Changes in gfs2:

- An iopen glock locking scheme rework that speeds up deletes of
   inodes accessed from multiple nodes.
 - Various bug fixes and debugging improvements.
 - Convert gfs2-glocks.txt to ReST.
 -----BEGIN PGP SIGNATURE-----
 
 iQJIBAABCAAyFiEEJZs3krPW0xkhLMTc1b+f6wMTZToFAl7eYjMUHGFncnVlbmJh
 QHJlZGhhdC5jb20ACgkQ1b+f6wMTZTr/9g//cJ6jgiD/+qzh0VzougVksVZIduAl
 RMB+kldOjBS2ORbyYM87Jm1tdyakgZuFO91XlSwChWRdC3Y2mqdaJIEE/kATqfY9
 7Frlw++SyFKLvIf04kDYGk2hXX+umXXYFfrIiKb0tzDSGkPRaARUb3RM4TRvlSDP
 /U0JlYA/4aXMUge+VpYsbpSGeqHNfEzmcmCyPXGmZYyh1MZ/RocMZFYEsP9NH82J
 l07fxowUd10LJPEmBajzjD2NmEvjdvF4gBCOfJVNIfOzCj0CwXL3vmtu1SUNOKr+
 em266EWZF89eOcvfdtE6xF0w81oCAK43wYRjIODSgI9JCLXGiOYmlWZVwZoqu5iy
 2GQDhj/taq3ObuVqjR5n6GYuqMoJ+D0LSD13qccDALq/Bdy4lq9TMLSdDbkrVIM/
 8BVn0nI5MzUlTV3mq6uxhU0HqtYDwUEiHWURWw6bYRug5OvQy3nbg/XZptYlnD87
 XQccE4ErjlgSHLiYx1YckFz/GG6ytrRAKl9bGMkZ0u2+XmDsH+iJJgLcaXCPUP9h
 /hhYagKI55UBDer7we4tppbu+gnJrg3PXlgImf53iMc7ia0KQHd+TfSIFkGPuydI
 aEKKhIQzje23JayMbPRnwPbNlw9zU1loPi7hPS3rCpDY+w8oawpFyIieEOcxJyEt
 pYkOt4BQi9LvpGc=
 =PsCY
 -----END PGP SIGNATURE-----

Merge tag 'gfs2-for-5.8' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2

Pull gfs2 updates from Andreas Gruenbacher:

 - An iopen glock locking scheme rework that speeds up deletes of inodes
   accessed from multiple nodes

 - Various bug fixes and debugging improvements

 - Convert gfs2-glocks.txt to ReST

* tag 'gfs2-for-5.8' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
  gfs2: fix use-after-free on transaction ail lists
  gfs2: new slab for transactions
  gfs2: initialize transaction tr_ailX_lists earlier
  gfs2: Smarter iopen glock waiting
  gfs2: Wake up when setting GLF_DEMOTE
  gfs2: Check inode generation number in delete_work_func
  gfs2: Move inode generation number check into gfs2_inode_lookup
  gfs2: Minor gfs2_lookup_by_inum cleanup
  gfs2: Try harder to delete inodes locally
  gfs2: Give up the iopen glock on contention
  gfs2: Turn gl_delete into a delayed work
  gfs2: Keep track of deleted inode generations in LVBs
  gfs2: Allow ASPACE glocks to also have an lvb
  gfs2: instrumentation wrt log_flush stuck
  gfs2: introduce new gfs2_glock_assert_withdraw
  gfs2: print mapping->nrpages in glock dump for address space glocks
  gfs2: Only do glock put in gfs2_create_inode for free inodes
  gfs2: Allow lock_nolock mount to specify jid=X
  gfs2: Don't ignore inode write errors during inode_go_sync
  docs: filesystems: convert gfs2-glocks.txt to ReST

Documentation/filesystems/gfs2-glocks.rst

@@ -1,5 +1,8 @@
-                   Glock internal locking rules
+.. SPDX-License-Identifier: GPL-2.0
-                  ------------------------------
+
+============================
+Glock internal locking rules
+============================
 
 This documents the basic principles of the glock state machine
 internals. Each glock (struct gfs2_glock in fs/gfs2/incore.h)
@@ -24,24 +27,28 @@ There are three lock states that users of the glock layer can request,
 namely shared (SH), deferred (DF) and exclusive (EX). Those translate
 to the following DLM lock modes:
 
-Glock mode    | DLM lock mode
+==========	====== =====================================================
-------------------------------
+Glock mode      DLM    lock mode
-    UN        |    IV/NL  Unlocked (no DLM lock associated with glock) or NL
+==========	====== =====================================================
-    SH        |    PR     (Protected read)
+    UN          IV/NL  Unlocked (no DLM lock associated with glock) or NL
-    DF        |    CW     (Concurrent write)
+    SH          PR     (Protected read)
-    EX        |    EX     (Exclusive)
+    DF          CW     (Concurrent write)
+    EX          EX     (Exclusive)
+==========	====== =====================================================
 
 Thus DF is basically a shared mode which is incompatible with the "normal"
 shared lock mode, SH. In GFS2 the DF mode is used exclusively for direct I/O
 operations. The glocks are basically a lock plus some routines which deal
 with cache management. The following rules apply for the cache:
 
-Glock mode   |  Cache data | Cache Metadata | Dirty Data | Dirty Metadata
+==========      ==========   ==============   ==========   ==============
---------------------------------------------------------------------------
+Glock mode      Cache data   Cache Metadata   Dirty Data   Dirty Metadata
-    UN       |     No      |       No       |     No     |      No
+==========      ==========   ==============   ==========   ==============
-    SH       |     Yes     |       Yes      |     No     |      No
+    UN             No              No             No            No
-    DF       |     No      |       Yes      |     No     |      No
+    SH             Yes             Yes            No            No
-    EX       |     Yes     |       Yes      |     Yes    |      Yes
+    DF             No              Yes            No            No
+    EX             Yes             Yes            Yes           Yes
+==========      ==========   ==============   ==========   ==============
 
 These rules are implemented using the various glock operations which
 are defined for each type of glock. Not all types of glocks use
@@ -49,21 +56,23 @@ all the modes. Only inode glocks use the DF mode for example.
 
 Table of glock operations and per type constants:
 
-Field            | Purpose
+=============      =============================================================
-----------------------------------------------------------------------------
+Field              Purpose
-go_xmote_th      | Called before remote state change (e.g. to sync dirty data)
+=============      =============================================================
-go_xmote_bh      | Called after remote state change (e.g. to refill cache)
+go_xmote_th        Called before remote state change (e.g. to sync dirty data)
-go_inval         | Called if remote state change requires invalidating the cache
+go_xmote_bh        Called after remote state change (e.g. to refill cache)
-go_demote_ok     | Returns boolean value of whether its ok to demote a glock
+go_inval           Called if remote state change requires invalidating the cache
-                 | (e.g. checks timeout, and that there is no cached data)
+go_demote_ok       Returns boolean value of whether its ok to demote a glock
-go_lock          | Called for the first local holder of a lock
+                   (e.g. checks timeout, and that there is no cached data)
-go_unlock        | Called on the final local unlock of a lock
+go_lock            Called for the first local holder of a lock
-go_dump          | Called to print content of object for debugfs file, or on
+go_unlock          Called on the final local unlock of a lock
-                 | error to dump glock to the log.
+go_dump            Called to print content of object for debugfs file, or on
-go_type          | The type of the glock, LM_TYPE_.....
+                   error to dump glock to the log.
-go_callback	 | Called if the DLM sends a callback to drop this lock
+go_type            The type of the glock, ``LM_TYPE_*``
-go_flags	 | GLOF_ASPACE is set, if the glock has an address space
+go_callback	   Called if the DLM sends a callback to drop this lock
-                 | associated with it
+go_flags	   GLOF_ASPACE is set, if the glock has an address space
+                   associated with it
+=============      =============================================================
 
 The minimum hold time for each lock is the time after a remote lock
 grant for which we ignore remote demote requests. This is in order to
@@ -82,21 +91,25 @@ rather than via the glock.
 
