Merge tag 'afs-next-20171113' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs

Pull AFS updates from David Howells:
 "kAFS filesystem driver overhaul.

  The major points of the overhaul are:

   (1) Preliminary groundwork is laid for supporting network-namespacing
       of kAFS. The remainder of the namespacing work requires some way
       to pass namespace information to submounts triggered by an
       automount. This requires something like the mount overhaul that's
       in progress.

   (2) sockaddr_rxrpc is used in preference to in_addr for holding
       addresses internally and add support for talking to the YFS VL
       server. With this, kAFS can do everything over IPv6 as well as
       IPv4 if it's talking to servers that support it.

   (3) Callback handling is overhauled to be generally passive rather
       than active. 'Callbacks' are promises by the server to tell us
       about data and metadata changes. Callbacks are now checked when
       we next touch an inode rather than actively going and looking for
       it where possible.

   (4) File access permit caching is overhauled to store the caching
       information per-inode rather than per-directory, shared over
       subordinate files. Whilst older AFS servers only allow ACLs on
       directories (shared to the files in that directory), newer AFS
       servers break that restriction.

       To improve memory usage and to make it easier to do mass-key
       removal, permit combinations are cached and shared.

   (5) Cell database management is overhauled to allow lighter locks to
       be used and to make cell records autonomous state machines that
       look after getting their own DNS records and cleaning themselves
       up, in particular preventing races in acquiring and relinquishing
       the fscache token for the cell.

   (6) Volume caching is overhauled. The afs_vlocation record is got rid
       of to simplify things and the superblock is now keyed on the cell
       and the numeric volume ID only. The volume record is tied to a
       superblock and normal superblock management is used to mediate
       the lifetime of the volume fscache token.

   (7) File server record caching is overhauled to make server records
       independent of cells and volumes. A server can be in multiple
       cells (in such a case, the administrator must make sure that the
       VL services for all cells correctly reflect the volumes shared
       between those cells).

       Server records are now indexed using the UUID of the server
       rather than the address since a server can have multiple
       addresses.

   (8) File server rotation is overhauled to handle VMOVED, VBUSY (and
       similar), VOFFLINE and VNOVOL indications and to handle rotation
       both of servers and addresses of those servers. The rotation will
       also wait and retry if the server says it is busy.

   (9) Data writeback is overhauled. Each inode no longer stores a list
       of modified sections tagged with the key that authorised it in
       favour of noting the modified region of a page in page->private
       and storing a list of keys that made modifications in the inode.

       This simplifies things and allows other keys to be used to
       actually write to the server if a key that made a modification
       becomes useless.

  (10) Writable mmap() is implemented. This allows a kernel to be build
       entirely on AFS.

  Note that Pre AFS-3.4 servers are no longer supported, though this can
  be added back if necessary (AFS-3.4 was released in 1998)"

* tag 'afs-next-20171113' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs: (35 commits)
  afs: Protect call->state changes against signals
  afs: Trace page dirty/clean
  afs: Implement shared-writeable mmap
  afs: Get rid of the afs_writeback record
  afs: Introduce a file-private data record
  afs: Use a dynamic port if 7001 is in use
  afs: Fix directory read/modify race
  afs: Trace the sending of pages
  afs: Trace the initiation and completion of client calls
  afs: Fix documentation on # vs % prefix in mount source specification
  afs: Fix total-length calculation for multiple-page send
  afs: Only progress call state at end of Tx phase from rxrpc callback
  afs: Make use of the YFS service upgrade to fully support IPv6
  afs: Overhaul volume and server record caching and fileserver rotation
  afs: Move server rotation code into its own file
  afs: Add an address list concept
  afs: Overhaul cell database management
  afs: Overhaul permit caching
  afs: Overhaul the callback handling
  afs: Rename struct afs_call server member to cm_server
  ...

