Unwritten extent conversion can recurse back into the filesystem due
to memory allocation. Memory reclaim requires I/O completions to be
processed to allow the callers to make progress. If the I/O
completion workqueue thread is doing the recursion, then we have a
deadlock situation.
Move unwritten extent completion into it's own workqueue so it
doesn't block I/O completions for normal delayed allocation or
overwrite data.
Signed-off-by: Dave Chinner <david@fromorbit.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Hi.
I introduced "is_partially_uptodate" aops for XFS.
A page can have multiple buffers and even if a page is not uptodate,
some buffers can be uptodate on pagesize != blocksize environment.
This aops checks that all buffers which correspond to a part of a file
that we want to read are uptodate. If so, we do not have to issue actual
read IO to HDD even if a page is not uptodate because the portion we
want to read are uptodate.
"block_is_partially_uptodate" function is already used by ext2/3/4.
With the following patch random read/write mixed workloads or random read
after random write workloads can be optimized and we can get performance
improvement.
I did a performance test using the sysbench.
#sysbench --num-threads=4 --max-requests=100000 --test=fileio --file-num=1 \
--file-block-size=8K --file-total-size=1G --file-test-mode=rndrw \
--file-fsync-freq=0 --file-rw-ratio=0.5 run
-2.6.29-rc6
Test execution summary:
total time: 123.8645s
total number of events: 100000
total time taken by event execution: 442.4994
per-request statistics:
min: 0.0000s
avg: 0.0044s
max: 0.3387s
approx. 95 percentile: 0.0118s
-2.6.29-rc6-patched
Test execution summary:
total time: 108.0757s
total number of events: 100000
total time taken by event execution: 417.7505
per-request statistics:
min: 0.0000s
avg: 0.0042s
max: 0.3217s
approx. 95 percentile: 0.0118s
arch: ia64
pagesize: 16k
blocksize: 4k
Signed-off-by: Hisashi Hifumi <hifumi.hisashi@oss.ntt.co.jp>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Felix Blyakher <felixb@sgi.com>
The whole machinery to wait on I/O completion is related to the I/O path
and should be there instead of in xfs_vnode.c. Also give the functions
more descriptive names.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Niv Sardi <xaiki@sgi.com>
There's just one caller of this helper, and it's much cleaner to just merge
the xfs_do_force_shutdown call into it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Niv Sardi <xaiki@sgi.com>
There are a few inode flags around that aren't used anywhere, so remove
them. Also update xfsidbg to display all used inode flags correctly.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Niv Sardi <xaiki@sgi.com>
Once the Linux inode and the XFS inode are combined, we cannot rely on
just check if the linux inode exists as a method of determining if it is
valid or not. Hence we should always call xfs_mark_inode_dirty_sync()
instead as it does the correct checks to determine if the liinux inode is
in a valid state or not.
SGI-PV: 988141
SGI-Modid: xfs-linux-melb:xfs-kern:32318a
Signed-off-by: David Chinner <david@fromorbit.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
With the help from some tracing I found that we try to map extents beyond
eof when doing a direct I/O read. It appears that the way to inform the
generic direct I/O path (ie do_direct_IO()) that we have breached eof is
to return an unmapped buffer from xfs_get_blocks_direct(). This will cause
do_direct_IO() to jump to the hole handling code where is will check for
eof and then abort.
This problem was found because a direct I/O read was trying to map beyond
eof and was encountering delayed allocations. The delayed allocations
beyond eof are speculative allocations and they didn't get converted when
the direct I/O flushed the file because there was only enough space in the
current AG to convert and write out the dirty pages within eof. Note that
xfs_iomap_write_allocate() wont necessarily convert all the delayed
allocation passed to it - it will return after allocating the first extent
- so if the delayed allocation extends beyond eof then it will stay that
way.
SGI-PV: 983683
SGI-Modid: xfs-linux-melb:xfs-kern:31929a
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
xfs_vtoi() is redundant and only unsed in small sections of code.
Replace them with widely used XFS_I() inline and kill xfs_vtoi().
SGI-PV: 981498
SGI-Modid: xfs-linux-melb:xfs-kern:31725a
Signed-off-by: David Chinner <david@fromorbit.com>
Signed-off-by: Niv Sardi <xaiki@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Like the page lock change, this also requires name change, so convert the
raw test_and_set bitop to a trylock.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Converting page lock to new locking bitops requires a change of page flag
operation naming, so we might as well convert it to something nicer
(!TestSetPageLocked_Lock => trylock_page, SetPageLocked => set_page_locked).
