On 4/12/17 11:14 AM, Darrick J. Wong wrote:
> On Wed, Apr 12, 2017 at 11:39:37PM +0800, Zorro Lang wrote:
>> There was a bug during log replay, the attr/attr3 leaf verifier
>> reported corruption when encountering a leaf attribute with a
>> count of 0 in the header, as below:
>>
>> Metadata corruption detected at xfs_attr3_leaf block 0x480988/0x1000
>>
>> commit f714016 from xfsprogs has fixed this bug. This test case
>> will emulate this corruption by xfs_db and use xfs_repair to fix
>> it.
>>
>> Signed-off-by: Zorro Lang <zlang@xxxxxxxxxx>
>> ---
>>
>> Hi,
>>
>> V2 did below changes:
>> 1) remove the requirement for xfs_db's write and addr commands
>> 2) check if xfs_db support write -d option
>> 3) turn to get filesystem block size from _filter_mkfs output
>> 4) add more comments
>>
>> Thanks,
>> Zorro
>>
>> tests/xfs/288 | 110 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
>> tests/xfs/288.out | 2 +
>> tests/xfs/group | 1 +
>> 3 files changed, 113 insertions(+)
>> create mode 100755 tests/xfs/288
>> create mode 100644 tests/xfs/288.out
>>
>> diff --git a/tests/xfs/288 b/tests/xfs/288
>> new file mode 100755
>> index 0000000..81e7b50
>> --- /dev/null
>> +++ b/tests/xfs/288
>> @@ -0,0 +1,110 @@
>> +#! /bin/bash
>> +# FS QA Test 288
>> +#
>> +# When an attribute leaf block count is 0, xfs_repair should junk
>> +# that leaf directly (as xfsprogs commit f714016).
>> +#
>> +#-----------------------------------------------------------------------
>> +# Copyright (c) 2017 Red Hat Inc. All Rights Reserved.
>> +#
>> +# This program is free software; you can redistribute it and/or
>> +# modify it under the terms of the GNU General Public License as
>> +# published by the Free Software Foundation.
>> +#
>> +# This program is distributed in the hope that it would be useful,
>> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
>> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>> +# GNU General Public License for more details.
>> +#
>> +# You should have received a copy of the GNU General Public License
>> +# along with this program; if not, write the Free Software Foundation,
>> +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
>> +#-----------------------------------------------------------------------
>> +#
>> +
>> +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 15
>> +
>> +_cleanup()
>> +{
>> + cd /
>> + rm -f $tmp.*
>> +}
>> +
>> +# get standard environment, filters and checks
>> +. ./common/rc
>> +. ./common/filter
>> +. ./common/attr
>> +
>> +# remove previous $seqres.full before test
>> +rm -f $seqres.full
>> +
>> +# V4 XFS attr type is 'attr'
>> +write_cmd="write"
>> +
>> +# Modify as appropriate.
>> +_supported_fs xfs
>> +_supported_os Linux
>> +_require_scratch
>> +_require_attrs
>> +
>> +_scratch_mkfs_xfs | _filter_mkfs 2>$tmp.mkfs >/dev/null
>> +. $tmp.mkfs
>> +
>> +# If test on XFS with CRC enabled, after change attr's hdr.count we need
>> +# to recalculate crc by use write -d option
>> +if [ $_fs_has_crcs -eq 1 ]; then
>> + # check if xfs_db support write -d option
>> + _scratch_xfs_db -x -c "sb" -c "write -d" 2>&1 | \
>> + grep -q "invalid option"
>> + if [ $? -eq 0 ];then
>> + _notrun "xfs_db doesn't support write -d option"
>> + fi
>> + # This's a trick to check if xfs_db support recalculating
>> + # CRCs of attr3
>> + _scratch_xfs_db -x -c "daddr 0" \
>> + -c "type attr3" \
>> + -c "write -d" 2>/dev/null | \
>> + grep -q "Cannot recalculate CRCs on this type of object"
>> + if [ $? -eq 0 ];then
>> + _notrun "xfs_db can't recalculate CRCs of attr3"
>> + fi
>> + # V5 XFS attr type is 'attr3'
>> + write_cmd="write -d"
>> +fi
Also, I think it might be a lot easier to just create the fs w/o CRCs,
I don't think we need CRCs to validate the xfs_repair fix.
>> +
>> +_scratch_mount
>> +
>> +touch $SCRATCH_MNT/$seq.attrfile
>> +inum=$(stat -c '%i' $SCRATCH_MNT/$seq.attrfile)
>> +
>> +# To get an attr block leaf, we need to extend attr format to extent
>> +# format at least, and the max inode size is half of filesystem
>> +# block size, so write half of block size attr to make sure attr
>> +# out of local format.
>> +maxisize=$((dbsize/2))
>> +$SETFATTR_PROG -n "user.testattr${seq}" \
>> + -v `python -c "print ($maxisize * 'a')"` \
>
> Need to check for the existence of python, or use xfs_io to write
> $maxisize bytes to a file and backtick-cat it into the command line.
or don't depend on python at all, do something like:
dd if=/dev/zero bs=$maxsize count=1 | attr -s testattr${seq} $SCRATCH_MNT/$seq.attrfile
or /anything/ else that doesn't depend on python, please :)
>> + $SCRATCH_MNT/$seq.attrfile
>> +
>> +_scratch_unmount
>> +# manually corrupt the XFS, by set the header count of attr to 0
>> +_scratch_xfs_db -x -c "inode $inum" \
>> + -c "addr a.bmx[0].startblock" \
>
> "ablock 0" ?
>
>> + -c "$write_cmd hdr.count 0" >> $seqres.full
>> +
>
> Uh... we need an xfs_repair -n run here to validate that repair actually
> finds a broken xattr block. Only then should we fix the fs. Remember
> that xfs_repair only tells us if it thinks the fs is still broken.
>
> --D
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
