From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from cn.fujitsu.com ([59.151.112.132]:62023 "EHLO heian.cn.fujitsu.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1751580AbdF2FVT (ORCPT ); Thu, 29 Jun 2017 01:21:19 -0400 Message-ID: <59548E48.5030405@cn.fujitsu.com> Date: Thu, 29 Jun 2017 13:21:12 +0800 From: Xiao Yang MIME-Version: 1.0 Subject: Re: [PATCH] xfs: test a race between dio reads and mapped writes References: <1498554232-20777-1-git-send-email-yangx.jy@cn.fujitsu.com> <20170628101329.GH23360@eguan.usersys.redhat.com> <59547B95.9000905@cn.fujitsu.com> <20170629043402.GJ23360@eguan.usersys.redhat.com> In-Reply-To: <20170629043402.GJ23360@eguan.usersys.redhat.com> Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Sender: fstests-owner@vger.kernel.org To: Eryu Guan Cc: fstests@vger.kernel.org List-ID: On 2017/06/29 12:34, Eryu Guan wrote: > On Thu, Jun 29, 2017 at 12:01:25PM +0800, Xiao Yang wrote: >> On 2017/06/28 18:13, Eryu Guan wrote: >>> On Tue, Jun 27, 2017 at 05:03:52PM +0800, Xiao Yang wrote: >>>> This regression test reproduces a race between a direct >>>> I/O read and a mapped write to a hole in a file. If the >>>> race occurs, it will trigger a BUG_ON(). This XFS bug >>>> has been fixed by: >>>> >>>> 04197b3 (xfs: don't BUG() on mixed direct and mapped I/O) >>>> >>>> Signed-off-by: Xiao Yang >>>> --- >>>> tests/xfs/422 | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ >>>> tests/xfs/422.out | 2 ++ >>>> tests/xfs/group | 1 + >>>> 3 files changed, 89 insertions(+) >>>> create mode 100755 tests/xfs/422 >>>> create mode 100644 tests/xfs/422.out >>>> >>>> diff --git a/tests/xfs/422 b/tests/xfs/422 >>>> new file mode 100755 >>>> index 0000000..a57ffab >>>> --- /dev/null >>>> +++ b/tests/xfs/422 >>>> @@ -0,0 +1,86 @@ >>>> +#! /bin/bash >>>> +# FS QA Test No. 422 >>>> +# >>>> +# Regression test for commit: >>>> +# 04197b3 ("xfs: don't BUG() on mixed direct and mapped I/O) >>> ^^ missing ending " :) >>>> +# >>>> +# This case tests a race between a direct I/O read and a >>>> +# mapped write to a hole in a file. If the race occurs, >>>> +# it will trigger a BUG_ON(). >>>> +# >>>> +#----------------------------------------------------------------------- >>>> +# Copyright (c) 2017 Fujitsu. All Rights Reserved. >>>> +# Author: Xiao Yang >>>> +# >>>> +# 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() >>>> +{ >>>> + rm -rf $tmp.* >>>> +} >>>> + >>>> +# get standard environment and checks >>>> +. ./common/rc >>>> + >>>> +# real QA test starts here >>>> +_supported_os Linux >>>> +_supported_fs xfs >>> I see nothing xfs specific in this test, I think it can be made generic >>> easily. >> Hi Eryu, >> >> Agreed. It could be moved into generic. >>>> +_require_scratch >>>> +_disable_dmesg_check >>> Better to add comments on ignoring dmesg check result. >> For xfs filesystem, this race triggers various warnings on some >> distributions(RHEL7) and upstream kernel. >> e.g, >> 1) This race triggers the following warnings on RHEL7.3GA >> ----------------------------------------------------------------------------------------------------------------------------- >> [ 0.297084] WARNING: at arch/x86/mm/ioremap.c:101 >> __ioremap_caller+0x238/0x320() > I'm not sure about this one, but the other xfs or iomap warnings are > "known issues" when mixing direct I/O and mmap I/O on the same file. > We've filtered out these intentional warnings in generic/095, and Brian > has just sent a patch[1] to move this filter to common/xfs and added it > to generic/247. I think you can do the same to this new test, filter out > these known warnings so we won't miss any other potential bug. > > I don't think disable dmesg check only because it triggers warnings is a > good