| Message ID | 1492064601-13298-1-git-send-email-zlang@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Apr 13, 2017 at 02:23:21PM +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@redhat.com> > --- > > Hi, > > V3 did below changes: > 1) Only test on V4 XFS. > 2) Use 'ablock 0' to instead of "a.bmx[0].startblock" > 3) Use "dd ... | attr -s ..." to instead of that python command in > "setfattr ..." > 4) Do "xfs_repair -n" to check if xfs_repair can find the corruption > we made. > 5) Use xfs_db to check if xfs_repair has fixed the corruption properly. > I check it by trying to locate 'ablock 0' again. Should I turn to > check if core.naextents == 0? > > Thanks, > Zorro > > tests/xfs/288 | 109 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/288.out | 2 + > tests/xfs/group | 1 + > 3 files changed, 112 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..5e6167f > --- /dev/null > +++ b/tests/xfs/288 > @@ -0,0 +1,109 @@ > +#! /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" This variable is not used in this test. > + > +# Modify as appropriate. > +_supported_fs xfs > +_supported_os Linux > +_require_scratch > +_require_attrs > + > +# Due to xfs_db's write -d option is still not stable, and there's no > +# plan to support attr3. So only run this case on V4 XFS. > +# Please update this if xfs_db get enough improvement in one day. > +if [ -z "$XFS_MKFS_HAS_NO_META_SUPPORT" ]; then > + mkfs_opts="-m crc=0" > +fi > +$MKFS_XFS_PROG -f $mkfs_opts $SCRATCH_DEV | _filter_mkfs 2>$tmp.mkfs >/dev/null > +. $tmp.mkfs _scratch_mkfs "$mkfs_opts" | _filter_mkfs ... If there's mkfs option conflicts between $MKFS_OPTIONS and local $mkfs_opts, _scratch_mkfs will try mkfs again with user specified $mkfs_opts only. > + > +_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 > +# or btree 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. > +dd if=/dev/zero bs=$((dbsize/2)) count=1 2>/dev/null | \ > + $ATTR_PROG -s testattr $SCRATCH_MNT/$seq.attrfile >/dev/null I think we can use perl here, as perl is widely used in fstests and no new dependency needed. Thanks, Eryu > + > +_scratch_unmount > +# manually corrupt the XFS, by set the header count of attr to 0 > +_scratch_xfs_db -x -c "inode $inum" \ > + -c "ablock 0" \ > + -c "write hdr.count 0" >> $seqres.full > + > +# make sure xfs_repair can find above corruption. If it can't, that > +# means we need to fix this bug on current xfs_repair > +_scratch_xfs_repair -n >> $seqres.full 2>&1 > +if [ $? -eq 0 ];then > + _fail "xfs_repair can't find the corruption" > +else > + # If xfs_repair can find this corruption, then this repair > + # should junk above leaf attribute and fix this XFS. > + _scratch_xfs_repair >> $seqres.full 2>&1 > + > + # Old xfs_repair maybe find and fix this corruption by > + # reset the first used heap value and the usedbytes cnt > + # in ablock 0. That's not what we want. So check if > + # xfs_repair has junked the whole ablock 0 by xfs_db. > + _scratch_xfs_db -x -c "inode $inum" -c "ablock 0" | \ > + grep -q "no attribute data" > + if [ $? -ne 0 ]; then > + _fail "xfs_repair didn't junk the empty attr leaf" > + fi > +fi > + > +echo "Silence is golden" > + > +# success, all done > +status=0 > +exit > diff --git a/tests/xfs/288.out b/tests/xfs/288.out > new file mode 100644 > index 0000000..2958a5c > --- /dev/null > +++ b/tests/xfs/288.out > @@ -0,0 +1,2 @@ > +QA output created by 288 > +Silence is golden > diff --git a/tests/xfs/group b/tests/xfs/group > index 75769f9..f27a7b6 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -285,6 +285,7 @@ > 285 dangerous_fuzzers dangerous_scrub > 286 dangerous_fuzzers