linux.git/fs/xfs, branch v2.6.26 Linux kernel source tree Fix reference counting race on log buffers 2008-07-11T18:37:18+00:00 Dave Chinner david@fromorbit.com 2008-07-11T07:43:55+00:00 49641f1acfdfd437ed9b0a70b86bf36626c02afe When we release the iclog, we do an atomic_dec_and_lock to determine if we are the last reference and need to trigger update of log headers and writeout. However, in xlog_state_get_iclog_space() we also need to check if we have the last reference count there. If we do, we release the log buffer, otherwise we decrement the reference count. But the compare and decrement in xlog_state_get_iclog_space() is not atomic, so both places can see a reference count of 2 and neither will release the iclog. That leads to a filesystem hang. Close the race by replacing the atomic_read() and atomic_dec() pair with atomic_add_unless() to ensure that they are executed atomically. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Tim Shimmin <tes@sgi.com> Tested-by: Eric Sandeen <sandeen@sandeen.net> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
When we release the iclog, we do an atomic_dec_and_lock to determine if
we are the last reference and need to trigger update of log headers and
writeout.  However, in xlog_state_get_iclog_space() we also need to
check if we have the last reference count there.  If we do, we release
the log buffer, otherwise we decrement the reference count.

But the compare and decrement in xlog_state_get_iclog_space() is not
atomic, so both places can see a reference count of 2 and neither will
release the iclog.  That leads to a filesystem hang.

Close the race by replacing the atomic_read() and atomic_dec() pair with
atomic_add_unless() to ensure that they are executed atomically.

Signed-off-by: Dave Chinner <david@fromorbit.com>
Reviewed-by: Tim Shimmin <tes@sgi.com>
Tested-by: Eric Sandeen <sandeen@sandeen.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[XFS] Fix memory corruption with small buffer reads 2008-05-23T08:12:49+00:00 Christoph Hellwig hch@infradead.org 2008-05-19T06:34:42+00:00 6ab455eeaff6893cd06da33843e840d888cdc04a When we have multiple buffers in a single page for a blocksize == pagesize filesystem we might overwrite the page contents if two callers hit it shortly after each other. To prevent that we need to keep the page locked until I/O is completed and the page marked uptodate. Thanks to Eric Sandeen for triaging this bug and finding a reproducible testcase and Dave Chinner for additional advice. This should fix kernel.org bz #10421. Tested-by: Eric Sandeen <sandeen@sandeen.net> SGI-PV: 981813 SGI-Modid: xfs-linux-melb:xfs-kern:31173a Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
When we have multiple buffers in a single page for a blocksize == pagesize
filesystem we might overwrite the page contents if two callers hit it
shortly after each other. To prevent that we need to keep the page locked
until I/O is completed and the page marked uptodate.

Thanks to Eric Sandeen for triaging this bug and finding a reproducible
testcase and Dave Chinner for additional advice.

This should fix kernel.org bz #10421.

Tested-by: Eric Sandeen <sandeen@sandeen.net>

SGI-PV: 981813
SGI-Modid: xfs-linux-melb:xfs-kern:31173a

Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] Fix inode list allocation size in writeback. 2008-05-23T05:26:15+00:00 David Chinner dgc@sgi.com 2008-05-20T01:30:15+00:00 c8f5f12e46f079a954d4f7163ba59dadee08ca26 We only need to allocate space for the number of inodes in the cluster when writing back inodes, not every byte in the inode cluster. This reduces the amount of memory needing to be allocated to 256 bytes instead of 64k. SGI-PV: 981949 SGI-Modid: xfs-linux-melb:xfs-kern:31182a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
We only need to allocate space for the number of inodes in the cluster
when writing back inodes, not every byte in the inode cluster. This
reduces the amount of memory needing to be allocated to 256 bytes instead
of 64k.

SGI-PV: 981949
SGI-Modid: xfs-linux-melb:xfs-kern:31182a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] Don't allow memory reclaim to wait on the filesystem in inode 2008-05-23T05:26:03+00:00 David Chinner dgc@sgi.com 2008-05-19T06:29:34+00:00 49383b0e98ad1f69ff4c816eb1961f703df12318 writeback If we allow memory reclaim to wait on the pages under writeback in inode cluster writeback we could deadlock because we are currently holding the ILOCK on the initial writeback inode which is needed in data I/O completion to change the file size or do unwritten extent conversion before the pages are taken out of writeback state. SGI-PV: 981091 SGI-Modid: xfs-linux-melb:xfs-kern:31015a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
writeback

If we allow memory reclaim to wait on the pages under writeback in inode
cluster writeback we could deadlock because we are currently holding the
ILOCK on the initial writeback inode which is needed in data I/O
completion to change the file size or do unwritten extent conversion
before the pages are taken out of writeback state.

