hello,
On 07/21/2016 06:41 PM, Eryu Guan wrote:
On Thu, Jul 21, 2016 at 03:30:25PM +0800, Wang Xiaoguang wrote:
Currently in btrfs, there is something wrong with fallocate(2)'s data
space reservation, it'll temporarily occupy more data space thant it
really needs, which in turn will impact other operations' data request.
In this test case, it runs write(2) and fallocate(2) in parallel and the
total needed data space for these two operations don't exceed whole fs
free data space, to see whether we will get any unexpected ENOSPC error.
Signed-off-by: Wang Xiaoguang <wangxg.f...@cn.fujitsu.com>
---
tests/generic/371 | 119 ++++++++++++++++++++++++++++++++++++++++++++++++++
tests/generic/371.out | 2 +
tests/generic/group | 1 +
3 files changed, 122 insertions(+)
create mode 100755 tests/generic/371
create mode 100644 tests/generic/371.out
diff --git a/tests/generic/371 b/tests/generic/371
new file mode 100755
index 0000000..b85327a
--- /dev/null
+++ b/tests/generic/371
@@ -0,0 +1,119 @@
+#! /bin/bash
+# FS QA Test 371
+#
+# Run write(2) and fallocate(2) in parallel and the total needed data space
+# for these operations don't exceed whole fs free data space, to see whether
+# we will get any unexpected ENOSPC error.
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2016 Fujitsu. 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
+
+# Modify as appropriate.
+_supported_fs generic
+_supported_os Linux
+_require_scratch
Need '_require_xfs_io_command "falloc"', otherwise test fails on ext3/2,
because they don't support fallocate(2).
OK, I see.
+
+_scratch_mkfs_sized $((256 * 1024 * 1024)) >> $seqres.full 2>&1
+_scratch_mount
+
+testfile1=$SCRATCH_MNT/testfile1
+testfile2=$SCRATCH_MNT/testfile2
+
+write_work()
+{
+ rm -f $testfile1
+ while [ 1 ]; do
+ $XFS_IO_PROG -f -c "pwrite 0 80M" $testfile1 >>$seqres.full 2>&1
+ grep "No space left on device" $seqres.full >/dev/null
+ if [ $? -eq 0 ]; then
+ echo "unexpected ENOSPC error occurs"
+ exit 1
+ fi
No need to grep for error message, just redirect stdout to /dev/null and
any error will appear in stderr, which will break golden image.
And $seqres.full is where we dump logs for debug purpose, tests should
dump output they need to somewhere like $tmp.<suffix>
I see.
+ rm -f $testfile1
+ done
+}
+
+fallocate_work()
+{
+ rm -f $testfile2
+ while [ 1 ]; do
+ $XFS_IO_PROG -f -c "falloc 0 80M" $testfile2 >>$seqres.full 2>&1
+ grep "No space left on device" $seqres.full >/dev/null
+ if [ $? -eq 0 ]; then
+ echo "unexpected ENOSPC error occurs"
+ exit 1
+ fi
+ rm -f $testfile2
+ done
+}
+
+run_time=$((180 * $TIME_FACTOR))
180s is too long time, I can reproduce it in around 10s on my test vm,
just loop for 100 times for each operation (pwrite and falloc)
+write_work &
+write_work_pid=$!
+fallocate_work &
+fallocate_work_pid=$!
+
+for ((elapsed_time = 0; elapsed_time < run_time; elapsed_time += 5)); do
+ kill -0 $write_work_pid >/dev/null 2>&1
+ if [ $? -ne 0 ]; then
+ kill $fallocate_work_pid >/dev/null 2>&1
+ break
+ fi
+
+ kill -0 $fallocate_work_pid >/dev/null 2>&1
+ if [ $? -ne 0 ]; then
+ kill $write_work_pid >/dev/null 2>&1
+ break
+ fi
+ sleep 5
+done
+
+kill $fallocate_work_pid >/dev/null 2>&1
+kill $write_work_pid >/dev/null 2>&1
+wait
+
+# wait un-finished xfs_io
+while ps aux | grep "xfs_io" | grep -qv grep; do
+ sleep 1
+done
And this seems unnecessarily complicated
So I'd write it as:
echo "Silence is golden"
for ((i=0; i<100; i++)); do
$XFS_IO_PROG -fc "pwrite 0 80M" $testfile1 >/dev/null
rm -f $testfile1
done &
pids=$!
for ((i=0; i<100; i++)); do
$XFS_IO_PROG -fc "falloc 0 80M" $testfile2 >/dev/null
rm -f $testfile2
done &
pids="$pids $!"
wait $pids
status=0
exit
btrfs failed due to extra "pwrite64: No space left on device" output from
xfs_io, and ext4 and xfs passed the test within 15s for me.
Thanks for you cleaner scripts.
New patch will be sent soon.
Regards,
Xiaoguang Wang
+
+echo "Silence is golden"
+status=0
+exit
diff --git a/tests/generic/371.out b/tests/generic/371.out
new file mode 100644
index 0000000..22ec8a2
--- /dev/null
+++ b/tests/generic/371.out
@@ -0,0 +1,2 @@
+QA output created by 371
+Silence is golden
diff --git a/tests/generic/group b/tests/generic/group
index 97ecb65..3d4a802 100644
--- a/tests/generic/group
+++ b/tests/generic/group
@@ -373,3 +373,4 @@
368 auto quick richacl
369 auto quick richacl
370 auto quick richacl
+371 auto enospc prealloc stress
So we can add 'quick' group and remove 'stress'.
Thanks,
Eryu
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
