Skip to contentSkip to conversation

ext4/048: Add new regression test

PATCH Superseded chenlong 2 participants 5 messages May 19, 2021 – May 21, 2021

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

chenlong

author
#0 lore raw

PATCH 1/1 ext4/048: Add new regression test

Check the block group zero and prevent initializing reserved inodes. But in some special cases, the reserved inode may not all belong to the group zero, it may exist into the second group if we format filesystem below.
mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda
So, it will end up triggering a false positive report of a corrupted file system

It's a regression test for commit a149d2a5cabb.

---
 tests/ext4/048     | 50 ++++++++++++++++++++++++++++++++++++++++++++++
 tests/ext4/048.out |  5 +++++
 tests/ext4/group   |  1 +
 3 files changed, 56 insertions(+)
 create mode 100755 tests/ext4/048
 create mode 100644 tests/ext4/048.out
tests/ext4/048 new file +50 -0
new file mode 100755
index 00000000..56801579
@@ -0,0 +1,50 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Huawei.  All Rights Reserved.
+#
+# FS QA Test 048
+#
+# Regression test for commit:
+# a149d2a5cabb(ext4: fix check to prevent false positive report of incorrect
+# used inodes)
+#
+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.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+_supported_fs ext4
+_require_scratch
+
+echo "+ create scratch fs"
+_scratch_mkfs_ext4 -b 4096 -g 8192 -N 1024 -I 4096 >> $seqres.full 2>&1
+
+echo "+ mount fs"
+_scratch_mount -o errors=remount-ro
+sleep 5
+
+echo "+ check mountpoint status"
+cat /proc/self/mounts | grep ${SCRATCH_MNT} | \
+    $AWK_PROG '{print $4}' | grep -oE '^rw,' | tee -a $seqres.full
+
+# success, all done
+status=0
+exit
tests/ext4/048.out new file +5 -0
new file mode 100644
index 00000000..16e50e86
@@ -0,0 +1,5 @@
+QA output created by 048
++ create scratch fs
++ mount fs
++ check mountpoint status
+rw,
tests/ext4/group +1 -0
index ceda2ba6..82b77efb 100644
@@ -50,6 +50,7 @@
 045 auto dir
 046 auto prealloc quick
 047 auto quick dax
+048 auto quick
 271 auto rw quick
 301 aio auto ioctl rw stress defrag
 302 aio auto ioctl rw stress defrag
Signature · 1 line
2.18.0.huawei.25
#1 lore raw
Quoted 57 lines · chenlong

Check the block group zero and prevent initializing reserved inodes. But in some special cases, the reserved inode may not all belong to the group zero, it may exist into the second group if we format filesystem below.
mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda
So, it will end up triggering a false positive report of a corrupted file system

It's a regression test for commit a149d2a5cabb.

---
 tests/ext4/048     | 50 ++++++++++++++++++++++++++++++++++++++++++++++
 tests/ext4/048.out |  5 +++++
 tests/ext4/group   |  1 +
 3 files changed, 56 insertions(+)
 create mode 100755 tests/ext4/048
 create mode 100644 tests/ext4/048.out
tests/ext4/048 new file +32 -0
new file mode 100755
index 00000000..56801579
@@ -0,0 +1,50 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Huawei.  All Rights Reserved.
+#
+# FS QA Test 048
+#
+# Regression test for commit:
+# a149d2a5cabb(ext4: fix check to prevent false positive report of incorrect
+# used inodes)
+#
+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.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+

tests/ext4/048 line 42

Quoted 10 lines
tests/ext4/048 new file +10 -0
+# real QA test starts here
+_supported_fs ext4
+_require_scratch
+
+echo "+ create scratch fs"
+_scratch_mkfs_ext4 -b 4096 -g 8192 -N 1024 -I 4096 >> $seqres.full 2>&1
+
+echo "+ mount fs"
+_scratch_mount -o errors=remount-ro
+sleep 5

Why do we need to sleep here? Won't _scratch_mount return nonzero if the (buggy) kernel thought the fs was corrupt?

Also, won't the grep below catch that the fs didn't mount?

tests/ext4/048 line 45

Quoted 3 lines
tests/ext4/048 new file +3 -0
+
+echo "+ check mountpoint status"
+cat /proc/self/mounts | grep ${SCRATCH_MNT} | \

Careful here, if someone sets SCRATCH_MNT=/u, this will match another mount for /usr. At the bare minimum you want 'grep -w' here, I think.