1 files changed, 312 insertions, 299 deletions

diff --git a/fs/afs/volume.c b/fs/afs/volume.c
index db73d6dad02b..684c48293353 100644
--- a/fs/afs/volume.c
+++ b/fs/afs/volume.c
@@ -10,19 +10,167 @@
  */
 
 #include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/init.h>
 #include <linux/slab.h>
-#include <linux/fs.h>
-#include <linux/pagemap.h>
-#include <linux/sched.h>
 #include "internal.h"
 
-static const char *afs_voltypes[] = { "R/W", "R/O", "BAK" };
+unsigned __read_mostly afs_volume_gc_delay = 10;
+unsigned __read_mostly afs_volume_record_life = 60 * 60;
+
+static const char *const afs_voltypes[] = { "R/W", "R/O", "BAK" };
+
+/*
+ * Allocate a volume record and load it up from a vldb record.
+ */
+static struct afs_volume *afs_alloc_volume(struct afs_mount_params *params,
+					   struct afs_vldb_entry *vldb,
+					   unsigned long type_mask)
+{
+	struct afs_server_list *slist;
+	struct afs_server *server;
+	struct afs_volume *volume;
+	int ret = -ENOMEM, nr_servers = 0, i, j;
+
+	for (i = 0; i < vldb->nr_servers; i++)
+		if (vldb->fs_mask[i] & type_mask)
+			nr_servers++;
+
+	volume = kzalloc(sizeof(struct afs_volume), GFP_KERNEL);
+	if (!volume)
+		goto error_0;
+
+	volume->vid		= vldb->vid[params->type];
+	volume->update_at	= ktime_get_real_seconds() + afs_volume_record_life;
+	volume->cell		= afs_get_cell(params->cell);
+	volume->type		= params->type;
+	volume->type_force	= params->force;
+	volume->name_len	= vldb->name_len;
+
+	atomic_set(&volume->usage, 1);
+	INIT_LIST_HEAD(&volume->proc_link);
+	rwlock_init(&volume->servers_lock);
+	memcpy(volume->name, vldb->name, vldb->name_len + 1);
+
+	slist = afs_alloc_server_list(params->cell, params->key, vldb, type_mask);
+	if (IS_ERR(slist)) {
+		ret = PTR_ERR(slist);
+		goto error_1;
+	}
+
+	refcount_set(&slist->usage, 1);
+	volume->servers = slist;
+
+	/* Make sure a records exists for each server this volume occupies. */
+	for (i = 0; i < nr_servers; i++) {
+		if (!(vldb->fs_mask[i] & type_mask))
+			continue;
+
+		server = afs_lookup_server(params->cell, params->key,
+					   &vldb->fs_server[i]);
+		if (IS_ERR(server)) {
+			ret = PTR_ERR(server);
+			if (ret == -ENOENT)
+				continue;
+			goto error_2;
+		}
+
+		/* Insertion-sort by server pointer */
+		for (j = 0; j < slist->nr_servers; j++)
+			if (slist->servers[j].server >= server)
+				break;
+		if (j < slist->nr_servers) {
+			if (slist->servers[j].server == server) {
+				afs_put_server(params->net, server);
+				continue;
+			}
+
+			memmove(slist->servers + j + 1,
+				slist->servers + j,
+				(slist->nr_servers - j) * sizeof(struct afs_server_entry));
+		}
+
+		slist->servers[j].server = server;
+		slist->nr_servers++;
+	}
+
+	if (slist->nr_servers == 0) {
+		ret = -EDESTADDRREQ;
+		goto error_2;
+	}
+
+	return volume;
+
+error_2:
+	afs_put_serverlist(params->net, slist);
+error_1:
+	kfree(volume);
+error_0:
+	return ERR_PTR(ret);
+}
 
