_PARM_DESC(max_user_congthresh,
+ "Global limit for the maximum congestion threshold an "
+ "unprivileged user can set");
+
 #define FUSE_SUPER_MAGIC 0x65735546
 
 #define FUSE_DEFAULT_BLKSIZE 512
 
+/** Maximum number of outstanding background requests */
+#define FUSE_DEFAULT_MAX_BACKGROUND 12
+
+/** Congestion starts at 75% of maximum */
+#define FUSE_DEFAULT_CONGESTION_THRESHOLD (FUSE_DEFAULT_MAX_BACKGROUND * 3 / 4)
+
 struct fuse_mount_data {
 	int fd;
 	unsigned rootmode;
@@ -115,14 +140,6 @@ static int fuse_remount_fs(struct super_block *sb, int *flags, char *data)
 	return 0;
 }
 
-void fuse_truncate(struct address_space *mapping, loff_t offset)
-{
-	/* See vmtruncate() */
-	unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-	truncate_inode_pages(mapping, offset);
-	unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-}
-
 void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr,
 				   u64 attr_valid)
 {
@@ -180,8 +197,7 @@ void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr,
 	spin_unlock(&fc->lock);
 
 	if (S_ISREG(inode->i_mode) && oldsize != attr->size) {
-		if (attr->size < oldsize)
-			fuse_truncate(inode->i_mapping, attr->size);
+		truncate_pagecache(inode, oldsize, attr->size);
 		invalidate_inode_pages2(inode->i_mapping);
 	}
 }
@@ -517,6 +533,8 @@ void fuse_conn_init(struct fuse_conn *fc)
 	INIT_LIST_HEAD(&fc->bg_queue);
 	INIT_LIST_HEAD(&fc->entry);
 	atomic_set(&fc->num_waiting, 0);
+	fc->max_background = FUSE_DEFAULT_MAX_BACKGROUND;
+	fc->congestion_threshold = FUSE_DEFAULT_CONGESTION_THRESHOLD;
 	fc->khctr = 0;
 	fc->polled_files = RB_ROOT;
 	fc->reqctr = 0;
@@ -727,6 +745,54 @@ static const struct super_operations fuse_super_operations = {
 	.show_options	= fuse_show_options,
 };
 
+static void sanitize_global_limit(unsigned *limit)
+{
+	if (*limit == 0)
+		*limit = ((num_physpages << PAGE_SHIFT) >> 13) /
+			 sizeof(struct fuse_req);
+
+	if (*limit >= 1 << 16)
+		*limit = (1 << 16) - 1;
+}
+
+static int set_global_limit(const char *val, struct kernel_param *kp)
+{
+	int rv;
+
+	rv = param_set_uint(val, kp);
+	if (rv)
+		return rv;
+
+	sanitize_global_limit((unsigned *)kp->arg);
+
+	return 0;
+}
+
+static void process_init_limits(struct fuse_conn *fc, struct fuse_init_out *arg)
+{
+	int cap_sys_admin = capable(CAP_SYS_ADMIN);
+
+	if (arg->minor < 13)
+		return;
+
+	sanitize_global_limit(&max_user_bgreq);
+	sanitize_global_limit(&max_user_congthresh);
+
+	if (arg->max_background) {
+		fc->max_background = arg->max_background;
+
+		if (!cap_sys_admin && fc->max_background > max_user_bgreq)
+			fc->max_background = max_user_bgreq;
+	}
+	if (arg->congestion_threshold) {
+		fc->congestion_threshold = arg->congestion_threshold;
+
+		if (!cap_sys_admin &&
+		    fc->congestion_threshold > max_user_congthresh)
+			fc->congestion_threshold = max_user_congthresh;
+	}
+}
+
 static void process_init_reply(struct fuse_conn *fc, struct fuse_req *req)
 {
 	struct fuse_init_out *arg = &req->misc.init_out;
@@ -736,6 +802,8 @@ static void process_init_reply(struct fuse_conn *fc, struct fuse_req *req)
 	else {
 		unsigned long ra_pages;
 
+		process_init_limits(fc, arg);
+
 		if (arg->minor >= 6) {
 			ra_pages = arg->max_readahead / PAGE_CACHE_SIZE;
 			if (arg->flags & FUSE_ASYNC_READ)
@@ -1150,6 +1218,9 @@ static int __init fuse_init(void)
 	if (res)
 		goto err_sysfs_cleanup;
 
+	sanitize_global_limit(&max_user_bgreq);
+	sanitize_global_limit(&max_user_congthresh);
+
 	return 0;
 
  err_sysfs_cleanup:
diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c
index 7ebae9a4ecc0..694b5d48f036 100644
--- a/fs/gfs2/aops.c
+++ b/fs/gfs2/aops.c
@@ -1135,6 +1135,7 @@ static const struct address_space_operations gfs2_writeback_aops = {
 	.direct_IO = gfs2_direct_IO,
 	.migratepage = buffer_migrate_page,
 	.is_partially_uptodate = block_is_partially_uptodate,
+	.error_remove_page = generic_error_remove_page,
 };
 
 static const struct address_space_operations gfs2_ordered_aops = {
@@ -1151,6 +1152,7 @@ static const struct address_space_operations gfs2_ordered_aops = {
 	.direct_IO = gfs2_direct_IO,
 	.migratepage = buffer_migrate_page,
 	.is_partially_uptodate = block_is_partially_uptodate,
+	.error_remove_page = generic_error_remove_page,
 };
 
 static const struct address_space_operations gfs2_jdata_aops = {
@@ -1166,6 +1168,7 @@ static const struct address_space_operations gfs2_jdata_aops = {
 	.invalidatepage = gfs2_invalidatepage,
 	.releasepage = gfs2_releasepage,
 	.is_partially_uptodate = block_is_partially_uptodate,
+	.error_remove_page = generic_error_remove_page,
 };
 
 void gfs2_set_aops(struct inode *inode)
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 166f38fbd246..4eb308aa3234 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -418,7 +418,7 @@ out:
 	return ret;
 }
 
-static struct vm_operations_struct gfs2_vm_ops = {
+static const struct vm_operations_struct gfs2_vm_ops = {
 	.fault = filemap_fault,
 	.page_mkwrite = gfs2_page_mkwrite,
 };
diff --git a/fs/gfs2/ops_inode.c b/fs/gfs2/ops_inode.c
index c3ac18054057..247436c10deb 100644
--- a/fs/gfs2/ops_inode.c
+++ b/fs/gfs2/ops_inode.c
@@ -12,7 +12,6 @@
 #include <linux/completion.h>
 #include <linux/buffer_head.h>
 #include <linux/namei.h>
-#include <linux/utsname.h>
 #include <linux/mm.h>
 #include <linux/xattr.h>
 #include <linux/posix_acl.h>
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 28c590b7c9da..8f1cfb02a6cb 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -179,7 +179,7 @@ static inline u64 gfs2_bit_search(const __le64 *ptr, u64 mask, u8 state)
  * always aligned to a 64 bit boundary.
  *
  * The size of the buffer is in bytes, but is it assumed that it is
- * always ok to to read a complete multiple of 64 bits at the end
+ * always ok to read a complete multiple of 64 bits at the end
  * of the block in case the end is no aligned to a natural boundary.
  *
  * Return: the block number (bitmap buffer scope) that was found
diff --git a/fs/hfs/mdb.c b/fs/hfs/mdb.c
index 7b6165f25fbe..8bbe03c3f6d5 100644
--- a/fs/hfs/mdb.c
+++ b/fs/hfs/mdb.c
@@ -344,10 +344,8 @@ void hfs_mdb_put(struct super_block *sb)
 	brelse(HFS_SB(sb)->mdb_bh);
 	brelse(HFS_SB(sb)->alt_mdb_bh);
 
-	if (HFS_SB(sb)->nls_io)
-		unload_nls(HFS_SB(sb)->nls_io);
-	if (HFS_SB(sb)->nls_disk)
-		unload_nls(HFS_SB(sb)->nls_disk);
+	unload_nls(HFS_SB(sb)->nls_io);
+	unload_nls(HFS_SB(sb)->nls_disk);
 
 	free_pages((unsigned long)HFS_SB(sb)->bitmap, PAGE_SIZE < 8192 ? 1 : 0);
 	kfree(HFS_SB(sb));
diff --git a/fs/hfsplus/super.c b/fs/hfsplus/super.c
index c0759fe0855b..43022f3d5148 100644
--- a/fs/hfsplus/super.c
+++ b/fs/hfsplus/super.c
@@ -229,8 +229,7 @@ static void hfsplus_put_super(struct super_block *sb)
 	iput(HFSPLUS_SB(sb).alloc_file);
 	iput(HFSPLUS_SB(sb).hidden_dir);
 	brelse(HFSPLUS_SB(sb).s_vhbh);
-	if (HFSPLUS_SB(sb).nls)
-		unload_nls(HFSPLUS_SB(sb).nls);
+	unload_nls(HFSPLUS_SB(sb).nls);
 	kfree(sb->s_fs_info);
 	sb->s_fs_info = NULL;
 
@@ -464,8 +463,7 @@ out:
 
 cleanup:
 	hfsplus_put_super(sb);
-	if (nls)
-		unload_nls(nls);
+	unload_nls(nls);
 	return err;
 }
 
diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index a93b885311d8..87a1258953b8 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -31,12 +31,10 @@
 #include <linux/statfs.h>
 #include <linux/security.h>
 #include <linux/ima.h>
+#include <linux/magic.h>
 
 #include <asm/uaccess.h>
 
-/* some random number */
-#define HUGETLBFS_MAGIC	0x958458f6
-
 static const struct super_operations hugetlbfs_ops;
 static const struct address_space_operations hugetlbfs_aops;
 const struct file_operations hugetlbfs_file_operations;
@@ -382,36 +380,11 @@ static void hugetlbfs_delete_inode(struct inode *inode)
 
 static void hugetlbfs_forget_inode(struct inode *inode) __releases(inode_lock)
 {
-	struct super_block *sb = inode->i_sb;
-
-	if (!hlist_unhashed(&inode->i_hash)) {
-		if (!(inode->i_state & (I_DIRTY|I_SYNC)))
-			list_move(&inode->i_list, &inode_unused);
-		inodes_stat.nr_unused++;
-		if (!sb || (sb->s_flags & MS_ACTIVE)) {
-			spin_unlock(&inode_lock);
-			return;
-		}
-		inode->i_state |= I_WILL_FREE;
-		spin_unlock(&inode_lock);
-		/*
-		 * write_inode_now is a noop as we set BDI_CAP_NO_WRITEBACK
-		 * in our backing_dev_info.
-		 */
-		write_inode_now(inode, 1);
-		spin_lock(&inode_lock);
-		inode->i_state &= ~I_WILL_FREE;
-		inodes_stat.nr_unused--;
-		hlist_del_init(&inode->i_hash);
+	if (generic_detach_inode(inode)) {
+		truncate_hugepages(inode, 0);
+		clear_inode(inode);
+		destroy_inode(inode);
 	}
-	list_del_init(&inode->i_list);
-	list_del_init(&inode->i_sb_list);
-	inode->i_state |= I_FREEING;
-	inodes_stat.nr_inodes--;
-	spin_unlock(&inode_lock);
-	truncate_hugepages(inode, 0);
-	clear_inode(inode);
-	destroy_inode(inode);
 }
 
 static void hugetlbfs_drop_inode(struct inode *inode)
@@ -507,6 +480,13 @@ static struct inode *hugetlbfs_get_inode(struct super_block *sb, uid_t uid,
 		inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
 		INIT_LIST_HEAD(&inode->i_mapping->private_list);
 		info = HUGETLBFS_I(inode);
+		/*
+		 * The policy is initialized here even if we are creating a
+		 * private inode because initialization simply creates an
+		 * an empty rb tree and calls spin_lock_init(), later when we
+		 * call mpol_free_shared_policy() it will just return because
+		 * the rb tree will still be empty.
+		 */
 		mpol_shared_policy_init(&info->policy, NULL);
 		switch (mode & S_IFMT) {
 		default:
@@ -937,7 +917,7 @@ static int can_do_hugetlb_shm(void)
 }
 
 struct file *hugetlb_file_setup(const char *name, size_t size, int acctflag,
-						struct user_struct **user)
+				struct user_struct **user, int creat_flags)
 {
 	int error = -ENOMEM;
 	struct file *file;
@@ -949,7 +929,7 @@ struct file *hugetlb_file_setup(const char *name, size_t size, int acctflag,
 	if (!hugetlbfs_vfsmount)
 		return ERR_PTR(-ENOENT);
 
-	if (!can_do_hugetlb_shm()) {
+	if (creat_flags == HUGETLB_SHMFS_INODE && !can_do_hugetlb_shm()) {
 		*user = current_user();
 		if (user_shm_lock(size, *user)) {
 			WARN_ONCE(1,
diff --git a/fs/inode.c b/fs/inode.c
index b2ba83d2c4e1..4d8e3be55976 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -14,6 +14,7 @@
 #include <linux/module.h>
 #include <linux/backing-dev.h>
 #include <linux/wait.h>
+#include <linux/rwsem.h>
 #include <linux/hash.h>
 #include <linux/swap.h>
 #include <linux/security.h>
@@ -87,14 +88,18 @@ static struct hlist_head *inode_hashtable __read_mostly;
 DEFINE_SPINLOCK(inode_lock);
 
 /*
- * iprune_mutex provides exclusion between the kswapd or try_to_free_pages
+ * iprune_sem provides exclusion between the kswapd or try_to_free_pages
  * icache shrinking path, and the umount path.  Without this exclusion,
  * by the time prune_icache calls iput for the inode whose pages it has
  * been invalidating, or by the time it calls clear_inode & destroy_inode
  * from its final dispose_list, the struct super_block they refer to
  * (for inode->i_sb->s_op) may already have been freed and reused.
+ *
+ * We make this an rwsem because the fastpath is icache shrinking. In
+ * some cases a filesystem may be doing a significant amount of work in
+ * its inode reclaim code, so this should improve parallelism.
  */
-static DEFINE_MUTEX(iprune_mutex);
+static DECLARE_RWSEM(iprune_sem);
 
 /*
  * Statistics gathering..
@@ -123,7 +128,7 @@ static void wake_up_inode(struct inode *inode)
 int inode_init_always(struct super_block *sb, struct inode *inode)
 {
 	static const struct address_space_operations empty_aops;
-	static struct inode_operations empty_iops;
+	static const struct inode_operations empty_iops;
 	static const struct file_operations empty_fops;
 	struct address_space *const mapping = &inode->i_data;
 
@@ -381,7 +386,7 @@ static int invalidate_list(struct list_head *head, struct list_head *dispose)
 		/*
 		 * We can reschedule here without worrying about the list's
 		 * consistency because the per-sb list of inodes must not
-		 * change during umount anymore, and because iprune_mutex keeps
+		 * change during umount anymore, and because iprune_sem keeps
 		 * shrink_icache_memory() away.
 		 */
 		cond_resched_lock(&inode_lock);
@@ -420,7 +425,7 @@ int invalidate_inodes(struct super_block *sb)
 	int busy;
 	LIST_HEAD(throw_away);
 
-	mutex_lock(&iprune_mutex);
+	down_write(&iprune_sem);
 	spin_lock(&inode_lock);
 	inotify_unmount_inodes(&sb->s_inodes);
 	fsnotify_unmount_inodes(&sb->s_inodes);
@@ -428,7 +433,7 @@ int invalidate_inodes(struct super_block *sb)
 	spin_unlock(&inode_lock);
 
 	dispose_list(&throw_away);
-	mutex_unlock(&iprune_mutex);
+	up_write(&iprune_sem);
 
 	return busy;
 }
@@ -467,7 +472,7 @@ static void prune_icache(int nr_to_scan)
 	int nr_scanned;
 	unsigned long reap = 0;
 
-	mutex_lock(&iprune_mutex);
+	down_read(&iprune_sem);
 	spin_lock(&inode_lock);
 	for (nr_scanned = 0; nr_scanned < nr_to_scan; nr_scanned++) {
 		struct inode *inode;
@@ -509,7 +514,7 @@ static void prune_icache(int nr_to_scan)
 	spin_unlock(&inode_lock);
 
 	dispose_list(&freeable);
-	mutex_unlock(&iprune_mutex);
+	up_read(&iprune_sem);
 }
 
 /*
@@ -695,13 +700,15 @@ void unlock_new_inode(struct inode *inode)
 	}
 #endif
 	/*
-	 * This is special!  We do not need the spinlock
-	 * when clearing I_LOCK, because we're guaranteed
-	 * that nobody else tries to do anything about the
-	 * state of the inode when it is locked, as we
-	 * just created it (so there can be no old holders
-	 * that haven't tested I_LOCK).
+	 * This is special!  We do not need the spinlock when clearing I_LOCK,
+	 * because we're guaranteed that nobody else tries to do anything about
+	 * the state of the inode when it is locked, as we just created it (so
+	 * there can be no old holders that haven't tested I_LOCK).
+	 * However we must emit the memory barrier so that other CPUs reliably
+	 * see the clearing of I_LOCK after the other inode initialisation has
+	 * completed.
 	 */
+	smp_mb();
 	WARN_ON((inode->i_state & (I_LOCK|I_NEW)) != (I_LOCK|I_NEW));
 	inode->i_state &= ~(I_LOCK|I_NEW);
 	wake_up_inode(inode);
@@ -1234,7 +1241,16 @@ void generic_delete_inode(struct inode *inode)
 }
 EXPORT_SYMBOL(generic_delete_inode);
 
-static void generic_forget_inode(struct inode *inode)
+/**
+ *	generic_detach_inode - remove inode from inode lists
+ *	@inode: inode to remove
+ *
+ *	Remove inode from inode lists, write it if it's dirty. This is just an
+ *	internal VFS helper exported for hugetlbfs. Do not use!
+ *
+ *	Returns 1 if inode should be completely destroyed.
+ */
+int generic_detach_inode(struct inode *inode)
 {
 	struct super_block *sb = inode->i_sb;
 
@@ -1244,7 +1260,7 @@ static void generic_forget_inode(struct inode *inode)
 		inodes_stat.nr_unused++;
 		if (sb->s_flags & MS_ACTIVE) {
 			spin_unlock(&inode_lock);
-			return;
+			return 0;
 		}
 		WARN_ON(inode->i_state & I_NEW);
 		inode->i_state |= I_WILL_FREE;
@@ -1262,6 +1278,14 @@ static void generic_forget_inode(struct inode *inode)
 	inode->i_state |= I_FREEING;
 	inodes_stat.nr_inodes--;
 	spin_unlock(&inode_lock);
+	return 1;
+}
+EXPORT_SYMBOL_GPL(generic_detach_inode);
+
+static void generic_forget_inode(struct inode *inode)
+{
+	if (!generic_detach_inode(inode))
+		return;
 	if (inode->i_data.nrpages)
 		truncate_inode_pages(&inode->i_data, 0);
 	clear_inode(inode);
@@ -1392,31 +1416,31 @@ void touch_atime(struct vfsmount *mnt, struct dentry *dentry)
 	struct inode *inode = dentry->d_inode;
 	struct timespec now;
 
-	if (mnt_want_write(mnt))
-		return;
 	if (inode->i_flags & S_NOATIME)
-		goto out;
+		return;
 	if (IS_NOATIME(inode))
-		goto out;
+		return;
 	if ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode))
-		goto out;
+		return;
 
 	if (mnt->mnt_flags & MNT_NOATIME)
-		goto out;
+		return;
 	if ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode))
-		goto out;
+		return;
 
 	now = current_fs_time(inode->i_sb);
 
 	if (!relatime_need_update(mnt, inode, now))
-		goto out;
+		return;
 
 	if (timespec_equal(&inode->i_atime, &now))
-		goto out;
+		return;
+
+	if (mnt_want_write(mnt))
+		return;
 
 	inode->i_atime = now;
 	mark_inode_dirty_sync(inode);
-out:
 	mnt_drop_write(mnt);
 }
 EXPORT_SYMBOL(touch_atime);
@@ -1437,34 +1461,37 @@ void file_update_time(struct file *file)
 {
 	struct inode *inode = file->f_path.dentry->d_inode;
 	struct timespec now;
-	int sync_it = 0;
-	int err;
+	enum { S_MTIME = 1, S_CTIME = 2, S_VERSION = 4 } sync_it = 0;
 
+	/* First try to exhaust all avenues to not sync */
 	if (IS_NOCMTIME(inode))
 		return;
 
-	err = mnt_want_write_file(file);
-	if (err)
-		return;
-
 	now = current_fs_time(inode->i_sb);
-	if (!timespec_equal(&inode->i_mtime, &now)) {
-		inode->i_mtime = now;
-		sync_it = 1;
-	}
+	if (!timespec_equal(&inode->i_mtime, &now))
+		sync_it = S_MTIME;
 
-	if (!timespec_equal(&inode->i_ctime, &now)) {
-		inode->i_ctime = now;
-		sync_it = 1;
-	}
+	if (!timespec_equal(&inode->i_ctime, &now))
+		sync_it |= S_CTIME;
 
-	if (IS_I_VERSION(inode)) {
-		inode_inc_iversion(inode);
-		sync_it = 1;
-	}
+	if (IS_I_VERSION(inode))
+		sync_it |= S_VERSION;
+
+	if (!sync_it)
+		return;
 
-	if (sync_it)
-		mark_inode_dirty_sync(inode);
+	/* Finally allowed to write? Takes lock. */
+	if (mnt_want_write_file(file))
+		return;
+
+	/* Only change inode inside the lock region */
+	if (sync_it & S_VERSION)
+		inode_inc_iversion(inode);
+	if (sync_it & S_CTIME)
+		inode->i_ctime = now;
+	if (sync_it & S_MTIME)
+		inode->i_mtime = now;
+	mark_inode_dirty_sync(inode);
 	mnt_drop_write(file->f_path.mnt);
 }
 EXPORT_SYMBOL(file_update_time);
@@ -1592,7 +1619,8 @@ void init_special_inode(struct inode *inode, umode_t mode, dev_t rdev)
 	else if (S_ISSOCK(mode))
 		inode->i_fop = &bad_sock_fops;
 	else
-		printk(KERN_DEBUG "init_special_inode: bogus i_mode (%o)\n",
-		       mode);
+		printk(KERN_DEBUG "init_special_inode: bogus i_mode (%o) for"
+				  " inode %s:%lu\n", mode, inode->i_sb->s_id,
+				  inode->i_ino);
 }
 EXPORT_SYMBOL(init_special_inode);
diff --git a/fs/internal.h b/fs/internal.h
index d55ef562f0bb..515175b8b72e 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -57,6 +57,7 @@ extern int check_unsafe_exec(struct linux_binprm *);
  * namespace.c
  */
 extern int copy_mount_options(const void __user *, unsigned long *);
+extern int copy_mount_string(const void __user *, char **);
 
 extern void free_vfsmnt(struct vfsmount *);
 extern struct vfsmount *alloc_vfsmnt(const char *);
diff --git a/fs/ioctl.c b/fs/ioctl.c
index 5612880fcbe7..7b17a14396ff 100644
--- a/fs/ioctl.c
+++ b/fs/ioctl.c
@@ -162,20 +162,21 @@ EXPORT_SYMBOL(fiemap_check_flags);
 static int fiemap_check_ranges(struct super_block *sb,
 			       u64 start, u64 len, u64 *new_len)
 {
+	u64 maxbytes = (u64) sb->s_maxbytes;
+
 	*new_len = len;
 
 	if (len == 0)
 		return -EINVAL;
 
-	if (start > sb->s_maxbytes)
+	if (start > maxbytes)
 		return -EFBIG;
 
 	/*
 	 * Shrink request scope to what the fs can actually handle.
 	 */
-	if ((len > sb->s_maxbytes) ||
-	    (sb->s_maxbytes - len) < start)
-		*new_len = sb->s_maxbytes - start;
+	if (len > maxbytes || (maxbytes - len) < start)
+		*new_len = maxbytes - start;
 
 	return 0;
 }
diff --git a/fs/isofs/inode.c b/fs/isofs/inode.c
index 85f96bc651c7..6b4dcd4f2943 100644
--- a/fs/isofs/inode.c
+++ b/fs/isofs/inode.c
@@ -46,10 +46,7 @@ static void isofs_put_super(struct super_block *sb)
 #ifdef CONFIG_JOLIET
 	lock_kernel();
 
-	if (sbi->s_nls_iocharset) {
-		unload_nls(sbi->s_nls_iocharset);
-		sbi->s_nls_iocharset = NULL;
-	}
+	unload_nls(sbi->s_nls_iocharset);
 
 	unlock_kernel();
 #endif
@@ -912,8 +909,7 @@ out_no_root:
 		printk(KERN_WARNING "%s: get root inode failed\n", __func__);
 out_no_inode:
 #ifdef CONFIG_JOLIET
-	if (sbi->s_nls_iocharset)
-		unload_nls(sbi->s_nls_iocharset);
+	unload_nls(sbi->s_nls_iocharset);
 #endif
 	goto out_freesbi;
 out_no_read:
diff --git a/fs/jbd/checkpoint.c b/fs/jbd/checkpoint.c
index 61f32f3868cd..b0435dd0654d 100644
--- a/fs/jbd/checkpoint.c
+++ b/fs/jbd/checkpoint.c
@@ -456,7 +456,7 @@ int cleanup_journal_tail(journal_t *journal)
 {
 	transaction_t * transaction;
 	tid_t		first_tid;
-	unsigned long	blocknr, freed;
+	unsigned int	blocknr, freed;
 
 	if (is_journal_aborted(journal))
 		return 1;
@@ -502,8 +502,8 @@ int cleanup_journal_tail(journal_t *journal)
 		freed = freed + journal->j_last - journal->j_first;
 
 	jbd_debug(1,
-		  "Cleaning journal tail from %d to %d (offset %lu), "
-		  "freeing %lu\n",
+		  "Cleaning journal tail from %d to %d (offset %u), "
+		  "freeing %u\n",
 		  journal->j_tail_sequence, first_tid, blocknr, freed);
 
 	journal->j_free += freed;
diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c
index 618e21c0b7a3..4bd882548c45 100644
--- a/fs/jbd/commit.c
+++ b/fs/jbd/commit.c
@@ -308,7 +308,7 @@ void journal_commit_transaction(journal_t *journal)
 	int bufs;
 	int flags;
 	int err;
-	unsigned long blocknr;
+	unsigned int blocknr;
 	ktime_t start_time;
 	u64 commit_time;
 	char *tagp = NULL;
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index f96f85092d1c..bd3c073b485d 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -276,7 +276,7 @@ static void journal_kill_thread(journal_t *journal)
 int journal_write_metadata_buffer(transaction_t *transaction,
 				  struct journal_head  *jh_in,
 				  struct journal_head **jh_out,
-				  unsigned long blocknr)
+				  unsigned int blocknr)
 {
 	int need_copy_out = 0;
 	int done_copy_out = 0;
@@ -567,9 +567,9 @@ int log_wait_commit(journal_t *journal, tid_t tid)
  * Log buffer allocation routines:
  */
 
-int journal_next_log_block(journal_t *journal, unsigned long *retp)
+int journal_next_log_block(journal_t *journal, unsigned int *retp)
 {
-	unsigned long blocknr;
+	unsigned int blocknr;
 
 	spin_lock(&journal->j_state_lock);
 	J_ASSERT(journal->j_free > 1);
@@ -590,11 +590,11 @@ int journal_next_log_block(journal_t *journal, unsigned long *retp)
  * this is a no-op.  If needed, we can use j_blk_offset - everything is
  * ready.
  */
-int journal_bmap(journal_t *journal, unsigned long blocknr,
-		 unsigned long *retp)
+int journal_bmap(journal_t *journal, unsigned int blocknr,
+		 unsigned int *retp)
 {
 	int err = 0;
-	unsigned long ret;
+	unsigned int ret;
 
 	if (journal->j_inode) {
 		ret = bmap(journal->j_inode, blocknr);
@@ -604,7 +604,7 @@ int journal_bmap(journal_t *journal, unsigned long blocknr,
 			char b[BDEVNAME_SIZE];
 
 			printk(KERN_ALERT "%s: journal block not found "
-					"at offset %lu on %s\n",
+					"at offset %u on %s\n",
 				__func__,
 				blocknr,
 				bdevname(journal->j_dev, b));
@@ -630,7 +630,7 @@ int journal_bmap(journal_t *journal, unsigned long blocknr,
 struct journal_head *journal_get_descriptor_buffer(journal_t *journal)
 {
 	struct buffer_head *bh;
-	unsigned long blocknr;
+	unsigned int blocknr;
 	int err;
 
 	err = journal_next_log_block(journal, &blocknr);
@@ -774,7 +774,7 @@ journal_t * journal_init_inode (struct inode *inode)
 	journal_t *journal = journal_init_common();
 	int err;
 	int n;
-	unsigned long blocknr;
+	unsigned int blocknr;
 
 	if (!journal)
 		return NULL;
@@ -846,12 +846,12 @@ static void journal_fail_superblock (journal_t *journal)
 static int journal_reset(journal_t *journal)
 {
 	journal_superblock_t *sb = journal->j_superblock;
-	unsigned long first, last;
+	unsigned int first, last;
 
 	first = be32_to_cpu(sb->s_first);
 	last = be32_to_cpu(sb->s_maxlen);
 	if (first + JFS_MIN_JOURNAL_BLOCKS > last + 1) {
-		printk(KERN_ERR "JBD: Journal too short (blocks %lu-%lu).\n",
+		printk(KERN_ERR "JBD: Journal too short (blocks %u-%u).\n",
 		       first, last);
 		journal_fail_superblock(journal);
 		return -EINVAL;
@@ -885,7 +885,7 @@ static int journal_reset(journal_t *journal)
  **/
 int journal_create(journal_t *journal)
 {
-	unsigned long blocknr;
+	unsigned int blocknr;
 	struct buffer_head *bh;
 	journal_superblock_t *sb;
 	int i, err;
@@ -969,14 +969,14 @@ void journal_update_superblock(journal_t *journal, int wait)
 	if (sb->s_start == 0 && journal->j_tail_sequence ==
 				journal->j_transaction_sequence) {
 		jbd_debug(1,"JBD: Skipping superblock update on recovered sb "
-			"(start %ld, seq %d, errno %d)\n",
+			"(start %u, seq %d, errno %d)\n",
 			journal->j_tail, journal->j_tail_sequence,
 			journal->j_errno);
 		goto out;
 	}
 
 	spin_lock(&journal->j_state_lock);
-	jbd_debug(1,"JBD: updating superblock (start %ld, seq %d, errno %d)\n",
+	jbd_debug(1,"JBD: updating superblock (start %u, seq %d, errno %d)\n",
 		  journal->j_tail, journal->j_tail_sequence, journal->j_errno);
 
 	sb->s_sequence = cpu_to_be32(journal->j_tail_sequence);
@@ -1371,7 +1371,7 @@ int journal_flush(journal_t *journal)
 {
 	int err = 0;
 	transaction_t *transaction = NULL;
-	unsigned long old_tail;
+	unsigned int old_tail;
 
 	spin_lock(&journal->j_state_lock);
 
diff --git a/fs/jbd/recovery.c b/fs/jbd/recovery.c
index db5e982c5ddf..cb1a49ae605e 100644
--- a/fs/jbd/recovery.c
+++ b/fs/jbd/recovery.c
@@ -70,7 +70,7 @@ static int do_readahead(journal_t *journal, unsigned int start)
 {
 	int err;
 	unsigned int max, nbufs, next;
-	unsigned long blocknr;
+	unsigned int blocknr;
 	struct buffer_head *bh;
 
 	struct buffer_head * bufs[MAXBUF];
@@ -132,7 +132,7 @@ static int jread(struct buffer_head **bhp, journal_t *journal,
 		 unsigned int offset)
 {
 	int err;
-	unsigned long blocknr;
+	unsigned int blocknr;
 	struct buffer_head *bh;
 
 	*bhp = NULL;
@@ -314,7 +314,7 @@ static int do_one_pass(journal_t *journal,
 			struct recovery_info *info, enum passtype pass)
 {
 	unsigned int		first_commit_ID, next_commit_ID;
-	unsigned long		next_log_block;
+	unsigned int		next_log_block;
 	int			err, success = 0;
 	journal_superblock_t *	sb;
 	journal_header_t *	tmp;
@@ -367,14 +367,14 @@ static int do_one_pass(journal_t *journal,
 			if (tid_geq(next_commit_ID, info->end_transaction))
 				break;
 
-		jbd_debug(2, "Scanning for sequence ID %u at %lu/%lu\n",
+		jbd_debug(2, "Scanning for sequence ID %u at %u/%u\n",
 			  next_commit_ID, next_log_block, journal->j_last);
 
 		/* Skip over each chunk of the transaction looking
 		 * either the next descriptor block or the final commit
 		 * record. */
 
-		jbd_debug(3, "JBD: checking block %ld\n", next_log_block);
+		jbd_debug(3, "JBD: checking block %u\n", next_log_block);
 		err = jread(&bh, journal, next_log_block);
 		if (err)
 			goto failed;
@@ -429,7 +429,7 @@ static int do_one_pass(journal_t *journal,
 			tagp = &bh->b_data[sizeof(journal_header_t)];
 			while ((tagp - bh->b_data +sizeof(journal_block_tag_t))
 			       <= journal->j_blocksize) {
-				unsigned long io_block;
+				unsigned int io_block;
 
 				tag = (journal_block_tag_t *) tagp;
 				flags = be32_to_cpu(tag->t_flags);
@@ -443,10 +443,10 @@ static int do_one_pass(journal_t *journal,
 					success = err;
 					printk (KERN_ERR
 						"JBD: IO error %d recovering "
-						"block %ld in log\n",
+						"block %u in log\n",
 						err, io_block);
 				} else {
-					unsigned long blocknr;
+					unsigned int blocknr;
 
 					J_ASSERT(obh != NULL);
 					blocknr = be32_to_cpu(tag->t_blocknr);
@@ -581,7 +581,7 @@ static int scan_revoke_records(journal_t *journal, struct buffer_head *bh,
 	max = be32_to_cpu(header->r_count);
 
 	while (offset < max) {
-		unsigned long blocknr;
+		unsigned int blocknr;
 		int err;
 
 		blocknr = be32_to_cpu(* ((__be32 *) (bh->b_data+offset)));
diff --git a/fs/jbd/revoke.c b/fs/jbd/revoke.c
index da6cd9bdaabc..ad717328343a 100644
--- a/fs/jbd/revoke.c
+++ b/fs/jbd/revoke.c
@@ -101,7 +101,7 @@ struct jbd_revoke_record_s
 {
 	struct list_head  hash;
 	tid_t		  sequence;	/* Used for recovery only */
-	unsigned long	  blocknr;
+	unsigned int	  blocknr;
 };
 
 
@@ -126,7 +126,7 @@ static void flush_descriptor(journal_t *, struct journal_head *, int, int);
 /* Utility functions to maintain the revoke table */
 
 /* Borrowed from buffer.c: this is a tried and tested block hash function */
-static inline int hash(journal_t *journal, unsigned long block)
+static inline int hash(journal_t *journal, unsigned int block)
 {
 	struct jbd_revoke_table_s *table = journal->j_revoke;
 	int hash_shift = table->hash_shift;
@@ -136,7 +136,7 @@ static inline int hash(journal_t *journal, unsigned long block)
 		(block << (hash_shift - 12))) & (table->hash_size - 1);
 }
 
-static int insert_revoke_hash(journal_t *journal, unsigned long blocknr,
+static int insert_revoke_hash(journal_t *journal, unsigned int blocknr,
 			      tid_t seq)
 {
 	struct list_head *hash_list;
@@ -166,7 +166,7 @@ oom:
 /* Find a revoke record in the journal's hash table. */
 
 static struct jbd_revoke_record_s *find_revoke_record(journal_t *journal,
-						      unsigned long blocknr)
+						      unsigned int blocknr)
 {
 	struct list_head *hash_list;
 	struct jbd_revoke_record_s *record;
@@ -332,7 +332,7 @@ void journal_destroy_revoke(journal_t *journal)
  * by one.
  */
 
-int journal_revoke(handle_t *handle, unsigned long blocknr,
+int journal_revoke(handle_t *handle, unsigned int blocknr,
 		   struct buffer_head *bh_in)
 {
 	struct buffer_head *bh = NULL;
@@ -401,7 +401,7 @@ int journal_revoke(handle_t *handle, unsigned long blocknr,
 		}
 	}
 
-	jbd_debug(2, "insert revoke for block %lu, bh_in=%p\n", blocknr, bh_in);
+	jbd_debug(2, "insert revoke for block %u, bh_in=%p\n", blocknr, bh_in);
 	err = insert_revoke_hash(journal, blocknr,
 				handle->h_transaction->t_tid);
 	BUFFER_TRACE(bh_in, "exit");
@@ -644,7 +644,7 @@ static void flush_descriptor(journal_t *journal,
  */
 
 int journal_set_revoke(journal_t *journal,
-		       unsigned long blocknr,
+		       unsigned int blocknr,
 		       tid_t sequence)
 {
 	struct jbd_revoke_record_s *record;
@@ -668,7 +668,7 @@ int journal_set_revoke(journal_t *journal,
  */
 
 int journal_test_revoke(journal_t *journal,
-			unsigned long blocknr,
+			unsigned int blocknr,
 			tid_t sequence)
 {
 	struct jbd_revoke_record_s *record;
diff --git a/fs/jbd/transaction.c b/fs/jbd/transaction.c
index c03ac11f74be..006f9ad838a2 100644
--- a/fs/jbd/transaction.c
+++ b/fs/jbd/transaction.c
@@ -56,7 +56,8 @@ get_transaction(journal_t *journal, transaction_t *transaction)
 	spin_lock_init(&transaction->t_handle_lock);
 
 	/* Set up the commit timer for the new transaction. */
-	journal->j_commit_timer.expires = round_jiffies(transaction->t_expires);
+	journal->j_commit_timer.expires =
+				round_jiffies_up(transaction->t_expires);
 	add_timer(&journal->j_commit_timer);
 
 	J_ASSERT(journal->j_running_transaction == NULL);
@@ -228,6 +229,8 @@ repeat_locked:
 		  __log_space_left(journal));
 	spin_unlock(&transaction->t_handle_lock);
 	spin_unlock(&journal->j_state_lock);
+
+	lock_map_acquire(&handle->h_lockdep_map);
 out:
 	if (unlikely(new_transaction))		/* It's usually NULL */
 		kfree(new_transaction);
@@ -292,9 +295,6 @@ handle_t *journal_start(journal_t *journal, int nblocks)
 		handle = ERR_PTR(err);
 		goto out;
 	}
-
-	lock_map_acquire(&handle->h_lockdep_map);
-
 out:
 	return handle;
 }
@@ -416,6 +416,7 @@ int journal_restart(handle_t *handle, int nblocks)
 	__log_start_commit(journal, transaction->t_tid);
 	spin_unlock(&journal->j_state_lock);
 
+	lock_map_release(&handle->h_lockdep_map);
 	handle->h_buffer_credits = nblocks;
 	ret = start_this_handle(journal, handle);
 	return ret;
diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c
index 5d70b3e6d49b..ca0f5eb62b20 100644
--- a/fs/jbd2/checkpoint.c
+++ b/fs/jbd2/checkpoint.c
@@ -643,6 +643,7 @@ out:
 
 int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
 {
+	struct transaction_chp_stats_s *stats;
 	transaction_t *transaction;
 	journal_t *journal;
 	int ret = 0;
@@ -679,6 +680,12 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
 
 	/* OK, that was the last buffer for the transaction: we can now
 	   safely remove this transaction from the log */
+	stats = &transaction->t_chp_stats;
+	if (stats->cs_chp_time)
+		stats->cs_chp_time = jbd2_time_diff(stats->cs_chp_time,
+						    jiffies);
+	trace_jbd2_checkpoint_stats(journal->j_fs_dev->bd_dev,
+				    transaction->t_tid, stats);
 
 	__jbd2_journal_drop_transaction(journal, transaction);
 	kfree(transaction);
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 0df600e9162d..d4cfd6d2779e 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -25,6 +25,7 @@
 #include <linux/writeback.h>
 #include <linux/backing-dev.h>
 #include <linux/bio.h>
+#include <linux/blkdev.h>
 #include <trace/events/jbd2.h>
 
 /*
@@ -133,8 +134,8 @@ static int journal_submit_commit_record(journal_t *journal,
 	bh->b_end_io = journal_end_buffer_io_sync;
 
 	if (journal->j_flags & JBD2_BARRIER &&
-		!JBD2_HAS_INCOMPAT_FEATURE(journal,
-					 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+	    !JBD2_HAS_INCOMPAT_FEATURE(journal,
+				       JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
 		set_buffer_ordered(bh);
 		barrier_done = 1;
 	}
@@ -409,10 +410,10 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	if (commit_transaction->t_synchronous_commit)
 		write_op = WRITE_SYNC_PLUG;
 	trace_jbd2_commit_locking(journal, commit_transaction);
-	stats.u.run.rs_wait = commit_transaction->t_max_wait;
-	stats.u.run.rs_locked = jiffies;
-	stats.u.run.rs_running = jbd2_time_diff(commit_transaction->t_start,
-						stats.u.run.rs_locked);
+	stats.run.rs_wait = commit_transaction->t_max_wait;
+	stats.run.rs_locked = jiffies;
+	stats.run.rs_running = jbd2_time_diff(commit_transaction->t_start,
+					      stats.run.rs_locked);
 
 	spin_lock(&commit_transaction->t_handle_lock);
 	while (commit_transaction->t_updates) {
@@ -485,9 +486,9 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	jbd2_journal_switch_revoke_table(journal);
 
 	trace_jbd2_commit_flushing(journal, commit_transaction);
-	stats.u.run.rs_flushing = jiffies;
-	stats.u.run.rs_locked = jbd2_time_diff(stats.u.run.rs_locked,
-					       stats.u.run.rs_flushing);
+	stats.run.rs_flushing = jiffies;
+	stats.run.rs_locked = jbd2_time_diff(stats.run.rs_locked,
+					     stats.run.rs_flushing);
 
 	commit_transaction->t_state = T_FLUSH;
 	journal->j_committing_transaction = commit_transaction;
@@ -522,11 +523,11 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	spin_unlock(&journal->j_state_lock);
 
 	trace_jbd2_commit_logging(journal, commit_transaction);
-	stats.u.run.rs_logging = jiffies;
-	stats.u.run.rs_flushing = jbd2_time_diff(stats.u.run.rs_flushing,
-						 stats.u.run.rs_logging);
-	stats.u.run.rs_blocks = commit_transaction->t_outstanding_credits;
-	stats.u.run.rs_blocks_logged = 0;
+	stats.run.rs_logging = jiffies;
+	stats.run.rs_flushing = jbd2_time_diff(stats.run.rs_flushing,
+					       stats.run.rs_logging);
+	stats.run.rs_blocks = commit_transaction->t_outstanding_credits;
+	stats.run.rs_blocks_logged = 0;
 
 	J_ASSERT(commit_transaction->t_nr_buffers <=
 		 commit_transaction->t_outstanding_credits);
@@ -694,7 +695,7 @@ start_journal_io:
 				submit_bh(write_op, bh);
 			}
 			cond_resched();
-			stats.u.run.rs_blocks_logged += bufs;
+			stats.run.rs_blocks_logged += bufs;
 
 			/* Force a new descriptor to be generated next
                            time round the loop. */
@@ -706,11 +707,13 @@ start_journal_io:
 	/* Done it all: now write the commit record asynchronously. */
 
 	if (JBD2_HAS_INCOMPAT_FEATURE(journal,
-		JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+				      JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
 		err = journal_submit_commit_record(journal, commit_transaction,
 						 &cbh, crc32_sum);
 		if (err)
 			__jbd2_journal_abort_hard(journal);
+		if (journal->j_flags & JBD2_BARRIER)
+			blkdev_issue_flush(journal->j_dev, NULL);
 	}
 
 	/*
@@ -833,7 +836,7 @@ wait_for_iobuf:
 	jbd_debug(3, "JBD: commit phase 5\n");
 
 	if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
-		JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+				       JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
 		err = journal_submit_commit_record(journal, commit_transaction,
 						&cbh, crc32_sum);
 		if (err)
@@ -985,33 +988,30 @@ restart_loop:
 	J_ASSERT(commit_transaction->t_state == T_COMMIT);
 
 	commit_transaction->t_start = jiffies;
-	stats.u.run.rs_logging = jbd2_time_diff(stats.u.run.rs_logging,
-						commit_transaction->t_start);
+	stats.run.rs_logging = jbd2_time_diff(stats.run.rs_logging,
+					      commit_transaction->t_start);
 
 	/*
-	 * File the transaction for history
+	 * File the transaction statistics
 	 */
-	stats.ts_type = JBD2_STATS_RUN;
 	stats.ts_tid = commit_transaction->t_tid;
-	stats.u.run.rs_handle_count = commit_transaction->t_handle_count;
-	spin_lock(&journal->j_history_lock);
-	memcpy(journal->j_history + journal->j_history_cur, &stats,
-			sizeof(stats));
-	if (++journal->j_history_cur == journal->j_history_max)
-		journal->j_history_cur = 0;
+	stats.run.rs_handle_count = commit_transaction->t_handle_count;
+	trace_jbd2_run_stats(journal->j_fs_dev->bd_dev,
+			     commit_transaction->t_tid, &stats.run);
 
 	/*
 	 * Calculate overall stats
 	 */
+	spin_lock(&journal->j_history_lock);
 	journal->j_stats.ts_tid++;
-	journal->j_stats.u.run.rs_wait += stats.u.run.rs_wait;
-	journal->j_stats.u.run.rs_running += stats.u.run.rs_running;
-	journal->j_stats.u.run.rs_locked += stats.u.run.rs_locked;
-	journal->j_stats.u.run.rs_flushing += stats.u.run.rs_flushing;
-	journal->j_stats.u.run.rs_logging += stats.u.run.rs_logging;
-	journal->j_stats.u.run.rs_handle_count += stats.u.run.rs_handle_count;
-	journal->j_stats.u.run.rs_blocks += stats.u.run.rs_blocks;
-	journal->j_stats.u.run.rs_blocks_logged += stats.u.run.rs_blocks_logged;
+	journal->j_stats.run.rs_wait += stats.run.rs_wait;
+	journal->j_stats.run.rs_running += stats.run.rs_running;
+	journal->j_stats.run.rs_locked += stats.run.rs_locked;
+	journal->j_stats.run.rs_flushing += stats.run.rs_flushing;
+	journal->j_stats.run.rs_logging += stats.run.rs_logging;
+	journal->j_stats.run.rs_handle_count += stats.run.rs_handle_count;
+	journal->j_stats.run.rs_blocks += stats.run.rs_blocks;
+	journal->j_stats.run.rs_blocks_logged += stats.run.rs_blocks_logged;
 	spin_unlock(&journal->j_history_lock);
 
 	commit_transaction->t_state = T_FINISHED;
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index e378cb383979..b0ab5219becb 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -136,10 +136,6 @@ static int kjournald2(void *arg)
 	journal->j_task = current;
 	wake_up(&journal->j_wait_done_commit);
 
-	printk(KERN_INFO "kjournald2 starting: pid %d, dev %s, "
-	       "commit interval %ld seconds\n", current->pid,
-	       journal->j_devname, journal->j_commit_interval / HZ);
-
 	/*
 	 * And now, wait forever for commit wakeup events.
 	 */
@@ -223,7 +219,8 @@ static int jbd2_journal_start_thread(journal_t *journal)
 {
 	struct task_struct *t;
 
-	t = kthread_run(kjournald2, journal, "kjournald2");
+	t = kthread_run(kjournald2, journal, "jbd2/%s",
+			journal->j_devname);
 	if (IS_ERR(t))
 		return PTR_ERR(t);
 
@@ -679,153 +676,6 @@ struct jbd2_stats_proc_session {
 	int max;
 };
 
-static void *jbd2_history_skip_empty(struct jbd2_stats_proc_session *s,
-					struct transaction_stats_s *ts,
-					int first)
-{
-	if (ts == s->stats + s->max)
-		ts = s->stats;
-	if (!first && ts == s->stats + s->start)
-		return NULL;
-	while (ts->ts_type == 0) {
-		ts++;
-		if (ts == s->stats + s->max)
-			ts = s->stats;
-		if (ts == s->stats + s->start)
-			return NULL;
-	}
-	return ts;
-
-}
-
-static void *jbd2_seq_history_start(struct seq_file *seq, loff_t *pos)
-{
-	struct jbd2_stats_proc_session *s = seq->private;
-	struct transaction_stats_s *ts;
-	int l = *pos;
-
-	if (l == 0)
-		return SEQ_START_TOKEN;
-	ts = jbd2_history_skip_empty(s, s->stats + s->start, 1);
-	if (!ts)
-		return NULL;
-	l--;
-	while (l) {
-		ts = jbd2_history_skip_empty(s, ++ts, 0);
-		if (!ts)
-			break;
-		l--;
-	}
-	return ts;
-}
-
-static void *jbd2_seq_history_next(struct seq_file *seq, void *v, loff_t *pos)
-{
-	struct jbd2_stats_proc_session *s = seq->private;
-	struct transaction_stats_s *ts = v;
-
-	++*pos;
-	if (v == SEQ_START_TOKEN)
-		return jbd2_history_skip_empty(s, s->stats + s->start, 1);
-	else
-		return jbd2_history_skip_empty(s, ++ts, 0);
-}
-
-static int jbd2_seq_history_show(struct seq_file *seq, void *v)
-{
-	struct transaction_stats_s *ts = v;
-	if (v == SEQ_START_TOKEN) {
-		seq_printf(seq, "%-4s %-5s %-5s %-5s %-5s %-5s %-5s %-6s %-5s "
-				"%-5s %-5s %-5s %-5s %-5s\n", "R/C", "tid",
-				"wait", "run", "lock", "flush", "log", "hndls",
-				"block", "inlog", "ctime", "write", "drop",
-				"close");
-		return 0;
-	}
-	if (ts->ts_type == JBD2_STATS_RUN)
-		seq_printf(seq, "%-4s %-5lu %-5u %-5u %-5u %-5u %-5u "
-				"%-6lu %-5lu %-5lu\n", "R", ts->ts_tid,
-				jiffies_to_msecs(ts->u.run.rs_wait),
-				jiffies_to_msecs(ts->u.run.rs_running),
-				jiffies_to_msecs(ts->u.run.rs_locked),
-				jiffies_to_msecs(ts->u.run.rs_flushing),
-				jiffies_to_msecs(ts->u.run.rs_logging),
-				ts->u.run.rs_handle_count,
-				ts->u.run.rs_blocks,
-				ts->u.run.rs_blocks_logged);
-	else if (ts->ts_type == JBD2_STATS_CHECKPOINT)
-		seq_printf(seq, "%-4s %-5lu %48s %-5u %-5lu %-5lu %-5lu\n",
-				"C", ts->ts_tid, " ",
-				jiffies_to_msecs(ts->u.chp.cs_chp_time),
-				ts->u.chp.cs_written, ts->u.chp.cs_dropped,
-				ts->u.chp.cs_forced_to_close);
-	else
-		J_ASSERT(0);
-	return 0;
-}
-
-static void jbd2_seq_history_stop(struct seq_file *seq, void *v)
-{
-}
-
-static struct seq_operations jbd2_seq_history_ops = {
-	.start  = jbd2_seq_history_start,
-	.next   = jbd2_seq_history_next,
-	.stop   = jbd2_seq_history_stop,
-	.show   = jbd2_seq_history_show,
-};
-
-static int jbd2_seq_history_open(struct inode *inode, struct file *file)
-{
-	journal_t *journal = PDE(inode)->data;
-	struct jbd2_stats_proc_session *s;
-	int rc, size;
-
-	s = kmalloc(sizeof(*s), GFP_KERNEL);
-	if (s == NULL)
-		return -ENOMEM;
-	size = sizeof(struct transaction_stats_s) * journal->j_history_max;
-	s->stats = kmalloc(size, GFP_KERNEL);
-	if (s->stats == NULL) {
-		kfree(s);
-		return -ENOMEM;
-	}
-	spin_lock(&journal->j_history_lock);
-	memcpy(s->stats, journal->j_history, size);
-	s->max = journal->j_history_max;
-	s->start = journal->j_history_cur % s->max;
-	spin_unlock(&journal->j_history_lock);
-
-	rc = seq_open(file, &jbd2_seq_history_ops);
-	if (rc == 0) {
-		struct seq_file *m = file->private_data;
-		m->private = s;
-	} else {
-		kfree(s->stats);
-		kfree(s);
-	}
-	return rc;
-
-}
-
-static int jbd2_seq_history_release(struct inode *inode, struct file *file)
-{
-	struct seq_file *seq = file->private_data;
-	struct jbd2_stats_proc_session *s = seq->private;
-
-	kfree(s->stats);
-	kfree(s);
-	return seq_release(inode, file);
-}
-
-static struct file_operations jbd2_seq_history_fops = {
-	.owner		= THIS_MODULE,
-	.open           = jbd2_seq_history_open,
-	.read           = seq_read,
-	.llseek         = seq_lseek,
-	.release        = jbd2_seq_history_release,
-};
-
 static void *jbd2_seq_info_start(struct seq_file *seq, loff_t *pos)
 {
 	return *pos ? NULL : SEQ_START_TOKEN;
@@ -842,29 +692,29 @@ static int jbd2_seq_info_show(struct seq_file *seq, void *v)
 
 	if (v != SEQ_START_TOKEN)
 		return 0;
-	seq_printf(seq, "%lu transaction, each upto %u blocks\n",
+	seq_printf(seq, "%lu transaction, each up to %u blocks\n",
 			s->stats->ts_tid,
 			s->journal->j_max_transaction_buffers);
 	if (s->stats->ts_tid == 0)
 		return 0;
 	seq_printf(seq, "average: \n  %ums waiting for transaction\n",
-	    jiffies_to_msecs(s->stats->u.run.rs_wait / s->stats->ts_tid));
+	    jiffies_to_msecs(s->stats->run.rs_wait / s->stats->ts_tid));
 	seq_printf(seq, "  %ums running transaction\n",
-	    jiffies_to_msecs(s->stats->u.run.rs_running / s->stats->ts_tid));
+	    jiffies_to_msecs(s->stats->run.rs_running / s->stats->ts_tid));
 	seq_printf(seq, "  %ums transaction was being locked\n",
-	    jiffies_to_msecs(s->stats->u.run.rs_locked / s->stats->ts_tid));
+	    jiffies_to_msecs(s->stats->run.rs_locked / s->stats->ts_tid));
 	seq_printf(seq, "  %ums flushing data (in ordered mode)\n",
-	    jiffies_to_msecs(s->stats->u.run.rs_flushing / s->stats->ts_tid));
+	    jiffies_to_msecs(s->stats->run.rs_flushing / s->stats->ts_tid));
 	seq_printf(seq, "  %ums logging transaction\n",
-	    jiffies_to_msecs(s->stats->u.run.rs_logging / s->stats->ts_tid));
+	    jiffies_to_msecs(s->stats->run.rs_logging / s->stats->ts_tid));
 	seq_printf(seq, "  %lluus average transaction commit time\n",
 		   div_u64(s->journal->j_average_commit_time, 1000));
 	seq_printf(seq, "  %lu handles per transaction\n",
-	    s->stats->u.run.rs_handle_count / s->stats->ts_tid);
+	    s->stats->run.rs_handle_count / s->stats->ts_tid);
 	seq_printf(seq, "  %lu blocks per transaction\n",
-	    s->stats->u.run.rs_blocks / s->stats->ts_tid);
+	    s->stats->run.rs_blocks / s->stats->ts_tid);
 	seq_printf(seq, "  %lu logged blocks per transaction\n",
-	    s->stats->u.run.rs_blocks_logged / s->stats->ts_tid);
+	    s->stats->run.rs_blocks_logged / s->stats->ts_tid);
 	return 0;
 }
 
@@ -872,7 +722,7 @@ static void jbd2_seq_info_stop(struct seq_file *seq, void *v)
 {
 }
 
-static struct seq_operations jbd2_seq_info_ops = {
+static const struct seq_operations jbd2_seq_info_ops = {
 	.start  = jbd2_seq_info_start,
 	.next   = jbd2_seq_info_next,
 	.stop   = jbd2_seq_info_stop,
@@ -920,7 +770,7 @@ static int jbd2_seq_info_release(struct inode *inode, struct file *file)
 	return seq_release(inode, file);
 }
 
-static struct file_operations jbd2_seq_info_fops = {
+static const struct file_operations jbd2_seq_info_fops = {
 	.owner		= THIS_MODULE,
 	.open           = jbd2_seq_info_open,
 	.read           = seq_read,
@@ -934,8 +784,6 @@ static void jbd2_stats_proc_init(journal_t *journal)
 {
 	journal->j_proc_entry = proc_mkdir(journal->j_devname, proc_jbd2_stats);
 	if (journal->j_proc_entry) {
-		proc_create_data("history", S_IRUGO, journal->j_proc_entry,
-				 &jbd2_seq_history_fops, journal);
 		proc_create_data("info", S_IRUGO, journal->j_proc_entry,
 				 &jbd2_seq_info_fops, journal);
 	}
@@ -944,27 +792,9 @@ static void jbd2_stats_proc_init(journal_t *journal)
 static void jbd2_stats_proc_exit(journal_t *journal)
 {
 	remove_proc_entry("info", journal->j_proc_entry);
-	remove_proc_entry("history", journal->j_proc_entry);
 	remove_proc_entry(journal->j_devname, proc_jbd2_stats);
 }
 
-static void journal_init_stats(journal_t *journal)
-{
-	int size;
-
-	if (!proc_jbd2_stats)
-		return;
-
-	journal->j_history_max = 100;
-	size = sizeof(struct transaction_stats_s) * journal->j_history_max;
-	journal->j_history = kzalloc(size, GFP_KERNEL);
-	if (!journal->j_history) {
-		journal->j_history_max = 0;
-		return;
-	}
-	spin_lock_init(&journal->j_history_lock);
-}
-
 /*
  * Management for journal control blocks: functions to create and
  * destroy journal_t structures, and to initialise and read existing
@@ -1009,7 +839,7 @@ static journal_t * journal_init_common (void)
 		goto fail;
 	}
 
-	journal_init_stats(journal);
+	spin_lock_init(&journal->j_history_lock);
 
 	return journal;
 fail:
@@ -1115,7 +945,7 @@ journal_t * jbd2_journal_init_inode (struct inode *inode)
 	while ((p = strchr(p, '/')))
 		*p = '!';
 	p = journal->j_devname + strlen(journal->j_devname);
-	sprintf(p, ":%lu", journal->j_inode->i_ino);
+	sprintf(p, "-%lu", journal->j_inode->i_ino);
 	jbd_debug(1,
 		  "journal %p: inode %s/%ld, size %Ld, bits %d, blksize %ld\n",
 		  journal, inode->i_sb->s_id, inode->i_ino,
@@ -1187,6 +1017,12 @@ static int journal_reset(journal_t *journal)
 
 	first = be32_to_cpu(sb->s_first);
 	last = be32_to_cpu(sb->s_maxlen);
+	if (first + JBD2_MIN_JOURNAL_BLOCKS > last + 1) {
+		printk(KERN_ERR "JBD: Journal too short (blocks %llu-%llu).\n",
+		       first, last);
+		journal_fail_superblock(journal);
+		return -EINVAL;
+	}
 
 	journal->j_first = first;
 	journal->j_last = last;
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 6213ac728f30..a0512700542f 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -57,7 +57,7 @@ jbd2_get_transaction(journal_t *journal, transaction_t *transaction)
 	INIT_LIST_HEAD(&transaction->t_private_list);
 
 	/* Set up the commit timer for the new transaction. */
-	journal->j_commit_timer.expires = round_jiffies(transaction->t_expires);
+	journal->j_commit_timer.expires = round_jiffies_up(transaction->t_expires);
 	add_timer(&journal->j_commit_timer);
 
 	J_ASSERT(journal->j_running_transaction == NULL);
@@ -238,6 +238,8 @@ repeat_locked:
 		  __jbd2_log_space_left(journal));
 	spin_unlock(&transaction->t_handle_lock);
 	spin_unlock(&journal->j_state_lock);
+
+	lock_map_acquire(&handle->h_lockdep_map);
 out:
 	if (unlikely(new_transaction))		/* It's usually NULL */
 		kfree(new_transaction);
@@ -303,8 +305,6 @@ handle_t *jbd2_journal_start(journal_t *journal, int nblocks)
 		handle = ERR_PTR(err);
 		goto out;
 	}
-
-	lock_map_acquire(&handle->h_lockdep_map);
 out:
 	return handle;
 }
@@ -426,6 +426,7 @@ int jbd2_journal_restart(handle_t *handle, int nblocks)
 	__jbd2_log_start_commit(journal, transaction->t_tid);
 	spin_unlock(&journal->j_state_lock);
 
+	lock_map_release(&handle->h_lockdep_map);
 	handle->h_buffer_credits = nblocks;
 	ret = start_this_handle(journal, handle);
 	return ret;
diff --git a/fs/jffs2/background.c b/fs/jffs2/background.c
index e9580104b6ba..3ff50da94789 100644
--- a/fs/jffs2/background.c
+++ b/fs/jffs2/background.c
@@ -15,6 +15,7 @@
 #include <linux/completion.h>
 #include <linux/sched.h>
 #include <linux/freezer.h>
+#include <linux/kthread.h>
 #include "nodelist.h"
 
 
@@ -31,7 +32,7 @@ void jffs2_garbage_collect_trigger(struct jffs2_sb_info *c)
 /* This must only ever be called when no GC thread is currently running */
 int jffs2_start_garbage_collect_thread(struct jffs2_sb_info *c)
 {
-	pid_t pid;
+	struct task_struct *tsk;
 	int ret = 0;
 
 	BUG_ON(c->gc_task);
@@ -39,15 +40,16 @@ int jffs2_start_garbage_collect_thread(struct jffs2_sb_info *c)
 	init_completion(&c->gc_thread_start);
 	init_completion(&c->gc_thread_exit);
 
-	pid = kernel_thread(jffs2_garbage_collect_thread, c, CLONE_FS|CLONE_FILES);
-	if (pid < 0) {
-		printk(KERN_WARNING "fork failed for JFFS2 garbage collect thread: %d\n", -pid);
+	tsk = kthread_run(jffs2_garbage_collect_thread, c, "jffs2_gcd_mtd%d", c->mtd->index);
+	if (IS_ERR(tsk)) {
+		printk(KERN_WARNING "fork failed for JFFS2 garbage collect thread: %ld\n", -PTR_ERR(tsk));
 		complete(&c->gc_thread_exit);
-		ret = pid;
+		ret = PTR_ERR(tsk);
 	} else {
 		/* Wait for it... */
-		D1(printk(KERN_DEBUG "JFFS2: Garbage collect thread is pid %d\n", pid));
+		D1(printk(KERN_DEBUG "JFFS2: Garbage collect thread is pid %d\n", tsk->pid));
 		wait_for_completion(&c->gc_thread_start);
+		ret = tsk->pid;
 	}
 
 	return ret;
@@ -71,7 +73,6 @@ static int jffs2_garbage_collect_thread(void *_c)
 {
 	struct jffs2_sb_info *c = _c;
 
-	daemonize("jffs2_gcd_mtd%d", c->mtd->index);
 	allow_signal(SIGKILL);
 	allow_signal(SIGSTOP);
 	allow_signal(SIGCONT);
@@ -107,6 +108,11 @@ static int jffs2_garbage_collect_thread(void *_c)
 		 * the GC thread get there first. */
 		schedule_timeout_interruptible(msecs_to_jiffies(50));
 
+		if (kthread_should_stop()) {
+			D1(printk(KERN_DEBUG "jffs2_garbage_collect_thread():  kthread_stop() called.\n"));
+			goto die;
+		}
+
 		/* Put_super will send a SIGKILL and then wait on the sem.
 		 */
 		while (signal_pending(current) || freezing(current)) {
diff --git a/fs/jffs2/malloc.c b/fs/jffs2/malloc.c
index 9eff2bdae8a7..c082868910f2 100644
--- a/fs/jffs2/malloc.c
+++ b/fs/jffs2/malloc.c
@@ -39,13 +39,13 @@ int __init jffs2_create_slab_caches(void)
 
 	raw_dirent_slab = kmem_cache_create("jffs2_raw_dirent",
 					    sizeof(struct jffs2_raw_dirent),
-					    0, 0, NULL);
+					    0, SLAB_HWCACHE_ALIGN, NULL);
 	if (!raw_dirent_slab)
 		goto err;
 
 	raw_inode_slab = kmem_cache_create("jffs2_raw_inode",
 					   sizeof(struct jffs2_raw_inode),
-					   0, 0, NULL);
+					   0, SLAB_HWCACHE_ALIGN, NULL);
 	if (!raw_inode_slab)
 		goto err;
 
diff --git a/fs/jffs2/super.c b/fs/jffs2/super.c
index 0035c021395a..9a80e8e595d0 100644
--- a/fs/jffs2/super.c
+++ b/fs/jffs2/super.c
@@ -123,7 +123,7 @@ static struct dentry *jffs2_get_parent(struct dentry *child)
 	return d_obtain_alias(jffs2_iget(child->d_inode->i_sb, pino));
 }
 
-static struct export_operations jffs2_export_ops = {
+static const struct export_operations jffs2_export_ops = {
 	.get_parent = jffs2_get_parent,
 	.fh_to_dentry = jffs2_fh_to_dentry,
 	.fh_to_parent = jffs2_fh_to_parent,
diff --git a/fs/jfs/super.c b/fs/jfs/super.c
index 37e6dcda8fc8..2234c73fc577 100644
--- a/fs/jfs/super.c
+++ b/fs/jfs/super.c
@@ -178,13 +178,11 @@ static void jfs_put_super(struct super_block *sb)
 	rc = jfs_umount(sb);
 	if (rc)
 		jfs_err("jfs_umount failed with return code %d", rc);
-	if (sbi->nls_tab)
-		unload_nls(sbi->nls_tab);
-	sbi->nls_tab = NULL;
+
+	unload_nls(sbi->nls_tab);
 
 	truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
 	iput(sbi->direct_inode);
-	sbi->direct_inode = NULL;
 
 	kfree(sbi);
 
@@ -347,8 +345,7 @@ static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
 
 	if (nls_map != (void *) -1) {
 		/* Discard old (if remount) */
-		if (sbi->nls_tab)
-			unload_nls(sbi->nls_tab);
+		unload_nls(sbi->nls_tab);
 		sbi->nls_tab = nls_map;
 	}
 	return 1;
diff --git a/fs/libfs.c b/fs/libfs.c
index dcec3d3ea64f..219576c52d80 100644
--- a/fs/libfs.c
+++ b/fs/libfs.c
@@ -527,14 +527,18 @@ ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
 				const void *from, size_t available)
 {
 	loff_t pos = *ppos;
+	size_t ret;
+
 	if (pos < 0)
 		return -EINVAL;
-	if (pos >= available)
+	if (pos >= available || !count)
 		return 0;
 	if (count > available - pos)
 		count = available - pos;
-	if (copy_to_user(to, from + pos, count))
+	ret = copy_to_user(to, from + pos, count);
+	if (ret == count)
 		return -EFAULT;
+	count -= ret;
 	*ppos = pos + count;
 	return count;
 }
@@ -735,10 +739,11 @@ ssize_t simple_attr_write(struct file *file, const char __user *buf,
 	if (copy_from_user(attr->set_buf, buf, size))
 		goto out;
 
-	ret = len; /* claim we got the whole input */
 	attr->set_buf[size] = '\0';
 	val = simple_strtol(attr->set_buf, NULL, 0);
-	attr->set(attr->data, val);
+	ret = attr->set(attr->data, val);
+	if (ret == 0)
+		ret = len; /* on success, claim we got the whole input */
 out:
 	mutex_unlock(&attr->mutex);
 	return ret;
diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c
index 1f3b0fc0d351..fc9032dc8862 100644
--- a/fs/lockd/clntlock.c
+++ b/fs/lockd/clntlock.c
@@ -166,7 +166,7 @@ __be32 nlmclnt_grant(const struct sockaddr *addr, const struct nlm_lock *lock)
 		 */
 		if (fl_blocked->fl_u.nfs_fl.owner->pid != lock->svid)
 			continue;
-		if (!nlm_cmp_addr(nlm_addr(block->b_host), addr))
+		if (!rpc_cmp_addr(nlm_addr(block->b_host), addr))
 			continue;
 		if (nfs_compare_fh(NFS_FH(fl_blocked->fl_file->f_path.dentry->d_inode) ,fh) != 0)
 			continue;
diff --git a/fs/lockd/clntproc.c b/fs/lockd/clntproc.c
index 4336adba952a..c81249fef11f 100644
--- a/fs/lockd/clntproc.c
+++ b/fs/lockd/clntproc.c
@@ -458,7 +458,7 @@ static void nlmclnt_locks_release_private(struct file_lock *fl)
 	nlm_put_lockowner(fl->fl_u.nfs_fl.owner);
 }
 
-static struct file_lock_operations nlmclnt_lock_ops = {
+static const struct file_lock_operations nlmclnt_lock_ops = {
 	.fl_copy_lock = nlmclnt_locks_copy_lock,
 	.fl_release_private = nlmclnt_locks_release_private,
 };
diff --git a/fs/lockd/host.c b/fs/lockd/host.c
index 7cb076ac6b45..4600c2037b8b 100644
--- a/fs/lockd/host.c
+++ b/fs/lockd/host.c
@@ -111,7 +111,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni)
 	 */
 	chain = &nlm_hosts[nlm_hash_address(ni->sap)];
 	hlist_for_each_entry(host, pos, chain, h_hash) {
-		if (!nlm_cmp_addr(nlm_addr(host), ni->sap))
+		if (!rpc_cmp_addr(nlm_addr(host), ni->sap))
 			continue;
 
 		/* See if we have an NSM handle for this client */
@@ -125,7 +125,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni)
 		if (host->h_server != ni->server)
 			continue;
 		if (ni->server &&
-		    !nlm_cmp_addr(nlm_srcaddr(host), ni->src_sap))
+		    !rpc_cmp_addr(nlm_srcaddr(host), ni->src_sap))
 			continue;
 
 		/* Move to head of hash chain. */
diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c
index 30c933188dd7..f956651d0f65 100644
--- a/fs/lockd/mon.c
+++ b/fs/lockd/mon.c
@@ -209,7 +209,7 @@ static struct nsm_handle *nsm_lookup_addr(const struct sockaddr *sap)
 	struct nsm_handle *nsm;
 
 	list_for_each_entry(nsm, &nsm_handles, sm_link)
-		if (nlm_cmp_addr(nsm_addr(nsm), sap))
+		if (rpc_cmp_addr(nsm_addr(nsm), sap))
 			return nsm;
 	return NULL;
 }
diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c
index e577a78d7bac..d1001790fa9a 100644
--- a/fs/lockd/svclock.c
+++ b/fs/lockd/svclock.c
@@ -705,7 +705,7 @@ static int nlmsvc_same_owner(struct file_lock *fl1, struct file_lock *fl2)
 	return fl1->fl_owner == fl2->fl_owner && fl1->fl_pid == fl2->fl_pid;
 }
 
-struct lock_manager_operations nlmsvc_lock_operations = {
+const struct lock_manager_operations nlmsvc_lock_operations = {
 	.fl_compare_owner = nlmsvc_same_owner,
 	.fl_notify = nlmsvc_notify_blocked,
 	.fl_grant = nlmsvc_grant_deferred,
diff --git a/fs/lockd/svcsubs.c b/fs/lockd/svcsubs.c
index 9e4d6aab611b..ad478da7ca63 100644
--- a/fs/lockd/svcsubs.c
+++ b/fs/lockd/svcsubs.c
@@ -417,7 +417,7 @@ EXPORT_SYMBOL_GPL(nlmsvc_unlock_all_by_sb);
 static int
 nlmsvc_match_ip(void *datap, struct nlm_host *host)
 {
-	return nlm_cmp_addr(nlm_srcaddr(host), datap);
+	return rpc_cmp_addr(nlm_srcaddr(host), datap);
 }
 
 /**
diff --git a/fs/lockd/xdr.c b/fs/lockd/xdr.c
index 0336f2beacde..b583ab0a4cbb 100644
--- a/fs/lockd/xdr.c
+++ b/fs/lockd/xdr.c
@@ -8,7 +8,6 @@
 
 #include <linux/types.h>
 #include <linux/sched.h>
-#include <linux/utsname.h>
 #include <linux/nfs.h>
 
 #include <linux/sunrpc/xdr.h>
diff --git a/fs/lockd/xdr4.c b/fs/lockd/xdr4.c
index e1d528653192..ad9dbbc9145d 100644
--- a/fs/lockd/xdr4.c
+++ b/fs/lockd/xdr4.c
@@ -9,7 +9,6 @@
 
 #include <linux/types.h>
 #include <linux/sched.h>
-#include <linux/utsname.h>
 #include <linux/nfs.h>
 
 #include <linux/sunrpc/xdr.h>
diff --git a/fs/locks.c b/fs/locks.c
index 19ee18a6829b..a8794f233bc9 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -434,7 +434,7 @@ static int lease_mylease_callback(struct file_lock *fl, struct file_lock *try)
 	return fl->fl_file == try->fl_file;
 }
 
-static struct lock_manager_operations lease_manager_ops = {
+static const struct lock_manager_operations lease_manager_ops = {
 	.fl_break = lease_break_callback,
 	.fl_release_private = lease_release_private_callback,
 	.fl_mylease = lease_mylease_callback,
diff --git a/fs/minix/dir.c b/fs/minix/dir.c
index d407e7a0b6fe..6198731d7fcd 100644
--- a/fs/minix/dir.c
+++ b/fs/minix/dir.c
@@ -308,14 +308,18 @@ int minix_delete_entry(struct minix_dir_entry *de, struct page *page)
 	struct inode *inode = (struct inode*)mapping->host;
 	char *kaddr = page_address(page);
 	loff_t pos = page_offset(page) + (char*)de - kaddr;
-	unsigned len = minix_sb(inode->i_sb)->s_dirsize;
+	struct minix_sb_info *sbi = minix_sb(inode->i_sb);
+	unsigned len = sbi->s_dirsize;
 	int err;
 
 	lock_page(page);
 	err = __minix_write_begin(NULL, mapping, pos, len,
 					AOP_FLAG_UNINTERRUPTIBLE, &page, NULL);
 	if (err == 0) {
-		de->inode = 0;
+		if (sbi->s_version == MINIX_V3)
+			((minix3_dirent *) de)->inode = 0;
+		else
+			de->inode = 0;
 		err = dir_commit_chunk(page, pos, len);
 	} else {
 		unlock_page(page);
@@ -440,7 +444,10 @@ void minix_set_link(struct minix_dir_entry *de, struct page *page,
 	err = __minix_write_begin(NULL, mapping, pos, sbi->s_dirsize,
 					AOP_FLAG_UNINTERRUPTIBLE, &page, NULL);
 	if (err == 0) {
-		de->inode = inode->i_ino;
+		if (sbi->s_version == MINIX_V3)
+			((minix3_dirent *) de)->inode = inode->i_ino;
+		else
+			de->inode = inode->i_ino;
 		err = dir_commit_chunk(page, pos, sbi->s_dirsize);
 	} else {
 		unlock_page(page);
@@ -470,7 +477,14 @@ ino_t minix_inode_by_name(struct dentry *dentry)
 	ino_t res = 0;
 
 	if (de) {
-		res = de->inode;
+		struct address_space *mapping = page->mapping;
+		struct inode *inode = mapping->host;
+		struct minix_sb_info *sbi = minix_sb(inode->i_sb);
+
+		if (sbi->s_version == MINIX_V3)
+			res = ((minix3_dirent *) de)->inode;
+		else
+			res = de->inode;
 		dir_put_page(page);
 	}
 	return res;
diff --git a/fs/namespace.c b/fs/namespace.c
index 7230787d18b0..bdc3cb4fd222 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -1640,7 +1640,7 @@ static int do_new_mount(struct path *path, char *type, int flags,
 {
 	struct vfsmount *mnt;
 
-	if (!type || !memchr(type, 0, PAGE_SIZE))
+	if (!type)
 		return -EINVAL;
 
 	/* we need capabilities... */
@@ -1871,6 +1871,23 @@ int copy_mount_options(const void __user * data, unsigned long *where)
 	return 0;
 }
 
+int copy_mount_string(const void __user *data, char **where)
+{
+	char *tmp;
+
+	if (!data) {
+		*where = NULL;
+		return 0;
+	}
+
+	tmp = strndup_user(data, PAGE_SIZE);
+	if (IS_ERR(tmp))
+		return PTR_ERR(tmp);
+
+	*where = tmp;
+	return 0;
+}
+
 /*
  * Flags is a 32-bit value that allows up to 31 non-fs dependent flags to
  * be given to the mount() call (ie: read-only, no-dev, no-suid etc).
@@ -1900,8 +1917,6 @@ long do_mount(char *dev_name, char *dir_name, char *type_page,
 
 	if (!dir_name || !*dir_name || !memchr(dir_name, 0, PAGE_SIZE))
 		return -EINVAL;
-	if (dev_name && !memchr(dev_name, 0, PAGE_SIZE))
-		return -EINVAL;
 
 	if (data_page)
 		((char *)data_page)[PAGE_SIZE - 1] = 0;
@@ -2070,40 +2085,42 @@ EXPORT_SYMBOL(create_mnt_ns);
 SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name,
 		char __user *, type, unsigned long, flags, void __user *, data)
 {
-	int retval;
+	int ret;
+	char *kernel_type;
+	char *kernel_dir;
+	char *kernel_dev;
 	unsigned long data_page;
-	unsigned long type_page;
-	unsigned long dev_page;
-	char *dir_page;
 
-	retval = copy_mount_options(type, &type_page);
-	if (retval < 0)
-		return retval;
+	ret = copy_mount_string(type, &kernel_type);
+	if (ret < 0)
+		goto out_type;
 
-	dir_page = getname(dir_name);
-	retval = PTR_ERR(dir_page);
-	if (IS_ERR(dir_page))
-		goto out1;
+	kernel_dir = getname(dir_name);
+	if (IS_ERR(kernel_dir)) {
+		ret = PTR_ERR(kernel_dir);
+		goto out_dir;
+	}
 
-	retval = copy_mount_options(dev_name, &dev_page);
-	if (retval < 0)
-		goto out2;
+	ret = copy_mount_string(dev_name, &kernel_dev);
+	if (ret < 0)
+		goto out_dev;
 
-	retval = copy_mount_options(data, &data_page);
-	if (retval < 0)
-		goto out3;
+	ret = copy_mount_options(data, &data_page);
+	if (ret < 0)
+		goto out_data;
 
-	retval = do_mount((char *)dev_page, dir_page, (char *)type_page,
-			  flags, (void *)data_page);
-	free_page(data_page);
+	ret = do_mount(kernel_dev, kernel_dir, kernel_type, flags,
+		(void *) data_page);
 
-out3:
-	free_page(dev_page);
-out2:
-	putname(dir_page);
-out1:
-	free_page(type_page);
-	return retval;
+	free_page(data_page);
+out_data:
+	kfree(kernel_dev);
+out_dev:
+	putname(kernel_dir);
+out_dir:
+	kfree(kernel_type);
+out_type:
+	return ret;
 }
 
 /*
diff --git a/fs/ncpfs/dir.c b/fs/ncpfs/dir.c
index 9c590722d87e..b8b5b30d53f0 100644
--- a/fs/ncpfs/dir.c
+++ b/fs/ncpfs/dir.c
@@ -1241,7 +1241,7 @@ ncp_date_unix2dos(int unix_date, __le16 *time, __le16 *date)
 		month = 2;
 	} else {
 		nl_day = (year & 3) || day <= 59 ? day : day - 1;
-		for (month = 0; month < 12; month++)
+		for (month = 1; month < 12; month++)
 			if (day_n[month] > nl_day)
 				break;
 	}
diff --git a/fs/ncpfs/inode.c b/fs/ncpfs/inode.c
index b99ce205b1bd..cf98da1be23e 100644
--- a/fs/ncpfs/inode.c
+++ b/fs/ncpfs/inode.c
@@ -746,16 +746,8 @@ static void ncp_put_super(struct super_block *sb)
 
 #ifdef CONFIG_NCPFS_NLS
 	/* unload the NLS charsets */
-	if (server->nls_vol)
-	{
-		unload_nls(server->nls_vol);
-		server->nls_vol = NULL;
-	}
-	if (server->nls_io)
-	{
-		unload_nls(server->nls_io);
-		server->nls_io = NULL;
-	}
+	unload_nls(server->nls_vol);
+	unload_nls(server->nls_io);
 #endif /* CONFIG_NCPFS_NLS */
 
 	if (server->info_filp)
diff --git a/fs/ncpfs/ioctl.c b/fs/ncpfs/ioctl.c
index fa038df63ac8..0d58caf4a6e1 100644
--- a/fs/ncpfs/ioctl.c
+++ b/fs/ncpfs/ioctl.c
@@ -223,10 +223,8 @@ ncp_set_charsets(struct ncp_server* server, struct ncp_nls_ioctl __user *arg)
 	oldset_io = server->nls_io;
 	server->nls_io = iocharset;
 
-	if (oldset_cp)
-		unload_nls(oldset_cp);
-	if (oldset_io)
-		unload_nls(oldset_io);
+	unload_nls(oldset_cp);
+	unload_nls(oldset_io);
 
 	return 0;
 }
@@ -442,7 +440,7 @@ static int __ncp_ioctl(struct inode *inode, struct file *filp,
 			if (dentry) {
 				struct inode* s_inode = dentry->d_inode;
 				
-				if (inode) {
+				if (s_inode) {
 					NCP_FINFO(s_inode)->volNumber = vnum;
 					NCP_FINFO(s_inode)->dirEntNum = de;
 					NCP_FINFO(s_inode)->DosDirNum = dosde;
diff --git a/fs/ncpfs/mmap.c b/fs/ncpfs/mmap.c
index 5d8dcb9ee326..15458decdb8a 100644
--- a/fs/ncpfs/mmap.c
+++ b/fs/ncpfs/mmap.c
@@ -95,7 +95,7 @@ static int ncp_file_mmap_fault(struct vm_area_struct *area,
 	return VM_FAULT_MAJOR;
 }
 
-static struct vm_operations_struct ncp_file_mmap =
+static const struct vm_operations_struct ncp_file_mmap =
 {
 	.fault = ncp_file_mmap_fault,
 };
diff --git a/fs/nfs/callback_xdr.c b/fs/nfs/callback_xdr.c
index e5a2dac5f715..76b0aa0f73bf 100644
--- a/fs/nfs/callback_xdr.c
+++ b/fs/nfs/callback_xdr.c
@@ -222,7 +222,7 @@ static unsigned decode_sessionid(struct xdr_stream *xdr,
 
 	p = read_buf(xdr, len);
 	if (unlikely(p == NULL))
-		return htonl(NFS4ERR_RESOURCE);;
+		return htonl(NFS4ERR_RESOURCE);
 
 	memcpy(sid->data, p, len);
 	return 0;
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index e350bd6a2334..99ea196f071f 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -648,8 +648,6 @@ static int nfs_start_lockd(struct nfs_server *server)
 		.hostname	= clp->cl_hostname,
 		.address	= (struct sockaddr *)&clp->cl_addr,
 		.addrlen	= clp->cl_addrlen,
-		.protocol	= server->flags & NFS_MOUNT_TCP ?
-						IPPROTO_TCP : IPPROTO_UDP,
 		.nfs_version	= clp->rpc_ops->version,
 		.noresvport	= server->flags & NFS_MOUNT_NORESVPORT ?
 					1 : 0,
@@ -660,6 +658,14 @@ static int nfs_start_lockd(struct nfs_server *server)
 	if (server->flags & NFS_MOUNT_NONLM)
 		return 0;
 
+	switch (clp->cl_proto) {
+		default:
+			nlm_init.protocol = IPPROTO_TCP;
+			break;
+		case XPRT_TRANSPORT_UDP:
+			nlm_init.protocol = IPPROTO_UDP;
+	}
+
 	host = nlmclnt_init(&nlm_init);
 	if (IS_ERR(host))
 		return PTR_ERR(host);
@@ -787,7 +793,7 @@ static int nfs_init_server(struct nfs_server *server,
 	dprintk("--> nfs_init_server()\n");
 
 #ifdef CONFIG_NFS_V3
-	if (data->flags & NFS_MOUNT_VER3)
+	if (data->version == 3)
 		cl_init.rpc_ops = &nfs_v3_clientops;
 #endif
 
@@ -933,10 +939,6 @@ static int nfs_probe_fsinfo(struct nfs_server *server, struct nfs_fh *mntfh, str
 		goto out_error;
 
 	nfs_server_set_fsinfo(server, &fsinfo);
-	error = bdi_init(&server->backing_dev_info);
-	if (error)
-		goto out_error;
-
 
 	/* Get some general file system info */
 	if (server->namelen == 0) {
@@ -968,6 +970,7 @@ static void nfs_server_copy_userdata(struct nfs_server *target, struct nfs_serve
 	target->acdirmin = source->acdirmin;
 	target->acdirmax = source->acdirmax;
 	target->caps = source->caps;
+	target->options = source->options;
 }
 
 /*
@@ -995,6 +998,12 @@ static struct nfs_server *nfs_alloc_server(void)
 		return NULL;
 	}
 
+	if (bdi_init(&server->backing_dev_info)) {
+		nfs_free_iostats(server->io_stats);
+		kfree(server);
+		return NULL;
+	}
+
 	return server;
 }
 
@@ -1171,7 +1180,7 @@ static int nfs4_init_client(struct nfs_client *clp,
 				      1, flags & NFS_MOUNT_NORESVPORT);
 	if (error < 0)
 		goto error;
-	memcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
+	strlcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
 
 	error = nfs_idmap_new(clp);
 	if (error < 0) {
@@ -1529,7 +1538,7 @@ static void *nfs_server_list_next(struct seq_file *p, void *v, loff_t *pos);
 static void nfs_server_list_stop(struct seq_file *p, void *v);
 static int nfs_server_list_show(struct seq_file *m, void *v);
 
-static struct seq_operations nfs_server_list_ops = {
+static const struct seq_operations nfs_server_list_ops = {
 	.start	= nfs_server_list_start,
 	.next	= nfs_server_list_next,
 	.stop	= nfs_server_list_stop,
@@ -1550,7 +1559,7 @@ static void *nfs_volume_list_next(struct seq_file *p, void *v, loff_t *pos);
 static void nfs_volume_list_stop(struct seq_file *p, void *v);
 static int nfs_volume_list_show(struct seq_file *m, void *v);
 
-static struct seq_operations nfs_volume_list_ops = {
+static const struct seq_operations nfs_volume_list_ops = {
 	.start	= nfs_volume_list_start,
 	.next	= nfs_volume_list_next,
 	.stop	= nfs_volume_list_stop,
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 5021b75d2d1e..f5fdd39e037a 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -59,7 +59,7 @@ static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl);
 static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl);
 static int nfs_setlease(struct file *file, long arg, struct file_lock **fl);
 
-static struct vm_operations_struct nfs_file_vm_ops;
+static const struct vm_operations_struct nfs_file_vm_ops;
 
 const struct file_operations nfs_file_operations = {
 	.llseek		= nfs_file_llseek,
@@ -525,6 +525,7 @@ const struct address_space_operations nfs_file_aops = {
 	.direct_IO = nfs_direct_IO,
 	.migratepage = nfs_migrate_page,
 	.launder_page = nfs_launder_page,
+	.error_remove_page = generic_error_remove_page,
 };
 
 /*
@@ -571,7 +572,7 @@ out_unlock:
 	return VM_FAULT_SIGBUS;
 }
 
-static struct vm_operations_struct nfs_file_vm_ops = {
+static const struct vm_operations_struct nfs_file_vm_ops = {
 	.fault = filemap_fault,
 	.page_mkwrite = nfs_vm_page_mkwrite,
 };
diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c
index 379be678cb7e..70fad69eb959 100644
--- a/fs/nfs/fscache.c
+++ b/fs/nfs/fscache.c
@@ -58,17 +58,34 @@ void nfs_fscache_release_client_cookie(struct nfs_client *clp)
 /*
  * Get the cache cookie for an NFS superblock.  We have to handle
  * uniquification here because the cache doesn't do it for us.
+ *
+ * The default uniquifier is just an empty string, but it may be overridden
+ * either by the 'fsc=xxx' option to mount, or by inheriting it from the parent
+ * superblock across an automount point of some nature.
  */
-void nfs_fscache_get_super_cookie(struct super_block *sb,
-				  struct nfs_parsed_mount_data *data)
+void nfs_fscache_get_super_cookie(struct super_block *sb, const char *uniq,
+				  struct nfs_clone_mount *mntdata)
 {
 	struct nfs_fscache_key *key, *xkey;
 	struct nfs_server *nfss = NFS_SB(sb);
 	struct rb_node **p, *parent;
-	const char *uniq = data->fscache_uniq ?: "";
 	int diff, ulen;
 
-	ulen = strlen(uniq);
+	if (uniq) {
+		ulen = strlen(uniq);
+	} else if (mntdata) {
+		struct nfs_server *mnt_s = NFS_SB(mntdata->sb);
+		if (mnt_s->fscache_key) {
+			uniq = mnt_s->fscache_key->key.uniquifier;
+			ulen = mnt_s->fscache_key->key.uniq_len;
+		}
+	}
+
+	if (!uniq) {
+		uniq = "";
+		ulen = 1;
+	}
+
 	key = kzalloc(sizeof(*key) + ulen, GFP_KERNEL);
 	if (!key)
 		return;
diff --git a/fs/nfs/fscache.h b/fs/nfs/fscache.h
index 6e809bb0ff08..b9c572d0679f 100644
--- a/fs/nfs/fscache.h
+++ b/fs/nfs/fscache.h
@@ -74,7 +74,8 @@ extern void nfs_fscache_get_client_cookie(struct nfs_client *);
 extern void nfs_fscache_release_client_cookie(struct nfs_client *);
 
 extern void nfs_fscache_get_super_cookie(struct super_block *,
-					 struct nfs_parsed_mount_data *);
+					 const char *,
+					 struct nfs_clone_mount *);
 extern void nfs_fscache_release_super_cookie(struct super_block *);
 
 extern void nfs_fscache_init_inode_cookie(struct inode *);
@@ -173,7 +174,8 @@ static inline void nfs_fscache_release_client_cookie(struct nfs_client *clp) {}
 
 static inline void nfs_fscache_get_super_cookie(
 	struct super_block *sb,
-	struct nfs_parsed_mount_data *data)
+	const char *uniq,
+	struct nfs_clone_mount *mntdata)
 {
 }
 static inline void nfs_fscache_release_super_cookie(struct super_block *sb) {}
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 060022b4651c..faa091865ad0 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -458,49 +458,21 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
  */
 static int nfs_vmtruncate(struct inode * inode, loff_t offset)
 {
-	if (i_size_read(inode) < offset) {
-		unsigned long limit;
-
-		limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
-		if (limit != RLIM_INFINITY && offset > limit)
-			goto out_sig;
-		if (offset > inode->i_sb->s_maxbytes)
-			goto out_big;
-		spin_lock(&inode->i_lock);
-		i_size_write(inode, offset);
-		spin_unlock(&inode->i_lock);
-	} else {
-		struct address_space *mapping = inode->i_mapping;
+	loff_t oldsize;
+	int err;
 
-		/*
-		 * truncation of in-use swapfiles is disallowed - it would
-		 * cause subsequent swapout to scribble on the now-freed
-		 * blocks.
-		 */
-		if (IS_SWAPFILE(inode))
-			return -ETXTBSY;
-		spin_lock(&inode->i_lock);
-		i_size_write(inode, offset);
-		spin_unlock(&inode->i_lock);
+	err = inode_newsize_ok(inode, offset);
+	if (err)
+		goto out;
 
-		/*
-		 * unmap_mapping_range is called twice, first simply for
-		 * efficiency so that truncate_inode_pages does fewer
-		 * single-page unmaps.  However after this first call, and
-		 * before truncate_inode_pages finishes, it is possible for
-		 * private pages to be COWed, which remain after
-		 * truncate_inode_pages finishes, hence the second
-		 * unmap_mapping_range call must be made for correctness.
-		 */
-		unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-		truncate_inode_pages(mapping, offset);
-		unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
-	}
-	return 0;
-out_sig:
-	send_sig(SIGXFSZ, current, 0);
-out_big:
-	return -EFBIG;
+	spin_lock(&inode->i_lock);
+	oldsize = inode->i_size;
+	i_size_write(inode, offset);
+	spin_unlock(&inode->i_lock);
+
+	truncate_pagecache(inode, oldsize, offset);
+out:
+	return err;
 }
 
 /**
diff --git a/fs/nfs/nfs2xdr.c b/fs/nfs/nfs2xdr.c
index c862c9340f9a..5e078b222b4e 100644
--- a/fs/nfs/nfs2xdr.c
+++ b/fs/nfs/nfs2xdr.c
@@ -13,7 +13,6 @@
 #include <linux/time.h>
 #include <linux/mm.h>
 #include <linux/slab.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/in.h>
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index ee6a13f05443..3f8881d1a050 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -7,7 +7,6 @@
  */
 
 #include <linux/mm.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/sunrpc/clnt.h>
diff --git a/fs/nfs/nfs3xdr.c b/fs/nfs/nfs3xdr.c
index 35869a4921f1..5fe5492fbd29 100644
--- a/fs/nfs/nfs3xdr.c
+++ b/fs/nfs/nfs3xdr.c
@@ -10,7 +10,6 @@
 #include <linux/time.h>
 #include <linux/mm.h>
 #include <linux/slab.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/in.h>
diff --git a/fs/nfs/nfs4namespace.c b/fs/nfs/nfs4namespace.c
index 2636c26d56fa..fa3408f20112 100644
--- a/fs/nfs/nfs4namespace.c
+++ b/fs/nfs/nfs4namespace.c
@@ -121,7 +121,7 @@ static struct vfsmount *try_location(struct nfs_clone_mount *mountdata,
 
 	mnt_path = nfs4_pathname_string(&location->rootpath, page2, PAGE_SIZE);
 	if (IS_ERR(mnt_path))
-		return mnt;
+		return ERR_CAST(mnt_path);
 	mountdata->mnt_path = mnt_path;
 	maxbuflen = mnt_path - 1 - page2;
 
@@ -132,15 +132,15 @@ static struct vfsmount *try_location(struct nfs_clone_mount *mountdata,
 		if (buf->len <= 0 || buf->len >= maxbuflen)
 			continue;
 
-		mountdata->addr = (struct sockaddr *)&addr;
-
 		if (memchr(buf->data, IPV6_SCOPE_DELIMITER, buf->len))
 			continue;
-		mountdata->addrlen = nfs_parse_server_name(buf->data,
-				buf->len,
-				mountdata->addr, mountdata->addrlen);
+
+		mountdata->addrlen = nfs_parse_server_name(buf->data, buf->len,
+				(struct sockaddr *)&addr, sizeof(addr));
 		if (mountdata->addrlen == 0)
 			continue;
+
+		mountdata->addr = (struct sockaddr *)&addr;
 		rpc_set_port(mountdata->addr, NFS_PORT);
 
 		memcpy(page2, buf->data, buf->len);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index be6544aef41f..ed7c269e2514 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -36,7 +36,6 @@
  */
 
 #include <linux/mm.h>
-#include <linux/utsname.h>
 #include <linux/delay.h>
 #include <linux/errno.h>
 #include <linux/string.h>
diff --git a/fs/nfs/nfs4renewd.c b/fs/nfs/nfs4renewd.c
index e27c6cef18f2..0156c01c212c 100644
--- a/fs/nfs/nfs4renewd.c
+++ b/fs/nfs/nfs4renewd.c
@@ -127,12 +127,6 @@ nfs4_schedule_state_renewal(struct nfs_client *clp)
 }
 
 void
-nfs4_renewd_prepare_shutdown(struct nfs_server *server)
-{
-	cancel_delayed_work(&server->nfs_client->cl_renewd);
-}
-
-void
 nfs4_kill_renewd(struct nfs_client *clp)
 {
 	cancel_delayed_work_sync(&clp->cl_renewd);
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 1434080aefeb..2ef4fecf3984 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -638,7 +638,7 @@ static void nfs4_fl_release_lock(struct file_lock *fl)
 	nfs4_put_lock_state(fl->fl_u.nfs4_fl.owner);
 }
 
-static struct file_lock_operations nfs4_fl_lock_ops = {
+static const struct file_lock_operations nfs4_fl_lock_ops = {
 	.fl_copy_lock = nfs4_fl_copy_lock,
 	.fl_release_private = nfs4_fl_release_lock,
 };
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index cfc30d362f94..83ad47cbdd8a 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -39,7 +39,6 @@
 #include <linux/time.h>
 #include <linux/mm.h>
 #include <linux/slab.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/in.h>
diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c
index 7be72d90d49d..ef583854d8d0 100644
--- a/fs/nfs/proc.c
+++ b/fs/nfs/proc.c
@@ -32,7 +32,6 @@
 #include <linux/slab.h>
 #include <linux/time.h>
 #include <linux/mm.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/in.h>
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index de935692d40d..90be551b80c1 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -728,6 +728,29 @@ static void nfs_umount_begin(struct super_block *sb)
 	unlock_kernel();
 }
 
+static struct nfs_parsed_mount_data *nfs_alloc_parsed_mount_data(unsigned int version)
+{
+	struct nfs_parsed_mount_data *data;
+
+	data = kzalloc(sizeof(*data), GFP_KERNEL);
+	if (data) {
+		data->rsize		= NFS_MAX_FILE_IO_SIZE;
+		data->wsize		= NFS_MAX_FILE_IO_SIZE;
+		data->acregmin		= NFS_DEF_ACREGMIN;
+		data->acregmax		= NFS_DEF_ACREGMAX;
+		data->acdirmin		= NFS_DEF_ACDIRMIN;
+		data->acdirmax		= NFS_DEF_ACDIRMAX;
+		data->mount_server.port	= NFS_UNSPEC_PORT;
+		data->nfs_server.port	= NFS_UNSPEC_PORT;
+		data->nfs_server.protocol = XPRT_TRANSPORT_TCP;
+		data->auth_flavors[0]	= RPC_AUTH_UNIX;
+		data->auth_flavor_len	= 1;
+		data->version		= version;
+		data->minorversion	= 0;
+	}
+	return data;
+}
+
 /*
  * Sanity-check a server address provided by the mount command.
  *
@@ -755,15 +778,13 @@ static int nfs_verify_server_address(struct sockaddr *addr)
  * Select between a default port value and a user-specified port value.
  * If a zero value is set, then autobind will be used.
  */
-static void nfs_set_default_port(struct sockaddr *sap, const int parsed_port,
+static void nfs_set_port(struct sockaddr *sap, int *port,
 				 const unsigned short default_port)
 {
-	unsigned short port = default_port;
-
-	if (parsed_port != NFS_UNSPEC_PORT)
-		port = parsed_port;
+	if (*port == NFS_UNSPEC_PORT)
+		*port = default_port;
 
-	rpc_set_port(sap, port);
+	rpc_set_port(sap, *port);
 }
 
 /*
@@ -1232,6 +1253,7 @@ static int nfs_parse_mount_options(char *raw,
 			default:
 				dfprintk(MOUNT, "NFS:   unrecognized "
 						"transport protocol\n");
+				kfree(string);
 				return 0;
 			}
 			break;
@@ -1430,10 +1452,13 @@ static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 	int status;
 
 	if (args->mount_server.version == 0) {
-		if (args->flags & NFS_MOUNT_VER3)
-			args->mount_server.version = NFS_MNT3_VERSION;
-		else
-			args->mount_server.version = NFS_MNT_VERSION;
+		switch (args->version) {
+			default:
+				args->mount_server.version = NFS_MNT3_VERSION;
+				break;
+			case 2:
+				args->mount_server.version = NFS_MNT_VERSION;
+		}
 	}
 	request.version = args->mount_server.version;
 
@@ -1451,7 +1476,7 @@ static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 		args->mount_server.addrlen = args->nfs_server.addrlen;
 	}
 	request.salen = args->mount_server.addrlen;
-	nfs_set_default_port(request.sap, args->mount_server.port, 0);
+	nfs_set_port(request.sap, &args->mount_server.port, 0);
 
 	/*
 	 * Now ask the mount server to map our export path
@@ -1634,20 +1659,6 @@ static int nfs_validate_mount_data(void *options,
 	if (data == NULL)
 		goto out_no_data;
 
-	args->flags		= (NFS_MOUNT_VER3 | NFS_MOUNT_TCP);
-	args->rsize		= NFS_MAX_FILE_IO_SIZE;
-	args->wsize		= NFS_MAX_FILE_IO_SIZE;
-	args->acregmin		= NFS_DEF_ACREGMIN;
-	args->acregmax		= NFS_DEF_ACREGMAX;
-	args->acdirmin		= NFS_DEF_ACDIRMIN;
-	args->acdirmax		= NFS_DEF_ACDIRMAX;
-	args->mount_server.port	= NFS_UNSPEC_PORT;
-	args->nfs_server.port	= NFS_UNSPEC_PORT;
-	args->nfs_server.protocol = XPRT_TRANSPORT_TCP;
-	args->auth_flavors[0]	= RPC_AUTH_UNIX;
-	args->auth_flavor_len	= 1;
-	args->minorversion	= 0;
-
 	switch (data->version) {
 	case 1:
 		data->namlen = 0;
@@ -1755,7 +1766,7 @@ static int nfs_validate_mount_data(void *options,
 			goto out_v4_not_compiled;
 #endif
 
-		nfs_set_default_port(sap, args->nfs_server.port, 0);
+		nfs_set_port(sap, &args->nfs_server.port, 0);
 
 		nfs_set_mount_transport_protocol(args);
 
@@ -1778,7 +1789,7 @@ static int nfs_validate_mount_data(void *options,
 	}
 
 #ifndef CONFIG_NFS_V3
-	if (args->flags & NFS_MOUNT_VER3)
+	if (args->version == 3)
 		goto out_v3_not_compiled;
 #endif /* !CONFIG_NFS_V3 */
 
@@ -1836,9 +1847,10 @@ nfs_compare_remount_data(struct nfs_server *nfss,
 	    data->acdirmin != nfss->acdirmin / HZ ||
 	    data->acdirmax != nfss->acdirmax / HZ ||
 	    data->timeo != (10U * nfss->client->cl_timeout->to_initval / HZ) ||
+	    data->nfs_server.port != nfss->port ||
 	    data->nfs_server.addrlen != nfss->nfs_client->cl_addrlen ||
-	    memcmp(&data->nfs_server.address, &nfss->nfs_client->cl_addr,
-		   data->nfs_server.addrlen) != 0)
+	    !rpc_cmp_addr((struct sockaddr *)&data->nfs_server.address,
+			  (struct sockaddr *)&nfss->nfs_client->cl_addr))
 		return -EINVAL;
 
 	return 0;
@@ -1881,6 +1893,7 @@ nfs_remount(struct super_block *sb, int *flags, char *raw_data)
 	data->acdirmin = nfss->acdirmin / HZ;
 	data->acdirmax = nfss->acdirmax / HZ;
 	data->timeo = 10U * nfss->client->cl_timeout->to_initval / HZ;
+	data->nfs_server.port = nfss->port;
 	data->nfs_server.addrlen = nfss->nfs_client->cl_addrlen;
 	memcpy(&data->nfs_server.address, &nfss->nfs_client->cl_addr,
 		data->nfs_server.addrlen);
@@ -1936,7 +1949,7 @@ static void nfs_fill_super(struct super_block *sb,
 	if (data->bsize)
 		sb->s_blocksize = nfs_block_size(data->bsize, &sb->s_blocksize_bits);
 
-	if (server->flags & NFS_MOUNT_VER3) {
+	if (server->nfs_client->rpc_ops->version == 3) {
 		/* The VFS shouldn't apply the umask to mode bits. We will do
 		 * so ourselves when necessary.
 		 */
@@ -1960,7 +1973,7 @@ static void nfs_clone_super(struct super_block *sb,
 	sb->s_blocksize = old_sb->s_blocksize;
 	sb->s_maxbytes = old_sb->s_maxbytes;
 
-	if (server->flags & NFS_MOUNT_VER3) {
+	if (server->nfs_client->rpc_ops->version == 3) {
 		/* The VFS shouldn't apply the umask to mode bits. We will do
 		 * so ourselves when necessary.
 		 */
@@ -2094,7 +2107,7 @@ static int nfs_get_sb(struct file_system_type *fs_type,
 	};
 	int error = -ENOMEM;
 
-	data = kzalloc(sizeof(*data), GFP_KERNEL);
+	data = nfs_alloc_parsed_mount_data(3);
 	mntfh = kzalloc(sizeof(*mntfh), GFP_KERNEL);
 	if (data == NULL || mntfh == NULL)
 		goto out_free_fh;
@@ -2144,7 +2157,8 @@ static int nfs_get_sb(struct file_system_type *fs_type,
 	if (!s->s_root) {
 		/* initial superblock/root creation */
 		nfs_fill_super(s, data);
-		nfs_fscache_get_super_cookie(s, data);
+		nfs_fscache_get_super_cookie(
+			s, data ? data->fscache_uniq : NULL, NULL);
 	}
 
 	mntroot = nfs_get_root(s, mntfh);
@@ -2190,8 +2204,8 @@ static void nfs_kill_super(struct super_block *s)
 {
 	struct nfs_server *server = NFS_SB(s);
 
-	bdi_unregister(&server->backing_dev_info);
 	kill_anon_super(s);
+	bdi_unregister(&server->backing_dev_info);
 	nfs_fscache_release_super_cookie(s);
 	nfs_free_server(server);
 }
@@ -2245,6 +2259,7 @@ static int nfs_xdev_get_sb(struct file_system_type *fs_type, int flags,
 	if (!s->s_root) {
 		/* initial superblock/root creation */
 		nfs_clone_super(s, data->sb);
+		nfs_fscache_get_super_cookie(s, NULL, data);
 	}
 
 	mntroot = nfs_get_root(s, data->fh);
@@ -2317,7 +2332,7 @@ static int nfs4_validate_text_mount_data(void *options,
 {
 	struct sockaddr *sap = (struct sockaddr *)&args->nfs_server.address;
 
-	nfs_set_default_port(sap, args->nfs_server.port, NFS_PORT);
+	nfs_set_port(sap, &args->nfs_server.port, NFS_PORT);
 
 	nfs_validate_transport_protocol(args);
 
@@ -2362,18 +2377,6 @@ static int nfs4_validate_mount_data(void *options,
 	if (data == NULL)
 		goto out_no_data;
 
-	args->rsize		= NFS_MAX_FILE_IO_SIZE;
-	args->wsize		= NFS_MAX_FILE_IO_SIZE;
-	args->acregmin		= NFS_DEF_ACREGMIN;
-	args->acregmax		= NFS_DEF_ACREGMAX;
-	args->acdirmin		= NFS_DEF_ACDIRMIN;
-	args->acdirmax		= NFS_DEF_ACDIRMAX;
-	args->nfs_server.port	= NFS_UNSPEC_PORT;
-	args->auth_flavors[0]	= RPC_AUTH_UNIX;
-	args->auth_flavor_len	= 1;
-	args->version		= 4;
-	args->minorversion	= 0;
-
 	switch (data->version) {
 	case 1:
 		if (data->host_addrlen > sizeof(args->nfs_server.address))
@@ -2508,7 +2511,8 @@ static int nfs4_remote_get_sb(struct file_system_type *fs_type,
 	if (!s->s_root) {
 		/* initial superblock/root creation */
 		nfs4_fill_super(s);
-		nfs_fscache_get_super_cookie(s, data);
+		nfs_fscache_get_super_cookie(
+			s, data ? data->fscache_uniq : NULL, NULL);
 	}
 
 	mntroot = nfs4_get_root(s, mntfh);
@@ -2656,7 +2660,7 @@ static int nfs4_get_sb(struct file_system_type *fs_type,
 	struct nfs_parsed_mount_data *data;
 	int error = -ENOMEM;
 
-	data = kzalloc(sizeof(*data), GFP_KERNEL);
+	data = nfs_alloc_parsed_mount_data(4);
 	if (data == NULL)
 		goto out_free_data;
 
@@ -2686,7 +2690,6 @@ static void nfs4_kill_super(struct super_block *sb)
 	dprintk("--> %s\n", __func__);
 	nfs_super_return_all_delegations(sb);
 	kill_anon_super(sb);
-	nfs4_renewd_prepare_shutdown(server);
 	nfs_fscache_release_super_cookie(sb);
 	nfs_free_server(server);
 	dprintk("<-- %s\n", __func__);
@@ -2741,6 +2744,7 @@ static int nfs4_xdev_get_sb(struct file_system_type *fs_type, int flags,
 	if (!s->s_root) {
 		/* initial superblock/root creation */
 		nfs4_clone_super(s, data->sb);
+		nfs_fscache_get_super_cookie(s, NULL, data);
 	}
 
 	mntroot = nfs4_get_root(s, data->fh);
@@ -2822,6 +2826,7 @@ static int nfs4_remote_referral_get_sb(struct file_system_type *fs_type,
 	if (!s->s_root) {
 		/* initial superblock/root creation */
 		nfs4_fill_super(s);
+		nfs_fscache_get_super_cookie(s, NULL, data);
 	}
 
 	mntroot = nfs4_get_root(s, &mntfh);
diff --git a/fs/nfsd/export.c b/fs/nfsd/export.c
index d9462643155c..c1c9e035d4a4 100644
--- a/fs/nfsd/export.c
+++ b/fs/nfsd/export.c
@@ -1341,6 +1341,8 @@ exp_pseudoroot(struct svc_rqst *rqstp, struct svc_fh *fhp)
 	if (rv)
 		goto out;
 	rv = check_nfsd_access(exp, rqstp);
+	if (rv)
+		fh_put(fhp);
 out:
 	exp_put(exp);
 	return rv;
@@ -1515,7 +1517,7 @@ static int e_show(struct seq_file *m, void *p)
 	return svc_export_show(m, &svc_export_cache, cp);
 }
 
-struct seq_operations nfs_exports_op = {
+const struct seq_operations nfs_exports_op = {
 	.start	= e_start,
 	.next	= e_next,
 	.stop	= e_stop,
diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
index 01d4ec1c88e0..edf926e1062f 100644
--- a/fs/nfsd/nfs3xdr.c
+++ b/fs/nfsd/nfs3xdr.c
@@ -814,17 +814,6 @@ encode_entry_baggage(struct nfsd3_readdirres *cd, __be32 *p, const char *name,
 	return p;
 }
 
-static __be32 *
-encode_entryplus_baggage(struct nfsd3_readdirres *cd, __be32 *p,
-		struct svc_fh *fhp)
-{
-	p = encode_post_op_attr(cd->rqstp, p, fhp);
-	*p++ = xdr_one;			/* yes, a file handle follows */
-	p = encode_fh(p, fhp);
-	fh_put(fhp);
-	return p;
-}
-
 static int
 compose_entry_fh(struct nfsd3_readdirres *cd, struct svc_fh *fhp,
 		const char *name, int namlen)
@@ -836,29 +825,54 @@ compose_entry_fh(struct nfsd3_readdirres *cd, struct svc_fh *fhp,
 	dparent = cd->fh.fh_dentry;
 	exp  = cd->fh.fh_export;
 
-	fh_init(fhp, NFS3_FHSIZE);
 	if (isdotent(name, namlen)) {
 		if (namlen == 2) {
 			dchild = dget_parent(dparent);
 			if (dchild == dparent) {
 				/* filesystem root - cannot return filehandle for ".." */
 				dput(dchild);
-				return 1;
+				return -ENOENT;
 			}
 		} else
 			dchild = dget(dparent);
 	} else
 		dchild = lookup_one_len(name, dparent, namlen);
 	if (IS_ERR(dchild))
-		return 1;
-	if (d_mountpoint(dchild) ||
-	    fh_compose(fhp, exp, dchild, &cd->fh) != 0 ||
-	    !dchild->d_inode)
-		rv = 1;
+		return -ENOENT;
+	rv = -ENOENT;
+	if (d_mountpoint(dchild))
+		goto out;
+	rv = fh_compose(fhp, exp, dchild, &cd->fh);
+	if (rv)
+		goto out;
+	if (!dchild->d_inode)
+		goto out;
+	rv = 0;
+out:
 	dput(dchild);
 	return rv;
 }
 
+__be32 *encode_entryplus_baggage(struct nfsd3_readdirres *cd, __be32 *p, const char *name, int namlen)
+{
+	struct svc_fh	fh;
+	int err;
+
+	fh_init(&fh, NFS3_FHSIZE);
+	err = compose_entry_fh(cd, &fh, name, namlen);
+	if (err) {
+		*p++ = 0;
+		*p++ = 0;
+		goto out;
+	}
+	p = encode_post_op_attr(cd->rqstp, p, &fh);
+	*p++ = xdr_one;			/* yes, a file handle follows */
+	p = encode_fh(p, &fh);
+out:
+	fh_put(&fh);
+	return p;
+}
+
 /*
  * Encode a directory entry. This one works for both normal readdir
  * and readdirplus.
@@ -929,16 +943,8 @@ encode_entry(struct readdir_cd *ccd, const char *name, int namlen,
 
 		p = encode_entry_baggage(cd, p, name, namlen, ino);
 
-		/* throw in readdirplus baggage */
-		if (plus) {
-			struct svc_fh	fh;
-
-			if (compose_entry_fh(cd, &fh, name, namlen) > 0) {
-				*p++ = 0;
-				*p++ = 0;
-			} else
-				p = encode_entryplus_baggage(cd, p, &fh);
-		}
+		if (plus)
+			p = encode_entryplus_baggage(cd, p, name, namlen);
 		num_entry_words = p - cd->buffer;
 	} else if (cd->rqstp->rq_respages[pn+1] != NULL) {
 		/* temporarily encode entry into next page, then move back to
@@ -951,17 +957,8 @@ encode_entry(struct readdir_cd *ccd, const char *name, int namlen,
 
 		p1 = encode_entry_baggage(cd, p1, name, namlen, ino);
 
