Re: [PATCH] xfs/126: fix that corrupt xattr might fail with a small probability

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Wed, Jan 08, 2020 at 05:27:58PM +0800, yu kuai wrote:
> The cmd used in xfs_db to corrupt xattr is "blocktrash -x 32
> -y $((blksz * 8)) -n8 -3", which means select random 8 bit from 32 to
> end of the block, and the changed bits are randomized. However,
> there is a small chance that corrupting xattr failed because irrelevant
> bits are chossen or the chooosen bits are not changed, which lead to
> output missmatch:
> QA output created by 126                    QA output created by 126
> + create scratch fs                         + create scratch fs
> + mount fs image                            + mount fs image
> + make some files                           + make some files
> + check fs                                  + check fs
> + check xattr                               + check xattr
> + corrupt xattr                             + corrupt xattr
> + mount image && modify xattr               + mount image && modify xattr
> + repair fs                                 + repair fs
> + mount image (2)                           + mount image (2)
> + chattr -R -i                              + chattr -R -i
> + modify xattr (2)                          + modify xattr (2)
>                                             > # file: tmp/scratch/attrfile
>                                             > user.x00000000="0000000000000000"
>                                             >
> + check fs (2)                              + check fs (2)
> 
> Fix the problem by adding a seed for random processing to select same
> bits each time, and inverting the selected bits.
> 
> Signed-off-by: yu kuai <yukuai3@huawei.com>
> ---
>  tests/xfs/126 | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/xfs/126 b/tests/xfs/126
> index 4f9f8cf9..9b57e58b 100755
> --- a/tests/xfs/126
> +++ b/tests/xfs/126
> @@ -37,7 +37,7 @@ test -n "${FORCE_FUZZ}" || _require_scratch_xfs_crc
>  _require_attrs
>  _require_populate_commands
>  _require_xfs_db_blocktrash_z_command
> -test -z "${FUZZ_ARGS}" && FUZZ_ARGS="-n 8 -3"
> +test -z "${FUZZ_ARGS}" && FUZZ_ARGS="-n 8 -2"

TBH I've wondered if blocktrash -3 and fuzz-random should snapshot the
buffer before randomizing it and try again if the contents don't change?
I suspect most of our fuzz tests expect "randomize the ____" to return
with ____ full of random junk, not the exact same contents as before.

--D

>  rm -f $seqres.full
>  
> @@ -72,7 +72,7 @@ echo "+ corrupt xattr"
>  loff=1
>  while true; do
>  	_scratch_xfs_db -x -c "inode ${inode}" -c "ablock ${loff}" -c "stack" | grep -q 'file attr block is unmapped' && break
> -	_scratch_xfs_db -x -c "inode ${inode}" -c "ablock ${loff}" -c "stack" -c "blocktrash -x 32 -y $((blksz * 8)) -z ${FUZZ_ARGS}" >> $seqres.full
> +	_scratch_xfs_db -x -c "inode ${inode}" -c "ablock ${loff}" -c "stack" -c "blocktrash -s 1024 -x 32 -y $((blksz * 8)) -z ${FUZZ_ARGS}" >> $seqres.full
>  	loff="$((loff + 1))"
>  done
>  
> -- 
> 2.17.2
>