drbd: Track the reasons to suspend IO in dedicated state bits

There are three ways to get IO suspended:

 * Loss of any access to data
 * Fence-peer-handler running
 * User requested to suspend IO

Track those in different bits, so that one condition clearing its
state bit does not interfere with the other two conditions.

Only when the user resumes IO he overrules all three bits.

The fact is hidden from the user, he sees only a single suspend
bit.
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>

drivers/block/drbd/drbd_int.h

@@ -1681,6 +1681,8 @@ void drbd_bcast_ee(struct drbd_conf *mdev,
 #define susp_MASK 1
 #define user_isp_MASK 1
 #define aftr_isp_MASK 1
+#define susp_nod_MASK 1
+#define susp_fen_MASK 1
 
 #define NS(T, S) \
 	({ union drbd_state mask; mask.i = 0; mask.T = T##_MASK; mask; }), \
@@ -2254,11 +2256,16 @@ static inline int drbd_state_is_stable(union drbd_state s)
 	return 1;
 }
 
+static inline int is_susp(union drbd_state s)
+{
+	return s.susp || s.susp_nod || s.susp_fen;
+}
+
 static inline int __inc_ap_bio_cond(struct drbd_conf *mdev)
 {
 	int mxb = drbd_get_max_buffers(mdev);
 
-	if (mdev->state.susp)
+	if (is_susp(mdev->state))
 		return 0;
 	if (test_bit(SUSPEND_IO, &mdev->flags))
 		return 0;

drivers/block/drbd/drbd_main.c

@@ -654,7 +654,7 @@ static void print_st(struct drbd_conf *mdev, char *name, union drbd_state ns)
 	    drbd_role_str(ns.peer),
 	    drbd_disk_str(ns.disk),
 	    drbd_disk_str(ns.pdsk),
-	    ns.susp ? 's' : 'r',
+	    is_susp(ns) ? 's' : 'r',
 	    ns.aftr_isp ? 'a' : '-',
 	    ns.peer_isp ? 'p' : '-',
 	    ns.user_isp ? 'u' : '-'
@@ -925,12 +925,12 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state
 	if (fp == FP_STONITH &&
 	    (ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.pdsk > D_OUTDATED) &&
 	    !(os.role == R_PRIMARY && os.conn < C_CONNECTED && os.pdsk > D_OUTDATED))
-		ns.susp = 1; /* Suspend IO while fence-peer handler runs (peer lost) */
+		ns.susp_fen = 1; /* Suspend IO while fence-peer handler runs (peer lost) */
 
 	if (mdev->sync_conf.on_no_data == OND_SUSPEND_IO &&
 	    (ns.role == R_PRIMARY && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE) &&
 	    !(os.role == R_PRIMARY && os.disk < D_UP_TO_DATE && os.pdsk < D_UP_TO_DATE))
-		ns.susp = 1; /* Suspend IO while no data available (no accessible data available) */
+		ns.susp_nod = 1; /* Suspend IO while no data available (no accessible data available) */
 
 	if (ns.aftr_isp || ns.peer_isp || ns.user_isp) {
 		if (ns.conn == C_SYNC_SOURCE)
@@ -1030,7 +1030,10 @@ int __drbd_set_state(struct drbd_conf *mdev,
 		PSC(conn);
 		PSC(disk);
 		PSC(pdsk);
-		PSC(susp);
+		if (is_susp(ns) != is_susp(os))
+			pbp += sprintf(pbp, "susp( %s -> %s ) ",
+				       drbd_susp_str(is_susp(os)),
+				       drbd_susp_str(is_susp(ns)));
 		PSC(aftr_isp);
 		PSC(peer_isp);
 		PSC(user_isp);
@@ -1218,6 +1221,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 {
 	enum drbd_fencing_p fp;
 	enum drbd_req_event what = nothing;
+	union drbd_state nsm = (union drbd_state){ .i = -1 };
 
 	if (os.conn != C_CONNECTED && ns.conn == C_CONNECTED) {
 		clear_bit(CRASHED_PRIMARY, &mdev->flags);
@@ -1241,19 +1245,21 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 	/* Here we have the actions that are performed after a
 	   state change. This function might sleep */
 
-	if (os.susp && ns.susp && mdev->sync_conf.on_no_data == OND_SUSPEND_IO) {
+	nsm.i = -1;
+	if (ns.susp_nod) {
 		if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) {
 			if (ns.conn == C_CONNECTED)
-				what = resend;
+				what = resend, nsm.susp_nod = 0;
 			else /* ns.conn > C_CONNECTED */
 				dev_err(DEV, "Unexpected Resynd going on!\n");
 		}
 
 		if (os.disk == D_ATTACHING && ns.disk > D_ATTACHING)
-			what = restart_frozen_disk_io;
+			what = restart_frozen_disk_io, nsm.susp_nod = 0;
+
 	}
 