dangerous_scrub dangerous_online_repair > 287 auto dump quota quick > +288 auto quick repair fuzzers > 290 auto rw prealloc quick ioctl zero > 291 auto repair > 292 auto mkfs quick > -- > 2.7.4 > > -- > To unsubscribe from this list: send the line "unsubscribe fstests" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/tests/xfs/288 b/tests/xfs/288 new file mode 100755 index 0000000..5e6167f --- /dev/null +++ b/tests/xfs/288 @@ -0,0 +1,109 @@ +#! /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 $seqres.full + +# V4 XFS attr type is 'attr' +write_cmd="write" + +# Modify as appropriate. +_supported_fs xfs +_supported_os Linux +_require_scratch +_require_attrs + +# Due to xfs_db's write -d option is still not stable, and there's no +# plan to support attr3. So only run this case on V4 XFS. +# Please update this if xfs_db get enough improvement in one day. +if [ -z "$XFS_MKFS_HAS_NO_META_SUPPORT" ]; then + mkfs_opts="-m crc=0" +fi +$MKFS_XFS_PROG -f $mkfs_opts $SCRATCH_DEV | _filter_mkfs 2>$tmp.mkfs >/dev/null +. $tmp.mkfs + +_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 +# or btree 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. +dd if=/dev/zero bs=$((dbsize/2)) count=1 2>/dev/null | \ + $ATTR_PROG -s testattr $SCRATCH_MNT/$seq.attrfile >/dev/null + +_scratch_unmount +# manually corrupt the XFS, by set the header count of attr to 0 +_scratch_xfs_db -x -c "inode $inum" \ + -c "ablock 0" \ + -c "write hdr.count 0" >> $seqres.full + +# make sure xfs_repair can find above corruption. If it can't, that +# means we need to fix this bug on current xfs_repair +_scratch_xfs_repair -n >> $seqres.full 2>&1 +if [ $? -eq 0 ];then + _fail "xfs_repair can't find the corruption" +else + # If xfs_repair can find this corruption, then this repair + # should junk above leaf attribute and fix this XFS. + _scratch_xfs_repair >> $seqres.full 2>&1 + + # Old xfs_repair maybe find and fix this corruption by + # reset the first used heap value and the usedbytes cnt + # in ablock 0. That's not what we want. So check if + # xfs_repair has junked the whole ablock 0 by xfs_db. + _scratch_xfs_db -x -c "inode $inum" -c "ablock 0" | \ + grep -q "no attribute data" + if [ $? -ne 0 ]; then + _fail "xfs_repair didn't junk the empty attr leaf" + fi +fi + +echo "Silence is golden" + +# success, all done +status=0 +exit diff --git a/tests/xfs/288.out b/tests/xfs/288.out new file mode 100644 index 0000000..2958a5c --- /dev/null +++ b/tests/xfs/288.out @@ -0,0 +1,2 @@ +QA output created by 288 +Silence is golden diff --git a/tests/xfs/group b/tests/xfs/group index 75769f9..f27a7b6 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -285,6 +285,7 @@ 285 dangerous_fuzzers dangerous_scrub 286 dangerous_fuzzers dangerous_scrub dangerous_online_repair 287 auto dump quota quick +288 auto quick repair fuzzers 290 auto rw prealloc quick ioctl zero 291 auto repair 292 auto mkfs quick
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@redhat.com> --- Hi, V3 did below changes: 1) Only test on V4 XFS. 2) Use 'ablock 0' to instead of "a.bmx[0].startblock" 3) Use "dd ... | attr -s ..." to instead of that python command in "setfattr ..." 4) Do "xfs_repair -n" to check if xfs_repair can find the corruption we made. 5) Use xfs_db to check if xfs_repair has fixed the corruption properly. I check it by trying to locate 'ablock 0' again. Should I turn to check if core.naextents == 0? Thanks, Zorro tests/xfs/288 | 109 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/288.out | 2 + tests/xfs/group | 1 + 3 files changed, 112 insertions(+) create mode 100755 tests/xfs/288 create mode 100644 tests/xfs/288.out