On Sat, Jul 21, 2012 at 11:46:00AM +0800, Liu Bo wrote:
> From: Zhou Bo <zhoub-fnst@xxxxxxxxxxxxxx>
>
> This patch adds btrfs snapshot function test to xfstests.
>
> Signed-off-by: Zhou Bo <zhoub-fnst@xxxxxxxxxxxxxx>
> ---
> 285 | 365 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 285.out | 2 +
> group | 1 +
> 3 files changed, 368 insertions(+), 0 deletions(-)
> create mode 100755 285
> create mode 100644 285.out
>
> diff --git a/285 b/285
> new file mode 100755
> index 0000000..d247af3
> --- /dev/null
> +++ b/285
> @@ -0,0 +1,365 @@
> +#! /bin/bash
> +# FS QA Test No. 285
> +#
> +# Test btrfs's subvolume and snapshot function
There already is one subvolume/snapshot test which is simple and basic.
The new one is much more extensive and this needs a more verbose
description
> +#
> +#-----------------------------------------------------------------------
> +# Copyright (c) 2012 Fujitsu. All Rights Reserved.
> +#
> +# 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
> +#
> +#-----------------------------------------------------------------------
> +#
> +# creator
> +owner=zhoub-fnst@xxxxxxxxxxxxxx
> +
> +n=0
> +seq=`basename $0`
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=0 # success is the default!
> +
> +_cleanup()
> +{
> + rm -f $tmp.*
> +}
> +
> +trap "_cleanup ; exit \$status" 0 1 2 3 15
> +
> +# get standard environment, filters and checks
> +. ./common.rc
> +. ./common.filter
> +
> +# real QA test starts here
> +_supported_fs btrfs
> +_supported_os Linux
> +_require_scratch
> +
> +_scratch_mkfs_sized `expr 1024 \* 1024 \* 1024` > /dev/null 2>&1
Just curious, is there a reason why you create a 1G filesystem? This
would imply a --mixed type of fs.
> +_scratch_mount
> +
> +_prepare_snapshot()
> +{
> + _scratch_remount > /dev/null
> + btrfs sub snap $SCRATCH_MNT $SCRATCH_MNT/basesnapshot > /dev/null 2>>$here/$seq.full
> + btrfs sub snap -r $SCRATCH_MNT $SCRATCH_MNT/readonlysnapshot > /dev/null 2>>$here/$seq.full
here and below: please type full subcommands ie.
btrfs subvolume snapshot
although the short ones are allowed, using full names is future-proof
when there could be another subcommand with the same short prefix.
> + _scratch_unmount > /dev/null 2>>$here/$seq.full
> + VALID_SUBVOLUME="basesnapshot"
> + VALID_RO_SUBVOLUME="readonlysnapshot"
> + SNAPSHOTSTR="snapshot"
> + FILE1="file1-"
> + FILE2="file2-"
> + MVFILE2="newfile2-"
> + DIR1="dir1-"
> + DIR2="dir2-"
> + MVDIR2="newdir2-"
> + MVSNAPSHOT="mvsnapshot-"
> + SRCSUBVOL="srcsubvol-"
> +}
> +
> +_parse_options()
> +{
> + SOURCE_TARGET="$1"
> + case $SOURCE_TARGET in
> + "1")
> + SOURCE_SUBVOLUME="$VALID_SUBVOLUME"
> + ;;
> + esac
> + SOURCE_READ="$2"
> + case $SOURCE_READ in
> + "1")
> + SOURCE_SUBVOLUME="$VALID_RO_SUBVOLUME"
> + ;;
> + esac
> + DESTINATION_TARGET="$3"
> + case $DESTINATION_TARGET in
> + "1")
> + DESTINATION_SUBVOLUME=$SNAPSHOTSTR$n
> + ;;
> + esac
> + DESTINATION_READ="$4"
> + case $DESTINATION_READ in
> + "1")
> + SNAPSHOTOPT_STR="-r"
not that it matters much, SNAPSHOT_OPT_STR would look more consistent
with other variable names, like MOUNT_OPT_STR
> + ;;
> + "2")
> + SNAPSHOTOPT_STR=""
> + ;;
> + esac
> + MOUNT_OPT="$5"
> + case $MOUNT_OPT in
> + "1")
> + MOUNT_OPT_STR=""
> + ;;
> + "2")
> + MOUNT_OPT_STR="-r"
> + ;;
> + "3")
> + MOUNT_OPT_STR="-o nodatacow"
> + ;;
> + esac
> + FILE_OPERATION_OPT="$6"
> + SNAPSHOT_ACTION_OPT="$7"
> + TEST_DIR1=$DIR1$n
