Merge commit 'v2.6.32-rc8'

16 files changed, 362 insertions, 92 deletions

diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index 69b355ae7f49..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_BTRFS_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_BTRFS_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_BTRFS_POSIX_ACL */
+#endif /* CONFIG_BTRFS_FS_POSIX_ACL */
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index c71abec0ab90..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
 	 */
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 1bb897ecdeeb..444b3e9b92a4 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1009,6 +1009,7 @@ 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;
@@ -1152,6 +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)
@@ -2373,7 +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_BTRFS_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/disk-io.c b/fs/btrfs/disk-io.c
index 100551a66c46..02b6afbd7450 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -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);
 
@@ -1087,6 +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;
 }
 
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index d0c4d584efad..94627c4cc193 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -1568,23 +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);
@@ -1604,9 +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,
@@ -2980,10 +2977,10 @@ static int maybe_allocate_chunk(struct btrfs_root *root,
 
 	free_space = btrfs_super_total_bytes(disk_super);
 	/*
-	 * we allow the metadata to grow to a max of either 5gb or 5% of the
+	 * we allow the metadata to grow to a max of either 10gb or 5% of the
 	 * space in the volume.
 	 */
-	min_metadata = min((u64)5 * 1024 * 1024 * 1024,
+	min_metadata = min((u64)10 * 1024 * 1024 * 1024,
 			     div64_u64(free_space * 5, 100));
 	if (info->total_bytes >= min_metadata) {
 		spin_unlock(&info->lock);
@@ -3690,6 +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;
@@ -4097,7 +4102,7 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 }
 
 enum btrfs_loop_type {
-	LOOP_CACHED_ONLY = 0,
+	LOOP_FIND_IDEAL = 0,
 	LOOP_CACHING_NOWAIT = 1,
 	LOOP_CACHING_WAIT = 2,
 	LOOP_ALLOC_CHUNK = 3,
@@ -4126,12 +4131,15 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_block_group_cache *block_group = NULL;
 	int empty_cluster = 2 * 1024 * 1024;
 	int allowed_chunk_alloc = 0;
+	int done_chunk_alloc = 0;
 	struct btrfs_space_info *space_info;
 	int last_ptr_loop = 0;
 	int loop = 0;
 	bool found_uncached_bg = false;
 	bool failed_cluster_refill = false;
 	bool failed_alloc = false;
+	u64 ideal_cache_percent = 0;
+	u64 ideal_cache_offset = 0;
 
 	WARN_ON(num_bytes < root->sectorsize);
 	btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
@@ -4167,14 +4175,19 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 		empty_cluster = 0;
 
 	if (search_start == hint_byte) {
+ideal_cache:
 		block_group = btrfs_lookup_block_group(root->fs_info,
 						       search_start);
 		/*
 		 * we don't want to use the block group if it doesn't match our
 		 * allocation bits, or if its not cached.
+		 *
+		 * However if we are re-searching with an ideal block group
+		 * picked out then we don't care that the block group is cached.
 		 */
 		if (block_group && block_group_bits(block_group, data) &&
-		    block_group_cache_done(block_group)) {
+		    (block_group->cached != BTRFS_CACHE_NO ||
+		     search_start == ideal_cache_offset)) {
 			down_read(&space_info->groups_sem);
 			if (list_empty(&block_group->list) ||
 			    block_group->ro) {
@@ -4186,13 +4199,13 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 				 */
 				btrfs_put_block_group(block_group);
 				up_read(&space_info->groups_sem);
-			} else
+			} else {
 				goto have_block_group;
+			}
 		} else if (block_group) {
 			btrfs_put_block_group(block_group);
 		}
 	}
-
 search:
 	down_read(&space_info->groups_sem);
 	list_for_each_entry(block_group, &space_info->block_groups, list) {
@@ -4204,28 +4217,45 @@ search:
 
