All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock
@ 2016-06-20 15:55 ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: linux-fsdevel; +Cc: cluster-devel

Hi,

This is a set of four patches from Andreas Gruenbacher that fix the
gfs2_lookup_by_inum deadlock problem, plus a few bugs we've uncovered
during testing, and a cleanup or two. I've been working with Andreas
for a while now, and we've both made several attempts to fix this
problem in the past, in regard to the transition of dinodes from the
"unlinked" to the "free" state. This is the latest attempt, and it
seems to be working well.

I sent the first 3 patches on 15 June, and Andreas had posted some of
this as well, but I wanted to add the fourth patch, plus cc linux-fsdevel.

Prior to that, a previous version was posted which included a patch to
VFS, but Al Viro didn't like that. That design was scrapped in favor
of this one, which is simpler and confined to GFS2. Hopefully Al will
like this implementation.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
Andreas Gruenbacher (4):
  gfs2: Fix gfs2_lookup_by_inum lock inversion
  gfs2: Get rid of gfs2_ilookup
  gfs2: Large-filesystem fix for 32-bit systems
  gfs2: Lock holder cleanup

 fs/gfs2/dentry.c     |   2 +-
 fs/gfs2/dir.c        |   3 +-
 fs/gfs2/export.c     |  11 -----
 fs/gfs2/file.c       |   2 +-
 fs/gfs2/glock.c      |  11 +----
 fs/gfs2/glock.h      |  10 ++++
 fs/gfs2/inode.c      | 128 ++++++++++++++++++++++++++++++++++-----------------
 fs/gfs2/inode.h      |   4 +-
 fs/gfs2/main.c       |   2 +-
 fs/gfs2/ops_fstype.c |   3 +-
 fs/gfs2/quota.c      |   2 +-
 fs/gfs2/rgrp.c       |   4 +-
 fs/gfs2/super.c      |  24 ++++++----
 13 files changed, 123 insertions(+), 83 deletions(-)

-- 
2.5.5


^ permalink raw reply	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock
@ 2016-06-20 15:55 ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: cluster-devel.redhat.com

Hi,

This is a set of four patches from Andreas Gruenbacher that fix the
gfs2_lookup_by_inum deadlock problem, plus a few bugs we've uncovered
during testing, and a cleanup or two. I've been working with Andreas
for a while now, and we've both made several attempts to fix this
problem in the past, in regard to the transition of dinodes from the
"unlinked" to the "free" state. This is the latest attempt, and it
seems to be working well.

I sent the first 3 patches on 15 June, and Andreas had posted some of
this as well, but I wanted to add the fourth patch, plus cc linux-fsdevel.

Prior to that, a previous version was posted which included a patch to
VFS, but Al Viro didn't like that. That design was scrapped in favor
of this one, which is simpler and confined to GFS2. Hopefully Al will
like this implementation.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
Andreas Gruenbacher (4):
  gfs2: Fix gfs2_lookup_by_inum lock inversion
  gfs2: Get rid of gfs2_ilookup
  gfs2: Large-filesystem fix for 32-bit systems
  gfs2: Lock holder cleanup

 fs/gfs2/dentry.c     |   2 +-
 fs/gfs2/dir.c        |   3 +-
 fs/gfs2/export.c     |  11 -----
 fs/gfs2/file.c       |   2 +-
 fs/gfs2/glock.c      |  11 +----
 fs/gfs2/glock.h      |  10 ++++
 fs/gfs2/inode.c      | 128 ++++++++++++++++++++++++++++++++++-----------------
 fs/gfs2/inode.h      |   4 +-
 fs/gfs2/main.c       |   2 +-
 fs/gfs2/ops_fstype.c |   3 +-
 fs/gfs2/quota.c      |   2 +-
 fs/gfs2/rgrp.c       |   4 +-
 fs/gfs2/super.c      |  24 ++++++----
 13 files changed, 123 insertions(+), 83 deletions(-)

-- 
2.5.5



^ permalink raw reply	[flat|nested] 12+ messages in thread
* [PATCH 1/4] gfs2: Fix gfs2_lookup_by_inum lock inversion
  2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
@ 2016-06-20 15:55   ` Bob Peterson
  -1 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: linux-fsdevel; +Cc: cluster-devel

From: Andreas Gruenbacher <agruenba@redhat.com>

The current gfs2_lookup_by_inum takes the glock of a presumed inode
identified by block number, verifies that the block is indeed an inode,
and then instantiates and reads the new inode via gfs2_inode_lookup.

However, instantiating a new inode may block on freeing a previous
instance of that inode (__wait_on_freeing_inode), and freeing an inode
requires to take the glock already held, leading to lock inversion and
deadlock.

Fix this by first instantiating the new inode, then verifying that the
block is an inode (if required), and then reading in the new inode, all
in gfs2_inode_lookup.

If the block we are looking for is not an inode, we discard the new
inode via iget_failed, which marks inodes as bad and unhashes them.
Other tasks waiting on that inode will get back a bad inode back from
ilookup or iget_locked; in that case, retry the lookup.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/dir.c        |   3 +-
 fs/gfs2/glock.c      |   4 +-
 fs/gfs2/inode.c      | 101 +++++++++++++++++++++++++++++++++++++--------------
 fs/gfs2/inode.h      |   3 +-
 fs/gfs2/ops_fstype.c |   3 +-
 5 files changed, 81 insertions(+), 33 deletions(-)

diff --git a/fs/gfs2/dir.c b/fs/gfs2/dir.c
index 4a01f30..1b02665 100644
--- a/fs/gfs2/dir.c
+++ b/fs/gfs2/dir.c
@@ -1660,7 +1660,8 @@ struct inode *gfs2_dir_search(struct inode *dir, const struct qstr *name,
 		brelse(bh);
 		if (fail_on_exist)
 			return ERR_PTR(-EEXIST);
-		inode = gfs2_inode_lookup(dir->i_sb, dtype, addr, formal_ino);
+		inode = gfs2_inode_lookup(dir->i_sb, dtype, addr, formal_ino,
+					  GFS2_BLKST_FREE /* ignore */);
 		if (!IS_ERR(inode))
 			GFS2_I(inode)->i_rahead = rahead;
 		return inode;
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 706fd93..ce46375 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -576,7 +576,7 @@ 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_name.ln_sbd;
 	struct gfs2_inode *ip;
-	struct inode *inode;
+	struct inode *inode = NULL;
 	u64 no_addr = gl->gl_name.ln_number;
 
 	/* If someone's using this glock to create a new dinode, the block must
@@ -590,7 +590,7 @@ static void delete_work_func(struct work_struct *work)
 
 	if (ip)
 		inode = gfs2_ilookup(sdp->sd_vfs, no_addr);
-	else
+	if (IS_ERR_OR_NULL(inode))
 		inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
 	if (inode && !IS_ERR(inode)) {
 		d_prune_aliases(inode);
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 21dc784..6d5c6bb 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -39,7 +39,33 @@
 
 struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr)
 {
-	return ilookup(sb, (unsigned long)no_addr);
+	struct inode *inode;
+
+repeat:
+	inode = ilookup(sb, no_addr);
+	if (!inode)
+		return inode;
+	if (is_bad_inode(inode)) {
+		iput(inode);
+		goto repeat;
+	}
+	return inode;
+}
+
+static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
+{
+	struct inode *inode;
+
+repeat:
+	inode = iget_locked(sb, no_addr);
+	if (!inode)
+		return inode;
+	if (is_bad_inode(inode)) {
+		iput(inode);
+		goto repeat;
+	}
+	GFS2_I(inode)->i_no_addr = no_addr;
+	return inode;
 }
 
 /**
@@ -78,26 +104,37 @@ static void gfs2_set_iop(struct inode *inode)
 /**
  * gfs2_inode_lookup - Lookup an inode
  * @sb: The super block
- * @no_addr: The inode number
  * @type: The type of the inode
+ * @no_addr: The inode number
+ * @no_formal_ino: The inode generation number
+ * @blktype: Requested block type (GFS2_BLKST_DINODE or GFS2_BLKST_UNLINKED;
+ *           GFS2_BLKST_FREE do indicate not to verify)
+ *
+ * If @type is DT_UNKNOWN, the inode type is fetched from disk.
+ *
+ * If @blktype is anything other than GFS2_BLKST_FREE (which is used as a
+ * placeholder because it doesn't otherwise make sense), the on-disk block type
+ * is verified to be @blktype.
  *
  * Returns: A VFS inode, or an error
  */
 
 struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
-				u64 no_addr, u64 no_formal_ino)
+				u64 no_addr, u64 no_formal_ino,
+				unsigned int blktype)
 {
 	struct inode *inode;
 	struct gfs2_inode *ip;
 	struct gfs2_glock *io_gl = NULL;
+	struct gfs2_holder i_gh;
+	bool unlock = false;
 	int error;
 
