]> pilppa.com Git - linux-2.6-omap-h63xx.git/commit
ext3: lighten up resize transaction requirements
authorEric Sandeen <sandeen@redhat.com>
Wed, 17 Oct 2007 06:30:28 +0000 (23:30 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Wed, 17 Oct 2007 15:43:01 +0000 (08:43 -0700)
commit1ad6ecf9146e85ccb4e0bce70b91a93f57145c72
tree03e90fbc705f02773def014dbd4abda4dabb5dd9
parentd58ae67813ff97030d2f47ff7d1e5f54e5d7c5b3
ext3: lighten up resize transaction requirements

When resizing online, setup_new_group_blocks attempts to reserve a
potentially very large transaction, depending on the current filesystem
geometry.  For some journal sizes, there may not be enough room for this
transaction, and the online resize will fail.

The patch below resizes & restarts the transaction as necessary while
setting up the new group, and should work with even the smallest journal.

Tested with something like:

[root@newbox ~]# dd if=/dev/zero of=fsfile bs=1024 count=32768
[root@newbox ~]# mkfs.ext3 -b 1024 fsfile 16384
[root@newbox ~]# mount -o loop fsfile mnt/
[root@newbox ~]# resize2fs /dev/loop0
resize2fs 1.40.2 (12-Jul-2007)
Filesystem at /dev/loop0 is mounted on /root/mnt; on-line resizing required
old desc_blocks = 1, new_desc_blocks = 1
Performing an on-line resize of /dev/loop0 to 32768 (1k) blocks.
resize2fs: No space left on device While trying to add group #2
[root@newbox ~]# dmesg | tail -n 1
JBD: resize2fs wants too many credits (258 > 256)
[root@newbox ~]#

With the below change, it works.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Acked-by: Andreas Dilger <adilger@clusterfs.com>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ext3/resize.c