 have_block_group:
 		if (unlikely(block_group->cached == BTRFS_CACHE_NO)) {
+			u64 free_percent;
+
+			free_percent = btrfs_block_group_used(&block_group->item);
+			free_percent *= 100;
+			free_percent = div64_u64(free_percent,
+						 block_group->key.offset);
+			free_percent = 100 - free_percent;
+			if (free_percent > ideal_cache_percent &&
+			    likely(!block_group->ro)) {
+				ideal_cache_offset = block_group->key.objectid;
+				ideal_cache_percent = free_percent;
+			}
+
 			/*
-			 * we want to start caching kthreads, but not too many
-			 * right off the bat so we don't overwhelm the system,
-			 * so only start them if there are less than 2 and we're
-			 * in the initial allocation phase.
+			 * We only want to start kthread caching if we are at
+			 * the point where we will wait for caching to make
+			 * progress, or if our ideal search is over and we've
+			 * found somebody to start caching.
 			 */
 			if (loop > LOOP_CACHING_NOWAIT ||
-			    atomic_read(&space_info->caching_threads) < 2) {
+			    (loop > LOOP_FIND_IDEAL &&
+			     atomic_read(&space_info->caching_threads) < 2)) {
 				ret = cache_block_group(block_group);
 				BUG_ON(ret);
 			}
-		}
-
-		cached = block_group_cache_done(block_group);
-		if (unlikely(!cached)) {
 			found_uncached_bg = true;
 
-			/* if we only want cached bgs, loop */
-			if (loop == LOOP_CACHED_ONLY)
+			/*
+			 * If loop is set for cached only, try the next block
+			 * group.
+			 */
+			if (loop == LOOP_FIND_IDEAL)
 				goto loop;
 		}
 
+		cached = block_group_cache_done(block_group);
+		if (unlikely(!cached))
+			found_uncached_bg = true;
+
 		if (unlikely(block_group->ro))
 			goto loop;
 
@@ -4405,9 +4435,11 @@ loop:
 	}
 	up_read(&space_info->groups_sem);
 
-	/* LOOP_CACHED_ONLY, only search fully cached block groups
-	 * LOOP_CACHING_NOWAIT, search partially cached block groups, but
-	 *			dont wait foR them to finish caching
+	/* LOOP_FIND_IDEAL, only search caching/cached bg's, and don't wait for
+	 *			for them to make caching progress.  Also
+	 *			determine the best possible bg to cache
+	 * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
+	 *			caching kthreads as we move along
 	 * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching
 	 * LOOP_ALLOC_CHUNK, force a chunk allocation and try again
 	 * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try
@@ -4416,12 +4448,47 @@ loop:
 	if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE &&
 	    (found_uncached_bg || empty_size || empty_cluster ||
 	     allowed_chunk_alloc)) {
-		if (found_uncached_bg) {
+		if (loop == LOOP_FIND_IDEAL && found_uncached_bg) {
 			found_uncached_bg = false;
-			if (loop < LOOP_CACHING_WAIT) {
-				loop++;
+			loop++;
+			if (!ideal_cache_percent &&
+			    atomic_read(&space_info->caching_threads))
 				goto search;
-			}
+
+			/*
+			 * 1 of the following 2 things have happened so far
+			 *
+			 * 1) We found an ideal block group for caching that
+			 * is mostly full and will cache quickly, so we might
+			 * as well wait for it.
+			 *
+			 * 2) We searched for cached only and we didn't find
+			 * anything, and we didn't start any caching kthreads
+			 * either, so chances are we will loop through and
+			 * start a couple caching kthreads, and then come back
+			 * around and just wait for them.  This will be slower
+			 * because we will have 2 caching kthreads reading at
+			 * the same time when we could have just started one
+			 * and waited for it to get far enough to give us an
+			 * allocation, so go ahead and go to the wait caching
+			 * loop.
+			 */
+			loop = LOOP_CACHING_WAIT;
+			search_start = ideal_cache_offset;
+			ideal_cache_percent = 0;
+			goto ideal_cache;
+		} else if (loop == LOOP_FIND_IDEAL) {
+			/*
+			 * Didn't find a uncached bg, wait on anything we find
+			 * next.
+			 */
+			loop = LOOP_CACHING_WAIT;
+			goto search;
+		}
+
+		if (loop < LOOP_CACHING_WAIT) {
+			loop++;
+			goto search;
 		}
 
 		if (loop == LOOP_ALLOC_CHUNK) {
@@ -4433,7 +4500,8 @@ loop:
 			ret = do_chunk_alloc(trans, root, num_bytes +
 					     2 * 1024 * 1024, data, 1);
 			allowed_chunk_alloc = 0;
-		} else {
+			done_chunk_alloc = 1;
+		} else if (!done_chunk_alloc) {
 			space_info->force_alloc = 1;
 		}
 
