[Cluster-devel] GFS2: Pre-pull patch posting (fixes)

[Cluster-devel] GFS2: Pre-pull patch posting (fixes)

[Steven Whitehouse]

Hi,

Here is the current content of the GFS2 -fixes git tree. Nothing
very exciting this time... some fixes for issues we've
had relating to flushing glocks/memory usage, plus a couple of
other fixes relating to statfs and the timely removal of inodes
which have been unlinked on a remote node,

Steve.

[Cluster-devel] GFS2: Pull request (fixes)

[Steven Whitehouse]

Hi,

Please consider pulling the following changes,

Steve.

----------------------------------------------------------------------------
The following changes since commit 4be3bd7849165e7efa6b0b35a23d6a3598d97465:
  Linus Torvalds (1):
        Linux 2.6.31-rc4

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes.git master

Benjamin Marzinski (5):
      GFS2: keep statfs info in sync on grows
      GFS2: Fix panic in glock memory shrinker
      GFS2: Don't put unlikely reclaim candidates on the reclaim list.
      GFS2: Fix incorrent statfs consistency check
      GFS2: remove dcache entries for remote deleted inodes

Steven Whitehouse (2):
      GFS2: Shrink the shrinker
      GFS2: Don't try and dealloc own inode

 fs/gfs2/aops.c   |   39 +++++++++++++++
 fs/gfs2/glock.c  |  138 +++++++++++++++++++++++++++++++++++------------------
 fs/gfs2/glock.h  |    3 +
 fs/gfs2/glops.c  |   21 ++++++++
 fs/gfs2/incore.h |    2 +
 fs/gfs2/rgrp.c   |   23 ++++------
 fs/gfs2/super.c  |   40 ++++++++++------
 fs/gfs2/super.h  |    4 ++
 8 files changed, 195 insertions(+), 75 deletions(-)

[Cluster-devel] [PATCH 1/7] GFS2: Shrink the shrinker

[Steven Whitehouse]

This patch removes some of the special cases that the shrinker
was trying to deal with. As a result we leave fewer items on
the list and none at all which cannot be demoted. This makes
the list scanning more efficient and solves some issues seen
with large numbers of inodes.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/glock.c |   23 +++++------------------
 1 files changed, 5 insertions(+), 18 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 297421c..fdb796c 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -1300,7 +1300,6 @@ static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 	struct gfs2_glock *gl;
 	int may_demote;
 	int nr_skipped = 0;
-	int got_ref = 0;
 	LIST_HEAD(skipped);
 
 	if (nr == 0)
@@ -1318,7 +1317,6 @@ static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 		/* Test for being demotable */
 		if (!test_and_set_bit(GLF_LOCK, &gl->gl_flags)) {
 			gfs2_glock_hold(gl);
-			got_ref = 1;
 			spin_unlock(&lru_lock);
 			spin_lock(&gl->gl_spin);
 			may_demote = demote_ok(gl);
@@ -1327,25 +1325,14 @@ static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 			if (may_demote) {
 				handle_callback(gl, LM_ST_UNLOCKED, 0);
 				nr--;
-				if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
-					gfs2_glock_put(gl);
-				got_ref = 0;
 			}
+			if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
+				gfs2_glock_put(gl);
 			spin_lock(&lru_lock);
-			if (may_demote)
-				continue;
-		}
-		if (list_empty(&gl->gl_lru) &&
-		    (atomic_read(&gl->gl_ref) <= (2 + got_ref))) {
-			nr_skipped++;
-			list_add(&gl->gl_lru, &skipped);
-		}
-		if (got_ref) {
-			spin_unlock(&lru_lock);
-			gfs2_glock_put(gl);
-			spin_lock(&lru_lock);
-			got_ref = 0;
+			continue;
 		}
+		nr_skipped++;
+		list_add(&gl->gl_lru, &skipped);
 	}
 	list_splice(&skipped, &lru_list);
 	atomic_add(nr_skipped, &lru_count);
-- 
1.6.2.2

[Cluster-devel] [PATCH 2/7] GFS2: keep statfs info in sync on grows

[Steven Whitehouse]

From: Benjamin Marzinski <bmarzins@redhat.com>

GFS2 wasn't syncing its statfs info on grows.  This causes a problem
when you grow the filesystem on multiple nodes.  GFS2 would calculate
the new space based on the resource groups (which are always current),
and then assume that the filesystem had grown the from the existing
statfs size.  If you grew the filesystem on two different nodes in a
short time, the second node wouldn't see the statfs size change from the
first node, and would assume that it was grown by a larger amount than
it was.  When all these changes were synced out, the total fileystem
size would be incorrect (the first grow would be counted twice).

