ocfs2_dlm: Flush dlm workqueue before starting to migrate

This is to prevent the condition in which a previously queued
up assert master asserts after we start the migration. Now
migration ensures the workqueue is flushed before proceeding
with migrating the lock to another node. This condition is
typically encountered during parallel umounts.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

fs/ocfs2/dlm/dlmmaster.c

@@ -1507,10 +1507,11 @@ int dlm_master_request_handler(struct o2net_msg *msg, u32 len, void *data)
 
 		/* take care of the easy cases up front */
 		spin_lock(&res->spinlock);
-		if (res->state & DLM_LOCK_RES_RECOVERING) {
+		if (res->state & (DLM_LOCK_RES_RECOVERING|
+				  DLM_LOCK_RES_MIGRATING)) {
 			spin_unlock(&res->spinlock);
 			mlog(0, "returning DLM_MASTER_RESP_ERROR since res is "
-			     "being recovered\n");
+			     "being recovered/migrated\n");
 			response = DLM_MASTER_RESP_ERROR;
 			if (mle)
 				kmem_cache_free(dlm_mle_cache, mle);
@@ -2493,6 +2494,9 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 	 * the lockres
 	 */
 
+	/* now that remote nodes are spinning on the MIGRATING flag,
+	 * ensure that all assert_master work is flushed. */
+	flush_workqueue(dlm->dlm_worker);
 
 	/* get an extra reference on the mle.
 	 * otherwise the assert_master from the new
@@ -2547,7 +2551,8 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 			    res->owner == target)
 				break;
 
-			mlog(0, "timed out during migration\n");
+			mlog(0, "%s:%.*s: timed out during migration\n",
+			     dlm->name, res->lockname.len, res->lockname.name);
 			/* avoid hang during shutdown when migrating lockres 
 			 * to a node which also goes down */
 			if (dlm_is_node_dead(dlm, target)) {
@@ -2555,20 +2560,19 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 				     "target %u is no longer up, restarting\n",
 				     dlm->name, res->lockname.len,
 				     res->lockname.name, target);
-				ret = -ERESTARTSYS;
+				ret = -EINVAL;
+				/* migration failed, detach and clean up mle */
+				dlm_mle_detach_hb_events(dlm, mle);
+				dlm_put_mle(mle);
+				dlm_put_mle_inuse(mle);
+				spin_lock(&res->spinlock);
+				res->state &= ~DLM_LOCK_RES_MIGRATING;
+				spin_unlock(&res->spinlock);
+				goto leave;
 			}
-		}
-		if (ret == -ERESTARTSYS) {
-			/* migration failed, detach and clean up mle */
-			dlm_mle_detach_hb_events(dlm, mle);
-			dlm_put_mle(mle);
-			dlm_put_mle_inuse(mle);
-			spin_lock(&res->spinlock);
-			res->state &= ~DLM_LOCK_RES_MIGRATING;
-			spin_unlock(&res->spinlock);
-			goto leave;
-		}
-		/* TODO: if node died: stop, clean up, return error */
+		} else
+			mlog(0, "%s:%.*s: caught signal during migration\n",
+			     dlm->name, res->lockname.len, res->lockname.name);
 	}
 
 	/* all done, set the owner, clear the flag */