--D

Quoted 31 lines
tests/ext4/048 new file +5 -0
+    $AWK_PROG '{print $4}' | grep -oE '^rw,' | tee -a $seqres.full
+
+# success, all done
+status=0
+exit
tests/ext4/048.out new file +5 -0
new file mode 100644
index 00000000..16e50e86
@@ -0,0 +1,5 @@
+QA output created by 048
++ create scratch fs
++ mount fs
++ check mountpoint status
+rw,
tests/ext4/group +1 -0
index ceda2ba6..82b77efb 100644
@@ -50,6 +50,7 @@
 045 auto dir
 046 auto prealloc quick
 047 auto quick dax
+048 auto quick
 271 auto rw quick
 301 aio auto ioctl rw stress defrag
 302 aio auto ioctl rw stress defrag
Signature · 1 line
2.18.0.huawei.25
#2 lore raw
Quoted 72 lines · Darrick J. Wong
Quoted 57 lines · chenlong

Check the block group zero and prevent initializing reserved inodes. But in some special cases, the reserved inode may not all belong to the group zero, it may exist into the second group if we format filesystem below.
mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda
So, it will end up triggering a false positive report of a corrupted file system

It's a regression test for commit a149d2a5cabb.

---
  tests/ext4/048     | 50 ++++++++++++++++++++++++++++++++++++++++++++++
  tests/ext4/048.out |  5 +++++
  tests/ext4/group   |  1 +
  3 files changed, 56 insertions(+)

create mode 100755 tests/ext4/048
create mode 100644 tests/ext4/048.out

tests/ext4/048 new file +32 -0
new file mode 100755
index 00000000..56801579
@@ -0,0 +1,50 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Huawei.  All Rights Reserved.
+#
+# FS QA Test 048
+#
+# Regression test for commit:
+# a149d2a5cabb(ext4: fix check to prevent false positive report of incorrect
+# used inodes)
+#
+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.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+

tests/ext4/048 line 42

Quoted 10 lines
tests/ext4/048 new file +10 -0
+# real QA test starts here
+_supported_fs ext4
+_require_scratch
+
+echo "+ create scratch fs"
+_scratch_mkfs_ext4 -b 4096 -g 8192 -N 1024 -I 4096 >> $seqres.full 2>&1
+
+echo "+ mount fs"
+_scratch_mount -o errors=remount-ro
+sleep 5

Why do we need to sleep here? Won't _scratch_mount return nonzero if the (buggy) kernel thought the fs was corrupt?

Quoted 2 lines

Also, won't the grep below catch that the fs didn't mount?

The default maximum wake-up delay of the lazy init thread of ext4 is EXT4_DEF_LI_MAX_START_DELAY (5s).

After mounting the file system, the lazy init thread will be awakened to perform group initialization at 0~5s

after the current time. If an error is detected during the initialization process, it will be set to read-only,

so the test case needs to delay a little while waiting for the init thread to execute.

in this testcase, _scratch_mount will be return zero, after a short period of time, the file system will be remounted as read-only.

In an environment with this bug, such as:

[root@testvm ~]# mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda

[root@testvm ~]# mount -o errors=remount-ro /dev/sda /mnt

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (rw,relatime,errors=remount-ro)

After the file system is mounted,file system status is rw,but a moment later, status will be change to ro.

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (ro,relatime,errors=remount-ro)

Quoted 9 lines

tests/ext4/048 line 45

Quoted 3 lines
tests/ext4/048 new file +3 -0
+
+echo "+ check mountpoint status"
+cat /proc/self/mounts | grep ${SCRATCH_MNT} | \

Careful here, if someone sets SCRATCH_MNT=/u, this will match another mount for /usr. At the bare minimum you want 'grep -w' here, I think.

--D

There is a risk here and needs to be fixed.

Quoted 33 lines
Quoted 31 lines
tests/ext4/048 new file +5 -0
+    $AWK_PROG '{print $4}' | grep -oE '^rw,' | tee -a $seqres.full
+
+# success, all done
+status=0
+exit
tests/ext4/048.out new file +5 -0
new file mode 100644
index 00000000..16e50e86
@@ -0,0 +1,5 @@
+QA output created by 048
++ create scratch fs
++ mount fs
++ check mountpoint status
+rw,
tests/ext4/group +1 -0
index ceda2ba6..82b77efb 100644
@@ -50,6 +50,7 @@
  045 auto dir
  046 auto prealloc quick
  047 auto quick dax
+048 auto quick
  271 auto rw quick
  301 aio auto ioctl rw stress defrag
  302 aio auto ioctl rw stress defrag
Signature · 1 line
2.18.0.huawei.25

.