 /*
- * lookup a volume by name
- * - this can be one of the following:
+ * Look up a VLDB record for a volume.
+ */
+static struct afs_vldb_entry *afs_vl_lookup_vldb(struct afs_cell *cell,
+						 struct key *key,
+						 const char *volname,
+						 size_t volnamesz)
+{
+	struct afs_addr_cursor ac;
+	struct afs_vldb_entry *vldb;
+	int ret;
+
+	ret = afs_set_vl_cursor(&ac, cell);
+	if (ret < 0)
+		return ERR_PTR(ret);
+
+	while (afs_iterate_addresses(&ac)) {
+		if (!test_bit(ac.index, &ac.alist->probed)) {
+			ret = afs_vl_get_capabilities(cell->net, &ac, key);
+			switch (ret) {
+			case VL_SERVICE:
+				clear_bit(ac.index, &ac.alist->yfs);
+				set_bit(ac.index, &ac.alist->probed);
+				ac.addr->srx_service = ret;
+				break;
+			case YFS_VL_SERVICE:
+				set_bit(ac.index, &ac.alist->yfs);
+				set_bit(ac.index, &ac.alist->probed);
+				ac.addr->srx_service = ret;
+				break;
+			}
+		}
+		
+		vldb = afs_vl_get_entry_by_name_u(cell->net, &ac, key,
+						  volname, volnamesz);
+		switch (ac.error) {
+		case 0:
+			afs_end_cursor(&ac);
+			return vldb;
+		case -ECONNABORTED:
+			ac.error = afs_abort_to_error(ac.abort_code);
+			goto error;
+		case -ENOMEM:
+		case -ENONET:
+			goto error;
+		case -ENETUNREACH:
+		case -EHOSTUNREACH:
+		case -ECONNREFUSED:
+			break;
+		default:
+			ac.error = -EIO;
+			goto error;
+		}
+	}
+
+error:
+	return ERR_PTR(afs_end_cursor(&ac));
+}
+
+/*
+ * Look up a volume in the VL server and create a candidate volume record for
+ * it.
+ *
+ * The volume name can be one of the following:
  *	"%[cell:]volume[.]"		R/W volume
  *	"#[cell:]volume[.]"		R/O or R/W volume (rwparent=0),
  *					 or R/W (rwparent=1) volume
@@ -42,353 +190,218 @@ static const char *afs_voltypes[] = { "R/W", "R/O", "BAK" };
  * - Rule 3: If parent volume is R/W, then only mount R/W volume unless
  *           explicitly told otherwise
  */
-struct afs_volume *afs_volume_lookup(struct afs_mount_params *params)
+struct afs_volume *afs_create_volume(struct afs_mount_params *params)
 {
-	struct afs_vlocation *vlocation = NULL;
-	struct afs_volume *volume = NULL;
-	struct afs_server *server = NULL;
-	char srvtmask;
-	int ret, loop;
-
-	_enter("{%*.*s,%d}",
-	       params->volnamesz, params->volnamesz, params->volname, params->rwpath);
-
-	/* lookup the volume location record */
-	vlocation = afs_vlocation_lookup(params->cell, params->key,
-					 params->volname, params->volnamesz);
-	if (IS_ERR(vlocation)) {
-		ret = PTR_ERR(vlocation);
-		vlocation = NULL;
-		goto error;
-	}
+	struct afs_vldb_entry *vldb;
+	struct afs_volume *volume;
+	unsigned long type_mask = 1UL << params->type;
 
-	/* make the final decision on the type we want */
-	ret = -ENOMEDIUM;
-	if (params->force && !(vlocation->vldb.vidmask & (1 << params->type)))
-		goto error;
+	vldb = afs_vl_lookup_vldb(params->cell, params->key,
+				  params->volname, params->volnamesz);
+	if (IS_ERR(vldb))
+		return ERR_CAST(vldb);
 
-	srvtmask = 0;
-	for (loop = 0; loop < vlocation->vldb.nservers; loop++)
-		srvtmask |= vlocation->vldb.srvtmask[loop];
+	if (test_bit(AFS_VLDB_QUERY_ERROR, &vldb->flags)) {
+		volume = ERR_PTR(vldb->error);
+		goto error;
+	}
 
