Skip to contentSkip to conversation

ext4: Test for resize overflow

PATCH In review Jan Kara 2 participants 8 messages May 29, 2018

lore
Share & tools
lore

Jan Kara

author
#0 lore raw

cover ext4: Test for resize overflow

Hello,

this series add a test for ext4 resize oveflow bug. It also contains a few other cleanups that Eryu wanted.

Honza

author
#1 lore raw

PATCH 1/6 dmhugedisk: Allow specifying of chunk size

Ext4 will want to use dmhugedisk infrastructure for testing resize bugs. Ext4 fs images are rather sparse (especially with smaller block sizes) so the current chunk size of 512 sectors leads to large space consumption. Allow test to specify chunk size.

---
 common/dmhugedisk | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
common/dmhugedisk +6 -1
index 4d3b63fcfcca..733b10cd2292 100644
@@ -31,6 +31,11 @@ _dmhugedisk_init()
 {
 	test -z "$1" && _fatal "must specify sector count to _dmhugedisk_init"
 	local dm_backing_dev=$SCRATCH_DEV
+	local chunk_size="$2"
+
+	if [ -z "$chunk_size" ]; then
+		chunk_size=512
+	fi
 
 	$DMSETUP_PROG remove huge-test > /dev/null 2>&1
 	$DMSETUP_PROG remove huge-test-zero > /dev/null 2>&1
@@ -41,7 +46,7 @@ _dmhugedisk_init()
 	DMHUGEDISK_DEV='/dev/mapper/huge-test'
 
 	DMHUGEDISK_ZERO_TABLE="0 $blk_dev_size zero"
-	DMHUGEDISK_DEV_TABLE="0 $blk_dev_size snapshot $DMHUGEDISK_ZERO $SCRATCH_DEV N 512"
+	DMHUGEDISK_DEV_TABLE="0 $blk_dev_size snapshot $DMHUGEDISK_ZERO $SCRATCH_DEV N $chunk_size"
 
 	$DMSETUP_PROG create huge-test-zero --table "$DMHUGEDISK_ZERO_TABLE" || \
 		_fatal "failed to create dm huge zero device"
Signature · 1 line
2.13.6
author
#2 lore raw

PATCH 2/6 ext4: Test for s_inodes_count overflow during fs resize

In reply to Jan Kara

Test for overflow of s_inodes_count during filesystem resizing.

---
 common/config      |   1 +
 tests/ext4/033     | 123 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/ext4/033.out |   6 +++
 tests/ext4/group   |   1 +
 4 files changed, 131 insertions(+)
 create mode 100755 tests/ext4/033
 create mode 100644 tests/ext4/033.out
common/config +1 -0
index 51380231aed3..6efc8f9fc4a5 100644
@@ -173,6 +173,7 @@ export INDENT_PROG="$(type -P indent)"
 export XFS_COPY_PROG="$(type -P xfs_copy)"
 export FSTRIM_PROG="$(type -P fstrim)"
 export DUMPE2FS_PROG="$(type -P dumpe2fs)"
+export RESIZE2FS_PROG="$(type -P resize2fs)"
 export FIO_PROG="$(type -P fio)"
 export FILEFRAG_PROG="$(type -P filefrag)"
 export E4DEFRAG_PROG="$(type -P e4defrag)"
tests/ext4/033 new file +123 -0

123 lines not on this page · raw message

tests/ext4/033.out new file +6 -0
new file mode 100644
index 000000000000..24c251cbb2b3
@@ -0,0 +1,6 @@
+QA output created by 033
+Figure out block size
+Format huge device
+Resizing to inode limit + 1...
+Resizing to max group count...
+Resizing to device size...
tests/ext4/group +1 -0
index 5bd15f82b3be..b850f568e674 100644
@@ -35,6 +35,7 @@
 030 auto quick dax
 031 auto quick dax
 032 auto quick ioctl resize
+033 auto ioctl resize
 271 auto rw quick
 301 aio auto ioctl rw stress defrag
 302 aio auto ioctl rw stress defrag
Signature · 1 line
2.13.6
author
#3 lore raw

PATCH 3/6 xfs/310: Cleanup

In reply to Jan Kara

Cleanup couple of things that were objected to when creating test ext4/033 out of this one.