This patch syncs makes GFS2 read in the statfs changes from disk before
a grow, and write them out after the grow, while the master statfs inode
is locked.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/aops.c  |   39 +++++++++++++++++++++++++++++++++++++++
 fs/gfs2/super.c |   39 +++++++++++++++++++++++++--------------
 fs/gfs2/super.h |    4 ++++
 3 files changed, 68 insertions(+), 14 deletions(-)

diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c
index 03ebb43..7ebae9a 100644
--- a/fs/gfs2/aops.c
+++ b/fs/gfs2/aops.c
@@ -624,6 +624,7 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 {
 	struct gfs2_inode *ip = GFS2_I(mapping->host);
 	struct gfs2_sbd *sdp = GFS2_SB(mapping->host);
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
 	unsigned int data_blocks = 0, ind_blocks = 0, rblocks;
 	int alloc_required;
 	int error = 0;
@@ -637,6 +638,14 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 	error = gfs2_glock_nq(&ip->i_gh);
 	if (unlikely(error))
 		goto out_uninit;
+	if (&ip->i_inode == sdp->sd_rindex) {
+		error = gfs2_glock_nq_init(m_ip->i_gl, LM_ST_EXCLUSIVE,
+					   GL_NOCACHE, &m_ip->i_gh);
+		if (unlikely(error)) {
+			gfs2_glock_dq(&ip->i_gh);
+			goto out_uninit;
+		}
+	}
 
 	error = gfs2_write_alloc_required(ip, pos, len, &alloc_required);
 	if (error)
@@ -667,6 +676,8 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 		rblocks += data_blocks ? data_blocks : 1;
 	if (ind_blocks || data_blocks)
 		rblocks += RES_STATFS + RES_QUOTA;
+	if (&ip->i_inode == sdp->sd_rindex)
+		rblocks += 2 * RES_STATFS;
 
 	error = gfs2_trans_begin(sdp, rblocks,
 				 PAGE_CACHE_SIZE/sdp->sd_sb.sb_bsize);
@@ -712,6 +723,10 @@ out_alloc_put:
 		gfs2_alloc_put(ip);
 	}
 out_unlock:
+	if (&ip->i_inode == sdp->sd_rindex) {
+		gfs2_glock_dq(&m_ip->i_gh);
+		gfs2_holder_uninit(&m_ip->i_gh);
+	}
 	gfs2_glock_dq(&ip->i_gh);
 out_uninit:
 	gfs2_holder_uninit(&ip->i_gh);
@@ -725,14 +740,21 @@ out_uninit:
 static void adjust_fs_space(struct inode *inode)
 {
 	struct gfs2_sbd *sdp = inode->i_sb->s_fs_info;
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
+	struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode);
 	struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master;
 	struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local;
+	struct buffer_head *m_bh, *l_bh;
 	u64 fs_total, new_free;
 
 	/* Total up the file system space, according to the latest rindex. */
 	fs_total = gfs2_ri_total(sdp);
+	if (gfs2_meta_inode_buffer(m_ip, &m_bh) != 0)
+		return;
 
 	spin_lock(&sdp->sd_statfs_spin);
+	gfs2_statfs_change_in(m_sc, m_bh->b_data +
+			      sizeof(struct gfs2_dinode));
 	if (fs_total > (m_sc->sc_total + l_sc->sc_total))
 		new_free = fs_total - (m_sc->sc_total + l_sc->sc_total);
 	else
@@ -741,6 +763,13 @@ static void adjust_fs_space(struct inode *inode)
 	fs_warn(sdp, "File system extended by %llu blocks.\n",
 		(unsigned long long)new_free);
 	gfs2_statfs_change(sdp, new_free, new_free, 0);
+
+	if (gfs2_meta_inode_buffer(l_ip, &l_bh) != 0)
+		goto out;
+	update_statfs(sdp, m_bh, l_bh);
+	brelse(l_bh);
+out:
+	brelse(m_bh);
 }
 
