[Cluster-devel] [GFS2 PATCH] GFS2: Adjust lock reference count

[Bob Peterson <rpeterso@redhat.com>]

Hi,

This patch closes a small timing window whereby a request to hold the
transaction glock can get stuck. The problem is that after the DLM has
granted the lock, it can get into a state whereby it doesn't transition
the glock to a held state, due to not having requeued the glock state
machine to finish the transition.

Steve: Feel free to adjust the patch description as you see fit.
I'm not feeling very creative at the moment.

Regards,

Bob Peterson
Red Hat File Systems

Signed-off-by: Bob Peterson <rpeterso@redhat.com> 
---
 fs/gfs2/glock.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 52f7478..67ac456 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -1047,9 +1047,13 @@ int gfs2_glock_nq(struct gfs2_holder *gh)
 
 	spin_lock(&gl->gl_spin);
 	add_to_queue(gh);
-	if ((LM_FLAG_NOEXP & gh->gh_flags) &&
-	    test_and_clear_bit(GLF_FROZEN, &gl->gl_flags))
+	if (unlikely((LM_FLAG_NOEXP & gh->gh_flags) &&
+		     test_and_clear_bit(GLF_FROZEN, &gl->gl_flags))) {
 		set_bit(GLF_REPLY_PENDING, &gl->gl_flags);
+		atomic_inc(&gl->gl_ref);
+		if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
+			atomic_dec(&gl->gl_ref);
+	}
 	run_queue(gl, 1);
 	spin_unlock(&gl->gl_spin);