> + TEST_DIR2=$DIR2$n
> + TEST_MVDIR2=$MVDIR2$n
> + TEST_FILE1=$FILE1$n
> + TEST_FILE2=$FILE2$n
> + TEST_MVFILE2=$MVFILE2$n
> + TEST_MVSNAPSHOT=$MVSNAPSHOT$n
> + SRC_SUBVOLUME=$SRCSUBVOL$n
> + n=$[n+1]
> +}
> +
> +_create_file()
> +{
> + mkdir $SRC_SUBVOLUME/$TEST_DIR1 $SRC_SUBVOLUME/$TEST_DIR2 > /dev/null
> + touch $SRC_SUBVOLUME/$TEST_FILE1 $SRC_SUBVOLUME/$TEST_FILE2 > /dev/null
> +}
> +
> +_do_file_operation()
> +{
> + btrfs filesystem balance $SCRATCH_MNT > /dev/null 2>&1 &
although 'btrfs filesystem balance /mnt' works, please use the newer
syntax 'btrfs filesystem balance start /mnt'
I find the function name misleading, why is balance called here? if you
really need to rebalance (both data and metadata?), put it into a
separate function and call when appropriate.
> + rm -rf $SRC_SUBVOLUME/$TEST_DIR1 $SRC_SUBVOLUME/$TEST_FILE1 > /dev/null
> + mv $SRC_SUBVOLUME/$TEST_DIR2 $SRC_SUBVOLUME/$TEST_MVDIR2 > /dev/null
> + mv $SRC_SUBVOLUME/$TEST_FILE2 $SRC_SUBVOLUME/$TEST_MVFILE2 > /dev/null
> +}
> +
> +_do_snapshot_action()
> +{
> + if [ "$SNAPSHOT_ACTION_OPT" == 2 ];then
> + btrfs subvolume delete $DESTINATION_SUBVOLUME > /dev/null 2>>$here/$seq.full
> + fi
> + if [ "$SNAPSHOT_ACTION_OPT" == 3 ];then
> + mv $DESTINATION_SUBVOLUME $TEST_MVSNAPSHOT > /dev/null 2>>$here/$seq.full
> + fi
> +}
> +
> +_check_snapshot()
> +{
> + if [ "$SNAPSHOT_ACTION_OPT" == 2 ];then
> + if [ -d "$DESTINATION_SUBVOLUME" ];then
> + echo "case $n fails, deleting snapshot fails." >> $here/$seq.full
> + status=1
> + fi
> + fi
> + if [ "$SNAPSHOT_ACTION_OPT" == 3 ];then
> + if [ ! -d "$TEST_MVSNAPSHOT" ];then
> + echo "case $n fails, renaming snapshot fails." >> $here/$seq.full
> + status=1
> + fi
> + fi
> +
> +}
> +
> +_check_file()
> +{
> + cd $DESTINATION_SUBVOLUME
> + if [ "$FILE_OPERATION_OPT" == 2 ];then
> + if [ -d "$TEST_DIR1" ];then
> + echo "case $n fails, before snapshot we delete dir in src, but it exists in snap" >> $here/$seq.full
> + status=1
> + fi
> + if [ -f "$TEST_FILE1" ];then
> + echo "case $n fails, before snapshot we delete file in src, but it exists in snap" >> $here/$seq.full
> + status=1
> + fi
> + if [ -d "$TEST_DIR2" -o ! -d "$TEST_MVDIR2" ];then
> + echo "case $n fails, before snapshot we rename dir in src, but it remains in snap" >> $here/$seq.full
> + status=1
> +
> + fi
> + if [ -f "$TEST_FILE2" -o ! -f "$TEST_MVFILE2" ];then
> + echo "case $n fails, before snapshot we rename file in src,but it remains in snap" >> $here/$seq.full
> + status=1
> + fi
> +
> + else
> + if [ ! -d "$TEST_DIR1" ];then
> + echo "case $n fails, after snapshot we delete dir in src, but it doesn't exist in snap" >> $here/$seq.full
> + status=1
> + fi
> + if [ ! -f "$TEST_FILE1" ];then
> + echo "case $n fails, after snapshot we delete file in src, but it doesn't exist in snap" >> $here/$seq.full
> + status=1
> +
> + fi
> + if [ ! -d "$TEST_DIR2" -o -d "$TEST_MVDIR2" ];then
> + echo "case $n fails, after snapshot we rename dir in src, but it also changes in snap" >> $here/$seq.full
> + status=1
> + fi
> + if [ ! -f "$TEST_FILE2" -o -f "$TEST_MVFILE2" ];then
> + echo "case $n fails, after snapshot we rename file in src, but it also changes in snap" >> $here/$seq.full
> + status=1
> + fi
> + fi
> + btrfs filesystem balance cancel $SCRATCH_MNT > /dev/null 2>&1
> + wait
balance cancel will block until it succeeds, what's the purpose of
'wait' here ? I didn't see any background process started.