 Locking rules for glock operations:
 
-Operation     |  GLF_LOCK bit lock held |  gl_lockref.lock spinlock held
+=============    ======================    =============================
--------------------------------------------------------------------------
+Operation        GLF_LOCK bit lock held    gl_lockref.lock spinlock held
-go_xmote_th   |       Yes               |       No
+=============    ======================    =============================
-go_xmote_bh   |       Yes               |       No
+go_xmote_th           Yes                       No
-go_inval      |       Yes               |       No
+go_xmote_bh           Yes                       No
-go_demote_ok  |       Sometimes         |       Yes
+go_inval              Yes                       No
-go_lock       |       Yes               |       No
+go_demote_ok          Sometimes                 Yes
-go_unlock     |       Yes               |       No
+go_lock               Yes                       No
-go_dump       |       Sometimes         |       Yes
+go_unlock             Yes                       No
-go_callback   |       Sometimes (N/A)   |       Yes
+go_dump               Sometimes                 Yes
+go_callback           Sometimes (N/A)           Yes
+=============    ======================    =============================
 
-N.B. Operations must not drop either the bit lock or the spinlock
+.. Note::
-if its held on entry. go_dump and do_demote_ok must never block.
+
-Note that go_dump will only be called if the glock's state
+   Operations must not drop either the bit lock or the spinlock
-indicates that it is caching uptodate data.
+   if its held on entry. go_dump and do_demote_ok must never block.
+   Note that go_dump will only be called if the glock's state
+   indicates that it is caching uptodate data.
 
 Glock locking order within GFS2:
 
@@ -104,7 +117,7 @@ Glock locking order within GFS2:
  2. Rename glock (for rename only)
  3. Inode glock(s)
     (Parents before children, inodes at "same level" with same parent in
-     lock number order)
+    lock number order)
  4. Rgrp glock(s) (for (de)allocation operations)
  5. Transaction glock (via gfs2_trans_begin) for non-read operations
  6. i_rw_mutex (if required)
@@ -117,8 +130,8 @@ determine the lifetime of the inode in question. Locking of inodes
 is on a per-inode basis. Locking of rgrps is on a per rgrp basis.
 In general we prefer to lock local locks prior to cluster locks.
 
-                            Glock Statistics
+Glock Statistics
-                           ------------------
+----------------
 
 The stats are divided into two sets: those relating to the
 super block and those relating to an individual glock. The
@@ -173,8 +186,8 @@ we'd like to get a better idea of these timings:
 1. To be able to better set the glock "min hold time"
 2. To spot performance issues more easily
 3. To improve the algorithm for selecting resource groups for
-allocation (to base it on lock wait time, rather than blindly
+   allocation (to base it on lock wait time, rather than blindly
-using a "try lock")
+   using a "try lock")
 
 Due to the smoothing action of the updates, a step change in
 some input quantity being sampled will only fully be taken
@@ -195,10 +208,13 @@ as possible. There are always inaccuracies in any
 measuring system, but I hope this is as accurate as we
 can reasonably make it.
 
-Per sb stats can be found here:
+Per sb stats can be found here::
-/sys/kernel/debug/gfs2/<fsname>/sbstats
+
-Per glock stats can be found here:
+    /sys/kernel/debug/gfs2/<fsname>/sbstats
-/sys/kernel/debug/gfs2/<fsname>/glstats
+
+Per glock stats can be found here::
+
+    /sys/kernel/debug/gfs2/<fsname>/glstats
 
 Assuming that debugfs is mounted on /sys/kernel/debug and also
 that <fsname> is replaced with the name of the gfs2 filesystem
@@ -206,14 +222,16 @@ in question.
 
 The abbreviations used in the output as are follows:
 
-srtt     - Smoothed round trip time for non-blocking dlm requests
+=========  ================================================================
-srttvar  - Variance estimate for srtt
+srtt       Smoothed round trip time for non blocking dlm requests
-srttb    - Smoothed round trip time for (potentially) blocking dlm requests
+srttvar    Variance estimate for srtt
-srttvarb - Variance estimate for srttb
+srttb      Smoothed round trip time for (potentially) blocking dlm requests
-sirt     - Smoothed inter-request time (for dlm requests)
+srttvarb   Variance estimate for srttb
-sirtvar  - Variance estimate for sirt
+sirt       Smoothed inter request time (for dlm requests)
-dlm      - Number of dlm requests made (dcnt in glstats file)
+sirtvar    Variance estimate for sirt
-queue    - Number of glock requests queued (qcnt in glstats file)
+dlm        Number of dlm requests made (dcnt in glstats file)
+queue      Number of glock requests queued (qcnt in glstats file)
+=========  ================================================================
 
 The sbstats file contains a set of these stats for each glock type (so 8 lines
 for each type) and for each cpu (one column per cpu). The glstats file contains
@@ -224,9 +242,12 @@ The gfs2_glock_lock_time tracepoint prints out the current values of the stats
 for the glock in question, along with some addition information on each dlm
 reply that is received:
 
-status - The status of the dlm request
+======   =======================================
-flags  - The dlm request flags
+status   The status of the dlm request
-tdiff  - The time taken by this specific request
+flags    The dlm request flags
+tdiff    The time taken by this specific request
+======   =======================================
+
 (remaining fields as per above list)
 
 

Documentation/filesystems/index.rst

@@ -88,6 +88,7 @@ Documentation for filesystem implementations.
    f2fs
    gfs2
    gfs2-uevents
+   gfs2-glocks
    hfs
    hfsplus
    hpfs

MAINTAINERS

@@ -7251,7 +7251,7 @@ L:	cluster-devel@redhat.com
 S:	Supported
 W:	http://sources.redhat.com/cluster/
 T:	git git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git
-F:	Documentation/filesystems/gfs2*.txt
+F:	Documentation/filesystems/gfs2*
 F:	fs/gfs2/
 F:	include/uapi/linux/gfs2_ondisk.h
 

fs/gfs2/export.c

@@ -134,7 +134,9 @@ static struct dentry *gfs2_get_dentry(struct super_block *sb,
 	struct gfs2_sbd *sdp = sb->s_fs_info;
 	struct inode *inode;
 
-	inode = gfs2_lookup_by_inum(sdp, inum->no_addr, &inum->no_formal_ino,
+	if (!inum->no_formal_ino)
+		return ERR_PTR(-ESTALE);
+	inode = gfs2_lookup_by_inum(sdp, inum->no_addr, inum->no_formal_ino,
 				    GFS2_BLKST_DINODE);
 	if (IS_ERR(inode))
 		return ERR_CAST(inode);

fs/gfs2/glock.c

@@ -125,12 +125,11 @@ static void gfs2_glock_dealloc(struct rcu_head *rcu)
 {
 	struct gfs2_glock *gl = container_of(rcu, struct gfs2_glock, gl_rcu);
 
