[PATCH typo-fixed] fstests: btrfs: 159 superblock corruption test case

[PATCH typo-fixed] fstests: btrfs: 159 superblock corruption test case

[Anand Jain]

Verify if the superblock corruption is handled correctly.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 tests/btrfs/159     | 142 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/btrfs/159.out |  35 +++++++++++++
 tests/btrfs/group   |   1 +
 3 files changed, 178 insertions(+)
 create mode 100755 tests/btrfs/159
 create mode 100644 tests/btrfs/159.out

diff --git a/tests/btrfs/159 b/tests/btrfs/159
new file mode 100755
index 000000000000..f42ba4b777e7
--- /dev/null
+++ b/tests/btrfs/159
@@ -0,0 +1,142 @@
+#! /bin/bash
+# FS QA Test 159
+#
+# Test if the superblock corruption is handled correctly.
+#   - Check and validate superblock csum in mount and scan context
+#   - Check and validate superblock for all disks in the fs
+#   - Make sure if the copies are really a copy of the primary superblock
+#
+#---------------------------------------------------------------------
+# Copyright (c) 2018 Oracle.  All Rights Reserved.
+# Author: Anand Jain <anand.jain@oracle.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"
+status=1	# failure is the default!
+
+_cleanup()
+{
+	_scratch_dev_pool_put
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/module
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+
+_supported_fs btrfs
+_supported_os Linux
+_require_scratch_dev_pool 2
+_require_loadable_fs_module "btrfs"
+_require_command "$WIPEFS_PROG" wipefs
+
+_scratch_dev_pool_get 2
+
+MNT_OPT=$(echo $MOUNT_OPTIONS | cut -d" " -f2-)
+DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
+DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
+
+wipe()
+{
+	$WIPEFS_PROG -a $DEV_GOOD > /dev/null 2>&1
+	$WIPEFS_PROG -a $DEV_BAD > /dev/null 2>&1
+}
+
+# If copy1 fsid does not match with primary fail the scan thus the mount as well
+check_copy1_fsid()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1_fsid\\n{" | tee -a $seqres.full
+
+	wipe
+	$MKFS_BTRFS_PROG -fq $DEV_GOOD
+	$MKFS_BTRFS_PROG -fq $DEV_BAD
+	_reload_fs_module "btrfs"
+
+	dd status=none of=$DEV_BAD if=$DEV_GOOD ibs=1 obs=1 skip=$bytenr seek=$bytenr count=4K|\
+		tee -a $seqres.full
+
+	_mount -o $MNT_OPT $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
+
+	echo -e "good\\n}" | tee -a $seqres.full
+}
+
+# As in Linux kernel 4.16 if the primary is corrupted mount will fail.
+# Which might change in the long run.
+check_primary()
+{
+	local bytenr=65536
+	echo -e "\\ncheck_primary\\n{" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#To corrupt a disk block, read in hex, write in dec
+	od -j$bytenr -N1 -An -x $DEV_BAD |\
+		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1|\
+		tee -a $seqres.full
+	_mount -o $MNT_OPT,device=$DEV_GOOD $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
+
+	echo -e "good\\n}" | tee -a $seqres.full
+}
+
+# If copy1 or copy2 is corrupted we still should be able to mount
+check_copy1()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1\\n{" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#corrupt the disk block bytenr, read in hex, write in dec
+	od -j$bytenr -N1 -An -x $DEV_BAD |\
+		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1|\
+		tee -a $seqres.full
+	_mount -o $MNT_OPT,device=$DEV_GOOD $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
+
+	echo -e "good\\n}" | tee -a $seqres.full
+
+	_scratch_unmount
+}
+
+check_copy1_fsid
+
+check_primary
+check_copy1
+
+echo -e "\\nReverse the good and bad device"
+# Generally devid 1 is used to read tree at the mount time, now reverse the
+# devid on which the corrupt superblock will reside.
+dev_tmp=$DEV_GOOD
+DEV_GOOD=$DEV_BAD
+DEV_BAD=$dev_tmp
+check_primary
+check_copy1
+
+status=0
+exit
diff --git a/tests/btrfs/159.out b/tests/btrfs/159.out
new file mode 100644
index 000000000000..f8f776f28870
--- /dev/null
+++ b/tests/btrfs/159.out
@@ -0,0 +1,35 @@
+QA output created by 159
+
+check_copy1_fsid
+{
+mount: wrong fs type, bad option, bad superblock on SCRATCH_DEV,
+       missing codepage or helper program, or other error
+
+       In some cases useful info is found in syslog - try
+       dmesg | tail or so.
+good
+}
+
+check_primary
+{
+mount: mount SCRATCH_DEV on /scratch failed: Structure needs cleaning
+good
+}
+
+check_copy1
+{
+good
+}
+
+Reverse the good and bad device
+
+check_primary
+{
+mount: mount SCRATCH_DEV on /scratch failed: Structure needs cleaning
+good
+}
+
+check_copy1
+{
+good
+}
diff --git a/tests/btrfs/group b/tests/btrfs/group
index 8007e07e9cbd..ba766f6b84f8 100644
--- a/tests/btrfs/group
+++ b/tests/btrfs/group
@@ -161,3 +161,4 @@
 156 auto quick trim
 157 auto quick raid
 158 auto quick raid scrub
+159 auto quick
-- 
2.7.0

Re: [PATCH typo-fixed] fstests: btrfs: 159 superblock corruption test case

[Eryu Guan]

On Thu, Mar 29, 2018 at 06:28:48PM +0800, Anand Jain wrote:
> Verify if the superblock corruption is handled correctly.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
>  tests/btrfs/159     | 142 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/159.out |  35 +++++++++++++
>  tests/btrfs/group   |   1 +
>  3 files changed, 178 insertions(+)
>  create mode 100755 tests/btrfs/159
>  create mode 100644 tests/btrfs/159.out
> 
> diff --git a/tests/btrfs/159 b/tests/btrfs/159
> new file mode 100755
> index 000000000000..f42ba4b777e7
> --- /dev/null
> +++ b/tests/btrfs/159
> @@ -0,0 +1,142 @@
> +#! /bin/bash
> +# FS QA Test 159
> +#
> +# Test if the superblock corruption is handled correctly.
> +#   - Check and validate superblock csum in mount and scan context
> +#   - Check and validate superblock for all disks in the fs
> +#   - Make sure if the copies are really a copy of the primary superblock
> +#
> +#---------------------------------------------------------------------
> +# Copyright (c) 2018 Oracle.  All Rights Reserved.
> +# Author: Anand Jain <anand.jain@oracle.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"
> +status=1	# failure is the default!

Please use './new btrfs' to generate new test template, which defines
variables like 'tmp' and 'here', these variables may not be used in the
test explicitly, but they could be used by some helper functions,
especially the "$tmp" var. (Or please don't remove them from template :)

> +
> +_cleanup()
> +{
> +	_scratch_dev_pool_put
> +}
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +. ./common/module
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +_supported_fs btrfs
> +_supported_os Linux
> +_require_scratch_dev_pool 2
> +_require_loadable_fs_module "btrfs"
> +_require_command "$WIPEFS_PROG" wipefs
> +
> +_scratch_dev_pool_get 2
> +
> +MNT_OPT=$(echo $MOUNT_OPTIONS | cut -d" " -f2-)

I'm not sure about the purpose of this MNT_OPT, and it could be empty
and causes problems in test. Please see below.

> +DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
> +DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
> +
> +wipe()
> +{
> +	$WIPEFS_PROG -a $DEV_GOOD > /dev/null 2>&1
> +	$WIPEFS_PROG -a $DEV_BAD > /dev/null 2>&1
> +}
> +
> +# If copy1 fsid does not match with primary fail the scan thus the mount as well
> +check_copy1_fsid()
> +{
> +	local bytenr=67108864
> +	echo -e "\\ncheck_copy1_fsid\\n{" | tee -a $seqres.full

Hmm, I don't think the "{ }" block is necessary in .out file, IMHO it
makes the diff context harder to read when test fails. Just echo the
test name would be fine.

> +
> +	wipe
> +	$MKFS_BTRFS_PROG -fq $DEV_GOOD
> +	$MKFS_BTRFS_PROG -fq $DEV_BAD
> +	_reload_fs_module "btrfs"
> +
> +	dd status=none of=$DEV_BAD if=$DEV_GOOD ibs=1 obs=1 skip=$bytenr seek=$bytenr count=4K|\
> +		tee -a $seqres.full
> +
> +	_mount -o $MNT_OPT $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool

When $MOUNT_OPTIONS is empty, this mount fails as

+mount: can't find /mnt/scratch in /etc/fstab

And I think we need to umount $SCRATCH_MNT here in case a buggy btrfs
module allowed the mount, as the next test assumes the device is not
mounted anywhere. Otherwise I see failures like:

+ERROR: /dev/loop1 is mounted
+mount failed

> +
> +	echo -e "good\\n}" | tee -a $seqres.full
> +}
> +
> +# As in Linux kernel 4.16 if the primary is corrupted mount will fail.
> +# Which might change in the long run.
> +check_primary()
> +{
> +	local bytenr=65536
> +	echo -e "\\ncheck_primary\\n{" | tee -a $seqres.full
> +
> +	wipe
> +	_scratch_pool_mkfs "-mraid1 -draid1"
> +	_reload_fs_module "btrfs"
> +
> +	#To corrupt a disk block, read in hex, write in dec
> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1|\
> +		tee -a $seqres.full
> +	_mount -o $MNT_OPT,device=$DEV_GOOD $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool

Same here, need to umount $SCRATCH_MNT.

Also, if a mount failure is expected (as it's recorded in the golden
image file), you need _filter_error_mount here, and _filter_scratch_pool
can be dropped, because _filter_error_mount will remove all
$SCRATCH_DEV/MNT from the mount output. For more details please see
commit 94d3a4f00cbd ("fstests: filter mount error message for EUCLEAN
and ESTALE").

Thanks,
Eryu

> +
> +	echo -e "good\\n}" | tee -a $seqres.full
> +}
> +
> +# If copy1 or copy2 is corrupted we still should be able to mount
> +check_copy1()
> +{
> +	local bytenr=67108864
> +	echo -e "\\ncheck_copy1\\n{" | tee -a $seqres.full
> +
> +	wipe
> +	_scratch_pool_mkfs "-mraid1 -draid1"
> +	_reload_fs_module "btrfs"
> +
> +	#corrupt the disk block bytenr, read in hex, write in dec
> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1|\
> +		tee -a $seqres.full
> +	_mount -o $MNT_OPT,device=$DEV_GOOD $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
> +
> +	echo -e "good\\n}" | tee -a $seqres.full
> +
> +	_scratch_unmount
> +}
> +
> +check_copy1_fsid
> +
> +check_primary
> +check_copy1
> +
> +echo -e "\\nReverse the good and bad device"
> +# Generally devid 1 is used to read tree at the mount time, now reverse the
> +# devid on which the corrupt superblock will reside.
> +dev_tmp=$DEV_GOOD
> +DEV_GOOD=$DEV_BAD
> +DEV_BAD=$dev_tmp
> +check_primary
> +check_copy1
> +
> +status=0
> +exit
> diff --git a/tests/btrfs/159.out b/tests/btrfs/159.out
> new file mode 100644
> index 000000000000..f8f776f28870
> --- /dev/null
> +++ b/tests/btrfs/159.out
> @@ -0,0 +1,35 @@
> +QA output created by 159
> +
> +check_copy1_fsid
> +{
> +mount: wrong fs type, bad option, bad superblock on SCRATCH_DEV,
> +       missing codepage or helper program, or other error
> +
> +       In some cases useful info is found in syslog - try
> +       dmesg | tail or so.
> +good
> +}
> +
> +check_primary
> +{
> +mount: mount SCRATCH_DEV on /scratch failed: Structure needs cleaning
> +good
> +}
> +
> +check_copy1
> +{
> +good
> +}
> +
> +Reverse the good and bad device
> +
> +check_primary
> +{
> +mount: mount SCRATCH_DEV on /scratch failed: Structure needs cleaning
> +good
> +}
> +
> +check_copy1
> +{
> +good
> +}
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index 8007e07e9cbd..ba766f6b84f8 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -161,3 +161,4 @@
>  156 auto quick trim
>  157 auto quick raid
>  158 auto quick raid scrub
> +159 auto quick
> -- 
> 2.7.0
> 
> --
> 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

Re: [PATCH typo-fixed] fstests: btrfs: 159 superblock corruption test case

[Anand Jain]

>> +seq=`basename $0`
>> +seqres=$RESULT_DIR/$seq
>> +echo "QA output created by $seq"
>> +status=1	# failure is the default!
> 
> Please use './new btrfs' to generate new test template, which defines
> variables like 'tmp' and 'here', these variables may not be used in the
> test explicitly, but they could be used by some helper functions,
> especially the "$tmp" var. (Or please don't remove them from template :)

  Ok. May be its a good idea to have an _init() to set mandatory stuffs?

>> +MNT_OPT=$(echo $MOUNT_OPTIONS | cut -d" " -f2-)

> I'm not sure about the purpose of this MNT_OPT, and it could be empty
> and causes problems in test. Please see below.

  Test case here need control of the device(s) used for the mount.
  Thats awkward I missed no mount-option secnario. Will fix.

>> +	echo -e "\\ncheck_copy1_fsid\\n{" | tee -a $seqres.full
> 
> Hmm, I don't think the "{ }" block is necessary in .out file, IMHO it
> makes the diff context harder to read when test fails. Just echo the
> test name would be fine.

  Ok. Will take that out.


>> +	_mount -o $MNT_OPT $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
> 
> When $MOUNT_OPTIONS is empty, this mount fails as
> 
> +mount: can't find /mnt/scratch in /etc/fstab
> 
> And I think we need to umount $SCRATCH_MNT here in case a buggy btrfs
> module allowed the mount, as the next test assumes the device is not
> mounted anywhere. Otherwise I see failures like:
> 
> +ERROR: /dev/loop1 is mounted
> +mount failed

  got it.

>> +
>> +	echo -e "good\\n}" | tee -a $seqres.full
>> +}
>> +
>> +# As in Linux kernel 4.16 if the primary is corrupted mount will fail.
>> +# Which might change in the long run.
>> +check_primary()
>> +{
>> +	local bytenr=65536
>> +	echo -e "\\ncheck_primary\\n{" | tee -a $seqres.full
>> +
>> +	wipe
>> +	_scratch_pool_mkfs "-mraid1 -draid1"
>> +	_reload_fs_module "btrfs"
>> +
>> +	#To corrupt a disk block, read in hex, write in dec
>> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
>> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1|\
>> +		tee -a $seqres.full
>> +	_mount -o $MNT_OPT,device=$DEV_GOOD $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
> 
> Same here, need to umount $SCRATCH_MNT.
> 
> Also, if a mount failure is expected (as it's recorded in the golden
> image file), you need _filter_error_mount here, and _filter_scratch_pool
> can be dropped, because _filter_error_mount will remove all
> $SCRATCH_DEV/MNT from the mount output. For more details please see
> commit 94d3a4f00cbd ("fstests: filter mount error message for EUCLEAN
> and ESTALE").

  I am fixing this in v2.

Thanks, Anand

[PATCH v2] fstests: btrfs/159 superblock corruption test case

[Anand Jain]

Verify if the superblock corruption is handled correctly.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
v1->v2:
 $subject slightly changed
 Added more info about the test-case
 Keep the stuff from the ./new btrfs
 Add mount_opt_minus_args() to get the options (if) set at the config file
 Move DEV_GOOD & DEV_BAD to where it starts to use
 To help debugging added run_check where possible
 Remove {} in the out file
 Use _filter_error_mount for mount fail cases other than -EINVAL

 tests/btrfs/159     | 177 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/btrfs/159.out |  23 +++++++
 tests/btrfs/group   |   1 +
 3 files changed, 201 insertions(+)
 create mode 100755 tests/btrfs/159
 create mode 100644 tests/btrfs/159.out

diff --git a/tests/btrfs/159 b/tests/btrfs/159
new file mode 100755
index 000000000000..521cfdab0242
--- /dev/null
+++ b/tests/btrfs/159
@@ -0,0 +1,177 @@
+#! /bin/bash
+# FS QA Test 159
+#
+# Test if the superblock corruption is handled correctly:
+# 	- Test fsid miss-match (csum ok) between primary and copy superblock
+#	Fixed by the ML patch:
+#	btrfs: check if the fsid in the primary sb and copy sb are same
+# 	- Test if the mount fails if the primary superblock csum is
+#		corrupted on any disk
+# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
+#	Fixed by the ML patches:
+#	btrfs: verify superblock checksum during scan
+#	btrfs: verify checksum for all devices in mount context
+#
+#---------------------------------------------------------------------
+# Copyright (c) 2018 Oracle.  All Rights Reserved.
+# Author: Anand Jain <anand.jain@oracle.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"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+	_scratch_dev_pool_put
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/module
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+
+_supported_fs btrfs
+_supported_os Linux
+_require_scratch_dev_pool 2
+_require_loadable_fs_module "btrfs"
+_require_command "$WIPEFS_PROG" wipefs
+
+_scratch_dev_pool_get 2
+
+mount_opt_minus_args()
+{
+	local nr
+	local mnt_opt=""
+
+	nr=`_scratch_mount_options | awk '{print NF}'`
+	if [ $nr -eq 4 ]; then
+		#gets only the mount option set in the config file
+		mnt_opt=`_scratch_mount_options | awk '{print $2}'`
+	fi
+	#Append the additional opts provide as func args.
+	#Make sure -o is not echo-ed if both config file mnt-option
+	#and the test case mnt-option are null.
+	if [ -z $mnt_opt ]; then
+		if [ ! -z $* ]; then
+			echo "-o $*"
+		fi
+	else
+		if [ -z $* ]; then
+			echo "-o $mnt_opt"
+		else
+			echo "-o $mnt_opt,$*"
+		fi
+	fi
+}
+
+wipe()
+{
+	$WIPEFS_PROG -a $DEV_GOOD > /dev/null 2>&1
+	$WIPEFS_PROG -a $DEV_BAD > /dev/null 2>&1
+}
+
+# Test for fsid miss-match (csum ok) with primary and copy superblock.
+check_copy1_fsid()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1_fsid\\n" | tee -a $seqres.full
+
+	wipe
+	$MKFS_BTRFS_PROG -fq $DEV_GOOD
+	$MKFS_BTRFS_PROG -fq $DEV_BAD
+	_reload_fs_module "btrfs"
+
+	run_check dd status=none of=$DEV_BAD if=$DEV_GOOD ibs=1 obs=1\
+			skip=$bytenr seek=$bytenr count=4K
+
+	#must fail
+	_mount `mount_opt_minus_args` $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
+	#If mount didn't fail
+	_scratch_unmount > /dev/null 2>&1
+}
+
+# Test if the mount fails if the primary superblock csum is corrupted.
+check_primary()
+{
+	local bytenr=65536
+	echo -e "\\ncheck_primary\\n" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#corrupt bytenr DEV_BAD
+	od -j$bytenr -N1 -An -x $DEV_BAD |\
+		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
+
+	#must fail
+	_mount `mount_opt_minus_args device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT 2>&1 |\
+	      _filter_error_mount
+
+	#If mount didn't fail
+	_scratch_unmount > /dev/null 2>&1
+}
+
+# If copy1 is corrupted we still should be able to mount, check that.
+check_copy1()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#corrupt bytenr DEV_BAD
+	od -j$bytenr -N1 -An -x $DEV_BAD |\
+		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
+
+	_mount `mount_opt_minus_args device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT 2>&1 |\
+		_filter_error_mount
+	_scratch_unmount
+}
+
+DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
+DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
+
+check_copy1_fsid
+check_primary
+check_copy1
+
+echo -e "\\nReverse good and bad devices"
+# For csum verification on devid 2, reverse the good and bad disks.
+DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
+DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
+check_primary
+check_copy1
+
+# success, all done
+status=0
+exit
diff --git a/tests/btrfs/159.out b/tests/btrfs/159.out
new file mode 100644
index 000000000000..3e7ba31929e6
--- /dev/null
+++ b/tests/btrfs/159.out
@@ -0,0 +1,23 @@
+QA output created by 159
+
+check_copy1_fsid
+
+mount: wrong fs type, bad option, bad superblock on SCRATCH_DEV,
+       missing codepage or helper program, or other error
+
+       In some cases useful info is found in syslog - try
+       dmesg | tail or so.
+
+check_primary
+
+mount: Structure needs cleaning
+
+check_copy1
+
+Reverse good and bad devices
+
+check_primary
+
+mount: Structure needs cleaning
+
+check_copy1
diff --git a/tests/btrfs/group b/tests/btrfs/group
index 8007e07e9cbd..ba766f6b84f8 100644
--- a/tests/btrfs/group
+++ b/tests/btrfs/group
@@ -161,3 +161,4 @@
 156 auto quick trim
 157 auto quick raid
 158 auto quick raid scrub
+159 auto quick
-- 
2.7.0

Re: [PATCH v2] fstests: btrfs/159 superblock corruption test case

[Eryu Guan]

On Thu, Apr 05, 2018 at 02:28:49PM +0800, Anand Jain wrote:
> Verify if the superblock corruption is handled correctly.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
> v1->v2:
>  $subject slightly changed
>  Added more info about the test-case
>  Keep the stuff from the ./new btrfs
>  Add mount_opt_minus_args() to get the options (if) set at the config file
>  Move DEV_GOOD & DEV_BAD to where it starts to use
>  To help debugging added run_check where possible
>  Remove {} in the out file
>  Use _filter_error_mount for mount fail cases other than -EINVAL
> 
>  tests/btrfs/159     | 177 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/159.out |  23 +++++++
>  tests/btrfs/group   |   1 +
>  3 files changed, 201 insertions(+)
>  create mode 100755 tests/btrfs/159
>  create mode 100644 tests/btrfs/159.out
> 
> diff --git a/tests/btrfs/159 b/tests/btrfs/159
> new file mode 100755
> index 000000000000..521cfdab0242
> --- /dev/null
> +++ b/tests/btrfs/159
> @@ -0,0 +1,177 @@
> +#! /bin/bash
> +# FS QA Test 159
> +#
> +# Test if the superblock corruption is handled correctly:
> +# 	- Test fsid miss-match (csum ok) between primary and copy superblock
> +#	Fixed by the ML patch:
> +#	btrfs: check if the fsid in the primary sb and copy sb are same
> +# 	- Test if the mount fails if the primary superblock csum is
> +#		corrupted on any disk
> +# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
> +#	Fixed by the ML patches:
> +#	btrfs: verify superblock checksum during scan
> +#	btrfs: verify checksum for all devices in mount context
> +#
> +#---------------------------------------------------------------------
> +# Copyright (c) 2018 Oracle.  All Rights Reserved.
> +# Author: Anand Jain <anand.jain@oracle.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"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> +	cd /
> +	rm -f $tmp.*
> +	_scratch_dev_pool_put
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +. ./common/module
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +_supported_fs btrfs
> +_supported_os Linux
> +_require_scratch_dev_pool 2
> +_require_loadable_fs_module "btrfs"
> +_require_command "$WIPEFS_PROG" wipefs
> +
> +_scratch_dev_pool_get 2
> +
> +mount_opt_minus_args()
> +{
> +	local nr
> +	local mnt_opt=""
> +
> +	nr=`_scratch_mount_options | awk '{print NF}'`
> +	if [ $nr -eq 4 ]; then

Seems this only works with "scratch_mount_options" returns something
like:

"-o opt dev mnt" or "-o opt1,opt2 dev mnt"

but not

"-oopt dev mnt" nor
"-o opt1 -o opt2 dev mnt" nor if $SELINUX_MOUNT_OPTIONS not empty.

Also if MOUNT_OPTIONS is "-oopt1 -oopt2", mount_opt_minus_args would
return something like "-o -oopt2,device=<dev>", which are not valid
mount options.

> +		#gets only the mount option set in the config file
> +		mnt_opt=`_scratch_mount_options | awk '{print $2}'`
> +	fi
> +	#Append the additional opts provide as func args.
> +	#Make sure -o is not echo-ed if both config file mnt-option
> +	#and the test case mnt-option are null.
> +	if [ -z $mnt_opt ]; then
> +		if [ ! -z $* ]; then
> +			echo "-o $*"
> +		fi
> +	else
> +		if [ -z $* ]; then
> +			echo "-o $mnt_opt"
> +		else
> +			echo "-o $mnt_opt,$*"
> +		fi
> +	fi
> +}
> +
> +wipe()
> +{
> +	$WIPEFS_PROG -a $DEV_GOOD > /dev/null 2>&1
> +	$WIPEFS_PROG -a $DEV_BAD > /dev/null 2>&1
> +}
> +
> +# Test for fsid miss-match (csum ok) with primary and copy superblock.
> +check_copy1_fsid()
> +{
> +	local bytenr=67108864
> +	echo -e "\\ncheck_copy1_fsid\\n" | tee -a $seqres.full
> +
> +	wipe
> +	$MKFS_BTRFS_PROG -fq $DEV_GOOD
> +	$MKFS_BTRFS_PROG -fq $DEV_BAD
> +	_reload_fs_module "btrfs"
> +
> +	run_check dd status=none of=$DEV_BAD if=$DEV_GOOD ibs=1 obs=1\
> +			skip=$bytenr seek=$bytenr count=4K
> +
> +	#must fail
> +	_mount `mount_opt_minus_args` $DEV_BAD $SCRATCH_MNT 2>&1 | _filter_scratch_pool
> +	#If mount didn't fail
> +	_scratch_unmount > /dev/null 2>&1
> +}
> +
> +# Test if the mount fails if the primary superblock csum is corrupted.
> +check_primary()
> +{
> +	local bytenr=65536
> +	echo -e "\\ncheck_primary\\n" | tee -a $seqres.full
> +
> +	wipe
> +	_scratch_pool_mkfs "-mraid1 -draid1"
> +	_reload_fs_module "btrfs"
> +
> +	#corrupt bytenr DEV_BAD
> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
> +
> +	#must fail
> +	_mount `mount_opt_minus_args device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT 2>&1 |\
> +	      _filter_error_mount

If I read the code correctly, the purpose of mount_opt_minus_args is to
remove the device and mount point mount arguments and keep other options
if there's any, then append "-o device=<dev>" as mount options.

If so, I think you could use "_common_dev_mount_options", e.g.

	_mount `_common_dev_mount_options -o device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT ...

But do mount options really matter in this test? Could we just mount
with "-o device=<dev>"?

Thanks,
Eryu

> +
> +	#If mount didn't fail
> +	_scratch_unmount > /dev/null 2>&1
> +}
> +
> +# If copy1 is corrupted we still should be able to mount, check that.
> +check_copy1()
> +{
> +	local bytenr=67108864
> +	echo -e "\\ncheck_copy1" | tee -a $seqres.full
> +
> +	wipe
> +	_scratch_pool_mkfs "-mraid1 -draid1"
> +	_reload_fs_module "btrfs"
> +
> +	#corrupt bytenr DEV_BAD
> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
> +
> +	_mount `mount_opt_minus_args device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT 2>&1 |\
> +		_filter_error_mount
> +	_scratch_unmount
> +}
> +
> +DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
> +DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
> +
> +check_copy1_fsid
> +check_primary
> +check_copy1
> +
> +echo -e "\\nReverse good and bad devices"
> +# For csum verification on devid 2, reverse the good and bad disks.
> +DEV_BAD=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
> +DEV_GOOD=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
> +check_primary
> +check_copy1
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/btrfs/159.out b/tests/btrfs/159.out
> new file mode 100644
> index 000000000000..3e7ba31929e6
> --- /dev/null
> +++ b/tests/btrfs/159.out
> @@ -0,0 +1,23 @@
> +QA output created by 159
> +
> +check_copy1_fsid
> +
> +mount: wrong fs type, bad option, bad superblock on SCRATCH_DEV,
> +       missing codepage or helper program, or other error
> +
> +       In some cases useful info is found in syslog - try
> +       dmesg | tail or so.
> +
> +check_primary
> +
> +mount: Structure needs cleaning
> +
> +check_copy1
> +
> +Reverse good and bad devices
> +
> +check_primary
> +
> +mount: Structure needs cleaning
> +
> +check_copy1
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index 8007e07e9cbd..ba766f6b84f8 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -161,3 +161,4 @@
>  156 auto quick trim
>  157 auto quick raid
>  158 auto quick raid scrub
> +159 auto quick
> -- 
> 2.7.0
> 
> --
> 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

Re: [PATCH v2] fstests: btrfs/159 superblock corruption test case

[Anand Jain]

>> +mount_opt_minus_args()
>> +{
>> +	local nr
>> +	local mnt_opt=""
>> +
>> +	nr=`_scratch_mount_options | awk '{print NF}'`
>> +	if [ $nr -eq 4 ]; then
> 
> Seems this only works with "scratch_mount_options" returns something
> like:
> 
> "-o opt dev mnt" or "-o opt1,opt2 dev mnt"
> 
> but not
> 
> "-oopt dev mnt" nor
> "-o opt1 -o opt2 dev mnt" nor if $SELINUX_MOUNT_OPTIONS not empty.
> 
> Also if MOUNT_OPTIONS is "-oopt1 -oopt2", mount_opt_minus_args would
> return something like "-o -oopt2,device=<dev>", which are not valid
> mount options.

  Oh. Yes. This function is getting complicated. I have dropped this
  function in v3 which is kind of much better.

>> +# Test if the mount fails if the primary superblock csum is corrupted.
>> +check_primary()
>> +{
>> +	local bytenr=65536
>> +	echo -e "\\ncheck_primary\\n" | tee -a $seqres.full
>> +
>> +	wipe
>> +	_scratch_pool_mkfs "-mraid1 -draid1"
>> +	_reload_fs_module "btrfs"
>> +
>> +	#corrupt bytenr DEV_BAD
>> +	od -j$bytenr -N1 -An -x $DEV_BAD |\
>> +		dd status=none of=$DEV_BAD ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
>> +
>> +	#must fail
>> +	_mount `mount_opt_minus_args device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT 2>&1 |\
>> +	      _filter_error_mount
> 
> If I read the code correctly, the purpose of mount_opt_minus_args is to
> remove the device and mount point mount arguments and keep other options
> if there's any, then append "-o device=<dev>" as mount options.
> 
> If so, I think you could use "_common_dev_mount_options", e.g.
> 
> 	_mount `_common_dev_mount_options -o device=$DEV_GOOD` $DEV_BAD $SCRATCH_MNT ...
> 
> But do mount options really matter in this test? Could we just mount
> with "-o device=<dev>"?

  In fact for this test it doesn't matter, yes we could just use
  "-o device=<dev>". But I thought its xfstests rule that it must be
  used. Anyway in V3, I am changing the device itself on which
  corrupted-SB would reside, so now we can let the default _scratch_mount
  do its job as usual. Which means it shall use the config mount options.
  V3 is in the ML for the review.

Thanks, Anand

> Thanks,
> Eryu

[PATCH v3] fstests: btrfs/159 superblock corruption test case

[Anand Jain]

Verify if the superblock corruption is handled correctly.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
v2->v3:
 Provide the disk to be corrupted as an arg, instead of swapping the devices,
  so drop mount_opt_minus_args().
 159.out slightly changed.
v1->v2:
 $subject slightly changed
 Added more info about the test-case
 Keep the stuff from the ./new btrfs
 Add mount_opt_minus_args() to get the options (if) set at the config file
 Move DEV_GOOD & DEV_BAD to where it starts to use
 To help debugging added run_check where possible
 Remove {} in the out file
 Use _filter_error_mount for mount fail cases other than -EINVAL

 tests/btrfs/159     | 149 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/btrfs/159.out |  21 ++++++++
 tests/btrfs/group   |   1 +
 3 files changed, 171 insertions(+)
 create mode 100755 tests/btrfs/159
 create mode 100644 tests/btrfs/159.out

diff --git a/tests/btrfs/159 b/tests/btrfs/159
new file mode 100755
index 000000000000..c3a50b58b0b9
--- /dev/null
+++ b/tests/btrfs/159
@@ -0,0 +1,149 @@
+#! /bin/bash
+# FS QA Test 159
+#
+# Test if the superblock corruption is handled correctly:
+# 	- Test fsid miss-match (csum ok) between primary and copy superblock
+#	Fixed by the ML patch:
+#	btrfs: check if the fsid in the primary sb and copy sb are same
+# 	- Test if the mount fails if the primary superblock csum is
+#		corrupted on any disk
+# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
+#	Fixed by the ML patches:
+#	btrfs: verify superblock checksum during scan
+#	btrfs: verify checksum for all devices in mount context
+#
+#---------------------------------------------------------------------
+# Copyright (c) 2018 Oracle.  All Rights Reserved.
+# Author: Anand Jain <anand.jain@oracle.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"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+	_scratch_dev_pool_put
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/module
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+
+_supported_fs btrfs
+_supported_os Linux
+_require_scratch_dev_pool 2
+_require_loadable_fs_module "btrfs"
+_require_command "$WIPEFS_PROG" wipefs
+
+_scratch_dev_pool_get 2
+
+wipe()
+{
+	$WIPEFS_PROG -a $SCRATCH_DEV_POOL > /dev/null 2>&1
+}
+
+# Test for fsid miss-match (csum ok) with primary and copy superblock.
+check_copy1_fsid()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1_fsid\\n" | tee -a $seqres.full
+
+	wipe
+	$MKFS_BTRFS_PROG -fq $EXTRA_DEV
+	_scratch_mkfs >> $seqres.full 2>&1
+	_reload_fs_module "btrfs"
+
+	run_check dd status=none of=$SCRATCH_DEV if=$EXTRA_DEV ibs=1 obs=1\
+			skip=$bytenr seek=$bytenr count=4K
+	$WIPEFS_PROG -a $EXTRA_DEV > /dev/null 2>&1
+
+	#must fail
+	_try_scratch_mount 2>&1 | _filter_scratch_pool
+	#If mount didn't fail
+	_scratch_unmount > /dev/null 2>&1
+}
+
+# Test if the mount fails if the primary superblock csum is corrupted.
+# arg 1 is dev to corrupt
+check_primary()
+{
+	local bytenr=65536
+	echo -e "\\ncheck_primary\\n" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#corrupt primary superblock bytenr
+	od -j$bytenr -N1 -An -x $1 |\
+		dd status=none of=$1 ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
+
+	#must fail
+	_try_scratch_mount "-o device=$EXTRA_DEV" 2>&1 | _filter_error_mount
+
+	#If mount didn't fail
+	_scratch_unmount > /dev/null 2>&1
+}
+
+# If copy1 is corrupted we still should be able to mount, check that.
+# arg 1 is dev to corrupt
+check_copy1()
+{
+	local bytenr=67108864
+	echo -e "\\ncheck_copy1" | tee -a $seqres.full
+
+	wipe
+	_scratch_pool_mkfs "-mraid1 -draid1"
+	_reload_fs_module "btrfs"
+
+	#corrupt copy1 superrblock bytenr
+	od -j$bytenr -N1 -An -x $1 |\
+		dd status=none of=$1 ibs=1 obs=1 seek=$bytenr count=1 conv=fsync
+
+	_try_scratch_mount "-o device=$EXTRA_DEV" 2>&1 | _filter_error_mount
+	_scratch_unmount
+}
+
+EXTRA_DEV=$(echo $SCRATCH_DEV_POOL | awk '{print $1}')
+if [ $EXTRA_DEV == $SCRATCH_DEV ]; then
+	EXTRA_DEV=$(echo $SCRATCH_DEV_POOL | awk '{print $2}')
+fi
+echo SCRATCH_DEV=$SCRATCH_DEV EXTRA_DEV=$EXTRA_DEV >> $seqres.full
+
+check_copy1_fsid
+check_primary $SCRATCH_DEV
+check_primary $EXTRA_DEV
+check_copy1 $SCRATCH_DEV
+check_copy1 $EXTRA_DEV
+
+# success, all done
+status=0
+exit
diff --git a/tests/btrfs/159.out b/tests/btrfs/159.out
new file mode 100644
index 000000000000..9daf9f515e87
--- /dev/null
+++ b/tests/btrfs/159.out
@@ -0,0 +1,21 @@
+QA output created by 159
+
+check_copy1_fsid
+
+mount: wrong fs type, bad option, bad superblock on SCRATCH_DEV,
+       missing codepage or helper program, or other error
+
+       In some cases useful info is found in syslog - try
+       dmesg | tail or so.
+
+check_primary
+
+mount: Structure needs cleaning
+
+check_primary
+
+mount: Structure needs cleaning
+
+check_copy1
+
+check_copy1
diff --git a/tests/btrfs/group b/tests/btrfs/group
index 8007e07e9cbd..ba766f6b84f8 100644
--- a/tests/btrfs/group
+++ b/tests/btrfs/group
@@ -161,3 +161,4 @@
 156 auto quick trim
 157 auto quick raid
 158 auto quick raid scrub
+159 auto quick
-- 
2.7.0

