drbd: Disable activity log updates when the whole device is out of sync

When the complete device is marked as out of sync, we can disable
updates of the on disk AL. Currently AL updates are only disabled
if one uses the "invalidate-remote" command on an unconnected,
primary device, or when at attach time all bits in the bitmap are
set.

As of now, AL updated do not get disabled when a all bits becomes
set due to application writes to an unconnected DRBD device.
While this is a missing feature, it is not considered important,
and might get added later.

BTW, after initializing a "one legged" DRBD device
drbdadm create-md resX
drbdadm -- --force primary resX
AL updates also get disabled, until the first connect.

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

drivers/block/drbd/drbd_bitmap.c

@@ -569,7 +569,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits)
  *
  * maybe bm_set should be atomic_t ?
  */
-static unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev)
+unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev)
 {
 	struct drbd_bitmap *b = mdev->bitmap;
 	unsigned long s;

drivers/block/drbd/drbd_int.h

@@ -863,6 +863,7 @@ enum {
 	CONN_DRY_RUN,		/* Expect disconnect after resync handshake. */
 	GOT_PING_ACK,		/* set when we receive a ping_ack packet, misc wait gets woken */
 	NEW_CUR_UUID,		/* Create new current UUID when thawing IO */
+	AL_SUSPENDED,		/* Activity logging is currently suspended. */
 };
 
 struct drbd_bitmap; /* opaque for drbd_conf */
@@ -1425,6 +1426,7 @@ extern unsigned long drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_
 /* bm_find_next variants for use while you hold drbd_bm_lock() */
 extern unsigned long _drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo);
 extern unsigned long _drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo);
+extern unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev);
 extern unsigned long drbd_bm_total_weight(struct drbd_conf *mdev);
 extern int drbd_bm_rs_done(struct drbd_conf *mdev);
 /* for receive_bitmap */

drivers/block/drbd/drbd_main.c

@@ -963,6 +963,12 @@ static void set_ov_position(struct drbd_conf *mdev, enum drbd_conns cs)
 	}
 }
 
+static void drbd_resume_al(struct drbd_conf *mdev)
+{
+	if (test_and_clear_bit(AL_SUSPENDED, &mdev->flags))
+		dev_info(DEV, "Resumed AL updates\n");
+}
+
 /**
  * __drbd_set_state() - Set a new DRBD state
  * @mdev:	DRBD device.
@@ -1160,6 +1166,10 @@ int __drbd_set_state(struct drbd_conf *mdev,
 	    ns.conn <= C_TEAR_DOWN && ns.conn >= C_TIMEOUT)
 		drbd_thread_restart_nowait(&mdev->receiver);
 
+	/* Resume AL writing if we get a connection */
+	if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED)
+		drbd_resume_al(mdev);
+
 	ascw = kmalloc(sizeof(*ascw), GFP_ATOMIC);
 	if (ascw) {
 		ascw->os = os;
@@ -2851,6 +2861,7 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev)
 	}
 
 	drbd_free_resources(mdev);
+	clear_bit(AL_SUSPENDED, &mdev->flags);
 
 	/*
 	 * currently we drbd_init_ee only on module load, so
@@ -3652,6 +3663,7 @@ int drbd_bmio_clear_n_write(struct drbd_conf *mdev)
 {
 	int rv = -EIO;
 
+	drbd_resume_al(mdev);
 	if (get_ldev_if_state(mdev, D_ATTACHING)) {
 		drbd_bm_clear_all(mdev);
 		rv = drbd_bm_write(mdev);

drivers/block/drbd/drbd_nl.c

@@ -777,6 +777,29 @@ static void drbd_reconfig_done(struct drbd_conf *mdev)
 	wake_up(&mdev->state_wait);
 }
 
+/* Make sure IO is suspended before calling this function(). */
+static void drbd_suspend_al(struct drbd_conf *mdev)
+{
+	int s = 0;
+
+	if (lc_try_lock(mdev->act_log)) {
+		drbd_al_shrink(mdev);
+		lc_unlock(mdev->act_log);
+	} else {
+		dev_warn(DEV, "Failed to lock al in drbd_suspend_al()\n");
+		return;
+	}
+
+	spin_lock_irq(&mdev->req_lock);
+	if (mdev->state.conn < C_CONNECTED)
+		s = !test_and_set_bit(AL_SUSPENDED, &mdev->flags);
+
+	spin_unlock_irq(&mdev->req_lock);
+
+	if (s)
+		dev_info(DEV, "Suspended AL updates\n");
+}
+
 /* does always return 0;
  * interesting return code is in reply->ret_code */
 static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp,
@@ -1113,6 +1136,9 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		drbd_al_to_on_disk_bm(mdev);
 	}
 
+	if (_drbd_bm_total_weight(mdev) == drbd_bm_bits(mdev))
+		drbd_suspend_al(mdev); /* IO is still suspended here... */
+
 	spin_lock_irq(&mdev->req_lock);
 	os = mdev->state;
 	ns.i = os.i;
@@ -1792,12 +1818,38 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl
 	return 0;
 }
 
+static int drbd_bmio_set_susp_al(struct drbd_conf *mdev)
+{
+	int rv;
+
+	rv = drbd_bmio_set_n_write(mdev);
+	drbd_suspend_al(mdev);
+	return rv;
+}
+
 static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp,
 				   struct drbd_nl_cfg_reply *reply)
 {
+	int retcode;
 
-	reply->ret_code = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S));
+	retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S), CS_ORDERED);
 
+	if (retcode < SS_SUCCESS) {
+		if (retcode == SS_NEED_CONNECTION && mdev->state.role == R_PRIMARY) {
+			/* The peer will get a resync upon connect anyways. Just make that
+			   into a full resync. */
+			retcode = drbd_request_state(mdev, NS(pdsk, D_INCONSISTENT));
+			if (retcode >= SS_SUCCESS) {
+				/* open coded drbd_bitmap_io() */
+				if (drbd_bitmap_io(mdev, &drbd_bmio_set_susp_al,
+						   "set_n_write from invalidate_peer"))
+					retcode = ERR_IO_MD_DISK;
+			}
+		} else
+			retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S));
+	}
+
+	reply->ret_code = retcode;
 	return 0;
 }
 

drivers/block/drbd/drbd_proc.c

@@ -203,7 +203,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
 			seq_printf(seq, "%2d: cs:Unconfigured\n", i);
 		} else {
 			seq_printf(seq,
-			   "%2d: cs:%s ro:%s/%s ds:%s/%s %c %c%c%c%c%c\n"
+			   "%2d: cs:%s ro:%s/%s ds:%s/%s %c %c%c%c%c%c%c\n"
 			   "    ns:%u nr:%u dw:%u dr:%u al:%u bm:%u "
 			   "lo:%d pe:%d ua:%d ap:%d ep:%d wo:%c",
 			   i, sn,
@@ -218,6 +218,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
 			   mdev->state.peer_isp ? 'p' : '-',
 			   mdev->state.user_isp ? 'u' : '-',
 			   mdev->congestion_reason ?: '-',
+			   test_bit(AL_SUSPENDED, &mdev->flags) ? 's' : '-',
 			   mdev->send_cnt/2,
 			   mdev->recv_cnt/2,
 			   mdev->writ_cnt/2,

drivers/block/drbd/drbd_req.c

@@ -94,7 +94,8 @@ static void _req_is_done(struct drbd_conf *mdev, struct drbd_request *req, const
 		 */
 		if (s & RQ_LOCAL_MASK) {
 			if (get_ldev_if_state(mdev, D_FAILED)) {
-				drbd_al_complete_io(mdev, req->sector);
+				if (s & RQ_IN_ACT_LOG)
+					drbd_al_complete_io(mdev, req->sector);
 				put_ldev(mdev);
 			} else if (__ratelimit(&drbd_ratelimit_state)) {
 				dev_warn(DEV, "Should have called drbd_al_complete_io(, %llu), "
@@ -802,8 +803,10 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio)
 	 * resync extent to finish, and, if necessary, pulls in the target
 	 * extent into the activity log, which involves further disk io because
 	 * of transactional on-disk meta data updates. */
-	if (rw == WRITE && local)
+	if (rw == WRITE && local && !test_bit(AL_SUSPENDED, &mdev->flags)) {
+		req->rq_state |= RQ_IN_ACT_LOG;
 		drbd_al_begin_io(mdev, sector);
+	}
 
 	remote = remote && (mdev->state.pdsk == D_UP_TO_DATE ||
 			    (mdev->state.pdsk == D_INCONSISTENT &&

drivers/block/drbd/drbd_req.h

@@ -189,6 +189,9 @@ enum drbd_req_state_bits {
 
 	/* Set when this is a write, clear for a read */
 	__RQ_WRITE,
+
+	/* Should call drbd_al_complete_io() for this request... */
+	__RQ_IN_ACT_LOG,
 };
 
 #define RQ_LOCAL_PENDING   (1UL << __RQ_LOCAL_PENDING)
@@ -208,6 +211,7 @@ enum drbd_req_state_bits {
 #define RQ_NET_MASK        (((1UL << __RQ_NET_MAX)-1) & ~RQ_LOCAL_MASK)
 
 #define RQ_WRITE           (1UL << __RQ_WRITE)
+#define RQ_IN_ACT_LOG      (1UL << __RQ_IN_ACT_LOG)
 
 /* For waking up the frozen transfer log mod_req() has to return if the request
    should be counted in the epoch object*/

drivers/block/drbd/drbd_worker.c

@@ -1273,7 +1273,7 @@ int w_restart_disk_io(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
 {
 	struct drbd_request *req = container_of(w, struct drbd_request, w);
 
-	if (bio_data_dir(req->master_bio) == WRITE)
+	if (bio_data_dir(req->master_bio) == WRITE && req->rq_state & RQ_IN_ACT_LOG)
 		drbd_al_begin_io(mdev, req->sector);
 	/* Calling drbd_al_begin_io() out of the worker might deadlocks
 	   theoretically. Practically it can not deadlock, since this is