reiserfs, kill-the-BKL: fix unsafe j_flush_mutex lock
Frederic Weisbecker [Mon, 13 Apr 2009 22:10:35 +0000 (00:10 +0200)]
Impact: fix a deadlock

The j_flush_mutex is acquired safely in journal.c:
if we can't take it, we free the reiserfs per superblock lock
and wait a bit.

But we have a remaining place in kupdate_transactions() where
j_flush_mutex is still acquired traditionnaly. Thus the following
scenario (warned by lockdep) can happen:

A B

mutex_lock(&write_lock) mutex_lock(&write_lock)
mutex_lock(&j_flush_mutex) mutex_lock(&j_flush_mutex) //block
mutex_unlock(&write_lock)
sleep...
mutex_lock(&write_lock) //deadlock

Fix this by using reiserfs_mutex_lock_safe() in kupdate_transactions().

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Alessio Igor Bogani <abogani@texware.it>
Cc: Jeff Mahoney <jeffm@suse.com>
LKML-Reference: <1239660635-12940-1-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

fs/reiserfs/journal.c

index 438c71f..0ba98ca 100644 (file)
@@ -1826,7 +1826,7 @@ static int kupdate_transactions(struct super_block *s,
        struct reiserfs_journal *journal = SB_JOURNAL(s);
        chunk.nr = 0;
 
-       mutex_lock(&journal->j_flush_mutex);
+       reiserfs_mutex_lock_safe(&journal->j_flush_mutex, s);
        if (!journal_list_still_alive(s, orig_trans_id)) {
                goto done;
        }