This also facilitates lockdeping of page lock.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On unwritten I/O completion, we fail to propagate an error when converting
the extent to a written extent. This means that the I/O silently fails.
propagate the error onto the ioend so that the inode is marked with an
error appropriately.
SGI-PV: 980084
SGI-Modid: xfs-linux-melb:xfs-kern:30826a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Niv Sardi <xaiki@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
We can just use xfs_ilock/xfs_iunlock instead and get rid of the ugly
bhv_vrwlock_t.
SGI-PV: 976035
SGI-Modid: xfs-linux-melb:xfs-kern:30533a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Use XFS_IS_REALTIME_INODE in more places, and #define it to 0 if
CONFIG_XFS_RT is off. This should be safe because mount checks in
xfs_rtmount_init:
so if we get mounted w/o CONFIG_XFS_RT, no realtime inodes should be
encountered after that.
Defining XFS_IS_REALTIME_INODE to 0 saves a bit of stack space,
presumeably gcc can optimize around the various "if (0)" type checks:
xfs_alloc_file_space -8 xfs_bmap_adjacent -16 xfs_bmapi -8
xfs_bmap_rtalloc -16 xfs_bunmapi -28 xfs_free_file_space -64 xfs_imap +8
<-- ? hmm. xfs_iomap_write_direct -12 xfs_qm_dqusage_adjust -4
xfs_qm_vop_chown_reserve -4
SGI-PV: 971186
SGI-Modid: xfs-linux-melb:xfs-kern:30014a
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
xfs_iocore_t is a structure embedded in xfs_inode. Except for one field it
just duplicates fields already in xfs_inode, and there is nothing this
abstraction buys us on XFS/Linux. This patch removes it and shrinks source
and binary size of xfs aswell as shrinking the size of xfs_inode by 60/44
bytes in debug/non-debug builds.
SGI-PV: 970852
SGI-Modid: xfs-linux-melb:xfs-kern:29754a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
Currently there is an indirection called ioops in the XFS data I/O path.
Various functions are called by functions pointers, but there is no
coherence in what this is for, and of course for XFS itself it's entirely
unused. This patch removes it instead and significantly reduces source and
binary size of XFS while making maintaince easier.
SGI-PV: 970841
SGI-Modid: xfs-linux-melb:xfs-kern:29737a
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Tim Shimmin <tes@sgi.com>
There is no reason to go through xfs_iomap for the BMAPI_UNWRITTEN because
it has nothing in common with the other cases. Instead check for the
shutdown filesystem in xfs_end_bio_unwritten and perform a direct call to
xfs_iomap_write_unwritten (which should be renamed to something more
sensible one day)
SGI-PV: 970241
SGI-Modid: xfs-linux-melb:xfs-kern:29681a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Donald Douwsma <donaldd@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
There is no reason to go into the iomap machinery just to get the right
block device for an inode. Instead look at the realtime flag in the inode
and grab the right device from the mount structure.
I created a new helper, xfs_find_bdev_for_inode instead of opencoding it
because I plan to use it in other places in the future.
SGI-PV: 970240
SGI-Modid: xfs-linux-melb:xfs-kern:29680a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Donald Douwsma <donaldd@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
Simplify vnode tracing calls by embedding function name & return addr in
the calling macro.
Also do a lot of vnode->inode renaming for consistency, while we're at it.
SGI-PV: 970335
SGI-Modid: xfs-linux-melb:xfs-kern:29650a
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
* 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6: (59 commits)
[XFS] eagerly remove vmap mappings to avoid upsetting Xen
[XFS] simplify validata_fields
[XFS] no longer using io_vnode, as was remaining from 23 cherrypick
[XFS] Remove STATIC which was missing from prior manual merge
[XFS] Put back the QUEUE_ORDERED_NONE test in the barrier check.
[XFS] Turn off XBF_ASYNC flag before re-reading superblock.
[XFS] avoid race in sync_inodes() that can fail to write out all dirty data
[XFS] This fix prevents bulkstat from spinning in an infinite loop.
[XFS] simplify xfs_create/mknod/symlink prototype
[XFS] avoid xfs_getattr in XFS_IOC_FSGETXATTR ioctl
[XFS] get_bulkall() could return incorrect inode state
[XFS] Kill unused IOMAP_EOF flag
[XFS] fix when DMAPI mount option processing happens
[XFS] ensure file size is logged on synchronous writes
[XFS] growlock should be a mutex
[XFS] replace some large xfs_log_priv.h macros by proper functions
[XFS] kill struct bhv_vfs
[XFS] move syncing related members from struct bhv_vfs to struct xfs_mount
[XFS] kill the vfs_flags member in struct bhv_vfs
[XFS] kill the vfs_fsid and vfs_altfsid members in struct bhv_vfs
...
