| Message ID | 2f102cc7-a6b2-e23d-9c56-de6c6bc5110d@suse.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Aug 16, 2016 at 02:30:21PM -0400, Jeff Mahoney wrote: > Ensure that we can unmount a read-only file system when quota rescan > is paused from a previous read-write mount. > > If the kernel has a separate bug where we are returning early while > waiting for the rescan worker, we can use that to un-hang the test, > and report both errors. > > This issue is resolved by the following patch for the Linux kernel: > "btrfs: properly track when rescan worker is running" > > Signed-off-by: Jeff Mahoney <jeffm@suse.com> > --- > tests/btrfs/131 | 100 ++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/btrfs/131.out | 3 ++ > tests/btrfs/group | 1 + > 3 files changed, 104 insertions(+) > create mode 100755 tests/btrfs/131 > create mode 100644 tests/btrfs/131.out > > diff --git a/tests/btrfs/131 b/tests/btrfs/131 > new file mode 100755 > index 0000000..56c38a2 > --- /dev/null > +++ b/tests/btrfs/131 > @@ -0,0 +1,100 @@ > +#! /bin/bash > +# FS QA Test 131 > +# > +# Test for bug where read-only mounts will hang on umount when > +# a qgroup rescan was paused. This also tests whether that hung > +# umount can be unhung by trying to make use of a separate bug that > +# means we can interrupt the wait for the rescan worker. If that > +# happens, we report both errors. > +# > +#----------------------------------------------------------------------- > +# Copyright (c) 2016 SUSE. 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 > + > +# remove previous $seqres.full before test > +rm -f $seqres.full > + > +# real QA test starts here > + > +# Modify as appropriate. > +_supported_fs btrfs > +_supported_os Linux > + > +# We'll exit with a quota rescan paused > +_require_scratch_nocheck > + > +_require_btrfs > +BTRFS_DEBUG_TREE_PROG="`set_prog_path btrfs-debug-tree`" > +_require_command "$BTRFS_DEBUG_TREE_PROG" btrfs-debug-tree > +TIMEOUT_PROG="`set_prog_path timeout`" > +_require_command "$TIMEOUT_PROG" timeout The "set_prog_path" calls belong to common/config, we only need "_require_command" calls in test. > + > +rm -f $seqres.full > +_scratch_mkfs >>$seqres.full 2>&1 > + > +_scratch_mount > +cp -aR /lib/modules/$(uname -r) $SCRATCH_MNT Consider populating the filesystem using _populate_fs or fsstress? > + > +# A qgroup rescan on an empty or small file system completes nearly > +# immediately. We need to ensure that it runs long enough that it will > +# be paused on umount. Snapshots slow down the rescan so we should see > +# the race without a lot of data. This is an arbitrary number that > +# works on a ramdisk so it should be sufficient for any storage. > +for n in $(seq 1 100); do > + _run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/$n Use tab for indention. > +done > +_run_btrfs_util_prog quota enable $SCRATCH_MNT > +_scratch_unmount > + > +echo "read-write umount completed" > + > +# Confirm that the rescan is paused > +if ! $BTRFS_DEBUG_TREE_PROG $SCRATCH_DEV | \ > + egrep -q 'flags ON|SCANNING|INCONSISTENT scan'; then > + echo "qgroup rescan not paused." > +fi > +_scratch_mount -r > + > +# If the bug exists, this will hang. If we can kill it, that's another bug. > +$TIMEOUT_PROG 10 umount $SCRATCH_MNT > +if test $? -eq 124 ; then > + echo "umount hung but was killed" I see this log with 4.8-rc1 kernel, is that expected? Thanks, Eryu > +fi > +echo "read-only umount completed" > + > +# success, all done > +status=0 > +exit > diff --git a/tests/btrfs/131.out b/tests/btrfs/131.out > new