 /**
@@ -763,6 +792,7 @@ static int gfs2_stuffed_write_end(struct inode *inode, struct buffer_head *dibh,
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
 	u64 to = pos + copied;
 	void *kaddr;
 	unsigned char *buf = dibh->b_data + sizeof(struct gfs2_dinode);
@@ -794,6 +824,10 @@ static int gfs2_stuffed_write_end(struct inode *inode, struct buffer_head *dibh,
 
 	brelse(dibh);
 	gfs2_trans_end(sdp);
+	if (inode == sdp->sd_rindex) {
+		gfs2_glock_dq(&m_ip->i_gh);
+		gfs2_holder_uninit(&m_ip->i_gh);
+	}
 	gfs2_glock_dq(&ip->i_gh);
 	gfs2_holder_uninit(&ip->i_gh);
 	return copied;
@@ -823,6 +857,7 @@ static int gfs2_write_end(struct file *file, struct address_space *mapping,
 	struct inode *inode = page->mapping->host;
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
 	struct buffer_head *dibh;
 	struct gfs2_alloc *al = ip->i_alloc;
 	unsigned int from = pos & (PAGE_CACHE_SIZE - 1);
@@ -865,6 +900,10 @@ failed:
 		gfs2_quota_unlock(ip);
 		gfs2_alloc_put(ip);
 	}
+	if (inode == sdp->sd_rindex) {
+		gfs2_glock_dq(&m_ip->i_gh);
+		gfs2_holder_uninit(&m_ip->i_gh);
+	}
 	gfs2_glock_dq(&ip->i_gh);
 	gfs2_holder_uninit(&ip->i_gh);
 	return ret;
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 0a68013..552e321 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -353,7 +353,7 @@ fail:
 	return error;
 }
 
-static void gfs2_statfs_change_in(struct gfs2_statfs_change_host *sc, const void *buf)
+void gfs2_statfs_change_in(struct gfs2_statfs_change_host *sc, const void *buf)
 {
 	const struct gfs2_statfs_change *str = buf;
 
@@ -441,6 +441,29 @@ void gfs2_statfs_change(struct gfs2_sbd *sdp, s64 total, s64 free,
 	brelse(l_bh);
 }
 
+void update_statfs(struct gfs2_sbd *sdp, struct buffer_head *m_bh,
+		   struct buffer_head *l_bh)
+{
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
+	struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode);
+	struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master;
+	struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local;
+
+	gfs2_trans_add_bh(l_ip->i_gl, l_bh, 1);
+
+	spin_lock(&sdp->sd_statfs_spin);
+	m_sc->sc_total += l_sc->sc_total;
+	m_sc->sc_free += l_sc->sc_free;
+	m_sc->sc_dinodes += l_sc->sc_dinodes;
+	memset(l_sc, 0, sizeof(struct gfs2_statfs_change));
+	memset(l_bh->b_data + sizeof(struct gfs2_dinode),
+	       0, sizeof(struct gfs2_statfs_change));
+	spin_unlock(&sdp->sd_statfs_spin);
+
+	gfs2_trans_add_bh(m_ip->i_gl, m_bh, 1);
+	gfs2_statfs_change_out(m_sc, m_bh->b_data + sizeof(struct gfs2_dinode));
+}
+
 int gfs2_statfs_sync(struct gfs2_sbd *sdp)
 {
 	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
@@ -477,19 +500,7 @@ int gfs2_statfs_sync(struct gfs2_sbd *sdp)
 	if (error)
 		goto out_bh2;
 
-	gfs2_trans_add_bh(l_ip->i_gl, l_bh, 1);
-
-	spin_lock(&sdp->sd_statfs_spin);
-	m_sc->sc_total += l_sc->sc_total;
-	m_sc->sc_free += l_sc->sc_free;
-	m_sc->sc_dinodes += l_sc->sc_dinodes;
-	memset(l_sc, 0, sizeof(struct gfs2_statfs_change));
-	memset(l_bh->b_data + sizeof(struct gfs2_dinode),
-	       0, sizeof(struct gfs2_statfs_change));
-	spin_unlock(&sdp->sd_statfs_spin);
-
-	gfs2_trans_add_bh(m_ip->i_gl, m_bh, 1);
-	gfs2_statfs_change_out(m_sc, m_bh->b_data + sizeof(struct gfs2_dinode));
+	update_statfs(sdp, m_bh, l_bh);
 
 	gfs2_trans_end(sdp);
 
