ocfs2: fix regression in ocfs2_read_blocks_sync()
We're panicing in ocfs2_read_blocks_sync() if a jbd-managed buffer is seen.
At first glance, this seems ok but in reality it can happen. My test case
was to just run 'exorcist'. A struct inode is being pushed out of memory but
is then re-read at a later time, before the buffer has been checkpointed by
jbd. This causes a BUG to be hit in ocfs2_read_blocks_sync().
Reviewed-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
diff --git a/fs/ocfs2/buffer_head_io.c b/fs/ocfs2/buffer_head_io.c
index 7e947c6..3a178ec 100644
--- a/fs/ocfs2/buffer_head_io.c
+++ b/fs/ocfs2/buffer_head_io.c
@@ -112,7 +112,7 @@
bh = bhs[i];
if (buffer_jbd(bh)) {
- mlog(ML_ERROR,
+ mlog(ML_BH_IO,
"trying to sync read a jbd "
"managed bh (blocknr = %llu), skipping\n",
(unsigned long long)bh->b_blocknr);
@@ -147,15 +147,10 @@
for (i = nr; i > 0; i--) {
bh = bhs[i - 1];
- if (buffer_jbd(bh)) {
- mlog(ML_ERROR,
- "the journal got the buffer while it was "
- "locked for io! (blocknr = %llu)\n",
- (unsigned long long)bh->b_blocknr);
- BUG();
- }
+ /* No need to wait on the buffer if it's managed by JBD. */
+ if (!buffer_jbd(bh))
+ wait_on_buffer(bh);
- wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
/* Status won't be cleared from here on out,
* so we can safely record this and loop back
@@ -251,8 +246,6 @@
ignore_cache = 1;
}
- /* XXX: Can we ever get this and *not* have the cached
- * flag set? */
if (buffer_jbd(bh)) {
if (ignore_cache)
mlog(ML_BH_IO, "trying to sync read a jbd "