file mode 100644 > index 0000000..845a501 > --- /dev/null > +++ b/tests/btrfs/131.out > @@ -0,0 +1,3 @@ > +QA output created by 131 > +read-write umount completed > +read-only umount completed > diff --git a/tests/btrfs/group b/tests/btrfs/group > index 6b29c05..929fa21 100644 > --- a/tests/btrfs/group > +++ b/tests/btrfs/group > @@ -133,3 +133,4 @@ > 128 auto quick send > 129 auto quick send > 130 auto clone send > +131 auto quick qgroup > -- > 1.8.5.6 > > > -- > Jeff Mahoney > SUSE Labs > -- > 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
At 08/17/2016 04:45 PM, Eryu Guan wrote: > On Tue, Aug 16, 2016 at 02:30:21PM -0400, Jeff Mahoney wrote: >> Ensure that we can unmount a read-only file system when quota rescan >> is paused from a previous read-write mount. >> >> If the kernel has a separate bug where we are returning early while >> waiting for the rescan worker, we can use that to un-hang the test, >> and report both errors. >> >> This issue is resolved by the following patch for the Linux kernel: >> "btrfs: properly track when rescan worker is running" >> >> Signed-off-by: Jeff Mahoney <jeffm@suse.com> >> --- >> tests/btrfs/131 | 100 ++++++++++++++++++++++++++++++++++++++++++++++++++++ >> tests/btrfs/131.out | 3 ++ >> tests/btrfs/group | 1 + >> 3 files changed, 104 insertions(+) >> create mode 100755 tests/btrfs/131 >> create mode 100644 tests/btrfs/131.out >> >> diff --git a/tests/btrfs/131 b/tests/btrfs/131 >> new file mode 100755 >> index 0000000..56c38a2 >> --- /dev/null >> +++ b/tests/btrfs/131 >> @@ -0,0 +1,100 @@ >> +#! /bin/bash >> +# FS QA Test 131 >> +# >> +# Test for bug where read-only mounts will hang on umount when >> +# a qgroup rescan was paused. This also tests whether that hung >> +# umount can be unhung by trying to make use of a separate bug that >> +# means we can interrupt the wait for the rescan worker. If that >> +# happens, we report both errors. >> +# >> +#----------------------------------------------------------------------- >> +# Copyright (c) 2016 SUSE. 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 >> + >> +# remove previous $seqres.full before test >> +rm -f $seqres.full >> + >> +# real QA test starts here >> + >> +# Modify as appropriate. >> +_supported_fs btrfs >> +_supported_os Linux >> + >> +# We'll exit with a quota rescan paused >> +_require_scratch_nocheck >> + >> +_require_btrfs >> +BTRFS_DEBUG_TREE_PROG="`set_prog_path btrfs-debug-tree`" >> +_require_command "$BTRFS_DEBUG_TREE_PROG" btrfs-debug-tree >> +TIMEOUT_PROG="`set_prog_path timeout`" >> +_require_command "$TIMEOUT_PROG" timeout > > The "set_prog_path" calls belong to common/config, we only need > "_require_command" calls in test. > >> + >> +rm -f $seqres.full >> +_scratch_mkfs >>$seqres.full 2>&1 >> + >> +_scratch_mount >> +cp -aR /lib/modules/$(uname -r) $SCRATCH_MNT > > Consider populating the filesystem using _populate_fs or fsstress? If only want to increase tree level, best solution is to create a lot of small 2K files. (Needs max_inline=2K incase user override it) _populate_fs with -s 2048 is for this case. If want normal files, -s 16384 would ensure it won't be inlined and still small enough for quick population. > >> + >> +# A qgroup rescan on an empty or small file system completes nearly >> +# immediately. We need to ensure that it runs long enough that it will >> +# be paused on umount. Snapshots slow down the rescan so we should see >> +# the race without a lot of data. This is an arbitrary number that >> +# works on a ramdisk so it should be sufficient for