-		/* throw in readdirplus baggage */
-		if (plus) {
-			struct svc_fh	fh;
-
-			if (compose_entry_fh(cd, &fh, name, namlen) > 0) {
-				/* zero out the filehandle */
-				*p1++ = 0;
-				*p1++ = 0;
-			} else
-				p1 = encode_entryplus_baggage(cd, p1, &fh);
-		}
+		if (plus)
+			p = encode_entryplus_baggage(cd, p1, name, namlen);
 
 		/* determine entry word length and lengths to go in pages */
 		num_entry_words = p1 - tmp;
diff --git a/fs/nfsd/nfs4acl.c b/fs/nfsd/nfs4acl.c
index 54b8b4140c8f..725d02f210e2 100644
--- a/fs/nfsd/nfs4acl.c
+++ b/fs/nfsd/nfs4acl.c
@@ -321,7 +321,7 @@ _posix_to_nfsv4_one(struct posix_acl *pacl, struct nfs4_acl *acl,
 	deny = ~pas.group & pas.other;
 	if (deny) {
 		ace->type = NFS4_ACE_ACCESS_DENIED_ACE_TYPE;
-		ace->flag = eflag | NFS4_ACE_IDENTIFIER_GROUP;
+		ace->flag = eflag;
 		ace->access_mask = deny_mask_from_posix(deny, flags);
 		ace->whotype = NFS4_ACL_WHO_GROUP;
 		ace++;
@@ -335,7 +335,7 @@ _posix_to_nfsv4_one(struct posix_acl *pacl, struct nfs4_acl *acl,
 		if (deny) {
 			ace->type = NFS4_ACE_ACCESS_DENIED_ACE_TYPE;
 			ace->flag = eflag | NFS4_ACE_IDENTIFIER_GROUP;
-			ace->access_mask = mask_from_posix(deny, flags);
+			ace->access_mask = deny_mask_from_posix(deny, flags);
 			ace->whotype = NFS4_ACL_WHO_NAMED;
 			ace->who = pa->e_id;
 			ace++;
diff --git a/fs/nfsd/nfs4callback.c b/fs/nfsd/nfs4callback.c
index 3fd23f7aceca..24e8d78f8dde 100644
--- a/fs/nfsd/nfs4callback.c
+++ b/fs/nfsd/nfs4callback.c
@@ -43,25 +43,30 @@
 #include <linux/sunrpc/xdr.h>
 #include <linux/sunrpc/svc.h>
 #include <linux/sunrpc/clnt.h>
+#include <linux/sunrpc/svcsock.h>
 #include <linux/nfsd/nfsd.h>
 #include <linux/nfsd/state.h>
 #include <linux/sunrpc/sched.h>
 #include <linux/nfs4.h>
+#include <linux/sunrpc/xprtsock.h>
 
 #define NFSDDBG_FACILITY                NFSDDBG_PROC
 
 #define NFSPROC4_CB_NULL 0
 #define NFSPROC4_CB_COMPOUND 1
+#define NFS4_STATEID_SIZE 16
 
 /* Index of predefined Linux callback client operations */
 
 enum {
-        NFSPROC4_CLNT_CB_NULL = 0,
+	NFSPROC4_CLNT_CB_NULL = 0,
 	NFSPROC4_CLNT_CB_RECALL,
+	NFSPROC4_CLNT_CB_SEQUENCE,
 };
 
 enum nfs_cb_opnum4 {
 	OP_CB_RECALL            = 4,
+	OP_CB_SEQUENCE          = 11,
 };
 
 #define NFS4_MAXTAGLEN		20
@@ -70,17 +75,29 @@ enum nfs_cb_opnum4 {
 #define NFS4_dec_cb_null_sz		0
 #define cb_compound_enc_hdr_sz		4
 #define cb_compound_dec_hdr_sz		(3 + (NFS4_MAXTAGLEN >> 2))
+#define sessionid_sz			(NFS4_MAX_SESSIONID_LEN >> 2)
+#define cb_sequence_enc_sz		(sessionid_sz + 4 +             \
+					1 /* no referring calls list yet */)
+#define cb_sequence_dec_sz		(op_dec_sz + sessionid_sz + 4)
+
 #define op_enc_sz			1
 #define op_dec_sz			2
 #define enc_nfs4_fh_sz			(1 + (NFS4_FHSIZE >> 2))
 #define enc_stateid_sz			(NFS4_STATEID_SIZE >> 2)
 #define NFS4_enc_cb_recall_sz		(cb_compound_enc_hdr_sz +       \
+					cb_sequence_enc_sz +            \
 					1 + enc_stateid_sz +            \
 					enc_nfs4_fh_sz)
 
 #define NFS4_dec_cb_recall_sz		(cb_compound_dec_hdr_sz  +      \
+					cb_sequence_dec_sz +            \
 					op_dec_sz)
 
+struct nfs4_rpc_args {
+	void				*args_op;
+	struct nfsd4_cb_sequence	args_seq;
+};
+
 /*
 * Generic encode routines from fs/nfs/nfs4xdr.c
 */
@@ -137,11 +154,13 @@ xdr_error:                                      \
 } while (0)
 
 struct nfs4_cb_compound_hdr {
-	int		status;
-	u32		ident;
+	/* args */
+	u32		ident;	/* minorversion 0 only */
 	u32		nops;
 	__be32		*nops_p;
 	u32		minorversion;
+	/* res */
+	int		status;
 	u32		taglen;
 	char		*tag;
 };
@@ -238,6 +257,27 @@ encode_cb_recall(struct xdr_stream *xdr, struct nfs4_delegation *dp,
 	hdr->nops++;
 }
 
+static void
+encode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *args,
+		   struct nfs4_cb_compound_hdr *hdr)
+{
+	__be32 *p;
+
+	if (hdr->minorversion == 0)
+		return;
+
+	RESERVE_SPACE(1 + NFS4_MAX_SESSIONID_LEN + 20);
+
+	WRITE32(OP_CB_SEQUENCE);
+	WRITEMEM(args->cbs_clp->cl_sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	WRITE32(args->cbs_clp->cl_cb_seq_nr);
+	WRITE32(0);		/* slotid, always 0 */
+	WRITE32(0);		/* highest slotid always 0 */
+	WRITE32(0);		/* cachethis always 0 */
+	WRITE32(0); /* FIXME: support referring_call_lists */
+	hdr->nops++;
+}
+
 static int
 nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -249,15 +289,19 @@ nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_delegation *args)
+nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p,
+		struct nfs4_rpc_args *rpc_args)
 {
 	struct xdr_stream xdr;
+	struct nfs4_delegation *args = rpc_args->args_op;
 	struct nfs4_cb_compound_hdr hdr = {
 		.ident = args->dl_ident,
+		.minorversion = rpc_args->args_seq.cbs_minorversion,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_cb_compound_hdr(&xdr, &hdr);
+	encode_cb_sequence(&xdr, &rpc_args->args_seq, &hdr);
 	encode_cb_recall(&xdr, args, &hdr);
 	encode_cb_nops(&hdr);
 	return 0;
@@ -299,6 +343,57 @@ decode_cb_op_hdr(struct xdr_stream *xdr, enum nfs_opnum4 expected)
 	return 0;
 }
 
+/*
+ * Our current back channel implmentation supports a single backchannel
+ * with a single slot.
+ */
+static int
+decode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *res,
+		   struct rpc_rqst *rqstp)
+{
+	struct nfs4_sessionid id;
+	int status;
+	u32 dummy;
+	__be32 *p;
+
+	if (res->cbs_minorversion == 0)
+		return 0;
+
+	status = decode_cb_op_hdr(xdr, OP_CB_SEQUENCE);
+	if (status)
+		return status;
+
+	/*
+	 * If the server returns different values for sessionID, slotID or
+	 * sequence number, the server is looney tunes.
+	 */
+	status = -ESERVERFAULT;
+
+	READ_BUF(NFS4_MAX_SESSIONID_LEN + 16);
+	memcpy(id.data, p, NFS4_MAX_SESSIONID_LEN);
+	p += XDR_QUADLEN(NFS4_MAX_SESSIONID_LEN);
+	if (memcmp(id.data, res->cbs_clp->cl_sessionid.data,
+		   NFS4_MAX_SESSIONID_LEN)) {
+		dprintk("%s Invalid session id\n", __func__);
+		goto out;
+	}
+	READ32(dummy);
+	if (dummy != res->cbs_clp->cl_cb_seq_nr) {
+		dprintk("%s Invalid sequence number\n", __func__);
+		goto out;
+	}
+	READ32(dummy); 	/* slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid slotid\n", __func__);
+		goto out;
+	}
+	/* FIXME: process highest slotid and target highest slotid */
+	status = 0;
+out:
+	return status;
+}
+
+
 static int
 nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -306,7 +401,8 @@ nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
+nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p,
+		struct nfsd4_cb_sequence *seq)
 {
 	struct xdr_stream xdr;
 	struct nfs4_cb_compound_hdr hdr;
@@ -316,6 +412,11 @@ nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
 	status = decode_cb_compound_hdr(&xdr, &hdr);
 	if (status)
 		goto out;
+	if (seq) {
+		status = decode_cb_sequence(&xdr, seq, rqstp);
+		if (status)
+			goto out;
+	}
 	status = decode_cb_op_hdr(&xdr, OP_CB_RECALL);
 out:
 	return status;
@@ -377,16 +478,15 @@ static int max_cb_time(void)
 
 int setup_callback_client(struct nfs4_client *clp)
 {
-	struct sockaddr_in	addr;
 	struct nfs4_cb_conn *cb = &clp->cl_cb_conn;
 	struct rpc_timeout	timeparms = {
 		.to_initval	= max_cb_time(),
 		.to_retries	= 0,
 	};
 	struct rpc_create_args args = {
-		.protocol	= IPPROTO_TCP,
-		.address	= (struct sockaddr *)&addr,
-		.addrsize	= sizeof(addr),
+		.protocol	= XPRT_TRANSPORT_TCP,
+		.address	= (struct sockaddr *) &cb->cb_addr,
+		.addrsize	= cb->cb_addrlen,
 		.timeout	= &timeparms,
 		.program	= &cb_program,
 		.prognumber	= cb->cb_prog,
@@ -399,13 +499,10 @@ int setup_callback_client(struct nfs4_client *clp)
 
 	if (!clp->cl_principal && (clp->cl_flavor >= RPC_AUTH_GSS_KRB5))
 		return -EINVAL;
-
-	/* Initialize address */
-	memset(&addr, 0, sizeof(addr));
-	addr.sin_family = AF_INET;
-	addr.sin_port = htons(cb->cb_port);
-	addr.sin_addr.s_addr = htonl(cb->cb_addr);
-
+	if (cb->cb_minorversion) {
+		args.bc_xprt = clp->cl_cb_xprt;
+		args.protocol = XPRT_TRANSPORT_BC_TCP;
+	}
 	/* Create RPC client */
 	client = rpc_create(&args);
 	if (IS_ERR(client)) {
@@ -439,42 +536,29 @@ static const struct rpc_call_ops nfsd4_cb_probe_ops = {
 	.rpc_call_done = nfsd4_cb_probe_done,
 };
 
-static struct rpc_cred *lookup_cb_cred(struct nfs4_cb_conn *cb)
-{
-	struct auth_cred acred = {
-		.machine_cred = 1
-	};
+static struct rpc_cred *callback_cred;
 
-	/*
-	 * Note in the gss case this doesn't actually have to wait for a
-	 * gss upcall (or any calls to the client); this just creates a
-	 * non-uptodate cred which the rpc state machine will fill in with
-	 * a refresh_upcall later.
-	 */
-	return rpcauth_lookup_credcache(cb->cb_client->cl_auth, &acred,
-							RPCAUTH_LOOKUP_NEW);
+int set_callback_cred(void)
+{
+	callback_cred = rpc_lookup_machine_cred();
+	if (!callback_cred)
+		return -ENOMEM;
+	return 0;
 }
 
+
 void do_probe_callback(struct nfs4_client *clp)
 {
 	struct nfs4_cb_conn *cb = &clp->cl_cb_conn;
 	struct rpc_message msg = {
 		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
 		.rpc_argp       = clp,
+		.rpc_cred	= callback_cred
 	};
-	struct rpc_cred *cred;
 	int status;
 
-	cred = lookup_cb_cred(cb);
-	if (IS_ERR(cred)) {
-		status = PTR_ERR(cred);
-		goto out;
-	}
-	cb->cb_cred = cred;
-	msg.rpc_cred = cb->cb_cred;
 	status = rpc_call_async(cb->cb_client, &msg, RPC_TASK_SOFT,
 				&nfsd4_cb_probe_ops, (void *)clp);
-out:
 	if (status) {
 		warn_no_callback_path(clp, status);
 		put_nfs4_client(clp);
@@ -503,11 +587,95 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	do_probe_callback(clp);
 }
 
+/*
+ * There's currently a single callback channel slot.
+ * If the slot is available, then mark it busy.  Otherwise, set the
+ * thread for sleeping on the callback RPC wait queue.
+ */
+static int nfsd41_cb_setup_sequence(struct nfs4_client *clp,
+		struct rpc_task *task)
+{
+	struct nfs4_rpc_args *args = task->tk_msg.rpc_argp;
+	u32 *ptr = (u32 *)clp->cl_sessionid.data;
+	int status = 0;
+
+	dprintk("%s: %u:%u:%u:%u\n", __func__,
+		ptr[0], ptr[1], ptr[2], ptr[3]);
+
+	if (test_and_set_bit(0, &clp->cl_cb_slot_busy) != 0) {
+		rpc_sleep_on(&clp->cl_cb_waitq, task, NULL);
+		dprintk("%s slot is busy\n", __func__);
+		status = -EAGAIN;
+		goto out;
+	}
+
+	/*
+	 * We'll need the clp during XDR encoding and decoding,
+	 * and the sequence during decoding to verify the reply
+	 */
+	args->args_seq.cbs_clp = clp;
+	task->tk_msg.rpc_resp = &args->args_seq;
+
+out:
+	dprintk("%s status=%d\n", __func__, status);
+	return status;
+}
+
+/*
+ * TODO: cb_sequence should support referring call lists, cachethis, multiple
+ * slots, and mark callback channel down on communication errors.
+ */
+static void nfsd4_cb_prepare(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_delegation *dp = calldata;
+	struct nfs4_client *clp = dp->dl_client;
+	struct nfs4_rpc_args *args = task->tk_msg.rpc_argp;
+	u32 minorversion = clp->cl_cb_conn.cb_minorversion;
+	int status = 0;
+
+	args->args_seq.cbs_minorversion = minorversion;
+	if (minorversion) {
+		status = nfsd41_cb_setup_sequence(clp, task);
+		if (status) {
+			if (status != -EAGAIN) {
+				/* terminate rpc task */
+				task->tk_status = status;
+				task->tk_action = NULL;
+			}
+			return;
+		}
+	}
+	rpc_call_start(task);
+}
+
+static void nfsd4_cb_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_delegation *dp = calldata;
+	struct nfs4_client *clp = dp->dl_client;
+
+	dprintk("%s: minorversion=%d\n", __func__,
+		clp->cl_cb_conn.cb_minorversion);
+
+	if (clp->cl_cb_conn.cb_minorversion) {
+		/* No need for lock, access serialized in nfsd4_cb_prepare */
+		++clp->cl_cb_seq_nr;
+		clear_bit(0, &clp->cl_cb_slot_busy);
+		rpc_wake_up_next(&clp->cl_cb_waitq);
+		dprintk("%s: freed slot, new seqid=%d\n", __func__,
+			clp->cl_cb_seq_nr);
+
+		/* We're done looking into the sequence information */
+		task->tk_msg.rpc_resp = NULL;
+	}
+}
+
 static void nfsd4_cb_recall_done(struct rpc_task *task, void *calldata)
 {
 	struct nfs4_delegation *dp = calldata;
 	struct nfs4_client *clp = dp->dl_client;
 
+	nfsd4_cb_done(task, calldata);
+
 	switch (task->tk_status) {
 	case -EIO:
 		/* Network partition? */
@@ -520,16 +688,19 @@ static void nfsd4_cb_recall_done(struct rpc_task *task, void *calldata)
 		break;
 	default:
 		/* success, or error we can't handle */
-		return;
+		goto done;
 	}
 	if (dp->dl_retries--) {
 		rpc_delay(task, 2*HZ);
 		task->tk_status = 0;
 		rpc_restart_call(task);
+		return;
 	} else {
 		atomic_set(&clp->cl_cb_conn.cb_set, 0);
 		warn_no_callback_path(clp, task->tk_status);
 	}
+done:
+	kfree(task->tk_msg.rpc_argp);
 }
 
 static void nfsd4_cb_recall_release(void *calldata)
@@ -542,6 +713,7 @@ static void nfsd4_cb_recall_release(void *calldata)
 }
 
 static const struct rpc_call_ops nfsd4_cb_recall_ops = {
+	.rpc_call_prepare = nfsd4_cb_prepare,
 	.rpc_call_done = nfsd4_cb_recall_done,
 	.rpc_release = nfsd4_cb_recall_release,
 };
@@ -554,17 +726,24 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 {
 	struct nfs4_client *clp = dp->dl_client;
 	struct rpc_clnt *clnt = clp->cl_cb_conn.cb_client;
+	struct nfs4_rpc_args *args;
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_RECALL],
-		.rpc_argp = dp,
-		.rpc_cred = clp->cl_cb_conn.cb_cred
+		.rpc_cred = callback_cred
 	};
-	int status;
+	int status = -ENOMEM;
 
+	args = kzalloc(sizeof(*args), GFP_KERNEL);
+	if (!args)
+		goto out;
+	args->args_op = dp;
+	msg.rpc_argp = args;
 	dp->dl_retries = 1;
 	status = rpc_call_async(clnt, &msg, RPC_TASK_SOFT,
 				&nfsd4_cb_recall_ops, dp);
+out:
 	if (status) {
+		kfree(args);
 		put_nfs4_client(clp);
 		nfs4_put_delegation(dp);
 	}
diff --git a/fs/nfsd/nfs4idmap.c b/fs/nfsd/nfs4idmap.c
index cdfa86fa1471..ba2c199592fd 100644
--- a/fs/nfsd/nfs4idmap.c
+++ b/fs/nfsd/nfs4idmap.c
@@ -38,7 +38,6 @@
 #include <linux/init.h>
 
 #include <linux/mm.h>
-#include <linux/utsname.h>
 #include <linux/errno.h>
 #include <linux/string.h>
 #include <linux/sunrpc/clnt.h>
diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
index 7c8801769a3c..bebc0c2e1b0a 100644
--- a/fs/nfsd/nfs4proc.c
+++ b/fs/nfsd/nfs4proc.c
@@ -68,7 +68,6 @@ check_attr_support(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		   u32 *bmval, u32 *writable)
 {
 	struct dentry *dentry = cstate->current_fh.fh_dentry;
-	struct svc_export *exp = cstate->current_fh.fh_export;
 
 	/*
 	 * Check about attributes are supported by the NFSv4 server or not.
@@ -80,17 +79,13 @@ check_attr_support(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		return nfserr_attrnotsupp;
 
 	/*
-	 * Check FATTR4_WORD0_ACL & FATTR4_WORD0_FS_LOCATIONS can be supported
+	 * Check FATTR4_WORD0_ACL can be supported
 	 * in current environment or not.
 	 */
 	if (bmval[0] & FATTR4_WORD0_ACL) {
 		if (!IS_POSIXACL(dentry->d_inode))
 			return nfserr_attrnotsupp;
 	}
-	if (bmval[0] & FATTR4_WORD0_FS_LOCATIONS) {
-		if (exp->ex_fslocs.locations == NULL)
-			return nfserr_attrnotsupp;
-	}
 
 	/*
 	 * According to spec, read-only attributes return ERR_INVAL.
@@ -123,6 +118,35 @@ nfsd4_check_open_attributes(struct svc_rqst *rqstp,
 	return status;
 }
 
+static int
+is_create_with_attrs(struct nfsd4_open *open)
+{
+	return open->op_create == NFS4_OPEN_CREATE
+		&& (open->op_createmode == NFS4_CREATE_UNCHECKED
+		    || open->op_createmode == NFS4_CREATE_GUARDED
+		    || open->op_createmode == NFS4_CREATE_EXCLUSIVE4_1);
+}
+
+/*
+ * if error occurs when setting the acl, just clear the acl bit
+ * in the returned attr bitmap.
+ */
+static void
+do_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
+		struct nfs4_acl *acl, u32 *bmval)
+{
+	__be32 status;
+
+	status = nfsd4_set_nfs4_acl(rqstp, fhp, acl);
+	if (status)
+		/*
+		 * We should probably fail the whole open at this point,
+		 * but we've already created the file, so it's too late;
+		 * So this seems the least of evils:
+		 */
+		bmval[0] &= ~FATTR4_WORD0_ACL;
+}
+
 static inline void
 fh_dup2(struct svc_fh *dst, struct svc_fh *src)
 {
@@ -206,6 +230,9 @@ do_open_lookup(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_o
 	if (status)
 		goto out;
 
+	if (is_create_with_attrs(open) && open->op_acl != NULL)
+		do_set_nfs4_acl(rqstp, &resfh, open->op_acl, open->op_bmval);
+
 	set_change_info(&open->op_cinfo, current_fh);
 	fh_dup2(current_fh, &resfh);
 
@@ -536,12 +563,17 @@ nfsd4_create(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		status = nfserr_badtype;
 	}
 
-	if (!status) {
-		fh_unlock(&cstate->current_fh);
-		set_change_info(&create->cr_cinfo, &cstate->current_fh);
-		fh_dup2(&cstate->current_fh, &resfh);
-	}
+	if (status)
+		goto out;
+
+	if (create->cr_acl != NULL)
+		do_set_nfs4_acl(rqstp, &resfh, create->cr_acl,
+				create->cr_bmval);
 
+	fh_unlock(&cstate->current_fh);
+	set_change_info(&create->cr_cinfo, &cstate->current_fh);
+	fh_dup2(&cstate->current_fh, &resfh);
+out:
 	fh_put(&resfh);
 	return status;
 }
@@ -947,34 +979,6 @@ static struct nfsd4_operation nfsd4_ops[];
 static const char *nfsd4_op_name(unsigned opnum);
 
 /*
- * This is a replay of a compound for which no cache entry pages
- * were used. Encode the sequence operation, and if cachethis is FALSE
- * encode the uncache rep error on the next operation.
- */
-static __be32
-nfsd4_enc_uncached_replay(struct nfsd4_compoundargs *args,
-			 struct nfsd4_compoundres *resp)
-{
-	struct nfsd4_op *op;
-
-	dprintk("--> %s resp->opcnt %d ce_cachethis %u \n", __func__,
-		resp->opcnt, resp->cstate.slot->sl_cache_entry.ce_cachethis);
-
-	/* Encode the replayed sequence operation */
-	BUG_ON(resp->opcnt != 1);
-	op = &args->ops[resp->opcnt - 1];
-	nfsd4_encode_operation(resp, op);
-
-	/*return nfserr_retry_uncached_rep in next operation. */
-	if (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0) {
-		op = &args->ops[resp->opcnt++];
-		op->status = nfserr_retry_uncached_rep;
-		nfsd4_encode_operation(resp, op);
-	}
-	return op->status;
-}
-
-/*
  * Enforce NFSv4.1 COMPOUND ordering rules.
  *
  * TODO:
@@ -1083,13 +1087,10 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 			BUG_ON(op->status == nfs_ok);
 
 encode_op:
-		/* Only from SEQUENCE or CREATE_SESSION */
+		/* Only from SEQUENCE */
 		if (resp->cstate.status == nfserr_replay_cache) {
 			dprintk("%s NFS4.1 replay from cache\n", __func__);
-			if (nfsd4_not_cached(resp))
-				status = nfsd4_enc_uncached_replay(args, resp);
-			else
-				status = op->status;
+			status = op->status;
 			goto out;
 		}
 		if (op->status == nfserr_replay_me) {
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 980a216a48c8..2153f9bdbebd 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -55,6 +55,7 @@
 #include <linux/lockd/bind.h>
 #include <linux/module.h>
 #include <linux/sunrpc/svcauth_gss.h>
+#include <linux/sunrpc/clnt.h>
 
 #define NFSDDBG_FACILITY                NFSDDBG_PROC
 
@@ -413,36 +414,65 @@ gen_sessionid(struct nfsd4_session *ses)
 }
 
 /*
- * Give the client the number of slots it requests bound by
- * NFSD_MAX_SLOTS_PER_SESSION and by sv_drc_max_pages.
+ * The protocol defines ca_maxresponssize_cached to include the size of
+ * the rpc header, but all we need to cache is the data starting after
+ * the end of the initial SEQUENCE operation--the rest we regenerate
+ * each time.  Therefore we can advertise a ca_maxresponssize_cached
+ * value that is the number of bytes in our cache plus a few additional
+ * bytes.  In order to stay on the safe side, and not promise more than
+ * we can cache, those additional bytes must be the minimum possible: 24
+ * bytes of rpc header (xid through accept state, with AUTH_NULL
+ * verifier), 12 for the compound header (with zero-length tag), and 44
+ * for the SEQUENCE op response:
+ */
+#define NFSD_MIN_HDR_SEQ_SZ  (24 + 12 + 44)
+
+/*
+ * Give the client the number of ca_maxresponsesize_cached slots it
+ * requests, of size bounded by NFSD_SLOT_CACHE_SIZE,
+ * NFSD_MAX_MEM_PER_SESSION, and nfsd_drc_max_mem. Do not allow more
+ * than NFSD_MAX_SLOTS_PER_SESSION.
  *
- * If we run out of pages (sv_drc_pages_used == sv_drc_max_pages) we
- * should (up to a point) re-negotiate active sessions and reduce their
- * slot usage to make rooom for new connections. For now we just fail the
- * create session.
+ * If we run out of reserved DRC memory we should (up to a point)
+ * re-negotiate active sessions and reduce their slot usage to make
+ * rooom for new connections. For now we just fail the create session.
  */
-static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan)
+static int set_forechannel_drc_size(struct nfsd4_channel_attrs *fchan)
 {
-	int status = 0, np = fchan->maxreqs * NFSD_PAGES_PER_SLOT;
+	int mem, size = fchan->maxresp_cached;
 
 	if (fchan->maxreqs < 1)
 		return nfserr_inval;
-	else if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION)
-		fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION;
 
-	spin_lock(&nfsd_serv->sv_lock);
-	if (np + nfsd_serv->sv_drc_pages_used > nfsd_serv->sv_drc_max_pages)
-		np = nfsd_serv->sv_drc_max_pages - nfsd_serv->sv_drc_pages_used;
-	nfsd_serv->sv_drc_pages_used += np;
-	spin_unlock(&nfsd_serv->sv_lock);
+	if (size < NFSD_MIN_HDR_SEQ_SZ)
+		size = NFSD_MIN_HDR_SEQ_SZ;
+	size -= NFSD_MIN_HDR_SEQ_SZ;
+	if (size > NFSD_SLOT_CACHE_SIZE)
+		size = NFSD_SLOT_CACHE_SIZE;
+
+	/* bound the maxreqs by NFSD_MAX_MEM_PER_SESSION */
+	mem = fchan->maxreqs * size;
+	if (mem > NFSD_MAX_MEM_PER_SESSION) {
+		fchan->maxreqs = NFSD_MAX_MEM_PER_SESSION / size;
+		if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION)
+			fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION;
+		mem = fchan->maxreqs * size;
+	}
 
-	if (np <= 0) {
-		status = nfserr_resource;
-		fchan->maxreqs = 0;
-	} else
-		fchan->maxreqs = np / NFSD_PAGES_PER_SLOT;
+	spin_lock(&nfsd_drc_lock);
+	/* bound the total session drc memory ussage */
+	if (mem + nfsd_drc_mem_used > nfsd_drc_max_mem) {
+		fchan->maxreqs = (nfsd_drc_max_mem - nfsd_drc_mem_used) / size;
+		mem = fchan->maxreqs * size;
+	}
+	nfsd_drc_mem_used += mem;
+	spin_unlock(&nfsd_drc_lock);
 
-	return status;
+	if (fchan->maxreqs == 0)
+		return nfserr_serverfault;
+
+	fchan->maxresp_cached = size + NFSD_MIN_HDR_SEQ_SZ;
+	return 0;
 }
 
 /*
@@ -466,36 +496,41 @@ static int init_forechannel_attrs(struct svc_rqst *rqstp,
 		fchan->maxresp_sz = maxcount;
 	session_fchan->maxresp_sz = fchan->maxresp_sz;
 
-	/* Set the max response cached size our default which is
-	 * a multiple of PAGE_SIZE and small */
-	session_fchan->maxresp_cached = NFSD_PAGES_PER_SLOT * PAGE_SIZE;
-	fchan->maxresp_cached = session_fchan->maxresp_cached;
-
 	/* Use the client's maxops if possible */
 	if (fchan->maxops > NFSD_MAX_OPS_PER_COMPOUND)
 		fchan->maxops = NFSD_MAX_OPS_PER_COMPOUND;
 	session_fchan->maxops = fchan->maxops;
 
-	/* try to use the client requested number of slots */
-	if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION)
-		fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION;
-
 	/* FIXME: Error means no more DRC pages so the server should
 	 * recover pages from existing sessions. For now fail session
 	 * creation.
 	 */
-	status = set_forechannel_maxreqs(fchan);
+	status = set_forechannel_drc_size(fchan);
 
+	session_fchan->maxresp_cached = fchan->maxresp_cached;
 	session_fchan->maxreqs = fchan->maxreqs;
+
+	dprintk("%s status %d\n", __func__, status);
 	return status;
 }
 
+static void
+free_session_slots(struct nfsd4_session *ses)
+{
+	int i;
+
+	for (i = 0; i < ses->se_fchannel.maxreqs; i++)
+		kfree(ses->se_slots[i]);
+}
+
 static int
 alloc_init_session(struct svc_rqst *rqstp, struct nfs4_client *clp,
 		   struct nfsd4_create_session *cses)
 {
 	struct nfsd4_session *new, tmp;
-	int idx, status = nfserr_resource, slotsize;
+	struct nfsd4_slot *sp;
+	int idx, slotsize, cachesize, i;
+	int status;
 
 	memset(&tmp, 0, sizeof(tmp));
 
@@ -506,14 +541,27 @@ alloc_init_session(struct svc_rqst *rqstp, struct nfs4_client *clp,
 	if (status)
 		goto out;
 
-	/* allocate struct nfsd4_session and slot table in one piece */
-	slotsize = tmp.se_fchannel.maxreqs * sizeof(struct nfsd4_slot);
+	BUILD_BUG_ON(NFSD_MAX_SLOTS_PER_SESSION * sizeof(struct nfsd4_slot)
+		     + sizeof(struct nfsd4_session) > PAGE_SIZE);
+
+	status = nfserr_serverfault;
+	/* allocate struct nfsd4_session and slot table pointers in one piece */
+	slotsize = tmp.se_fchannel.maxreqs * sizeof(struct nfsd4_slot *);
 	new = kzalloc(sizeof(*new) + slotsize, GFP_KERNEL);
 	if (!new)
 		goto out;
 
 	memcpy(new, &tmp, sizeof(*new));
 
+	/* allocate each struct nfsd4_slot and data cache in one piece */
+	cachesize = new->se_fchannel.maxresp_cached - NFSD_MIN_HDR_SEQ_SZ;
+	for (i = 0; i < new->se_fchannel.maxreqs; i++) {
+		sp = kzalloc(sizeof(*sp) + cachesize, GFP_KERNEL);
+		if (!sp)
+			goto out_free;
+		new->se_slots[i] = sp;
+	}
+
 	new->se_client = clp;
 	gen_sessionid(new);
 	idx = hash_sessionid(&new->se_sessionid);
@@ -530,6 +578,10 @@ alloc_init_session(struct svc_rqst *rqstp, struct nfs4_client *clp,
 	status = nfs_ok;
 out:
 	return status;
+out_free:
+	free_session_slots(new);
+	kfree(new);
+	goto out;
 }
 
 /* caller must hold sessionid_lock */
@@ -572,19 +624,16 @@ release_session(struct nfsd4_session *ses)
 	nfsd4_put_session(ses);
 }
 
-static void nfsd4_release_respages(struct page **respages, short resused);
-
 void
 free_session(struct kref *kref)
 {
 	struct nfsd4_session *ses;
-	int i;
 
 	ses = container_of(kref, struct nfsd4_session, se_ref);
-	for (i = 0; i < ses->se_fchannel.maxreqs; i++) {
-		struct nfsd4_cache_entry *e = &ses->se_slots[i].sl_cache_entry;
-		nfsd4_release_respages(e->ce_respages, e->ce_resused);
-	}
+	spin_lock(&nfsd_drc_lock);
+	nfsd_drc_mem_used -= ses->se_fchannel.maxreqs * NFSD_SLOT_CACHE_SIZE;
+	spin_unlock(&nfsd_drc_lock);
+	free_session_slots(ses);
 	kfree(ses);
 }
 
@@ -647,18 +696,14 @@ shutdown_callback_client(struct nfs4_client *clp)
 		clp->cl_cb_conn.cb_client = NULL;
 		rpc_shutdown_client(clnt);
 	}
-	if (clp->cl_cb_conn.cb_cred) {
-		put_rpccred(clp->cl_cb_conn.cb_cred);
-		clp->cl_cb_conn.cb_cred = NULL;
-	}
 }
 
 static inline void
 free_client(struct nfs4_client *clp)
 {
 	shutdown_callback_client(clp);
-	nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages,
-			     clp->cl_slot.sl_cache_entry.ce_resused);
+	if (clp->cl_cb_xprt)
+		svc_xprt_put(clp->cl_cb_xprt);
 	if (clp->cl_cred.cr_group_info)
 		put_group_info(clp->cl_cred.cr_group_info);
 	kfree(clp->cl_principal);
@@ -714,25 +759,6 @@ expire_client(struct nfs4_client *clp)
 	put_nfs4_client(clp);
 }
 
-static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
-{
-	struct nfs4_client *clp;
-
-	clp = alloc_client(name);
-	if (clp == NULL)
-		return NULL;
-	memcpy(clp->cl_recdir, recdir, HEXDIR_LEN);
-	atomic_set(&clp->cl_count, 1);
-	atomic_set(&clp->cl_cb_conn.cb_set, 0);
-	INIT_LIST_HEAD(&clp->cl_idhash);
-	INIT_LIST_HEAD(&clp->cl_strhash);
-	INIT_LIST_HEAD(&clp->cl_openowners);
-	INIT_LIST_HEAD(&clp->cl_delegations);
-	INIT_LIST_HEAD(&clp->cl_sessions);
-	INIT_LIST_HEAD(&clp->cl_lru);
-	return clp;
-}
-
 static void copy_verf(struct nfs4_client *target, nfs4_verifier *source)
 {
 	memcpy(target->cl_verifier.data, source->data,
@@ -795,6 +821,46 @@ static void gen_confirm(struct nfs4_client *clp)
 	*p++ = i++;
 }
 
+static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir,
+		struct svc_rqst *rqstp, nfs4_verifier *verf)
+{
+	struct nfs4_client *clp;
+	struct sockaddr *sa = svc_addr(rqstp);
+	char *princ;
+
+	clp = alloc_client(name);
+	if (clp == NULL)
+		return NULL;
+
+	princ = svc_gss_principal(rqstp);
+	if (princ) {
+		clp->cl_principal = kstrdup(princ, GFP_KERNEL);
+		if (clp->cl_principal == NULL) {
+			free_client(clp);
+			return NULL;
+		}
+	}
+
+	memcpy(clp->cl_recdir, recdir, HEXDIR_LEN);
+	atomic_set(&clp->cl_count, 1);
+	atomic_set(&clp->cl_cb_conn.cb_set, 0);
+	INIT_LIST_HEAD(&clp->cl_idhash);
+	INIT_LIST_HEAD(&clp->cl_strhash);
+	INIT_LIST_HEAD(&clp->cl_openowners);
+	INIT_LIST_HEAD(&clp->cl_delegations);
+	INIT_LIST_HEAD(&clp->cl_sessions);
+	INIT_LIST_HEAD(&clp->cl_lru);
+	clear_bit(0, &clp->cl_cb_slot_busy);
+	rpc_init_wait_queue(&clp->cl_cb_waitq, "Backchannel slot table");
+	copy_verf(clp, verf);
+	rpc_copy_addr((struct sockaddr *) &clp->cl_addr, sa);
+	clp->cl_flavor = rqstp->rq_flavor;
+	copy_cred(&clp->cl_cred, &rqstp->rq_cred);
+	gen_confirm(clp);
+
+	return clp;
+}
+
 static int check_name(struct xdr_netobj name)
 {
 	if (name.len == 0) 
@@ -902,93 +968,40 @@ find_unconfirmed_client_by_str(const char *dname, unsigned int hashval,
 	return NULL;
 }
 
-/* a helper function for parse_callback */
-static int
-parse_octet(unsigned int *lenp, char **addrp)
-{
-	unsigned int len = *lenp;
-	char *p = *addrp;
-	int n = -1;
-	char c;
-
-	for (;;) {
-		if (!len)
-			break;
-		len--;
-		c = *p++;
-		if (c == '.')
-			break;
-		if ((c < '0') || (c > '9')) {
-			n = -1;
-			break;
-		}
-		if (n < 0)
-			n = 0;
-		n = (n * 10) + (c - '0');
-		if (n > 255) {
-			n = -1;
-			break;
-		}
-	}
-	*lenp = len;
-	*addrp = p;
-	return n;
-}
-
-/* parse and set the setclientid ipv4 callback address */
-static int
-parse_ipv4(unsigned int addr_len, char *addr_val, unsigned int *cbaddrp, unsigned short *cbportp)
-{
-	int temp = 0;
-	u32 cbaddr = 0;
-	u16 cbport = 0;
-	u32 addrlen = addr_len;
-	char *addr = addr_val;
-	int i, shift;
-
-	/* ipaddress */
-	shift = 24;
-	for(i = 4; i > 0  ; i--) {
-		if ((temp = parse_octet(&addrlen, &addr)) < 0) {
-			return 0;
-		}
-		cbaddr |= (temp << shift);
-		if (shift > 0)
-		shift -= 8;
-	}
-	*cbaddrp = cbaddr;
-
-	/* port */
-	shift = 8;
-	for(i = 2; i > 0  ; i--) {
-		if ((temp = parse_octet(&addrlen, &addr)) < 0) {
-			return 0;
-		}
-		cbport |= (temp << shift);
-		if (shift > 0)
-			shift -= 8;
-	}
-	*cbportp = cbport;
-	return 1;
-}
-
 static void
-gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se)
+gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, u32 scopeid)
 {
 	struct nfs4_cb_conn *cb = &clp->cl_cb_conn;
-
-	/* Currently, we only support tcp for the callback channel */
-	if ((se->se_callback_netid_len != 3) || memcmp((char *)se->se_callback_netid_val, "tcp", 3))
+	unsigned short expected_family;
+
+	/* Currently, we only support tcp and tcp6 for the callback channel */
+	if (se->se_callback_netid_len == 3 &&
+	    !memcmp(se->se_callback_netid_val, "tcp", 3))
+		expected_family = AF_INET;
+	else if (se->se_callback_netid_len == 4 &&
+		 !memcmp(se->se_callback_netid_val, "tcp6", 4))
+		expected_family = AF_INET6;
+	else
 		goto out_err;
 
-	if ( !(parse_ipv4(se->se_callback_addr_len, se->se_callback_addr_val,
-	                 &cb->cb_addr, &cb->cb_port)))
+	cb->cb_addrlen = rpc_uaddr2sockaddr(se->se_callback_addr_val,
+					    se->se_callback_addr_len,
+					    (struct sockaddr *) &cb->cb_addr,
+					    sizeof(cb->cb_addr));
+
+	if (!cb->cb_addrlen || cb->cb_addr.ss_family != expected_family)
 		goto out_err;
+
+	if (cb->cb_addr.ss_family == AF_INET6)
+		((struct sockaddr_in6 *) &cb->cb_addr)->sin6_scope_id = scopeid;
+
 	cb->cb_minorversion = 0;
 	cb->cb_prog = se->se_callback_prog;
 	cb->cb_ident = se->se_callback_ident;
 	return;
 out_err:
+	cb->cb_addr.ss_family = AF_UNSPEC;
+	cb->cb_addrlen = 0;
 	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
 		"will not receive delegations\n",
 		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);
@@ -996,175 +1009,87 @@ out_err:
 	return;
 }
 
-void
-nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp)
-{
-	struct nfsd4_compoundres *resp = rqstp->rq_resp;
-
-	resp->cstate.statp = statp;
-}
-
 /*
- * Dereference the result pages.
+ * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
  */
-static void
-nfsd4_release_respages(struct page **respages, short resused)
+void
+nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
 {
-	int i;
+	struct nfsd4_slot *slot = resp->cstate.slot;
+	unsigned int base;
 
-	dprintk("--> %s\n", __func__);
-	for (i = 0; i < resused; i++) {
-		if (!respages[i])
-			continue;
-		put_page(respages[i]);
-		respages[i] = NULL;
-	}
-}
+	dprintk("--> %s slot %p\n", __func__, slot);
 
-static void
-nfsd4_copy_pages(struct page **topages, struct page **frompages, short count)
-{
-	int i;
+	slot->sl_opcnt = resp->opcnt;
+	slot->sl_status = resp->cstate.status;
 
-	for (i = 0; i < count; i++) {
-		topages[i] = frompages[i];
-		if (!topages[i])
-			continue;
-		get_page(topages[i]);
+	if (nfsd4_not_cached(resp)) {
+		slot->sl_datalen = 0;
+		return;
 	}
+	slot->sl_datalen = (char *)resp->p - (char *)resp->cstate.datap;
+	base = (char *)resp->cstate.datap -
+					(char *)resp->xbuf->head[0].iov_base;
+	if (read_bytes_from_xdr_buf(resp->xbuf, base, slot->sl_data,
+				    slot->sl_datalen))
+		WARN("%s: sessions DRC could not cache compound\n", __func__);
+	return;
 }
 
 /*
- * Cache the reply pages up to NFSD_PAGES_PER_SLOT + 1, clearing the previous
- * pages. We add a page to NFSD_PAGES_PER_SLOT for the case where the total
- * length of the XDR response is less than se_fmaxresp_cached
- * (NFSD_PAGES_PER_SLOT * PAGE_SIZE) but the xdr_buf pages is used for a
- * of the reply (e.g. readdir).
+ * Encode the replay sequence operation from the slot values.
+ * If cachethis is FALSE encode the uncached rep error on the next
+ * operation which sets resp->p and increments resp->opcnt for
+ * nfs4svc_encode_compoundres.
  *
- * Store the base and length of the rq_req.head[0] page
- * of the NFSv4.1 data, just past the rpc header.
  */
-void
-nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
+static __be32
+nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
+			  struct nfsd4_compoundres *resp)
 {
-	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
-	struct svc_rqst *rqstp = resp->rqstp;
-	struct nfsd4_compoundargs *args = rqstp->rq_argp;
-	struct nfsd4_op *op = &args->ops[resp->opcnt];
-	struct kvec *resv = &rqstp->rq_res.head[0];
-
-	dprintk("--> %s entry %p\n", __func__, entry);
-
-	/* Don't cache a failed OP_SEQUENCE. */
-	if (resp->opcnt == 1 && op->opnum == OP_SEQUENCE && resp->cstate.status)
-		return;
+	struct nfsd4_op *op;
+	struct nfsd4_slot *slot = resp->cstate.slot;
 
-	nfsd4_release_respages(entry->ce_respages, entry->ce_resused);
-	entry->ce_opcnt = resp->opcnt;
-	entry->ce_status = resp->cstate.status;
+	dprintk("--> %s resp->opcnt %d cachethis %u \n", __func__,
+		resp->opcnt, resp->cstate.slot->sl_cachethis);
 
-	/*
-	 * Don't need a page to cache just the sequence operation - the slot
-	 * does this for us!
-	 */
+	/* Encode the replayed sequence operation */
+	op = &args->ops[resp->opcnt - 1];
+	nfsd4_encode_operation(resp, op);
 
-	if (nfsd4_not_cached(resp)) {
-		entry->ce_resused = 0;
-		entry->ce_rpchdrlen = 0;
-		dprintk("%s Just cache SEQUENCE. ce_cachethis %d\n", __func__,
-			resp->cstate.slot->sl_cache_entry.ce_cachethis);
-		return;
-	}
-	entry->ce_resused = rqstp->rq_resused;
-	if (entry->ce_resused > NFSD_PAGES_PER_SLOT + 1)
-		entry->ce_resused = NFSD_PAGES_PER_SLOT + 1;
-	nfsd4_copy_pages(entry->ce_respages, rqstp->rq_respages,
-			 entry->ce_resused);
-	entry->ce_datav.iov_base = resp->cstate.statp;
-	entry->ce_datav.iov_len = resv->iov_len - ((char *)resp->cstate.statp -
-				(char *)page_address(rqstp->rq_respages[0]));
-	/* Current request rpc header length*/
-	entry->ce_rpchdrlen = (char *)resp->cstate.statp -
-				(char *)page_address(rqstp->rq_respages[0]);
-}
-
-/*
- * We keep the rpc header, but take the nfs reply from the replycache.
- */
-static int
-nfsd41_copy_replay_data(struct nfsd4_compoundres *resp,
-			struct nfsd4_cache_entry *entry)
-{
-	struct svc_rqst *rqstp = resp->rqstp;
-	struct kvec *resv = &resp->rqstp->rq_res.head[0];
-	int len;
-
-	/* Current request rpc header length*/
-	len = (char *)resp->cstate.statp -
-			(char *)page_address(rqstp->rq_respages[0]);
-	if (entry->ce_datav.iov_len + len > PAGE_SIZE) {
-		dprintk("%s v41 cached reply too large (%Zd).\n", __func__,
-			entry->ce_datav.iov_len);
-		return 0;
+	/* Return nfserr_retry_uncached_rep in next operation. */
+	if (args->opcnt > 1 && slot->sl_cachethis == 0) {
+		op = &args->ops[resp->opcnt++];
+		op->status = nfserr_retry_uncached_rep;
+		nfsd4_encode_operation(resp, op);
 	}
-	/* copy the cached reply nfsd data past the current rpc header */
-	memcpy((char *)resv->iov_base + len, entry->ce_datav.iov_base,
-		entry->ce_datav.iov_len);
-	resv->iov_len = len + entry->ce_datav.iov_len;
-	return 1;
+	return op->status;
 }
 
 /*
- * Keep the first page of the replay. Copy the NFSv4.1 data from the first
- * cached page.  Replace any futher replay pages from the cache.
+ * The sequence operation is not cached because we can use the slot and
+ * session values.
  */
 __be32
 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
 			 struct nfsd4_sequence *seq)
 {
-	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
+	struct nfsd4_slot *slot = resp->cstate.slot;
 	__be32 status;
 
-	dprintk("--> %s entry %p\n", __func__, entry);
-
-	/*
-	 * If this is just the sequence operation, we did not keep
-	 * a page in the cache entry because we can just use the
-	 * slot info stored in struct nfsd4_sequence that was checked
-	 * against the slot in nfsd4_sequence().
-	 *
-	 * This occurs when seq->cachethis is FALSE, or when the client
-	 * session inactivity timer fires and a solo sequence operation
-	 * is sent (lease renewal).
-	 */
-	if (seq && nfsd4_not_cached(resp)) {
-		seq->maxslots = resp->cstate.session->se_fchannel.maxreqs;
-		return nfs_ok;
-	}
-
-	if (!nfsd41_copy_replay_data(resp, entry)) {
-		/*
-		 * Not enough room to use the replay rpc header, send the
-		 * cached header. Release all the allocated result pages.
-		 */
-		svc_free_res_pages(resp->rqstp);
-		nfsd4_copy_pages(resp->rqstp->rq_respages, entry->ce_respages,
-			entry->ce_resused);
-	} else {
-		/* Release all but the first allocated result page */
+	dprintk("--> %s slot %p\n", __func__, slot);
 
-		resp->rqstp->rq_resused--;
-		svc_free_res_pages(resp->rqstp);
+	/* Either returns 0 or nfserr_retry_uncached */
+	status = nfsd4_enc_sequence_replay(resp->rqstp->rq_argp, resp);
+	if (status == nfserr_retry_uncached_rep)
+		return status;
 
-		nfsd4_copy_pages(&resp->rqstp->rq_respages[1],
-				 &entry->ce_respages[1],
-				 entry->ce_resused - 1);
-	}
+	/* The sequence operation has been encoded, cstate->datap set. */
+	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
 
-	resp->rqstp->rq_resused = entry->ce_resused;
-	resp->opcnt = entry->ce_opcnt;
-	resp->cstate.iovlen = entry->ce_datav.iov_len + entry->ce_rpchdrlen;
-	status = entry->ce_status;
+	resp->opcnt = slot->sl_opcnt;
+	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
+	status = slot->sl_status;
 
 	return status;
 }
@@ -1194,13 +1119,15 @@ nfsd4_exchange_id(struct svc_rqst *rqstp,
 	int status;
 	unsigned int		strhashval;
 	char			dname[HEXDIR_LEN];
+	char			addr_str[INET6_ADDRSTRLEN];
 	nfs4_verifier		verf = exid->verifier;
-	u32			ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct sockaddr		*sa = svc_addr(rqstp);
 
+	rpc_ntop(sa, addr_str, sizeof(addr_str));
 	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
-		" ip_addr=%u flags %x, spa_how %d\n",
+		"ip_addr=%s flags %x, spa_how %d\n",
 		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
-		ip_addr, exid->flags, exid->spa_how);
+		addr_str, exid->flags, exid->spa_how);
 
 	if (!check_name(exid->clname) || (exid->flags & ~EXCHGID4_FLAG_MASK_A))
 		return nfserr_inval;
@@ -1281,28 +1208,23 @@ nfsd4_exchange_id(struct svc_rqst *rqstp,
 
 out_new:
 	/* Normal case */
-	new = create_client(exid->clname, dname);
+	new = create_client(exid->clname, dname, rqstp, &verf);
 	if (new == NULL) {
-		status = nfserr_resource;
+		status = nfserr_serverfault;
 		goto out;
 	}
 
-	copy_verf(new, &verf);
-	copy_cred(&new->cl_cred, &rqstp->rq_cred);
-	new->cl_addr = ip_addr;
 	gen_clid(new);
-	gen_confirm(new);
 	add_to_unconfirmed(new, strhashval);
 out_copy:
 	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
 	exid->clientid.cl_id = new->cl_clientid.cl_id;
 
-	new->cl_slot.sl_seqid = 0;
 	exid->seqid = 1;
 	nfsd4_set_ex_flags(new, exid);
 
 	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
-		new->cl_slot.sl_seqid, new->cl_exchange_flags);
+		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
 	status = nfs_ok;
 
 out:
@@ -1313,40 +1235,60 @@ error:
 }
 
 static int
-check_slot_seqid(u32 seqid, struct nfsd4_slot *slot)
+check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
 {
-	dprintk("%s enter. seqid %d slot->sl_seqid %d\n", __func__, seqid,
-		slot->sl_seqid);
+	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
+		slot_seqid);
 
 	/* The slot is in use, and no response has been sent. */
-	if (slot->sl_inuse) {
-		if (seqid == slot->sl_seqid)
+	if (slot_inuse) {
+		if (seqid == slot_seqid)
 			return nfserr_jukebox;
 		else
 			return nfserr_seq_misordered;
 	}
 	/* Normal */
-	if (likely(seqid == slot->sl_seqid + 1))
+	if (likely(seqid == slot_seqid + 1))
 		return nfs_ok;
 	/* Replay */
-	if (seqid == slot->sl_seqid)
+	if (seqid == slot_seqid)
 		return nfserr_replay_cache;
 	/* Wraparound */
-	if (seqid == 1 && (slot->sl_seqid + 1) == 0)
+	if (seqid == 1 && (slot_seqid + 1) == 0)
 		return nfs_ok;
 	/* Misordered replay or misordered new request */
 	return nfserr_seq_misordered;
 }
 
+/*
+ * Cache the create session result into the create session single DRC
+ * slot cache by saving the xdr structure. sl_seqid has been set.
+ * Do this for solo or embedded create session operations.
+ */
+static void
+nfsd4_cache_create_session(struct nfsd4_create_session *cr_ses,
+			   struct nfsd4_clid_slot *slot, int nfserr)
+{
+	slot->sl_status = nfserr;
+	memcpy(&slot->sl_cr_ses, cr_ses, sizeof(*cr_ses));
+}
+
+static __be32
+nfsd4_replay_create_session(struct nfsd4_create_session *cr_ses,
+			    struct nfsd4_clid_slot *slot)
+{
+	memcpy(cr_ses, &slot->sl_cr_ses, sizeof(*cr_ses));
+	return slot->sl_status;
+}
+
 __be32
 nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_compound_state *cstate,
 		     struct nfsd4_create_session *cr_ses)
 {
-	u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
-	struct nfsd4_compoundres *resp = rqstp->rq_resp;
+	struct sockaddr *sa = svc_addr(rqstp);
 	struct nfs4_client *conf, *unconf;
-	struct nfsd4_slot *slot = NULL;
+	struct nfsd4_clid_slot *cs_slot = NULL;
 	int status = 0;
 
 	nfs4_lock_state();
@@ -1354,40 +1296,38 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 	conf = find_confirmed_client(&cr_ses->clientid);
 
 	if (conf) {
-		slot = &conf->cl_slot;
-		status = check_slot_seqid(cr_ses->seqid, slot);
+		cs_slot = &conf->cl_cs_slot;
+		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
 		if (status == nfserr_replay_cache) {
 			dprintk("Got a create_session replay! seqid= %d\n",
-				slot->sl_seqid);
-			cstate->slot = slot;
-			cstate->status = status;
+				cs_slot->sl_seqid);
 			/* Return the cached reply status */
-			status = nfsd4_replay_cache_entry(resp, NULL);
+			status = nfsd4_replay_create_session(cr_ses, cs_slot);
 			goto out;
-		} else if (cr_ses->seqid != conf->cl_slot.sl_seqid + 1) {
+		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
 			status = nfserr_seq_misordered;
 			dprintk("Sequence misordered!\n");
 			dprintk("Expected seqid= %d but got seqid= %d\n",
-				slot->sl_seqid, cr_ses->seqid);
+				cs_slot->sl_seqid, cr_ses->seqid);
 			goto out;
 		}
-		conf->cl_slot.sl_seqid++;
+		cs_slot->sl_seqid++;
 	} else if (unconf) {
 		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
-		    (ip_addr != unconf->cl_addr)) {
+		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
 			status = nfserr_clid_inuse;
 			goto out;
 		}
 
-		slot = &unconf->cl_slot;
-		status = check_slot_seqid(cr_ses->seqid, slot);
+		cs_slot = &unconf->cl_cs_slot;
+		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
 		if (status) {
 			/* an unconfirmed replay returns misordered */
 			status = nfserr_seq_misordered;
-			goto out;
+			goto out_cache;
 		}
 
-		slot->sl_seqid++; /* from 0 to 1 */
+		cs_slot->sl_seqid++; /* from 0 to 1 */
 		move_to_confirmed(unconf);
 
 		/*
@@ -1396,6 +1336,19 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		cr_ses->flags &= ~SESSION4_PERSIST;
 		cr_ses->flags &= ~SESSION4_RDMA;
 
+		if (cr_ses->flags & SESSION4_BACK_CHAN) {
+			unconf->cl_cb_xprt = rqstp->rq_xprt;
+			svc_xprt_get(unconf->cl_cb_xprt);
+			rpc_copy_addr(
+				(struct sockaddr *)&unconf->cl_cb_conn.cb_addr,
+				sa);
+			unconf->cl_cb_conn.cb_addrlen = svc_addr_len(sa);
+			unconf->cl_cb_conn.cb_minorversion =
+				cstate->minorversion;
+			unconf->cl_cb_conn.cb_prog = cr_ses->callback_prog;
+			unconf->cl_cb_seq_nr = 1;
+			nfsd4_probe_callback(unconf);
+		}
 		conf = unconf;
 	} else {
 		status = nfserr_stale_clientid;
@@ -1408,12 +1361,11 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 
 	memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data,
 	       NFS4_MAX_SESSIONID_LEN);
-	cr_ses->seqid = slot->sl_seqid;
+	cr_ses->seqid = cs_slot->sl_seqid;
 
-	slot->sl_inuse = true;
-	cstate->slot = slot;
-	/* Ensure a page is used for the cache */
-	slot->sl_cache_entry.ce_cachethis = 1;
+out_cache:
+	/* cache solo and embedded create sessions under the state lock */
+	nfsd4_cache_create_session(cr_ses, cs_slot, status);
 out:
 	nfs4_unlock_state();
 	dprintk("%s returns %d\n", __func__, ntohl(status));
@@ -1478,18 +1430,23 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	if (seq->slotid >= session->se_fchannel.maxreqs)
 		goto out;
 
-	slot = &session->se_slots[seq->slotid];
+	slot = session->se_slots[seq->slotid];
 	dprintk("%s: slotid %d\n", __func__, seq->slotid);
 
-	status = check_slot_seqid(seq->seqid, slot);
+	/* We do not negotiate the number of slots yet, so set the
+	 * maxslots to the session maxreqs which is used to encode
+	 * sr_highest_slotid and the sr_target_slot id to maxslots */
+	seq->maxslots = session->se_fchannel.maxreqs;
+
+	status = check_slot_seqid(seq->seqid, slot->sl_seqid, slot->sl_inuse);
 	if (status == nfserr_replay_cache) {
 		cstate->slot = slot;
 		cstate->session = session;
 		/* Return the cached reply status and set cstate->status
-		 * for nfsd4_svc_encode_compoundres processing */
+		 * for nfsd4_proc_compound processing */
 		status = nfsd4_replay_cache_entry(resp, seq);
 		cstate->status = nfserr_replay_cache;
-		goto replay_cache;
+		goto out;
 	}
 	if (status)
 		goto out;
@@ -1497,23 +1454,23 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	/* Success! bump slot seqid */
 	slot->sl_inuse = true;
 	slot->sl_seqid = seq->seqid;
-	slot->sl_cache_entry.ce_cachethis = seq->cachethis;
-	/* Always set the cache entry cachethis for solo sequence */
-	if (nfsd4_is_solo_sequence(resp))
-		slot->sl_cache_entry.ce_cachethis = 1;
+	slot->sl_cachethis = seq->cachethis;
 
 	cstate->slot = slot;
 	cstate->session = session;
 
-replay_cache:
-	/* Renew the clientid on success and on replay.
-	 * Hold a session reference until done processing the compound:
+	/* Hold a session reference until done processing the compound:
 	 * nfsd4_put_session called only if the cstate slot is set.
 	 */
-	renew_client(session->se_client);
 	nfsd4_get_session(session);
 out:
 	spin_unlock(&sessionid_lock);
+	/* Renew the clientid on success and on replay */
+	if (cstate->session) {
+		nfs4_lock_state();
+		renew_client(session->se_client);
+		nfs4_unlock_state();
+	}
 	dprintk("%s: return %d\n", __func__, ntohl(status));
 	return status;
 }
@@ -1522,7 +1479,7 @@ __be32
 nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		  struct nfsd4_setclientid *setclid)
 {
-	struct sockaddr_in	*sin = svc_addr_in(rqstp);
+	struct sockaddr		*sa = svc_addr(rqstp);
 	struct xdr_netobj 	clname = { 
 		.len = setclid->se_namelen,
 		.data = setclid->se_name,
@@ -1531,7 +1488,6 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	unsigned int 		strhashval;
 	struct nfs4_client	*conf, *unconf, *new;
 	__be32 			status;
-	char			*princ;
 	char                    dname[HEXDIR_LEN];
 	
 	if (!check_name(clname))
@@ -1554,8 +1510,11 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		/* RFC 3530 14.2.33 CASE 0: */
 		status = nfserr_clid_inuse;
 		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
-			dprintk("NFSD: setclientid: string in use by client"
-				" at %pI4\n", &conf->cl_addr);
+			char addr_str[INET6_ADDRSTRLEN];
+			rpc_ntop((struct sockaddr *) &conf->cl_addr, addr_str,
+				 sizeof(addr_str));
+			dprintk("NFSD: setclientid: string in use by client "
+				"at %s\n", addr_str);
 			goto out;
 		}
 	}
@@ -1573,7 +1532,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		 */
 		if (unconf)
 			expire_client(unconf);
-		new = create_client(clname, dname);
+		new = create_client(clname, dname, rqstp, &clverifier);
 		if (new == NULL)
 			goto out;
 		gen_clid(new);
@@ -1590,7 +1549,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 			 */
 			expire_client(unconf);
 		}
-		new = create_client(clname, dname);
+		new = create_client(clname, dname, rqstp, &clverifier);
 		if (new == NULL)
 			goto out;
 		copy_clid(new, conf);
@@ -1600,7 +1559,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		 * probable client reboot; state will be removed if
 		 * confirmed.
 		 */
-		new = create_client(clname, dname);
+		new = create_client(clname, dname, rqstp, &clverifier);
 		if (new == NULL)
 			goto out;
 		gen_clid(new);
@@ -1611,25 +1570,12 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		 * confirmed.
 		 */
 		expire_client(unconf);
-		new = create_client(clname, dname);
+		new = create_client(clname, dname, rqstp, &clverifier);
 		if (new == NULL)
 			goto out;
 		gen_clid(new);
 	}
-	copy_verf(new, &clverifier);
-	new->cl_addr = sin->sin_addr.s_addr;
-	new->cl_flavor = rqstp->rq_flavor;
-	princ = svc_gss_principal(rqstp);
-	if (princ) {
-		new->cl_principal = kstrdup(princ, GFP_KERNEL);
-		if (new->cl_principal == NULL) {
-			free_client(new);
-			goto out;
-		}
-	}
-	copy_cred(&new->cl_cred, &rqstp->rq_cred);
-	gen_confirm(new);
-	gen_callback(new, setclid);
+	gen_callback(new, setclid, rpc_get_scope_id(sa));
 	add_to_unconfirmed(new, strhashval);
 	setclid->se_clientid.cl_boot = new->cl_clientid.cl_boot;
 	setclid->se_clientid.cl_id = new->cl_clientid.cl_id;
@@ -1651,7 +1597,7 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 			 struct nfsd4_compound_state *cstate,
 			 struct nfsd4_setclientid_confirm *setclientid_confirm)
 {
-	struct sockaddr_in *sin = svc_addr_in(rqstp);
+	struct sockaddr *sa = svc_addr(rqstp);
 	struct nfs4_client *conf, *unconf;
 	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
 	clientid_t * clid = &setclientid_confirm->sc_clientid;
@@ -1670,9 +1616,9 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 	unconf = find_unconfirmed_client(clid);
 
 	status = nfserr_clid_inuse;
-	if (conf && conf->cl_addr != sin->sin_addr.s_addr)
+	if (conf && !rpc_cmp_addr((struct sockaddr *) &conf->cl_addr, sa))
 		goto out;
-	if (unconf && unconf->cl_addr != sin->sin_addr.s_addr)
+	if (unconf && !rpc_cmp_addr((struct sockaddr *) &unconf->cl_addr, sa))
 		goto out;
 
 	/*
@@ -2163,7 +2109,7 @@ int nfsd_change_deleg_cb(struct file_lock **onlist, int arg)
 		return -EAGAIN;
 }
 
-static struct lock_manager_operations nfsd_lease_mng_ops = {
+static const struct lock_manager_operations nfsd_lease_mng_ops = {
 	.fl_break = nfsd_break_deleg_cb,
 	.fl_release_private = nfsd_release_deleg_cb,
 	.fl_copy_lock = nfsd_copy_lock_deleg_cb,
@@ -3368,7 +3314,7 @@ nfs4_transform_lock_offset(struct file_lock *lock)
 
 /* Hack!: For now, we're defining this just so we can use a pointer to it
  * as a unique cookie to identify our (NFSv4's) posix locks. */
-static struct lock_manager_operations nfsd_posix_mng_ops  = {
+static const struct lock_manager_operations nfsd_posix_mng_ops  = {
 };
 
 static inline void
@@ -4072,7 +4018,7 @@ set_max_delegations(void)
 
 /* initialization to perform when the nfsd service is started: */
 
-static void
+static int
 __nfs4_state_start(void)
 {
 	unsigned long grace_time;
@@ -4084,19 +4030,26 @@ __nfs4_state_start(void)
 	printk(KERN_INFO "NFSD: starting %ld-second grace period\n",
 	       grace_time/HZ);
 	laundry_wq = create_singlethread_workqueue("nfsd4");
+	if (laundry_wq == NULL)
+		return -ENOMEM;
 	queue_delayed_work(laundry_wq, &laundromat_work, grace_time);
 	set_max_delegations();
+	return set_callback_cred();
 }
 
-void
+int
 nfs4_state_start(void)
 {
+	int ret;
+
 	if (nfs4_init)
-		return;
+		return 0;
 	nfsd4_load_reboot_recovery_data();
-	__nfs4_state_start();
+	ret = __nfs4_state_start();
+	if (ret)
+		return ret;
 	nfs4_init = 1;
-	return;
+	return 0;
 }
 
 time_t
diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index 2dcc7feaa6ff..0fbd50cee1f6 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -1599,7 +1599,8 @@ static __be32 nfsd4_encode_fs_location4(struct nfsd4_fs_location *location,
 static char *nfsd4_path(struct svc_rqst *rqstp, struct svc_export *exp, __be32 *stat)
 {
 	struct svc_fh tmp_fh;
-	char *path, *rootpath;
+	char *path = NULL, *rootpath;
+	size_t rootlen;
 
 	fh_init(&tmp_fh, NFS4_FHSIZE);
 	*stat = exp_pseudoroot(rqstp, &tmp_fh);
@@ -1609,14 +1610,18 @@ static char *nfsd4_path(struct svc_rqst *rqstp, struct svc_export *exp, __be32 *
 
 	path = exp->ex_pathname;
 
-	if (strncmp(path, rootpath, strlen(rootpath))) {
+	rootlen = strlen(rootpath);
+	if (strncmp(path, rootpath, rootlen)) {
 		dprintk("nfsd: fs_locations failed;"
 			"%s is not contained in %s\n", path, rootpath);
 		*stat = nfserr_notsupp;
-		return NULL;
+		path = NULL;
+		goto out;
 	}
-
-	return path + strlen(rootpath);
+	path += rootlen;
+out:
+	fh_put(&tmp_fh);
+	return path;
 }
 
 /*
@@ -1793,11 +1798,6 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
 				goto out_nfserr;
 		}
 	}
-	if (bmval0 & FATTR4_WORD0_FS_LOCATIONS) {
-		if (exp->ex_fslocs.locations == NULL) {
-			bmval0 &= ~FATTR4_WORD0_FS_LOCATIONS;
-		}
-	}
 	if ((buflen -= 16) < 0)
 		goto out_resource;
 
@@ -1825,8 +1825,6 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
 			goto out_resource;
 		if (!aclsupport)
 			word0 &= ~FATTR4_WORD0_ACL;
-		if (!exp->ex_fslocs.locations)
-			word0 &= ~FATTR4_WORD0_FS_LOCATIONS;
 		if (!word2) {
 			WRITE32(2);
 			WRITE32(word0);
@@ -3064,6 +3062,7 @@ nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
 	WRITE32(0);
 
 	ADJUST_ARGS();
+	resp->cstate.datap = p; /* DRC cache data pointer */
 	return 0;
 }
 
@@ -3166,7 +3165,7 @@ static int nfsd4_check_drc_limit(struct nfsd4_compoundres *resp)
 		return status;
 
 	session = resp->cstate.session;
-	if (session == NULL || slot->sl_cache_entry.ce_cachethis == 0)
+	if (session == NULL || slot->sl_cachethis == 0)
 		return status;
 
 	if (resp->opcnt >= args->opcnt)
@@ -3291,6 +3290,7 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 	/*
 	 * All that remains is to write the tag and operation count...
 	 */
+	struct nfsd4_compound_state *cs = &resp->cstate;
 	struct kvec *iov;
 	p = resp->tagp;
 	*p++ = htonl(resp->taglen);
@@ -3304,17 +3304,11 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 		iov = &rqstp->rq_res.head[0];
 	iov->iov_len = ((char*)resp->p) - (char*)iov->iov_base;
 	BUG_ON(iov->iov_len > PAGE_SIZE);
-	if (nfsd4_has_session(&resp->cstate)) {
-		if (resp->cstate.status == nfserr_replay_cache &&
-				!nfsd4_not_cached(resp)) {
-			iov->iov_len = resp->cstate.iovlen;
-		} else {
-			nfsd4_store_cache_entry(resp);
-			dprintk("%s: SET SLOT STATE TO AVAILABLE\n", __func__);
-			resp->cstate.slot->sl_inuse = 0;
-		}
-		if (resp->cstate.session)
-			nfsd4_put_session(resp->cstate.session);
+	if (nfsd4_has_session(cs) && cs->status != nfserr_replay_cache) {
+		nfsd4_store_cache_entry(resp);
+		dprintk("%s: SET SLOT STATE TO AVAILABLE\n", __func__);
+		resp->cstate.slot->sl_inuse = false;
+		nfsd4_put_session(resp->cstate.session);
 	}
 	return 1;
 }
diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index 7e906c5b7671..5c01fc148ce8 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -174,12 +174,13 @@ static const struct file_operations exports_operations = {
 };
 
 extern int nfsd_pool_stats_open(struct inode *inode, struct file *file);
+extern int nfsd_pool_stats_release(struct inode *inode, struct file *file);
 
-static struct file_operations pool_stats_operations = {
+static const struct file_operations pool_stats_operations = {
 	.open		= nfsd_pool_stats_open,
 	.read		= seq_read,
 	.llseek		= seq_lseek,
-	.release	= seq_release,
+	.release	= nfsd_pool_stats_release,
 	.owner		= THIS_MODULE,
 };
 
@@ -776,10 +777,7 @@ static ssize_t write_pool_threads(struct file *file, char *buf, size_t size)
 		size -= len;
 		mesg += len;
 	}
-
-	mutex_unlock(&nfsd_mutex);
-	return (mesg-buf);
-
+	rv = mesg - buf;
 out_free:
 	kfree(nthreads);
 	mutex_unlock(&nfsd_mutex);
diff --git a/fs/nfsd/nfsfh.c b/fs/nfsd/nfsfh.c
index 8847f3fbfc1e..01965b2f3a76 100644
--- a/fs/nfsd/nfsfh.c
+++ b/fs/nfsd/nfsfh.c
@@ -397,44 +397,51 @@ static inline void _fh_update_old(struct dentry *dentry,
 		fh->ofh_dirino = 0;
 }
 
-__be32
-fh_compose(struct svc_fh *fhp, struct svc_export *exp, struct dentry *dentry,
-	   struct svc_fh *ref_fh)
+static bool is_root_export(struct svc_export *exp)
 {
-	/* ref_fh is a reference file handle.
-	 * if it is non-null and for the same filesystem, then we should compose
-	 * a filehandle which is of the same version, where possible.
-	 * Currently, that means that if ref_fh->fh_handle.fh_version == 0xca
-	 * Then create a 32byte filehandle using nfs_fhbase_old
-	 *
-	 */
+	return exp->ex_path.dentry == exp->ex_path.dentry->d_sb->s_root;
+}
 
-	u8 version;
-	u8 fsid_type = 0;
-	struct inode * inode = dentry->d_inode;
-	struct dentry *parent = dentry->d_parent;
-	__u32 *datap;
-	dev_t ex_dev = exp->ex_path.dentry->d_inode->i_sb->s_dev;
-	int root_export = (exp->ex_path.dentry == exp->ex_path.dentry->d_sb->s_root);
+static struct super_block *exp_sb(struct svc_export *exp)
+{
+	return exp->ex_path.dentry->d_inode->i_sb;
+}
 
-	dprintk("nfsd: fh_compose(exp %02x:%02x/%ld %s/%s, ino=%ld)\n",
-		MAJOR(ex_dev), MINOR(ex_dev),
-		(long) exp->ex_path.dentry->d_inode->i_ino,
-		parent->d_name.name, dentry->d_name.name,
-		(inode ? inode->i_ino : 0));
+static bool fsid_type_ok_for_exp(u8 fsid_type, struct svc_export *exp)
+{
+	switch (fsid_type) {
+	case FSID_DEV:
+		if (!old_valid_dev(exp_sb(exp)->s_dev))
+			return 0;
+		/* FALL THROUGH */
+	case FSID_MAJOR_MINOR:
+	case FSID_ENCODE_DEV:
+		return exp_sb(exp)->s_type->fs_flags & FS_REQUIRES_DEV;
+	case FSID_NUM:
+		return exp->ex_flags & NFSEXP_FSID;
+	case FSID_UUID8:
+	case FSID_UUID16:
+		if (!is_root_export(exp))
+			return 0;
+		/* fall through */
+	case FSID_UUID4_INUM:
+	case FSID_UUID16_INUM:
+		return exp->ex_uuid != NULL;
+	}
+	return 1;
+}
 
-	/* Choose filehandle version and fsid type based on
-	 * the reference filehandle (if it is in the same export)
-	 * or the export options.
-	 */
- retry:
+
+static void set_version_and_fsid_type(struct svc_fh *fhp, struct svc_export *exp, struct svc_fh *ref_fh)
+{
+	u8 version;
+	u8 fsid_type;
+retry:
 	version = 1;
 	if (ref_fh && ref_fh->fh_export == exp) {
 		version = ref_fh->fh_handle.fh_version;
 		fsid_type = ref_fh->fh_handle.fh_fsid_type;
 
-		if (ref_fh == fhp)
-			fh_put(ref_fh);
 		ref_fh = NULL;
 
 		switch (version) {
@@ -447,58 +454,66 @@ fh_compose(struct svc_fh *fhp, struct svc_export *exp, struct dentry *dentry,
 			goto retry;
 		}
 
-		/* Need to check that this type works for this
-		 * export point.  As the fsid -> filesystem mapping
-		 * was guided by user-space, there is no guarantee
-		 * that the filesystem actually supports that fsid
-		 * type. If it doesn't we loop around again without
-		 * ref_fh set.
+		/*
+		 * As the fsid -> filesystem mapping was guided by
+		 * user-space, there is no guarantee that the filesystem
+		 * actually supports that fsid type. If it doesn't we
+		 * loop around again without ref_fh set.
 		 */
-		switch(fsid_type) {
-		case FSID_DEV:
-			if (!old_valid_dev(ex_dev))
-				goto retry;
-			/* FALL THROUGH */
-		case FSID_MAJOR_MINOR:
-		case FSID_ENCODE_DEV:
-			if (!(exp->ex_path.dentry->d_inode->i_sb->s_type->fs_flags
-			      & FS_REQUIRES_DEV))
-				goto retry;
-			break;
-		case FSID_NUM:
-			if (! (exp->ex_flags & NFSEXP_FSID))
-				goto retry;
-			break;
-		case FSID_UUID8:
-		case FSID_UUID16:
-			if (!root_export)
-				goto retry;
-			/* fall through */
-		case FSID_UUID4_INUM:
-		case FSID_UUID16_INUM:
-			if (exp->ex_uuid == NULL)
-				goto retry;
-			break;
-		}
+		if (!fsid_type_ok_for_exp(fsid_type, exp))
+			goto retry;
 	} else if (exp->ex_flags & NFSEXP_FSID) {
 		fsid_type = FSID_NUM;
 	} else if (exp->ex_uuid) {
 		if (fhp->fh_maxsize >= 64) {
-			if (root_export)
+			if (is_root_export(exp))
 				fsid_type = FSID_UUID16;
 			else
 				fsid_type = FSID_UUID16_INUM;
 		} else {
-			if (root_export)
+			if (is_root_export(exp))
 				fsid_type = FSID_UUID8;
 			else
 				fsid_type = FSID_UUID4_INUM;
 		}
-	} else if (!old_valid_dev(ex_dev))
+	} else if (!old_valid_dev(exp_sb(exp)->s_dev))
 		/* for newer device numbers, we must use a newer fsid format */
 		fsid_type = FSID_ENCODE_DEV;
 	else
 		fsid_type = FSID_DEV;
+	fhp->fh_handle.fh_version = version;
+	if (version)
+		fhp->fh_handle.fh_fsid_type = fsid_type;
+}
+
+__be32
+fh_compose(struct svc_fh *fhp, struct svc_export *exp, struct dentry *dentry,
+	   struct svc_fh *ref_fh)
+{
+	/* ref_fh is a reference file handle.
+	 * if it is non-null and for the same filesystem, then we should compose
+	 * a filehandle which is of the same version, where possible.
+	 * Currently, that means that if ref_fh->fh_handle.fh_version == 0xca
+	 * Then create a 32byte filehandle using nfs_fhbase_old
+	 *
+	 */
+
+	struct inode * inode = dentry->d_inode;
+	struct dentry *parent = dentry->d_parent;
+	__u32 *datap;
+	dev_t ex_dev = exp_sb(exp)->s_dev;
+
+	dprintk("nfsd: fh_compose(exp %02x:%02x/%ld %s/%s, ino=%ld)\n",
+		MAJOR(ex_dev), MINOR(ex_dev),
+		(long) exp->ex_path.dentry->d_inode->i_ino,
+		parent->d_name.name, dentry->d_name.name,
+		(inode ? inode->i_ino : 0));
+
+	/* Choose filehandle version and fsid type based on
+	 * the reference filehandle (if it is in the same export)
+	 * or the export options.
+	 */
+	 set_version_and_fsid_type(fhp, exp, ref_fh);
 
 	if (ref_fh == fhp)
 		fh_put(ref_fh);
@@ -516,7 +531,7 @@ fh_compose(struct svc_fh *fhp, struct svc_export *exp, struct dentry *dentry,
 	fhp->fh_export = exp;
 	cache_get(&exp->h);
 
-	if (version == 0xca) {
+	if (fhp->fh_handle.fh_version == 0xca) {
 		/* old style filehandle please */
 		memset(&fhp->fh_handle.fh_base, 0, NFS_FHSIZE);
 		fhp->fh_handle.fh_size = NFS_FHSIZE;
@@ -530,22 +545,22 @@ fh_compose(struct svc_fh *fhp, struct svc_export *exp, struct dentry *dentry,
 			_fh_update_old(dentry, exp, &fhp->fh_handle);
 	} else {
 		int len;
-		fhp->fh_handle.fh_version = 1;
 		fhp->fh_handle.fh_auth_type = 0;
 		datap = fhp->fh_handle.fh_auth+0;
-		fhp->fh_handle.fh_fsid_type = fsid_type;
-		mk_fsid(fsid_type, datap, ex_dev,
+		mk_fsid(fhp->fh_handle.fh_fsid_type, datap, ex_dev,
 			exp->ex_path.dentry->d_inode->i_ino,
 			exp->ex_fsid, exp->ex_uuid);
 
-		len = key_len(fsid_type);
+		len = key_len(fhp->fh_handle.fh_fsid_type);
 		datap += len/4;
 		fhp->fh_handle.fh_size = 4 + len;
 
 		if (inode)
 			_fh_update(fhp, exp, dentry);
-		if (fhp->fh_handle.fh_fileid_type == 255)
+		if (fhp->fh_handle.fh_fileid_type == 255) {
+			fh_put(fhp);
 			return nfserr_opnotsupp;
+		}
 	}
 
 	return 0;
@@ -639,8 +654,7 @@ enum fsid_source fsid_source(struct svc_fh *fhp)
 	case FSID_DEV:
 	case FSID_ENCODE_DEV:
 	case FSID_MAJOR_MINOR:
-		if (fhp->fh_export->ex_path.dentry->d_inode->i_sb->s_type->fs_flags
-		    & FS_REQUIRES_DEV)
+		if (exp_sb(fhp->fh_export)->s_type->fs_flags & FS_REQUIRES_DEV)
 			return FSIDSOURCE_DEV;
 		break;
 	case FSID_NUM:
diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index 24d58adfe5fd..67ea83eedd43 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -34,6 +34,7 @@
 #include <linux/nfsd/syscall.h>
 #include <linux/lockd/bind.h>
 #include <linux/nfsacl.h>
+#include <linux/seq_file.h>
 
 #define NFSDDBG_FACILITY	NFSDDBG_SVC
 
@@ -66,6 +67,16 @@ struct timeval			nfssvc_boot;
 DEFINE_MUTEX(nfsd_mutex);
 struct svc_serv 		*nfsd_serv;
 
+/*
+ * nfsd_drc_lock protects nfsd_drc_max_pages and nfsd_drc_pages_used.
+ * nfsd_drc_max_pages limits the total amount of memory available for
+ * version 4.1 DRC caches.
+ * nfsd_drc_pages_used tracks the current version 4.1 DRC memory usage.
+ */
+spinlock_t	nfsd_drc_lock;
+unsigned int	nfsd_drc_max_mem;
+unsigned int	nfsd_drc_mem_used;
+
 #if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
 static struct svc_stat	nfsd_acl_svcstats;
 static struct svc_version *	nfsd_acl_version[] = {
@@ -235,13 +246,12 @@ void nfsd_reset_versions(void)
  */
 static void set_max_drc(void)
 {
-	/* The percent of nr_free_buffer_pages used by the V4.1 server DRC */
-	#define NFSD_DRC_SIZE_SHIFT	7
-	nfsd_serv->sv_drc_max_pages = nr_free_buffer_pages()
-						>> NFSD_DRC_SIZE_SHIFT;
-	nfsd_serv->sv_drc_pages_used = 0;
-	dprintk("%s svc_drc_max_pages %u\n", __func__,
-		nfsd_serv->sv_drc_max_pages);
+	#define NFSD_DRC_SIZE_SHIFT	10
+	nfsd_drc_max_mem = (nr_free_buffer_pages()
+					>> NFSD_DRC_SIZE_SHIFT) * PAGE_SIZE;
+	nfsd_drc_mem_used = 0;
+	spin_lock_init(&nfsd_drc_lock);
+	dprintk("%s nfsd_drc_max_mem %u \n", __func__, nfsd_drc_max_mem);
 }
 
 int nfsd_create_serv(void)
@@ -401,7 +411,9 @@ nfsd_svc(unsigned short port, int nrservs)
 	error =	nfsd_racache_init(2*nrservs);
 	if (error<0)
 		goto out;
-	nfs4_state_start();
+	error = nfs4_state_start();
+	if (error)
+		goto out;
 
 	nfsd_reset_versions();
 
@@ -569,10 +581,6 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
 		+ rqstp->rq_res.head[0].iov_len;
 	rqstp->rq_res.head[0].iov_len += sizeof(__be32);
 
-	/* NFSv4.1 DRC requires statp */
-	if (rqstp->rq_vers == 4)
-		nfsd4_set_statp(rqstp, statp);
-
 	/* Now call the procedure handler, and encode NFS status. */
 	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
 	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
@@ -607,7 +615,25 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
 
 int nfsd_pool_stats_open(struct inode *inode, struct file *file)
 {
-	if (nfsd_serv == NULL)
+	int ret;
+	mutex_lock(&nfsd_mutex);
+	if (nfsd_serv == NULL) {
+		mutex_unlock(&nfsd_mutex);
 		return -ENODEV;
-	return svc_pool_stats_open(nfsd_serv, file);
+	}
+	/* bump up the psudo refcount while traversing */
+	svc_get(nfsd_serv);
+	ret = svc_pool_stats_open(nfsd_serv, file);
+	mutex_unlock(&nfsd_mutex);
+	return ret;
+}
+
+int nfsd_pool_stats_release(struct inode *inode, struct file *file)
+{
+	int ret = seq_release(inode, file);
+	mutex_lock(&nfsd_mutex);
+	/* this function really, really should have been called svc_put() */
+	svc_destroy(nfsd_serv);
+	mutex_unlock(&nfsd_mutex);
+	return ret;
 }
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 8fa09bfbcba7..a293f0273263 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -89,6 +89,12 @@ struct raparm_hbucket {
 #define RAPARM_HASH_MASK	(RAPARM_HASH_SIZE-1)
 static struct raparm_hbucket	raparm_hash[RAPARM_HASH_SIZE];
 
+static inline int
+nfsd_v4client(struct svc_rqst *rq)
+{
+    return rq->rq_prog == NFS_PROGRAM && rq->rq_vers == 4;
+}
+
 /* 
  * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
  * a mount point.
@@ -115,7 +121,8 @@ nfsd_cross_mnt(struct svc_rqst *rqstp, struct dentry **dpp,
 		path_put(&path);
 		goto out;
 	}
-	if ((exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
+	if (nfsd_v4client(rqstp) ||
+		(exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
 		/* successfully crossed mount point */
 		/*
 		 * This is subtle: path.dentry is *not* on path.mnt
diff --git a/fs/nilfs2/btnode.c b/fs/nilfs2/btnode.c
index c668bca579c1..5941958f1e47 100644
--- a/fs/nilfs2/btnode.c
+++ b/fs/nilfs2/btnode.c
@@ -36,6 +36,7 @@
 
 void nilfs_btnode_cache_init_once(struct address_space *btnc)
 {
+	memset(btnc, 0, sizeof(*btnc));
 	INIT_RADIX_TREE(&btnc->page_tree, GFP_ATOMIC);
 	spin_lock_init(&btnc->tree_lock);
 	INIT_LIST_HEAD(&btnc->private_list);
@@ -46,7 +47,7 @@ void nilfs_btnode_cache_init_once(struct address_space *btnc)
 	INIT_LIST_HEAD(&btnc->i_mmap_nonlinear);
 }
 
-static struct address_space_operations def_btnode_aops = {
+static const struct address_space_operations def_btnode_aops = {
 	.sync_page		= block_sync_page,
 };
 
diff --git a/fs/nilfs2/dir.c b/fs/nilfs2/dir.c
index 1a4fa04cf071..e097099bfc8f 100644
--- a/fs/nilfs2/dir.c
+++ b/fs/nilfs2/dir.c
@@ -697,7 +697,7 @@ not_empty:
 	return 0;
 }
 
-struct file_operations nilfs_dir_operations = {
+const struct file_operations nilfs_dir_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
 	.readdir	= nilfs_readdir,
diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c
index 6bd84a0d8238..30292df443ce 100644
--- a/fs/nilfs2/file.c
+++ b/fs/nilfs2/file.c
@@ -117,7 +117,7 @@ static int nilfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	return 0;
 }
 
-struct vm_operations_struct nilfs_file_vm_ops = {
+static const struct vm_operations_struct nilfs_file_vm_ops = {
 	.fault		= filemap_fault,
 	.page_mkwrite	= nilfs_page_mkwrite,
 };
@@ -134,7 +134,7 @@ static int nilfs_file_mmap(struct file *file, struct vm_area_struct *vma)
  * We have mostly NULL's here: the current defaults are ok for
  * the nilfs filesystem.
  */
-struct file_operations nilfs_file_operations = {
+const struct file_operations nilfs_file_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= do_sync_read,
 	.write		= do_sync_write,
@@ -151,7 +151,7 @@ struct file_operations nilfs_file_operations = {
 	.splice_read	= generic_file_splice_read,
 };
 
-struct inode_operations nilfs_file_inode_operations = {
+const struct inode_operations nilfs_file_inode_operations = {
 	.truncate	= nilfs_truncate,
 	.setattr	= nilfs_setattr,
 	.permission     = nilfs_permission,
diff --git a/fs/nilfs2/gcinode.c b/fs/nilfs2/gcinode.c
index 1b3c2bb20da9..e6de0a27ab5d 100644
--- a/fs/nilfs2/gcinode.c
+++ b/fs/nilfs2/gcinode.c
@@ -52,7 +52,7 @@
 #include "dat.h"
 #include "ifile.h"
 
-static struct address_space_operations def_gcinode_aops = {
+static const struct address_space_operations def_gcinode_aops = {
 	.sync_page		= block_sync_page,
 };
 
diff --git a/fs/nilfs2/inode.c b/fs/nilfs2/inode.c
index 807e584b163d..5040220c3732 100644
--- a/fs/nilfs2/inode.c
+++ b/fs/nilfs2/inode.c
@@ -238,7 +238,7 @@ nilfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
 	return size;
 }
 
-struct address_space_operations nilfs_aops = {
+const struct address_space_operations nilfs_aops = {
 	.writepage		= nilfs_writepage,
 	.readpage		= nilfs_readpage,
 	.sync_page		= block_sync_page,
@@ -400,6 +400,7 @@ int nilfs_read_inode_common(struct inode *inode,
 	ii->i_dir_acl = S_ISREG(inode->i_mode) ?
 		0 : le32_to_cpu(raw_inode->i_dir_acl);
 #endif
+	ii->i_dir_start_lookup = 0;
 	ii->i_cno = 0;
 	inode->i_generation = le32_to_cpu(raw_inode->i_generation);
 
diff --git a/fs/nilfs2/mdt.c b/fs/nilfs2/mdt.c
index 156bf6091a96..f6326112d647 100644
--- a/fs/nilfs2/mdt.c
+++ b/fs/nilfs2/mdt.c
@@ -427,13 +427,13 @@ nilfs_mdt_write_page(struct page *page, struct writeback_control *wbc)
 }
 
 
-static struct address_space_operations def_mdt_aops = {
+static const struct address_space_operations def_mdt_aops = {
 	.writepage		= nilfs_mdt_write_page,
 	.sync_page		= block_sync_page,
 };
 
-static struct inode_operations def_mdt_iops;
-static struct file_operations def_mdt_fops;
+static const struct inode_operations def_mdt_iops;
+static const struct file_operations def_mdt_fops;
 
 /*
  * NILFS2 uses pseudo inodes for meta data files such as DAT, cpfile, sufile,
diff --git a/fs/nilfs2/namei.c b/fs/nilfs2/namei.c
index df70dadb336f..ed02e886fa79 100644
--- a/fs/nilfs2/namei.c
+++ b/fs/nilfs2/namei.c
@@ -448,7 +448,7 @@ out:
 	return err;
 }
 
-struct inode_operations nilfs_dir_inode_operations = {
+const struct inode_operations nilfs_dir_inode_operations = {
 	.create		= nilfs_create,
 	.lookup		= nilfs_lookup,
 	.link		= nilfs_link,
@@ -462,12 +462,12 @@ struct inode_operations nilfs_dir_inode_operations = {
 	.permission	= nilfs_permission,
 };
 
-struct inode_operations nilfs_special_inode_operations = {
+const struct inode_operations nilfs_special_inode_operations = {
 	.setattr	= nilfs_setattr,
 	.permission	= nilfs_permission,
 };
 
-struct inode_operations nilfs_symlink_inode_operations = {
+const struct inode_operations nilfs_symlink_inode_operations = {
 	.readlink	= generic_readlink,
 	.follow_link	= page_follow_link_light,
 	.put_link	= page_put_link,
diff --git a/fs/nilfs2/nilfs.h b/fs/nilfs2/nilfs.h
index 724c63766e82..4da6f67e9a91 100644
--- a/fs/nilfs2/nilfs.h
+++ b/fs/nilfs2/nilfs.h
@@ -294,13 +294,13 @@ void nilfs_clear_gcdat_inode(struct the_nilfs *);
 /*
  * Inodes and files operations
  */
-extern struct file_operations nilfs_dir_operations;
-extern struct inode_operations nilfs_file_inode_operations;
-extern struct file_operations nilfs_file_operations;
-extern struct address_space_operations nilfs_aops;
-extern struct inode_operations nilfs_dir_inode_operations;
-extern struct inode_operations nilfs_special_inode_operations;
-extern struct inode_operations nilfs_symlink_inode_operations;
+extern const struct file_operations nilfs_dir_operations;
+extern const struct inode_operations nilfs_file_inode_operations;
+extern const struct file_operations nilfs_file_operations;
+extern const struct address_space_operations nilfs_aops;
+extern const struct inode_operations nilfs_dir_inode_operations;
+extern const struct inode_operations nilfs_special_inode_operations;
+extern const struct inode_operations nilfs_symlink_inode_operations;
 
 /*
  * filesystem type
diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
index 55f3d6b60732..644e66727dd0 100644
--- a/fs/nilfs2/super.c
+++ b/fs/nilfs2/super.c
@@ -504,7 +504,7 @@ static int nilfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
 	return 0;
 }
 
-static struct super_operations nilfs_sops = {
+static const struct super_operations nilfs_sops = {
 	.alloc_inode    = nilfs_alloc_inode,
 	.destroy_inode  = nilfs_destroy_inode,
 	.dirty_inode    = nilfs_dirty_inode,
@@ -560,7 +560,7 @@ nilfs_fh_to_parent(struct super_block *sb, struct fid *fid, int fh_len,
 				    nilfs_nfs_get_inode);
 }
 
-static struct export_operations nilfs_export_ops = {
+static const struct export_operations nilfs_export_ops = {
 	.fh_to_dentry = nilfs_fh_to_dentry,
 	.fh_to_parent = nilfs_fh_to_parent,
 	.get_parent = nilfs_get_parent,
diff --git a/fs/nls/nls_base.c b/fs/nls/nls_base.c
index 477d37d83b31..44a88a9fa2c8 100644
--- a/fs/nls/nls_base.c
+++ b/fs/nls/nls_base.c
@@ -124,10 +124,10 @@ int utf8s_to_utf16s(const u8 *s, int len, wchar_t *pwcs)
 	while (*s && len > 0) {
 		if (*s & 0x80) {
 			size = utf8_to_utf32(s, len, &u);
-			if (size < 0) {
-				/* Ignore character and move on */
-				size = 1;
-			} else if (u >= PLANE_SIZE) {
+			if (size < 0)
+				return -EINVAL;
+
+			if (u >= PLANE_SIZE) {
 				u -= PLANE_SIZE;
 				*op++ = (wchar_t) (SURROGATE_PAIR |
 						((u >> 10) & SURROGATE_BITS));
@@ -270,7 +270,8 @@ struct nls_table *load_nls(char *charset)
 
 void unload_nls(struct nls_table *nls)
 {
-	module_put(nls->owner);
+	if (nls)
+		module_put(nls->owner);
 }
 
 static const wchar_t charset2uni[256] = {
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index 828a889be909..7e54e52964dd 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -91,6 +91,7 @@ static int dnotify_handle_event(struct fsnotify_group *group,
 	struct dnotify_struct *dn;
 	struct dnotify_struct **prev;
 	struct fown_struct *fown;
+	__u32 test_mask = event->mask & ~FS_EVENT_ON_CHILD;
 
 	to_tell = event->to_tell;
 
@@ -106,7 +107,7 @@ static int dnotify_handle_event(struct fsnotify_group *group,
 	spin_lock(&entry->lock);
 	prev = &dnentry->dn;
 	while ((dn = *prev) != NULL) {
-		if ((dn->dn_mask & event->mask) == 0) {
+		if ((dn->dn_mask & test_mask) == 0) {
 			prev = &dn->dn_next;
 			continue;
 		}
diff --git a/fs/notify/inode_mark.c b/fs/notify/inode_mark.c
index c8a07c65482b..3165d85aada2 100644
--- a/fs/notify/inode_mark.c
+++ b/fs/notify/inode_mark.c
@@ -324,11 +324,11 @@ int fsnotify_add_mark(struct fsnotify_mark_entry *entry,
 	spin_lock(&group->mark_lock);
 	spin_lock(&inode->i_lock);
 
-	entry->group = group;
-	entry->inode = inode;
-
 	lentry = fsnotify_find_mark_entry(group, inode);
 	if (!lentry) {
+		entry->group = group;
+		entry->inode = inode;
+
 		hlist_add_head(&entry->i_list, &inode->i_fsnotify_mark_entries);
 		list_add(&entry->g_list, &group->mark_entries);
 
diff --git a/fs/notify/notification.c b/fs/notify/notification.c
index 3816d5750dd5..b8bf53b4c108 100644
--- a/fs/notify/notification.c
+++ b/fs/notify/notification.c
@@ -143,7 +143,7 @@ static bool event_compare(struct fsnotify_event *old, struct fsnotify_event *new
 			/* remember, after old was put on the wait_q we aren't
 			 * allowed to look at the inode any more, only thing
 			 * left to check was if the file_name is the same */
-			if (old->name_len &&
+			if (!old->name_len ||
 			    !strcmp(old->file_name, new->file_name))
 				return true;
 			break;
diff --git a/fs/ntfs/aops.c b/fs/ntfs/aops.c
index b38f944f0667..cfce53cb65d7 100644
--- a/fs/ntfs/aops.c
+++ b/fs/ntfs/aops.c
@@ -1550,6 +1550,7 @@ const struct address_space_operations ntfs_aops = {
 	.migratepage	= buffer_migrate_page,	/* Move a page cache page from
 						   one physical page to an
 						   other. */
+	.error_remove_page = generic_error_remove_page,
 };
 
 /**
@@ -1569,6 +1570,7 @@ const struct address_space_operations ntfs_mst_aops = {
 	.migratepage	= buffer_migrate_page,	/* Move a page cache page from
 						   one physical page to an
 						   other. */
+	.error_remove_page = generic_error_remove_page,
 };
 
 #ifdef NTFS_RW
diff --git a/fs/ntfs/file.c b/fs/ntfs/file.c
index 4350d4993b18..663c0e341f8b 100644
--- a/fs/ntfs/file.c
+++ b/fs/ntfs/file.c
@@ -2146,46 +2146,6 @@ static ssize_t ntfs_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 }
 
 /**
- * ntfs_file_writev -
- *
- * Basically the same as generic_file_writev() except that it ends up calling
- * ntfs_file_aio_write_nolock() instead of __generic_file_aio_write_nolock().
- */
-static ssize_t ntfs_file_writev(struct file *file, const struct iovec *iov,
-		unsigned long nr_segs, loff_t *ppos)
-{
-	struct address_space *mapping = file->f_mapping;
-	struct inode *inode = mapping->host;
-	struct kiocb kiocb;
-	ssize_t ret;
-
-	mutex_lock(&inode->i_mutex);
-	init_sync_kiocb(&kiocb, file);
-	ret = ntfs_file_aio_write_nolock(&kiocb, iov, nr_segs, ppos);
-	if (ret == -EIOCBQUEUED)
-		ret = wait_on_sync_kiocb(&kiocb);
-	mutex_unlock(&inode->i_mutex);
-	if (ret > 0) {
-		int err = generic_write_sync(file, *ppos - ret, ret);
-		if (err < 0)
-			ret = err;
-	}
-	return ret;
-}
-
-/**
- * ntfs_file_write - simple wrapper for ntfs_file_writev()
- */
-static ssize_t ntfs_file_write(struct file *file, const char __user *buf,
-		size_t count, loff_t *ppos)
-{
-	struct iovec local_iov = { .iov_base = (void __user *)buf,
-				   .iov_len = count };
-
-	return ntfs_file_writev(file, &local_iov, 1, ppos);
-}
-
-/**
  * ntfs_file_fsync - sync a file to disk
  * @filp:	file to be synced
  * @dentry:	dentry describing the file to sync
@@ -2247,7 +2207,7 @@ const struct file_operations ntfs_file_ops = {
 	.read		= do_sync_read,		 /* Read from file. */
 	.aio_read	= generic_file_aio_read, /* Async read from file. */
 #ifdef NTFS_RW
-	.write		= ntfs_file_write,	 /* Write to file. */
+	.write		= do_sync_write,	 /* Write to file. */
 	.aio_write	= ntfs_file_aio_write,	 /* Async write to file. */
 	/*.release	= ,*/			 /* Last file is closed.  See
 						    fs/ext2/file.c::
diff --git a/fs/ntfs/layout.h b/fs/ntfs/layout.h
index 50931b1ce4b9..8b2549f672bf 100644
--- a/fs/ntfs/layout.h
+++ b/fs/ntfs/layout.h
@@ -829,7 +829,7 @@ enum {
 	/* Note, FILE_ATTR_VALID_SET_FLAGS masks out the old DOS VolId, the
 	   F_A_DEVICE, F_A_DIRECTORY, F_A_SPARSE_FILE, F_A_REPARSE_POINT,
 	   F_A_COMPRESSED, and F_A_ENCRYPTED and preserves the rest.  This mask
-	   is used to to obtain all flags that are valid for setting. */
+	   is used to obtain all flags that are valid for setting. */
 	/*
 	 * The flag FILE_ATTR_DUP_FILENAME_INDEX_PRESENT is present in all
 	 * FILENAME_ATTR attributes but not in the STANDARD_INFORMATION
diff --git a/fs/ntfs/malloc.h b/fs/ntfs/malloc.h
index cd0be3f5c3cd..a44b14cbceeb 100644
--- a/fs/ntfs/malloc.h
+++ b/fs/ntfs/malloc.h
@@ -47,7 +47,7 @@ static inline void *__ntfs_malloc(unsigned long size, gfp_t gfp_mask)
 		return kmalloc(PAGE_SIZE, gfp_mask & ~__GFP_HIGHMEM);
 		/* return (void *)__get_free_page(gfp_mask); */
 	}
-	if (likely(size >> PAGE_SHIFT < num_physpages))
+	if (likely((size >> PAGE_SHIFT) < totalram_pages))
 		return __vmalloc(size, gfp_mask, PAGE_KERNEL);
 	return NULL;
 }
diff --git a/fs/ntfs/super.c b/fs/ntfs/super.c
index abaaa1cbf8de..80b04770e8e9 100644
--- a/fs/ntfs/super.c
+++ b/fs/ntfs/super.c
@@ -201,8 +201,7 @@ use_utf8:
 						v, old_nls->charset);
 				nls_map = old_nls;
 			} else /* nls_map */ {
-				if (old_nls)
-					unload_nls(old_nls);
+				unload_nls(old_nls);
 			}
 		} else if (!strcmp(p, "utf8")) {
 			bool val = false;
@@ -2427,10 +2426,9 @@ static void ntfs_put_super(struct super_block *sb)
 		ntfs_free(vol->upcase);
 		vol->upcase = NULL;
 	}
-	if (vol->nls_map) {
-		unload_nls(vol->nls_map);
-		vol->nls_map = NULL;
-	}
+
+	unload_nls(vol->nls_map);
+
 	sb->s_fs_info = NULL;
 	kfree(vol);
 
diff --git a/fs/ocfs2/Makefile b/fs/ocfs2/Makefile
index 01596079dd63..31f25ce32c97 100644
--- a/fs/ocfs2/Makefile
+++ b/fs/ocfs2/Makefile
@@ -28,6 +28,7 @@ ocfs2-objs := \
 	locks.o			\
 	mmap.o 			\
 	namei.o 		\
+	refcounttree.o		\
 	resize.o		\
 	slot_map.o 		\
 	suballoc.o 		\
diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index ab513ddaeff2..38a42f5d59ff 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -49,10 +49,21 @@
 #include "super.h"
 #include "uptodate.h"
 #include "xattr.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
+enum ocfs2_contig_type {
+	CONTIG_NONE = 0,
+	CONTIG_LEFT,
+	CONTIG_RIGHT,
+	CONTIG_LEFTRIGHT,
+};
 
+static enum ocfs2_contig_type
+	ocfs2_extent_rec_contig(struct super_block *sb,
+				struct ocfs2_extent_rec *ext,
+				struct ocfs2_extent_rec *insert_rec);
 /*
  * Operations for a specific extent tree type.
  *
@@ -79,18 +90,30 @@ struct ocfs2_extent_tree_operations {
 	 * that value.  new_clusters is the delta, and must be
 	 * added to the total.  Required.
 	 */
-	void (*eo_update_clusters)(struct inode *inode,
-				   struct ocfs2_extent_tree *et,
+	void (*eo_update_clusters)(struct ocfs2_extent_tree *et,
 				   u32 new_clusters);
 
 	/*
+	 * If this extent tree is supported by an extent map, insert
+	 * a record into the map.
+	 */
+	void (*eo_extent_map_insert)(struct ocfs2_extent_tree *et,
+				     struct ocfs2_extent_rec *rec);
+
+	/*
+	 * If this extent tree is supported by an extent map, truncate the
+	 * map to clusters,
+	 */
+	void (*eo_extent_map_truncate)(struct ocfs2_extent_tree *et,
+				       u32 clusters);
+
+	/*
 	 * If ->eo_insert_check() exists, it is called before rec is
 	 * inserted into the extent tree.  It is optional.
 	 */
-	int (*eo_insert_check)(struct inode *inode,
-			       struct ocfs2_extent_tree *et,
+	int (*eo_insert_check)(struct ocfs2_extent_tree *et,
 			       struct ocfs2_extent_rec *rec);
-	int (*eo_sanity_check)(struct inode *inode, struct ocfs2_extent_tree *et);
+	int (*eo_sanity_check)(struct ocfs2_extent_tree *et);
 
 	/*
 	 * --------------------------------------------------------------
@@ -109,8 +132,17 @@ struct ocfs2_extent_tree_operations {
 	 * it exists.  If it does not, et->et_max_leaf_clusters is set
 	 * to 0 (unlimited).  Optional.
 	 */
-	void (*eo_fill_max_leaf_clusters)(struct inode *inode,
-					  struct ocfs2_extent_tree *et);
+	void (*eo_fill_max_leaf_clusters)(struct ocfs2_extent_tree *et);
+
+	/*
+	 * ->eo_extent_contig test whether the 2 ocfs2_extent_rec
+	 * are contiguous or not. Optional. Don't need to set it if use
+	 * ocfs2_extent_rec as the tree leaf.
+	 */
+	enum ocfs2_contig_type
+		(*eo_extent_contig)(struct ocfs2_extent_tree *et,
+				    struct ocfs2_extent_rec *ext,
+				    struct ocfs2_extent_rec *insert_rec);
 };
 
 
@@ -121,19 +153,22 @@ struct ocfs2_extent_tree_operations {
 static u64 ocfs2_dinode_get_last_eb_blk(struct ocfs2_extent_tree *et);
 static void ocfs2_dinode_set_last_eb_blk(struct ocfs2_extent_tree *et,
 					 u64 blkno);
-static void ocfs2_dinode_update_clusters(struct inode *inode,
-					 struct ocfs2_extent_tree *et,
+static void ocfs2_dinode_update_clusters(struct ocfs2_extent_tree *et,
 					 u32 clusters);
-static int ocfs2_dinode_insert_check(struct inode *inode,
-				     struct ocfs2_extent_tree *et,
+static void ocfs2_dinode_extent_map_insert(struct ocfs2_extent_tree *et,
+					   struct ocfs2_extent_rec *rec);
+static void ocfs2_dinode_extent_map_truncate(struct ocfs2_extent_tree *et,
+					     u32 clusters);
+static int ocfs2_dinode_insert_check(struct ocfs2_extent_tree *et,
 				     struct ocfs2_extent_rec *rec);
-static int ocfs2_dinode_sanity_check(struct inode *inode,
-				     struct ocfs2_extent_tree *et);
+static int ocfs2_dinode_sanity_check(struct ocfs2_extent_tree *et);
 static void ocfs2_dinode_fill_root_el(struct ocfs2_extent_tree *et);
 static struct ocfs2_extent_tree_operations ocfs2_dinode_et_ops = {
 	.eo_set_last_eb_blk	= ocfs2_dinode_set_last_eb_blk,
 	.eo_get_last_eb_blk	= ocfs2_dinode_get_last_eb_blk,
 	.eo_update_clusters	= ocfs2_dinode_update_clusters,
+	.eo_extent_map_insert	= ocfs2_dinode_extent_map_insert,
+	.eo_extent_map_truncate	= ocfs2_dinode_extent_map_truncate,
 	.eo_insert_check	= ocfs2_dinode_insert_check,
 	.eo_sanity_check	= ocfs2_dinode_sanity_check,
 	.eo_fill_root_el	= ocfs2_dinode_fill_root_el,
@@ -156,40 +191,53 @@ static u64 ocfs2_dinode_get_last_eb_blk(struct ocfs2_extent_tree *et)
 	return le64_to_cpu(di->i_last_eb_blk);
 }
 
-static void ocfs2_dinode_update_clusters(struct inode *inode,
-					 struct ocfs2_extent_tree *et,
+static void ocfs2_dinode_update_clusters(struct ocfs2_extent_tree *et,
 					 u32 clusters)
 {
+	struct ocfs2_inode_info *oi = cache_info_to_inode(et->et_ci);
 	struct ocfs2_dinode *di = et->et_object;
 
 	le32_add_cpu(&di->i_clusters, clusters);
-	spin_lock(&OCFS2_I(inode)->ip_lock);
-	OCFS2_I(inode)->ip_clusters = le32_to_cpu(di->i_clusters);
-	spin_unlock(&OCFS2_I(inode)->ip_lock);
+	spin_lock(&oi->ip_lock);
+	oi->ip_clusters = le32_to_cpu(di->i_clusters);
+	spin_unlock(&oi->ip_lock);
 }
 
-static int ocfs2_dinode_insert_check(struct inode *inode,
-				     struct ocfs2_extent_tree *et,
+static void ocfs2_dinode_extent_map_insert(struct ocfs2_extent_tree *et,
+					   struct ocfs2_extent_rec *rec)
+{
+	struct inode *inode = &cache_info_to_inode(et->et_ci)->vfs_inode;
+
+	ocfs2_extent_map_insert_rec(inode, rec);
+}
+
+static void ocfs2_dinode_extent_map_truncate(struct ocfs2_extent_tree *et,
+					     u32 clusters)
+{
+	struct inode *inode = &cache_info_to_inode(et->et_ci)->vfs_inode;
+
+	ocfs2_extent_map_trunc(inode, clusters);
+}
+
+static int ocfs2_dinode_insert_check(struct ocfs2_extent_tree *et,
 				     struct ocfs2_extent_rec *rec)
 {
-	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_inode_info *oi = cache_info_to_inode(et->et_ci);
+	struct ocfs2_super *osb = OCFS2_SB(oi->vfs_inode.i_sb);
 
-	BUG_ON(OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL);
+	BUG_ON(oi->ip_dyn_features & OCFS2_INLINE_DATA_FL);
 	mlog_bug_on_msg(!ocfs2_sparse_alloc(osb) &&
-			(OCFS2_I(inode)->ip_clusters !=
-			 le32_to_cpu(rec->e_cpos)),
+			(oi->ip_clusters != le32_to_cpu(rec->e_cpos)),
 			"Device %s, asking for sparse allocation: inode %llu, "
 			"cpos %u, clusters %u\n",
 			osb->dev_str,
-			(unsigned long long)OCFS2_I(inode)->ip_blkno,
-			rec->e_cpos,
-			OCFS2_I(inode)->ip_clusters);
+			(unsigned long long)oi->ip_blkno,
+			rec->e_cpos, oi->ip_clusters);
 
 	return 0;
 }
 
-static int ocfs2_dinode_sanity_check(struct inode *inode,
-				     struct ocfs2_extent_tree *et)
+static int ocfs2_dinode_sanity_check(struct ocfs2_extent_tree *et)
 {
 	struct ocfs2_dinode *di = et->et_object;
 
@@ -229,8 +277,7 @@ static u64 ocfs2_xattr_value_get_last_eb_blk(struct ocfs2_extent_tree *et)
 	return le64_to_cpu(vb->vb_xv->xr_last_eb_blk);
 }
 
-static void ocfs2_xattr_value_update_clusters(struct inode *inode,
-					      struct ocfs2_extent_tree *et,
+static void ocfs2_xattr_value_update_clusters(struct ocfs2_extent_tree *et,
 					      u32 clusters)
 {
 	struct ocfs2_xattr_value_buf *vb = et->et_object;
@@ -252,12 +299,11 @@ static void ocfs2_xattr_tree_fill_root_el(struct ocfs2_extent_tree *et)
 	et->et_root_el = &xb->xb_attrs.xb_root.xt_list;
 }
 
-static void ocfs2_xattr_tree_fill_max_leaf_clusters(struct inode *inode,
-						    struct ocfs2_extent_tree *et)
+static void ocfs2_xattr_tree_fill_max_leaf_clusters(struct ocfs2_extent_tree *et)
 {
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
 	et->et_max_leaf_clusters =
-		ocfs2_clusters_for_bytes(inode->i_sb,
-					 OCFS2_MAX_XATTR_TREE_LEAF_SIZE);
+		ocfs2_clusters_for_bytes(sb, OCFS2_MAX_XATTR_TREE_LEAF_SIZE);
 }
 
 static void ocfs2_xattr_tree_set_last_eb_blk(struct ocfs2_extent_tree *et,
@@ -277,8 +323,7 @@ static u64 ocfs2_xattr_tree_get_last_eb_blk(struct ocfs2_extent_tree *et)
 	return le64_to_cpu(xt->xt_last_eb_blk);
 }
 
-static void ocfs2_xattr_tree_update_clusters(struct inode *inode,
-					     struct ocfs2_extent_tree *et,
+static void ocfs2_xattr_tree_update_clusters(struct ocfs2_extent_tree *et,
 					     u32 clusters)
 {
 	struct ocfs2_xattr_block *xb = et->et_object;
@@ -309,8 +354,7 @@ static u64 ocfs2_dx_root_get_last_eb_blk(struct ocfs2_extent_tree *et)
 	return le64_to_cpu(dx_root->dr_last_eb_blk);
 }
 
-static void ocfs2_dx_root_update_clusters(struct inode *inode,
-					  struct ocfs2_extent_tree *et,
+static void ocfs2_dx_root_update_clusters(struct ocfs2_extent_tree *et,
 					  u32 clusters)
 {
 	struct ocfs2_dx_root_block *dx_root = et->et_object;
@@ -318,8 +362,7 @@ static void ocfs2_dx_root_update_clusters(struct inode *inode,
 	le32_add_cpu(&dx_root->dr_clusters, clusters);
 }
 
-static int ocfs2_dx_root_sanity_check(struct inode *inode,
-				      struct ocfs2_extent_tree *et)
+static int ocfs2_dx_root_sanity_check(struct ocfs2_extent_tree *et)
 {
 	struct ocfs2_dx_root_block *dx_root = et->et_object;
 
@@ -343,8 +386,54 @@ static struct ocfs2_extent_tree_operations ocfs2_dx_root_et_ops = {
 	.eo_fill_root_el	= ocfs2_dx_root_fill_root_el,
 };
 
