Re: ext3: kernel BUG at fs/jbd/journal.c:412!

[Theodore Tso <tytso@mit.edu>]

On Thu, Nov 06, 2008 at 08:12:27AM -0800, Arthur Jones wrote:
> Hi Aneesh, ...
> 
> On Thu, Nov 06, 2008 at 07:50:25AM -0800, Aneesh Kumar K.V wrote:
> > I get this with ext4-patchqueue. I guess we have some ext3 patches queued there.
> 
> This could be related to a patch I just posted.
> 
> See the thread on linux-ext4 called "ext3: slow symlink corruption
> on umount" for details on how this patch came about...

No this is the other ext3 patch we have in the patch tree.  I see the
problem, we're calling __log_space_left in a diagnostic printk after
we've released the j_state_lock.  Here's the incremental fix:

diff --git a/fs/jbd/checkpoint.c b/fs/jbd/checkpoint.c
index 5e856de..18e5137 100644
--- a/fs/jbd/checkpoint.c
+++ b/fs/jbd/checkpoint.c
@@ -115,7 +115,7 @@ static int __try_to_free_cp_buf(struct journal_head *jh)
  */
 void __log_wait_for_space(journal_t *journal)
 {
-	int nblocks;
+	int nblocks, space_left;
 	assert_spin_locked(&journal->j_state_lock);
 
 	nblocks = jbd_space_needed(journal);
@@ -139,7 +139,8 @@ void __log_wait_for_space(journal_t *journal)
 		spin_lock(&journal->j_state_lock);
 		spin_lock(&journal->j_list_lock);
 		nblocks = jbd_space_needed(journal);
-		if (__log_space_left(journal) < nblocks) {
+		space_left = __log_space_left(journal);
+		if (space_left < nblocks) {
 			int chkpt = journal->j_checkpoint_transactions != NULL;
 			int tid = 0;
 
@@ -157,8 +158,7 @@ void __log_wait_for_space(journal_t *journal)
 			} else {
 				printk(KERN_ERR "%s: needed %d blocks and "
 				       "only had %d space available\n",
-				       __func__, nblocks,
-				       __log_space_left(journal));
+				       __func__, nblocks, space_left);
 				printk(KERN_ERR "%s: no way to get more "
 				       "journal space\n", __func__);
 				WARN_ON(1);


This is in a "should never happen path", though.  What were you doing
to trigger it?

						- Ted