Re: [PATCH v3] fstests: btrfs/159 superblock corruption test case

[Eryu Guan]

On Mon, Apr 09, 2018 at 01:28:30PM +0800, Anand Jain wrote:
> Verify if the superblock corruption is handled correctly.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
> v2->v3:
>  Provide the disk to be corrupted as an arg, instead of swapping the devices,
>   so drop mount_opt_minus_args().
>  159.out slightly changed.
> v1->v2:
>  $subject slightly changed
>  Added more info about the test-case
>  Keep the stuff from the ./new btrfs
>  Add mount_opt_minus_args() to get the options (if) set at the config file
>  Move DEV_GOOD & DEV_BAD to where it starts to use
>  To help debugging added run_check where possible
>  Remove {} in the out file
>  Use _filter_error_mount for mount fail cases other than -EINVAL
> 
>  tests/btrfs/159     | 149 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/159.out |  21 ++++++++
>  tests/btrfs/group   |   1 +
>  3 files changed, 171 insertions(+)
>  create mode 100755 tests/btrfs/159
>  create mode 100644 tests/btrfs/159.out
> 
> diff --git a/tests/btrfs/159 b/tests/btrfs/159
> new file mode 100755
> index 000000000000..c3a50b58b0b9
> --- /dev/null
> +++ b/tests/btrfs/159
> @@ -0,0 +1,149 @@
> +#! /bin/bash
> +# FS QA Test 159
> +#
> +# Test if the superblock corruption is handled correctly:
> +# 	- Test fsid miss-match (csum ok) between primary and copy superblock
> +#	Fixed by the ML patch:
> +#	btrfs: check if the fsid in the primary sb and copy sb are same
> +# 	- Test if the mount fails if the primary superblock csum is
> +#		corrupted on any disk
> +# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
> +#	Fixed by the ML patches:
> +#	btrfs: verify superblock checksum during scan
> +#	btrfs: verify checksum for all devices in mount context

Do you have a tree that I can pull from? I want to make sure the test
does pass on patched kernel, but the patchset doesn't apply on v4.16
kernel.

Thanks,
Eryu

Re: [PATCH v3] fstests: btrfs/159 superblock corruption test case

[Anand Jain]

>> +# Test if the superblock corruption is handled correctly:
>> +# 	- Test fsid miss-match (csum ok) between primary and copy superblock
>> +#	Fixed by the ML patch:
>> +#	btrfs: check if the fsid in the primary sb and copy sb are same
>> +# 	- Test if the mount fails if the primary superblock csum is
>> +#		corrupted on any disk
>> +# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
>> +#	Fixed by the ML patches:
>> +#	btrfs: verify superblock checksum during scan
>> +#	btrfs: verify checksum for all devices in mount context
> 
> Do you have a tree that I can pull from? I want to make sure the test
> does pass on patched kernel, but the patchset doesn't apply on v4.16
> kernel.

We have new discussions on whether to check for the alien-superblock and
the superblock-checksum at the mount and scan time respectively. And
depending on its outcome this test-case should be modified as well. So
can you please defer this fstest patch, for now, I shall send a revised
fstest patch when kernel patches gets integrated.

In any case, if you want to give a try, those patches are base on kdave 
repo at [1].
  [1]
  https://github.com/kdave/btrfs-devel.git misc-next

Thanks, Anand

> Thanks,
> Eryu
> --
> 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
> 

Re: [PATCH v3] fstests: btrfs/159 superblock corruption test case

[Eryu Guan]

On Sat, Apr 14, 2018 at 06:43:49AM +0800, Anand Jain wrote:
> 
> > > +# Test if the superblock corruption is handled correctly:
> > > +# 	- Test fsid miss-match (csum ok) between primary and copy superblock
> > > +#	Fixed by the ML patch:
> > > +#	btrfs: check if the fsid in the primary sb and copy sb are same
> > > +# 	- Test if the mount fails if the primary superblock csum is
> > > +#		corrupted on any disk
> > > +# 	- Test if the mount does not fail if the copy1 sb csum is corrupted
> > > +#	Fixed by the ML patches:
> > > +#	btrfs: verify superblock checksum during scan
> > > +#	btrfs: verify checksum for all devices in mount context
> > 
> > Do you have a tree that I can pull from? I want to make sure the test
> > does pass on patched kernel, but the patchset doesn't apply on v4.16
> > kernel.
> 
> We have new discussions on whether to check for the alien-superblock and
> the superblock-checksum at the mount and scan time respectively. And
> depending on its outcome this test-case should be modified as well. So
> can you please defer this fstest patch, for now, I shall send a revised
> fstest patch when kernel patches gets integrated.

Thanks for the heads-up, I'll drop it for now.

> 
> In any case, if you want to give a try, those patches are base on kdave repo
> at [1].
>  [1]
>  https://github.com/kdave/btrfs-devel.git misc-next

Thanks!

Eryu