+	/* Make the final decision on the type we want */
+	volume = ERR_PTR(-ENOMEDIUM);
 	if (params->force) {
-		if (!(srvtmask & (1 << params->type)))
+		if (!(vldb->flags & type_mask))
 			goto error;
-	} else if (srvtmask & AFS_VOL_VTM_RO) {
+	} else if (test_bit(AFS_VLDB_HAS_RO, &vldb->flags)) {
 		params->type = AFSVL_ROVOL;
-	} else if (srvtmask & AFS_VOL_VTM_RW) {
+	} else if (test_bit(AFS_VLDB_HAS_RW, &vldb->flags)) {
 		params->type = AFSVL_RWVOL;
 	} else {
 		goto error;
 	}
 
-	down_write(&params->cell->vl_sem);
+	type_mask = 1UL << params->type;
+	volume = afs_alloc_volume(params, vldb, type_mask);
 
-	/* is the volume already active? */
-	if (vlocation->vols[params->type]) {
-		/* yes - re-use it */
-		volume = vlocation->vols[params->type];
-		afs_get_volume(volume);
-		goto success;
-	}
+error:
+	kfree(vldb);
+	return volume;
+}
 
-	/* create a new volume record */
-	_debug("creating new volume record");
+/*
+ * Destroy a volume record
+ */
+static void afs_destroy_volume(struct afs_net *net, struct afs_volume *volume)
+{
+	_enter("%p", volume);
 
-	ret = -ENOMEM;
-	volume = kzalloc(sizeof(struct afs_volume), GFP_KERNEL);
-	if (!volume)
-		goto error_up;
+#ifdef CONFIG_AFS_FSCACHE
+	ASSERTCMP(volume->cache, ==, NULL);
+#endif
 
-	atomic_set(&volume->usage, 1);
-	volume->type		= params->type;
-	volume->type_force	= params->force;
-	volume->cell		= params->cell;
-	volume->vid		= vlocation->vldb.vid[params->type];
-
-	init_rwsem(&volume->server_sem);
-
-	/* look up all the applicable server records */
-	for (loop = 0; loop < 8; loop++) {
-		if (vlocation->vldb.srvtmask[loop] & (1 << volume->type)) {
-			server = afs_lookup_server(
-			       volume->cell, &vlocation->vldb.servers[loop]);
-			if (IS_ERR(server)) {
-				ret = PTR_ERR(server);
-				goto error_discard;
-			}
+	afs_put_serverlist(net, volume->servers);
+	afs_put_cell(net, volume->cell);
+	kfree(volume);
 
-			volume->servers[volume->nservers] = server;
-			volume->nservers++;
-		}
+	_leave(" [destroyed]");
+}
+
+/*
+ * Drop a reference on a volume record.
+ */
+void afs_put_volume(struct afs_cell *cell, struct afs_volume *volume)
+{
+	if (volume) {
+		_enter("%s", volume->name);
+
+		if (atomic_dec_and_test(&volume->usage))
+			afs_destroy_volume(cell->net, volume);
 	}
+}
 
-	/* attach the cache and volume location */
+/*
+ * Activate a volume.
+ */
+void afs_activate_volume(struct afs_volume *volume)
+{
 #ifdef CONFIG_AFS_FSCACHE
-	volume->cache = fscache_acquire_cookie(vlocation->cache,
+	volume->cache = fscache_acquire_cookie(volume->cell->cache,
 					       &afs_volume_cache_index_def,
 					       volume, true);
 #endif
-	afs_get_vlocation(vlocation);
-	volume->vlocation = vlocation;
-
-	vlocation->vols[volume->type] = volume;
-
-success:
-	_debug("kAFS selected %s volume %08x",
-	       afs_voltypes[volume->type], volume->vid);
-	up_write(&params->cell->vl_sem);
-	afs_put_vlocation(vlocation);
-	_leave(" = %p", volume);
-	return volume;
-
-	/* clean up */
-error_up:
-	up_write(&params->cell->vl_sem);
-error:
-	afs_put_vlocation(vlocation);
-	_leave(" = %d", ret);
-	return ERR_PTR(ret);
-
-error_discard:
-	up_write(&params->cell->vl_sem);
-
-	for (loop = volume->nservers - 1; loop >= 0; loop--)
-		afs_put_server(volume->servers[loop]);
 