-	if (fp == FP_STONITH && ns.susp) {
+	if (ns.susp_fen) {
 		/* case1: The outdate peer handler is successful: */
 		if (os.pdsk > D_OUTDATED  && ns.pdsk <= D_OUTDATED) {
 			tl_clear(mdev);
@@ -1263,20 +1269,22 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 				drbd_md_sync(mdev);
 			}
 			spin_lock_irq(&mdev->req_lock);
-			_drbd_set_state(_NS(mdev, susp, 0), CS_VERBOSE, NULL);
+			_drbd_set_state(_NS(mdev, susp_fen, 0), CS_VERBOSE, NULL);
 			spin_unlock_irq(&mdev->req_lock);
 		}
 		/* case2: The connection was established again: */
 		if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) {
 			clear_bit(NEW_CUR_UUID, &mdev->flags);
 			what = resend;
+			nsm.susp_fen = 0;
 		}
 	}
 
 	if (what != nothing) {
 		spin_lock_irq(&mdev->req_lock);
 		_tl_restart(mdev, what);
-		_drbd_set_state(_NS(mdev, susp, 0), CS_VERBOSE, NULL);
+		nsm.i &= mdev->state.i;
+		_drbd_set_state(mdev, nsm, CS_VERBOSE, NULL);
 		spin_unlock_irq(&mdev->req_lock);
 	}
 
@@ -1298,7 +1306,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 		if (get_ldev(mdev)) {
 			if ((ns.role == R_PRIMARY || ns.peer == R_PRIMARY) &&
 			    mdev->ldev->md.uuid[UI_BITMAP] == 0 && ns.disk >= D_UP_TO_DATE) {
-				if (mdev->state.susp) {
+				if (is_susp(mdev->state)) {
 					set_bit(NEW_CUR_UUID, &mdev->flags);
 				} else {
 					drbd_uuid_new_current(mdev);
@@ -1417,7 +1425,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 		resume_next_sg(mdev);
 
 	/* free tl_hash if we Got thawed and are C_STANDALONE */
-	if (ns.conn == C_STANDALONE && ns.susp == 0 && mdev->tl_hash)
+	if (ns.conn == C_STANDALONE && !is_susp(ns) && mdev->tl_hash)
 		drbd_free_tl_hash(mdev);
 
 	/* Upon network connection, we need to start the receiver */
@@ -2732,7 +2740,9 @@ static void drbd_set_defaults(struct drbd_conf *mdev)
 		  .conn = C_STANDALONE,
 		  .disk = D_DISKLESS,
 		  .pdsk = D_UNKNOWN,
-		  .susp = 0
+		  .susp = 0,
+		  .susp_nod = 0,
+		  .susp_fen = 0
 		} };
 }
 

drivers/block/drbd/drbd_nl.c

@@ -209,7 +209,8 @@ enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev)
 		put_ldev(mdev);
 	} else {
 		dev_warn(DEV, "Not fencing peer, I'm not even Consistent myself.\n");
-		return mdev->state.pdsk;
+		nps = mdev->state.pdsk;
+		goto out;
 	}
 
 	r = drbd_khelper(mdev, "fence-peer");
@@ -256,6 +257,14 @@ enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev)
 
 	dev_info(DEV, "fence-peer helper returned %d (%s)\n",
 			(r>>8) & 0xff, ex_to_string);
+
+out:
+	if (mdev->state.susp_fen && nps >= D_UNKNOWN) {
+		/* The handler was not successful... unfreeze here, the
+		   state engine can not unfreeze... */
+		_drbd_request_state(mdev, NS(susp_fen, 0), CS_VERBOSE);
+	}
+
 	return nps;
 }
 
@@ -550,7 +559,7 @@ char *ppsize(char *buf, unsigned long long size)
 void drbd_suspend_io(struct drbd_conf *mdev)
 {
 	set_bit(SUSPEND_IO, &mdev->flags);
-	if (mdev->state.susp)
+	if (is_susp(mdev->state))
 		return;
 	wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt));
 }
@@ -1016,7 +1025,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 
 	drbd_suspend_io(mdev);
 	/* also wait for the last barrier ack. */
-	wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_pending_cnt) || mdev->state.susp);
+	wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_pending_cnt) || is_susp(mdev->state));
 	/* and for any other previously queued work */
 	drbd_flush_workqueue(mdev);
 
@@ -1114,8 +1123,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		clear_bit(CRASHED_PRIMARY, &mdev->flags);
 
 	if (drbd_md_test_flag(mdev->ldev, MDF_PRIMARY_IND) &&
-	    !(mdev->state.role == R_PRIMARY && mdev->state.susp &&
-	      mdev->sync_conf.on_no_data == OND_SUSPEND_IO)) {
+	    !(mdev->state.role == R_PRIMARY && mdev->state.susp_nod)) {
 		set_bit(CRASHED_PRIMARY, &mdev->flags);
 		cp_discovered = 1;
 	}
@@ -1939,7 +1947,7 @@ static int drbd_nl_resume_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		drbd_md_sync(mdev);
 	}
 	drbd_suspend_io(mdev);
-	reply->ret_code = drbd_request_state(mdev, NS(susp, 0));
+	reply->ret_code = drbd_request_state(mdev, NS3(susp, 0, susp_nod, 0, susp_fen, 0));
 	if (reply->ret_code == SS_SUCCESS) {
 		if (mdev->state.conn < C_CONNECTED)
 			tl_clear(mdev);

drivers/block/drbd/drbd_proc.c

@@ -213,7 +213,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
 			   drbd_disk_str(mdev->state.pdsk),
 			   (mdev->net_conf == NULL ? ' ' :
 			    (mdev->net_conf->wire_protocol - DRBD_PROT_A+'A')),
-			   mdev->state.susp ? 's' : 'r',
+			   is_susp(mdev->state) ? 's' : 'r',
 			   mdev->state.aftr_isp ? 'a' : '-',
 			   mdev->state.peer_isp ? 'p' : '-',
 			   mdev->state.user_isp ? 'u' : '-',

drivers/block/drbd/drbd_receiver.c

@@ -3315,7 +3315,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned
 	if ((nconn == C_CONNECTED || nconn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
 		ns.disk = mdev->new_state_tmp.disk;
 	cs_flags = CS_VERBOSE + (oconn < C_CONNECTED && nconn >= C_CONNECTED ? 0 : CS_HARD);
-	if (ns.pdsk == D_CONSISTENT && ns.susp && nconn == C_CONNECTED && oconn < C_CONNECTED &&
+	if (ns.pdsk == D_CONSISTENT && is_susp(ns) && nconn == C_CONNECTED && oconn < C_CONNECTED &&
 	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
 		/* Do not allow tl_restart(resend) for a rebooted peer. We can only allow this
 		   for temporal network outages! */
@@ -3829,7 +3829,7 @@ static void drbd_disconnect(struct drbd_conf *mdev)
 	kfree(mdev->p_uuid);
 	mdev->p_uuid = NULL;
 
-	if (!mdev->state.susp)
+	if (!is_susp(mdev->state))
 		tl_clear(mdev);
 
 	dev_info(DEV, "Connection closed\n");
@@ -3858,7 +3858,7 @@ static void drbd_disconnect(struct drbd_conf *mdev)
 	if (os.conn == C_DISCONNECTING) {
 		wait_event(mdev->net_cnt_wait, atomic_read(&mdev->net_cnt) == 0);
 
-		if (!mdev->state.susp) {
+		if (!is_susp(mdev->state)) {
 			/* we must not free the tl_hash
 			 * while application io is still on the fly */
 			wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt));

drivers/block/drbd/drbd_req.c

@@ -287,7 +287,7 @@ static void _req_may_be_done_not_susp(struct drbd_request *req, struct bio_and_e
 {
 	struct drbd_conf *mdev = req->mdev;
 
-	if (!mdev->state.susp)
+	if (!is_susp(mdev->state))
 		_req_may_be_done(req, m);
 }
 
@@ -812,7 +812,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio)
 			    (mdev->state.pdsk == D_INCONSISTENT &&
 			     mdev->state.conn >= C_CONNECTED));
 
-	if (!(local || remote) && !mdev->state.susp) {
+	if (!(local || remote) && !is_susp(mdev->state)) {
 		dev_err(DEV, "IO ERROR: neither local nor remote disk\n");
 		goto fail_free_complete;
 	}
@@ -838,7 +838,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio)
 	/* GOOD, everything prepared, grab the spin_lock */
 	spin_lock_irq(&mdev->req_lock);
 
-	if (mdev->state.susp) {
+	if (is_susp(mdev->state)) {
 		/* If we got suspended, use the retry mechanism of
 		   generic_make_request() to restart processing of this
 		   bio. In the next call to drbd_make_request_26

include/linux/drbd.h

@@ -232,13 +232,17 @@ union drbd_state {
 		unsigned conn:5 ;   /* 17/32	 cstates */
 		unsigned disk:4 ;   /* 8/16	 from D_DISKLESS to D_UP_TO_DATE */
 		unsigned pdsk:4 ;   /* 8/16	 from D_DISKLESS to D_UP_TO_DATE */
-		unsigned susp:1 ;   /* 2/2	 IO suspended  no/yes */
+		unsigned susp:1 ;   /* 2/2	 IO suspended no/yes (by user) */
 		unsigned aftr_isp:1 ; /* isp .. imposed sync pause */
 		unsigned peer_isp:1 ;
 		unsigned user_isp:1 ;
-		unsigned _pad:11;   /* 0	 unused */
+		unsigned susp_nod:1 ; /* IO suspended because no data */
+		unsigned susp_fen:1 ; /* IO suspended because fence peer handler runs*/
+		unsigned _pad:9;   /* 0	 unused */
 #elif defined(__BIG_ENDIAN_BITFIELD)
-		unsigned _pad:11;   /* 0	 unused */
+		unsigned _pad:9;
+		unsigned susp_fen:1 ;
+		unsigned susp_nod:1 ;
 		unsigned user_isp:1 ;
 		unsigned peer_isp:1 ;
 		unsigned aftr_isp:1 ; /* isp .. imposed sync pause */