diff --git a/fs/gfs2/super.h b/fs/gfs2/super.h
index b56413e..22e0417 100644
--- a/fs/gfs2/super.h
+++ b/fs/gfs2/super.h
@@ -40,6 +40,10 @@ extern int gfs2_make_fs_rw(struct gfs2_sbd *sdp);
 extern int gfs2_statfs_init(struct gfs2_sbd *sdp);
 extern void gfs2_statfs_change(struct gfs2_sbd *sdp, s64 total, s64 free,
 			       s64 dinodes);
+extern void gfs2_statfs_change_in(struct gfs2_statfs_change_host *sc,
+				  const void *buf);
+extern void update_statfs(struct gfs2_sbd *sdp, struct buffer_head *m_bh,
+			  struct buffer_head *l_bh);
 extern int gfs2_statfs_sync(struct gfs2_sbd *sdp);
 
 extern int gfs2_freeze_fs(struct gfs2_sbd *sdp);
-- 
1.6.2.2

[Cluster-devel] [PATCH 3/7] GFS2: Fix panic in glock memory shrinker

[Steven Whitehouse]

From: Benjamin Marzinski <bmarzins@redhat.com>

It is possible for gfs2_shrink_glock_memory() to check a glock for
demotion
that's in the process of being freed by gfs2_glock_put().  In this case,
gfs2_shrink_glock_memory() will acquire a new reference to this glock,
and
then try to free the glock itself when it drops the refernce.  To solve
this, gfs2_shrink_glock_memory() just needs to check if the glock is in
the process of being freed, and if so skip it without ever unlocking the
lru_lock.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Acked-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/glock.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index fdb796c..827136e 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -1314,6 +1314,10 @@ static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 		list_del_init(&gl->gl_lru);
 		atomic_dec(&lru_count);
 
+		/* Check if glock is about to be freed */
+		if (atomic_read(&gl->gl_ref) == 0)
+			continue;
+
 		/* Test for being demotable */
 		if (!test_and_set_bit(GLF_LOCK, &gl->gl_flags)) {
 			gfs2_glock_hold(gl);
-- 
1.6.2.2

[Cluster-devel] [PATCH 4/7] GFS2: Don't try and dealloc own inode

[Steven Whitehouse]

When searching for unlinked, but still allocated inodes during block
allocation, avoid the block relating to the inode that is doing the
allocation. This fixes a hang caused when an unlinked, but still
open, inode tries to allocate some more blocks and lands up
finding itself during the search for deallocatable inodes.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/rgrp.c |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index daa4ae3..5e50741 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -961,7 +961,8 @@ static int try_rgrp_fit(struct gfs2_rgrpd *rgd, struct gfs2_alloc *al)
  * Returns: The inode, if one has been found
  */
 
-static struct inode *try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked)
+static struct inode *try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked,
+				     u64 skip)
 {
 	struct inode *inode;
 	u32 goal = 0, block;
@@ -985,6 +986,8 @@ static struct inode *try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked)
 		goal++;
 		if (*last_unlinked != NO_BLOCK && no_addr <= *last_unlinked)
 			continue;
+		if (no_addr == skip)
+			continue;
 		*last_unlinked = no_addr;
 		inode = gfs2_inode_lookup(rgd->rd_sbd->sd_vfs, DT_UNKNOWN,
 					  no_addr, -1, 1);
@@ -1104,7 +1107,7 @@ static struct inode *get_local_rgrp(struct gfs2_inode *ip, u64 *last_unlinked)
 			if (try_rgrp_fit(rgd, al))
 				goto out;
 			if (rgd->rd_flags & GFS2_RDF_CHECK)
-				inode = try_rgrp_unlink(rgd, last_unlinked);
+				inode = try_rgrp_unlink(rgd, last_unlinked, ip->i_no_addr);
 			if (!rg_locked)
 				gfs2_glock_dq_uninit(&al->al_rgd_gh);
 			if (inode)
@@ -1138,7 +1141,7 @@ static struct inode *get_local_rgrp(struct gfs2_inode *ip, u64 *last_unlinked)
 			if (try_rgrp_fit(rgd, al))
 				goto out;
 			if (rgd->rd_flags & GFS2_RDF_CHECK)
-				inode = try_rgrp_unlink(rgd, last_unlinked);
+				inode = try_rgrp_unlink(rgd, last_unlinked, ip->i_no_addr);
 			if (!rg_locked)
 				gfs2_glock_dq_uninit(&al->al_rgd_gh);
 			if (inode)
-- 
1.6.2.2

[Cluster-devel] [PATCH 5/7] GFS2: Don't put unlikely reclaim candidates on the reclaim list.

