resize2fs: exit fix_sb_journal_backup early for external journal

Resizing a filesystem with an external journal fails when it tries to read inode 0:

# touch testfs
# truncate testfs 1342177280
# touch testjournal
# truncate testjournal 134217728
# mke2fs -O journal_dev testjournal 
# losetup /dev/loop0 testjournal 
# mkfs.ext4 -J device=/dev/loop0 testfs 127680
# resize2fs testfs 
resize2fs 1.41.9 (22-Aug-2009)
Resizing the filesystem on testfs to 327680 (4k) blocks.
resize2fs: Illegal inode number while trying to resize testfs
Please run 'e2fsck -fy testfs' to fix the filesystem
after the aborted resize operation.

I think the right, simple thing to do is just bail out early for an external journal here, as there are no backup blocks to update.

---

index 5a1eb2a..2d4a14f 100644
@@ -1887,6 +1887,10 @@ static errcode_t fix_sb_journal_backup(ext2_filsys fs)
 	if (!(fs->super->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL))
 		return 0;
 
+	/* External journal? Nothing to do. */
+	if (fs->super->s_journal_dev && !fs->super->s_journal_inum)
+		return 0;
+
 	retval = ext2fs_read_inode(fs, fs->super->s_journal_inum, &inode);
 	if (retval)
 		return retval;

Thank you for the patch, I've refined your testcase:

#!/bin/sh
touch testfs
truncate -s2G testfs
touch testjournal
truncate -s128M testjournal
LOLOG=losetup -f
losetup "${LOLOG}" testjournal
mke2fs -O journal_dev "${LOLOG}"
LOFS=losetup -f
losetup "${LOFS}" testfs
mkfs.ext4 -J device="${LOLOG}" "${LOFS}" 127680
echo -e "\n\nThis will fail\n"
resize2fs "${LOFS}"
echo -e "\nFsck to clear\n"
fsck -fy "${LOFS}"
echo -e "\n\nIf run in the build dir of a the current version/git checkout with the patch applied and build... ./resize/resize2fs will work.\n"
./resize/resize2fs "${LOFS}"
losetup -d "${LOLOG}"
losetup -d "${LOFS}"

Insert data checks/etc as desired.

I've run this test case a few times on my system. It looks like it's the right thing to do, but before I run it on a drive that actually has data I care about I'd like to make sure it's doesn't have some kind of unforeseen side effects.

#!/bin/sh
touch testfs
truncate -s1500M testfs
touch testjournal
truncate -s128M testjournal
LOLOG=losetup -f
losetup "${LOLOG}" testjournal
mke2fs -O journal_dev "${LOLOG}"
LOFS=losetup -f
losetup "${LOFS}" testfs
mkfs.ext4 -J device="${LOLOG}" "${LOFS}" 200000
mkdir /tmp/testfs
mount "${LOFS}" /tmp/testfs
echo "Filesystem mounted, copying /usr/src/linux/ to (try to) fill the device." find /usr/src/linux/ -depth -print0 | cpio -p0mud /tmp/testfs pushd /tmp/testfs
echo "Generating checksums"
find ./ -type f -print0 | xargs -0I = md5sum = > /tmp/check.md5 popd
df | grep "/tmp/testfs"
wc /tmp/check.md5
umount /tmp/testfs
fsck -fy "${LOFS}"
resize2fs "${LOFS}"
fsck -fy "${LOFS}"
./resize/resize2fs "${LOFS}"
mount "${LOFS}" /tmp/testfs
pushd /tmp/testfs
df | grep "/tmp/testfs"
md5sum -c /tmp/check.md5 | sed '/OK[\r\n]*$/d'
if [ "$?" == "0" ] ; then echo "MD5 check PASSED" ; fi
popd
umount /tmp/testfs
rm -rf /tmp/testfs /tmp/check.md5
losetup -d "${LOLOG}"
losetup -d "${LOFS}"
rm testfs testjournal
echo "Cleanup Complete"

I really think the patch is fine, but you can wait 'til Ted merges it for the official stamp of approval... :)

-eric

I agree it looks like the right thing to do there, but I'm not sure if there's some other effect it's supposed to have.

However just based on that small section I should probably be safe if I use the latest RC kernel (for other ext4 reasons that sound important) and then do an fsck -f after the resize.

Thanks, applied to the e2fsprogs maint branch.

- Ted