+static void ocfs2_refcount_tree_fill_root_el(struct ocfs2_extent_tree *et)
+{
+	struct ocfs2_refcount_block *rb = et->et_object;
+
+	et->et_root_el = &rb->rf_list;
+}
+
+static void ocfs2_refcount_tree_set_last_eb_blk(struct ocfs2_extent_tree *et,
+						u64 blkno)
+{
+	struct ocfs2_refcount_block *rb = et->et_object;
+
+	rb->rf_last_eb_blk = cpu_to_le64(blkno);
+}
+
+static u64 ocfs2_refcount_tree_get_last_eb_blk(struct ocfs2_extent_tree *et)
+{
+	struct ocfs2_refcount_block *rb = et->et_object;
+
+	return le64_to_cpu(rb->rf_last_eb_blk);
+}
+
+static void ocfs2_refcount_tree_update_clusters(struct ocfs2_extent_tree *et,
+						u32 clusters)
+{
+	struct ocfs2_refcount_block *rb = et->et_object;
+
+	le32_add_cpu(&rb->rf_clusters, clusters);
+}
+
+static enum ocfs2_contig_type
+ocfs2_refcount_tree_extent_contig(struct ocfs2_extent_tree *et,
+				  struct ocfs2_extent_rec *ext,
+				  struct ocfs2_extent_rec *insert_rec)
+{
+	return CONTIG_NONE;
+}
+
+static struct ocfs2_extent_tree_operations ocfs2_refcount_tree_et_ops = {
+	.eo_set_last_eb_blk	= ocfs2_refcount_tree_set_last_eb_blk,
+	.eo_get_last_eb_blk	= ocfs2_refcount_tree_get_last_eb_blk,
+	.eo_update_clusters	= ocfs2_refcount_tree_update_clusters,
+	.eo_fill_root_el	= ocfs2_refcount_tree_fill_root_el,
+	.eo_extent_contig	= ocfs2_refcount_tree_extent_contig,
+};
+
 static void __ocfs2_init_extent_tree(struct ocfs2_extent_tree *et,
-				     struct inode *inode,
+				     struct ocfs2_caching_info *ci,
 				     struct buffer_head *bh,
 				     ocfs2_journal_access_func access,
 				     void *obj,
@@ -352,6 +441,7 @@ static void __ocfs2_init_extent_tree(struct ocfs2_extent_tree *et,
 {
 	et->et_ops = ops;
 	et->et_root_bh = bh;
+	et->et_ci = ci;
 	et->et_root_journal_access = access;
 	if (!obj)
 		obj = (void *)bh->b_data;
@@ -361,41 +451,49 @@ static void __ocfs2_init_extent_tree(struct ocfs2_extent_tree *et,
 	if (!et->et_ops->eo_fill_max_leaf_clusters)
 		et->et_max_leaf_clusters = 0;
 	else
-		et->et_ops->eo_fill_max_leaf_clusters(inode, et);
+		et->et_ops->eo_fill_max_leaf_clusters(et);
 }
 
 void ocfs2_init_dinode_extent_tree(struct ocfs2_extent_tree *et,
-				   struct inode *inode,
+				   struct ocfs2_caching_info *ci,
 				   struct buffer_head *bh)
 {
-	__ocfs2_init_extent_tree(et, inode, bh, ocfs2_journal_access_di,
+	__ocfs2_init_extent_tree(et, ci, bh, ocfs2_journal_access_di,
 				 NULL, &ocfs2_dinode_et_ops);
 }
 
 void ocfs2_init_xattr_tree_extent_tree(struct ocfs2_extent_tree *et,
-				       struct inode *inode,
+				       struct ocfs2_caching_info *ci,
 				       struct buffer_head *bh)
 {
-	__ocfs2_init_extent_tree(et, inode, bh, ocfs2_journal_access_xb,
+	__ocfs2_init_extent_tree(et, ci, bh, ocfs2_journal_access_xb,
 				 NULL, &ocfs2_xattr_tree_et_ops);
 }
 
 void ocfs2_init_xattr_value_extent_tree(struct ocfs2_extent_tree *et,
-					struct inode *inode,
+					struct ocfs2_caching_info *ci,
 					struct ocfs2_xattr_value_buf *vb)
 {
-	__ocfs2_init_extent_tree(et, inode, vb->vb_bh, vb->vb_access, vb,
+	__ocfs2_init_extent_tree(et, ci, vb->vb_bh, vb->vb_access, vb,
 				 &ocfs2_xattr_value_et_ops);
 }
 
 void ocfs2_init_dx_root_extent_tree(struct ocfs2_extent_tree *et,
-				    struct inode *inode,
+				    struct ocfs2_caching_info *ci,
 				    struct buffer_head *bh)
 {
-	__ocfs2_init_extent_tree(et, inode, bh, ocfs2_journal_access_dr,
+	__ocfs2_init_extent_tree(et, ci, bh, ocfs2_journal_access_dr,
 				 NULL, &ocfs2_dx_root_et_ops);
 }
 
+void ocfs2_init_refcount_extent_tree(struct ocfs2_extent_tree *et,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *bh)
+{
+	__ocfs2_init_extent_tree(et, ci, bh, ocfs2_journal_access_rb,
+				 NULL, &ocfs2_refcount_tree_et_ops);
+}
+
 static inline void ocfs2_et_set_last_eb_blk(struct ocfs2_extent_tree *et,
 					    u64 new_last_eb_blk)
 {
@@ -407,78 +505,71 @@ static inline u64 ocfs2_et_get_last_eb_blk(struct ocfs2_extent_tree *et)
 	return et->et_ops->eo_get_last_eb_blk(et);
 }
 
-static inline void ocfs2_et_update_clusters(struct inode *inode,
-					    struct ocfs2_extent_tree *et,
+static inline void ocfs2_et_update_clusters(struct ocfs2_extent_tree *et,
 					    u32 clusters)
 {
-	et->et_ops->eo_update_clusters(inode, et, clusters);
+	et->et_ops->eo_update_clusters(et, clusters);
+}
+
+static inline void ocfs2_et_extent_map_insert(struct ocfs2_extent_tree *et,
+					      struct ocfs2_extent_rec *rec)
+{
+	if (et->et_ops->eo_extent_map_insert)
+		et->et_ops->eo_extent_map_insert(et, rec);
+}
+
+static inline void ocfs2_et_extent_map_truncate(struct ocfs2_extent_tree *et,
+						u32 clusters)
+{
+	if (et->et_ops->eo_extent_map_truncate)
+		et->et_ops->eo_extent_map_truncate(et, clusters);
 }
 
 static inline int ocfs2_et_root_journal_access(handle_t *handle,
-					       struct inode *inode,
 					       struct ocfs2_extent_tree *et,
 					       int type)
 {
-	return et->et_root_journal_access(handle, inode, et->et_root_bh,
+	return et->et_root_journal_access(handle, et->et_ci, et->et_root_bh,
 					  type);
 }
 
-static inline int ocfs2_et_insert_check(struct inode *inode,
-					struct ocfs2_extent_tree *et,
+static inline enum ocfs2_contig_type
+	ocfs2_et_extent_contig(struct ocfs2_extent_tree *et,
+			       struct ocfs2_extent_rec *rec,
+			       struct ocfs2_extent_rec *insert_rec)
+{
+	if (et->et_ops->eo_extent_contig)
+		return et->et_ops->eo_extent_contig(et, rec, insert_rec);
+
+	return ocfs2_extent_rec_contig(
+				ocfs2_metadata_cache_get_super(et->et_ci),
+				rec, insert_rec);
+}
+
+static inline int ocfs2_et_insert_check(struct ocfs2_extent_tree *et,
 					struct ocfs2_extent_rec *rec)
 {
 	int ret = 0;
 
 	if (et->et_ops->eo_insert_check)
-		ret = et->et_ops->eo_insert_check(inode, et, rec);
+		ret = et->et_ops->eo_insert_check(et, rec);
 	return ret;
 }
 
-static inline int ocfs2_et_sanity_check(struct inode *inode,
-					struct ocfs2_extent_tree *et)
+static inline int ocfs2_et_sanity_check(struct ocfs2_extent_tree *et)
 {
 	int ret = 0;
 
 	if (et->et_ops->eo_sanity_check)
-		ret = et->et_ops->eo_sanity_check(inode, et);
+		ret = et->et_ops->eo_sanity_check(et);
 	return ret;
 }
 
 static void ocfs2_free_truncate_context(struct ocfs2_truncate_context *tc);
 static int ocfs2_cache_extent_block_free(struct ocfs2_cached_dealloc_ctxt *ctxt,
 					 struct ocfs2_extent_block *eb);
-
-/*
- * Structures which describe a path through a btree, and functions to
- * manipulate them.
- *
- * The idea here is to be as generic as possible with the tree
- * manipulation code.
- */
-struct ocfs2_path_item {
-	struct buffer_head		*bh;
-	struct ocfs2_extent_list	*el;
-};
-
-#define OCFS2_MAX_PATH_DEPTH	5
-
-struct ocfs2_path {
-	int				p_tree_depth;
-	ocfs2_journal_access_func	p_root_access;
-	struct ocfs2_path_item		p_node[OCFS2_MAX_PATH_DEPTH];
-};
-
-#define path_root_bh(_path) ((_path)->p_node[0].bh)
-#define path_root_el(_path) ((_path)->p_node[0].el)
-#define path_root_access(_path)((_path)->p_root_access)
-#define path_leaf_bh(_path) ((_path)->p_node[(_path)->p_tree_depth].bh)
-#define path_leaf_el(_path) ((_path)->p_node[(_path)->p_tree_depth].el)
-#define path_num_items(_path) ((_path)->p_tree_depth + 1)
-
-static int ocfs2_find_path(struct inode *inode, struct ocfs2_path *path,
-			   u32 cpos);
-static void ocfs2_adjust_rightmost_records(struct inode *inode,
-					   handle_t *handle,
+static void ocfs2_adjust_rightmost_records(handle_t *handle,
+					   struct ocfs2_extent_tree *et,
 					   struct ocfs2_path *path,
 					   struct ocfs2_extent_rec *insert_rec);
 /*
@@ -486,7 +577,7 @@ static void ocfs2_adjust_rightmost_records(struct inode *inode,
  * to build another path. Generally, this involves freeing the buffer
  * heads.
  */
-static void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root)
+void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root)
 {
 	int i, start = 0, depth = 0;
 	struct ocfs2_path_item *node;
@@ -515,7 +606,7 @@ static void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root)
 	path->p_tree_depth = depth;
 }
 
-static void ocfs2_free_path(struct ocfs2_path *path)
+void ocfs2_free_path(struct ocfs2_path *path)
 {
 	if (path) {
 		ocfs2_reinit_path(path, 0);
@@ -613,13 +704,13 @@ static struct ocfs2_path *ocfs2_new_path(struct buffer_head *root_bh,
 	return path;
 }
 
-static struct ocfs2_path *ocfs2_new_path_from_path(struct ocfs2_path *path)
+struct ocfs2_path *ocfs2_new_path_from_path(struct ocfs2_path *path)
 {
 	return ocfs2_new_path(path_root_bh(path), path_root_el(path),
 			      path_root_access(path));
 }
 
-static struct ocfs2_path *ocfs2_new_path_from_et(struct ocfs2_extent_tree *et)
+struct ocfs2_path *ocfs2_new_path_from_et(struct ocfs2_extent_tree *et)
 {
 	return ocfs2_new_path(et->et_root_bh, et->et_root_el,
 			      et->et_root_journal_access);
@@ -632,10 +723,10 @@ static struct ocfs2_path *ocfs2_new_path_from_et(struct ocfs2_extent_tree *et)
  * I don't like the way this function's name looks next to
  * ocfs2_journal_access_path(), but I don't have a better one.
  */
-static int ocfs2_path_bh_journal_access(handle_t *handle,
-					struct inode *inode,
-					struct ocfs2_path *path,
-					int idx)
+int ocfs2_path_bh_journal_access(handle_t *handle,
+				 struct ocfs2_caching_info *ci,
+				 struct ocfs2_path *path,
+				 int idx)
 {
 	ocfs2_journal_access_func access = path_root_access(path);
 
@@ -645,15 +736,16 @@ static int ocfs2_path_bh_journal_access(handle_t *handle,
 	if (idx)
 		access = ocfs2_journal_access_eb;
 
-	return access(handle, inode, path->p_node[idx].bh,
+	return access(handle, ci, path->p_node[idx].bh,
 		      OCFS2_JOURNAL_ACCESS_WRITE);
 }
 
 /*
  * Convenience function to journal all components in a path.
  */
-static int ocfs2_journal_access_path(struct inode *inode, handle_t *handle,
-				     struct ocfs2_path *path)
+int ocfs2_journal_access_path(struct ocfs2_caching_info *ci,
+			      handle_t *handle,
+			      struct ocfs2_path *path)
 {
 	int i, ret = 0;
 
@@ -661,7 +753,7 @@ static int ocfs2_journal_access_path(struct inode *inode, handle_t *handle,
 		goto out;
 
 	for(i = 0; i < path_num_items(path); i++) {
-		ret = ocfs2_path_bh_journal_access(handle, inode, path, i);
+		ret = ocfs2_path_bh_journal_access(handle, ci, path, i);
 		if (ret < 0) {
 			mlog_errno(ret);
 			goto out;
@@ -702,17 +794,9 @@ int ocfs2_search_extent_list(struct ocfs2_extent_list *el, u32 v_cluster)
 	return ret;
 }
 
-enum ocfs2_contig_type {
-	CONTIG_NONE = 0,
-	CONTIG_LEFT,
-	CONTIG_RIGHT,
-	CONTIG_LEFTRIGHT,
-};
-
-
 /*
  * NOTE: ocfs2_block_extent_contig(), ocfs2_extents_adjacent() and
- * ocfs2_extent_contig only work properly against leaf nodes!
+ * ocfs2_extent_rec_contig only work properly against leaf nodes!
  */
 static int ocfs2_block_extent_contig(struct super_block *sb,
 				     struct ocfs2_extent_rec *ext,
@@ -738,9 +822,9 @@ static int ocfs2_extents_adjacent(struct ocfs2_extent_rec *left,
 }
 
 static enum ocfs2_contig_type
-	ocfs2_extent_contig(struct inode *inode,
-			    struct ocfs2_extent_rec *ext,
-			    struct ocfs2_extent_rec *insert_rec)
+	ocfs2_extent_rec_contig(struct super_block *sb,
+				struct ocfs2_extent_rec *ext,
+				struct ocfs2_extent_rec *insert_rec)
 {
 	u64 blkno = le64_to_cpu(insert_rec->e_blkno);
 
@@ -753,12 +837,12 @@ static enum ocfs2_contig_type
 		return CONTIG_NONE;
 
 	if (ocfs2_extents_adjacent(ext, insert_rec) &&
-	    ocfs2_block_extent_contig(inode->i_sb, ext, blkno))
+	    ocfs2_block_extent_contig(sb, ext, blkno))
 			return CONTIG_RIGHT;
 
 	blkno = le64_to_cpu(ext->e_blkno);
 	if (ocfs2_extents_adjacent(insert_rec, ext) &&
-	    ocfs2_block_extent_contig(inode->i_sb, insert_rec, blkno))
+	    ocfs2_block_extent_contig(sb, insert_rec, blkno))
 		return CONTIG_LEFT;
 
 	return CONTIG_NONE;
@@ -853,13 +937,13 @@ static int ocfs2_validate_extent_block(struct super_block *sb,
 	return 0;
 }
 
-int ocfs2_read_extent_block(struct inode *inode, u64 eb_blkno,
+int ocfs2_read_extent_block(struct ocfs2_caching_info *ci, u64 eb_blkno,
 			    struct buffer_head **bh)
 {
 	int rc;
 	struct buffer_head *tmp = *bh;
 
-	rc = ocfs2_read_block(inode, eb_blkno, &tmp,
+	rc = ocfs2_read_block(ci, eb_blkno, &tmp,
 			      ocfs2_validate_extent_block);
 
 	/* If ocfs2_read_block() got us a new bh, pass it up. */
@@ -874,7 +958,6 @@ int ocfs2_read_extent_block(struct inode *inode, u64 eb_blkno,
  * How many free extents have we got before we need more meta data?
  */
 int ocfs2_num_free_extents(struct ocfs2_super *osb,
-			   struct inode *inode,
 			   struct ocfs2_extent_tree *et)
 {
 	int retval;
@@ -889,7 +972,8 @@ int ocfs2_num_free_extents(struct ocfs2_super *osb,
 	last_eb_blk = ocfs2_et_get_last_eb_blk(et);
 
 	if (last_eb_blk) {
-		retval = ocfs2_read_extent_block(inode, last_eb_blk, &eb_bh);
+		retval = ocfs2_read_extent_block(et->et_ci, last_eb_blk,
+						 &eb_bh);
 		if (retval < 0) {
 			mlog_errno(retval);
 			goto bail;
@@ -913,9 +997,8 @@ bail:
  * sets h_signature, h_blkno, h_suballoc_bit, h_suballoc_slot, and
  * l_count for you
  */
-static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
-				     handle_t *handle,
-				     struct inode *inode,
+static int ocfs2_create_new_meta_bhs(handle_t *handle,
+				     struct ocfs2_extent_tree *et,
 				     int wanted,
 				     struct ocfs2_alloc_context *meta_ac,
 				     struct buffer_head *bhs[])
@@ -924,6 +1007,8 @@ static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
 	u16 suballoc_bit_start;
 	u32 num_got;
 	u64 first_blkno;
+	struct ocfs2_super *osb =
+		OCFS2_SB(ocfs2_metadata_cache_get_super(et->et_ci));
 	struct ocfs2_extent_block *eb;
 
 	mlog_entry_void();
@@ -949,9 +1034,10 @@ static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
 				mlog_errno(status);
 				goto bail;
 			}
-			ocfs2_set_new_buffer_uptodate(inode, bhs[i]);
+			ocfs2_set_new_buffer_uptodate(et->et_ci, bhs[i]);
 
-			status = ocfs2_journal_access_eb(handle, inode, bhs[i],
+			status = ocfs2_journal_access_eb(handle, et->et_ci,
+							 bhs[i],
 							 OCFS2_JOURNAL_ACCESS_CREATE);
 			if (status < 0) {
 				mlog_errno(status);
@@ -1023,7 +1109,6 @@ static inline u32 ocfs2_sum_rightmost_rec(struct ocfs2_extent_list  *el)
  * extent block's rightmost record.
  */
 static int ocfs2_adjust_rightmost_branch(handle_t *handle,
-					 struct inode *inode,
 					 struct ocfs2_extent_tree *et)
 {
 	int status;
@@ -1037,7 +1122,7 @@ static int ocfs2_adjust_rightmost_branch(handle_t *handle,
 		return status;
 	}
 
-	status = ocfs2_find_path(inode, path, UINT_MAX);
+	status = ocfs2_find_path(et->et_ci, path, UINT_MAX);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out;
@@ -1050,7 +1135,7 @@ static int ocfs2_adjust_rightmost_branch(handle_t *handle,
 		goto out;
 	}
 
-	status = ocfs2_journal_access_path(inode, handle, path);
+	status = ocfs2_journal_access_path(et->et_ci, handle, path);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out;
@@ -1059,7 +1144,7 @@ static int ocfs2_adjust_rightmost_branch(handle_t *handle,
 	el = path_leaf_el(path);
 	rec = &el->l_recs[le32_to_cpu(el->l_next_free_rec) - 1];
 
-	ocfs2_adjust_rightmost_records(inode, handle, path, rec);
+	ocfs2_adjust_rightmost_records(handle, et, path, rec);
 
 out:
 	ocfs2_free_path(path);
@@ -1068,7 +1153,7 @@ out:
 
 /*
  * Add an entire tree branch to our inode. eb_bh is the extent block
- * to start at, if we don't want to start the branch at the dinode
+ * to start at, if we don't want to start the branch at the root
  * structure.
  *
  * last_eb_bh is required as we have to update it's next_leaf pointer
@@ -1077,9 +1162,7 @@ out:
  * the new branch will be 'empty' in the sense that every block will
  * contain a single record with cluster count == 0.
  */
-static int ocfs2_add_branch(struct ocfs2_super *osb,
-			    handle_t *handle,
-			    struct inode *inode,
+static int ocfs2_add_branch(handle_t *handle,
 			    struct ocfs2_extent_tree *et,
 			    struct buffer_head *eb_bh,
 			    struct buffer_head **last_eb_bh,
@@ -1123,7 +1206,7 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
 	if (root_end > new_cpos) {
 		mlog(0, "adjust the cluster end from %u to %u\n",
 		     root_end, new_cpos);
-		status = ocfs2_adjust_rightmost_branch(handle, inode, et);
+		status = ocfs2_adjust_rightmost_branch(handle, et);
 		if (status) {
 			mlog_errno(status);
 			goto bail;
@@ -1139,7 +1222,7 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	status = ocfs2_create_new_meta_bhs(osb, handle, inode, new_blocks,
+	status = ocfs2_create_new_meta_bhs(handle, et, new_blocks,
 					   meta_ac, new_eb_bhs);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1161,7 +1244,7 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
 		BUG_ON(!OCFS2_IS_VALID_EXTENT_BLOCK(eb));
 		eb_el = &eb->h_list;
 
-		status = ocfs2_journal_access_eb(handle, inode, bh,
+		status = ocfs2_journal_access_eb(handle, et->et_ci, bh,
 						 OCFS2_JOURNAL_ACCESS_CREATE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -1201,20 +1284,20 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
 	 * journal_dirty erroring as it won't unless we've aborted the
 	 * handle (in which case we would never be here) so reserving
 	 * the write with journal_access is all we need to do. */
-	status = ocfs2_journal_access_eb(handle, inode, *last_eb_bh,
+	status = ocfs2_journal_access_eb(handle, et->et_ci, *last_eb_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
 	}
-	status = ocfs2_et_root_journal_access(handle, inode, et,
+	status = ocfs2_et_root_journal_access(handle, et,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
 	}
 	if (eb_bh) {
-		status = ocfs2_journal_access_eb(handle, inode, eb_bh,
+		status = ocfs2_journal_access_eb(handle, et->et_ci, eb_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -1274,9 +1357,7 @@ bail:
  * returns back the new extent block so you can add a branch to it
  * after this call.
  */
-static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
-				  handle_t *handle,
-				  struct inode *inode,
+static int ocfs2_shift_tree_depth(handle_t *handle,
 				  struct ocfs2_extent_tree *et,
 				  struct ocfs2_alloc_context *meta_ac,
 				  struct buffer_head **ret_new_eb_bh)
@@ -1290,7 +1371,7 @@ static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
 
 	mlog_entry_void();
 
-	status = ocfs2_create_new_meta_bhs(osb, handle, inode, 1, meta_ac,
+	status = ocfs2_create_new_meta_bhs(handle, et, 1, meta_ac,
 					   &new_eb_bh);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1304,7 +1385,7 @@ static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
 	eb_el = &eb->h_list;
 	root_el = et->et_root_el;
 
-	status = ocfs2_journal_access_eb(handle, inode, new_eb_bh,
+	status = ocfs2_journal_access_eb(handle, et->et_ci, new_eb_bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1323,7 +1404,7 @@ static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	status = ocfs2_et_root_journal_access(handle, inode, et,
+	status = ocfs2_et_root_journal_access(handle, et,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1379,9 +1460,7 @@ bail:
  *
  * return status < 0 indicates an error.
  */
-static int ocfs2_find_branch_target(struct ocfs2_super *osb,
-				    struct inode *inode,
-				    struct ocfs2_extent_tree *et,
+static int ocfs2_find_branch_target(struct ocfs2_extent_tree *et,
 				    struct buffer_head **target_bh)
 {
 	int status = 0, i;
@@ -1399,19 +1478,21 @@ static int ocfs2_find_branch_target(struct ocfs2_super *osb,
 
 	while(le16_to_cpu(el->l_tree_depth) > 1) {
 		if (le16_to_cpu(el->l_next_free_rec) == 0) {
-			ocfs2_error(inode->i_sb, "Dinode %llu has empty "
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu has empty "
 				    "extent list (next_free_rec == 0)",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno);
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci));
 			status = -EIO;
 			goto bail;
 		}
 		i = le16_to_cpu(el->l_next_free_rec) - 1;
 		blkno = le64_to_cpu(el->l_recs[i].e_blkno);
 		if (!blkno) {
-			ocfs2_error(inode->i_sb, "Dinode %llu has extent "
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu has extent "
 				    "list where extent # %d has no physical "
 				    "block start",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno, i);
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci), i);
 			status = -EIO;
 			goto bail;
 		}
@@ -1419,7 +1500,7 @@ static int ocfs2_find_branch_target(struct ocfs2_super *osb,
 		brelse(bh);
 		bh = NULL;
 
-		status = ocfs2_read_extent_block(inode, blkno, &bh);
+		status = ocfs2_read_extent_block(et->et_ci, blkno, &bh);
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail;
@@ -1460,20 +1541,18 @@ bail:
  *
  * *last_eb_bh will be updated by ocfs2_add_branch().
  */
-static int ocfs2_grow_tree(struct inode *inode, handle_t *handle,
-			   struct ocfs2_extent_tree *et, int *final_depth,
-			   struct buffer_head **last_eb_bh,
+static int ocfs2_grow_tree(handle_t *handle, struct ocfs2_extent_tree *et,
+			   int *final_depth, struct buffer_head **last_eb_bh,
 			   struct ocfs2_alloc_context *meta_ac)
 {
 	int ret, shift;
 	struct ocfs2_extent_list *el = et->et_root_el;
 	int depth = le16_to_cpu(el->l_tree_depth);
-	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct buffer_head *bh = NULL;
 
 	BUG_ON(meta_ac == NULL);
 
-	shift = ocfs2_find_branch_target(osb, inode, et, &bh);
+	shift = ocfs2_find_branch_target(et, &bh);
 	if (shift < 0) {
 		ret = shift;
 		mlog_errno(ret);
@@ -1490,8 +1569,7 @@ static int ocfs2_grow_tree(struct inode *inode, handle_t *handle,
 		/* ocfs2_shift_tree_depth will return us a buffer with
 		 * the new extent block (so we can pass that to
 		 * ocfs2_add_branch). */
-		ret = ocfs2_shift_tree_depth(osb, handle, inode, et,
-					     meta_ac, &bh);
+		ret = ocfs2_shift_tree_depth(handle, et, meta_ac, &bh);
 		if (ret < 0) {
 			mlog_errno(ret);
 			goto out;
@@ -1517,7 +1595,7 @@ static int ocfs2_grow_tree(struct inode *inode, handle_t *handle,
 	/* call ocfs2_add_branch to add the final part of the tree with
 	 * the new data. */
 	mlog(0, "add branch. bh = %p\n", bh);
-	ret = ocfs2_add_branch(osb, handle, inode, et, bh, last_eb_bh,
+	ret = ocfs2_add_branch(handle, et, bh, last_eb_bh,
 			       meta_ac);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -1687,7 +1765,7 @@ set_and_inc:
  *
  * The array index of the subtree root is passed back.
  */
-static int ocfs2_find_subtree_root(struct inode *inode,
+static int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
 				   struct ocfs2_path *left,
 				   struct ocfs2_path *right)
 {
@@ -1705,10 +1783,10 @@ static int ocfs2_find_subtree_root(struct inode *inode,
 		 * The caller didn't pass two adjacent paths.
 		 */
 		mlog_bug_on_msg(i > left->p_tree_depth,
-				"Inode %lu, left depth %u, right depth %u\n"
+				"Owner %llu, left depth %u, right depth %u\n"
 				"left leaf blk %llu, right leaf blk %llu\n",
-				inode->i_ino, left->p_tree_depth,
-				right->p_tree_depth,
+				(unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+				left->p_tree_depth, right->p_tree_depth,
 				(unsigned long long)path_leaf_bh(left)->b_blocknr,
 				(unsigned long long)path_leaf_bh(right)->b_blocknr);
 	} while (left->p_node[i].bh->b_blocknr ==
@@ -1725,7 +1803,7 @@ typedef void (path_insert_t)(void *, struct buffer_head *);
  * This code can be called with a cpos larger than the tree, in which
  * case it will return the rightmost path.
  */
-static int __ocfs2_find_path(struct inode *inode,
+static int __ocfs2_find_path(struct ocfs2_caching_info *ci,
 			     struct ocfs2_extent_list *root_el, u32 cpos,
 			     path_insert_t *func, void *data)
 {
@@ -1736,15 +1814,14 @@ static int __ocfs2_find_path(struct inode *inode,
 	struct ocfs2_extent_block *eb;
 	struct ocfs2_extent_list *el;
 	struct ocfs2_extent_rec *rec;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 
 	el = root_el;
 	while (el->l_tree_depth) {
 		if (le16_to_cpu(el->l_next_free_rec) == 0) {
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu has empty extent list at "
+			ocfs2_error(ocfs2_metadata_cache_get_super(ci),
+				    "Owner %llu has empty extent list at "
 				    "depth %u\n",
-				    (unsigned long long)oi->ip_blkno,
+				    (unsigned long long)ocfs2_metadata_cache_owner(ci),
 				    le16_to_cpu(el->l_tree_depth));
 			ret = -EROFS;
 			goto out;
@@ -1767,10 +1844,10 @@ static int __ocfs2_find_path(struct inode *inode,
 
 		blkno = le64_to_cpu(el->l_recs[i].e_blkno);
 		if (blkno == 0) {
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu has bad blkno in extent list "
+			ocfs2_error(ocfs2_metadata_cache_get_super(ci),
+				    "Owner %llu has bad blkno in extent list "
 				    "at depth %u (index %d)\n",
-				    (unsigned long long)oi->ip_blkno,
+				    (unsigned long long)ocfs2_metadata_cache_owner(ci),
 				    le16_to_cpu(el->l_tree_depth), i);
 			ret = -EROFS;
 			goto out;
@@ -1778,7 +1855,7 @@ static int __ocfs2_find_path(struct inode *inode,
 
 		brelse(bh);
 		bh = NULL;
-		ret = ocfs2_read_extent_block(inode, blkno, &bh);
+		ret = ocfs2_read_extent_block(ci, blkno, &bh);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -1789,10 +1866,10 @@ static int __ocfs2_find_path(struct inode *inode,
 
 		if (le16_to_cpu(el->l_next_free_rec) >
 		    le16_to_cpu(el->l_count)) {
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu has bad count in extent list "
+			ocfs2_error(ocfs2_metadata_cache_get_super(ci),
+				    "Owner %llu has bad count in extent list "
 				    "at block %llu (next free=%u, count=%u)\n",
-				    (unsigned long long)oi->ip_blkno,
+				    (unsigned long long)ocfs2_metadata_cache_owner(ci),
 				    (unsigned long long)bh->b_blocknr,
 				    le16_to_cpu(el->l_next_free_rec),
 				    le16_to_cpu(el->l_count));
@@ -1836,14 +1913,14 @@ static void find_path_ins(void *data, struct buffer_head *bh)
 	ocfs2_path_insert_eb(fp->path, fp->index, bh);
 	fp->index++;
 }
-static int ocfs2_find_path(struct inode *inode, struct ocfs2_path *path,
-			   u32 cpos)
+int ocfs2_find_path(struct ocfs2_caching_info *ci,
+		    struct ocfs2_path *path, u32 cpos)
 {
 	struct find_path_data data;
 
 	data.index = 1;
 	data.path = path;
-	return __ocfs2_find_path(inode, path_root_el(path), cpos,
+	return __ocfs2_find_path(ci, path_root_el(path), cpos,
 				 find_path_ins, &data);
 }
 
@@ -1868,13 +1945,14 @@ static void find_leaf_ins(void *data, struct buffer_head *bh)
  *
  * This function doesn't handle non btree extent lists.
  */
-int ocfs2_find_leaf(struct inode *inode, struct ocfs2_extent_list *root_el,
-		    u32 cpos, struct buffer_head **leaf_bh)
+int ocfs2_find_leaf(struct ocfs2_caching_info *ci,
+		    struct ocfs2_extent_list *root_el, u32 cpos,
+		    struct buffer_head **leaf_bh)
 {
 	int ret;
 	struct buffer_head *bh = NULL;
 
-	ret = __ocfs2_find_path(inode, root_el, cpos, find_leaf_ins, &bh);
+	ret = __ocfs2_find_path(ci, root_el, cpos, find_leaf_ins, &bh);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -1980,7 +2058,7 @@ static void ocfs2_adjust_root_records(struct ocfs2_extent_list *root_el,
  *   - When we've adjusted the last extent record in the left path leaf and the
  *     1st extent record in the right path leaf during cross extent block merge.
  */
-static void ocfs2_complete_edge_insert(struct inode *inode, handle_t *handle,
+static void ocfs2_complete_edge_insert(handle_t *handle,
 				       struct ocfs2_path *left_path,
 				       struct ocfs2_path *right_path,
 				       int subtree_index)
@@ -2058,8 +2136,8 @@ static void ocfs2_complete_edge_insert(struct inode *inode, handle_t *handle,
 		mlog_errno(ret);
 }
 
-static int ocfs2_rotate_subtree_right(struct inode *inode,
-				      handle_t *handle,
+static int ocfs2_rotate_subtree_right(handle_t *handle,
+				      struct ocfs2_extent_tree *et,
 				      struct ocfs2_path *left_path,
 				      struct ocfs2_path *right_path,
 				      int subtree_index)
@@ -2075,10 +2153,10 @@ static int ocfs2_rotate_subtree_right(struct inode *inode,
 	left_el = path_leaf_el(left_path);
 
 	if (left_el->l_next_free_rec != left_el->l_count) {
-		ocfs2_error(inode->i_sb,
+		ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
 			    "Inode %llu has non-full interior leaf node %llu"
 			    "(next free = %u)",
-			    (unsigned long long)OCFS2_I(inode)->ip_blkno,
+			    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 			    (unsigned long long)left_leaf_bh->b_blocknr,
 			    le16_to_cpu(left_el->l_next_free_rec));
 		return -EROFS;
@@ -2094,7 +2172,7 @@ static int ocfs2_rotate_subtree_right(struct inode *inode,
 	root_bh = left_path->p_node[subtree_index].bh;
 	BUG_ON(root_bh != right_path->p_node[subtree_index].bh);
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, right_path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, right_path,
 					   subtree_index);
 	if (ret) {
 		mlog_errno(ret);
@@ -2102,14 +2180,14 @@ static int ocfs2_rotate_subtree_right(struct inode *inode,
 	}
 
 	for(i = subtree_index + 1; i < path_num_items(right_path); i++) {
-		ret = ocfs2_path_bh_journal_access(handle, inode,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 						   right_path, i);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		ret = ocfs2_path_bh_journal_access(handle, inode,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 						   left_path, i);
 		if (ret) {
 			mlog_errno(ret);
@@ -2123,7 +2201,7 @@ static int ocfs2_rotate_subtree_right(struct inode *inode,
 	/* This is a code error, not a disk corruption. */
 	mlog_bug_on_msg(!right_el->l_next_free_rec, "Inode %llu: Rotate fails "
 			"because rightmost leaf block %llu is empty\n",
-			(unsigned long long)OCFS2_I(inode)->ip_blkno,
+			(unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 			(unsigned long long)right_leaf_bh->b_blocknr);
 
 	ocfs2_create_empty_extent(right_el);
@@ -2157,8 +2235,8 @@ static int ocfs2_rotate_subtree_right(struct inode *inode,
 		goto out;
 	}
 
-	ocfs2_complete_edge_insert(inode, handle, left_path, right_path,
-				subtree_index);
+	ocfs2_complete_edge_insert(handle, left_path, right_path,
+				   subtree_index);
 
 out:
 	return ret;
@@ -2248,10 +2326,18 @@ static int ocfs2_extend_rotate_transaction(handle_t *handle, int subtree_depth,
 					   int op_credits,
 					   struct ocfs2_path *path)
 {
+	int ret;
 	int credits = (path->p_tree_depth - subtree_depth) * 2 + 1 + op_credits;
 
-	if (handle->h_buffer_credits < credits)
-		return ocfs2_extend_trans(handle, credits);
+	if (handle->h_buffer_credits < credits) {
+		ret = ocfs2_extend_trans(handle,
+					 credits - handle->h_buffer_credits);
+		if (ret)
+			return ret;
+
+		if (unlikely(handle->h_buffer_credits < credits))
+			return ocfs2_extend_trans(handle, credits);
+	}
 
 	return 0;
 }
@@ -2321,8 +2407,8 @@ static int ocfs2_leftmost_rec_contains(struct ocfs2_extent_list *el, u32 cpos)
  *   *ret_left_path will contain a valid path which can be passed to
  *   ocfs2_insert_path().
  */
-static int ocfs2_rotate_tree_right(struct inode *inode,
-				   handle_t *handle,
+static int ocfs2_rotate_tree_right(handle_t *handle,
+				   struct ocfs2_extent_tree *et,
 				   enum ocfs2_split_type split,
 				   u32 insert_cpos,
 				   struct ocfs2_path *right_path,
@@ -2331,6 +2417,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 	int ret, start, orig_credits = handle->h_buffer_credits;
 	u32 cpos;
 	struct ocfs2_path *left_path = NULL;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
 
 	*ret_left_path = NULL;
 
@@ -2341,7 +2428,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path, &cpos);
+	ret = ocfs2_find_cpos_for_left_leaf(sb, right_path, &cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -2379,7 +2466,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 		mlog(0, "Rotating a tree: ins. cpos: %u, left path cpos: %u\n",
 		     insert_cpos, cpos);
 
-		ret = ocfs2_find_path(inode, left_path, cpos);
+		ret = ocfs2_find_path(et->et_ci, left_path, cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -2387,10 +2474,11 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 
 		mlog_bug_on_msg(path_leaf_bh(left_path) ==
 				path_leaf_bh(right_path),
-				"Inode %lu: error during insert of %u "
+				"Owner %llu: error during insert of %u "
 				"(left path cpos %u) results in two identical "
 				"paths ending at %llu\n",
-				inode->i_ino, insert_cpos, cpos,
+				(unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+				insert_cpos, cpos,
 				(unsigned long long)
 				path_leaf_bh(left_path)->b_blocknr);
 
@@ -2416,7 +2504,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 			goto out_ret_path;
 		}
 
-		start = ocfs2_find_subtree_root(inode, left_path, right_path);
+		start = ocfs2_find_subtree_root(et, left_path, right_path);
 
 		mlog(0, "Subtree root at index %d (blk %llu, depth %d)\n",
 		     start,
@@ -2430,7 +2518,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 			goto out;
 		}
 
-		ret = ocfs2_rotate_subtree_right(inode, handle, left_path,
+		ret = ocfs2_rotate_subtree_right(handle, et, left_path,
 						 right_path, start);
 		if (ret) {
 			mlog_errno(ret);
@@ -2462,8 +2550,7 @@ static int ocfs2_rotate_tree_right(struct inode *inode,
 		 */
 		ocfs2_mv_path(right_path, left_path);
 
-		ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path,
-						    &cpos);
+		ret = ocfs2_find_cpos_for_left_leaf(sb, right_path, &cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -2477,7 +2564,8 @@ out_ret_path:
 	return ret;
 }
 
-static int ocfs2_update_edge_lengths(struct inode *inode, handle_t *handle,
+static int ocfs2_update_edge_lengths(handle_t *handle,
+				     struct ocfs2_extent_tree *et,
 				     int subtree_index, struct ocfs2_path *path)
 {
 	int i, idx, ret;
@@ -2502,7 +2590,7 @@ static int ocfs2_update_edge_lengths(struct inode *inode, handle_t *handle,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_path(inode, handle, path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, path);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -2532,7 +2620,8 @@ out:
 	return ret;
 }
 
-static void ocfs2_unlink_path(struct inode *inode, handle_t *handle,
+static void ocfs2_unlink_path(handle_t *handle,
+			      struct ocfs2_extent_tree *et,
 			      struct ocfs2_cached_dealloc_ctxt *dealloc,
 			      struct ocfs2_path *path, int unlink_start)
 {
@@ -2554,12 +2643,12 @@ static void ocfs2_unlink_path(struct inode *inode, handle_t *handle,
 			mlog(ML_ERROR,
 			     "Inode %llu, attempted to remove extent block "
 			     "%llu with %u records\n",
-			     (unsigned long long)OCFS2_I(inode)->ip_blkno,
+			     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 			     (unsigned long long)le64_to_cpu(eb->h_blkno),
 			     le16_to_cpu(el->l_next_free_rec));
 
 			ocfs2_journal_dirty(handle, bh);
-			ocfs2_remove_from_cache(inode, bh);
+			ocfs2_remove_from_cache(et->et_ci, bh);
 			continue;
 		}
 
@@ -2572,11 +2661,12 @@ static void ocfs2_unlink_path(struct inode *inode, handle_t *handle,
 		if (ret)
 			mlog_errno(ret);
 
-		ocfs2_remove_from_cache(inode, bh);
+		ocfs2_remove_from_cache(et->et_ci, bh);
 	}
 }
 
-static void ocfs2_unlink_subtree(struct inode *inode, handle_t *handle,
+static void ocfs2_unlink_subtree(handle_t *handle,
+				 struct ocfs2_extent_tree *et,
 				 struct ocfs2_path *left_path,
 				 struct ocfs2_path *right_path,
 				 int subtree_index,
@@ -2607,17 +2697,17 @@ static void ocfs2_unlink_subtree(struct inode *inode, handle_t *handle,
 	ocfs2_journal_dirty(handle, root_bh);
 	ocfs2_journal_dirty(handle, path_leaf_bh(left_path));
 
-	ocfs2_unlink_path(inode, handle, dealloc, right_path,
+	ocfs2_unlink_path(handle, et, dealloc, right_path,
 			  subtree_index + 1);
 }
 
-static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
+static int ocfs2_rotate_subtree_left(handle_t *handle,
+				     struct ocfs2_extent_tree *et,
 				     struct ocfs2_path *left_path,
 				     struct ocfs2_path *right_path,
 				     int subtree_index,
 				     struct ocfs2_cached_dealloc_ctxt *dealloc,
-				     int *deleted,
-				     struct ocfs2_extent_tree *et)
+				     int *deleted)
 {
 	int ret, i, del_right_subtree = 0, right_has_empty = 0;
 	struct buffer_head *root_bh, *et_root_bh = path_root_bh(right_path);
@@ -2653,7 +2743,7 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 			return -EAGAIN;
 
 		if (le16_to_cpu(right_leaf_el->l_next_free_rec) > 1) {
-			ret = ocfs2_journal_access_eb(handle, inode,
+			ret = ocfs2_journal_access_eb(handle, et->et_ci,
 						      path_leaf_bh(right_path),
 						      OCFS2_JOURNAL_ACCESS_WRITE);
 			if (ret) {
@@ -2672,7 +2762,7 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 		 * We have to update i_last_eb_blk during the meta
 		 * data delete.
 		 */
-		ret = ocfs2_et_root_journal_access(handle, inode, et,
+		ret = ocfs2_et_root_journal_access(handle, et,
 						   OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
 			mlog_errno(ret);
@@ -2688,7 +2778,7 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 	 */
 	BUG_ON(right_has_empty && !del_right_subtree);
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, right_path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, right_path,
 					   subtree_index);
 	if (ret) {
 		mlog_errno(ret);
@@ -2696,14 +2786,14 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 	}
 
 	for(i = subtree_index + 1; i < path_num_items(right_path); i++) {
-		ret = ocfs2_path_bh_journal_access(handle, inode,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 						   right_path, i);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		ret = ocfs2_path_bh_journal_access(handle, inode,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 						   left_path, i);
 		if (ret) {
 			mlog_errno(ret);
@@ -2740,9 +2830,9 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 		mlog_errno(ret);
 
 	if (del_right_subtree) {
-		ocfs2_unlink_subtree(inode, handle, left_path, right_path,
+		ocfs2_unlink_subtree(handle, et, left_path, right_path,
 				     subtree_index, dealloc);
-		ret = ocfs2_update_edge_lengths(inode, handle, subtree_index,
+		ret = ocfs2_update_edge_lengths(handle, et, subtree_index,
 						left_path);
 		if (ret) {
 			mlog_errno(ret);
@@ -2766,7 +2856,7 @@ static int ocfs2_rotate_subtree_left(struct inode *inode, handle_t *handle,
 
 		*deleted = 1;
 	} else
-		ocfs2_complete_edge_insert(inode, handle, left_path, right_path,
+		ocfs2_complete_edge_insert(handle, left_path, right_path,
 					   subtree_index);
 
 out:
@@ -2852,8 +2942,8 @@ out:
 	return ret;
 }
 
-static int ocfs2_rotate_rightmost_leaf_left(struct inode *inode,
-					    handle_t *handle,
+static int ocfs2_rotate_rightmost_leaf_left(handle_t *handle,
+					    struct ocfs2_extent_tree *et,
 					    struct ocfs2_path *path)
 {
 	int ret;
@@ -2863,7 +2953,7 @@ static int ocfs2_rotate_rightmost_leaf_left(struct inode *inode,
 	if (!ocfs2_is_empty_extent(&el->l_recs[0]))
 		return 0;
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, path,
 					   path_num_items(path) - 1);
 	if (ret) {
 		mlog_errno(ret);
@@ -2880,24 +2970,24 @@ out:
 	return ret;
 }
 
-static int __ocfs2_rotate_tree_left(struct inode *inode,
-				    handle_t *handle, int orig_credits,
+static int __ocfs2_rotate_tree_left(handle_t *handle,
+				    struct ocfs2_extent_tree *et,
+				    int orig_credits,
 				    struct ocfs2_path *path,
 				    struct ocfs2_cached_dealloc_ctxt *dealloc,
-				    struct ocfs2_path **empty_extent_path,
-				    struct ocfs2_extent_tree *et)
+				    struct ocfs2_path **empty_extent_path)
 {
 	int ret, subtree_root, deleted;
 	u32 right_cpos;
 	struct ocfs2_path *left_path = NULL;
 	struct ocfs2_path *right_path = NULL;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
 
 	BUG_ON(!ocfs2_is_empty_extent(&(path_leaf_el(path)->l_recs[0])));
 
 	*empty_extent_path = NULL;
 
-	ret = ocfs2_find_cpos_for_right_leaf(inode->i_sb, path,
-					     &right_cpos);
+	ret = ocfs2_find_cpos_for_right_leaf(sb, path, &right_cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -2920,13 +3010,13 @@ static int __ocfs2_rotate_tree_left(struct inode *inode,
 	}
 
 	while (right_cpos) {
-		ret = ocfs2_find_path(inode, right_path, right_cpos);
+		ret = ocfs2_find_path(et->et_ci, right_path, right_cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		subtree_root = ocfs2_find_subtree_root(inode, left_path,
+		subtree_root = ocfs2_find_subtree_root(et, left_path,
 						       right_path);
 
 		mlog(0, "Subtree root at index %d (blk %llu, depth %d)\n",
@@ -2946,16 +3036,16 @@ static int __ocfs2_rotate_tree_left(struct inode *inode,
 		 * Caller might still want to make changes to the
 		 * tree root, so re-add it to the journal here.
 		 */
-		ret = ocfs2_path_bh_journal_access(handle, inode,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 						   left_path, 0);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		ret = ocfs2_rotate_subtree_left(inode, handle, left_path,
+		ret = ocfs2_rotate_subtree_left(handle, et, left_path,
 						right_path, subtree_root,
-						dealloc, &deleted, et);
+						dealloc, &deleted);
 		if (ret == -EAGAIN) {
 			/*
 			 * The rotation has to temporarily stop due to
@@ -2982,7 +3072,7 @@ static int __ocfs2_rotate_tree_left(struct inode *inode,
 
 		ocfs2_mv_path(left_path, right_path);
 
-		ret = ocfs2_find_cpos_for_right_leaf(inode->i_sb, left_path,
+		ret = ocfs2_find_cpos_for_right_leaf(sb, left_path,
 						     &right_cpos);
 		if (ret) {
 			mlog_errno(ret);
@@ -2997,10 +3087,10 @@ out:
 	return ret;
 }
 
-static int ocfs2_remove_rightmost_path(struct inode *inode, handle_t *handle,
+static int ocfs2_remove_rightmost_path(handle_t *handle,
+				struct ocfs2_extent_tree *et,
 				struct ocfs2_path *path,
-				struct ocfs2_cached_dealloc_ctxt *dealloc,
-				struct ocfs2_extent_tree *et)
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
 {
 	int ret, subtree_index;
 	u32 cpos;
@@ -3009,7 +3099,7 @@ static int ocfs2_remove_rightmost_path(struct inode *inode, handle_t *handle,
 	struct ocfs2_extent_list *el;
 
 
-	ret = ocfs2_et_sanity_check(inode, et);
+	ret = ocfs2_et_sanity_check(et);
 	if (ret)
 		goto out;
 	/*
@@ -3024,13 +3114,14 @@ static int ocfs2_remove_rightmost_path(struct inode *inode, handle_t *handle,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_path(inode, handle, path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, path);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
 	}
 
-	ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, path, &cpos);
+	ret = ocfs2_find_cpos_for_left_leaf(ocfs2_metadata_cache_get_super(et->et_ci),
+					    path, &cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -3048,23 +3139,23 @@ static int ocfs2_remove_rightmost_path(struct inode *inode, handle_t *handle,
 			goto out;
 		}
 
-		ret = ocfs2_find_path(inode, left_path, cpos);
+		ret = ocfs2_find_path(et->et_ci, left_path, cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		ret = ocfs2_journal_access_path(inode, handle, left_path);
+		ret = ocfs2_journal_access_path(et->et_ci, handle, left_path);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		subtree_index = ocfs2_find_subtree_root(inode, left_path, path);
+		subtree_index = ocfs2_find_subtree_root(et, left_path, path);
 
-		ocfs2_unlink_subtree(inode, handle, left_path, path,
+		ocfs2_unlink_subtree(handle, et, left_path, path,
 				     subtree_index, dealloc);
-		ret = ocfs2_update_edge_lengths(inode, handle, subtree_index,
+		ret = ocfs2_update_edge_lengths(handle, et, subtree_index,
 						left_path);
 		if (ret) {
 			mlog_errno(ret);
@@ -3078,10 +3169,10 @@ static int ocfs2_remove_rightmost_path(struct inode *inode, handle_t *handle,
 		 * 'path' is also the leftmost path which
 		 * means it must be the only one. This gets
 		 * handled differently because we want to
-		 * revert the inode back to having extents
+		 * revert the root back to having extents
 		 * in-line.
 		 */
-		ocfs2_unlink_path(inode, handle, dealloc, path, 1);
+		ocfs2_unlink_path(handle, et, dealloc, path, 1);
 
 		el = et->et_root_el;
 		el->l_tree_depth = 0;
@@ -3114,10 +3205,10 @@ out:
  * the rightmost tree leaf record is removed so the caller is
  * responsible for detecting and correcting that.
  */
-static int ocfs2_rotate_tree_left(struct inode *inode, handle_t *handle,
+static int ocfs2_rotate_tree_left(handle_t *handle,
+				  struct ocfs2_extent_tree *et,
 				  struct ocfs2_path *path,
-				  struct ocfs2_cached_dealloc_ctxt *dealloc,
-				  struct ocfs2_extent_tree *et)
+				  struct ocfs2_cached_dealloc_ctxt *dealloc)
 {
 	int ret, orig_credits = handle->h_buffer_credits;
 	struct ocfs2_path *tmp_path = NULL, *restart_path = NULL;
@@ -3134,8 +3225,7 @@ rightmost_no_delete:
 		 * Inline extents. This is trivially handled, so do
 		 * it up front.
 		 */
-		ret = ocfs2_rotate_rightmost_leaf_left(inode, handle,
-						       path);
+		ret = ocfs2_rotate_rightmost_leaf_left(handle, et, path);
 		if (ret)
 			mlog_errno(ret);
 		goto out;
@@ -3151,7 +3241,7 @@ rightmost_no_delete:
 	 *
 	 *  1) is handled via ocfs2_rotate_rightmost_leaf_left()
 	 *  2a) we need the left branch so that we can update it with the unlink
-	 *  2b) we need to bring the inode back to inline extents.
+	 *  2b) we need to bring the root back to inline extents.
 	 */
 
 	eb = (struct ocfs2_extent_block *)path_leaf_bh(path)->b_data;
@@ -3167,9 +3257,9 @@ rightmost_no_delete:
 
 		if (le16_to_cpu(el->l_next_free_rec) == 0) {
 			ret = -EIO;
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu has empty extent block at %llu",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno,
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu has empty extent block at %llu",
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 				    (unsigned long long)le64_to_cpu(eb->h_blkno));
 			goto out;
 		}
@@ -3183,8 +3273,8 @@ rightmost_no_delete:
 		 * nonempty list.
 		 */
 
-		ret = ocfs2_remove_rightmost_path(inode, handle, path,
-						  dealloc, et);
+		ret = ocfs2_remove_rightmost_path(handle, et, path,
+						  dealloc);
 		if (ret)
 			mlog_errno(ret);
 		goto out;
@@ -3195,8 +3285,8 @@ rightmost_no_delete:
 	 * and restarting from there.
 	 */
 try_rotate:
-	ret = __ocfs2_rotate_tree_left(inode, handle, orig_credits, path,
-				       dealloc, &restart_path, et);
+	ret = __ocfs2_rotate_tree_left(handle, et, orig_credits, path,
+				       dealloc, &restart_path);
 	if (ret && ret != -EAGAIN) {
 		mlog_errno(ret);
 		goto out;
@@ -3206,9 +3296,9 @@ try_rotate:
 		tmp_path = restart_path;
 		restart_path = NULL;
 
-		ret = __ocfs2_rotate_tree_left(inode, handle, orig_credits,
+		ret = __ocfs2_rotate_tree_left(handle, et, orig_credits,
 					       tmp_path, dealloc,
-					       &restart_path, et);
+					       &restart_path);
 		if (ret && ret != -EAGAIN) {
 			mlog_errno(ret);
 			goto out;
@@ -3259,7 +3349,7 @@ static void ocfs2_cleanup_merge(struct ocfs2_extent_list *el,
 	}
 }
 
-static int ocfs2_get_right_path(struct inode *inode,
+static int ocfs2_get_right_path(struct ocfs2_extent_tree *et,
 				struct ocfs2_path *left_path,
 				struct ocfs2_path **ret_right_path)
 {
@@ -3276,8 +3366,8 @@ static int ocfs2_get_right_path(struct inode *inode,
 	left_el = path_leaf_el(left_path);
 	BUG_ON(left_el->l_next_free_rec != left_el->l_count);
 
-	ret = ocfs2_find_cpos_for_right_leaf(inode->i_sb, left_path,
-					     &right_cpos);
+	ret = ocfs2_find_cpos_for_right_leaf(ocfs2_metadata_cache_get_super(et->et_ci),
+					     left_path, &right_cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -3293,7 +3383,7 @@ static int ocfs2_get_right_path(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_path(inode, right_path, right_cpos);
+	ret = ocfs2_find_path(et->et_ci, right_path, right_cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -3313,9 +3403,9 @@ out:
  * For index == l_count - 1, the "next" means the 1st extent rec of the
  * next extent block.
  */
-static int ocfs2_merge_rec_right(struct inode *inode,
-				 struct ocfs2_path *left_path,
+static int ocfs2_merge_rec_right(struct ocfs2_path *left_path,
 				 handle_t *handle,
+				 struct ocfs2_extent_tree *et,
 				 struct ocfs2_extent_rec *split_rec,
 				 int index)
 {
@@ -3336,7 +3426,7 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 	if (index == le16_to_cpu(el->l_next_free_rec) - 1 &&
 	    le16_to_cpu(el->l_next_free_rec) == le16_to_cpu(el->l_count)) {
 		/* we meet with a cross extent block merge. */
-		ret = ocfs2_get_right_path(inode, left_path, &right_path);
+		ret = ocfs2_get_right_path(et, left_path, &right_path);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -3355,8 +3445,8 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 		       le16_to_cpu(left_rec->e_leaf_clusters) !=
 		       le32_to_cpu(right_rec->e_cpos));
 
-		subtree_index = ocfs2_find_subtree_root(inode,
-							left_path, right_path);
+		subtree_index = ocfs2_find_subtree_root(et, left_path,
+							right_path);
 
 		ret = ocfs2_extend_rotate_transaction(handle, subtree_index,
 						      handle->h_buffer_credits,
@@ -3369,7 +3459,7 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 		root_bh = left_path->p_node[subtree_index].bh;
 		BUG_ON(root_bh != right_path->p_node[subtree_index].bh);
 
-		ret = ocfs2_path_bh_journal_access(handle, inode, right_path,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci, right_path,
 						   subtree_index);
 		if (ret) {
 			mlog_errno(ret);
@@ -3378,14 +3468,14 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 
 		for (i = subtree_index + 1;
 		     i < path_num_items(right_path); i++) {
-			ret = ocfs2_path_bh_journal_access(handle, inode,
+			ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 							   right_path, i);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
 			}
 
-			ret = ocfs2_path_bh_journal_access(handle, inode,
+			ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 							   left_path, i);
 			if (ret) {
 				mlog_errno(ret);
@@ -3398,7 +3488,7 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 		right_rec = &el->l_recs[index + 1];
 	}
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, left_path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, left_path,
 					   path_num_items(left_path) - 1);
 	if (ret) {
 		mlog_errno(ret);
@@ -3409,7 +3499,8 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 
 	le32_add_cpu(&right_rec->e_cpos, -split_clusters);
 	le64_add_cpu(&right_rec->e_blkno,
-		     -ocfs2_clusters_to_blocks(inode->i_sb, split_clusters));
+		     -ocfs2_clusters_to_blocks(ocfs2_metadata_cache_get_super(et->et_ci),
+					       split_clusters));
 	le16_add_cpu(&right_rec->e_leaf_clusters, split_clusters);
 
 	ocfs2_cleanup_merge(el, index);
@@ -3423,8 +3514,8 @@ static int ocfs2_merge_rec_right(struct inode *inode,
 		if (ret)
 			mlog_errno(ret);
 
-		ocfs2_complete_edge_insert(inode, handle, left_path,
-					   right_path, subtree_index);
+		ocfs2_complete_edge_insert(handle, left_path, right_path,
+					   subtree_index);
 	}
 out:
 	if (right_path)
@@ -3432,7 +3523,7 @@ out:
 	return ret;
 }
 
-static int ocfs2_get_left_path(struct inode *inode,
+static int ocfs2_get_left_path(struct ocfs2_extent_tree *et,
 			       struct ocfs2_path *right_path,
 			       struct ocfs2_path **ret_left_path)
 {
@@ -3445,7 +3536,7 @@ static int ocfs2_get_left_path(struct inode *inode,
 	/* This function shouldn't be called for non-trees. */
 	BUG_ON(right_path->p_tree_depth == 0);
 
-	ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb,
+	ret = ocfs2_find_cpos_for_left_leaf(ocfs2_metadata_cache_get_super(et->et_ci),
 					    right_path, &left_cpos);
 	if (ret) {
 		mlog_errno(ret);
@@ -3462,7 +3553,7 @@ static int ocfs2_get_left_path(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_path(inode, left_path, left_cpos);
+	ret = ocfs2_find_path(et->et_ci, left_path, left_cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -3485,12 +3576,11 @@ out:
  * remove the rightmost leaf extent block in the right_path and change
  * the right path to indicate the new rightmost path.
  */
-static int ocfs2_merge_rec_left(struct inode *inode,
-				struct ocfs2_path *right_path,
+static int ocfs2_merge_rec_left(struct ocfs2_path *right_path,
 				handle_t *handle,
+				struct ocfs2_extent_tree *et,
 				struct ocfs2_extent_rec *split_rec,
 				struct ocfs2_cached_dealloc_ctxt *dealloc,
-				struct ocfs2_extent_tree *et,
 				int index)
 {
 	int ret, i, subtree_index = 0, has_empty_extent = 0;
@@ -3508,7 +3598,7 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 	right_rec = &el->l_recs[index];
 	if (index == 0) {
 		/* we meet with a cross extent block merge. */
-		ret = ocfs2_get_left_path(inode, right_path, &left_path);
+		ret = ocfs2_get_left_path(et, right_path, &left_path);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -3524,8 +3614,8 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 		       le16_to_cpu(left_rec->e_leaf_clusters) !=
 		       le32_to_cpu(split_rec->e_cpos));
 
-		subtree_index = ocfs2_find_subtree_root(inode,
-							left_path, right_path);
+		subtree_index = ocfs2_find_subtree_root(et, left_path,
+							right_path);
 
 		ret = ocfs2_extend_rotate_transaction(handle, subtree_index,
 						      handle->h_buffer_credits,
@@ -3538,7 +3628,7 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 		root_bh = left_path->p_node[subtree_index].bh;
 		BUG_ON(root_bh != right_path->p_node[subtree_index].bh);
 
-		ret = ocfs2_path_bh_journal_access(handle, inode, right_path,
+		ret = ocfs2_path_bh_journal_access(handle, et->et_ci, right_path,
 						   subtree_index);
 		if (ret) {
 			mlog_errno(ret);
@@ -3547,14 +3637,14 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 
 		for (i = subtree_index + 1;
 		     i < path_num_items(right_path); i++) {
-			ret = ocfs2_path_bh_journal_access(handle, inode,
+			ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 							   right_path, i);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
 			}
 
-			ret = ocfs2_path_bh_journal_access(handle, inode,
+			ret = ocfs2_path_bh_journal_access(handle, et->et_ci,
 							   left_path, i);
 			if (ret) {
 				mlog_errno(ret);
@@ -3567,7 +3657,7 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 			has_empty_extent = 1;
 	}
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, right_path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, right_path,
 					   path_num_items(right_path) - 1);
 	if (ret) {
 		mlog_errno(ret);
@@ -3586,7 +3676,8 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 
 	le32_add_cpu(&right_rec->e_cpos, split_clusters);
 	le64_add_cpu(&right_rec->e_blkno,
-		     ocfs2_clusters_to_blocks(inode->i_sb, split_clusters));
+		     ocfs2_clusters_to_blocks(ocfs2_metadata_cache_get_super(et->et_ci),
+					      split_clusters));
 	le16_add_cpu(&right_rec->e_leaf_clusters, -split_clusters);
 
 	ocfs2_cleanup_merge(el, index);
@@ -3608,9 +3699,9 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 		if (le16_to_cpu(right_rec->e_leaf_clusters) == 0 &&
 		    le16_to_cpu(el->l_next_free_rec) == 1) {
 
-			ret = ocfs2_remove_rightmost_path(inode, handle,
+			ret = ocfs2_remove_rightmost_path(handle, et,
 							  right_path,
-							  dealloc, et);
+							  dealloc);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
@@ -3622,7 +3713,7 @@ static int ocfs2_merge_rec_left(struct inode *inode,
 			ocfs2_mv_path(right_path, left_path);
 			left_path = NULL;
 		} else
-			ocfs2_complete_edge_insert(inode, handle, left_path,
+			ocfs2_complete_edge_insert(handle, left_path,
 						   right_path, subtree_index);
 	}
 out:
@@ -3631,15 +3722,13 @@ out:
 	return ret;
 }
 
-static int ocfs2_try_to_merge_extent(struct inode *inode,
-				     handle_t *handle,
+static int ocfs2_try_to_merge_extent(handle_t *handle,
+				     struct ocfs2_extent_tree *et,
 				     struct ocfs2_path *path,
 				     int split_index,
 				     struct ocfs2_extent_rec *split_rec,
 				     struct ocfs2_cached_dealloc_ctxt *dealloc,
-				     struct ocfs2_merge_ctxt *ctxt,
-				     struct ocfs2_extent_tree *et)
-
+				     struct ocfs2_merge_ctxt *ctxt)
 {
 	int ret = 0;
 	struct ocfs2_extent_list *el = path_leaf_el(path);
@@ -3655,8 +3744,7 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 		 * extents - having more than one in a leaf is
 		 * illegal.
 		 */
-		ret = ocfs2_rotate_tree_left(inode, handle, path,
-					     dealloc, et);
+		ret = ocfs2_rotate_tree_left(handle, et, path, dealloc);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -3685,8 +3773,7 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 		 * prevoius extent block. It is more efficient and easier
 		 * if we do merge_right first and merge_left later.
 		 */
-		ret = ocfs2_merge_rec_right(inode, path,
-					    handle, split_rec,
+		ret = ocfs2_merge_rec_right(path, handle, et, split_rec,
 					    split_index);
 		if (ret) {
 			mlog_errno(ret);
@@ -3699,8 +3786,7 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 		BUG_ON(!ocfs2_is_empty_extent(&el->l_recs[0]));
 
 		/* The merge left us with an empty extent, remove it. */
-		ret = ocfs2_rotate_tree_left(inode, handle, path,
-					     dealloc, et);
+		ret = ocfs2_rotate_tree_left(handle, et, path, dealloc);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -3712,18 +3798,15 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 		 * Note that we don't pass split_rec here on purpose -
 		 * we've merged it into the rec already.
 		 */
-		ret = ocfs2_merge_rec_left(inode, path,
-					   handle, rec,
-					   dealloc, et,
-					   split_index);
+		ret = ocfs2_merge_rec_left(path, handle, et, rec,
+					   dealloc, split_index);
 
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
-		ret = ocfs2_rotate_tree_left(inode, handle, path,
-					     dealloc, et);
+		ret = ocfs2_rotate_tree_left(handle, et, path, dealloc);
 		/*
 		 * Error from this last rotate is not critical, so
 		 * print but don't bubble it up.
@@ -3740,19 +3823,16 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 		 * the record on the left (hence the left merge).
 		 */
 		if (ctxt->c_contig_type == CONTIG_RIGHT) {
-			ret = ocfs2_merge_rec_left(inode,
-						   path,
-						   handle, split_rec,
-						   dealloc, et,
+			ret = ocfs2_merge_rec_left(path, handle, et,
+						   split_rec, dealloc,
 						   split_index);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
 			}
 		} else {
-			ret = ocfs2_merge_rec_right(inode,
-						    path,
-						    handle, split_rec,
+			ret = ocfs2_merge_rec_right(path, handle,
+						    et, split_rec,
 						    split_index);
 			if (ret) {
 				mlog_errno(ret);
@@ -3765,8 +3845,8 @@ static int ocfs2_try_to_merge_extent(struct inode *inode,
 			 * The merge may have left an empty extent in
 			 * our leaf. Try to rotate it away.
 			 */
-			ret = ocfs2_rotate_tree_left(inode, handle, path,
-						     dealloc, et);
+			ret = ocfs2_rotate_tree_left(handle, et, path,
+						     dealloc);
 			if (ret)
 				mlog_errno(ret);
 			ret = 0;
@@ -3812,10 +3892,10 @@ static void ocfs2_subtract_from_rec(struct super_block *sb,
  * list. If this leaf is part of an allocation tree, it is assumed
  * that the tree above has been prepared.
  */
-static void ocfs2_insert_at_leaf(struct ocfs2_extent_rec *insert_rec,
+static void ocfs2_insert_at_leaf(struct ocfs2_extent_tree *et,
+				 struct ocfs2_extent_rec *insert_rec,
 				 struct ocfs2_extent_list *el,
-				 struct ocfs2_insert_type *insert,
-				 struct inode *inode)
+				 struct ocfs2_insert_type *insert)
 {
 	int i = insert->ins_contig_index;
 	unsigned int range;
@@ -3827,7 +3907,8 @@ static void ocfs2_insert_at_leaf(struct ocfs2_extent_rec *insert_rec,
 		i = ocfs2_search_extent_list(el, le32_to_cpu(insert_rec->e_cpos));
 		BUG_ON(i == -1);
 		rec = &el->l_recs[i];
-		ocfs2_subtract_from_rec(inode->i_sb, insert->ins_split, rec,
+		ocfs2_subtract_from_rec(ocfs2_metadata_cache_get_super(et->et_ci),
+					insert->ins_split, rec,
 					insert_rec);
 		goto rotate;
 	}
@@ -3869,10 +3950,10 @@ static void ocfs2_insert_at_leaf(struct ocfs2_extent_rec *insert_rec,
 
 		mlog_bug_on_msg(le16_to_cpu(el->l_next_free_rec) >=
 				le16_to_cpu(el->l_count),
-				"inode %lu, depth %u, count %u, next free %u, "
+				"owner %llu, depth %u, count %u, next free %u, "
 				"rec.cpos %u, rec.clusters %u, "
 				"insert.cpos %u, insert.clusters %u\n",
-				inode->i_ino,
+				ocfs2_metadata_cache_owner(et->et_ci),
 				le16_to_cpu(el->l_tree_depth),
 				le16_to_cpu(el->l_count),
 				le16_to_cpu(el->l_next_free_rec),
@@ -3900,8 +3981,8 @@ rotate:
 	ocfs2_rotate_leaf(el, insert_rec);
 }
 
-static void ocfs2_adjust_rightmost_records(struct inode *inode,
-					   handle_t *handle,
+static void ocfs2_adjust_rightmost_records(handle_t *handle,
+					   struct ocfs2_extent_tree *et,
 					   struct ocfs2_path *path,
 					   struct ocfs2_extent_rec *insert_rec)
 {
@@ -3919,9 +4000,9 @@ static void ocfs2_adjust_rightmost_records(struct inode *inode,
 
 		next_free = le16_to_cpu(el->l_next_free_rec);
 		if (next_free == 0) {
-			ocfs2_error(inode->i_sb,
-				    "Dinode %llu has a bad extent list",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno);
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu has a bad extent list",
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci));
 			ret = -EIO;
 			return;
 		}
@@ -3941,7 +4022,8 @@ static void ocfs2_adjust_rightmost_records(struct inode *inode,
 	}
 }
 
-static int ocfs2_append_rec_to_path(struct inode *inode, handle_t *handle,
+static int ocfs2_append_rec_to_path(handle_t *handle,
+				    struct ocfs2_extent_tree *et,
 				    struct ocfs2_extent_rec *insert_rec,
 				    struct ocfs2_path *right_path,
 				    struct ocfs2_path **ret_left_path)
@@ -3969,8 +4051,8 @@ static int ocfs2_append_rec_to_path(struct inode *inode, handle_t *handle,
 	    (next_free == 1 && ocfs2_is_empty_extent(&el->l_recs[0]))) {
 		u32 left_cpos;
 
-		ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path,
-						    &left_cpos);
+		ret = ocfs2_find_cpos_for_left_leaf(ocfs2_metadata_cache_get_super(et->et_ci),
+						    right_path, &left_cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -3992,7 +4074,8 @@ static int ocfs2_append_rec_to_path(struct inode *inode, handle_t *handle,
 				goto out;
 			}
 
-			ret = ocfs2_find_path(inode, left_path, left_cpos);
+			ret = ocfs2_find_path(et->et_ci, left_path,
+					      left_cpos);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
@@ -4005,13 +4088,13 @@ static int ocfs2_append_rec_to_path(struct inode *inode, handle_t *handle,
 		}
 	}
 
-	ret = ocfs2_journal_access_path(inode, handle, right_path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, right_path);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
 	}
 
-	ocfs2_adjust_rightmost_records(inode, handle, right_path, insert_rec);
+	ocfs2_adjust_rightmost_records(handle, et, right_path, insert_rec);
 
 	*ret_left_path = left_path;
 	ret = 0;
@@ -4022,7 +4105,7 @@ out:
 	return ret;
 }
 
-static void ocfs2_split_record(struct inode *inode,
+static void ocfs2_split_record(struct ocfs2_extent_tree *et,
 			       struct ocfs2_path *left_path,
 			       struct ocfs2_path *right_path,
 			       struct ocfs2_extent_rec *split_rec,
@@ -4095,7 +4178,8 @@ static void ocfs2_split_record(struct inode *inode,
 	}
 
 	rec = &el->l_recs[index];
-	ocfs2_subtract_from_rec(inode->i_sb, split, rec, split_rec);
+	ocfs2_subtract_from_rec(ocfs2_metadata_cache_get_super(et->et_ci),
+				split, rec, split_rec);
 	ocfs2_rotate_leaf(insert_el, split_rec);
 }
 
@@ -4107,8 +4191,8 @@ static void ocfs2_split_record(struct inode *inode,
  * in. left_path should only be passed in if we need to update that
  * portion of the tree after an edge insert.
  */
-static int ocfs2_insert_path(struct inode *inode,
-			     handle_t *handle,
+static int ocfs2_insert_path(handle_t *handle,
+			     struct ocfs2_extent_tree *et,
 			     struct ocfs2_path *left_path,
 			     struct ocfs2_path *right_path,
 			     struct ocfs2_extent_rec *insert_rec,
@@ -4134,7 +4218,7 @@ static int ocfs2_insert_path(struct inode *inode,
 			goto out;
 		}
 
-		ret = ocfs2_journal_access_path(inode, handle, left_path);
+		ret = ocfs2_journal_access_path(et->et_ci, handle, left_path);
 		if (ret < 0) {
 			mlog_errno(ret);
 			goto out;
@@ -4145,7 +4229,7 @@ static int ocfs2_insert_path(struct inode *inode,
 	 * Pass both paths to the journal. The majority of inserts
 	 * will be touching all components anyway.
 	 */
-	ret = ocfs2_journal_access_path(inode, handle, right_path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, right_path);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out;
@@ -4157,7 +4241,7 @@ static int ocfs2_insert_path(struct inode *inode,
 		 * of splits, but it's easier to just let one separate
 		 * function sort it all out.
 		 */
-		ocfs2_split_record(inode, left_path, right_path,
+		ocfs2_split_record(et, left_path, right_path,
 				   insert_rec, insert->ins_split);
 
 		/*
@@ -4171,8 +4255,8 @@ static int ocfs2_insert_path(struct inode *inode,
 			if (ret)
 				mlog_errno(ret);
 	} else
-		ocfs2_insert_at_leaf(insert_rec, path_leaf_el(right_path),
-				     insert, inode);
+		ocfs2_insert_at_leaf(et, insert_rec, path_leaf_el(right_path),
+				     insert);
 
 	ret = ocfs2_journal_dirty(handle, leaf_bh);
 	if (ret)
@@ -4185,10 +4269,10 @@ static int ocfs2_insert_path(struct inode *inode,
 		 *
 		 * XXX: Should we extend the transaction here?
 		 */
-		subtree_index = ocfs2_find_subtree_root(inode, left_path,
+		subtree_index = ocfs2_find_subtree_root(et, left_path,
 							right_path);
-		ocfs2_complete_edge_insert(inode, handle, left_path,
-					   right_path, subtree_index);
+		ocfs2_complete_edge_insert(handle, left_path, right_path,
+					   subtree_index);
 	}
 
 	ret = 0;
@@ -4196,8 +4280,7 @@ out:
 	return ret;
 }
 
-static int ocfs2_do_insert_extent(struct inode *inode,
-				  handle_t *handle,
+static int ocfs2_do_insert_extent(handle_t *handle,
 				  struct ocfs2_extent_tree *et,
 				  struct ocfs2_extent_rec *insert_rec,
 				  struct ocfs2_insert_type *type)
@@ -4210,7 +4293,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 
 	el = et->et_root_el;
 
-	ret = ocfs2_et_root_journal_access(handle, inode, et,
+	ret = ocfs2_et_root_journal_access(handle, et,
 					   OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -4218,7 +4301,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 	}
 
 	if (le16_to_cpu(el->l_tree_depth) == 0) {
-		ocfs2_insert_at_leaf(insert_rec, el, type, inode);
+		ocfs2_insert_at_leaf(et, insert_rec, el, type);
 		goto out_update_clusters;
 	}
 
@@ -4241,7 +4324,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 		cpos = UINT_MAX;
 	}
 
-	ret = ocfs2_find_path(inode, right_path, cpos);
+	ret = ocfs2_find_path(et->et_ci, right_path, cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -4260,7 +4343,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 	 * can wind up skipping both of these two special cases...
 	 */
 	if (rotate) {
-		ret = ocfs2_rotate_tree_right(inode, handle, type->ins_split,
+		ret = ocfs2_rotate_tree_right(handle, et, type->ins_split,
 					      le32_to_cpu(insert_rec->e_cpos),
 					      right_path, &left_path);
 		if (ret) {
@@ -4272,7 +4355,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 		 * ocfs2_rotate_tree_right() might have extended the
 		 * transaction without re-journaling our tree root.
 		 */
-		ret = ocfs2_et_root_journal_access(handle, inode, et,
+		ret = ocfs2_et_root_journal_access(handle, et,
 						   OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
 			mlog_errno(ret);
@@ -4280,7 +4363,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 		}
 	} else if (type->ins_appending == APPEND_TAIL
 		   && type->ins_contig != CONTIG_LEFT) {
-		ret = ocfs2_append_rec_to_path(inode, handle, insert_rec,
+		ret = ocfs2_append_rec_to_path(handle, et, insert_rec,
 					       right_path, &left_path);
 		if (ret) {
 			mlog_errno(ret);
@@ -4288,7 +4371,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 		}
 	}
 
-	ret = ocfs2_insert_path(inode, handle, left_path, right_path,
+	ret = ocfs2_insert_path(handle, et, left_path, right_path,
 				insert_rec, type);
 	if (ret) {
 		mlog_errno(ret);
@@ -4297,7 +4380,7 @@ static int ocfs2_do_insert_extent(struct inode *inode,
 
 out_update_clusters:
 	if (type->ins_split == SPLIT_NONE)
-		ocfs2_et_update_clusters(inode, et,
+		ocfs2_et_update_clusters(et,
 					 le16_to_cpu(insert_rec->e_leaf_clusters));
 
 	ret = ocfs2_journal_dirty(handle, et->et_root_bh);
@@ -4312,7 +4395,8 @@ out:
 }
 
 static enum ocfs2_contig_type
-ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
+ocfs2_figure_merge_contig_type(struct ocfs2_extent_tree *et,
+			       struct ocfs2_path *path,
 			       struct ocfs2_extent_list *el, int index,
 			       struct ocfs2_extent_rec *split_rec)
 {
@@ -4324,12 +4408,12 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 	struct ocfs2_path *left_path = NULL, *right_path = NULL;
 	struct buffer_head *bh;
 	struct ocfs2_extent_block *eb;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
 
 	if (index > 0) {
 		rec = &el->l_recs[index - 1];
 	} else if (path->p_tree_depth > 0) {
-		status = ocfs2_find_cpos_for_left_leaf(inode->i_sb,
-						       path, &left_cpos);
+		status = ocfs2_find_cpos_for_left_leaf(sb, path, &left_cpos);
 		if (status)
 			goto out;
 
@@ -4338,7 +4422,8 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 			if (!left_path)
 				goto out;
 
-			status = ocfs2_find_path(inode, left_path, left_cpos);
+			status = ocfs2_find_path(et->et_ci, left_path,
+						 left_cpos);
 			if (status)
 				goto out;
 
@@ -4348,7 +4433,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 			    le16_to_cpu(new_el->l_count)) {
 				bh = path_leaf_bh(left_path);
 				eb = (struct ocfs2_extent_block *)bh->b_data;
-				ocfs2_error(inode->i_sb,
+				ocfs2_error(sb,
 					    "Extent block #%llu has an "
 					    "invalid l_next_free_rec of "
 					    "%d.  It should have "
@@ -4373,7 +4458,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 			if (split_rec->e_cpos == el->l_recs[index].e_cpos)
 				ret = CONTIG_RIGHT;
 		} else {
-			ret = ocfs2_extent_contig(inode, rec, split_rec);
+			ret = ocfs2_et_extent_contig(et, rec, split_rec);
 		}
 	}
 
@@ -4382,8 +4467,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 		rec = &el->l_recs[index + 1];
 	else if (le16_to_cpu(el->l_next_free_rec) == le16_to_cpu(el->l_count) &&
 		 path->p_tree_depth > 0) {
-		status = ocfs2_find_cpos_for_right_leaf(inode->i_sb,
-							path, &right_cpos);
+		status = ocfs2_find_cpos_for_right_leaf(sb, path, &right_cpos);
 		if (status)
 			goto out;
 
@@ -4394,7 +4478,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 		if (!right_path)
 			goto out;
 
-		status = ocfs2_find_path(inode, right_path, right_cpos);
+		status = ocfs2_find_path(et->et_ci, right_path, right_cpos);
 		if (status)
 			goto out;
 
@@ -4404,7 +4488,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 			if (le16_to_cpu(new_el->l_next_free_rec) <= 1) {
 				bh = path_leaf_bh(right_path);
 				eb = (struct ocfs2_extent_block *)bh->b_data;
-				ocfs2_error(inode->i_sb,
+				ocfs2_error(sb,
 					    "Extent block #%llu has an "
 					    "invalid l_next_free_rec of %d",
 					    (unsigned long long)le64_to_cpu(eb->h_blkno),
@@ -4419,7 +4503,7 @@ ocfs2_figure_merge_contig_type(struct inode *inode, struct ocfs2_path *path,
 	if (rec) {
 		enum ocfs2_contig_type contig_type;
 
-		contig_type = ocfs2_extent_contig(inode, rec, split_rec);
+		contig_type = ocfs2_et_extent_contig(et, rec, split_rec);
 
 		if (contig_type == CONTIG_LEFT && ret == CONTIG_RIGHT)
 			ret = CONTIG_LEFTRIGHT;
@@ -4436,11 +4520,10 @@ out:
 	return ret;
 }
 
-static void ocfs2_figure_contig_type(struct inode *inode,
+static void ocfs2_figure_contig_type(struct ocfs2_extent_tree *et,
 				     struct ocfs2_insert_type *insert,
 				     struct ocfs2_extent_list *el,
-				     struct ocfs2_extent_rec *insert_rec,
-				     struct ocfs2_extent_tree *et)
+				     struct ocfs2_extent_rec *insert_rec)
 {
 	int i;
 	enum ocfs2_contig_type contig_type = CONTIG_NONE;
@@ -4448,8 +4531,8 @@ static void ocfs2_figure_contig_type(struct inode *inode,
 	BUG_ON(le16_to_cpu(el->l_tree_depth) != 0);
 
 	for(i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
-		contig_type = ocfs2_extent_contig(inode, &el->l_recs[i],
-						  insert_rec);
+		contig_type = ocfs2_et_extent_contig(et, &el->l_recs[i],
+						     insert_rec);
 		if (contig_type != CONTIG_NONE) {
 			insert->ins_contig_index = i;
 			break;
@@ -4530,8 +4613,7 @@ set_tail_append:
  * All of the information is stored on the ocfs2_insert_type
  * structure.
  */
-static int ocfs2_figure_insert_type(struct inode *inode,
-				    struct ocfs2_extent_tree *et,
+static int ocfs2_figure_insert_type(struct ocfs2_extent_tree *et,
 				    struct buffer_head **last_eb_bh,
 				    struct ocfs2_extent_rec *insert_rec,
 				    int *free_records,
@@ -4555,7 +4637,7 @@ static int ocfs2_figure_insert_type(struct inode *inode,
 		 * ocfs2_figure_insert_type() and ocfs2_add_branch()
 		 * may want it later.
 		 */
-		ret = ocfs2_read_extent_block(inode,
+		ret = ocfs2_read_extent_block(et->et_ci,
 					      ocfs2_et_get_last_eb_blk(et),
 					      &bh);
 		if (ret) {
@@ -4578,7 +4660,7 @@ static int ocfs2_figure_insert_type(struct inode *inode,
 		le16_to_cpu(el->l_next_free_rec);
 
 	if (!insert->ins_tree_depth) {
-		ocfs2_figure_contig_type(inode, insert, el, insert_rec, et);
+		ocfs2_figure_contig_type(et, insert, el, insert_rec);
 		ocfs2_figure_appending_type(insert, el, insert_rec);
 		return 0;
 	}
@@ -4596,7 +4678,7 @@ static int ocfs2_figure_insert_type(struct inode *inode,
 	 * us the rightmost tree path. This is accounted for below in
 	 * the appending code.
 	 */
-	ret = ocfs2_find_path(inode, path, le32_to_cpu(insert_rec->e_cpos));
+	ret = ocfs2_find_path(et->et_ci, path, le32_to_cpu(insert_rec->e_cpos));
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -4612,7 +4694,7 @@ static int ocfs2_figure_insert_type(struct inode *inode,
          *     into two types of appends: simple record append, or a
          *     rotate inside the tail leaf.
 	 */
-	ocfs2_figure_contig_type(inode, insert, el, insert_rec, et);
+	ocfs2_figure_contig_type(et, insert, el, insert_rec);
 
 	/*
 	 * The insert code isn't quite ready to deal with all cases of
@@ -4657,13 +4739,11 @@ out:
 }
 
 /*
- * Insert an extent into an inode btree.
+ * Insert an extent into a btree.
  *
- * The caller needs to update fe->i_clusters
+ * The caller needs to update the owning btree's cluster count.
  */
-int ocfs2_insert_extent(struct ocfs2_super *osb,
-			handle_t *handle,
-			struct inode *inode,
+int ocfs2_insert_extent(handle_t *handle,
 			struct ocfs2_extent_tree *et,
 			u32 cpos,
 			u64 start_blk,
@@ -4677,21 +4757,22 @@ int ocfs2_insert_extent(struct ocfs2_super *osb,
 	struct ocfs2_insert_type insert = {0, };
 	struct ocfs2_extent_rec rec;
 
-	mlog(0, "add %u clusters at position %u to inode %llu\n",
-	     new_clusters, cpos, (unsigned long long)OCFS2_I(inode)->ip_blkno);
+	mlog(0, "add %u clusters at position %u to owner %llu\n",
+	     new_clusters, cpos,
+	     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci));
 
 	memset(&rec, 0, sizeof(rec));
 	rec.e_cpos = cpu_to_le32(cpos);
 	rec.e_blkno = cpu_to_le64(start_blk);
 	rec.e_leaf_clusters = cpu_to_le16(new_clusters);
 	rec.e_flags = flags;
-	status = ocfs2_et_insert_check(inode, et, &rec);
+	status = ocfs2_et_insert_check(et, &rec);
 	if (status) {
 		mlog_errno(status);
 		goto bail;
 	}
 
-	status = ocfs2_figure_insert_type(inode, et, &last_eb_bh, &rec,
+	status = ocfs2_figure_insert_type(et, &last_eb_bh, &rec,
 					  &free_records, &insert);
 	if (status < 0) {
 		mlog_errno(status);
@@ -4705,7 +4786,7 @@ int ocfs2_insert_extent(struct ocfs2_super *osb,
 	     free_records, insert.ins_tree_depth);
 
 	if (insert.ins_contig == CONTIG_NONE && free_records == 0) {
-		status = ocfs2_grow_tree(inode, handle, et,
+		status = ocfs2_grow_tree(handle, et,
 					 &insert.ins_tree_depth, &last_eb_bh,
 					 meta_ac);
 		if (status) {
@@ -4715,11 +4796,11 @@ int ocfs2_insert_extent(struct ocfs2_super *osb,
 	}
 
 	/* Finally, we can add clusters. This might rotate the tree for us. */
-	status = ocfs2_do_insert_extent(inode, handle, et, &rec, &insert);
+	status = ocfs2_do_insert_extent(handle, et, &rec, &insert);
 	if (status < 0)
 		mlog_errno(status);
-	else if (et->et_ops == &ocfs2_dinode_et_ops)
-		ocfs2_extent_map_insert_rec(inode, &rec);
+	else
+		ocfs2_et_extent_map_insert(et, &rec);
 
 bail:
 	brelse(last_eb_bh);
@@ -4735,13 +4816,11 @@ bail:
  * it is not limited to the file storage. Any extent tree can use this
  * function if it implements the proper ocfs2_extent_tree.
  */
-int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb,
-				struct inode *inode,
+int ocfs2_add_clusters_in_btree(handle_t *handle,
+				struct ocfs2_extent_tree *et,
 				u32 *logical_offset,
 				u32 clusters_to_add,
 				int mark_unwritten,
-				struct ocfs2_extent_tree *et,
-				handle_t *handle,
 				struct ocfs2_alloc_context *data_ac,
 				struct ocfs2_alloc_context *meta_ac,
 				enum ocfs2_alloc_restarted *reason_ret)
@@ -4752,13 +4831,15 @@ int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb,
 	u32 bit_off, num_bits;
 	u64 block;
 	u8 flags = 0;
+	struct ocfs2_super *osb =
+		OCFS2_SB(ocfs2_metadata_cache_get_super(et->et_ci));
 
 	BUG_ON(!clusters_to_add);
 
 	if (mark_unwritten)
 		flags = OCFS2_EXT_UNWRITTEN;
 
-	free_extents = ocfs2_num_free_extents(osb, inode, et);
+	free_extents = ocfs2_num_free_extents(osb, et);
 	if (free_extents < 0) {
 		status = free_extents;
 		mlog_errno(status);
@@ -4795,7 +4876,7 @@ int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb,
 	BUG_ON(num_bits > clusters_to_add);
 
 	/* reserve our write early -- insert_extent may update the tree root */
-	status = ocfs2_et_root_journal_access(handle, inode, et,
+	status = ocfs2_et_root_journal_access(handle, et,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -4803,10 +4884,10 @@ int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb,
 	}
 
 	block = ocfs2_clusters_to_blocks(osb->sb, bit_off);
-	mlog(0, "Allocating %u clusters at block %u for inode %llu\n",
-	     num_bits, bit_off, (unsigned long long)OCFS2_I(inode)->ip_blkno);
-	status = ocfs2_insert_extent(osb, handle, inode, et,
-				     *logical_offset, block,
+	mlog(0, "Allocating %u clusters at block %u for owner %llu\n",
+	     num_bits, bit_off,
+	     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci));
+	status = ocfs2_insert_extent(handle, et, *logical_offset, block,
 				     num_bits, flags, meta_ac);
 	if (status < 0) {
 		mlog_errno(status);
@@ -4856,10 +4937,9 @@ static void ocfs2_make_right_split_rec(struct super_block *sb,
 	split_rec->e_flags = rec->e_flags;
 }
 
-static int ocfs2_split_and_insert(struct inode *inode,
-				  handle_t *handle,
-				  struct ocfs2_path *path,
+static int ocfs2_split_and_insert(handle_t *handle,
 				  struct ocfs2_extent_tree *et,
+				  struct ocfs2_path *path,
 				  struct buffer_head **last_eb_bh,
 				  int split_index,
 				  struct ocfs2_extent_rec *orig_split_rec,
@@ -4892,7 +4972,7 @@ leftright:
 
 	if (le16_to_cpu(rightmost_el->l_next_free_rec) ==
 	    le16_to_cpu(rightmost_el->l_count)) {
-		ret = ocfs2_grow_tree(inode, handle, et,
+		ret = ocfs2_grow_tree(handle, et,
 				      &depth, last_eb_bh, meta_ac);
 		if (ret) {
 			mlog_errno(ret);
@@ -4921,8 +5001,8 @@ leftright:
 		 */
 		insert.ins_split = SPLIT_RIGHT;
 
-		ocfs2_make_right_split_rec(inode->i_sb, &tmprec, insert_range,
-					   &rec);
+		ocfs2_make_right_split_rec(ocfs2_metadata_cache_get_super(et->et_ci),
+					   &tmprec, insert_range, &rec);
 
 		split_rec = tmprec;
 
@@ -4930,7 +5010,7 @@ leftright:
 		do_leftright = 1;
 	}
 
-	ret = ocfs2_do_insert_extent(inode, handle, et, &split_rec, &insert);
+	ret = ocfs2_do_insert_extent(handle, et, &split_rec, &insert);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -4946,7 +5026,7 @@ leftright:
 		ocfs2_reinit_path(path, 1);
 
 		cpos = le32_to_cpu(split_rec.e_cpos);
-		ret = ocfs2_find_path(inode, path, cpos);
+		ret = ocfs2_find_path(et->et_ci, path, cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -4961,8 +5041,8 @@ out:
 	return ret;
 }
 
-static int ocfs2_replace_extent_rec(struct inode *inode,
-				    handle_t *handle,
+static int ocfs2_replace_extent_rec(handle_t *handle,
+				    struct ocfs2_extent_tree *et,
 				    struct ocfs2_path *path,
 				    struct ocfs2_extent_list *el,
 				    int split_index,
@@ -4970,7 +5050,7 @@ static int ocfs2_replace_extent_rec(struct inode *inode,
 {
 	int ret;
 
-	ret = ocfs2_path_bh_journal_access(handle, inode, path,
+	ret = ocfs2_path_bh_journal_access(handle, et->et_ci, path,
 					   path_num_items(path) - 1);
 	if (ret) {
 		mlog_errno(ret);
@@ -4985,9 +5065,8 @@ out:
 }
 
 /*
- * Mark part or all of the extent record at split_index in the leaf
- * pointed to by path as written. This removes the unwritten
- * extent flag.
+ * Split part or all of the extent record at split_index in the leaf
+ * pointed to by path. Merge with the contiguous extent record if needed.
  *
  * Care is taken to handle contiguousness so as to not grow the tree.
  *
@@ -5004,14 +5083,13 @@ out:
  * have been brought into cache (and pinned via the journal), so the
  * extra overhead is not expressed in terms of disk reads.
  */
-static int __ocfs2_mark_extent_written(struct inode *inode,
-				       struct ocfs2_extent_tree *et,
-				       handle_t *handle,
-				       struct ocfs2_path *path,
-				       int split_index,
-				       struct ocfs2_extent_rec *split_rec,
-				       struct ocfs2_alloc_context *meta_ac,
-				       struct ocfs2_cached_dealloc_ctxt *dealloc)
+int ocfs2_split_extent(handle_t *handle,
+		       struct ocfs2_extent_tree *et,
+		       struct ocfs2_path *path,
+		       int split_index,
+		       struct ocfs2_extent_rec *split_rec,
+		       struct ocfs2_alloc_context *meta_ac,
+		       struct ocfs2_cached_dealloc_ctxt *dealloc)
 {
 	int ret = 0;
 	struct ocfs2_extent_list *el = path_leaf_el(path);
@@ -5020,12 +5098,6 @@ static int __ocfs2_mark_extent_written(struct inode *inode,
 	struct ocfs2_merge_ctxt ctxt;
 	struct ocfs2_extent_list *rightmost_el;
 
-	if (!(rec->e_flags & OCFS2_EXT_UNWRITTEN)) {
-		ret = -EIO;
-		mlog_errno(ret);
-		goto out;
-	}
-
 	if (le32_to_cpu(rec->e_cpos) > le32_to_cpu(split_rec->e_cpos) ||
 	    ((le32_to_cpu(rec->e_cpos) + le16_to_cpu(rec->e_leaf_clusters)) <
 	     (le32_to_cpu(split_rec->e_cpos) + le16_to_cpu(split_rec->e_leaf_clusters)))) {
@@ -5034,19 +5106,19 @@ static int __ocfs2_mark_extent_written(struct inode *inode,
 		goto out;
 	}
 
-	ctxt.c_contig_type = ocfs2_figure_merge_contig_type(inode, path, el,
+	ctxt.c_contig_type = ocfs2_figure_merge_contig_type(et, path, el,
 							    split_index,
 							    split_rec);
 
 	/*
 	 * The core merge / split code wants to know how much room is
-	 * left in this inodes allocation tree, so we pass the
+	 * left in this allocation tree, so we pass the
 	 * rightmost extent list.
 	 */
 	if (path->p_tree_depth) {
 		struct ocfs2_extent_block *eb;
 
-		ret = ocfs2_read_extent_block(inode,
+		ret = ocfs2_read_extent_block(et->et_ci,
 					      ocfs2_et_get_last_eb_blk(et),
 					      &last_eb_bh);
 		if (ret) {
@@ -5073,19 +5145,18 @@ static int __ocfs2_mark_extent_written(struct inode *inode,
 
 	if (ctxt.c_contig_type == CONTIG_NONE) {
 		if (ctxt.c_split_covers_rec)
-			ret = ocfs2_replace_extent_rec(inode, handle,
-						       path, el,
+			ret = ocfs2_replace_extent_rec(handle, et, path, el,
 						       split_index, split_rec);
 		else
-			ret = ocfs2_split_and_insert(inode, handle, path, et,
+			ret = ocfs2_split_and_insert(handle, et, path,
 						     &last_eb_bh, split_index,
 						     split_rec, meta_ac);
 		if (ret)
 			mlog_errno(ret);
 	} else {
-		ret = ocfs2_try_to_merge_extent(inode, handle, path,
+		ret = ocfs2_try_to_merge_extent(handle, et, path,
 						split_index, split_rec,
-						dealloc, &ctxt, et);
+						dealloc, &ctxt);
 		if (ret)
 			mlog_errno(ret);
 	}
@@ -5096,46 +5167,31 @@ out:
 }
 
 /*
- * Mark the already-existing extent at cpos as written for len clusters.
+ * Change the flags of the already-existing extent at cpos for len clusters.
+ *
+ * new_flags: the flags we want to set.
+ * clear_flags: the flags we want to clear.
+ * phys: the new physical offset we want this new extent starts from.
  *
  * If the existing extent is larger than the request, initiate a
  * split. An attempt will be made at merging with adjacent extents.
  *
  * The caller is responsible for passing down meta_ac if we'll need it.
  */
-int ocfs2_mark_extent_written(struct inode *inode,
-			      struct ocfs2_extent_tree *et,
-			      handle_t *handle, u32 cpos, u32 len, u32 phys,
-			      struct ocfs2_alloc_context *meta_ac,
-			      struct ocfs2_cached_dealloc_ctxt *dealloc)
+int ocfs2_change_extent_flag(handle_t *handle,
+			     struct ocfs2_extent_tree *et,
+			     u32 cpos, u32 len, u32 phys,
+			     struct ocfs2_alloc_context *meta_ac,
+			     struct ocfs2_cached_dealloc_ctxt *dealloc,
+			     int new_flags, int clear_flags)
 {
 	int ret, index;
-	u64 start_blkno = ocfs2_clusters_to_blocks(inode->i_sb, phys);
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
+	u64 start_blkno = ocfs2_clusters_to_blocks(sb, phys);
 	struct ocfs2_extent_rec split_rec;
 	struct ocfs2_path *left_path = NULL;
 	struct ocfs2_extent_list *el;
-
-	mlog(0, "Inode %lu cpos %u, len %u, phys %u (%llu)\n",
-	     inode->i_ino, cpos, len, phys, (unsigned long long)start_blkno);
-
-	if (!ocfs2_writes_unwritten_extents(OCFS2_SB(inode->i_sb))) {
-		ocfs2_error(inode->i_sb, "Inode %llu has unwritten extents "
-			    "that are being written to, but the feature bit "
-			    "is not set in the super block.",
-			    (unsigned long long)OCFS2_I(inode)->ip_blkno);
-		ret = -EROFS;
-		goto out;
-	}
-
-	/*
-	 * XXX: This should be fixed up so that we just re-insert the
-	 * next extent records.
-	 *
-	 * XXX: This is a hack on the extent tree, maybe it should be
-	 * an op?
-	 */
-	if (et->et_ops == &ocfs2_dinode_et_ops)
-		ocfs2_extent_map_trunc(inode, 0);
+	struct ocfs2_extent_rec *rec;
 
 	left_path = ocfs2_new_path_from_et(et);
 	if (!left_path) {
@@ -5144,7 +5200,7 @@ int ocfs2_mark_extent_written(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_path(inode, left_path, cpos);
+	ret = ocfs2_find_path(et->et_ci, left_path, cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -5153,34 +5209,102 @@ int ocfs2_mark_extent_written(struct inode *inode,
 
 	index = ocfs2_search_extent_list(el, cpos);
 	if (index == -1 || index >= le16_to_cpu(el->l_next_free_rec)) {
-		ocfs2_error(inode->i_sb,
-			    "Inode %llu has an extent at cpos %u which can no "
+		ocfs2_error(sb,
+			    "Owner %llu has an extent at cpos %u which can no "
 			    "longer be found.\n",
-			    (unsigned long long)OCFS2_I(inode)->ip_blkno, cpos);
+			     (unsigned long long)
+			     ocfs2_metadata_cache_owner(et->et_ci), cpos);
 		ret = -EROFS;
 		goto out;
 	}
 
+	ret = -EIO;
+	rec = &el->l_recs[index];
+	if (new_flags && (rec->e_flags & new_flags)) {
+		mlog(ML_ERROR, "Owner %llu tried to set %d flags on an "
+		     "extent that already had them",
+		     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+		     new_flags);
+		goto out;
+	}
+
+	if (clear_flags && !(rec->e_flags & clear_flags)) {
+		mlog(ML_ERROR, "Owner %llu tried to clear %d flags on an "
+		     "extent that didn't have them",
+		     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+		     clear_flags);
+		goto out;
+	}
+
 	memset(&split_rec, 0, sizeof(struct ocfs2_extent_rec));
 	split_rec.e_cpos = cpu_to_le32(cpos);
 	split_rec.e_leaf_clusters = cpu_to_le16(len);
 	split_rec.e_blkno = cpu_to_le64(start_blkno);
-	split_rec.e_flags = path_leaf_el(left_path)->l_recs[index].e_flags;
-	split_rec.e_flags &= ~OCFS2_EXT_UNWRITTEN;
-
-	ret = __ocfs2_mark_extent_written(inode, et, handle, left_path,
-					  index, &split_rec, meta_ac,
-					  dealloc);
+	split_rec.e_flags = rec->e_flags;
+	if (new_flags)
+		split_rec.e_flags |= new_flags;
+	if (clear_flags)
+		split_rec.e_flags &= ~clear_flags;
+
+	ret = ocfs2_split_extent(handle, et, left_path,
+				 index, &split_rec, meta_ac,
+				 dealloc);
 	if (ret)
 		mlog_errno(ret);
 
 out:
 	ocfs2_free_path(left_path);
 	return ret;
+
 }
 
-static int ocfs2_split_tree(struct inode *inode, struct ocfs2_extent_tree *et,
-			    handle_t *handle, struct ocfs2_path *path,
+/*
+ * Mark the already-existing extent at cpos as written for len clusters.
+ * This removes the unwritten extent flag.
+ *
+ * If the existing extent is larger than the request, initiate a
+ * split. An attempt will be made at merging with adjacent extents.
+ *
+ * The caller is responsible for passing down meta_ac if we'll need it.
+ */
+int ocfs2_mark_extent_written(struct inode *inode,
+			      struct ocfs2_extent_tree *et,
+			      handle_t *handle, u32 cpos, u32 len, u32 phys,
+			      struct ocfs2_alloc_context *meta_ac,
+			      struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret;
+
+	mlog(0, "Inode %lu cpos %u, len %u, phys clusters %u\n",
+	     inode->i_ino, cpos, len, phys);
+
+	if (!ocfs2_writes_unwritten_extents(OCFS2_SB(inode->i_sb))) {
+		ocfs2_error(inode->i_sb, "Inode %llu has unwritten extents "
+			    "that are being written to, but the feature bit "
+			    "is not set in the super block.",
+			    (unsigned long long)OCFS2_I(inode)->ip_blkno);
+		ret = -EROFS;
+		goto out;
+	}
+
+	/*
+	 * XXX: This should be fixed up so that we just re-insert the
+	 * next extent records.
+	 */
+	ocfs2_et_extent_map_truncate(et, 0);
+
+	ret = ocfs2_change_extent_flag(handle, et, cpos,
+				       len, phys, meta_ac, dealloc,
+				       0, OCFS2_EXT_UNWRITTEN);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	return ret;
+}
+
+static int ocfs2_split_tree(handle_t *handle, struct ocfs2_extent_tree *et,
+			    struct ocfs2_path *path,
 			    int index, u32 new_range,
 			    struct ocfs2_alloc_context *meta_ac)
 {
@@ -5197,11 +5321,12 @@ static int ocfs2_split_tree(struct inode *inode, struct ocfs2_extent_tree *et,
 	 */
 	el = path_leaf_el(path);
 	rec = &el->l_recs[index];
-	ocfs2_make_right_split_rec(inode->i_sb, &split_rec, new_range, rec);
+	ocfs2_make_right_split_rec(ocfs2_metadata_cache_get_super(et->et_ci),
+				   &split_rec, new_range, rec);
 
 	depth = path->p_tree_depth;
 	if (depth > 0) {
-		ret = ocfs2_read_extent_block(inode,
+		ret = ocfs2_read_extent_block(et->et_ci,
 					      ocfs2_et_get_last_eb_blk(et),
 					      &last_eb_bh);
 		if (ret < 0) {
@@ -5224,7 +5349,7 @@ static int ocfs2_split_tree(struct inode *inode, struct ocfs2_extent_tree *et,
 
 	if (le16_to_cpu(rightmost_el->l_next_free_rec) ==
 	    le16_to_cpu(rightmost_el->l_count)) {
-		ret = ocfs2_grow_tree(inode, handle, et, &depth, &last_eb_bh,
+		ret = ocfs2_grow_tree(handle, et, &depth, &last_eb_bh,
 				      meta_ac);
 		if (ret) {
 			mlog_errno(ret);
@@ -5238,7 +5363,7 @@ static int ocfs2_split_tree(struct inode *inode, struct ocfs2_extent_tree *et,
 	insert.ins_split = SPLIT_RIGHT;
 	insert.ins_tree_depth = depth;
 
-	ret = ocfs2_do_insert_extent(inode, handle, et, &split_rec, &insert);
+	ret = ocfs2_do_insert_extent(handle, et, &split_rec, &insert);
 	if (ret)
 		mlog_errno(ret);
 
@@ -5247,23 +5372,23 @@ out:
 	return ret;
 }
 
-static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
+static int ocfs2_truncate_rec(handle_t *handle,
+			      struct ocfs2_extent_tree *et,
 			      struct ocfs2_path *path, int index,
 			      struct ocfs2_cached_dealloc_ctxt *dealloc,
-			      u32 cpos, u32 len,
-			      struct ocfs2_extent_tree *et)
+			      u32 cpos, u32 len)
 {
 	int ret;
 	u32 left_cpos, rec_range, trunc_range;
 	int wants_rotate = 0, is_rightmost_tree_rec = 0;
-	struct super_block *sb = inode->i_sb;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
 	struct ocfs2_path *left_path = NULL;
 	struct ocfs2_extent_list *el = path_leaf_el(path);
 	struct ocfs2_extent_rec *rec;
 	struct ocfs2_extent_block *eb;
 
 	if (ocfs2_is_empty_extent(&el->l_recs[0]) && index > 0) {
-		ret = ocfs2_rotate_tree_left(inode, handle, path, dealloc, et);
+		ret = ocfs2_rotate_tree_left(handle, et, path, dealloc);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -5295,14 +5420,13 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 		 * by this leaf and the one to it's left.
 		 *
 		 * There are two cases we can skip:
-		 *   1) Path is the leftmost one in our inode tree.
+		 *   1) Path is the leftmost one in our btree.
 		 *   2) The leaf is rightmost and will be empty after
 		 *      we remove the extent record - the rotate code
 		 *      knows how to update the newly formed edge.
 		 */
 
-		ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, path,
-						    &left_cpos);
+		ret = ocfs2_find_cpos_for_left_leaf(sb, path, &left_cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -5316,7 +5440,8 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 				goto out;
 			}
 
-			ret = ocfs2_find_path(inode, left_path, left_cpos);
+			ret = ocfs2_find_path(et->et_ci, left_path,
+					      left_cpos);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
@@ -5332,13 +5457,13 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_path(inode, handle, path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, path);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_path(inode, handle, left_path);
+	ret = ocfs2_journal_access_path(et->et_ci, handle, left_path);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -5361,7 +5486,7 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 			 * be deleted by the rotate code.
 			 */
 			rec = &el->l_recs[next_free - 1];
-			ocfs2_adjust_rightmost_records(inode, handle, path,
+			ocfs2_adjust_rightmost_records(handle, et, path,
 						       rec);
 		}
 	} else if (le32_to_cpu(rec->e_cpos) == cpos) {
@@ -5373,11 +5498,12 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 		/* Remove rightmost portion of the record */
 		le16_add_cpu(&rec->e_leaf_clusters, -len);
 		if (is_rightmost_tree_rec)
-			ocfs2_adjust_rightmost_records(inode, handle, path, rec);
+			ocfs2_adjust_rightmost_records(handle, et, path, rec);
 	} else {
 		/* Caller should have trapped this. */
-		mlog(ML_ERROR, "Inode %llu: Invalid record truncate: (%u, %u) "
-		     "(%u, %u)\n", (unsigned long long)OCFS2_I(inode)->ip_blkno,
+		mlog(ML_ERROR, "Owner %llu: Invalid record truncate: (%u, %u) "
+		     "(%u, %u)\n",
+		     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 		     le32_to_cpu(rec->e_cpos),
 		     le16_to_cpu(rec->e_leaf_clusters), cpos, len);
 		BUG();
@@ -5386,14 +5512,14 @@ static int ocfs2_truncate_rec(struct inode *inode, handle_t *handle,
 	if (left_path) {
 		int subtree_index;
 
-		subtree_index = ocfs2_find_subtree_root(inode, left_path, path);
-		ocfs2_complete_edge_insert(inode, handle, left_path, path,
+		subtree_index = ocfs2_find_subtree_root(et, left_path, path);
+		ocfs2_complete_edge_insert(handle, left_path, path,
 					   subtree_index);
 	}
 
 	ocfs2_journal_dirty(handle, path_leaf_bh(path));
 
-	ret = ocfs2_rotate_tree_left(inode, handle, path, dealloc, et);
+	ret = ocfs2_rotate_tree_left(handle, et, path, dealloc);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -5404,9 +5530,9 @@ out:
 	return ret;
 }
 
-int ocfs2_remove_extent(struct inode *inode,
+int ocfs2_remove_extent(handle_t *handle,
 			struct ocfs2_extent_tree *et,
-			u32 cpos, u32 len, handle_t *handle,
+			u32 cpos, u32 len,
 			struct ocfs2_alloc_context *meta_ac,
 			struct ocfs2_cached_dealloc_ctxt *dealloc)
 {
@@ -5416,7 +5542,11 @@ int ocfs2_remove_extent(struct inode *inode,
 	struct ocfs2_extent_list *el;
 	struct ocfs2_path *path = NULL;
 
-	ocfs2_extent_map_trunc(inode, 0);
+	/*
+	 * XXX: Why are we truncating to 0 instead of wherever this
+	 * affects us?
+	 */
+	ocfs2_et_extent_map_truncate(et, 0);
 
 	path = ocfs2_new_path_from_et(et);
 	if (!path) {
@@ -5425,7 +5555,7 @@ int ocfs2_remove_extent(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_path(inode, path, cpos);
+	ret = ocfs2_find_path(et->et_ci, path, cpos);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -5434,10 +5564,11 @@ int ocfs2_remove_extent(struct inode *inode,
 	el = path_leaf_el(path);
 	index = ocfs2_search_extent_list(el, cpos);
 	if (index == -1 || index >= le16_to_cpu(el->l_next_free_rec)) {
-		ocfs2_error(inode->i_sb,
-			    "Inode %llu has an extent at cpos %u which can no "
+		ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+			    "Owner %llu has an extent at cpos %u which can no "
 			    "longer be found.\n",
-			    (unsigned long long)OCFS2_I(inode)->ip_blkno, cpos);
+			    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+			    cpos);
 		ret = -EROFS;
 		goto out;
 	}
@@ -5464,20 +5595,21 @@ int ocfs2_remove_extent(struct inode *inode,
 
 	BUG_ON(cpos < le32_to_cpu(rec->e_cpos) || trunc_range > rec_range);
 
-	mlog(0, "Inode %llu, remove (cpos %u, len %u). Existing index %d "
+	mlog(0, "Owner %llu, remove (cpos %u, len %u). Existing index %d "
 	     "(cpos %u, len %u)\n",
-	     (unsigned long long)OCFS2_I(inode)->ip_blkno, cpos, len, index,
+	     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
+	     cpos, len, index,
 	     le32_to_cpu(rec->e_cpos), ocfs2_rec_clusters(el, rec));
 
 	if (le32_to_cpu(rec->e_cpos) == cpos || rec_range == trunc_range) {
-		ret = ocfs2_truncate_rec(inode, handle, path, index, dealloc,
-					 cpos, len, et);
+		ret = ocfs2_truncate_rec(handle, et, path, index, dealloc,
+					 cpos, len);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 	} else {
-		ret = ocfs2_split_tree(inode, et, handle, path, index,
+		ret = ocfs2_split_tree(handle, et, path, index,
 				       trunc_range, meta_ac);
 		if (ret) {
 			mlog_errno(ret);
@@ -5490,7 +5622,7 @@ int ocfs2_remove_extent(struct inode *inode,
 		 */
 		ocfs2_reinit_path(path, 1);
 
-		ret = ocfs2_find_path(inode, path, cpos);
+		ret = ocfs2_find_path(et->et_ci, path, cpos);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -5499,9 +5631,9 @@ int ocfs2_remove_extent(struct inode *inode,
 		el = path_leaf_el(path);
 		index = ocfs2_search_extent_list(el, cpos);
 		if (index == -1 || index >= le16_to_cpu(el->l_next_free_rec)) {
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu: split at cpos %u lost record.",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno,
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu: split at cpos %u lost record.",
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 				    cpos);
 			ret = -EROFS;
 			goto out;
@@ -5515,18 +5647,18 @@ int ocfs2_remove_extent(struct inode *inode,
 		rec_range = le32_to_cpu(rec->e_cpos) +
 			ocfs2_rec_clusters(el, rec);
 		if (rec_range != trunc_range) {
-			ocfs2_error(inode->i_sb,
-				    "Inode %llu: error after split at cpos %u"
+			ocfs2_error(ocfs2_metadata_cache_get_super(et->et_ci),
+				    "Owner %llu: error after split at cpos %u"
 				    "trunc len %u, existing record is (%u,%u)",
-				    (unsigned long long)OCFS2_I(inode)->ip_blkno,
+				    (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 				    cpos, len, le32_to_cpu(rec->e_cpos),
 				    ocfs2_rec_clusters(el, rec));
 			ret = -EROFS;
 			goto out;
 		}
 
-		ret = ocfs2_truncate_rec(inode, handle, path, index, dealloc,
-					 cpos, len, et);
+		ret = ocfs2_truncate_rec(handle, et, path, index, dealloc,
+					 cpos, len);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -5573,7 +5705,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_et_root_journal_access(handle, inode, et,
+	ret = ocfs2_et_root_journal_access(handle, et,
 					   OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -5583,14 +5715,13 @@ int ocfs2_remove_btree_range(struct inode *inode,
 	vfs_dq_free_space_nodirty(inode,
 				  ocfs2_clusters_to_bytes(inode->i_sb, len));
 
-	ret = ocfs2_remove_extent(inode, et, cpos, len, handle, meta_ac,
-				  dealloc);
+	ret = ocfs2_remove_extent(handle, et, cpos, len, meta_ac, dealloc);
 	if (ret) {
 		mlog_errno(ret);
 		goto out_commit;
 	}
 
