| Message ID | 1492011577-13436-1-git-send-email-zlang@redhat.com (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Headers | show |
On 4/12/17 10:39 AM, Zorro Lang wrote: > +# This repair should junk above leaf attribute and fix this XFS, then > +# SCRATCH_DEV shouldn't be corrupted after this case done. > +_repair_scratch_fs >> $seqres.full > + > +echo "Silence is golden" > + as mentioned on IRC, I don't think you have any validation here. If you are running with an xfs_repair that does not find the corruption, then neither the repair above nor the post-test xfs_repair -n will find any problems, and the test will incorrectly pass. -Eric -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
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@redhat.com> > --- > > 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 > + > +_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. > + $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 > +# This repair should junk above leaf attribute and fix this XFS, then > +# SCRATCH_DEV shouldn't be corrupted after this case done. > +_repair_scratch_fs >> $seqres.full > + > +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 linux-xfs" 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 linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
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@redhat.com> >> --- >> >> 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@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Apr 12, 2017 at 11:09:35AM -0500, Eric Sandeen wrote: > > > On 4/12/17 10:39 AM, Zorro Lang wrote: > > +# This repair should junk above leaf attribute and fix this XFS, then > > +# SCRATCH_DEV shouldn't be corrupted after this case done. > > +_repair_scratch_fs >> $seqres.full > > + > > +echo "Silence is golden" > > + > > as mentioned on IRC, I don't think you have any validation here. > > If you are running with an xfs_repair that does not find the corruption, > then neither the repair above nor the post-test xfs_repair -n will > find any problems, and the test will incorrectly pass. Sure, you're right, I shouldn't validate the xfs_repair by xfs_repair. I'll change that. Thanks, Zorro > > -Eric > -- > To unsubscribe from this list: send the line "unsubscribe linux-xfs" 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 linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Apr 12, 2017 at 09:14:34AM -0700, 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@redhat.com> > > --- > > > > 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 > > + > > +_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. As you and Eric all don't like this 'python' way, I'll change it :) > > > + $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" ? Anything weird ? > > > + -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. Yes, I'll change it. Thanks, Zorro > > --D > > > +# This repair should junk above leaf attribute and fix this XFS, then > > +# SCRATCH_DEV shouldn't be corrupted after this case done. > > +_repair_scratch_fs >> $seqres.full > > + > > +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 linux-xfs" 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 linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Apr 13, 2017 at 12:27:07AM +0800, Zorro Lang wrote: > On Wed, Apr 12, 2017 at 09:14:34AM -0700, 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@redhat.com> > > > --- > > > > > > 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 > > > + > > > +_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. > > As you and Eric all don't like this 'python' way, I'll change it :) > > > > > > + $SCRATCH_MNT/$seq.attrfile > > > + > > > +_scratch_unmount > > > +# manually corrupt the XFS, by set the header count of attr to 0 > > > +_scratch_xfs_db -x -c "inode $inum" \ > > > + -c "addr a.bmx[0].startblock" \ > > > > "ablock 0" ? > > Anything weird ? Well... if the attr fork should somehow be in btree format then 'ablock 0' will handle it just fine, whereas directly reading the extents-format extent array will break. --D > > > > > + -c "$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. > > Yes, I'll change it. > > Thanks, > Zorro > > > > > --D > > > > > +# This repair should junk above leaf attribute and fix this XFS, then > > > +# SCRATCH_DEV shouldn't be corrupted after this case done. > > > +_repair_scratch_fs >> $seqres.full > > > + > > > +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 linux-xfs" 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 linux-xfs" 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 linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Apr 12, 2017 at 09:55:35AM -0700, Darrick J. Wong wrote: > On Thu, Apr 13, 2017 at 12:27:07AM +0800, Zorro Lang wrote: > > On Wed, Apr 12, 2017 at 09:14:34AM -0700, 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@redhat.com> > > > > --- > > > > > > > > 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 > > > > + > > > > +_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. > > > > As you and Eric all don't like this 'python' way, I'll change it :) > > > > > > > > > + $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" ? > > > > Anything weird ? > > Well... if the attr fork should somehow be in btree format then 'ablock 0' > will handle it just fine, whereas directly reading the extents-format > extent array will break. Wow, yes, that's really helpful! Thanks so much :) > > --D > > > > > > > > + -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. > > > > Yes, I'll change it. > > > > Thanks, > > Zorro > > > > > > > > --D > > > > > > > +# This repair should junk above leaf attribute and fix this XFS, then > > > > +# SCRATCH_DEV shouldn't be corrupted after this case done. > > > > +_repair_scratch_fs >> $seqres.full > > > > + > > > > +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 linux-xfs" 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 linux-xfs" 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 linux-xfs" 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..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 + +_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')"` \ + $SCRATCH_MNT/$seq.attrfile + +_scratch_unmount +# manually corrupt the XFS, by set the header count of attr to 0 +_scratch_xfs_db -x -c "inode $inum" \ + -c "addr a.bmx[0].startblock" \ + -c "$write_cmd hdr.count 0" >> $seqres.full + +# This repair should junk above leaf attribute and fix this XFS, then +# SCRATCH_DEV shouldn't be corrupted after this case done. +_repair_scratch_fs >> $seqres.full + +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, 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