Patchwork [4/4] e2fsck: regression tests for INCOMPAT_MMP feature

login
register
mail settings
Submitter Andreas Dilger
Date Sept. 23, 2011, 11:38 p.m.
Message ID <1316821124-7461-4-git-send-email-adilger@whamcloud.com>
Download mbox | patch
Permalink /patch/116199/
State Accepted
Headers show

Comments

Andreas Dilger - Sept. 23, 2011, 11:38 p.m.
Add tests for the MMP feature - creating a filesystem with mke2fs
and MMP enabled, enable/disable MMP with tune2fs, disabling the
e2fsck MMP flag with tune2fs after a failed e2fsck, and e2fsck
checking and fixing a corrupt MMP block.

The MMP tests need to be run from a real disk, not tmpfs, because
tmpfs doesn't support O_DIRECT reads, which MMP uses to ensure
that reads from the MMP block are not filled from the page cache.
Using a local disk does not slow down the tests noticably, since
they wait to detect if the MMP block is being modified.

Signed-off-by: Andreas Dilger <adilger@whamcloud.com>

#
# ERROR: trailing whitespace
# #19: FILE: tests/f_mmp/expect.1:13:
# +Superblock backups stored on blocks: $
#
# ERROR: Invalid UTF-8, patch and commit message should be encoded in UTF-8
# #22: FILE: tests/f_mmp/expect.1:16:
# +Allocating group tables: 0/21/2   done
#                              ^
#
# ERROR: trailing whitespace
# #22: FILE: tests/f_mmp/expect.1:16:
# +Allocating group tables: 0/2^H^H^H1/2^H^H^H   ^H^H^Hdone                            $
#
# ERROR: Invalid UTF-8, patch and commit message should be encoded in UTF-8
# #23: FILE: tests/f_mmp/expect.1:17:
# +Writing inode tables: 0/21/2   done
#                           ^
#
# ERROR: trailing whitespace
# #23: FILE: tests/f_mmp/expect.1:17:
# +Writing inode tables: 0/2^H^H^H1/2^H^H^H   ^H^H^Hdone                            $
#
# ERROR: Invalid UTF-8, patch and commit message should be encoded in UTF-8
# #25: FILE: tests/f_mmp/expect.1:19:
# +Writing superblocks and filesystem accounting information: 0/21/2   done
#                                                                ^
#
# ERROR: trailing whitespace
# #28: FILE: tests/f_mmp/expect.1:22:
# + $
#
# total: 7 errors, 0 warnings, 289 lines checked
#
# Your patch has style problems, please review.  If any of these errors
# are false positives report them to the maintainer, see
# CHECKPATCH in MAINTAINERS.
---
 tests/f_mmp/expect.1         |   79 ++++++++++++++++++++++++++++++++++++++++++
 tests/f_mmp/script           |   15 ++++++++
 tests/f_mmp_1on/name         |    1 +
 tests/f_mmp_1on/script       |   40 +++++++++++++++++++++
 tests/f_mmp_2off/name        |    1 +
 tests/f_mmp_2off/script      |   40 +++++++++++++++++++++
 tests/f_mmp_e2fsck/name      |    1 +
 tests/f_mmp_e2fsck/script    |   67 +++++++++++++++++++++++++++++++++++
 tests/f_mmp_garbage/expect.1 |    9 +++++
 tests/f_mmp_garbage/expect.2 |    7 ++++
 tests/f_mmp_garbage/name     |    1 +
 tests/f_mmp_garbage/script   |   28 +++++++++++++++
 12 files changed, 289 insertions(+), 0 deletions(-)
 create mode 100644 tests/f_mmp/expect.1
 create mode 100644 tests/f_mmp/script
 create mode 100644 tests/f_mmp_1on/name
 create mode 100644 tests/f_mmp_1on/script
 create mode 100644 tests/f_mmp_2off/name
 create mode 100644 tests/f_mmp_2off/script
 create mode 100644 tests/f_mmp_e2fsck/name
 create mode 100644 tests/f_mmp_e2fsck/script
 create mode 100644 tests/f_mmp_garbage/expect.1
 create mode 100644 tests/f_mmp_garbage/expect.2
 create mode 100644 tests/f_mmp_garbage/name
 create mode 100644 tests/f_mmp_garbage/script
Andreas Dilger - Sept. 23, 2011, 11:51 p.m.
On 2011-09-23, at 5:38 PM, Andreas Dilger wrote:
> Add tests for the MMP feature - creating a filesystem with mke2fs
> and MMP enabled, enable/disable MMP with tune2fs, disabling the
> e2fsck MMP flag with tune2fs after a failed e2fsck, and e2fsck
> checking and fixing a corrupt MMP block.
> 
> The MMP tests need to be run from a real disk, not tmpfs, because
> tmpfs doesn't support O_DIRECT reads, which MMP uses to ensure
> that reads from the MMP block are not filled from the page cache.
> Using a local disk does not slow down the tests noticably, since
> they wait to detect if the MMP block is being modified.
> 
> Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
> 
> #
> # ERROR: trailing whitespace
> # #19: FILE: tests/f_mmp/expect.1:13:
> # +Superblock backups stored on blocks: $
> #
> # ERROR: Invalid UTF-8, patch and commit message should be encoded in UTF-8
> # #22: FILE: tests/f_mmp/expect.1:16:
> # +Allocating group tables: 0/21/2   done
> #                              ^

