1 From: Jeff Mahoney <jeffm@suse.com>
2 Subject: [PATCH] ext3: always mark super uptodate before dirty
5 The superblock's bh is something of an exception. It is only read
6 during mount and is only released during unmount. The in-memory
7 copy is invariably the most recent one.
9 If a write error occurs while syncing the superblock, it will be marked
10 !uptodate. When another error occurs, ext3_error will invoke
11 ext3_commit_super, which will mark the superblock dirty and try to
12 sync it out again. If the buffer is !uptodate, then mark_buffer_dirty
13 will issue a warning, but continue anyway.
15 This patch marks it uptodate before writing it out. This doesn't really
16 change anything other than silencing the warning in mark_buffer_dirty.
17 If the write succeeds, good. Otherwise, it will just have uptodate
20 Signed-off-by: Jeff Mahoney <jeffm@suse.com>
22 fs/ext3/super.c | 7 +++++++
23 1 file changed, 7 insertions(+)
27 @@ -2265,6 +2265,13 @@ static void ext3_commit_super (struct su
28 es->s_free_blocks_count = cpu_to_le32(ext3_count_free_blocks(sb));
29 es->s_free_inodes_count = cpu_to_le32(ext3_count_free_inodes(sb));
30 BUFFER_TRACE(sbh, "marking dirty");
32 + /* We only read the superblock once. The in-memory version is
33 + * always the most recent. If ext3_error is called after a
34 + * superblock write failure, it will be !uptodate. This write
35 + * will likely fail also, but it avoids the WARN_ON in
36 + * mark_buffer_dirty. */
37 + set_buffer_uptodate(sbh);
38 mark_buffer_dirty(sbh);
40 sync_dirty_buffer(sbh);