| Message ID | 1471265786-24970-1-git-send-email-bfoster@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Aug 15, 2016 at 08:56:26AM -0400, Brian Foster wrote: > XFS had a bug that lead to a possible out-of-order log recovery > situation (e.g., replay a stale modification from the log over more > recent metadata in destination buffer). This resulted in false > corruption reports during log recovery and thus mount failure. > > This condition is caused by system crash or filesystem shutdown shortly > after a successful log recovery. Add a test to run a combined workload, > fs shutdown and log recovery loop known to reproduce the problem on > affected kernels. > > Signed-off-by: Brian Foster <bfoster@redhat.com> > --- > ping > v3: > - More generic test fixups. > v2: http://marc.info/?l=linux-xfs&m=147126275528704&w=2 > - Use $KILLALL_PROG for killall command. > - Convert to generic test. > v1: http://marc.info/?l=linux-xfs&m=147100402211629&w=2 > > tests/generic/999 | 89 +++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/generic/999.out | 2 ++ > tests/generic/group | 1 + > 3 files changed, 92 insertions(+) > create mode 100755 tests/generic/999 > create mode 100644 tests/generic/999.out > > diff --git a/tests/generic/999 b/tests/generic/999 > new file mode 100755 > index 0000000..57bb39c > --- /dev/null > +++ b/tests/generic/999 > @@ -0,0 +1,89 @@ > +#! /bin/bash > +# FS QA Test No. 999 > +# > +# Test XFS log recovery ordering on v5 superblock filesystems. XFS had a problem > +# where it would incorrectly replay older modifications from the log over more > +# recent versions of metadata due to failure to update metadata LSNs during log > +# recovery. This could result in false positive reports of corruption during log > +# recovery and permanent mount failure. > +# > +# To test this situation, run frequent shutdowns immediately after log recovery. > +# Ensure that log recovery does not recover stale modifications and cause > +# spurious corruption reports and/or mount failures. > +# > +#----------------------------------------------------------------------- > +# Copyright (c) 2016 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.* > + $KILLALL_PROG -9 fsstress > /dev/null 2>&1 > + _scratch_unmount > /dev/null 2>&1 > +} > + > +# get standard environment, filters and checks > +. ./common/rc > + > +# Modify as appropriate. > +_supported_fs generic > +_supported_os Linux > + > +_require_scratch > +_require_scratch_shutdown > +_require_command "$KILLALL_PROG" > + > +rm -f $seqres.full > + > +echo "Silence is golden." > + > +_scratch_mkfs >> $seqres.full 2>&1 > +_scratch_mount || _fail "mount failed" > + > +for i in $(seq 1 50); do > + ($FSSTRESS_PROG -d $SCRATCH_MNT -n 999999 -p 4 >> $seqres.full &) \ > + > /dev/null 2>&1 > + > + # purposely include 0 second sleeps to test shutdown immediately after > + # recovery > + sleep $((RANDOM % 3)) > + $XFS_IO_PROG -xc shutdown $SCRATCH_MNT > + > + ps -e | grep fsstress > /dev/null 2>&1 > + while [ $? -eq 0 ]; do > + $KILLALL_PROG -9 fsstress > /dev/null 2>&1 > + wait > /dev/null 2>&1 > + ps -e | grep fsstress > /dev/null 2>&1 > + done > + > + # quit if mount fails so we don't shutdown the host fs > + _scratch_cycle_mount || _fail "cycle mount failed" > +done > + > +# success, all done > +status=0 > +exit > diff --git a/tests/generic/999.out b/tests/generic/999.out > new file mode 100644 > index 0000000..d254382 > --- /dev/null > +++ b/tests/generic/999.out > @@ -0,0 +1,2 @@ > +QA output created by 999 > +Silence is golden. > diff --git a/tests/generic/group b/tests/generic/group > index 4acae99..605a244 100644 > --- a/tests/generic/group > +++ b/tests/generic/group > @@ -375,3 +375,4 @@ > 370 auto quick richacl > 371 auto quick enospc prealloc > 372 auto quick clone > +999 auto log metadata > -- > 2.5.5 > > -- > 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
On Tue, Sep 27, 2016 at 09:12:49AM -0400, Brian Foster wrote: > On Mon, Aug 15, 2016 at 08:56:26AM -0400, Brian Foster wrote: > > XFS had a bug that lead to a possible out-of-order log recovery > > situation (e.g., replay a stale modification from the log over more > > recent metadata in destination buffer). This resulted in false > > corruption reports during log recovery and thus mount failure. > > > > This condition is caused by system crash or filesystem shutdown shortly > > after a successful log recovery. Add a test to run a combined workload, > > fs shutdown and log recovery loop known to reproduce the problem on > > affected kernels. > > > > Signed-off-by: Brian Foster <bfoster@redhat.com> > > --- > > > > ping It's pending in my stage tree, because it crashes current upstream kernel, and Dave wants the fixes go upstream first, so the test won't crash the test machine and interrupt the test. I noticed the fixes are in xfs tree for-next branch, I think we're ready to include this test in next fstests update. Thanks, Eryu -- 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
On Tue, Sep 27, 2016 at 09:37:30PM +0800, Eryu Guan wrote: > On Tue, Sep 27, 2016 at 09:12:49AM -0400, Brian Foster wrote: > > On Mon, Aug 15, 2016 at 08:56:26AM -0400, Brian Foster wrote: > > > XFS had a bug that lead to a possible out-of-order log recovery > > > situation (e.g., replay a stale modification from the log over more > > > recent metadata in destination buffer). This resulted in false > > > corruption reports during log recovery and thus mount failure. > > > > > > This condition is caused by system crash or filesystem shutdown shortly > > > after a successful log recovery. Add a test to run a combined workload, > > > fs shutdown and log recovery loop known to reproduce the problem on > > > affected kernels. > > > > > > Signed-off-by: Brian Foster <bfoster@redhat.com> > > > --- > > > > > > > ping > > It's pending in my stage tree, because it crashes current upstream > kernel, and Dave wants the fixes go upstream first, so the test won't > crash the test machine and interrupt the test. > > I noticed the fixes are in xfs tree for-next branch, I think we're ready > to include this test in next fstests update. Yup, it's good to go. Cheers, Dave.
