btrfs: regression test for btrfs extent merge
diff mbox

Message ID 1413423740-7022-1-git-send-email-quwenruo@cn.fujitsu.com
State New, archived
Headers show

Commit Message

Qu Wenruo Oct. 16, 2014, 1:42 a.m. UTC
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

Comments

Filipe Manana Oct. 16, 2014, 7:44 a.m. UTC | #1
On Thu, Oct 16, 2014 at 2:42 AM, Qu Wenruo <quwenruo@cn.fujitsu.com> 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>

Tested-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>

Thanks very much for this Qu.

> ---
>  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`
> +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
> +
> +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'
> +
> +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
Eryu Guan Oct. 16, 2014, 8:13 a.m. UTC | #2
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.

> +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?

> +
> +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
> +
> +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
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Qu Wenruo Oct. 16, 2014, 8:32 a.m. UTC | #3
-------- 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?10?16? 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

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Dave Chinner Nov. 10, 2014, 1:35 a.m. UTC | #4
On Thu, Oct 16, 2014 at 04:32:06PM +0800, Qu Wenruo wrote:
> -------- Original Message --------
> Subject: Re: [PATCH] btrfs: regression test for btrfs extent merge
> From: Eryu Guan <eguan@redhat.com>
> >>+
> >>+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.

Can you please send a patch to fix btrfs/077 as well?

Cheers,

Dave.

Patch
diff mbox

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`
+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
+
+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'
+
+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