SGI-PV: 981091
SGI-Modid: xfs-linux-melb:xfs-kern:31015a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] Fix fsync() b0rkage. 2008-05-23T05:25:25+00:00 David Chinner dgc@sgi.com 2008-05-19T06:29:46+00:00 978b7237123d007b9fa983af6e0e2fa8f97f9934 xfs_fsync() fails to wait for data I/O completion before checking if the inode is dirty or clean to decide whether to log the inode or not. This misses inode size updates when the data flushed by the fsync() is extending the file. Hence, like fdatasync(), we need to wait for I/o completion first, then check the inode for cleanliness. Doing so makes the behaviour of xfs_fsync() identical for fsync and fdatasync and we *always* use synchronous semantics if the inode is dirty. Therefore also kill the differences and remove the unused flags from the xfs_fsync function and callers. SGI-PV: 981296 SGI-Modid: xfs-linux-melb:xfs-kern:31033a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
xfs_fsync() fails to wait for data I/O completion before checking if the
inode is dirty or clean to decide whether to log the inode or not. This
misses inode size updates when the data flushed by the fsync() is
extending the file.

Hence, like fdatasync(), we need to wait for I/o completion first, then
check the inode for cleanliness. Doing so makes the behaviour of
xfs_fsync() identical for fsync and fdatasync and we *always* use
synchronous semantics if the inode is dirty. Therefore also kill the
differences and remove the unused flags from the xfs_fsync function and
callers.

SGI-PV: 981296
SGI-Modid: xfs-linux-melb:xfs-kern:31033a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] Include linux/random.h in all builds, not just debug builds. 2008-04-30T08:17:44+00:00 David Chinner dgc@sgi.com 2008-04-30T08:15:34+00:00 a94477da38e0b261a7ecea71f4c95a3bcd5be69c SGI-PV: 979416 SGI-Modid: xfs-linux-melb:xfs-kern:31008a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
SGI-PV: 979416
SGI-Modid: xfs-linux-melb:xfs-kern:31008a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] Fix build failure after enabling CONFIG_XFS_DEBUG 2008-04-29T06:08:44+00:00 Stephen Rothwell sfr@canb.auug.org.au 2008-04-22T05:26:13+00:00 adaa693b845373296631766176ebf0f73a342e10 Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] remove dmapi cruft in xfs_file.c 2008-04-29T06:08:27+00:00 Christoph Hellwig hch@lst.de 2008-04-21T08:11:13+00:00 c5acbaf43da139fe014d78d1f0ca7754fa856ddb The dmapi cruft in xfs_file.c is totally out of date in mainline vs CVS, and at this point just removing this code which can't be used on mainline at all seems to be the best option to keep it maintainable. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
The dmapi cruft in xfs_file.c is totally out of date in mainline vs
CVS, and at this point just removing this code which can't be used on
mainline at all seems to be the best option to keep it maintainable.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] remove sendfile leftovers 2008-04-29T06:08:14+00:00 Christoph Hellwig hch@lst.de 2008-04-21T07:25:35+00:00 3a738a5c73e0617d11b27ac46dd6a1a8f752017b Remove the last sendfile leftovers in mainline. This code is already gone in CVS. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
Remove the last sendfile leftovers in mainline.  This code is already
gone in CVS.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
[XFS] allow enabling CONFIG_XFS_DEBUG 2008-04-29T06:07:48+00:00 Christoph Hellwig hch@lst.de 2008-04-21T07:22:27+00:00 7788fae6cce616fe2c624273fcfe54cf50f5c38b Back when I first submitted XFS for mainline inclusion we made the decision that the debug code is far to extensive to be accidentally enabled by users in mainline. But then again it's often quite useful to track problems down and hacking the makefile all the time is rather annoying. Given all the debug options with even more overhead like lockdep or DEBUG_PAGE_ALLOC users (or rather developers) should know by now what they're doing. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Lachlan McIlroy <lachlan@sgi.com> 
Back when I first submitted XFS for mainline inclusion we made the
decision that the debug code is far to extensive to be accidentally
enabled by users in mainline.  But then again it's often quite useful
to track problems down and hacking the makefile all the time is rather
annoying.  Given all the debug options with even more overhead like
lockdep or DEBUG_PAGE_ALLOC users (or rather developers) should know
by now what they're doing.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>