-	ocfs2_et_update_clusters(inode, et, -len);
+	ocfs2_et_update_clusters(et, -len);
 
 	ret = ocfs2_journal_dirty(handle, et->et_root_bh);
 	if (ret) {
@@ -5690,7 +5821,7 @@ int ocfs2_truncate_log_append(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	status = ocfs2_journal_access_di(handle, tl_inode, tl_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(tl_inode), tl_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -5752,7 +5883,7 @@ static int ocfs2_replay_truncate_records(struct ocfs2_super *osb,
 	while (i >= 0) {
 		/* Caller has given us at least enough credits to
 		 * update the truncate log dinode */
-		status = ocfs2_journal_access_di(handle, tl_inode, tl_bh,
+		status = ocfs2_journal_access_di(handle, INODE_CACHE(tl_inode), tl_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -6010,7 +6141,7 @@ int ocfs2_begin_truncate_log_recovery(struct ocfs2_super *osb,
 		tl->tl_used = 0;
 
 		ocfs2_compute_meta_ecc(osb->sb, tl_bh->b_data, &di->i_check);
-		status = ocfs2_write_block(osb, tl_bh, tl_inode);
+		status = ocfs2_write_block(osb, tl_bh, INODE_CACHE(tl_inode));
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail;
@@ -6400,9 +6531,9 @@ ocfs2_find_per_slot_free_list(int type,
 	return fl;
 }
 
-static int ocfs2_cache_block_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
-				     int type, int slot, u64 blkno,
-				     unsigned int bit)
+int ocfs2_cache_block_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
+			      int type, int slot, u64 blkno,
+			      unsigned int bit)
 {
 	int ret;
 	struct ocfs2_per_slot_free_list *fl;
@@ -6518,7 +6649,7 @@ static int ocfs2_find_new_last_ext_blk(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_find_leaf(inode, path_root_el(path), cpos, &bh);
+	ret = ocfs2_find_leaf(INODE_CACHE(inode), path_root_el(path), cpos, &bh);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -6551,7 +6682,7 @@ out:
  */
 static int ocfs2_trim_tree(struct inode *inode, struct ocfs2_path *path,
 			   handle_t *handle, struct ocfs2_truncate_context *tc,
-			   u32 clusters_to_del, u64 *delete_start)
+			   u32 clusters_to_del, u64 *delete_start, u8 *flags)
 {
 	int ret, i, index = path->p_tree_depth;
 	u32 new_edge = 0;
@@ -6561,6 +6692,7 @@ static int ocfs2_trim_tree(struct inode *inode, struct ocfs2_path *path,
 	struct ocfs2_extent_rec *rec;
 
 	*delete_start = 0;
+	*flags = 0;
 
 	while (index >= 0) {
 		bh = path->p_node[index].bh;
@@ -6648,6 +6780,7 @@ find_tail_record:
 			*delete_start = le64_to_cpu(rec->e_blkno)
 				+ ocfs2_clusters_to_blocks(inode->i_sb,
 					le16_to_cpu(rec->e_leaf_clusters));
+			*flags = rec->e_flags;
 
 			/*
 			 * If it's now empty, remove this record.
@@ -6719,7 +6852,7 @@ delete:
 
 			mlog(0, "deleting this extent block.\n");
 
-			ocfs2_remove_from_cache(inode, bh);
+			ocfs2_remove_from_cache(INODE_CACHE(inode), bh);
 
 			BUG_ON(ocfs2_rec_clusters(el, &el->l_recs[0]));
 			BUG_ON(le32_to_cpu(el->l_recs[0].e_cpos));
@@ -6747,7 +6880,8 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 			     struct buffer_head *fe_bh,
 			     handle_t *handle,
 			     struct ocfs2_truncate_context *tc,
-			     struct ocfs2_path *path)
+			     struct ocfs2_path *path,
+			     struct ocfs2_alloc_context *meta_ac)
 {
 	int status;
 	struct ocfs2_dinode *fe;
@@ -6755,6 +6889,7 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 	struct ocfs2_extent_list *el;
 	struct buffer_head *last_eb_bh = NULL;
 	u64 delete_blk = 0;
+	u8 rec_flags;
 
 	fe = (struct ocfs2_dinode *) fe_bh->b_data;
 
@@ -6769,14 +6904,14 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 	 * Each component will be touched, so we might as well journal
 	 * here to avoid having to handle errors later.
 	 */
-	status = ocfs2_journal_access_path(inode, handle, path);
+	status = ocfs2_journal_access_path(INODE_CACHE(inode), handle, path);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
 	}
 
 	if (last_eb_bh) {
-		status = ocfs2_journal_access_eb(handle, inode, last_eb_bh,
+		status = ocfs2_journal_access_eb(handle, INODE_CACHE(inode), last_eb_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -6810,7 +6945,7 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 	inode->i_blocks = ocfs2_inode_sector_count(inode);
 
 	status = ocfs2_trim_tree(inode, path, handle, tc,
-				 clusters_to_del, &delete_blk);
+				 clusters_to_del, &delete_blk, &rec_flags);
 	if (status) {
 		mlog_errno(status);
 		goto bail;
@@ -6842,8 +6977,16 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 	}
 
 	if (delete_blk) {
-		status = ocfs2_truncate_log_append(osb, handle, delete_blk,
-						   clusters_to_del);
+		if (rec_flags & OCFS2_EXT_REFCOUNTED)
+			status = ocfs2_decrease_refcount(inode, handle,
+					ocfs2_blocks_to_clusters(osb->sb,
+								 delete_blk),
+					clusters_to_del, meta_ac,
+					&tc->tc_dealloc, 1);
+		else
+			status = ocfs2_truncate_log_append(osb, handle,
+							   delete_blk,
+							   clusters_to_del);
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail;
@@ -6863,9 +7006,9 @@ static int ocfs2_zero_func(handle_t *handle, struct buffer_head *bh)
 	return 0;
 }
 
-static void ocfs2_map_and_dirty_page(struct inode *inode, handle_t *handle,
-				     unsigned int from, unsigned int to,
-				     struct page *page, int zero, u64 *phys)
+void ocfs2_map_and_dirty_page(struct inode *inode, handle_t *handle,
+			      unsigned int from, unsigned int to,
+			      struct page *page, int zero, u64 *phys)
 {
 	int ret, partial = 0;
 
@@ -6933,20 +7076,16 @@ out:
 		ocfs2_unlock_and_free_pages(pages, numpages);
 }
 
-static int ocfs2_grab_eof_pages(struct inode *inode, loff_t start, loff_t end,
-				struct page **pages, int *num)
+int ocfs2_grab_pages(struct inode *inode, loff_t start, loff_t end,
+		     struct page **pages, int *num)
 {
 	int numpages, ret = 0;
-	struct super_block *sb = inode->i_sb;
 	struct address_space *mapping = inode->i_mapping;
 	unsigned long index;
 	loff_t last_page_bytes;
 
 	BUG_ON(start > end);
 
-	BUG_ON(start >> OCFS2_SB(sb)->s_clustersize_bits !=
-	       (end - 1) >> OCFS2_SB(sb)->s_clustersize_bits);
-
 	numpages = 0;
 	last_page_bytes = PAGE_ALIGN(end);
 	index = start >> PAGE_CACHE_SHIFT;
@@ -6974,6 +7113,17 @@ out:
 	return ret;
 }
 
+static int ocfs2_grab_eof_pages(struct inode *inode, loff_t start, loff_t end,
+				struct page **pages, int *num)
+{
+	struct super_block *sb = inode->i_sb;
+
+	BUG_ON(start >> OCFS2_SB(sb)->s_clustersize_bits !=
+	       (end - 1) >> OCFS2_SB(sb)->s_clustersize_bits);
+
+	return ocfs2_grab_pages(inode, start, end, pages, num);
+}
+
 /*
  * Zero the area past i_size but still within an allocated
  * cluster. This avoids exposing nonzero data on subsequent file
@@ -7138,7 +7288,7 @@ int ocfs2_convert_inline_data_to_extents(struct inode *inode,
 		goto out_unlock;
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -7218,9 +7368,8 @@ int ocfs2_convert_inline_data_to_extents(struct inode *inode,
 		 * this proves to be false, we could always re-build
 		 * the in-inode data from our pages.
 		 */
-		ocfs2_init_dinode_extent_tree(&et, inode, di_bh);
-		ret = ocfs2_insert_extent(osb, handle, inode, &et,
-					  0, block, 1, 0, NULL);
+		ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), di_bh);
+		ret = ocfs2_insert_extent(handle, &et, 0, block, 1, 0, NULL);
 		if (ret) {
 			mlog_errno(ret);
 			goto out_commit;
@@ -7262,11 +7411,14 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 {
 	int status, i, credits, tl_sem = 0;
 	u32 clusters_to_del, new_highest_cpos, range;
+	u64 blkno = 0;
 	struct ocfs2_extent_list *el;
 	handle_t *handle = NULL;
 	struct inode *tl_inode = osb->osb_tl_inode;
 	struct ocfs2_path *path = NULL;
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)fe_bh->b_data;
+	struct ocfs2_alloc_context *meta_ac = NULL;
+	struct ocfs2_refcount_tree *ref_tree = NULL;
 
 	mlog_entry_void();
 
@@ -7292,10 +7444,12 @@ start:
 		goto bail;
 	}
 
+	credits = 0;
+
 	/*
 	 * Truncate always works against the rightmost tree branch.
 	 */
-	status = ocfs2_find_path(inode, path, UINT_MAX);
+	status = ocfs2_find_path(INODE_CACHE(inode), path, UINT_MAX);
 	if (status) {
 		mlog_errno(status);
 		goto bail;
@@ -7332,10 +7486,15 @@ start:
 		clusters_to_del = 0;
 	} else if (le32_to_cpu(el->l_recs[i].e_cpos) >= new_highest_cpos) {
 		clusters_to_del = ocfs2_rec_clusters(el, &el->l_recs[i]);
+		blkno = le64_to_cpu(el->l_recs[i].e_blkno);
 	} else if (range > new_highest_cpos) {
 		clusters_to_del = (ocfs2_rec_clusters(el, &el->l_recs[i]) +
 				   le32_to_cpu(el->l_recs[i].e_cpos)) -
 				  new_highest_cpos;
+		blkno = le64_to_cpu(el->l_recs[i].e_blkno) +
+			ocfs2_clusters_to_blocks(inode->i_sb,
+				ocfs2_rec_clusters(el, &el->l_recs[i]) -
+				clusters_to_del);
 	} else {
 		status = 0;
 		goto bail;
@@ -7344,6 +7503,29 @@ start:
 	mlog(0, "clusters_to_del = %u in this pass, tail blk=%llu\n",
 	     clusters_to_del, (unsigned long long)path_leaf_bh(path)->b_blocknr);
 
+	if (el->l_recs[i].e_flags & OCFS2_EXT_REFCOUNTED && clusters_to_del) {
+		BUG_ON(!(OCFS2_I(inode)->ip_dyn_features &
+			 OCFS2_HAS_REFCOUNT_FL));
+
+		status = ocfs2_lock_refcount_tree(osb,
+						le64_to_cpu(di->i_refcount_loc),
+						1, &ref_tree, NULL);
+		if (status) {
+			mlog_errno(status);
+			goto bail;
+		}
+
+		status = ocfs2_prepare_refcount_change_for_del(inode, fe_bh,
+							       blkno,
+							       clusters_to_del,
+							       &credits,
+							       &meta_ac);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
 	mutex_lock(&tl_inode->i_mutex);
 	tl_sem = 1;
 	/* ocfs2_truncate_log_needs_flush guarantees us at least one
@@ -7357,7 +7539,7 @@ start:
 		}
 	}
 
-	credits = ocfs2_calc_tree_trunc_credits(osb->sb, clusters_to_del,
+	credits += ocfs2_calc_tree_trunc_credits(osb->sb, clusters_to_del,
 						(struct ocfs2_dinode *)fe_bh->b_data,
 						el);
 	handle = ocfs2_start_trans(osb, credits);
@@ -7369,7 +7551,7 @@ start:
 	}
 
 	status = ocfs2_do_truncate(osb, clusters_to_del, inode, fe_bh, handle,
-				   tc, path);
+				   tc, path, meta_ac);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -7383,6 +7565,16 @@ start:
 
 	ocfs2_reinit_path(path, 1);
 
+	if (meta_ac) {
+		ocfs2_free_alloc_context(meta_ac);
+		meta_ac = NULL;
+	}
+
+	if (ref_tree) {
+		ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+		ref_tree = NULL;
+	}
+
 	/*
 	 * The check above will catch the case where we've truncated
 	 * away all allocation.
@@ -7399,6 +7591,12 @@ bail:
 	if (handle)
 		ocfs2_commit_trans(osb, handle);
 
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+
+	if (ref_tree)
+		ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+
 	ocfs2_run_deallocs(osb, &tc->tc_dealloc);
 
 	ocfs2_free_path(path);
@@ -7445,7 +7643,7 @@ int ocfs2_prepare_truncate(struct ocfs2_super *osb,
 	ocfs2_init_dealloc_ctxt(&(*tc)->tc_dealloc);
 
 	if (fe->id2.i_list.l_tree_depth) {
-		status = ocfs2_read_extent_block(inode,
+		status = ocfs2_read_extent_block(INODE_CACHE(inode),
 						 le64_to_cpu(fe->i_last_eb_blk),
 						 &last_eb_bh);
 		if (status < 0) {
@@ -7507,7 +7705,7 @@ int ocfs2_truncate_inline(struct inode *inode, struct buffer_head *di_bh,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
diff --git a/fs/ocfs2/alloc.h b/fs/ocfs2/alloc.h
index 353254ba29e1..9c122d574464 100644
--- a/fs/ocfs2/alloc.h
+++ b/fs/ocfs2/alloc.h
@@ -45,7 +45,8 @@
  *
  * ocfs2_extent_tree contains info for the root of the b-tree, it must have a
  * root ocfs2_extent_list and a root_bh so that they can be used in the b-tree
- * functions.  With metadata ecc, we now call different journal_access
+ * functions.  It needs the ocfs2_caching_info structure associated with
+ * I/O on the tree.  With metadata ecc, we now call different journal_access
  * functions for each type of metadata, so it must have the
  * root_journal_access function.
  * ocfs2_extent_tree_operations abstract the normal operations we do for
@@ -56,6 +57,7 @@ struct ocfs2_extent_tree {
 	struct ocfs2_extent_tree_operations	*et_ops;
 	struct buffer_head			*et_root_bh;
 	struct ocfs2_extent_list		*et_root_el;
+	struct ocfs2_caching_info		*et_ci;
 	ocfs2_journal_access_func		et_root_journal_access;
 	void					*et_object;
 	unsigned int				et_max_leaf_clusters;
@@ -66,31 +68,32 @@ struct ocfs2_extent_tree {
  * specified object buffer.
  */
 void ocfs2_init_dinode_extent_tree(struct ocfs2_extent_tree *et,
-				   struct inode *inode,
+				   struct ocfs2_caching_info *ci,
 				   struct buffer_head *bh);
 void ocfs2_init_xattr_tree_extent_tree(struct ocfs2_extent_tree *et,
-				       struct inode *inode,
+				       struct ocfs2_caching_info *ci,
 				       struct buffer_head *bh);
 struct ocfs2_xattr_value_buf;
 void ocfs2_init_xattr_value_extent_tree(struct ocfs2_extent_tree *et,
-					struct inode *inode,
+					struct ocfs2_caching_info *ci,
 					struct ocfs2_xattr_value_buf *vb);
 void ocfs2_init_dx_root_extent_tree(struct ocfs2_extent_tree *et,
-				    struct inode *inode,
+				    struct ocfs2_caching_info *ci,
 				    struct buffer_head *bh);
+void ocfs2_init_refcount_extent_tree(struct ocfs2_extent_tree *et,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *bh);
 
 /*
  * Read an extent block into *bh.  If *bh is NULL, a bh will be
  * allocated.  This is a cached read.  The extent block will be validated
  * with ocfs2_validate_extent_block().
  */
-int ocfs2_read_extent_block(struct inode *inode, u64 eb_blkno,
+int ocfs2_read_extent_block(struct ocfs2_caching_info *ci, u64 eb_blkno,
 			    struct buffer_head **bh);
 
 struct ocfs2_alloc_context;
-int ocfs2_insert_extent(struct ocfs2_super *osb,
-			handle_t *handle,
-			struct inode *inode,
+int ocfs2_insert_extent(handle_t *handle,
 			struct ocfs2_extent_tree *et,
 			u32 cpos,
 			u64 start_blk,
@@ -103,25 +106,36 @@ enum ocfs2_alloc_restarted {
 	RESTART_TRANS,
 	RESTART_META
 };
-int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb,
-				struct inode *inode,
+int ocfs2_add_clusters_in_btree(handle_t *handle,
+				struct ocfs2_extent_tree *et,
 				u32 *logical_offset,
 				u32 clusters_to_add,
 				int mark_unwritten,
-				struct ocfs2_extent_tree *et,
-				handle_t *handle,
 				struct ocfs2_alloc_context *data_ac,
 				struct ocfs2_alloc_context *meta_ac,
 				enum ocfs2_alloc_restarted *reason_ret);
 struct ocfs2_cached_dealloc_ctxt;
+struct ocfs2_path;
+int ocfs2_split_extent(handle_t *handle,
+		       struct ocfs2_extent_tree *et,
+		       struct ocfs2_path *path,
+		       int split_index,
+		       struct ocfs2_extent_rec *split_rec,
+		       struct ocfs2_alloc_context *meta_ac,
+		       struct ocfs2_cached_dealloc_ctxt *dealloc);
 int ocfs2_mark_extent_written(struct inode *inode,
 			      struct ocfs2_extent_tree *et,
 			      handle_t *handle, u32 cpos, u32 len, u32 phys,
 			      struct ocfs2_alloc_context *meta_ac,
 			      struct ocfs2_cached_dealloc_ctxt *dealloc);
-int ocfs2_remove_extent(struct inode *inode,
-			struct ocfs2_extent_tree *et,
-			u32 cpos, u32 len, handle_t *handle,
+int ocfs2_change_extent_flag(handle_t *handle,
+			     struct ocfs2_extent_tree *et,
+			     u32 cpos, u32 len, u32 phys,
+			     struct ocfs2_alloc_context *meta_ac,
+			     struct ocfs2_cached_dealloc_ctxt *dealloc,
+			     int new_flags, int clear_flags);
+int ocfs2_remove_extent(handle_t *handle, struct ocfs2_extent_tree *et,
+			u32 cpos, u32 len,
 			struct ocfs2_alloc_context *meta_ac,
 			struct ocfs2_cached_dealloc_ctxt *dealloc);
 int ocfs2_remove_btree_range(struct inode *inode,
@@ -130,7 +144,6 @@ int ocfs2_remove_btree_range(struct inode *inode,
 			     struct ocfs2_cached_dealloc_ctxt *dealloc);
 
 int ocfs2_num_free_extents(struct ocfs2_super *osb,
-			   struct inode *inode,
 			   struct ocfs2_extent_tree *et);
 
 /*
@@ -195,6 +208,9 @@ static inline void ocfs2_init_dealloc_ctxt(struct ocfs2_cached_dealloc_ctxt *c)
 }
 int ocfs2_cache_cluster_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
 				u64 blkno, unsigned int bit);
+int ocfs2_cache_block_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
+			      int type, int slot, u64 blkno,
+			      unsigned int bit);
 static inline int ocfs2_dealloc_has_cluster(struct ocfs2_cached_dealloc_ctxt *c)
 {
 	return c->c_global_allocator != NULL;
@@ -222,8 +238,9 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 int ocfs2_truncate_inline(struct inode *inode, struct buffer_head *di_bh,
 			  unsigned int start, unsigned int end, int trunc);
 
-int ocfs2_find_leaf(struct inode *inode, struct ocfs2_extent_list *root_el,
-		    u32 cpos, struct buffer_head **leaf_bh);
+int ocfs2_find_leaf(struct ocfs2_caching_info *ci,
+		    struct ocfs2_extent_list *root_el, u32 cpos,
+		    struct buffer_head **leaf_bh);
 int ocfs2_search_extent_list(struct ocfs2_extent_list *el, u32 v_cluster);
 
 /*
@@ -254,4 +271,50 @@ static inline int ocfs2_is_empty_extent(struct ocfs2_extent_rec *rec)
 	return !rec->e_leaf_clusters;
 }
 
+int ocfs2_grab_pages(struct inode *inode, loff_t start, loff_t end,
+		     struct page **pages, int *num);
+void ocfs2_map_and_dirty_page(struct inode *inode, handle_t *handle,
+			      unsigned int from, unsigned int to,
+			      struct page *page, int zero, u64 *phys);
+/*
+ * Structures which describe a path through a btree, and functions to
+ * manipulate them.
+ *
+ * The idea here is to be as generic as possible with the tree
+ * manipulation code.
+ */
+struct ocfs2_path_item {
+	struct buffer_head		*bh;
+	struct ocfs2_extent_list	*el;
+};
+
+#define OCFS2_MAX_PATH_DEPTH	5
+
+struct ocfs2_path {
+	int				p_tree_depth;
+	ocfs2_journal_access_func	p_root_access;
+	struct ocfs2_path_item		p_node[OCFS2_MAX_PATH_DEPTH];
+};
+
+#define path_root_bh(_path) ((_path)->p_node[0].bh)
+#define path_root_el(_path) ((_path)->p_node[0].el)
+#define path_root_access(_path)((_path)->p_root_access)
+#define path_leaf_bh(_path) ((_path)->p_node[(_path)->p_tree_depth].bh)
+#define path_leaf_el(_path) ((_path)->p_node[(_path)->p_tree_depth].el)
+#define path_num_items(_path) ((_path)->p_tree_depth + 1)
+
+void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root);
+void ocfs2_free_path(struct ocfs2_path *path);
+int ocfs2_find_path(struct ocfs2_caching_info *ci,
+		    struct ocfs2_path *path,
+		    u32 cpos);
+struct ocfs2_path *ocfs2_new_path_from_path(struct ocfs2_path *path);
+struct ocfs2_path *ocfs2_new_path_from_et(struct ocfs2_extent_tree *et);
+int ocfs2_path_bh_journal_access(handle_t *handle,
+				 struct ocfs2_caching_info *ci,
+				 struct ocfs2_path *path,
+				 int idx);
+int ocfs2_journal_access_path(struct ocfs2_caching_info *ci,
+			      handle_t *handle,
+			      struct ocfs2_path *path);
 #endif /* OCFS2_ALLOC_H */
diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 8a1e61545f41..deb2b132ae5e 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -44,6 +44,7 @@
 #include "suballoc.h"
 #include "super.h"
 #include "symlink.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
@@ -126,8 +127,8 @@ bail:
 	return err;
 }
 
-static int ocfs2_get_block(struct inode *inode, sector_t iblock,
-			   struct buffer_head *bh_result, int create)
+int ocfs2_get_block(struct inode *inode, sector_t iblock,
+		    struct buffer_head *bh_result, int create)
 {
 	int err = 0;
 	unsigned int ext_flags;
@@ -590,6 +591,8 @@ static int ocfs2_direct_IO_get_blocks(struct inode *inode, sector_t iblock,
 		goto bail;
 	}
 
+	/* We should already CoW the refcounted extent. */
+	BUG_ON(ext_flags & OCFS2_EXT_REFCOUNTED);
 	/*
 	 * get_more_blocks() expects us to describe a hole by clearing
 	 * the mapped bit on bh_result().
@@ -687,6 +690,10 @@ static ssize_t ocfs2_direct_IO(int rw,
 	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
 		return 0;
 
+	/* Fallback to buffered I/O if we are appending. */
+	if (i_size_read(inode) <= offset)
+		return 0;
+
 	ret = blockdev_direct_IO_no_locking(rw, iocb, inode,
 					    inode->i_sb->s_bdev, iov, offset,
 					    nr_segs, 
@@ -1259,7 +1266,8 @@ static int ocfs2_write_cluster(struct address_space *mapping,
 			goto out;
 		}
 	} else if (unwritten) {
-		ocfs2_init_dinode_extent_tree(&et, inode, wc->w_di_bh);
+		ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode),
+					      wc->w_di_bh);
 		ret = ocfs2_mark_extent_written(inode, &et,
 						wc->w_handle, cpos, 1, phys,
 						meta_ac, &wc->w_dealloc);
@@ -1448,6 +1456,9 @@ static int ocfs2_populate_write_desc(struct inode *inode,
 				goto out;
 			}
 
+			/* We should already CoW the refcountd extent. */
+			BUG_ON(ext_flags & OCFS2_EXT_REFCOUNTED);
+
 			/*
 			 * Assume worst case - that we're writing in
 			 * the middle of the extent.
@@ -1528,7 +1539,7 @@ static int ocfs2_write_begin_inline(struct address_space *mapping,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, wc->w_di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), wc->w_di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		ocfs2_commit_trans(osb, handle);
@@ -1699,6 +1710,19 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
 		goto out;
 	}
 
+	ret = ocfs2_check_range_for_refcount(inode, pos, len);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	} else if (ret == 1) {
+		ret = ocfs2_refcount_cow(inode, di_bh,
+					 wc->w_cpos, wc->w_clen, UINT_MAX);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
 	ret = ocfs2_populate_write_desc(inode, wc, &clusters_to_alloc,
 					&extents_to_split);
 	if (ret) {
@@ -1726,7 +1750,8 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
 		     (long long)i_size_read(inode), le32_to_cpu(di->i_clusters),
 		     clusters_to_alloc, extents_to_split);
 
-		ocfs2_init_dinode_extent_tree(&et, inode, wc->w_di_bh);
+		ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode),
+					      wc->w_di_bh);
 		ret = ocfs2_lock_allocators(inode, &et,
 					    clusters_to_alloc, extents_to_split,
 					    &data_ac, &meta_ac);
@@ -1773,7 +1798,7 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
 	 * We don't want this to fail in ocfs2_write_end(), so do it
 	 * here.
 	 */
-	ret = ocfs2_journal_access_di(handle, inode, wc->w_di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), wc->w_di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1997,4 +2022,5 @@ const struct address_space_operations ocfs2_aops = {
 	.releasepage		= ocfs2_releasepage,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate	= block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h
index 503e49232e11..c48e93ffc513 100644
--- a/fs/ocfs2/aops.h
+++ b/fs/ocfs2/aops.h
@@ -57,6 +57,8 @@ int ocfs2_read_inline_data(struct inode *inode, struct page *page,
 			   struct buffer_head *di_bh);
 int ocfs2_size_fits_inline_data(struct buffer_head *di_bh, u64 new_size);
 
+int ocfs2_get_block(struct inode *inode, sector_t iblock,
+		    struct buffer_head *bh_result, int create);
 /* all ocfs2_dio_end_io()'s fault */
 #define ocfs2_iocb_is_rw_locked(iocb) \
 	test_bit(0, (unsigned long *)&iocb->private)
diff --git a/fs/ocfs2/buffer_head_io.c b/fs/ocfs2/buffer_head_io.c
index 15c8e6deee2e..d43d34a1dd31 100644
--- a/fs/ocfs2/buffer_head_io.c
+++ b/fs/ocfs2/buffer_head_io.c
@@ -52,12 +52,12 @@ enum ocfs2_state_bits {
 BUFFER_FNS(NeedsValidate, needs_validate);
 
 int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
-		      struct inode *inode)
+		      struct ocfs2_caching_info *ci)
 {
 	int ret = 0;
 
-	mlog_entry("(bh->b_blocknr = %llu, inode=%p)\n",
-		   (unsigned long long)bh->b_blocknr, inode);
+	mlog_entry("(bh->b_blocknr = %llu, ci=%p)\n",
+		   (unsigned long long)bh->b_blocknr, ci);
 
 	BUG_ON(bh->b_blocknr < OCFS2_SUPER_BLOCK_BLKNO);
 	BUG_ON(buffer_jbd(bh));
@@ -70,7 +70,7 @@ int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
 		goto out;
 	}
 
-	mutex_lock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_lock(ci);
 
 	lock_buffer(bh);
 	set_buffer_uptodate(bh);
@@ -85,7 +85,7 @@ int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
 	wait_on_buffer(bh);
 
 	if (buffer_uptodate(bh)) {
-		ocfs2_set_buffer_uptodate(inode, bh);
+		ocfs2_set_buffer_uptodate(ci, bh);
 	} else {
 		/* We don't need to remove the clustered uptodate
 		 * information for this bh as it's not marked locally
@@ -94,7 +94,7 @@ int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
 		put_bh(bh);
 	}
 
-	mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_unlock(ci);
 out:
 	mlog_exit(ret);
 	return ret;
@@ -177,7 +177,7 @@ bail:
 	return status;
 }
 
-int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
+int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
 		      struct buffer_head *bhs[], int flags,
 		      int (*validate)(struct super_block *sb,
 				      struct buffer_head *bh))
@@ -185,11 +185,12 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 	int status = 0;
 	int i, ignore_cache = 0;
 	struct buffer_head *bh;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
 
-	mlog_entry("(inode=%p, block=(%llu), nr=(%d), flags=%d)\n",
-		   inode, (unsigned long long)block, nr, flags);
+	mlog_entry("(ci=%p, block=(%llu), nr=(%d), flags=%d)\n",
+		   ci, (unsigned long long)block, nr, flags);
 
-	BUG_ON(!inode);
+	BUG_ON(!ci);
 	BUG_ON((flags & OCFS2_BH_READAHEAD) &&
 	       (flags & OCFS2_BH_IGNORE_CACHE));
 
@@ -212,12 +213,12 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 		goto bail;
 	}
 
-	mutex_lock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_lock(ci);
 	for (i = 0 ; i < nr ; i++) {
 		if (bhs[i] == NULL) {
-			bhs[i] = sb_getblk(inode->i_sb, block++);
+			bhs[i] = sb_getblk(sb, block++);
 			if (bhs[i] == NULL) {
-				mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
+				ocfs2_metadata_cache_io_unlock(ci);
 				status = -EIO;
 				mlog_errno(status);
 				goto bail;
@@ -250,11 +251,11 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 		 *    before our is-it-in-flight check.
 		 */
 
-		if (!ignore_cache && !ocfs2_buffer_uptodate(inode, bh)) {
+		if (!ignore_cache && !ocfs2_buffer_uptodate(ci, bh)) {
 			mlog(ML_UPTODATE,
-			     "bh (%llu), inode %llu not uptodate\n",
+			     "bh (%llu), owner %llu not uptodate\n",
 			     (unsigned long long)bh->b_blocknr,
-			     (unsigned long long)OCFS2_I(inode)->ip_blkno);
+			     (unsigned long long)ocfs2_metadata_cache_owner(ci));
 			/* We're using ignore_cache here to say
 			 * "go to disk" */
 			ignore_cache = 1;
@@ -283,7 +284,7 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 			 * previously submitted request than we are
 			 * done here. */
 			if ((flags & OCFS2_BH_READAHEAD)
-			    && ocfs2_buffer_read_ahead(inode, bh))
+			    && ocfs2_buffer_read_ahead(ci, bh))
 				continue;
 
 			lock_buffer(bh);
@@ -305,7 +306,7 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 			 * buffer lock. */
 			if (!(flags & OCFS2_BH_IGNORE_CACHE)
 			    && !(flags & OCFS2_BH_READAHEAD)
-			    && ocfs2_buffer_uptodate(inode, bh)) {
+			    && ocfs2_buffer_uptodate(ci, bh)) {
 				unlock_buffer(bh);
 				continue;
 			}
@@ -327,7 +328,7 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 
 		if (!(flags & OCFS2_BH_READAHEAD)) {
 			/* We know this can't have changed as we hold the
-			 * inode sem. Avoid doing any work on the bh if the
+			 * owner sem. Avoid doing any work on the bh if the
 			 * journal has it. */
 			if (!buffer_jbd(bh))
 				wait_on_buffer(bh);
@@ -351,7 +352,7 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 				 * that better not have changed */
 				BUG_ON(buffer_jbd(bh));
 				clear_buffer_needs_validate(bh);
-				status = validate(inode->i_sb, bh);
+				status = validate(sb, bh);
 				if (status) {
 					put_bh(bh);
 					bhs[i] = NULL;
@@ -363,9 +364,9 @@ int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
 		/* Always set the buffer in the cache, even if it was
 		 * a forced read, or read-ahead which hasn't yet
 		 * completed. */
-		ocfs2_set_buffer_uptodate(inode, bh);
+		ocfs2_set_buffer_uptodate(ci, bh);
 	}
-	mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_unlock(ci);
 
 	mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n", 
 	     (unsigned long long)block, nr,
@@ -399,7 +400,7 @@ static void ocfs2_check_super_or_backup(struct super_block *sb,
 
 /*
  * Write super block and backups doesn't need to collaborate with journal,
- * so we don't need to lock ip_io_mutex and inode doesn't need to bea passed
+ * so we don't need to lock ip_io_mutex and ci doesn't need to bea passed
  * into this function.
  */
 int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
diff --git a/fs/ocfs2/buffer_head_io.h b/fs/ocfs2/buffer_head_io.h
index c75d682dadd8..b97bcc6dde7c 100644
--- a/fs/ocfs2/buffer_head_io.h
+++ b/fs/ocfs2/buffer_head_io.h
@@ -33,7 +33,7 @@ void ocfs2_end_buffer_io_sync(struct buffer_head *bh,
 
 int ocfs2_write_block(struct ocfs2_super          *osb,
 		      struct buffer_head  *bh,
-		      struct inode        *inode);
+		      struct ocfs2_caching_info   *ci);
 int ocfs2_read_blocks_sync(struct ocfs2_super *osb, u64 block,
 			   unsigned int nr, struct buffer_head *bhs[]);
 
@@ -44,7 +44,7 @@ int ocfs2_read_blocks_sync(struct ocfs2_super *osb, u64 block,
  * be set even for a READAHEAD call, as it marks the buffer for later
  * validation.
  */
-int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
+int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
 		      struct buffer_head *bhs[], int flags,
 		      int (*validate)(struct super_block *sb,
 				      struct buffer_head *bh));
@@ -55,7 +55,7 @@ int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
 #define OCFS2_BH_IGNORE_CACHE      1
 #define OCFS2_BH_READAHEAD         8
 
-static inline int ocfs2_read_block(struct inode *inode, u64 off,
+static inline int ocfs2_read_block(struct ocfs2_caching_info *ci, u64 off,
 				   struct buffer_head **bh,
 				   int (*validate)(struct super_block *sb,
 						   struct buffer_head *bh))
@@ -68,7 +68,7 @@ static inline int ocfs2_read_block(struct inode *inode, u64 off,
 		goto bail;
 	}
 
-	status = ocfs2_read_blocks(inode, off, 1, bh, 0, validate);
+	status = ocfs2_read_blocks(ci, off, 1, bh, 0, validate);
 
 bail:
 	return status;
diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c
index 09cc25d04611..c452d116b892 100644
--- a/fs/ocfs2/cluster/heartbeat.c
+++ b/fs/ocfs2/cluster/heartbeat.c
@@ -966,7 +966,7 @@ static ssize_t o2hb_debug_read(struct file *file, char __user *buf,
 }
 #endif  /* CONFIG_DEBUG_FS */
 
-static struct file_operations o2hb_debug_fops = {
+static const struct file_operations o2hb_debug_fops = {
 	.open =		o2hb_debug_open,
 	.release =	o2hb_debug_release,
 	.read =		o2hb_debug_read,
diff --git a/fs/ocfs2/cluster/masklog.c b/fs/ocfs2/cluster/masklog.c
index 96df5416993e..1cd2934de615 100644
--- a/fs/ocfs2/cluster/masklog.c
+++ b/fs/ocfs2/cluster/masklog.c
@@ -111,6 +111,7 @@ static struct mlog_attribute mlog_attrs[MLOG_MAX_BITS] = {
 	define_mask(EXPORT),
 	define_mask(XATTR),
 	define_mask(QUOTA),
+	define_mask(REFCOUNT),
 	define_mask(ERROR),
 	define_mask(NOTICE),
 	define_mask(KTHREAD),
diff --git a/fs/ocfs2/cluster/masklog.h b/fs/ocfs2/cluster/masklog.h
index 696c32e50716..9b4d11726cf2 100644
--- a/fs/ocfs2/cluster/masklog.h
+++ b/fs/ocfs2/cluster/masklog.h
@@ -113,6 +113,7 @@
 #define ML_EXPORT	0x0000000010000000ULL /* ocfs2 export operations */
 #define ML_XATTR	0x0000000020000000ULL /* ocfs2 extended attributes */
 #define ML_QUOTA	0x0000000040000000ULL /* ocfs2 quota operations */
+#define ML_REFCOUNT	0x0000000080000000ULL /* refcount tree operations */
 /* bits that are infrequently given and frequently matched in the high word */
 #define ML_ERROR	0x0000000100000000ULL /* sent to KERN_ERR */
 #define ML_NOTICE	0x0000000200000000ULL /* setn to KERN_NOTICE */
diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
index f8424874fa07..da794bc07a6c 100644
--- a/fs/ocfs2/cluster/netdebug.c
+++ b/fs/ocfs2/cluster/netdebug.c
@@ -163,7 +163,7 @@ static void nst_seq_stop(struct seq_file *seq, void *v)
 {
 }
 
-static struct seq_operations nst_seq_ops = {
+static const struct seq_operations nst_seq_ops = {
 	.start = nst_seq_start,
 	.next = nst_seq_next,
 	.stop = nst_seq_stop,
@@ -207,7 +207,7 @@ static int nst_fop_release(struct inode *inode, struct file *file)
 	return seq_release_private(inode, file);
 }
 
-static struct file_operations nst_seq_fops = {
+static const struct file_operations nst_seq_fops = {
 	.open = nst_fop_open,
 	.read = seq_read,
 	.llseek = seq_lseek,
@@ -344,7 +344,7 @@ static void sc_seq_stop(struct seq_file *seq, void *v)
 {
 }
 
-static struct seq_operations sc_seq_ops = {
+static const struct seq_operations sc_seq_ops = {
 	.start = sc_seq_start,
 	.next = sc_seq_next,
 	.stop = sc_seq_stop,
@@ -388,7 +388,7 @@ static int sc_fop_release(struct inode *inode, struct file *file)
 	return seq_release_private(inode, file);
 }
 
-static struct file_operations sc_seq_fops = {
+static const struct file_operations sc_seq_fops = {
 	.open = sc_fop_open,
 	.read = seq_read,
 	.llseek = seq_lseek,
diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c
index b358f3bf896d..28c3ec238796 100644
--- a/fs/ocfs2/dir.c
+++ b/fs/ocfs2/dir.c
@@ -176,7 +176,7 @@ static int ocfs2_dx_dir_link_trailer(struct inode *dir, handle_t *handle,
 	struct ocfs2_dx_root_block *dx_root;
 	struct ocfs2_dir_block_trailer *trailer;
 
-	ret = ocfs2_journal_access_dr(handle, dir, dx_root_bh,
+	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -564,7 +564,8 @@ static int ocfs2_read_dir_block_direct(struct inode *dir, u64 phys,
 	int ret;
 	struct buffer_head *tmp = *bh;
 
-	ret = ocfs2_read_block(dir, phys, &tmp, ocfs2_validate_dir_block);
+	ret = ocfs2_read_block(INODE_CACHE(dir), phys, &tmp,
+			       ocfs2_validate_dir_block);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -622,7 +623,8 @@ static int ocfs2_read_dx_root(struct inode *dir, struct ocfs2_dinode *di,
 	u64 blkno = le64_to_cpu(di->i_dx_root);
 	struct buffer_head *tmp = *dx_root_bh;
 
-	ret = ocfs2_read_block(dir, blkno, &tmp, ocfs2_validate_dx_root);
+	ret = ocfs2_read_block(INODE_CACHE(dir), blkno, &tmp,
+			       ocfs2_validate_dx_root);
 
 	/* If ocfs2_read_block() got us a new bh, pass it up. */
 	if (!ret && !*dx_root_bh)
@@ -662,7 +664,8 @@ static int ocfs2_read_dx_leaf(struct inode *dir, u64 blkno,
 	int ret;
 	struct buffer_head *tmp = *dx_leaf_bh;
 
-	ret = ocfs2_read_block(dir, blkno, &tmp, ocfs2_validate_dx_leaf);
+	ret = ocfs2_read_block(INODE_CACHE(dir), blkno, &tmp,
+			       ocfs2_validate_dx_leaf);
 
 	/* If ocfs2_read_block() got us a new bh, pass it up. */
 	if (!ret && !*dx_leaf_bh)
@@ -680,7 +683,7 @@ static int ocfs2_read_dx_leaves(struct inode *dir, u64 start, int num,
 {
 	int ret;
 
-	ret = ocfs2_read_blocks(dir, start, num, dx_leaf_bhs, 0,
+	ret = ocfs2_read_blocks(INODE_CACHE(dir), start, num, dx_leaf_bhs, 0,
 				ocfs2_validate_dx_leaf);
 	if (ret)
 		mlog_errno(ret);
@@ -802,7 +805,8 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 	struct ocfs2_extent_rec *rec = NULL;
 
 	if (el->l_tree_depth) {
-		ret = ocfs2_find_leaf(inode, el, major_hash, &eb_bh);
+		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, major_hash,
+				      &eb_bh);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -1133,7 +1137,8 @@ int ocfs2_update_entry(struct inode *dir, handle_t *handle,
 	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
 		access = ocfs2_journal_access_di;
 
-	ret = access(handle, dir, de_bh, OCFS2_JOURNAL_ACCESS_WRITE);
+	ret = access(handle, INODE_CACHE(dir), de_bh,
+		     OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -1176,7 +1181,7 @@ static int __ocfs2_delete_entry(handle_t *handle, struct inode *dir,
 			goto bail;
 		}
 		if (de == de_del)  {
-			status = access(handle, dir, bh,
+			status = access(handle, INODE_CACHE(dir), bh,
 					OCFS2_JOURNAL_ACCESS_WRITE);
 			if (status < 0) {
 				status = -EIO;
@@ -1326,7 +1331,7 @@ static int ocfs2_delete_entry_dx(handle_t *handle, struct inode *dir,
 	 * the entry count needs to be updated. Also, we might be
 	 * adding to the start of the free list.
 	 */
-	ret = ocfs2_journal_access_dr(handle, dir, dx_root_bh,
+	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1334,7 +1339,7 @@ static int ocfs2_delete_entry_dx(handle_t *handle, struct inode *dir,
 	}
 
 	if (!ocfs2_dx_root_inline(dx_root)) {
-		ret = ocfs2_journal_access_dl(handle, dir,
+		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
 					      lookup->dl_dx_leaf_bh,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
@@ -1493,7 +1498,7 @@ static int __ocfs2_dx_dir_leaf_insert(struct inode *dir, handle_t *handle,
 	int ret;
 	struct ocfs2_dx_leaf *dx_leaf;
 
-	ret = ocfs2_journal_access_dl(handle, dir, dx_leaf_bh,
+	ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), dx_leaf_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1523,7 +1528,7 @@ static int ocfs2_dx_dir_insert(struct inode *dir, handle_t *handle,
 	struct ocfs2_dx_root_block *dx_root;
 	struct buffer_head *dx_root_bh = lookup->dl_dx_root_bh;
 
-	ret = ocfs2_journal_access_dr(handle, dir, dx_root_bh,
+	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1645,11 +1650,13 @@ int __ocfs2_add_entry(handle_t *handle,
 		 */
 		if (ocfs2_free_list_at_root(lookup)) {
 			bh = lookup->dl_dx_root_bh;
-			retval = ocfs2_journal_access_dr(handle, dir, bh,
+			retval = ocfs2_journal_access_dr(handle,
+						 INODE_CACHE(dir), bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		} else {
 			bh = lookup->dl_prev_leaf_bh;
-			retval = ocfs2_journal_access_db(handle, dir, bh,
+			retval = ocfs2_journal_access_db(handle,
+						 INODE_CACHE(dir), bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		}
 		if (retval) {
@@ -1700,11 +1707,13 @@ int __ocfs2_add_entry(handle_t *handle,
 			}
 
 			if (insert_bh == parent_fe_bh)
-				status = ocfs2_journal_access_di(handle, dir,
+				status = ocfs2_journal_access_di(handle,
+								 INODE_CACHE(dir),
 								 insert_bh,
 								 OCFS2_JOURNAL_ACCESS_WRITE);
 			else {
-				status = ocfs2_journal_access_db(handle, dir,
+				status = ocfs2_journal_access_db(handle,
+								 INODE_CACHE(dir),
 								 insert_bh,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 
@@ -2280,7 +2289,7 @@ static int ocfs2_fill_new_dir_id(struct ocfs2_super *osb,
 	struct ocfs2_inline_data *data = &di->id2.i_data;
 	unsigned int size = le16_to_cpu(data->id_count);
 
-	ret = ocfs2_journal_access_di(handle, inode, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -2332,9 +2341,9 @@ static int ocfs2_fill_new_dir_el(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	ocfs2_set_new_buffer_uptodate(inode, new_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), new_bh);
 
-	status = ocfs2_journal_access_db(handle, inode, new_bh,
+	status = ocfs2_journal_access_db(handle, INODE_CACHE(inode), new_bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -2418,9 +2427,9 @@ static int ocfs2_dx_dir_attach_index(struct ocfs2_super *osb,
 		ret = -EIO;
 		goto out;
 	}
-	ocfs2_set_new_buffer_uptodate(dir, dx_root_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), dx_root_bh);
 
-	ret = ocfs2_journal_access_dr(handle, dir, dx_root_bh,
+	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
 				      OCFS2_JOURNAL_ACCESS_CREATE);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -2454,7 +2463,7 @@ static int ocfs2_dx_dir_attach_index(struct ocfs2_super *osb,
 	if (ret)
 		mlog_errno(ret);
 
-	ret = ocfs2_journal_access_di(handle, dir, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
 				      OCFS2_JOURNAL_ACCESS_CREATE);
 	if (ret) {
 		mlog_errno(ret);
@@ -2495,9 +2504,9 @@ static int ocfs2_dx_dir_format_cluster(struct ocfs2_super *osb,
 		}
 		dx_leaves[i] = bh;
 
-		ocfs2_set_new_buffer_uptodate(dir, bh);
+		ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), bh);
 
-		ret = ocfs2_journal_access_dl(handle, dir, bh,
+		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), bh,
 					      OCFS2_JOURNAL_ACCESS_CREATE);
 		if (ret < 0) {
 			mlog_errno(ret);
@@ -2582,7 +2591,6 @@ static int ocfs2_dx_dir_new_cluster(struct inode *dir,
 {
 	int ret;
 	u64 phys_blkno;
-	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
 
 	ret = __ocfs2_dx_dir_new_cluster(dir, cpos, handle, data_ac, dx_leaves,
 					 num_dx_leaves, &phys_blkno);
@@ -2591,7 +2599,7 @@ static int ocfs2_dx_dir_new_cluster(struct inode *dir,
 		goto out;
 	}
 
-	ret = ocfs2_insert_extent(osb, handle, dir, et, cpos, phys_blkno, 1, 0,
+	ret = ocfs2_insert_extent(handle, et, cpos, phys_blkno, 1, 0,
 				  meta_ac);
 	if (ret)
 		mlog_errno(ret);
@@ -2895,7 +2903,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 	struct ocfs2_extent_tree dx_et;
 	int did_quota = 0, bytes_allocated = 0;
 
-	ocfs2_init_dinode_extent_tree(&et, dir, di_bh);
+	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(dir), di_bh);
 
 	alloc = ocfs2_clusters_for_bytes(sb, bytes);
 	dx_alloc = 0;
@@ -3005,9 +3013,9 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 		goto out_commit;
 	}
 
-	ocfs2_set_new_buffer_uptodate(dir, dirdata_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), dirdata_bh);
 
-	ret = ocfs2_journal_access_db(handle, dir, dirdata_bh,
+	ret = ocfs2_journal_access_db(handle, INODE_CACHE(dir), dirdata_bh,
 				      OCFS2_JOURNAL_ACCESS_CREATE);
 	if (ret) {
 		mlog_errno(ret);
@@ -3060,7 +3068,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 	 * We let the later dirent insert modify c/mtime - to the user
 	 * the data hasn't changed.
 	 */
-	ret = ocfs2_journal_access_di(handle, dir, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
 				      OCFS2_JOURNAL_ACCESS_CREATE);
 	if (ret) {
 		mlog_errno(ret);
@@ -3085,7 +3093,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 	 * This should never fail as our extent list is empty and all
 	 * related blocks have been journaled already.
 	 */
-	ret = ocfs2_insert_extent(osb, handle, dir, &et, 0, blkno, len,
+	ret = ocfs2_insert_extent(handle, &et, 0, blkno, len,
 				  0, NULL);
 	if (ret) {
 		mlog_errno(ret);
@@ -3117,8 +3125,10 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 			ocfs2_dx_dir_index_root_block(dir, dx_root_bh,
 						      dirdata_bh);
 		} else {
-			ocfs2_init_dx_root_extent_tree(&dx_et, dir, dx_root_bh);
-			ret = ocfs2_insert_extent(osb, handle, dir, &dx_et, 0,
+			ocfs2_init_dx_root_extent_tree(&dx_et,
+						       INODE_CACHE(dir),
+						       dx_root_bh);
+			ret = ocfs2_insert_extent(handle, &dx_et, 0,
 						  dx_insert_blkno, 1, 0, NULL);
 			if (ret)
 				mlog_errno(ret);
@@ -3138,7 +3148,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 		}
 		blkno = ocfs2_clusters_to_blocks(dir->i_sb, bit_off);
 
-		ret = ocfs2_insert_extent(osb, handle, dir, &et, 1,
+		ret = ocfs2_insert_extent(handle, &et, 1,
 					  blkno, len, 0, NULL);
 		if (ret) {
 			mlog_errno(ret);
@@ -3337,8 +3347,9 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 	spin_lock(&OCFS2_I(dir)->ip_lock);
 	if (dir_i_size == ocfs2_clusters_to_bytes(sb, OCFS2_I(dir)->ip_clusters)) {
 		spin_unlock(&OCFS2_I(dir)->ip_lock);
-		ocfs2_init_dinode_extent_tree(&et, dir, parent_fe_bh);
-		num_free_extents = ocfs2_num_free_extents(osb, dir, &et);
+		ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(dir),
+					      parent_fe_bh);
+		num_free_extents = ocfs2_num_free_extents(osb, &et);
 		if (num_free_extents < 0) {
 			status = num_free_extents;
 			mlog_errno(status);
@@ -3387,9 +3398,9 @@ do_extend:
 		goto bail;
 	}
 
-	ocfs2_set_new_buffer_uptodate(dir, new_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), new_bh);
 
-	status = ocfs2_journal_access_db(handle, dir, new_bh,
+	status = ocfs2_journal_access_db(handle, INODE_CACHE(dir), new_bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -3829,7 +3840,7 @@ static int ocfs2_dx_dir_rebalance(struct ocfs2_super *osb, struct inode *dir,
 	     (unsigned long long)OCFS2_I(dir)->ip_blkno,
 	     (unsigned long long)leaf_blkno, insert_hash);
 
-	ocfs2_init_dx_root_extent_tree(&et, dir, dx_root_bh);
+	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
 
 	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
 	/*
@@ -3885,7 +3896,7 @@ static int ocfs2_dx_dir_rebalance(struct ocfs2_super *osb, struct inode *dir,
 	}
 	did_quota = 1;
 
-	ret = ocfs2_journal_access_dl(handle, dir, dx_leaf_bh,
+	ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), dx_leaf_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -3949,7 +3960,8 @@ static int ocfs2_dx_dir_rebalance(struct ocfs2_super *osb, struct inode *dir,
 	}
 
 	for (i = 0; i < num_dx_leaves; i++) {
-		ret = ocfs2_journal_access_dl(handle, dir, orig_dx_leaves[i],
+		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
+					      orig_dx_leaves[i],
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
 			mlog_errno(ret);
@@ -4165,7 +4177,7 @@ static int ocfs2_expand_inline_dx_root(struct inode *dir,
 	 * failure to add the dx_root_bh to the journal won't result
 	 * us losing clusters.
 	 */
-	ret = ocfs2_journal_access_dr(handle, dir, dx_root_bh,
+	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -4207,9 +4219,8 @@ static int ocfs2_expand_inline_dx_root(struct inode *dir,
 
 	/* This should never fail considering we start with an empty
 	 * dx_root. */
-	ocfs2_init_dx_root_extent_tree(&et, dir, dx_root_bh);
-	ret = ocfs2_insert_extent(osb, handle, dir, &et, 0,
-				  insert_blkno, 1, 0, NULL);
+	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
+	ret = ocfs2_insert_extent(handle, &et, 0, insert_blkno, 1, 0, NULL);
 	if (ret)
 		mlog_errno(ret);
 	did_quota = 0;
@@ -4469,7 +4480,7 @@ static int ocfs2_dx_dir_remove_index(struct inode *dir,
 		goto out_unlock;
 	}
 
-	ret = ocfs2_journal_access_di(handle, dir, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -4532,7 +4543,7 @@ int ocfs2_dx_dir_truncate(struct inode *dir, struct buffer_head *di_bh)
 	if (ocfs2_dx_root_inline(dx_root))
 		goto remove_index;
 
-	ocfs2_init_dx_root_extent_tree(&et, dir, dx_root_bh);
+	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
 
 	/* XXX: What if dr_clusters is too large? */
 	while (le32_to_cpu(dx_root->dr_clusters)) {
@@ -4565,7 +4576,7 @@ remove_index:
 		goto out;
 	}
 
-	ocfs2_remove_from_cache(dir, dx_root_bh);
+	ocfs2_remove_from_cache(INODE_CACHE(dir), dx_root_bh);
 out:
 	ocfs2_schedule_truncate_log_flush(osb, 1);
 	ocfs2_run_deallocs(osb, &dealloc);
diff --git a/fs/ocfs2/dlm/dlmast.c b/fs/ocfs2/dlm/dlmast.c
index 81eff8e58322..01cf8cc3d286 100644
--- a/fs/ocfs2/dlm/dlmast.c
+++ b/fs/ocfs2/dlm/dlmast.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlm/dlmconvert.c b/fs/ocfs2/dlm/dlmconvert.c
index 75997b4deaf3..ca96bce50e18 100644
--- a/fs/ocfs2/dlm/dlmconvert.c
+++ b/fs/ocfs2/dlm/dlmconvert.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlm/dlmdebug.c b/fs/ocfs2/dlm/dlmdebug.c
index df52f706f669..42b0bad7a612 100644
--- a/fs/ocfs2/dlm/dlmdebug.c
+++ b/fs/ocfs2/dlm/dlmdebug.c
@@ -27,7 +27,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/sysctl.h>
 #include <linux/spinlock.h>
 #include <linux/debugfs.h>
@@ -479,7 +478,7 @@ bail:
 	return -ENOMEM;
 }
 
-static struct file_operations debug_purgelist_fops = {
+static const struct file_operations debug_purgelist_fops = {
 	.open =		debug_purgelist_open,
 	.release =	debug_buffer_release,
 	.read =		debug_buffer_read,
@@ -539,7 +538,7 @@ bail:
 	return -ENOMEM;
 }
 
-static struct file_operations debug_mle_fops = {
+static const struct file_operations debug_mle_fops = {
 	.open =		debug_mle_open,
 	.release =	debug_buffer_release,
 	.read =		debug_buffer_read,
@@ -683,7 +682,7 @@ static int lockres_seq_show(struct seq_file *s, void *v)
 	return 0;
 }
 
-static struct seq_operations debug_lockres_ops = {
+static const struct seq_operations debug_lockres_ops = {
 	.start =	lockres_seq_start,
 	.stop =		lockres_seq_stop,
 	.next =		lockres_seq_next,
@@ -742,7 +741,7 @@ static int debug_lockres_release(struct inode *inode, struct file *file)
 	return seq_release_private(inode, file);
 }
 
-static struct file_operations debug_lockres_fops = {
+static const struct file_operations debug_lockres_fops = {
 	.open =		debug_lockres_open,
 	.release =	debug_lockres_release,
 	.read =		seq_read,
@@ -926,7 +925,7 @@ bail:
 	return -ENOMEM;
 }
 
-static struct file_operations debug_state_fops = {
+static const struct file_operations debug_state_fops = {
 	.open =		debug_state_open,
 	.release =	debug_buffer_release,
 	.read =		debug_buffer_read,
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 4d9e6b288dd8..0334000676d3 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -28,7 +28,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/spinlock.h>
 #include <linux/delay.h>
diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c
index 83a9f2972ac8..437698e9465f 100644
--- a/fs/ocfs2/dlm/dlmlock.c
+++ b/fs/ocfs2/dlm/dlmlock.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index f8b653fcd4dd..83bcaf266b35 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 43e6e3280569..d9fa3d22e17c 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c
index d490b66ad9d7..52ec020ea78b 100644
--- a/fs/ocfs2/dlm/dlmthread.c
+++ b/fs/ocfs2/dlm/dlmthread.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
@@ -212,14 +211,18 @@ static int dlm_purge_lockres(struct dlm_ctxt *dlm,
 		spin_lock(&dlm->spinlock);
 	}
 
+	spin_lock(&res->spinlock);
 	if (!list_empty(&res->purge)) {
 		mlog(0, "removing lockres %.*s:%p from purgelist, "
 		     "master = %d\n", res->lockname.len, res->lockname.name,
 		     res, master);
 		list_del_init(&res->purge);
+		spin_unlock(&res->spinlock);
 		dlm_lockres_put(res);
 		dlm->purge_count--;
-	}
+	} else
+		spin_unlock(&res->spinlock);
+
 	__dlm_unhash_lockres(res);
 
 	/* lockres is not in the hash now.  drop the flag and wake up
diff --git a/fs/ocfs2/dlm/dlmunlock.c b/fs/ocfs2/dlm/dlmunlock.c
index 756f5b0998e0..00f53b2aea76 100644
--- a/fs/ocfs2/dlm/dlmunlock.c
+++ b/fs/ocfs2/dlm/dlmunlock.c
@@ -30,7 +30,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/sysctl.h>
 #include <linux/random.h>
diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index 110bb57c46ab..0d38d67194cb 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -53,6 +53,7 @@
 #include "super.h"
 #include "uptodate.h"
 #include "quota.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
@@ -110,6 +111,11 @@ static void ocfs2_dentry_post_unlock(struct ocfs2_super *osb,
 
 static void ocfs2_set_qinfo_lvb(struct ocfs2_lock_res *lockres);
 
+static int ocfs2_check_refcount_downconvert(struct ocfs2_lock_res *lockres,
+					    int new_level);
+static int ocfs2_refcount_convert_worker(struct ocfs2_lock_res *lockres,
+					 int blocking);
+
 #define mlog_meta_lvb(__level, __lockres) ocfs2_dump_meta_lvb_info(__level, __PRETTY_FUNCTION__, __LINE__, __lockres)
 
 /* This aids in debugging situations where a bad LVB might be involved. */
@@ -278,6 +284,12 @@ static struct ocfs2_lock_res_ops ocfs2_qinfo_lops = {
 	.flags		= LOCK_TYPE_REQUIRES_REFRESH | LOCK_TYPE_USES_LVB,
 };
 
+static struct ocfs2_lock_res_ops ocfs2_refcount_block_lops = {
+	.check_downconvert = ocfs2_check_refcount_downconvert,
+	.downconvert_worker = ocfs2_refcount_convert_worker,
+	.flags		= 0,
+};
+
 static inline int ocfs2_is_inode_lock(struct ocfs2_lock_res *lockres)
 {
 	return lockres->l_type == OCFS2_LOCK_TYPE_META ||
@@ -306,6 +318,12 @@ static inline struct ocfs2_mem_dqinfo *ocfs2_lock_res_qinfo(struct ocfs2_lock_re
 	return (struct ocfs2_mem_dqinfo *)lockres->l_priv;
 }
 
+static inline struct ocfs2_refcount_tree *
+ocfs2_lock_res_refcount_tree(struct ocfs2_lock_res *res)
+{
+	return container_of(res, struct ocfs2_refcount_tree, rf_lockres);
+}
+
 static inline struct ocfs2_super *ocfs2_get_lockres_osb(struct ocfs2_lock_res *lockres)
 {
 	if (lockres->l_ops->get_osb)
@@ -693,6 +711,17 @@ void ocfs2_qinfo_lock_res_init(struct ocfs2_lock_res *lockres,
 				   info);
 }
 
+void ocfs2_refcount_lock_res_init(struct ocfs2_lock_res *lockres,
+				  struct ocfs2_super *osb, u64 ref_blkno,
+				  unsigned int generation)
+{
+	ocfs2_lock_res_init_once(lockres);
+	ocfs2_build_lock_name(OCFS2_LOCK_TYPE_REFCOUNT, ref_blkno,
+			      generation, lockres->l_name);
+	ocfs2_lock_res_init_common(osb, lockres, OCFS2_LOCK_TYPE_REFCOUNT,
+				   &ocfs2_refcount_block_lops, osb);
+}
+
 void ocfs2_lock_res_free(struct ocfs2_lock_res *res)
 {
 	mlog_entry_void();
@@ -1548,8 +1577,10 @@ int ocfs2_rw_lock(struct inode *inode, int write)
 	     (unsigned long long)OCFS2_I(inode)->ip_blkno,
 	     write ? "EXMODE" : "PRMODE");
 
-	if (ocfs2_mount_local(osb))
+	if (ocfs2_mount_local(osb)) {
+		mlog_exit(0);
 		return 0;
+	}
 
 	lockres = &OCFS2_I(inode)->ip_rw_lockres;
 
@@ -2127,7 +2158,7 @@ static int ocfs2_inode_lock_update(struct inode *inode,
 
 	/* This will discard any caching information we might have had
 	 * for the inode metadata. */
-	ocfs2_metadata_cache_purge(inode);
+	ocfs2_metadata_cache_purge(INODE_CACHE(inode));
 
 	ocfs2_extent_map_trunc(inode, 0);
 
@@ -3009,6 +3040,7 @@ static void ocfs2_unlock_ast(void *opaque, int error)
 		     "unlock_action %d\n", error, lockres->l_name,
 		     lockres->l_unlock_action);
 		spin_unlock_irqrestore(&lockres->l_lock, flags);
+		mlog_exit_void();
 		return;
 	}
 
@@ -3495,11 +3527,11 @@ out:
 	return UNBLOCK_CONTINUE;
 }
 
-static int ocfs2_check_meta_downconvert(struct ocfs2_lock_res *lockres,
-					int new_level)
+static int ocfs2_ci_checkpointed(struct ocfs2_caching_info *ci,
+				 struct ocfs2_lock_res *lockres,
+				 int new_level)
 {
-	struct inode *inode = ocfs2_lock_res_inode(lockres);
-	int checkpointed = ocfs2_inode_fully_checkpointed(inode);
+	int checkpointed = ocfs2_ci_fully_checkpointed(ci);
 
 	BUG_ON(new_level != DLM_LOCK_NL && new_level != DLM_LOCK_PR);
 	BUG_ON(lockres->l_level != DLM_LOCK_EX && !checkpointed);
@@ -3507,10 +3539,18 @@ static int ocfs2_check_meta_downconvert(struct ocfs2_lock_res *lockres,
 	if (checkpointed)
 		return 1;
 
-	ocfs2_start_checkpoint(OCFS2_SB(inode->i_sb));
+	ocfs2_start_checkpoint(OCFS2_SB(ocfs2_metadata_cache_get_super(ci)));
 	return 0;
 }
 
+static int ocfs2_check_meta_downconvert(struct ocfs2_lock_res *lockres,
+					int new_level)
+{
+	struct inode *inode = ocfs2_lock_res_inode(lockres);
+
+	return ocfs2_ci_checkpointed(INODE_CACHE(inode), lockres, new_level);
+}
+
 static void ocfs2_set_meta_lvb(struct ocfs2_lock_res *lockres)
 {
 	struct inode *inode = ocfs2_lock_res_inode(lockres);
@@ -3640,6 +3680,26 @@ static int ocfs2_dentry_convert_worker(struct ocfs2_lock_res *lockres,
 	return UNBLOCK_CONTINUE_POST;
 }
 
+static int ocfs2_check_refcount_downconvert(struct ocfs2_lock_res *lockres,
+					    int new_level)
+{
+	struct ocfs2_refcount_tree *tree =
+				ocfs2_lock_res_refcount_tree(lockres);
+
+	return ocfs2_ci_checkpointed(&tree->rf_ci, lockres, new_level);
+}
+
+static int ocfs2_refcount_convert_worker(struct ocfs2_lock_res *lockres,
+					 int blocking)
+{
+	struct ocfs2_refcount_tree *tree =
+				ocfs2_lock_res_refcount_tree(lockres);
+
+	ocfs2_metadata_cache_purge(&tree->rf_ci);
+
+	return UNBLOCK_CONTINUE;
+}
+
 static void ocfs2_set_qinfo_lvb(struct ocfs2_lock_res *lockres)
 {
 	struct ocfs2_qinfo_lvb *lvb;
@@ -3752,6 +3812,37 @@ bail:
 	return status;
 }
 
+int ocfs2_refcount_lock(struct ocfs2_refcount_tree *ref_tree, int ex)
+{
+	int status;
+	int level = ex ? DLM_LOCK_EX : DLM_LOCK_PR;
+	struct ocfs2_lock_res *lockres = &ref_tree->rf_lockres;
+	struct ocfs2_super *osb = lockres->l_priv;
+
+
+	if (ocfs2_is_hard_readonly(osb))
+		return -EROFS;
+
+	if (ocfs2_mount_local(osb))
+		return 0;
+
+	status = ocfs2_cluster_lock(osb, lockres, level, 0, 0);
+	if (status < 0)
+		mlog_errno(status);
+
+	return status;
+}
+
+void ocfs2_refcount_unlock(struct ocfs2_refcount_tree *ref_tree, int ex)
+{
+	int level = ex ? DLM_LOCK_EX : DLM_LOCK_PR;
+	struct ocfs2_lock_res *lockres = &ref_tree->rf_lockres;
+	struct ocfs2_super *osb = lockres->l_priv;
+
+	if (!ocfs2_mount_local(osb))
+		ocfs2_cluster_unlock(osb, lockres, level);
+}
+
 /*
  * This is the filesystem locking protocol.  It provides the lock handling
  * hooks for the underlying DLM.  It has a maximum version number.
diff --git a/fs/ocfs2/dlmglue.h b/fs/ocfs2/dlmglue.h
index 7553836931de..d1ce48e1b3d6 100644
--- a/fs/ocfs2/dlmglue.h
+++ b/fs/ocfs2/dlmglue.h
@@ -101,6 +101,9 @@ void ocfs2_file_lock_res_init(struct ocfs2_lock_res *lockres,
 struct ocfs2_mem_dqinfo;
 void ocfs2_qinfo_lock_res_init(struct ocfs2_lock_res *lockres,
                                struct ocfs2_mem_dqinfo *info);
+void ocfs2_refcount_lock_res_init(struct ocfs2_lock_res *lockres,
+				  struct ocfs2_super *osb, u64 ref_blkno,
+				  unsigned int generation);
 void ocfs2_lock_res_free(struct ocfs2_lock_res *res);
 int ocfs2_create_new_inode_locks(struct inode *inode);
 int ocfs2_drop_inode_locks(struct inode *inode);
@@ -148,6 +151,9 @@ int ocfs2_file_lock(struct file *file, int ex, int trylock);
 void ocfs2_file_unlock(struct file *file);
 int ocfs2_qinfo_lock(struct ocfs2_mem_dqinfo *oinfo, int ex);
 void ocfs2_qinfo_unlock(struct ocfs2_mem_dqinfo *oinfo, int ex);
+struct ocfs2_refcount_tree;
+int ocfs2_refcount_lock(struct ocfs2_refcount_tree *ref_tree, int ex);
+void ocfs2_refcount_unlock(struct ocfs2_refcount_tree *ref_tree, int ex);
 
 
 void ocfs2_mark_lockres_freeing(struct ocfs2_lock_res *lockres);
diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index f2bb1a04d253..843db64e9d4a 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -293,7 +293,7 @@ static int ocfs2_last_eb_is_empty(struct inode *inode,
 	struct ocfs2_extent_block *eb;
 	struct ocfs2_extent_list *el;
 
-	ret = ocfs2_read_extent_block(inode, last_eb_blk, &eb_bh);
+	ret = ocfs2_read_extent_block(INODE_CACHE(inode), last_eb_blk, &eb_bh);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
@@ -353,11 +353,11 @@ static int ocfs2_search_for_hole_index(struct ocfs2_extent_list *el,
  * eb_bh is NULL. Otherwise, eb_bh should point to the extent block
  * containing el.
  */
-static int ocfs2_figure_hole_clusters(struct inode *inode,
-				      struct ocfs2_extent_list *el,
-				      struct buffer_head *eb_bh,
-				      u32 v_cluster,
-				      u32 *num_clusters)
+int ocfs2_figure_hole_clusters(struct ocfs2_caching_info *ci,
+			       struct ocfs2_extent_list *el,
+			       struct buffer_head *eb_bh,
+			       u32 v_cluster,
+			       u32 *num_clusters)
 {
 	int ret, i;
 	struct buffer_head *next_eb_bh = NULL;
@@ -375,7 +375,7 @@ static int ocfs2_figure_hole_clusters(struct inode *inode,
 		if (le64_to_cpu(eb->h_next_leaf_blk) == 0ULL)
 			goto no_more_extents;
 
-		ret = ocfs2_read_extent_block(inode,
+		ret = ocfs2_read_extent_block(ci,
 					      le64_to_cpu(eb->h_next_leaf_blk),
 					      &next_eb_bh);
 		if (ret) {
@@ -428,7 +428,8 @@ static int ocfs2_get_clusters_nocache(struct inode *inode,
 	tree_height = le16_to_cpu(el->l_tree_depth);
 
 	if (tree_height > 0) {
-		ret = ocfs2_find_leaf(inode, el, v_cluster, &eb_bh);
+		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, v_cluster,
+				      &eb_bh);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -455,7 +456,8 @@ static int ocfs2_get_clusters_nocache(struct inode *inode,
 		 * field.
 		 */
 		if (hole_len) {
-			ret = ocfs2_figure_hole_clusters(inode, el, eb_bh,
+			ret = ocfs2_figure_hole_clusters(INODE_CACHE(inode),
+							 el, eb_bh,
 							 v_cluster, &len);
 			if (ret) {
 				mlog_errno(ret);
@@ -539,7 +541,8 @@ static void ocfs2_relative_extent_offsets(struct super_block *sb,
 
 int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster,
 			     u32 *p_cluster, u32 *num_clusters,
-			     struct ocfs2_extent_list *el)
+			     struct ocfs2_extent_list *el,
+			     unsigned int *extent_flags)
 {
 	int ret = 0, i;
 	struct buffer_head *eb_bh = NULL;
@@ -548,7 +551,8 @@ int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster,
 	u32 coff;
 
 	if (el->l_tree_depth) {
-		ret = ocfs2_find_leaf(inode, el, v_cluster, &eb_bh);
+		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, v_cluster,
+				      &eb_bh);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -590,6 +594,9 @@ int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster,
 		*p_cluster = *p_cluster + coff;
 		if (num_clusters)
 			*num_clusters = ocfs2_rec_clusters(el, rec) - coff;
+
+		if (extent_flags)
+			*extent_flags = rec->e_flags;
 	}
 out:
 	if (eb_bh)
@@ -862,8 +869,8 @@ int ocfs2_read_virt_blocks(struct inode *inode, u64 v_block, int nr,
 			BUG_ON(bhs[done + i]->b_blocknr != (p_block + i));
 		}
 
-		rc = ocfs2_read_blocks(inode, p_block, count, bhs + done,
-				       flags, validate);
+		rc = ocfs2_read_blocks(INODE_CACHE(inode), p_block, count,
+				       bhs + done, flags, validate);
 		if (rc) {
 			mlog_errno(rc);
 			break;
diff --git a/fs/ocfs2/extent_map.h b/fs/ocfs2/extent_map.h
index b7dd9731b462..e79d41c2c909 100644
--- a/fs/ocfs2/extent_map.h
+++ b/fs/ocfs2/extent_map.h
@@ -55,12 +55,18 @@ int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 
 int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster,
 			     u32 *p_cluster, u32 *num_clusters,
-			     struct ocfs2_extent_list *el);
+			     struct ocfs2_extent_list *el,
+			     unsigned int *extent_flags);
 
 int ocfs2_read_virt_blocks(struct inode *inode, u64 v_block, int nr,
 			   struct buffer_head *bhs[], int flags,
 			   int (*validate)(struct super_block *sb,
 					   struct buffer_head *bh));
+int ocfs2_figure_hole_clusters(struct ocfs2_caching_info *ci,
+			       struct ocfs2_extent_list *el,
+			       struct buffer_head *eb_bh,
+			       u32 v_cluster,
+			       u32 *num_clusters);
 static inline int ocfs2_read_virt_block(struct inode *inode, u64 v_block,
 					struct buffer_head **bh,
 					int (*validate)(struct super_block *sb,
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 221c5e98957b..89fc8ee1f5a5 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -59,6 +59,7 @@
 #include "xattr.h"
 #include "acl.h"
 #include "quota.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
@@ -259,7 +260,7 @@ int ocfs2_update_inode_atime(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -334,6 +335,39 @@ out:
 	return ret;
 }
 
+static int ocfs2_cow_file_pos(struct inode *inode,
+			      struct buffer_head *fe_bh,
+			      u64 offset)
+{
+	int status;
+	u32 phys, cpos = offset >> OCFS2_SB(inode->i_sb)->s_clustersize_bits;
+	unsigned int num_clusters = 0;
+	unsigned int ext_flags = 0;
+
+	/*
+	 * If the new offset is aligned to the range of the cluster, there is
+	 * no space for ocfs2_zero_range_for_truncate to fill, so no need to
+	 * CoW either.
+	 */
+	if ((offset & (OCFS2_SB(inode->i_sb)->s_clustersize - 1)) == 0)
+		return 0;
+
+	status = ocfs2_get_clusters(inode, cpos, &phys,
+				    &num_clusters, &ext_flags);
+	if (status) {
+		mlog_errno(status);
+		goto out;
+	}
+
+	if (!(ext_flags & OCFS2_EXT_REFCOUNTED))
+		goto out;
+
+	return ocfs2_refcount_cow(inode, fe_bh, cpos, 1, cpos+1);
+
+out:
+	return status;
+}
+
 static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 				     struct inode *inode,
 				     struct buffer_head *fe_bh,
@@ -346,6 +380,17 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 
 	mlog_entry_void();
 
+	/*
+	 * We need to CoW the cluster contains the offset if it is reflinked
+	 * since we will call ocfs2_zero_range_for_truncate later which will
+	 * write "0" from offset to the end of the cluster.
+	 */
+	status = ocfs2_cow_file_pos(inode, fe_bh, new_i_size);
+	if (status) {
+		mlog_errno(status);
+		return status;
+	}
+
 	/* TODO: This needs to actually orphan the inode in this
 	 * transaction. */
 
@@ -356,7 +401,7 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 		goto out;
 	}
 
-	status = ocfs2_journal_access_di(handle, inode, fe_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -486,6 +531,8 @@ bail_unlock_sem:
 	up_write(&OCFS2_I(inode)->ip_alloc_sem);
 
 bail:
+	if (!status && OCFS2_I(inode)->ip_clusters == 0)
+		status = ocfs2_try_remove_refcount_tree(inode, di_bh);
 
 	mlog_exit(status);
 	return status;
@@ -515,11 +562,10 @@ int ocfs2_add_inode_data(struct ocfs2_super *osb,
 	int ret;
 	struct ocfs2_extent_tree et;
 
-	ocfs2_init_dinode_extent_tree(&et, inode, fe_bh);
-	ret = ocfs2_add_clusters_in_btree(osb, inode, logical_offset,
-					   clusters_to_add, mark_unwritten,
-					   &et, handle,
-					   data_ac, meta_ac, reason_ret);
+	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), fe_bh);
+	ret = ocfs2_add_clusters_in_btree(handle, &et, logical_offset,
+					  clusters_to_add, mark_unwritten,
+					  data_ac, meta_ac, reason_ret);
 
 	return ret;
 }
@@ -564,7 +610,7 @@ restart_all:
 	     (unsigned long long)OCFS2_I(inode)->ip_blkno,
 	     (long long)i_size_read(inode), le32_to_cpu(fe->i_clusters),
 	     clusters_to_add);
-	ocfs2_init_dinode_extent_tree(&et, inode, bh);
+	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), bh);
 	status = ocfs2_lock_allocators(inode, &et, clusters_to_add, 0,
 				       &data_ac, &meta_ac);
 	if (status) {
@@ -593,7 +639,7 @@ restarted_transaction:
 	/* reserve a write to the file entry early on - that we if we
 	 * run out of credits in the allocation path, we can still
 	 * update i_size. */
-	status = ocfs2_journal_access_di(handle, inode, bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1131,7 +1177,7 @@ static int __ocfs2_write_remove_suid(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -1395,7 +1441,7 @@ static int ocfs2_remove_inode_range(struct inode *inode,
 	struct address_space *mapping = inode->i_mapping;
 	struct ocfs2_extent_tree et;
 
-	ocfs2_init_dinode_extent_tree(&et, inode, di_bh);
+	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), di_bh);
 	ocfs2_init_dealloc_ctxt(&dealloc);
 
 	if (byte_len == 0)
@@ -1657,6 +1703,70 @@ static long ocfs2_fallocate(struct inode *inode, int mode, loff_t offset,
 					 OCFS2_IOC_RESVSP64, &sr, change_size);
 }
 
+int ocfs2_check_range_for_refcount(struct inode *inode, loff_t pos,
+				   size_t count)
+{
+	int ret = 0;
+	unsigned int extent_flags;
+	u32 cpos, clusters, extent_len, phys_cpos;
+	struct super_block *sb = inode->i_sb;
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)) ||
+	    !(OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL))
+		return 0;
+
+	cpos = pos >> OCFS2_SB(sb)->s_clustersize_bits;
+	clusters = ocfs2_clusters_for_bytes(sb, pos + count) - cpos;
+
+	while (clusters) {
+		ret = ocfs2_get_clusters(inode, cpos, &phys_cpos, &extent_len,
+					 &extent_flags);
+		if (ret < 0) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		if (phys_cpos && (extent_flags & OCFS2_EXT_REFCOUNTED)) {
+			ret = 1;
+			break;
+		}
+
+		if (extent_len > clusters)
+			extent_len = clusters;
+
+		clusters -= extent_len;
+		cpos += extent_len;
+	}
+out:
+	return ret;
+}
+
+static int ocfs2_prepare_inode_for_refcount(struct inode *inode,
+					    loff_t pos, size_t count,
+					    int *meta_level)
+{
+	int ret;
+	struct buffer_head *di_bh = NULL;
+	u32 cpos = pos >> OCFS2_SB(inode->i_sb)->s_clustersize_bits;
+	u32 clusters =
+		ocfs2_clusters_for_bytes(inode->i_sb, pos + count) - cpos;
+
+	ret = ocfs2_inode_lock(inode, &di_bh, 1);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	*meta_level = 1;
+
+	ret = ocfs2_refcount_cow(inode, di_bh, cpos, clusters, UINT_MAX);
+	if (ret)
+		mlog_errno(ret);
+out:
+	brelse(di_bh);
+	return ret;
+}
+
 static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 					 loff_t *ppos,
 					 size_t count,
@@ -1713,6 +1823,22 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 
 		end = saved_pos + count;
 
+		ret = ocfs2_check_range_for_refcount(inode, saved_pos, count);
+		if (ret == 1) {
+			ocfs2_inode_unlock(inode, meta_level);
+			meta_level = -1;
+
+			ret = ocfs2_prepare_inode_for_refcount(inode,
+							       saved_pos,
+							       count,
+							       &meta_level);
+		}
+
+		if (ret < 0) {
+			mlog_errno(ret);
+			goto out_unlock;
+		}
+
 		/*
 		 * Skip the O_DIRECT checks if we don't need
 		 * them.
@@ -1759,7 +1885,8 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 		*ppos = saved_pos;
 
 out_unlock:
-	ocfs2_inode_unlock(inode, meta_level);
+	if (meta_level >= 0)
+		ocfs2_inode_unlock(inode, meta_level);
 
 out:
 	return ret;
diff --git a/fs/ocfs2/file.h b/fs/ocfs2/file.h
index 172f9fbc9fc7..d66cf4f7c70e 100644
--- a/fs/ocfs2/file.h
+++ b/fs/ocfs2/file.h
@@ -69,4 +69,6 @@ int ocfs2_update_inode_atime(struct inode *inode,
 int ocfs2_change_file_space(struct file *file, unsigned int cmd,
 			    struct ocfs2_space_resv *sr);
 
+int ocfs2_check_range_for_refcount(struct inode *inode, loff_t pos,
+				   size_t count);
 #endif /* OCFS2_FILE_H */
diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
index 4dc8890ba316..0297fb8982b8 100644
--- a/fs/ocfs2/inode.c
+++ b/fs/ocfs2/inode.c
@@ -53,6 +53,7 @@
 #include "sysfile.h"
 #include "uptodate.h"
 #include "xattr.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
@@ -562,7 +563,8 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 			goto out;
 		}
 
-		status = ocfs2_journal_access_di(handle, inode, fe_bh,
+		status = ocfs2_journal_access_di(handle, INODE_CACHE(inode),
+						 fe_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -646,7 +648,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 	}
 
 	/* set the inodes dtime */
-	status = ocfs2_journal_access_di(handle, inode, di_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -662,7 +664,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 		goto bail_commit;
 	}
 
-	ocfs2_remove_from_cache(inode, di_bh);
+	ocfs2_remove_from_cache(INODE_CACHE(inode), di_bh);
 	vfs_dq_free_inode(inode);
 
 	status = ocfs2_free_dinode(handle, inode_alloc_inode,
@@ -781,6 +783,12 @@ static int ocfs2_wipe_inode(struct inode *inode,
 		goto bail_unlock_dir;
 	}
 
+	status = ocfs2_remove_refcount_tree(inode, di_bh);
+	if (status < 0) {
+		mlog_errno(status);
+		goto bail_unlock_dir;
+	}
+
 	status = ocfs2_remove_inode(inode, di_bh, orphan_dir_inode,
 				    orphan_dir_bh);
 	if (status < 0)
@@ -1112,13 +1120,14 @@ void ocfs2_clear_inode(struct inode *inode)
 	ocfs2_lock_res_free(&oi->ip_inode_lockres);
 	ocfs2_lock_res_free(&oi->ip_open_lockres);
 
-	ocfs2_metadata_cache_purge(inode);
+	ocfs2_metadata_cache_exit(INODE_CACHE(inode));
 
-	mlog_bug_on_msg(oi->ip_metadata_cache.ci_num_cached,
+	mlog_bug_on_msg(INODE_CACHE(inode)->ci_num_cached,
 			"Clear inode of %llu, inode has %u cache items\n",
-			(unsigned long long)oi->ip_blkno, oi->ip_metadata_cache.ci_num_cached);
+			(unsigned long long)oi->ip_blkno,
+			INODE_CACHE(inode)->ci_num_cached);
 
-	mlog_bug_on_msg(!(oi->ip_flags & OCFS2_INODE_CACHE_INLINE),
+	mlog_bug_on_msg(!(INODE_CACHE(inode)->ci_flags & OCFS2_CACHE_FL_INLINE),
 			"Clear inode of %llu, inode has a bad flag\n",
 			(unsigned long long)oi->ip_blkno);
 
@@ -1145,9 +1154,7 @@ void ocfs2_clear_inode(struct inode *inode)
 			(unsigned long long)oi->ip_blkno, oi->ip_open_count);
 
 	/* Clear all other flags. */
-	oi->ip_flags = OCFS2_INODE_CACHE_INLINE;
-	oi->ip_created_trans = 0;
-	oi->ip_last_trans = 0;
+	oi->ip_flags = 0;
 	oi->ip_dir_start_lookup = 0;
 	oi->ip_blkno = 0ULL;
 
@@ -1239,7 +1246,7 @@ int ocfs2_mark_inode_dirty(handle_t *handle,
 	mlog_entry("(inode %llu)\n",
 		   (unsigned long long)OCFS2_I(inode)->ip_blkno);
 
-	status = ocfs2_journal_access_di(handle, inode, bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1380,8 +1387,8 @@ int ocfs2_read_inode_block_full(struct inode *inode, struct buffer_head **bh,
 	int rc;
 	struct buffer_head *tmp = *bh;
 
-	rc = ocfs2_read_blocks(inode, OCFS2_I(inode)->ip_blkno, 1, &tmp,
-			       flags, ocfs2_validate_inode_block);
+	rc = ocfs2_read_blocks(INODE_CACHE(inode), OCFS2_I(inode)->ip_blkno,
+			       1, &tmp, flags, ocfs2_validate_inode_block);
 
 	/* If ocfs2_read_blocks() got us a new bh, pass it up. */
 	if (!rc && !*bh)
@@ -1394,3 +1401,56 @@ int ocfs2_read_inode_block(struct inode *inode, struct buffer_head **bh)
 {
 	return ocfs2_read_inode_block_full(inode, bh, 0);
 }
+
+
+static u64 ocfs2_inode_cache_owner(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	return oi->ip_blkno;
+}
+
+static struct super_block *ocfs2_inode_cache_get_super(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	return oi->vfs_inode.i_sb;
+}
+
+static void ocfs2_inode_cache_lock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	spin_lock(&oi->ip_lock);
+}
+
+static void ocfs2_inode_cache_unlock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	spin_unlock(&oi->ip_lock);
+}
+
+static void ocfs2_inode_cache_io_lock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	mutex_lock(&oi->ip_io_mutex);
+}
+
+static void ocfs2_inode_cache_io_unlock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_inode_info *oi = cache_info_to_inode(ci);
+
+	mutex_unlock(&oi->ip_io_mutex);
+}
+
+const struct ocfs2_caching_operations ocfs2_inode_caching_ops = {
+	.co_owner		= ocfs2_inode_cache_owner,
+	.co_get_super		= ocfs2_inode_cache_get_super,
+	.co_cache_lock		= ocfs2_inode_cache_lock,
+	.co_cache_unlock	= ocfs2_inode_cache_unlock,
+	.co_io_lock		= ocfs2_inode_cache_io_lock,
+	.co_io_unlock		= ocfs2_inode_cache_io_unlock,
+};
+
diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h
index ea71525aad41..ba4fe07b293c 100644
--- a/fs/ocfs2/inode.h
+++ b/fs/ocfs2/inode.h
@@ -60,12 +60,6 @@ struct ocfs2_inode_info
 
 	u32				ip_dir_start_lookup;
 
-	/* next two are protected by trans_inc_lock */
-	/* which transaction were we created on? Zero if none. */
-	unsigned long			ip_created_trans;
-	/* last transaction we were a part of. */
-	unsigned long			ip_last_trans;
-
 	struct ocfs2_caching_info	ip_metadata_cache;
 
 	struct ocfs2_extent_map		ip_extent_map;
@@ -106,8 +100,6 @@ struct ocfs2_inode_info
 #define OCFS2_INODE_MAYBE_ORPHANED	0x00000020
 /* Does someone have the file open O_DIRECT */
 #define OCFS2_INODE_OPEN_DIRECT		0x00000040
-/* Indicates that the metadata cache should be used as an array. */
-#define OCFS2_INODE_CACHE_INLINE	0x00000080
 
 static inline struct ocfs2_inode_info *OCFS2_I(struct inode *inode)
 {
@@ -120,6 +112,12 @@ static inline struct ocfs2_inode_info *OCFS2_I(struct inode *inode)
 extern struct kmem_cache *ocfs2_inode_cache;
 
 extern const struct address_space_operations ocfs2_aops;
+extern const struct ocfs2_caching_operations ocfs2_inode_caching_ops;
+
+static inline struct ocfs2_caching_info *INODE_CACHE(struct inode *inode)
+{
+	return &OCFS2_I(inode)->ip_metadata_cache;
+}
 
 void ocfs2_clear_inode(struct inode *inode);
 void ocfs2_delete_inode(struct inode *inode);
@@ -172,4 +170,10 @@ int ocfs2_read_inode_block(struct inode *inode, struct buffer_head **bh);
 /* The same, but can be passed OCFS2_BH_* flags */
 int ocfs2_read_inode_block_full(struct inode *inode, struct buffer_head **bh,
 				int flags);
+
+static inline struct ocfs2_inode_info *cache_info_to_inode(struct ocfs2_caching_info *ci)
+{
+	return container_of(ci, struct ocfs2_inode_info, ip_metadata_cache);
+}
+
 #endif /* OCFS2_INODE_H */
diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 467b413bec21..31fbb0619510 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -21,6 +21,7 @@
 #include "ocfs2_fs.h"
 #include "ioctl.h"
 #include "resize.h"
+#include "refcounttree.h"
 
 #include <linux/ext2_fs.h>
 
@@ -115,6 +116,9 @@ long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 	int status;
 	struct ocfs2_space_resv sr;
 	struct ocfs2_new_group_input input;
+	struct reflink_arguments args;
+	const char *old_path, *new_path;
+	bool preserve;
 
 	switch (cmd) {
 	case OCFS2_IOC_GETFLAGS:
@@ -160,6 +164,15 @@ long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 			return -EFAULT;
 
 		return ocfs2_group_add(inode, &input);
+	case OCFS2_IOC_REFLINK:
+		if (copy_from_user(&args, (struct reflink_arguments *)arg,
+				   sizeof(args)))
+			return -EFAULT;
+		old_path = (const char *)(unsigned long)args.old_path;
+		new_path = (const char *)(unsigned long)args.new_path;
+		preserve = (args.preserve != 0);
+
+		return ocfs2_reflink_ioctl(inode, old_path, new_path, preserve);
 	default:
 		return -ENOTTY;
 	}
@@ -182,6 +195,7 @@ long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg)
 	case OCFS2_IOC_GROUP_EXTEND:
 	case OCFS2_IOC_GROUP_ADD:
 	case OCFS2_IOC_GROUP_ADD64:
+	case OCFS2_IOC_REFLINK:
 		break;
 	default:
 		return -ENOIOCTLCMD;
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index c48b93ac6b65..54c16b66327e 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -48,6 +48,7 @@
 #include "slot_map.h"
 #include "super.h"
 #include "sysfile.h"
+#include "uptodate.h"
 #include "quota.h"
 
 #include "buffer_head_io.h"
@@ -554,6 +555,14 @@ static struct ocfs2_triggers eb_triggers = {
 	.ot_offset	= offsetof(struct ocfs2_extent_block, h_check),
 };
 
+static struct ocfs2_triggers rb_triggers = {
+	.ot_triggers = {
+		.t_commit = ocfs2_commit_trigger,
+		.t_abort = ocfs2_abort_trigger,
+	},
+	.ot_offset	= offsetof(struct ocfs2_refcount_block, rf_check),
+};
+
 static struct ocfs2_triggers gd_triggers = {
 	.ot_triggers = {
 		.t_commit = ocfs2_commit_trigger,
@@ -601,14 +610,16 @@ static struct ocfs2_triggers dl_triggers = {
 };
 
 static int __ocfs2_journal_access(handle_t *handle,
-				  struct inode *inode,
+				  struct ocfs2_caching_info *ci,
 				  struct buffer_head *bh,
 				  struct ocfs2_triggers *triggers,
 				  int type)
 {
 	int status;
+	struct ocfs2_super *osb =
+		OCFS2_SB(ocfs2_metadata_cache_get_super(ci));
 
-	BUG_ON(!inode);
+	BUG_ON(!ci || !ci->ci_ops);
 	BUG_ON(!handle);
 	BUG_ON(!bh);
 
@@ -627,15 +638,15 @@ static int __ocfs2_journal_access(handle_t *handle,
 		BUG();
 	}
 
-	/* Set the current transaction information on the inode so
+	/* Set the current transaction information on the ci so
 	 * that the locking code knows whether it can drop it's locks
-	 * on this inode or not. We're protected from the commit
+	 * on this ci or not. We're protected from the commit
 	 * thread updating the current transaction id until
 	 * ocfs2_commit_trans() because ocfs2_start_trans() took
 	 * j_trans_barrier for us. */
-	ocfs2_set_inode_lock_trans(OCFS2_SB(inode->i_sb)->journal, inode);
+	ocfs2_set_ci_lock_trans(osb->journal, ci);
 
-	mutex_lock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_lock(ci);
 	switch (type) {
 	case OCFS2_JOURNAL_ACCESS_CREATE:
 	case OCFS2_JOURNAL_ACCESS_WRITE:
@@ -650,9 +661,9 @@ static int __ocfs2_journal_access(handle_t *handle,
 		status = -EINVAL;
 		mlog(ML_ERROR, "Uknown access type!\n");
 	}
-	if (!status && ocfs2_meta_ecc(OCFS2_SB(inode->i_sb)) && triggers)
+	if (!status && ocfs2_meta_ecc(osb) && triggers)
 		jbd2_journal_set_triggers(bh, &triggers->ot_triggers);
-	mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
+	ocfs2_metadata_cache_io_unlock(ci);
 
 	if (status < 0)
 		mlog(ML_ERROR, "Error %d getting %d access to buffer!\n",
@@ -662,66 +673,65 @@ static int __ocfs2_journal_access(handle_t *handle,
 	return status;
 }
 
-int ocfs2_journal_access_di(handle_t *handle, struct inode *inode,
-			       struct buffer_head *bh, int type)
+int ocfs2_journal_access_di(handle_t *handle, struct ocfs2_caching_info *ci,
+			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &di_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &di_triggers, type);
 }
 
-int ocfs2_journal_access_eb(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_eb(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &eb_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &eb_triggers, type);
 }
 
-int ocfs2_journal_access_gd(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_rb(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &gd_triggers,
+	return __ocfs2_journal_access(handle, ci, bh, &rb_triggers,
 				      type);
 }
 
-int ocfs2_journal_access_db(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_gd(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &db_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &gd_triggers, type);
 }
 
-int ocfs2_journal_access_xb(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_db(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &xb_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &db_triggers, type);
 }
 
-int ocfs2_journal_access_dq(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_xb(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &dq_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &xb_triggers, type);
 }
 
-int ocfs2_journal_access_dr(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_dq(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &dr_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &dq_triggers, type);
 }
 
-int ocfs2_journal_access_dl(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_dr(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, &dl_triggers,
-				      type);
+	return __ocfs2_journal_access(handle, ci, bh, &dr_triggers, type);
+}
+
+int ocfs2_journal_access_dl(handle_t *handle, struct ocfs2_caching_info *ci,
+			    struct buffer_head *bh, int type)
+{
+	return __ocfs2_journal_access(handle, ci, bh, &dl_triggers, type);
 }
 
-int ocfs2_journal_access(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access(handle_t *handle, struct ocfs2_caching_info *ci,
 			 struct buffer_head *bh, int type)
 {
-	return __ocfs2_journal_access(handle, inode, bh, NULL, type);
+	return __ocfs2_journal_access(handle, ci, bh, NULL, type);
 }
 
 int ocfs2_journal_dirty(handle_t *handle,
@@ -898,7 +908,7 @@ static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
 		ocfs2_bump_recovery_generation(fe);
 
 	ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &fe->i_check);
-	status = ocfs2_write_block(osb, bh, journal->j_inode);
+	status = ocfs2_write_block(osb, bh, INODE_CACHE(journal->j_inode));
 	if (status < 0)
 		mlog_errno(status);
 
@@ -1642,7 +1652,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
 					ocfs2_get_recovery_generation(fe);
 
 	ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &fe->i_check);
-	status = ocfs2_write_block(osb, bh, inode);
+	status = ocfs2_write_block(osb, bh, INODE_CACHE(inode));
 	if (status < 0)
 		mlog_errno(status);
 
diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h
index 2c3222aec622..3f74e09b0d80 100644
--- a/fs/ocfs2/journal.h
+++ b/fs/ocfs2/journal.h
@@ -90,56 +90,66 @@ static inline unsigned long ocfs2_inc_trans_id(struct ocfs2_journal *j)
 	return old_id;
 }
 
-static inline void ocfs2_set_inode_lock_trans(struct ocfs2_journal *journal,
-					      struct inode *inode)
+static inline void ocfs2_set_ci_lock_trans(struct ocfs2_journal *journal,
+					   struct ocfs2_caching_info *ci)
 {
 	spin_lock(&trans_inc_lock);
-	OCFS2_I(inode)->ip_last_trans = journal->j_trans_id;
+	ci->ci_last_trans = journal->j_trans_id;
 	spin_unlock(&trans_inc_lock);
 }
 
 /* Used to figure out whether it's safe to drop a metadata lock on an
- * inode. Returns true if all the inodes changes have been
+ * cached object. Returns true if all the object's changes have been
  * checkpointed to disk. You should be holding the spinlock on the
  * metadata lock while calling this to be sure that nobody can take
  * the lock and put it on another transaction. */
-static inline int ocfs2_inode_fully_checkpointed(struct inode *inode)
+static inline int ocfs2_ci_fully_checkpointed(struct ocfs2_caching_info *ci)
 {
 	int ret;
-	struct ocfs2_journal *journal = OCFS2_SB(inode->i_sb)->journal;
+	struct ocfs2_journal *journal =
+		OCFS2_SB(ocfs2_metadata_cache_get_super(ci))->journal;
 
 	spin_lock(&trans_inc_lock);
-	ret = time_after(journal->j_trans_id, OCFS2_I(inode)->ip_last_trans);
+	ret = time_after(journal->j_trans_id, ci->ci_last_trans);
 	spin_unlock(&trans_inc_lock);
 	return ret;
 }
 
-/* convenience function to check if an inode is still new (has never
- * hit disk) Will do you a favor and set created_trans = 0 when you've
- * been checkpointed.  returns '1' if the inode is still new. */
-static inline int ocfs2_inode_is_new(struct inode *inode)
+/* convenience function to check if an object backed by struct
+ * ocfs2_caching_info  is still new (has never hit disk) Will do you a
+ * favor and set created_trans = 0 when you've
+ * been checkpointed.  returns '1' if the ci is still new. */
+static inline int ocfs2_ci_is_new(struct ocfs2_caching_info *ci)
 {
 	int ret;
+	struct ocfs2_journal *journal =
+		OCFS2_SB(ocfs2_metadata_cache_get_super(ci))->journal;
 
+	spin_lock(&trans_inc_lock);
+	ret = !(time_after(journal->j_trans_id, ci->ci_created_trans));
+	if (!ret)
+		ci->ci_created_trans = 0;
+	spin_unlock(&trans_inc_lock);
+	return ret;
+}
+
+/* Wrapper for inodes so we can check system files */
+static inline int ocfs2_inode_is_new(struct inode *inode)
+{
 	/* System files are never "new" as they're written out by
 	 * mkfs. This helps us early during mount, before we have the
 	 * journal open and j_trans_id could be junk. */
 	if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SYSTEM_FILE)
 		return 0;
-	spin_lock(&trans_inc_lock);
-	ret = !(time_after(OCFS2_SB(inode->i_sb)->journal->j_trans_id,
-			   OCFS2_I(inode)->ip_created_trans));
-	if (!ret)
-		OCFS2_I(inode)->ip_created_trans = 0;
-	spin_unlock(&trans_inc_lock);
-	return ret;
+
+	return ocfs2_ci_is_new(INODE_CACHE(inode));
 }
 
-static inline void ocfs2_inode_set_new(struct ocfs2_super *osb,
-				       struct inode *inode)
+static inline void ocfs2_ci_set_new(struct ocfs2_super *osb,
+				    struct ocfs2_caching_info *ci)
 {
 	spin_lock(&trans_inc_lock);
-	OCFS2_I(inode)->ip_created_trans = osb->journal->j_trans_id;
+	ci->ci_created_trans = osb->journal->j_trans_id;
 	spin_unlock(&trans_inc_lock);
 }
 
@@ -200,7 +210,7 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode)
 	if (ocfs2_mount_local(osb))
 		return;
 
-	if (!ocfs2_inode_fully_checkpointed(inode)) {
+	if (!ocfs2_ci_fully_checkpointed(INODE_CACHE(inode))) {
 		/* WARNING: This only kicks off a single
 		 * checkpoint. If someone races you and adds more
 		 * metadata to the journal, you won't know, and will
@@ -210,7 +220,7 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode)
 		ocfs2_start_checkpoint(osb);
 
 		wait_event(osb->journal->j_checkpointed,
-			   ocfs2_inode_fully_checkpointed(inode));
+			   ocfs2_ci_fully_checkpointed(INODE_CACHE(inode)));
 	}
 }
 
@@ -266,31 +276,34 @@ int			     ocfs2_extend_trans(handle_t *handle, int nblocks);
 
 
 /* ocfs2_inode */
-int ocfs2_journal_access_di(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_di(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* ocfs2_extent_block */
-int ocfs2_journal_access_eb(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_eb(handle_t *handle, struct ocfs2_caching_info *ci,
+			    struct buffer_head *bh, int type);
+/* ocfs2_refcount_block */
+int ocfs2_journal_access_rb(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* ocfs2_group_desc */
-int ocfs2_journal_access_gd(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_gd(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* ocfs2_xattr_block */
-int ocfs2_journal_access_xb(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_xb(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* quota blocks */
-int ocfs2_journal_access_dq(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_dq(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* dirblock */
-int ocfs2_journal_access_db(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_db(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* ocfs2_dx_root_block */
-int ocfs2_journal_access_dr(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_dr(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* ocfs2_dx_leaf */
-int ocfs2_journal_access_dl(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access_dl(handle_t *handle, struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh, int type);
 /* Anything that has no ecc */
-int ocfs2_journal_access(handle_t *handle, struct inode *inode,
+int ocfs2_journal_access(handle_t *handle, struct ocfs2_caching_info *ci,
 			 struct buffer_head *bh, int type);
 
 /*
@@ -477,6 +490,23 @@ static inline int ocfs2_calc_dxi_expand_credits(struct super_block *sb)
 	return credits;
 }
 
+/* inode update, new refcount block and its allocation credits. */
+#define OCFS2_REFCOUNT_TREE_CREATE_CREDITS (OCFS2_INODE_UPDATE_CREDITS + 1 \
+					    + OCFS2_SUBALLOC_ALLOC)
+
+/* inode and the refcount block update. */
+#define OCFS2_REFCOUNT_TREE_SET_CREDITS (OCFS2_INODE_UPDATE_CREDITS + 1)
+
+/*
+ * inode and the refcount block update.
+ * It doesn't include the credits for sub alloc change.
+ * So if we need to free the bit, OCFS2_SUBALLOC_FREE needs to be added.
+ */
+#define OCFS2_REFCOUNT_TREE_REMOVE_CREDITS (OCFS2_INODE_UPDATE_CREDITS + 1)
+
+/* 2 metadata alloc, 2 new blocks and root refcount block */
+#define OCFS2_EXPAND_REFCOUNT_TREE_CREDITS (OCFS2_SUBALLOC_ALLOC * 2 + 3)
+
 /*
  * Please note that the caller must make sure that root_el is the root
  * of extent tree. So for an inode, it should be &fe->id2.i_list. Otherwise
diff --git a/fs/ocfs2/localalloc.c b/fs/ocfs2/localalloc.c
index bac7e6abaf47..ac10f83edb95 100644
--- a/fs/ocfs2/localalloc.c
+++ b/fs/ocfs2/localalloc.c
@@ -297,8 +297,8 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	}
 	memcpy(alloc_copy, alloc, bh->b_size);
 
-	status = ocfs2_journal_access_di(handle, local_alloc_inode, bh,
-					 OCFS2_JOURNAL_ACCESS_WRITE);
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(local_alloc_inode),
+					 bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out_commit;
@@ -392,7 +392,7 @@ int ocfs2_begin_local_alloc_recovery(struct ocfs2_super *osb,
 	ocfs2_clear_local_alloc(alloc);
 
 	ocfs2_compute_meta_ecc(osb->sb, alloc_bh->b_data, &alloc->i_check);
-	status = ocfs2_write_block(osb, alloc_bh, inode);
+	status = ocfs2_write_block(osb, alloc_bh, INODE_CACHE(inode));
 	if (status < 0)
 		mlog_errno(status);
 
@@ -678,7 +678,8 @@ int ocfs2_claim_local_alloc_bits(struct ocfs2_super *osb,
 	 * delete bits from it! */
 	*num_bits = bits_wanted;
 
-	status = ocfs2_journal_access_di(handle, local_alloc_inode,
+	status = ocfs2_journal_access_di(handle,
+					 INODE_CACHE(local_alloc_inode),
 					 osb->local_alloc_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
@@ -1156,7 +1157,8 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
 	}
 	memcpy(alloc_copy, alloc, osb->local_alloc_bh->b_size);
 
-	status = ocfs2_journal_access_di(handle, local_alloc_inode,
+	status = ocfs2_journal_access_di(handle,
+					 INODE_CACHE(local_alloc_inode),
 					 osb->local_alloc_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
diff --git a/fs/ocfs2/mmap.c b/fs/ocfs2/mmap.c
index b606496b72ec..39737613424a 100644
--- a/fs/ocfs2/mmap.c
+++ b/fs/ocfs2/mmap.c
@@ -202,7 +202,7 @@ out:
 	return ret;
 }
 
-static struct vm_operations_struct ocfs2_file_vm_ops = {
+static const struct vm_operations_struct ocfs2_file_vm_ops = {
 	.fault		= ocfs2_fault,
 	.page_mkwrite	= ocfs2_page_mkwrite,
 };
diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index 8601f934010b..f010b22b1c44 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -69,7 +69,6 @@
 static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 			      struct inode *dir,
 			      struct inode *inode,
-			      struct dentry *dentry,
 			      dev_t dev,
 			      struct buffer_head **new_fe_bh,
 			      struct buffer_head *parent_fe_bh,
@@ -78,7 +77,7 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 
 static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
 				    struct inode **ret_orphan_dir,
-				    struct inode *inode,
+				    u64 blkno,
 				    char *name,
 				    struct ocfs2_dir_lookup_result *lookup);
 
@@ -358,8 +357,12 @@ static int ocfs2_mknod(struct inode *dir,
 	}
 	did_quota_inode = 1;
 
+	mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry,
+		   inode->i_mode, (unsigned long)dev, dentry->d_name.len,
+		   dentry->d_name.name);
+
 	/* do the real work now. */
-	status = ocfs2_mknod_locked(osb, dir, inode, dentry, dev,
+	status = ocfs2_mknod_locked(osb, dir, inode, dev,
 				    &new_fe_bh, parent_fe_bh, handle,
 				    inode_ac);
 	if (status < 0) {
@@ -375,7 +378,8 @@ static int ocfs2_mknod(struct inode *dir,
 			goto leave;
 		}
 
-		status = ocfs2_journal_access_di(handle, dir, parent_fe_bh,
+		status = ocfs2_journal_access_di(handle, INODE_CACHE(dir),
+						 parent_fe_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -465,7 +469,6 @@ leave:
 static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 			      struct inode *dir,
 			      struct inode *inode,
-			      struct dentry *dentry,
 			      dev_t dev,
 			      struct buffer_head **new_fe_bh,
 			      struct buffer_head *parent_fe_bh,
@@ -479,10 +482,6 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 	u16 suballoc_bit;
 	u16 feat;
 
-	mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry,
-		   inode->i_mode, (unsigned long)dev, dentry->d_name.len,
-		   dentry->d_name.name);
-
 	*new_fe_bh = NULL;
 
 	status = ocfs2_claim_new_inode(osb, handle, dir, parent_fe_bh,
@@ -507,9 +506,10 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 		mlog_errno(status);
 		goto leave;
 	}
-	ocfs2_set_new_buffer_uptodate(inode, *new_fe_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), *new_fe_bh);
 
-	status = ocfs2_journal_access_di(handle, inode, *new_fe_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode),
+					 *new_fe_bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -565,7 +565,7 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 	}
 
 	ocfs2_populate_inode(inode, fe, 1);
-	ocfs2_inode_set_new(osb, inode);
+	ocfs2_ci_set_new(osb, INODE_CACHE(inode));
 	if (!ocfs2_mount_local(osb)) {
 		status = ocfs2_create_new_inode_locks(inode);
 		if (status < 0)
@@ -682,7 +682,7 @@ static int ocfs2_link(struct dentry *old_dentry,
 		goto out_unlock_inode;
 	}
 
-	err = ocfs2_journal_access_di(handle, inode, fe_bh,
+	err = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (err < 0) {
 		mlog_errno(err);
@@ -850,7 +850,8 @@ static int ocfs2_unlink(struct inode *dir,
 	}
 
 	if (inode_is_unlinkable(inode)) {
-		status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, inode,
+		status = ocfs2_prepare_orphan_dir(osb, &orphan_dir,
+						  OCFS2_I(inode)->ip_blkno,
 						  orphan_name, &orphan_insert);
 		if (status < 0) {
 			mlog_errno(status);
@@ -866,7 +867,7 @@ static int ocfs2_unlink(struct inode *dir,
 		goto leave;
 	}
 
-	status = ocfs2_journal_access_di(handle, inode, fe_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1241,9 +1242,8 @@ static int ocfs2_rename(struct inode *old_dir,
 
 		if (S_ISDIR(new_inode->i_mode) || (new_inode->i_nlink == 1)) {
 			status = ocfs2_prepare_orphan_dir(osb, &orphan_dir,
-							  new_inode,
-							  orphan_name,
-							  &orphan_insert);
+						OCFS2_I(new_inode)->ip_blkno,
+						orphan_name, &orphan_insert);
 			if (status < 0) {
 				mlog_errno(status);
 				goto bail;
@@ -1284,7 +1284,8 @@ static int ocfs2_rename(struct inode *old_dir,
 				goto bail;
 			}
 		}
-		status = ocfs2_journal_access_di(handle, new_inode, newfe_bh,
+		status = ocfs2_journal_access_di(handle, INODE_CACHE(new_inode),
+						 newfe_bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -1331,7 +1332,8 @@ static int ocfs2_rename(struct inode *old_dir,
 	old_inode->i_ctime = CURRENT_TIME;
 	mark_inode_dirty(old_inode);
 
-	status = ocfs2_journal_access_di(handle, old_inode, old_inode_bh,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(old_inode),
+					 old_inode_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status >= 0) {
 		old_di = (struct ocfs2_dinode *) old_inode_bh->b_data;
@@ -1407,9 +1409,10 @@ static int ocfs2_rename(struct inode *old_dir,
 			     (int)old_dir_nlink, old_dir->i_nlink);
 		} else {
 			struct ocfs2_dinode *fe;
-			status = ocfs2_journal_access_di(handle, old_dir,
-						      old_dir_bh,
-						      OCFS2_JOURNAL_ACCESS_WRITE);
+			status = ocfs2_journal_access_di(handle,
+							 INODE_CACHE(old_dir),
+							 old_dir_bh,
+							 OCFS2_JOURNAL_ACCESS_WRITE);
 			fe = (struct ocfs2_dinode *) old_dir_bh->b_data;
 			ocfs2_set_links_count(fe, old_dir->i_nlink);
 			status = ocfs2_journal_dirty(handle, old_dir_bh);
@@ -1527,9 +1530,11 @@ static int ocfs2_create_symlink_data(struct ocfs2_super *osb,
 			mlog_errno(status);
 			goto bail;
 		}
-		ocfs2_set_new_buffer_uptodate(inode, bhs[virtual]);
+		ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode),
+					      bhs[virtual]);
 
-		status = ocfs2_journal_access(handle, inode, bhs[virtual],
+		status = ocfs2_journal_access(handle, INODE_CACHE(inode),
+					      bhs[virtual],
 					      OCFS2_JOURNAL_ACCESS_CREATE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -1692,7 +1697,11 @@ static int ocfs2_symlink(struct inode *dir,
 	}
 	did_quota_inode = 1;
 
-	status = ocfs2_mknod_locked(osb, dir, inode, dentry,
+	mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry,
+		   inode->i_mode, dentry->d_name.len,
+		   dentry->d_name.name);
+
+	status = ocfs2_mknod_locked(osb, dir, inode,
 				    0, &new_fe_bh, parent_fe_bh, handle,
 				    inode_ac);
 	if (status < 0) {
@@ -1842,7 +1851,7 @@ bail:
 
 static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
 				    struct inode **ret_orphan_dir,
-				    struct inode *inode,
+				    u64 blkno,
 				    char *name,
 				    struct ocfs2_dir_lookup_result *lookup)
 {
@@ -1850,7 +1859,7 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
 	struct buffer_head *orphan_dir_bh = NULL;
 	int status = 0;
 
-	status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, name);
+	status = ocfs2_blkno_stringify(blkno, name);
 	if (status < 0) {
 		mlog_errno(status);
 		return status;
@@ -1917,7 +1926,9 @@ static int ocfs2_orphan_add(struct ocfs2_super *osb,
 		goto leave;
 	}
 
-	status = ocfs2_journal_access_di(handle, orphan_dir_inode, orphan_dir_bh,
+	status = ocfs2_journal_access_di(handle,
+					 INODE_CACHE(orphan_dir_inode),
+					 orphan_dir_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -2002,7 +2013,9 @@ int ocfs2_orphan_del(struct ocfs2_super *osb,
 		goto leave;
 	}
 
-	status = ocfs2_journal_access_di(handle,orphan_dir_inode,  orphan_dir_bh,
+	status = ocfs2_journal_access_di(handle,
+					 INODE_CACHE(orphan_dir_inode),
+					 orphan_dir_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -2028,6 +2041,274 @@ leave:
 	return status;
 }
 
