[0/7] Abort cleanup fixes

Message

Josef Bacik Nov. 21, 2018, 7:05 p.m. UTC

A new xfstests that really hammers on transaction aborts (generic/495 I think?)
uncovered a lot of random issues.  Some of these were introduced with the new
delayed refs rsv patches, others were just exposed by them, such as the pending
bg stuff.  With these patches in place I stopped getting all the random
leftovers and WARN_ON()'s when running whichever xfstest that was and things are
much smoother now.  Thanks,

Josef

Comments

David Sterba Jan. 14, 2019, 11:55 a.m. UTC | #1

On Wed, Nov 21, 2018 at 02:05:38PM -0500, Josef Bacik wrote:
> A new xfstests that really hammers on transaction aborts (generic/495 I think?)

The test number is 475 and it would be really useful to have a sample of
the stacktraces next to the patches that fix the problems. I was seeing
random failures in 475 but took me a while to find this patchset that
fixed them.

I'm queuing the fixes without cleanups to 5.0-rc.

David Sterba Jan. 28, 2019, 10:06 p.m. UTC | #2

On Wed, Nov 21, 2018 at 02:05:38PM -0500, Josef Bacik wrote:
> A new xfstests that really hammers on transaction aborts (generic/495 I think?)
> uncovered a lot of random issues.  Some of these were introduced with the new
> delayed refs rsv patches, others were just exposed by them, such as the pending
> bg stuff.  With these patches in place I stopped getting all the random
> leftovers and WARN_ON()'s when running whichever xfstest that was and things are
> much smoother now.  Thanks,

FYI

in 5.0-rc:

- btrfs: wait on ordered extents on abort cleanup
- btrfs: handle delayed ref head accounting cleanup in abort

in misc-next:

- btrfs: make btrfs_destroy_delayed_refs use btrfs_delete_ref_head
- btrfs: make btrfs_destroy_delayed_refs use btrfs_delayed_ref_lock
- btrfs: call btrfs_create_pending_block_groups unconditionally

reworked as https://patchwork.kernel.org/patch/10784039/ :

- btrfs: cleanup pending bgs on transaction abort
- btrfs: just delete pending bgs if we are aborted