tests/generic/459

#! /bin/bash
# FS QA Test 459
#
# Test buffer filesystem error recovery during a full overcommited dm-thin device.
#
# When a dm-thin device reaches its full capacity, but the virtual device still
# shows available space, the filesystem should be able to handle such cases
# failing its operation without locking up.
#
# This test has been created first to cover a XFS problem where it loops
# indefinitely in xfsaild due items still in AIL. The buffers containing such
# items couldn't be resubmitted because the items were flush locked.
# But, once this doesn't require any special filesystem feature to be executed,
# this has been integrated as a generic test.
#
# This test might hang the filesystem when ran on an unpatched kernel
#
#-----------------------------------------------------------------------
# Copyright (c) 2017 Red Hat, Inc. 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
#-----------------------------------------------------------------------
#

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.*
	$UMOUNT_PROG $SCRATCH_MNT >>$seqres.full 2>&1
	$LVM_PROG vgremove -ff $vgname >>$seqres.full 2>&1
	$LVM_PROG pvremove -ff $SCRATCH_DEV >>$seqres.full 2>&1
}

# get standard environment, filters and checks
. ./common/rc

# real QA test starts here

# This tests for filesystem lockup not consistency, so don't check for fs
# consistency after test
_supported_fs generic
_supported_os Linux
_require_scratch_nocheck
_require_dm_target thin-pool
_require_dm_target snapshot
_require_command $LVM_PROG lvm
_require_command "$THIN_CHECK_PROG" thin_check
_require_freeze

# remove previous $seqres.full before test
rm -f $seqres.full

vgname=vg_$seq
lvname=lv_$seq
poolname=pool_$seq
snapname=snap_$seq
origpsize=100
virtsize=200
newpsize=200

# Ensure we have enough disk space
_scratch_mkfs_sized $((250 * 1024 * 1024)) >>$seqres.full 2>&1

# Create a 100MB dm-thin POOL
$LVM_PROG pvcreate -f $SCRATCH_DEV >>$seqres.full 2>&1
$LVM_PROG vgcreate -f $vgname $SCRATCH_DEV >>$seqres.full 2>&1

$LVM_PROG lvcreate  --thinpool $poolname  --errorwhenfull y \
		    --zero n -L $origpsize \
		    --poolmetadatasize 4M $vgname >>$seqres.full 2>&1

# Create a overprovisioned 200MB dm-thin virt. device
$LVM_PROG lvcreate  --virtualsize $virtsize \
		    -T $vgname/$poolname \
		    -n $lvname >>$seqres.full 2>&1

_mkfs_dev /dev/mapper/$vgname-$lvname >>$seqres.full 2>&1


# Running the test over the original volume doesn't reproduce the problem
# reliably, while, running it over a snapshot, makes the problem 100%
# reproducible, so, create a snapshot and run the test over it.
$LVM_PROG lvcreate  -k n -s $vgname/$lvname \
		    -n $snapname >>$seqres.full 2>&1

_mount /dev/mapper/$vgname-$snapname $SCRATCH_MNT

# Consume all space available in the volume and freeze to ensure everything
# required to make the fs consistent is flushed to disk.
$XFS_IO_PROG -f -d -c 'pwrite -b 1m 0 120m' $SCRATCH_MNT/f1 >>$seqres.full 2>&1

# In XFS, this freeze will never complete until the dm-thin POOL device is
# extended. It is expected, and is only used so xfsaild is triggered to
# flush AIL items, other filesystems usually get remounted as read-only during
# the above write process.
xfs_freeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
freezeid=$!

# Wait enough so xfsaild can run
sleep 10

# Make some extra space available so the freeze above can proceed
$LVM_PROG lvextend -L $newpsize $vgname/$poolname >>$seqres.full 2>&1

wait $freezeid
ret=$?

# Different filesystems will handle the lack of real space in different ways,
# some will remount the filesystem in read-only mode, some will not. These tests
# will check if:
#	- The filesystem turns into Read-Only and reject further writes
#	- The filesystem stays in Read-Write mode, but can be frozen/thawed
#	  without getting stuck.
if [ $ret -ne 0 ]; then
	# freeze failed, filesystem should reject further writes and remount
	# as readonly. Sometimes the previous write process won't trigger
	# ro-remount, e.g. on ext3/4, do additional touch here to make sure
	# filesystems see the metadata I/O error.
	touch $SCRATCH_MNT/newfile >/dev/null 2>&1
	ISRO=$(_fs_options /dev/mapper/$vgname-$snapname | grep -w "ro")
	if [ -n "$ISRO" ]; then
		echo "Test OK"
	else
		echo "Freeze failed and FS isn't Read-Only. Test Failed"
		status=1
		exit
	fi
else
	# Try to thaw the filesystem, and complete test if if succeed.
	# NOTE: This will hang on affected XFS filesystems.
	xfs_freeze -u $SCRATCH_MNT >>$seqres.full 2>&1
	echo "Test OK"
fi

status=0
exit