[Steven Whitehouse]

From: Benjamin Marzinski <bmarzins@redhat.com>

GFS2 was placing far too many glocks on the reclaim list that were not good
candidates for freeing up from cache.  These locks would sit there and
repeatedly get scanned to see if they could be reclaimed, wasting a lot
of time when there was memory pressure. This fix does more checks on the
locks to see if they are actually likely to be removable from cache.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/glock.c |   72 +++++++++++++++++++++++++++++++++++--------------------
 1 files changed, 46 insertions(+), 26 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 827136e..f041a89 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -174,6 +174,26 @@ static void gfs2_glock_hold(struct gfs2_glock *gl)
 }
 
 /**
+ * demote_ok - Check to see if it's ok to unlock a glock
+ * @gl: the glock
+ *
+ * Returns: 1 if it's ok
+ */
+
+static int demote_ok(const struct gfs2_glock *gl)
+{
+	const struct gfs2_glock_operations *glops = gl->gl_ops;
+
+	if (gl->gl_state == LM_ST_UNLOCKED)
+		return 0;
+	if (!list_empty(&gl->gl_holders))
+		return 0;
+	if (glops->go_demote_ok)
+		return glops->go_demote_ok(gl);
+	return 1;
+}
+
+/**
  * gfs2_glock_schedule_for_reclaim - Add a glock to the reclaim list
  * @gl: the glock
  *
@@ -181,8 +201,13 @@ static void gfs2_glock_hold(struct gfs2_glock *gl)
 
 static void gfs2_glock_schedule_for_reclaim(struct gfs2_glock *gl)
 {
+	int may_reclaim;
+	may_reclaim = (demote_ok(gl) &&
+		       (atomic_read(&gl->gl_ref) == 1 ||
+			(gl->gl_name.ln_type == LM_TYPE_INODE &&
+			 atomic_read(&gl->gl_ref) <= 2)));
 	spin_lock(&lru_lock);
-	if (list_empty(&gl->gl_lru) && gl->gl_state != LM_ST_UNLOCKED) {
+	if (list_empty(&gl->gl_lru) && may_reclaim) {
 		list_add_tail(&gl->gl_lru, &lru_list);
 		atomic_inc(&lru_count);
 	}
@@ -190,6 +215,21 @@ static void gfs2_glock_schedule_for_reclaim(struct gfs2_glock *gl)
 }
 
 /**
+ * gfs2_glock_put_nolock() - Decrement reference count on glock
+ * @gl: The glock to put
+ *
+ * This function should only be used if the caller has its own reference
+ * to the glock, in addition to the one it is dropping.
+ */
+
+static void gfs2_glock_put_nolock(struct gfs2_glock *gl)
+{
+	if (atomic_dec_and_test(&gl->gl_ref))
+		GLOCK_BUG_ON(gl, 1);
+	gfs2_glock_schedule_for_reclaim(gl);
+}
+
+/**
  * gfs2_glock_put() - Decrement reference count on glock
  * @gl: The glock to put
  *
@@ -214,9 +254,9 @@ int gfs2_glock_put(struct gfs2_glock *gl)
 		rv = 1;
 		goto out;
 	}
-	/* 1 for being hashed, 1 for having state != LM_ST_UNLOCKED */
-	if (atomic_read(&gl->gl_ref) == 2)
-		gfs2_glock_schedule_for_reclaim(gl);
+	spin_lock(&gl->gl_spin);
+	gfs2_glock_schedule_for_reclaim(gl);
+	spin_unlock(&gl->gl_spin);
 	write_unlock(gl_lock_addr(gl->gl_hash));
 out:
 	return rv;
@@ -398,7 +438,7 @@ static void state_change(struct gfs2_glock *gl, unsigned int new_state)
 		if (held2)
 			gfs2_glock_hold(gl);
 		else
-			gfs2_glock_put(gl);
+			gfs2_glock_put_nolock(gl);
 	}
 
 	gl->gl_state = new_state;
@@ -633,7 +673,7 @@ out:
 out_sched:
 	gfs2_glock_hold(gl);
 	if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
-		gfs2_glock_put(gl);
+		gfs2_glock_put_nolock(gl);
 out_unlock:
 	clear_bit(GLF_LOCK, &gl->gl_flags);
 	goto out;
@@ -1274,26 +1314,6 @@ void gfs2_glock_complete(struct gfs2_glock *gl, int ret)
 		gfs2_glock_put(gl);
 }
 