-	inode = iget_locked(sb, (unsigned long)no_addr);
+	inode = gfs2_iget(sb, no_addr);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
 
 	ip = GFS2_I(inode);
-	ip->i_no_addr = no_addr;
 
 	if (inode->i_state & I_NEW) {
 		struct gfs2_sbd *sdp = GFS2_SB(inode);
@@ -112,10 +149,30 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		if (unlikely(error))
 			goto fail_put;
 
+		if (type == DT_UNKNOWN || blktype != GFS2_BLKST_FREE) {
+			/*
+			 * The GL_SKIP flag indicates to skip reading the inode
+			 * block.  We read the inode with gfs2_inode_refresh
+			 * after possibly checking the block type.
+			 */
+			error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE,
+						   GL_SKIP, &i_gh);
+			if (error)
+				goto fail_put;
+			unlock = true;
+
+			if (blktype != GFS2_BLKST_FREE) {
+				error = gfs2_check_blk_type(sdp, no_addr,
+							    blktype);
+				if (error)
+					goto fail_put;
+			}
+		}
+
 		set_bit(GIF_INVALID, &ip->i_flags);
 		error = gfs2_glock_nq_init(io_gl, LM_ST_SHARED, GL_EXACT, &ip->i_iopen_gh);
 		if (unlikely(error))
-			goto fail_iopen;
+			goto fail_put;
 
 		ip->i_iopen_gh.gh_gl->gl_object = ip;
 		gfs2_glock_put(io_gl);
@@ -134,6 +191,8 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		unlock_new_inode(inode);
 	}
 
+	if (unlock)
+		gfs2_glock_dq_uninit(&i_gh);
 	return inode;
 
 fail_refresh:
@@ -141,10 +200,11 @@ fail_refresh:
 	ip->i_iopen_gh.gh_gl->gl_object = NULL;
 	gfs2_glock_dq_wait(&ip->i_iopen_gh);
 	gfs2_holder_uninit(&ip->i_iopen_gh);
-fail_iopen:
+fail_put:
 	if (io_gl)
 		gfs2_glock_put(io_gl);
-fail_put:
+	if (unlock)
+		gfs2_glock_dq_uninit(&i_gh);
 	ip->i_gl->gl_object = NULL;
 fail:
 	iget_failed(inode);
@@ -155,23 +215,12 @@ struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 				  u64 *no_formal_ino, unsigned int blktype)
 {
 	struct super_block *sb = sdp->sd_vfs;
-	struct gfs2_holder i_gh;
-	struct inode *inode = NULL;
+	struct inode *inode;
 	int error;
 
-	/* Must not read in block until block type is verified */
-	error = gfs2_glock_nq_num(sdp, no_addr, &gfs2_inode_glops,
-				  LM_ST_EXCLUSIVE, GL_SKIP, &i_gh);
-	if (error)
-		return ERR_PTR(error);
-
-	error = gfs2_check_blk_type(sdp, no_addr, blktype);
-	if (error)
-		goto fail;
-
-	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, 0);
+	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, 0, blktype);
 	if (IS_ERR(inode))
-		goto fail;
+		return inode;
 
 	/* Two extra checks for NFS only */
 	if (no_formal_ino) {
@@ -182,16 +231,12 @@ struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 		error = -EIO;
 		if (GFS2_I(inode)->i_diskflags & GFS2_DIF_SYSTEM)
 			goto fail_iput;
-
-		error = 0;
 	}
+	return inode;
 
-fail:
-	gfs2_glock_dq_uninit(&i_gh);
-	return error ? ERR_PTR(error) : inode;
 fail_iput:
 	iput(inode);
-	goto fail;
+	return ERR_PTR(error);
 }
 
 
diff --git a/fs/gfs2/inode.h b/fs/gfs2/inode.h
index e1af0d4..443b46c 100644
--- a/fs/gfs2/inode.h
+++ b/fs/gfs2/inode.h
@@ -94,7 +94,8 @@ err:
 }
 
 extern struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned type, 
-				       u64 no_addr, u64 no_formal_ino);
+				       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,
 					 unsigned int blktype);
diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index 4546360..b8f6fc9 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -454,7 +454,8 @@ static int gfs2_lookup_root(struct super_block *sb, struct dentry **dptr,
 	struct dentry *dentry;
 	struct inode *inode;
 
-	inode = gfs2_inode_lookup(sb, DT_DIR, no_addr, 0);
+	inode = gfs2_inode_lookup(sb, DT_DIR, no_addr, 0,
+				  GFS2_BLKST_FREE /* ignore */);
 	if (IS_ERR(inode)) {
 		fs_err(sdp, "can't read in %s inode: %ld\n", name, PTR_ERR(inode));
 		return PTR_ERR(inode);
-- 
2.5.5


^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 1/4] gfs2: Fix gfs2_lookup_by_inum lock inversion
@ 2016-06-20 15:55   ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: cluster-devel.redhat.com

From: Andreas Gruenbacher <agruenba@redhat.com>

The current gfs2_lookup_by_inum takes the glock of a presumed inode
identified by block number, verifies that the block is indeed an inode,
and then instantiates and reads the new inode via gfs2_inode_lookup.

However, instantiating a new inode may block on freeing a previous
instance of that inode (__wait_on_freeing_inode), and freeing an inode
requires to take the glock already held, leading to lock inversion and
deadlock.

Fix this by first instantiating the new inode, then verifying that the
block is an inode (if required), and then reading in the new inode, all
in gfs2_inode_lookup.

If the block we are looking for is not an inode, we discard the new
inode via iget_failed, which marks inodes as bad and unhashes them.
Other tasks waiting on that inode will get back a bad inode back from
ilookup or iget_locked; in that case, retry the lookup.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/dir.c        |   3 +-
 fs/gfs2/glock.c      |   4 +-
 fs/gfs2/inode.c      | 101 +++++++++++++++++++++++++++++++++++++--------------
 fs/gfs2/inode.h      |   3 +-
 fs/gfs2/ops_fstype.c |   3 +-
 5 files changed, 81 insertions(+), 33 deletions(-)

diff --git a/fs/gfs2/dir.c b/fs/gfs2/dir.c
index 4a01f30..1b02665 100644
--- a/fs/gfs2/dir.c
+++ b/fs/gfs2/dir.c
@@ -1660,7 +1660,8 @@ struct inode *gfs2_dir_search(struct inode *dir, const struct qstr *name,
 		brelse(bh);
 		if (fail_on_exist)
 			return ERR_PTR(-EEXIST);
-		inode = gfs2_inode_lookup(dir->i_sb, dtype, addr, formal_ino);
+		inode = gfs2_inode_lookup(dir->i_sb, dtype, addr, formal_ino,
+					  GFS2_BLKST_FREE /* ignore */);
 		if (!IS_ERR(inode))
 			GFS2_I(inode)->i_rahead = rahead;
 		return inode;
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 706fd93..ce46375 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -576,7 +576,7 @@ 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_name.ln_sbd;
 	struct gfs2_inode *ip;
-	struct inode *inode;
+	struct inode *inode = NULL;
 	u64 no_addr = gl->gl_name.ln_number;
 
 	/* If someone's using this glock to create a new dinode, the block must
@@ -590,7 +590,7 @@ static void delete_work_func(struct work_struct *work)
 
 	if (ip)
 		inode = gfs2_ilookup(sdp->sd_vfs, no_addr);
-	else
+	if (IS_ERR_OR_NULL(inode))
 		inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
 	if (inode && !IS_ERR(inode)) {
 		d_prune_aliases(inode);
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 21dc784..6d5c6bb 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -39,7 +39,33 @@
 
 struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr)
 {
-	return ilookup(sb, (unsigned long)no_addr);
+	struct inode *inode;
+
+repeat:
+	inode = ilookup(sb, no_addr);
+	if (!inode)
+		return inode;
+	if (is_bad_inode(inode)) {
+		iput(inode);
+		goto repeat;
+	}
+	return inode;
+}
+
+static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
+{
+	struct inode *inode;
+
+repeat:
+	inode = iget_locked(sb, no_addr);
+	if (!inode)
+		return inode;
+	if (is_bad_inode(inode)) {
+		iput(inode);
+		goto repeat;
+	}
+	GFS2_I(inode)->i_no_addr = no_addr;
+	return inode;
 }
 
 /**
@@ -78,26 +104,37 @@ static void gfs2_set_iop(struct inode *inode)
 /**
  * gfs2_inode_lookup - Lookup an inode
  * @sb: The super block
- * @no_addr: The inode number
  * @type: The type of the inode
+ * @no_addr: The inode number
+ * @no_formal_ino: The inode generation number
+ * @blktype: Requested block type (GFS2_BLKST_DINODE or GFS2_BLKST_UNLINKED;
+ *           GFS2_BLKST_FREE do indicate not to verify)
+ *
+ * If @type is DT_UNKNOWN, the inode type is fetched from disk.
+ *
+ * If @blktype is anything other than GFS2_BLKST_FREE (which is used as a
+ * placeholder because it doesn't otherwise make sense), the on-disk block type
+ * is verified to be @blktype.
  *
  * Returns: A VFS inode, or an error
  */
 
 struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
-				u64 no_addr, u64 no_formal_ino)
+				u64 no_addr, u64 no_formal_ino,
+				unsigned int blktype)
 {
 	struct inode *inode;
 	struct gfs2_inode *ip;
 	struct gfs2_glock *io_gl = NULL;
+	struct gfs2_holder i_gh;
+	bool unlock = false;
 	int error;
 