-	kfree(volume);
-	goto error;
+	write_lock(&volume->cell->proc_lock);
+	list_add_tail(&volume->proc_link, &volume->cell->proc_volumes);
+	write_unlock(&volume->cell->proc_lock);
 }
 
 /*
- * destroy a volume record
+ * Deactivate a volume.
  */
-void afs_put_volume(struct afs_volume *volume)
+void afs_deactivate_volume(struct afs_volume *volume)
 {
-	struct afs_vlocation *vlocation;
-	int loop;
-
-	if (!volume)
-		return;
-
-	_enter("%p", volume);
+	_enter("%s", volume->name);
 
-	ASSERTCMP(atomic_read(&volume->usage), >, 0);
+	write_lock(&volume->cell->proc_lock);
+	list_del_init(&volume->proc_link);
+	write_unlock(&volume->cell->proc_lock);
 
-	vlocation = volume->vlocation;
-
-	/* to prevent a race, the decrement and the dequeue must be effectively
-	 * atomic */
-	down_write(&vlocation->cell->vl_sem);
-
-	if (likely(!atomic_dec_and_test(&volume->usage))) {
-		up_write(&vlocation->cell->vl_sem);
-		_leave("");
-		return;
-	}
-
-	vlocation->vols[volume->type] = NULL;
-
-	up_write(&vlocation->cell->vl_sem);
-
-	/* finish cleaning up the volume */
 #ifdef CONFIG_AFS_FSCACHE
-	fscache_relinquish_cookie(volume->cache, 0);
+	fscache_relinquish_cookie(volume->cache,
+				  test_bit(AFS_VOLUME_DELETED, &volume->flags));
+	volume->cache = NULL;
 #endif
-	afs_put_vlocation(vlocation);
-
-	for (loop = volume->nservers - 1; loop >= 0; loop--)
-		afs_put_server(volume->servers[loop]);
-
-	kfree(volume);
 
-	_leave(" [destroyed]");
+	_leave("");
 }
 
 /*
- * pick a server to use to try accessing this volume
- * - returns with an elevated usage count on the server chosen
+ * Query the VL service to update the volume status.
  */
-struct afs_server *afs_volume_pick_fileserver(struct afs_vnode *vnode)
+static int afs_update_volume_status(struct afs_volume *volume, struct key *key)
 {
-	struct afs_volume *volume = vnode->volume;
-	struct afs_server *server;
-	int ret, state, loop;
+	struct afs_server_list *new, *old, *discard;
+	struct afs_vldb_entry *vldb;
+	char idbuf[16];
+	int ret, idsz;
+
+	_enter("");
+
+	/* We look up an ID by passing it as a decimal string in the
+	 * operation's name parameter.
+	 */
+	idsz = sprintf(idbuf, "%u", volume->vid);
 
-	_enter("%s", volume->vlocation->vldb.name);
+	vldb = afs_vl_lookup_vldb(volume->cell, key, idbuf, idsz);
+	if (IS_ERR(vldb)) {
+		ret = PTR_ERR(vldb);
+		goto error;
+	}
 
-	/* stick with the server we're already using if we can */
-	if (vnode->server && vnode->server->fs_state == 0) {
-		afs_get_server(vnode->server);
-		_leave(" = %p [current]", vnode->server);
-		return vnode->server;
+	/* See if the volume got renamed. */
+	if (vldb->name_len != volume->name_len ||
+	    memcmp(vldb->name, volume->name, vldb->name_len) != 0) {
+		/* TODO: Use RCU'd string. */
+		memcpy(volume->name, vldb->name, AFS_MAXVOLNAME);
+		volume->name_len = vldb->name_len;
 	}
 
