drbd: fix resync finished detection

This fixes one recent regresion,
and one long existing bug.

The bug:
drbd_try_clear_on_disk_bm() assumed that all "count" bits have to be
accounted in the resync extent corresponding to the start sector.

Since we allow application requests to cross our "extent" boundaries,
this assumption is no longer true, resulting in possible misaccounting,
scary messages
("BAD! sector=12345s enr=6 rs_left=-7 rs_failed=0 count=58 cstate=..."),
and potentially, if the last bit to be cleared during resync would
reside in previously misaccounted resync extent, the resync would never
be recognized as finished, but would be "stalled" forever, even though
all blocks are in sync again and all bits have been cleared...

The regression was introduced by
    drbd: get rid of atomic update on disk bitmap works

For an "empty" resync (rs_total == 0), we must not "finish" the
resync on the SyncSource before the SyncTarget knows all relevant
information (sync uuid).  We need to wait for the full round-trip,
the SyncTarget will then explicitly notify us.

Also for normal, non-empty resyncs (rs_total > 0), the resync-finished
condition needs to be tested before the schedule() in wait_for_work, or
it is likely to be missed.

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

1 files changed, 22 insertions, 20 deletions

diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 47bc84017b5b..bafb62eb22c9 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -1740,11 +1740,20 @@ void drbd_start_resync(struct drbd_device *device, enum drbd_conns side)
 			device->rs_mark_time[i] = now;
 		}
 		_drbd_pause_after(device);
+		/* Forget potentially stale cached per resync extent bit-counts.
+		 * Open coded drbd_rs_cancel_all(device), we already have IRQs
+		 * disabled, and know the disk state is ok. */
+		spin_lock(&device->al_lock);
+		lc_reset(device->resync);
+		device->resync_locked = 0;
+		device->resync_wenr = LC_FREE;
+		spin_unlock(&device->al_lock);
 	}
 	write_unlock(&global_state_lock);
 	spin_unlock_irq(&device->resource->req_lock);
 
 	if (r == SS_SUCCESS) {
+		wake_up(&device->al_wait); /* for lc_reset() above */
 		/* reset rs_last_bcast when a resync or verify is started,
 		 * to deal with potential jiffies wrap. */
 		device->rs_last_bcast = jiffies - HZ;
@@ -1807,36 +1816,22 @@ void drbd_start_resync(struct drbd_device *device, enum drbd_conns side)
 static void update_on_disk_bitmap(struct drbd_device *device)
 {
 	struct sib_info sib = { .sib_reason = SIB_SYNC_PROGRESS, };
+	bool resync_done = test_and_clear_bit(RS_DONE, &device->flags);
 	device->rs_last_bcast = jiffies;
 
 	if (!get_ldev(device))
 		return;
 
 	drbd_bm_write_lazy(device, 0);
-	if (drbd_bm_total_weight(device) <= device->rs_failed)
+	if (resync_done && is_sync_state(device->state.conn))
 		drbd_resync_finished(device);
+
 	drbd_bcast_event(device, &sib);
 	/* update timestamp, in case it took a while to write out stuff */
 	device->rs_last_bcast = jiffies;
 	put_ldev(device);
 }
 
-bool wants_lazy_bitmap_update(struct drbd_device *device)
-{
-	enum drbd_conns connection_state = device->state.conn;
-	return
-	/* only do a lazy writeout, if device is in some resync state */
-	   (connection_state == C_SYNC_SOURCE
-	||  connection_state == C_SYNC_TARGET
-	||  connection_state == C_PAUSED_SYNC_S
-	||  connection_state == C_PAUSED_SYNC_T) &&
-	/* AND
-	 * either we just finished, or the last lazy update
-	 * was some time ago already. */
-	   (drbd_bm_total_weight(device) <= device->rs_failed
-	||  time_after(jiffies, device->rs_last_bcast + 2*HZ));
-}
-
 static void try_update_all_on_disk_bitmaps(struct drbd_connection *connection)
 {
 	struct drbd_peer_device *peer_device;
@@ -1845,8 +1840,9 @@ static void try_update_all_on_disk_bitmaps(struct drbd_connection *connection)
 	rcu_read_lock();
 	idr_for_each_entry(&connection->peer_devices, peer_device, vnr) {
 		struct drbd_device *device = peer_device->device;
-		if (!wants_lazy_bitmap_update(device))
+		if (!test_and_clear_bit(RS_PROGRESS, &device->flags))
 			continue;
+
 		kref_get(&device->kref);
 		rcu_read_unlock();
 		update_on_disk_bitmap(device);
@@ -1930,15 +1926,18 @@ static void wait_for_work(struct drbd_connection *connection, struct list_head *
 		if (send_barrier)
 			maybe_send_barrier(connection,
 					connection->send.current_epoch_nr + 1);
+
+		if (test_bit(CONN_RS_PROGRESS, &connection->flags))
+			break;
+
 		/* drbd_send() may have called flush_signals() */
 		if (get_t_state(&connection->worker) != RUNNING)
 			break;
+
 		schedule();
 		/* may be woken up for other things but new work, too,
 		 * e.g. if the current epoch got closed.
 		 * In which case we send the barrier above. */
-
-		try_update_all_on_disk_bitmaps(connection);
 	}
 	finish_wait(&connection->sender_work.q_wait, &wait);
 
@@ -1973,6 +1972,9 @@ int drbd_worker(struct drbd_thread *thi)
 		if (list_empty(&work_list))
 			wait_for_work(connection, &work_list);
 
+		if (test_and_clear_bit(CONN_RS_PROGRESS, &connection->flags))
+			try_update_all_on_disk_bitmaps(connection);
+
 		if (signal_pending(current)) {
 			flush_signals(current);
 			if (get_t_state(thi) == RUNNING) {