-	if (gl->gl_ops->go_flags & GLOF_ASPACE) {
+	kfree(gl->gl_lksb.sb_lvbptr);
+	if (gl->gl_ops->go_flags & GLOF_ASPACE)
 		kmem_cache_free(gfs2_glock_aspace_cachep, gl);
-	} else {
+	else
-		kfree(gl->gl_lksb.sb_lvbptr);
 		kmem_cache_free(gfs2_glock_cachep, gl);
-	}
 }
 
 /**
@@ -164,7 +163,7 @@ void gfs2_glock_free(struct gfs2_glock *gl)
 {
 	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
 
-	BUG_ON(atomic_read(&gl->gl_revokes));
+	gfs2_glock_assert_withdraw(gl, atomic_read(&gl->gl_revokes) == 0);
 	rhashtable_remove_fast(&gl_hash_table, &gl->gl_node, ht_parms);
 	smp_mb();
 	wake_up_glock(gl);
@@ -465,6 +464,15 @@ static void state_change(struct gfs2_glock *gl, unsigned int new_state)
 	gl->gl_tchange = jiffies;
 }
 
+static void gfs2_set_demote(struct gfs2_glock *gl)
+{
+	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
+
+	set_bit(GLF_DEMOTE, &gl->gl_flags);
+	smp_mb();
+	wake_up(&sdp->sd_async_glock_wait);
+}
+
 static void gfs2_demote_wake(struct gfs2_glock *gl)
 {
 	gl->gl_demote_state = LM_ST_EXCLUSIVE;
@@ -626,7 +634,8 @@ __acquires(&gl->gl_lockref.lock)
 		 */
 		if ((atomic_read(&gl->gl_ail_count) != 0) &&
 		    (!cmpxchg(&sdp->sd_log_error, 0, -EIO))) {
-			gfs2_assert_warn(sdp, !atomic_read(&gl->gl_ail_count));
+			gfs2_glock_assert_warn(gl,
+					       !atomic_read(&gl->gl_ail_count));
 			gfs2_dump_glock(NULL, gl, true);
 		}
 		glops->go_inval(gl, target == LM_ST_DEFERRED ? 0 : DIO_METADATA);
@@ -756,20 +765,127 @@ out_unlock:
 	return;
 }
 
+void gfs2_inode_remember_delete(struct gfs2_glock *gl, u64 generation)
+{
+	struct gfs2_inode_lvb *ri = (void *)gl->gl_lksb.sb_lvbptr;
+
+	if (ri->ri_magic == 0)
+		ri->ri_magic = cpu_to_be32(GFS2_MAGIC);
+	if (ri->ri_magic == cpu_to_be32(GFS2_MAGIC))
+		ri->ri_generation_deleted = cpu_to_be64(generation);
+}
+
+bool gfs2_inode_already_deleted(struct gfs2_glock *gl, u64 generation)
+{
+	struct gfs2_inode_lvb *ri = (void *)gl->gl_lksb.sb_lvbptr;
+
+	if (ri->ri_magic != cpu_to_be32(GFS2_MAGIC))
+		return false;
+	return generation <= be64_to_cpu(ri->ri_generation_deleted);
+}
+
+static void gfs2_glock_poke(struct gfs2_glock *gl)
+{
+	int flags = LM_FLAG_TRY_1CB | LM_FLAG_ANY | GL_SKIP;
+	struct gfs2_holder gh;
+	int error;
+
+	error = gfs2_glock_nq_init(gl, LM_ST_SHARED, flags, &gh);
+	if (!error)
+		gfs2_glock_dq(&gh);
+}
+
+static bool gfs2_try_evict(struct gfs2_glock *gl)
+{
+	struct gfs2_inode *ip;
+	bool evicted = false;
+
+	/*
+	 * If there is contention on the iopen glock and we have an inode, try
+	 * to grab and release the inode so that it can be evicted.  This will
+	 * allow the remote node to go ahead and delete the inode without us
+	 * having to do it, which will avoid rgrp glock thrashing.
+	 *
+	 * The remote node is likely still holding the corresponding inode
+	 * glock, so it will run before we get to verify that the delete has
+	 * happened below.
+	 */
+	spin_lock(&gl->gl_lockref.lock);
+	ip = gl->gl_object;
+	if (ip && !igrab(&ip->i_inode))
+		ip = NULL;
+	spin_unlock(&gl->gl_lockref.lock);
+	if (ip) {
+		struct gfs2_glock *inode_gl = NULL;
+
+		gl->gl_no_formal_ino = ip->i_no_formal_ino;
+		set_bit(GIF_DEFERRED_DELETE, &ip->i_flags);
+		d_prune_aliases(&ip->i_inode);
+		iput(&ip->i_inode);
+
+		/* If the inode was evicted, gl->gl_object will now be NULL. */
+		spin_lock(&gl->gl_lockref.lock);
+		ip = gl->gl_object;
+		if (ip) {
+			inode_gl = ip->i_gl;
+			lockref_get(&inode_gl->gl_lockref);
+			clear_bit(GIF_DEFERRED_DELETE, &ip->i_flags);
+		}
+		spin_unlock(&gl->gl_lockref.lock);
+		if (inode_gl) {
+			gfs2_glock_poke(inode_gl);
+			gfs2_glock_put(inode_gl);
+		}
+		evicted = !ip;
+	}
+	return evicted;
+}
+
 static void delete_work_func(struct work_struct *work)
 {
-	struct gfs2_glock *gl = container_of(work, struct gfs2_glock, gl_delete);
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct gfs2_glock *gl = container_of(dwork, struct gfs2_glock, gl_delete);
 	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
 	struct inode *inode;
 	u64 no_addr = gl->gl_name.ln_number;
 
+	spin_lock(&gl->gl_lockref.lock);
+	clear_bit(GLF_PENDING_DELETE, &gl->gl_flags);
+	spin_unlock(&gl->gl_lockref.lock);
+
 	/* If someone's using this glock to create a new dinode, the block must
 	   have been freed by another node, then re-used, in which case our
 	   iopen callback is too late after the fact. Ignore it. */
 	if (test_bit(GLF_INODE_CREATING, &gl->gl_flags))
 		goto out;
 
