Re: [PATCH 1/2] xfs: test what happens when we reset the root dir and it has xattrs

[Brian Foster <bfoster@redhat.com>]

On Tue, Apr 27, 2021 at 09:08:43PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <djwong@kernel.org>
> 
> Make sure that we can reset the root directory and the xattrs are erased
> properly.
> 
> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> ---
>  tests/xfs/757     |   63 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/xfs/757.out |    7 ++++++
>  tests/xfs/group   |    1 +
>  3 files changed, 71 insertions(+)
>  create mode 100755 tests/xfs/757
>  create mode 100644 tests/xfs/757.out
> 
> 
> diff --git a/tests/xfs/757 b/tests/xfs/757
> new file mode 100755
> index 00000000..0b9914f6
> --- /dev/null
> +++ b/tests/xfs/757
> @@ -0,0 +1,63 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (c) 2021 Oracle.  All Rights Reserved.
> +#
> +# FS QA Test No. 757
> +#
> +# Make sure that attrs are handled properly when repair has to reset the root
> +# directory.
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 7 15
> +
> +_cleanup()
> +{
> +	cd /
> +	rm -rf $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +. ./common/populate
> +. ./common/fuzzy
> +
> +# real QA test starts here
> +_supported_fs xfs
> +_require_scratch_nocheck
> +
> +rm -f $seqres.full
> +
> +echo "Format and populate btree attr root dir"
> +_scratch_mkfs > "$seqres.full" 2>&1
> +_scratch_mount
> +
> +blksz="$(stat -f -c '%s' "${SCRATCH_MNT}")"
> +__populate_create_attr "${SCRATCH_MNT}" "$((64 * blksz / 40))" true
> +_scratch_unmount
> +
> +echo "Break the root directory"
> +_scratch_xfs_fuzz_metadata_field core.mode zeroes 'sb 0' 'addr rootino' >> $seqres.full 2>&1
> +
> +echo "Detect bad root directory"
> +_scratch_xfs_repair -n >> $seqres.full 2>&1 && \
> +	echo "Should have detected bad root dir"
> +
> +echo "Fix bad root directory"
> +_scratch_xfs_repair >> $seqres.full 2>&1
> +
> +echo "Detect fixed root directory"
> +_scratch_xfs_repair -n >> $seqres.full 2>&1
> +
> +echo "Mount test"
> +_scratch_mount
> +

Is the regression test here that attrs are erased after this sequence
(as suggested in the commit log), or that the fs mounts, or both? I'm
basically just wondering if we should also dump the xattrs on the root
dir as a last step (and expect NULL output)..? That aside:

Reviewed-by: Brian Foster <bfoster@redhat.com>

> +# success, all done
> +status=0
> +exit
> diff --git a/tests/xfs/757.out b/tests/xfs/757.out
> new file mode 100644
> index 00000000..9f3aed5a
> --- /dev/null
> +++ b/tests/xfs/757.out
> @@ -0,0 +1,7 @@
> +QA output created by 757
> +Format and populate btree attr root dir
> +Break the root directory
> +Detect bad root directory
> +Fix bad root directory
> +Detect fixed root directory
> +Mount test
> diff --git a/tests/xfs/group b/tests/xfs/group
> index 731f869c..76e31167 100644
> --- a/tests/xfs/group
> +++ b/tests/xfs/group
> @@ -528,5 +528,6 @@
>  537 auto quick
>  538 auto stress
>  539 auto quick mount
> +757 auto quick attr repair
>  908 auto quick bigtime
>  909 auto quick bigtime quota
>