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index 2c726b7b9faa..ccbdcb54ec5d 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -208,7 +208,7 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
 	write_lock(&tree->lock);
 	em = lookup_extent_mapping(tree, start, len);
 
-	WARN_ON(em->start != start || !em);
+	WARN_ON(!em || em->start != start);
 
 	if (!em)
 		goto out;
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 2d623aa0625f..06550affbd27 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1086,8 +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) {
@@ -1137,6 +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
@@ -1144,6 +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) {
@@ -1153,13 +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
 	 */
@@ -1188,14 +1195,18 @@ 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:
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 5c2caad76212..cb2849f03251 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -1296,7 +1296,7 @@ again:
 			window_start = entry->offset;
 			window_free = entry->bytes;
 			last = entry;
-			max_extent = 0;
+			max_extent = entry->bytes;
 		} else {
 			last = next;
 			window_free += next->bytes;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 9e138b793dc7..b3ad168a0bfc 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -538,7 +538,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	struct extent_io_tree *io_tree;
-	int ret;
+	int ret = 0;
 
 	if (list_empty(&async_cow->extents))
 		return 0;
@@ -552,6 +552,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 
 		io_tree = &BTRFS_I(inode)->io_tree;
 
+retry:
 		/* did the compression code fall back to uncompressed IO? */
 		if (!async_extent->pages) {
 			int page_started = 0;
@@ -562,11 +563,11 @@ static noinline int submit_compressed_extents(struct inode *inode,
 				    async_extent->ram_size - 1, GFP_NOFS);
 
 			/* allocate blocks */
-			cow_file_range(inode, async_cow->locked_page,
-				       async_extent->start,
-				       async_extent->start +
-				       async_extent->ram_size - 1,
-				       &page_started, &nr_written, 0);
+			ret = cow_file_range(inode, async_cow->locked_page,
+					     async_extent->start,
+					     async_extent->start +
+					     async_extent->ram_size - 1,
+					     &page_started, &nr_written, 0);
 
 			/*
 			 * if page_started, cow_file_range inserted an
@@ -574,7 +575,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 			 * and IO for us.  Otherwise, we need to submit
 			 * all those pages down to the drive.
 			 */
-			if (!page_started)
+			if (!page_started && !ret)
 				extent_write_locked_range(io_tree,
 						  inode, async_extent->start,
 						  async_extent->start +
@@ -602,7 +603,21 @@ static noinline int submit_compressed_extents(struct inode *inode,
 					   async_extent->compressed_size,
 					   0, alloc_hint,
 					   (u64)-1, &ins, 1);
-		BUG_ON(ret);
+		if (ret) {
+			int i;
+			for (i = 0; i < async_extent->nr_pages; i++) {
+				WARN_ON(async_extent->pages[i]->mapping);
+				page_cache_release(async_extent->pages[i]);
+			}
+			kfree(async_extent->pages);
+			async_extent->nr_pages = 0;
+			async_extent->pages = NULL;
+			unlock_extent(io_tree, async_extent->start,
+				      async_extent->start +
+				      async_extent->ram_size - 1, GFP_NOFS);
+			goto retry;
+		}
+
 		em = alloc_extent_map(GFP_NOFS);
 		em->start = async_extent->start;
 		em->len = async_extent->ram_size;
@@ -743,8 +758,22 @@ static noinline int cow_file_range(struct inode *inode,
 	em = search_extent_mapping(&BTRFS_I(inode)->extent_tree,
 				   start, num_bytes);
 	if (em) {
-		alloc_hint = em->block_start;
-		free_extent_map(em);
+		/*
+		 * if block start isn't an actual block number then find the
+		 * first block in this inode and use that as a hint.  If that
+		 * block is also bogus then just don't worry about it.
+		 */
+		if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
+			free_extent_map(em);
+			em = search_extent_mapping(em_tree, 0, 0);
+			if (em && em->block_start < EXTENT_MAP_LAST_BYTE)
+				alloc_hint = em->block_start;
+			if (em)
+				free_extent_map(em);
+		} else {
+			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);
@@ -2474,7 +2503,19 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 
 	root = BTRFS_I(dir)->root;
 
