From: Mark Tinguely <tinguely@sgi.com>
To: stable@vger.kernel.org
Cc: xfs@oss.sgi.com
Subject: [3.0-stable PATCH 33/36] xfs: check for stale inode before acquiring iflock on push
Date: Mon, 03 Dec 2012 17:42:41 -0600 [thread overview]
Message-ID: <20121203144311.853211105@sgi.com> (raw)
In-Reply-To: 20121203144208.143464631@sgi.com
[-- Attachment #1: 103 --]
[-- Type: text/plain, Size: 2295 bytes --]
From: Brian Foster <bfoster@redhat.com>
Upstream commit: 9a3a5dab63461b84213052888bf38a962b22d035
An inode in the AIL can be flush locked and marked stale if
a cluster free transaction occurs at the right time. The
inode item is then marked as flushing, which causes xfsaild
to spin and leaves the filesystem stalled. This is
reproduced by running xfstests 273 in a loop for an
extended period of time.
Check for stale inodes before the flush lock. This marks
the inode as pinned, leads to a log flush and allows the
filesystem to proceed.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
fs/xfs/xfs_inode_item.c | 26 +++++++++++++++-----------
1 file changed, 15 insertions(+), 11 deletions(-)
Index: b/fs/xfs/xfs_inode_item.c
===================================================================
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -588,6 +588,21 @@ xfs_inode_item_trylock(
if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
return XFS_ITEM_LOCKED;
+ /*
+ * Re-check the pincount now that we stabilized the value by
+ * taking the ilock.
+ */
+ if (xfs_ipincount(ip) > 0) {
+ xfs_iunlock(ip, XFS_ILOCK_SHARED);
+ return XFS_ITEM_PINNED;
+ }
+
+ /* Stale items should force out the iclog */
+ if (ip->i_flags & XFS_ISTALE) {
+ xfs_iunlock(ip, XFS_ILOCK_SHARED);
+ return XFS_ITEM_PINNED;
+ }
+
if (!xfs_iflock_nowait(ip)) {
/*
* inode has already been flushed to the backing buffer,
@@ -597,17 +612,6 @@ xfs_inode_item_trylock(
return XFS_ITEM_PUSHBUF;
}
- /* Stale items should force out the iclog */
- if (ip->i_flags & XFS_ISTALE) {
- xfs_ifunlock(ip);
- /*
- * we hold the AIL lock - notify the unlock routine of this
- * so it doesn't try to get the lock again.
- */
- xfs_iunlock(ip, XFS_ILOCK_SHARED|XFS_IUNLOCK_NONOTIFY);
- return XFS_ITEM_PINNED;
- }
-
#ifdef DEBUG
if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
ASSERT(iip->ili_format.ilf_fields != 0);
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
next prev parent reply other threads:[~2012-12-03 23:40 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-12-03 23:42 [3.0-stable PATCH 00/36] Proposed 3.0-stable bug patches Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 01/36] xfs: fix possible overflow in xfs_ioc_trim() Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 02/36] xfs: fix allocation length overflow in xfs_bmapi_write() Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 03/36] xfs: mark the xfssyncd workqueue as non-reentrant Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 04/36] xfs: xfs_trans_add_item() - dont assign in ASSERT() when compare is intended Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 05/36] xfs: only take the ILOCK in xfs_reclaim_inode() Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 06/36] xfs: fallback to vmalloc for large buffers in xfs_attrmulti_attr_get Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 07/36] xfs: fallback to vmalloc for large buffers in xfs_getbmap Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 08/36] xfs: fix deadlock in xfs_rtfree_extent Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 09/36] xfs: Fix open flag handling in open_by_handle code Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 10/36] xfs: Account log unmount transaction correctly Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 11/36] xfs: fix fstrim offset calculations Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 12/36] xfs: dont fill statvfs with project quota for a directory Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 13/36] xfs: Ensure inode reclaim can run during quotacheck Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 14/36] xfs: use shared ilock mode for direct IO writes by default Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 15/36] xfs: punch all delalloc blocks beyond EOF on write failure Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 16/36] xfs: page type check in writeback only checks last buffer Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 17/36] xfs: punch new delalloc blocks out of failed writes inside EOF Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 18/36] xfs: dont assert on delalloc regions beyond EOF Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 19/36] xfs: limit specualtive delalloc to maxioffset Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 20/36] xfs: Use preallocation for inodes with extsz hints Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 21/36] xfs: Dont allocate new buffers on every call to _xfs_buf_find Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 22/36] xfs: clean up buffer allocation Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 23/36] xfs: fix buffer lookup race on allocation failure Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 24/36] xfs: use iolock on XFS_IOC_ALLOCSP calls Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 25/36] xfs: Properly exclude IO type flags from buffer flags Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 26/36] xfs: flush outstanding buffers on log mount failure Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 27/36] xfs: protect xfs_sync_worker with s_umount semaphore Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 28/36] xfs: fix memory reclaim deadlock on agi buffer Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 29/36] xfs: xfs_vm_writepage clear iomap_valid when Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 30/36] xfs: fix allocbt cursor leak in xfs_alloc_ag_vextent_near Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 31/36] xfs: shutdown xfs_sync_worker before the log Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 32/36] xfs: really fix the cursor leak in xfs_alloc_ag_vextent_near Mark Tinguely
2012-12-03 23:42 ` Mark Tinguely [this message]
2012-12-03 23:42 ` [3.0-stable PATCH 34/36] xfs: stop the sync worker before xfs_unmountfs Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 35/36] xfs: zero allocation_args on the kernel stack Mark Tinguely
2012-12-03 23:42 ` [3.0-stable PATCH 36/36] xfs: only update the last_sync_lsn when a transaction completes Mark Tinguely
2012-12-04 21:44 ` [3.0-stable PATCH 00/36] Proposed 3.0-stable bug patches Ben Myers
2012-12-05 21:45 ` Dave Chinner
2012-12-06 17:27 ` Mark Tinguely
2012-12-07 10:06 ` Dave Chinner
2012-12-07 21:15 ` Ben Myers
2012-12-08 12:06 ` Christoph Hellwig
2012-12-08 19:12 ` Greg KH
2012-12-10 0:24 ` Dave Chinner
2012-12-10 22:03 ` Ben Myers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20121203144311.853211105@sgi.com \
--to=tinguely@sgi.com \
--cc=stable@vger.kernel.org \
--cc=xfs@oss.sgi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox