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@xxxxxxxx>
---
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@xxxxxxxxxxxxxxx
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@xxxxxxxxxxxxxxx
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@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