Sigh, my bad.  I installed an automatic git prepare_commit_msg hook that
runs checkpatch.pl against the patch, but it doesn't like the format of
the expect output (which is not C code and probably should be ignored).

Please delete this stuff out of the comments, or I can resubmit if you want.

Cheers, Andreas
--
Andreas Dilger 
Principal Engineer
Whamcloud, Inc.



--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Theodore Ts'o - Sept. 24, 2011, 6:51 p.m.
On Fri, Sep 23, 2011 at 05:38:44PM -0600, Andreas Dilger wrote:
> Add tests for the MMP feature - creating a filesystem with mke2fs
> and MMP enabled, enable/disable MMP with tune2fs, disabling the
> e2fsck MMP flag with tune2fs after a failed e2fsck, and e2fsck
> checking and fixing a corrupt MMP block.
> 
> The MMP tests need to be run from a real disk, not tmpfs, because
> tmpfs doesn't support O_DIRECT reads, which MMP uses to ensure
> that reads from the MMP block are not filled from the page cache.
> Using a local disk does not slow down the tests noticably, since
> they wait to detect if the MMP block is being modified.
> 
> Signed-off-by: Andreas Dilger <adilger@whamcloud.com>

Applied, although I needed to refrob the test names.  The first
character denotes not the test is for:

*) f_* tests are for e2fsck
*) m_* tests are for tune2fs
*) u_* tests are for e2undofs

This allows someone to test mke2fs by running the command:

	(cd tests; ./test_script m)

						- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Andreas Dilger - Sept. 25, 2011, 6:04 a.m.
On 2011-09-24, at 12:51 PM, Ted Ts'o wrote:
> On Fri, Sep 23, 2011 at 05:38:44PM -0600, Andreas Dilger wrote:
>> Add tests for the MMP feature - creating a filesystem with mke2fs
>> and MMP enabled, enable/disable MMP with tune2fs, disabling the
>> e2fsck MMP flag with tune2fs after a failed e2fsck, and e2fsck
>> checking and fixing a corrupt MMP block.
>> 
>> The MMP tests need to be run from a real disk, not tmpfs, because
>> tmpfs doesn't support O_DIRECT reads, which MMP uses to ensure
>> that reads from the MMP block are not filled from the page cache.
>> Using a local disk does not slow down the tests noticably, since
>> they wait to detect if the MMP block is being modified.
>> 
>> Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
> 
> Applied, although I needed to refrob the test names.  The first
> character denotes not the test is for:
> 
> *) f_* tests are for e2fsck
> *) m_* tests are for tune2fs

You mean m_* for mke2fs and t_* for tune2fs, I think.

> *) u_* tests are for e2undofs
> 
> This allows someone to test mke2fs by running the command:

FYI, the reason that I ordered the MMP tests in that way is because it
makes sense to verify that mke2fs can create a filesystem with the MMP
feature before using a filesystem created by mke2fs with MMP to test if
e2fsck handles it correctly.

At this point, it is unlikely that there will be a problem with this in
the future, because the code is working today, and at worst it means that
the e2fsck f_mmp_* tests might incorrectly pass before m_mmp_* fails, or
something, so I don't insist on changing it back.

In any case, thanks for merging the patches.

Cheers, Andreas
--
Andreas Dilger 
Principal Engineer
Whamcloud, Inc.



--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Theodore Ts'o - Sept. 25, 2011, 12:05 p.m.
On Sun, Sep 25, 2011 at 12:04:14AM -0600, Andreas Dilger wrote:
> FYI, the reason that I ordered the MMP tests in that way is because it
> makes sense to verify that mke2fs can create a filesystem with the MMP
> feature before using a filesystem created by mke2fs with MMP to test if
> e2fsck handles it correctly.