-	down_read(&volume->server_sem);
+	/* See if the volume's server list got updated. */
+	new = afs_alloc_server_list(volume->cell, key,
+				      vldb, (1 << volume->type));
+	if (IS_ERR(new)) {
+		ret = PTR_ERR(new);
+		goto error_vldb;
+	}
 
-	/* handle the no-server case */
-	if (volume->nservers == 0) {
-		ret = volume->rjservers ? -ENOMEDIUM : -ESTALE;
-		up_read(&volume->server_sem);
-		_leave(" = %d [no servers]", ret);
-		return ERR_PTR(ret);
+	write_lock(&volume->servers_lock);
+
+	discard = new;
+	old = volume->servers;
+	if (afs_annotate_server_list(new, old)) {
+		new->seq = volume->servers_seq + 1;
+		volume->servers = new;
+		smp_wmb();
+		volume->servers_seq++;
+		discard = old;
 	}
 
-	/* basically, just search the list for the first live server and use
-	 * that */
+	volume->update_at = ktime_get_real_seconds() + afs_volume_record_life;
+	clear_bit(AFS_VOLUME_NEEDS_UPDATE, &volume->flags);
+	write_unlock(&volume->servers_lock);
 	ret = 0;
-	for (loop = 0; loop < volume->nservers; loop++) {
-		server = volume->servers[loop];
-		state = server->fs_state;
 
-		_debug("consider %d [%d]", loop, state);
+	afs_put_serverlist(volume->cell->net, discard);
+error_vldb:
+	kfree(vldb);
+error:
+	_leave(" = %d", ret);
+	return ret;
+}
 
-		switch (state) {
-			/* found an apparently healthy server */
-		case 0:
-			afs_get_server(server);
-			up_read(&volume->server_sem);
-			_leave(" = %p (picked %08x)",
-			       server, ntohl(server->addr.s_addr));
-			return server;
+/*
+ * Make sure the volume record is up to date.
+ */
+int afs_check_volume_status(struct afs_volume *volume, struct key *key)
+{
+	time64_t now = ktime_get_real_seconds();
+	int ret, retries = 0;
 
-		case -ENETUNREACH:
-			if (ret == 0)
-				ret = state;
-			break;
+	_enter("");
 
-		case -EHOSTUNREACH:
-			if (ret == 0 ||
-			    ret == -ENETUNREACH)
-				ret = state;
-			break;
+	if (volume->update_at <= now)
+		set_bit(AFS_VOLUME_NEEDS_UPDATE, &volume->flags);
 
-		case -ECONNREFUSED:
-			if (ret == 0 ||
-			    ret == -ENETUNREACH ||
-			    ret == -EHOSTUNREACH)
-				ret = state;
-			break;
+retry:
+	if (!test_bit(AFS_VOLUME_NEEDS_UPDATE, &volume->flags) &&
+	    !test_bit(AFS_VOLUME_WAIT, &volume->flags)) {
+		_leave(" = 0");
+		return 0;
+	}
 
-		default:
-		case -EREMOTEIO:
-			if (ret == 0 ||
-			    ret == -ENETUNREACH ||
-			    ret == -EHOSTUNREACH ||
-			    ret == -ECONNREFUSED)
-				ret = state;
-			break;
-		}
+	if (!test_and_set_bit_lock(AFS_VOLUME_UPDATING, &volume->flags)) {
+		ret = afs_update_volume_status(volume, key);
+		clear_bit_unlock(AFS_VOLUME_WAIT, &volume->flags);
+		clear_bit_unlock(AFS_VOLUME_UPDATING, &volume->flags);
+		wake_up_bit(&volume->flags, AFS_VOLUME_WAIT);
+		_leave(" = %d", ret);
+		return ret;
 	}
 
