GFS2: Only wait for demote when last holder is dequeued
authorBob Peterson <rpeterso@redhat.com>
Fri, 20 Jun 2014 13:36:41 +0000 (09:36 -0400)
committerSteven Whitehouse <swhiteho@redhat.com>
Fri, 18 Jul 2014 10:12:14 +0000 (11:12 +0100)
Function gfs2_glock_dq_wait is supposed to dequeue a glock and then
wait for the lock to be demoted. The problem is, if this is a shared
lock, its demote will depend on the other holders, which means you
might end up waiting forever because the other process is blocked.
This problem is especially apparent when dealing with nested flocks.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
fs/gfs2/glock.c

index c355f7320e448bfe30a8c325b2448d707f8c19be..278fae5b6982f81ca3ee8c17d830ad28d2ab47c6 100644 (file)
@@ -1128,7 +1128,9 @@ void gfs2_glock_dq_wait(struct gfs2_holder *gh)
        struct gfs2_glock *gl = gh->gh_gl;
        gfs2_glock_dq(gh);
        might_sleep();
-       wait_on_bit(&gl->gl_flags, GLF_DEMOTE, gfs2_glock_demote_wait, TASK_UNINTERRUPTIBLE);
+       if (!find_first_holder(gl))
+               wait_on_bit(&gl->gl_flags, GLF_DEMOTE, gfs2_glock_demote_wait,
+                           TASK_UNINTERRUPTIBLE);
 }
 
 /**