xfs: rework xfs_buf_bio_endio error handling
authorDave Chinner <dchinner@redhat.com>
Wed, 1 Oct 2014 23:04:31 +0000 (09:04 +1000)
committerDave Chinner <david@fromorbit.com>
Wed, 1 Oct 2014 23:04:31 +0000 (09:04 +1000)
commit61be9c529a4a715ab8679e9ca82bc3790c7ab66c
tree3e8d1c757133eb7c844cd4ba8815f4a6bc0218cb
parente8aaba9a783c8e5d2c58ebe69650ea31b91bb745
xfs: rework xfs_buf_bio_endio error handling

Currently the report of a bio error from completion
immediately marks the buffer with an error. The issue is that this
is racy w.r.t. synchronous IO - the submitter can see b_error being
set before the IO is complete, and hence we cannot differentiate
between submission failures and completion failures.

Add an internal b_io_error field protected by the b_lock to catch IO
completion errors, and only propagate that to the buffer during
final IO completion handling. Hence we can tell in xfs_buf_iorequest
if we've had a submission failure bey checking bp->b_error before
dropping our b_io_remaining reference - that reference will prevent
b_io_error values from being propagated to b_error in the event that
completion races with submission.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
fs/xfs/xfs_buf.c
fs/xfs/xfs_buf.h