On Mon, Feb 5, 2024 at 8:56 AM Zorro Lang <zlang@xxxxxxxxxx> wrote:
>
> On Mon, Feb 05, 2024 at 02:16:26PM +0800, Zorro Lang wrote:
> > On Mon, Feb 05, 2024 at 02:00:16PM +0800, Zorro Lang wrote:
> > > On Mon, Feb 05, 2024 at 01:45:02PM +0800, Zorro Lang wrote:
> > > > On Sun, Feb 04, 2024 at 10:48:47PM +0800, Zorro Lang wrote:
> > > > > Hi all,
> > > > >
> > > > > The for-next branch of the xfstests repository at:
> > > > >
> > > > > git://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git
> > > > >
> > > > > has just been updated and tagged as v2024.02.04 release.
> > > > >
> > > > > Release Notes:
> > > > > 1) There's 4 new test cases in this release -- btrfs/310, generic/740, xfs/601,
> > > > > and xfs/606.
> > > > > 2) btrfs/303 is removed.
> > > > > 3) Supports XFS metadump V1 and V2 testing, refer to common/xfs_metadump_tests
> > > > > new file (might be renamed to common/metadump if more other fs need).
> > > > > 4) Some fixes and updates for UBIFS, BCACHEFS and NFS testing, welcome more
> > > > > updates for more filesystems.
> > > > > 5) Some updates/fixes for source building, if anyone find building issues in
> > > > > this release, please report to us.
> > > > >
> > > > > More details refer to below, thanks for all your contributions!
> > > > >
> > > > > Thanks,
> > > > > Zorro
> > > > >
> > > > > The new head of the for-next branch is commit:
> > > > >
> > > > > 9b63adb69add t_snapshot_deleted_subvolume: add check for BTRFS_IOC_SNAP_DESTROY_V2
> > > > >
> > > > > New commits ([N]ew, [U]pdate, [D]elete):
> > > > >
> > > > > Alexander Aring (1):
> > > > > [N] [56625c897fd9] generic: add fcntl lock corner cases tests
> > > > >
> > > > > Anthony Iliopoulos (2):
> > > > > [U] [42f3cb7e56e3] src/af_unix: drop unused offsetof macro definition
> > > > > [U] [79d98d94e58c] build: remove unused m4 macros
> > > > >
> > > > > Boris Burkov (1):
> > > > > [D] [a6e196af46cc] btrfs: Remove btrfs/303
> > > > >
> > > > > Darrick J. Wong (10):
> > > > > [U] [e7eadc3ee835] generic/256: constrain runtime with TIME_FACTOR
> > > > > [U] [8c8be04c1065] common/xfs: simplify maximum metadump format detection
> > > > > [U] [42f797643529] common/populate: always metadump full metadata blocks
> > > > > [U] [56ccc8cc69d1] xfs/336: fix omitted -a and -o in metadump call
> > > > > [N] [85c1e0f518ff] common: refactor metadump v1 and v2 tests
> > > > > [U] [1bdc3a5f1c9a] xfs/{129,234,253,605}: disable metadump v1 testing with external devices
> > > > > [U] [329671426287] xfs/503: test metadump obfuscation, not progressbars
> > > > > [N] [c31a500edc04] xfs/503: split copy and metadump into two tests
> > > > > [U] [7e2bf5955c2d] common/xfs: only pass -l in _xfs_mdrestore for v2 metadumps
> > > > > [U] [6f4223b6865f] xfs/122: fix for xfs_attr_shortform removal in 6.8
> > > >
> > > > Note: A regression problem on this release sometimes after running xfs/605 or
> > > > xfs/129, it only affect XFS testing, it might kill all later testing, I'm
> > > > looking into it, and fix ASAP, please linux-xfs don't not update to this
> > > > release temporarily, wait for a fix. Sorry for this trouble.
> > >
> > > !!! It's a critical dangerous bug! Please don't use this release on xfs testing, it might
> > > kill your system!
> > >
> > > The problem comes from common/xfs_metadump_tests:_cleanup_verify_metadump():
> > >
> > > rm -f "$XFS_METADUMP_FILE" "$XFS_METADUMP_IMG"*
> > >
> > > The XFS_METADUMP_FILE and XFS_METADUMP_IMG are set in _setup_verify_metadump(), but
> > > if the _setup_verify_metadump isn't called due to someone case _notrun before calling
> > > it, its _cleanup will still do _cleanup_verify_metadump, likes this:
> > >
> > > _cleanup()
> > > {
> > > cd /
> > > rm -r -f $tmp.*
> > > _cleanup_verify_metadump
> > >
> > > Then boooomb, it does "rm -f *" in "/" .... The "*" after "rm" must be written
> > > carefully. I'll send a patch to fix this problem soon, please don't update
> > > fstests until the fix be merged.
> >
> > I decide to make a respin, I think this bug is dangerous enough due to it might kill
> > the whole testing system. If someone use fstests without the fix, that's too dangerous.
> >
> > I think most of people haven't shifted to the new release, and this version hasn't
> > been merged into master branch. So I decide to do a respin with a manually fix to
> > avoid this problem ASAP. Then we can do more jobs later. Sorry for this trouble,
> > hope it hasn't affected you.
>
> I've reset the for-next branch to last v2024.01.14 version. Please don't worry, the
> patches already got RVBs are in my local branch, I'll make a new release in this
> week soon, after some discussion and testing.
>
Would you consider including my ovl test fix in this release:
https://lore.kernel.org/fstests/20240203081228.1725872-1-amir73il@xxxxxxxxx/
Without this fix, overlay/084 fails on master and on v.6.7.3.
Thanks,
Amir.