-	inode = iget_locked(sb, (unsigned long)no_addr);
+	inode = gfs2_iget(sb, no_addr);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
 
 	ip = GFS2_I(inode);
-	ip->i_no_addr = no_addr;
 
 	if (inode->i_state & I_NEW) {
 		struct gfs2_sbd *sdp = GFS2_SB(inode);
@@ -112,10 +149,30 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		if (unlikely(error))
 			goto fail_put;
 
+		if (type == DT_UNKNOWN || blktype != GFS2_BLKST_FREE) {
+			/*
+			 * The GL_SKIP flag indicates to skip reading the inode
+			 * block.  We read the inode with gfs2_inode_refresh
+			 * after possibly checking the block type.
+			 */
+			error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE,
+						   GL_SKIP, &i_gh);
+			if (error)
+				goto fail_put;
+			unlock = true;
+
+			if (blktype != GFS2_BLKST_FREE) {
+				error = gfs2_check_blk_type(sdp, no_addr,
+							    blktype);
+				if (error)
+					goto fail_put;
+			}
+		}
+
 		set_bit(GIF_INVALID, &ip->i_flags);
 		error = gfs2_glock_nq_init(io_gl, LM_ST_SHARED, GL_EXACT, &ip->i_iopen_gh);
 		if (unlikely(error))
-			goto fail_iopen;
+			goto fail_put;
 
 		ip->i_iopen_gh.gh_gl->gl_object = ip;
 		gfs2_glock_put(io_gl);
@@ -134,6 +191,8 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		unlock_new_inode(inode);
 	}
 
+	if (unlock)
+		gfs2_glock_dq_uninit(&i_gh);
 	return inode;
 
 fail_refresh:
@@ -141,10 +200,11 @@ fail_refresh:
 	ip->i_iopen_gh.gh_gl->gl_object = NULL;
 	gfs2_glock_dq_wait(&ip->i_iopen_gh);
 	gfs2_holder_uninit(&ip->i_iopen_gh);
-fail_iopen:
+fail_put:
 	if (io_gl)
 		gfs2_glock_put(io_gl);
-fail_put:
+	if (unlock)
+		gfs2_glock_dq_uninit(&i_gh);
 	ip->i_gl->gl_object = NULL;
 fail:
 	iget_failed(inode);
@@ -155,23 +215,12 @@ struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 				  u64 *no_formal_ino, unsigned int blktype)
 {
 	struct super_block *sb = sdp->sd_vfs;
-	struct gfs2_holder i_gh;
-	struct inode *inode = NULL;
+	struct inode *inode;
 	int error;
 
-	/* Must not read in block until block type is verified */
-	error = gfs2_glock_nq_num(sdp, no_addr, &gfs2_inode_glops,
-				  LM_ST_EXCLUSIVE, GL_SKIP, &i_gh);
-	if (error)
-		return ERR_PTR(error);
-
-	error = gfs2_check_blk_type(sdp, no_addr, blktype);
-	if (error)
-		goto fail;
-
-	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, 0);
+	inode = gfs2_inode_lookup(sb, DT_UNKNOWN, no_addr, 0, blktype);
 	if (IS_ERR(inode))
-		goto fail;
+		return inode;
 
 	/* Two extra checks for NFS only */
 	if (no_formal_ino) {
@@ -182,16 +231,12 @@ struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 		error = -EIO;
 		if (GFS2_I(inode)->i_diskflags & GFS2_DIF_SYSTEM)
 			goto fail_iput;
-
-		error = 0;
 	}
+	return inode;
 
-fail:
-	gfs2_glock_dq_uninit(&i_gh);
-	return error ? ERR_PTR(error) : inode;
 fail_iput:
 	iput(inode);
-	goto fail;
+	return ERR_PTR(error);
 }
 
 
diff --git a/fs/gfs2/inode.h b/fs/gfs2/inode.h
index e1af0d4..443b46c 100644
--- a/fs/gfs2/inode.h
+++ b/fs/gfs2/inode.h
@@ -94,7 +94,8 @@ err:
 }
 
 extern struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned type, 
-				       u64 no_addr, u64 no_formal_ino);
+				       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,
 					 unsigned int blktype);
diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index 4546360..b8f6fc9 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -454,7 +454,8 @@ static int gfs2_lookup_root(struct super_block *sb, struct dentry **dptr,
 	struct dentry *dentry;
 	struct inode *inode;
 
-	inode = gfs2_inode_lookup(sb, DT_DIR, no_addr, 0);
+	inode = gfs2_inode_lookup(sb, DT_DIR, no_addr, 0,
+				  GFS2_BLKST_FREE /* ignore */);
 	if (IS_ERR(inode)) {
 		fs_err(sdp, "can't read in %s inode: %ld\n", name, PTR_ERR(inode));
 		return PTR_ERR(inode);
-- 
2.5.5



^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [PATCH 2/4] gfs2: Get rid of gfs2_ilookup
  2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
@ 2016-06-20 15:55   ` Bob Peterson
  -1 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: linux-fsdevel; +Cc: cluster-devel

From: Andreas Gruenbacher <agruenba@redhat.com>

Now that gfs2_lookup_by_inum only takes the inode glock for new inodes
(and not for cached inodes anymore), there no longer is a need to
optimize the cached-inode case in gfs2_get_dentry or delete_work_func,
and gfs2_ilookup can be removed.

In addition, gfs2_get_dentry wasn't checking the GFS2_DIF_SYSTEM flag in
i_diskflags in the gfs2_ilookup case (see gfs2_lookup_by_inum); this
inconsistency goes away as well.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/export.c | 11 -----------
 fs/gfs2/glock.c  | 11 ++---------
 fs/gfs2/inode.c  | 15 ---------------
 fs/gfs2/inode.h  |  1 -
 4 files changed, 2 insertions(+), 36 deletions(-)

diff --git a/fs/gfs2/export.c b/fs/gfs2/export.c
index d5bda85..a332f3c 100644
--- a/fs/gfs2/export.c
+++ b/fs/gfs2/export.c
@@ -137,21 +137,10 @@ static struct dentry *gfs2_get_dentry(struct super_block *sb,
 	struct gfs2_sbd *sdp = sb->s_fs_info;
 	struct inode *inode;
 
-	inode = gfs2_ilookup(sb, inum->no_addr);
-	if (inode) {
-		if (GFS2_I(inode)->i_no_formal_ino != inum->no_formal_ino) {
-			iput(inode);
-			return ERR_PTR(-ESTALE);
-		}
-		goto out_inode;
-	}
-
 	inode = gfs2_lookup_by_inum(sdp, inum->no_addr, &inum->no_formal_ino,
 				    GFS2_BLKST_DINODE);
 	if (IS_ERR(inode))
 		return ERR_CAST(inode);
-
-out_inode:
 	return d_obtain_alias(inode);
 }
 
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index ce46375..1138a61 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -575,8 +575,7 @@ 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_name.ln_sbd;
-	struct gfs2_inode *ip;
-	struct inode *inode = NULL;
+	struct inode *inode;
 	u64 no_addr = gl->gl_name.ln_number;
 
 	/* If someone's using this glock to create a new dinode, the block must
@@ -585,13 +584,7 @@ static void delete_work_func(struct work_struct *work)
 	if (test_bit(GLF_INODE_CREATING, &gl->gl_flags))
 		goto out;
 
-	ip = gl->gl_object;
-	/* Note: Unsafe to dereference ip as we don't hold right refs/locks */
-
-	if (ip)
-		inode = gfs2_ilookup(sdp->sd_vfs, no_addr);
-	if (IS_ERR_OR_NULL(inode))
-		inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
+	inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
 	if (inode && !IS_ERR(inode)) {
 		d_prune_aliases(inode);
 		iput(inode);
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 6d5c6bb..ebff26e 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -37,21 +37,6 @@
 #include "super.h"
 #include "glops.h"
 
-struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr)
-{
-	struct inode *inode;
-
-repeat:
-	inode = ilookup(sb, no_addr);
-	if (!inode)
-		return inode;
-	if (is_bad_inode(inode)) {
-		iput(inode);
-		goto repeat;
-	}
-	return inode;
-}
-
 static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
 {
 	struct inode *inode;
diff --git a/fs/gfs2/inode.h b/fs/gfs2/inode.h
index 443b46c..7710dfd 100644
--- a/fs/gfs2/inode.h
+++ b/fs/gfs2/inode.h
@@ -99,7 +99,6 @@ extern struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned type,
 extern struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 					 u64 *no_formal_ino,
 					 unsigned int blktype);
-extern struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr);
 
 extern int gfs2_inode_refresh(struct gfs2_inode *ip);
 
-- 
2.5.5


^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 2/4] gfs2: Get rid of gfs2_ilookup
@ 2016-06-20 15:55   ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: cluster-devel.redhat.com

From: Andreas Gruenbacher <agruenba@redhat.com>

Now that gfs2_lookup_by_inum only takes the inode glock for new inodes
(and not for cached inodes anymore), there no longer is a need to
optimize the cached-inode case in gfs2_get_dentry or delete_work_func,
and gfs2_ilookup can be removed.