#3 lore raw
Quoted 81 lines · Chenlong (H)
Quoted 72 lines · Darrick J. Wong
Quoted 57 lines · chenlong

Check the block group zero and prevent initializing reserved inodes. But in some special cases, the reserved inode may not all belong to the group zero, it may exist into the second group if we format filesystem below.
mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda
So, it will end up triggering a false positive report of a corrupted file system

It's a regression test for commit a149d2a5cabb.

---
  tests/ext4/048     | 50 ++++++++++++++++++++++++++++++++++++++++++++++
  tests/ext4/048.out |  5 +++++
  tests/ext4/group   |  1 +
  3 files changed, 56 insertions(+)

create mode 100755 tests/ext4/048
create mode 100644 tests/ext4/048.out

tests/ext4/048 new file +32 -0
new file mode 100755
index 00000000..56801579
@@ -0,0 +1,50 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Huawei.  All Rights Reserved.
+#
+# FS QA Test 048
+#
+# Regression test for commit:
+# a149d2a5cabb(ext4: fix check to prevent false positive report of incorrect
+# used inodes)
+#
+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.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+

tests/ext4/048 line 42

Quoted 10 lines
tests/ext4/048 new file +10 -0
+# real QA test starts here
+_supported_fs ext4
+_require_scratch
+
+echo "+ create scratch fs"
+_scratch_mkfs_ext4 -b 4096 -g 8192 -N 1024 -I 4096 >> $seqres.full 2>&1
+
+echo "+ mount fs"
+_scratch_mount -o errors=remount-ro
+sleep 5

Why do we need to sleep here? Won't _scratch_mount return nonzero if the (buggy) kernel thought the fs was corrupt?

Quoted 2 lines

Also, won't the grep below catch that the fs didn't mount?

The default maximum wake-up delay of the lazy init thread of ext4 is EXT4_DEF_LI_MAX_START_DELAY (5s).

Quoted 8 lines

After mounting the file system, the lazy init thread will be awakened to perform group initialization at 0~5s

after the current time. If an error is detected during the initialization process, it will be set to read-only,

so the test case needs to delay a little while waiting for the init thread to execute.

Oh, that's why there's a delay. Could you paste this explanation into the test case as a comment right before the 'sleep 5', please?

Also, if the delay is configurable you might want to set it to a known value at the start of this test.

--D

Quoted 65 lines

in this testcase, _scratch_mount will be return zero, after a short period of time, the file system will be remounted as read-only.

In an environment with this bug, such as:

[root@testvm ~]# mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda

[root@testvm ~]# mount -o errors=remount-ro /dev/sda /mnt

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (rw,relatime,errors=remount-ro)

After the file system is mounted,file system status is rw,but a moment later, status will be change to ro.

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (ro,relatime,errors=remount-ro)

Quoted 9 lines

tests/ext4/048 line 45

Quoted 3 lines
tests/ext4/048 new file +3 -0
+
+echo "+ check mountpoint status"
+cat /proc/self/mounts | grep ${SCRATCH_MNT} | \

Careful here, if someone sets SCRATCH_MNT=/u, this will match another mount for /usr. At the bare minimum you want 'grep -w' here, I think.

--D

There is a risk here and needs to be fixed.

Quoted 33 lines
Quoted 31 lines
tests/ext4/048 new file +5 -0
+    $AWK_PROG '{print $4}' | grep -oE '^rw,' | tee -a $seqres.full
+
+# success, all done
+status=0
+exit
tests/ext4/048.out new file +5 -0
new file mode 100644
index 00000000..16e50e86
@@ -0,0 +1,5 @@
+QA output created by 048
++ create scratch fs
++ mount fs
++ check mountpoint status
+rw,
tests/ext4/group +1 -0
index ceda2ba6..82b77efb 100644
@@ -50,6 +50,7 @@
  045 auto dir
  046 auto prealloc quick
  047 auto quick dax
+048 auto quick
  271 auto rw quick
  301 aio auto ioctl rw stress defrag
  302 aio auto ioctl rw stress defrag
Signature · 1 line
2.18.0.huawei.25

.