> + cd ..
are you always sure that this will succeed? this is an unexpected
side-effect of the function, none of the _check_* does that.
> +}
> +
> +_test_snapshot_ro()
> +{
> + btrfs sub snap $SNAPSHOTOPT_STR $SOURCE_SUBVOLUME $DESTINATION_SUBVOLUME > /dev/null 2>&1
> + if [ "$?" == 0 ];then
> + if [ "$MOUNT_OPT" != 1 ];then
> + echo "case $n fails, btrfs snapshot fails." >> $here/$seq.full
> + status=1
> + else
> + if [ "$DESTINATION_READ" == 1 ];then
> + if [ -w "$DESTINATION_SUBVOLUME" ];then
> + echo "case $n fails, snapshot should be readonly." >> $here/$seq.full
> + status=1
> + fi
> + else
> + if [ ! -w "$DESTINATION_SUBVOLUME" ];then
> + echo "case $n fails, snapshot should be writable." >> $here/$seq.full
> + status=1
> + fi
> + fi
> + fi
> +
unnecessary newline
> + fi
> +}
> +
> +_test_snapshot_operation()
> +{
> + btrfs sub create $SRC_SUBVOLUME > /dev/null 2>>$here/$seq.full
> + _create_file
> + if [ "$FILE_OPERATION_OPT" == 2 ];then
> + _do_file_operation
> + fi
> + btrfs sub snap $SRC_SUBVOLUME $DESTINATION_SUBVOLUME > /dev/null 2>>$here/$seq.full
> + if [ "$?" == 0 ];then
> + if [ "$FILE_OPERATION_OPT" == 3 ];then
> + _do_file_operation
> + fi
> + if [ "$FILE_OPERATION_OPT" -eq "2" -o "$FILE_OPERATION_OPT" -eq "3" ];then
> + _check_file
> + fi
> + _do_snapshot_action
> + _check_snapshot
> + else
> + status=1
> + echo "case $n fails, btrfs snapshot fails." >> $here/$seq.full
> + fi
> +}
> +
> +_test_process()
> +{
> + _scratch_mount "$MOUNT_OPT_STR"
> + cd $SCRATCH_MNT
> + if [ "$FILE_OPERATION_OPT" == 0 ];then
> + _test_snapshot_ro
> + else
> + _test_snapshot_operation
> + fi
> + cd ~
what's the purpose? no other xfstest does this so it's not a common
thing AFAICT. I think you can safely cd into $TEST_MNT if you realy need
to
> + _scratch_unmount
> + _check_test_fs
> + if [ "$?" != 0 ];then
> + echo "case $n fails, btrfsck fails." >> $here/$seq.full
echo ".............. fsck ........."