-	/* no available servers
-	 * - TODO: handle the no active servers case better
-	 */
-	up_read(&volume->server_sem);
-	_leave(" = %d", ret);
-	return ERR_PTR(ret);
-}
+	if (!test_bit(AFS_VOLUME_WAIT, &volume->flags)) {
+		_leave(" = 0 [no wait]");
+		return 0;
+	}
 
-/*
- * release a server after use
- * - releases the ref on the server struct that was acquired by picking
- * - records result of using a particular server to access a volume
- * - return 0 to try again, 1 if okay or to issue error
- * - the caller must release the server struct if result was 0
- */
-int afs_volume_release_fileserver(struct afs_vnode *vnode,
-				  struct afs_server *server,
-				  int result)
-{
-	struct afs_volume *volume = vnode->volume;
-	unsigned loop;
-
-	_enter("%s,%08x,%d",
-	       volume->vlocation->vldb.name, ntohl(server->addr.s_addr),
-	       result);
-
-	switch (result) {
-		/* success */
-	case 0:
-		server->fs_act_jif = jiffies;
-		server->fs_state = 0;
-		_leave("");
-		return 1;
-
-		/* the fileserver denied all knowledge of the volume */
-	case -ENOMEDIUM:
-		server->fs_act_jif = jiffies;
-		down_write(&volume->server_sem);
-
-		/* firstly, find where the server is in the active list (if it
-		 * is) */
-		for (loop = 0; loop < volume->nservers; loop++)
-			if (volume->servers[loop] == server)
-				goto present;
-
-		/* no longer there - may have been discarded by another op */
-		goto try_next_server_upw;
-
-	present:
-		volume->nservers--;
-		memmove(&volume->servers[loop],
-			&volume->servers[loop + 1],
-			sizeof(volume->servers[loop]) *
-			(volume->nservers - loop));
-		volume->servers[volume->nservers] = NULL;
-		afs_put_server(server);
-		volume->rjservers++;
-
-		if (volume->nservers > 0)
-			/* another server might acknowledge its existence */
-			goto try_next_server_upw;
-
-		/* handle the case where all the fileservers have rejected the
-		 * volume
-		 * - TODO: try asking the fileservers for volume information
-		 * - TODO: contact the VL server again to see if the volume is
-		 *         no longer registered
-		 */
-		up_write(&volume->server_sem);
-		afs_put_server(server);
-		_leave(" [completely rejected]");
-		return 1;
-
-		/* problem reaching the server */
-	case -ENETUNREACH:
-	case -EHOSTUNREACH:
-	case -ECONNREFUSED:
-	case -ETIME:
-	case -ETIMEDOUT:
-	case -EREMOTEIO:
-		/* mark the server as dead
-		 * TODO: vary dead timeout depending on error
-		 */
-		spin_lock(&server->fs_lock);
-		if (!server->fs_state) {
-			server->fs_dead_jif = jiffies + HZ * 10;
-			server->fs_state = result;
-			printk("kAFS: SERVER DEAD state=%d\n", result);
-		}
-		spin_unlock(&server->fs_lock);
-		goto try_next_server;
-
-		/* miscellaneous error */
-	default:
-		server->fs_act_jif = jiffies;
-	case -ENOMEM:
-	case -ENONET:
-		/* tell the caller to accept the result */
-		afs_put_server(server);
-		_leave(" [local failure]");
-		return 1;
+	ret = wait_on_bit(&volume->flags, AFS_VOLUME_WAIT, TASK_INTERRUPTIBLE);
+	if (ret == -ERESTARTSYS) {
+		_leave(" = %d", ret);
+		return ret;
 	}
 
-	/* tell the caller to loop around and try the next server */
-try_next_server_upw:
-	up_write(&volume->server_sem);
-try_next_server:
-	afs_put_server(server);
-	_leave(" [try next server]");
-	return 0;
+	retries++;
+	if (retries == 4) {
+		_leave(" = -ESTALE");
+		return -ESTALE;
+	}
+	goto retry;
 }