In addition, gfs2_get_dentry wasn't checking the GFS2_DIF_SYSTEM flag in
i_diskflags in the gfs2_ilookup case (see gfs2_lookup_by_inum); this
inconsistency goes away as well.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/export.c | 11 -----------
 fs/gfs2/glock.c  | 11 ++---------
 fs/gfs2/inode.c  | 15 ---------------
 fs/gfs2/inode.h  |  1 -
 4 files changed, 2 insertions(+), 36 deletions(-)

diff --git a/fs/gfs2/export.c b/fs/gfs2/export.c
index d5bda85..a332f3c 100644
--- a/fs/gfs2/export.c
+++ b/fs/gfs2/export.c
@@ -137,21 +137,10 @@ static struct dentry *gfs2_get_dentry(struct super_block *sb,
 	struct gfs2_sbd *sdp = sb->s_fs_info;
 	struct inode *inode;
 
-	inode = gfs2_ilookup(sb, inum->no_addr);
-	if (inode) {
-		if (GFS2_I(inode)->i_no_formal_ino != inum->no_formal_ino) {
-			iput(inode);
-			return ERR_PTR(-ESTALE);
-		}
-		goto out_inode;
-	}
-
 	inode = gfs2_lookup_by_inum(sdp, inum->no_addr, &inum->no_formal_ino,
 				    GFS2_BLKST_DINODE);
 	if (IS_ERR(inode))
 		return ERR_CAST(inode);
-
-out_inode:
 	return d_obtain_alias(inode);
 }
 
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index ce46375..1138a61 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -575,8 +575,7 @@ 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_name.ln_sbd;
-	struct gfs2_inode *ip;
-	struct inode *inode = NULL;
+	struct inode *inode;
 	u64 no_addr = gl->gl_name.ln_number;
 
 	/* If someone's using this glock to create a new dinode, the block must
@@ -585,13 +584,7 @@ static void delete_work_func(struct work_struct *work)
 	if (test_bit(GLF_INODE_CREATING, &gl->gl_flags))
 		goto out;
 
-	ip = gl->gl_object;
-	/* Note: Unsafe to dereference ip as we don't hold right refs/locks */
-
-	if (ip)
-		inode = gfs2_ilookup(sdp->sd_vfs, no_addr);
-	if (IS_ERR_OR_NULL(inode))
-		inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
+	inode = gfs2_lookup_by_inum(sdp, no_addr, NULL, GFS2_BLKST_UNLINKED);
 	if (inode && !IS_ERR(inode)) {
 		d_prune_aliases(inode);
 		iput(inode);
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 6d5c6bb..ebff26e 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -37,21 +37,6 @@
 #include "super.h"
 #include "glops.h"
 
-struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr)
-{
-	struct inode *inode;
-
-repeat:
-	inode = ilookup(sb, no_addr);
-	if (!inode)
-		return inode;
-	if (is_bad_inode(inode)) {
-		iput(inode);
-		goto repeat;
-	}
-	return inode;
-}
-
 static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
 {
 	struct inode *inode;
diff --git a/fs/gfs2/inode.h b/fs/gfs2/inode.h
index 443b46c..7710dfd 100644
--- a/fs/gfs2/inode.h
+++ b/fs/gfs2/inode.h
@@ -99,7 +99,6 @@ extern struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned type,
 extern struct inode *gfs2_lookup_by_inum(struct gfs2_sbd *sdp, u64 no_addr,
 					 u64 *no_formal_ino,
 					 unsigned int blktype);
-extern struct inode *gfs2_ilookup(struct super_block *sb, u64 no_addr);
 
 extern int gfs2_inode_refresh(struct gfs2_inode *ip);
 
-- 
2.5.5



^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [PATCH 3/4] gfs2: Large-filesystem fix for 32-bit systems
  2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
@ 2016-06-20 15:55   ` Bob Peterson
  -1 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: linux-fsdevel; +Cc: cluster-devel

From: Andreas Gruenbacher <agruenba@redhat.com>

Commit ff34245d switched from iget5_locked to iget_locked among other
things, but iget_locked doesn't work for filesystems larger than 2^32
blocks on 32-bit systems.  Switch back to iget5_locked.  Filesystems
larger than 2^32 blocks are unrealistic to work well on 32-bit systems,
so this is mostly a code cleanliness fix.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/inode.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index ebff26e..481b649 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -37,19 +37,34 @@
 #include "super.h"
 #include "glops.h"
 
+static int iget_test(struct inode *inode, void *opaque)
+{
+	u64 no_addr = *(u64 *)opaque;
+
+	return GFS2_I(inode)->i_no_addr == no_addr;
+}
+
+static int iget_set(struct inode *inode, void *opaque)
+{
+	u64 no_addr = *(u64 *)opaque;
+
+	GFS2_I(inode)->i_no_addr = no_addr;
+	inode->i_ino = no_addr;
+	return 0;
+}
+
 static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
 {
 	struct inode *inode;
 
 repeat:
-	inode = iget_locked(sb, no_addr);
+	inode = iget5_locked(sb, no_addr, iget_test, iget_set, &no_addr);
 	if (!inode)
 		return inode;
 	if (is_bad_inode(inode)) {
 		iput(inode);
 		goto repeat;
 	}
-	GFS2_I(inode)->i_no_addr = no_addr;
 	return inode;
 }
 
-- 
2.5.5


^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 3/4] gfs2: Large-filesystem fix for 32-bit systems
@ 2016-06-20 15:55   ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: cluster-devel.redhat.com

From: Andreas Gruenbacher <agruenba@redhat.com>

Commit ff34245d switched from iget5_locked to iget_locked among other
things, but iget_locked doesn't work for filesystems larger than 2^32
blocks on 32-bit systems.  Switch back to iget5_locked.  Filesystems
larger than 2^32 blocks are unrealistic to work well on 32-bit systems,
so this is mostly a code cleanliness fix.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/inode.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index ebff26e..481b649 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -37,19 +37,34 @@
 #include "super.h"
 #include "glops.h"
 
+static int iget_test(struct inode *inode, void *opaque)
+{
+	u64 no_addr = *(u64 *)opaque;
+
+	return GFS2_I(inode)->i_no_addr == no_addr;
+}
+
+static int iget_set(struct inode *inode, void *opaque)
+{
+	u64 no_addr = *(u64 *)opaque;
+
+	GFS2_I(inode)->i_no_addr = no_addr;
+	inode->i_ino = no_addr;
+	return 0;
+}
+
 static struct inode *gfs2_iget(struct super_block *sb, u64 no_addr)
 {
 	struct inode *inode;
 
 repeat:
-	inode = iget_locked(sb, no_addr);
+	inode = iget5_locked(sb, no_addr, iget_test, iget_set, &no_addr);
 	if (!inode)
 		return inode;
 	if (is_bad_inode(inode)) {
 		iput(inode);
 		goto repeat;
 	}
-	GFS2_I(inode)->i_no_addr = no_addr;
 	return inode;
 }
 
-- 
2.5.5



^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [PATCH 4/4] gfs2: Lock holder cleanup
  2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
@ 2016-06-20 15:55   ` Bob Peterson
  -1 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: linux-fsdevel; +Cc: cluster-devel

From: Andreas Gruenbacher <agruenba@redhat.com>

Make the code more readable by cleaning up the different ways of
initializing lock holders and checking for initialized lock holders:
mark lock holders as uninitialized by setting the holder's glock to NULL
(gfs2_holder_mark_uninitialized) instead of zeroing out the entire
object or using a separate flag.  Recognize initialized holders by their
non-NULL glock (gfs2_holder_initialized).  Don't zero out holder objects
which are immeditiately initialized via gfs2_holder_init or
gfs2_glock_nq_init.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/dentry.c |  2 +-
 fs/gfs2/file.c   |  2 +-
 fs/gfs2/glock.c  |  2 +-
 fs/gfs2/glock.h  | 10 ++++++++++
 fs/gfs2/inode.c  | 33 +++++++++++++++------------------
 fs/gfs2/main.c   |  2 +-
 fs/gfs2/quota.c  |  2 +-
 fs/gfs2/rgrp.c   |  4 ++--
 fs/gfs2/super.c  | 24 ++++++++++++++----------
 9 files changed, 46 insertions(+), 35 deletions(-)

diff --git a/fs/gfs2/dentry.c b/fs/gfs2/dentry.c
index 30822b1..5173b98 100644
--- a/fs/gfs2/dentry.c
+++ b/fs/gfs2/dentry.c
@@ -117,7 +117,7 @@ static int gfs2_dentry_delete(const struct dentry *dentry)
 		return 0;
 