Miklos Szeredi <miklos@szeredi.hu> and me identified a writeback bug:
> The following strange behavior can be observed:
>
> 1. large file is written
> 2. after 30 seconds, nr_dirty goes down by 1024
> 3. then for some time (< 30 sec) nothing happens (disk idle)
> 4. then nr_dirty again goes down by 1024
> 5. repeat from 3. until whole file is written
>
> So basically a 4Mbyte chunk of the file is written every 30 seconds.
> I'm quite sure this is not the intended behavior.
It can be produced by the following test scheme:
# cat bin/test-writeback.sh
grep nr_dirty /proc/vmstat
echo 1 > /proc/sys/fs/inode_debug
dd if=/dev/zero of=/var/x bs=1K count=204800&
while true; do grep nr_dirty /proc/vmstat; sleep 1; done
# bin/test-writeback.sh
nr_dirty 19207
nr_dirty 19207
nr_dirty 30924
204800+0 records in
204800+0 records out
209715200 bytes (210 MB) copied, 1.58363 seconds, 132 MB/s
nr_dirty 47150
nr_dirty 47141
nr_dirty 47142
nr_dirty 47142
nr_dirty 47142
nr_dirty 47142
nr_dirty 47205
nr_dirty 47214
nr_dirty 47214
nr_dirty 47214
nr_dirty 47214
nr_dirty 47214
nr_dirty 47215
nr_dirty 47216
nr_dirty 47216
nr_dirty 47216
nr_dirty 47154
nr_dirty 47143
nr_dirty 47143
nr_dirty 47143
nr_dirty 47143
nr_dirty 47143
nr_dirty 47142
nr_dirty 47142
nr_dirty 47142
nr_dirty 47142
nr_dirty 47134
nr_dirty 47134
nr_dirty 47135
nr_dirty 47135
nr_dirty 47135
nr_dirty 46097 <== -1038
nr_dirty 46098
nr_dirty 46098
nr_dirty 46098
[...]
nr_dirty 46091
nr_dirty 46092
nr_dirty 46092
nr_dirty 45069 <== -1023
nr_dirty 45056
nr_dirty 45056
nr_dirty 45056
[...]
nr_dirty 37822
nr_dirty 36799 <== -1023
[...]
nr_dirty 36781
nr_dirty 35758 <== -1023
[...]
nr_dirty 34708
nr_dirty 33672 <== -1024
[...]
nr_dirty 33692
nr_dirty 32669 <== -1023
% ls -li /var/x
847824 -rw-r--r-- 1 root root 200M 2007-08-12 04:12 /var/x
% dmesg|grep 847824 # generated by a debug printk
[ 529.263184] redirtied inode 847824 line 548
[ 564.250872] redirtied inode 847824 line 548
[ 594.272797] redirtied inode 847824 line 548
[ 629.231330] redirtied inode 847824 line 548
[ 659.224674] redirtied inode 847824 line 548
[ 689.219890] redirtied inode 847824 line 548
[ 724.226655] redirtied inode 847824 line 548
[ 759.198568] redirtied inode 847824 line 548
# line 548 in fs/fs-writeback.c:
543 if (wbc->pages_skipped != pages_skipped) {
544 /*
545 * writeback is not making progress due to locked
546 * buffers. Skip this inode for now.
547 */
548 redirty_tail(inode);
549 }
More debug efforts show that __block_write_full_page()
never has the chance to call submit_bh() for that big dirty file:
the buffer head is *clean*. So basicly no page io is issued by
__block_write_full_page(), hence pages_skipped goes up.
Also the comment in generic_sync_sb_inodes():
544 /*
545 * writeback is not making progress due to locked
546 * buffers. Skip this inode for now.
547 */
and the comment in __block_write_full_page():
1713 /*
1714 * The page was marked dirty, but the buffers were
1715 * clean. Someone wrote them back by hand with
1716 * ll_rw_block/submit_bh. A rare case.
1717 */
do not quite agree with each other. The page writeback should be skipped for
'locked buffer', but here it is 'clean buffer'!
This patch fixes this bug. Though I'm not sure why __block_write_full_page()
is called only to do nothing and who actually issued the writeback for us.
This is the two possible new behaviors after the patch:
1) pretty nice: wait 30s and write ALL:)
2) not so good:
- during the dd: ~16M
- after 30s: ~4M
- after 5s: ~4M
- after 5s: ~176M
The next patch will fix case (2).