There are other e2fsck tests that use mke2fs, and for better or for
worse I've never worried about test ordering.  If mke2fs blows up for
whatever reason, then some f_* tests will fail for reasons that have
nothing to do with e2fsck.  Since the test suite doesn't take that
long to run, most of the time people will let the tests run to
completion and the wise developer will notice the string of mke2fs
failures, and decide to tackle that first.  (Or it will be obvious the
moment he or she looks like the failed log files.)

       	     	       	    	       - Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/tests/f_mmp/expect.1 b/tests/f_mmp/expect.1
new file mode 100644
index 0000000..3a00815
--- /dev/null
+++ b/tests/f_mmp/expect.1
@@ -0,0 +1,79 @@ 
+Filesystem label=
+OS type: Linux
+Block size=2048 (log=1)
+Fragment size=2048 (log=1)
+Stride=0 blocks, Stripe width=0 blocks
+16384 inodes, 32768 blocks
+1638 blocks (5.00%) reserved for the super user
+First data block=0
+Maximum filesystem blocks=33554432
+2 block groups
+16384 blocks per group, 16384 fragments per group
+8192 inodes per group
+Superblock backups stored on blocks: 
+	16384
+
+Allocating group tables: 0/21/2   done                            
+Writing inode tables: 0/21/2   done                            
+Multiple mount protection is enabled with update interval 5 seconds.
+Writing superblocks and filesystem accounting information: 0/21/2   done
+
+Filesystem features: ext_attr resize_inode dir_index filetype mmp sparse_super
+ 
+Pass 1: Checking inodes, blocks, and sizes
+Pass 2: Checking directory structure
+Pass 3: Checking directory connectivity
+Pass 4: Checking reference counts
+Pass 5: Checking group summary information
+test_filesys: 11/16384 files (0.0% non-contiguous), 1105/32768 blocks
+Exit status is 0
+
+Filesystem volume name:   <none>
+Last mounted on:          <not available>
+Filesystem magic number:  0xEF53
+Filesystem revision #:    1 (dynamic)
+Filesystem features:      ext_attr resize_inode dir_index filetype mmp sparse_super
+Default mount options:    (none)
+Filesystem state:         clean
+Errors behavior:          Continue
+Filesystem OS type:       Linux
+Inode count:              16384
+Block count:              32768
+Reserved block count:     1638
+Free blocks:              31663
+Free inodes:              16373
+First block:              0
+Block size:               2048
+Fragment size:            2048
+Reserved GDT blocks:      31
+Blocks per group:         16384
+Fragments per group:      16384
+Inodes per group:         8192
+Inode blocks per group:   512
+Mount count:              0
+Check interval:           15552000 (6 months)
+Reserved blocks uid:      0
+Reserved blocks gid:      0
+First inode:              11
+Inode size:	          128
+Default directory hash:   half_md4
+MMP block number:         557
+MMP update interval:      5
+
+
+Group 0: (Blocks 0-16383)
+  Primary superblock at 0, Group descriptors at 1-1
+  Reserved GDT blocks at 2-32
+  Block bitmap at 33 (+33), Inode bitmap at 34 (+34)
+  Inode table at 35-546 (+35)
+  15826 free blocks, 8181 free inodes, 2 directories
+  Free blocks: 558-16383
+  Free inodes: 12-8192
+Group 1: (Blocks 16384-32767)
+  Backup superblock at 16384, Group descriptors at 16385-16385
+  Reserved GDT blocks at 16386-16416
+  Block bitmap at 16417 (+33), Inode bitmap at 16418 (+34)
+  Inode table at 16419-16930 (+35)
+  15837 free blocks, 8192 free inodes, 0 directories
+  Free blocks: 16931-32767
+  Free inodes: 8193-16384
diff --git a/tests/f_mmp/script b/tests/f_mmp/script
new file mode 100644