-/**
- * demote_ok - Check to see if it's ok to unlock a glock
- * @gl: the glock
- *
- * Returns: 1 if it's ok
- */
-
-static int demote_ok(const struct gfs2_glock *gl)
-{
-	const struct gfs2_glock_operations *glops = gl->gl_ops;
-
-	if (gl->gl_state == LM_ST_UNLOCKED)
-		return 0;
-	if (!list_empty(&gl->gl_holders))
-		return 0;
-	if (glops->go_demote_ok)
-		return glops->go_demote_ok(gl);
-	return 1;
-}
-
 
 static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 {
-- 
1.6.2.2

[Cluster-devel] [PATCH 6/7] GFS2: Fix incorrent statfs consistency check

[Steven Whitehouse]

From: Benjamin Marzinski <bmarzins@redhat.com>

Since both linked and unlinked inodes are counted by rgd->rd_dinodes, It
makes no sense to count them with the used data blocks (first check that
I changed), it makes sense to count them with the linked inodes (second
check), and it makes no sense to care if there are more unlinked inodes
than linked ones. This fixes these errors.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/rgrp.c |   14 +++-----------
 1 files changed, 3 insertions(+), 11 deletions(-)

diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 5e50741..fba7957 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -285,27 +285,19 @@ void gfs2_rgrp_verify(struct gfs2_rgrpd *rgd)
 	}
 
 	tmp = rgd->rd_data - rgd->rd_free - rgd->rd_dinodes;
-	if (count[1] + count[2] != tmp) {
+	if (count[1] != tmp) {
 		if (gfs2_consist_rgrpd(rgd))
 			fs_err(sdp, "used data mismatch:  %u != %u\n",
 			       count[1], tmp);
 		return;
 	}
 
-	if (count[3] != rgd->rd_dinodes) {
+	if (count[2] + count[3] != rgd->rd_dinodes) {
 		if (gfs2_consist_rgrpd(rgd))
 			fs_err(sdp, "used metadata mismatch:  %u != %u\n",
-			       count[3], rgd->rd_dinodes);
+			       count[2] + count[3], rgd->rd_dinodes);
 		return;
 	}
-
-	if (count[2] > count[3]) {
-		if (gfs2_consist_rgrpd(rgd))
-			fs_err(sdp, "unlinked inodes > inodes:  %u\n",
-			       count[2]);
-		return;
-	}
-
 }
 
 static inline int rgrp_contains_block(struct gfs2_rgrpd *rgd, u64 block)
-- 
1.6.2.2

[Cluster-devel] [PATCH 7/7] GFS2: remove dcache entries for remote deleted inodes

[Steven Whitehouse]

From: Benjamin Marzinski <bmarzins@redhat.com>

When a file is deleted from a gfs2 filesystem on one node, a dcache
entry for it may still exist on other nodes in the cluster. If this
happens, gfs2 will be unable to free this file on disk. Because of this,
it's possible to have a gfs2 filesystem with no files on it and no free
space. With this patch, when a node receives a callback notifying it
that the file is being deleted on another node, it schedules a new
workqueue thread to remove the file's dcache entry.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
---
 fs/gfs2/glock.c  |   43 ++++++++++++++++++++++++++++++++++++++-----
 fs/gfs2/glock.h  |    3 +++
 fs/gfs2/glops.c  |   21 +++++++++++++++++++++
 fs/gfs2/incore.h |    2 ++
 fs/gfs2/super.c  |    1 +
 5 files changed, 65 insertions(+), 5 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index f041a89..8b674b1 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -63,6 +63,7 @@ static void do_xmote(struct gfs2_glock *gl, struct gfs2_holder *gh, unsigned int
 static DECLARE_RWSEM(gfs2_umount_flush_sem);
 static struct dentry *gfs2_root;
 static struct workqueue_struct *glock_workqueue;
+struct workqueue_struct *gfs2_delete_workqueue;
 static LIST_HEAD(lru_list);
 static atomic_t lru_count = ATOMIC_INIT(0);
 static DEFINE_SPINLOCK(lru_lock);
@@ -167,7 +168,7 @@ static void glock_free(struct gfs2_glock *gl)
  *
  */
 
