[PATCH v4.2 00/32] xfstests: test the nfs/cifs/btrfs/xfs reflink/dedupe ioctls

[Darrick J. Wong]

On Sat, Feb 13, 2016 at 12:26:59PM +1100, Dave Chinner wrote:
> On Thu, Feb 11, 2016 at 03:39:16PM -0800, Darrick J. Wong wrote:
> > Dave Chinner: I've renumbered the new tests and pushed to github[3] if
> > you'd like to pull.  See the pull request at the end of this message.
> > 
> > This is a patch set against the reflink/dedupe test cases in xfstests.
> > The first three patches fix errors in the existing reflink tests, some
> > of which are from Christoph Hellwig.
> > 
> > The next eight patches fix style errors, refactor commonly used code,
> > remove unnecessary clutter, and add missing _require* checks to the
> > existing reflink tests.
> > 
> > Patch 12 adds a test checking that unwritten extent conversion does
> > NOT happen after a directio write to an unwritten extent hits a disk
> > error.   Due to a bug in the VFS directio code, ext4 can disclose
> > stale disk contents if an aio dio write fails; XFS suffers this
> > problem for any failing dio write to an unwritten extent.  Christoph's
> > kernel patchset titled "vfs/xfs: directio updates to ease COW handling
> > V2" (and a separate ext4 warning cleanup) is needed to fix this.
> > 
> > Patches 13-31 add more reflink tests focusing on correct CoW behavior
> > particularly with the CoW extent size hint enabled.  It also provides
> > a few regression tests for bugs that have been hit while running XFS
> > reflink, a few tests of the quota accounting when various reflink
> > operations happen, and a few tests for get_bmapx to ensure that what
> > it reports is at least somewhat accurate.
> > 
> > Patch 25 adds a few basic reverse-mapping tests for XFS.
> > 
> > If you're going to start using this mess, you probably ought to just
> > pull from my github trees for kernel[1], xfsprogs[2], xfstests[3],
> > xfs-docs[4], and man-pages[5].  All tests should pass on XFS, YMWV on
> > btrfs and ocfs2.
> > 
> > Comments and questions are, as always, welcome.
> 
> I haven't worked out which patch causes this, but:
> 
> xfs/246  - output mismatch (see /home/dave/src/xfstests-dev/results//xfs_1k/xfs/246.out.bad)
>     --- tests/xfs/246.out       2016-02-13 09:46:01.419169115 +1100
>     +++ /home/dave/src/xfstests-dev/results//xfs_1k/xfs/246.out.bad     2016-02-13 11:36:05.205851863 +1100
>     @@ -3,4 +3,5 @@
>      Create the original files
>      Dump extents after sync
>      Hole CoW extents:
>     -SCRATCH_MNT/test-246/file1: no extents
>     +bmap: invalid option -- 'c'
>     +bmap [-adlpv] [-n nx] -- print block mapping for an XFS file
>     ...
>     (Run 'diff -u tests/xfs/246.out /home/dave/src/xfstests-dev/results//xfs_1k/xfs/246.out.bad'  to see the entire diff)
> 
> This test neds a check that the "-c" option in the bmap command
> exists. Not sure if you can use '_requires_xfs_io_command "bmap -c"'
> here or whether this requires more help...
> 
> Followup patch, anyway, because I'm about to push out everything
> that was in your second pull req (i.e. with the aiocp fix).

xfs_io "bmap -c" is a new switch that dumps the CoW fork of a file.
AFAICT there isn't any way to check that an xfs_io command supports
a particular switch, though it wouldn't be difficult to add a check.

--D

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david at fromorbit.com