Use _require_scratch_nocheck instead of recreating scratch fs before exiting. Avoid needless cleanup of dmhugedev on exit - _cleanup takes care of that. Use _scratch_unmount where possible.

---
 tests/xfs/310 | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
tests/xfs/310 +2 -4
index f33a15ad028c..5c1f20ce453c 100755
@@ -35,7 +35,6 @@ _cleanup()
 	umount $SCRATCH_MNT > /dev/null 2>&1
 	_dmhugedisk_cleanup
 	rm -rf $tmp.*
-	_scratch_mkfs >/dev/null 2>&1
 }
 
 # get standard environment, filters and checks
@@ -46,7 +45,7 @@ _cleanup()
 # real QA test starts here
 _supported_os Linux
 _supported_fs xfs
-_require_scratch
+_require_scratch_nocheck
 _require_xfs_scratch_rmapbt
 _require_xfs_io_command "falloc"
 
@@ -60,7 +59,7 @@ _scratch_mount >> $seqres.full
 testdir=$SCRATCH_MNT/test-$seq
 blksz="$(_get_block_size $SCRATCH_MNT)"
 
-umount $SCRATCH_MNT
+_scratch_unmount
 
 echo "Format huge device"
 nr_blks=2100000	# 2^21 plus a little more
@@ -110,7 +109,6 @@ test $nr_rmaps -eq 1 || xfs_db -c 'agf 0' -c 'addr rmaproot' -c 'p' $DMHUGEDISK_
 echo "Check and fake-repair huge filesystem again" | tee -a $seqres.full
 $XFS_DB_PROG -c 'check' $DMHUGEDISK_DEV
 $XFS_REPAIR_PROG -n $DMHUGEDISK_DEV >> $seqres.full 2>&1
-_dmhugedisk_cleanup
 
 echo "Done"
 
Signature · 1 line
2.13.6
Reviewed-by on v1 3/6
#4 lore raw

Re: [PATCH 3/6] xfs/310: Cleanup

Quoted 8 lines · Jan Kara

Cleanup couple of things that were objected to when creating test ext4/033 out of this one.

Use _require_scratch_nocheck instead of recreating scratch fs before exiting. Avoid needless cleanup of dmhugedev on exit - _cleanup takes care of that. Use _scratch_unmount where possible.

Looks good.

Signature · 2 lines
Dave Chinner
[email protected]
author
#5 lore raw

PATCH 4/6 ext4/032: Fix up resize2fs usage

In reply to Jan Kara

The test uses resize2fs(8) without proper feature test macro and also without specifying full path to it. Fix that.

---
 tests/ext4/032 | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
tests/ext4/032 +2 -1
index 22cd533be8e8..13faa26a0f9f 100755
@@ -72,7 +72,7 @@ ext4_online_resize()
 
 	echo "+++ resize fs to $final_size" | tee -a $seqres.full
 
-	resize2fs -f ${LOOP_DEVICE} $final_size >$tmp.resize2fs 2>&1
+	$RESIZE2FS_PROG -f ${LOOP_DEVICE} $final_size >$tmp.resize2fs 2>&1
 	if [ $? -ne 0 ]; then
 		if [ $check_if_supported -eq 1 ]; then
 			grep -iq "operation not supported" $tmp.resize2fs \
@@ -114,6 +114,7 @@ _require_scratch
 # We use resize_inode to make sure that block group descriptor table
 # can be extended.
 _require_scratch_ext4_feature "bigalloc,resize_inode"
+_require_command "$RESIZE2FS_PROG" resize2fs
 
 _scratch_mkfs >>$seqres.full 2>&1
 _scratch_mount
Signature · 1 line
2.13.6
author
#6 lore raw

PATCH 5/6 ext4/010: Fixup resize2fs and dumpe2fs usage

In reply to Jan Kara

Use $RESIZE2FS_PROG and $DUMPE2FS_PROG instead of names directly to make progs be executed with full path. Also add proper feature test macros.

---
 tests/ext4/010 | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
tests/ext4/010 +6 -4
index c877bd7af43f..37ba2aaa7c85 100755
@@ -47,8 +47,10 @@ _supported_fs ext4
 _supported_os Linux
 
 _require_scratch
+_require_dumpe2fs
 test -n "${FORCE_FUZZ}" || _require_scratch_ext4_crc
 _require_attrs
+_require_command "$RESIZE2FS_PROG" resize2fs
 
 rm -f $seqres.full
 TESTDIR="${SCRATCH_MNT}/scratchdir"
@@ -56,9 +58,9 @@ TESTFILE="${TESTDIR}/testfile"
 
 echo "+ create scratch fs"
 _scratch_mkfs_ext4 > /dev/null 2>&1
-dumpe2fs -g "${SCRATCH_DEV}" > /dev/null 2>&1 || _notrun "dumpe2fs -g not supported"
-resize2fs -M "${SCRATCH_DEV}" >> $seqres.full 2>&1
-nr_groups="$(dumpe2fs -g "${SCRATCH_DEV}" 2> /dev/null | tail -n 1 | cut -d : -f 1)"
+$DUMPE2FS_PROG -g "${SCRATCH_DEV}" > /dev/null 2>&1 || _notrun "dumpe2fs -g not supported"
+$RESIZE2FS_PROG -M "${SCRATCH_DEV}" >> $seqres.full 2>&1
+nr_groups="$($DUMPE2FS_PROG -g "${SCRATCH_DEV}" 2> /dev/null | tail -n 1 | cut -d : -f 1)"
 
 echo "+ mount fs image"
 _scratch_mount
@@ -74,7 +76,7 @@ echo "+ check fs"
 e2fsck -fn "${SCRATCH_DEV}" >> $seqres.full 2>&1 || _fail "fsck should not fail"
 
 echo "+ corrupt image"
-dumpe2fs -g "${SCRATCH_DEV}" 2>/dev/null | awk -F ':' '{if (int($6) > 0) {print $6}}' | while read blk; do
+$DUMPE2FS_PROG -g "${SCRATCH_DEV}" 2>/dev/null | awk -F ':' '{if (int($6) > 0) {print $6}}' | while read blk; do
 	debugfs -w -n -R "zap_block ${blk}" "${SCRATCH_DEV}" >> $seqres.full 2>&1 || _fail "inode bitmap fuzz failed"
 done
 
Signature · 1 line
2.13.6
author
#7 lore raw

PATCH 6/6 ext4/306: Fixup resize2fs usage

In reply to Jan Kara

Use $RESIZE2FS_PROG instead of direct name to call program with full path. Also add an appropriate feature test.

---
 tests/ext4/306 | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
tests/ext4/306 +2 -1
index 7d00bad6c1ce..5b3744d0597c 100755
@@ -44,6 +44,7 @@ _supported_fs ext4
 _supported_os Linux
 
 _require_scratch
+_require_command "$RESIZE2FS_PROG" resize2fs
 
 rm -f $seqres.full
 
@@ -67,7 +68,7 @@ df -h $SCRATCH_MNT >> $seqres.full
 
 # Grow fs by 512m
 echo "Resize to 1g"
-resize2fs $SCRATCH_DEV 1g >> $seqres.full 2>&1 || _fail "Could not resize to 1g"
+$RESIZE2FS_PROG $SCRATCH_DEV 1g >> $seqres.full 2>&1 || _fail "Could not resize to 1g"
 df -h $SCRATCH_MNT >> $seqres.full
 
 # See if we can add more blocks to the files
Signature · 1 line
2.13.6