 	ginode = GFS2_I(d_inode(dentry));
-	if (!ginode->i_iopen_gh.gh_gl)
+	if (!gfs2_holder_initialized(&ginode->i_iopen_gh))
 		return 0;
 
 	if (test_bit(GLF_DEMOTE, &ginode->i_iopen_gh.gh_gl->gl_flags))
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index e0f98e4..320e65e 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -1098,7 +1098,7 @@ static void do_unflock(struct file *file, struct file_lock *fl)
 
 	mutex_lock(&fp->f_fl_mutex);
 	locks_lock_file_wait(file, fl);
-	if (fl_gh->gh_gl) {
+	if (gfs2_holder_initialized(fl_gh)) {
 		gfs2_glock_dq(fl_gh);
 		gfs2_holder_uninit(fl_gh);
 	}
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 1138a61..3a90b2b 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -801,7 +801,7 @@ void gfs2_holder_uninit(struct gfs2_holder *gh)
 {
 	put_pid(gh->gh_owner_pid);
 	gfs2_glock_put(gh->gh_gl);
-	gh->gh_gl = NULL;
+	gfs2_holder_mark_uninitialized(gh);
 	gh->gh_ip = 0;
 }
 
diff --git a/fs/gfs2/glock.h b/fs/gfs2/glock.h
index 46ab67f..ab1ef32 100644
--- a/fs/gfs2/glock.h
+++ b/fs/gfs2/glock.h
@@ -247,4 +247,14 @@ extern void gfs2_unregister_debugfs(void);
 
 extern const struct lm_lockops gfs2_dlm_ops;
 
+static inline void gfs2_holder_mark_uninitialized(struct gfs2_holder *gh)
+{
+	gh->gh_gl = NULL;
+}
+
+static inline bool gfs2_holder_initialized(struct gfs2_holder *gh)
+{
+	return gh->gh_gl;
+}
+
 #endif /* __GLOCK_DOT_H__ */
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 481b649..de54d60 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -127,9 +127,9 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 	struct gfs2_inode *ip;
 	struct gfs2_glock *io_gl = NULL;
 	struct gfs2_holder i_gh;
-	bool unlock = false;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&i_gh);
 	inode = gfs2_iget(sb, no_addr);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
@@ -159,7 +159,6 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 						   GL_SKIP, &i_gh);
 			if (error)
 				goto fail_put;
-			unlock = true;
 
 			if (blktype != GFS2_BLKST_FREE) {
 				error = gfs2_check_blk_type(sdp, no_addr,
@@ -191,7 +190,7 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		unlock_new_inode(inode);
 	}
 
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 	return inode;
 
@@ -203,7 +202,7 @@ fail_refresh:
 fail_put:
 	if (io_gl)
 		gfs2_glock_put(io_gl);
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 	ip->i_gl->gl_object = NULL;
 fail:
@@ -281,8 +280,8 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 	struct gfs2_holder d_gh;
 	int error = 0;
 	struct inode *inode = NULL;
-	int unlock = 0;
 
+	gfs2_holder_mark_uninitialized(&d_gh);
 	if (!name->len || name->len > GFS2_FNAMESIZE)
 		return ERR_PTR(-ENAMETOOLONG);
 
@@ -297,7 +296,6 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 		error = gfs2_glock_nq_init(dip->i_gl, LM_ST_SHARED, 0, &d_gh);
 		if (error)
 			return ERR_PTR(error);
-		unlock = 1;
 	}
 
 	if (!is_root) {
@@ -310,7 +308,7 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 	if (IS_ERR(inode))
 		error = PTR_ERR(inode);
 out:
-	if (unlock)
+	if (gfs2_holder_initialized(&d_gh))
 		gfs2_glock_dq_uninit(&d_gh);
 	if (error == -ENOENT)
 		return NULL;
@@ -1354,7 +1352,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 	struct gfs2_inode *ip = GFS2_I(d_inode(odentry));
 	struct gfs2_inode *nip = NULL;
 	struct gfs2_sbd *sdp = GFS2_SB(odir);
-	struct gfs2_holder ghs[5], r_gh = { .gh_gl = NULL, };
+	struct gfs2_holder ghs[5], r_gh;
 	struct gfs2_rgrpd *nrgd;
 	unsigned int num_gh;
 	int dir_rename = 0;
@@ -1362,6 +1360,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 	unsigned int x;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&r_gh);
 	if (d_really_is_positive(ndentry)) {
 		nip = GFS2_I(d_inode(ndentry));
 		if (ip == nip)
@@ -1551,7 +1550,7 @@ out_gunlock:
 		gfs2_holder_uninit(ghs + x);
 	}
 out_gunlock_r:
-	if (r_gh.gh_gl)
+	if (gfs2_holder_initialized(&r_gh))
 		gfs2_glock_dq_uninit(&r_gh);
 out:
 	return error;
@@ -1577,13 +1576,14 @@ static int gfs2_exchange(struct inode *odir, struct dentry *odentry,
 	struct gfs2_inode *oip = GFS2_I(odentry->d_inode);
 	struct gfs2_inode *nip = GFS2_I(ndentry->d_inode);
 	struct gfs2_sbd *sdp = GFS2_SB(odir);
-	struct gfs2_holder ghs[5], r_gh = { .gh_gl = NULL, };
+	struct gfs2_holder ghs[5], r_gh;
 	unsigned int num_gh;
 	unsigned int x;
 	umode_t old_mode = oip->i_inode.i_mode;
 	umode_t new_mode = nip->i_inode.i_mode;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&r_gh);
 	error = gfs2_rindex_update(sdp);
 	if (error)
 		return error;
@@ -1691,7 +1691,7 @@ out_gunlock:
 		gfs2_holder_uninit(ghs + x);
 	}
 out_gunlock_r:
-	if (r_gh.gh_gl)
+	if (gfs2_holder_initialized(&r_gh))
 		gfs2_glock_dq_uninit(&r_gh);
 out:
 	return error;
@@ -1788,9 +1788,8 @@ int gfs2_permission(struct inode *inode, int mask)
 	struct gfs2_inode *ip;
 	struct gfs2_holder i_gh;
 	int error;
-	int unlock = 0;
-
 
+	gfs2_holder_mark_uninitialized(&i_gh);
 	ip = GFS2_I(inode);
 	if (gfs2_glock_is_locked_by_me(ip->i_gl) == NULL) {
 		if (mask & MAY_NOT_BLOCK)
@@ -1798,14 +1797,13 @@ int gfs2_permission(struct inode *inode, int mask)
 		error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY, &i_gh);
 		if (error)
 			return error;
-		unlock = 1;
 	}
 
 	if ((mask & MAY_WRITE) && IS_IMMUTABLE(inode))
 		error = -EACCES;
 	else
 		error = generic_permission(inode, mask);
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 
 	return error;
@@ -1977,17 +1975,16 @@ static int gfs2_getattr(struct vfsmount *mnt, struct dentry *dentry,
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_holder gh;
 	int error;
-	int unlock = 0;
 
+	gfs2_holder_mark_uninitialized(&gh);
 	if (gfs2_glock_is_locked_by_me(ip->i_gl) == NULL) {
 		error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY, &gh);
 		if (error)
 			return error;
-		unlock = 1;
 	}
 
 	generic_fillattr(inode, stat);
-	if (unlock)
+	if (gfs2_holder_initialized(&gh))
 		gfs2_glock_dq_uninit(&gh);
 
 	return 0;
diff --git a/fs/gfs2/main.c b/fs/gfs2/main.c
index 615f675..74fd013 100644
--- a/fs/gfs2/main.c
+++ b/fs/gfs2/main.c
@@ -45,7 +45,7 @@ static void gfs2_init_inode_once(void *foo)
 	memset(&ip->i_res, 0, sizeof(ip->i_res));
 	RB_CLEAR_NODE(&ip->i_res.rs_node);
 	ip->i_hash_cache = NULL;
-	ip->i_iopen_gh.gh_gl = NULL;
+	gfs2_holder_mark_uninitialized(&ip->i_iopen_gh);
 }
 
 static void gfs2_init_glock_once(void *foo)
diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c
index ce7d69a..6c657b2 100644
--- a/fs/gfs2/quota.c
+++ b/fs/gfs2/quota.c
@@ -883,7 +883,7 @@ static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
 	gfs2_write_calc_reserv(ip, sizeof(struct gfs2_quota),
 			      &data_blocks, &ind_blocks);
 
-	ghs = kcalloc(num_qd, sizeof(struct gfs2_holder), GFP_NOFS);
+	ghs = kmalloc(num_qd * sizeof(struct gfs2_holder), GFP_NOFS);
 	if (!ghs)
 		return -ENOMEM;
 
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 960aaf4..fba38ca 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -2100,7 +2100,7 @@ void gfs2_inplace_release(struct gfs2_inode *ip)
 {
 	struct gfs2_blkreserv *rs = &ip->i_res;
 
-	if (rs->rs_rgd_gh.gh_gl)
+	if (gfs2_holder_initialized(&rs->rs_rgd_gh))
 		gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
 }
 