-	inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
+	if (test_bit(GLF_DEMOTE, &gl->gl_flags)) {
+		/*
+		 * If we can evict the inode, give the remote node trying to
+		 * delete the inode some time before verifying that the delete
+		 * has happened.  Otherwise, if we cause contention on the inode glock
+		 * immediately, the remote node will think that we still have
+		 * the inode in use, and so it will give up waiting.
+		 *
+		 * If we can't evict the inode, signal to the remote node that
+		 * the inode is still in use.  We'll later try to delete the
+		 * inode locally in gfs2_evict_inode.
+		 *
+		 * FIXME: We only need to verify that the remote node has
+		 * deleted the inode because nodes before this remote delete
+		 * rework won't cooperate.  At a later time, when we no longer
+		 * care about compatibility with such nodes, we can skip this
+		 * step entirely.
+		 */
+		if (gfs2_try_evict(gl)) {
+			if (gfs2_queue_delete_work(gl, 5 * HZ))
+				return;
+		}
+		goto out;
+	}
+
+	inode = gfs2_lookup_by_inum(sdp, no_addr, gl->gl_no_formal_ino,
+				    GFS2_BLKST_UNLINKED);
 	if (!IS_ERR_OR_NULL(inode)) {
 		d_prune_aliases(inode);
 		iput(inode);
@@ -800,7 +916,7 @@ static void glock_work_func(struct work_struct *work)
 
 		if (!delay) {
 			clear_bit(GLF_PENDING_DEMOTE, &gl->gl_flags);
-			set_bit(GLF_DEMOTE, &gl->gl_flags);
+			gfs2_set_demote(gl);
 		}
 	}
 	run_queue(gl, 0);
@@ -931,7 +1047,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
 	gl->gl_object = NULL;
 	gl->gl_hold_time = GL_GLOCK_DFT_HOLD;
 	INIT_DELAYED_WORK(&gl->gl_work, glock_work_func);
-	INIT_WORK(&gl->gl_delete, delete_work_func);
+	INIT_DELAYED_WORK(&gl->gl_delete, delete_work_func);
 
 	mapping = gfs2_glock2aspace(gl);
 	if (mapping) {
@@ -1145,9 +1261,10 @@ wait_for_dlm:
 static void handle_callback(struct gfs2_glock *gl, unsigned int state,
 			    unsigned long delay, bool remote)
 {
-	int bit = delay ? GLF_PENDING_DEMOTE : GLF_DEMOTE;
+	if (delay)
-
+		set_bit(GLF_PENDING_DEMOTE, &gl->gl_flags);
-	set_bit(bit, &gl->gl_flags);
+	else
+		gfs2_set_demote(gl);
 	if (gl->gl_demote_state == LM_ST_EXCLUSIVE) {
 		gl->gl_demote_state = state;
 		gl->gl_demote_time = jiffies;
@@ -1754,6 +1871,44 @@ static void glock_hash_walk(glock_examiner examiner, const struct gfs2_sbd *sdp)
 	rhashtable_walk_exit(&iter);
 }
 
+bool gfs2_queue_delete_work(struct gfs2_glock *gl, unsigned long delay)
+{
+	bool queued;
+
+	spin_lock(&gl->gl_lockref.lock);
+	queued = queue_delayed_work(gfs2_delete_workqueue,
+				    &gl->gl_delete, delay);
+	if (queued)
+		set_bit(GLF_PENDING_DELETE, &gl->gl_flags);
+	spin_unlock(&gl->gl_lockref.lock);
+	return queued;
+}
+
+void gfs2_cancel_delete_work(struct gfs2_glock *gl)
+{
+	if (cancel_delayed_work_sync(&gl->gl_delete)) {
+		clear_bit(GLF_PENDING_DELETE, &gl->gl_flags);
+		gfs2_glock_put(gl);
+	}
+}
+
+bool gfs2_delete_work_queued(const struct gfs2_glock *gl)
+{
+	return test_bit(GLF_PENDING_DELETE, &gl->gl_flags);
+}
+
+static void flush_delete_work(struct gfs2_glock *gl)
+{
+	flush_delayed_work(&gl->gl_delete);
+	gfs2_glock_queue_work(gl, 0);
+}
+
+void gfs2_flush_delete_work(struct gfs2_sbd *sdp)
+{
+	glock_hash_walk(flush_delete_work, sdp);
+	flush_workqueue(gfs2_delete_workqueue);
+}
+
 /**
  * thaw_glock - thaw out a glock which has an unprocessed reply waiting
  * @gl: The glock to thaw
@@ -1836,7 +1991,7 @@ void gfs2_glock_finish_truncate(struct gfs2_inode *ip)
 	int ret;
 
 	ret = gfs2_truncatei_resume(ip);
-	gfs2_assert_withdraw(gl->gl_name.ln_sbd, ret == 0);
+	gfs2_glock_assert_withdraw(gl, ret == 0);
 
 	spin_lock(&gl->gl_lockref.lock);
 	clear_bit(GLF_LOCK, &gl->gl_flags);
@@ -1978,7 +2133,13 @@ void gfs2_dump_glock(struct seq_file *seq, struct gfs2_glock *gl, bool fsid)
 	char gflags_buf[32];
 	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
 	char fs_id_buf[sizeof(sdp->sd_fsname) + 7];
+	unsigned long nrpages = 0;
 
+	if (gl->gl_ops->go_flags & GLOF_ASPACE) {
+		struct address_space *mapping = gfs2_glock2aspace(gl);
+
+		nrpages = mapping->nrpages;
+	}
 	memset(fs_id_buf, 0, sizeof(fs_id_buf));
 	if (fsid && sdp) /* safety precaution */
 		sprintf(fs_id_buf, "fsid=%s: ", sdp->sd_fsname);
@@ -1987,15 +2148,16 @@ void gfs2_dump_glock(struct seq_file *seq, struct gfs2_glock *gl, bool fsid)
 	if (!test_bit(GLF_DEMOTE, &gl->gl_flags))
 		dtime = 0;
 	gfs2_print_dbg(seq, "%sG:  s:%s n:%u/%llx f:%s t:%s d:%s/%llu a:%d "
-		       "v:%d r:%d m:%ld\n", fs_id_buf, state2str(gl->gl_state),
+		       "v:%d r:%d m:%ld p:%lu\n",
-		  gl->gl_name.ln_type,
+		       fs_id_buf, state2str(gl->gl_state),
-		  (unsigned long long)gl->gl_name.ln_number,
+		       gl->gl_name.ln_type,
-		  gflags2str(gflags_buf, gl),
+		       (unsigned long long)gl->gl_name.ln_number,
-		  state2str(gl->gl_target),
+		       gflags2str(gflags_buf, gl),
-		  state2str(gl->gl_demote_state), dtime,
+		       state2str(gl->gl_target),
-		  atomic_read(&gl->gl_ail_count),
+		       state2str(gl->gl_demote_state), dtime,
-		  atomic_read(&gl->gl_revokes),
+		       atomic_read(&gl->gl_ail_count),
-		  (int)gl->gl_lockref.count, gl->gl_hold_time);
+		       atomic_read(&gl->gl_revokes),
+		       (int)gl->gl_lockref.count, gl->gl_hold_time, nrpages);
 
 	list_for_each_entry(gh, &gl->gl_holders, gh_list)
 		dump_holder(seq, gh, fs_id_buf);

fs/gfs2/glock.h

@@ -205,6 +205,15 @@ extern void gfs2_dump_glock(struct seq_file *seq, struct gfs2_glock *gl,
 #define GLOCK_BUG_ON(gl,x) do { if (unlikely(x)) {		\
 			gfs2_dump_glock(NULL, gl, true);	\
 			BUG(); } } while(0)
+#define gfs2_glock_assert_warn(gl, x) do { if (unlikely(!(x))) {	\
+			gfs2_dump_glock(NULL, gl, true);		\
+			gfs2_assert_warn((gl)->gl_name.ln_sbd, (x)); } } \
+	while (0)
+#define gfs2_glock_assert_withdraw(gl, x) do { if (unlikely(!(x))) {	\
+			gfs2_dump_glock(NULL, gl, true);		\
+			gfs2_assert_withdraw((gl)->gl_name.ln_sbd, (x)); } } \
+	while (0)
+
 extern __printf(2, 3)
 void gfs2_print_dbg(struct seq_file *seq, const char *fmt, ...);
 