+	/*
+	 * 5 items for unlink inode
+	 * 1 for orphan
+	 */
+	ret = btrfs_reserve_metadata_space(root, 6);
+	if (ret)
+		return ret;
+
 	trans = btrfs_start_transaction(root, 1);
+	if (IS_ERR(trans)) {
+		btrfs_unreserve_metadata_space(root, 6);
+		return PTR_ERR(trans);
+	}
 
 	btrfs_set_trans_block_group(trans, dir);
 
@@ -2489,6 +2530,7 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 	nr = trans->blocks_used;
 
 	btrfs_end_transaction_throttle(trans, root);
+	btrfs_unreserve_metadata_space(root, 6);
 	btrfs_btree_balance_dirty(root, nr);
 	return ret;
 }
@@ -2569,7 +2611,16 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	    inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
 		return -ENOTEMPTY;
 
+	ret = btrfs_reserve_metadata_space(root, 5);
+	if (ret)
+		return ret;
+
 	trans = btrfs_start_transaction(root, 1);
+	if (IS_ERR(trans)) {
+		btrfs_unreserve_metadata_space(root, 5);
+		return PTR_ERR(trans);
+	}
+
 	btrfs_set_trans_block_group(trans, dir);
 
 	if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
@@ -2592,6 +2643,7 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 out:
 	nr = trans->blocks_used;
 	ret = btrfs_end_transaction_throttle(trans, root);
+	btrfs_unreserve_metadata_space(root, 5);
 	btrfs_btree_balance_dirty(root, nr);
 
 	if (ret && !err)
@@ -3032,12 +3084,22 @@ 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;
@@ -3070,6 +3132,10 @@ again:
 		goto again;
 	}
 
+	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);
@@ -3088,6 +3154,9 @@ 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:
@@ -3111,7 +3180,9 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
 	if (size <= hole_start)
 		return 0;
 
-	btrfs_truncate_page(inode->i_mapping, inode->i_size);
+	err = btrfs_truncate_page(inode->i_mapping, inode->i_size);
+	if (err)
+		return err;
 
 	while (1) {
 		struct btrfs_ordered_extent *ordered;
@@ -3480,6 +3551,7 @@ 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;
@@ -4980,7 +5052,9 @@ again:
 	set_page_dirty(page);
 	SetPageUptodate(page);
 
-	BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
+	BTRFS_I(inode)->last_trans = root->fs_info->generation;
+	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
+
 	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 out_unlock:
@@ -5005,7 +5079,9 @@ 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);
@@ -5100,9 +5176,11 @@ 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;
+	ei->root = NULL;
 	spin_lock_init(&ei->accounting_lock);
 	btrfs_ordered_inode_tree_init(&ei->ordered_tree);
 	INIT_LIST_HEAD(&ei->i_orphan);
@@ -5119,6 +5197,14 @@ void btrfs_destroy_inode(struct inode *inode)
 	WARN_ON(inode->i_data.nrpages);
 
 	/*
+	 * This can happen where we create an inode, but somebody else also
+	 * created the same inode and we need to destroy the one we already
+	 * created.
+	 */
+	if (!root)
+		goto free;
+
+	/*
 	 * Make sure we're properly removed from the ordered operation
 	 * lists.
 	 */
@@ -5153,6 +5239,7 @@ void btrfs_destroy_inode(struct inode *inode)
 	}
 	inode_tree_del(inode);
 	btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
+free:
 	kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
 }
 
@@ -5258,11 +5345,14 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		return -ENOTEMPTY;
 
