drbd: Implemented real timeout checking for request processing time

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>

drivers/block/drbd/drbd_int.h

@@ -990,6 +990,7 @@ struct drbd_conf {
 	struct timer_list resync_timer;
 	struct timer_list md_sync_timer;
 	struct timer_list start_resync_timer;
+	struct timer_list request_timer;
 #ifdef DRBD_DEBUG_MD_SYNC
 	struct {
 		unsigned int line;

drivers/block/drbd/drbd_main.c

@@ -3017,12 +3017,15 @@ void drbd_init_set_defaults(struct drbd_conf *mdev)
 	init_timer(&mdev->resync_timer);
 	init_timer(&mdev->md_sync_timer);
 	init_timer(&mdev->start_resync_timer);
+	init_timer(&mdev->request_timer);
 	mdev->resync_timer.function = resync_timer_fn;
 	mdev->resync_timer.data = (unsigned long) mdev;
 	mdev->md_sync_timer.function = md_sync_timer_fn;
 	mdev->md_sync_timer.data = (unsigned long) mdev;
 	mdev->start_resync_timer.function = start_resync_timer_fn;
 	mdev->start_resync_timer.data = (unsigned long) mdev;
+	mdev->request_timer.function = request_timer_fn;
+	mdev->request_timer.data = (unsigned long) mdev;
 
 	init_waitqueue_head(&mdev->misc_wait);
 	init_waitqueue_head(&mdev->state_wait);

drivers/block/drbd/drbd_receiver.c

@@ -912,6 +912,7 @@ static int drbd_connect(struct drbd_conf *mdev)
 	drbd_send_state(mdev);
 	clear_bit(USE_DEGR_WFC_T, &mdev->flags);
 	clear_bit(RESIZE_PENDING, &mdev->flags);
+	mod_timer(&mdev->request_timer, jiffies + HZ); /* just start it here. */
 
 	return 1;
 
@@ -3822,6 +3823,8 @@ static void drbd_disconnect(struct drbd_conf *mdev)
 	atomic_set(&mdev->rs_pending_cnt, 0);
 	wake_up(&mdev->misc_wait);
 
+	del_timer(&mdev->request_timer);
+
 	/* make sure syncer is stopped and w_resume_next_sg queued */
 	del_timer_sync(&mdev->resync_timer);
 	resync_timer_fn((unsigned long)mdev);

drivers/block/drbd/drbd_req.c

@@ -1194,3 +1194,42 @@ int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct
 	}
 	return limit;
 }
+
+void request_timer_fn(unsigned long data)
+{
+	struct drbd_conf *mdev = (struct drbd_conf *) data;
+	struct drbd_request *req; /* oldest request */
+	struct list_head *le;
+	unsigned long et = 0; /* effective timeout = ko_count * timeout */
+
+	if (get_net_conf(mdev)) {
+		et = mdev->net_conf->timeout*HZ/10 * mdev->net_conf->ko_count;
+		put_net_conf(mdev);
+	}
+	if (!et || mdev->state.conn < C_WF_REPORT_PARAMS)
+		return; /* Recurring timer stopped */
+
+	spin_lock_irq(&mdev->req_lock);
+	le = &mdev->oldest_tle->requests;
+	if (list_empty(le)) {
+		spin_unlock_irq(&mdev->req_lock);
+		mod_timer(&mdev->request_timer, jiffies + et);
+		return;
+	}
+
+	le = le->prev;
+	req = list_entry(le, struct drbd_request, tl_requests);
+	if (time_is_before_eq_jiffies(req->start_time + et)) {
+		if (req->rq_state & RQ_NET_PENDING) {
+			dev_warn(DEV, "Remote failed to finish a request within ko-count * timeout\n");
+			_drbd_set_state(_NS(mdev, conn, C_TIMEOUT), CS_VERBOSE, NULL);
+		} else {
+			dev_warn(DEV, "Local backing block device frozen?\n");
+			mod_timer(&mdev->request_timer, jiffies + et);
+		}
+	} else {
+		mod_timer(&mdev->request_timer, req->start_time + et);
+	}
+
+	spin_unlock_irq(&mdev->req_lock);
+}

drivers/block/drbd/drbd_req.h

@@ -322,6 +322,7 @@ extern int __req_mod(struct drbd_request *req, enum drbd_req_event what,
 		struct bio_and_error *m);
 extern void complete_master_bio(struct drbd_conf *mdev,
 		struct bio_and_error *m);
+extern void request_timer_fn(unsigned long data);
 
 /* use this if you don't want to deal with calling complete_master_bio()
  * outside the spinlock, e.g. when walking some list on cleanup. */