idea :) Hi Eryu It sounds reasonable to me. I will filter out these known warnings instead of disabling dmesg check. > Thanks, > Eryu > > [1] [PATCH] generic/247: filter out expected XFS warnings for mixed mmap/direct I/O > >> [ 648.745481] WARNING: at fs/xfs/xfs_aops.c:1244 >> xfs_vm_releasepage+0xcb/0x100 [xfs]() >> [ 648.745694] WARNING: at fs/xfs/xfs_file.c:377 >> xfs_file_aio_read+0x2dd/0x340 [xfs]() >> ----------------------------------------------------------------------------------------------------------------------------- >> 2) It triggers the following warnings on RHEL7.4Beta >> ---------------------------------------------------------------------------------------------------------------------------- >> [ 497.575319] WARNING: CPU: 0 PID: 26865 at fs/xfs/xfs_file.c:361 >> xfs_file_dio_aio_read+0x1ba/0x2a0 [xfs] >> [ 497.577651] WARNING: CPU: 0 PID: 26865 at fs/xfs/xfs_aops.c:1285 >> __xfs_get_blocks+0x8ca/0x8e0 [xfs] Could we add __xfs_get_blocks into filter_xfs_dmesg? :-) This known warning was introduced by: 04197b3 (xfs: don't BUG() on mixed direct and mapped I/O) It has been removed by: acdda3a (xfs: use iomap_dio_rw) Thanks, Xiao Yang >> ---------------------------------------------------------------------------------------------------------------------------- >> 3) It triggers the following warnings on upstream kernel >> ---------------------------------------------------------------------------------------------------------------------------- >> [ 57.946882] WARNING: CPU: 1 PID: 4053 at fs/iomap.c:890 >> iomap_dio_rw+0x37a/0x390 >> [ 57.947066] WARNING: CPU: 1 PID: 4053 at fs/iomap.c:766 >> iomap_dio_actor+0xca/0x370 >> ---------------------------------------------------------------------------------------------------------------------------- >> >> It focus on the oops due to this race, and the warnings are different based >> on different kernel. >> So we may ignore dmesg check. >> >> I will add comments on ignoring dmesg check result as you said. >> >> Thanks, >> Xiao Yang >>> Thanks, >>> Eryu >>> >>>> +_require_xfs_io_command "truncate" >>>> +_require_xfs_io_command "fpunch" >>>> + >>>> +rm -f $seqres.full >>>> + >>>> +# format and mount >>>> +_scratch_mkfs> $seqres.full 2>&1 >>>> +_scratch_mount>> $seqres.full 2>&1 >>>> + >>>> +filesz=$((65536 * 2)) >>>> + >>>> +# create a test file with a hole >>>> +$XFS_IO_PROG -f -c "truncate $((filesz * 2))" $SCRATCH_MNT/file>> $seqres.full >>>> + >>>> +# run a background dio read to a hole in a loop >>>> +while true; do >>>> + $XFS_IO_PROG -d -c "pread 0 $filesz" $SCRATCH_MNT/file> /dev/null 2>&1 >>>> +done& >>>> + >>>> +dread_pid=$! >>>> + >>>> +# run mapped write to the same hole as dio read >>>> +for i in `seq 0 999`; do >>>> + $XFS_IO_PROG -c "mmap 0 $filesz" -c "mwrite 0 $filesz" $SCRATCH_MNT/file \ >>>> + > /dev/null >>>> + $XFS_IO_PROG -c "fpunch 0 $filesz" $SCRATCH_MNT/file> /dev/null >>>> +done >>>> + >>>> +kill -9 $dread_pid> /dev/null 2>&1 >>>> +wait $dread_pid> /dev/null 2>&1 >>>> + >>>> +echo "Silence is golden" >>>> + >>>> +# success, all done >>>> +status=0 >>>> +exit >>>> diff --git a/tests/xfs/422.out b/tests/xfs/422.out >>>> new file mode 100644 >>>> index 0000000..f70693f >>>> --- /dev/null >>>> +++ b/tests/xfs/422.out >>>> @@ -0,0 +1,2 @@ >>>> +QA output created by 422 >>>> +Silence is golden >>>> diff --git a/tests/xfs/group b/tests/xfs/group >>>> index 2a3b950..f43ffcf 100644 >>>> --- a/tests/xfs/group >>>> +++ b/tests/xfs/group >>>> @@ -419,3 +419,4 @@ >>>> 419 auto quick >>>> 420 auto quick clone dedupe >>>> 421 auto quick clone dedupe >>>> +422 auto quick dangerous >>>> -- >>>> 1.8.3.1 >>>> >>>> >>>> >>>> -- >>>> 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 >>> . >>> >> >> > > . >