diff --git a/tests/generic/999 b/tests/generic/999 new file mode 100755 index 0000000..57bb39c --- /dev/null +++ b/tests/generic/999 @@ -0,0 +1,89 @@ +#! /bin/bash +# FS QA Test No. 999 +# +# Test XFS log recovery ordering on v5 superblock filesystems. XFS had a problem +# where it would incorrectly replay older modifications from the log over more +# recent versions of metadata due to failure to update metadata LSNs during log +# recovery. This could result in false positive reports of corruption during log +# recovery and permanent mount failure. +# +# To test this situation, run frequent shutdowns immediately after log recovery. +# Ensure that log recovery does not recover stale modifications and cause +# spurious corruption reports and/or mount failures. +# +#----------------------------------------------------------------------- +# Copyright (c) 2016 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.* + $KILLALL_PROG -9 fsstress > /dev/null 2>&1 + _scratch_unmount > /dev/null 2>&1 +} + +# get standard environment, filters and checks +. ./common/rc + +# Modify as appropriate. +_supported_fs generic +_supported_os Linux + +_require_scratch +_require_scratch_shutdown +_require_command "$KILLALL_PROG" + +rm -f $seqres.full + +echo "Silence is golden." + +_scratch_mkfs >> $seqres.full 2>&1 +_scratch_mount || _fail "mount failed" + +for i in $(seq 1 50); do + ($FSSTRESS_PROG -d $SCRATCH_MNT -n 999999 -p 4 >> $seqres.full &) \ + > /dev/null 2>&1 + + # purposely include 0 second sleeps to test shutdown immediately after + # recovery + sleep $((RANDOM % 3)) + $XFS_IO_PROG -xc shutdown $SCRATCH_MNT + + ps -e | grep fsstress > /dev/null 2>&1 + while [ $? -eq 0 ]; do + $KILLALL_PROG -9 fsstress > /dev/null 2>&1 + wait > /dev/null 2>&1 + ps -e | grep fsstress > /dev/null 2>&1 + done + + # quit if mount fails so we don't shutdown the host fs + _scratch_cycle_mount || _fail "cycle mount failed" +done + +# success, all done +status=0 +exit diff --git a/tests/generic/999.out b/tests/generic/999.out new file mode 100644 index 0000000..d254382 --- /dev/null +++ b/tests/generic/999.out @@ -0,0 +1,2 @@ +QA output created by 999 +Silence is golden. diff --git a/tests/generic/group b/tests/generic/group index 4acae99..605a244 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -375,3 +375,4 @@ 370 auto quick richacl 371 auto quick enospc prealloc 372 auto quick clone +999 auto log metadata
XFS had a bug that lead to a possible out-of-order log recovery situation (e.g., replay a stale modification from the log over more recent metadata in destination buffer). This resulted in false corruption reports during log recovery and thus mount failure. This condition is caused by system crash or filesystem shutdown shortly after a successful log recovery. Add a test to run a combined workload, fs shutdown and log recovery loop known to reproduce the problem on affected kernels. Signed-off-by: Brian Foster <bfoster@redhat.com> --- v3: - More generic test fixups. v2: http://marc.info/?l=linux-xfs&m=147126275528704&w=2 - Use $KILLALL_PROG for killall command. - Convert to generic test. v1: http://marc.info/?l=linux-xfs&m=147100402211629&w=2 tests/generic/999 | 89 +++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/999.out | 2 ++ tests/generic/group | 1 + 3 files changed, 92 insertions(+) create mode 100755 tests/generic/999 create mode 100644 tests/generic/999.out