@@ -235,6 +244,10 @@ static inline int gfs2_glock_nq_init(struct gfs2_glock *gl,
 
 extern void gfs2_glock_cb(struct gfs2_glock *gl, unsigned int state);
 extern void gfs2_glock_complete(struct gfs2_glock *gl, int ret);
+extern bool gfs2_queue_delete_work(struct gfs2_glock *gl, unsigned long delay);
+extern void gfs2_cancel_delete_work(struct gfs2_glock *gl);
+extern bool gfs2_delete_work_queued(const struct gfs2_glock *gl);
+extern void gfs2_flush_delete_work(struct gfs2_sbd *sdp);
 extern void gfs2_gl_hash_clear(struct gfs2_sbd *sdp);
 extern void gfs2_glock_finish_truncate(struct gfs2_inode *ip);
 extern void gfs2_glock_thaw(struct gfs2_sbd *sdp);
@@ -306,4 +319,7 @@ static inline void glock_clear_object(struct gfs2_glock *gl, void *object)
 	spin_unlock(&gl->gl_lockref.lock);
 }
 
+extern void gfs2_inode_remember_delete(struct gfs2_glock *gl, u64 generation);
+extern bool gfs2_inode_already_deleted(struct gfs2_glock *gl, u64 generation);
+
 #endif /* __GLOCK_DOT_H__ */

fs/gfs2/glops.c

@@ -91,6 +91,8 @@ static int gfs2_ail_empty_gl(struct gfs2_glock *gl)
 	memset(&tr, 0, sizeof(tr));
 	INIT_LIST_HEAD(&tr.tr_buf);
 	INIT_LIST_HEAD(&tr.tr_databuf);
+	INIT_LIST_HEAD(&tr.tr_ail1_list);
+	INIT_LIST_HEAD(&tr.tr_ail2_list);
 	tr.tr_revokes = atomic_read(&gl->gl_ail_count);
 
 	if (!tr.tr_revokes) {
@@ -268,7 +270,7 @@ static int inode_go_sync(struct gfs2_glock *gl)
 	struct gfs2_inode *ip = gfs2_glock2inode(gl);
 	int isreg = ip && S_ISREG(ip->i_inode.i_mode);
 	struct address_space *metamapping = gfs2_glock2aspace(gl);
-	int error = 0;
+	int error = 0, ret;
 
 	if (isreg) {
 		if (test_and_clear_bit(GIF_SW_PAGED, &ip->i_flags))
@@ -289,8 +291,10 @@ static int inode_go_sync(struct gfs2_glock *gl)
 		error = filemap_fdatawait(mapping);
 		mapping_set_error(mapping, error);
 	}
-	error = filemap_fdatawait(metamapping);
+	ret = filemap_fdatawait(metamapping);
-	mapping_set_error(metamapping, error);
+	mapping_set_error(metamapping, ret);
+	if (!error)
+		error = ret;
 	gfs2_ail_empty_gl(gl);
 	/*
 	 * Writeback of the data mapping may cause the dirty flag to be set
@@ -608,11 +612,17 @@ static void iopen_go_callback(struct gfs2_glock *gl, bool remote)
 	if (gl->gl_demote_state == LM_ST_UNLOCKED &&
 	    gl->gl_state == LM_ST_SHARED && ip) {
 		gl->gl_lockref.count++;
-		if (queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0)
+		if (!queue_delayed_work(gfs2_delete_workqueue,
+					&gl->gl_delete, 0))
 			gl->gl_lockref.count--;
 	}
 }
 
+static int iopen_go_demote_ok(const struct gfs2_glock *gl)
+{
+       return !gfs2_delete_work_queued(gl);
+}
+
 /**
  * inode_go_free - wake up anyone waiting for dlm's unlock ast to free it
  * @gl: glock being freed
@@ -692,7 +702,7 @@ const struct gfs2_glock_operations gfs2_inode_glops = {
 	.go_lock = inode_go_lock,
 	.go_dump = inode_go_dump,
 	.go_type = LM_TYPE_INODE,
-	.go_flags = GLOF_ASPACE | GLOF_LRU,
+	.go_flags = GLOF_ASPACE | GLOF_LRU | GLOF_LVB,
 	.go_free = inode_go_free,
 };
 
@@ -716,6 +726,7 @@ const struct gfs2_glock_operations gfs2_freeze_glops = {
 const struct gfs2_glock_operations gfs2_iopen_glops = {
 	.go_type = LM_TYPE_IOPEN,
 	.go_callback = iopen_go_callback,
+	.go_demote_ok = iopen_go_demote_ok,
 	.go_flags = GLOF_LRU | GLOF_NONDISK,
 };
 

fs/gfs2/incore.h

@@ -345,6 +345,7 @@ enum {
 	GLF_OBJECT			= 14, /* Used only for tracing */
 	GLF_BLOCKING			= 15,
 	GLF_INODE_CREATING		= 16, /* Inode creation occurring */
+	GLF_PENDING_DELETE		= 17,
 	GLF_FREEING			= 18, /* Wait for glock to be freed */
 };
 
@@ -378,8 +379,11 @@ struct gfs2_glock {
 	atomic_t gl_revokes;
 	struct delayed_work gl_work;
 	union {
-		/* For inode and iopen glocks only */
+		/* For iopen glocks only */
-		struct work_struct gl_delete;
+		struct {
+			struct delayed_work gl_delete;
+			u64 gl_no_formal_ino;
+		};
 		/* For rgrp glocks only */
 		struct {
 			loff_t start;
@@ -398,6 +402,7 @@ enum {
 	GIF_ORDERED		= 4,
 	GIF_FREE_VFS_INODE      = 5,
 	GIF_GLOP_PENDING	= 6,
+	GIF_DEFERRED_DELETE	= 7,
 };
 
 struct gfs2_inode {

fs/gfs2/inode.c

@@ -115,6 +115,10 @@ static void gfs2_set_iop(struct inode *inode)
  * placeholder because it doesn't otherwise make sense), the on-disk block type
  * is verified to be @blktype.
  *
+ * When @no_formal_ino is non-zero, this function will return ERR_PTR(-ESTALE)
+ * if it detects that @no_formal_ino doesn't match the actual inode generation
+ * number.  However, it doesn't always know unless @type is DT_UNKNOWN.
+ *
  * Returns: A VFS inode, or an error
  */
 
@@ -158,6 +162,11 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 			if (error)
 				goto fail;
 
+			error = -ESTALE;
+			if (no_formal_ino &&
+			    gfs2_inode_already_deleted(ip->i_gl, no_formal_ino))
+				goto fail;
+
 			if (blktype != GFS2_BLKST_FREE) {
 				error = gfs2_check_blk_type(sdp, no_addr,
 							    blktype);
@@ -171,6 +180,7 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		error = gfs2_glock_nq_init(io_gl, LM_ST_SHARED, GL_EXACT, &ip->i_iopen_gh);
 		if (unlikely(error))
 			goto fail;
+		gfs2_cancel_delete_work(ip->i_iopen_gh.gh_gl);
 		glock_set_object(ip->i_iopen_gh.gh_gl, ip);
 		gfs2_glock_put(io_gl);
 		io_gl = NULL;
@@ -189,13 +199,23 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 			inode->i_mode = DT2IF(type);
 		}
 
-		gfs2_set_iop(inode);
+		if (gfs2_holder_initialized(&i_gh))
+			gfs2_glock_dq_uninit(&i_gh);
 