+int ocfs2_create_inode_in_orphan(struct inode *dir,
+				 int mode,
+				 struct inode **new_inode)
+{
+	int status, did_quota_inode = 0;
+	struct inode *inode = NULL;
+	struct inode *orphan_dir = NULL;
+	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
+	struct ocfs2_dinode *di = NULL;
+	handle_t *handle = NULL;
+	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
+	struct buffer_head *parent_di_bh = NULL;
+	struct buffer_head *new_di_bh = NULL;
+	struct ocfs2_alloc_context *inode_ac = NULL;
+	struct ocfs2_dir_lookup_result orphan_insert = { NULL, };
+
+	status = ocfs2_inode_lock(dir, &parent_di_bh, 1);
+	if (status < 0) {
+		if (status != -ENOENT)
+			mlog_errno(status);
+		return status;
+	}
+
+	/*
+	 * We give the orphan dir the root blkno to fake an orphan name,
+	 * and allocate enough space for our insertion.
+	 */
+	status = ocfs2_prepare_orphan_dir(osb, &orphan_dir,
+					  osb->root_blkno,
+					  orphan_name, &orphan_insert);
+	if (status < 0) {
+		mlog_errno(status);
+		goto leave;
+	}
+
+	/* reserve an inode spot */
+	status = ocfs2_reserve_new_inode(osb, &inode_ac);
+	if (status < 0) {
+		if (status != -ENOSPC)
+			mlog_errno(status);
+		goto leave;
+	}
+
+	inode = ocfs2_get_init_inode(dir, mode);
+	if (!inode) {
+		status = -ENOMEM;
+		mlog_errno(status);
+		goto leave;
+	}
+
+	handle = ocfs2_start_trans(osb, ocfs2_mknod_credits(osb->sb, 0, 0));
+	if (IS_ERR(handle)) {
+		status = PTR_ERR(handle);
+		handle = NULL;
+		mlog_errno(status);
+		goto leave;
+	}
+
+	/* We don't use standard VFS wrapper because we don't want vfs_dq_init
+	 * to be called. */
+	if (sb_any_quota_active(osb->sb) &&
+	    osb->sb->dq_op->alloc_inode(inode, 1) == NO_QUOTA) {
+		status = -EDQUOT;
+		goto leave;
+	}
+	did_quota_inode = 1;
+
+	/* do the real work now. */
+	status = ocfs2_mknod_locked(osb, dir, inode,
+				    0, &new_di_bh, parent_di_bh, handle,
+				    inode_ac);
+	if (status < 0) {
+		mlog_errno(status);
+		goto leave;
+	}
+
+	status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, orphan_name);
+	if (status < 0) {
+		mlog_errno(status);
+		goto leave;
+	}
+
+	di = (struct ocfs2_dinode *)new_di_bh->b_data;
+	status = ocfs2_orphan_add(osb, handle, inode, di, orphan_name,
+				  &orphan_insert, orphan_dir);
+	if (status < 0) {
+		mlog_errno(status);
+		goto leave;
+	}
+
+	/* get open lock so that only nodes can't remove it from orphan dir. */
+	status = ocfs2_open_lock(inode);
+	if (status < 0)
+		mlog_errno(status);
+
+leave:
+	if (status < 0 && did_quota_inode)
+		vfs_dq_free_inode(inode);
+	if (handle)
+		ocfs2_commit_trans(osb, handle);
+
+	if (orphan_dir) {
+		/* This was locked for us in ocfs2_prepare_orphan_dir() */
+		ocfs2_inode_unlock(orphan_dir, 1);
+		mutex_unlock(&orphan_dir->i_mutex);
+		iput(orphan_dir);
+	}
+
+	if (status == -ENOSPC)
+		mlog(0, "Disk is full\n");
+
+	if ((status < 0) && inode) {
+		clear_nlink(inode);
+		iput(inode);
+	}
+
+	if (inode_ac)
+		ocfs2_free_alloc_context(inode_ac);
+
+	brelse(new_di_bh);
+
+	if (!status)
+		*new_inode = inode;
+
+	ocfs2_free_dir_lookup_result(&orphan_insert);
+
+	ocfs2_inode_unlock(dir, 1);
+	brelse(parent_di_bh);
+	return status;
+}
+
+int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
+				   struct inode *inode,
+				   struct dentry *dentry)
+{
+	int status = 0;
+	struct buffer_head *parent_di_bh = NULL;
+	handle_t *handle = NULL;
+	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
+	struct ocfs2_dinode *dir_di, *di;
+	struct inode *orphan_dir_inode = NULL;
+	struct buffer_head *orphan_dir_bh = NULL;
+	struct buffer_head *di_bh = NULL;
+	struct ocfs2_dir_lookup_result lookup = { NULL, };
+
+	mlog_entry("(0x%p, 0x%p, %.*s')\n", dir, dentry,
+		   dentry->d_name.len, dentry->d_name.name);
+
+	status = ocfs2_inode_lock(dir, &parent_di_bh, 1);
+	if (status < 0) {
+		if (status != -ENOENT)
+			mlog_errno(status);
+		return status;
+	}
+
+	dir_di = (struct ocfs2_dinode *) parent_di_bh->b_data;
+	if (!dir_di->i_links_count) {
+		/* can't make a file in a deleted directory. */
+		status = -ENOENT;
+		goto leave;
+	}
+
+	status = ocfs2_check_dir_for_entry(dir, dentry->d_name.name,
+					   dentry->d_name.len);
+	if (status)
+		goto leave;
+
+	/* get a spot inside the dir. */
+	status = ocfs2_prepare_dir_for_insert(osb, dir, parent_di_bh,
+					      dentry->d_name.name,
+					      dentry->d_name.len, &lookup);
+	if (status < 0) {
+		mlog_errno(status);
+		goto leave;
+	}
+
+	orphan_dir_inode = ocfs2_get_system_file_inode(osb,
+						       ORPHAN_DIR_SYSTEM_INODE,
+						       osb->slot_num);
+	if (!orphan_dir_inode) {
+		status = -EEXIST;
+		mlog_errno(status);
+		goto leave;
+	}
+
+	mutex_lock(&orphan_dir_inode->i_mutex);
+
+	status = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1);
+	if (status < 0) {
+		mlog_errno(status);
+		mutex_unlock(&orphan_dir_inode->i_mutex);
+		iput(orphan_dir_inode);
+		goto leave;
+	}
+
+	status = ocfs2_read_inode_block(inode, &di_bh);
+	if (status < 0) {
+		mlog_errno(status);
+		goto orphan_unlock;
+	}
+
+	handle = ocfs2_start_trans(osb, ocfs2_rename_credits(osb->sb));
+	if (IS_ERR(handle)) {
+		status = PTR_ERR(handle);
+		handle = NULL;
+		mlog_errno(status);
+		goto orphan_unlock;
+	}
+
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode),
+					 di_bh, OCFS2_JOURNAL_ACCESS_WRITE);
+	if (status < 0) {
+		mlog_errno(status);
+		goto out_commit;
+	}
+
+	status = ocfs2_orphan_del(osb, handle, orphan_dir_inode, inode,
+				  orphan_dir_bh);
+	if (status < 0) {
+		mlog_errno(status);
+		goto out_commit;
+	}
+
+	di = (struct ocfs2_dinode *)di_bh->b_data;
+	le32_add_cpu(&di->i_flags, -OCFS2_ORPHANED_FL);
+	di->i_orphaned_slot = 0;
+	ocfs2_journal_dirty(handle, di_bh);
+
+	status = ocfs2_add_entry(handle, dentry, inode,
+				 OCFS2_I(inode)->ip_blkno, parent_di_bh,
+				 &lookup);
+	if (status < 0) {
+		mlog_errno(status);
+		goto out_commit;
+	}
+
+	status = ocfs2_dentry_attach_lock(dentry, inode,
+					  OCFS2_I(dir)->ip_blkno);
+	if (status) {
+		mlog_errno(status);
+		goto out_commit;
+	}
+
+	insert_inode_hash(inode);
+	dentry->d_op = &ocfs2_dentry_ops;
+	d_instantiate(dentry, inode);
+	status = 0;
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+orphan_unlock:
+	ocfs2_inode_unlock(orphan_dir_inode, 1);
+	mutex_unlock(&orphan_dir_inode->i_mutex);
+	iput(orphan_dir_inode);
+leave:
+
+	ocfs2_inode_unlock(dir, 1);
+
+	brelse(di_bh);
+	brelse(parent_di_bh);
+	brelse(orphan_dir_bh);
+
+	ocfs2_free_dir_lookup_result(&lookup);
+
+	mlog_exit(status);
+
+	return status;
+}
+
 const struct inode_operations ocfs2_dir_iops = {
 	.create		= ocfs2_create,
 	.lookup		= ocfs2_lookup,
diff --git a/fs/ocfs2/namei.h b/fs/ocfs2/namei.h
index 688aef64c879..e5d059d4f115 100644
--- a/fs/ocfs2/namei.h
+++ b/fs/ocfs2/namei.h
@@ -35,5 +35,11 @@ int ocfs2_orphan_del(struct ocfs2_super *osb,
 		     struct inode *orphan_dir_inode,
 		     struct inode *inode,
 		     struct buffer_head *orphan_dir_bh);
+int ocfs2_create_inode_in_orphan(struct inode *dir,
+				 int mode,
+				 struct inode **new_inode);
+int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
+				   struct inode *new_inode,
+				   struct dentry *new_dentry);
 
 #endif /* OCFS2_NAMEI_H */
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index 39e1d5a39505..eae404602424 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -51,20 +51,51 @@
 /* For struct ocfs2_blockcheck_stats */
 #include "blockcheck.h"
 
+
+/* Caching of metadata buffers */
+
 /* Most user visible OCFS2 inodes will have very few pieces of
  * metadata, but larger files (including bitmaps, etc) must be taken
  * into account when designing an access scheme. We allow a small
  * amount of inlined blocks to be stored on an array and grow the
  * structure into a rb tree when necessary. */
-#define OCFS2_INODE_MAX_CACHE_ARRAY 2
+#define OCFS2_CACHE_INFO_MAX_ARRAY 2
+
+/* Flags for ocfs2_caching_info */
+
+enum ocfs2_caching_info_flags {
+	/* Indicates that the metadata cache is using the inline array */
+	OCFS2_CACHE_FL_INLINE	= 1<<1,
+};
 
+struct ocfs2_caching_operations;
 struct ocfs2_caching_info {
+	/*
+	 * The parent structure provides the locks, but because the
+	 * parent structure can differ, it provides locking operations
+	 * to struct ocfs2_caching_info.
+	 */
+	const struct ocfs2_caching_operations *ci_ops;
+
+	/* next two are protected by trans_inc_lock */
+	/* which transaction were we created on? Zero if none. */
+	unsigned long		ci_created_trans;
+	/* last transaction we were a part of. */
+	unsigned long		ci_last_trans;
+
+	/* Cache structures */
+	unsigned int		ci_flags;
 	unsigned int		ci_num_cached;
 	union {
-		sector_t	ci_array[OCFS2_INODE_MAX_CACHE_ARRAY];
+	sector_t	ci_array[OCFS2_CACHE_INFO_MAX_ARRAY];
 		struct rb_root	ci_tree;
 	} ci_cache;
 };
+/*
+ * Need this prototype here instead of in uptodate.h because journal.h
+ * uses it.
+ */
+struct super_block *ocfs2_metadata_cache_get_super(struct ocfs2_caching_info *ci);
 
 /* this limits us to 256 nodes
  * if we need more, we can do a kmalloc for the map */
@@ -377,12 +408,17 @@ struct ocfs2_super
 
 	/* the group we used to allocate inodes. */
 	u64				osb_inode_alloc_group;
+
+	/* rb tree root for refcount lock. */
+	struct rb_root	osb_rf_lock_tree;
+	struct ocfs2_refcount_tree *osb_ref_tree_lru;
 };
 
 #define OCFS2_SB(sb)	    ((struct ocfs2_super *)(sb)->s_fs_info)
 
 /* Useful typedef for passing around journal access functions */
-typedef int (*ocfs2_journal_access_func)(handle_t *handle, struct inode *inode,
+typedef int (*ocfs2_journal_access_func)(handle_t *handle,
+					 struct ocfs2_caching_info *ci,
 					 struct buffer_head *bh, int type);
 
 static inline int ocfs2_should_order_data(struct inode *inode)
@@ -480,6 +516,13 @@ static inline void ocfs2_add_links_count(struct ocfs2_dinode *di, int n)
 	ocfs2_set_links_count(di, links);
 }
 
+static inline int ocfs2_refcount_tree(struct ocfs2_super *osb)
+{
+	if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE)
+		return 1;
+	return 0;
+}
+
 /* set / clear functions because cluster events can make these happen
  * in parallel so we want the transitions to be atomic. this also
  * means that any future flags osb_flags must be protected by spinlock
@@ -578,6 +621,9 @@ static inline int ocfs2_uses_extended_slot_map(struct ocfs2_super *osb)
 #define OCFS2_IS_VALID_DX_LEAF(ptr)					\
 	(!strcmp((ptr)->dl_signature, OCFS2_DX_LEAF_SIGNATURE))
 
+#define OCFS2_IS_VALID_REFCOUNT_BLOCK(ptr)				\
+	(!strcmp((ptr)->rf_signature, OCFS2_REFCOUNT_BLOCK_SIGNATURE))
+
 static inline unsigned long ino_from_blkno(struct super_block *sb,
 					   u64 blkno)
 {
diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index 7ab6e9e5e77c..e9431e4a5e7c 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -68,6 +68,7 @@
 #define OCFS2_DIR_TRAILER_SIGNATURE	"DIRTRL1"
 #define OCFS2_DX_ROOT_SIGNATURE		"DXDIR01"
 #define OCFS2_DX_LEAF_SIGNATURE		"DXLEAF1"
+#define OCFS2_REFCOUNT_BLOCK_SIGNATURE	"REFCNT1"
 
 /* Compatibility flags */
 #define OCFS2_HAS_COMPAT_FEATURE(sb,mask)			\
@@ -98,7 +99,8 @@
 					 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
 					 | OCFS2_FEATURE_INCOMPAT_XATTR \
 					 | OCFS2_FEATURE_INCOMPAT_META_ECC \
-					 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS)
+					 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \
+					 | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE)
 #define OCFS2_FEATURE_RO_COMPAT_SUPP	(OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
 					 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
 					 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
@@ -160,6 +162,9 @@
 /* Metadata checksum and error correction */
 #define OCFS2_FEATURE_INCOMPAT_META_ECC		0x0800
 
+/* Refcount tree support */
+#define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE	0x1000
+
 /*
  * backup superblock flag is used to indicate that this volume
  * has backup superblocks.
@@ -223,6 +228,7 @@
 #define OCFS2_HAS_XATTR_FL	(0x0002)
 #define OCFS2_INLINE_XATTR_FL	(0x0004)
 #define OCFS2_INDEXED_DIR_FL	(0x0008)
+#define OCFS2_HAS_REFCOUNT_FL   (0x0010)
 
 /* Inode attributes, keep in sync with EXT2 */
 #define OCFS2_SECRM_FL		(0x00000001)	/* Secure deletion */
@@ -241,8 +247,11 @@
 /*
  * Extent record flags (e_node.leaf.flags)
  */
-#define OCFS2_EXT_UNWRITTEN	(0x01)	/* Extent is allocated but
-					 * unwritten */
+#define OCFS2_EXT_UNWRITTEN		(0x01)	/* Extent is allocated but
+						 * unwritten */
+#define OCFS2_EXT_REFCOUNTED		(0x02)  /* Extent is reference
+						 * counted in an associated
+						 * refcount tree */
 
 /*
  * ioctl commands
@@ -292,6 +301,15 @@ struct ocfs2_new_group_input {
 #define OCFS2_IOC_GROUP_ADD	_IOW('o', 2,struct ocfs2_new_group_input)
 #define OCFS2_IOC_GROUP_ADD64	_IOW('o', 3,struct ocfs2_new_group_input)
 
+/* Used to pass 2 file names to reflink. */
+struct reflink_arguments {
+	__u64 old_path;
+	__u64 new_path;
+	__u64 preserve;
+};
+#define OCFS2_IOC_REFLINK	_IOW('o', 4, struct reflink_arguments)
+
+
 /*
  * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
  */
@@ -717,7 +735,8 @@ struct ocfs2_dinode {
 	__le64 i_xattr_loc;
 /*80*/	struct ocfs2_block_check i_check;	/* Error checking */
 /*88*/	__le64 i_dx_root;		/* Pointer to dir index root block */
-	__le64 i_reserved2[5];
+/*90*/	__le64 i_refcount_loc;
+	__le64 i_reserved2[4];
 /*B8*/	union {
 		__le64 i_pad1;		/* Generic way to refer to this
 					   64bit union */
@@ -901,6 +920,60 @@ struct ocfs2_group_desc
 /*40*/	__u8    bg_bitmap[0];
 };
 
+struct ocfs2_refcount_rec {
+/*00*/	__le64 r_cpos;		/* Physical offset, in clusters */
+	__le32 r_clusters;	/* Clusters covered by this extent */
+	__le32 r_refcount;	/* Reference count of this extent */
+/*10*/
+};
+#define OCFS2_32BIT_POS_MASK		(0xffffffffULL)
+
+#define OCFS2_REFCOUNT_LEAF_FL          (0x00000001)
+#define OCFS2_REFCOUNT_TREE_FL          (0x00000002)
+
+struct ocfs2_refcount_list {
+/*00*/	__le16 rl_count;	/* Maximum number of entries possible
+				   in rl_records */
+	__le16 rl_used;		/* Current number of used records */
+	__le32 rl_reserved2;
+	__le64 rl_reserved1;	/* Pad to sizeof(ocfs2_refcount_record) */
+/*10*/	struct ocfs2_refcount_rec rl_recs[0];	/* Refcount records */
+};
+
+
+struct ocfs2_refcount_block {
+/*00*/	__u8 rf_signature[8];		/* Signature for verification */
+	__le16 rf_suballoc_slot;	/* Slot suballocator this block
+					   belongs to */
+	__le16 rf_suballoc_bit;		/* Bit offset in suballocator
+					   block group */
+	__le32 rf_fs_generation;	/* Must match superblock */
+/*10*/	__le64 rf_blkno;		/* Offset on disk, in blocks */
+	__le64 rf_parent;		/* Parent block, only valid if
+					   OCFS2_REFCOUNT_LEAF_FL is set in
+					   rf_flags */
+/*20*/	struct ocfs2_block_check rf_check;	/* Error checking */
+	__le64 rf_last_eb_blk;		/* Pointer to last extent block */
+/*30*/	__le32 rf_count;		/* Number of inodes sharing this
+					   refcount tree */
+	__le32 rf_flags;		/* See the flags above */
+	__le32 rf_clusters;		/* clusters covered by refcount tree. */
+	__le32 rf_cpos;			/* cluster offset in refcount tree.*/
+/*40*/	__le32 rf_generation;		/* generation number. all be the same
+					 * for the same refcount tree. */
+	__le32 rf_reserved0;
+	__le64 rf_reserved1[7];
+/*80*/	union {
+		struct ocfs2_refcount_list rf_records;  /* List of refcount
+							  records */
+		struct ocfs2_extent_list rf_list;	/* Extent record list,
+							only valid if
+							OCFS2_REFCOUNT_TREE_FL
+							is set in rf_flags */
+	};
+/* Actual on-disk size is one block */
+};
+
 /*
  * On disk extended attribute structure for OCFS2.
  */
@@ -1312,6 +1385,32 @@ static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
 
 	return size / sizeof(struct ocfs2_extent_rec);
 }
+
+static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb)
+{
+	int size;
+
+	size = sb->s_blocksize -
+		offsetof(struct ocfs2_refcount_block, rf_list.l_recs);
+
+	return size / sizeof(struct ocfs2_extent_rec);
+}
+
+static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb)
+{
+	int size;
+
+	size = sb->s_blocksize -
+		offsetof(struct ocfs2_refcount_block, rf_records.rl_recs);
+
+	return size / sizeof(struct ocfs2_refcount_rec);
+}
+
+static inline u32
+ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec)
+{
+	return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
+}
 #else
 static inline int ocfs2_fast_symlink_chars(int blocksize)
 {
diff --git a/fs/ocfs2/ocfs2_lockid.h b/fs/ocfs2/ocfs2_lockid.h
index c212cf5a2bdf..d277aabf5dfb 100644
--- a/fs/ocfs2/ocfs2_lockid.h
+++ b/fs/ocfs2/ocfs2_lockid.h
@@ -49,6 +49,7 @@ enum ocfs2_lock_type {
 	OCFS2_LOCK_TYPE_QINFO,
 	OCFS2_LOCK_TYPE_NFS_SYNC,
 	OCFS2_LOCK_TYPE_ORPHAN_SCAN,
+	OCFS2_LOCK_TYPE_REFCOUNT,
 	OCFS2_NUM_LOCK_TYPES
 };
 
@@ -89,6 +90,9 @@ static inline char ocfs2_lock_type_char(enum ocfs2_lock_type type)
 		case OCFS2_LOCK_TYPE_ORPHAN_SCAN:
 			c = 'P';
 			break;
+		case OCFS2_LOCK_TYPE_REFCOUNT:
+			c = 'T';
+			break;
 		default:
 			c = '\0';
 	}
@@ -110,6 +114,7 @@ static char *ocfs2_lock_type_strings[] = {
 	[OCFS2_LOCK_TYPE_QINFO] = "Quota",
 	[OCFS2_LOCK_TYPE_NFS_SYNC] = "NFSSync",
 	[OCFS2_LOCK_TYPE_ORPHAN_SCAN] = "OrphanScan",
+	[OCFS2_LOCK_TYPE_REFCOUNT] = "Refcount",
 };
 
 static inline const char *ocfs2_lock_type_string(enum ocfs2_lock_type type)
diff --git a/fs/ocfs2/quota.h b/fs/ocfs2/quota.h
index 3fb96fcd4c81..e5df9d170b0c 100644
--- a/fs/ocfs2/quota.h
+++ b/fs/ocfs2/quota.h
@@ -109,7 +109,7 @@ void ocfs2_unlock_global_qf(struct ocfs2_mem_dqinfo *oinfo, int ex);
 int ocfs2_read_quota_block(struct inode *inode, u64 v_block,
 			   struct buffer_head **bh);
 
-extern struct dquot_operations ocfs2_quota_operations;
+extern const struct dquot_operations ocfs2_quota_operations;
 extern struct quota_format_type ocfs2_quota_format;
 
 int ocfs2_quota_setup(void);
diff --git a/fs/ocfs2/quota_global.c b/fs/ocfs2/quota_global.c
index 44f2a5e1d042..b437dc0c4cad 100644
--- a/fs/ocfs2/quota_global.c
+++ b/fs/ocfs2/quota_global.c
@@ -154,7 +154,7 @@ static int ocfs2_get_quota_block(struct inode *inode, int block,
 		err = -EIO;
 		mlog_errno(err);
 	}
-	return err;;
+	return err;
 }
 
 /* Read data from global quotafile - avoid pagecache and such because we cannot
@@ -253,8 +253,9 @@ ssize_t ocfs2_quota_write(struct super_block *sb, int type,
 	flush_dcache_page(bh->b_page);
 	set_buffer_uptodate(bh);
 	unlock_buffer(bh);
-	ocfs2_set_buffer_uptodate(gqinode, bh);
-	err = ocfs2_journal_access_dq(handle, gqinode, bh, ja_type);
+	ocfs2_set_buffer_uptodate(INODE_CACHE(gqinode), bh);
+	err = ocfs2_journal_access_dq(handle, INODE_CACHE(gqinode), bh,
+				      ja_type);
 	if (err < 0) {
 		brelse(bh);
 		goto out;
@@ -849,7 +850,7 @@ static void ocfs2_destroy_dquot(struct dquot *dquot)
 	kmem_cache_free(ocfs2_dquot_cachep, dquot);
 }
 
-struct dquot_operations ocfs2_quota_operations = {
+const struct dquot_operations ocfs2_quota_operations = {
 	.initialize	= dquot_initialize,
 	.drop		= dquot_drop,
 	.alloc_space	= dquot_alloc_space,
diff --git a/fs/ocfs2/quota_local.c b/fs/ocfs2/quota_local.c
index bdb09cb6e1fe..1a2c50a759fa 100644
--- a/fs/ocfs2/quota_local.c
+++ b/fs/ocfs2/quota_local.c
@@ -108,7 +108,7 @@ static int ocfs2_modify_bh(struct inode *inode, struct buffer_head *bh,
 		mlog_errno(status);
 		return status;
 	}
-	status = ocfs2_journal_access_dq(handle, inode, bh,
+	status = ocfs2_journal_access_dq(handle, INODE_CACHE(inode), bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -510,7 +510,8 @@ static int ocfs2_recover_local_quota_file(struct inode *lqinode,
 				goto out_commit;
 			}
 			/* Release local quota file entry */
-			status = ocfs2_journal_access_dq(handle, lqinode,
+			status = ocfs2_journal_access_dq(handle,
+					INODE_CACHE(lqinode),
 					qbh, OCFS2_JOURNAL_ACCESS_WRITE);
 			if (status < 0) {
 				mlog_errno(status);
@@ -619,7 +620,8 @@ int ocfs2_finish_quota_recovery(struct ocfs2_super *osb,
 			mlog_errno(status);
 			goto out_bh;
 		}
-		status = ocfs2_journal_access_dq(handle, lqinode, bh,
+		status = ocfs2_journal_access_dq(handle, INODE_CACHE(lqinode),
+						 bh,
 						 OCFS2_JOURNAL_ACCESS_WRITE);
 		if (status < 0) {
 			mlog_errno(status);
@@ -993,8 +995,8 @@ static struct ocfs2_quota_chunk *ocfs2_local_quota_add_chunk(
 		goto out_trans;
 	}
 	dchunk = (struct ocfs2_local_disk_chunk *)bh->b_data;
-	ocfs2_set_new_buffer_uptodate(lqinode, bh);
-	status = ocfs2_journal_access_dq(handle, lqinode, bh,
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(lqinode), bh);
+	status = ocfs2_journal_access_dq(handle, INODE_CACHE(lqinode), bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1027,8 +1029,8 @@ static struct ocfs2_quota_chunk *ocfs2_local_quota_add_chunk(
 		mlog_errno(status);
 		goto out_trans;
 	}
-	ocfs2_set_new_buffer_uptodate(lqinode, dbh);
-	status = ocfs2_journal_access_dq(handle, lqinode, dbh,
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(lqinode), dbh);
+	status = ocfs2_journal_access_dq(handle, INODE_CACHE(lqinode), dbh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1131,7 +1133,7 @@ static struct ocfs2_quota_chunk *ocfs2_extend_local_quota_file(
 		mlog_errno(status);
 		goto out;
 	}
-	ocfs2_set_new_buffer_uptodate(lqinode, bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(lqinode), bh);
 
 	/* Local quota info, chunk header and the new block we initialize */
 	handle = ocfs2_start_trans(OCFS2_SB(sb),
@@ -1143,7 +1145,7 @@ static struct ocfs2_quota_chunk *ocfs2_extend_local_quota_file(
 		goto out;
 	}
 	/* Zero created block */
-	status = ocfs2_journal_access_dq(handle, lqinode, bh,
+	status = ocfs2_journal_access_dq(handle, INODE_CACHE(lqinode), bh,
 				 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1158,7 +1160,8 @@ static struct ocfs2_quota_chunk *ocfs2_extend_local_quota_file(
 		goto out_trans;
 	}
 	/* Update chunk header */
-	status = ocfs2_journal_access_dq(handle, lqinode, chunk->qc_headerbh,
+	status = ocfs2_journal_access_dq(handle, INODE_CACHE(lqinode),
+					 chunk->qc_headerbh,
 				 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1292,7 +1295,8 @@ static int ocfs2_local_release_dquot(struct dquot *dquot)
 		goto out;
 	}
 
-	status = ocfs2_journal_access_dq(handle, sb_dqopt(sb)->files[type],
+	status = ocfs2_journal_access_dq(handle,
+			INODE_CACHE(sb_dqopt(sb)->files[type]),
 			od->dq_chunk->qc_headerbh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
new file mode 100644
index 000000000000..60287fc56bcb
--- /dev/null
+++ b/fs/ocfs2/refcounttree.c
@@ -0,0 +1,4313 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * refcounttree.c
+ *
+ * Copyright (C) 2009 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ */
+
+#include <linux/sort.h>
+#define MLOG_MASK_PREFIX ML_REFCOUNT
+#include <cluster/masklog.h>
+#include "ocfs2.h"
+#include "inode.h"
+#include "alloc.h"
+#include "suballoc.h"
+#include "journal.h"
+#include "uptodate.h"
+#include "super.h"
+#include "buffer_head_io.h"
+#include "blockcheck.h"
+#include "refcounttree.h"
+#include "sysfile.h"
+#include "dlmglue.h"
+#include "extent_map.h"
+#include "aops.h"
+#include "xattr.h"
+#include "namei.h"
+
+#include <linux/bio.h>
+#include <linux/blkdev.h>
+#include <linux/gfp.h>
+#include <linux/slab.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
+#include <linux/swap.h>
+#include <linux/security.h>
+#include <linux/fsnotify.h>
+#include <linux/quotaops.h>
+#include <linux/namei.h>
+#include <linux/mount.h>
+
+struct ocfs2_cow_context {
+	struct inode *inode;
+	u32 cow_start;
+	u32 cow_len;
+	struct ocfs2_extent_tree data_et;
+	struct ocfs2_refcount_tree *ref_tree;
+	struct buffer_head *ref_root_bh;
+	struct ocfs2_alloc_context *meta_ac;
+	struct ocfs2_alloc_context *data_ac;
+	struct ocfs2_cached_dealloc_ctxt dealloc;
+	void *cow_object;
+	struct ocfs2_post_refcount *post_refcount;
+	int extra_credits;
+	int (*get_clusters)(struct ocfs2_cow_context *context,
+			    u32 v_cluster, u32 *p_cluster,
+			    u32 *num_clusters,
+			    unsigned int *extent_flags);
+	int (*cow_duplicate_clusters)(handle_t *handle,
+				      struct ocfs2_cow_context *context,
+				      u32 cpos, u32 old_cluster,
+				      u32 new_cluster, u32 new_len);
+};
+
+static inline struct ocfs2_refcount_tree *
+cache_info_to_refcount(struct ocfs2_caching_info *ci)
+{
+	return container_of(ci, struct ocfs2_refcount_tree, rf_ci);
+}
+
+static int ocfs2_validate_refcount_block(struct super_block *sb,
+					 struct buffer_head *bh)
+{
+	int rc;
+	struct ocfs2_refcount_block *rb =
+		(struct ocfs2_refcount_block *)bh->b_data;
+
+	mlog(0, "Validating refcount block %llu\n",
+	     (unsigned long long)bh->b_blocknr);
+
+	BUG_ON(!buffer_uptodate(bh));
+
+	/*
+	 * If the ecc fails, we return the error but otherwise
+	 * leave the filesystem running.  We know any error is
+	 * local to this block.
+	 */
+	rc = ocfs2_validate_meta_ecc(sb, bh->b_data, &rb->rf_check);
+	if (rc) {
+		mlog(ML_ERROR, "Checksum failed for refcount block %llu\n",
+		     (unsigned long long)bh->b_blocknr);
+		return rc;
+	}
+
+
+	if (!OCFS2_IS_VALID_REFCOUNT_BLOCK(rb)) {
+		ocfs2_error(sb,
+			    "Refcount block #%llu has bad signature %.*s",
+			    (unsigned long long)bh->b_blocknr, 7,
+			    rb->rf_signature);
+		return -EINVAL;
+	}
+
+	if (le64_to_cpu(rb->rf_blkno) != bh->b_blocknr) {
+		ocfs2_error(sb,
+			    "Refcount block #%llu has an invalid rf_blkno "
+			    "of %llu",
+			    (unsigned long long)bh->b_blocknr,
+			    (unsigned long long)le64_to_cpu(rb->rf_blkno));
+		return -EINVAL;
+	}
+
+	if (le32_to_cpu(rb->rf_fs_generation) != OCFS2_SB(sb)->fs_generation) {
+		ocfs2_error(sb,
+			    "Refcount block #%llu has an invalid "
+			    "rf_fs_generation of #%u",
+			    (unsigned long long)bh->b_blocknr,
+			    le32_to_cpu(rb->rf_fs_generation));
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ocfs2_read_refcount_block(struct ocfs2_caching_info *ci,
+				     u64 rb_blkno,
+				     struct buffer_head **bh)
+{
+	int rc;
+	struct buffer_head *tmp = *bh;
+
+	rc = ocfs2_read_block(ci, rb_blkno, &tmp,
+			      ocfs2_validate_refcount_block);
+
+	/* If ocfs2_read_block() got us a new bh, pass it up. */
+	if (!rc && !*bh)
+		*bh = tmp;
+
+	return rc;
+}
+
+static u64 ocfs2_refcount_cache_owner(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	return rf->rf_blkno;
+}
+
+static struct super_block *
+ocfs2_refcount_cache_get_super(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	return rf->rf_sb;
+}
+
+static void ocfs2_refcount_cache_lock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	spin_lock(&rf->rf_lock);
+}
+
+static void ocfs2_refcount_cache_unlock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	spin_unlock(&rf->rf_lock);
+}
+
+static void ocfs2_refcount_cache_io_lock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	mutex_lock(&rf->rf_io_mutex);
+}
+
+static void ocfs2_refcount_cache_io_unlock(struct ocfs2_caching_info *ci)
+{
+	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);
+
+	mutex_unlock(&rf->rf_io_mutex);
+}
+
+static const struct ocfs2_caching_operations ocfs2_refcount_caching_ops = {
+	.co_owner		= ocfs2_refcount_cache_owner,
+	.co_get_super		= ocfs2_refcount_cache_get_super,
+	.co_cache_lock		= ocfs2_refcount_cache_lock,
+	.co_cache_unlock	= ocfs2_refcount_cache_unlock,
+	.co_io_lock		= ocfs2_refcount_cache_io_lock,
+	.co_io_unlock		= ocfs2_refcount_cache_io_unlock,
+};
+
+static struct ocfs2_refcount_tree *
+ocfs2_find_refcount_tree(struct ocfs2_super *osb, u64 blkno)
+{
+	struct rb_node *n = osb->osb_rf_lock_tree.rb_node;
+	struct ocfs2_refcount_tree *tree = NULL;
+
+	while (n) {
+		tree = rb_entry(n, struct ocfs2_refcount_tree, rf_node);
+
+		if (blkno < tree->rf_blkno)
+			n = n->rb_left;
+		else if (blkno > tree->rf_blkno)
+			n = n->rb_right;
+		else
+			return tree;
+	}
+
+	return NULL;
+}
+
+/* osb_lock is already locked. */
+static void ocfs2_insert_refcount_tree(struct ocfs2_super *osb,
+				       struct ocfs2_refcount_tree *new)
+{
+	u64 rf_blkno = new->rf_blkno;
+	struct rb_node *parent = NULL;
+	struct rb_node **p = &osb->osb_rf_lock_tree.rb_node;
+	struct ocfs2_refcount_tree *tmp;
+
+	while (*p) {
+		parent = *p;
+
+		tmp = rb_entry(parent, struct ocfs2_refcount_tree,
+			       rf_node);
+
+		if (rf_blkno < tmp->rf_blkno)
+			p = &(*p)->rb_left;
+		else if (rf_blkno > tmp->rf_blkno)
+			p = &(*p)->rb_right;
+		else {
+			/* This should never happen! */
+			mlog(ML_ERROR, "Duplicate refcount block %llu found!\n",
+			     (unsigned long long)rf_blkno);
+			BUG();
+		}
+	}
+
+	rb_link_node(&new->rf_node, parent, p);
+	rb_insert_color(&new->rf_node, &osb->osb_rf_lock_tree);
+}
+
+static void ocfs2_free_refcount_tree(struct ocfs2_refcount_tree *tree)
+{
+	ocfs2_metadata_cache_exit(&tree->rf_ci);
+	ocfs2_simple_drop_lockres(OCFS2_SB(tree->rf_sb), &tree->rf_lockres);
+	ocfs2_lock_res_free(&tree->rf_lockres);
+	kfree(tree);
+}
+
+static inline void
+ocfs2_erase_refcount_tree_from_list_no_lock(struct ocfs2_super *osb,
+					struct ocfs2_refcount_tree *tree)
+{
+	rb_erase(&tree->rf_node, &osb->osb_rf_lock_tree);
+	if (osb->osb_ref_tree_lru && osb->osb_ref_tree_lru == tree)
+		osb->osb_ref_tree_lru = NULL;
+}
+
+static void ocfs2_erase_refcount_tree_from_list(struct ocfs2_super *osb,
+					struct ocfs2_refcount_tree *tree)
+{
+	spin_lock(&osb->osb_lock);
+	ocfs2_erase_refcount_tree_from_list_no_lock(osb, tree);
+	spin_unlock(&osb->osb_lock);
+}
+
+void ocfs2_kref_remove_refcount_tree(struct kref *kref)
+{
+	struct ocfs2_refcount_tree *tree =
+		container_of(kref, struct ocfs2_refcount_tree, rf_getcnt);
+
+	ocfs2_free_refcount_tree(tree);
+}
+
+static inline void
+ocfs2_refcount_tree_get(struct ocfs2_refcount_tree *tree)
+{
+	kref_get(&tree->rf_getcnt);
+}
+
+static inline void
+ocfs2_refcount_tree_put(struct ocfs2_refcount_tree *tree)
+{
+	kref_put(&tree->rf_getcnt, ocfs2_kref_remove_refcount_tree);
+}
+
+static inline void ocfs2_init_refcount_tree_ci(struct ocfs2_refcount_tree *new,
+					       struct super_block *sb)
+{
+	ocfs2_metadata_cache_init(&new->rf_ci, &ocfs2_refcount_caching_ops);
+	mutex_init(&new->rf_io_mutex);
+	new->rf_sb = sb;
+	spin_lock_init(&new->rf_lock);
+}
+
+static inline void ocfs2_init_refcount_tree_lock(struct ocfs2_super *osb,
+					struct ocfs2_refcount_tree *new,
+					u64 rf_blkno, u32 generation)
+{
+	init_rwsem(&new->rf_sem);
+	ocfs2_refcount_lock_res_init(&new->rf_lockres, osb,
+				     rf_blkno, generation);
+}
+
+static struct ocfs2_refcount_tree*
+ocfs2_allocate_refcount_tree(struct ocfs2_super *osb, u64 rf_blkno)
+{
+	struct ocfs2_refcount_tree *new;
+
+	new = kzalloc(sizeof(struct ocfs2_refcount_tree), GFP_NOFS);
+	if (!new)
+		return NULL;
+
+	new->rf_blkno = rf_blkno;
+	kref_init(&new->rf_getcnt);
+	ocfs2_init_refcount_tree_ci(new, osb->sb);
+
+	return new;
+}
+
+static int ocfs2_get_refcount_tree(struct ocfs2_super *osb, u64 rf_blkno,
+				   struct ocfs2_refcount_tree **ret_tree)
+{
+	int ret = 0;
+	struct ocfs2_refcount_tree *tree, *new = NULL;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_block *ref_rb;
+
+	spin_lock(&osb->osb_lock);
+	if (osb->osb_ref_tree_lru &&
+	    osb->osb_ref_tree_lru->rf_blkno == rf_blkno)
+		tree = osb->osb_ref_tree_lru;
+	else
+		tree = ocfs2_find_refcount_tree(osb, rf_blkno);
+	if (tree)
+		goto out;
+
+	spin_unlock(&osb->osb_lock);
+
+	new = ocfs2_allocate_refcount_tree(osb, rf_blkno);
+	if (!new) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		return ret;
+	}
+	/*
+	 * We need the generation to create the refcount tree lock and since
+	 * it isn't changed during the tree modification, we are safe here to
+	 * read without protection.
+	 * We also have to purge the cache after we create the lock since the
+	 * refcount block may have the stale data. It can only be trusted when
+	 * we hold the refcount lock.
+	 */
+	ret = ocfs2_read_refcount_block(&new->rf_ci, rf_blkno, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		ocfs2_metadata_cache_exit(&new->rf_ci);
+		kfree(new);
+		return ret;
+	}
+
+	ref_rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	new->rf_generation = le32_to_cpu(ref_rb->rf_generation);
+	ocfs2_init_refcount_tree_lock(osb, new, rf_blkno,
+				      new->rf_generation);
+	ocfs2_metadata_cache_purge(&new->rf_ci);
+
+	spin_lock(&osb->osb_lock);
+	tree = ocfs2_find_refcount_tree(osb, rf_blkno);
+	if (tree)
+		goto out;
+
+	ocfs2_insert_refcount_tree(osb, new);
+
+	tree = new;
+	new = NULL;
+
+out:
+	*ret_tree = tree;
+
+	osb->osb_ref_tree_lru = tree;
+
+	spin_unlock(&osb->osb_lock);
+
+	if (new)
+		ocfs2_free_refcount_tree(new);
+
+	brelse(ref_root_bh);
+	return ret;
+}
+
+static int ocfs2_get_refcount_block(struct inode *inode, u64 *ref_blkno)
+{
+	int ret;
+	struct buffer_head *di_bh = NULL;
+	struct ocfs2_dinode *di;
+
+	ret = ocfs2_read_inode_block(inode, &di_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	BUG_ON(!(OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+	di = (struct ocfs2_dinode *)di_bh->b_data;
+	*ref_blkno = le64_to_cpu(di->i_refcount_loc);
+	brelse(di_bh);
+out:
+	return ret;
+}
+
+static int __ocfs2_lock_refcount_tree(struct ocfs2_super *osb,
+				      struct ocfs2_refcount_tree *tree, int rw)
+{
+	int ret;
+
+	ret = ocfs2_refcount_lock(tree, rw);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (rw)
+		down_write(&tree->rf_sem);
+	else
+		down_read(&tree->rf_sem);
+
+out:
+	return ret;
+}
+
+/*
+ * Lock the refcount tree pointed by ref_blkno and return the tree.
+ * In most case, we lock the tree and read the refcount block.
+ * So read it here if the caller really needs it.
+ *
+ * If the tree has been re-created by other node, it will free the
+ * old one and re-create it.
+ */
+int ocfs2_lock_refcount_tree(struct ocfs2_super *osb,
+			     u64 ref_blkno, int rw,
+			     struct ocfs2_refcount_tree **ret_tree,
+			     struct buffer_head **ref_bh)
+{
+	int ret, delete_tree = 0;
+	struct ocfs2_refcount_tree *tree = NULL;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_block *rb;
+
+again:
+	ret = ocfs2_get_refcount_tree(osb, ref_blkno, &tree);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	ocfs2_refcount_tree_get(tree);
+
+	ret = __ocfs2_lock_refcount_tree(osb, tree, rw);
+	if (ret) {
+		mlog_errno(ret);
+		ocfs2_refcount_tree_put(tree);
+		goto out;
+	}
+
+	ret = ocfs2_read_refcount_block(&tree->rf_ci, tree->rf_blkno,
+					&ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		ocfs2_unlock_refcount_tree(osb, tree, rw);
+		ocfs2_refcount_tree_put(tree);
+		goto out;
+	}
+
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	/*
+	 * If the refcount block has been freed and re-created, we may need
+	 * to recreate the refcount tree also.
+	 *
+	 * Here we just remove the tree from the rb-tree, and the last
+	 * kref holder will unlock and delete this refcount_tree.
+	 * Then we goto "again" and ocfs2_get_refcount_tree will create
+	 * the new refcount tree for us.
+	 */
+	if (tree->rf_generation != le32_to_cpu(rb->rf_generation)) {
+		if (!tree->rf_removed) {
+			ocfs2_erase_refcount_tree_from_list(osb, tree);
+			tree->rf_removed = 1;
+			delete_tree = 1;
+		}
+
+		ocfs2_unlock_refcount_tree(osb, tree, rw);
+		/*
+		 * We get an extra reference when we create the refcount
+		 * tree, so another put will destroy it.
+		 */
+		if (delete_tree)
+			ocfs2_refcount_tree_put(tree);
+		brelse(ref_root_bh);
+		ref_root_bh = NULL;
+		goto again;
+	}
+
+	*ret_tree = tree;
+	if (ref_bh) {
+		*ref_bh = ref_root_bh;
+		ref_root_bh = NULL;
+	}
+out:
+	brelse(ref_root_bh);
+	return ret;
+}
+
+int ocfs2_lock_refcount_tree_by_inode(struct inode *inode, int rw,
+				      struct ocfs2_refcount_tree **ret_tree,
+				      struct buffer_head **ref_bh)
+{
+	int ret;
+	u64 ref_blkno;
+
+	ret = ocfs2_get_refcount_block(inode, &ref_blkno);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	return ocfs2_lock_refcount_tree(OCFS2_SB(inode->i_sb), ref_blkno,
+					rw, ret_tree, ref_bh);
+}
+
+void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb,
+				struct ocfs2_refcount_tree *tree, int rw)
+{
+	if (rw)
+		up_write(&tree->rf_sem);
+	else
+		up_read(&tree->rf_sem);
+
+	ocfs2_refcount_unlock(tree, rw);
+	ocfs2_refcount_tree_put(tree);
+}
+
+void ocfs2_purge_refcount_trees(struct ocfs2_super *osb)
+{
+	struct rb_node *node;
+	struct ocfs2_refcount_tree *tree;
+	struct rb_root *root = &osb->osb_rf_lock_tree;
+
+	while ((node = rb_last(root)) != NULL) {
+		tree = rb_entry(node, struct ocfs2_refcount_tree, rf_node);
+
+		mlog(0, "Purge tree %llu\n",
+		     (unsigned long long) tree->rf_blkno);
+
+		rb_erase(&tree->rf_node, root);
+		ocfs2_free_refcount_tree(tree);
+	}
+}
+
+/*
+ * Create a refcount tree for an inode.
+ * We take for granted that the inode is already locked.
+ */
+static int ocfs2_create_refcount_tree(struct inode *inode,
+				      struct buffer_head *di_bh)
+{
+	int ret;
+	handle_t *handle = NULL;
+	struct ocfs2_alloc_context *meta_ac = NULL;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct buffer_head *new_bh = NULL;
+	struct ocfs2_refcount_block *rb;
+	struct ocfs2_refcount_tree *new_tree = NULL, *tree = NULL;
+	u16 suballoc_bit_start;
+	u32 num_got;
+	u64 first_blkno;
+
+	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+
+	mlog(0, "create tree for inode %lu\n", inode->i_ino);
+
+	ret = ocfs2_reserve_new_metadata_blocks(osb, 1, &meta_ac);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_CREATE_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = ocfs2_claim_metadata(osb, handle, meta_ac, 1,
+				   &suballoc_bit_start, &num_got,
+				   &first_blkno);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	new_tree = ocfs2_allocate_refcount_tree(osb, first_blkno);
+	if (!new_tree) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	new_bh = sb_getblk(inode->i_sb, first_blkno);
+	ocfs2_set_new_buffer_uptodate(&new_tree->rf_ci, new_bh);
+
+	ret = ocfs2_journal_access_rb(handle, &new_tree->rf_ci, new_bh,
+				      OCFS2_JOURNAL_ACCESS_CREATE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	/* Initialize ocfs2_refcount_block. */
+	rb = (struct ocfs2_refcount_block *)new_bh->b_data;
+	memset(rb, 0, inode->i_sb->s_blocksize);
+	strcpy((void *)rb, OCFS2_REFCOUNT_BLOCK_SIGNATURE);
+	rb->rf_suballoc_slot = cpu_to_le16(osb->slot_num);
+	rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
+	rb->rf_fs_generation = cpu_to_le32(osb->fs_generation);
+	rb->rf_blkno = cpu_to_le64(first_blkno);
+	rb->rf_count = cpu_to_le32(1);
+	rb->rf_records.rl_count =
+			cpu_to_le16(ocfs2_refcount_recs_per_rb(osb->sb));
+	spin_lock(&osb->osb_lock);
+	rb->rf_generation = osb->s_next_generation++;
+	spin_unlock(&osb->osb_lock);
+
+	ocfs2_journal_dirty(handle, new_bh);
+
+	spin_lock(&oi->ip_lock);
+	oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
+	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+	di->i_refcount_loc = cpu_to_le64(first_blkno);
+	spin_unlock(&oi->ip_lock);
+
+	mlog(0, "created tree for inode %lu, refblock %llu\n",
+	     inode->i_ino, (unsigned long long)first_blkno);
+
+	ocfs2_journal_dirty(handle, di_bh);
+
+	/*
+	 * We have to init the tree lock here since it will use
+	 * the generation number to create it.
+	 */
+	new_tree->rf_generation = le32_to_cpu(rb->rf_generation);
+	ocfs2_init_refcount_tree_lock(osb, new_tree, first_blkno,
+				      new_tree->rf_generation);
+
+	spin_lock(&osb->osb_lock);
+	tree = ocfs2_find_refcount_tree(osb, first_blkno);
+
+	/*
+	 * We've just created a new refcount tree in this block.  If
+	 * we found a refcount tree on the ocfs2_super, it must be
+	 * one we just deleted.  We free the old tree before
+	 * inserting the new tree.
+	 */
+	BUG_ON(tree && tree->rf_generation == new_tree->rf_generation);
+	if (tree)
+		ocfs2_erase_refcount_tree_from_list_no_lock(osb, tree);
+	ocfs2_insert_refcount_tree(osb, new_tree);
+	spin_unlock(&osb->osb_lock);
+	new_tree = NULL;
+	if (tree)
+		ocfs2_refcount_tree_put(tree);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+
+out:
+	if (new_tree) {
+		ocfs2_metadata_cache_exit(&new_tree->rf_ci);
+		kfree(new_tree);
+	}
+
+	brelse(new_bh);
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+
+	return ret;
+}
+
+static int ocfs2_set_refcount_tree(struct inode *inode,
+				   struct buffer_head *di_bh,
+				   u64 refcount_loc)
+{
+	int ret;
+	handle_t *handle = NULL;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_block *rb;
+	struct ocfs2_refcount_tree *ref_tree;
+
+	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+
+	ret = ocfs2_lock_refcount_tree(osb, refcount_loc, 1,
+				       &ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_SET_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, &ref_tree->rf_ci, ref_root_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	le32_add_cpu(&rb->rf_count, 1);
+
+	ocfs2_journal_dirty(handle, ref_root_bh);
+
+	spin_lock(&oi->ip_lock);
+	oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
+	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+	di->i_refcount_loc = cpu_to_le64(refcount_loc);
+	spin_unlock(&oi->ip_lock);
+	ocfs2_journal_dirty(handle, di_bh);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+out:
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	brelse(ref_root_bh);
+
+	return ret;
+}
+
+int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh)
+{
+	int ret, delete_tree = 0;
+	handle_t *handle = NULL;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_refcount_block *rb;
+	struct inode *alloc_inode = NULL;
+	struct buffer_head *alloc_bh = NULL;
+	struct buffer_head *blk_bh = NULL;
+	struct ocfs2_refcount_tree *ref_tree;
+	int credits = OCFS2_REFCOUNT_TREE_REMOVE_CREDITS;
+	u64 blk = 0, bg_blkno = 0, ref_blkno = le64_to_cpu(di->i_refcount_loc);
+	u16 bit = 0;
+
+	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL))
+		return 0;
+
+	BUG_ON(!ref_blkno);
+	ret = ocfs2_lock_refcount_tree(osb, ref_blkno, 1, &ref_tree, &blk_bh);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	rb = (struct ocfs2_refcount_block *)blk_bh->b_data;
+
+	/*
+	 * If we are the last user, we need to free the block.
+	 * So lock the allocator ahead.
+	 */
+	if (le32_to_cpu(rb->rf_count) == 1) {
+		blk = le64_to_cpu(rb->rf_blkno);
+		bit = le16_to_cpu(rb->rf_suballoc_bit);
+		bg_blkno = ocfs2_which_suballoc_group(blk, bit);
+
+		alloc_inode = ocfs2_get_system_file_inode(osb,
+					EXTENT_ALLOC_SYSTEM_INODE,
+					le16_to_cpu(rb->rf_suballoc_slot));
+		if (!alloc_inode) {
+			ret = -ENOMEM;
+			mlog_errno(ret);
+			goto out;
+		}
+		mutex_lock(&alloc_inode->i_mutex);
+
+		ret = ocfs2_inode_lock(alloc_inode, &alloc_bh, 1);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_mutex;
+		}
+
+		credits += OCFS2_SUBALLOC_FREE;
+	}
+
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, &ref_tree->rf_ci, blk_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	spin_lock(&oi->ip_lock);
+	oi->ip_dyn_features &= ~OCFS2_HAS_REFCOUNT_FL;
+	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+	di->i_refcount_loc = 0;
+	spin_unlock(&oi->ip_lock);
+	ocfs2_journal_dirty(handle, di_bh);
+
+	le32_add_cpu(&rb->rf_count , -1);
+	ocfs2_journal_dirty(handle, blk_bh);
+
+	if (!rb->rf_count) {
+		delete_tree = 1;
+		ocfs2_erase_refcount_tree_from_list(osb, ref_tree);
+		ret = ocfs2_free_suballoc_bits(handle, alloc_inode,
+					       alloc_bh, bit, bg_blkno, 1);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+out_unlock:
+	if (alloc_inode) {
+		ocfs2_inode_unlock(alloc_inode, 1);
+		brelse(alloc_bh);
+	}
+out_mutex:
+	if (alloc_inode) {
+		mutex_unlock(&alloc_inode->i_mutex);
+		iput(alloc_inode);
+	}
+out:
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	if (delete_tree)
+		ocfs2_refcount_tree_put(ref_tree);
+	brelse(blk_bh);
+
+	return ret;
+}
+
+static void ocfs2_find_refcount_rec_in_rl(struct ocfs2_caching_info *ci,
+					  struct buffer_head *ref_leaf_bh,
+					  u64 cpos, unsigned int len,
+					  struct ocfs2_refcount_rec *ret_rec,
+					  int *index)
+{
+	int i = 0;
+	struct ocfs2_refcount_block *rb =
+		(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_rec *rec = NULL;
+
+	for (; i < le16_to_cpu(rb->rf_records.rl_used); i++) {
+		rec = &rb->rf_records.rl_recs[i];
+
+		if (le64_to_cpu(rec->r_cpos) +
+		    le32_to_cpu(rec->r_clusters) <= cpos)
+			continue;
+		else if (le64_to_cpu(rec->r_cpos) > cpos)
+			break;
+
+		/* ok, cpos fail in this rec. Just return. */
+		if (ret_rec)
+			*ret_rec = *rec;
+		goto out;
+	}
+
+	if (ret_rec) {
+		/* We meet with a hole here, so fake the rec. */
+		ret_rec->r_cpos = cpu_to_le64(cpos);
+		ret_rec->r_refcount = 0;
+		if (i < le16_to_cpu(rb->rf_records.rl_used) &&
+		    le64_to_cpu(rec->r_cpos) < cpos + len)
+			ret_rec->r_clusters =
+				cpu_to_le32(le64_to_cpu(rec->r_cpos) - cpos);
+		else
+			ret_rec->r_clusters = cpu_to_le32(len);
+	}
+
+out:
+	*index = i;
+}
+
+/*
+ * Try to remove refcount tree. The mechanism is:
+ * 1) Check whether i_clusters == 0, if no, exit.
+ * 2) check whether we have i_xattr_loc in dinode. if yes, exit.
+ * 3) Check whether we have inline xattr stored outside, if yes, exit.
+ * 4) Remove the tree.
+ */
+int ocfs2_try_remove_refcount_tree(struct inode *inode,
+				   struct buffer_head *di_bh)
+{
+	int ret;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+	down_write(&oi->ip_xattr_sem);
+	down_write(&oi->ip_alloc_sem);
+
+	if (oi->ip_clusters)
+		goto out;
+
+	if ((oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) && di->i_xattr_loc)
+		goto out;
+
+	if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL &&
+	    ocfs2_has_inline_xattr_value_outside(inode, di))
+		goto out;
+
+	ret = ocfs2_remove_refcount_tree(inode, di_bh);
+	if (ret)
+		mlog_errno(ret);
+out:
+	up_write(&oi->ip_alloc_sem);
+	up_write(&oi->ip_xattr_sem);
+	return 0;
+}
+
+/*
+ * Given a cpos and len, try to find the refcount record which contains cpos.
+ * 1. If cpos can be found in one refcount record, return the record.
+ * 2. If cpos can't be found, return a fake record which start from cpos
+ *    and end at a small value between cpos+len and start of the next record.
+ *    This fake record has r_refcount = 0.
+ */
+static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
+				  struct buffer_head *ref_root_bh,
+				  u64 cpos, unsigned int len,
+				  struct ocfs2_refcount_rec *ret_rec,
+				  int *index,
+				  struct buffer_head **ret_bh)
+{
+	int ret = 0, i, found;
+	u32 low_cpos;
+	struct ocfs2_extent_list *el;
+	struct ocfs2_extent_rec *tmp, *rec = NULL;
+	struct ocfs2_extent_block *eb;
+	struct buffer_head *eb_bh = NULL, *ref_leaf_bh = NULL;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+	if (!(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)) {
+		ocfs2_find_refcount_rec_in_rl(ci, ref_root_bh, cpos, len,
+					      ret_rec, index);
+		*ret_bh = ref_root_bh;
+		get_bh(ref_root_bh);
+		return 0;
+	}
+
+	el = &rb->rf_list;
+	low_cpos = cpos & OCFS2_32BIT_POS_MASK;
+
+	if (el->l_tree_depth) {
+		ret = ocfs2_find_leaf(ci, el, low_cpos, &eb_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
+		el = &eb->h_list;
+
+		if (el->l_tree_depth) {
+			ocfs2_error(sb,
+			"refcount tree %llu has non zero tree "
+			"depth in leaf btree tree block %llu\n",
+			(unsigned long long)ocfs2_metadata_cache_owner(ci),
+			(unsigned long long)eb_bh->b_blocknr);
+			ret = -EROFS;
+			goto out;
+		}
+	}
+
+	found = 0;
+	for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
+		rec = &el->l_recs[i];
+
+		if (le32_to_cpu(rec->e_cpos) <= low_cpos) {
+			found = 1;
+			break;
+		}
+	}
+
+	/* adjust len when we have ocfs2_extent_rec after it. */
+	if (found && i < le16_to_cpu(el->l_next_free_rec) - 1) {
+		tmp = &el->l_recs[i+1];
+
+		if (le32_to_cpu(tmp->e_cpos) < cpos + len)
+			len = le32_to_cpu(tmp->e_cpos) - cpos;
+	}
+
+	ret = ocfs2_read_refcount_block(ci, le64_to_cpu(rec->e_blkno),
+					&ref_leaf_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ocfs2_find_refcount_rec_in_rl(ci, ref_leaf_bh, cpos, len,
+				      ret_rec, index);
+	*ret_bh = ref_leaf_bh;
+out:
+	brelse(eb_bh);
+	return ret;
+}
+
+enum ocfs2_ref_rec_contig {
+	REF_CONTIG_NONE = 0,
+	REF_CONTIG_LEFT,
+	REF_CONTIG_RIGHT,
+	REF_CONTIG_LEFTRIGHT,
+};
+
+static enum ocfs2_ref_rec_contig
+	ocfs2_refcount_rec_adjacent(struct ocfs2_refcount_block *rb,
+				    int index)
+{
+	if ((rb->rf_records.rl_recs[index].r_refcount ==
+	    rb->rf_records.rl_recs[index + 1].r_refcount) &&
+	    (le64_to_cpu(rb->rf_records.rl_recs[index].r_cpos) +
+	    le32_to_cpu(rb->rf_records.rl_recs[index].r_clusters) ==
+	    le64_to_cpu(rb->rf_records.rl_recs[index + 1].r_cpos)))
+		return REF_CONTIG_RIGHT;
+
+	return REF_CONTIG_NONE;
+}
+
+static enum ocfs2_ref_rec_contig
+	ocfs2_refcount_rec_contig(struct ocfs2_refcount_block *rb,
+				  int index)
+{
+	enum ocfs2_ref_rec_contig ret = REF_CONTIG_NONE;
+
+	if (index < le16_to_cpu(rb->rf_records.rl_used) - 1)
+		ret = ocfs2_refcount_rec_adjacent(rb, index);
+
+	if (index > 0) {
+		enum ocfs2_ref_rec_contig tmp;
+
+		tmp = ocfs2_refcount_rec_adjacent(rb, index - 1);
+
+		if (tmp == REF_CONTIG_RIGHT) {
+			if (ret == REF_CONTIG_RIGHT)
+				ret = REF_CONTIG_LEFTRIGHT;
+			else
+				ret = REF_CONTIG_LEFT;
+		}
+	}
+
+	return ret;
+}
+
+static void ocfs2_rotate_refcount_rec_left(struct ocfs2_refcount_block *rb,
+					   int index)
+{
+	BUG_ON(rb->rf_records.rl_recs[index].r_refcount !=
+	       rb->rf_records.rl_recs[index+1].r_refcount);
+
+	le32_add_cpu(&rb->rf_records.rl_recs[index].r_clusters,
+		     le32_to_cpu(rb->rf_records.rl_recs[index+1].r_clusters));
+
+	if (index < le16_to_cpu(rb->rf_records.rl_used) - 2)
+		memmove(&rb->rf_records.rl_recs[index + 1],
+			&rb->rf_records.rl_recs[index + 2],
+			sizeof(struct ocfs2_refcount_rec) *
+			(le16_to_cpu(rb->rf_records.rl_used) - index - 2));
+
+	memset(&rb->rf_records.rl_recs[le16_to_cpu(rb->rf_records.rl_used) - 1],
+	       0, sizeof(struct ocfs2_refcount_rec));
+	le16_add_cpu(&rb->rf_records.rl_used, -1);
+}
+
+/*
+ * Merge the refcount rec if we are contiguous with the adjacent recs.
+ */
+static void ocfs2_refcount_rec_merge(struct ocfs2_refcount_block *rb,
+				     int index)
+{
+	enum ocfs2_ref_rec_contig contig =
+				ocfs2_refcount_rec_contig(rb, index);
+
+	if (contig == REF_CONTIG_NONE)
+		return;
+
+	if (contig == REF_CONTIG_LEFT || contig == REF_CONTIG_LEFTRIGHT) {
+		BUG_ON(index == 0);
+		index--;
+	}
+
+	ocfs2_rotate_refcount_rec_left(rb, index);
+
+	if (contig == REF_CONTIG_LEFTRIGHT)
+		ocfs2_rotate_refcount_rec_left(rb, index);
+}
+
+/*
+ * Change the refcount indexed by "index" in ref_bh.
+ * If refcount reaches 0, remove it.
+ */
+static int ocfs2_change_refcount_rec(handle_t *handle,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *ref_leaf_bh,
+				     int index, int merge, int change)
+{
+	int ret;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_list *rl = &rb->rf_records;
+	struct ocfs2_refcount_rec *rec = &rl->rl_recs[index];
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "change index %d, old count %u, change %d\n", index,
+	     le32_to_cpu(rec->r_refcount), change);
+	le32_add_cpu(&rec->r_refcount, change);
+
+	if (!rec->r_refcount) {
+		if (index != le16_to_cpu(rl->rl_used) - 1) {
+			memmove(rec, rec + 1,
+				(le16_to_cpu(rl->rl_used) - index - 1) *
+				sizeof(struct ocfs2_refcount_rec));
+			memset(&rl->rl_recs[le16_to_cpu(rl->rl_used) - 1],
+			       0, sizeof(struct ocfs2_refcount_rec));
+		}
+
+		le16_add_cpu(&rl->rl_used, -1);
+	} else if (merge)
+		ocfs2_refcount_rec_merge(rb, index);
+
+	ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
+	if (ret)
+		mlog_errno(ret);
+out:
+	return ret;
+}
+
+static int ocfs2_expand_inline_ref_root(handle_t *handle,
+					struct ocfs2_caching_info *ci,
+					struct buffer_head *ref_root_bh,
+					struct buffer_head **ref_leaf_bh,
+					struct ocfs2_alloc_context *meta_ac)
+{
+	int ret;
+	u16 suballoc_bit_start;
+	u32 num_got;
+	u64 blkno;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct buffer_head *new_bh = NULL;
+	struct ocfs2_refcount_block *new_rb;
+	struct ocfs2_refcount_block *root_rb =
+			(struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_claim_metadata(OCFS2_SB(sb), handle, meta_ac, 1,
+				   &suballoc_bit_start, &num_got,
+				   &blkno);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	new_bh = sb_getblk(sb, blkno);
+	if (new_bh == NULL) {
+		ret = -EIO;
+		mlog_errno(ret);
+		goto out;
+	}
+	ocfs2_set_new_buffer_uptodate(ci, new_bh);
+
+	ret = ocfs2_journal_access_rb(handle, ci, new_bh,
+				      OCFS2_JOURNAL_ACCESS_CREATE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * Initialize ocfs2_refcount_block.
+	 * It should contain the same information as the old root.
+	 * so just memcpy it and change the corresponding field.
+	 */
+	memcpy(new_bh->b_data, ref_root_bh->b_data, sb->s_blocksize);
+
+	new_rb = (struct ocfs2_refcount_block *)new_bh->b_data;
+	new_rb->rf_suballoc_slot = cpu_to_le16(OCFS2_SB(sb)->slot_num);
+	new_rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
+	new_rb->rf_blkno = cpu_to_le64(blkno);
+	new_rb->rf_cpos = cpu_to_le32(0);
+	new_rb->rf_parent = cpu_to_le64(ref_root_bh->b_blocknr);
+	new_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_LEAF_FL);
+	ocfs2_journal_dirty(handle, new_bh);
+
+	/* Now change the root. */
+	memset(&root_rb->rf_list, 0, sb->s_blocksize -
+	       offsetof(struct ocfs2_refcount_block, rf_list));
+	root_rb->rf_list.l_count = cpu_to_le16(ocfs2_extent_recs_per_rb(sb));
+	root_rb->rf_clusters = cpu_to_le32(1);
+	root_rb->rf_list.l_next_free_rec = cpu_to_le16(1);
+	root_rb->rf_list.l_recs[0].e_blkno = cpu_to_le64(blkno);
+	root_rb->rf_list.l_recs[0].e_leaf_clusters = cpu_to_le16(1);
+	root_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_TREE_FL);
+
+	ocfs2_journal_dirty(handle, ref_root_bh);
+
+	mlog(0, "new leaf block %llu, used %u\n", (unsigned long long)blkno,
+	     le16_to_cpu(new_rb->rf_records.rl_used));
+
+	*ref_leaf_bh = new_bh;
+	new_bh = NULL;
+out:
+	brelse(new_bh);
+	return ret;
+}
+
+static int ocfs2_refcount_rec_no_intersect(struct ocfs2_refcount_rec *prev,
+					   struct ocfs2_refcount_rec *next)
+{
+	if (ocfs2_get_ref_rec_low_cpos(prev) + le32_to_cpu(prev->r_clusters) <=
+		ocfs2_get_ref_rec_low_cpos(next))
+		return 1;
+
+	return 0;
+}
+
+static int cmp_refcount_rec_by_low_cpos(const void *a, const void *b)
+{
+	const struct ocfs2_refcount_rec *l = a, *r = b;
+	u32 l_cpos = ocfs2_get_ref_rec_low_cpos(l);
+	u32 r_cpos = ocfs2_get_ref_rec_low_cpos(r);
+
+	if (l_cpos > r_cpos)
+		return 1;
+	if (l_cpos < r_cpos)
+		return -1;
+	return 0;
+}
+
+static int cmp_refcount_rec_by_cpos(const void *a, const void *b)
+{
+	const struct ocfs2_refcount_rec *l = a, *r = b;
+	u64 l_cpos = le64_to_cpu(l->r_cpos);
+	u64 r_cpos = le64_to_cpu(r->r_cpos);
+
+	if (l_cpos > r_cpos)
+		return 1;
+	if (l_cpos < r_cpos)
+		return -1;
+	return 0;
+}
+
+static void swap_refcount_rec(void *a, void *b, int size)
+{
+	struct ocfs2_refcount_rec *l = a, *r = b, tmp;
+
+	tmp = *(struct ocfs2_refcount_rec *)l;
+	*(struct ocfs2_refcount_rec *)l =
+			*(struct ocfs2_refcount_rec *)r;
+	*(struct ocfs2_refcount_rec *)r = tmp;
+}
+
+/*
+ * The refcount cpos are ordered by their 64bit cpos,
+ * But we will use the low 32 bit to be the e_cpos in the b-tree.
+ * So we need to make sure that this pos isn't intersected with others.
+ *
+ * Note: The refcount block is already sorted by their low 32 bit cpos,
+ *       So just try the middle pos first, and we will exit when we find
+ *       the good position.
+ */
+static int ocfs2_find_refcount_split_pos(struct ocfs2_refcount_list *rl,
+					 u32 *split_pos, int *split_index)
+{
+	int num_used = le16_to_cpu(rl->rl_used);
+	int delta, middle = num_used / 2;
+
+	for (delta = 0; delta < middle; delta++) {
+		/* Let's check delta earlier than middle */
+		if (ocfs2_refcount_rec_no_intersect(
+					&rl->rl_recs[middle - delta - 1],
+					&rl->rl_recs[middle - delta])) {
+			*split_index = middle - delta;
+			break;
+		}
+
+		/* For even counts, don't walk off the end */
+		if ((middle + delta + 1) == num_used)
+			continue;
+
+		/* Now try delta past middle */
+		if (ocfs2_refcount_rec_no_intersect(
+					&rl->rl_recs[middle + delta],
+					&rl->rl_recs[middle + delta + 1])) {
+			*split_index = middle + delta + 1;
+			break;
+		}
+	}
+
+	if (delta >= middle)
+		return -ENOSPC;
+
+	*split_pos = ocfs2_get_ref_rec_low_cpos(&rl->rl_recs[*split_index]);
+	return 0;
+}
+
+static int ocfs2_divide_leaf_refcount_block(struct buffer_head *ref_leaf_bh,
+					    struct buffer_head *new_bh,
+					    u32 *split_cpos)
+{
+	int split_index = 0, num_moved, ret;
+	u32 cpos = 0;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_list *rl = &rb->rf_records;
+	struct ocfs2_refcount_block *new_rb =
+			(struct ocfs2_refcount_block *)new_bh->b_data;
+	struct ocfs2_refcount_list *new_rl = &new_rb->rf_records;
+
+	mlog(0, "split old leaf refcount block %llu, count = %u, used = %u\n",
+	     (unsigned long long)ref_leaf_bh->b_blocknr,
+	     le32_to_cpu(rl->rl_count), le32_to_cpu(rl->rl_used));
+
+	/*
+	 * XXX: Improvement later.
+	 * If we know all the high 32 bit cpos is the same, no need to sort.
+	 *
+	 * In order to make the whole process safe, we do:
+	 * 1. sort the entries by their low 32 bit cpos first so that we can
+	 *    find the split cpos easily.
+	 * 2. call ocfs2_insert_extent to insert the new refcount block.
+	 * 3. move the refcount rec to the new block.
+	 * 4. sort the entries by their 64 bit cpos.
+	 * 5. dirty the new_rb and rb.
+	 */
+	sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
+	     sizeof(struct ocfs2_refcount_rec),
+	     cmp_refcount_rec_by_low_cpos, swap_refcount_rec);
+
+	ret = ocfs2_find_refcount_split_pos(rl, &cpos, &split_index);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	new_rb->rf_cpos = cpu_to_le32(cpos);
+
+	/* move refcount records starting from split_index to the new block. */
+	num_moved = le16_to_cpu(rl->rl_used) - split_index;
+	memcpy(new_rl->rl_recs, &rl->rl_recs[split_index],
+	       num_moved * sizeof(struct ocfs2_refcount_rec));
+
+	/*ok, remove the entries we just moved over to the other block. */
+	memset(&rl->rl_recs[split_index], 0,
+	       num_moved * sizeof(struct ocfs2_refcount_rec));
+
+	/* change old and new rl_used accordingly. */
+	le16_add_cpu(&rl->rl_used, -num_moved);
+	new_rl->rl_used = cpu_to_le32(num_moved);
+
+	sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
+	     sizeof(struct ocfs2_refcount_rec),
+	     cmp_refcount_rec_by_cpos, swap_refcount_rec);
+
+	sort(&new_rl->rl_recs, le16_to_cpu(new_rl->rl_used),
+	     sizeof(struct ocfs2_refcount_rec),
+	     cmp_refcount_rec_by_cpos, swap_refcount_rec);
+
+	*split_cpos = cpos;
+	return 0;
+}
+
+static int ocfs2_new_leaf_refcount_block(handle_t *handle,
+					 struct ocfs2_caching_info *ci,
+					 struct buffer_head *ref_root_bh,
+					 struct buffer_head *ref_leaf_bh,
+					 struct ocfs2_alloc_context *meta_ac)
+{
+	int ret;
+	u16 suballoc_bit_start;
+	u32 num_got, new_cpos;
+	u64 blkno;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct ocfs2_refcount_block *root_rb =
+			(struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	struct buffer_head *new_bh = NULL;
+	struct ocfs2_refcount_block *new_rb;
+	struct ocfs2_extent_tree ref_et;
+
+	BUG_ON(!(le32_to_cpu(root_rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL));
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_claim_metadata(OCFS2_SB(sb), handle, meta_ac, 1,
+				   &suballoc_bit_start, &num_got,
+				   &blkno);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	new_bh = sb_getblk(sb, blkno);
+	if (new_bh == NULL) {
+		ret = -EIO;
+		mlog_errno(ret);
+		goto out;
+	}
+	ocfs2_set_new_buffer_uptodate(ci, new_bh);
+
+	ret = ocfs2_journal_access_rb(handle, ci, new_bh,
+				      OCFS2_JOURNAL_ACCESS_CREATE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/* Initialize ocfs2_refcount_block. */
+	new_rb = (struct ocfs2_refcount_block *)new_bh->b_data;
+	memset(new_rb, 0, sb->s_blocksize);
+	strcpy((void *)new_rb, OCFS2_REFCOUNT_BLOCK_SIGNATURE);
+	new_rb->rf_suballoc_slot = cpu_to_le16(OCFS2_SB(sb)->slot_num);
+	new_rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
+	new_rb->rf_fs_generation = cpu_to_le32(OCFS2_SB(sb)->fs_generation);
+	new_rb->rf_blkno = cpu_to_le64(blkno);
+	new_rb->rf_parent = cpu_to_le64(ref_root_bh->b_blocknr);
+	new_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_LEAF_FL);
+	new_rb->rf_records.rl_count =
+				cpu_to_le16(ocfs2_refcount_recs_per_rb(sb));
+	new_rb->rf_generation = root_rb->rf_generation;
+
+	ret = ocfs2_divide_leaf_refcount_block(ref_leaf_bh, new_bh, &new_cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ocfs2_journal_dirty(handle, ref_leaf_bh);
+	ocfs2_journal_dirty(handle, new_bh);
+
+	ocfs2_init_refcount_extent_tree(&ref_et, ci, ref_root_bh);
+
+	mlog(0, "insert new leaf block %llu at %u\n",
+	     (unsigned long long)new_bh->b_blocknr, new_cpos);
+
+	/* Insert the new leaf block with the specific offset cpos. */
+	ret = ocfs2_insert_extent(handle, &ref_et, new_cpos, new_bh->b_blocknr,
+				  1, 0, meta_ac);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	brelse(new_bh);
+	return ret;
+}
+
+static int ocfs2_expand_refcount_tree(handle_t *handle,
+				      struct ocfs2_caching_info *ci,
+				      struct buffer_head *ref_root_bh,
+				      struct buffer_head *ref_leaf_bh,
+				      struct ocfs2_alloc_context *meta_ac)
+{
+	int ret;
+	struct buffer_head *expand_bh = NULL;
+
+	if (ref_root_bh == ref_leaf_bh) {
+		/*
+		 * the old root bh hasn't been expanded to a b-tree,
+		 * so expand it first.
+		 */
+		ret = ocfs2_expand_inline_ref_root(handle, ci, ref_root_bh,
+						   &expand_bh, meta_ac);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	} else {
+		expand_bh = ref_leaf_bh;
+		get_bh(expand_bh);
+	}
+
+
+	/* Now add a new refcount block into the tree.*/
+	ret = ocfs2_new_leaf_refcount_block(handle, ci, ref_root_bh,
+					    expand_bh, meta_ac);
+	if (ret)
+		mlog_errno(ret);
+out:
+	brelse(expand_bh);
+	return ret;
+}
+
+/*
+ * Adjust the extent rec in b-tree representing ref_leaf_bh.
+ *
+ * Only called when we have inserted a new refcount rec at index 0
+ * which means ocfs2_extent_rec.e_cpos may need some change.
+ */
+static int ocfs2_adjust_refcount_rec(handle_t *handle,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *ref_root_bh,
+				     struct buffer_head *ref_leaf_bh,
+				     struct ocfs2_refcount_rec *rec)
+{
+	int ret = 0, i;
+	u32 new_cpos, old_cpos;
+	struct ocfs2_path *path = NULL;
+	struct ocfs2_extent_tree et;
+	struct ocfs2_refcount_block *rb =
+		(struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	struct ocfs2_extent_list *el;
+
+	if (!(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL))
+		goto out;
+
+	rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	old_cpos = le32_to_cpu(rb->rf_cpos);
+	new_cpos = le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
+	if (old_cpos <= new_cpos)
+		goto out;
+
+	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
+
+	path = ocfs2_new_path_from_et(&et);
+	if (!path) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_find_path(ci, path, old_cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * 2 more credits, one for the leaf refcount block, one for
+	 * the extent block contains the extent rec.
+	 */
+	ret = ocfs2_extend_trans(handle, handle->h_buffer_credits + 2);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_eb(handle, ci, path_leaf_bh(path),
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/* change the leaf extent block first. */
+	el = path_leaf_el(path);
+
+	for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++)
+		if (le32_to_cpu(el->l_recs[i].e_cpos) == old_cpos)
+			break;
+
+	BUG_ON(i == le16_to_cpu(el->l_next_free_rec));
+
+	el->l_recs[i].e_cpos = cpu_to_le32(new_cpos);
+
+	/* change the r_cpos in the leaf block. */
+	rb->rf_cpos = cpu_to_le32(new_cpos);
+
+	ocfs2_journal_dirty(handle, path_leaf_bh(path));
+	ocfs2_journal_dirty(handle, ref_leaf_bh);
+
+out:
+	ocfs2_free_path(path);
+	return ret;
+}
+
+static int ocfs2_insert_refcount_rec(handle_t *handle,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *ref_root_bh,
+				     struct buffer_head *ref_leaf_bh,
+				     struct ocfs2_refcount_rec *rec,
+				     int index, int merge,
+				     struct ocfs2_alloc_context *meta_ac)
+{
+	int ret;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_list *rf_list = &rb->rf_records;
+	struct buffer_head *new_bh = NULL;
+
+	BUG_ON(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL);
+
+	if (rf_list->rl_used == rf_list->rl_count) {
+		u64 cpos = le64_to_cpu(rec->r_cpos);
+		u32 len = le32_to_cpu(rec->r_clusters);
+
+		ret = ocfs2_expand_refcount_tree(handle, ci, ref_root_bh,
+						 ref_leaf_bh, meta_ac);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+					     cpos, len, NULL, &index,
+					     &new_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		ref_leaf_bh = new_bh;
+		rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+		rf_list = &rb->rf_records;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (index < le16_to_cpu(rf_list->rl_used))
+		memmove(&rf_list->rl_recs[index + 1],
+			&rf_list->rl_recs[index],
+			(le16_to_cpu(rf_list->rl_used) - index) *
+			 sizeof(struct ocfs2_refcount_rec));
+
+	mlog(0, "insert refcount record start %llu, len %u, count %u "
+	     "to leaf block %llu at index %d\n",
+	     (unsigned long long)le64_to_cpu(rec->r_cpos),
+	     le32_to_cpu(rec->r_clusters), le32_to_cpu(rec->r_refcount),
+	     (unsigned long long)ref_leaf_bh->b_blocknr, index);
+
+	rf_list->rl_recs[index] = *rec;
+
+	le16_add_cpu(&rf_list->rl_used, 1);
+
+	if (merge)
+		ocfs2_refcount_rec_merge(rb, index);
+
+	ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (index == 0) {
+		ret = ocfs2_adjust_refcount_rec(handle, ci,
+						ref_root_bh,
+						ref_leaf_bh, rec);
+		if (ret)
+			mlog_errno(ret);
+	}
+out:
+	brelse(new_bh);
+	return ret;
+}
+
+/*
+ * Split the refcount_rec indexed by "index" in ref_leaf_bh.
+ * This is much simple than our b-tree code.
+ * split_rec is the new refcount rec we want to insert.
+ * If split_rec->r_refcount > 0, we are changing the refcount(in case we
+ * increase refcount or decrease a refcount to non-zero).
+ * If split_rec->r_refcount == 0, we are punching a hole in current refcount
+ * rec( in case we decrease a refcount to zero).
+ */
+static int ocfs2_split_refcount_rec(handle_t *handle,
+				    struct ocfs2_caching_info *ci,
+				    struct buffer_head *ref_root_bh,
+				    struct buffer_head *ref_leaf_bh,
+				    struct ocfs2_refcount_rec *split_rec,
+				    int index, int merge,
+				    struct ocfs2_alloc_context *meta_ac,
+				    struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret, recs_need;
+	u32 len;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_list *rf_list = &rb->rf_records;
+	struct ocfs2_refcount_rec *orig_rec = &rf_list->rl_recs[index];
+	struct ocfs2_refcount_rec *tail_rec = NULL;
+	struct buffer_head *new_bh = NULL;
+
+	BUG_ON(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL);
+
+	mlog(0, "original r_pos %llu, cluster %u, split %llu, cluster %u\n",
+	     le64_to_cpu(orig_rec->r_cpos), le32_to_cpu(orig_rec->r_clusters),
+	     le64_to_cpu(split_rec->r_cpos),
+	     le32_to_cpu(split_rec->r_clusters));
+
+	/*
+	 * If we just need to split the header or tail clusters,
+	 * no more recs are needed, just split is OK.
+	 * Otherwise we at least need one new recs.
+	 */
+	if (!split_rec->r_refcount &&
+	    (split_rec->r_cpos == orig_rec->r_cpos ||
+	     le64_to_cpu(split_rec->r_cpos) +
+	     le32_to_cpu(split_rec->r_clusters) ==
+	     le64_to_cpu(orig_rec->r_cpos) + le32_to_cpu(orig_rec->r_clusters)))
+		recs_need = 0;
+	else
+		recs_need = 1;
+
+	/*
+	 * We need one more rec if we split in the middle and the new rec have
+	 * some refcount in it.
+	 */
+	if (split_rec->r_refcount &&
+	    (split_rec->r_cpos != orig_rec->r_cpos &&
+	     le64_to_cpu(split_rec->r_cpos) +
+	     le32_to_cpu(split_rec->r_clusters) !=
+	     le64_to_cpu(orig_rec->r_cpos) + le32_to_cpu(orig_rec->r_clusters)))
+		recs_need++;
+
+	/* If the leaf block don't have enough record, expand it. */
+	if (le16_to_cpu(rf_list->rl_used) + recs_need > rf_list->rl_count) {
+		struct ocfs2_refcount_rec tmp_rec;
+		u64 cpos = le64_to_cpu(orig_rec->r_cpos);
+		len = le32_to_cpu(orig_rec->r_clusters);
+		ret = ocfs2_expand_refcount_tree(handle, ci, ref_root_bh,
+						 ref_leaf_bh, meta_ac);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		/*
+		 * We have to re-get it since now cpos may be moved to
+		 * another leaf block.
+		 */
+		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+					     cpos, len, &tmp_rec, &index,
+					     &new_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		ref_leaf_bh = new_bh;
+		rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+		rf_list = &rb->rf_records;
+		orig_rec = &rf_list->rl_recs[index];
+	}
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * We have calculated out how many new records we need and store
+	 * in recs_need, so spare enough space first by moving the records
+	 * after "index" to the end.
+	 */
+	if (index != le16_to_cpu(rf_list->rl_used) - 1)
+		memmove(&rf_list->rl_recs[index + 1 + recs_need],
+			&rf_list->rl_recs[index + 1],
+			(le16_to_cpu(rf_list->rl_used) - index - 1) *
+			 sizeof(struct ocfs2_refcount_rec));
+
+	len = (le64_to_cpu(orig_rec->r_cpos) +
+	      le32_to_cpu(orig_rec->r_clusters)) -
+	      (le64_to_cpu(split_rec->r_cpos) +
+	      le32_to_cpu(split_rec->r_clusters));
+
+	/*
+	 * If we have "len", the we will split in the tail and move it
+	 * to the end of the space we have just spared.
+	 */
+	if (len) {
+		tail_rec = &rf_list->rl_recs[index + recs_need];
+
+		memcpy(tail_rec, orig_rec, sizeof(struct ocfs2_refcount_rec));
+		le64_add_cpu(&tail_rec->r_cpos,
+			     le32_to_cpu(tail_rec->r_clusters) - len);
+		tail_rec->r_clusters = le32_to_cpu(len);
+	}
+
+	/*
+	 * If the split pos isn't the same as the original one, we need to
+	 * split in the head.
+	 *
+	 * Note: We have the chance that split_rec.r_refcount = 0,
+	 * recs_need = 0 and len > 0, which means we just cut the head from
+	 * the orig_rec and in that case we have done some modification in
+	 * orig_rec above, so the check for r_cpos is faked.
+	 */
+	if (split_rec->r_cpos != orig_rec->r_cpos && tail_rec != orig_rec) {
+		len = le64_to_cpu(split_rec->r_cpos) -
+		      le64_to_cpu(orig_rec->r_cpos);
+		orig_rec->r_clusters = cpu_to_le32(len);
+		index++;
+	}
+
+	le16_add_cpu(&rf_list->rl_used, recs_need);
+
+	if (split_rec->r_refcount) {
+		rf_list->rl_recs[index] = *split_rec;
+		mlog(0, "insert refcount record start %llu, len %u, count %u "
+		     "to leaf block %llu at index %d\n",
+		     (unsigned long long)le64_to_cpu(split_rec->r_cpos),
+		     le32_to_cpu(split_rec->r_clusters),
+		     le32_to_cpu(split_rec->r_refcount),
+		     (unsigned long long)ref_leaf_bh->b_blocknr, index);
+
+		if (merge)
+			ocfs2_refcount_rec_merge(rb, index);
+	}
+
+	ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	brelse(new_bh);
+	return ret;
+}
+
+static int __ocfs2_increase_refcount(handle_t *handle,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *ref_root_bh,
+				     u64 cpos, u32 len, int merge,
+				     struct ocfs2_alloc_context *meta_ac,
+				     struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret = 0, index;
+	struct buffer_head *ref_leaf_bh = NULL;
+	struct ocfs2_refcount_rec rec;
+	unsigned int set_len = 0;
+
+	mlog(0, "Tree owner %llu, add refcount start %llu, len %u\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
+	     (unsigned long long)cpos, len);
+
+	while (len) {
+		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+					     cpos, len, &rec, &index,
+					     &ref_leaf_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		set_len = le32_to_cpu(rec.r_clusters);
+
+		/*
+		 * Here we may meet with 3 situations:
+		 *
+		 * 1. If we find an already existing record, and the length
+		 *    is the same, cool, we just need to increase the r_refcount
+		 *    and it is OK.
+		 * 2. If we find a hole, just insert it with r_refcount = 1.
+		 * 3. If we are in the middle of one extent record, split
+		 *    it.
+		 */
+		if (rec.r_refcount && le64_to_cpu(rec.r_cpos) == cpos &&
+		    set_len <= len) {
+			mlog(0, "increase refcount rec, start %llu, len %u, "
+			     "count %u\n", (unsigned long long)cpos, set_len,
+			     le32_to_cpu(rec.r_refcount));
+			ret = ocfs2_change_refcount_rec(handle, ci,
+							ref_leaf_bh, index,
+							merge, 1);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+		} else if (!rec.r_refcount) {
+			rec.r_refcount = cpu_to_le32(1);
+
+			mlog(0, "insert refcount rec, start %llu, len %u\n",
+			     (unsigned long long)le64_to_cpu(rec.r_cpos),
+			     set_len);
+			ret = ocfs2_insert_refcount_rec(handle, ci, ref_root_bh,
+							ref_leaf_bh,
+							&rec, index,
+							merge, meta_ac);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+		} else  {
+			set_len = min((u64)(cpos + len),
+				      le64_to_cpu(rec.r_cpos) + set_len) - cpos;
+			rec.r_cpos = cpu_to_le64(cpos);
+			rec.r_clusters = cpu_to_le32(set_len);
+			le32_add_cpu(&rec.r_refcount, 1);
+
+			mlog(0, "split refcount rec, start %llu, "
+			     "len %u, count %u\n",
+			     (unsigned long long)le64_to_cpu(rec.r_cpos),
+			     set_len, le32_to_cpu(rec.r_refcount));
+			ret = ocfs2_split_refcount_rec(handle, ci,
+						       ref_root_bh, ref_leaf_bh,
+						       &rec, index, merge,
+						       meta_ac, dealloc);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+		}
+
+		cpos += set_len;
+		len -= set_len;
+		brelse(ref_leaf_bh);
+		ref_leaf_bh = NULL;
+	}
+
+out:
+	brelse(ref_leaf_bh);
+	return ret;
+}
+
+static int ocfs2_remove_refcount_extent(handle_t *handle,
+				struct ocfs2_caching_info *ci,
+				struct buffer_head *ref_root_bh,
+				struct buffer_head *ref_leaf_bh,
+				struct ocfs2_alloc_context *meta_ac,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_extent_tree et;
+
+	BUG_ON(rb->rf_records.rl_used);
+
+	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
+	ret = ocfs2_remove_extent(handle, &et, le32_to_cpu(rb->rf_cpos),
+				  1, meta_ac, dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ocfs2_remove_from_cache(ci, ref_leaf_bh);
+
+	/*
+	 * add the freed block to the dealloc so that it will be freed
+	 * when we run dealloc.
+	 */
+	ret = ocfs2_cache_block_dealloc(dealloc, EXTENT_ALLOC_SYSTEM_INODE,
+					le16_to_cpu(rb->rf_suballoc_slot),
+					le64_to_cpu(rb->rf_blkno),
+					le16_to_cpu(rb->rf_suballoc_bit));
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+	le32_add_cpu(&rb->rf_clusters, -1);
+
+	/*
+	 * check whether we need to restore the root refcount block if
+	 * there is no leaf extent block at atll.
+	 */
+	if (!rb->rf_list.l_next_free_rec) {
+		BUG_ON(rb->rf_clusters);
+
+		mlog(0, "reset refcount tree root %llu to be a record block.\n",
+		     (unsigned long long)ref_root_bh->b_blocknr);
+
+		rb->rf_flags = 0;
+		rb->rf_parent = 0;
+		rb->rf_cpos = 0;
+		memset(&rb->rf_records, 0, sb->s_blocksize -
+		       offsetof(struct ocfs2_refcount_block, rf_records));
+		rb->rf_records.rl_count =
+				cpu_to_le16(ocfs2_refcount_recs_per_rb(sb));
+	}
+
+	ocfs2_journal_dirty(handle, ref_root_bh);
+
+out:
+	return ret;
+}
+
+int ocfs2_increase_refcount(handle_t *handle,
+			    struct ocfs2_caching_info *ci,
+			    struct buffer_head *ref_root_bh,
+			    u64 cpos, u32 len,
+			    struct ocfs2_alloc_context *meta_ac,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	return __ocfs2_increase_refcount(handle, ci, ref_root_bh,
+					 cpos, len, 1,
+					 meta_ac, dealloc);
+}
+
+static int ocfs2_decrease_refcount_rec(handle_t *handle,
+				struct ocfs2_caching_info *ci,
+				struct buffer_head *ref_root_bh,
+				struct buffer_head *ref_leaf_bh,
+				int index, u64 cpos, unsigned int len,
+				struct ocfs2_alloc_context *meta_ac,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+	struct ocfs2_refcount_rec *rec = &rb->rf_records.rl_recs[index];
+
+	BUG_ON(cpos < le64_to_cpu(rec->r_cpos));
+	BUG_ON(cpos + len >
+	       le64_to_cpu(rec->r_cpos) + le32_to_cpu(rec->r_clusters));
+
+	if (cpos == le64_to_cpu(rec->r_cpos) &&
+	    len == le32_to_cpu(rec->r_clusters))
+		ret = ocfs2_change_refcount_rec(handle, ci,
+						ref_leaf_bh, index, 1, -1);
+	else {
+		struct ocfs2_refcount_rec split = *rec;
+		split.r_cpos = cpu_to_le64(cpos);
+		split.r_clusters = cpu_to_le32(len);
+
+		le32_add_cpu(&split.r_refcount, -1);
+
+		mlog(0, "split refcount rec, start %llu, "
+		     "len %u, count %u, original start %llu, len %u\n",
+		     (unsigned long long)le64_to_cpu(split.r_cpos),
+		     len, le32_to_cpu(split.r_refcount),
+		     (unsigned long long)le64_to_cpu(rec->r_cpos),
+		     le32_to_cpu(rec->r_clusters));
+		ret = ocfs2_split_refcount_rec(handle, ci,
+					       ref_root_bh, ref_leaf_bh,
+					       &split, index, 1,
+					       meta_ac, dealloc);
+	}
+
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/* Remove the leaf refcount block if it contains no refcount record. */
+	if (!rb->rf_records.rl_used && ref_leaf_bh != ref_root_bh) {
+		ret = ocfs2_remove_refcount_extent(handle, ci, ref_root_bh,
+						   ref_leaf_bh, meta_ac,
+						   dealloc);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out:
+	return ret;
+}
+
+static int __ocfs2_decrease_refcount(handle_t *handle,
+				     struct ocfs2_caching_info *ci,
+				     struct buffer_head *ref_root_bh,
+				     u64 cpos, u32 len,
+				     struct ocfs2_alloc_context *meta_ac,
+				     struct ocfs2_cached_dealloc_ctxt *dealloc,
+				     int delete)
+{
+	int ret = 0, index = 0;
+	struct ocfs2_refcount_rec rec;
+	unsigned int r_count = 0, r_len;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct buffer_head *ref_leaf_bh = NULL;
+
+	mlog(0, "Tree owner %llu, decrease refcount start %llu, "
+	     "len %u, delete %u\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
+	     (unsigned long long)cpos, len, delete);
+
+	while (len) {
+		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+					     cpos, len, &rec, &index,
+					     &ref_leaf_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		r_count = le32_to_cpu(rec.r_refcount);
+		BUG_ON(r_count == 0);
+		if (!delete)
+			BUG_ON(r_count > 1);
+
+		r_len = min((u64)(cpos + len), le64_to_cpu(rec.r_cpos) +
+			      le32_to_cpu(rec.r_clusters)) - cpos;
+
+		ret = ocfs2_decrease_refcount_rec(handle, ci, ref_root_bh,
+						  ref_leaf_bh, index,
+						  cpos, r_len,
+						  meta_ac, dealloc);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		if (le32_to_cpu(rec.r_refcount) == 1 && delete) {
+			ret = ocfs2_cache_cluster_dealloc(dealloc,
+					  ocfs2_clusters_to_blocks(sb, cpos),
+							  r_len);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+		}
+
+		cpos += r_len;
+		len -= r_len;
+		brelse(ref_leaf_bh);
+		ref_leaf_bh = NULL;
+	}
+
+out:
+	brelse(ref_leaf_bh);
+	return ret;
+}
+
+/* Caller must hold refcount tree lock. */
+int ocfs2_decrease_refcount(struct inode *inode,
+			    handle_t *handle, u32 cpos, u32 len,
+			    struct ocfs2_alloc_context *meta_ac,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc,
+			    int delete)
+{
+	int ret;
+	u64 ref_blkno;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_tree *tree;
+
+	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+	ret = ocfs2_get_refcount_block(inode, &ref_blkno);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb), ref_blkno, &tree);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_read_refcount_block(&tree->rf_ci, tree->rf_blkno,
+					&ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = __ocfs2_decrease_refcount(handle, &tree->rf_ci, ref_root_bh,
+					cpos, len, meta_ac, dealloc, delete);
+	if (ret)
+		mlog_errno(ret);
+out:
+	brelse(ref_root_bh);
+	return ret;
+}
+
+/*
+ * Mark the already-existing extent at cpos as refcounted for len clusters.
+ * This adds the refcount extent flag.
+ *
+ * If the existing extent is larger than the request, initiate a
+ * split. An attempt will be made at merging with adjacent extents.
+ *
+ * The caller is responsible for passing down meta_ac if we'll need it.
+ */
+static int ocfs2_mark_extent_refcounted(struct inode *inode,
+				struct ocfs2_extent_tree *et,
+				handle_t *handle, u32 cpos,
+				u32 len, u32 phys,
+				struct ocfs2_alloc_context *meta_ac,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret;
+
+	mlog(0, "Inode %lu refcount tree cpos %u, len %u, phys cluster %u\n",
+	     inode->i_ino, cpos, len, phys);
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+			    "tree, but the feature bit is not set in the "
+			    "super block.", inode->i_ino);
+		ret = -EROFS;
+		goto out;
+	}
+
+	ret = ocfs2_change_extent_flag(handle, et, cpos,
+				       len, phys, meta_ac, dealloc,
+				       OCFS2_EXT_REFCOUNTED, 0);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	return ret;
+}
+
+/*
+ * Given some contiguous physical clusters, calculate what we need
+ * for modifying their refcount.
+ */
+static int ocfs2_calc_refcount_meta_credits(struct super_block *sb,
+					    struct ocfs2_caching_info *ci,
+					    struct buffer_head *ref_root_bh,
+					    u64 start_cpos,
+					    u32 clusters,
+					    int *meta_add,
+					    int *credits)
+{
+	int ret = 0, index, ref_blocks = 0, recs_add = 0;
+	u64 cpos = start_cpos;
+	struct ocfs2_refcount_block *rb;
+	struct ocfs2_refcount_rec rec;
+	struct buffer_head *ref_leaf_bh = NULL, *prev_bh = NULL;
+	u32 len;
+
+	mlog(0, "start_cpos %llu, clusters %u\n",
+	     (unsigned long long)start_cpos, clusters);
+	while (clusters) {
+		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+					     cpos, clusters, &rec,
+					     &index, &ref_leaf_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		if (ref_leaf_bh != prev_bh) {
+			/*
+			 * Now we encounter a new leaf block, so calculate
+			 * whether we need to extend the old leaf.
+			 */
+			if (prev_bh) {
+				rb = (struct ocfs2_refcount_block *)
+							prev_bh->b_data;
+
+				if (le64_to_cpu(rb->rf_records.rl_used) +
+				    recs_add >
+				    le16_to_cpu(rb->rf_records.rl_count))
+					ref_blocks++;
+			}
+
+			recs_add = 0;
+			*credits += 1;
+			brelse(prev_bh);
+			prev_bh = ref_leaf_bh;
+			get_bh(prev_bh);
+		}
+
+		rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+
+		mlog(0, "recs_add %d,cpos %llu, clusters %u, rec->r_cpos %llu,"
+		     "rec->r_clusters %u, rec->r_refcount %u, index %d\n",
+		     recs_add, (unsigned long long)cpos, clusters,
+		     (unsigned long long)le64_to_cpu(rec.r_cpos),
+		     le32_to_cpu(rec.r_clusters),
+		     le32_to_cpu(rec.r_refcount), index);
+
+		len = min((u64)cpos + clusters, le64_to_cpu(rec.r_cpos) +
+			  le32_to_cpu(rec.r_clusters)) - cpos;
+		/*
+		 * If the refcount rec already exist, cool. We just need
+		 * to check whether there is a split. Otherwise we just need
+		 * to increase the refcount.
+		 * If we will insert one, increases recs_add.
+		 *
+		 * We record all the records which will be inserted to the
+		 * same refcount block, so that we can tell exactly whether
+		 * we need a new refcount block or not.
+		 */
+		if (rec.r_refcount) {
+			/* Check whether we need a split at the beginning. */
+			if (cpos == start_cpos &&
+			    cpos != le64_to_cpu(rec.r_cpos))
+				recs_add++;
+
+			/* Check whether we need a split in the end. */
+			if (cpos + clusters < le64_to_cpu(rec.r_cpos) +
+			    le32_to_cpu(rec.r_clusters))
+				recs_add++;
+		} else
+			recs_add++;
+
+		brelse(ref_leaf_bh);
+		ref_leaf_bh = NULL;
+		clusters -= len;
+		cpos += len;
+	}
+
+	if (prev_bh) {
+		rb = (struct ocfs2_refcount_block *)prev_bh->b_data;
+
+		if (le64_to_cpu(rb->rf_records.rl_used) + recs_add >
+		    le16_to_cpu(rb->rf_records.rl_count))
+			ref_blocks++;
+
+		*credits += 1;
+	}
+
+	if (!ref_blocks)
+		goto out;
+
+	mlog(0, "we need ref_blocks %d\n", ref_blocks);
+	*meta_add += ref_blocks;
+	*credits += ref_blocks;
+
+	/*
+	 * So we may need ref_blocks to insert into the tree.
+	 * That also means we need to change the b-tree and add that number
+	 * of records since we never merge them.
+	 * We need one more block for expansion since the new created leaf
+	 * block is also full and needs split.
+	 */
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL) {
+		struct ocfs2_extent_tree et;
+
+		ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
+		*meta_add += ocfs2_extend_meta_needed(et.et_root_el);
+		*credits += ocfs2_calc_extend_credits(sb,
+						      et.et_root_el,
+						      ref_blocks);
+	} else {
+		*credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+		*meta_add += 1;
+	}
+
+out:
+	brelse(ref_leaf_bh);
+	brelse(prev_bh);
+	return ret;
+}
+
+/*
+ * For refcount tree, we will decrease some contiguous clusters
+ * refcount count, so just go through it to see how many blocks
+ * we gonna touch and whether we need to create new blocks.
+ *
+ * Normally the refcount blocks store these refcount should be
+ * continguous also, so that we can get the number easily.
+ * As for meta_ac, we will at most add split 2 refcount record and
+ * 2 more refcount block, so just check it in a rough way.
+ *
+ * Caller must hold refcount tree lock.
+ */
+int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
+					  struct buffer_head *di_bh,
+					  u64 phys_blkno,
+					  u32 clusters,
+					  int *credits,
+					  struct ocfs2_alloc_context **meta_ac)
+{
+	int ret, ref_blocks = 0;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_tree *tree;
+	u64 start_cpos = ocfs2_blocks_to_clusters(inode->i_sb, phys_blkno);
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+			    "tree, but the feature bit is not set in the "
+			    "super block.", inode->i_ino);
+		ret = -EROFS;
+		goto out;
+	}
+
+	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+	ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb),
+				      le64_to_cpu(di->i_refcount_loc), &tree);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_read_refcount_block(&tree->rf_ci,
+					le64_to_cpu(di->i_refcount_loc),
+					&ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
+					       &tree->rf_ci,
+					       ref_root_bh,
+					       start_cpos, clusters,
+					       &ref_blocks, credits);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "reserve new metadata %d, credits = %d\n",
+	     ref_blocks, *credits);
+
+	if (ref_blocks) {
+		ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
+							ref_blocks, meta_ac);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out:
+	brelse(ref_root_bh);
+	return ret;
+}
+
+#define	MAX_CONTIG_BYTES	1048576
+
+static inline unsigned int ocfs2_cow_contig_clusters(struct super_block *sb)
+{
+	return ocfs2_clusters_for_bytes(sb, MAX_CONTIG_BYTES);
+}
+
+static inline unsigned int ocfs2_cow_contig_mask(struct super_block *sb)
+{
+	return ~(ocfs2_cow_contig_clusters(sb) - 1);
+}
+
+/*
+ * Given an extent that starts at 'start' and an I/O that starts at 'cpos',
+ * find an offset (start + (n * contig_clusters)) that is closest to cpos
+ * while still being less than or equal to it.
+ *
+ * The goal is to break the extent at a multiple of contig_clusters.
+ */
+static inline unsigned int ocfs2_cow_align_start(struct super_block *sb,
+						 unsigned int start,
+						 unsigned int cpos)
+{
+	BUG_ON(start > cpos);
+
+	return start + ((cpos - start) & ocfs2_cow_contig_mask(sb));
+}
+
+/*
+ * Given a cluster count of len, pad it out so that it is a multiple
+ * of contig_clusters.
+ */
+static inline unsigned int ocfs2_cow_align_length(struct super_block *sb,
+						  unsigned int len)
+{
+	unsigned int padded =
+		(len + (ocfs2_cow_contig_clusters(sb) - 1)) &
+		ocfs2_cow_contig_mask(sb);
+
+	/* Did we wrap? */
+	if (padded < len)
+		padded = UINT_MAX;
+
+	return padded;
+}
+
+/*
+ * Calculate out the start and number of virtual clusters we need to to CoW.
+ *
+ * cpos is vitual start cluster position we want to do CoW in a
+ * file and write_len is the cluster length.
+ * max_cpos is the place where we want to stop CoW intentionally.
+ *
+ * Normal we will start CoW from the beginning of extent record cotaining cpos.
+ * We try to break up extents on boundaries of MAX_CONTIG_BYTES so that we
+ * get good I/O from the resulting extent tree.
+ */
+static int ocfs2_refcount_cal_cow_clusters(struct inode *inode,
+					   struct ocfs2_extent_list *el,
+					   u32 cpos,
+					   u32 write_len,
+					   u32 max_cpos,
+					   u32 *cow_start,
+					   u32 *cow_len)
+{
+	int ret = 0;
+	int tree_height = le16_to_cpu(el->l_tree_depth), i;
+	struct buffer_head *eb_bh = NULL;
+	struct ocfs2_extent_block *eb = NULL;
+	struct ocfs2_extent_rec *rec;
+	unsigned int want_clusters, rec_end = 0;
+	int contig_clusters = ocfs2_cow_contig_clusters(inode->i_sb);
+	int leaf_clusters;
+
+	BUG_ON(cpos + write_len > max_cpos);
+
+	if (tree_height > 0) {
+		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, cpos, &eb_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
+		el = &eb->h_list;
+
+		if (el->l_tree_depth) {
+			ocfs2_error(inode->i_sb,
+				    "Inode %lu has non zero tree depth in "
+				    "leaf block %llu\n", inode->i_ino,
+				    (unsigned long long)eb_bh->b_blocknr);
+			ret = -EROFS;
+			goto out;
+		}
+	}
+
+	*cow_len = 0;
+	for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
+		rec = &el->l_recs[i];
+
+		if (ocfs2_is_empty_extent(rec)) {
+			mlog_bug_on_msg(i != 0, "Inode %lu has empty record in "
+					"index %d\n", inode->i_ino, i);
+			continue;
+		}
+
+		if (le32_to_cpu(rec->e_cpos) +
+		    le16_to_cpu(rec->e_leaf_clusters) <= cpos)
+			continue;
+
+		if (*cow_len == 0) {
+			/*
+			 * We should find a refcounted record in the
+			 * first pass.
+			 */
+			BUG_ON(!(rec->e_flags & OCFS2_EXT_REFCOUNTED));
+			*cow_start = le32_to_cpu(rec->e_cpos);
+		}
+
+		/*
+		 * If we encounter a hole, a non-refcounted record or
+		 * pass the max_cpos, stop the search.
+		 */
+		if ((!(rec->e_flags & OCFS2_EXT_REFCOUNTED)) ||
+		    (*cow_len && rec_end != le32_to_cpu(rec->e_cpos)) ||
+		    (max_cpos <= le32_to_cpu(rec->e_cpos)))
+			break;
+
+		leaf_clusters = le16_to_cpu(rec->e_leaf_clusters);
+		rec_end = le32_to_cpu(rec->e_cpos) + leaf_clusters;
+		if (rec_end > max_cpos) {
+			rec_end = max_cpos;
+			leaf_clusters = rec_end - le32_to_cpu(rec->e_cpos);
+		}
+
+		/*
+		 * How many clusters do we actually need from
+		 * this extent?  First we see how many we actually
+		 * need to complete the write.  If that's smaller
+		 * than contig_clusters, we try for contig_clusters.
+		 */
+		if (!*cow_len)
+			want_clusters = write_len;
+		else
+			want_clusters = (cpos + write_len) -
+				(*cow_start + *cow_len);
+		if (want_clusters < contig_clusters)
+			want_clusters = contig_clusters;
+
+		/*
+		 * If the write does not cover the whole extent, we
+		 * need to calculate how we're going to split the extent.
+		 * We try to do it on contig_clusters boundaries.
+		 *
+		 * Any extent smaller than contig_clusters will be
+		 * CoWed in its entirety.
+		 */
+		if (leaf_clusters <= contig_clusters)
+			*cow_len += leaf_clusters;
+		else if (*cow_len || (*cow_start == cpos)) {
+			/*
+			 * This extent needs to be CoW'd from its
+			 * beginning, so all we have to do is compute
+			 * how many clusters to grab.  We align
+			 * want_clusters to the edge of contig_clusters
+			 * to get better I/O.
+			 */
+			want_clusters = ocfs2_cow_align_length(inode->i_sb,
+							       want_clusters);
+
+			if (leaf_clusters < want_clusters)
+				*cow_len += leaf_clusters;
+			else
+				*cow_len += want_clusters;
+		} else if ((*cow_start + contig_clusters) >=
+			   (cpos + write_len)) {
+			/*
+			 * Breaking off contig_clusters at the front
+			 * of the extent will cover our write.  That's
+			 * easy.
+			 */
+			*cow_len = contig_clusters;
+		} else if ((rec_end - cpos) <= contig_clusters) {
+			/*
+			 * Breaking off contig_clusters at the tail of
+			 * this extent will cover cpos.
+			 */
+			*cow_start = rec_end - contig_clusters;
+			*cow_len = contig_clusters;
+		} else if ((rec_end - cpos) <= want_clusters) {
+			/*
+			 * While we can't fit the entire write in this
+			 * extent, we know that the write goes from cpos
+			 * to the end of the extent.  Break that off.
+			 * We try to break it at some multiple of
+			 * contig_clusters from the front of the extent.
+			 * Failing that (ie, cpos is within
+			 * contig_clusters of the front), we'll CoW the
+			 * entire extent.
+			 */
+			*cow_start = ocfs2_cow_align_start(inode->i_sb,
+							   *cow_start, cpos);
+			*cow_len = rec_end - *cow_start;
+		} else {
+			/*
+			 * Ok, the entire write lives in the middle of
+			 * this extent.  Let's try to slice the extent up
+			 * nicely.  Optimally, our CoW region starts at
+			 * m*contig_clusters from the beginning of the
+			 * extent and goes for n*contig_clusters,
+			 * covering the entire write.
+			 */
+			*cow_start = ocfs2_cow_align_start(inode->i_sb,
+							   *cow_start, cpos);
+
+			want_clusters = (cpos + write_len) - *cow_start;
+			want_clusters = ocfs2_cow_align_length(inode->i_sb,
+							       want_clusters);
+			if (*cow_start + want_clusters <= rec_end)
+				*cow_len = want_clusters;
+			else
+				*cow_len = rec_end - *cow_start;
+		}
+
+		/* Have we covered our entire write yet? */
+		if ((*cow_start + *cow_len) >= (cpos + write_len))
+			break;
+
+		/*
+		 * If we reach the end of the extent block and don't get enough
+		 * clusters, continue with the next extent block if possible.
+		 */
+		if (i + 1 == le16_to_cpu(el->l_next_free_rec) &&
+		    eb && eb->h_next_leaf_blk) {
+			brelse(eb_bh);
+			eb_bh = NULL;
+
+			ret = ocfs2_read_extent_block(INODE_CACHE(inode),
+					       le64_to_cpu(eb->h_next_leaf_blk),
+					       &eb_bh);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+
+			eb = (struct ocfs2_extent_block *) eb_bh->b_data;
+			el = &eb->h_list;
+			i = -1;
+		}
+	}
+
+out:
+	brelse(eb_bh);
+	return ret;
+}
+
+/*
+ * Prepare meta_ac, data_ac and calculate credits when we want to add some
+ * num_clusters in data_tree "et" and change the refcount for the old
+ * clusters(starting form p_cluster) in the refcount tree.
+ *
+ * Note:
+ * 1. since we may split the old tree, so we at most will need num_clusters + 2
+ *    more new leaf records.
+ * 2. In some case, we may not need to reserve new clusters(e.g, reflink), so
+ *    just give data_ac = NULL.
+ */
+static int ocfs2_lock_refcount_allocators(struct super_block *sb,
+					u32 p_cluster, u32 num_clusters,
+					struct ocfs2_extent_tree *et,
+					struct ocfs2_caching_info *ref_ci,
+					struct buffer_head *ref_root_bh,
+					struct ocfs2_alloc_context **meta_ac,
+					struct ocfs2_alloc_context **data_ac,
+					int *credits)
+{
+	int ret = 0, meta_add = 0;
+	int num_free_extents = ocfs2_num_free_extents(OCFS2_SB(sb), et);
+
+	if (num_free_extents < 0) {
+		ret = num_free_extents;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (num_free_extents < num_clusters + 2)
+		meta_add =
+			ocfs2_extend_meta_needed(et->et_root_el);
+
+	*credits += ocfs2_calc_extend_credits(sb, et->et_root_el,
+					      num_clusters + 2);
+
+	ret = ocfs2_calc_refcount_meta_credits(sb, ref_ci, ref_root_bh,
+					       p_cluster, num_clusters,
+					       &meta_add, credits);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "reserve new metadata %d, clusters %u, credits = %d\n",
+	     meta_add, num_clusters, *credits);
+	ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(sb), meta_add,
+						meta_ac);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (data_ac) {
+		ret = ocfs2_reserve_clusters(OCFS2_SB(sb), num_clusters,
+					     data_ac);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out:
+	if (ret) {
+		if (*meta_ac) {
+			ocfs2_free_alloc_context(*meta_ac);
+			*meta_ac = NULL;
+		}
+	}
+
+	return ret;
+}
+
+static int ocfs2_clear_cow_buffer(handle_t *handle, struct buffer_head *bh)
+{
+	BUG_ON(buffer_dirty(bh));
+
+	clear_buffer_mapped(bh);
+
+	return 0;
+}
+
+static int ocfs2_duplicate_clusters_by_page(handle_t *handle,
+					    struct ocfs2_cow_context *context,
+					    u32 cpos, u32 old_cluster,
+					    u32 new_cluster, u32 new_len)
+{
+	int ret = 0, partial;
+	struct ocfs2_caching_info *ci = context->data_et.et_ci;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
+	struct page *page;
+	pgoff_t page_index;
+	unsigned int from, to;
+	loff_t offset, end, map_end;
+	struct address_space *mapping = context->inode->i_mapping;
+
+	mlog(0, "old_cluster %u, new %u, len %u at offset %u\n", old_cluster,
+	     new_cluster, new_len, cpos);
+
+	offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
+	end = offset + (new_len << OCFS2_SB(sb)->s_clustersize_bits);
+
+	while (offset < end) {
+		page_index = offset >> PAGE_CACHE_SHIFT;
+		map_end = (page_index + 1) << PAGE_CACHE_SHIFT;
+		if (map_end > end)
+			map_end = end;
+
+		/* from, to is the offset within the page. */
+		from = offset & (PAGE_CACHE_SIZE - 1);
+		to = PAGE_CACHE_SIZE;
+		if (map_end & (PAGE_CACHE_SIZE - 1))
+			to = map_end & (PAGE_CACHE_SIZE - 1);
+
+		page = grab_cache_page(mapping, page_index);
+
+		/* This page can't be dirtied before we CoW it out. */
+		BUG_ON(PageDirty(page));
+
+		if (!PageUptodate(page)) {
+			ret = block_read_full_page(page, ocfs2_get_block);
+			if (ret) {
+				mlog_errno(ret);
+				goto unlock;
+			}
+			lock_page(page);
+		}
+
+		if (page_has_buffers(page)) {
+			ret = walk_page_buffers(handle, page_buffers(page),
+						from, to, &partial,
+						ocfs2_clear_cow_buffer);
+			if (ret) {
+				mlog_errno(ret);
+				goto unlock;
+			}
+		}
+
+		ocfs2_map_and_dirty_page(context->inode,
+					 handle, from, to,
+					 page, 0, &new_block);
+		mark_page_accessed(page);
+unlock:
+		unlock_page(page);
+		page_cache_release(page);
+		page = NULL;
+		offset = map_end;
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+static int ocfs2_duplicate_clusters_by_jbd(handle_t *handle,
+					   struct ocfs2_cow_context *context,
+					   u32 cpos, u32 old_cluster,
+					   u32 new_cluster, u32 new_len)
+{
+	int ret = 0;
+	struct super_block *sb = context->inode->i_sb;
+	struct ocfs2_caching_info *ci = context->data_et.et_ci;
+	int i, blocks = ocfs2_clusters_to_blocks(sb, new_len);
+	u64 old_block = ocfs2_clusters_to_blocks(sb, old_cluster);
+	u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
+	struct ocfs2_super *osb = OCFS2_SB(sb);
+	struct buffer_head *old_bh = NULL;
+	struct buffer_head *new_bh = NULL;
+
+	mlog(0, "old_cluster %u, new %u, len %u\n", old_cluster,
+	     new_cluster, new_len);
+
+	for (i = 0; i < blocks; i++, old_block++, new_block++) {
+		new_bh = sb_getblk(osb->sb, new_block);
+		if (new_bh == NULL) {
+			ret = -EIO;
+			mlog_errno(ret);
+			break;
+		}
+
+		ocfs2_set_new_buffer_uptodate(ci, new_bh);
+
+		ret = ocfs2_read_block(ci, old_block, &old_bh, NULL);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		ret = ocfs2_journal_access(handle, ci, new_bh,
+					   OCFS2_JOURNAL_ACCESS_CREATE);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		memcpy(new_bh->b_data, old_bh->b_data, sb->s_blocksize);
+		ret = ocfs2_journal_dirty(handle, new_bh);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		brelse(new_bh);
+		brelse(old_bh);
+		new_bh = NULL;
+		old_bh = NULL;
+	}
+
+	brelse(new_bh);
+	brelse(old_bh);
+	return ret;
+}
+
+static int ocfs2_clear_ext_refcount(handle_t *handle,
+				    struct ocfs2_extent_tree *et,
+				    u32 cpos, u32 p_cluster, u32 len,
+				    unsigned int ext_flags,
+				    struct ocfs2_alloc_context *meta_ac,
+				    struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret, index;
+	struct ocfs2_extent_rec replace_rec;
+	struct ocfs2_path *path = NULL;
+	struct ocfs2_extent_list *el;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
+	u64 ino = ocfs2_metadata_cache_owner(et->et_ci);
+
+	mlog(0, "inode %llu cpos %u, len %u, p_cluster %u, ext_flags %u\n",
+	     (unsigned long long)ino, cpos, len, p_cluster, ext_flags);
+
+	memset(&replace_rec, 0, sizeof(replace_rec));
+	replace_rec.e_cpos = cpu_to_le32(cpos);
+	replace_rec.e_leaf_clusters = cpu_to_le16(len);
+	replace_rec.e_blkno = cpu_to_le64(ocfs2_clusters_to_blocks(sb,
+								   p_cluster));
+	replace_rec.e_flags = ext_flags;
+	replace_rec.e_flags &= ~OCFS2_EXT_REFCOUNTED;
+
+	path = ocfs2_new_path_from_et(et);
+	if (!path) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_find_path(et->et_ci, path, cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	el = path_leaf_el(path);
+
+	index = ocfs2_search_extent_list(el, cpos);
+	if (index == -1 || index >= le16_to_cpu(el->l_next_free_rec)) {
+		ocfs2_error(sb,
+			    "Inode %llu has an extent at cpos %u which can no "
+			    "longer be found.\n",
+			    (unsigned long long)ino, cpos);
+		ret = -EROFS;
+		goto out;
+	}
+
+	ret = ocfs2_split_extent(handle, et, path, index,
+				 &replace_rec, meta_ac, dealloc);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	ocfs2_free_path(path);
+	return ret;
+}
+
+static int ocfs2_replace_clusters(handle_t *handle,
+				  struct ocfs2_cow_context *context,
+				  u32 cpos, u32 old,
+				  u32 new, u32 len,
+				  unsigned int ext_flags)
+{
+	int ret;
+	struct ocfs2_caching_info *ci = context->data_et.et_ci;
+	u64 ino = ocfs2_metadata_cache_owner(ci);
+
+	mlog(0, "inode %llu, cpos %u, old %u, new %u, len %u, ext_flags %u\n",
+	     (unsigned long long)ino, cpos, old, new, len, ext_flags);
+
+	/*If the old clusters is unwritten, no need to duplicate. */
+	if (!(ext_flags & OCFS2_EXT_UNWRITTEN)) {
+		ret = context->cow_duplicate_clusters(handle, context, cpos,
+						      old, new, len);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	ret = ocfs2_clear_ext_refcount(handle, &context->data_et,
+				       cpos, new, len, ext_flags,
+				       context->meta_ac, &context->dealloc);
+	if (ret)
+		mlog_errno(ret);
+out:
+	return ret;
+}
+
+static int ocfs2_cow_sync_writeback(struct super_block *sb,
+				    struct ocfs2_cow_context *context,
+				    u32 cpos, u32 num_clusters)
+{
+	int ret = 0;
+	loff_t offset, end, map_end;
+	pgoff_t page_index;
+	struct page *page;
+
+	if (ocfs2_should_order_data(context->inode))
+		return 0;
+
+	offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
+	end = offset + (num_clusters << OCFS2_SB(sb)->s_clustersize_bits);
+
+	ret = filemap_fdatawrite_range(context->inode->i_mapping,
+				       offset, end - 1);
+	if (ret < 0) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	while (offset < end) {
+		page_index = offset >> PAGE_CACHE_SHIFT;
+		map_end = (page_index + 1) << PAGE_CACHE_SHIFT;
+		if (map_end > end)
+			map_end = end;
+
+		page = grab_cache_page(context->inode->i_mapping, page_index);
+		BUG_ON(!page);
+
+		wait_on_page_writeback(page);
+		if (PageError(page)) {
+			ret = -EIO;
+			mlog_errno(ret);
+		} else
+			mark_page_accessed(page);
+
+		unlock_page(page);
+		page_cache_release(page);
+		page = NULL;
+		offset = map_end;
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+static int ocfs2_di_get_clusters(struct ocfs2_cow_context *context,
+				 u32 v_cluster, u32 *p_cluster,
+				 u32 *num_clusters,
+				 unsigned int *extent_flags)
+{
+	return ocfs2_get_clusters(context->inode, v_cluster, p_cluster,
+				  num_clusters, extent_flags);
+}
+
+static int ocfs2_make_clusters_writable(struct super_block *sb,
+					struct ocfs2_cow_context *context,
+					u32 cpos, u32 p_cluster,
+					u32 num_clusters, unsigned int e_flags)
+{
+	int ret, delete, index, credits =  0;
+	u32 new_bit, new_len;
+	unsigned int set_len;
+	struct ocfs2_super *osb = OCFS2_SB(sb);
+	handle_t *handle;
+	struct buffer_head *ref_leaf_bh = NULL;
+	struct ocfs2_caching_info *ref_ci = &context->ref_tree->rf_ci;
+	struct ocfs2_refcount_rec rec;
+
+	mlog(0, "cpos %u, p_cluster %u, num_clusters %u, e_flags %u\n",
+	     cpos, p_cluster, num_clusters, e_flags);
+
+	ret = ocfs2_lock_refcount_allocators(sb, p_cluster, num_clusters,
+					     &context->data_et,
+					     ref_ci,
+					     context->ref_root_bh,
+					     &context->meta_ac,
+					     &context->data_ac, &credits);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	if (context->post_refcount)
+		credits += context->post_refcount->credits;
+
+	credits += context->extra_credits;
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	while (num_clusters) {
+		ret = ocfs2_get_refcount_rec(ref_ci, context->ref_root_bh,
+					     p_cluster, num_clusters,
+					     &rec, &index, &ref_leaf_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_commit;
+		}
+
+		BUG_ON(!rec.r_refcount);
+		set_len = min((u64)p_cluster + num_clusters,
+			      le64_to_cpu(rec.r_cpos) +
+			      le32_to_cpu(rec.r_clusters)) - p_cluster;
+
+		/*
+		 * There are many different situation here.
+		 * 1. If refcount == 1, remove the flag and don't COW.
+		 * 2. If refcount > 1, allocate clusters.
+		 *    Here we may not allocate r_len once at a time, so continue
+		 *    until we reach num_clusters.
+		 */
+		if (le32_to_cpu(rec.r_refcount) == 1) {
+			delete = 0;
+			ret = ocfs2_clear_ext_refcount(handle,
+						       &context->data_et,
+						       cpos, p_cluster,
+						       set_len, e_flags,
+						       context->meta_ac,
+						       &context->dealloc);
+			if (ret) {
+				mlog_errno(ret);
+				goto out_commit;
+			}
+		} else {
+			delete = 1;
+
+			ret = __ocfs2_claim_clusters(osb, handle,
+						     context->data_ac,
+						     1, set_len,
+						     &new_bit, &new_len);
+			if (ret) {
+				mlog_errno(ret);
+				goto out_commit;
+			}
+
+			ret = ocfs2_replace_clusters(handle, context,
+						     cpos, p_cluster, new_bit,
+						     new_len, e_flags);
+			if (ret) {
+				mlog_errno(ret);
+				goto out_commit;
+			}
+			set_len = new_len;
+		}
+
+		ret = __ocfs2_decrease_refcount(handle, ref_ci,
+						context->ref_root_bh,
+						p_cluster, set_len,
+						context->meta_ac,
+						&context->dealloc, delete);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_commit;
+		}
+
+		cpos += set_len;
+		p_cluster += set_len;
+		num_clusters -= set_len;
+		brelse(ref_leaf_bh);
+		ref_leaf_bh = NULL;
+	}
+
+	/* handle any post_cow action. */
+	if (context->post_refcount && context->post_refcount->func) {
+		ret = context->post_refcount->func(context->inode, handle,
+						context->post_refcount->para);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_commit;
+		}
+	}
+
+	/*
+	 * Here we should write the new page out first if we are
+	 * in write-back mode.
+	 */
+	if (context->get_clusters == ocfs2_di_get_clusters) {
+		ret = ocfs2_cow_sync_writeback(sb, context, cpos, num_clusters);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+
+out:
+	if (context->data_ac) {
+		ocfs2_free_alloc_context(context->data_ac);
+		context->data_ac = NULL;
+	}
+	if (context->meta_ac) {
+		ocfs2_free_alloc_context(context->meta_ac);
+		context->meta_ac = NULL;
+	}
+	brelse(ref_leaf_bh);
+
+	return ret;
+}
+
+static int ocfs2_replace_cow(struct ocfs2_cow_context *context)
+{
+	int ret = 0;
+	struct inode *inode = context->inode;
+	u32 cow_start = context->cow_start, cow_len = context->cow_len;
+	u32 p_cluster, num_clusters;
+	unsigned int ext_flags;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+			    "tree, but the feature bit is not set in the "
+			    "super block.", inode->i_ino);
+		return -EROFS;
+	}
+
+	ocfs2_init_dealloc_ctxt(&context->dealloc);
+
+	while (cow_len) {
+		ret = context->get_clusters(context, cow_start, &p_cluster,
+					    &num_clusters, &ext_flags);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		BUG_ON(!(ext_flags & OCFS2_EXT_REFCOUNTED));
+
+		if (cow_len < num_clusters)
+			num_clusters = cow_len;
+
+		ret = ocfs2_make_clusters_writable(inode->i_sb, context,
+						   cow_start, p_cluster,
+						   num_clusters, ext_flags);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		cow_len -= num_clusters;
+		cow_start += num_clusters;
+	}
+
+	if (ocfs2_dealloc_has_cluster(&context->dealloc)) {
+		ocfs2_schedule_truncate_log_flush(osb, 1);
+		ocfs2_run_deallocs(osb, &context->dealloc);
+	}
+
+	return ret;
+}
+
+/*
+ * Starting at cpos, try to CoW write_len clusters.  Don't CoW
+ * past max_cpos.  This will stop when it runs into a hole or an
+ * unrefcounted extent.
+ */
+static int ocfs2_refcount_cow_hunk(struct inode *inode,
+				   struct buffer_head *di_bh,
+				   u32 cpos, u32 write_len, u32 max_cpos)
+{
+	int ret;
+	u32 cow_start = 0, cow_len = 0;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_tree *ref_tree;
+	struct ocfs2_cow_context *context = NULL;
+
+	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+	ret = ocfs2_refcount_cal_cow_clusters(inode, &di->id2.i_list,
+					      cpos, write_len, max_cpos,
+					      &cow_start, &cow_len);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "CoW inode %lu, cpos %u, write_len %u, cow_start %u, "
+	     "cow_len %u\n", inode->i_ino,
+	     cpos, write_len, cow_start, cow_len);
+
+	BUG_ON(cow_len == 0);
+
+	context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
+	if (!context) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
+				       1, &ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	context->inode = inode;
+	context->cow_start = cow_start;
+	context->cow_len = cow_len;
+	context->ref_tree = ref_tree;
+	context->ref_root_bh = ref_root_bh;
+	context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_page;
+	context->get_clusters = ocfs2_di_get_clusters;
+
+	ocfs2_init_dinode_extent_tree(&context->data_et,
+				      INODE_CACHE(inode), di_bh);
+
+	ret = ocfs2_replace_cow(context);
+	if (ret)
+		mlog_errno(ret);
+
+	/*
+	 * truncate the extent map here since no matter whether we meet with
+	 * any error during the action, we shouldn't trust cached extent map
+	 * any more.
+	 */
+	ocfs2_extent_map_trunc(inode, cow_start);
+
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	brelse(ref_root_bh);
+out:
+	kfree(context);
+	return ret;
+}
+
+/*
+ * CoW any and all clusters between cpos and cpos+write_len.
+ * Don't CoW past max_cpos.  If this returns successfully, all
+ * clusters between cpos and cpos+write_len are safe to modify.
+ */
+int ocfs2_refcount_cow(struct inode *inode,
+		       struct buffer_head *di_bh,
+		       u32 cpos, u32 write_len, u32 max_cpos)
+{
+	int ret = 0;
+	u32 p_cluster, num_clusters;
+	unsigned int ext_flags;
+
+	while (write_len) {
+		ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
+					 &num_clusters, &ext_flags);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		if (write_len < num_clusters)
+			num_clusters = write_len;
+
+		if (ext_flags & OCFS2_EXT_REFCOUNTED) {
+			ret = ocfs2_refcount_cow_hunk(inode, di_bh, cpos,
+						      num_clusters, max_cpos);
+			if (ret) {
+				mlog_errno(ret);
+				break;
+			}
+		}
+
+		write_len -= num_clusters;
+		cpos += num_clusters;
+	}
+
+	return ret;
+}
+
+static int ocfs2_xattr_value_get_clusters(struct ocfs2_cow_context *context,
+					  u32 v_cluster, u32 *p_cluster,
+					  u32 *num_clusters,
+					  unsigned int *extent_flags)
+{
+	struct inode *inode = context->inode;
+	struct ocfs2_xattr_value_root *xv = context->cow_object;
+
+	return ocfs2_xattr_get_clusters(inode, v_cluster, p_cluster,
+					num_clusters, &xv->xr_list,
+					extent_flags);
+}
+
+/*
+ * Given a xattr value root, calculate the most meta/credits we need for
+ * refcount tree change if we truncate it to 0.
+ */
+int ocfs2_refcounted_xattr_delete_need(struct inode *inode,
+				       struct ocfs2_caching_info *ref_ci,
+				       struct buffer_head *ref_root_bh,
+				       struct ocfs2_xattr_value_root *xv,
+				       int *meta_add, int *credits)
+{
+	int ret = 0, index, ref_blocks = 0;
+	u32 p_cluster, num_clusters;
+	u32 cpos = 0, clusters = le32_to_cpu(xv->xr_clusters);
+	struct ocfs2_refcount_block *rb;
+	struct ocfs2_refcount_rec rec;
+	struct buffer_head *ref_leaf_bh = NULL;
+
+	while (cpos < clusters) {
+		ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
+					       &num_clusters, &xv->xr_list,
+					       NULL);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		cpos += num_clusters;
+
+		while (num_clusters) {
+			ret = ocfs2_get_refcount_rec(ref_ci, ref_root_bh,
+						     p_cluster, num_clusters,
+						     &rec, &index,
+						     &ref_leaf_bh);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+
+			BUG_ON(!rec.r_refcount);
+
+			rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+
+			/*
+			 * We really don't know whether the other clusters is in
+			 * this refcount block or not, so just take the worst
+			 * case that all the clusters are in this block and each
+			 * one will split a refcount rec, so totally we need
+			 * clusters * 2 new refcount rec.
+			 */
+			if (le64_to_cpu(rb->rf_records.rl_used) + clusters * 2 >
+			    le16_to_cpu(rb->rf_records.rl_count))
+				ref_blocks++;
+
+			*credits += 1;
+			brelse(ref_leaf_bh);
+			ref_leaf_bh = NULL;
+
+			if (num_clusters <= le32_to_cpu(rec.r_clusters))
+				break;
+			else
+				num_clusters -= le32_to_cpu(rec.r_clusters);
+			p_cluster += num_clusters;
+		}
+	}
+
+	*meta_add += ref_blocks;
+	if (!ref_blocks)
+		goto out;
+
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)
+		*credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+	else {
+		struct ocfs2_extent_tree et;
+
+		ocfs2_init_refcount_extent_tree(&et, ref_ci, ref_root_bh);
+		*credits += ocfs2_calc_extend_credits(inode->i_sb,
+						      et.et_root_el,
+						      ref_blocks);
+	}
+
+out:
+	brelse(ref_leaf_bh);
+	return ret;
+}
+
+/*
+ * Do CoW for xattr.
+ */
+int ocfs2_refcount_cow_xattr(struct inode *inode,
+			     struct ocfs2_dinode *di,
+			     struct ocfs2_xattr_value_buf *vb,
+			     struct ocfs2_refcount_tree *ref_tree,
+			     struct buffer_head *ref_root_bh,
+			     u32 cpos, u32 write_len,
+			     struct ocfs2_post_refcount *post)
+{
+	int ret;
+	struct ocfs2_xattr_value_root *xv = vb->vb_xv;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_cow_context *context = NULL;
+	u32 cow_start, cow_len;
+
+	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+	ret = ocfs2_refcount_cal_cow_clusters(inode, &xv->xr_list,
+					      cpos, write_len, UINT_MAX,
+					      &cow_start, &cow_len);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	BUG_ON(cow_len == 0);
+
+	context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
+	if (!context) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	context->inode = inode;
+	context->cow_start = cow_start;
+	context->cow_len = cow_len;
+	context->ref_tree = ref_tree;
+	context->ref_root_bh = ref_root_bh;;
+	context->cow_object = xv;
+
+	context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_jbd;
+	/* We need the extra credits for duplicate_clusters by jbd. */
+	context->extra_credits =
+		ocfs2_clusters_to_blocks(inode->i_sb, 1) * cow_len;
+	context->get_clusters = ocfs2_xattr_value_get_clusters;
+	context->post_refcount = post;
+
+	ocfs2_init_xattr_value_extent_tree(&context->data_et,
+					   INODE_CACHE(inode), vb);
+
+	ret = ocfs2_replace_cow(context);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	kfree(context);
+	return ret;
+}
+
+/*
+ * Insert a new extent into refcount tree and mark a extent rec
+ * as refcounted in the dinode tree.
+ */
+int ocfs2_add_refcount_flag(struct inode *inode,
+			    struct ocfs2_extent_tree *data_et,
+			    struct ocfs2_caching_info *ref_ci,
+			    struct buffer_head *ref_root_bh,
+			    u32 cpos, u32 p_cluster, u32 num_clusters,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc,
+			    struct ocfs2_post_refcount *post)
+{
+	int ret;
+	handle_t *handle;
+	int credits = 1, ref_blocks = 0;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_alloc_context *meta_ac = NULL;
+
+	ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
+					       ref_ci, ref_root_bh,
+					       p_cluster, num_clusters,
+					       &ref_blocks, &credits);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "reserve new metadata %d, credits = %d\n",
+	     ref_blocks, credits);
+
+	if (ref_blocks) {
+		ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
+							ref_blocks, &meta_ac);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	if (post)
+		credits += post->credits;
+
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_mark_extent_refcounted(inode, data_et, handle,
+					   cpos, num_clusters, p_cluster,
+					   meta_ac, dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = __ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
+					p_cluster, num_clusters, 0,
+					meta_ac, dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	if (post && post->func) {
+		ret = post->func(inode, handle, post->para);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+out:
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+	return ret;
+}
+
+static int ocfs2_change_ctime(struct inode *inode,
+			      struct buffer_head *di_bh)
+{
+	int ret;
+	handle_t *handle;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+	handle = ocfs2_start_trans(OCFS2_SB(inode->i_sb),
+				   OCFS2_INODE_UPDATE_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	inode->i_ctime = CURRENT_TIME;
+	di->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec);
+	di->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec);
+
+	ocfs2_journal_dirty(handle, di_bh);
+
+out_commit:
+	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
+out:
+	return ret;
+}
+
+static int ocfs2_attach_refcount_tree(struct inode *inode,
+				      struct buffer_head *di_bh)
+{
+	int ret, data_changed = 0;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_refcount_tree *ref_tree;
+	unsigned int ext_flags;
+	loff_t size;
+	u32 cpos, num_clusters, clusters, p_cluster;
+	struct ocfs2_cached_dealloc_ctxt dealloc;
+	struct ocfs2_extent_tree di_et;
+
+	ocfs2_init_dealloc_ctxt(&dealloc);
+
+	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL)) {
+		ret = ocfs2_create_refcount_tree(inode, di_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	BUG_ON(!di->i_refcount_loc);
+	ret = ocfs2_lock_refcount_tree(osb,
+				       le64_to_cpu(di->i_refcount_loc), 1,
+				       &ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ocfs2_init_dinode_extent_tree(&di_et, INODE_CACHE(inode), di_bh);
+
+	size = i_size_read(inode);
+	clusters = ocfs2_clusters_for_bytes(inode->i_sb, size);
+
+	cpos = 0;
+	while (cpos < clusters) {
+		ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
+					 &num_clusters, &ext_flags);
+
+		if (p_cluster && !(ext_flags & OCFS2_EXT_REFCOUNTED)) {
+			ret = ocfs2_add_refcount_flag(inode, &di_et,
+						      &ref_tree->rf_ci,
+						      ref_root_bh, cpos,
+						      p_cluster, num_clusters,
+						      &dealloc, NULL);
+			if (ret) {
+				mlog_errno(ret);
+				goto unlock;
+			}
+
+			data_changed = 1;
+		}
+		cpos += num_clusters;
+	}
+
+	if (oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
+		ret = ocfs2_xattr_attach_refcount_tree(inode, di_bh,
+						       &ref_tree->rf_ci,
+						       ref_root_bh,
+						       &dealloc);
+		if (ret) {
+			mlog_errno(ret);
+			goto unlock;
+		}
+	}
+
+	if (data_changed) {
+		ret = ocfs2_change_ctime(inode, di_bh);
+		if (ret)
+			mlog_errno(ret);
+	}
+
+unlock:
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	brelse(ref_root_bh);
+
+	if (!ret && ocfs2_dealloc_has_cluster(&dealloc)) {
+		ocfs2_schedule_truncate_log_flush(osb, 1);
+		ocfs2_run_deallocs(osb, &dealloc);
+	}
+out:
+	/*
+	 * Empty the extent map so that we may get the right extent
+	 * record from the disk.
+	 */
+	ocfs2_extent_map_trunc(inode, 0);
+
+	return ret;
+}
+
+static int ocfs2_add_refcounted_extent(struct inode *inode,
+				   struct ocfs2_extent_tree *et,
+				   struct ocfs2_caching_info *ref_ci,
+				   struct buffer_head *ref_root_bh,
+				   u32 cpos, u32 p_cluster, u32 num_clusters,
+				   unsigned int ext_flags,
+				   struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret;
+	handle_t *handle;
+	int credits = 0;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_alloc_context *meta_ac = NULL;
+
+	ret = ocfs2_lock_refcount_allocators(inode->i_sb,
+					     p_cluster, num_clusters,
+					     et, ref_ci,
+					     ref_root_bh, &meta_ac,
+					     NULL, &credits);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_insert_extent(handle, et, cpos,
+			cpu_to_le64(ocfs2_clusters_to_blocks(inode->i_sb,
+							     p_cluster)),
+			num_clusters, ext_flags, meta_ac);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
+				      p_cluster, num_clusters,
+				      meta_ac, dealloc);
+	if (ret)
+		mlog_errno(ret);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+out:
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+	return ret;
+}
+
+static int ocfs2_duplicate_extent_list(struct inode *s_inode,
+				struct inode *t_inode,
+				struct buffer_head *t_bh,
+				struct ocfs2_caching_info *ref_ci,
+				struct buffer_head *ref_root_bh,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret = 0;
+	u32 p_cluster, num_clusters, clusters, cpos;
+	loff_t size;
+	unsigned int ext_flags;
+	struct ocfs2_extent_tree et;
+
+	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(t_inode), t_bh);
+
+	size = i_size_read(s_inode);
+	clusters = ocfs2_clusters_for_bytes(s_inode->i_sb, size);
+
+	cpos = 0;
+	while (cpos < clusters) {
+		ret = ocfs2_get_clusters(s_inode, cpos, &p_cluster,
+					 &num_clusters, &ext_flags);
+
+		if (p_cluster) {
+			ret = ocfs2_add_refcounted_extent(t_inode, &et,
+							  ref_ci, ref_root_bh,
+							  cpos, p_cluster,
+							  num_clusters,
+							  ext_flags,
+							  dealloc);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+		}
+
+		cpos += num_clusters;
+	}
+
+out:
+	return ret;
+}
+
+/*
+ * change the new file's attributes to the src.
+ *
+ * reflink creates a snapshot of a file, that means the attributes
+ * must be identical except for three exceptions - nlink, ino, and ctime.
+ */
+static int ocfs2_complete_reflink(struct inode *s_inode,
+				  struct buffer_head *s_bh,
+				  struct inode *t_inode,
+				  struct buffer_head *t_bh,
+				  bool preserve)
+{
+	int ret;
+	handle_t *handle;
+	struct ocfs2_dinode *s_di = (struct ocfs2_dinode *)s_bh->b_data;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)t_bh->b_data;
+	loff_t size = i_size_read(s_inode);
+
+	handle = ocfs2_start_trans(OCFS2_SB(t_inode->i_sb),
+				   OCFS2_INODE_UPDATE_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		return ret;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(t_inode), t_bh,
+				      OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	spin_lock(&OCFS2_I(t_inode)->ip_lock);
+	OCFS2_I(t_inode)->ip_clusters = OCFS2_I(s_inode)->ip_clusters;
+	OCFS2_I(t_inode)->ip_attr = OCFS2_I(s_inode)->ip_attr;
+	OCFS2_I(t_inode)->ip_dyn_features = OCFS2_I(s_inode)->ip_dyn_features;
+	spin_unlock(&OCFS2_I(t_inode)->ip_lock);
+	i_size_write(t_inode, size);
+
+	di->i_xattr_inline_size = s_di->i_xattr_inline_size;
+	di->i_clusters = s_di->i_clusters;
+	di->i_size = s_di->i_size;
+	di->i_dyn_features = s_di->i_dyn_features;
+	di->i_attr = s_di->i_attr;
+
+	if (preserve) {
+		di->i_uid = s_di->i_uid;
+		di->i_gid = s_di->i_gid;
+		di->i_mode = s_di->i_mode;
+
+		/*
+		 * update time.
+		 * we want mtime to appear identical to the source and
+		 * update ctime.
+		 */
+		t_inode->i_ctime = CURRENT_TIME;
+
+		di->i_ctime = cpu_to_le64(t_inode->i_ctime.tv_sec);
+		di->i_ctime_nsec = cpu_to_le32(t_inode->i_ctime.tv_nsec);
+
+		t_inode->i_mtime = s_inode->i_mtime;
+		di->i_mtime = s_di->i_mtime;
+		di->i_mtime_nsec = s_di->i_mtime_nsec;
+	}
+
+	ocfs2_journal_dirty(handle, t_bh);
+
+out_commit:
+	ocfs2_commit_trans(OCFS2_SB(t_inode->i_sb), handle);
+	return ret;
+}
+
+static int ocfs2_create_reflink_node(struct inode *s_inode,
+				     struct buffer_head *s_bh,
+				     struct inode *t_inode,
+				     struct buffer_head *t_bh,
+				     bool preserve)
+{
+	int ret;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_cached_dealloc_ctxt dealloc;
+	struct ocfs2_super *osb = OCFS2_SB(s_inode->i_sb);
+	struct ocfs2_refcount_block *rb;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)s_bh->b_data;
+	struct ocfs2_refcount_tree *ref_tree;
+
+	ocfs2_init_dealloc_ctxt(&dealloc);
+
+	ret = ocfs2_set_refcount_tree(t_inode, t_bh,
+				      le64_to_cpu(di->i_refcount_loc));
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
+				       1, &ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+	ret = ocfs2_duplicate_extent_list(s_inode, t_inode, t_bh,
+					  &ref_tree->rf_ci, ref_root_bh,
+					  &dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_unlock_refcount;
+	}
+
+	ret = ocfs2_complete_reflink(s_inode, s_bh, t_inode, t_bh, preserve);
+	if (ret)
+		mlog_errno(ret);
+
+out_unlock_refcount:
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	brelse(ref_root_bh);
+out:
+	if (ocfs2_dealloc_has_cluster(&dealloc)) {
+		ocfs2_schedule_truncate_log_flush(osb, 1);
+		ocfs2_run_deallocs(osb, &dealloc);
+	}
+
+	return ret;
+}
+
+static int __ocfs2_reflink(struct dentry *old_dentry,
+			   struct buffer_head *old_bh,
+			   struct inode *new_inode,
+			   bool preserve)
+{
+	int ret;
+	struct inode *inode = old_dentry->d_inode;
+	struct buffer_head *new_bh = NULL;
+
+	ret = filemap_fdatawrite(inode->i_mapping);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_attach_refcount_tree(inode, old_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mutex_lock(&new_inode->i_mutex);
+	ret = ocfs2_inode_lock(new_inode, &new_bh, 1);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	ret = ocfs2_create_reflink_node(inode, old_bh,
+					new_inode, new_bh, preserve);
+	if (ret) {
+		mlog_errno(ret);
+		goto inode_unlock;
+	}
+
+	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
+		ret = ocfs2_reflink_xattrs(inode, old_bh,
+					   new_inode, new_bh,
+					   preserve);
+		if (ret)
+			mlog_errno(ret);
+	}
+inode_unlock:
+	ocfs2_inode_unlock(new_inode, 1);
+	brelse(new_bh);
+out_unlock:
+	mutex_unlock(&new_inode->i_mutex);
+out:
+	if (!ret) {
+		ret = filemap_fdatawait(inode->i_mapping);
+		if (ret)
+			mlog_errno(ret);
+	}
+	return ret;
+}
+
+static int ocfs2_reflink(struct dentry *old_dentry, struct inode *dir,
+			 struct dentry *new_dentry, bool preserve)
+{
+	int error;
+	struct inode *inode = old_dentry->d_inode;
+	struct buffer_head *old_bh = NULL;
+	struct inode *new_orphan_inode = NULL;
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
+		return -EOPNOTSUPP;
+
+	error = ocfs2_create_inode_in_orphan(dir, inode->i_mode,
+					     &new_orphan_inode);
+	if (error) {
+		mlog_errno(error);
+		goto out;
+	}
+
+	error = ocfs2_inode_lock(inode, &old_bh, 1);
+	if (error) {
+		mlog_errno(error);
+		goto out;
+	}
+
+	down_write(&OCFS2_I(inode)->ip_xattr_sem);
+	down_write(&OCFS2_I(inode)->ip_alloc_sem);
+	error = __ocfs2_reflink(old_dentry, old_bh,
+				new_orphan_inode, preserve);
+	up_write(&OCFS2_I(inode)->ip_alloc_sem);
+	up_write(&OCFS2_I(inode)->ip_xattr_sem);
+
+	ocfs2_inode_unlock(inode, 1);
+	brelse(old_bh);
+
+	if (error) {
+		mlog_errno(error);
+		goto out;
+	}
+
+	/* If the security isn't preserved, we need to re-initialize them. */
+	if (!preserve) {
+		error = ocfs2_init_security_and_acl(dir, new_orphan_inode);
+		if (error)
+			mlog_errno(error);
+	}
+out:
+	if (!error) {
+		error = ocfs2_mv_orphaned_inode_to_new(dir, new_orphan_inode,
+						       new_dentry);
+		if (error)
+			mlog_errno(error);
+	}
+
+	if (new_orphan_inode) {
+		/*
+		 * We need to open_unlock the inode no matter whether we
+		 * succeed or not, so that other nodes can delete it later.
+		 */
+		ocfs2_open_unlock(new_orphan_inode);
+		if (error)
+			iput(new_orphan_inode);
+	}
+
+	return error;
+}
+
+/*
+ * Below here are the bits used by OCFS2_IOC_REFLINK() to fake
+ * sys_reflink().  This will go away when vfs_reflink() exists in
+ * fs/namei.c.
+ */
+
+/* copied from may_create in VFS. */
+static inline int ocfs2_may_create(struct inode *dir, struct dentry *child)
+{
+	if (child->d_inode)
+		return -EEXIST;
+	if (IS_DEADDIR(dir))
+		return -ENOENT;
+	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
+}
+
+/* copied from user_path_parent. */
+static int ocfs2_user_path_parent(const char __user *path,
+				  struct nameidata *nd, char **name)
+{
+	char *s = getname(path);
+	int error;
+
+	if (IS_ERR(s))
+		return PTR_ERR(s);
+
+	error = path_lookup(s, LOOKUP_PARENT, nd);
+	if (error)
+		putname(s);
+	else
+		*name = s;
+
+	return error;
+}
+
+/**
+ * ocfs2_vfs_reflink - Create a reference-counted link
+ *
+ * @old_dentry:        source dentry + inode
+ * @dir:       directory to create the target
+ * @new_dentry:        target dentry
+ * @preserve:  if true, preserve all file attributes
+ */
+int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
+		      struct dentry *new_dentry, bool preserve)
+{
+	struct inode *inode = old_dentry->d_inode;
+	int error;
+
+	if (!inode)
+		return -ENOENT;
+
+	error = ocfs2_may_create(dir, new_dentry);
+	if (error)
+		return error;
+
+	if (dir->i_sb != inode->i_sb)
+		return -EXDEV;
+
+	/*
+	 * A reflink to an append-only or immutable file cannot be created.
+	 */
+	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
+		return -EPERM;
+
+	/* Only regular files can be reflinked. */
+	if (!S_ISREG(inode->i_mode))
+		return -EPERM;
+
+	/*
+	 * If the caller wants to preserve ownership, they require the
+	 * rights to do so.
+	 */
+	if (preserve) {
+		if ((current_fsuid() != inode->i_uid) && !capable(CAP_CHOWN))
+			return -EPERM;
+		if (!in_group_p(inode->i_gid) && !capable(CAP_CHOWN))
+			return -EPERM;
+	}
+
+	/*
+	 * If the caller is modifying any aspect of the attributes, they
+	 * are not creating a snapshot.  They need read permission on the
+	 * file.
+	 */
+	if (!preserve) {
+		error = inode_permission(inode, MAY_READ);
+		if (error)
+			return error;
+	}
+
+	mutex_lock(&inode->i_mutex);
+	vfs_dq_init(dir);
+	error = ocfs2_reflink(old_dentry, dir, new_dentry, preserve);
+	mutex_unlock(&inode->i_mutex);
+	if (!error)
+		fsnotify_create(dir, new_dentry);
+	return error;
+}
+/*
+ * Most codes are copied from sys_linkat.
+ */
+int ocfs2_reflink_ioctl(struct inode *inode,
+			const char __user *oldname,
+			const char __user *newname,
+			bool preserve)
+{
+	struct dentry *new_dentry;
+	struct nameidata nd;
+	struct path old_path;
+	int error;
+	char *to = NULL;
+
+	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
+		return -EOPNOTSUPP;
+
+	error = user_path_at(AT_FDCWD, oldname, 0, &old_path);
+	if (error) {
+		mlog_errno(error);
+		return error;
+	}
+
+	error = ocfs2_user_path_parent(newname, &nd, &to);
+	if (error) {
+		mlog_errno(error);
+		goto out;
+	}
+
+	error = -EXDEV;
+	if (old_path.mnt != nd.path.mnt)
+		goto out_release;
+	new_dentry = lookup_create(&nd, 0);
+	error = PTR_ERR(new_dentry);
+	if (IS_ERR(new_dentry)) {
+		mlog_errno(error);
+		goto out_unlock;
+	}
+
+	error = mnt_want_write(nd.path.mnt);
+	if (error) {
+		mlog_errno(error);
+		goto out_dput;
+	}
+
+	error = ocfs2_vfs_reflink(old_path.dentry,
+				  nd.path.dentry->d_inode,
+				  new_dentry, preserve);
+	mnt_drop_write(nd.path.mnt);
+out_dput:
+	dput(new_dentry);
+out_unlock:
+	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
+out_release:
+	path_put(&nd.path);
+	putname(to);
+out:
+	path_put(&old_path);
+
+	return error;
+}
diff --git a/fs/ocfs2/refcounttree.h b/fs/ocfs2/refcounttree.h
new file mode 100644
index 000000000000..c1d19b1d3ecc
--- /dev/null
+++ b/fs/ocfs2/refcounttree.h
@@ -0,0 +1,106 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * refcounttree.h
+ *
+ * Copyright (C) 2009 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ */
+#ifndef OCFS2_REFCOUNTTREE_H
+#define OCFS2_REFCOUNTTREE_H
+
+struct ocfs2_refcount_tree {
+	struct rb_node rf_node;
+	u64 rf_blkno;
+	u32 rf_generation;
+	struct rw_semaphore rf_sem;
+	struct ocfs2_lock_res rf_lockres;
+	struct kref rf_getcnt;
+	int rf_removed;
+
+	/* the following 4 fields are used by caching_info. */
+	struct ocfs2_caching_info rf_ci;
+	spinlock_t rf_lock;
+	struct mutex rf_io_mutex;
+	struct super_block *rf_sb;
+};
+
+void ocfs2_purge_refcount_trees(struct ocfs2_super *osb);
+int ocfs2_lock_refcount_tree(struct ocfs2_super *osb, u64 ref_blkno, int rw,
+			     struct ocfs2_refcount_tree **tree,
+			     struct buffer_head **ref_bh);
+void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb,
+				struct ocfs2_refcount_tree *tree,
+				int rw);
+
+int ocfs2_decrease_refcount(struct inode *inode,
+			    handle_t *handle, u32 cpos, u32 len,
+			    struct ocfs2_alloc_context *meta_ac,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc,
+			    int delete);
+int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
+					  struct buffer_head *di_bh,
+					  u64 phys_blkno,
+					  u32 clusters,
+					  int *credits,
+					  struct ocfs2_alloc_context **meta_ac);
+int ocfs2_refcount_cow(struct inode *inode, struct buffer_head *di_bh,
+		       u32 cpos, u32 write_len, u32 max_cpos);
+
+typedef int (ocfs2_post_refcount_func)(struct inode *inode,
+				       handle_t *handle,
+				       void *para);
+/*
+ * Some refcount caller need to do more work after we modify the data b-tree
+ * during refcount operation(including CoW and add refcount flag), and make the
+ * transaction complete. So it must give us this structure so that we can do it
+ * within our transaction.
+ *
+ */
+struct ocfs2_post_refcount {
+	int credits;			/* credits it need for journal. */
+	ocfs2_post_refcount_func *func;	/* real function. */
+	void *para;
+};
+
+int ocfs2_refcounted_xattr_delete_need(struct inode *inode,
+				       struct ocfs2_caching_info *ref_ci,
+				       struct buffer_head *ref_root_bh,
+				       struct ocfs2_xattr_value_root *xv,
+				       int *meta_add, int *credits);
+int ocfs2_refcount_cow_xattr(struct inode *inode,
+			     struct ocfs2_dinode *di,
+			     struct ocfs2_xattr_value_buf *vb,
+			     struct ocfs2_refcount_tree *ref_tree,
+			     struct buffer_head *ref_root_bh,
+			     u32 cpos, u32 write_len,
+			     struct ocfs2_post_refcount *post);
+int ocfs2_add_refcount_flag(struct inode *inode,
+			    struct ocfs2_extent_tree *data_et,
+			    struct ocfs2_caching_info *ref_ci,
+			    struct buffer_head *ref_root_bh,
+			    u32 cpos, u32 p_cluster, u32 num_clusters,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc,
+			    struct ocfs2_post_refcount *post);
+int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh);
+int ocfs2_try_remove_refcount_tree(struct inode *inode,
+				   struct buffer_head *di_bh);
+int ocfs2_increase_refcount(handle_t *handle,
+			    struct ocfs2_caching_info *ci,
+			    struct buffer_head *ref_root_bh,
+			    u64 cpos, u32 len,
+			    struct ocfs2_alloc_context *meta_ac,
+			    struct ocfs2_cached_dealloc_ctxt *dealloc);
+int ocfs2_reflink_ioctl(struct inode *inode,
+			const char __user *oldname,
+			const char __user *newname,
+			bool preserve);
+#endif /* OCFS2_REFCOUNTTREE_H */
diff --git a/fs/ocfs2/resize.c b/fs/ocfs2/resize.c
index 424adaa5f900..3c3d673a4d20 100644
--- a/fs/ocfs2/resize.c
+++ b/fs/ocfs2/resize.c
@@ -106,8 +106,8 @@ static int ocfs2_update_last_group_and_inode(handle_t *handle,
 	mlog_entry("(new_clusters=%d, first_new_cluster = %u)\n",
 		   new_clusters, first_new_cluster);
 
