Re: [ANNOUNCE] xfs-linux: for-next updated to 6a94b1acda7e

["Darrick J. Wong" <djwong@kernel.org>]

On Wed, Apr 24, 2024 at 08:45:23AM +1000, Dave Chinner wrote:
> On Tue, Apr 23, 2024 at 03:46:24PM +0530, Chandan Babu R wrote:
> > Hi folks,
> > 
> > The for-next branch of the xfs-linux repository at:
> > 
> > 	https://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git
> > 
> > has just been updated.
> > 
> > Patches often get missed, so please check if your outstanding patches
> > were in this update. If they have not been in this update, please
> > resubmit them to linux-xfs@vger.kernel.org so they can be picked up in
> > the next update.
> > 
> > The new head of the for-next branch is commit:
> > 
> > 6a94b1acda7e xfs: reinstate delalloc for RT inodes (if sb_rextsize == 1)
> 
> I've just noticed a regression in for-next - it was there prior to
> this update, but I hadn't run a 1kB block size fstests run in a
> while so I've only just noticed it. It is 100% reproducable, and may
> well be a problem with the partial filter matches in the test rather
> than a kernel bug...
> 
> SECTION       -- xfs_1k
> FSTYP         -- xfs (debug)
> PLATFORM      -- Linux/x86_64 test1 6.9.0-rc5-dgc+ #219 SMP PREEMPT_DYNAMIC Wed Apr 24 08:30:50 AEST 2024
> MKFS_OPTIONS  -- -f -m rmapbt=1 -b size=1k /dev/vdb
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/vdb /mnt/scratch
> 
> xfs/348 19s ... - output mismatch (see /home/dave/src/xfstests-dev/results//xfs_1k/xfs/348.out.bad)
>     --- tests/xfs/348.out       2022-12-21 15:53:25.579041081 +1100
>     +++ /home/dave/src/xfstests-dev/results//xfs_1k/xfs/348.out.bad     2024-04-24 08:34:43.718525603 +1000
>     @@ -2,7 +2,7 @@
>      ===== Find inode by file type:
>      dt=1 => FIFO_INO
>      dt=2 => CHRDEV_INO
>     -dt=4 => DIR_INO
>     +dt=4 => PARENT_INO108928

Given:

pino=$(ls -id $testdir | awk '{print $1}')
echo "s/inode $pino/PARENT_INO/" >> $inode_filter

I'd say that pino=35 would then result in:

s/inode 35/PARENT_INO/

Which will then match and transform:

inode 35108928

into:

PARENT_INO108928

So yeah, I agree that the partial filter matches are a problem.

--D

>      dt=6 => BLKDEV_INO
>      dt=10 => DATA_INO
>     ...
>     (Run 'diff -u /home/dave/src/xfstests-dev/tests/xfs/348.out /home/dave/src/xfstests-dev/results//xfs_1k/xfs/348.out.bad'  to see the entire diff)
> Failures: xfs/348
> Failed 1 of 1 tests
> 
> xfsprogs version installed on this test VM is:
> 
> $ xfs_repair -V
> xfs_repair version 6.4.0
> $
> 
> -Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
>