| Message ID | 1413423740-7022-1-git-send-email-quwenruo@cn.fujitsu.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
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
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 fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
-------- 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 [ $ddret -ne 0 ]; then >> + echo "Extent merge bug detected" >> + status=1 >> + exit >> +else >> + echo "Silence is golden" >> + status=0 >> + exit >> +fi >> diff --git a/tests/btrfs/078.out b/tests/btrfs/078.out >> new file mode 100644 >> index 0000000..b8acea8 >> --- /dev/null >> +++ b/tests/btrfs/078.out >> @@ -0,0 +1,2 @@ >> +QA output created by 078 >> +Silence is golden >> diff --git a/tests/btrfs/group b/tests/btrfs/group >> index 9adf862..40e7430 100644 >> --- a/tests/btrfs/group >> +++ b/tests/btrfs/group >> @@ -80,3 +80,4 @@ >> 075 auto quick subvol >> 076 auto quick >> 077 auto quick >> +078 auto >> -- >> 2.1.2 >> >> -- >> To unsubscribe from this list: send the line "unsubscribe fstests" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Oct 16, 2014 at 04:32:06PM +0800, Qu Wenruo wrote: > -------- Original Message -------- > Subject: Re: [PATCH] btrfs: regression test for btrfs extent merge > From: Eryu Guan <eguan@redhat.com> > >>+ > >>+tmp=`mktemp -d` > >I'm a bit concerned about this. Some functions in common/rc use > >$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.
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
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