-	ret = ocfs2_journal_access_gd(handle, bm_inode, group_bh,
-				      OCFS2_JOURNAL_ACCESS_WRITE);
+	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(bm_inode),
+				      group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out;
@@ -141,7 +141,7 @@ static int ocfs2_update_last_group_and_inode(handle_t *handle,
 	}
 
 	/* update the inode accordingly. */
-	ret = ocfs2_journal_access_di(handle, bm_inode, bm_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(bm_inode), bm_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -514,7 +514,7 @@ int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
 		goto out_unlock;
 	}
 
-	ocfs2_set_new_buffer_uptodate(inode, group_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), group_bh);
 
 	ret = ocfs2_verify_group_and_input(main_bm_inode, fe, input, group_bh);
 	if (ret) {
@@ -536,8 +536,8 @@ int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
 	cl = &fe->id2.i_chain;
 	cr = &cl->cl_recs[input->chain];
 
-	ret = ocfs2_journal_access_gd(handle, main_bm_inode, group_bh,
-				      OCFS2_JOURNAL_ACCESS_WRITE);
+	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(main_bm_inode),
+				      group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out_commit;
@@ -552,8 +552,8 @@ int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
 		goto out_commit;
 	}
 
-	ret = ocfs2_journal_access_di(handle, main_bm_inode, main_bm_bh,
-				      OCFS2_JOURNAL_ACCESS_WRITE);
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(main_bm_inode),
+				      main_bm_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out_commit;
diff --git a/fs/ocfs2/slot_map.c b/fs/ocfs2/slot_map.c
index 40661e7824e9..bfbd7e9e949f 100644
--- a/fs/ocfs2/slot_map.c
+++ b/fs/ocfs2/slot_map.c
@@ -150,8 +150,8 @@ int ocfs2_refresh_slot_info(struct ocfs2_super *osb)
 	 * be !NULL.  Thus, ocfs2_read_blocks() will ignore blocknr.  If
 	 * this is not true, the read of -1 (UINT64_MAX) will fail.
 	 */
-	ret = ocfs2_read_blocks(si->si_inode, -1, si->si_blocks, si->si_bh,
-				OCFS2_BH_IGNORE_CACHE, NULL);
+	ret = ocfs2_read_blocks(INODE_CACHE(si->si_inode), -1, si->si_blocks,
+				si->si_bh, OCFS2_BH_IGNORE_CACHE, NULL);
 	if (ret == 0) {
 		spin_lock(&osb->osb_lock);
 		ocfs2_update_slot_info(si);
@@ -213,7 +213,7 @@ static int ocfs2_update_disk_slot(struct ocfs2_super *osb,
 		ocfs2_update_disk_slot_old(si, slot_num, &bh);
 	spin_unlock(&osb->osb_lock);
 
-	status = ocfs2_write_block(osb, bh, si->si_inode);
+	status = ocfs2_write_block(osb, bh, INODE_CACHE(si->si_inode));
 	if (status < 0)
 		mlog_errno(status);
 
@@ -404,8 +404,8 @@ static int ocfs2_map_slot_buffers(struct ocfs2_super *osb,
 		     (unsigned long long)blkno);
 
 		bh = NULL;  /* Acquire a fresh bh */
-		status = ocfs2_read_blocks(si->si_inode, blkno, 1, &bh,
-					   OCFS2_BH_IGNORE_CACHE, NULL);
+		status = ocfs2_read_blocks(INODE_CACHE(si->si_inode), blkno,
+					   1, &bh, OCFS2_BH_IGNORE_CACHE, NULL);
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail;
diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
index 73a16d4666dc..c30b644d9572 100644
--- a/fs/ocfs2/suballoc.c
+++ b/fs/ocfs2/suballoc.c
@@ -310,7 +310,7 @@ int ocfs2_read_group_descriptor(struct inode *inode, struct ocfs2_dinode *di,
 	int rc;
 	struct buffer_head *tmp = *bh;
 
-	rc = ocfs2_read_block(inode, gd_blkno, &tmp,
+	rc = ocfs2_read_block(INODE_CACHE(inode), gd_blkno, &tmp,
 			      ocfs2_validate_group_descriptor);
 	if (rc)
 		goto out;
@@ -352,7 +352,7 @@ static int ocfs2_block_group_fill(handle_t *handle,
 	}
 
 	status = ocfs2_journal_access_gd(handle,
-					 alloc_inode,
+					 INODE_CACHE(alloc_inode),
 					 bg_bh,
 					 OCFS2_JOURNAL_ACCESS_CREATE);
 	if (status < 0) {
@@ -476,7 +476,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 		mlog_errno(status);
 		goto bail;
 	}
-	ocfs2_set_new_buffer_uptodate(alloc_inode, bg_bh);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(alloc_inode), bg_bh);
 
 	status = ocfs2_block_group_fill(handle,
 					alloc_inode,
@@ -491,7 +491,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 
 	bg = (struct ocfs2_group_desc *) bg_bh->b_data;
 
-	status = ocfs2_journal_access_di(handle, alloc_inode,
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(alloc_inode),
 					 bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1033,7 +1033,7 @@ static inline int ocfs2_block_group_set_bits(handle_t *handle,
 		journal_type = OCFS2_JOURNAL_ACCESS_UNDO;
 
 	status = ocfs2_journal_access_gd(handle,
-					 alloc_inode,
+					 INODE_CACHE(alloc_inode),
 					 group_bh,
 					 journal_type);
 	if (status < 0) {
@@ -1106,7 +1106,8 @@ static int ocfs2_relink_block_group(handle_t *handle,
 	bg_ptr = le64_to_cpu(bg->bg_next_group);
 	prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
 
-	status = ocfs2_journal_access_gd(handle, alloc_inode, prev_bg_bh,
+	status = ocfs2_journal_access_gd(handle, INODE_CACHE(alloc_inode),
+					 prev_bg_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -1121,8 +1122,8 @@ static int ocfs2_relink_block_group(handle_t *handle,
 		goto out_rollback;
 	}
 
-	status = ocfs2_journal_access_gd(handle, alloc_inode, bg_bh,
-					 OCFS2_JOURNAL_ACCESS_WRITE);
+	status = ocfs2_journal_access_gd(handle, INODE_CACHE(alloc_inode),
+					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out_rollback;
@@ -1136,8 +1137,8 @@ static int ocfs2_relink_block_group(handle_t *handle,
 		goto out_rollback;
 	}
 
-	status = ocfs2_journal_access_di(handle, alloc_inode, fe_bh,
-					 OCFS2_JOURNAL_ACCESS_WRITE);
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(alloc_inode),
+					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out_rollback;
@@ -1288,7 +1289,7 @@ static int ocfs2_alloc_dinode_update_counts(struct inode *inode,
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
 	struct ocfs2_chain_list *cl = (struct ocfs2_chain_list *) &di->id2.i_chain;
 
-	ret = ocfs2_journal_access_di(handle, inode, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -1461,7 +1462,7 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
 	/* Ok, claim our bits now: set the info on dinode, chainlist
 	 * and then the group */
 	status = ocfs2_journal_access_di(handle,
-					 alloc_inode,
+					 INODE_CACHE(alloc_inode),
 					 ac->ac_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
@@ -1907,8 +1908,8 @@ static inline int ocfs2_block_group_clear_bits(handle_t *handle,
 	if (ocfs2_is_cluster_bitmap(alloc_inode))
 		journal_type = OCFS2_JOURNAL_ACCESS_UNDO;
 
-	status = ocfs2_journal_access_gd(handle, alloc_inode, group_bh,
-					 journal_type);
+	status = ocfs2_journal_access_gd(handle, INODE_CACHE(alloc_inode),
+					 group_bh, journal_type);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -1993,8 +1994,8 @@ int ocfs2_free_suballoc_bits(handle_t *handle,
 		goto bail;
 	}
 
-	status = ocfs2_journal_access_di(handle, alloc_inode, alloc_bh,
-					 OCFS2_JOURNAL_ACCESS_WRITE);
+	status = ocfs2_journal_access_di(handle, INODE_CACHE(alloc_inode),
+					 alloc_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -2151,7 +2152,7 @@ int ocfs2_lock_allocators(struct inode *inode,
 
 	BUG_ON(clusters_to_add != 0 && data_ac == NULL);
 
-	num_free_extents = ocfs2_num_free_extents(osb, inode, et);
+	num_free_extents = ocfs2_num_free_extents(osb, et);
 	if (num_free_extents < 0) {
 		ret = num_free_extents;
 		mlog_errno(ret);
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index a3f8871d21fd..c0e48aeebb1c 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -28,7 +28,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
-#include <linux/utsname.h>
 #include <linux/init.h>
 #include <linux/random.h>
 #include <linux/statfs.h>
@@ -69,6 +68,7 @@
 #include "ver.h"
 #include "xattr.h"
 #include "quota.h"
+#include "refcounttree.h"
 
 #include "buffer_head_io.h"
 
@@ -373,7 +373,7 @@ static ssize_t ocfs2_debug_read(struct file *file, char __user *buf,
 }
 #endif	/* CONFIG_DEBUG_FS */
 
-static struct file_operations ocfs2_osb_debug_fops = {
+static const struct file_operations ocfs2_osb_debug_fops = {
 	.open =		ocfs2_osb_debug_open,
 	.release =	ocfs2_debug_release,
 	.read =		ocfs2_debug_read,
@@ -965,7 +965,7 @@ static int ocfs2_quota_off(struct super_block *sb, int type, int remount)
 	return vfs_quota_disable(sb, type, DQUOT_LIMITS_ENABLED);
 }
 
-static struct quotactl_ops ocfs2_quotactl_ops = {
+static const struct quotactl_ops ocfs2_quotactl_ops = {
 	.quota_on	= ocfs2_quota_on,
 	.quota_off	= ocfs2_quota_off,
 	.quota_sync	= vfs_quota_sync,
@@ -1668,8 +1668,6 @@ static void ocfs2_inode_init_once(void *data)
 	spin_lock_init(&oi->ip_lock);
 	ocfs2_extent_map_init(&oi->vfs_inode);
 	INIT_LIST_HEAD(&oi->ip_io_markers);
-	oi->ip_created_trans = 0;
-	oi->ip_last_trans = 0;
 	oi->ip_dir_start_lookup = 0;
 
 	init_rwsem(&oi->ip_alloc_sem);
@@ -1683,7 +1681,8 @@ static void ocfs2_inode_init_once(void *data)
 	ocfs2_lock_res_init_once(&oi->ip_inode_lockres);
 	ocfs2_lock_res_init_once(&oi->ip_open_lockres);
 
-	ocfs2_metadata_cache_init(&oi->vfs_inode);
+	ocfs2_metadata_cache_init(INODE_CACHE(&oi->vfs_inode),
+				  &ocfs2_inode_caching_ops);
 
 	inode_init_once(&oi->vfs_inode);
 }
@@ -1859,6 +1858,8 @@ static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err)
 
 	ocfs2_sync_blockdev(sb);
 
+	ocfs2_purge_refcount_trees(osb);
+
 	/* No cluster connection means we've failed during mount, so skip
 	 * all the steps which depended on that to complete. */
 	if (osb->cconn) {
@@ -2065,6 +2066,8 @@ static int ocfs2_initialize_super(struct super_block *sb,
 		goto bail;
 	}
 
+	osb->osb_rf_lock_tree = RB_ROOT;
+
 	osb->s_feature_compat =
 		le32_to_cpu(OCFS2_RAW_SB(di)->s_feature_compat);
 	osb->s_feature_ro_compat =
@@ -2490,7 +2493,8 @@ void __ocfs2_abort(struct super_block* sb,
 	/* Force a panic(). This stinks, but it's better than letting
 	 * things continue without having a proper hard readonly
 	 * here. */
-	OCFS2_SB(sb)->s_mount_opt |= OCFS2_MOUNT_ERRORS_PANIC;
+	if (!ocfs2_mount_local(OCFS2_SB(sb)))
+		OCFS2_SB(sb)->s_mount_opt |= OCFS2_MOUNT_ERRORS_PANIC;
 	ocfs2_handle_error(sb);
 }
 
diff --git a/fs/ocfs2/symlink.c b/fs/ocfs2/symlink.c
index 579dd1b1110f..e3421030a69f 100644
--- a/fs/ocfs2/symlink.c
+++ b/fs/ocfs2/symlink.c
@@ -38,7 +38,6 @@
 #include <linux/types.h>
 #include <linux/slab.h>
 #include <linux/pagemap.h>
-#include <linux/utsname.h>
 #include <linux/namei.h>
 
 #define MLOG_MASK_PREFIX ML_NAMEI
diff --git a/fs/ocfs2/uptodate.c b/fs/ocfs2/uptodate.c
index 187b99ff0368..b6284f235d2f 100644
--- a/fs/ocfs2/uptodate.c
+++ b/fs/ocfs2/uptodate.c
@@ -75,15 +75,77 @@ struct ocfs2_meta_cache_item {
 
 static struct kmem_cache *ocfs2_uptodate_cachep = NULL;
 
-void ocfs2_metadata_cache_init(struct inode *inode)
+u64 ocfs2_metadata_cache_owner(struct ocfs2_caching_info *ci)
 {
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
+	BUG_ON(!ci || !ci->ci_ops);
 
-	oi->ip_flags |= OCFS2_INODE_CACHE_INLINE;
+	return ci->ci_ops->co_owner(ci);
+}
+
+struct super_block *ocfs2_metadata_cache_get_super(struct ocfs2_caching_info *ci)
+{
+	BUG_ON(!ci || !ci->ci_ops);
+
+	return ci->ci_ops->co_get_super(ci);
+}
+
+static void ocfs2_metadata_cache_lock(struct ocfs2_caching_info *ci)
+{
+	BUG_ON(!ci || !ci->ci_ops);
+
+	ci->ci_ops->co_cache_lock(ci);
+}
+
+static void ocfs2_metadata_cache_unlock(struct ocfs2_caching_info *ci)
+{
+	BUG_ON(!ci || !ci->ci_ops);
+
+	ci->ci_ops->co_cache_unlock(ci);
+}
+
+void ocfs2_metadata_cache_io_lock(struct ocfs2_caching_info *ci)
+{
+	BUG_ON(!ci || !ci->ci_ops);
+
+	ci->ci_ops->co_io_lock(ci);
+}
+
+void ocfs2_metadata_cache_io_unlock(struct ocfs2_caching_info *ci)
+{
+	BUG_ON(!ci || !ci->ci_ops);
+
+	ci->ci_ops->co_io_unlock(ci);
+}
+
+
+static void ocfs2_metadata_cache_reset(struct ocfs2_caching_info *ci,
+				       int clear)
+{
+	ci->ci_flags |= OCFS2_CACHE_FL_INLINE;
 	ci->ci_num_cached = 0;
+
+	if (clear) {
+		ci->ci_created_trans = 0;
+		ci->ci_last_trans = 0;
+	}
+}
+
+void ocfs2_metadata_cache_init(struct ocfs2_caching_info *ci,
+			       const struct ocfs2_caching_operations *ops)
+{
+	BUG_ON(!ops);
+
+	ci->ci_ops = ops;
+	ocfs2_metadata_cache_reset(ci, 1);
 }
 
+void ocfs2_metadata_cache_exit(struct ocfs2_caching_info *ci)
+{
+	ocfs2_metadata_cache_purge(ci);
+	ocfs2_metadata_cache_reset(ci, 1);
+}
+
+
 /* No lock taken here as 'root' is not expected to be visible to other
  * processes. */
 static unsigned int ocfs2_purge_copied_metadata_tree(struct rb_root *root)
@@ -112,19 +174,20 @@ static unsigned int ocfs2_purge_copied_metadata_tree(struct rb_root *root)
  * This function is a few more lines longer than necessary due to some
  * accounting done here, but I think it's worth tracking down those
  * bugs sooner -- Mark */
-void ocfs2_metadata_cache_purge(struct inode *inode)
+void ocfs2_metadata_cache_purge(struct ocfs2_caching_info *ci)
 {
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	unsigned int tree, to_purge, purged;
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
 	struct rb_root root = RB_ROOT;
 
-	spin_lock(&oi->ip_lock);
-	tree = !(oi->ip_flags & OCFS2_INODE_CACHE_INLINE);
+	BUG_ON(!ci || !ci->ci_ops);
+
+	ocfs2_metadata_cache_lock(ci);
+	tree = !(ci->ci_flags & OCFS2_CACHE_FL_INLINE);
 	to_purge = ci->ci_num_cached;
 
-	mlog(0, "Purge %u %s items from Inode %llu\n", to_purge,
-	     tree ? "array" : "tree", (unsigned long long)oi->ip_blkno);
+	mlog(0, "Purge %u %s items from Owner %llu\n", to_purge,
+	     tree ? "array" : "tree",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci));
 
 	/* If we're a tree, save off the root so that we can safely
 	 * initialize the cache. We do the work to free tree members
@@ -132,16 +195,17 @@ void ocfs2_metadata_cache_purge(struct inode *inode)
 	if (tree)
 		root = ci->ci_cache.ci_tree;
 
-	ocfs2_metadata_cache_init(inode);
-	spin_unlock(&oi->ip_lock);
+	ocfs2_metadata_cache_reset(ci, 0);
+	ocfs2_metadata_cache_unlock(ci);
 
 	purged = ocfs2_purge_copied_metadata_tree(&root);
 	/* If possible, track the number wiped so that we can more
 	 * easily detect counting errors. Unfortunately, this is only
 	 * meaningful for trees. */
 	if (tree && purged != to_purge)
-		mlog(ML_ERROR, "Inode %llu, count = %u, purged = %u\n",
-		     (unsigned long long)oi->ip_blkno, to_purge, purged);
+		mlog(ML_ERROR, "Owner %llu, count = %u, purged = %u\n",
+		     (unsigned long long)ocfs2_metadata_cache_owner(ci),
+		     to_purge, purged);
 }
 
 /* Returns the index in the cache array, -1 if not found.
@@ -182,27 +246,25 @@ ocfs2_search_cache_tree(struct ocfs2_caching_info *ci,
 	return NULL;
 }
 
-static int ocfs2_buffer_cached(struct ocfs2_inode_info *oi,
+static int ocfs2_buffer_cached(struct ocfs2_caching_info *ci,
 			       struct buffer_head *bh)
 {
 	int index = -1;
 	struct ocfs2_meta_cache_item *item = NULL;
 
-	spin_lock(&oi->ip_lock);
+	ocfs2_metadata_cache_lock(ci);
 
-	mlog(0, "Inode %llu, query block %llu (inline = %u)\n",
-	     (unsigned long long)oi->ip_blkno,
+	mlog(0, "Owner %llu, query block %llu (inline = %u)\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
 	     (unsigned long long) bh->b_blocknr,
-	     !!(oi->ip_flags & OCFS2_INODE_CACHE_INLINE));
+	     !!(ci->ci_flags & OCFS2_CACHE_FL_INLINE));
 
-	if (oi->ip_flags & OCFS2_INODE_CACHE_INLINE)
-		index = ocfs2_search_cache_array(&oi->ip_metadata_cache,
-						 bh->b_blocknr);
+	if (ci->ci_flags & OCFS2_CACHE_FL_INLINE)
+		index = ocfs2_search_cache_array(ci, bh->b_blocknr);
 	else
-		item = ocfs2_search_cache_tree(&oi->ip_metadata_cache,
-					       bh->b_blocknr);
+		item = ocfs2_search_cache_tree(ci, bh->b_blocknr);
 
-	spin_unlock(&oi->ip_lock);
+	ocfs2_metadata_cache_unlock(ci);
 
 	mlog(0, "index = %d, item = %p\n", index, item);
 
@@ -214,7 +276,7 @@ static int ocfs2_buffer_cached(struct ocfs2_inode_info *oi,
  * 
  * This can be called under lock_buffer()
  */
-int ocfs2_buffer_uptodate(struct inode *inode,
+int ocfs2_buffer_uptodate(struct ocfs2_caching_info *ci,
 			  struct buffer_head *bh)
 {
 	/* Doesn't matter if the bh is in our cache or not -- if it's
@@ -230,24 +292,24 @@ int ocfs2_buffer_uptodate(struct inode *inode,
 
 	/* Ok, locally the buffer is marked as up to date, now search
 	 * our cache to see if we can trust that. */
-	return ocfs2_buffer_cached(OCFS2_I(inode), bh);
+	return ocfs2_buffer_cached(ci, bh);
 }
 
-/* 
+/*
  * Determine whether a buffer is currently out on a read-ahead request.
- * ip_io_sem should be held to serialize submitters with the logic here.
+ * ci_io_sem should be held to serialize submitters with the logic here.
  */
-int ocfs2_buffer_read_ahead(struct inode *inode,
+int ocfs2_buffer_read_ahead(struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh)
 {
-	return buffer_locked(bh) && ocfs2_buffer_cached(OCFS2_I(inode), bh);
+	return buffer_locked(bh) && ocfs2_buffer_cached(ci, bh);
 }
 
 /* Requires ip_lock */
 static void ocfs2_append_cache_array(struct ocfs2_caching_info *ci,
 				     sector_t block)
 {
-	BUG_ON(ci->ci_num_cached >= OCFS2_INODE_MAX_CACHE_ARRAY);
+	BUG_ON(ci->ci_num_cached >= OCFS2_CACHE_INFO_MAX_ARRAY);
 
 	mlog(0, "block %llu takes position %u\n", (unsigned long long) block,
 	     ci->ci_num_cached);
@@ -292,66 +354,64 @@ static void __ocfs2_insert_cache_tree(struct ocfs2_caching_info *ci,
 	ci->ci_num_cached++;
 }
 
-static inline int ocfs2_insert_can_use_array(struct ocfs2_inode_info *oi,
-					     struct ocfs2_caching_info *ci)
+/* co_cache_lock() must be held */
+static inline int ocfs2_insert_can_use_array(struct ocfs2_caching_info *ci)
 {
-	assert_spin_locked(&oi->ip_lock);
-
-	return (oi->ip_flags & OCFS2_INODE_CACHE_INLINE) &&
-		(ci->ci_num_cached < OCFS2_INODE_MAX_CACHE_ARRAY);
+	return (ci->ci_flags & OCFS2_CACHE_FL_INLINE) &&
+		(ci->ci_num_cached < OCFS2_CACHE_INFO_MAX_ARRAY);
 }
 
-/* tree should be exactly OCFS2_INODE_MAX_CACHE_ARRAY wide. NULL the
+/* tree should be exactly OCFS2_CACHE_INFO_MAX_ARRAY wide. NULL the
  * pointers in tree after we use them - this allows caller to detect
- * when to free in case of error. */
-static void ocfs2_expand_cache(struct ocfs2_inode_info *oi,
+ * when to free in case of error.
+ *
+ * The co_cache_lock() must be held. */
+static void ocfs2_expand_cache(struct ocfs2_caching_info *ci,
 			       struct ocfs2_meta_cache_item **tree)
 {
 	int i;
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
 
-	mlog_bug_on_msg(ci->ci_num_cached != OCFS2_INODE_MAX_CACHE_ARRAY,
-			"Inode %llu, num cached = %u, should be %u\n",
-			(unsigned long long)oi->ip_blkno, ci->ci_num_cached,
-			OCFS2_INODE_MAX_CACHE_ARRAY);
-	mlog_bug_on_msg(!(oi->ip_flags & OCFS2_INODE_CACHE_INLINE),
-			"Inode %llu not marked as inline anymore!\n",
-			(unsigned long long)oi->ip_blkno);
-	assert_spin_locked(&oi->ip_lock);
+	mlog_bug_on_msg(ci->ci_num_cached != OCFS2_CACHE_INFO_MAX_ARRAY,
+			"Owner %llu, num cached = %u, should be %u\n",
+			(unsigned long long)ocfs2_metadata_cache_owner(ci),
+			ci->ci_num_cached, OCFS2_CACHE_INFO_MAX_ARRAY);
+	mlog_bug_on_msg(!(ci->ci_flags & OCFS2_CACHE_FL_INLINE),
+			"Owner %llu not marked as inline anymore!\n",
+			(unsigned long long)ocfs2_metadata_cache_owner(ci));
 
 	/* Be careful to initialize the tree members *first* because
 	 * once the ci_tree is used, the array is junk... */
-	for(i = 0; i < OCFS2_INODE_MAX_CACHE_ARRAY; i++)
+	for (i = 0; i < OCFS2_CACHE_INFO_MAX_ARRAY; i++)
 		tree[i]->c_block = ci->ci_cache.ci_array[i];
 
-	oi->ip_flags &= ~OCFS2_INODE_CACHE_INLINE;
+	ci->ci_flags &= ~OCFS2_CACHE_FL_INLINE;
 	ci->ci_cache.ci_tree = RB_ROOT;
 	/* this will be set again by __ocfs2_insert_cache_tree */
 	ci->ci_num_cached = 0;
 
-	for(i = 0; i < OCFS2_INODE_MAX_CACHE_ARRAY; i++) {
+	for (i = 0; i < OCFS2_CACHE_INFO_MAX_ARRAY; i++) {
 		__ocfs2_insert_cache_tree(ci, tree[i]);
 		tree[i] = NULL;
 	}
 
 	mlog(0, "Expanded %llu to a tree cache: flags 0x%x, num = %u\n",
-	     (unsigned long long)oi->ip_blkno, oi->ip_flags, ci->ci_num_cached);
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
+	     ci->ci_flags, ci->ci_num_cached);
 }
 
 /* Slow path function - memory allocation is necessary. See the
  * comment above ocfs2_set_buffer_uptodate for more information. */
-static void __ocfs2_set_buffer_uptodate(struct ocfs2_inode_info *oi,
+static void __ocfs2_set_buffer_uptodate(struct ocfs2_caching_info *ci,
 					sector_t block,
 					int expand_tree)
 {
 	int i;
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
 	struct ocfs2_meta_cache_item *new = NULL;
-	struct ocfs2_meta_cache_item *tree[OCFS2_INODE_MAX_CACHE_ARRAY] =
+	struct ocfs2_meta_cache_item *tree[OCFS2_CACHE_INFO_MAX_ARRAY] =
 		{ NULL, };
 
-	mlog(0, "Inode %llu, block %llu, expand = %d\n",
-	     (unsigned long long)oi->ip_blkno,
+	mlog(0, "Owner %llu, block %llu, expand = %d\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
 	     (unsigned long long)block, expand_tree);
 
 	new = kmem_cache_alloc(ocfs2_uptodate_cachep, GFP_NOFS);
@@ -364,7 +424,7 @@ static void __ocfs2_set_buffer_uptodate(struct ocfs2_inode_info *oi,
 	if (expand_tree) {
 		/* Do *not* allocate an array here - the removal code
 		 * has no way of tracking that. */
-		for(i = 0; i < OCFS2_INODE_MAX_CACHE_ARRAY; i++) {
+		for (i = 0; i < OCFS2_CACHE_INFO_MAX_ARRAY; i++) {
 			tree[i] = kmem_cache_alloc(ocfs2_uptodate_cachep,
 						   GFP_NOFS);
 			if (!tree[i]) {
@@ -376,21 +436,21 @@ static void __ocfs2_set_buffer_uptodate(struct ocfs2_inode_info *oi,
 		}
 	}
 
-	spin_lock(&oi->ip_lock);
-	if (ocfs2_insert_can_use_array(oi, ci)) {
+	ocfs2_metadata_cache_lock(ci);
+	if (ocfs2_insert_can_use_array(ci)) {
 		mlog(0, "Someone cleared the tree underneath us\n");
 		/* Ok, items were removed from the cache in between
 		 * locks. Detect this and revert back to the fast path */
 		ocfs2_append_cache_array(ci, block);
-		spin_unlock(&oi->ip_lock);
+		ocfs2_metadata_cache_unlock(ci);
 		goto out_free;
 	}
 
 	if (expand_tree)
-		ocfs2_expand_cache(oi, tree);
+		ocfs2_expand_cache(ci, tree);
 
 	__ocfs2_insert_cache_tree(ci, new);
-	spin_unlock(&oi->ip_lock);
+	ocfs2_metadata_cache_unlock(ci);
 
 	new = NULL;
 out_free:
@@ -400,14 +460,14 @@ out_free:
 	/* If these were used, then ocfs2_expand_cache re-set them to
 	 * NULL for us. */
 	if (tree[0]) {
-		for(i = 0; i < OCFS2_INODE_MAX_CACHE_ARRAY; i++)
+		for (i = 0; i < OCFS2_CACHE_INFO_MAX_ARRAY; i++)
 			if (tree[i])
 				kmem_cache_free(ocfs2_uptodate_cachep,
 						tree[i]);
 	}
 }
 
-/* Item insertion is guarded by ip_io_mutex, so the insertion path takes
+/* Item insertion is guarded by co_io_lock(), so the insertion path takes
  * advantage of this by not rechecking for a duplicate insert during
  * the slow case. Additionally, if the cache needs to be bumped up to
  * a tree, the code will not recheck after acquiring the lock --
@@ -425,59 +485,55 @@ out_free:
  * Readahead buffers can be passed in here before the I/O request is
  * completed.
  */
-void ocfs2_set_buffer_uptodate(struct inode *inode,
+void ocfs2_set_buffer_uptodate(struct ocfs2_caching_info *ci,
 			       struct buffer_head *bh)
 {
 	int expand;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
 
 	/* The block may very well exist in our cache already, so avoid
 	 * doing any more work in that case. */
-	if (ocfs2_buffer_cached(oi, bh))
+	if (ocfs2_buffer_cached(ci, bh))
 		return;
 
-	mlog(0, "Inode %llu, inserting block %llu\n",
-	     (unsigned long long)oi->ip_blkno,
+	mlog(0, "Owner %llu, inserting block %llu\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
 	     (unsigned long long)bh->b_blocknr);
 
 	/* No need to recheck under spinlock - insertion is guarded by
-	 * ip_io_mutex */
-	spin_lock(&oi->ip_lock);
-	if (ocfs2_insert_can_use_array(oi, ci)) {
+	 * co_io_lock() */
+	ocfs2_metadata_cache_lock(ci);
+	if (ocfs2_insert_can_use_array(ci)) {
 		/* Fast case - it's an array and there's a free
 		 * spot. */
 		ocfs2_append_cache_array(ci, bh->b_blocknr);
-		spin_unlock(&oi->ip_lock);
+		ocfs2_metadata_cache_unlock(ci);
 		return;
 	}
 
 	expand = 0;
-	if (oi->ip_flags & OCFS2_INODE_CACHE_INLINE) {
+	if (ci->ci_flags & OCFS2_CACHE_FL_INLINE) {
 		/* We need to bump things up to a tree. */
 		expand = 1;
 	}
-	spin_unlock(&oi->ip_lock);
+	ocfs2_metadata_cache_unlock(ci);
 
-	__ocfs2_set_buffer_uptodate(oi, bh->b_blocknr, expand);
+	__ocfs2_set_buffer_uptodate(ci, bh->b_blocknr, expand);
 }
 
 /* Called against a newly allocated buffer. Most likely nobody should
  * be able to read this sort of metadata while it's still being
- * allocated, but this is careful to take ip_io_mutex anyway. */
-void ocfs2_set_new_buffer_uptodate(struct inode *inode,
+ * allocated, but this is careful to take co_io_lock() anyway. */
+void ocfs2_set_new_buffer_uptodate(struct ocfs2_caching_info *ci,
 				   struct buffer_head *bh)
 {
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-
 	/* This should definitely *not* exist in our cache */
-	BUG_ON(ocfs2_buffer_cached(oi, bh));
+	BUG_ON(ocfs2_buffer_cached(ci, bh));
 
 	set_buffer_uptodate(bh);
 
-	mutex_lock(&oi->ip_io_mutex);
-	ocfs2_set_buffer_uptodate(inode, bh);
-	mutex_unlock(&oi->ip_io_mutex);
+	ocfs2_metadata_cache_io_lock(ci);
+	ocfs2_set_buffer_uptodate(ci, bh);
+	ocfs2_metadata_cache_io_unlock(ci);
 }
 
 /* Requires ip_lock. */
@@ -487,7 +543,7 @@ static void ocfs2_remove_metadata_array(struct ocfs2_caching_info *ci,
 	sector_t *array = ci->ci_cache.ci_array;
 	int bytes;
 
-	BUG_ON(index < 0 || index >= OCFS2_INODE_MAX_CACHE_ARRAY);
+	BUG_ON(index < 0 || index >= OCFS2_CACHE_INFO_MAX_ARRAY);
 	BUG_ON(index >= ci->ci_num_cached);
 	BUG_ON(!ci->ci_num_cached);
 
@@ -515,21 +571,19 @@ static void ocfs2_remove_metadata_tree(struct ocfs2_caching_info *ci,
 	ci->ci_num_cached--;
 }
 
-static void ocfs2_remove_block_from_cache(struct inode *inode,
+static void ocfs2_remove_block_from_cache(struct ocfs2_caching_info *ci,
 					  sector_t block)
 {
 	int index;
 	struct ocfs2_meta_cache_item *item = NULL;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
 
-	spin_lock(&oi->ip_lock);
-	mlog(0, "Inode %llu, remove %llu, items = %u, array = %u\n",
-	     (unsigned long long)oi->ip_blkno,
+	ocfs2_metadata_cache_lock(ci);
+	mlog(0, "Owner %llu, remove %llu, items = %u, array = %u\n",
+	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
 	     (unsigned long long) block, ci->ci_num_cached,
-	     oi->ip_flags & OCFS2_INODE_CACHE_INLINE);
+	     ci->ci_flags & OCFS2_CACHE_FL_INLINE);
 
-	if (oi->ip_flags & OCFS2_INODE_CACHE_INLINE) {
+	if (ci->ci_flags & OCFS2_CACHE_FL_INLINE) {
 		index = ocfs2_search_cache_array(ci, block);
 		if (index != -1)
 			ocfs2_remove_metadata_array(ci, index);
@@ -538,7 +592,7 @@ static void ocfs2_remove_block_from_cache(struct inode *inode,
 		if (item)
 			ocfs2_remove_metadata_tree(ci, item);
 	}
-	spin_unlock(&oi->ip_lock);
+	ocfs2_metadata_cache_unlock(ci);
 
 	if (item)
 		kmem_cache_free(ocfs2_uptodate_cachep, item);
@@ -549,23 +603,24 @@ static void ocfs2_remove_block_from_cache(struct inode *inode,
  * bother reverting things to an inlined array in the case of a remove
  * which moves us back under the limit.
  */
-void ocfs2_remove_from_cache(struct inode *inode,
+void ocfs2_remove_from_cache(struct ocfs2_caching_info *ci,
 			     struct buffer_head *bh)
 {
 	sector_t block = bh->b_blocknr;
 
-	ocfs2_remove_block_from_cache(inode, block);
+	ocfs2_remove_block_from_cache(ci, block);
 }
 
 /* Called when we remove xattr clusters from an inode. */
-void ocfs2_remove_xattr_clusters_from_cache(struct inode *inode,
+void ocfs2_remove_xattr_clusters_from_cache(struct ocfs2_caching_info *ci,
 					    sector_t block,
 					    u32 c_len)
 {
-	unsigned int i, b_len = ocfs2_clusters_to_blocks(inode->i_sb, 1) * c_len;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	unsigned int i, b_len = ocfs2_clusters_to_blocks(sb, 1) * c_len;
 
 	for (i = 0; i < b_len; i++, block++)
-		ocfs2_remove_block_from_cache(inode, block);
+		ocfs2_remove_block_from_cache(ci, block);
 }
 
 int __init init_ocfs2_uptodate_cache(void)
@@ -577,7 +632,7 @@ int __init init_ocfs2_uptodate_cache(void)
 		return -ENOMEM;
 
 	mlog(0, "%u inlined cache items per inode.\n",
-	     OCFS2_INODE_MAX_CACHE_ARRAY);
+	     OCFS2_CACHE_INFO_MAX_ARRAY);
 
 	return 0;
 }
diff --git a/fs/ocfs2/uptodate.h b/fs/ocfs2/uptodate.h
index 531b4b3a0c47..0d826fe2da0d 100644
--- a/fs/ocfs2/uptodate.h
+++ b/fs/ocfs2/uptodate.h
@@ -26,24 +26,59 @@
 #ifndef OCFS2_UPTODATE_H
 #define OCFS2_UPTODATE_H
 
+/*
+ * The caching code relies on locking provided by the user of
+ * struct ocfs2_caching_info.  These operations connect that up.
+ */
+struct ocfs2_caching_operations {
+	/*
+	 * A u64 representing the owning structure.  Usually this
+	 * is the block number (i_blkno or whatnot).  This is used so
+	 * that caching log messages can identify the owning structure.
+	 */
+	u64	(*co_owner)(struct ocfs2_caching_info *ci);
+
+	/* The superblock is needed during I/O. */
+	struct super_block *(*co_get_super)(struct ocfs2_caching_info *ci);
+	/*
+	 * Lock and unlock the caching data.  These will not sleep, and
+	 * should probably be spinlocks.
+	 */
+	void	(*co_cache_lock)(struct ocfs2_caching_info *ci);
+	void	(*co_cache_unlock)(struct ocfs2_caching_info *ci);
+
+	/*
+	 * Lock and unlock for disk I/O.  These will sleep, and should
+	 * be mutexes.
+	 */
+	void	(*co_io_lock)(struct ocfs2_caching_info *ci);
+	void	(*co_io_unlock)(struct ocfs2_caching_info *ci);
+};
+
 int __init init_ocfs2_uptodate_cache(void);
 void exit_ocfs2_uptodate_cache(void);
 
-void ocfs2_metadata_cache_init(struct inode *inode);
-void ocfs2_metadata_cache_purge(struct inode *inode);
+void ocfs2_metadata_cache_init(struct ocfs2_caching_info *ci,
+			       const struct ocfs2_caching_operations *ops);
+void ocfs2_metadata_cache_purge(struct ocfs2_caching_info *ci);
+void ocfs2_metadata_cache_exit(struct ocfs2_caching_info *ci);
+
+u64 ocfs2_metadata_cache_owner(struct ocfs2_caching_info *ci);
+void ocfs2_metadata_cache_io_lock(struct ocfs2_caching_info *ci);
+void ocfs2_metadata_cache_io_unlock(struct ocfs2_caching_info *ci);
 
-int ocfs2_buffer_uptodate(struct inode *inode,
+int ocfs2_buffer_uptodate(struct ocfs2_caching_info *ci,
 			  struct buffer_head *bh);
-void ocfs2_set_buffer_uptodate(struct inode *inode,
+void ocfs2_set_buffer_uptodate(struct ocfs2_caching_info *ci,
 			       struct buffer_head *bh);
-void ocfs2_set_new_buffer_uptodate(struct inode *inode,
+void ocfs2_set_new_buffer_uptodate(struct ocfs2_caching_info *ci,
 				   struct buffer_head *bh);
-void ocfs2_remove_from_cache(struct inode *inode,
+void ocfs2_remove_from_cache(struct ocfs2_caching_info *ci,
 			     struct buffer_head *bh);
-void ocfs2_remove_xattr_clusters_from_cache(struct inode *inode,
+void ocfs2_remove_xattr_clusters_from_cache(struct ocfs2_caching_info *ci,
 					    sector_t block,
 					    u32 c_len);
-int ocfs2_buffer_read_ahead(struct inode *inode,
+int ocfs2_buffer_read_ahead(struct ocfs2_caching_info *ci,
 			    struct buffer_head *bh);
 
 #endif /* OCFS2_UPTODATE_H */
diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index d1a27cda984f..fe3419068df2 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -55,7 +55,8 @@
 #include "buffer_head_io.h"
 #include "super.h"
 #include "xattr.h"
-
+#include "refcounttree.h"
+#include "acl.h"
 
 struct ocfs2_xattr_def_value_root {
 	struct ocfs2_xattr_value_root	xv;
@@ -140,7 +141,7 @@ struct ocfs2_xattr_search {
 	int not_found;
 };
 
-static int ocfs2_xattr_bucket_get_name_value(struct inode *inode,
+static int ocfs2_xattr_bucket_get_name_value(struct super_block *sb,
 					     struct ocfs2_xattr_header *xh,
 					     int index,
 					     int *block_off,
@@ -157,7 +158,7 @@ static int ocfs2_xattr_index_block_find(struct inode *inode,
 					struct ocfs2_xattr_search *xs);
 
 static int ocfs2_xattr_tree_list_index_block(struct inode *inode,
-					struct ocfs2_xattr_tree_root *xt,
+					struct buffer_head *blk_bh,
 					char *buffer,
 					size_t buffer_size);
 
@@ -170,12 +171,42 @@ static int ocfs2_xattr_set_entry_index_block(struct inode *inode,
 					     struct ocfs2_xattr_search *xs,
 					     struct ocfs2_xattr_set_ctxt *ctxt);
 
-static int ocfs2_delete_xattr_index_block(struct inode *inode,
-					  struct buffer_head *xb_bh);
+typedef int (xattr_tree_rec_func)(struct inode *inode,
+				  struct buffer_head *root_bh,
+				  u64 blkno, u32 cpos, u32 len, void *para);
+static int ocfs2_iterate_xattr_index_block(struct inode *inode,
+					   struct buffer_head *root_bh,
+					   xattr_tree_rec_func *rec_func,
+					   void *para);
+static int ocfs2_delete_xattr_in_bucket(struct inode *inode,
+					struct ocfs2_xattr_bucket *bucket,
+					void *para);
+static int ocfs2_rm_xattr_cluster(struct inode *inode,
+				  struct buffer_head *root_bh,
+				  u64 blkno,
+				  u32 cpos,
+				  u32 len,
+				  void *para);
+
 static int ocfs2_mv_xattr_buckets(struct inode *inode, handle_t *handle,
 				  u64 src_blk, u64 last_blk, u64 to_blk,
 				  unsigned int start_bucket,
 				  u32 *first_hash);
+static int ocfs2_prepare_refcount_xattr(struct inode *inode,
+					struct ocfs2_dinode *di,
+					struct ocfs2_xattr_info *xi,
+					struct ocfs2_xattr_search *xis,
+					struct ocfs2_xattr_search *xbs,
+					struct ocfs2_refcount_tree **ref_tree,
+					int *meta_need,
+					int *credits);
+static int ocfs2_get_xattr_tree_value_root(struct super_block *sb,
+					   struct ocfs2_xattr_bucket *bucket,
+					   int offset,
+					   struct ocfs2_xattr_value_root **xv,
+					   struct buffer_head **bh);
+static int ocfs2_xattr_security_set(struct inode *inode, const char *name,
+				    const void *value, size_t size, int flags);
 
 static inline u16 ocfs2_xattr_buckets_per_cluster(struct ocfs2_super *osb)
 {
@@ -254,9 +285,9 @@ static int ocfs2_init_xattr_bucket(struct ocfs2_xattr_bucket *bucket,
 			break;
 		}
 
-		if (!ocfs2_buffer_uptodate(bucket->bu_inode,
+		if (!ocfs2_buffer_uptodate(INODE_CACHE(bucket->bu_inode),
 					   bucket->bu_bhs[i]))
-			ocfs2_set_new_buffer_uptodate(bucket->bu_inode,
+			ocfs2_set_new_buffer_uptodate(INODE_CACHE(bucket->bu_inode),
 						      bucket->bu_bhs[i]);
 	}
 
@@ -271,7 +302,7 @@ static int ocfs2_read_xattr_bucket(struct ocfs2_xattr_bucket *bucket,
 {
 	int rc;
 
-	rc = ocfs2_read_blocks(bucket->bu_inode, xb_blkno,
+	rc = ocfs2_read_blocks(INODE_CACHE(bucket->bu_inode), xb_blkno,
 			       bucket->bu_blocks, bucket->bu_bhs, 0,
 			       NULL);
 	if (!rc) {
@@ -297,7 +328,8 @@ static int ocfs2_xattr_bucket_journal_access(handle_t *handle,
 	int i, rc = 0;
 
 	for (i = 0; i < bucket->bu_blocks; i++) {
-		rc = ocfs2_journal_access(handle, bucket->bu_inode,
+		rc = ocfs2_journal_access(handle,
+					  INODE_CACHE(bucket->bu_inode),
 					  bucket->bu_bhs[i], type);
 		if (rc) {
 			mlog_errno(rc);
@@ -399,7 +431,7 @@ static int ocfs2_read_xattr_block(struct inode *inode, u64 xb_blkno,
 	int rc;
 	struct buffer_head *tmp = *bh;
 
-	rc = ocfs2_read_block(inode, xb_blkno, &tmp,
+	rc = ocfs2_read_block(INODE_CACHE(inode), xb_blkno, &tmp,
 			      ocfs2_validate_xattr_block);
 
 	/* If ocfs2_read_block() got us a new bh, pass it up. */
@@ -596,15 +628,14 @@ static int ocfs2_xattr_extend_allocation(struct inode *inode,
 	int status = 0;
 	handle_t *handle = ctxt->handle;
 	enum ocfs2_alloc_restarted why;
-	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	u32 prev_clusters, logical_start = le32_to_cpu(vb->vb_xv->xr_clusters);
 	struct ocfs2_extent_tree et;
 
 	mlog(0, "(clusters_to_add for xattr= %u)\n", clusters_to_add);
 
-	ocfs2_init_xattr_value_extent_tree(&et, inode, vb);
+	ocfs2_init_xattr_value_extent_tree(&et, INODE_CACHE(inode), vb);
 
-	status = vb->vb_access(handle, inode, vb->vb_bh,
+	status = vb->vb_access(handle, INODE_CACHE(inode), vb->vb_bh,
 			      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
@@ -612,13 +643,11 @@ static int ocfs2_xattr_extend_allocation(struct inode *inode,
 	}
 
 	prev_clusters = le32_to_cpu(vb->vb_xv->xr_clusters);
-	status = ocfs2_add_clusters_in_btree(osb,
-					     inode,
+	status = ocfs2_add_clusters_in_btree(handle,
+					     &et,
 					     &logical_start,
 					     clusters_to_add,
 					     0,
-					     &et,
-					     handle,
 					     ctxt->data_ac,
 					     ctxt->meta_ac,
 					     &why);
@@ -649,6 +678,7 @@ leave:
 static int __ocfs2_remove_xattr_range(struct inode *inode,
 				      struct ocfs2_xattr_value_buf *vb,
 				      u32 cpos, u32 phys_cpos, u32 len,
+				      unsigned int ext_flags,
 				      struct ocfs2_xattr_set_ctxt *ctxt)
 {
 	int ret;
@@ -656,16 +686,16 @@ static int __ocfs2_remove_xattr_range(struct inode *inode,
 	handle_t *handle = ctxt->handle;
 	struct ocfs2_extent_tree et;
 
-	ocfs2_init_xattr_value_extent_tree(&et, inode, vb);
+	ocfs2_init_xattr_value_extent_tree(&et, INODE_CACHE(inode), vb);
 
-	ret = vb->vb_access(handle, inode, vb->vb_bh,
+	ret = vb->vb_access(handle, INODE_CACHE(inode), vb->vb_bh,
 			    OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
 		goto out;
 	}
 
-	ret = ocfs2_remove_extent(inode, &et, cpos, len, handle, ctxt->meta_ac,
+	ret = ocfs2_remove_extent(handle, &et, cpos, len, ctxt->meta_ac,
 				  &ctxt->dealloc);
 	if (ret) {
 		mlog_errno(ret);
@@ -680,7 +710,14 @@ static int __ocfs2_remove_xattr_range(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_cache_cluster_dealloc(&ctxt->dealloc, phys_blkno, len);
+	if (ext_flags & OCFS2_EXT_REFCOUNTED)
+		ret = ocfs2_decrease_refcount(inode, handle,
+					ocfs2_blocks_to_clusters(inode->i_sb,
+								 phys_blkno),
+					len, ctxt->meta_ac, &ctxt->dealloc, 1);
+	else
+		ret = ocfs2_cache_cluster_dealloc(&ctxt->dealloc,
+						  phys_blkno, len);
 	if (ret)
 		mlog_errno(ret);
 
@@ -695,6 +732,7 @@ static int ocfs2_xattr_shrink_size(struct inode *inode,
 				   struct ocfs2_xattr_set_ctxt *ctxt)
 {
 	int ret = 0;
+	unsigned int ext_flags;
 	u32 trunc_len, cpos, phys_cpos, alloc_size;
 	u64 block;
 
@@ -706,7 +744,7 @@ static int ocfs2_xattr_shrink_size(struct inode *inode,
 	while (trunc_len) {
 		ret = ocfs2_xattr_get_clusters(inode, cpos, &phys_cpos,
 					       &alloc_size,
-					       &vb->vb_xv->xr_list);
+					       &vb->vb_xv->xr_list, &ext_flags);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -717,15 +755,15 @@ static int ocfs2_xattr_shrink_size(struct inode *inode,
 
 		ret = __ocfs2_remove_xattr_range(inode, vb, cpos,
 						 phys_cpos, alloc_size,
-						 ctxt);
+						 ext_flags, ctxt);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
 		block = ocfs2_clusters_to_blocks(inode->i_sb, phys_cpos);
-		ocfs2_remove_xattr_clusters_from_cache(inode, block,
-						       alloc_size);
+		ocfs2_remove_xattr_clusters_from_cache(INODE_CACHE(inode),
+						       block, alloc_size);
 		cpos += alloc_size;
 		trunc_len -= alloc_size;
 	}
@@ -810,6 +848,23 @@ static int ocfs2_xattr_list_entries(struct inode *inode,
 	return result;
 }
 
+int ocfs2_has_inline_xattr_value_outside(struct inode *inode,
+					 struct ocfs2_dinode *di)
+{
+	struct ocfs2_xattr_header *xh;
+	int i;
+
+	xh = (struct ocfs2_xattr_header *)
+		 ((void *)di + inode->i_sb->s_blocksize -
+		 le16_to_cpu(di->i_xattr_inline_size));
+
+	for (i = 0; i < le16_to_cpu(xh->xh_count); i++)
+		if (!ocfs2_xattr_is_local(&xh->xh_entries[i]))
+			return 1;
+
+	return 0;
+}
+
 static int ocfs2_xattr_ibody_list(struct inode *inode,
 				  struct ocfs2_dinode *di,
 				  char *buffer,
@@ -855,11 +910,9 @@ static int ocfs2_xattr_block_list(struct inode *inode,
 		struct ocfs2_xattr_header *header = &xb->xb_attrs.xb_header;
 		ret = ocfs2_xattr_list_entries(inode, header,
 					       buffer, buffer_size);
-	} else {
-		struct ocfs2_xattr_tree_root *xt = &xb->xb_attrs.xb_root;
-		ret = ocfs2_xattr_tree_list_index_block(inode, xt,
+	} else
+		ret = ocfs2_xattr_tree_list_index_block(inode, blk_bh,
 						   buffer, buffer_size);
-	}
 
 	brelse(blk_bh);
 
@@ -961,7 +1014,7 @@ static int ocfs2_xattr_get_value_outside(struct inode *inode,
 	cpos = 0;
 	while (cpos < clusters) {
 		ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
-					       &num_clusters, el);
+					       &num_clusters, el, NULL);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -970,7 +1023,8 @@ static int ocfs2_xattr_get_value_outside(struct inode *inode,
 		blkno = ocfs2_clusters_to_blocks(inode->i_sb, p_cluster);
 		/* Copy ocfs2_xattr_value */
 		for (i = 0; i < num_clusters * bpc; i++, blkno++) {
-			ret = ocfs2_read_block(inode, blkno, &bh, NULL);
+			ret = ocfs2_read_block(INODE_CACHE(inode), blkno,
+					       &bh, NULL);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
@@ -1085,7 +1139,7 @@ static int ocfs2_xattr_block_get(struct inode *inode,
 		i = xs->here - xs->header->xh_entries;
 
 		if (le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED) {
-			ret = ocfs2_xattr_bucket_get_name_value(inode,
+			ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb,
 								bucket_xh(xs->bucket),
 								i,
 								&block_off,
@@ -1183,7 +1237,7 @@ static int ocfs2_xattr_get(struct inode *inode,
 
 static int __ocfs2_xattr_set_value_outside(struct inode *inode,
 					   handle_t *handle,
-					   struct ocfs2_xattr_value_root *xv,
+					   struct ocfs2_xattr_value_buf *vb,
 					   const void *value,
 					   int value_len)
 {
@@ -1194,28 +1248,34 @@ static int __ocfs2_xattr_set_value_outside(struct inode *inode,
 	u32 clusters = ocfs2_clusters_for_bytes(inode->i_sb, value_len);
 	u64 blkno;
 	struct buffer_head *bh = NULL;
+	unsigned int ext_flags;
+	struct ocfs2_xattr_value_root *xv = vb->vb_xv;
 
 	BUG_ON(clusters > le32_to_cpu(xv->xr_clusters));
 
 	while (cpos < clusters) {
 		ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
-					       &num_clusters, &xv->xr_list);
+					       &num_clusters, &xv->xr_list,
+					       &ext_flags);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
 
+		BUG_ON(ext_flags & OCFS2_EXT_REFCOUNTED);
+
 		blkno = ocfs2_clusters_to_blocks(inode->i_sb, p_cluster);
 
 		for (i = 0; i < num_clusters * bpc; i++, blkno++) {
-			ret = ocfs2_read_block(inode, blkno, &bh, NULL);
+			ret = ocfs2_read_block(INODE_CACHE(inode), blkno,
+					       &bh, NULL);
 			if (ret) {
 				mlog_errno(ret);
 				goto out;
 			}
 
 			ret = ocfs2_journal_access(handle,
-						   inode,
+						   INODE_CACHE(inode),
 						   bh,
 						   OCFS2_JOURNAL_ACCESS_WRITE);
 			if (ret < 0) {
@@ -1266,7 +1326,7 @@ static int ocfs2_xattr_cleanup(struct inode *inode,
 	void *val = xs->base + offs;
 	size_t size = OCFS2_XATTR_SIZE(name_len) + OCFS2_XATTR_ROOT_SIZE;
 
-	ret = vb->vb_access(handle, inode, vb->vb_bh,
+	ret = vb->vb_access(handle, INODE_CACHE(inode), vb->vb_bh,
 			    OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1294,7 +1354,7 @@ static int ocfs2_xattr_update_entry(struct inode *inode,
 {
 	int ret;
 
-	ret = vb->vb_access(handle, inode, vb->vb_bh,
+	ret = vb->vb_access(handle, INODE_CACHE(inode), vb->vb_bh,
 			    OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1355,7 +1415,7 @@ static int ocfs2_xattr_set_value_outside(struct inode *inode,
 		mlog_errno(ret);
 		return ret;
 	}
-	ret = __ocfs2_xattr_set_value_outside(inode, ctxt->handle, vb->vb_xv,
+	ret = __ocfs2_xattr_set_value_outside(inode, ctxt->handle, vb,
 					      xi->value, xi->value_len);
 	if (ret < 0)
 		mlog_errno(ret);
@@ -1594,7 +1654,7 @@ static int ocfs2_xattr_set_entry(struct inode *inode,
 
 				ret = __ocfs2_xattr_set_value_outside(inode,
 								handle,
-								vb.vb_xv,
+								&vb,
 								xi->value,
 								xi->value_len);
 				if (ret < 0)
@@ -1615,7 +1675,7 @@ static int ocfs2_xattr_set_entry(struct inode *inode,
 		}
 	}
 
-	ret = ocfs2_journal_access_di(handle, inode, xs->inode_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), xs->inode_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1623,7 +1683,7 @@ static int ocfs2_xattr_set_entry(struct inode *inode,
 	}
 
 	if (!(flag & OCFS2_INLINE_XATTR_FL)) {
-		ret = vb.vb_access(handle, inode, vb.vb_bh,
+		ret = vb.vb_access(handle, INODE_CACHE(inode), vb.vb_bh,
 				   OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
 			mlog_errno(ret);
@@ -1700,51 +1760,112 @@ out:
 	return ret;
 }
 
+/*
+ * In xattr remove, if it is stored outside and refcounted, we may have
+ * the chance to split the refcount tree. So need the allocators.
+ */
+static int ocfs2_lock_xattr_remove_allocators(struct inode *inode,
+					struct ocfs2_xattr_value_root *xv,
+					struct ocfs2_caching_info *ref_ci,
+					struct buffer_head *ref_root_bh,
+					struct ocfs2_alloc_context **meta_ac,
+					int *ref_credits)
+{
+	int ret, meta_add = 0;
+	u32 p_cluster, num_clusters;
+	unsigned int ext_flags;
+
+	*ref_credits = 0;
+	ret = ocfs2_xattr_get_clusters(inode, 0, &p_cluster,
+				       &num_clusters,
+				       &xv->xr_list,
+				       &ext_flags);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (!(ext_flags & OCFS2_EXT_REFCOUNTED))
+		goto out;
+
+	ret = ocfs2_refcounted_xattr_delete_need(inode, ref_ci,
+						 ref_root_bh, xv,
+						 &meta_add, ref_credits);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
+						meta_add, meta_ac);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	return ret;
+}
+
 static int ocfs2_remove_value_outside(struct inode*inode,
 				      struct ocfs2_xattr_value_buf *vb,
-				      struct ocfs2_xattr_header *header)
+				      struct ocfs2_xattr_header *header,
+				      struct ocfs2_caching_info *ref_ci,
+				      struct buffer_head *ref_root_bh)
 {
-	int ret = 0, i;
+	int ret = 0, i, ref_credits;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_xattr_set_ctxt ctxt = { NULL, NULL, };
+	void *val;
 
 	ocfs2_init_dealloc_ctxt(&ctxt.dealloc);
 
-	ctxt.handle = ocfs2_start_trans(osb,
-					ocfs2_remove_extent_credits(osb->sb));
-	if (IS_ERR(ctxt.handle)) {
-		ret = PTR_ERR(ctxt.handle);
-		mlog_errno(ret);
-		goto out;
-	}
-
 	for (i = 0; i < le16_to_cpu(header->xh_count); i++) {
 		struct ocfs2_xattr_entry *entry = &header->xh_entries[i];
 
-		if (!ocfs2_xattr_is_local(entry)) {
-			void *val;
+		if (ocfs2_xattr_is_local(entry))
+			continue;
 
-			val = (void *)header +
-				le16_to_cpu(entry->xe_name_offset);
-			vb->vb_xv = (struct ocfs2_xattr_value_root *)
-				(val + OCFS2_XATTR_SIZE(entry->xe_name_len));
-			ret = ocfs2_xattr_value_truncate(inode, vb, 0, &ctxt);
-			if (ret < 0) {
-				mlog_errno(ret);
-				break;
-			}
+		val = (void *)header +
+			le16_to_cpu(entry->xe_name_offset);
+		vb->vb_xv = (struct ocfs2_xattr_value_root *)
+			(val + OCFS2_XATTR_SIZE(entry->xe_name_len));
+
+		ret = ocfs2_lock_xattr_remove_allocators(inode, vb->vb_xv,
+							 ref_ci, ref_root_bh,
+							 &ctxt.meta_ac,
+							 &ref_credits);
+
+		ctxt.handle = ocfs2_start_trans(osb, ref_credits +
+					ocfs2_remove_extent_credits(osb->sb));
+		if (IS_ERR(ctxt.handle)) {
+			ret = PTR_ERR(ctxt.handle);
+			mlog_errno(ret);
+			break;
+		}
+
+		ret = ocfs2_xattr_value_truncate(inode, vb, 0, &ctxt);
+		if (ret < 0) {
+			mlog_errno(ret);
+			break;
+		}
+
+		ocfs2_commit_trans(osb, ctxt.handle);
+		if (ctxt.meta_ac) {
+			ocfs2_free_alloc_context(ctxt.meta_ac);
+			ctxt.meta_ac = NULL;
 		}
 	}
 
-	ocfs2_commit_trans(osb, ctxt.handle);
+	if (ctxt.meta_ac)
+		ocfs2_free_alloc_context(ctxt.meta_ac);
 	ocfs2_schedule_truncate_log_flush(osb, 1);
 	ocfs2_run_deallocs(osb, &ctxt.dealloc);
-out:
 	return ret;
 }
 
 static int ocfs2_xattr_ibody_remove(struct inode *inode,
-				    struct buffer_head *di_bh)
+				    struct buffer_head *di_bh,
+				    struct ocfs2_caching_info *ref_ci,
+				    struct buffer_head *ref_root_bh)
 {
 
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
@@ -1759,13 +1880,21 @@ static int ocfs2_xattr_ibody_remove(struct inode *inode,
 		 ((void *)di + inode->i_sb->s_blocksize -
 		 le16_to_cpu(di->i_xattr_inline_size));
 
-	ret = ocfs2_remove_value_outside(inode, &vb, header);
+	ret = ocfs2_remove_value_outside(inode, &vb, header,
+					 ref_ci, ref_root_bh);
 
 	return ret;
 }
 
+struct ocfs2_rm_xattr_bucket_para {
+	struct ocfs2_caching_info *ref_ci;
+	struct buffer_head *ref_root_bh;
+};
+
 static int ocfs2_xattr_block_remove(struct inode *inode,
-				    struct buffer_head *blk_bh)
+				    struct buffer_head *blk_bh,
+				    struct ocfs2_caching_info *ref_ci,
+				    struct buffer_head *ref_root_bh)
 {
 	struct ocfs2_xattr_block *xb;
 	int ret = 0;
@@ -1773,19 +1902,29 @@ static int ocfs2_xattr_block_remove(struct inode *inode,
 		.vb_bh = blk_bh,
 		.vb_access = ocfs2_journal_access_xb,
 	};
+	struct ocfs2_rm_xattr_bucket_para args = {
+		.ref_ci = ref_ci,
+		.ref_root_bh = ref_root_bh,
+	};
 
 	xb = (struct ocfs2_xattr_block *)blk_bh->b_data;
 	if (!(le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED)) {
 		struct ocfs2_xattr_header *header = &(xb->xb_attrs.xb_header);
-		ret = ocfs2_remove_value_outside(inode, &vb, header);
+		ret = ocfs2_remove_value_outside(inode, &vb, header,
+						 ref_ci, ref_root_bh);
 	} else
-		ret = ocfs2_delete_xattr_index_block(inode, blk_bh);
+		ret = ocfs2_iterate_xattr_index_block(inode,
+						blk_bh,
+						ocfs2_rm_xattr_cluster,
+						&args);
 
 	return ret;
 }
 
 static int ocfs2_xattr_free_block(struct inode *inode,
-				  u64 block)
+				  u64 block,
+				  struct ocfs2_caching_info *ref_ci,
+				  struct buffer_head *ref_root_bh)
 {
 	struct inode *xb_alloc_inode;
 	struct buffer_head *xb_alloc_bh = NULL;
@@ -1803,7 +1942,7 @@ static int ocfs2_xattr_free_block(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_xattr_block_remove(inode, blk_bh);
+	ret = ocfs2_xattr_block_remove(inode, blk_bh, ref_ci, ref_root_bh);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out;
@@ -1863,6 +2002,9 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 {
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+	struct ocfs2_refcount_tree *ref_tree = NULL;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_caching_info *ref_ci = NULL;
 	handle_t *handle;
 	int ret;
 
@@ -1872,8 +2014,21 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 	if (!(oi->ip_dyn_features & OCFS2_HAS_XATTR_FL))
 		return 0;
 
+	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL) {
+		ret = ocfs2_lock_refcount_tree(OCFS2_SB(inode->i_sb),
+					       le64_to_cpu(di->i_refcount_loc),
+					       1, &ref_tree, &ref_root_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+		ref_ci = &ref_tree->rf_ci;
+
+	}
+
 	if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL) {
-		ret = ocfs2_xattr_ibody_remove(inode, di_bh);
+		ret = ocfs2_xattr_ibody_remove(inode, di_bh,
+					       ref_ci, ref_root_bh);
 		if (ret < 0) {
 			mlog_errno(ret);
 			goto out;
@@ -1882,7 +2037,8 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 
 	if (di->i_xattr_loc) {
 		ret = ocfs2_xattr_free_block(inode,
-					     le64_to_cpu(di->i_xattr_loc));
+					     le64_to_cpu(di->i_xattr_loc),
+					     ref_ci, ref_root_bh);
 		if (ret < 0) {
 			mlog_errno(ret);
 			goto out;
@@ -1896,7 +2052,7 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 		mlog_errno(ret);
 		goto out;
 	}
-	ret = ocfs2_journal_access_di(handle, inode, di_bh,
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -1916,6 +2072,9 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 out_commit:
 	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
 out:
+	if (ref_tree)
+		ocfs2_unlock_refcount_tree(OCFS2_SB(inode->i_sb), ref_tree, 1);
+	brelse(ref_root_bh);
 	return ret;
 }
 
@@ -2083,6 +2242,84 @@ cleanup:
 	return ret;
 }
 
+static int ocfs2_create_xattr_block(handle_t *handle,
+				    struct inode *inode,
+				    struct buffer_head *inode_bh,
+				    struct ocfs2_alloc_context *meta_ac,
+				    struct buffer_head **ret_bh,
+				    int indexed)
+{
+	int ret;
+	u16 suballoc_bit_start;
+	u32 num_got;
+	u64 first_blkno;
+	struct ocfs2_dinode *di =  (struct ocfs2_dinode *)inode_bh->b_data;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct buffer_head *new_bh = NULL;
+	struct ocfs2_xattr_block *xblk;
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), inode_bh,
+				      OCFS2_JOURNAL_ACCESS_CREATE);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto end;
+	}
+
+	ret = ocfs2_claim_metadata(osb, handle, meta_ac, 1,
+				   &suballoc_bit_start, &num_got,
+				   &first_blkno);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto end;
+	}
+
+	new_bh = sb_getblk(inode->i_sb, first_blkno);
+	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), new_bh);
+
+	ret = ocfs2_journal_access_xb(handle, INODE_CACHE(inode),
+				      new_bh,
+				      OCFS2_JOURNAL_ACCESS_CREATE);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto end;
+	}
+
+	/* Initialize ocfs2_xattr_block */
+	xblk = (struct ocfs2_xattr_block *)new_bh->b_data;
+	memset(xblk, 0, inode->i_sb->s_blocksize);
+	strcpy((void *)xblk, OCFS2_XATTR_BLOCK_SIGNATURE);
+	xblk->xb_suballoc_slot = cpu_to_le16(osb->slot_num);
+	xblk->xb_suballoc_bit = cpu_to_le16(suballoc_bit_start);
+	xblk->xb_fs_generation = cpu_to_le32(osb->fs_generation);
+	xblk->xb_blkno = cpu_to_le64(first_blkno);
+
+	if (indexed) {
+		struct ocfs2_xattr_tree_root *xr = &xblk->xb_attrs.xb_root;
+		xr->xt_clusters = cpu_to_le32(1);
+		xr->xt_last_eb_blk = 0;
+		xr->xt_list.l_tree_depth = 0;
+		xr->xt_list.l_count = cpu_to_le16(
+					ocfs2_xattr_recs_per_xb(inode->i_sb));
+		xr->xt_list.l_next_free_rec = cpu_to_le16(1);
+		xblk->xb_flags = cpu_to_le16(OCFS2_XATTR_INDEXED);
+	}
+
+	ret = ocfs2_journal_dirty(handle, new_bh);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto end;
+	}
+	di->i_xattr_loc = cpu_to_le64(first_blkno);
+	ocfs2_journal_dirty(handle, inode_bh);
+
+	*ret_bh = new_bh;
+	new_bh = NULL;
+
+end:
+	brelse(new_bh);
+	return ret;
+}
+
 /*
  * ocfs2_xattr_block_set()
  *
@@ -2095,63 +2332,24 @@ static int ocfs2_xattr_block_set(struct inode *inode,
 				 struct ocfs2_xattr_set_ctxt *ctxt)
 {
 	struct buffer_head *new_bh = NULL;
-	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_dinode *di =  (struct ocfs2_dinode *)xs->inode_bh->b_data;
 	handle_t *handle = ctxt->handle;
 	struct ocfs2_xattr_block *xblk = NULL;
-	u16 suballoc_bit_start;
-	u32 num_got;
-	u64 first_blkno;
 	int ret;
 
 	if (!xs->xattr_bh) {
-		ret = ocfs2_journal_access_di(handle, inode, xs->inode_bh,
-					      OCFS2_JOURNAL_ACCESS_CREATE);
-		if (ret < 0) {
-			mlog_errno(ret);
-			goto end;
-		}
-
-		ret = ocfs2_claim_metadata(osb, handle, ctxt->meta_ac, 1,
-					   &suballoc_bit_start, &num_got,
-					   &first_blkno);
-		if (ret < 0) {
-			mlog_errno(ret);
-			goto end;
-		}
-
-		new_bh = sb_getblk(inode->i_sb, first_blkno);
-		ocfs2_set_new_buffer_uptodate(inode, new_bh);
-
-		ret = ocfs2_journal_access_xb(handle, inode, new_bh,
-					      OCFS2_JOURNAL_ACCESS_CREATE);
-		if (ret < 0) {
+		ret = ocfs2_create_xattr_block(handle, inode, xs->inode_bh,
+					       ctxt->meta_ac, &new_bh, 0);
+		if (ret) {
 			mlog_errno(ret);
 			goto end;
 		}
 
-		/* Initialize ocfs2_xattr_block */
 		xs->xattr_bh = new_bh;
-		xblk = (struct ocfs2_xattr_block *)new_bh->b_data;
-		memset(xblk, 0, inode->i_sb->s_blocksize);
-		strcpy((void *)xblk, OCFS2_XATTR_BLOCK_SIGNATURE);
-		xblk->xb_suballoc_slot = cpu_to_le16(osb->slot_num);
-		xblk->xb_suballoc_bit = cpu_to_le16(suballoc_bit_start);
-		xblk->xb_fs_generation = cpu_to_le32(osb->fs_generation);
-		xblk->xb_blkno = cpu_to_le64(first_blkno);
-
+		xblk = (struct ocfs2_xattr_block *)xs->xattr_bh->b_data;
 		xs->header = &xblk->xb_attrs.xb_header;
 		xs->base = (void *)xs->header;
 		xs->end = (void *)xblk + inode->i_sb->s_blocksize;
 		xs->here = xs->header->xh_entries;
-
-		ret = ocfs2_journal_dirty(handle, new_bh);
-		if (ret < 0) {
-			mlog_errno(ret);
-			goto end;
-		}
-		di->i_xattr_loc = cpu_to_le64(first_blkno);
-		ocfs2_journal_dirty(handle, xs->inode_bh);
 	} else
 		xblk = (struct ocfs2_xattr_block *)xs->xattr_bh->b_data;
 
@@ -2273,7 +2471,7 @@ static int ocfs2_calc_xattr_set_need(struct inode *inode,
 		old_in_xb = 1;
 
 		if (le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED) {
-			ret = ocfs2_xattr_bucket_get_name_value(inode,
+			ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb,
 							bucket_xh(xbs->bucket),
 							i, &block_off,
 							&name_offset);
@@ -2428,6 +2626,7 @@ static int ocfs2_init_xattr_set_ctxt(struct inode *inode,
 				     struct ocfs2_xattr_search *xis,
 				     struct ocfs2_xattr_search *xbs,
 				     struct ocfs2_xattr_set_ctxt *ctxt,
+				     int extra_meta,
 				     int *credits)
 {
 	int clusters_add, meta_add, ret;
@@ -2444,6 +2643,7 @@ static int ocfs2_init_xattr_set_ctxt(struct inode *inode,
 		return ret;
 	}
 
+	meta_add += extra_meta;
 	mlog(0, "Set xattr %s, reserve meta blocks = %d, clusters = %d, "
 	     "credits = %d\n", xi->name, meta_add, clusters_add, *credits);
 