> + status=1
> + fi
> +}
> +
> +rm -f $here/$seq.full
> +_prepare_snapshot
> +
> +# Src subvol | Src readonly | Des subvol | Des Readonly | Mount opt | File operation | Snap operation
> +
> +# case 1
> +# Valid Src | RO | Valid Des | RO | Default | Nothing | Create
> +_parse_options 1 1 1 1 1 0 0
> +_test_process
_parse_options and _test_process below are always called in tandem,
could be better to call just one function if there isn't a special
reason to call them separately here. (but of course it makes sense to
have the functionality separated)
I'd like to see some strings instead of a bunch of numbers. this also
ensures that the comment above the test case does not lie. there are too
many of them and it seems to easy to make a typo/mistake.
the comment itself could be higher-level, something like 'test snapshots
when we delete the original under -o nocow'.
let's see how it would appear:
_test valid_src ro valid_dest default nothing create
IMHO more clear to a random test reader. also this way ordering of the
option does not matter, less error prone and somehow easier to extend,
eg. when we woud like to add more mount options to test like 'nodatasum'.
> +
> +# case 2
> +# Valid Src | RO | Valid Des | Writable | Default | Nothing | Create
> +_parse_options 1 1 1 2 1 0 0
> +_test_process
> +
> +# case 3
> +# Valid Src | Writable | Valid Des | RO | Default | Nothing | Create
> +_parse_options 1 2 1 1 1 0 0
> +_test_process
> +
> +# case 4
> +# Valid Src | Writable | Valid Des | Writable | Default | Nothing | Create
> +_parse_options 1 2 1 2 1 0 0
> +_test_process
> +
> +# case 5
> +# Valid Src | RO | Valid Des | RO | -r | Nothing | Create
> +_parse_options 1 1 1 1 2 0 0
> +_test_process
> +
> +# case 6
> +# Valid Src | RO | Valid Des | Writable | -r | Nothing | Create
> +_parse_options 1 1 1 2 2 0 0
> +_test_process
> +
> +# case 7
> +# Valid Src | Writable | Valid Des | RO | -r | Nothing | Create
> +_parse_options 1 2 1 1 2 0 0
> +_test_process
> +
> +# case 8
> +# Valid Src | Writable | Valid Des | Writable | -r | Nothing | Create
> +_parse_options 1 2 1 2 2 0 0
> +_test_process
> +
> +# case 9
> +# Valid Src | Writable | Valid Des | Writable | Default | Before snap | Delete
> +_parse_options 1 2 1 2 1 2 2
> +_test_process
> +
> +# case 10
> +# Valid Src | Writable | Valid Des | Writable | Default | Before snap | Rename
> +_parse_options 1 2 1 2 1 2 3
> +_test_process
> +
> +# case 11
> +# Valid Src | Writable | Valid Des | Writable | Default | After snap | Delete
> +_parse_options 1 2 1 2 1 3 2
> +_test_process
> +
> +# case 12
> +# Valid Src | Writable | Valid Des | Writable | Default | After snap | Rename
> +_parse_options 1 2 1 2 1 3 3
> +_test_process
> +
> +# case 13
> +# Valid Src | Writable | Valid Des | Writable | no datacow | Before snap | Delete
> +_parse_options 1 2 1 2 3 2 2
> +_test_process
> +
> +# case 14
> +# Valid Src | Writable | Valid Des | Writable | no datacow | Before snap | Rename
> +_parse_options 1 2 1 2 3 2 3
> +_test_process
> +
> +# case 15
> +# Valid Src | Writable | Valid Des | Writable | no datacow | After snap | Delete
> +_parse_options 1 2 1 2 3 3 2
> +_test_process
> +
> +# case 16
> +# Valid Src | Writable | Valid Des | Writable | no datacow | After snap | Rename
> +_parse_options 1 2 1 2 3 3 3
> +_test_process
> +
> +echo "Silence is golden"
> +exit
> diff --git a/285.out b/285.out
> new file mode 100644
> index 0000000..ab12da9
> --- /dev/null
> +++ b/285.out
> @@ -0,0 +1,2 @@
> +QA output created by 285
> +Silence is golden
> diff --git a/group b/group
> index 104ed35..0b33178 100644
> --- a/group
> +++ b/group
> @@ -403,3 +403,4 @@ deprecated
> 282 dump ioctl auto quick
> 283 dump ioctl auto quick
> 284 auto
> +285 auto
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