#4 lore raw
Quoted 91 lines · Darrick J. Wong
Quoted 81 lines · Chenlong (H)
Quoted 72 lines · Darrick J. Wong
Quoted 57 lines · chenlong

Check the block group zero and prevent initializing reserved inodes. But in some special cases, the reserved inode may not all belong to the group zero, it may exist into the second group if we format filesystem below.
mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda
So, it will end up triggering a false positive report of a corrupted file system

It's a regression test for commit a149d2a5cabb.

---
   tests/ext4/048     | 50 ++++++++++++++++++++++++++++++++++++++++++++++
   tests/ext4/048.out |  5 +++++
   tests/ext4/group   |  1 +
   3 files changed, 56 insertions(+)

create mode 100755 tests/ext4/048
create mode 100644 tests/ext4/048.out

tests/ext4/048 new file +32 -0
new file mode 100755
index 00000000..56801579
@@ -0,0 +1,50 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Huawei.  All Rights Reserved.
+#
+# FS QA Test 048
+#
+# Regression test for commit:
+# a149d2a5cabb(ext4: fix check to prevent false positive report of incorrect
+# used inodes)
+#
+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.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+

tests/ext4/048 line 42

Quoted 10 lines
tests/ext4/048 new file +10 -0
+# real QA test starts here
+_supported_fs ext4
+_require_scratch
+
+echo "+ create scratch fs"
+_scratch_mkfs_ext4 -b 4096 -g 8192 -N 1024 -I 4096 >> $seqres.full 2>&1
+
+echo "+ mount fs"
+_scratch_mount -o errors=remount-ro
+sleep 5

Why do we need to sleep here? Won't _scratch_mount return nonzero if the (buggy) kernel thought the fs was corrupt?

Quoted 2 lines

Also, won't the grep below catch that the fs didn't mount?

The default maximum wake-up delay of the lazy init thread of ext4 is EXT4_DEF_LI_MAX_START_DELAY (5s).

Quoted 8 lines

After mounting the file system, the lazy init thread will be awakened to perform group initialization at 0~5s

after the current time. If an error is detected during the initialization process, it will be set to read-only,

so the test case needs to delay a little while waiting for the init thread to execute.

Quoted 8 lines

Oh, that's why there's a delay. Could you paste this explanation into the test case as a comment right before the 'sleep 5', please?

Also, if the delay is configurable you might want to set it to a known value at the start of this test.

--D

Thank you,here really need to add this explanation and set the sleep time to a configurable value.

Quoted 67 lines
Quoted 27 lines

in this testcase, _scratch_mount will be return zero, after a short period of time, the file system will be remounted as read-only.

In an environment with this bug, such as:

[root@testvm ~]# mkfs.ext4 -b 4096 -g 8192 -N 1024 -I 4096 /dev/sda

[root@testvm ~]# mount -o errors=remount-ro /dev/sda /mnt

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (rw,relatime,errors=remount-ro)

After the file system is mounted,file system status is rw,but a moment later, status will be change to ro.

[root@testvm ~]# mount | grep /dev/sda

/dev/sda on /mnt type ext4 (ro,relatime,errors=remount-ro)

Quoted 7 lines

tests/ext4/048 line 45

Quoted 3 lines
tests/ext4/048 new file +3 -0
+
+echo "+ check mountpoint status"
+cat /proc/self/mounts | grep ${SCRATCH_MNT} | \

Careful here, if someone sets SCRATCH_MNT=/u, this will match another mount for /usr. At the bare minimum you want 'grep -w' here, I think.

Quoted 38 lines
Quoted 2 lines

--D

There is a risk here and needs to be fixed.

Quoted 33 lines
Quoted 31 lines
tests/ext4/048 new file +5 -0
+    $AWK_PROG '{print $4}' | grep -oE '^rw,' | tee -a $seqres.full
+
+# success, all done
+status=0
+exit
tests/ext4/048.out new file +5 -0
new file mode 100644
index 00000000..16e50e86
@@ -0,0 +1,5 @@
+QA output created by 048
++ create scratch fs
++ mount fs
++ check mountpoint status
+rw,
tests/ext4/group +1 -0
index ceda2ba6..82b77efb 100644
@@ -50,6 +50,7 @@
   045 auto dir
   046 auto prealloc quick
   047 auto quick dax
+048 auto quick
   271 auto rw quick
   301 aio auto ioctl rw stress defrag
   302 aio auto ioctl rw stress defrag
Signature · 1 line
2.18.0.huawei.25

.

.