@@ -2598,7 +2798,7 @@ static int __ocfs2_xattr_set_handle(struct inode *inode,
 
 	if (!ret) {
 		/* Update inode ctime. */
-		ret = ocfs2_journal_access_di(ctxt->handle, inode,
+		ret = ocfs2_journal_access_di(ctxt->handle, INODE_CACHE(inode),
 					      xis->inode_bh,
 					      OCFS2_JOURNAL_ACCESS_WRITE);
 		if (ret) {
@@ -2711,10 +2911,11 @@ int ocfs2_xattr_set(struct inode *inode,
 {
 	struct buffer_head *di_bh = NULL;
 	struct ocfs2_dinode *di;
-	int ret, credits;
+	int ret, credits, ref_meta = 0, ref_credits = 0;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct inode *tl_inode = osb->osb_tl_inode;
 	struct ocfs2_xattr_set_ctxt ctxt = { NULL, NULL, };
+	struct ocfs2_refcount_tree *ref_tree = NULL;
 
 	struct ocfs2_xattr_info xi = {
 		.name_index = name_index,
@@ -2779,6 +2980,17 @@ int ocfs2_xattr_set(struct inode *inode,
 			goto cleanup;
 	}
 
+	/* Check whether the value is refcounted and do some prepartion. */
+	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL &&
+	    (!xis.not_found || !xbs.not_found)) {
+		ret = ocfs2_prepare_refcount_xattr(inode, di, &xi,
+						   &xis, &xbs, &ref_tree,
+						   &ref_meta, &ref_credits);
+		if (ret) {
+			mlog_errno(ret);
+			goto cleanup;
+		}
+	}
 
 	mutex_lock(&tl_inode->i_mutex);
 
@@ -2793,7 +3005,7 @@ int ocfs2_xattr_set(struct inode *inode,
 	mutex_unlock(&tl_inode->i_mutex);
 
 	ret = ocfs2_init_xattr_set_ctxt(inode, di, &xi, &xis,
-					&xbs, &ctxt, &credits);
+					&xbs, &ctxt, ref_meta, &credits);
 	if (ret) {
 		mlog_errno(ret);
 		goto cleanup;
@@ -2801,7 +3013,7 @@ int ocfs2_xattr_set(struct inode *inode,
 
 	/* we need to update inode's ctime field, so add credit for it. */
 	credits += OCFS2_INODE_UPDATE_CREDITS;
-	ctxt.handle = ocfs2_start_trans(osb, credits);
+	ctxt.handle = ocfs2_start_trans(osb, credits + ref_credits);
 	if (IS_ERR(ctxt.handle)) {
 		ret = PTR_ERR(ctxt.handle);
 		mlog_errno(ret);
@@ -2819,8 +3031,16 @@ int ocfs2_xattr_set(struct inode *inode,
 	if (ocfs2_dealloc_has_cluster(&ctxt.dealloc))
 		ocfs2_schedule_truncate_log_flush(osb, 1);
 	ocfs2_run_deallocs(osb, &ctxt.dealloc);
+
 cleanup:
+	if (ref_tree)
+		ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
 	up_write(&OCFS2_I(inode)->ip_xattr_sem);
+	if (!value && !ret) {
+		ret = ocfs2_try_remove_refcount_tree(inode, di_bh);
+		if (ret)
+			mlog_errno(ret);
+	}
 	ocfs2_inode_unlock(inode, 1);
 cleanup_nolock:
 	brelse(di_bh);
@@ -2849,7 +3069,8 @@ static int ocfs2_xattr_get_rec(struct inode *inode,
 	u64 e_blkno = 0;
 
 	if (el->l_tree_depth) {
-		ret = ocfs2_find_leaf(inode, el, name_hash, &eb_bh);
+		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, name_hash,
+				      &eb_bh);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
@@ -2931,7 +3152,7 @@ static int ocfs2_find_xe_in_bucket(struct inode *inode,
 		if (cmp)
 			continue;
 
-		ret = ocfs2_xattr_bucket_get_name_value(inode,
+		ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb,
 							xh,
 							i,
 							&block_off,
@@ -3175,7 +3396,7 @@ struct ocfs2_xattr_tree_list {
 	size_t result;
 };
 
-static int ocfs2_xattr_bucket_get_name_value(struct inode *inode,
+static int ocfs2_xattr_bucket_get_name_value(struct super_block *sb,
 					     struct ocfs2_xattr_header *xh,
 					     int index,
 					     int *block_off,
@@ -3188,8 +3409,8 @@ static int ocfs2_xattr_bucket_get_name_value(struct inode *inode,
 
 	name_offset = le16_to_cpu(xh->xh_entries[index].xe_name_offset);
 
-	*block_off = name_offset >> inode->i_sb->s_blocksize_bits;
-	*new_offset = name_offset % inode->i_sb->s_blocksize;
+	*block_off = name_offset >> sb->s_blocksize_bits;
+	*new_offset = name_offset % sb->s_blocksize;
 
 	return 0;
 }
@@ -3209,7 +3430,7 @@ static int ocfs2_list_xattr_bucket(struct inode *inode,
 		prefix = ocfs2_xattr_prefix(type);
 
 		if (prefix) {
-			ret = ocfs2_xattr_bucket_get_name_value(inode,
+			ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb,
 								bucket_xh(bucket),
 								i,
 								&block_off,
@@ -3232,22 +3453,19 @@ static int ocfs2_list_xattr_bucket(struct inode *inode,
 	return ret;
 }
 
-static int ocfs2_xattr_tree_list_index_block(struct inode *inode,
-					     struct ocfs2_xattr_tree_root *xt,
-					     char *buffer,
-					     size_t buffer_size)
+static int ocfs2_iterate_xattr_index_block(struct inode *inode,
+					   struct buffer_head *blk_bh,
+					   xattr_tree_rec_func *rec_func,
+					   void *para)
 {
-	struct ocfs2_extent_list *el = &xt->xt_list;
+	struct ocfs2_xattr_block *xb =
+			(struct ocfs2_xattr_block *)blk_bh->b_data;
+	struct ocfs2_extent_list *el = &xb->xb_attrs.xb_root.xt_list;
 	int ret = 0;
 	u32 name_hash = UINT_MAX, e_cpos = 0, num_clusters = 0;
 	u64 p_blkno = 0;
-	struct ocfs2_xattr_tree_list xl = {
-		.buffer = buffer,
-		.buffer_size = buffer_size,
-		.result = 0,
-	};
 
-	if (le16_to_cpu(el->l_next_free_rec) == 0)
+	if (!el->l_next_free_rec || !rec_func)
 		return 0;
 
 	while (name_hash > 0) {
@@ -3255,16 +3473,15 @@ static int ocfs2_xattr_tree_list_index_block(struct inode *inode,
 					  &e_cpos, &num_clusters, el);
 		if (ret) {
 			mlog_errno(ret);
-			goto out;
+			break;
 		}
 
-		ret = ocfs2_iterate_xattr_buckets(inode, p_blkno, num_clusters,
-						  ocfs2_list_xattr_bucket,
-						  &xl);
+		ret = rec_func(inode, blk_bh, p_blkno, e_cpos,
+			       num_clusters, para);
 		if (ret) {
 			if (ret != -ERANGE)
 				mlog_errno(ret);
-			goto out;
+			break;
 		}
 
 		if (e_cpos == 0)
@@ -3273,6 +3490,37 @@ static int ocfs2_xattr_tree_list_index_block(struct inode *inode,
 		name_hash = e_cpos - 1;
 	}
 
+	return ret;
+
+}
+
+static int ocfs2_list_xattr_tree_rec(struct inode *inode,
+				     struct buffer_head *root_bh,
+				     u64 blkno, u32 cpos, u32 len, void *para)
+{
+	return ocfs2_iterate_xattr_buckets(inode, blkno, len,
+					   ocfs2_list_xattr_bucket, para);
+}
+
+static int ocfs2_xattr_tree_list_index_block(struct inode *inode,
+					     struct buffer_head *blk_bh,
+					     char *buffer,
+					     size_t buffer_size)
+{
+	int ret;
+	struct ocfs2_xattr_tree_list xl = {
+		.buffer = buffer,
+		.buffer_size = buffer_size,
+		.result = 0,
+	};
+
+	ret = ocfs2_iterate_xattr_index_block(inode, blk_bh,
+					      ocfs2_list_xattr_tree_rec, &xl);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
 	ret = xl.result;
 out:
 	return ret;
@@ -3426,7 +3674,7 @@ static int ocfs2_xattr_create_index_block(struct inode *inode,
 	 */
 	down_write(&oi->ip_alloc_sem);
 
-	ret = ocfs2_journal_access_xb(handle, inode, xb_bh,
+	ret = ocfs2_journal_access_xb(handle, INODE_CACHE(inode), xb_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
@@ -4263,9 +4511,9 @@ static int ocfs2_add_new_xattr_cluster(struct inode *inode,
 	     (unsigned long long)OCFS2_I(inode)->ip_blkno,
 	     prev_cpos, (unsigned long long)bucket_blkno(first));
 
-	ocfs2_init_xattr_tree_extent_tree(&et, inode, root_bh);
+	ocfs2_init_xattr_tree_extent_tree(&et, INODE_CACHE(inode), root_bh);
 
-	ret = ocfs2_journal_access_xb(handle, inode, root_bh,
+	ret = ocfs2_journal_access_xb(handle, INODE_CACHE(inode), root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -4319,7 +4567,7 @@ static int ocfs2_add_new_xattr_cluster(struct inode *inode,
 
 	mlog(0, "Insert %u clusters at block %llu for xattr at %u\n",
 	     num_bits, (unsigned long long)block, v_start);
-	ret = ocfs2_insert_extent(osb, handle, inode, &et, v_start, block,
+	ret = ocfs2_insert_extent(handle, &et, v_start, block,
 				  num_bits, 0, ctxt->meta_ac);
 	if (ret < 0) {
 		mlog_errno(ret);
@@ -4798,10 +5046,13 @@ static int ocfs2_xattr_bucket_set_value_outside(struct inode *inode,
 	struct ocfs2_xattr_entry *xe = xs->here;
 	struct ocfs2_xattr_header *xh = bucket_xh(xs->bucket);
 	void *base;
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_access = ocfs2_journal_access,
+	};
 
 	BUG_ON(!xs->base || !xe || ocfs2_xattr_is_local(xe));
 
-	ret = ocfs2_xattr_bucket_get_name_value(inode, xh,
+	ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb, xh,
 						xe - xh->xh_entries,
 						&block_off,
 						&offset);
@@ -4814,8 +5065,10 @@ static int ocfs2_xattr_bucket_set_value_outside(struct inode *inode,
 	xv = (struct ocfs2_xattr_value_root *)(base + offset +
 		 OCFS2_XATTR_SIZE(xe->xe_name_len));
 
+	vb.vb_xv = xv;
+	vb.vb_bh = xs->bucket->bu_bhs[block_off];
 	ret = __ocfs2_xattr_set_value_outside(inode, handle,
-					      xv, val, value_len);
+					      &vb, val, value_len);
 	if (ret)
 		mlog_errno(ret);
 out:
@@ -4826,7 +5079,8 @@ static int ocfs2_rm_xattr_cluster(struct inode *inode,
 				  struct buffer_head *root_bh,
 				  u64 blkno,
 				  u32 cpos,
-				  u32 len)
+				  u32 len,
+				  void *para)
 {
 	int ret;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
@@ -4838,14 +5092,22 @@ static int ocfs2_rm_xattr_cluster(struct inode *inode,
 	struct ocfs2_cached_dealloc_ctxt dealloc;
 	struct ocfs2_extent_tree et;
 
-	ocfs2_init_xattr_tree_extent_tree(&et, inode, root_bh);
+	ret = ocfs2_iterate_xattr_buckets(inode, blkno, len,
+					  ocfs2_delete_xattr_in_bucket, para);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	ocfs2_init_xattr_tree_extent_tree(&et, INODE_CACHE(inode), root_bh);
 
 	ocfs2_init_dealloc_ctxt(&dealloc);
 
 	mlog(0, "rm xattr extent rec at %u len = %u, start from %llu\n",
 	     cpos, len, (unsigned long long)blkno);
 
-	ocfs2_remove_xattr_clusters_from_cache(inode, blkno, len);
+	ocfs2_remove_xattr_clusters_from_cache(INODE_CACHE(inode), blkno,
+					       len);
 
 	ret = ocfs2_lock_allocators(inode, &et, 0, 1, NULL, &meta_ac);
 	if (ret) {
@@ -4870,14 +5132,14 @@ static int ocfs2_rm_xattr_cluster(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_journal_access_xb(handle, inode, root_bh,
+	ret = ocfs2_journal_access_xb(handle, INODE_CACHE(inode), root_bh,
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (ret) {
 		mlog_errno(ret);
 		goto out_commit;
 	}
 
-	ret = ocfs2_remove_extent(inode, &et, cpos, len, handle, meta_ac,
+	ret = ocfs2_remove_extent(handle, &et, cpos, len, meta_ac,
 				  &dealloc);
 	if (ret) {
 		mlog_errno(ret);
@@ -5220,7 +5482,7 @@ static int ocfs2_delete_xattr_in_bucket(struct inode *inode,
 					struct ocfs2_xattr_bucket *bucket,
 					void *para)
 {
-	int ret = 0;
+	int ret = 0, ref_credits;
 	struct ocfs2_xattr_header *xh = bucket_xh(bucket);
 	u16 i;
 	struct ocfs2_xattr_entry *xe;
@@ -5228,7 +5490,9 @@ static int ocfs2_delete_xattr_in_bucket(struct inode *inode,
 	struct ocfs2_xattr_set_ctxt ctxt = {NULL, NULL,};
 	int credits = ocfs2_remove_extent_credits(osb->sb) +
 		ocfs2_blocks_per_xattr_bucket(inode->i_sb);
-
+	struct ocfs2_xattr_value_root *xv;
+	struct ocfs2_rm_xattr_bucket_para *args =
+			(struct ocfs2_rm_xattr_bucket_para *)para;
 
 	ocfs2_init_dealloc_ctxt(&ctxt.dealloc);
 
@@ -5237,7 +5501,16 @@ static int ocfs2_delete_xattr_in_bucket(struct inode *inode,
 		if (ocfs2_xattr_is_local(xe))
 			continue;
 
-		ctxt.handle = ocfs2_start_trans(osb, credits);
+		ret = ocfs2_get_xattr_tree_value_root(inode->i_sb, bucket,
+						      i, &xv, NULL);
+
+		ret = ocfs2_lock_xattr_remove_allocators(inode, xv,
+							 args->ref_ci,
+							 args->ref_root_bh,
+							 &ctxt.meta_ac,
+							 &ref_credits);
+
+		ctxt.handle = ocfs2_start_trans(osb, credits + ref_credits);
 		if (IS_ERR(ctxt.handle)) {
 			ret = PTR_ERR(ctxt.handle);
 			mlog_errno(ret);
@@ -5248,57 +5521,1439 @@ static int ocfs2_delete_xattr_in_bucket(struct inode *inode,
 							i, 0, &ctxt);
 
 		ocfs2_commit_trans(osb, ctxt.handle);
+		if (ctxt.meta_ac) {
+			ocfs2_free_alloc_context(ctxt.meta_ac);
+			ctxt.meta_ac = NULL;
+		}
 		if (ret) {
 			mlog_errno(ret);
 			break;
 		}
 	}
 
+	if (ctxt.meta_ac)
+		ocfs2_free_alloc_context(ctxt.meta_ac);
 	ocfs2_schedule_truncate_log_flush(osb, 1);
 	ocfs2_run_deallocs(osb, &ctxt.dealloc);
 	return ret;
 }
 
-static int ocfs2_delete_xattr_index_block(struct inode *inode,
-					  struct buffer_head *xb_bh)
+/*
+ * Whenever we modify a xattr value root in the bucket(e.g, CoW
+ * or change the extent record flag), we need to recalculate
+ * the metaecc for the whole bucket. So it is done here.
+ *
+ * Note:
+ * We have to give the extra credits for the caller.
+ */
+static int ocfs2_xattr_bucket_post_refcount(struct inode *inode,
+					    handle_t *handle,
+					    void *para)
+{
+	int ret;
+	struct ocfs2_xattr_bucket *bucket =
+			(struct ocfs2_xattr_bucket *)para;
+
+	ret = ocfs2_xattr_bucket_journal_access(handle, bucket,
+						OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	ocfs2_xattr_bucket_journal_dirty(handle, bucket);
+
+	return 0;
+}
+
+/*
+ * Special action we need if the xattr value is refcounted.
+ *
+ * 1. If the xattr is refcounted, lock the tree.
+ * 2. CoW the xattr if we are setting the new value and the value
+ *    will be stored outside.
+ * 3. In other case, decrease_refcount will work for us, so just
+ *    lock the refcount tree, calculate the meta and credits is OK.
+ *
+ * We have to do CoW before ocfs2_init_xattr_set_ctxt since
+ * currently CoW is a completed transaction, while this function
+ * will also lock the allocators and let us deadlock. So we will
+ * CoW the whole xattr value.
+ */
+static int ocfs2_prepare_refcount_xattr(struct inode *inode,
+					struct ocfs2_dinode *di,
+					struct ocfs2_xattr_info *xi,
+					struct ocfs2_xattr_search *xis,
+					struct ocfs2_xattr_search *xbs,
+					struct ocfs2_refcount_tree **ref_tree,
+					int *meta_add,
+					int *credits)
 {
-	struct ocfs2_xattr_block *xb =
-			(struct ocfs2_xattr_block *)xb_bh->b_data;
-	struct ocfs2_extent_list *el = &xb->xb_attrs.xb_root.xt_list;
 	int ret = 0;
-	u32 name_hash = UINT_MAX, e_cpos, num_clusters;
-	u64 p_blkno;
+	struct ocfs2_xattr_block *xb;
+	struct ocfs2_xattr_entry *xe;
+	char *base;
+	u32 p_cluster, num_clusters;
+	unsigned int ext_flags;
+	int name_offset, name_len;
+	struct ocfs2_xattr_value_buf vb;
+	struct ocfs2_xattr_bucket *bucket = NULL;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_post_refcount refcount;
+	struct ocfs2_post_refcount *p = NULL;
+	struct buffer_head *ref_root_bh = NULL;
 
-	if (le16_to_cpu(el->l_next_free_rec) == 0)
-		return 0;
+	if (!xis->not_found) {
+		xe = xis->here;
+		name_offset = le16_to_cpu(xe->xe_name_offset);
+		name_len = OCFS2_XATTR_SIZE(xe->xe_name_len);
+		base = xis->base;
+		vb.vb_bh = xis->inode_bh;
+		vb.vb_access = ocfs2_journal_access_di;
+	} else {
+		int i, block_off = 0;
+		xb = (struct ocfs2_xattr_block *)xbs->xattr_bh->b_data;
+		xe = xbs->here;
+		name_offset = le16_to_cpu(xe->xe_name_offset);
+		name_len = OCFS2_XATTR_SIZE(xe->xe_name_len);
+		i = xbs->here - xbs->header->xh_entries;
 
-	while (name_hash > 0) {
-		ret = ocfs2_xattr_get_rec(inode, name_hash, &p_blkno,
-					  &e_cpos, &num_clusters, el);
+		if (le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED) {
+			ret = ocfs2_xattr_bucket_get_name_value(inode->i_sb,
+							bucket_xh(xbs->bucket),
+							i, &block_off,
+							&name_offset);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+			base = bucket_block(xbs->bucket, block_off);
+			vb.vb_bh = xbs->bucket->bu_bhs[block_off];
+			vb.vb_access = ocfs2_journal_access;
+
+			if (ocfs2_meta_ecc(osb)) {
+				/*create parameters for ocfs2_post_refcount. */
+				bucket = xbs->bucket;
+				refcount.credits = bucket->bu_blocks;
+				refcount.para = bucket;
+				refcount.func =
+					ocfs2_xattr_bucket_post_refcount;
+				p = &refcount;
+			}
+		} else {
+			base = xbs->base;
+			vb.vb_bh = xbs->xattr_bh;
+			vb.vb_access = ocfs2_journal_access_xb;
+		}
+	}
+
+	if (ocfs2_xattr_is_local(xe))
+		goto out;
+
+	vb.vb_xv = (struct ocfs2_xattr_value_root *)
+				(base + name_offset + name_len);
+
+	ret = ocfs2_xattr_get_clusters(inode, 0, &p_cluster,
+				       &num_clusters, &vb.vb_xv->xr_list,
+				       &ext_flags);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * We just need to check the 1st extent record, since we always
+	 * CoW the whole xattr. So there shouldn't be a xattr with
+	 * some REFCOUNT extent recs after the 1st one.
+	 */
+	if (!(ext_flags & OCFS2_EXT_REFCOUNTED))
+		goto out;
+
+	ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
+				       1, ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * If we are deleting the xattr or the new size will be stored inside,
+	 * cool, leave it there, the xattr truncate process will remove them
+	 * for us(it still needs the refcount tree lock and the meta, credits).
+	 * And the worse case is that every cluster truncate will split the
+	 * refcount tree, and make the original extent become 3. So we will need
+	 * 2 * cluster more extent recs at most.
+	 */
+	if (!xi->value || xi->value_len <= OCFS2_XATTR_INLINE_SIZE) {
+
+		ret = ocfs2_refcounted_xattr_delete_need(inode,
+							 &(*ref_tree)->rf_ci,
+							 ref_root_bh, vb.vb_xv,
+							 meta_add, credits);
+		if (ret)
+			mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_refcount_cow_xattr(inode, di, &vb,
+				       *ref_tree, ref_root_bh, 0,
+				       le32_to_cpu(vb.vb_xv->xr_clusters), p);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	brelse(ref_root_bh);
+	return ret;
+}
+
+/*
+ * Add the REFCOUNTED flags for all the extent rec in ocfs2_xattr_value_root.
+ * The physical clusters will be added to refcount tree.
+ */
+static int ocfs2_xattr_value_attach_refcount(struct inode *inode,
+				struct ocfs2_xattr_value_root *xv,
+				struct ocfs2_extent_tree *value_et,
+				struct ocfs2_caching_info *ref_ci,
+				struct buffer_head *ref_root_bh,
+				struct ocfs2_cached_dealloc_ctxt *dealloc,
+				struct ocfs2_post_refcount *refcount)
+{
+	int ret = 0;
+	u32 clusters = le32_to_cpu(xv->xr_clusters);
+	u32 cpos, p_cluster, num_clusters;
+	struct ocfs2_extent_list *el = &xv->xr_list;
+	unsigned int ext_flags;
+
+	cpos = 0;
+	while (cpos < clusters) {
+		ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
+					       &num_clusters, el, &ext_flags);
+
+		cpos += num_clusters;
+		if ((ext_flags & OCFS2_EXT_REFCOUNTED))
+			continue;
+
+		BUG_ON(!p_cluster);
+
+		ret = ocfs2_add_refcount_flag(inode, value_et,
+					      ref_ci, ref_root_bh,
+					      cpos - num_clusters,
+					      p_cluster, num_clusters,
+					      dealloc, refcount);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+	}
+
+	return ret;
+}
+
+/*
+ * Given a normal ocfs2_xattr_header, refcount all the entries which
+ * have value stored outside.
+ * Used for xattrs stored in inode and ocfs2_xattr_block.
+ */
+static int ocfs2_xattr_attach_refcount_normal(struct inode *inode,
+				struct ocfs2_xattr_value_buf *vb,
+				struct ocfs2_xattr_header *header,
+				struct ocfs2_caching_info *ref_ci,
+				struct buffer_head *ref_root_bh,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+
+	struct ocfs2_xattr_entry *xe;
+	struct ocfs2_xattr_value_root *xv;
+	struct ocfs2_extent_tree et;
+	int i, ret = 0;
+
+	for (i = 0; i < le16_to_cpu(header->xh_count); i++) {
+		xe = &header->xh_entries[i];
+
+		if (ocfs2_xattr_is_local(xe))
+			continue;
+
+		xv = (struct ocfs2_xattr_value_root *)((void *)header +
+			le16_to_cpu(xe->xe_name_offset) +
+			OCFS2_XATTR_SIZE(xe->xe_name_len));
+
+		vb->vb_xv = xv;
+		ocfs2_init_xattr_value_extent_tree(&et, INODE_CACHE(inode), vb);
+
+		ret = ocfs2_xattr_value_attach_refcount(inode, xv, &et,
+							ref_ci, ref_root_bh,
+							dealloc, NULL);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static int ocfs2_xattr_inline_attach_refcount(struct inode *inode,
+				struct buffer_head *fe_bh,
+				struct ocfs2_caching_info *ref_ci,
+				struct buffer_head *ref_root_bh,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)fe_bh->b_data;
+	struct ocfs2_xattr_header *header = (struct ocfs2_xattr_header *)
+				(fe_bh->b_data + inode->i_sb->s_blocksize -
+				le16_to_cpu(di->i_xattr_inline_size));
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_bh = fe_bh,
+		.vb_access = ocfs2_journal_access_di,
+	};
+
+	return ocfs2_xattr_attach_refcount_normal(inode, &vb, header,
+						  ref_ci, ref_root_bh, dealloc);
+}
+
+struct ocfs2_xattr_tree_value_refcount_para {
+	struct ocfs2_caching_info *ref_ci;
+	struct buffer_head *ref_root_bh;
+	struct ocfs2_cached_dealloc_ctxt *dealloc;
+};
+
+static int ocfs2_get_xattr_tree_value_root(struct super_block *sb,
+					   struct ocfs2_xattr_bucket *bucket,
+					   int offset,
+					   struct ocfs2_xattr_value_root **xv,
+					   struct buffer_head **bh)
+{
+	int ret, block_off, name_offset;
+	struct ocfs2_xattr_header *xh = bucket_xh(bucket);
+	struct ocfs2_xattr_entry *xe = &xh->xh_entries[offset];
+	void *base;
+
+	ret = ocfs2_xattr_bucket_get_name_value(sb,
+						bucket_xh(bucket),
+						offset,
+						&block_off,
+						&name_offset);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	base = bucket_block(bucket, block_off);
+
+	*xv = (struct ocfs2_xattr_value_root *)(base + name_offset +
+			 OCFS2_XATTR_SIZE(xe->xe_name_len));
+
+	if (bh)
+		*bh = bucket->bu_bhs[block_off];
+out:
+	return ret;
+}
+
+/*
+ * For a given xattr bucket, refcount all the entries which
+ * have value stored outside.
+ */
+static int ocfs2_xattr_bucket_value_refcount(struct inode *inode,
+					     struct ocfs2_xattr_bucket *bucket,
+					     void *para)
+{
+	int i, ret = 0;
+	struct ocfs2_extent_tree et;
+	struct ocfs2_xattr_tree_value_refcount_para *ref =
+			(struct ocfs2_xattr_tree_value_refcount_para *)para;
+	struct ocfs2_xattr_header *xh =
+			(struct ocfs2_xattr_header *)bucket->bu_bhs[0]->b_data;
+	struct ocfs2_xattr_entry *xe;
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_access = ocfs2_journal_access,
+	};
+	struct ocfs2_post_refcount refcount = {
+		.credits = bucket->bu_blocks,
+		.para = bucket,
+		.func = ocfs2_xattr_bucket_post_refcount,
+	};
+	struct ocfs2_post_refcount *p = NULL;
+
+	/* We only need post_refcount if we support metaecc. */
+	if (ocfs2_meta_ecc(OCFS2_SB(inode->i_sb)))
+		p = &refcount;
+
+	mlog(0, "refcount bucket %llu, count = %u\n",
+	     (unsigned long long)bucket_blkno(bucket),
+	     le16_to_cpu(xh->xh_count));
+	for (i = 0; i < le16_to_cpu(xh->xh_count); i++) {
+		xe = &xh->xh_entries[i];
+
+		if (ocfs2_xattr_is_local(xe))
+			continue;
+
+		ret = ocfs2_get_xattr_tree_value_root(inode->i_sb, bucket, i,
+						      &vb.vb_xv, &vb.vb_bh);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		ocfs2_init_xattr_value_extent_tree(&et,
+						   INODE_CACHE(inode), &vb);
+
+		ret = ocfs2_xattr_value_attach_refcount(inode, vb.vb_xv,
+							&et, ref->ref_ci,
+							ref->ref_root_bh,
+							ref->dealloc, p);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+	}
+
+	return ret;
+
+}
+
+static int ocfs2_refcount_xattr_tree_rec(struct inode *inode,
+				     struct buffer_head *root_bh,
+				     u64 blkno, u32 cpos, u32 len, void *para)
+{
+	return ocfs2_iterate_xattr_buckets(inode, blkno, len,
+					   ocfs2_xattr_bucket_value_refcount,
+					   para);
+}
+
+static int ocfs2_xattr_block_attach_refcount(struct inode *inode,
+				struct buffer_head *blk_bh,
+				struct ocfs2_caching_info *ref_ci,
+				struct buffer_head *ref_root_bh,
+				struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret = 0;
+	struct ocfs2_xattr_block *xb =
+				(struct ocfs2_xattr_block *)blk_bh->b_data;
+
+	if (!(le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED)) {
+		struct ocfs2_xattr_header *header = &xb->xb_attrs.xb_header;
+		struct ocfs2_xattr_value_buf vb = {
+			.vb_bh = blk_bh,
+			.vb_access = ocfs2_journal_access_xb,
+		};
+
+		ret = ocfs2_xattr_attach_refcount_normal(inode, &vb, header,
+							 ref_ci, ref_root_bh,
+							 dealloc);
+	} else {
+		struct ocfs2_xattr_tree_value_refcount_para para = {
+			.ref_ci = ref_ci,
+			.ref_root_bh = ref_root_bh,
+			.dealloc = dealloc,
+		};
+
+		ret = ocfs2_iterate_xattr_index_block(inode, blk_bh,
+						ocfs2_refcount_xattr_tree_rec,
+						&para);
+	}
+
+	return ret;
+}
+
+int ocfs2_xattr_attach_refcount_tree(struct inode *inode,
+				     struct buffer_head *fe_bh,
+				     struct ocfs2_caching_info *ref_ci,
+				     struct buffer_head *ref_root_bh,
+				     struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	int ret = 0;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)fe_bh->b_data;
+	struct buffer_head *blk_bh = NULL;
+
+	if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL) {
+		ret = ocfs2_xattr_inline_attach_refcount(inode, fe_bh,
+							 ref_ci, ref_root_bh,
+							 dealloc);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
+	}
+
+	if (!di->i_xattr_loc)
+		goto out;
+
+	ret = ocfs2_read_xattr_block(inode, le64_to_cpu(di->i_xattr_loc),
+				     &blk_bh);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_xattr_block_attach_refcount(inode, blk_bh, ref_ci,
+						ref_root_bh, dealloc);
+	if (ret)
+		mlog_errno(ret);
+
+	brelse(blk_bh);
+out:
+
+	return ret;
+}
+
+typedef int (should_xattr_reflinked)(struct ocfs2_xattr_entry *xe);
+/*
+ * Store the information we need in xattr reflink.
+ * old_bh and new_bh are inode bh for the old and new inode.
+ */
+struct ocfs2_xattr_reflink {
+	struct inode *old_inode;
+	struct inode *new_inode;
+	struct buffer_head *old_bh;
+	struct buffer_head *new_bh;
+	struct ocfs2_caching_info *ref_ci;
+	struct buffer_head *ref_root_bh;
+	struct ocfs2_cached_dealloc_ctxt *dealloc;
+	should_xattr_reflinked *xattr_reflinked;
+};
+
+/*
+ * Given a xattr header and xe offset,
+ * return the proper xv and the corresponding bh.
+ * xattr in inode, block and xattr tree have different implementaions.
+ */
+typedef int (get_xattr_value_root)(struct super_block *sb,
+				   struct buffer_head *bh,
+				   struct ocfs2_xattr_header *xh,
+				   int offset,
+				   struct ocfs2_xattr_value_root **xv,
+				   struct buffer_head **ret_bh,
+				   void *para);
+
+/*
+ * Calculate all the xattr value root metadata stored in this xattr header and
+ * credits we need if we create them from the scratch.
+ * We use get_xattr_value_root so that all types of xattr container can use it.
+ */
+static int ocfs2_value_metas_in_xattr_header(struct super_block *sb,
+					     struct buffer_head *bh,
+					     struct ocfs2_xattr_header *xh,
+					     int *metas, int *credits,
+					     int *num_recs,
+					     get_xattr_value_root *func,
+					     void *para)
+{
+	int i, ret = 0;
+	struct ocfs2_xattr_value_root *xv;
+	struct ocfs2_xattr_entry *xe;
+
+	for (i = 0; i < le16_to_cpu(xh->xh_count); i++) {
+		xe = &xh->xh_entries[i];
+		if (ocfs2_xattr_is_local(xe))
+			continue;
+
+		ret = func(sb, bh, xh, i, &xv, NULL, para);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		*metas += le16_to_cpu(xv->xr_list.l_tree_depth) *
+			  le16_to_cpu(xv->xr_list.l_next_free_rec);
+
+		*credits += ocfs2_calc_extend_credits(sb,
+						&def_xv.xv.xr_list,
+						le32_to_cpu(xv->xr_clusters));
+
+		/*
+		 * If the value is a tree with depth > 1, We don't go deep
+		 * to the extent block, so just calculate a maximum record num.
+		 */
+		if (!xv->xr_list.l_tree_depth)
+			*num_recs += xv->xr_list.l_next_free_rec;
+		else
+			*num_recs += ocfs2_clusters_for_bytes(sb,
+							      XATTR_SIZE_MAX);
+	}
+
+	return ret;
+}
+
+/* Used by xattr inode and block to return the right xv and buffer_head. */
+static int ocfs2_get_xattr_value_root(struct super_block *sb,
+				      struct buffer_head *bh,
+				      struct ocfs2_xattr_header *xh,
+				      int offset,
+				      struct ocfs2_xattr_value_root **xv,
+				      struct buffer_head **ret_bh,
+				      void *para)
+{
+	struct ocfs2_xattr_entry *xe = &xh->xh_entries[offset];
+
+	*xv = (struct ocfs2_xattr_value_root *)((void *)xh +
+		le16_to_cpu(xe->xe_name_offset) +
+		OCFS2_XATTR_SIZE(xe->xe_name_len));
+
+	if (ret_bh)
+		*ret_bh = bh;
+
+	return 0;
+}
+
+/*
+ * Lock the meta_ac and caculate how much credits we need for reflink xattrs.
+ * It is only used for inline xattr and xattr block.
+ */
+static int ocfs2_reflink_lock_xattr_allocators(struct ocfs2_super *osb,
+					struct ocfs2_xattr_header *xh,
+					struct buffer_head *ref_root_bh,
+					int *credits,
+					struct ocfs2_alloc_context **meta_ac)
+{
+	int ret, meta_add = 0, num_recs = 0;
+	struct ocfs2_refcount_block *rb =
+			(struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+	*credits = 0;
+
+	ret = ocfs2_value_metas_in_xattr_header(osb->sb, NULL, xh,
+						&meta_add, credits, &num_recs,
+						ocfs2_get_xattr_value_root,
+						NULL);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * We need to add/modify num_recs in refcount tree, so just calculate
+	 * an approximate number we need for refcount tree change.
+	 * Sometimes we need to split the tree, and after split,  half recs
+	 * will be moved to the new block, and a new block can only provide
+	 * half number of recs. So we multiple new blocks by 2.
+	 */
+	num_recs = num_recs / ocfs2_refcount_recs_per_rb(osb->sb) * 2;
+	meta_add += num_recs;
+	*credits += num_recs + num_recs * OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)
+		*credits += le16_to_cpu(rb->rf_list.l_tree_depth) *
+			    le16_to_cpu(rb->rf_list.l_next_free_rec) + 1;
+	else
+		*credits += 1;
+
+	ret = ocfs2_reserve_new_metadata_blocks(osb, meta_add, meta_ac);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	return ret;
+}
 
-		ret = ocfs2_iterate_xattr_buckets(inode, p_blkno, num_clusters,
-						  ocfs2_delete_xattr_in_bucket,
-						  NULL);
+/*
+ * Given a xattr header, reflink all the xattrs in this container.
+ * It can be used for inode, block and bucket.
+ *
+ * NOTE:
+ * Before we call this function, the caller has memcpy the xattr in
+ * old_xh to the new_xh.
+ *
+ * If args.xattr_reflinked is set, call it to decide whether the xe should
+ * be reflinked or not. If not, remove it from the new xattr header.
+ */
+static int ocfs2_reflink_xattr_header(handle_t *handle,
+				      struct ocfs2_xattr_reflink *args,
+				      struct buffer_head *old_bh,
+				      struct ocfs2_xattr_header *xh,
+				      struct buffer_head *new_bh,
+				      struct ocfs2_xattr_header *new_xh,
+				      struct ocfs2_xattr_value_buf *vb,
+				      struct ocfs2_alloc_context *meta_ac,
+				      get_xattr_value_root *func,
+				      void *para)
+{
+	int ret = 0, i, j;
+	struct super_block *sb = args->old_inode->i_sb;
+	struct buffer_head *value_bh;
+	struct ocfs2_xattr_entry *xe, *last;
+	struct ocfs2_xattr_value_root *xv, *new_xv;
+	struct ocfs2_extent_tree data_et;
+	u32 clusters, cpos, p_cluster, num_clusters;
+	unsigned int ext_flags = 0;
+
+	mlog(0, "reflink xattr in container %llu, count = %u\n",
+	     (unsigned long long)old_bh->b_blocknr, le16_to_cpu(xh->xh_count));
+
+	last = &new_xh->xh_entries[le16_to_cpu(new_xh->xh_count)];
+	for (i = 0, j = 0; i < le16_to_cpu(xh->xh_count); i++, j++) {
+		xe = &xh->xh_entries[i];
+
+		if (args->xattr_reflinked && !args->xattr_reflinked(xe)) {
+			xe = &new_xh->xh_entries[j];
+
+			le16_add_cpu(&new_xh->xh_count, -1);
+			if (new_xh->xh_count) {
+				memmove(xe, xe + 1,
+					(void *)last - (void *)xe);
+				memset(last, 0,
+				       sizeof(struct ocfs2_xattr_entry));
+			}
+
+			/*
+			 * We don't want j to increase in the next round since
+			 * it is already moved ahead.
+			 */
+			j--;
+			continue;
+		}
+
+		if (ocfs2_xattr_is_local(xe))
+			continue;
+
+		ret = func(sb, old_bh, xh, i, &xv, NULL, para);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		ret = func(sb, new_bh, new_xh, j, &new_xv, &value_bh, para);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		/*
+		 * For the xattr which has l_tree_depth = 0, all the extent
+		 * recs have already be copied to the new xh with the
+		 * propriate OCFS2_EXT_REFCOUNTED flag we just need to
+		 * increase the refount count int the refcount tree.
+		 *
+		 * For the xattr which has l_tree_depth > 0, we need
+		 * to initialize it to the empty default value root,
+		 * and then insert the extents one by one.
+		 */
+		if (xv->xr_list.l_tree_depth) {
+			memcpy(new_xv, &def_xv, sizeof(def_xv));
+			vb->vb_xv = new_xv;
+			vb->vb_bh = value_bh;
+			ocfs2_init_xattr_value_extent_tree(&data_et,
+					INODE_CACHE(args->new_inode), vb);
+		}
+
+		clusters = le32_to_cpu(xv->xr_clusters);
+		cpos = 0;
+		while (cpos < clusters) {
+			ret = ocfs2_xattr_get_clusters(args->old_inode,
+						       cpos,
+						       &p_cluster,
+						       &num_clusters,
+						       &xv->xr_list,
+						       &ext_flags);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+
+			BUG_ON(!p_cluster);
+
+			if (xv->xr_list.l_tree_depth) {
+				ret = ocfs2_insert_extent(handle,
+						&data_et, cpos,
+						ocfs2_clusters_to_blocks(
+							args->old_inode->i_sb,
+							p_cluster),
+						num_clusters, ext_flags,
+						meta_ac);
+				if (ret) {
+					mlog_errno(ret);
+					goto out;
+				}
+			}
+
+			ret = ocfs2_increase_refcount(handle, args->ref_ci,
+						      args->ref_root_bh,
+						      p_cluster, num_clusters,
+						      meta_ac, args->dealloc);
+			if (ret) {
+				mlog_errno(ret);
+				goto out;
+			}
+
+			cpos += num_clusters;
+		}
+	}
+
+out:
+	return ret;
+}
+
+static int ocfs2_reflink_xattr_inline(struct ocfs2_xattr_reflink *args)
+{
+	int ret = 0, credits = 0;
+	handle_t *handle;
+	struct ocfs2_super *osb = OCFS2_SB(args->old_inode->i_sb);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)args->old_bh->b_data;
+	int inline_size = le16_to_cpu(di->i_xattr_inline_size);
+	int header_off = osb->sb->s_blocksize - inline_size;
+	struct ocfs2_xattr_header *xh = (struct ocfs2_xattr_header *)
+					(args->old_bh->b_data + header_off);
+	struct ocfs2_xattr_header *new_xh = (struct ocfs2_xattr_header *)
+					(args->new_bh->b_data + header_off);
+	struct ocfs2_alloc_context *meta_ac = NULL;
+	struct ocfs2_inode_info *new_oi;
+	struct ocfs2_dinode *new_di;
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_bh = args->new_bh,
+		.vb_access = ocfs2_journal_access_di,
+	};
+
+	ret = ocfs2_reflink_lock_xattr_allocators(osb, xh, args->ref_root_bh,
+						  &credits, &meta_ac);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_journal_access_di(handle, INODE_CACHE(args->new_inode),
+				      args->new_bh, OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	memcpy(args->new_bh->b_data + header_off,
+	       args->old_bh->b_data + header_off, inline_size);
+
+	new_di = (struct ocfs2_dinode *)args->new_bh->b_data;
+	new_di->i_xattr_inline_size = cpu_to_le16(inline_size);
+
+	ret = ocfs2_reflink_xattr_header(handle, args, args->old_bh, xh,
+					 args->new_bh, new_xh, &vb, meta_ac,
+					 ocfs2_get_xattr_value_root, NULL);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	new_oi = OCFS2_I(args->new_inode);
+	spin_lock(&new_oi->ip_lock);
+	new_oi->ip_dyn_features |= OCFS2_HAS_XATTR_FL | OCFS2_INLINE_XATTR_FL;
+	new_di->i_dyn_features = cpu_to_le16(new_oi->ip_dyn_features);
+	spin_unlock(&new_oi->ip_lock);
+
+	ocfs2_journal_dirty(handle, args->new_bh);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+
+out:
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+	return ret;
+}
+
+static int ocfs2_create_empty_xattr_block(struct inode *inode,
+					  struct buffer_head *fe_bh,
+					  struct buffer_head **ret_bh,
+					  int indexed)
+{
+	int ret;
+	handle_t *handle;
+	struct ocfs2_alloc_context *meta_ac;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+
+	ret = ocfs2_reserve_new_metadata_blocks(osb, 1, &meta_ac);
+	if (ret < 0) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	handle = ocfs2_start_trans(osb, OCFS2_XATTR_BLOCK_CREATE_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	mlog(0, "create new xattr block for inode %llu, index = %d\n",
+	     (unsigned long long)fe_bh->b_blocknr, indexed);
+	ret = ocfs2_create_xattr_block(handle, inode, fe_bh,
+				       meta_ac, ret_bh, indexed);
+	if (ret)
+		mlog_errno(ret);
+
+	ocfs2_commit_trans(osb, handle);
+out:
+	ocfs2_free_alloc_context(meta_ac);
+	return ret;
+}
+
+static int ocfs2_reflink_xattr_block(struct ocfs2_xattr_reflink *args,
+				     struct buffer_head *blk_bh,
+				     struct buffer_head *new_blk_bh)
+{
+	int ret = 0, credits = 0;
+	handle_t *handle;
+	struct ocfs2_inode_info *new_oi = OCFS2_I(args->new_inode);
+	struct ocfs2_dinode *new_di;
+	struct ocfs2_super *osb = OCFS2_SB(args->new_inode->i_sb);
+	int header_off = offsetof(struct ocfs2_xattr_block, xb_attrs.xb_header);
+	struct ocfs2_xattr_block *xb =
+			(struct ocfs2_xattr_block *)blk_bh->b_data;
+	struct ocfs2_xattr_header *xh = &xb->xb_attrs.xb_header;
+	struct ocfs2_xattr_block *new_xb =
+			(struct ocfs2_xattr_block *)new_blk_bh->b_data;
+	struct ocfs2_xattr_header *new_xh = &new_xb->xb_attrs.xb_header;
+	struct ocfs2_alloc_context *meta_ac;
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_bh = new_blk_bh,
+		.vb_access = ocfs2_journal_access_xb,
+	};
+
+	ret = ocfs2_reflink_lock_xattr_allocators(osb, xh, args->ref_root_bh,
+						  &credits, &meta_ac);
+	if (ret) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	/* One more credits in case we need to add xattr flags in new inode. */
+	handle = ocfs2_start_trans(osb, credits + 1);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (!(new_oi->ip_dyn_features & OCFS2_HAS_XATTR_FL)) {
+		ret = ocfs2_journal_access_di(handle,
+					      INODE_CACHE(args->new_inode),
+					      args->new_bh,
+					      OCFS2_JOURNAL_ACCESS_WRITE);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_commit;
+		}
+	}
+
+	ret = ocfs2_journal_access_xb(handle, INODE_CACHE(args->new_inode),
+				      new_blk_bh, OCFS2_JOURNAL_ACCESS_WRITE);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	memcpy(new_blk_bh->b_data + header_off, blk_bh->b_data + header_off,
+	       osb->sb->s_blocksize - header_off);
+
+	ret = ocfs2_reflink_xattr_header(handle, args, blk_bh, xh,
+					 new_blk_bh, new_xh, &vb, meta_ac,
+					 ocfs2_get_xattr_value_root, NULL);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ocfs2_journal_dirty(handle, new_blk_bh);
+
+	if (!(new_oi->ip_dyn_features & OCFS2_HAS_XATTR_FL)) {
+		new_di = (struct ocfs2_dinode *)args->new_bh->b_data;
+		spin_lock(&new_oi->ip_lock);
+		new_oi->ip_dyn_features |= OCFS2_HAS_XATTR_FL;
+		new_di->i_dyn_features = cpu_to_le16(new_oi->ip_dyn_features);
+		spin_unlock(&new_oi->ip_lock);
+
+		ocfs2_journal_dirty(handle, args->new_bh);
+	}
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+
+out:
+	ocfs2_free_alloc_context(meta_ac);
+	return ret;
+}
+
+struct ocfs2_reflink_xattr_tree_args {
+	struct ocfs2_xattr_reflink *reflink;
+	struct buffer_head *old_blk_bh;
+	struct buffer_head *new_blk_bh;
+	struct ocfs2_xattr_bucket *old_bucket;
+	struct ocfs2_xattr_bucket *new_bucket;
+};
+
+/*
+ * NOTE:
+ * We have to handle the case that both old bucket and new bucket
+ * will call this function to get the right ret_bh.
+ * So The caller must give us the right bh.
+ */
+static int ocfs2_get_reflink_xattr_value_root(struct super_block *sb,
+					struct buffer_head *bh,
+					struct ocfs2_xattr_header *xh,
+					int offset,
+					struct ocfs2_xattr_value_root **xv,
+					struct buffer_head **ret_bh,
+					void *para)
+{
+	struct ocfs2_reflink_xattr_tree_args *args =
+			(struct ocfs2_reflink_xattr_tree_args *)para;
+	struct ocfs2_xattr_bucket *bucket;
+
+	if (bh == args->old_bucket->bu_bhs[0])
+		bucket = args->old_bucket;
+	else
+		bucket = args->new_bucket;
+
+	return ocfs2_get_xattr_tree_value_root(sb, bucket, offset,
+					       xv, ret_bh);
+}
+
+struct ocfs2_value_tree_metas {
+	int num_metas;
+	int credits;
+	int num_recs;
+};
+
+static int ocfs2_value_tree_metas_in_bucket(struct super_block *sb,
+					struct buffer_head *bh,
+					struct ocfs2_xattr_header *xh,
+					int offset,
+					struct ocfs2_xattr_value_root **xv,
+					struct buffer_head **ret_bh,
+					void *para)
+{
+	struct ocfs2_xattr_bucket *bucket =
+				(struct ocfs2_xattr_bucket *)para;
+
+	return ocfs2_get_xattr_tree_value_root(sb, bucket, offset,
+					       xv, ret_bh);
+}
+
+static int ocfs2_calc_value_tree_metas(struct inode *inode,
+				      struct ocfs2_xattr_bucket *bucket,
+				      void *para)
+{
+	struct ocfs2_value_tree_metas *metas =
+			(struct ocfs2_value_tree_metas *)para;
+	struct ocfs2_xattr_header *xh =
+			(struct ocfs2_xattr_header *)bucket->bu_bhs[0]->b_data;
+
+	/* Add the credits for this bucket first. */
+	metas->credits += bucket->bu_blocks;
+	return ocfs2_value_metas_in_xattr_header(inode->i_sb, bucket->bu_bhs[0],
+					xh, &metas->num_metas,
+					&metas->credits, &metas->num_recs,
+					ocfs2_value_tree_metas_in_bucket,
+					bucket);
+}
+
+/*
+ * Given a xattr extent rec starting from blkno and having len clusters,
+ * iterate all the buckets calculate how much metadata we need for reflinking
+ * all the ocfs2_xattr_value_root and lock the allocators accordingly.
+ */
+static int ocfs2_lock_reflink_xattr_rec_allocators(
+				struct ocfs2_reflink_xattr_tree_args *args,
+				struct ocfs2_extent_tree *xt_et,
+				u64 blkno, u32 len, int *credits,
+				struct ocfs2_alloc_context **meta_ac,
+				struct ocfs2_alloc_context **data_ac)
+{
+	int ret, num_free_extents;
+	struct ocfs2_value_tree_metas metas;
+	struct ocfs2_super *osb = OCFS2_SB(args->reflink->old_inode->i_sb);
+	struct ocfs2_refcount_block *rb;
+
+	memset(&metas, 0, sizeof(metas));
+
+	ret = ocfs2_iterate_xattr_buckets(args->reflink->old_inode, blkno, len,
+					  ocfs2_calc_value_tree_metas, &metas);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	*credits = metas.credits;
+
+	/*
+	 * Calculate we need for refcount tree change.
+	 *
+	 * We need to add/modify num_recs in refcount tree, so just calculate
+	 * an approximate number we need for refcount tree change.
+	 * Sometimes we need to split the tree, and after split,  half recs
+	 * will be moved to the new block, and a new block can only provide
+	 * half number of recs. So we multiple new blocks by 2.
+	 * In the end, we have to add credits for modifying the already
+	 * existed refcount block.
+	 */
+	rb = (struct ocfs2_refcount_block *)args->reflink->ref_root_bh->b_data;
+	metas.num_recs =
+		(metas.num_recs + ocfs2_refcount_recs_per_rb(osb->sb) - 1) /
+		 ocfs2_refcount_recs_per_rb(osb->sb) * 2;
+	metas.num_metas += metas.num_recs;
+	*credits += metas.num_recs +
+		    metas.num_recs * OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)
+		*credits += le16_to_cpu(rb->rf_list.l_tree_depth) *
+			    le16_to_cpu(rb->rf_list.l_next_free_rec) + 1;
+	else
+		*credits += 1;
+
+	/* count in the xattr tree change. */
+	num_free_extents = ocfs2_num_free_extents(osb, xt_et);
+	if (num_free_extents < 0) {
+		ret = num_free_extents;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (num_free_extents < len)
+		metas.num_metas += ocfs2_extend_meta_needed(xt_et->et_root_el);
+
+	*credits += ocfs2_calc_extend_credits(osb->sb,
+					      xt_et->et_root_el, len);
+
+	if (metas.num_metas) {
+		ret = ocfs2_reserve_new_metadata_blocks(osb, metas.num_metas,
+							meta_ac);
 		if (ret) {
 			mlog_errno(ret);
 			goto out;
 		}
+	}
 
-		ret = ocfs2_rm_xattr_cluster(inode, xb_bh,
-					     p_blkno, e_cpos, num_clusters);
+	if (len) {
+		ret = ocfs2_reserve_clusters(osb, len, data_ac);
+		if (ret)
+			mlog_errno(ret);
+	}
+out:
+	if (ret) {
+		if (*meta_ac) {
+			ocfs2_free_alloc_context(*meta_ac);
+			meta_ac = NULL;
+		}
+	}
+
+	return ret;
+}
+
+static int ocfs2_reflink_xattr_buckets(handle_t *handle,
+				u64 blkno, u64 new_blkno, u32 clusters,
+				struct ocfs2_alloc_context *meta_ac,
+				struct ocfs2_alloc_context *data_ac,
+				struct ocfs2_reflink_xattr_tree_args *args)
+{
+	int i, j, ret = 0;
+	struct super_block *sb = args->reflink->old_inode->i_sb;
+	u32 bpc = ocfs2_xattr_buckets_per_cluster(OCFS2_SB(sb));
+	u32 num_buckets = clusters * bpc;
+	int bpb = args->old_bucket->bu_blocks;
+	struct ocfs2_xattr_value_buf vb = {
+		.vb_access = ocfs2_journal_access,
+	};
+
+	for (i = 0; i < num_buckets; i++, blkno += bpb, new_blkno += bpb) {
+		ret = ocfs2_read_xattr_bucket(args->old_bucket, blkno);
 		if (ret) {
 			mlog_errno(ret);
 			break;
 		}
 
-		if (e_cpos == 0)
+		ret = ocfs2_init_xattr_bucket(args->new_bucket, new_blkno);
+		if (ret) {
+			mlog_errno(ret);
 			break;
+		}
 
-		name_hash = e_cpos - 1;
+		/*
+		 * The real bucket num in this series of blocks is stored
+		 * in the 1st bucket.
+		 */
+		if (i == 0)
+			num_buckets = le16_to_cpu(
+				bucket_xh(args->old_bucket)->xh_num_buckets);
+
+		ret = ocfs2_xattr_bucket_journal_access(handle,
+						args->new_bucket,
+						OCFS2_JOURNAL_ACCESS_CREATE);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		for (j = 0; j < bpb; j++)
+			memcpy(bucket_block(args->new_bucket, j),
+			       bucket_block(args->old_bucket, j),
+			       sb->s_blocksize);
+
+		ocfs2_xattr_bucket_journal_dirty(handle, args->new_bucket);
+
+		ret = ocfs2_reflink_xattr_header(handle, args->reflink,
+					args->old_bucket->bu_bhs[0],
+					bucket_xh(args->old_bucket),
+					args->new_bucket->bu_bhs[0],
+					bucket_xh(args->new_bucket),
+					&vb, meta_ac,
+					ocfs2_get_reflink_xattr_value_root,
+					args);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		/*
+		 * Re-access and dirty the bucket to calculate metaecc.
+		 * Because we may extend the transaction in reflink_xattr_header
+		 * which will let the already accessed block gone.
+		 */
+		ret = ocfs2_xattr_bucket_journal_access(handle,
+						args->new_bucket,
+						OCFS2_JOURNAL_ACCESS_WRITE);
+		if (ret) {
+			mlog_errno(ret);
+			break;
+		}
+
+		ocfs2_xattr_bucket_journal_dirty(handle, args->new_bucket);
+		ocfs2_xattr_bucket_relse(args->old_bucket);
+		ocfs2_xattr_bucket_relse(args->new_bucket);
+	}
+
+	ocfs2_xattr_bucket_relse(args->old_bucket);
+	ocfs2_xattr_bucket_relse(args->new_bucket);
+	return ret;
+}
+/*
+ * Create the same xattr extent record in the new inode's xattr tree.
+ */
+static int ocfs2_reflink_xattr_rec(struct inode *inode,
+				   struct buffer_head *root_bh,
+				   u64 blkno,
+				   u32 cpos,
+				   u32 len,
+				   void *para)
+{
+	int ret, credits = 0;
+	u32 p_cluster, num_clusters;
+	u64 new_blkno;
+	handle_t *handle;
+	struct ocfs2_reflink_xattr_tree_args *args =
+			(struct ocfs2_reflink_xattr_tree_args *)para;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct ocfs2_alloc_context *meta_ac = NULL;
+	struct ocfs2_alloc_context *data_ac = NULL;
+	struct ocfs2_extent_tree et;
+
+	ocfs2_init_xattr_tree_extent_tree(&et,
+					  INODE_CACHE(args->reflink->new_inode),
+					  args->new_blk_bh);
+
+	ret = ocfs2_lock_reflink_xattr_rec_allocators(args, &et, blkno,
+						      len, &credits,
+						      &meta_ac, &data_ac);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	handle = ocfs2_start_trans(osb, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_claim_clusters(osb, handle, data_ac,
+				   len, &p_cluster, &num_clusters);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	new_blkno = ocfs2_clusters_to_blocks(osb->sb, p_cluster);
+
+	mlog(0, "reflink xattr buckets %llu to %llu, len %u\n",
+	     (unsigned long long)blkno, (unsigned long long)new_blkno, len);
+	ret = ocfs2_reflink_xattr_buckets(handle, blkno, new_blkno, len,
+					  meta_ac, data_ac, args);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	mlog(0, "insert new xattr extent rec start %llu len %u to %u\n",
+	     (unsigned long long)new_blkno, len, cpos);
+	ret = ocfs2_insert_extent(handle, &et, cpos, new_blkno,
+				  len, 0, meta_ac);
+	if (ret)
+		mlog_errno(ret);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+
+out:
+	if (meta_ac)
+		ocfs2_free_alloc_context(meta_ac);
+	if (data_ac)
+		ocfs2_free_alloc_context(data_ac);
+	return ret;
+}
+
+/*
+ * Create reflinked xattr buckets.
+ * We will add bucket one by one, and refcount all the xattrs in the bucket
+ * if they are stored outside.
+ */
+static int ocfs2_reflink_xattr_tree(struct ocfs2_xattr_reflink *args,
+				    struct buffer_head *blk_bh,
+				    struct buffer_head *new_blk_bh)
+{
+	int ret;
+	struct ocfs2_reflink_xattr_tree_args para;
+
+	memset(&para, 0, sizeof(para));
+	para.reflink = args;
+	para.old_blk_bh = blk_bh;
+	para.new_blk_bh = new_blk_bh;
+
+	para.old_bucket = ocfs2_xattr_bucket_new(args->old_inode);
+	if (!para.old_bucket) {
+		mlog_errno(-ENOMEM);
+		return -ENOMEM;
+	}
+
+	para.new_bucket = ocfs2_xattr_bucket_new(args->new_inode);
+	if (!para.new_bucket) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_iterate_xattr_index_block(args->old_inode, blk_bh,
+					      ocfs2_reflink_xattr_rec,
+					      &para);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	ocfs2_xattr_bucket_free(para.old_bucket);
+	ocfs2_xattr_bucket_free(para.new_bucket);
+	return ret;
+}
+
+static int ocfs2_reflink_xattr_in_block(struct ocfs2_xattr_reflink *args,
+					struct buffer_head *blk_bh)
+{
+	int ret, indexed = 0;
+	struct buffer_head *new_blk_bh = NULL;
+	struct ocfs2_xattr_block *xb =
+			(struct ocfs2_xattr_block *)blk_bh->b_data;
+
+
+	if (le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED)
+		indexed = 1;
+
+	ret = ocfs2_create_empty_xattr_block(args->new_inode, args->new_bh,
+					     &new_blk_bh, indexed);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	if (!(le16_to_cpu(xb->xb_flags) & OCFS2_XATTR_INDEXED))
+		ret = ocfs2_reflink_xattr_block(args, blk_bh, new_blk_bh);
+	else
+		ret = ocfs2_reflink_xattr_tree(args, blk_bh, new_blk_bh);
+	if (ret)
+		mlog_errno(ret);
+
+out:
+	brelse(new_blk_bh);
+	return ret;
+}
+
+static int ocfs2_reflink_xattr_no_security(struct ocfs2_xattr_entry *xe)
+{
+	int type = ocfs2_xattr_get_type(xe);
+
+	return type != OCFS2_XATTR_INDEX_SECURITY &&
+	       type != OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS &&
+	       type != OCFS2_XATTR_INDEX_POSIX_ACL_DEFAULT;
+}
+
+int ocfs2_reflink_xattrs(struct inode *old_inode,
+			 struct buffer_head *old_bh,
+			 struct inode *new_inode,
+			 struct buffer_head *new_bh,
+			 bool preserve_security)
+{
+	int ret;
+	struct ocfs2_xattr_reflink args;
+	struct ocfs2_inode_info *oi = OCFS2_I(old_inode);
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *)old_bh->b_data;
+	struct buffer_head *blk_bh = NULL;
+	struct ocfs2_cached_dealloc_ctxt dealloc;
+	struct ocfs2_refcount_tree *ref_tree;
+	struct buffer_head *ref_root_bh = NULL;
+
+	ret = ocfs2_lock_refcount_tree(OCFS2_SB(old_inode->i_sb),
+				       le64_to_cpu(di->i_refcount_loc),
+				       1, &ref_tree, &ref_root_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ocfs2_init_dealloc_ctxt(&dealloc);
+
+	args.old_inode = old_inode;
+	args.new_inode = new_inode;
+	args.old_bh = old_bh;
+	args.new_bh = new_bh;
+	args.ref_ci = &ref_tree->rf_ci;
+	args.ref_root_bh = ref_root_bh;
+	args.dealloc = &dealloc;
+	if (preserve_security)
+		args.xattr_reflinked = NULL;
+	else
+		args.xattr_reflinked = ocfs2_reflink_xattr_no_security;
+
+	if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL) {
+		ret = ocfs2_reflink_xattr_inline(&args);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_unlock;
+		}
+	}
+
+	if (!di->i_xattr_loc)
+		goto out_unlock;
+
+	ret = ocfs2_read_xattr_block(old_inode, le64_to_cpu(di->i_xattr_loc),
+				     &blk_bh);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	ret = ocfs2_reflink_xattr_in_block(&args, blk_bh);
+	if (ret)
+		mlog_errno(ret);
+
+	brelse(blk_bh);
+
+out_unlock:
+	ocfs2_unlock_refcount_tree(OCFS2_SB(old_inode->i_sb),
+				   ref_tree, 1);
+	brelse(ref_root_bh);
+
+	if (ocfs2_dealloc_has_cluster(&dealloc)) {
+		ocfs2_schedule_truncate_log_flush(OCFS2_SB(old_inode->i_sb), 1);
+		ocfs2_run_deallocs(OCFS2_SB(old_inode->i_sb), &dealloc);
 	}
 
 out:
@@ -5306,6 +6961,51 @@ out:
 }
 
 /*
+ * Initialize security and acl for a already created inode.
+ * Used for reflink a non-preserve-security file.
+ *
+ * It uses common api like ocfs2_xattr_set, so the caller
+ * must not hold any lock expect i_mutex.
+ */
+int ocfs2_init_security_and_acl(struct inode *dir,
+				struct inode *inode)
+{
+	int ret = 0;
+	struct buffer_head *dir_bh = NULL;
+	struct ocfs2_security_xattr_info si = {
+		.enable = 1,
+	};
+
+	ret = ocfs2_init_security_get(inode, dir, &si);
+	if (!ret) {
+		ret = ocfs2_xattr_security_set(inode, si.name,
+					       si.value, si.value_len,
+					       XATTR_CREATE);
+		if (ret) {
+			mlog_errno(ret);
+			goto leave;
+		}
+	} else if (ret != -EOPNOTSUPP) {
+		mlog_errno(ret);
+		goto leave;
+	}
+
+	ret = ocfs2_inode_lock(dir, &dir_bh, 0);
+	if (ret) {
+		mlog_errno(ret);
+		goto leave;
+	}
+
+	ret = ocfs2_init_acl(NULL, inode, dir, NULL, dir_bh, NULL, NULL);
+	if (ret)
+		mlog_errno(ret);
+
+	ocfs2_inode_unlock(dir, 0);
+	brelse(dir_bh);
+leave:
+	return ret;
+}
+/*
  * 'security' attributes support
  */
 static size_t ocfs2_xattr_security_list(struct inode *inode, char *list,
diff --git a/fs/ocfs2/xattr.h b/fs/ocfs2/xattr.h
index 1ca7e9a1b7bc..08e36389f56d 100644
--- a/fs/ocfs2/xattr.h
+++ b/fs/ocfs2/xattr.h
@@ -55,6 +55,8 @@ int ocfs2_xattr_set_handle(handle_t *, struct inode *, struct buffer_head *,
 			   int, const char *, const void *, size_t, int,
 			   struct ocfs2_alloc_context *,
 			   struct ocfs2_alloc_context *);
+int ocfs2_has_inline_xattr_value_outside(struct inode *inode,
+					 struct ocfs2_dinode *di);
 int ocfs2_xattr_remove(struct inode *, struct buffer_head *);
 int ocfs2_init_security_get(struct inode *, struct inode *,
 			    struct ocfs2_security_xattr_info *);
@@ -83,5 +85,16 @@ struct ocfs2_xattr_value_buf {
 	struct ocfs2_xattr_value_root	*vb_xv;
 };
 
-
+int ocfs2_xattr_attach_refcount_tree(struct inode *inode,
+				     struct buffer_head *fe_bh,
+				     struct ocfs2_caching_info *ref_ci,
+				     struct buffer_head *ref_root_bh,
+				     struct ocfs2_cached_dealloc_ctxt *dealloc);
+int ocfs2_reflink_xattrs(struct inode *old_inode,
+			 struct buffer_head *old_bh,
+			 struct inode *new_inode,
+			 struct buffer_head *new_bh,
+			 bool preserve_security);
+int ocfs2_init_security_and_acl(struct inode *dir,
+				struct inode *inode);
 #endif /* OCFS2_XATTR_H */
diff --git a/fs/omfs/dir.c b/fs/omfs/dir.c
index c7275cfbdcfb..b42d62419034 100644
--- a/fs/omfs/dir.c
+++ b/fs/omfs/dir.c
@@ -489,7 +489,7 @@ out:
 	return ret;
 }
 
-struct inode_operations omfs_dir_inops = {
+const struct inode_operations omfs_dir_inops = {
 	.lookup = omfs_lookup,
 	.mkdir = omfs_mkdir,
 	.rename = omfs_rename,
@@ -498,7 +498,7 @@ struct inode_operations omfs_dir_inops = {
 	.rmdir = omfs_rmdir,
 };
 
-struct file_operations omfs_dir_operations = {
+const struct file_operations omfs_dir_operations = {
 	.read = generic_read_dir,
 	.readdir = omfs_readdir,
 	.llseek = generic_file_llseek,
diff --git a/fs/omfs/file.c b/fs/omfs/file.c
index d17e774eaf45..399487c09364 100644
--- a/fs/omfs/file.c
+++ b/fs/omfs/file.c
@@ -322,7 +322,7 @@ static sector_t omfs_bmap(struct address_space *mapping, sector_t block)
 	return generic_block_bmap(mapping, block, omfs_get_block);
 }
 
-struct file_operations omfs_file_operations = {
+const struct file_operations omfs_file_operations = {
 	.llseek = generic_file_llseek,
 	.read = do_sync_read,
 	.write = do_sync_write,
@@ -333,11 +333,11 @@ struct file_operations omfs_file_operations = {
 	.splice_read = generic_file_splice_read,
 };
 
-struct inode_operations omfs_file_inops = {
+const struct inode_operations omfs_file_inops = {
 	.truncate = omfs_truncate
 };
 
-struct address_space_operations omfs_aops = {
+const struct address_space_operations omfs_aops = {
 	.readpage = omfs_readpage,
 	.readpages = omfs_readpages,
 	.writepage = omfs_writepage,
diff --git a/fs/omfs/inode.c b/fs/omfs/inode.c
index 379ae5fb4411..f3b7c1541f3a 100644
--- a/fs/omfs/inode.c
+++ b/fs/omfs/inode.c
@@ -278,7 +278,7 @@ static int omfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	return 0;
 }
 
-static struct super_operations omfs_sops = {
+static const struct super_operations omfs_sops = {
 	.write_inode	= omfs_write_inode,
 	.delete_inode	= omfs_delete_inode,
 	.put_super	= omfs_put_super,
diff --git a/fs/omfs/omfs.h b/fs/omfs/omfs.h
index 2bc0f0670406..ebe2fdbe535e 100644
--- a/fs/omfs/omfs.h
+++ b/fs/omfs/omfs.h
@@ -44,16 +44,16 @@ extern int omfs_allocate_range(struct super_block *sb, int min_request,
 extern int omfs_clear_range(struct super_block *sb, u64 block, int count);
 
 /* dir.c */
-extern struct file_operations omfs_dir_operations;
-extern struct inode_operations omfs_dir_inops;
+extern const struct file_operations omfs_dir_operations;
+extern const struct inode_operations omfs_dir_inops;
 extern int omfs_make_empty(struct inode *inode, struct super_block *sb);
 extern int omfs_is_bad(struct omfs_sb_info *sbi, struct omfs_header *header,
 			u64 fsblock);
 
 /* file.c */
-extern struct file_operations omfs_file_operations;
-extern struct inode_operations omfs_file_inops;
-extern struct address_space_operations omfs_aops;
+extern const struct file_operations omfs_file_operations;
+extern const struct inode_operations omfs_file_inops;
+extern const struct address_space_operations omfs_aops;
 extern void omfs_make_empty_table(struct buffer_head *bh, int offset);
 extern int omfs_shrink_inode(struct inode *inode);
 
diff --git a/fs/open.c b/fs/open.c
index 31191bf513e4..4f01e06227c6 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -290,10 +290,9 @@ out:
 	return error;
 }
 
-SYSCALL_DEFINE2(truncate, const char __user *, path, unsigned long, length)
+SYSCALL_DEFINE2(truncate, const char __user *, path, long, length)
 {
-	/* on 32-bit boxen it will cut the range 2^31--2^32-1 off */
-	return do_sys_truncate(path, (long)length);
+	return do_sys_truncate(path, length);
 }
 
 static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index fbeaddf595d3..7b685e10cbad 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -581,7 +581,7 @@ try_scan:
 		}
 
 		if (from + size > get_capacity(disk)) {
-			struct block_device_operations *bdops = disk->fops;
+			const struct block_device_operations *bdops = disk->fops;
 			unsigned long long capacity;
 
 			printk(KERN_WARNING
diff --git a/fs/pipe.c b/fs/pipe.c
index 52c415114838..ae17d026aaa3 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -777,36 +777,55 @@ pipe_rdwr_release(struct inode *inode, struct file *filp)
 static int
 pipe_read_open(struct inode *inode, struct file *filp)
 {
-	/* We could have perhaps used atomic_t, but this and friends
-	   below are the only places.  So it doesn't seem worthwhile.  */
+	int ret = -ENOENT;
+
 	mutex_lock(&inode->i_mutex);
-	inode->i_pipe->readers++;
+
+	if (inode->i_pipe) {
+		ret = 0;
+		inode->i_pipe->readers++;
+	}
+
 	mutex_unlock(&inode->i_mutex);
 
-	return 0;
+	return ret;
 }
 
 static int
 pipe_write_open(struct inode *inode, struct file *filp)
 {
+	int ret = -ENOENT;
+
 	mutex_lock(&inode->i_mutex);
-	inode->i_pipe->writers++;
+
+	if (inode->i_pipe) {
+		ret = 0;
+		inode->i_pipe->writers++;
+	}
+
 	mutex_unlock(&inode->i_mutex);
 
-	return 0;
+	return ret;
 }
 
 static int
 pipe_rdwr_open(struct inode *inode, struct file *filp)
 {
+	int ret = -ENOENT;
+
 	mutex_lock(&inode->i_mutex);
-	if (filp->f_mode & FMODE_READ)
-		inode->i_pipe->readers++;
-	if (filp->f_mode & FMODE_WRITE)
-		inode->i_pipe->writers++;
+
+	if (inode->i_pipe) {
+		ret = 0;
+		if (filp->f_mode & FMODE_READ)
+			inode->i_pipe->readers++;
+		if (filp->f_mode & FMODE_WRITE)
+			inode->i_pipe->writers++;
+	}
+
 	mutex_unlock(&inode->i_mutex);
 
-	return 0;
+	return ret;
 }
 
 /*
diff --git a/fs/proc/array.c b/fs/proc/array.c
index 762aea9c9c71..e209f64ab27b 100644
--- a/fs/proc/array.c
+++ b/fs/proc/array.c
@@ -82,6 +82,7 @@
 #include <linux/pid_namespace.h>
 #include <linux/ptrace.h>
 #include <linux/tracehook.h>
+#include <linux/swapops.h>
 
 #include <asm/pgtable.h>
 #include <asm/processor.h>
@@ -321,6 +322,94 @@ static inline void task_context_switch_counts(struct seq_file *m,
 			p->nivcsw);
 }
 
+#ifdef CONFIG_MMU
+
+struct stack_stats {
+	struct vm_area_struct *vma;
+	unsigned long	startpage;
+	unsigned long	usage;
+};
+
+static int stack_usage_pte_range(pmd_t *pmd, unsigned long addr,
+				unsigned long end, struct mm_walk *walk)
+{
+	struct stack_stats *ss = walk->private;
+	struct vm_area_struct *vma = ss->vma;
+	pte_t *pte, ptent;
+	spinlock_t *ptl;
+	int ret = 0;
+
+	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
+	for (; addr != end; pte++, addr += PAGE_SIZE) {
+		ptent = *pte;
+
+#ifdef CONFIG_STACK_GROWSUP
+		if (pte_present(ptent) || is_swap_pte(ptent))
+			ss->usage = addr - ss->startpage + PAGE_SIZE;
+#else
+		if (pte_present(ptent) || is_swap_pte(ptent)) {
+			ss->usage = ss->startpage - addr + PAGE_SIZE;
+			pte++;
+			ret = 1;
+			break;
+		}
+#endif
+	}
+	pte_unmap_unlock(pte - 1, ptl);
+	cond_resched();
+	return ret;
+}
+
+static inline unsigned long get_stack_usage_in_bytes(struct vm_area_struct *vma,
+				struct task_struct *task)
+{
+	struct stack_stats ss;
+	struct mm_walk stack_walk = {
+		.pmd_entry = stack_usage_pte_range,
+		.mm = vma->vm_mm,
+		.private = &ss,
+	};
+
+	if (!vma->vm_mm || is_vm_hugetlb_page(vma))
+		return 0;
+
+	ss.vma = vma;
+	ss.startpage = task->stack_start & PAGE_MASK;
+	ss.usage = 0;
+
+#ifdef CONFIG_STACK_GROWSUP
+	walk_page_range(KSTK_ESP(task) & PAGE_MASK, vma->vm_end,
+		&stack_walk);
+#else
+	walk_page_range(vma->vm_start, (KSTK_ESP(task) & PAGE_MASK) + PAGE_SIZE,
+		&stack_walk);
+#endif
+	return ss.usage;
+}
+
+static inline void task_show_stack_usage(struct seq_file *m,
+						struct task_struct *task)
+{
+	struct vm_area_struct	*vma;
+	struct mm_struct	*mm = get_task_mm(task);
+
+	if (mm) {
+		down_read(&mm->mmap_sem);
+		vma = find_vma(mm, task->stack_start);
+		if (vma)
+			seq_printf(m, "Stack usage:\t%lu kB\n",
+				get_stack_usage_in_bytes(vma, task) >> 10);
+
+		up_read(&mm->mmap_sem);
+		mmput(mm);
+	}
+}
+#else
+static void task_show_stack_usage(struct seq_file *m, struct task_struct *task)
+{
+}
+#endif		/* CONFIG_MMU */
+
 static void task_cpus_allowed(struct seq_file *m, struct task_struct *task)
 {
 	seq_printf(m, "Cpus_allowed:\t");
@@ -351,6 +440,7 @@ int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
 	task_show_regs(m, task);
 #endif
 	task_context_switch_counts(m, task);
+	task_show_stack_usage(m, task);
 	return 0;
 }
 
@@ -492,7 +582,7 @@ static int do_task_stat(struct seq_file *m, struct pid_namespace *ns,
 		rsslim,
 		mm ? mm->start_code : 0,
 		mm ? mm->end_code : 0,
-		(permitted && mm) ? mm->start_stack : 0,
+		(permitted) ? task->stack_start : 0,
 		esp,
 		eip,
 		/* The signal information here is obsolete.
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 6f742f6658a9..837469a96598 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -447,7 +447,7 @@ static int proc_oom_score(struct task_struct *task, char *buffer)
 
 	do_posix_clock_monotonic_gettime(&uptime);
 	read_lock(&tasklist_lock);
-	points = badness(task, uptime.tv_sec);
+	points = badness(task->group_leader, uptime.tv_sec);
 	read_unlock(&tasklist_lock);
 	return sprintf(buffer, "%lu\n", points);
 }
@@ -458,7 +458,7 @@ struct limit_names {
 };
 
 static const struct limit_names lnames[RLIM_NLIMITS] = {
-	[RLIMIT_CPU] = {"Max cpu time", "ms"},
+	[RLIMIT_CPU] = {"Max cpu time", "seconds"},
 	[RLIMIT_FSIZE] = {"Max file size", "bytes"},
 	[RLIMIT_DATA] = {"Max data size", "bytes"},
 	[RLIMIT_STACK] = {"Max stack size", "bytes"},
@@ -999,11 +999,17 @@ static ssize_t oom_adjust_read(struct file *file, char __user *buf,
 	struct task_struct *task = get_proc_task(file->f_path.dentry->d_inode);
 	char buffer[PROC_NUMBUF];
 	size_t len;
-	int oom_adjust;
+	int oom_adjust = OOM_DISABLE;
+	unsigned long flags;
 
 	if (!task)
 		return -ESRCH;
-	oom_adjust = task->oomkilladj;
+
+	if (lock_task_sighand(task, &flags)) {
+		oom_adjust = task->signal->oom_adj;
+		unlock_task_sighand(task, &flags);
+	}
+
 	put_task_struct(task);
 
 	len = snprintf(buffer, sizeof(buffer), "%i\n", oom_adjust);
@@ -1015,32 +1021,44 @@ static ssize_t oom_adjust_write(struct file *file, const char __user *buf,
 				size_t count, loff_t *ppos)
 {
 	struct task_struct *task;
-	char buffer[PROC_NUMBUF], *end;
-	int oom_adjust;
+	char buffer[PROC_NUMBUF];
+	long oom_adjust;
+	unsigned long flags;
+	int err;
 
 	memset(buffer, 0, sizeof(buffer));
 	if (count > sizeof(buffer) - 1)
 		count = sizeof(buffer) - 1;
 	if (copy_from_user(buffer, buf, count))
 		return -EFAULT;
-	oom_adjust = simple_strtol(buffer, &end, 0);
+
+	err = strict_strtol(strstrip(buffer), 0, &oom_adjust);
+	if (err)
+		return -EINVAL;
 	if ((oom_adjust < OOM_ADJUST_MIN || oom_adjust > OOM_ADJUST_MAX) &&
 	     oom_adjust != OOM_DISABLE)
 		return -EINVAL;
-	if (*end == '\n')
-		end++;
+
 	task = get_proc_task(file->f_path.dentry->d_inode);
 	if (!task)
 		return -ESRCH;
-	if (oom_adjust < task->oomkilladj && !capable(CAP_SYS_RESOURCE)) {
+	if (!lock_task_sighand(task, &flags)) {
+		put_task_struct(task);
+		return -ESRCH;
+	}
+
+	if (oom_adjust < task->signal->oom_adj && !capable(CAP_SYS_RESOURCE)) {
+		unlock_task_sighand(task, &flags);
 		put_task_struct(task);
 		return -EACCES;
 	}
-	task->oomkilladj = oom_adjust;
+
+	task->signal->oom_adj = oom_adjust;
+
+	unlock_task_sighand(task, &flags);
 	put_task_struct(task);
-	if (end - buffer == 0)
-		return -EIO;
-	return end - buffer;
+
+	return count;
 }
 
 static const struct file_operations proc_oom_adjust_operations = {
@@ -1169,17 +1187,16 @@ static ssize_t proc_fault_inject_write(struct file * file,
 		count = sizeof(buffer) - 1;
 	if (copy_from_user(buffer, buf, count))
 		return -EFAULT;
-	make_it_fail = simple_strtol(buffer, &end, 0);
-	if (*end == '\n')
-		end++;
+	make_it_fail = simple_strtol(strstrip(buffer), &end, 0);
+	if (*end)
+		return -EINVAL;
 	task = get_proc_task(file->f_dentry->d_inode);
 	if (!task)
 		return -ESRCH;
 	task->make_it_fail = make_it_fail;
 	put_task_struct(task);
-	if (end - buffer == 0)
-		return -EIO;
-	return end - buffer;
+
+	return count;
 }
 
 static const struct file_operations proc_fault_inject_operations = {
@@ -2586,9 +2603,6 @@ static void proc_flush_task_mnt(struct vfsmount *mnt, pid_t pid, pid_t tgid)
 		dput(dentry);
 	}
 
-	if (tgid == 0)
-		goto out;
-
 	name.name = buf;
 	name.len = snprintf(buf, sizeof(buf), "%d", tgid);
 	leader = d_hash_and_lookup(mnt->mnt_root, &name);
@@ -2645,17 +2659,16 @@ out:
 void proc_flush_task(struct task_struct *task)
 {
 	int i;
-	struct pid *pid, *tgid = NULL;
+	struct pid *pid, *tgid;
 	struct upid *upid;
 
 	pid = task_pid(task);
-	if (thread_group_leader(task))
-		tgid = task_tgid(task);
+	tgid = task_tgid(task);
 
 	for (i = 0; i <= pid->level; i++) {
 		upid = &pid->numbers[i];
 		proc_flush_task_mnt(upid->ns->proc_mnt, upid->nr,
-			tgid ? tgid->numbers[i].nr : 0);
+					tgid->numbers[i].nr);
 	}
 
 	upid = &pid->numbers[pid->level];
diff --git a/fs/proc/kcore.c b/fs/proc/kcore.c
index 59b43a068872..a44a7897fd4d 100644
--- a/fs/proc/kcore.c
+++ b/fs/proc/kcore.c
@@ -17,9 +17,14 @@
 #include <linux/elfcore.h>
 #include <linux/vmalloc.h>
 #include <linux/highmem.h>
+#include <linux/bootmem.h>
 #include <linux/init.h>
 #include <asm/uaccess.h>
 #include <asm/io.h>
+#include <linux/list.h>
+#include <linux/ioport.h>
+#include <linux/memory.h>
+#include <asm/sections.h>
 
 #define CORE_STR "CORE"
 
@@ -29,17 +34,6 @@
 
 static struct proc_dir_entry *proc_root_kcore;
 
-static int open_kcore(struct inode * inode, struct file * filp)
-{
-	return capable(CAP_SYS_RAWIO) ? 0 : -EPERM;
-}
-
-static ssize_t read_kcore(struct file *, char __user *, size_t, loff_t *);
-
-static const struct file_operations proc_kcore_operations = {
-	.read		= read_kcore,
-	.open		= open_kcore,
-};
 
 #ifndef kc_vaddr_to_offset
 #define	kc_vaddr_to_offset(v) ((v) - PAGE_OFFSET)
@@ -57,18 +51,19 @@ struct memelfnote
 	void *data;
 };
 
-static struct kcore_list *kclist;
+static LIST_HEAD(kclist_head);
 static DEFINE_RWLOCK(kclist_lock);
+static int kcore_need_update = 1;
 
 void
-kclist_add(struct kcore_list *new, void *addr, size_t size)
+kclist_add(struct kcore_list *new, void *addr, size_t size, int type)
 {
 	new->addr = (unsigned long)addr;
 	new->size = size;
+	new->type = type;
 
 	write_lock(&kclist_lock);
-	new->next = kclist;
-	kclist = new;
+	list_add_tail(&new->list, &kclist_head);
 	write_unlock(&kclist_lock);
 }
 
@@ -80,7 +75,7 @@ static size_t get_kcore_size(int *nphdr, size_t *elf_buflen)
 	*nphdr = 1; /* PT_NOTE */
 	size = 0;
 
-	for (m=kclist; m; m=m->next) {
+	list_for_each_entry(m, &kclist_head, list) {
 		try = kc_vaddr_to_offset((size_t)m->addr + m->size);
 		if (try > size)
 			size = try;
@@ -97,6 +92,177 @@ static size_t get_kcore_size(int *nphdr, size_t *elf_buflen)
 	return size + *elf_buflen;
 }
 
+static void free_kclist_ents(struct list_head *head)
+{
+	struct kcore_list *tmp, *pos;
+
+	list_for_each_entry_safe(pos, tmp, head, list) {
+		list_del(&pos->list);
+		kfree(pos);
+	}
+}
+/*
+ * Replace all KCORE_RAM/KCORE_VMEMMAP information with passed list.
+ */
+static void __kcore_update_ram(struct list_head *list)
+{
+	int nphdr;
+	size_t size;
+	struct kcore_list *tmp, *pos;
+	LIST_HEAD(garbage);
+
+	write_lock(&kclist_lock);
+	if (kcore_need_update) {
+		list_for_each_entry_safe(pos, tmp, &kclist_head, list) {
+			if (pos->type == KCORE_RAM
+				|| pos->type == KCORE_VMEMMAP)
+				list_move(&pos->list, &garbage);
+		}
+		list_splice_tail(list, &kclist_head);
+	} else
+		list_splice(list, &garbage);
+	kcore_need_update = 0;
+	proc_root_kcore->size = get_kcore_size(&nphdr, &size);
+	write_unlock(&kclist_lock);
+
+	free_kclist_ents(&garbage);
+}
+
+
+#ifdef CONFIG_HIGHMEM
+/*
+ * If no highmem, we can assume [0...max_low_pfn) continuous range of memory
+ * because memory hole is not as big as !HIGHMEM case.
+ * (HIGHMEM is special because part of memory is _invisible_ from the kernel.)
+ */
+static int kcore_update_ram(void)
+{
+	LIST_HEAD(head);
+	struct kcore_list *ent;
+	int ret = 0;
+
+	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
+	if (!ent)
+		return -ENOMEM;
+	ent->addr = (unsigned long)__va(0);
+	ent->size = max_low_pfn << PAGE_SHIFT;
+	ent->type = KCORE_RAM;
+	list_add(&ent->list, &head);
+	__kcore_update_ram(&head);
+	return ret;
+}
+
+#else /* !CONFIG_HIGHMEM */
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+/* calculate vmemmap's address from given system ram pfn and register it */
+int get_sparsemem_vmemmap_info(struct kcore_list *ent, struct list_head *head)
+{
+	unsigned long pfn = __pa(ent->addr) >> PAGE_SHIFT;
+	unsigned long nr_pages = ent->size >> PAGE_SHIFT;
+	unsigned long start, end;
+	struct kcore_list *vmm, *tmp;
+
+
+	start = ((unsigned long)pfn_to_page(pfn)) & PAGE_MASK;
+	end = ((unsigned long)pfn_to_page(pfn + nr_pages)) - 1;
+	end = ALIGN(end, PAGE_SIZE);
+	/* overlap check (because we have to align page */
+	list_for_each_entry(tmp, head, list) {
+		if (tmp->type != KCORE_VMEMMAP)
+			continue;
+		if (start < tmp->addr + tmp->size)
+			if (end > tmp->addr)
+				end = tmp->addr;
+	}
+	if (start < end) {
+		vmm = kmalloc(sizeof(*vmm), GFP_KERNEL);
+		if (!vmm)
+			return 0;
+		vmm->addr = start;
+		vmm->size = end - start;
+		vmm->type = KCORE_VMEMMAP;
+		list_add_tail(&vmm->list, head);
+	}
+	return 1;
+
+}
+#else
+int get_sparsemem_vmemmap_info(struct kcore_list *ent, struct list_head *head)
+{
+	return 1;
+}
+
+#endif
+
+static int
+kclist_add_private(unsigned long pfn, unsigned long nr_pages, void *arg)
+{
+	struct list_head *head = (struct list_head *)arg;
+	struct kcore_list *ent;
+
+	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
+	if (!ent)
+		return -ENOMEM;
+	ent->addr = (unsigned long)__va((pfn << PAGE_SHIFT));
+	ent->size = nr_pages << PAGE_SHIFT;
+
+	/* Sanity check: Can happen in 32bit arch...maybe */
+	if (ent->addr < (unsigned long) __va(0))
+		goto free_out;
+
+	/* cut not-mapped area. ....from ppc-32 code. */
+	if (ULONG_MAX - ent->addr < ent->size)
+		ent->size = ULONG_MAX - ent->addr;
+
+	/* cut when vmalloc() area is higher than direct-map area */
+	if (VMALLOC_START > (unsigned long)__va(0)) {
+		if (ent->addr > VMALLOC_START)
+			goto free_out;
+		if (VMALLOC_START - ent->addr < ent->size)
+			ent->size = VMALLOC_START - ent->addr;
+	}
+
+	ent->type = KCORE_RAM;
+	list_add_tail(&ent->list, head);
+
+	if (!get_sparsemem_vmemmap_info(ent, head)) {
+		list_del(&ent->list);
+		goto free_out;
+	}
+
+	return 0;
+free_out:
+	kfree(ent);
+	return 1;
+}
+
+static int kcore_update_ram(void)
+{
+	int nid, ret;
+	unsigned long end_pfn;
+	LIST_HEAD(head);
+
+	/* Not inialized....update now */
+	/* find out "max pfn" */
+	end_pfn = 0;
+	for_each_node_state(nid, N_HIGH_MEMORY) {
+		unsigned long node_end;
+		node_end  = NODE_DATA(nid)->node_start_pfn +
+			NODE_DATA(nid)->node_spanned_pages;
+		if (end_pfn < node_end)
+			end_pfn = node_end;
+	}
+	/* scan 0 to max_pfn */
+	ret = walk_system_ram_range(0, end_pfn, &head, kclist_add_private);
+	if (ret) {
+		free_kclist_ents(&head);
+		return -ENOMEM;
+	}
+	__kcore_update_ram(&head);
+	return ret;
+}
+#endif /* CONFIG_HIGHMEM */
 
 /*****************************************************************************/
 /*
@@ -192,7 +358,7 @@ static void elf_kcore_store_hdr(char *bufp, int nphdr, int dataoff)
 	nhdr->p_align	= 0;
 
 	/* setup ELF PT_LOAD program header for every area */
-	for (m=kclist; m; m=m->next) {
+	list_for_each_entry(m, &kclist_head, list) {
 		phdr = (struct elf_phdr *) bufp;
 		bufp += sizeof(struct elf_phdr);
 		offset += sizeof(struct elf_phdr);
@@ -265,7 +431,8 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
 	unsigned long start;
 
 	read_lock(&kclist_lock);
-	proc_root_kcore->size = size = get_kcore_size(&nphdr, &elf_buflen);
+	size = get_kcore_size(&nphdr, &elf_buflen);
+
 	if (buflen == 0 || *fpos >= size) {
 		read_unlock(&kclist_lock);
 		return 0;
@@ -317,7 +484,7 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
 		struct kcore_list *m;
 
 		read_lock(&kclist_lock);
-		for (m=kclist; m; m=m->next) {
+		list_for_each_entry(m, &kclist_head, list) {
 			if (start >= m->addr && start < (m->addr+m->size))
 				break;
 		}
@@ -326,45 +493,14 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
 		if (m == NULL) {
 			if (clear_user(buffer, tsz))
 				return -EFAULT;
-		} else if (is_vmalloc_addr((void *)start)) {
+		} else if (is_vmalloc_or_module_addr((void *)start)) {
 			char * elf_buf;
-			struct vm_struct *m;
-			unsigned long curstart = start;
-			unsigned long cursize = tsz;
 
 			elf_buf = kzalloc(tsz, GFP_KERNEL);
 			if (!elf_buf)
 				return -ENOMEM;
-
-			read_lock(&vmlist_lock);
-			for (m=vmlist; m && cursize; m=m->next) {
-				unsigned long vmstart;
-				unsigned long vmsize;
-				unsigned long msize = m->size - PAGE_SIZE;
-
-				if (((unsigned long)m->addr + msize) < 
-								curstart)
-					continue;
-				if ((unsigned long)m->addr > (curstart + 
-								cursize))
-					break;
-				vmstart = (curstart < (unsigned long)m->addr ? 
-					(unsigned long)m->addr : curstart);
-				if (((unsigned long)m->addr + msize) > 
-							(curstart + cursize))
-					vmsize = curstart + cursize - vmstart;
-				else
-					vmsize = (unsigned long)m->addr + 
-							msize - vmstart;
-				curstart = vmstart + vmsize;
-				cursize -= vmsize;
-				/* don't dump ioremap'd stuff! (TA) */
-				if (m->flags & VM_IOREMAP)
-					continue;
-				memcpy(elf_buf + (vmstart - start),
-					(char *)vmstart, vmsize);
-			}
-			read_unlock(&vmlist_lock);
+			vread(elf_buf, (char *)start, tsz);
+			/* we have to zero-fill user buffer even if no read */
 			if (copy_to_user(buffer, elf_buf, tsz)) {
 				kfree(elf_buf);
 				return -EFAULT;
@@ -402,12 +538,96 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
 	return acc;
 }
 
+
+static int open_kcore(struct inode *inode, struct file *filp)
+{
+	if (!capable(CAP_SYS_RAWIO))
+		return -EPERM;
+	if (kcore_need_update)
+		kcore_update_ram();
+	if (i_size_read(inode) != proc_root_kcore->size) {
+		mutex_lock(&inode->i_mutex);
+		i_size_write(inode, proc_root_kcore->size);
+		mutex_unlock(&inode->i_mutex);
+	}
+	return 0;
+}
+
+
+static const struct file_operations proc_kcore_operations = {
+	.read		= read_kcore,
+	.open		= open_kcore,
+};
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+/* just remember that we have to update kcore */
+static int __meminit kcore_callback(struct notifier_block *self,
+				    unsigned long action, void *arg)
+{
+	switch (action) {
+	case MEM_ONLINE:
+	case MEM_OFFLINE:
+		write_lock(&kclist_lock);
+		kcore_need_update = 1;
+		write_unlock(&kclist_lock);
+	}
+	return NOTIFY_OK;
+}
+#endif
+
+
+static struct kcore_list kcore_vmalloc;
+
+#ifdef CONFIG_ARCH_PROC_KCORE_TEXT
+static struct kcore_list kcore_text;
+/*
+ * If defined, special segment is used for mapping kernel text instead of
+ * direct-map area. We need to create special TEXT section.
+ */
+static void __init proc_kcore_text_init(void)
+{
+	kclist_add(&kcore_text, _stext, _end - _stext, KCORE_TEXT);
+}
+#else
+static void __init proc_kcore_text_init(void)
+{
+}
+#endif
+
+#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
+/*
+ * MODULES_VADDR has no intersection with VMALLOC_ADDR.
+ */
+struct kcore_list kcore_modules;
+static void __init add_modules_range(void)
+{
+	kclist_add(&kcore_modules, (void *)MODULES_VADDR,
+			MODULES_END - MODULES_VADDR, KCORE_VMALLOC);
+}
+#else
+static void __init add_modules_range(void)
+{
+}
+#endif
+
 static int __init proc_kcore_init(void)
 {
-	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL, &proc_kcore_operations);
-	if (proc_root_kcore)
-		proc_root_kcore->size =
-				(size_t)high_memory - PAGE_OFFSET + PAGE_SIZE;
+	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL,
+				      &proc_kcore_operations);
+	if (!proc_root_kcore) {
+		printk(KERN_ERR "couldn't create /proc/kcore\n");
+		return 0; /* Always returns 0. */
+	}
+	/* Store text area if it's special */
+	proc_kcore_text_init();
+	/* Store vmalloc area */
+	kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
+		VMALLOC_END - VMALLOC_START, KCORE_VMALLOC);
+	add_modules_range();
+	/* Store direct-map area from physical memory map */
+	kcore_update_ram();
+	hotplug_memory_notifier(kcore_callback, 0);
+
 	return 0;
 }
 module_init(proc_kcore_init);
diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
index d5c410d47fae..c7bff4f603ff 100644
--- a/fs/proc/meminfo.c
+++ b/fs/proc/meminfo.c
@@ -81,9 +81,11 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
 		"Writeback:      %8lu kB\n"
 		"AnonPages:      %8lu kB\n"
 		"Mapped:         %8lu kB\n"
+		"Shmem:          %8lu kB\n"
 		"Slab:           %8lu kB\n"
 		"SReclaimable:   %8lu kB\n"
 		"SUnreclaim:     %8lu kB\n"
+		"KernelStack:    %8lu kB\n"
 		"PageTables:     %8lu kB\n"
 #ifdef CONFIG_QUICKLIST
 		"Quicklists:     %8lu kB\n"
@@ -95,7 +97,11 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
 		"Committed_AS:   %8lu kB\n"
 		"VmallocTotal:   %8lu kB\n"
 		"VmallocUsed:    %8lu kB\n"
-		"VmallocChunk:   %8lu kB\n",
+		"VmallocChunk:   %8lu kB\n"
+#ifdef CONFIG_MEMORY_FAILURE
+		"HardwareCorrupted: %8lu kB\n"
+#endif
+		,
 		K(i.totalram),
 		K(i.freeram),
 		K(i.bufferram),
@@ -124,10 +130,12 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
 		K(global_page_state(NR_WRITEBACK)),
 		K(global_page_state(NR_ANON_PAGES)),
 		K(global_page_state(NR_FILE_MAPPED)),
+		K(global_page_state(NR_SHMEM)),
 		K(global_page_state(NR_SLAB_RECLAIMABLE) +
 				global_page_state(NR_SLAB_UNRECLAIMABLE)),
 		K(global_page_state(NR_SLAB_RECLAIMABLE)),
 		K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
+		global_page_state(NR_KERNEL_STACK) * THREAD_SIZE / 1024,
 		K(global_page_state(NR_PAGETABLE)),
 #ifdef CONFIG_QUICKLIST
 		K(quicklist_total_size()),
@@ -140,6 +148,9 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
 		(unsigned long)VMALLOC_TOTAL >> 10,
 		vmi.used >> 10,
 		vmi.largest_chunk >> 10
+#ifdef CONFIG_MEMORY_FAILURE
+		,atomic_long_read(&mce_bad_pages) << (PAGE_SHIFT - 10)
+#endif
 		);
 
 	hugetlb_report_meminfo(m);
diff --git a/fs/proc/nommu.c b/fs/proc/nommu.c
index 7e14d1a04001..9fe7d7ebe115 100644
--- a/fs/proc/nommu.c
+++ b/fs/proc/nommu.c
@@ -109,7 +109,7 @@ static void *nommu_region_list_next(struct seq_file *m, void *v, loff_t *pos)
 	return rb_next((struct rb_node *) v);
 }
 
-static struct seq_operations proc_nommu_region_list_seqop = {
+static const struct seq_operations proc_nommu_region_list_seqop = {
 	.start	= nommu_region_list_start,
 	.next	= nommu_region_list_next,
 	.stop	= nommu_region_list_stop,
diff --git a/fs/proc/page.c b/fs/proc/page.c
index 2707c6c7a20f..5033ce0d254b 100644
--- a/fs/proc/page.c
+++ b/fs/proc/page.c
@@ -2,6 +2,7 @@
 #include <linux/compiler.h>
 #include <linux/fs.h>
 #include <linux/init.h>
+#include <linux/ksm.h>
 #include <linux/mm.h>
 #include <linux/mmzone.h>
 #include <linux/proc_fs.h>
@@ -93,8 +94,11 @@ static const struct file_operations proc_kpagecount_operations = {
 #define KPF_COMPOUND_TAIL	16
 #define KPF_HUGE		17
 #define KPF_UNEVICTABLE		18
+#define KPF_HWPOISON		19
 #define KPF_NOPAGE		20
 
+#define KPF_KSM			21
+
 /* kernel hacking assistances
  * WARNING: subject to change, never rely on them!
  */
@@ -137,6 +141,8 @@ static u64 get_uflags(struct page *page)
 		u |= 1 << KPF_MMAP;
 	if (PageAnon(page))
 		u |= 1 << KPF_ANON;
+	if (PageKsm(page))
+		u |= 1 << KPF_KSM;
 
 	/*
 	 * compound pages: export both head/tail info
@@ -175,6 +181,10 @@ static u64 get_uflags(struct page *page)
 	u |= kpf_copy_bit(k, KPF_UNEVICTABLE,	PG_unevictable);
 	u |= kpf_copy_bit(k, KPF_MLOCKED,	PG_mlocked);
 
+#ifdef CONFIG_MEMORY_FAILURE
+	u |= kpf_copy_bit(k, KPF_HWPOISON,	PG_hwpoison);
+#endif
+
 #ifdef CONFIG_IA64_UNCACHED_ALLOCATOR
 	u |= kpf_copy_bit(k, KPF_UNCACHED,	PG_uncached);
 #endif
diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
index 9b1e4e9a16bf..f667e8aeabdf 100644
--- a/fs/proc/proc_sysctl.c
+++ b/fs/proc/proc_sysctl.c
@@ -153,7 +153,7 @@ static ssize_t proc_sys_call_handler(struct file *filp, void __user *buf,
 
 	/* careful: calling conventions are nasty here */
 	res = count;
-	error = table->proc_handler(table, write, filp, buf, &res, ppos);
+	error = table->proc_handler(table, write, buf, &res, ppos);
 	if (!error)
 		error = res;
 out:
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 9bd8be1d235c..2a1bef9203c6 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -243,6 +243,25 @@ static void show_map_vma(struct seq_file *m, struct vm_area_struct *vma)
 				} else if (vma->vm_start <= mm->start_stack &&
 					   vma->vm_end >= mm->start_stack) {
 					name = "[stack]";
+				} else {
+					unsigned long stack_start;
+					struct proc_maps_private *pmp;
+
+					pmp = m->private;
+					stack_start = pmp->task->stack_start;
+
+					if (vma->vm_start <= stack_start &&
+					    vma->vm_end >= stack_start) {
+						pad_len_spaces(m, len);
+						seq_printf(m,
+						 "[threadstack:%08lx]",
+#ifdef CONFIG_STACK_GROWSUP
+						 vma->vm_end - stack_start
+#else
+						 stack_start - vma->vm_start
+#endif
+						);
+					}
 				}
 			} else {
 				name = "[vdso]";
@@ -465,23 +484,28 @@ static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr,
 	return 0;
 }
 
+#define CLEAR_REFS_ALL 1
+#define CLEAR_REFS_ANON 2
+#define CLEAR_REFS_MAPPED 3
+
 static ssize_t clear_refs_write(struct file *file, const char __user *buf,
 				size_t count, loff_t *ppos)
 {
 	struct task_struct *task;
-	char buffer[PROC_NUMBUF], *end;
+	char buffer[PROC_NUMBUF];
 	struct mm_struct *mm;
 	struct vm_area_struct *vma;
+	long type;
 
 	memset(buffer, 0, sizeof(buffer));
 	if (count > sizeof(buffer) - 1)
 		count = sizeof(buffer) - 1;
 	if (copy_from_user(buffer, buf, count))
 		return -EFAULT;
-	if (!simple_strtol(buffer, &end, 0))
+	if (strict_strtol(strstrip(buffer), 10, &type))
+		return -EINVAL;
+	if (type < CLEAR_REFS_ALL || type > CLEAR_REFS_MAPPED)
 		return -EINVAL;
-	if (*end == '\n')
-		end++;
 	task = get_proc_task(file->f_path.dentry->d_inode);
 	if (!task)
 		return -ESRCH;
@@ -494,18 +518,31 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
 		down_read(&mm->mmap_sem);
 		for (vma = mm->mmap; vma; vma = vma->vm_next) {
 			clear_refs_walk.private = vma;
-			if (!is_vm_hugetlb_page(vma))
-				walk_page_range(vma->vm_start, vma->vm_end,
-						&clear_refs_walk);
+			if (is_vm_hugetlb_page(vma))
+				continue;
+			/*
+			 * Writing 1 to /proc/pid/clear_refs affects all pages.
+			 *
+			 * Writing 2 to /proc/pid/clear_refs only affects
+			 * Anonymous pages.
+			 *
+			 * Writing 3 to /proc/pid/clear_refs only affects file
+			 * mapped pages.
+			 */
+			if (type == CLEAR_REFS_ANON && vma->vm_file)
+				continue;
+			if (type == CLEAR_REFS_MAPPED && !vma->vm_file)
+				continue;
+			walk_page_range(vma->vm_start, vma->vm_end,
+					&clear_refs_walk);
 		}
 		flush_tlb_mm(mm);
 		up_read(&mm->mmap_sem);
 		mmput(mm);
 	}
 	put_task_struct(task);
-	if (end - buffer == 0)
-		return -EIO;
-	return end - buffer;
+
+	return count;
 }
 
 const struct file_operations proc_clear_refs_operations = {
diff --git a/fs/proc/uptime.c b/fs/proc/uptime.c
index 0c10a0b3f146..766b1d456050 100644
--- a/fs/proc/uptime.c
+++ b/fs/proc/uptime.c
@@ -4,13 +4,18 @@
 #include <linux/sched.h>
 #include <linux/seq_file.h>
 #include <linux/time.h>
+#include <linux/kernel_stat.h>
 #include <asm/cputime.h>
 
 static int uptime_proc_show(struct seq_file *m, void *v)
 {
 	struct timespec uptime;
 	struct timespec idle;
-	cputime_t idletime = cputime_add(init_task.utime, init_task.stime);
+	int i;
+	cputime_t idletime = cputime_zero;
+
+	for_each_possible_cpu(i)
+		idletime = cputime64_add(idletime, kstat_cpu(i).cpustat.idle);
 
 	do_posix_clock_monotonic_gettime(&uptime);
 	monotonic_to_bootbased(&uptime);
diff --git a/fs/qnx4/Kconfig b/fs/qnx4/Kconfig
index be8e0e1445b6..5f6089994042 100644
--- a/fs/qnx4/Kconfig
+++ b/fs/qnx4/Kconfig
@@ -6,20 +6,9 @@ config QNX4FS_FS
 	  QNX 4 and QNX 6 (the latter is also called QNX RTP).
 	  Further information is available at <http://www.qnx.com/>.
 	  Say Y if you intend to mount QNX hard disks or floppies.
-	  Unless you say Y to "QNX4FS read-write support" below, you will
-	  only be able to read these file systems.
 
 	  To compile this file system support as a module, choose M here: the
 	  module will be called qnx4.
 