-		unlock_new_inode(inode);
+		gfs2_set_iop(inode);
 	}
 
-	if (gfs2_holder_initialized(&i_gh))
+	if (no_formal_ino && ip->i_no_formal_ino &&
-		gfs2_glock_dq_uninit(&i_gh);
+	    no_formal_ino != ip->i_no_formal_ino) {
+		if (inode->i_state & I_NEW)
+			goto fail;
+		iput(inode);
+		return ERR_PTR(-ESTALE);
+	}
+
+	if (inode->i_state & I_NEW)
+		unlock_new_inode(inode);
+
 	return inode;
 
 fail:
@@ -207,23 +227,26 @@ fail:
 	return ERR_PTR(error);
 }
 
+/**
+ * gfs2_lookup_by_inum - look up an inode by inode number
+ * @sdp: The super block
+ * @no_addr: The inode number
+ * @no_formal_ino: The inode generation number (0 for any)
+ * @blktype: Requested block type (see gfs2_inode_lookup)
+ */
 struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
-				  u64 *no_formal_ino, unsigned int blktype)
+				  u64 no_formal_ino, unsigned int blktype)
 {
 	struct super_block *sb = sdp->sd_vfs;
 	struct inode *inode;
 	int error;
 
-	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, 0, blktype);
+	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, no_formal_ino,
+				  blktype);
 	if (IS_ERR(inode))
 		return inode;
 
-	/* Two extra checks for NFS only */
 	if (no_formal_ino) {
-		error = -ESTALE;
-		if (GFS2_I(inode)->i_no_formal_ino != *no_formal_ino)
-			goto fail_iput;
-
 		error = -EIO;
 		if (GFS2_I(inode)->i_diskflags & GFS2_DIF_SYSTEM)
 			goto fail_iput;
@@ -725,6 +748,7 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry,
 	if (error)
 		goto fail_gunlock2;
 
+	gfs2_cancel_delete_work(ip->i_iopen_gh.gh_gl);
 	glock_set_object(ip->i_iopen_gh.gh_gl, ip);
 	gfs2_set_iop(inode);
 	insert_inode_hash(inode);
@@ -781,7 +805,8 @@ fail_gunlock2:
 fail_free_inode:
 	if (ip->i_gl) {
 		glock_clear_object(ip->i_gl, ip);
-		gfs2_glock_put(ip->i_gl);
+		if (free_vfs_inode) /* else evict will do the put for us */
+			gfs2_glock_put(ip->i_gl);
 	}
 	gfs2_rs_delete(ip, NULL);
 	gfs2_qa_put(ip);

fs/gfs2/inode.h

@@ -92,7 +92,7 @@ extern struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned type,
 				       u64 no_addr, u64 no_formal_ino,
 				       unsigned int blktype);
 extern struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
-					 u64 *no_formal_ino,
+					 u64 no_formal_ino,
 					 unsigned int blktype);
 
 extern int gfs2_inode_refresh(struct gfs2_inode *ip);

fs/gfs2/log.c

@@ -30,6 +30,7 @@
 #include "util.h"
 #include "dir.h"
 #include "trace_gfs2.h"
+#include "trans.h"
 
 static void gfs2_log_shutdown(struct gfs2_sbd *sdp);
 
@@ -145,9 +146,6 @@ static void dump_ail_list(struct gfs2_sbd *sdp)
 	struct gfs2_bufdata *bd;
 	struct buffer_head *bh;
 
-	fs_err(sdp, "Error: In gfs2_ail1_flush for ten minutes! t=%d\n",
-	       current->journal_info ? 1 : 0);
-
 	list_for_each_entry_reverse(tr, &sdp->sd_ail1_list, tr_list) {
 		list_for_each_entry_reverse(bd, &tr->tr_ail1_list,
 					    bd_ail_st_list) {
@@ -197,6 +195,8 @@ void gfs2_ail1_flush(struct gfs2_sbd *sdp, struct writeback_control *wbc)
 restart:
 	ret = 0;
 	if (time_after(jiffies, flush_start + (HZ * 600))) {
+		fs_err(sdp, "Error: In %s for ten minutes! t=%d\n",
+		       __func__, current->journal_info ? 1 : 0);
 		dump_ail_list(sdp);
 		goto out;
 	}
@@ -379,7 +379,7 @@ static void ail2_empty(struct gfs2_sbd *sdp, unsigned int new_tail)
 		list_del(&tr->tr_list);
 		gfs2_assert_warn(sdp, list_empty(&tr->tr_ail1_list));
 		gfs2_assert_warn(sdp, list_empty(&tr->tr_ail2_list));
-		kfree(tr);
+		gfs2_trans_free(sdp, tr);
 	}
 
 	spin_unlock(&sdp->sd_ail_lock);
@@ -864,18 +864,40 @@ static void ail_drain(struct gfs2_sbd *sdp)
 		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail1_list);
 		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail2_list);
 		list_del(&tr->tr_list);
-		kfree(tr);
+		gfs2_trans_free(sdp, tr);
 	}
 	while (!list_empty(&sdp->sd_ail2_list)) {
 		tr = list_first_entry(&sdp->sd_ail2_list, struct gfs2_trans,
 				      tr_list);
 		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail2_list);
 		list_del(&tr->tr_list);
-		kfree(tr);
+		gfs2_trans_free(sdp, tr);
 	}
 	spin_unlock(&sdp->sd_ail_lock);
 }
 
+/**
+ * empty_ail1_list - try to start IO and empty the ail1 list
+ * @sdp: Pointer to GFS2 superblock
+ */
+static void empty_ail1_list(struct gfs2_sbd *sdp)
+{
+	unsigned long start = jiffies;
+
+	for (;;) {
+		if (time_after(jiffies, start + (HZ * 600))) {
+			fs_err(sdp, "Error: In %s for 10 minutes! t=%d\n",
+			       __func__, current->journal_info ? 1 : 0);
+			dump_ail_list(sdp);
+			return;
+		}
+		gfs2_ail1_start(sdp);
+		gfs2_ail1_wait(sdp);
+		if (gfs2_ail1_empty(sdp, 0))
+			return;
+	}
+}
+
 /**
  * gfs2_log_flush - flush incore transaction(s)
  * @sdp: the filesystem
@@ -912,8 +934,6 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 	tr = sdp->sd_log_tr;
 	if (tr) {
 		sdp->sd_log_tr = NULL;
-		INIT_LIST_HEAD(&tr->tr_ail1_list);
-		INIT_LIST_HEAD(&tr->tr_ail2_list);
 		tr->tr_first = sdp->sd_log_flush_head;
 		if (unlikely (state == SFS_FROZEN))
 			if (gfs2_assert_withdraw_delayed(sdp,
@@ -965,12 +985,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 
 	if (!(flags & GFS2_LOG_HEAD_FLUSH_NORMAL)) {
 		if (!sdp->sd_log_idle) {
-			for (;;) {
+			empty_ail1_list(sdp);
-				gfs2_ail1_start(sdp);
-				gfs2_ail1_wait(sdp);
-				if (gfs2_ail1_empty(sdp, 0))
-					break;
-			}
 			if (gfs2_withdrawn(sdp))
 				goto out;
 			atomic_dec(&sdp->sd_log_blks_free); /* Adjust for unreserved buffer */
@@ -994,7 +1009,7 @@ out:
 	trace_gfs2_log_flush(sdp, 0, flags);
 	up_write(&sdp->sd_log_flush_lock);
 
