Re: [PATCH] btrfs: regression test for btrfs extent merge

[Qu Wenruo <quwenruo@cn.fujitsu.com>]

-------- Original Message --------
Subject: Re: [PATCH] btrfs: regression test for btrfs extent merge
From: Eryu Guan <eguan@redhat.com>
To: Qu Wenruo <quwenruo@cn.fujitsu.com>
Date: 2014年10月16日 16:13
> On Thu, Oct 16, 2014 at 09:42:20AM +0800, Qu Wenruo wrote:
>> The following kernel commit introduced the bug:
>> 51f395ad btrfs: Use right extent length when inserting overlap extent map.
>>
>> When btrfs commit race with btrfs_get_extent(), merge_extent_mapping()
>> may build up a new extent which length overflows and cause extent map
>> insert fail, causing the caller get a -EEXIST error.
>>
>> This regression is fixed by the following patches:
>> btrfs: Fix and enhance merge_extent_mapping() to insert best fitted extent map
>> btrfs: Fix the wrong condition judgment about subset extent map
>>
>> Cc: Filipe Manana <fdmanana@suse.com>
>> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
>> ---
>>   tests/btrfs/078     | 128 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   tests/btrfs/078.out |   2 +
>>   tests/btrfs/group   |   1 +
>>   3 files changed, 131 insertions(+)
>>   create mode 100755 tests/btrfs/078
>>   create mode 100644 tests/btrfs/078.out
>>
>> diff --git a/tests/btrfs/078 b/tests/btrfs/078
>> new file mode 100755
>> index 0000000..bdc20af
>> --- /dev/null
>> +++ b/tests/btrfs/078
>> @@ -0,0 +1,128 @@
>> +#! /bin/bash
>> +# FS QA Test No. btrfs/078
>> +#
>> +# Do write along with fiemap ioctl.
>> +# Regression test for the kernel comit:
>> +# 51f395ad btrfs: Use right extent length when inserting overlap extent map.
>> +#
>> +# When calling fiemap(without SYNC flag) and btrfs fs is commiting,
>> +# it will cause race condition and cause btrfs to generate a wrong extent
>> +# whose len is overflow and fail to insert into the extent map tree,
>> +# returning -EEXIST.
>> +#
>> +# Fixed by the following patches (not merged in mainline yet):
>> +# btrfs: Fix and enhance merge_extent_mapping() to insert best fitted extent map
>> +# btrfs: Fix the wrong condition judgment about subset extent map
>> +#
>> +#-----------------------------------------------------------------------
>> +# Copyright (C) 2014 Fujitsu All Rights Reserved.
>> +# Author: Qu Wenruo <quwenruo@cn.fujitsu.com>
>> +#
>> +# 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"
>> +
>> +tmp=`mktemp -d`
> I'm a bit concerned about this. Some functions in common/rc use
> $tmp.some_sufix as temp file. If $tmp defined as a dir here, I'm not
> sure if it will cause any trouble.
It seems that most tests uses /tmp/$$ as tmp, only some exception like 
btrfs/077 uses mktemp -d.
So I'd better use the old /tmp/$$ method.
>
>> +status=1	# failure is the default!
>> +trap "_cleanup; exit \$status" 0 1 2 3 15
>> +
>> +_cleanup()
>> +{
>> +	kill $dd_pid &> /dev/null
>> +	kill $fiemap_pid &> /dev/null
>> +	wait
>> +	rm -fr $test_file
>> +	rm -fr $tmp
>> +}
>> +
>> +# get standard environment, filters and checks
>> +. ./common/rc
>> +. ./common/filter
>> +
>> +# real QA test starts here
>> +_supported_fs btrfs
>> +_supported_os Linux
>> +_require_scratch
>> +_need_to_be_root
>> +# Since xfs_io's fiemap always use SYNC flag and can't be unset,
>> +# we must use filefrag to call fiemap without SYNC flag.
>> +_require_command "/usr/sbin/filefrag"
>> +_require_xfs_io_command "falloc"
>> +
>> +test_file=$SCRATCH_MNT/testfile
>> +filesize=$((10 * 1024 * 1024 * 1024)) #10G size
>> +buffersize=$((1024 * 1024)) # 1M bs for dd
>> +count=$(($filesize / $buffersize))
>> +testfile=$SCRATCH_MNT/testfile
> test_file and testfile are both set to $SCRATCH_MNT/testfile and you
> use both $test_file and $testfile through the test, seems you can use
> only one?
Oh, my mistake, should only be one testfile.
>
>> +
>> +rm -f $seqres.full
>> +rm -f $test_file
>> +
>> +_scratch_mkfs >>$seqres.full 2>&1
>> +_scratch_mount
>> +$XFS_IO_PROG -f -c "falloc 0 $filesize" $testfile || \
>> +	_fail 'need device larger than 10G'
> You can use _require_fs_space() helper here.
>
> Thanks,
> Eryu
Thanks for pointing out this, really helps a lot.

Thanks,
Qu
>> +
>> +dd_work() {
>> +	out=$1
>> +	dd if=/dev/zero of=$out bs=$buffersize count=$count \
>> +	   conv=notrunc &> /dev/null
>> +}
>> +
>> +# There is a bug for e2fsprogs, at least in version 1.42.9, filefrag will
>> +# leak the return value, so we can't judge return value only,
>> +# but also to filter the output
>> +_filter_error() {
>> +	# when filefrag fails FIEMAP ioctl, it will fall back to FIBMAP,
>> +	# which is not supported by btrfs and will report "FIBMAP: strerr()"
>> +	# However due to other btrfs commits, kernel will return -ENOTTY
>> +	# or -EINVAL, so only grep for "FIBMAP" for max compatibility.
>> +	grep "FIBMAP"
>> +}
>> +
>> +fiemap_work() {
>> +	filename=$1
>> +	while true; do
>> +		filefrag $filename 2> $tmp/output 1> /dev/null
>> +		ret=$?
>> +		err=`cat $tmp/output | _filter_error`
>> +		if [ $ret -ne 0 -o -n "$err" ]; then
>> +			kill $dd_pid
>> +			return 1
>> +		fi
>> +	done
>> +}
>> +
>> +dd_work $testfile &
>> +dd_pid=$!
>> +fiemap_work $testfile &
>> +fiemap_pid=$!
>> +wait $dd_pid
>> +ddret=$?
>> +kill $fiemap_pid &> /dev/null
>> +wait $fiemap_pid
>> +
>> +if [ $ddret -ne 0 ]; then
>> +	echo "Extent merge bug detected"
>> +	status=1
>> +	exit
>> +else
>> +	echo "Silence is golden"
>> +	status=0
>> +	exit
>> +fi
>> diff --git a/tests/btrfs/078.out b/tests/btrfs/078.out
>> new file mode 100644
>> index 0000000..b8acea8
>> --- /dev/null
>> +++ b/tests/btrfs/078.out
>> @@ -0,0 +1,2 @@
>> +QA output created by 078
>> +Silence is golden
>> diff --git a/tests/btrfs/group b/tests/btrfs/group
>> index 9adf862..40e7430 100644
>> --- a/tests/btrfs/group
>> +++ b/tests/btrfs/group
>> @@ -80,3 +80,4 @@
>>   075 auto quick subvol
>>   076 auto quick
>>   077 auto quick
>> +078 auto
>> -- 
>> 2.1.2
>>
>> --
>> 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