@@ -2600,7 +2600,7 @@ void gfs2_rlist_alloc(struct gfs2_rgrp_list *rlist, unsigned int state)
 {
 	unsigned int x;
 
-	rlist->rl_ghs = kcalloc(rlist->rl_rgrps, sizeof(struct gfs2_holder),
+	rlist->rl_ghs = kmalloc(rlist->rl_rgrps * sizeof(struct gfs2_holder),
 				GFP_NOFS | __GFP_NOFAIL);
 	for (x = 0; x < rlist->rl_rgrps; x++)
 		gfs2_holder_init(rlist->rl_rgd[x]->rd_gl,
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 9b2ff353..3a7e60b 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -855,7 +855,7 @@ static int gfs2_make_fs_ro(struct gfs2_sbd *sdp)
 	wait_event(sdp->sd_reserving_log_wait, atomic_read(&sdp->sd_reserving_log) == 0);
 	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_blks_free) == sdp->sd_jdesc->jd_blocks);
 
-	if (freeze_gh.gh_gl)
+	if (gfs2_holder_initialized(&freeze_gh))
 		gfs2_glock_dq_uninit(&freeze_gh);
 
 	gfs2_quota_cleanup(sdp);
@@ -1033,7 +1033,7 @@ static int gfs2_unfreeze(struct super_block *sb)
 
 	mutex_lock(&sdp->sd_freeze_mutex);
         if (atomic_read(&sdp->sd_freeze_state) != SFS_FROZEN ||
-	    sdp->sd_freeze_gh.gh_gl == NULL) {
+	    !gfs2_holder_initialized(&sdp->sd_freeze_gh)) {
 		mutex_unlock(&sdp->sd_freeze_mutex);
                 return 0;
 	}
@@ -1084,9 +1084,11 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 	int error = 0, err;
 
 	memset(sc, 0, sizeof(struct gfs2_statfs_change_host));
-	gha = kcalloc(slots, sizeof(struct gfs2_holder), GFP_KERNEL);
+	gha = kmalloc(slots * sizeof(struct gfs2_holder), GFP_KERNEL);
 	if (!gha)
 		return -ENOMEM;
+	for (x = 0; x < slots; x++)
+		gfs2_holder_mark_uninitialized(gha + x);
 
 	rgd_next = gfs2_rgrpd_get_first(sdp);
 
@@ -1096,7 +1098,7 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 		for (x = 0; x < slots; x++) {
 			gh = gha + x;
 
-			if (gh->gh_gl && gfs2_glock_poll(gh)) {
+			if (gfs2_holder_initialized(gh) && gfs2_glock_poll(gh)) {
 				err = gfs2_glock_wait(gh);
 				if (err) {
 					gfs2_holder_uninit(gh);
@@ -1109,7 +1111,7 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 				}
 			}
 
-			if (gh->gh_gl)
+			if (gfs2_holder_initialized(gh))
 				done = 0;
 			else if (rgd_next && !error) {
 				error = gfs2_glock_nq_init(rgd_next->rd_gl,
@@ -1304,9 +1306,11 @@ static int gfs2_drop_inode(struct inode *inode)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 
-	if (!test_bit(GIF_FREE_VFS_INODE, &ip->i_flags) && inode->i_nlink) {
+	if (!test_bit(GIF_FREE_VFS_INODE, &ip->i_flags) &&
+	    inode->i_nlink &&
+	    gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl;
-		if (gl && test_bit(GLF_DEMOTE, &gl->gl_flags))
+		if (test_bit(GLF_DEMOTE, &gl->gl_flags))
 			clear_nlink(inode);
 	}
 	return generic_drop_inode(inode);
@@ -1551,7 +1555,7 @@ static void gfs2_evict_inode(struct inode *inode)
 			goto out_truncate;
 	}
 
-	if (ip->i_iopen_gh.gh_gl &&
+	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;
 		gfs2_glock_dq_wait(&ip->i_iopen_gh);
@@ -1610,7 +1614,7 @@ out_unlock:
 	if (gfs2_rs_active(&ip->i_res))
 		gfs2_rs_deltree(&ip->i_res);
 
-	if (ip->i_iopen_gh.gh_gl) {
+	if (gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) {
 			ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 			gfs2_glock_dq_wait(&ip->i_iopen_gh);
@@ -1632,7 +1636,7 @@ out:
 	gfs2_glock_add_to_lru(ip->i_gl);
 	gfs2_glock_put(ip->i_gl);
 	ip->i_gl = NULL;
-	if (ip->i_iopen_gh.gh_gl) {
+	if (gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		ip->i_iopen_gh.gh_gl->gl_object = NULL;
 		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 		gfs2_glock_dq_wait(&ip->i_iopen_gh);
-- 
2.5.5


^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 4/4] gfs2: Lock holder cleanup
@ 2016-06-20 15:55   ` Bob Peterson
  0 siblings, 0 replies; 12+ messages in thread
From: Bob Peterson @ 2016-06-20 15:55 UTC (permalink / raw)
  To: cluster-devel.redhat.com

From: Andreas Gruenbacher <agruenba@redhat.com>

Make the code more readable by cleaning up the different ways of
initializing lock holders and checking for initialized lock holders:
mark lock holders as uninitialized by setting the holder's glock to NULL
(gfs2_holder_mark_uninitialized) instead of zeroing out the entire
object or using a separate flag.  Recognize initialized holders by their
non-NULL glock (gfs2_holder_initialized).  Don't zero out holder objects
which are immeditiately initialized via gfs2_holder_init or
gfs2_glock_nq_init.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/dentry.c |  2 +-
 fs/gfs2/file.c   |  2 +-
 fs/gfs2/glock.c  |  2 +-
 fs/gfs2/glock.h  | 10 ++++++++++
 fs/gfs2/inode.c  | 33 +++++++++++++++------------------
 fs/gfs2/main.c   |  2 +-
 fs/gfs2/quota.c  |  2 +-
 fs/gfs2/rgrp.c   |  4 ++--
 fs/gfs2/super.c  | 24 ++++++++++++++----------
 9 files changed, 46 insertions(+), 35 deletions(-)

diff --git a/fs/gfs2/dentry.c b/fs/gfs2/dentry.c
index 30822b1..5173b98 100644
--- a/fs/gfs2/dentry.c
+++ b/fs/gfs2/dentry.c
@@ -117,7 +117,7 @@ static int gfs2_dentry_delete(const struct dentry *dentry)
 		return 0;
 
 	ginode = GFS2_I(d_inode(dentry));
-	if (!ginode->i_iopen_gh.gh_gl)
+	if (!gfs2_holder_initialized(&ginode->i_iopen_gh))
 		return 0;
 
 	if (test_bit(GLF_DEMOTE, &ginode->i_iopen_gh.gh_gl->gl_flags))
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index e0f98e4..320e65e 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -1098,7 +1098,7 @@ static void do_unflock(struct file *file, struct file_lock *fl)
 
 	mutex_lock(&fp->f_fl_mutex);
 	locks_lock_file_wait(file, fl);
-	if (fl_gh->gh_gl) {
+	if (gfs2_holder_initialized(fl_gh)) {
 		gfs2_glock_dq(fl_gh);
 		gfs2_holder_uninit(fl_gh);
 	}
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 1138a61..3a90b2b 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -801,7 +801,7 @@ void gfs2_holder_uninit(struct gfs2_holder *gh)
 {
 	put_pid(gh->gh_owner_pid);
 	gfs2_glock_put(gh->gh_gl);
-	gh->gh_gl = NULL;
+	gfs2_holder_mark_uninitialized(gh);
 	gh->gh_ip = 0;
 }
 
diff --git a/fs/gfs2/glock.h b/fs/gfs2/glock.h
index 46ab67f..ab1ef32 100644
--- a/fs/gfs2/glock.h
+++ b/fs/gfs2/glock.h
@@ -247,4 +247,14 @@ extern void gfs2_unregister_debugfs(void);
 
 extern const struct lm_lockops gfs2_dlm_ops;
 
+static inline void gfs2_holder_mark_uninitialized(struct gfs2_holder *gh)
+{
+	gh->gh_gl = NULL;
+}
+
+static inline bool gfs2_holder_initialized(struct gfs2_holder *gh)
+{
+	return gh->gh_gl;
+}
+
 #endif /* __GLOCK_DOT_H__ */
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 481b649..de54d60 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -127,9 +127,9 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 	struct gfs2_inode *ip;
 	struct gfs2_glock *io_gl = NULL;
 	struct gfs2_holder i_gh;
-	bool unlock = false;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&i_gh);
 	inode = gfs2_iget(sb, no_addr);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
@@ -159,7 +159,6 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 						   GL_SKIP, &i_gh);
 			if (error)
 				goto fail_put;
-			unlock = true;
 
 			if (blktype != GFS2_BLKST_FREE) {
 				error = gfs2_check_blk_type(sdp, no_addr,
@@ -191,7 +190,7 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 		unlock_new_inode(inode);
 	}
 
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 	return inode;
 