-	kfree(tr);
+	gfs2_trans_free(sdp, tr);
 }
 
 /**
@@ -1003,8 +1018,10 @@ out:
  * @new: New transaction to be merged
  */
 
-static void gfs2_merge_trans(struct gfs2_trans *old, struct gfs2_trans *new)
+static void gfs2_merge_trans(struct gfs2_sbd *sdp, struct gfs2_trans *new)
 {
+	struct gfs2_trans *old = sdp->sd_log_tr;
+
 	WARN_ON_ONCE(!test_bit(TR_ATTACHED, &old->tr_flags));
 
 	old->tr_num_buf_new	+= new->tr_num_buf_new;
@@ -1016,6 +1033,11 @@ static void gfs2_merge_trans(struct gfs2_trans *old, struct gfs2_trans *new)
 
 	list_splice_tail_init(&new->tr_databuf, &old->tr_databuf);
 	list_splice_tail_init(&new->tr_buf, &old->tr_buf);
+
+	spin_lock(&sdp->sd_ail_lock);
+	list_splice_tail_init(&new->tr_ail1_list, &old->tr_ail1_list);
+	list_splice_tail_init(&new->tr_ail2_list, &old->tr_ail2_list);
+	spin_unlock(&sdp->sd_ail_lock);
 }
 
 static void log_refund(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
@@ -1027,7 +1049,7 @@ static void log_refund(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
 	gfs2_log_lock(sdp);
 
 	if (sdp->sd_log_tr) {
-		gfs2_merge_trans(sdp->sd_log_tr, tr);
+		gfs2_merge_trans(sdp, tr);
 	} else if (tr->tr_num_buf_new || tr->tr_num_databuf_new) {
 		gfs2_assert_withdraw(sdp, test_bit(TR_ALLOCED, &tr->tr_flags));
 		sdp->sd_log_tr = tr;

fs/gfs2/main.c

@@ -143,6 +143,12 @@ static int __init init_gfs2_fs(void)
 	if (!gfs2_qadata_cachep)
 		goto fail_cachep7;
 
+	gfs2_trans_cachep = kmem_cache_create("gfs2_trans",
+					       sizeof(struct gfs2_trans),
+					       0, 0, NULL);
+	if (!gfs2_trans_cachep)
+		goto fail_cachep8;
+
 	error = register_shrinker(&gfs2_qd_shrinker);
 	if (error)
 		goto fail_shrinker;
@@ -194,6 +200,8 @@ fail_fs2:
 fail_fs1:
 	unregister_shrinker(&gfs2_qd_shrinker);
 fail_shrinker:
+	kmem_cache_destroy(gfs2_trans_cachep);
+fail_cachep8:
 	kmem_cache_destroy(gfs2_qadata_cachep);
 fail_cachep7:
 	kmem_cache_destroy(gfs2_quotad_cachep);
@@ -236,6 +244,7 @@ static void __exit exit_gfs2_fs(void)
 	rcu_barrier();
 
 	mempool_destroy(gfs2_page_pool);
+	kmem_cache_destroy(gfs2_trans_cachep);
 	kmem_cache_destroy(gfs2_qadata_cachep);
 	kmem_cache_destroy(gfs2_quotad_cachep);
 	kmem_cache_destroy(gfs2_rgrpd_cachep);

fs/gfs2/ops_fstype.c

@@ -880,7 +880,7 @@ fail:
 }
 
 static const match_table_t nolock_tokens = {
-	{ Opt_jid, "jid=%d\n", },
+	{ Opt_jid, "jid=%d", },
 	{ Opt_err, NULL },
 };
 

fs/gfs2/rgrp.c

@@ -1835,7 +1835,7 @@ static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip
 		 */
 		ip = gl->gl_object;
 
-		if (ip || queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0)
+		if (ip || !gfs2_queue_delete_work(gl, 0))
 			gfs2_glock_put(gl);
 		else
 			found++;

fs/gfs2/super.c

@@ -626,7 +626,7 @@ int gfs2_make_fs_ro(struct gfs2_sbd *sdp)
 		}
 	}
 
-	flush_workqueue(gfs2_delete_workqueue);
+	gfs2_flush_delete_work(sdp);
 	if (!log_write_allowed && current == sdp->sd_quotad_process)
 		fs_warn(sdp, "The quotad daemon is withdrawing.\n");
 	else if (sdp->sd_quotad_process)
@@ -1054,7 +1054,7 @@ static int gfs2_drop_inode(struct inode *inode)
 		struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl;
 
 		gfs2_glock_hold(gl);
-		if (queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0)
+		if (!gfs2_queue_delete_work(gl, 0))
 			gfs2_glock_queue_put(gl);
 		return false;
 	}
@@ -1258,6 +1258,55 @@ static void gfs2_glock_put_eventually(struct gfs2_glock *gl)
 		gfs2_glock_put(gl);
 }
 
+static bool gfs2_upgrade_iopen_glock(struct inode *inode)
+{
+	struct gfs2_inode *ip = GFS2_I(inode);
+	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	struct gfs2_holder *gh = &ip->i_iopen_gh;
+	long timeout = 5 * HZ;
+	int error;
+
+	gh->gh_flags |= GL_NOCACHE;
+	gfs2_glock_dq_wait(gh);
+
+	/*
+	 * If there are no other lock holders, we'll get the lock immediately.
+	 * Otherwise, the other nodes holding the lock will be notified about
+	 * our locking request.  If they don't have the inode open, they'll
+	 * evict the cached inode and release the lock.  Otherwise, if they
+	 * poke the inode glock, we'll take this as an indication that they
+	 * still need the iopen glock and that they'll take care of deleting
+	 * the inode when they're done.  As a last resort, if another node
+	 * keeps holding the iopen glock without showing any activity on the
+	 * inode glock, we'll eventually time out.
+	 *
+	 * Note that we're passing the LM_FLAG_TRY_1CB flag to the first
+	 * locking request as an optimization to notify lock holders as soon as
+	 * possible.  Without that flag, they'd be notified implicitly by the
+	 * second locking request.
+	 */
+
+	gfs2_holder_reinit(LM_ST_EXCLUSIVE, LM_FLAG_TRY_1CB | GL_NOCACHE, gh);
+	error = gfs2_glock_nq(gh);
+	if (error != GLR_TRYFAILED)
+		return !error;
+
+	gfs2_holder_reinit(LM_ST_EXCLUSIVE, GL_ASYNC | GL_NOCACHE, gh);
+	error = gfs2_glock_nq(gh);
+	if (error)
+		return false;
+
+	timeout = wait_event_interruptible_timeout(sdp->sd_async_glock_wait,
+		!test_bit(HIF_WAIT, &gh->gh_iflags) ||
+		test_bit(GLF_DEMOTE, &ip->i_gl->gl_flags),
+		timeout);
+	if (!test_bit(HIF_HOLDER, &gh->gh_iflags)) {
+		gfs2_glock_dq(gh);
+		return false;
+	}
+	return true;
+}
+
 /**
  * gfs2_evict_inode - Remove an inode from cache
  * @inode: The inode to evict
@@ -1299,9 +1348,12 @@ static void gfs2_evict_inode(struct inode *inode)
 	if (test_bit(GIF_ALLOC_FAILED, &ip->i_flags)) {
 		BUG_ON(!gfs2_glock_is_locked_by_me(ip->i_gl));
 		gfs2_holder_mark_uninitialized(&gh);
-		goto alloc_failed;
+		goto out_delete;
 	}
 