index 0000000..1547463
--- /dev/null
+++ b/tests/f_mmp/script
@@ -0,0 +1,15 @@ 
+DESCRIPTION="enable MMP during mke2fs"
+FS_SIZE=65536
+MKE2FS_DEVICE_SECTSIZE=2048
+export MKE2FS_DEVICE_SECTSIZE
+TMPFILE=test.img
+> $TMPFILE
+stat -f $TMPFILE | grep -q "Type: tmpfs"
+if [ $? == 0 ]; then
+	rm -f $TMPFILE
+	echo "skipped for tmpfs (no O_DIRECT support)"
+	return 0
+fi
+MKE2FS_OPTS="-O mmp"
+. $cmd_dir/run_mke2fs
+unset MKE2FS_DEVICE_SECTSIZE
diff --git a/tests/f_mmp_1on/name b/tests/f_mmp_1on/name
new file mode 100644
index 0000000..62e16b2
--- /dev/null
+++ b/tests/f_mmp_1on/name
@@ -0,0 +1 @@ 
+enable MMP using tune2fs
diff --git a/tests/f_mmp_1on/script b/tests/f_mmp_1on/script
new file mode 100644
index 0000000..ee9884f
--- /dev/null
+++ b/tests/f_mmp_1on/script
@@ -0,0 +1,40 @@ 
+FSCK_OPT=-yf
+
+TMPFILE=test.img
+rm -f $test_name.failed $test_name.ok
+> $TMPFILE
+
+stat -f $TMPFILE | grep -q "Type: tmpfs"
+if [ $? == 0 ] ; then
+	rm -f $TMPFILE
+	echo "skipped for tmpfs (no O_DIRECT support)"
+	return 0
+fi
+
+$MKE2FS -q -F -o Linux -b 1024 $TMPFILE 100 > $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "mke2fs failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+$TUNE2FS -O mmp -E mmp_update_interval=1 $TMPFILE >> $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "tune2fs -O mmp failed with $status" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+$FSCK $FSCK_OPT $TMPFILE >> $test_name.log 2>&1
+status=$?
+if [ "$status" = 0 ] ; then
+	echo "ok"
+	touch $test_name.ok
+else
+	echo "e2fsck with MMP enabled failed with $status" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+rm -f $TMPFILE
diff --git a/tests/f_mmp_2off/name b/tests/f_mmp_2off/name
new file mode 100644
index 0000000..d7cac51
--- /dev/null
+++ b/tests/f_mmp_2off/name
@@ -0,0 +1 @@ 
+disable MMP using tune2fs
diff --git a/tests/f_mmp_2off/script b/tests/f_mmp_2off/script
new file mode 100644
index 0000000..ec9f71e
--- /dev/null
+++ b/tests/f_mmp_2off/script
@@ -0,0 +1,40 @@ 
+FSCK_OPT=-yf
+
+TMPFILE=test.img
+rm -f $test_name.failed $test_name.ok
+> $TMPFILE
+
+stat -f $TMPFILE | grep -q "Type: tmpfs"
+if [ $? == 0 ]; then
+	rm -f $TMPFILE
+	echo "skipped for tmpfs (no O_DIRECT support)"
+	return 0
+fi
+
+$MKE2FS -q -F -o Linux -b 1024 -O mmp $TMPFILE 100 > $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "mke2fs -O mmp failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+$TUNE2FS -O ^mmp $TMPFILE > $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "tune2fs -O ^mmp failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+$FSCK $FSCK_OPT $TMPFILE > $test_name.log 2>&1
+status=$?
+if [ "$status" = 0 ] ; then
+	echo "ok"
+	touch $test_name.ok
+else
+	echo "e2fsck after MMP disable failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+rm -f $TMPFILE
diff --git a/tests/f_mmp_e2fsck/name b/tests/f_mmp_e2fsck/name
new file mode 100644
index 0000000..20d66d6
--- /dev/null
+++ b/tests/f_mmp_e2fsck/name
@@ -0,0 +1 @@ 
+disable MMP with tune2fs after e2fsck killed
diff --git a/tests/f_mmp_e2fsck/script b/tests/f_mmp_e2fsck/script
new file mode 100644
index 0000000..548734a
--- /dev/null
+++ b/tests/f_mmp_e2fsck/script
@@ -0,0 +1,67 @@ 
+FSCK_OPT=-yf
+
+TMPFILE=test.img
+rm -f $test_name.failed $test_name.ok
+> $TMPFILE
+
+stat -f $TMPFILE | grep -q "Type: tmpfs"
+if [ $? == 0 ]; then
+	rm -f $TMPFILE
+	echo "skipped for tmpfs (no O_DIRECT support)"
+	return 0
+fi
+
+echo "make the test image ..." > $test_name.log
+$MKE2FS -q -F -o Linux -b 1024 -O mmp -E mmp_update_interval=1 $TMPFILE 100 >> $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "mke2fs -O mmp failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+# this will cause debugfs to create the $test_name.mark file once it has
+# passed the MMP startup, then continue reading input until it is killed
+MARKFILE=$test_name.new
+rm -f $MARKFILE
+echo "set mmp sequence to EXT2_MMP_SEQ_FSCK..." >> $test_name.log
+( { echo dump_mmp; echo "dump_inode <2> $MARKFILE"; cat /dev/zero; } |
+	$DEBUGFS -w $TMPFILE >> $test_name.log 2>&1 & ) > /dev/null 2>&1 &
+echo "wait until debugfs has started ..." >> $test_name.log
+while [ ! -e $MARKFILE ]; do
+	sleep 1
+done
+rm -f $MARKFILE
+echo "kill debugfs abruptly (simulates e2fsck failure) ..." >> $test_name.log
+killall -9 debugfs >> $test_name.log
+
+
+echo "e2fsck (should fail mmp_seq = EXT2_MMP_SEQ_FSCK) ..." >> $test_name.log
+$FSCK $FSCK_OPT $TMPFILE >> $test_name.log 2>&1
+status=$?
+if [ "$status" == 0 ] ; then
+	echo "e2fsck with MMP as EXT2_MMP_SEQ_FSCK ran!" > $test_name.failed
+	echo "failed"
+	return 1
+fi
+
+echo "clear mmp_seq with tune2fs ..." >> $test_name.log
+$TUNE2FS -f -E clear_mmp $TMPFILE >> $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "tune2fs clearing EXT2_MMP_SEQ_FSCK failed" > $test_name.failed
+	echo "failed"
+	return 1
+fi
+
+echo "run e2fsck again (should pass with clean mmp_seq) ..." >> $test_name.log
+$FSCK $FSCK_OPT $TMPFILE >> $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "e2fsck after clearing EXT2_MMP_SEQ_FSCK failed"> $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+echo "ok"
+rm -f $TMPFILE
diff --git a/tests/f_mmp_garbage/expect.1 b/tests/f_mmp_garbage/expect.1
new file mode 100644
index 0000000..4ee5cfb
--- /dev/null
+++ b/tests/f_mmp_garbage/expect.1
@@ -0,0 +1,9 @@ 
+Superblock has invalid MMP magic.  Fix? yes
+
+Pass 1: Checking inodes, blocks, and sizes
+Pass 2: Checking directory structure
+Pass 3: Checking directory connectivity
+Pass 4: Checking reference counts
+Pass 5: Checking group summary information
+test_filesys: 11/16 files (0.0% non-contiguous), 22/100 blocks
+Exit status is 0
diff --git a/tests/f_mmp_garbage/expect.2 b/tests/f_mmp_garbage/expect.2
new file mode 100644
index 0000000..3bf3869
--- /dev/null
+++ b/tests/f_mmp_garbage/expect.2
@@ -0,0 +1,7 @@ 
+Pass 1: Checking inodes, blocks, and sizes
+Pass 2: Checking directory structure
+Pass 3: Checking directory connectivity
+Pass 4: Checking reference counts
+Pass 5: Checking group summary information
+test_filesys: 11/16 files (0.0% non-contiguous), 22/100 blocks
+Exit status is 0
diff --git a/tests/f_mmp_garbage/name b/tests/f_mmp_garbage/name
new file mode 100644
index 0000000..17e0b14
--- /dev/null
+++ b/tests/f_mmp_garbage/name
@@ -0,0 +1 @@ 
+repair MMP when it is corrupted
diff --git a/tests/f_mmp_garbage/script b/tests/f_mmp_garbage/script
new file mode 100644
index 0000000..3c80032
--- /dev/null
+++ b/tests/f_mmp_garbage/script
@@ -0,0 +1,28 @@ 
+FSCK_OPT=-yf
+
+TMPFILE=test.img
+rm -f $test_name.failed $test_name.ok
+> $TMPFILE
+
+stat -f $TMPFILE | grep -q "Type: tmpfs"
+if [ $? == 0 ] ; then
+	rm -f $TMPFILE
+	echo "skipped for tmpfs (no O_DIRECT support)"
+	return 0
+fi
+
+echo "make the test image ..." > $test_name.log
+$MKE2FS -q -F -o Linux -b 1024 -O mmp -E mmp_update_interval=1 $TMPFILE 100 >> $test_name.log 2>&1
+status=$?
+if [ "$status" != 0 ] ; then
+	echo "mke2fs -O mmp failed" > $test_name.failed
+	echo "failed"
+	return $status
+fi
+
+# create a corrupted image
+echo "modify the mmp sequence ..." >> $test_name.log
+$DEBUGFS -w -R "set_mmp_value magic 0x12345678" $TMPFILE >> $test_name.log 2>&1
+
+SKIP_GUNZIP=true
+. $cmd_dir/run_e2fsck