any storage. >> +for n in $(seq 1 100); do >> + _run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/$n > > Use tab for indention. > >> +done >> +_run_btrfs_util_prog quota enable $SCRATCH_MNT >> +_scratch_unmount >> + >> +echo "read-write umount completed" >> + >> +# Confirm that the rescan is paused >> +if ! $BTRFS_DEBUG_TREE_PROG $SCRATCH_DEV | \ >> + egrep -q 'flags ON|SCANNING|INCONSISTENT scan'; then >> + echo "qgroup rescan not paused." >> +fi I'm a little concerned of this operation. As debug-tree output can change at any time, since it doesn't expect it to be used as a stable tool to exam some flag. Although I don't have a better alternative though. Thanks, Qu >> +_scratch_mount -r >> + >> +# If the bug exists, this will hang. If we can kill it, that's another bug. >> +$TIMEOUT_PROG 10 umount $SCRATCH_MNT >> +if test $? -eq 124 ; then >> + echo "umount hung but was killed" > > I see this log with 4.8-rc1 kernel, is that expected? > > Thanks, > Eryu > >> +fi >> +echo "read-only umount completed" >> + >> +# success, all done >> +status=0 >> +exit >> diff --git a/tests/btrfs/131.out b/tests/btrfs/131.out >> new file mode 100644 >> index 0000000..845a501 >> --- /dev/null >> +++ b/tests/btrfs/131.out >> @@ -0,0 +1,3 @@ >> +QA output created by 131 >> +read-write umount completed >> +read-only umount completed >> diff --git a/tests/btrfs/group b/tests/btrfs/group >> index 6b29c05..929fa21 100644 >> --- a/tests/btrfs/group >> +++ b/tests/btrfs/group >> @@ -133,3 +133,4 @@ >> 128 auto quick send >> 129 auto quick send >> 130 auto clone send >> +131 auto quick qgroup >> -- >> 1.8.5.6 >> >> >> -- >> Jeff Mahoney >> SUSE Labs >> -- >> To unsubscribe from this list: send the line "unsubscribe fstests" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html > -- > To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 8/17/16 4:45 AM, Eryu Guan wrote: > On Tue, Aug 16, 2016 at 02:30:21PM -0400, Jeff Mahoney wrote: >> +_require_btrfs >> +BTRFS_DEBUG_TREE_PROG="`set_prog_path btrfs-debug-tree`" >> +_require_command "$BTRFS_DEBUG_TREE_PROG" btrfs-debug-tree >> +TIMEOUT_PROG="`set_prog_path timeout`" >> +_require_command "$TIMEOUT_PROG" timeout > > The "set_prog_path" calls belong to common/config, we only need > "_require_command" calls in test. Ok, then we should probably fix up the existing cases that do this in tests/ since this and the cp /lib/modules thing were copied from other test cases in the repo. >> + >> +rm -f $seqres.full >> +_scratch_mkfs >>$seqres.full 2>&1 >> + >> +_scratch_mount >> +cp -aR /lib/modules/$(uname -r) $SCRATCH_MNT > > Consider populating the filesystem using _populate_fs or fsstress? Same. >> + >> +# A qgroup rescan on an empty or small file system completes nearly >> +# immediately. We need to ensure that it runs long enough that it will >> +# be paused on umount. Snapshots slow down the rescan so we should see >> +# the race without a lot of data. This is an arbitrary number that >> +# works on a ramdisk so it should be sufficient for any storage. >> +for n in $(seq 1 100); do >> + _run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/$n > > Use tab for indention. Ok. >> +done >> +_run_btrfs_util_prog quota enable $SCRATCH_MNT >> +_scratch_unmount >> + >> +echo "read-write umount completed" >> + >> +# Confirm that the rescan is paused >> +if ! $BTRFS_DEBUG_TREE_PROG $SCRATCH_DEV | \ >> + egrep -q 'flags ON|SCANNING|INCONSISTENT scan'; then >> + echo "qgroup rescan not paused." >> +fi >> +_scratch_mount -r >> + >> +# If the bug exists, this will hang. If we can kill it, that's another bug. >> +$TIMEOUT_PROG 10 umount $SCRATCH_MNT >> +if test $? -eq 124 ; then >> + echo "umount hung but was killed" > > I see this log with 4.8-rc1 kernel, is that expected? Yes, the patch was posted this week and hasn't landed yet. Thanks, -Jeff