Cc: David Chinner <dgc@sgi.com>
Cc: Ken Chen <kenchen@google.com>
Signed-off-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Because we cherrypicked SGI-Modid xfs-linux-melb:xfs-kern:29675a
and it depended on the sgi mod which removed io_vnode (which was
not cherrypicked in 23) it was hand modified.
This fixes things back up (to the originial mod) now we have moved
on again.
Reviewed-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
struct bhv_vnode is on it's way out, so move the trace buffer to the XFS
inode. Note that this makes the tracing macros rather misnamed, but this
kind of fallout will be fixed up incrementally later on.
SGI-PV: 969608
SGI-Modid: xfs-linux-melb:xfs-kern:29498a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
struct bhv_vnode is on it's way out, so move the I/O count to the XFS
inode.
SGI-PV: 969608
SGI-Modid: xfs-linux-melb:xfs-kern:29497a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
All flags previously handled at the vnode level are not in the xfs_inode
where we already have a flags mechanisms and free bits for flags
previously in the vnode.
SGI-PV: 969608
SGI-Modid: xfs-linux-melb:xfs-kern:29495a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
As bi_end_io is only called once when the reqeust is complete,
the 'size' argument is now redundant. Remove it.
Now there is no need for bio_endio to subtract the size completed
from bi_size. So don't do that either.
While we are at it, change bi_end_io to return void.
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
- in xfs_probe_cluster rename the inner len to pg_len. There's no harm
here because the outer len isn't used after the inner len comes into
existence but it keeps the code clean.
- in xfs_da_do_buf remove the inner i because they don't overlap
and they are both the same type.
SGI-PV: 968555
SGI-Modid: xfs-linux-melb:xfs-kern:29311a
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
If we have multiple unwritten extents within a single page, we fail to
tell the I/o completion construction handlers we need a new handle for the
second and subsequent blocks in the page. While we still issue the I/O
correctly, we do not have the correct ranges recorded in the ioend
structures and hence when we go to convert the unwritten extents we screw
it up.
Make sure we start a new ioend every time the mapping changes so that we
convert the correct ranges on I/O completion.
SGI-PV: 964647
SGI-Modid: xfs-linux-melb:xfs-kern:28797a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Tim Shimmin <tes@sgi.com>
Currently we do not wait on extent conversion to occur, and hence we can
return to userspace from a synchronous direct I/O write without having
completed all the actions in the write. Hence a read after the write may
see zeroes (unwritten extent) rather than the data that was written.
Block the I/O completion by triggering a synchronous workqueue flush to
ensure that the conversion has occurred before we return to userspace.
SGI-PV: 964092
SGI-Modid: xfs-linux-melb:xfs-kern:28775a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
The recent fix for preventing NULL files from being left around does not
update the file size corectly in all cases. The missing case is a write
extending the file that does not need to allocate a block.
In that case we used a read mapping of the extent which forced the use of
the read I/O completion handler instead of the write I/O completion
handle. Hence the file size was not updated on I/O completion.
SGI-PV: 965068
SGI-Modid: xfs-linux-melb:xfs-kern:28657a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Nathan Scott <nscott@aconex.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
The problem that has been addressed is that of synchronising updates of
the file size with writes that extend a file. Without the fix the update
of a file's size, as a result of a write beyond eof, is independent of
when the cached data is flushed to disk. Often the file size update would
be written to the filesystem log before the data is flushed to disk. When
a system crashes between these two events and the filesystem log is
replayed on mount the file's size will be set but since the contents never
made it to disk the file is full of holes. If some of the cached data was
flushed to disk then it may just be a section of the file at the end that
has holes.
There are existing fixes to help alleviate this problem, particularly in
the case where a file has been truncated, that force cached data to be
flushed to disk when the file is closed. If the system crashes while the
file(s) are still open then this flushing will never occur.
The fix that we have implemented is to introduce a second file size,
called the in-memory file size, that represents the current file size as
viewed by the user. The existing file size, called the on-disk file size,
is the one that get's written to the filesystem log and we only update it
when it is safe to do so. When we write to a file beyond eof we only
update the in- memory file size in the write operation. Later when the I/O
operation, that flushes the cached data to disk completes, an I/O
completion routine will update the on-disk file size. The on-disk file
size will be updated to the maximum offset of the I/O or to the value of
the in-memory file size if the I/O includes eof.
SGI-PV: 958522
SGI-Modid: xfs-linux-melb:xfs-kern:28322a
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
Don't hide buffer_unwritten behind buffer_delay() and remove the hack that
clears unexpected buffer_unwritten() states now that it can't happen.
