[PATCH v3 0/2] xfs: quotacheck vs. dquot reclaim deadlock

[Brian Foster <bfoster@redhat.com>]

Hi all,

This incorporates Dave's feedback from review of v2 with one exception:

- After taking another look at xfs_qm_quotacheck(), it appears to
  intentionally jump to the normal exit path of the function in various
  error situations. I therefore opted to not to create a new error label
  for post delwri submit errors, since we still do the cancellation in
  the case of success as well. If desired, we can append a broader
  refactoring patch for that function, separate from the patches that
  fix bugs.

Brian

v3:
- Dropped the rfc/experiment patch.
- Created a _delwri_cancel() helper function.
- Refactored _delwri_pushbuf() to expect an unlocked buffer.
- Updated _delwri_pushbuf() with more detailed comments around delwri
  queue logic and reference counting.
v2: http://www.spinics.net/lists/linux-xfs/msg04483.html
- Added quotacheck error handling fixup patch.
- Push buffers with flush locked dquots for deadlock avoidance rather
  than bypass dquot reclaim.
- Added RFC patch for quotacheck early buffer list submission.
v1: http://www.spinics.net/lists/linux-xfs/msg04304.html

Brian Foster (2):
  xfs: fix up quotacheck buffer list error handling
  xfs: push buffer of flush locked dquot to avoid quotacheck deadlock

 fs/xfs/xfs_buf.c   | 84 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_buf.h   |  2 ++
 fs/xfs/xfs_qm.c    | 35 ++++++++++++++++++-----
 fs/xfs/xfs_trace.h |  1 +
 4 files changed, 115 insertions(+), 7 deletions(-)

-- 
2.7.4