diff --git a/tests/btrfs/131 b/tests/btrfs/131 new file mode 100755 index 0000000..56c38a2 --- /dev/null +++ b/tests/btrfs/131 @@ -0,0 +1,100 @@ +#! /bin/bash +# FS QA Test 131 +# +# Test for bug where read-only mounts will hang on umount when +# a qgroup rescan was paused. This also tests whether that hung +# umount can be unhung by trying to make use of a separate bug that +# means we can interrupt the wait for the rescan worker. If that +# happens, we report both errors. +# +#----------------------------------------------------------------------- +# Copyright (c) 2016 SUSE. 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 + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs btrfs +_supported_os Linux + +# We'll exit with a quota rescan paused +_require_scratch_nocheck + +_require_btrfs +BTRFS_DEBUG_TREE_PROG="`set_prog_path btrfs-debug-tree`" +_require_command "$BTRFS_DEBUG_TREE_PROG" btrfs-debug-tree +TIMEOUT_PROG="`set_prog_path timeout`" +_require_command "$TIMEOUT_PROG" timeout + +rm -f $seqres.full +_scratch_mkfs >>$seqres.full 2>&1 + +_scratch_mount +cp -aR /lib/modules/$(uname -r) $SCRATCH_MNT + +# A qgroup rescan on an empty or small file system completes nearly +# immediately. We need to ensure that it runs long enough that it will +# be paused on umount. Snapshots slow down the rescan so we should see +# the race without a lot of data. This is an arbitrary number that +# works on a ramdisk so it should be sufficient for any storage. +for n in $(seq 1 100); do + _run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/$n +done +_run_btrfs_util_prog quota enable $SCRATCH_MNT +_scratch_unmount + +echo "read-write umount completed" + +# Confirm that the rescan is paused +if ! $BTRFS_DEBUG_TREE_PROG $SCRATCH_DEV | \ + egrep -q 'flags ON|SCANNING|INCONSISTENT scan'; then + echo "qgroup rescan not paused." +fi +_scratch_mount -r + +# If the bug exists, this will hang. If we can kill it, that's another bug. +$TIMEOUT_PROG 10 umount $SCRATCH_MNT +if test $? -eq 124 ; then + echo "umount hung but was killed" +fi +echo "read-only umount completed" + +# success, all done +status=0 +exit diff --git a/tests/btrfs/131.out b/tests/btrfs/131.out new file mode 100644 index 0000000..845a501 --- /dev/null +++ b/tests/btrfs/131.out @@ -0,0 +1,3 @@ +QA output created by 131 +read-write umount completed +read-only umount completed diff --git a/tests/btrfs/group b/tests/btrfs/group index 6b29c05..929fa21 100644 --- a/tests/btrfs/group +++ b/tests/btrfs/group @@ -133,3 +133,4 @@ 128 auto quick send 129 auto quick send 130 auto clone send +131 auto quick qgroup
Ensure that we can unmount a read-only file system when quota rescan is paused from a previous read-write mount. If the kernel has a separate bug where we are returning early while waiting for the rescan worker, we can use that to un-hang the test, and report both errors. This issue is resolved by the following patch for the Linux kernel: "btrfs: properly track when rescan worker is running" Signed-off-by: Jeff Mahoney <jeffm@suse.com> --- tests/btrfs/131 | 100 ++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/131.out | 3 ++ tests/btrfs/group | 1 + 3 files changed, 104 insertions(+) create mode 100755 tests/btrfs/131 create mode 100644 tests/btrfs/131.out