@@ -203,7 +202,7 @@ fail_refresh:
 fail_put:
 	if (io_gl)
 		gfs2_glock_put(io_gl);
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 	ip->i_gl->gl_object = NULL;
 fail:
@@ -281,8 +280,8 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 	struct gfs2_holder d_gh;
 	int error = 0;
 	struct inode *inode = NULL;
-	int unlock = 0;
 
+	gfs2_holder_mark_uninitialized(&d_gh);
 	if (!name->len || name->len > GFS2_FNAMESIZE)
 		return ERR_PTR(-ENAMETOOLONG);
 
@@ -297,7 +296,6 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 		error = gfs2_glock_nq_init(dip->i_gl, LM_ST_SHARED, 0, &d_gh);
 		if (error)
 			return ERR_PTR(error);
-		unlock = 1;
 	}
 
 	if (!is_root) {
@@ -310,7 +308,7 @@ struct inode *gfs2_lookupi(struct inode *dir, const struct qstr *name,
 	if (IS_ERR(inode))
 		error = PTR_ERR(inode);
 out:
-	if (unlock)
+	if (gfs2_holder_initialized(&d_gh))
 		gfs2_glock_dq_uninit(&d_gh);
 	if (error == -ENOENT)
 		return NULL;
@@ -1354,7 +1352,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 	struct gfs2_inode *ip = GFS2_I(d_inode(odentry));
 	struct gfs2_inode *nip = NULL;
 	struct gfs2_sbd *sdp = GFS2_SB(odir);
-	struct gfs2_holder ghs[5], r_gh = { .gh_gl = NULL, };
+	struct gfs2_holder ghs[5], r_gh;
 	struct gfs2_rgrpd *nrgd;
 	unsigned int num_gh;
 	int dir_rename = 0;
@@ -1362,6 +1360,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 	unsigned int x;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&r_gh);
 	if (d_really_is_positive(ndentry)) {
 		nip = GFS2_I(d_inode(ndentry));
 		if (ip == nip)
@@ -1551,7 +1550,7 @@ out_gunlock:
 		gfs2_holder_uninit(ghs + x);
 	}
 out_gunlock_r:
-	if (r_gh.gh_gl)
+	if (gfs2_holder_initialized(&r_gh))
 		gfs2_glock_dq_uninit(&r_gh);
 out:
 	return error;
@@ -1577,13 +1576,14 @@ static int gfs2_exchange(struct inode *odir, struct dentry *odentry,
 	struct gfs2_inode *oip = GFS2_I(odentry->d_inode);
 	struct gfs2_inode *nip = GFS2_I(ndentry->d_inode);
 	struct gfs2_sbd *sdp = GFS2_SB(odir);
-	struct gfs2_holder ghs[5], r_gh = { .gh_gl = NULL, };
+	struct gfs2_holder ghs[5], r_gh;
 	unsigned int num_gh;
 	unsigned int x;
 	umode_t old_mode = oip->i_inode.i_mode;
 	umode_t new_mode = nip->i_inode.i_mode;
 	int error;
 
+	gfs2_holder_mark_uninitialized(&r_gh);
 	error = gfs2_rindex_update(sdp);
 	if (error)
 		return error;
@@ -1691,7 +1691,7 @@ out_gunlock:
 		gfs2_holder_uninit(ghs + x);
 	}
 out_gunlock_r:
-	if (r_gh.gh_gl)
+	if (gfs2_holder_initialized(&r_gh))
 		gfs2_glock_dq_uninit(&r_gh);
 out:
 	return error;
@@ -1788,9 +1788,8 @@ int gfs2_permission(struct inode *inode, int mask)
 	struct gfs2_inode *ip;
 	struct gfs2_holder i_gh;
 	int error;
-	int unlock = 0;
-
 
+	gfs2_holder_mark_uninitialized(&i_gh);
 	ip = GFS2_I(inode);
 	if (gfs2_glock_is_locked_by_me(ip->i_gl) == NULL) {
 		if (mask & MAY_NOT_BLOCK)
@@ -1798,14 +1797,13 @@ int gfs2_permission(struct inode *inode, int mask)
 		error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY, &i_gh);
 		if (error)
 			return error;
-		unlock = 1;
 	}
 
 	if ((mask & MAY_WRITE) && IS_IMMUTABLE(inode))
 		error = -EACCES;
 	else
 		error = generic_permission(inode, mask);
-	if (unlock)
+	if (gfs2_holder_initialized(&i_gh))
 		gfs2_glock_dq_uninit(&i_gh);
 
 	return error;
@@ -1977,17 +1975,16 @@ static int gfs2_getattr(struct vfsmount *mnt, struct dentry *dentry,
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_holder gh;
 	int error;
-	int unlock = 0;
 
+	gfs2_holder_mark_uninitialized(&gh);
 	if (gfs2_glock_is_locked_by_me(ip->i_gl) == NULL) {
 		error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY, &gh);
 		if (error)
 			return error;
-		unlock = 1;
 	}
 
 	generic_fillattr(inode, stat);
-	if (unlock)
+	if (gfs2_holder_initialized(&gh))
 		gfs2_glock_dq_uninit(&gh);
 
 	return 0;
diff --git a/fs/gfs2/main.c b/fs/gfs2/main.c
index 615f675..74fd013 100644
--- a/fs/gfs2/main.c
+++ b/fs/gfs2/main.c
@@ -45,7 +45,7 @@ static void gfs2_init_inode_once(void *foo)
 	memset(&ip->i_res, 0, sizeof(ip->i_res));
 	RB_CLEAR_NODE(&ip->i_res.rs_node);
 	ip->i_hash_cache = NULL;
-	ip->i_iopen_gh.gh_gl = NULL;
+	gfs2_holder_mark_uninitialized(&ip->i_iopen_gh);
 }
 
 static void gfs2_init_glock_once(void *foo)
diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c
index ce7d69a..6c657b2 100644
--- a/fs/gfs2/quota.c
+++ b/fs/gfs2/quota.c
@@ -883,7 +883,7 @@ static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
 	gfs2_write_calc_reserv(ip, sizeof(struct gfs2_quota),
 			      &data_blocks, &ind_blocks);
 
-	ghs = kcalloc(num_qd, sizeof(struct gfs2_holder), GFP_NOFS);
+	ghs = kmalloc(num_qd * sizeof(struct gfs2_holder), GFP_NOFS);
 	if (!ghs)
 		return -ENOMEM;
 
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 960aaf4..fba38ca 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -2100,7 +2100,7 @@ void gfs2_inplace_release(struct gfs2_inode *ip)
 {
 	struct gfs2_blkreserv *rs = &ip->i_res;
 
-	if (rs->rs_rgd_gh.gh_gl)
+	if (gfs2_holder_initialized(&rs->rs_rgd_gh))
 		gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
 }
 
@@ -2600,7 +2600,7 @@ void gfs2_rlist_alloc(struct gfs2_rgrp_list *rlist, unsigned int state)
 {
 	unsigned int x;
 
-	rlist->rl_ghs = kcalloc(rlist->rl_rgrps, sizeof(struct gfs2_holder),
+	rlist->rl_ghs = kmalloc(rlist->rl_rgrps * sizeof(struct gfs2_holder),
 				GFP_NOFS | __GFP_NOFAIL);
 	for (x = 0; x < rlist->rl_rgrps; x++)
 		gfs2_holder_init(rlist->rl_rgd[x]->rd_gl,
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 9b2ff353..3a7e60b 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -855,7 +855,7 @@ static int gfs2_make_fs_ro(struct gfs2_sbd *sdp)
 	wait_event(sdp->sd_reserving_log_wait, atomic_read(&sdp->sd_reserving_log) == 0);
 	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_blks_free) == sdp->sd_jdesc->jd_blocks);
 
-	if (freeze_gh.gh_gl)
+	if (gfs2_holder_initialized(&freeze_gh))
 		gfs2_glock_dq_uninit(&freeze_gh);
 
 	gfs2_quota_cleanup(sdp);
@@ -1033,7 +1033,7 @@ static int gfs2_unfreeze(struct super_block *sb)
 
 	mutex_lock(&sdp->sd_freeze_mutex);
         if (atomic_read(&sdp->sd_freeze_state) != SFS_FROZEN ||
-	    sdp->sd_freeze_gh.gh_gl == NULL) {
+	    !gfs2_holder_initialized(&sdp->sd_freeze_gh)) {
 		mutex_unlock(&sdp->sd_freeze_mutex);
                 return 0;
 	}
@@ -1084,9 +1084,11 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 	int error = 0, err;
 
 	memset(sc, 0, sizeof(struct gfs2_statfs_change_host));
-	gha = kcalloc(slots, sizeof(struct gfs2_holder), GFP_KERNEL);
+	gha = kmalloc(slots * sizeof(struct gfs2_holder), GFP_KERNEL);
 	if (!gha)
 		return -ENOMEM;
+	for (x = 0; x < slots; x++)
+		gfs2_holder_mark_uninitialized(gha + x);
 
 	rgd_next = gfs2_rgrpd_get_first(sdp);
 