+	if (test_bit(GIF_DEFERRED_DELETE, &ip->i_flags))
+		goto out;
+
 	/* Deletes should never happen under memory pressure anymore.  */
 	if (WARN_ON_ONCE(current->flags & PF_MEMALLOC))
 		goto out;
@@ -1315,6 +1367,8 @@ static void gfs2_evict_inode(struct inode *inode)
 		goto out;
 	}
 
+	if (gfs2_inode_already_deleted(ip->i_gl, ip->i_no_formal_ino))
+		goto out_truncate;
 	error = gfs2_check_blk_type(sdp, ip->i_no_addr, GFS2_BLKST_UNLINKED);
 	if (error)
 		goto out_truncate;
@@ -1331,16 +1385,13 @@ static void gfs2_evict_inode(struct inode *inode)
 	if (inode->i_nlink)
 		goto out_truncate;
 
-alloc_failed:
+out_delete:
 	if (gfs2_holder_initialized(&ip->i_iopen_gh) &&
 	    test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) {
-		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
+		if (!gfs2_upgrade_iopen_glock(inode)) {
-		gfs2_glock_dq_wait(&ip->i_iopen_gh);
+			gfs2_holder_uninit(&ip->i_iopen_gh);
-		gfs2_holder_reinit(LM_ST_EXCLUSIVE, LM_FLAG_TRY_1CB | GL_NOCACHE,
-				   &ip->i_iopen_gh);
-		error = gfs2_glock_nq(&ip->i_iopen_gh);
-		if (error)
 			goto out_truncate;
+		}
 	}
 
 	if (S_ISDIR(inode->i_mode) &&
@@ -1368,6 +1419,7 @@ alloc_failed:
 	   that subsequent inode creates don't see an old gl_object. */
 	glock_clear_object(ip->i_gl, ip);
 	error = gfs2_dinode_dealloc(ip);
+	gfs2_inode_remember_delete(ip->i_gl, ip->i_no_formal_ino);
 	goto out_unlock;
 
 out_truncate:

fs/gfs2/trans.c

@@ -37,7 +37,7 @@ int gfs2_trans_begin(struct gfs2_sbd *sdp, unsigned int blocks,
 	if (!test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags))
 		return -EROFS;
 
-	tr = kzalloc(sizeof(struct gfs2_trans), GFP_NOFS);
+	tr = kmem_cache_zalloc(gfs2_trans_cachep, GFP_NOFS);
 	if (!tr)
 		return -ENOMEM;
 
@@ -52,6 +52,8 @@ int gfs2_trans_begin(struct gfs2_sbd *sdp, unsigned int blocks,
 		tr->tr_reserved += gfs2_struct2blk(sdp, revokes);
 	INIT_LIST_HEAD(&tr->tr_databuf);
 	INIT_LIST_HEAD(&tr->tr_buf);
+	INIT_LIST_HEAD(&tr->tr_ail1_list);
+	INIT_LIST_HEAD(&tr->tr_ail2_list);
 
 	sb_start_intwrite(sdp->sd_vfs);
 
@@ -65,7 +67,7 @@ int gfs2_trans_begin(struct gfs2_sbd *sdp, unsigned int blocks,
 
 fail:
 	sb_end_intwrite(sdp->sd_vfs);
-	kfree(tr);
+	kmem_cache_free(gfs2_trans_cachep, tr);
 
 	return error;
 }
@@ -93,7 +95,7 @@ void gfs2_trans_end(struct gfs2_sbd *sdp)
 	if (!test_bit(TR_TOUCHED, &tr->tr_flags)) {
 		gfs2_log_release(sdp, tr->tr_reserved);
 		if (alloced) {
-			kfree(tr);
+			gfs2_trans_free(sdp, tr);
 			sb_end_intwrite(sdp->sd_vfs);
 		}
 		return;
@@ -109,7 +111,7 @@ void gfs2_trans_end(struct gfs2_sbd *sdp)
 
 	gfs2_log_commit(sdp, tr);
 	if (alloced && !test_bit(TR_ATTACHED, &tr->tr_flags))
-		kfree(tr);
+		gfs2_trans_free(sdp, tr);
 	up_read(&sdp->sd_log_flush_lock);
 
 	if (sdp->sd_vfs->s_flags & SB_SYNCHRONOUS)
@@ -276,3 +278,14 @@ void gfs2_trans_remove_revoke(struct gfs2_sbd *sdp, u64 blkno, unsigned int len)
 	gfs2_log_unlock(sdp);
 }
 
+void gfs2_trans_free(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
+{
+	if (tr == NULL)
+		return;
+
+	gfs2_assert_warn(sdp, list_empty(&tr->tr_ail1_list));
+	gfs2_assert_warn(sdp, list_empty(&tr->tr_ail2_list));
+	gfs2_assert_warn(sdp, list_empty(&tr->tr_databuf));
+	gfs2_assert_warn(sdp, list_empty(&tr->tr_buf));
+	kmem_cache_free(gfs2_trans_cachep, tr);
+}

fs/gfs2/trans.h

@@ -42,5 +42,6 @@ extern void gfs2_trans_add_data(struct gfs2_glock *gl, struct buffer_head *bh);
 extern void gfs2_trans_add_meta(struct gfs2_glock *gl, struct buffer_head *bh);
 extern void gfs2_trans_add_revoke(struct gfs2_sbd *sdp, struct gfs2_bufdata *bd);
 extern void gfs2_trans_remove_revoke(struct gfs2_sbd *sdp, u64 blkno, unsigned int len);
+extern void gfs2_trans_free(struct gfs2_sbd *sdp, struct gfs2_trans *tr);
 
 #endif /* __TRANS_DOT_H__ */

fs/gfs2/util.c

@@ -32,6 +32,7 @@ struct kmem_cache *gfs2_bufdata_cachep __read_mostly;
 struct kmem_cache *gfs2_rgrpd_cachep __read_mostly;
 struct kmem_cache *gfs2_quotad_cachep __read_mostly;
 struct kmem_cache *gfs2_qadata_cachep __read_mostly;
+struct kmem_cache *gfs2_trans_cachep __read_mostly;
 mempool_t *gfs2_page_pool __read_mostly;
 
 void gfs2_assert_i(struct gfs2_sbd *sdp)

fs/gfs2/util.h

@@ -172,6 +172,7 @@ extern struct kmem_cache *gfs2_bufdata_cachep;
 extern struct kmem_cache *gfs2_rgrpd_cachep;
 extern struct kmem_cache *gfs2_quotad_cachep;
 extern struct kmem_cache *gfs2_qadata_cachep;
+extern struct kmem_cache *gfs2_trans_cachep;
 extern mempool_t *gfs2_page_pool;
 extern struct workqueue_struct *gfs2_control_wq;
 

include/uapi/linux/gfs2_ondisk.h

@@ -171,6 +171,12 @@ struct gfs2_rindex {
 #define GFS2_RGF_NOALLOC	0x00000008
 #define GFS2_RGF_TRIMMED	0x00000010
 
+struct gfs2_inode_lvb {
+	__be32 ri_magic;
+	__be32 __pad;
+	__be64 ri_generation_deleted;
+};
+
 struct gfs2_rgrp_lvb {
 	__be32 rl_magic;
 	__be32 rl_flags;