 	  If you don't know whether you need it, then you don't need it:
 	  answer N.
-
-config QNX4FS_RW
-	bool "QNX4FS write support (DANGEROUS)"
-	depends on QNX4FS_FS && EXPERIMENTAL && BROKEN
-	help
-	  Say Y if you want to test write support for QNX4 file systems.
-
-	  It's currently broken, so for now:
-	  answer N.
diff --git a/fs/qnx4/Makefile b/fs/qnx4/Makefile
index e4d408cc5473..4a283b3f87f8 100644
--- a/fs/qnx4/Makefile
+++ b/fs/qnx4/Makefile
@@ -4,4 +4,4 @@
 
 obj-$(CONFIG_QNX4FS_FS) += qnx4.o
 
-qnx4-objs := inode.o dir.o namei.o file.o bitmap.o truncate.o
+qnx4-objs := inode.o dir.o namei.o bitmap.o
diff --git a/fs/qnx4/bitmap.c b/fs/qnx4/bitmap.c
index e1cd061a25f7..0afba069d567 100644
--- a/fs/qnx4/bitmap.c
+++ b/fs/qnx4/bitmap.c
@@ -78,84 +78,3 @@ unsigned long qnx4_count_free_blocks(struct super_block *sb)
 
 	return total_free;
 }
-
-#ifdef CONFIG_QNX4FS_RW
-
-int qnx4_is_free(struct super_block *sb, long block)
-{
-	int start = le32_to_cpu(qnx4_sb(sb)->BitMap->di_first_xtnt.xtnt_blk) - 1;
-	int size = le32_to_cpu(qnx4_sb(sb)->BitMap->di_size);
-	struct buffer_head *bh;
-	const char *g;
-	int ret = -EIO;
-
-	start += block / (QNX4_BLOCK_SIZE * 8);
-	QNX4DEBUG(("qnx4: is_free requesting block [%lu], bitmap in block [%lu]\n",
-		   (unsigned long) block, (unsigned long) start));
-	(void) size;		/* CHECKME */
-	bh = sb_bread(sb, start);
-	if (bh == NULL) {
-		return -EIO;
-	}
-	g = bh->b_data + (block % QNX4_BLOCK_SIZE);
-	if (((*g) & (1 << (block % 8))) == 0) {
-		QNX4DEBUG(("qnx4: is_free -> block is free\n"));
-		ret = 1;
-	} else {
-		QNX4DEBUG(("qnx4: is_free -> block is busy\n"));
-		ret = 0;
-	}
-	brelse(bh);
-
-	return ret;
-}
-
-int qnx4_set_bitmap(struct super_block *sb, long block, int busy)
-{
-	int start = le32_to_cpu(qnx4_sb(sb)->BitMap->di_first_xtnt.xtnt_blk) - 1;
-	int size = le32_to_cpu(qnx4_sb(sb)->BitMap->di_size);
-	struct buffer_head *bh;
-	char *g;
-
-	start += block / (QNX4_BLOCK_SIZE * 8);
-	QNX4DEBUG(("qnx4: set_bitmap requesting block [%lu], bitmap in block [%lu]\n",
-		   (unsigned long) block, (unsigned long) start));
-	(void) size;		/* CHECKME */
-	bh = sb_bread(sb, start);
-	if (bh == NULL) {
-		return -EIO;
-	}
-	g = bh->b_data + (block % QNX4_BLOCK_SIZE);
-	if (busy == 0) {
-		(*g) &= ~(1 << (block % 8));
-	} else {
-		(*g) |= (1 << (block % 8));
-	}
-	mark_buffer_dirty(bh);
-	brelse(bh);
-
-	return 0;
-}
-
-static void qnx4_clear_inode(struct inode *inode)
-{
-	struct qnx4_inode_entry *qnx4_ino = qnx4_raw_inode(inode);
-	/* What for? */
-	memset(qnx4_ino->di_fname, 0, sizeof qnx4_ino->di_fname);
-	qnx4_ino->di_size = 0;
-	qnx4_ino->di_num_xtnts = 0;
-	qnx4_ino->di_mode = 0;
-	qnx4_ino->di_status = 0;
-}
-
-void qnx4_free_inode(struct inode *inode)
-{
-	if (inode->i_ino < 1) {
-		printk("free_inode: inode 0 or nonexistent inode\n");
-		return;
-	}
-	qnx4_clear_inode(inode);
-	clear_inode(inode);
-}
-
-#endif
diff --git a/fs/qnx4/dir.c b/fs/qnx4/dir.c
index 003c68f3238b..86cc39cb1398 100644
--- a/fs/qnx4/dir.c
+++ b/fs/qnx4/dir.c
@@ -85,9 +85,4 @@ const struct file_operations qnx4_dir_operations =
 const struct inode_operations qnx4_dir_inode_operations =
 {
 	.lookup		= qnx4_lookup,
-#ifdef CONFIG_QNX4FS_RW
-	.create		= qnx4_create,
-	.unlink		= qnx4_unlink,
-	.rmdir		= qnx4_rmdir,
-#endif
 };
diff --git a/fs/qnx4/file.c b/fs/qnx4/file.c
deleted file mode 100644
index 09b170ac936c..000000000000
--- a/fs/qnx4/file.c
+++ /dev/null
@@ -1,40 +0,0 @@
-/*
- * QNX4 file system, Linux implementation.
- *
- * Version : 0.2.1
- *
- * Using parts of the xiafs filesystem.
- *
- * History :
- *
- * 25-05-1998 by Richard Frowijn : first release.
- * 21-06-1998 by Frank Denis : wrote qnx4_readpage to use generic_file_read.
- * 27-06-1998 by Frank Denis : file overwriting.
- */
-
-#include "qnx4.h"
-
-/*
- * We have mostly NULL's here: the current defaults are ok for
- * the qnx4 filesystem.
- */
-const struct file_operations qnx4_file_operations =
-{
-	.llseek		= generic_file_llseek,
-	.read		= do_sync_read,
-	.aio_read	= generic_file_aio_read,
-	.mmap		= generic_file_mmap,
-	.splice_read	= generic_file_splice_read,
-#ifdef CONFIG_QNX4FS_RW
-	.write		= do_sync_write,
-	.aio_write	= generic_file_aio_write,
-	.fsync		= simple_fsync,
-#endif
-};
-
-const struct inode_operations qnx4_file_inode_operations =
-{
-#ifdef CONFIG_QNX4FS_RW
-	.truncate	= qnx4_truncate,
-#endif
-};
diff --git a/fs/qnx4/inode.c b/fs/qnx4/inode.c
index 681df5fcd161..d2cd1798d8c4 100644
--- a/fs/qnx4/inode.c
+++ b/fs/qnx4/inode.c
@@ -28,73 +28,6 @@
 
 static const struct super_operations qnx4_sops;
 
-#ifdef CONFIG_QNX4FS_RW
-
-static void qnx4_delete_inode(struct inode *inode)
-{
-	QNX4DEBUG(("qnx4: deleting inode [%lu]\n", (unsigned long) inode->i_ino));
-	truncate_inode_pages(&inode->i_data, 0);
-	inode->i_size = 0;
-	qnx4_truncate(inode);
-	lock_kernel();
-	qnx4_free_inode(inode);
-	unlock_kernel();
-}
-
-static int qnx4_write_inode(struct inode *inode, int do_sync)
-{
-	struct qnx4_inode_entry *raw_inode;
-	int block, ino;
-	struct buffer_head *bh;
-	ino = inode->i_ino;
-
-	QNX4DEBUG(("qnx4: write inode 1.\n"));
-	if (inode->i_nlink == 0) {
-		return 0;
-	}
-	if (!ino) {
-		printk("qnx4: bad inode number on dev %s: %d is out of range\n",
-		       inode->i_sb->s_id, ino);
-		return -EIO;
-	}
-	QNX4DEBUG(("qnx4: write inode 2.\n"));
-	block = ino / QNX4_INODES_PER_BLOCK;
-	lock_kernel();
-	if (!(bh = sb_bread(inode->i_sb, block))) {
-		printk("qnx4: major problem: unable to read inode from dev "
-		       "%s\n", inode->i_sb->s_id);
-		unlock_kernel();
-		return -EIO;
-	}
-	raw_inode = ((struct qnx4_inode_entry *) bh->b_data) +
-	    (ino % QNX4_INODES_PER_BLOCK);
-	raw_inode->di_mode  = cpu_to_le16(inode->i_mode);
-	raw_inode->di_uid   = cpu_to_le16(fs_high2lowuid(inode->i_uid));
-	raw_inode->di_gid   = cpu_to_le16(fs_high2lowgid(inode->i_gid));
-	raw_inode->di_nlink = cpu_to_le16(inode->i_nlink);
-	raw_inode->di_size  = cpu_to_le32(inode->i_size);
-	raw_inode->di_mtime = cpu_to_le32(inode->i_mtime.tv_sec);
-	raw_inode->di_atime = cpu_to_le32(inode->i_atime.tv_sec);
-	raw_inode->di_ctime = cpu_to_le32(inode->i_ctime.tv_sec);
-	raw_inode->di_first_xtnt.xtnt_size = cpu_to_le32(inode->i_blocks);
-	mark_buffer_dirty(bh);
-	if (do_sync) {
-		sync_dirty_buffer(bh);
-		if (buffer_req(bh) && !buffer_uptodate(bh)) {
-			printk("qnx4: IO error syncing inode [%s:%08x]\n",
-					inode->i_sb->s_id, ino);
-			brelse(bh);
-			unlock_kernel();
-			return -EIO;
-		}
-	}
-	brelse(bh);
-	unlock_kernel();
-	return 0;
-}
-
-#endif
-
 static void qnx4_put_super(struct super_block *sb);
 static struct inode *qnx4_alloc_inode(struct super_block *sb);
 static void qnx4_destroy_inode(struct inode *inode);
@@ -108,10 +41,6 @@ static const struct super_operations qnx4_sops =
 	.put_super	= qnx4_put_super,
 	.statfs		= qnx4_statfs,
 	.remount_fs	= qnx4_remount,
-#ifdef CONFIG_QNX4FS_RW
-	.write_inode	= qnx4_write_inode,
-	.delete_inode	= qnx4_delete_inode,
-#endif
 };
 
 static int qnx4_remount(struct super_block *sb, int *flags, char *data)
@@ -120,15 +49,7 @@ static int qnx4_remount(struct super_block *sb, int *flags, char *data)
 
 	qs = qnx4_sb(sb);
 	qs->Version = QNX4_VERSION;
-#ifndef CONFIG_QNX4FS_RW
 	*flags |= MS_RDONLY;
-#endif
-	if (*flags & MS_RDONLY) {
-		return 0;
-	}
-
-	mark_buffer_dirty(qs->sb_buf);
-
 	return 0;
 }
 
@@ -354,9 +275,7 @@ static int qnx4_fill_super(struct super_block *s, void *data, int silent)
 	}
 	s->s_op = &qnx4_sops;
 	s->s_magic = QNX4_SUPER_MAGIC;
-#ifndef CONFIG_QNX4FS_RW
 	s->s_flags |= MS_RDONLY;	/* Yup, read-only yet */
-#endif
 	qnx4_sb(s)->sb_buf = bh;
 	qnx4_sb(s)->sb = (struct qnx4_super_block *) bh->b_data;
 
@@ -489,8 +408,7 @@ struct inode *qnx4_iget(struct super_block *sb, unsigned long ino)
 
 	memcpy(qnx4_inode, raw_inode, QNX4_DIR_ENTRY_SIZE);
 	if (S_ISREG(inode->i_mode)) {
-		inode->i_op = &qnx4_file_inode_operations;
-		inode->i_fop = &qnx4_file_operations;
+		inode->i_fop = &generic_ro_fops;
 		inode->i_mapping->a_ops = &qnx4_aops;
 		qnx4_i(inode)->mmu_private = inode->i_size;
 	} else if (S_ISDIR(inode->i_mode)) {
diff --git a/fs/qnx4/namei.c b/fs/qnx4/namei.c
index 5972ed214937..ae1e7edbacd6 100644
--- a/fs/qnx4/namei.c
+++ b/fs/qnx4/namei.c
@@ -134,108 +134,3 @@ out:
 
 	return NULL;
 }
-
-#ifdef CONFIG_QNX4FS_RW
-int qnx4_create(struct inode *dir, struct dentry *dentry, int mode,
-		struct nameidata *nd)
-{
-	QNX4DEBUG(("qnx4: qnx4_create\n"));
-	if (dir == NULL) {
-		return -ENOENT;
-	}
-	return -ENOSPC;
-}
-
-int qnx4_rmdir(struct inode *dir, struct dentry *dentry)
-{
-	struct buffer_head *bh;
-	struct qnx4_inode_entry *de;
-	struct inode *inode;
-	int retval;
-	int ino;
-
-	QNX4DEBUG(("qnx4: qnx4_rmdir [%s]\n", dentry->d_name.name));
-	lock_kernel();
-	bh = qnx4_find_entry(dentry->d_name.len, dir, dentry->d_name.name,
-			     &de, &ino);
-	if (bh == NULL) {
-		unlock_kernel();
-		return -ENOENT;
-	}
-	inode = dentry->d_inode;
-	if (inode->i_ino != ino) {
-		retval = -EIO;
-		goto end_rmdir;
-	}
-#if 0
-	if (!empty_dir(inode)) {
-		retval = -ENOTEMPTY;
-		goto end_rmdir;
-	}
-#endif
-	if (inode->i_nlink != 2) {
-		QNX4DEBUG(("empty directory has nlink!=2 (%d)\n", inode->i_nlink));
-	}
-	QNX4DEBUG(("qnx4: deleting directory\n"));
-	de->di_status = 0;
-	memset(de->di_fname, 0, sizeof de->di_fname);
-	de->di_mode = 0;
-	mark_buffer_dirty_inode(bh, dir);
-	clear_nlink(inode);
-	mark_inode_dirty(inode);
-	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME_SEC;
-	inode_dec_link_count(dir);
-	retval = 0;
-
-      end_rmdir:
-	brelse(bh);
-
-	unlock_kernel();
-	return retval;
-}
-
-int qnx4_unlink(struct inode *dir, struct dentry *dentry)
-{
-	struct buffer_head *bh;
-	struct qnx4_inode_entry *de;
-	struct inode *inode;
-	int retval;
-	int ino;
-
-	QNX4DEBUG(("qnx4: qnx4_unlink [%s]\n", dentry->d_name.name));
-	lock_kernel();
-	bh = qnx4_find_entry(dentry->d_name.len, dir, dentry->d_name.name,
-			     &de, &ino);
-	if (bh == NULL) {
-		unlock_kernel();
-		return -ENOENT;
-	}
-	inode = dentry->d_inode;
-	if (inode->i_ino != ino) {
-		retval = -EIO;
-		goto end_unlink;
-	}
-	retval = -EPERM;
-	if (!inode->i_nlink) {
-		QNX4DEBUG(("Deleting nonexistent file (%s:%lu), %d\n",
-			   inode->i_sb->s_id,
-			   inode->i_ino, inode->i_nlink));
-		inode->i_nlink = 1;
-	}
-	de->di_status = 0;
-	memset(de->di_fname, 0, sizeof de->di_fname);
-	de->di_mode = 0;
-	mark_buffer_dirty_inode(bh, dir);
-	dir->i_ctime = dir->i_mtime = CURRENT_TIME_SEC;
-	mark_inode_dirty(dir);
-	inode->i_ctime = dir->i_ctime;
-	inode_dec_link_count(inode);
-	retval = 0;
-
-end_unlink:
-	unlock_kernel();
-	brelse(bh);
-
-	return retval;
-}
-#endif
diff --git a/fs/qnx4/qnx4.h b/fs/qnx4/qnx4.h
index 9efc089454f6..33a60858203b 100644
--- a/fs/qnx4/qnx4.h
+++ b/fs/qnx4/qnx4.h
@@ -29,17 +29,9 @@ extern unsigned long qnx4_block_map(struct inode *inode, long iblock);
 
 extern struct buffer_head *qnx4_bread(struct inode *, int, int);
 
-extern const struct inode_operations qnx4_file_inode_operations;
 extern const struct inode_operations qnx4_dir_inode_operations;
-extern const struct file_operations qnx4_file_operations;
 extern const struct file_operations qnx4_dir_operations;
 extern int qnx4_is_free(struct super_block *sb, long block);
-extern int qnx4_set_bitmap(struct super_block *sb, long block, int busy);
-extern int qnx4_create(struct inode *inode, struct dentry *dentry, int mode, struct nameidata *nd);
-extern void qnx4_truncate(struct inode *inode);
-extern void qnx4_free_inode(struct inode *inode);
-extern int qnx4_unlink(struct inode *dir, struct dentry *dentry);
-extern int qnx4_rmdir(struct inode *dir, struct dentry *dentry);
 
 static inline struct qnx4_sb_info *qnx4_sb(struct super_block *sb)
 {
diff --git a/fs/qnx4/truncate.c b/fs/qnx4/truncate.c
deleted file mode 100644
index d94d9ee241fe..000000000000
--- a/fs/qnx4/truncate.c
+++ /dev/null
@@ -1,34 +0,0 @@
-/* 
- * QNX4 file system, Linux implementation.
- * 
- * Version : 0.1
- * 
- * Using parts of the xiafs filesystem.
- * 
- * History :
- * 
- * 30-06-1998 by Frank DENIS : ugly filler.
- */
-
-#include <linux/smp_lock.h>
-#include "qnx4.h"
-
-#ifdef CONFIG_QNX4FS_RW
-
-void qnx4_truncate(struct inode *inode)
-{
-	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
-	      S_ISLNK(inode->i_mode))) {
-		return;
-	}
-	lock_kernel();
-	if (!(S_ISDIR(inode->i_mode))) {
-		/* TODO */
-	}
-	QNX4DEBUG(("qnx4: qnx4_truncate called\n"));
-	inode->i_mtime = inode->i_ctime = CURRENT_TIME_SEC;
-	mark_inode_dirty(inode);
-	unlock_kernel();
-}
-
-#endif
diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 38f7bd559f35..39b49c42a7ed 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -1839,7 +1839,7 @@ EXPORT_SYMBOL(dquot_commit_info);
 /*
  * Definitions of diskquota operations.
  */
-struct dquot_operations dquot_operations = {
+const struct dquot_operations dquot_operations = {
 	.initialize	= dquot_initialize,
 	.drop		= dquot_drop,
 	.alloc_space	= dquot_alloc_space,
@@ -2461,7 +2461,7 @@ out:
 }
 EXPORT_SYMBOL(vfs_set_dqinfo);
 
-struct quotactl_ops vfs_quotactl_ops = {
+const struct quotactl_ops vfs_quotactl_ops = {
 	.quota_on	= vfs_quota_on,
 	.quota_off	= vfs_quota_off,
 	.quota_sync	= vfs_quota_sync,
diff --git a/fs/ramfs/file-nommu.c b/fs/ramfs/file-nommu.c
index 11f0c06316de..32fae4040ebf 100644
--- a/fs/ramfs/file-nommu.c
+++ b/fs/ramfs/file-nommu.c
@@ -69,14 +69,11 @@ int ramfs_nommu_expand_for_mapping(struct inode *inode, size_t newsize)
 	/* make various checks */
 	order = get_order(newsize);
 	if (unlikely(order >= MAX_ORDER))
-		goto too_big;
+		return -EFBIG;
 
-	limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
-	if (limit != RLIM_INFINITY && newsize > limit)
-		goto fsize_exceeded;
-
-	if (newsize > inode->i_sb->s_maxbytes)
-		goto too_big;
+	ret = inode_newsize_ok(inode, newsize);
+	if (ret)
+		return ret;
 
 	i_size_write(inode, newsize);
 
@@ -118,12 +115,7 @@ int ramfs_nommu_expand_for_mapping(struct inode *inode, size_t newsize)
 
 	return 0;
 
- fsize_exceeded:
-	send_sig(SIGXFSZ, current, 0);
- too_big:
-	return -EFBIG;
-
- add_error:
+add_error:
 	while (loop < npages)
 		__free_page(pages + loop++);
 	return ret;
diff --git a/fs/ramfs/inode.c b/fs/ramfs/inode.c
index a7f0110fca4c..a6090aa1a7c1 100644
--- a/fs/ramfs/inode.c
+++ b/fs/ramfs/inode.c
@@ -34,12 +34,10 @@
 #include <linux/ramfs.h>
 #include <linux/sched.h>
 #include <linux/parser.h>
+#include <linux/magic.h>
 #include <asm/uaccess.h>
 #include "internal.h"
 
-/* some random number */
-#define RAMFS_MAGIC	0x858458f6
-
 #define RAMFS_DEFAULT_MODE	0755
 
 static const struct super_operations ramfs_ops;
diff --git a/fs/read_write.c b/fs/read_write.c
index 6c8c55dec2bc..3ac28987f22a 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -839,9 +839,6 @@ static ssize_t do_sendfile(int out_fd, int in_fd, loff_t *ppos,
 		max = min(in_inode->i_sb->s_maxbytes, out_inode->i_sb->s_maxbytes);
 
 	pos = *ppos;
-	retval = -EINVAL;
-	if (unlikely(pos < 0))
-		goto fput_out;
 	if (unlikely(pos + count > max)) {
 		retval = -EOVERFLOW;
 		if (pos >= max)
diff --git a/fs/reiserfs/super.c b/fs/reiserfs/super.c
index 7adea74d6a8a..f0ad05f38022 100644
--- a/fs/reiserfs/super.c
+++ b/fs/reiserfs/super.c
@@ -612,7 +612,7 @@ static int reiserfs_mark_dquot_dirty(struct dquot *);
 static int reiserfs_write_info(struct super_block *, int);
 static int reiserfs_quota_on(struct super_block *, int, int, char *, int);
 
-static struct dquot_operations reiserfs_quota_operations = {
+static const struct dquot_operations reiserfs_quota_operations = {
 	.initialize = dquot_initialize,
 	.drop = dquot_drop,
 	.alloc_space = dquot_alloc_space,
@@ -629,7 +629,7 @@ static struct dquot_operations reiserfs_quota_operations = {
 	.destroy_dquot	= dquot_destroy,
 };
 
-static struct quotactl_ops reiserfs_qctl_operations = {
+static const struct quotactl_ops reiserfs_qctl_operations = {
 	.quota_on = reiserfs_quota_on,
 	.quota_off = vfs_quota_off,
 	.quota_sync = vfs_quota_sync,
diff --git a/fs/romfs/storage.c b/fs/romfs/storage.c
index b3208adf8e71..71e2b4d50a0a 100644
--- a/fs/romfs/storage.c
+++ b/fs/romfs/storage.c
@@ -253,11 +253,11 @@ ssize_t romfs_dev_strnlen(struct super_block *sb,
 
 #ifdef CONFIG_ROMFS_ON_MTD
 	if (sb->s_mtd)
-		return romfs_mtd_strnlen(sb, pos, limit);
+		return romfs_mtd_strnlen(sb, pos, maxlen);
 #endif
 #ifdef CONFIG_ROMFS_ON_BLOCK
 	if (sb->s_bdev)
-		return romfs_blk_strnlen(sb, pos, limit);
+		return romfs_blk_strnlen(sb, pos, maxlen);
 #endif
 	return -EIO;
 }
diff --git a/fs/romfs/super.c b/fs/romfs/super.c
index 4ab3c03d8f95..c117fa80d1e9 100644
--- a/fs/romfs/super.c
+++ b/fs/romfs/super.c
@@ -284,7 +284,7 @@ static const struct file_operations romfs_dir_operations = {
 	.readdir	= romfs_readdir,
 };
 
-static struct inode_operations romfs_dir_inode_operations = {
+static const struct inode_operations romfs_dir_inode_operations = {
 	.lookup		= romfs_lookup,
 };
 
@@ -528,7 +528,7 @@ static int romfs_fill_super(struct super_block *sb, void *data, int silent)
 	pos = (ROMFH_SIZE + len + 1 + ROMFH_PAD) & ROMFH_MASK;
 
 	root = romfs_iget(sb, pos);
-	if (!root)
+	if (IS_ERR(root))
 		goto error;
 
 	sb->s_root = d_alloc_root(root);
diff --git a/fs/select.c b/fs/select.c
index 8084834e123e..fd38ce2e32e3 100644
--- a/fs/select.c
+++ b/fs/select.c
@@ -15,6 +15,7 @@
  */
 
 #include <linux/kernel.h>
+#include <linux/sched.h>
 #include <linux/syscalls.h>
 #include <linux/module.h>
 #include <linux/slab.h>
@@ -41,22 +42,28 @@
  * better solutions..
  */
 
+#define MAX_SLACK	(100 * NSEC_PER_MSEC)
+
 static long __estimate_accuracy(struct timespec *tv)
 {
 	long slack;
 	int divfactor = 1000;
 
+	if (tv->tv_sec < 0)
+		return 0;
+
 	if (task_nice(current) > 0)
 		divfactor = divfactor / 5;
 
+	if (tv->tv_sec > MAX_SLACK / (NSEC_PER_SEC/divfactor))
+		return MAX_SLACK;
+
 	slack = tv->tv_nsec / divfactor;
 	slack += tv->tv_sec * (NSEC_PER_SEC/divfactor);
 
-	if (slack > 100 * NSEC_PER_MSEC)
-		slack =  100 * NSEC_PER_MSEC;
+	if (slack > MAX_SLACK)
+		return MAX_SLACK;
 
-	if (slack < 0)
-		slack = 0;
 	return slack;
 }
 
diff --git a/fs/seq_file.c b/fs/seq_file.c
index 6c959275f2d0..eae7d9dbf3ff 100644
--- a/fs/seq_file.c
+++ b/fs/seq_file.c
@@ -429,20 +429,21 @@ EXPORT_SYMBOL(mangle_path);
  */
 int seq_path(struct seq_file *m, struct path *path, char *esc)
 {
-	if (m->count < m->size) {
-		char *s = m->buf + m->count;
-		char *p = d_path(path, s, m->size - m->count);
+	char *buf;
+	size_t size = seq_get_buf(m, &buf);
+	int res = -1;
+
+	if (size) {
+		char *p = d_path(path, buf, size);
 		if (!IS_ERR(p)) {
-			s = mangle_path(s, p, esc);
-			if (s) {
-				p = m->buf + m->count;
-				m->count = s - m->buf;
-				return s - p;
-			}
+			char *end = mangle_path(buf, p, esc);
+			if (end)
+				res = end - buf;
 		}
 	}
-	m->count = m->size;
-	return -1;
+	seq_commit(m, res);
+
+	return res;
 }
 EXPORT_SYMBOL(seq_path);
 
@@ -454,26 +455,28 @@ EXPORT_SYMBOL(seq_path);
 int seq_path_root(struct seq_file *m, struct path *path, struct path *root,
 		  char *esc)
 {
-	int err = -ENAMETOOLONG;
-	if (m->count < m->size) {
-		char *s = m->buf + m->count;
+	char *buf;
+	size_t size = seq_get_buf(m, &buf);
+	int res = -ENAMETOOLONG;
+
+	if (size) {
 		char *p;
 
 		spin_lock(&dcache_lock);
-		p = __d_path(path, root, s, m->size - m->count);
+		p = __d_path(path, root, buf, size);
 		spin_unlock(&dcache_lock);
-		err = PTR_ERR(p);
+		res = PTR_ERR(p);
 		if (!IS_ERR(p)) {
-			s = mangle_path(s, p, esc);
-			if (s) {
-				p = m->buf + m->count;
-				m->count = s - m->buf;
-				return 0;
-			}
+			char *end = mangle_path(buf, p, esc);
+			if (end)
+				res = end - buf;
+			else
+				res = -ENAMETOOLONG;
 		}
 	}
-	m->count = m->size;
-	return err;
+	seq_commit(m, res);
+
+	return res < 0 ? res : 0;
 }
 
 /*
@@ -481,20 +484,21 @@ int seq_path_root(struct seq_file *m, struct path *path, struct path *root,
  */
 int seq_dentry(struct seq_file *m, struct dentry *dentry, char *esc)
 {
-	if (m->count < m->size) {
-		char *s = m->buf + m->count;
-		char *p = dentry_path(dentry, s, m->size - m->count);
+	char *buf;
+	size_t size = seq_get_buf(m, &buf);
+	int res = -1;
+
+	if (size) {
+		char *p = dentry_path(dentry, buf, size);
 		if (!IS_ERR(p)) {
-			s = mangle_path(s, p, esc);
-			if (s) {
-				p = m->buf + m->count;
-				m->count = s - m->buf;
-				return s - p;
-			}
+			char *end = mangle_path(buf, p, esc);
+			if (end)
+				res = end - buf;
 		}
 	}
-	m->count = m->size;
-	return -1;
+	seq_commit(m, res);
+
+	return res;
 }
 
 int seq_bitmap(struct seq_file *m, const unsigned long *bits,
diff --git a/fs/smbfs/inode.c b/fs/smbfs/inode.c
index 1402d2d54f52..1c4c8f089970 100644
--- a/fs/smbfs/inode.c
+++ b/fs/smbfs/inode.c
@@ -459,14 +459,8 @@ smb_show_options(struct seq_file *s, struct vfsmount *m)
 static void
 smb_unload_nls(struct smb_sb_info *server)
 {
-	if (server->remote_nls) {
-		unload_nls(server->remote_nls);
-		server->remote_nls = NULL;
-	}
-	if (server->local_nls) {
-		unload_nls(server->local_nls);
-		server->local_nls = NULL;
-	}
+	unload_nls(server->remote_nls);
+	unload_nls(server->local_nls);
 }
 
 static void
diff --git a/fs/smbfs/proc.c b/fs/smbfs/proc.c
index 9468168b9af5..71c29b6670b4 100644
--- a/fs/smbfs/proc.c
+++ b/fs/smbfs/proc.c
@@ -509,7 +509,7 @@ date_unix2dos(struct smb_sb_info *server,
 		month = 2;
 	} else {
 		nl_day = (year & 3) || day <= 59 ? day : day - 1;
-		for (month = 0; month < 12; month++)
+		for (month = 1; month < 12; month++)
 			if (day_n[month] > nl_day)
 				break;
 	}
diff --git a/fs/squashfs/super.c b/fs/squashfs/super.c
index cb5fc57e370b..6c197ef53add 100644
--- a/fs/squashfs/super.c
+++ b/fs/squashfs/super.c
@@ -44,7 +44,7 @@
 #include "squashfs.h"
 
 static struct file_system_type squashfs_fs_type;
-static struct super_operations squashfs_super_ops;
+static const struct super_operations squashfs_super_ops;
 
 static int supported_squashfs_filesystem(short major, short minor, short comp)
 {
@@ -444,7 +444,7 @@ static struct file_system_type squashfs_fs_type = {
 	.fs_flags = FS_REQUIRES_DEV
 };
 
-static struct super_operations squashfs_super_ops = {
+static const struct super_operations squashfs_super_ops = {
 	.alloc_inode = squashfs_alloc_inode,
 	.destroy_inode = squashfs_destroy_inode,
 	.statfs = squashfs_statfs,
diff --git a/fs/super.c b/fs/super.c
index b03fea8fbfb6..19eb70b374bc 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -54,7 +54,7 @@ DEFINE_SPINLOCK(sb_lock);
 static struct super_block *alloc_super(struct file_system_type *type)
 {
 	struct super_block *s = kzalloc(sizeof(struct super_block),  GFP_USER);
-	static struct super_operations default_op;
+	static const struct super_operations default_op;
 
 	if (s) {
 		if (security_sb_alloc(s)) {
@@ -465,6 +465,48 @@ rescan:
 }
 
 EXPORT_SYMBOL(get_super);
+
+/**
+ * get_active_super - get an active reference to the superblock of a device
+ * @bdev: device to get the superblock for
+ *
+ * Scans the superblock list and finds the superblock of the file system
+ * mounted on the device given.  Returns the superblock with an active
+ * reference and s_umount held exclusively or %NULL if none was found.
+ */
+struct super_block *get_active_super(struct block_device *bdev)
+{
+	struct super_block *sb;
+
+	if (!bdev)
+		return NULL;
+
+	spin_lock(&sb_lock);
+	list_for_each_entry(sb, &super_blocks, s_list) {
+		if (sb->s_bdev != bdev)
+			continue;
+
+		sb->s_count++;
+		spin_unlock(&sb_lock);
+		down_write(&sb->s_umount);
+		if (sb->s_root) {
+			spin_lock(&sb_lock);
+			if (sb->s_count > S_BIAS) {
+				atomic_inc(&sb->s_active);
+				sb->s_count--;
+				spin_unlock(&sb_lock);
+				return sb;
+			}
+			spin_unlock(&sb_lock);
+		}
+		up_write(&sb->s_umount);
+		put_super(sb);
+		yield();
+		spin_lock(&sb_lock);
+	}
+	spin_unlock(&sb_lock);
+	return NULL;
+}
  
 struct super_block * user_get_super(dev_t dev)
 {
@@ -527,11 +569,15 @@ int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
 {
 	int retval;
 	int remount_rw;
-	
+
+	if (sb->s_frozen != SB_UNFROZEN)
+		return -EBUSY;
+
 #ifdef CONFIG_BLOCK
 	if (!(flags & MS_RDONLY) && bdev_read_only(sb->s_bdev))
 		return -EACCES;
 #endif
+
 	if (flags & MS_RDONLY)
 		acct_auto_close(sb);
 	shrink_dcache_sb(sb);
@@ -743,9 +789,14 @@ int get_sb_bdev(struct file_system_type *fs_type,
 	 * will protect the lockfs code from trying to start a snapshot
 	 * while we are mounting
 	 */
-	down(&bdev->bd_mount_sem);
+	mutex_lock(&bdev->bd_fsfreeze_mutex);
+	if (bdev->bd_fsfreeze_count > 0) {
+		mutex_unlock(&bdev->bd_fsfreeze_mutex);
+		error = -EBUSY;
+		goto error_bdev;
+	}
 	s = sget(fs_type, test_bdev_super, set_bdev_super, bdev);
-	up(&bdev->bd_mount_sem);
+	mutex_unlock(&bdev->bd_fsfreeze_mutex);
 	if (IS_ERR(s))
 		goto error_s;
 
@@ -892,6 +943,16 @@ vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void
  	if (error)
  		goto out_sb;
 
+	/*
+	 * filesystems should never set s_maxbytes larger than MAX_LFS_FILESIZE
+	 * but s_maxbytes was an unsigned long long for many releases. Throw
+	 * this warning for a little while to try and catch filesystems that
+	 * violate this rule. This warning should be either removed or
+	 * converted to a BUG() in 2.6.34.
+	 */
+	WARN((mnt->mnt_sb->s_maxbytes < 0), "%s set sb->s_maxbytes to "
+		"negative value (%lld)\n", type->name, mnt->mnt_sb->s_maxbytes);
+
 	mnt->mnt_mountpoint = mnt->mnt_root;
 	mnt->mnt_parent = mnt;
 	up_write(&mnt->mnt_sb->s_umount);
diff --git a/fs/sync.c b/fs/sync.c
index c08467a5d7cb..d104591b066b 100644
--- a/fs/sync.c
+++ b/fs/sync.c
@@ -183,6 +183,7 @@ int file_fsync(struct file *filp, struct dentry *dentry, int datasync)
 		ret = err;
 	return ret;
 }
+EXPORT_SYMBOL(file_fsync);
 
 /**
  * vfs_fsync_range - helper to sync a range of data & metadata to disk
diff --git a/fs/sysfs/bin.c b/fs/sysfs/bin.c
index 2524714bece1..60c702bc10ae 100644
--- a/fs/sysfs/bin.c
+++ b/fs/sysfs/bin.c
@@ -40,7 +40,7 @@ struct bin_buffer {
 	struct mutex			mutex;
 	void				*buffer;
 	int				mmapped;
-	struct vm_operations_struct 	*vm_ops;
+	const struct vm_operations_struct *vm_ops;
 	struct file			*file;
 	struct hlist_node		list;
 };
@@ -331,7 +331,7 @@ static int bin_migrate(struct vm_area_struct *vma, const nodemask_t *from,
 }
 #endif
 
-static struct vm_operations_struct bin_vm_ops = {
+static const struct vm_operations_struct bin_vm_ops = {
 	.open		= bin_vma_open,
 	.close		= bin_vma_close,
 	.fault		= bin_fault,
diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
index 0050fc40e8c9..5fad489ce5bc 100644
--- a/fs/sysfs/dir.c
+++ b/fs/sysfs/dir.c
@@ -894,7 +894,8 @@ int sysfs_move_dir(struct kobject *kobj, struct kobject *new_parent_kobj)
 
 	mutex_lock(&sysfs_rename_mutex);
 	BUG_ON(!sd->s_parent);
-	new_parent_sd = new_parent_kobj->sd ? new_parent_kobj->sd : &sysfs_root;
+	new_parent_sd = (new_parent_kobj && new_parent_kobj->sd) ?
+		new_parent_kobj->sd : &sysfs_root;
 
 	error = 0;
 	if (sd->s_parent == new_parent_sd)
diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index 561a9c050cef..f5ea4680f15f 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -268,7 +268,7 @@ static int sysfs_get_open_dirent(struct sysfs_dirent *sd,
 	struct sysfs_open_dirent *od, *new_od = NULL;
 
  retry:
-	spin_lock(&sysfs_open_dirent_lock);
+	spin_lock_irq(&sysfs_open_dirent_lock);
 
 	if (!sd->s_attr.open && new_od) {
 		sd->s_attr.open = new_od;
@@ -281,7 +281,7 @@ static int sysfs_get_open_dirent(struct sysfs_dirent *sd,
 		list_add_tail(&buffer->list, &od->buffers);
 	}
 
-	spin_unlock(&sysfs_open_dirent_lock);
+	spin_unlock_irq(&sysfs_open_dirent_lock);
 
 	if (od) {
 		kfree(new_od);
@@ -315,8 +315,9 @@ static void sysfs_put_open_dirent(struct sysfs_dirent *sd,
 				  struct sysfs_buffer *buffer)
 {
 	struct sysfs_open_dirent *od = sd->s_attr.open;
+	unsigned long flags;
 
-	spin_lock(&sysfs_open_dirent_lock);
+	spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
 
 	list_del(&buffer->list);
 	if (atomic_dec_and_test(&od->refcnt))
@@ -324,7 +325,7 @@ static void sysfs_put_open_dirent(struct sysfs_dirent *sd,
 	else
 		od = NULL;
 
-	spin_unlock(&sysfs_open_dirent_lock);
+	spin_unlock_irqrestore(&sysfs_open_dirent_lock, flags);
 
 	kfree(od);
 }
@@ -456,8 +457,9 @@ static unsigned int sysfs_poll(struct file *filp, poll_table *wait)
 void sysfs_notify_dirent(struct sysfs_dirent *sd)
 {
 	struct sysfs_open_dirent *od;
+	unsigned long flags;
 
-	spin_lock(&sysfs_open_dirent_lock);
+	spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
 
 	od = sd->s_attr.open;
 	if (od) {
@@ -465,7 +467,7 @@ void sysfs_notify_dirent(struct sysfs_dirent *sd)
 		wake_up_interruptible(&od->poll);
 	}
 
-	spin_unlock(&sysfs_open_dirent_lock);
+	spin_unlock_irqrestore(&sysfs_open_dirent_lock, flags);
 }
 EXPORT_SYMBOL_GPL(sysfs_notify_dirent);
 
diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index ee1ce68fd98b..076ca50e9933 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -715,7 +715,7 @@ long long ubifs_get_free_space_nolock(struct ubifs_info *c)
  * ubifs_get_free_space - return amount of free space.
  * @c: UBIFS file-system description object
  *
- * This function calculates and retuns amount of free space to report to
+ * This function calculates and returns amount of free space to report to
  * user-space.
  */
 long long ubifs_get_free_space(struct ubifs_info *c)
diff --git a/fs/ubifs/commit.c b/fs/ubifs/commit.c
index f3a7945527fb..4775af401167 100644
--- a/fs/ubifs/commit.c
+++ b/fs/ubifs/commit.c
@@ -510,7 +510,7 @@ int dbg_check_old_index(struct ubifs_info *c, struct ubifs_zbranch *zroot)
 	int lnum, offs, len, err = 0, uninitialized_var(last_level), child_cnt;
 	int first = 1, iip;
 	struct ubifs_debug_info *d = c->dbg;
-	union ubifs_key lower_key, upper_key, l_key, u_key;
+	union ubifs_key uninitialized_var(lower_key), upper_key, l_key, u_key;
 	unsigned long long uninitialized_var(last_sqnum);
 	struct ubifs_idx_node *idx;
 	struct list_head list;
diff --git a/fs/ubifs/debug.c b/fs/ubifs/debug.c
index ce2cd8343618..dbc093afd946 100644
--- a/fs/ubifs/debug.c
+++ b/fs/ubifs/debug.c
@@ -210,6 +210,20 @@ const char *dbg_cstate(int cmt_state)
 	}
 }
 
+const char *dbg_jhead(int jhead)
+{
+	switch (jhead) {
+	case GCHD:
+		return "0 (GC)";
+	case BASEHD:
+		return "1 (base)";
+	case DATAHD:
+		return "2 (data)";
+	default:
+		return "unknown journal head";
+	}
+}
+
 static void dump_ch(const struct ubifs_ch *ch)
 {
 	printk(KERN_DEBUG "\tmagic          %#x\n", le32_to_cpu(ch->magic));
@@ -623,8 +637,9 @@ void dbg_dump_budg(struct ubifs_info *c)
 	/* If we are in R/O mode, journal heads do not exist */
 	if (c->jheads)
 		for (i = 0; i < c->jhead_cnt; i++)
-			printk(KERN_DEBUG "\tjhead %d\t LEB %d\n",
-			       c->jheads[i].wbuf.jhead, c->jheads[i].wbuf.lnum);
+			printk(KERN_DEBUG "\tjhead %s\t LEB %d\n",
+			       dbg_jhead(c->jheads[i].wbuf.jhead),
+			       c->jheads[i].wbuf.lnum);
 	for (rb = rb_first(&c->buds); rb; rb = rb_next(rb)) {
 		bud = rb_entry(rb, struct ubifs_bud, rb);
 		printk(KERN_DEBUG "\tbud LEB %d\n", bud->lnum);
@@ -648,9 +663,90 @@ void dbg_dump_budg(struct ubifs_info *c)
 
 void dbg_dump_lprop(const struct ubifs_info *c, const struct ubifs_lprops *lp)
 {
-	printk(KERN_DEBUG "LEB %d lprops: free %d, dirty %d (used %d), "
-	       "flags %#x\n", lp->lnum, lp->free, lp->dirty,
-	       c->leb_size - lp->free - lp->dirty, lp->flags);
+	int i, spc, dark = 0, dead = 0;
+	struct rb_node *rb;
+	struct ubifs_bud *bud;
+
+	spc = lp->free + lp->dirty;
+	if (spc < c->dead_wm)
+		dead = spc;
+	else
+		dark = ubifs_calc_dark(c, spc);
+
+	if (lp->flags & LPROPS_INDEX)
+		printk(KERN_DEBUG "LEB %-7d free %-8d dirty %-8d used %-8d "
+		       "free + dirty %-8d flags %#x (", lp->lnum, lp->free,
+		       lp->dirty, c->leb_size - spc, spc, lp->flags);
+	else
+		printk(KERN_DEBUG "LEB %-7d free %-8d dirty %-8d used %-8d "
+		       "free + dirty %-8d dark %-4d dead %-4d nodes fit %-3d "
+		       "flags %#-4x (", lp->lnum, lp->free, lp->dirty,
+		       c->leb_size - spc, spc, dark, dead,
+		       (int)(spc / UBIFS_MAX_NODE_SZ), lp->flags);
+
+	if (lp->flags & LPROPS_TAKEN) {
+		if (lp->flags & LPROPS_INDEX)
+			printk(KERN_CONT "index, taken");
+		else
+			printk(KERN_CONT "taken");
+	} else {
+		const char *s;
+
+		if (lp->flags & LPROPS_INDEX) {
+			switch (lp->flags & LPROPS_CAT_MASK) {
+			case LPROPS_DIRTY_IDX:
+				s = "dirty index";
+				break;
+			case LPROPS_FRDI_IDX:
+				s = "freeable index";
+				break;
+			default:
+				s = "index";
+			}
+		} else {
+			switch (lp->flags & LPROPS_CAT_MASK) {
+			case LPROPS_UNCAT:
+				s = "not categorized";
+				break;
+			case LPROPS_DIRTY:
+				s = "dirty";
+				break;
+			case LPROPS_FREE:
+				s = "free";
+				break;
+			case LPROPS_EMPTY:
+				s = "empty";
+				break;
+			case LPROPS_FREEABLE:
+				s = "freeable";
+				break;
+			default:
+				s = NULL;
+				break;
+			}
+		}
+		printk(KERN_CONT "%s", s);
+	}
+
+	for (rb = rb_first((struct rb_root *)&c->buds); rb; rb = rb_next(rb)) {
+		bud = rb_entry(rb, struct ubifs_bud, rb);
+		if (bud->lnum == lp->lnum) {
+			int head = 0;
+			for (i = 0; i < c->jhead_cnt; i++) {
+				if (lp->lnum == c->jheads[i].wbuf.lnum) {
+					printk(KERN_CONT ", jhead %s",
+					       dbg_jhead(i));
+					head = 1;
+				}
+			}
+			if (!head)
+				printk(KERN_CONT ", bud of jhead %s",
+				       dbg_jhead(bud->jhead));
+		}
+	}
+	if (lp->lnum == c->gc_lnum)
+		printk(KERN_CONT ", GC LEB");
+	printk(KERN_CONT ")\n");
 }
 
 void dbg_dump_lprops(struct ubifs_info *c)
@@ -724,7 +820,7 @@ void dbg_dump_leb(const struct ubifs_info *c, int lnum)
 
 	printk(KERN_DEBUG "(pid %d) start dumping LEB %d\n",
 	       current->pid, lnum);
-	sleb = ubifs_scan(c, lnum, 0, c->dbg->buf);
+	sleb = ubifs_scan(c, lnum, 0, c->dbg->buf, 0);
 	if (IS_ERR(sleb)) {
 		ubifs_err("scan error %d", (int)PTR_ERR(sleb));
 		return;
@@ -909,8 +1005,10 @@ out:
 	ubifs_msg("saved lprops statistics dump");
 	dbg_dump_lstats(&d->saved_lst);
 	ubifs_get_lp_stats(c, &lst);
+
 	ubifs_msg("current lprops statistics dump");
-	dbg_dump_lstats(&d->saved_lst);
+	dbg_dump_lstats(&lst);
+
 	spin_lock(&c->space_lock);
 	dbg_dump_budg(c);
 	spin_unlock(&c->space_lock);
diff --git a/fs/ubifs/debug.h b/fs/ubifs/debug.h
index c1cd73b2e06e..29d960101ea6 100644
--- a/fs/ubifs/debug.h
+++ b/fs/ubifs/debug.h
@@ -271,6 +271,7 @@ void ubifs_debugging_exit(struct ubifs_info *c);
 /* Dump functions */
 const char *dbg_ntype(int type);
 const char *dbg_cstate(int cmt_state);
+const char *dbg_jhead(int jhead);
 const char *dbg_get_key_dump(const struct ubifs_info *c,
 			     const union ubifs_key *key);
 void dbg_dump_inode(const struct ubifs_info *c, const struct inode *inode);
@@ -321,6 +322,8 @@ void dbg_check_heap(struct ubifs_info *c, struct ubifs_lpt_heap *heap, int cat,
 int dbg_check_lprops(struct ubifs_info *c);
 int dbg_check_lpt_nodes(struct ubifs_info *c, struct ubifs_cnode *cnode,
 			int row, int col);
+int dbg_check_inode_size(struct ubifs_info *c, const struct inode *inode,
+			 loff_t size);
 
 /* Force the use of in-the-gaps method for testing */
 
@@ -425,6 +428,7 @@ void dbg_debugfs_exit_fs(struct ubifs_info *c);
 
 #define dbg_ntype(type)                        ""
 #define dbg_cstate(cmt_state)                  ""
+#define dbg_jhead(jhead)                       ""
 #define dbg_get_key_dump(c, key)               ({})
 #define dbg_dump_inode(c, inode)               ({})
 #define dbg_dump_node(c, node)                 ({})
@@ -460,6 +464,7 @@ void dbg_debugfs_exit_fs(struct ubifs_info *c);
 #define dbg_check_heap(c, heap, cat, add_pos)      ({})
 #define dbg_check_lprops(c)                        0
 #define dbg_check_lpt_nodes(c, cnode, row, col)    0
+#define dbg_check_inode_size(c, inode, size)       0
 #define dbg_force_in_the_gaps_enabled              0
 #define dbg_force_in_the_gaps()                    0
 #define dbg_failure_mode                           0
diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 6d34dc7e33e1..1009adc8d602 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -21,34 +21,32 @@
  */
 
 /*
- * This file implements VFS file and inode operations of regular files, device
+ * This file implements VFS file and inode operations for regular files, device
  * nodes and symlinks as well as address space operations.
  *
- * UBIFS uses 2 page flags: PG_private and PG_checked. PG_private is set if the
- * page is dirty and is used for budgeting purposes - dirty pages should not be
- * budgeted. The PG_checked flag is set if full budgeting is required for the
- * page e.g., when it corresponds to a file hole or it is just beyond the file
- * size. The budgeting is done in 'ubifs_write_begin()', because it is OK to
- * fail in this function, and the budget is released in 'ubifs_write_end()'. So
- * the PG_private and PG_checked flags carry the information about how the page
- * was budgeted, to make it possible to release the budget properly.
+ * UBIFS uses 2 page flags: @PG_private and @PG_checked. @PG_private is set if
+ * the page is dirty and is used for optimization purposes - dirty pages are
+ * not budgeted so the flag shows that 'ubifs_write_end()' should not release
+ * the budget for this page. The @PG_checked flag is set if full budgeting is
+ * required for the page e.g., when it corresponds to a file hole or it is
+ * beyond the file size. The budgeting is done in 'ubifs_write_begin()', because
+ * it is OK to fail in this function, and the budget is released in
+ * 'ubifs_write_end()'. So the @PG_private and @PG_checked flags carry
+ * information about how the page was budgeted, to make it possible to release
+ * the budget properly.
  *
- * A thing to keep in mind: inode's 'i_mutex' is locked in most VFS operations
- * we implement. However, this is not true for '->writepage()', which might be
- * called with 'i_mutex' unlocked. For example, when pdflush is performing
- * write-back, it calls 'writepage()' with unlocked 'i_mutex', although the
- * inode has 'I_LOCK' flag in this case. At "normal" work-paths 'i_mutex' is
- * locked in '->writepage', e.g. in "sys_write -> alloc_pages -> direct reclaim
- * path'. So, in '->writepage()' we are only guaranteed that the page is
- * locked.
+ * A thing to keep in mind: inode @i_mutex is locked in most VFS operations we
+ * implement. However, this is not true for 'ubifs_writepage()', which may be
+ * called with @i_mutex unlocked. For example, when pdflush is doing background
+ * write-back, it calls 'ubifs_writepage()' with unlocked @i_mutex. At "normal"
+ * work-paths the @i_mutex is locked in 'ubifs_writepage()', e.g. in the
+ * "sys_write -> alloc_pages -> direct reclaim path". So, in 'ubifs_writepage()'
+ * we are only guaranteed that the page is locked.
  *
- * Similarly, 'i_mutex' does not have to be locked in readpage(), e.g.,
- * readahead path does not have it locked ("sys_read -> generic_file_aio_read
- * -> ondemand_readahead -> readpage"). In case of readahead, 'I_LOCK' flag is
- * not set as well. However, UBIFS disables readahead.
- *
- * This, for example means that there might be 2 concurrent '->writepage()'
- * calls for the same inode, but different inode dirty pages.
+ * Similarly, @i_mutex is not always locked in 'ubifs_readpage()', e.g., the
+ * read-ahead path does not lock it ("sys_read -> generic_file_aio_read ->
+ * ondemand_readahead -> readpage"). In case of readahead, @I_LOCK flag is not
+ * set as well. However, UBIFS disables readahead.
  */
 
 #include "ubifs.h"
@@ -449,9 +447,9 @@ static int ubifs_write_begin(struct file *file, struct address_space *mapping,
 			/*
 			 * We change whole page so no need to load it. But we
 			 * have to set the @PG_checked flag to make the further
-			 * code the page is new. This might be not true, but it
-			 * is better to budget more that to read the page from
-			 * the media.
+			 * code know that the page is new. This might be not
+			 * true, but it is better to budget more than to read
+			 * the page from the media.
 			 */
 			SetPageChecked(page);
 			skipped_read = 1;
@@ -497,8 +495,8 @@ static int ubifs_write_begin(struct file *file, struct address_space *mapping,
 	}
 
 	/*
-	 * Whee, we aquired budgeting quickly - without involving
-	 * garbage-collection, committing or forceing write-back. We return
+	 * Whee, we acquired budgeting quickly - without involving
+	 * garbage-collection, committing or forcing write-back. We return
 	 * with @ui->ui_mutex locked if we are appending pages, and unlocked
 	 * otherwise. This is an optimization (slightly hacky though).
 	 */
@@ -562,7 +560,7 @@ static int ubifs_write_end(struct file *file, struct address_space *mapping,
 
 		/*
 		 * Return 0 to force VFS to repeat the whole operation, or the
-		 * error code if 'do_readpage()' failes.
+		 * error code if 'do_readpage()' fails.
 		 */
 		copied = do_readpage(page);
 		goto out;
@@ -1175,11 +1173,11 @@ static int do_truncation(struct ubifs_info *c, struct inode *inode,
 	ui->ui_size = inode->i_size;
 	/* Truncation changes inode [mc]time */
 	inode->i_mtime = inode->i_ctime = ubifs_current_time(inode);
-	/* The other attributes may be changed at the same time as well */
+	/* Other attributes may be changed at the same time as well */
 	do_attr_changes(inode, attr);
-
 	err = ubifs_jnl_truncate(c, inode, old_size, new_size);
 	mutex_unlock(&ui->ui_mutex);
+
 out_budg:
 	if (budgeted)
 		ubifs_release_budget(c, &req);
@@ -1536,7 +1534,7 @@ out_unlock:
 	return err;
 }
 
-static struct vm_operations_struct ubifs_file_vm_ops = {
+static const struct vm_operations_struct ubifs_file_vm_ops = {
 	.fault        = filemap_fault,
 	.page_mkwrite = ubifs_vm_page_mkwrite,
 };
diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
index f0f5f15d384e..618c2701d3a7 100644
--- a/fs/ubifs/gc.c
+++ b/fs/ubifs/gc.c
@@ -529,7 +529,7 @@ int ubifs_garbage_collect_leb(struct ubifs_info *c, struct ubifs_lprops *lp)
 	 * We scan the entire LEB even though we only really need to scan up to
 	 * (c->leb_size - lp->free).
 	 */
-	sleb = ubifs_scan(c, lnum, 0, c->sbuf);
+	sleb = ubifs_scan(c, lnum, 0, c->sbuf, 0);
 	if (IS_ERR(sleb))
 		return PTR_ERR(sleb);
 
diff --git a/fs/ubifs/io.c b/fs/ubifs/io.c
index 762a7d6cec73..e589fedaf1ef 100644
--- a/fs/ubifs/io.c
+++ b/fs/ubifs/io.c
@@ -297,7 +297,7 @@ static enum hrtimer_restart wbuf_timer_callback_nolock(struct hrtimer *timer)
 {
 	struct ubifs_wbuf *wbuf = container_of(timer, struct ubifs_wbuf, timer);
 
-	dbg_io("jhead %d", wbuf->jhead);
+	dbg_io("jhead %s", dbg_jhead(wbuf->jhead));
 	wbuf->need_sync = 1;
 	wbuf->c->need_wbuf_sync = 1;
 	ubifs_wake_up_bgt(wbuf->c);
@@ -314,7 +314,8 @@ static void new_wbuf_timer_nolock(struct ubifs_wbuf *wbuf)
 
 	if (wbuf->no_timer)
 		return;
-	dbg_io("set timer for jhead %d, %llu-%llu millisecs", wbuf->jhead,
+	dbg_io("set timer for jhead %s, %llu-%llu millisecs",
+	       dbg_jhead(wbuf->jhead),
 	       div_u64(ktime_to_ns(wbuf->softlimit), USEC_PER_SEC),
 	       div_u64(ktime_to_ns(wbuf->softlimit) + wbuf->delta,
 		       USEC_PER_SEC));
@@ -351,8 +352,8 @@ int ubifs_wbuf_sync_nolock(struct ubifs_wbuf *wbuf)
 		/* Write-buffer is empty or not seeked */
 		return 0;
 
-	dbg_io("LEB %d:%d, %d bytes, jhead %d",
-	       wbuf->lnum, wbuf->offs, wbuf->used, wbuf->jhead);
+	dbg_io("LEB %d:%d, %d bytes, jhead %s",
+	       wbuf->lnum, wbuf->offs, wbuf->used, dbg_jhead(wbuf->jhead));
 	ubifs_assert(!(c->vfs_sb->s_flags & MS_RDONLY));
 	ubifs_assert(!(wbuf->avail & 7));
 	ubifs_assert(wbuf->offs + c->min_io_size <= c->leb_size);
@@ -401,7 +402,7 @@ int ubifs_wbuf_seek_nolock(struct ubifs_wbuf *wbuf, int lnum, int offs,
 {
 	const struct ubifs_info *c = wbuf->c;
 
-	dbg_io("LEB %d:%d, jhead %d", lnum, offs, wbuf->jhead);
+	dbg_io("LEB %d:%d, jhead %s", lnum, offs, dbg_jhead(wbuf->jhead));
 	ubifs_assert(lnum >= 0 && lnum < c->leb_cnt);
 	ubifs_assert(offs >= 0 && offs <= c->leb_size);
 	ubifs_assert(offs % c->min_io_size == 0 && !(offs & 7));
@@ -508,9 +509,9 @@ int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len)
 	struct ubifs_info *c = wbuf->c;
 	int err, written, n, aligned_len = ALIGN(len, 8), offs;
 
-	dbg_io("%d bytes (%s) to jhead %d wbuf at LEB %d:%d", len,
-	       dbg_ntype(((struct ubifs_ch *)buf)->node_type), wbuf->jhead,
-	       wbuf->lnum, wbuf->offs + wbuf->used);
+	dbg_io("%d bytes (%s) to jhead %s wbuf at LEB %d:%d", len,
+	       dbg_ntype(((struct ubifs_ch *)buf)->node_type),
+	       dbg_jhead(wbuf->jhead), wbuf->lnum, wbuf->offs + wbuf->used);
 	ubifs_assert(len > 0 && wbuf->lnum >= 0 && wbuf->lnum < c->leb_cnt);
 	ubifs_assert(wbuf->offs >= 0 && wbuf->offs % c->min_io_size == 0);
 	ubifs_assert(!(wbuf->offs & 7) && wbuf->offs <= c->leb_size);
@@ -535,8 +536,8 @@ int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len)
 		memcpy(wbuf->buf + wbuf->used, buf, len);
 
 		if (aligned_len == wbuf->avail) {
-			dbg_io("flush jhead %d wbuf to LEB %d:%d",
-			       wbuf->jhead, wbuf->lnum, wbuf->offs);
+			dbg_io("flush jhead %s wbuf to LEB %d:%d",
+			       dbg_jhead(wbuf->jhead), wbuf->lnum, wbuf->offs);
 			err = ubi_leb_write(c->ubi, wbuf->lnum, wbuf->buf,
 					    wbuf->offs, c->min_io_size,
 					    wbuf->dtype);
@@ -564,8 +565,8 @@ int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len)
 	 * minimal I/O unit. We have to fill and flush write-buffer and switch
 	 * to the next min. I/O unit.
 	 */
-	dbg_io("flush jhead %d wbuf to LEB %d:%d",
-	       wbuf->jhead, wbuf->lnum, wbuf->offs);
+	dbg_io("flush jhead %s wbuf to LEB %d:%d",
+	       dbg_jhead(wbuf->jhead), wbuf->lnum, wbuf->offs);
 	memcpy(wbuf->buf + wbuf->used, buf, wbuf->avail);
 	err = ubi_leb_write(c->ubi, wbuf->lnum, wbuf->buf, wbuf->offs,
 			    c->min_io_size, wbuf->dtype);
@@ -698,8 +699,8 @@ int ubifs_read_node_wbuf(struct ubifs_wbuf *wbuf, void *buf, int type, int len,
 	int err, rlen, overlap;
 	struct ubifs_ch *ch = buf;
 
-	dbg_io("LEB %d:%d, %s, length %d, jhead %d", lnum, offs,
-	       dbg_ntype(type), len, wbuf->jhead);
+	dbg_io("LEB %d:%d, %s, length %d, jhead %s", lnum, offs,
+	       dbg_ntype(type), len, dbg_jhead(wbuf->jhead));
 	ubifs_assert(wbuf && lnum >= 0 && lnum < c->leb_cnt && offs >= 0);
 	ubifs_assert(!(offs & 7) && offs < c->leb_size);
 	ubifs_assert(type >= 0 && type < UBIFS_NODE_TYPES_CNT);
diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 64b5f3a309f5..d321baeca68d 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -158,7 +158,7 @@ again:
 	 * some. But the write-buffer mutex has to be unlocked because
 	 * GC also takes it.
 	 */
-	dbg_jnl("no free space  jhead %d, run GC", jhead);
+	dbg_jnl("no free space in jhead %s, run GC", dbg_jhead(jhead));
 	mutex_unlock(&wbuf->io_mutex);
 
 	lnum = ubifs_garbage_collect(c, 0);
@@ -173,7 +173,8 @@ again:
 		 * because we dropped @wbuf->io_mutex, so try once
 		 * again.
 		 */
-		dbg_jnl("GC couldn't make a free LEB for jhead %d", jhead);
+		dbg_jnl("GC couldn't make a free LEB for jhead %s",
+			dbg_jhead(jhead));
 		if (retries++ < 2) {
 			dbg_jnl("retry (%d)", retries);
 			goto again;
@@ -184,7 +185,7 @@ again:
 	}
 
 	mutex_lock_nested(&wbuf->io_mutex, wbuf->jhead);
-	dbg_jnl("got LEB %d for jhead %d", lnum, jhead);
+	dbg_jnl("got LEB %d for jhead %s", lnum, dbg_jhead(jhead));
 	avail = c->leb_size - wbuf->offs - wbuf->used;
 
 	if (wbuf->lnum != -1 && avail >= len) {
@@ -255,7 +256,8 @@ static int write_node(struct ubifs_info *c, int jhead, void *node, int len,
 	*lnum = c->jheads[jhead].wbuf.lnum;
 	*offs = c->jheads[jhead].wbuf.offs + c->jheads[jhead].wbuf.used;
 
-	dbg_jnl("jhead %d, LEB %d:%d, len %d", jhead, *lnum, *offs, len);
+	dbg_jnl("jhead %s, LEB %d:%d, len %d",
+		dbg_jhead(jhead), *lnum, *offs, len);
 	ubifs_prepare_node(c, node, len, 0);
 
 	return ubifs_wbuf_write_nolock(wbuf, node, len);
@@ -285,7 +287,8 @@ static int write_head(struct ubifs_info *c, int jhead, void *buf, int len,
 
 	*lnum = c->jheads[jhead].wbuf.lnum;
 	*offs = c->jheads[jhead].wbuf.offs + c->jheads[jhead].wbuf.used;
-	dbg_jnl("jhead %d, LEB %d:%d, len %d", jhead, *lnum, *offs, len);
+	dbg_jnl("jhead %s, LEB %d:%d, len %d",
+		dbg_jhead(jhead), *lnum, *offs, len);
 
 	err = ubifs_wbuf_write_nolock(wbuf, buf, len);
 	if (err)
diff --git a/fs/ubifs/key.h b/fs/ubifs/key.h
index 5fa27ea031ba..0f530c684f0b 100644
--- a/fs/ubifs/key.h
+++ b/fs/ubifs/key.h
@@ -229,23 +229,6 @@ static inline void xent_key_init(const struct ubifs_info *c,
 }
 
 /**
- * xent_key_init_hash - initialize extended attribute entry key without
- *                      re-calculating hash function.
- * @c: UBIFS file-system description object
- * @key: key to initialize
- * @inum: host inode number
- * @hash: extended attribute entry name hash
- */
-static inline void xent_key_init_hash(const struct ubifs_info *c,
-				      union ubifs_key *key, ino_t inum,
-				      uint32_t hash)
-{
-	ubifs_assert(!(hash & ~UBIFS_S_KEY_HASH_MASK));
-	key->u32[0] = inum;
-	key->u32[1] = hash | (UBIFS_XENT_KEY << UBIFS_S_KEY_HASH_BITS);
-}
-
-/**
  * xent_key_init_flash - initialize on-flash extended attribute entry key.
  * @c: UBIFS file-system description object
  * @k: key to initialize
@@ -295,22 +278,15 @@ static inline void data_key_init(const struct ubifs_info *c,
 }
 
 /**
- * data_key_init_flash - initialize on-flash data key.
+ * highest_data_key - get the highest possible data key for an inode.
  * @c: UBIFS file-system description object
- * @k: key to initialize
+ * @key: key to initialize
  * @inum: inode number
- * @block: block number
  */
-static inline void data_key_init_flash(const struct ubifs_info *c, void *k,
-				       ino_t inum, unsigned int block)
+static inline void highest_data_key(const struct ubifs_info *c,
+				   union ubifs_key *key, ino_t inum)
 {
-	union ubifs_key *key = k;
-
-	ubifs_assert(!(block & ~UBIFS_S_KEY_BLOCK_MASK));
-	key->j32[0] = cpu_to_le32(inum);
-	key->j32[1] = cpu_to_le32(block |
-				  (UBIFS_DATA_KEY << UBIFS_S_KEY_BLOCK_BITS));
-	memset(k + 8, 0, UBIFS_MAX_KEY_LEN - 8);
+	data_key_init(c, key, inum, UBIFS_S_KEY_BLOCK_MASK);
 }
 
 /**
@@ -554,4 +530,5 @@ static inline unsigned long long key_max_inode_size(const struct ubifs_info *c)
 		return 0;
 	}
 }
+
 #endif /* !__UBIFS_KEY_H__ */
diff --git a/fs/ubifs/log.c b/fs/ubifs/log.c
index 56e33772a1ee..c345e125f42c 100644
--- a/fs/ubifs/log.c
+++ b/fs/ubifs/log.c
@@ -169,8 +169,8 @@ void ubifs_add_bud(struct ubifs_info *c, struct ubifs_bud *bud)
 	 */
 	c->bud_bytes += c->leb_size - bud->start;
 
-	dbg_log("LEB %d:%d, jhead %d, bud_bytes %lld", bud->lnum,
-		bud->start, bud->jhead, c->bud_bytes);
+	dbg_log("LEB %d:%d, jhead %s, bud_bytes %lld", bud->lnum,
+		bud->start, dbg_jhead(bud->jhead), c->bud_bytes);
 	spin_unlock(&c->buds_lock);
 }
 
@@ -355,16 +355,16 @@ static void remove_buds(struct ubifs_info *c)
 			 * heads (non-closed buds).
 			 */
 			c->cmt_bud_bytes += wbuf->offs - bud->start;
-			dbg_log("preserve %d:%d, jhead %d, bud bytes %d, "
+			dbg_log("preserve %d:%d, jhead %s, bud bytes %d, "
 				"cmt_bud_bytes %lld", bud->lnum, bud->start,
-				bud->jhead, wbuf->offs - bud->start,
+				dbg_jhead(bud->jhead), wbuf->offs - bud->start,
 				c->cmt_bud_bytes);
 			bud->start = wbuf->offs;
 		} else {
 			c->cmt_bud_bytes += c->leb_size - bud->start;
-			dbg_log("remove %d:%d, jhead %d, bud bytes %d, "
+			dbg_log("remove %d:%d, jhead %s, bud bytes %d, "
 				"cmt_bud_bytes %lld", bud->lnum, bud->start,
-				bud->jhead, c->leb_size - bud->start,
+				dbg_jhead(bud->jhead), c->leb_size - bud->start,
 				c->cmt_bud_bytes);
 			rb_erase(p1, &c->buds);
 			/*
@@ -429,7 +429,8 @@ int ubifs_log_start_commit(struct ubifs_info *c, int *ltail_lnum)
 		if (lnum == -1 || offs == c->leb_size)
 			continue;
 
-		dbg_log("add ref to LEB %d:%d for jhead %d", lnum, offs, i);
+		dbg_log("add ref to LEB %d:%d for jhead %s",
+			lnum, offs, dbg_jhead(i));
 		ref = buf + len;
 		ref->ch.node_type = UBIFS_REF_NODE;
 		ref->lnum = cpu_to_le32(lnum);
@@ -695,7 +696,7 @@ int ubifs_consolidate_log(struct ubifs_info *c)
 	lnum = c->ltail_lnum;
 	write_lnum = lnum;
 	while (1) {
-		sleb = ubifs_scan(c, lnum, 0, c->sbuf);
+		sleb = ubifs_scan(c, lnum, 0, c->sbuf, 0);
 		if (IS_ERR(sleb)) {
 			err = PTR_ERR(sleb);
 			goto out_free;
diff --git a/fs/ubifs/lprops.c b/fs/ubifs/lprops.c
index 4cdd284dea56..4d4ca388889b 100644
--- a/fs/ubifs/lprops.c
+++ b/fs/ubifs/lprops.c
@@ -281,7 +281,7 @@ void ubifs_add_to_cat(struct ubifs_info *c, struct ubifs_lprops *lprops,
 	case LPROPS_FREE:
 		if (add_to_lpt_heap(c, lprops, cat))
 			break;
-		/* No more room on heap so make it uncategorized */
+		/* No more room on heap so make it un-categorized */
 		cat = LPROPS_UNCAT;
 		/* Fall through */
 	case LPROPS_UNCAT:
@@ -375,8 +375,8 @@ void ubifs_replace_cat(struct ubifs_info *c, struct ubifs_lprops *old_lprops,
  * @lprops: LEB properties
  *
  * A LEB may have fallen off of the bottom of a heap, and ended up as
- * uncategorized even though it has enough space for us now. If that is the case
- * this function will put the LEB back onto a heap.
+ * un-categorized even though it has enough space for us now. If that is the
+ * case this function will put the LEB back onto a heap.
  */
 void ubifs_ensure_cat(struct ubifs_info *c, struct ubifs_lprops *lprops)
 {
@@ -436,10 +436,10 @@ int ubifs_categorize_lprops(const struct ubifs_info *c,
 /**
  * change_category - change LEB properties category.
  * @c: UBIFS file-system description object
- * @lprops: LEB properties to recategorize
+ * @lprops: LEB properties to re-categorize
  *
  * LEB properties are categorized to enable fast find operations. When the LEB
- * properties change they must be recategorized.
+ * properties change they must be re-categorized.
  */
 static void change_category(struct ubifs_info *c, struct ubifs_lprops *lprops)
 {
@@ -461,21 +461,18 @@ static void change_category(struct ubifs_info *c, struct ubifs_lprops *lprops)
 }
 
 /**
- * calc_dark - calculate LEB dark space size.
+ * ubifs_calc_dark - calculate LEB dark space size.
  * @c: the UBIFS file-system description object
  * @spc: amount of free and dirty space in the LEB
  *
- * This function calculates amount of dark space in an LEB which has @spc bytes
- * of free and dirty space. Returns the calculations result.
+ * This function calculates and returns amount of dark space in an LEB which
+ * has @spc bytes of free and dirty space.
  *
- * Dark space is the space which is not always usable - it depends on which
- * nodes are written in which order. E.g., if an LEB has only 512 free bytes,
- * it is dark space, because it cannot fit a large data node. So UBIFS cannot
- * count on this LEB and treat these 512 bytes as usable because it is not true
- * if, for example, only big chunks of uncompressible data will be written to
- * the FS.
+ * UBIFS is trying to account the space which might not be usable, and this
+ * space is called "dark space". For example, if an LEB has only %512 free
+ * bytes, it is dark space, because it cannot fit a large data node.
  */
-static int calc_dark(struct ubifs_info *c, int spc)
+int ubifs_calc_dark(const struct ubifs_info *c, int spc)
 {
 	ubifs_assert(!(spc & 7));
 
@@ -518,7 +515,7 @@ static int is_lprops_dirty(struct ubifs_info *c, struct ubifs_lprops *lprops)
  * @free: new free space amount
  * @dirty: new dirty space amount
  * @flags: new flags
- * @idx_gc_cnt: change to the count of idx_gc list
+ * @idx_gc_cnt: change to the count of @idx_gc list
  *
  * This function changes LEB properties (@free, @dirty or @flag). However, the
  * property which has the %LPROPS_NC value is not changed. Returns a pointer to
@@ -535,7 +532,7 @@ const struct ubifs_lprops *ubifs_change_lp(struct ubifs_info *c,
 {
 	/*
 	 * This is the only function that is allowed to change lprops, so we
-	 * discard the const qualifier.
+	 * discard the "const" qualifier.
 	 */
 	struct ubifs_lprops *lprops = (struct ubifs_lprops *)lp;
 
@@ -575,7 +572,7 @@ const struct ubifs_lprops *ubifs_change_lp(struct ubifs_info *c,
 		if (old_spc < c->dead_wm)
 			c->lst.total_dead -= old_spc;
 		else
-			c->lst.total_dark -= calc_dark(c, old_spc);
+			c->lst.total_dark -= ubifs_calc_dark(c, old_spc);
 
 		c->lst.total_used -= c->leb_size - old_spc;
 	}
@@ -616,7 +613,7 @@ const struct ubifs_lprops *ubifs_change_lp(struct ubifs_info *c,
 		if (new_spc < c->dead_wm)
 			c->lst.total_dead += new_spc;
 		else
-			c->lst.total_dark += calc_dark(c, new_spc);
+			c->lst.total_dark += ubifs_calc_dark(c, new_spc);
 
 		c->lst.total_used += c->leb_size - new_spc;
 	}
@@ -1096,7 +1093,7 @@ static int scan_check_cb(struct ubifs_info *c,
 		}
 	}
 
-	sleb = ubifs_scan(c, lnum, 0, c->dbg->buf);
+	sleb = ubifs_scan(c, lnum, 0, c->dbg->buf, 0);
 	if (IS_ERR(sleb)) {
 		/*
 		 * After an unclean unmount, empty and freeable LEBs
@@ -1107,7 +1104,7 @@ static int scan_check_cb(struct ubifs_info *c,
 				  "- continuing checking");
 			lst->empty_lebs += 1;
 			lst->total_free += c->leb_size;
-			lst->total_dark += calc_dark(c, c->leb_size);
+			lst->total_dark += ubifs_calc_dark(c, c->leb_size);
 			return LPT_SCAN_CONTINUE;
 		}
 
@@ -1117,7 +1114,7 @@ static int scan_check_cb(struct ubifs_info *c,
 				  "- continuing checking");
 			lst->total_free  += lp->free;
 			lst->total_dirty += lp->dirty;
-			lst->total_dark  +=  calc_dark(c, c->leb_size);
+			lst->total_dark  +=  ubifs_calc_dark(c, c->leb_size);
 			return LPT_SCAN_CONTINUE;
 		}
 		data->err = PTR_ERR(sleb);
@@ -1235,7 +1232,7 @@ static int scan_check_cb(struct ubifs_info *c,
 		if (spc < c->dead_wm)
 			lst->total_dead += spc;
 		else
-			lst->total_dark += calc_dark(c, spc);
+			lst->total_dark += ubifs_calc_dark(c, spc);
 	}
 
 	ubifs_scan_destroy(sleb);
diff --git a/fs/ubifs/master.c b/fs/ubifs/master.c
index a88f33801b98..28beaeedadc0 100644
--- a/fs/ubifs/master.c
+++ b/fs/ubifs/master.c
@@ -29,7 +29,8 @@
  * @c: UBIFS file-system description object
  *
  * This function scans the master node LEBs and search for the latest master
- * node. Returns zero in case of success and a negative error code in case of
+ * node. Returns zero in case of success, %-EUCLEAN if there master area is
+ * corrupted and requires recovery, and a negative error code in case of
  * failure.
  */
 static int scan_for_master(struct ubifs_info *c)
@@ -40,7 +41,7 @@ static int scan_for_master(struct ubifs_info *c)
 
 	lnum = UBIFS_MST_LNUM;
 
-	sleb = ubifs_scan(c, lnum, 0, c->sbuf);
+	sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1);
 	if (IS_ERR(sleb))
 		return PTR_ERR(sleb);
 	nodes_cnt = sleb->nodes_cnt;
@@ -48,7 +49,7 @@ static int scan_for_master(struct ubifs_info *c)
 		snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node,
 				  list);
 		if (snod->type != UBIFS_MST_NODE)
-			goto out;
+			goto out_dump;
 		memcpy(c->mst_node, snod->node, snod->len);
 		offs = snod->offs;
 	}
@@ -56,7 +57,7 @@ static int scan_for_master(struct ubifs_info *c)
 
 	lnum += 1;
 
-	sleb = ubifs_scan(c, lnum, 0, c->sbuf);
+	sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1);
 	if (IS_ERR(sleb))
 		return PTR_ERR(sleb);
 	if (sleb->nodes_cnt != nodes_cnt)
@@ -65,7 +66,7 @@ static int scan_for_master(struct ubifs_info *c)
 		goto out;
 	snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node, list);
 	if (snod->type != UBIFS_MST_NODE)
-		goto out;
+		goto out_dump;
 	if (snod->offs != offs)
 		goto out;
 	if (memcmp((void *)c->mst_node + UBIFS_CH_SZ,
@@ -78,6 +79,12 @@ static int scan_for_master(struct ubifs_info *c)
 
 out:
 	ubifs_scan_destroy(sleb);
+	return -EUCLEAN;
+
+out_dump:
+	ubifs_err("unexpected node type %d master LEB %d:%d",
+		  snod->type, lnum, snod->offs);
+	ubifs_scan_destroy(sleb);
 	return -EINVAL;
 }
 
@@ -256,7 +263,8 @@ int ubifs_read_master(struct ubifs_info *c)
 
 	err = scan_for_master(c);
 	if (err) {
-		err = ubifs_recover_master_node(c);
+		if (err == -EUCLEAN)
+			err = ubifs_recover_master_node(c);
 		if (err)
 			/*
 			 * Note, we do not free 'c->mst_node' here because the
diff --git a/fs/ubifs/orphan.c b/fs/ubifs/orphan.c
index 152a7b34a141..82009c74b6a3 100644
--- a/fs/ubifs/orphan.c
+++ b/fs/ubifs/orphan.c
@@ -670,9 +670,10 @@ static int kill_orphans(struct ubifs_info *c)
 		struct ubifs_scan_leb *sleb;
 
 		dbg_rcvry("LEB %d", lnum);
-		sleb = ubifs_scan(c, lnum, 0, c->sbuf);
+		sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1);
 		if (IS_ERR(sleb)) {
-			sleb = ubifs_recover_leb(c, lnum, 0, c->sbuf, 0);
+			if (PTR_ERR(sleb) == -EUCLEAN)
+				sleb = ubifs_recover_leb(c, lnum, 0, c->sbuf, 0);
 			if (IS_ERR(sleb)) {
 				err = PTR_ERR(sleb);
 				break;
@@ -899,7 +900,7 @@ static int dbg_scan_orphans(struct ubifs_info *c, struct check_info *ci)
 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
 		struct ubifs_scan_leb *sleb;
 
-		sleb = ubifs_scan(c, lnum, 0, c->dbg->buf);
+		sleb = ubifs_scan(c, lnum, 0, c->dbg->buf, 0);
 		if (IS_ERR(sleb)) {
 			err = PTR_ERR(sleb);
 			break;
diff --git a/fs/ubifs/recovery.c b/fs/ubifs/recovery.c
index e5f6cf8a1155..f94ddf7efba0 100644
--- a/fs/ubifs/recovery.c
+++ b/fs/ubifs/recovery.c
@@ -286,7 +286,7 @@ int ubifs_recover_master_node(struct ubifs_info *c)
 		mst = mst2;
 	}
 
-	dbg_rcvry("recovered master node from LEB %d",
+	ubifs_msg("recovered master node from LEB %d",
 		  (mst == mst1 ? UBIFS_MST_LNUM : UBIFS_MST_LNUM + 1));
 
 	memcpy(c->mst_node, mst, UBIFS_MST_NODE_SZ);
@@ -790,7 +790,7 @@ struct ubifs_scan_leb *ubifs_recover_log_leb(struct ubifs_info *c, int lnum,
 		 * We can only recover at the end of the log, so check that the
 		 * next log LEB is empty or out of date.
 		 */
-		sleb = ubifs_scan(c, next_lnum, 0, sbuf);
+		sleb = ubifs_scan(c, next_lnum, 0, sbuf, 0);
 		if (IS_ERR(sleb))
 			return sleb;
 		if (sleb->nodes_cnt) {
diff --git a/fs/ubifs/replay.c b/fs/ubifs/replay.c
index 2970500f32df..5c2d6d759a3e 100644
--- a/fs/ubifs/replay.c
+++ b/fs/ubifs/replay.c
@@ -506,7 +506,7 @@ static int replay_bud(struct ubifs_info *c, int lnum, int offs, int jhead,
 	if (c->need_recovery)
 		sleb = ubifs_recover_leb(c, lnum, offs, c->sbuf, jhead != GCHD);
 	else
-		sleb = ubifs_scan(c, lnum, offs, c->sbuf);
+		sleb = ubifs_scan(c, lnum, offs, c->sbuf, 0);
 	if (IS_ERR(sleb))
 		return PTR_ERR(sleb);
 
@@ -836,8 +836,8 @@ static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf)
 	const struct ubifs_cs_node *node;
 
 	dbg_mnt("replay log LEB %d:%d", lnum, offs);
-	sleb = ubifs_scan(c, lnum, offs, sbuf);
-	if (IS_ERR(sleb) ) {
+	sleb = ubifs_scan(c, lnum, offs, sbuf, c->need_recovery);
+	if (IS_ERR(sleb)) {
 		if (PTR_ERR(sleb) != -EUCLEAN || !c->need_recovery)
 			return PTR_ERR(sleb);
 		sleb = ubifs_recover_log_leb(c, lnum, offs, sbuf);
diff --git a/fs/ubifs/scan.c b/fs/ubifs/scan.c
index 892ebfee4fe5..96c525384191 100644
--- a/fs/ubifs/scan.c
+++ b/fs/ubifs/scan.c
@@ -108,10 +108,9 @@ int ubifs_scan_a_node(const struct ubifs_info *c, void *buf, int len, int lnum,
 
 		/* Make the node pads to 8-byte boundary */
 		if ((node_len + pad_len) & 7) {
-			if (!quiet) {
+			if (!quiet)
 				dbg_err("bad padding length %d - %d",
 					offs, offs + node_len + pad_len);
-			}
 			return SCANNED_A_BAD_PAD_NODE;
 		}
 
@@ -253,15 +252,19 @@ void ubifs_scanned_corruption(const struct ubifs_info *c, int lnum, int offs,
  * @c: UBIFS file-system description object
  * @lnum: logical eraseblock number
  * @offs: offset to start at (usually zero)
- * @sbuf: scan buffer (must be c->leb_size)
+ * @sbuf: scan buffer (must be of @c->leb_size bytes in size)
+ * @quiet: print no messages
  *
  * This function scans LEB number @lnum and returns complete information about
  * its contents. Returns the scaned information in case of success and,
  * %-EUCLEAN if the LEB neads recovery, and other negative error codes in case
  * of failure.
+ *
+ * If @quiet is non-zero, this function does not print large and scary
+ * error messages and flash dumps in case of errors.
  */
 struct ubifs_scan_leb *ubifs_scan(const struct ubifs_info *c, int lnum,
-				  int offs, void *sbuf)
+				  int offs, void *sbuf, int quiet)
 {
 	void *buf = sbuf + offs;
 	int err, len = c->leb_size - offs;
@@ -280,7 +283,7 @@ struct ubifs_scan_leb *ubifs_scan(const struct ubifs_info *c, int lnum,
 
 		cond_resched();
 
-		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 0);
+		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, quiet);
 		if (ret > 0) {
 			/* Padding bytes or a valid padding node */
 			offs += ret;
@@ -320,7 +323,9 @@ struct ubifs_scan_leb *ubifs_scan(const struct ubifs_info *c, int lnum,
 	}
 
 	if (offs % c->min_io_size) {
-		ubifs_err("empty space starts at non-aligned offset %d", offs);
+		if (!quiet)
+			ubifs_err("empty space starts at non-aligned offset %d",
+				  offs);
 		goto corrupted;;
 	}
 
@@ -331,18 +336,25 @@ struct ubifs_scan_leb *ubifs_scan(const struct ubifs_info *c, int lnum,
 			break;
 	for (; len; offs++, buf++, len--)
 		if (*(uint8_t *)buf != 0xff) {
-			ubifs_err("corrupt empty space at LEB %d:%d",
-				  lnum, offs);
+			if (!quiet)
+				ubifs_err("corrupt empty space at LEB %d:%d",
+					  lnum, offs);
 			goto corrupted;
 		}
 
 	return sleb;
 
 corrupted:
-	ubifs_scanned_corruption(c, lnum, offs, buf);
+	if (!quiet) {
+		ubifs_scanned_corruption(c, lnum, offs, buf);
+		ubifs_err("LEB %d scanning failed", lnum);
+	}
 	err = -EUCLEAN;
+	ubifs_scan_destroy(sleb);
+	return ERR_PTR(err);
+
 error:
-	ubifs_err("LEB %d scanning failed", lnum);
+	ubifs_err("LEB %d scanning failed, error %d", lnum, err);
 	ubifs_scan_destroy(sleb);
 	return ERR_PTR(err);
 }
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index c4af069df1ad..333e181ee987 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -36,7 +36,6 @@
 #include <linux/mount.h>
 #include <linux/math64.h>
 #include <linux/writeback.h>
-#include <linux/smp_lock.h>
 #include "ubifs.h"
 
 /*
@@ -318,6 +317,8 @@ static int ubifs_write_inode(struct inode *inode, int wait)
 		if (err)
 			ubifs_err("can't write inode %lu, error %d",
 				  inode->i_ino, err);
+		else
+			err = dbg_check_inode_size(c, inode, ui->ui_size);
 	}
 
 	ui->dirty = 0;
@@ -448,17 +449,6 @@ static int ubifs_sync_fs(struct super_block *sb, int wait)
 		return 0;
 
 	/*
-	 * VFS calls '->sync_fs()' before synchronizing all dirty inodes and
-	 * pages, so synchronize them first, then commit the journal. Strictly
-	 * speaking, it is not necessary to commit the journal here,
-	 * synchronizing write-buffers would be enough. But committing makes
-	 * UBIFS free space predictions much more accurate, so we want to let
-	 * the user be able to get more accurate results of 'statfs()' after
-	 * they synchronize the file system.
-	 */
-	sync_inodes_sb(sb);
-
-	/*
 	 * Synchronize write buffers, because 'ubifs_run_commit()' does not
 	 * do this if it waits for an already running commit.
 	 */
@@ -468,6 +458,13 @@ static int ubifs_sync_fs(struct super_block *sb, int wait)
 			return err;
 	}
 
+	/*
+	 * Strictly speaking, it is not necessary to commit the journal here,
+	 * synchronizing write-buffers would be enough. But committing makes
+	 * UBIFS free space predictions much more accurate, so we want to let
+	 * the user be able to get more accurate results of 'statfs()' after
+	 * they synchronize the file system.
+	 */
 	err = ubifs_run_commit(c);
 	if (err)
 		return err;
@@ -1720,8 +1717,6 @@ static void ubifs_put_super(struct super_block *sb)
 	ubifs_msg("un-mount UBI device %d, volume %d", c->vi.ubi_num,
 		  c->vi.vol_id);
 
-	lock_kernel();
-
 	/*
 	 * The following asserts are only valid if there has not been a failure
 	 * of the media. For example, there will be dirty inodes if we failed
@@ -1786,8 +1781,6 @@ static void ubifs_put_super(struct super_block *sb)
 	ubi_close_volume(c->ubi);
 	mutex_unlock(&c->umount_mutex);
 	kfree(c);
-
-	unlock_kernel();
 }
 
 static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
@@ -1803,22 +1796,17 @@ static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
 		return err;
 	}
 
-	lock_kernel();
 	if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
 		if (c->ro_media) {
 			ubifs_msg("cannot re-mount due to prior errors");
-			unlock_kernel();
 			return -EROFS;
 		}
 		err = ubifs_remount_rw(c);
-		if (err) {
-			unlock_kernel();
+		if (err)
 			return err;
-		}
 	} else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
 		if (c->ro_media) {
 			ubifs_msg("cannot re-mount due to prior errors");
-			unlock_kernel();
 			return -EROFS;
 		}
 		ubifs_remount_ro(c);
@@ -1833,7 +1821,6 @@ static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
 	}
 
 	ubifs_assert(c->lst.taken_empty_lebs > 0);
-	unlock_kernel();
 	return 0;
 }
 
diff --git a/fs/ubifs/tnc.c b/fs/ubifs/tnc.c
index f249f7b0d656..e5b1a7d00fa0 100644
--- a/fs/ubifs/tnc.c
+++ b/fs/ubifs/tnc.c
@@ -1159,8 +1159,8 @@ static struct ubifs_znode *dirty_cow_bottom_up(struct ubifs_info *c,
  *   o exact match, i.e. the found zero-level znode contains key @key, then %1
  *     is returned and slot number of the matched branch is stored in @n;
  *   o not exact match, which means that zero-level znode does not contain
- *     @key, then %0 is returned and slot number of the closed branch is stored
- *     in  @n;
+ *     @key, then %0 is returned and slot number of the closest branch is stored
+ *     in @n;
  *   o @key is so small that it is even less than the lowest key of the
  *     leftmost zero-level node, then %0 is returned and %0 is stored in @n.
  *
@@ -1433,7 +1433,7 @@ static int maybe_leb_gced(struct ubifs_info *c, int lnum, int gc_seq1)
  * @lnum: LEB number is returned here
  * @offs: offset is returned here
  *
- * This function look up and reads node with key @key. The caller has to make
+ * This function looks up and reads node with key @key. The caller has to make
  * sure the @node buffer is large enough to fit the node. Returns zero in case
  * of success, %-ENOENT if the node was not found, and a negative error code in
  * case of failure. The node location can be returned in @lnum and @offs.
@@ -3268,3 +3268,73 @@ out_unlock:
 	mutex_unlock(&c->tnc_mutex);
 	return err;
 }
+
+#ifdef CONFIG_UBIFS_FS_DEBUG
+
+/**
+ * dbg_check_inode_size - check if inode size is correct.
+ * @c: UBIFS file-system description object
+ * @inum: inode number
+ * @size: inode size
+ *
+ * This function makes sure that the inode size (@size) is correct and it does
+ * not have any pages beyond @size. Returns zero if the inode is OK, %-EINVAL
+ * if it has a data page beyond @size, and other negative error code in case of
+ * other errors.
+ */
+int dbg_check_inode_size(struct ubifs_info *c, const struct inode *inode,
+			 loff_t size)
+{
+	int err, n;
+	union ubifs_key from_key, to_key, *key;
+	struct ubifs_znode *znode;
+	unsigned int block;
+
+	if (!S_ISREG(inode->i_mode))
+		return 0;
+	if (!(ubifs_chk_flags & UBIFS_CHK_GEN))
+		return 0;
+
+	block = (size + UBIFS_BLOCK_SIZE - 1) >> UBIFS_BLOCK_SHIFT;
+	data_key_init(c, &from_key, inode->i_ino, block);
+	highest_data_key(c, &to_key, inode->i_ino);
+
+	mutex_lock(&c->tnc_mutex);
+	err = ubifs_lookup_level0(c, &from_key, &znode, &n);
+	if (err < 0)
+		goto out_unlock;
+
+	if (err) {
+		err = -EINVAL;
+		key = &from_key;
+		goto out_dump;
+	}
+
+	err = tnc_next(c, &znode, &n);
+	if (err == -ENOENT) {
+		err = 0;
+		goto out_unlock;
+	}
+	if (err < 0)
+		goto out_unlock;
+
+	ubifs_assert(err == 0);
+	key = &znode->zbranch[n].key;
+	if (!key_in_range(c, key, &from_key, &to_key))
+		goto out_unlock;
+
+out_dump:
+	block = key_block(c, key);
+	ubifs_err("inode %lu has size %lld, but there are data at offset %lld "
+		  "(data key %s)", (unsigned long)inode->i_ino, size,
+		  ((loff_t)block) << UBIFS_BLOCK_SHIFT, DBGKEY(key));
+	dbg_dump_inode(c, inode);
+	dbg_dump_stack();
+	err = -EINVAL;
+
+out_unlock:
+	mutex_unlock(&c->tnc_mutex);
+	return err;
+}
+
+#endif /* CONFIG_UBIFS_FS_DEBUG */
diff --git a/fs/ubifs/tnc_commit.c b/fs/ubifs/tnc_commit.c
index fde8d127c768..53288e5d604e 100644
--- a/fs/ubifs/tnc_commit.c
+++ b/fs/ubifs/tnc_commit.c
@@ -245,7 +245,7 @@ static int layout_leb_in_gaps(struct ubifs_info *c, int *p)
 	 * it is more comprehensive and less efficient than is needed for this
 	 * purpose.
 	 */
-	sleb = ubifs_scan(c, lnum, 0, c->ileb_buf);
+	sleb = ubifs_scan(c, lnum, 0, c->ileb_buf, 0);
 	c->ileb_len = 0;
 	if (IS_ERR(sleb))
 		return PTR_ERR(sleb);
diff --git a/fs/ubifs/ubifs-media.h b/fs/ubifs/ubifs-media.h
index 3eee07e0c495..191ca7863fe7 100644
--- a/fs/ubifs/ubifs-media.h
+++ b/fs/ubifs/ubifs-media.h
@@ -135,6 +135,13 @@
 /* The key is always at the same position in all keyed nodes */
 #define UBIFS_KEY_OFFSET offsetof(struct ubifs_ino_node, key)
 
+/* Garbage collector journal head number */
+#define UBIFS_GC_HEAD   0
+/* Base journal head number */
+#define UBIFS_BASE_HEAD 1
+/* Data journal head number */
+#define UBIFS_DATA_HEAD 2
+
 /*
  * LEB Properties Tree node types.
  *
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index a29349094422..b2d976366a46 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -105,12 +105,10 @@
 /* Number of non-data journal heads */
 #define NONDATA_JHEADS_CNT 2
 
-/* Garbage collector head */
-#define GCHD   0
-/* Base journal head number */
-#define BASEHD 1
-/* First "general purpose" journal head */
-#define DATAHD 2
+/* Shorter names for journal head numbers for internal usage */
+#define GCHD   UBIFS_GC_HEAD
+#define BASEHD UBIFS_BASE_HEAD
+#define DATAHD UBIFS_DATA_HEAD
 
 /* 'No change' value for 'ubifs_change_lp()' */
 #define LPROPS_NC 0x80000001
@@ -1451,7 +1449,7 @@ int ubifs_sync_wbufs_by_inode(struct ubifs_info *c, struct inode *inode);
 
 /* scan.c */
 struct ubifs_scan_leb *ubifs_scan(const struct ubifs_info *c, int lnum,
-				  int offs, void *sbuf);
+				  int offs, void *sbuf, int quiet);
 void ubifs_scan_destroy(struct ubifs_scan_leb *sleb);
 int ubifs_scan_a_node(const struct ubifs_info *c, void *buf, int len, int lnum,
 		      int offs, int quiet);
@@ -1676,6 +1674,7 @@ const struct ubifs_lprops *ubifs_fast_find_free(struct ubifs_info *c);
 const struct ubifs_lprops *ubifs_fast_find_empty(struct ubifs_info *c);
 const struct ubifs_lprops *ubifs_fast_find_freeable(struct ubifs_info *c);
 const struct ubifs_lprops *ubifs_fast_find_frdi_idx(struct ubifs_info *c);
+int ubifs_calc_dark(const struct ubifs_info *c, int spc);
 
 /* file.c */
 int ubifs_fsync(struct file *file, struct dentry *dentry, int datasync);
diff --git a/fs/ubifs/xattr.c b/fs/ubifs/xattr.c
index adafcf556531..195830f47569 100644
--- a/fs/ubifs/xattr.c
+++ b/fs/ubifs/xattr.c
@@ -78,9 +78,9 @@ enum {
 	SECURITY_XATTR,
 };
 
-static struct inode_operations none_inode_operations;
-static struct address_space_operations none_address_operations;
-static struct file_operations none_file_operations;
+static const struct inode_operations none_inode_operations;
+static const struct address_space_operations none_address_operations;
+static const struct file_operations none_file_operations;
 
 /**
  * create_xattr - create an extended attribute.
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index d5e5559e31db..c2e30eea74dc 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -186,19 +186,37 @@ xfs_destroy_ioend(
 }
 
 /*
+ * If the end of the current ioend is beyond the current EOF,
+ * return the new EOF value, otherwise zero.
+ */
+STATIC xfs_fsize_t
+xfs_ioend_new_eof(
+	xfs_ioend_t		*ioend)
+{
+	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
+	xfs_fsize_t		isize;
+	xfs_fsize_t		bsize;
+
+	bsize = ioend->io_offset + ioend->io_size;
+	isize = MAX(ip->i_size, ip->i_new_size);
+	isize = MIN(isize, bsize);
+	return isize > ip->i_d.di_size ? isize : 0;
+}
+
+/*
  * Update on-disk file size now that data has been written to disk.
  * The current in-memory file size is i_size.  If a write is beyond
  * eof i_new_size will be the intended file size until i_size is
  * updated.  If this write does not extend all the way to the valid
  * file size then restrict this update to the end of the write.
  */
+
 STATIC void
 xfs_setfilesize(
 	xfs_ioend_t		*ioend)
 {
 	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
 	xfs_fsize_t		isize;
-	xfs_fsize_t		bsize;
 
 	ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
 	ASSERT(ioend->io_type != IOMAP_READ);
@@ -206,16 +224,10 @@ xfs_setfilesize(
 	if (unlikely(ioend->io_error))
 		return;
 
-	bsize = ioend->io_offset + ioend->io_size;
-
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-
-	isize = MAX(ip->i_size, ip->i_new_size);
-	isize = MIN(isize, bsize);
-
-	if (ip->i_d.di_size < isize) {
+	isize = xfs_ioend_new_eof(ioend);
+	if (isize) {
 		ip->i_d.di_size = isize;
-		ip->i_update_core = 1;
 		xfs_mark_inode_dirty_sync(ip);
 	}
 
@@ -404,10 +416,16 @@ xfs_submit_ioend_bio(
 	struct bio	*bio)
 {
 	atomic_inc(&ioend->io_remaining);
-
 	bio->bi_private = ioend;
 	bio->bi_end_io = xfs_end_bio;
 
+	/*
+	 * If the I/O is beyond EOF we mark the inode dirty immediately
+	 * but don't update the inode size until I/O completion.
+	 */
+	if (xfs_ioend_new_eof(ioend))
+		xfs_mark_inode_dirty_sync(XFS_I(ioend->io_inode));
+
 	submit_bio(WRITE, bio);
 	ASSERT(!bio_flagged(bio, BIO_EOPNOTSUPP));
 	bio_put(bio);
@@ -1635,4 +1653,5 @@ const struct address_space_operations xfs_address_space_operations = {
 	.direct_IO		= xfs_vm_direct_IO,
 	.migratepage		= buffer_migrate_page,
 	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
 };
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 988d8f87bc0f..eff61e2732af 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -42,7 +42,7 @@
 
 #include <linux/dcache.h>
 
-static struct vm_operations_struct xfs_file_vm_ops;
+static const struct vm_operations_struct xfs_file_vm_ops;
 
 STATIC ssize_t
 xfs_file_aio_read(
@@ -176,14 +176,7 @@ xfs_file_fsync(
 	struct dentry		*dentry,
 	int			datasync)
 {
-	struct inode		*inode = dentry->d_inode;
-	struct xfs_inode	*ip = XFS_I(inode);
-	int			error;
-
-	/* capture size updates in I/O completion before writing the inode. */
-	error = filemap_fdatawait(inode->i_mapping);
-	if (error)
-		return error;
+	struct xfs_inode	*ip = XFS_I(dentry->d_inode);
 
 	xfs_iflags_clear(ip, XFS_ITRUNCATED);
 	return -xfs_fsync(ip);
@@ -280,7 +273,7 @@ const struct file_operations xfs_dir_file_operations = {
 	.fsync		= xfs_file_fsync,
 };
 
-static struct vm_operations_struct xfs_file_vm_ops = {
+static const struct vm_operations_struct xfs_file_vm_ops = {
 	.fault		= filemap_fault,
 	.page_mkwrite	= xfs_vm_page_mkwrite,
 };
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index da0159d99f82..cd42ef78f6b5 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -57,19 +57,22 @@
 #include <linux/fiemap.h>
 
 /*
- * Bring the atime in the XFS inode uptodate.
- * Used before logging the inode to disk or when the Linux inode goes away.
+ * Bring the timestamps in the XFS inode uptodate.
+ *
+ * Used before writing the inode to disk.
  */
 void
-xfs_synchronize_atime(
+xfs_synchronize_times(
 	xfs_inode_t	*ip)
 {
 	struct inode	*inode = VFS_I(ip);
 
-	if (!(inode->i_state & I_CLEAR)) {
-		ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
-		ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
-	}
+	ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
+	ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
+	ip->i_d.di_ctime.t_sec = (__int32_t)inode->i_ctime.tv_sec;
+	ip->i_d.di_ctime.t_nsec = (__int32_t)inode->i_ctime.tv_nsec;
+	ip->i_d.di_mtime.t_sec = (__int32_t)inode->i_mtime.tv_sec;
+	ip->i_d.di_mtime.t_nsec = (__int32_t)inode->i_mtime.tv_nsec;
 }
 
 /*
@@ -106,32 +109,20 @@ xfs_ichgtime(
 	if ((flags & XFS_ICHGTIME_MOD) &&
 	    !timespec_equal(&inode->i_mtime, &tv)) {
 		inode->i_mtime = tv;
-		ip->i_d.di_mtime.t_sec = (__int32_t)tv.tv_sec;
-		ip->i_d.di_mtime.t_nsec = (__int32_t)tv.tv_nsec;
 		sync_it = 1;
 	}
 	if ((flags & XFS_ICHGTIME_CHG) &&
 	    !timespec_equal(&inode->i_ctime, &tv)) {
 		inode->i_ctime = tv;
-		ip->i_d.di_ctime.t_sec = (__int32_t)tv.tv_sec;
-		ip->i_d.di_ctime.t_nsec = (__int32_t)tv.tv_nsec;
 		sync_it = 1;
 	}
 
 	/*
-	 * We update the i_update_core field _after_ changing
-	 * the timestamps in order to coordinate properly with
-	 * xfs_iflush() so that we don't lose timestamp updates.
-	 * This keeps us from having to hold the inode lock
-	 * while doing this.  We use the SYNCHRONIZE macro to
-	 * ensure that the compiler does not reorder the update
-	 * of i_update_core above the timestamp updates above.
+	 * Update complete - now make sure everyone knows that the inode
+	 * is dirty.
 	 */
-	if (sync_it) {
-		SYNCHRONIZE();
-		ip->i_update_core = 1;
+	if (sync_it)
 		xfs_mark_inode_dirty_sync(ip);
-	}
 }
 
 /*
@@ -506,10 +497,8 @@ xfs_vn_getattr(
 	stat->gid = ip->i_d.di_gid;
 	stat->ino = ip->i_ino;
 	stat->atime = inode->i_atime;
-	stat->mtime.tv_sec = ip->i_d.di_mtime.t_sec;
-	stat->mtime.tv_nsec = ip->i_d.di_mtime.t_nsec;
-	stat->ctime.tv_sec = ip->i_d.di_ctime.t_sec;
-	stat->ctime.tv_nsec = ip->i_d.di_ctime.t_nsec;
+	stat->mtime = inode->i_mtime;
+	stat->ctime = inode->i_ctime;
 	stat->blocks =
 		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);
 
diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c
index 49e4a6aea73c..072050f8d346 100644
--- a/fs/xfs/linux-2.6/xfs_lrw.c
+++ b/fs/xfs/linux-2.6/xfs_lrw.c
@@ -667,7 +667,7 @@ start:
 		xip->i_new_size = new_size;
 
 	if (likely(!(ioflags & IO_INVIS)))
-		xfs_ichgtime(xip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+		file_update_time(file);
 
 	/*
 	 * If the offset is beyond the size of the file, we have a couple
diff --git a/fs/xfs/linux-2.6/xfs_quotaops.c b/fs/xfs/linux-2.6/xfs_quotaops.c
index cb6e2cca214f..9e41f91aa269 100644
--- a/fs/xfs/linux-2.6/xfs_quotaops.c
+++ b/fs/xfs/linux-2.6/xfs_quotaops.c
@@ -150,7 +150,7 @@ xfs_fs_set_xquota(
 	return -xfs_qm_scall_setqlim(mp, id, xfs_quota_type(type), fdq);
 }
 
-struct quotactl_ops xfs_quotactl_operations = {
+const struct quotactl_ops xfs_quotactl_operations = {
 	.quota_sync		= xfs_fs_quota_sync,
 	.get_xstate		= xfs_fs_get_xstate,
 	.set_xstate		= xfs_fs_set_xstate,
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 5d7c60ac77b4..18a4b8e11df2 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -67,7 +67,7 @@
 #include <linux/freezer.h>
 #include <linux/parser.h>
 
-static struct super_operations xfs_super_operations;
+static const struct super_operations xfs_super_operations;
 static kmem_zone_t *xfs_ioend_zone;
 mempool_t *xfs_ioend_pool;
 
@@ -977,6 +977,28 @@ xfs_fs_inode_init_once(
 }
 
 /*
+ * Dirty the XFS inode when mark_inode_dirty_sync() is called so that
+ * we catch unlogged VFS level updates to the inode. Care must be taken
+ * here - the transaction code calls mark_inode_dirty_sync() to mark the
+ * VFS inode dirty in a transaction and clears the i_update_core field;
+ * it must clear the field after calling mark_inode_dirty_sync() to
+ * correctly indicate that the dirty state has been propagated into the
+ * inode log item.
+ *
+ * We need the barrier() to maintain correct ordering between unlogged
+ * updates and the transaction commit code that clears the i_update_core
+ * field. This requires all updates to be completed before marking the
+ * inode dirty.
+ */
+STATIC void
+xfs_fs_dirty_inode(
+	struct inode	*inode)
+{
+	barrier();
+	XFS_I(inode)->i_update_core = 1;
+}
+
+/*
  * Attempt to flush the inode, this will actually fail
  * if the inode is pinned, but we dirty the inode again
  * at the point when it is unpinned after a log write,
@@ -1126,7 +1148,7 @@ xfs_fs_put_super(
 }
 
 STATIC int
-xfs_fs_sync_super(
+xfs_fs_sync_fs(
 	struct super_block	*sb,
 	int			wait)
 {
@@ -1134,23 +1156,23 @@ xfs_fs_sync_super(
 	int			error;
 
 	/*
-	 * Treat a sync operation like a freeze.  This is to work
-	 * around a race in sync_inodes() which works in two phases
-	 * - an asynchronous flush, which can write out an inode
-	 * without waiting for file size updates to complete, and a
-	 * synchronous flush, which wont do anything because the
-	 * async flush removed the inode's dirty flag.  Also
-	 * sync_inodes() will not see any files that just have
-	 * outstanding transactions to be flushed because we don't
-	 * dirty the Linux inode until after the transaction I/O
-	 * completes.
+	 * Not much we can do for the first async pass.  Writing out the
+	 * superblock would be counter-productive as we are going to redirty
+	 * when writing out other data and metadata (and writing out a single
+	 * block is quite fast anyway).
+	 *
+	 * Try to asynchronously kick off quota syncing at least.
 	 */
-	if (wait || unlikely(sb->s_frozen == SB_FREEZE_WRITE))
-		error = xfs_quiesce_data(mp);
-	else
-		error = xfs_sync_fsdata(mp, 0);
+	if (!wait) {
+		xfs_qm_sync(mp, SYNC_TRYLOCK);
+		return 0;
+	}
+
+	error = xfs_quiesce_data(mp);
+	if (error)
+		return -error;
 
-	if (unlikely(laptop_mode)) {
+	if (laptop_mode) {
 		int	prev_sync_seq = mp->m_sync_seq;
 
 		/*
@@ -1169,7 +1191,7 @@ xfs_fs_sync_super(
 				mp->m_sync_seq != prev_sync_seq);
 	}
 
-	return -error;
+	return 0;
 }
 
 STATIC int
@@ -1536,13 +1558,14 @@ xfs_fs_get_sb(
 			   mnt);
 }
 
-static struct super_operations xfs_super_operations = {
+static const struct super_operations xfs_super_operations = {
 	.alloc_inode		= xfs_fs_alloc_inode,
 	.destroy_inode		= xfs_fs_destroy_inode,
+	.dirty_inode		= xfs_fs_dirty_inode,
 	.write_inode		= xfs_fs_write_inode,
 	.clear_inode		= xfs_fs_clear_inode,
 	.put_super		= xfs_fs_put_super,
-	.sync_fs		= xfs_fs_sync_super,
+	.sync_fs		= xfs_fs_sync_fs,
 	.freeze_fs		= xfs_fs_freeze,
 	.statfs			= xfs_fs_statfs,
 	.remount_fs		= xfs_fs_remount,
diff --git a/fs/xfs/linux-2.6/xfs_super.h b/fs/xfs/linux-2.6/xfs_super.h
index 5a2ea3a21781..18175ebd58ed 100644
--- a/fs/xfs/linux-2.6/xfs_super.h
+++ b/fs/xfs/linux-2.6/xfs_super.h
@@ -93,7 +93,7 @@ extern void xfs_blkdev_issue_flush(struct xfs_buftarg *);
 
 extern const struct export_operations xfs_export_operations;
 extern struct xattr_handler *xfs_xattr_handlers[];
-extern struct quotactl_ops xfs_quotactl_operations;
+extern const struct quotactl_ops xfs_quotactl_operations;
 
 #define XFS_M(sb)		((struct xfs_mount *)((sb)->s_fs_info))
 
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 320be6aea492..961df0a22c78 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -309,11 +309,15 @@ xfs_sync_attr(
 STATIC int
 xfs_commit_dummy_trans(
 	struct xfs_mount	*mp,
-	uint			log_flags)
+	uint			flags)
 {
 	struct xfs_inode	*ip = mp->m_rootip;
 	struct xfs_trans	*tp;
 	int			error;
+	int			log_flags = XFS_LOG_FORCE;
+
+	if (flags & SYNC_WAIT)
+		log_flags |= XFS_LOG_SYNC;
 
 	/*
 	 * Put a dummy transaction in the log to tell recovery
@@ -331,13 +335,12 @@ xfs_commit_dummy_trans(
 	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
 	xfs_trans_ihold(tp, ip);
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	/* XXX(hch): ignoring the error here.. */
 	error = xfs_trans_commit(tp, 0);
-
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 
+	/* the log force ensures this transaction is pushed to disk */
 	xfs_log_force(mp, 0, log_flags);
-	return 0;
+	return error;
 }
 
 int
@@ -385,7 +388,20 @@ xfs_sync_fsdata(
 	else
 		XFS_BUF_ASYNC(bp);
 
-	return xfs_bwrite(mp, bp);
+	error = xfs_bwrite(mp, bp);
+	if (error)
+		return error;
+
+	/*
+	 * If this is a data integrity sync make sure all pending buffers
+	 * are flushed out for the log coverage check below.
+	 */
+	if (flags & SYNC_WAIT)
+		xfs_flush_buftarg(mp->m_ddev_targp, 1);
+
+	if (xfs_log_need_covered(mp))
+		error = xfs_commit_dummy_trans(mp, flags);
+	return error;
 
  out_brelse:
 	xfs_buf_relse(bp);
@@ -419,14 +435,16 @@ xfs_quiesce_data(
 	/* push non-blocking */
 	xfs_sync_data(mp, 0);
 	xfs_qm_sync(mp, SYNC_TRYLOCK);
-	xfs_filestream_flush(mp);
 
-	/* push and block */
+	/* push and block till complete */
 	xfs_sync_data(mp, SYNC_WAIT);
 	xfs_qm_sync(mp, SYNC_WAIT);
 
+	/* drop inode references pinned by filestreams */
+	xfs_filestream_flush(mp);
+
 	/* write superblock and hoover up shutdown errors */
-	error = xfs_sync_fsdata(mp, 0);
+	error = xfs_sync_fsdata(mp, SYNC_WAIT);
 
 	/* flush data-only devices */
 	if (mp->m_rtdev_targp)
@@ -570,8 +588,6 @@ xfs_sync_worker(
 		/* dgc: errors ignored here */
 		error = xfs_qm_sync(mp, SYNC_TRYLOCK);
 		error = xfs_sync_fsdata(mp, SYNC_TRYLOCK);
-		if (xfs_log_need_covered(mp))
-			error = xfs_commit_dummy_trans(mp, XFS_LOG_FORCE);
 	}
 	mp->m_sync_seq++;
 	wake_up(&mp->m_wait_single_sync_task);
diff --git a/fs/xfs/linux-2.6/xfs_sysctl.c b/fs/xfs/linux-2.6/xfs_sysctl.c
index 916c0ffb6083..c5bc67c4e3bb 100644
--- a/fs/xfs/linux-2.6/xfs_sysctl.c
+++ b/fs/xfs/linux-2.6/xfs_sysctl.c
@@ -26,7 +26,6 @@ STATIC int
 xfs_stats_clear_proc_handler(
 	ctl_table	*ctl,
 	int		write,
-	struct file	*filp,
 	void		__user *buffer,
 	size_t		*lenp,
 	loff_t		*ppos)
@@ -34,7 +33,7 @@ xfs_stats_clear_proc_handler(
 	int		c, ret, *valp = ctl->data;
 	__uint32_t	vn_active;
 
-	ret = proc_dointvec_minmax(ctl, write, filp, buffer, lenp, ppos);
+	ret = proc_dointvec_minmax(ctl, write, buffer, lenp, ppos);
 
 	if (!ret && write && *valp) {
 		printk("XFS Clearing xfsstats\n");
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 7465f9ee125f..ab89a7e94a0f 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -206,10 +206,10 @@ xfs_swap_extents(
 	 * process that the file was not changed out from
 	 * under it.
 	 */
-	if ((sbp->bs_ctime.tv_sec != ip->i_d.di_ctime.t_sec) ||
-	    (sbp->bs_ctime.tv_nsec != ip->i_d.di_ctime.t_nsec) ||
-	    (sbp->bs_mtime.tv_sec != ip->i_d.di_mtime.t_sec) ||
-	    (sbp->bs_mtime.tv_nsec != ip->i_d.di_mtime.t_nsec)) {
+	if ((sbp->bs_ctime.tv_sec != VFS_I(ip)->i_ctime.tv_sec) ||
+	    (sbp->bs_ctime.tv_nsec != VFS_I(ip)->i_ctime.tv_nsec) ||
+	    (sbp->bs_mtime.tv_sec != VFS_I(ip)->i_mtime.tv_sec) ||
+	    (sbp->bs_mtime.tv_nsec != VFS_I(ip)->i_mtime.tv_nsec)) {
 		error = XFS_ERROR(EBUSY);
 		goto out_unlock;
 	}
diff --git a/fs/xfs/xfs_dir2_leaf.c b/fs/xfs/xfs_dir2_leaf.c
index fa913e459442..41ad537c49e9 100644
--- a/fs/xfs/xfs_dir2_leaf.c
+++ b/fs/xfs/xfs_dir2_leaf.c
@@ -854,6 +854,7 @@ xfs_dir2_leaf_getdents(
 			 */
 			ra_want = howmany(bufsize + mp->m_dirblksize,
 					  mp->m_sb.sb_blocksize) - 1;
+			ASSERT(ra_want >= 0);
 
 			/*
 			 * If we don't have as many as we want, and we haven't
@@ -1088,7 +1089,8 @@ xfs_dir2_leaf_getdents(
 		 */
 		ptr += length;
 		curoff += length;
-		bufsize -= length;
+		/* bufsize may have just been a guess; don't go negative */
+		bufsize = bufsize > length ? bufsize - length : 0;
 	}
 
 	/*
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
index c4ea51b55dce..f52ac276277e 100644
--- a/fs/xfs/xfs_fs.h
+++ b/fs/xfs/xfs_fs.h
@@ -117,7 +117,7 @@ struct getbmapx {
 #define BMV_IF_VALID	\
 	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|BMV_IF_DELALLOC)
 
-/*	bmv_oflags values - returned for for each non-header segment */
+/*	bmv_oflags values - returned for each non-header segment */
 #define BMV_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
 #define BMV_OF_DELALLOC		0x2	/* segment = delayed allocation */
 #define BMV_OF_LAST		0x4	/* segment is the last in the file */
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index c1dc7ef5a1d8..b92a4fa2a0a1 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -3068,9 +3068,9 @@ xfs_iflush_int(
 	SYNCHRONIZE();
 
 	/*
-	 * Make sure to get the latest atime from the Linux inode.
+	 * Make sure to get the latest timestamps from the Linux inode.
 	 */
-	xfs_synchronize_atime(ip);
+	xfs_synchronize_times(ip);
 
 	if (XFS_TEST_ERROR(be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC,
 			       mp, XFS_ERRTAG_IFLUSH_1, XFS_RANDOM_IFLUSH_1)) {
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 0b38b9a869ec..41555de1d1db 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -504,7 +504,7 @@ void		xfs_ichgtime(xfs_inode_t *, int);
 void		xfs_lock_inodes(xfs_inode_t **, int, uint);
 void		xfs_lock_two_inodes(xfs_inode_t *, xfs_inode_t *, uint);
 
-void		xfs_synchronize_atime(xfs_inode_t *);
+void		xfs_synchronize_times(xfs_inode_t *);
 void		xfs_mark_inode_dirty_sync(xfs_inode_t *);
 
 #if defined(XFS_INODE_TRACE)
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 47d5b663c37e..9794b876d6ff 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -232,6 +232,15 @@ xfs_inode_item_format(
 	nvecs	     = 1;
 
 	/*
+	 * Make sure the linux inode is dirty. We do this before
+	 * clearing i_update_core as the VFS will call back into
+	 * XFS here and set i_update_core, so we need to dirty the
+	 * inode first so that the ordering of i_update_core and
+	 * unlogged modifications still works as described below.
+	 */
+	xfs_mark_inode_dirty_sync(ip);
+
+	/*
 	 * Clear i_update_core if the timestamps (or any other
 	 * non-transactional modification) need flushing/logging
 	 * and we're about to log them with the rest of the core.
@@ -263,14 +272,9 @@ xfs_inode_item_format(
 	}
 
 	/*
-	 * Make sure to get the latest atime from the Linux inode.
+	 * Make sure to get the latest timestamps from the Linux inode.
 	 */
-	xfs_synchronize_atime(ip);
-
-	/*
-	 * make sure the linux inode is dirty
-	 */
-	xfs_mark_inode_dirty_sync(ip);
+	xfs_synchronize_times(ip);
 
 	vecp->i_addr = (xfs_caddr_t)&ip->i_d;
 	vecp->i_len  = sizeof(struct xfs_icdinode);
diff --git a/fs/xfs/xfs_itable.c b/fs/xfs/xfs_itable.c
index b68f9107e26c..62efab2f3839 100644
--- a/fs/xfs/xfs_itable.c
+++ b/fs/xfs/xfs_itable.c
@@ -59,6 +59,7 @@ xfs_bulkstat_one_iget(
 {
 	xfs_icdinode_t	*dic;	/* dinode core info pointer */
 	xfs_inode_t	*ip;		/* incore inode pointer */
+	struct inode	*inode;
 	int		error;
 
 	error = xfs_iget(mp, NULL, ino,
@@ -72,6 +73,7 @@ xfs_bulkstat_one_iget(
 	ASSERT(ip->i_imap.im_blkno != 0);
 
 	dic = &ip->i_d;
+	inode = VFS_I(ip);
 
 	/* xfs_iget returns the following without needing
 	 * further change.
@@ -83,16 +85,19 @@ xfs_bulkstat_one_iget(
 	buf->bs_uid = dic->di_uid;
 	buf->bs_gid = dic->di_gid;
 	buf->bs_size = dic->di_size;
+
 	/*
-	 * We are reading the atime from the Linux inode because the
-	 * dinode might not be uptodate.
+	 * We need to read the timestamps from the Linux inode because
+	 * the VFS keeps writing directly into the inode structure instead
+	 * of telling us about the updates.
 	 */
-	buf->bs_atime.tv_sec = VFS_I(ip)->i_atime.tv_sec;
-	buf->bs_atime.tv_nsec = VFS_I(ip)->i_atime.tv_nsec;
-	buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
-	buf->bs_mtime.tv_nsec = dic->di_mtime.t_nsec;
-	buf->bs_ctime.tv_sec = dic->di_ctime.t_sec;
-	buf->bs_ctime.tv_nsec = dic->di_ctime.t_nsec;
+	buf->bs_atime.tv_sec = inode->i_atime.tv_sec;
+	buf->bs_atime.tv_nsec = inode->i_atime.tv_nsec;
+	buf->bs_mtime.tv_sec = inode->i_mtime.tv_sec;
+	buf->bs_mtime.tv_nsec = inode->i_mtime.tv_nsec;
+	buf->bs_ctime.tv_sec = inode->i_ctime.tv_sec;
+	buf->bs_ctime.tv_nsec = inode->i_ctime.tv_nsec;
+
 	buf->bs_xflags = xfs_ip2xflags(ip);
 	buf->bs_extsize = dic->di_extsize << mp->m_sb.sb_blocklog;
 	buf->bs_extents = dic->di_nextents;
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index a434f287962d..b572f7e840e0 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -2476,12 +2476,6 @@ xfs_reclaim(
 	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);
 
 	/*
-	 * Make sure the atime in the XFS inode is correct before freeing the
-	 * Linux inode.
-	 */
-	xfs_synchronize_atime(ip);
-
-	/*
 	 * If we have nothing to flush with this inode then complete the
 	 * teardown now, otherwise break the link between the xfs inode and the
 	 * linux inode and clean up the xfs inode later. This avoids flushing