From: Eryu Guan <eguan@redhat.com>
To: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com>
Cc: fstests@vger.kernel.org, linux-btrfs@vger.kernel.org
Subject: Re: [PATCH] generic/371: run write(2) and fallocate(2) in parallel
Date: Thu, 21 Jul 2016 18:41:00 +0800 [thread overview]
Message-ID: <20160721104100.GD27776@eguan.usersys.redhat.com> (raw)
In-Reply-To: <20160721073025.25557-1-wangxg.fnst@cn.fujitsu.com>
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.fnst@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).
> +
> +_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>
> + 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.
> +
> +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
next prev parent reply other threads:[~2016-07-21 10:41 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-07-21 7:30 [PATCH] generic/371: run write(2) and fallocate(2) in parallel Wang Xiaoguang
2016-07-21 10:41 ` Eryu Guan [this message]
2016-07-21 23:05 ` Dave Chinner
2016-07-25 7:10 ` Wang Xiaoguang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20160721104100.GD27776@eguan.usersys.redhat.com \
--to=eguan@redhat.com \
--cc=fstests@vger.kernel.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=wangxg.fnst@cn.fujitsu.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.