diff options
author | Ryan Ding <ryan.ding@oracle.com> | 2016-03-25 14:21:15 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2016-03-25 16:37:42 -0700 |
commit | e63890f38ade9497b5609ddeb7f52df0fe55ea15 (patch) | |
tree | 6374b96dfa45477fc53cb6033cddc5621d5936c7 /fs/ocfs2/file.c | |
parent | f1f973ffce96a47c2b3f142e91eccef5bf22f699 (diff) | |
download | linux-stable-e63890f38ade9497b5609ddeb7f52df0fe55ea15.tar.gz linux-stable-e63890f38ade9497b5609ddeb7f52df0fe55ea15.tar.bz2 linux-stable-e63890f38ade9497b5609ddeb7f52df0fe55ea15.zip |
ocfs2: fix ip_unaligned_aio deadlock with dio work queue
In the current implementation of unaligned aio+dio, lock order behave as
follow:
in user process context:
-> call io_submit()
-> get i_mutex
<== window1
-> get ip_unaligned_aio
-> submit direct io to block device
-> release i_mutex
-> io_submit() return
in dio work queue context(the work queue is created in __blockdev_direct_IO):
-> release ip_unaligned_aio
<== window2
-> get i_mutex
-> clear unwritten flag & change i_size
-> release i_mutex
There is a limitation to the thread number of dio work queue. 256 at
default. If all 256 thread are in the above 'window2' stage, and there
is a user process in the 'window1' stage, the system will became
deadlock. Since the user process hold i_mutex to wait ip_unaligned_aio
lock, while there is a direct bio hold ip_unaligned_aio mutex who is
waiting for a dio work queue thread to be schedule. But all the dio
work queue thread is waiting for i_mutex lock in 'window2'.
This case only happened in a test which send a large number(more than
256) of aio at one io_submit() call.
My design is to remove ip_unaligned_aio lock. Change it to a sync io
instead. Just like ip_unaligned_aio lock, serialize the unaligned aio
dio.
[akpm@linux-foundation.org: remove OCFS2_IOCB_UNALIGNED_IO, per Junxiao Bi]
Signed-off-by: Ryan Ding <ryan.ding@oracle.com>
Reviewed-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'fs/ocfs2/file.c')
-rw-r--r-- | fs/ocfs2/file.c | 27 |
1 files changed, 9 insertions, 18 deletions
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 1ab182321b18..c18ab45f8d21 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -2178,7 +2178,7 @@ static ssize_t ocfs2_file_write_iter(struct kiocb *iocb, struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); int full_coherency = !(osb->s_mount_opt & OCFS2_MOUNT_COHERENCY_BUFFERED); - int unaligned_dio = 0; + void *saved_ki_complete = NULL; int append_write = ((iocb->ki_pos + count) >= i_size_read(inode) ? 1 : 0); @@ -2241,17 +2241,12 @@ static ssize_t ocfs2_file_write_iter(struct kiocb *iocb, goto out; } - if (direct_io && !is_sync_kiocb(iocb)) - unaligned_dio = ocfs2_is_io_unaligned(inode, count, iocb->ki_pos); - - if (unaligned_dio) { + if (direct_io && !is_sync_kiocb(iocb) && + ocfs2_is_io_unaligned(inode, count, iocb->ki_pos)) { /* - * Wait on previous unaligned aio to complete before - * proceeding. + * Make it a sync io if it's an unaligned aio. */ - mutex_lock(&OCFS2_I(inode)->ip_unaligned_aio); - /* Mark the iocb as needing an unlock in ocfs2_dio_end_io */ - ocfs2_iocb_set_unaligned_aio(iocb); + saved_ki_complete = xchg(&iocb->ki_complete, NULL); } /* communicate with ocfs2_dio_end_io */ @@ -2272,11 +2267,10 @@ static ssize_t ocfs2_file_write_iter(struct kiocb *iocb, */ if ((written == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) { rw_level = -1; - unaligned_dio = 0; } if (unlikely(written <= 0)) - goto no_sync; + goto out; if (((file->f_flags & O_DSYNC) && !direct_io) || IS_SYNC(inode)) { @@ -2298,13 +2292,10 @@ static ssize_t ocfs2_file_write_iter(struct kiocb *iocb, iocb->ki_pos - 1); } -no_sync: - if (unaligned_dio && ocfs2_iocb_is_unaligned_aio(iocb)) { - ocfs2_iocb_clear_unaligned_aio(iocb); - mutex_unlock(&OCFS2_I(inode)->ip_unaligned_aio); - } - out: + if (saved_ki_complete) + xchg(&iocb->ki_complete, saved_ki_complete); + if (rw_level != -1) ocfs2_rw_unlock(inode, rw_level); |