-static void gfs2_glock_hold(struct gfs2_glock *gl)
+void gfs2_glock_hold(struct gfs2_glock *gl)
 {
 	GLOCK_BUG_ON(gl, atomic_read(&gl->gl_ref) == 0);
 	atomic_inc(&gl->gl_ref);
@@ -222,7 +223,7 @@ static void gfs2_glock_schedule_for_reclaim(struct gfs2_glock *gl)
  * to the glock, in addition to the one it is dropping.
  */
 
-static void gfs2_glock_put_nolock(struct gfs2_glock *gl)
+void gfs2_glock_put_nolock(struct gfs2_glock *gl)
 {
 	if (atomic_dec_and_test(&gl->gl_ref))
 		GLOCK_BUG_ON(gl, 1);
@@ -679,6 +680,29 @@ out_unlock:
 	goto out;
 }
 
+static void delete_work_func(struct work_struct *work)
+{
+	struct gfs2_glock *gl = container_of(work, struct gfs2_glock, gl_delete);
+	struct gfs2_sbd *sdp = gl->gl_sbd;
+	struct gfs2_inode *ip = NULL;
+	struct inode *inode;
+	u64 no_addr = 0;
+
+	spin_lock(&gl->gl_spin);
+	ip = (struct gfs2_inode *)gl->gl_object;
+	if (ip)
+		no_addr = ip->i_no_addr;
+	spin_unlock(&gl->gl_spin);
+	if (ip) {
+		inode = gfs2_ilookup(sdp->sd_vfs, no_addr);
+		if (inode) {
+			d_prune_aliases(inode);
+			iput(inode);
+		}
+	}
+	gfs2_glock_put(gl);
+}
+
 static void glock_work_func(struct work_struct *work)
 {
 	unsigned long delay = 0;
@@ -757,6 +781,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
 	gl->gl_sbd = sdp;
 	gl->gl_aspace = NULL;
 	INIT_DELAYED_WORK(&gl->gl_work, glock_work_func);
+	INIT_WORK(&gl->gl_delete, delete_work_func);
 
 	/* If this glock protects actual on-disk data or metadata blocks,
 	   create a VFS inode to manage the pages/buffers holding them. */
@@ -898,6 +923,8 @@ static void handle_callback(struct gfs2_glock *gl, unsigned int state,
 			gl->gl_demote_state != state) {
 		gl->gl_demote_state = LM_ST_UNLOCKED;
 	}
+	if (gl->gl_ops->go_callback)
+		gl->gl_ops->go_callback(gl);
 	trace_gfs2_demote_rq(gl);
 }
 
@@ -1344,14 +1371,14 @@ static int gfs2_shrink_glock_memory(int nr, gfp_t gfp_mask)
 			spin_unlock(&lru_lock);
 			spin_lock(&gl->gl_spin);
 			may_demote = demote_ok(gl);
-			spin_unlock(&gl->gl_spin);
-			clear_bit(GLF_LOCK, &gl->gl_flags);
 			if (may_demote) {
 				handle_callback(gl, LM_ST_UNLOCKED, 0);
 				nr--;
 			}
 			if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
-				gfs2_glock_put(gl);
+				gfs2_glock_put_nolock(gl);
+			spin_unlock(&gl->gl_spin);
+			clear_bit(GLF_LOCK, &gl->gl_flags);
 			spin_lock(&lru_lock);
 			continue;
 		}
@@ -1738,6 +1765,11 @@ int __init gfs2_glock_init(void)
 	glock_workqueue = create_workqueue("glock_workqueue");
 	if (IS_ERR(glock_workqueue))
 		return PTR_ERR(glock_workqueue);
+	gfs2_delete_workqueue = create_workqueue("delete_workqueue");
+	if (IS_ERR(gfs2_delete_workqueue)) {
+		destroy_workqueue(glock_workqueue);
+		return PTR_ERR(gfs2_delete_workqueue);
+	}
 
 	register_shrinker(&glock_shrinker);
 
@@ -1748,6 +1780,7 @@ void gfs2_glock_exit(void)
 {
 	unregister_shrinker(&glock_shrinker);
 	destroy_workqueue(glock_workqueue);
+	destroy_workqueue(gfs2_delete_workqueue);
 }
 
 static int gfs2_glock_iter_next(struct gfs2_glock_iter *gi)
diff --git a/fs/gfs2/glock.h b/fs/gfs2/glock.h
index a602a28..c609894 100644
--- a/fs/gfs2/glock.h
+++ b/fs/gfs2/glock.h
@@ -143,6 +143,7 @@ struct lm_lockops {
 