Signed-off-by: Dave Chinner <dgc@sgi.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Cc: Timothy Shimmin <tes@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When writing less than a filesystem block of data into an unwritten extent
via buffered I/O, __xfs_get_blocks fails to set the buffer new flag. As a
result, the generic code will not zero either edge of the block resulting
in garbage being written to disk either side of the real data. Set the
buffer new state on bufferd writes to unwritten extents to ensure that
zeroing occurs.
SGI-PV: 960328
SGI-Modid: xfs-linux-melb:xfs-kern:28000a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
gcc-4.1 and more recent aggressively inline static functions which
increases XFS stack usage by ~15% in critical paths. Prevent this from
occurring by adding noinline to the STATIC definition.
Also uninline some functions that are too large to be inlined and were
causing problems with CONFIG_FORCED_INLINING=y.
Finally, clean up all the different users of inline, __inline and
__inline__ and put them under one STATIC_INLINE macro. For debug kernels
the STATIC_INLINE macro uninlines those functions.
SGI-PV: 957159
SGI-Modid: xfs-linux-melb:xfs-kern:27585a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: David Chatterton <chatz@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
XFS appears to call clear_page_dirty to get the mapping tree dirty tag
set correctly at the same time the page dirty flag is cleared. I note
that this can be done by set_page_writeback() if we clear the dirty flag
on the page first when we are writing back the entire page.
Hence it seems to me that the XFS call to clear_page_dirty() could
easily be substituted by clear_page_dirty_for_io() followed by a call to
set_page_writeback() to get the mapping tree tags set correctly after
the page has been marked clean.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The only time it is safe to call aio_complete() is when the ->ki_retry
function returns -EIOCBQUEUED to the AIO core. direct_io_worker() has
historically done this by relying on its caller to translate positive return
codes into -EIOCBQUEUED for the aio case. It did this by trying to keep
conditionals in sync. direct_io_worker() knew when finished_one_bio() was
going to call aio_complete(). It would reverse the test and wait and free the
dio in the cases it thought that finished_one_bio() wasn't going to.
Not surprisingly, it ended up getting it wrong. 'ret' could be a negative
errno from the submission path but it failed to communicate this to
finished_one_bio(). direct_io_worker() would return < 0, it's callers
wouldn't raise -EIOCBQUEUED, and aio_complete() would be called. In the
future finished_one_bio()'s tests wouldn't reflect this and aio_complete()
would be called for a second time which can manifest as an oops.
The previous cleanups have whittled the sync and async completion paths down
to the point where we can collapse them and clearly reassert the invariant
that we must only call aio_complete() after returning -EIOCBQUEUED.
direct_io_worker() will only return -EIOCBQUEUED when it is not the last to
drop the dio refcount and the aio bio completion path will only call
aio_complete() when it is the last to drop the dio refcount.
direct_io_worker() can ensure that it is the last to drop the reference count
by waiting for bios to drain. It does this for sync ops, of course, and for
partial dio writes that must fall back to buffered and for aio ops that saw
errors during submission.
This means that operations that end up waiting, even if they were issued as
aio ops, will not call aio_complete() from dio. Instead we return the return
code of the operation and let the aio core call aio_complete(). This is
purposely done to fix a bug where AIO DIO file extensions would call
aio_complete() before their callers have a chance to update i_size.
Now that direct_io_worker() is explicitly returning -EIOCBQUEUED its callers
no longer have to translate for it. XFS needs to be careful not to free
resources that will be used during AIO completion if -EIOCBQUEUED is returned.
We maintain the previous behaviour of trying to write fs metadata for O_SYNC
aio+dio writes.
Signed-off-by: Zach Brown <zach.brown@oracle.com>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Cc: Suparna Bhattacharya <suparna@in.ibm.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Cc: <xfs-masters@oss.sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
handling.
SGI-PV: 955302
SGI-Modid: xfs-linux-melb:xfs-kern:26804a
Signed-off-by: Nathan Scott <nathans@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
__blockdev_direct_IO for the DIO_OWN_LOCKING case for direct I/O reads
since it drops and reacquires the i_mutex while holding the iolock and
this violates the locking order.
SGI-PV: 955696
SGI-Modid: xfs-linux-melb:xfs-kern:26898a
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
Signed-off-by: David Chatterton <chatz@sgi.com>
Same as with already do with the file operations: keep them in .rodata and
prevents people from doing runtime patching.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Steven French <sfrench@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
truncate down followed by delayed allocation (buffered writes) - worst
case scenario for the notorious NULL files problem. This reduces the
window where we are exposed to that problem significantly.
SGI-PV: 917976
SGI-Modid: xfs-linux-melb:xfs-kern:26100a
Signed-off-by: Nathan Scott <nathans@sgi.com>