 	/*
-	 * 2 items for dir items
-	 * 1 item for orphan entry
-	 * 1 item for ref
+	 * We want to reserve the absolute worst case amount of items.  So if
+	 * both inodes are subvols and we need to unlink them then that would
+	 * require 4 item modifications, but if they are both normal inodes it
+	 * would require 5 item modifications, so we'll assume their normal
+	 * inodes.  So 5 * 2 is 10, plus 1 for the new link, so 11 total items
+	 * should cover the worst case number of items we'll modify.
 	 */
-	ret = btrfs_reserve_metadata_space(root, 4);
+	ret = btrfs_reserve_metadata_space(root, 11);
 	if (ret)
 		return ret;
 
@@ -5378,7 +5468,7 @@ out_fail:
 	if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
 		up_read(&root->fs_info->subvol_sem);
 
-	btrfs_unreserve_metadata_space(root, 4);
+	btrfs_unreserve_metadata_space(root, 11);
 	return ret;
 }
 
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 9351428f30e2..67fa2d29d663 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -159,7 +159,6 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 	write_extent_buffer(l, item, ptr, sizeof(*item));
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 out:
-	btrfs_release_path(root, path);
 	btrfs_free_path(path);
 	return ret;
 }
@@ -332,7 +331,6 @@ int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	BUG_ON(refs != 0);
 	ret = btrfs_del_item(trans, root, path);
 out:
-	btrfs_release_path(root, path);
 	btrfs_free_path(path);
 	return ret;
 }
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 9de9b2236419..752a5463bf53 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -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 +349,7 @@ 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_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 	sb->s_flags |= MS_POSIXACL;
 #endif
 
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 0b8f36d4400a..c207e8c32c9b 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -163,8 +163,14 @@ static void wait_current_trans(struct btrfs_root *root)
 	}
 }
 
+enum btrfs_trans_type {
+	TRANS_START,
+	TRANS_JOIN,
+	TRANS_USERSPACE,
+};
+
 static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
-					     int num_blocks, int wait)
+					     int num_blocks, int type)
 {
 	struct btrfs_trans_handle *h =
 		kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
@@ -172,7 +178,8 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 
 	mutex_lock(&root->fs_info->trans_mutex);
 	if (!root->fs_info->log_root_recovering &&
-	    ((wait == 1 && !root->fs_info->open_ioctl_trans) || wait == 2))
+	    ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
+	     type == TRANS_USERSPACE))
 		wait_current_trans(root);
 	ret = join_transaction(root);
 	BUG_ON(ret);
@@ -186,7 +193,7 @@ 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)
+	if (!current->journal_info && type != TRANS_USERSPACE)
 		current->journal_info = h;
 
 	root->fs_info->running_transaction->use_count++;
@@ -198,18 +205,18 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 						   int num_blocks)
 {
-	return start_transaction(root, num_blocks, 1);
+	return start_transaction(root, num_blocks, TRANS_START);
 }
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
 						   int num_blocks)
 {
-	return start_transaction(root, num_blocks, 0);
+	return start_transaction(root, num_blocks, TRANS_JOIN);
 }
 
 struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
 							 int num_blocks)
 {
-	return start_transaction(r, num_blocks, 2);
+	return start_transaction(r, num_blocks, TRANS_USERSPACE);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -344,10 +351,10 @@ 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;
@@ -394,6 +401,29 @@ 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);
@@ -424,6 +454,22 @@ 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)
 {
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 4edfdc2acc5f..741666a7676a 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -1980,6 +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;
@@ -1994,12 +1995,13 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
 		wait_log_commit(trans, root, root->log_transid - 1);
 
-	while (root->log_multiple_pids) {
+	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;
@@ -2012,12 +2014,16 @@ 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;
@@ -2046,6 +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);
@@ -2065,6 +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;
@@ -2073,6 +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);
@@ -2092,9 +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();
@@ -2862,6 +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
@@ -2901,6 +2930,11 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	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);
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/xattr.c b/fs/btrfs/xattr.c
index b0fc93f95fd0..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_BTRFS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 	&btrfs_xattr_acl_access_handler,
 	&btrfs_xattr_acl_default_handler,
 #endif