 #define GLR_TRYFAILED		13
 
+extern struct workqueue_struct *gfs2_delete_workqueue;
 static inline struct gfs2_holder *gfs2_glock_is_locked_by_me(struct gfs2_glock *gl)
 {
 	struct gfs2_holder *gh;
@@ -191,6 +192,8 @@ static inline int gfs2_glock_is_blocking(struct gfs2_glock *gl)
 int gfs2_glock_get(struct gfs2_sbd *sdp,
 		   u64 number, const struct gfs2_glock_operations *glops,
 		   int create, struct gfs2_glock **glp);
+void gfs2_glock_hold(struct gfs2_glock *gl);
+void gfs2_glock_put_nolock(struct gfs2_glock *gl);
 int gfs2_glock_put(struct gfs2_glock *gl);
 void gfs2_holder_init(struct gfs2_glock *gl, unsigned int state, unsigned flags,
 		      struct gfs2_holder *gh);
diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c
index d5e4ab1..6985eef 100644
--- a/fs/gfs2/glops.c
+++ b/fs/gfs2/glops.c
@@ -323,6 +323,7 @@ static void trans_go_sync(struct gfs2_glock *gl)
 
 	if (gl->gl_state != LM_ST_UNLOCKED &&
 	    test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags)) {
+		flush_workqueue(gfs2_delete_workqueue);
 		gfs2_meta_syncfs(sdp);
 		gfs2_log_shutdown(sdp);
 	}
@@ -372,6 +373,25 @@ static int trans_go_demote_ok(const struct gfs2_glock *gl)
 	return 0;
 }
 
+/**
+ * iopen_go_callback - schedule the dcache entry for the inode to be deleted
+ * @gl: the glock
+ *
+ * gl_spin lock is held while calling this
+ */
+static void iopen_go_callback(struct gfs2_glock *gl)
+{
+	struct gfs2_inode *ip = (struct gfs2_inode *)gl->gl_object;
+
+	if (gl->gl_demote_state == LM_ST_UNLOCKED &&
+	    gl->gl_state == LM_ST_SHARED &&
+	    ip && test_bit(GIF_USER, &ip->i_flags)) {
+		gfs2_glock_hold(gl);
+		if (queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0)
+			gfs2_glock_put_nolock(gl);
+	}
+}
+
 const struct gfs2_glock_operations gfs2_meta_glops = {
 	.go_type = LM_TYPE_META,
 };
@@ -406,6 +426,7 @@ const struct gfs2_glock_operations gfs2_trans_glops = {
 
 const struct gfs2_glock_operations gfs2_iopen_glops = {
 	.go_type = LM_TYPE_IOPEN,
+	.go_callback = iopen_go_callback,
 };
 
 const struct gfs2_glock_operations gfs2_flock_glops = {
diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h
index 225347f..61801ad 100644
--- a/fs/gfs2/incore.h
+++ b/fs/gfs2/incore.h
@@ -159,6 +159,7 @@ struct gfs2_glock_operations {
 	int (*go_lock) (struct gfs2_holder *gh);
 	void (*go_unlock) (struct gfs2_holder *gh);
 	int (*go_dump)(struct seq_file *seq, const struct gfs2_glock *gl);
+	void (*go_callback) (struct gfs2_glock *gl);
 	const int go_type;
 	const unsigned long go_min_hold_time;
 };
@@ -228,6 +229,7 @@ struct gfs2_glock {
 	struct list_head gl_ail_list;
 	atomic_t gl_ail_count;
 	struct delayed_work gl_work;
+	struct work_struct gl_delete;
 };
 
 #define GFS2_MIN_LVB_SIZE 32	/* Min size of LVB that gfs2 supports */
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 552e321..f522bb0 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -691,6 +691,7 @@ static int gfs2_make_fs_ro(struct gfs2_sbd *sdp)
 	struct gfs2_holder t_gh;
 	int error;
 
+	flush_workqueue(gfs2_delete_workqueue);
 	gfs2_quota_sync(sdp);
 	gfs2_statfs_sync(sdp);
 
-- 
1.6.2.2