[PATCH 0/2] xfs: minor inode unlink debug enhancements

[PATCH 0/2] xfs: minor inode unlink debug enhancements

[Brian Foster]

The following couple of patches provide a bit more debug info in the inode
unlink code paths. This data proved useful for the investigation around a
recent (downstream/distro) link/unlink race problem:

http://oss.sgi.com/archives/xfs/2013-04/msg00342.html

Brian

Brian Foster (1):
  xfs: tracepoints for xfs_iunlink() and xfs_iunlink_remove()

Eric Sandeen (1):
  xfs: enhance warning on xfs_imap() failure in xfs_iunlink_remove()

 fs/xfs/xfs_inode.c |   10 ++++++++--
 fs/xfs/xfs_trace.h |    2 ++
 2 files changed, 10 insertions(+), 2 deletions(-)

-- 
1.7.7.6

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 1/2] xfs: tracepoints for xfs_iunlink() and xfs_iunlink_remove()

[Brian Foster]

Add tracepoints for xfs_iunlink() and xfs_iunlink_remove(). These
tracepoints are useful for debugging inode link (and unlinked list
issues).

Signed-off-by: Brian Foster <bfoster@redhat.com>
---
 fs/xfs/xfs_inode.c |    4 ++++
 fs/xfs/xfs_trace.h |    2 ++
 2 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 4f20165..c22f980 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1519,6 +1519,8 @@ xfs_iunlink(
 
 	mp = tp->t_mountp;
 
+	trace_xfs_iunlink(ip);
+
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
@@ -1598,6 +1600,8 @@ xfs_iunlink_remove(
 	mp = tp->t_mountp;
 	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
 
+	trace_xfs_iunlink_remove(ip);
+
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index aa4db33..43bd572 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -586,6 +586,8 @@ DEFINE_INODE_EVENT(xfs_file_fsync);
 DEFINE_INODE_EVENT(xfs_destroy_inode);
 DEFINE_INODE_EVENT(xfs_evict_inode);
 DEFINE_INODE_EVENT(xfs_update_time);
+DEFINE_INODE_EVENT(xfs_iunlink);
+DEFINE_INODE_EVENT(xfs_iunlink_remove);
 
 DEFINE_INODE_EVENT(xfs_dquot_dqalloc);
 DEFINE_INODE_EVENT(xfs_dquot_dqdetach);
-- 
1.7.7.6

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 2/2] xfs: enhance warning on xfs_imap() failure in xfs_iunlink_remove()

[Brian Foster]

From: Eric Sandeen <sandeen@redhat.com>

Include the global ino, ag number and agino in the warning message
emitted should xfs_imap() fail in the xfs_iunlink_remove() code
path.

Signed-off-by: Brian Foster <bfoster@redhat.com>
---
 fs/xfs/xfs_inode.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index c22f980..a4fe989 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1678,8 +1678,10 @@ xfs_iunlink_remove(
 			error = xfs_imap(mp, tp, next_ino, &imap, 0);
 			if (error) {
 				xfs_warn(mp,
-	"%s: xfs_imap returned error %d.",
-					 __func__, error);
+					"%s: xfs_imap() returned error %d "
+					"for inode 0x%llx ag %d agino 0x%x",
+					__func__, error, ip->i_ino, agno, agino);
+
 				return error;
 			}
 
-- 
1.7.7.6

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfs: tracepoints for xfs_iunlink() and xfs_iunlink_remove()

[Dave Chinner]

On Thu, Apr 25, 2013 at 11:41:29AM -0400, Brian Foster wrote:
> Add tracepoints for xfs_iunlink() and xfs_iunlink_remove(). These
> tracepoints are useful for debugging inode link (and unlinked list
> issues).
> 
> Signed-off-by: Brian Foster <bfoster@redhat.com>

Looks good.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 2/2] xfs: enhance warning on xfs_imap() failure in xfs_iunlink_remove()

[Dave Chinner]

On Thu, Apr 25, 2013 at 11:41:30AM -0400, Brian Foster wrote:
> From: Eric Sandeen <sandeen@redhat.com>
> 
> Include the global ino, ag number and agino in the warning message
> emitted should xfs_imap() fail in the xfs_iunlink_remove() code
> path.
> 
> Signed-off-by: Brian Foster <bfoster@redhat.com>
> ---
>  fs/xfs/xfs_inode.c |    6 ++++--
>  1 files changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
> index c22f980..a4fe989 100644
> --- a/fs/xfs/xfs_inode.c
> +++ b/fs/xfs/xfs_inode.c
> @@ -1678,8 +1678,10 @@ xfs_iunlink_remove(
>  			error = xfs_imap(mp, tp, next_ino, &imap, 0);
>  			if (error) {
>  				xfs_warn(mp,
> -	"%s: xfs_imap returned error %d.",
> -					 __func__, error);
> +					"%s: xfs_imap() returned error %d "
> +					"for inode 0x%llx ag %d agino 0x%x",
> +					__func__, error, ip->i_ino, agno, agino);
> +

The fact the format string is now split across 2 lines makes it hard
to grep for. The indentation should be a clue that the warning code
is intentionally formatted that way so that long format strings
don't get wrapped. i.e:

 				xfs_warn(mp,
-	"%s: xfs_imap returned error %d.",
-					 __func__, error);
+	"%s: xfs_imap() returned error %d for inode 0x%llx ag %d agino 0x%x",
+					__func__, error, ip->i_ino,
+					agno, agino);

fits into 80 columns and is much easier to read than the highly
indented version split across 2 lines....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs