On Sun, Jun 27, 2021 at 11:33:54PM +0800, Eryu Guan wrote: > Hi all, > > The master branch of the xfstests repository at: > > git://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git > > have just been updated. The only thing in this update is the "move test group > lists into test files" patchset from Darrick J. Wong, which moves each test's > group information into the test itself, and automatically generate the group > files as part of the make process. Please refer to the description quated from > Darrick's PR cover letter below [1]. > > This may break some old habits while using fstests, some heads up below .... > And this update probably may introduce regressions and bugs, please help report > and/or fix bugs if you hit any. Yup, generic/019 is the first regression I've noticed: generic/019 46s ... - output mismatch (see /home/dave/src/xfstests-dev/results//xfs/generic/019.out.bad) --- tests/generic/019.out 2014-04-08 09:43:28.506999723 +1000 +++ /home/dave/src/xfstests-dev/results//xfs/generic/019.out.bad 2021-06-28 10:39:30.653077176 +1000 @@ -7,3 +7,5 @@ Force SCRATCH_DEV device failure Make SCRATCH_DEV device operable again Disallow global fail_make_request feature +/home/dave/src/xfstests-dev/tests/generic/019: line 56: poweron_scratch_dev: command not found +Disallow global fail_make_request feature ... (Run 'diff -u /home/dave/src/xfstests-dev/tests/generic/019.out /home/dave/src/xfstests-dev/results//xfs/generic/019.out.bad' to see the entire diff) I suspect that _cleanup output on EXIT is now being captured to the $seq.out file, because I can't see how this _cleanup function ever worked - poweron_scratch_dev() doesn't exist and AFAICT it never has.... Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx
