Fix reference counting race on log buffers

When we release the iclog, we do an atomic_dec_and_lock to determine if
we are the last reference and need to trigger update of log headers and
writeout.  However, in xlog_state_get_iclog_space() we also need to
check if we have the last reference count there.  If we do, we release
the log buffer, otherwise we decrement the reference count.

But the compare and decrement in xlog_state_get_iclog_space() is not
atomic, so both places can see a reference count of 2 and neither will
release the iclog.  That leads to a filesystem hang.

Close the race by replacing the atomic_read() and atomic_dec() pair with
atomic_add_unless() to ensure that they are executed atomically.

Signed-off-by: Dave Chinner <david@fromorbit.com>
Reviewed-by: Tim Shimmin <tes@sgi.com>
Tested-by: Eric Sandeen <sandeen@sandeen.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index afaee301b0ee1fa2e1c491f5bd2109ed277fabf1..ad3d26ddfe3160739908362ab2f2054d0bfb530e 100644 (file)
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -2427,13 +2427,20 @@ restart:
        if (iclog->ic_size - iclog->ic_offset < 2*sizeof(xlog_op_header_t)) {
                xlog_state_switch_iclogs(log, iclog, iclog->ic_size);
 
-               /* If I'm the only one writing to this iclog, sync it to disk */
-               if (atomic_read(&iclog->ic_refcnt) == 1) {
+               /*
+                * If I'm the only one writing to this iclog, sync it to disk.
+                * We need to do an atomic compare and decrement here to avoid
+                * racing with concurrent atomic_dec_and_lock() calls in
+                * xlog_state_release_iclog() when there is more than one
+                * reference to the iclog.
+                */
+               if (!atomic_add_unless(&iclog->ic_refcnt, -1, 1)) {
+                       /* we are the only one */
                        spin_unlock(&log->l_icloglock);
-                       if ((error = xlog_state_release_iclog(log, iclog)))
+                       error = xlog_state_release_iclog(log, iclog);
+                       if (error)
                                return error;
                } else {
-                       atomic_dec(&iclog->ic_refcnt);
                        spin_unlock(&log->l_icloglock);
                }
                goto restart;