@@ -1096,7 +1098,7 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 		for (x = 0; x < slots; x++) {
 			gh = gha + x;
 
-			if (gh->gh_gl && gfs2_glock_poll(gh)) {
+			if (gfs2_holder_initialized(gh) && gfs2_glock_poll(gh)) {
 				err = gfs2_glock_wait(gh);
 				if (err) {
 					gfs2_holder_uninit(gh);
@@ -1109,7 +1111,7 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host
 				}
 			}
 
-			if (gh->gh_gl)
+			if (gfs2_holder_initialized(gh))
 				done = 0;
 			else if (rgd_next && !error) {
 				error = gfs2_glock_nq_init(rgd_next->rd_gl,
@@ -1304,9 +1306,11 @@ static int gfs2_drop_inode(struct inode *inode)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 
-	if (!test_bit(GIF_FREE_VFS_INODE, &ip->i_flags) && inode->i_nlink) {
+	if (!test_bit(GIF_FREE_VFS_INODE, &ip->i_flags) &&
+	    inode->i_nlink &&
+	    gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl;
-		if (gl && test_bit(GLF_DEMOTE, &gl->gl_flags))
+		if (test_bit(GLF_DEMOTE, &gl->gl_flags))
 			clear_nlink(inode);
 	}
 	return generic_drop_inode(inode);
@@ -1551,7 +1555,7 @@ static void gfs2_evict_inode(struct inode *inode)
 			goto out_truncate;
 	}
 
-	if (ip->i_iopen_gh.gh_gl &&
+	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;
 		gfs2_glock_dq_wait(&ip->i_iopen_gh);
@@ -1610,7 +1614,7 @@ out_unlock:
 	if (gfs2_rs_active(&ip->i_res))
 		gfs2_rs_deltree(&ip->i_res);
 
-	if (ip->i_iopen_gh.gh_gl) {
+	if (gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) {
 			ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 			gfs2_glock_dq_wait(&ip->i_iopen_gh);
@@ -1632,7 +1636,7 @@ out:
 	gfs2_glock_add_to_lru(ip->i_gl);
 	gfs2_glock_put(ip->i_gl);
 	ip->i_gl = NULL;
-	if (ip->i_iopen_gh.gh_gl) {
+	if (gfs2_holder_initialized(&ip->i_iopen_gh)) {
 		ip->i_iopen_gh.gh_gl->gl_object = NULL;
 		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 		gfs2_glock_dq_wait(&ip->i_iopen_gh);
-- 
2.5.5



^ permalink raw reply related	[flat|nested] 12+ messages in thread
* Re: [Cluster-devel] [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock
  2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
@ 2016-06-24 11:04   ` Steven Whitehouse
  -1 siblings, 0 replies; 12+ messages in thread
From: Steven Whitehouse @ 2016-06-24 11:04 UTC (permalink / raw)
  To: Bob Peterson, linux-fsdevel; +Cc: cluster-devel, Alexander Viro

Copying in Al for comment. The patches look good to me though from the 
GFS2 perspective. Lets make sure that we do test them very carefully, 
since this is a very tricky area to get right,

Steve.

On 20/06/16 16:55, Bob Peterson wrote:
> Hi,
>
> This is a set of four patches from Andreas Gruenbacher that fix the
> gfs2_lookup_by_inum deadlock problem, plus a few bugs we've uncovered
> during testing, and a cleanup or two. I've been working with Andreas
> for a while now, and we've both made several attempts to fix this
> problem in the past, in regard to the transition of dinodes from the
> "unlinked" to the "free" state. This is the latest attempt, and it
> seems to be working well.
>
> I sent the first 3 patches on 15 June, and Andreas had posted some of
> this as well, but I wanted to add the fourth patch, plus cc linux-fsdevel.
>
> Prior to that, a previous version was posted which included a patch to
> VFS, but Al Viro didn't like that. That design was scrapped in favor
> of this one, which is simpler and confined to GFS2. Hopefully Al will
> like this implementation.
>
> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
> Signed-off-by: Bob Peterson <rpeterso@redhat.com>
> ---
> Andreas Gruenbacher (4):
>    gfs2: Fix gfs2_lookup_by_inum lock inversion
>    gfs2: Get rid of gfs2_ilookup
>    gfs2: Large-filesystem fix for 32-bit systems
>    gfs2: Lock holder cleanup
>
>   fs/gfs2/dentry.c     |   2 +-
>   fs/gfs2/dir.c        |   3 +-
>   fs/gfs2/export.c     |  11 -----
>   fs/gfs2/file.c       |   2 +-
>   fs/gfs2/glock.c      |  11 +----
>   fs/gfs2/glock.h      |  10 ++++
>   fs/gfs2/inode.c      | 128 ++++++++++++++++++++++++++++++++++-----------------
>   fs/gfs2/inode.h      |   4 +-
>   fs/gfs2/main.c       |   2 +-
>   fs/gfs2/ops_fstype.c |   3 +-
>   fs/gfs2/quota.c      |   2 +-
>   fs/gfs2/rgrp.c       |   4 +-
>   fs/gfs2/super.c      |  24 ++++++----
>   13 files changed, 123 insertions(+), 83 deletions(-)
>


^ permalink raw reply	[flat|nested] 12+ messages in thread
* [Cluster-devel] [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock
@ 2016-06-24 11:04   ` Steven Whitehouse
  0 siblings, 0 replies; 12+ messages in thread
From: Steven Whitehouse @ 2016-06-24 11:04 UTC (permalink / raw)
  To: cluster-devel.redhat.com

Copying in Al for comment. The patches look good to me though from the 
GFS2 perspective. Lets make sure that we do test them very carefully, 
since this is a very tricky area to get right,

Steve.

On 20/06/16 16:55, Bob Peterson wrote:
> Hi,
>
> This is a set of four patches from Andreas Gruenbacher that fix the
> gfs2_lookup_by_inum deadlock problem, plus a few bugs we've uncovered
> during testing, and a cleanup or two. I've been working with Andreas
> for a while now, and we've both made several attempts to fix this
> problem in the past, in regard to the transition of dinodes from the
> "unlinked" to the "free" state. This is the latest attempt, and it
> seems to be working well.
>
> I sent the first 3 patches on 15 June, and Andreas had posted some of
> this as well, but I wanted to add the fourth patch, plus cc linux-fsdevel.
>
> Prior to that, a previous version was posted which included a patch to
> VFS, but Al Viro didn't like that. That design was scrapped in favor
> of this one, which is simpler and confined to GFS2. Hopefully Al will
> like this implementation.
>
> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
> Signed-off-by: Bob Peterson <rpeterso@redhat.com>
> ---
> Andreas Gruenbacher (4):
>    gfs2: Fix gfs2_lookup_by_inum lock inversion
>    gfs2: Get rid of gfs2_ilookup
>    gfs2: Large-filesystem fix for 32-bit systems
>    gfs2: Lock holder cleanup
>
>   fs/gfs2/dentry.c     |   2 +-
>   fs/gfs2/dir.c        |   3 +-
>   fs/gfs2/export.c     |  11 -----
>   fs/gfs2/file.c       |   2 +-
>   fs/gfs2/glock.c      |  11 +----
>   fs/gfs2/glock.h      |  10 ++++
>   fs/gfs2/inode.c      | 128 ++++++++++++++++++++++++++++++++++-----------------
>   fs/gfs2/inode.h      |   4 +-
>   fs/gfs2/main.c       |   2 +-
>   fs/gfs2/ops_fstype.c |   3 +-
>   fs/gfs2/quota.c      |   2 +-
>   fs/gfs2/rgrp.c       |   4 +-
>   fs/gfs2/super.c      |  24 ++++++----
>   13 files changed, 123 insertions(+), 83 deletions(-)
>



^ permalink raw reply	[flat|nested] 12+ messages in thread
end of thread, other threads:[~2016-06-24 11:04 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-06-20 15:55 [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock Bob Peterson
2016-06-20 15:55 ` [Cluster-devel] " Bob Peterson
2016-06-20 15:55 ` [PATCH 1/4] gfs2: Fix gfs2_lookup_by_inum lock inversion Bob Peterson
2016-06-20 15:55   ` [Cluster-devel] " Bob Peterson
2016-06-20 15:55 ` [PATCH 2/4] gfs2: Get rid of gfs2_ilookup Bob Peterson
2016-06-20 15:55   ` [Cluster-devel] " Bob Peterson
2016-06-20 15:55 ` [PATCH 3/4] gfs2: Large-filesystem fix for 32-bit systems Bob Peterson
2016-06-20 15:55   ` [Cluster-devel] " Bob Peterson
2016-06-20 15:55 ` [PATCH 4/4] gfs2: Lock holder cleanup Bob Peterson
2016-06-20 15:55   ` [Cluster-devel] " Bob Peterson
2016-06-24 11:04 ` [Cluster-devel] [PATCH 0/4] Patches for gfs2_lookup_by_inum deadlock Steven Whitehouse
2016-06-24 11:04   ` Steven Whitehouse

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.