5 files changed, 195 insertions, 152 deletions

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index d5e0906..903d581 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -56,7 +56,7 @@ MODULE_PARM_DESC(max_retries, "max number of retries a command may have");
 static int nvme_char_major;
 module_param(nvme_char_major, int, 0);
 
-static unsigned long default_ps_max_latency_us = 25000;
+static unsigned long default_ps_max_latency_us = 100000;
 module_param(default_ps_max_latency_us, ulong, 0644);
 MODULE_PARM_DESC(default_ps_max_latency_us,
 		 "max power saving latency for new devices; use PM QOS to change per device");
@@ -925,6 +925,29 @@ static int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 }
 
 #ifdef CONFIG_BLK_DEV_INTEGRITY
+static void nvme_prep_integrity(struct gendisk *disk, struct nvme_id_ns *id,
+		u16 bs)
+{
+	struct nvme_ns *ns = disk->private_data;
+	u16 old_ms = ns->ms;
+	u8 pi_type = 0;
+
+	ns->ms = le16_to_cpu(id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ms);
+	ns->ext = ns->ms && (id->flbas & NVME_NS_FLBAS_META_EXT);
+
+	/* PI implementation requires metadata equal t10 pi tuple size */
+	if (ns->ms == sizeof(struct t10_pi_tuple))
+		pi_type = id->dps & NVME_NS_DPS_PI_MASK;
+
+	if (blk_get_integrity(disk) &&
+	    (ns->pi_type != pi_type || ns->ms != old_ms ||
+	     bs != queue_logical_block_size(disk->queue) ||
+	     (ns->ms && ns->ext)))
+		blk_integrity_unregister(disk);
+
+	ns->pi_type = pi_type;
+}
+
 static void nvme_init_integrity(struct nvme_ns *ns)
 {
 	struct blk_integrity integrity;
@@ -951,6 +974,10 @@ static void nvme_init_integrity(struct nvme_ns *ns)
 	blk_queue_max_integrity_segments(ns->queue, 1);
 }
 #else
+static void nvme_prep_integrity(struct gendisk *disk, struct nvme_id_ns *id,
+		u16 bs)
+{
+}
 static void nvme_init_integrity(struct nvme_ns *ns)
 {
 }
@@ -997,37 +1024,22 @@ static int nvme_revalidate_ns(struct nvme_ns *ns, struct nvme_id_ns **id)
 static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id)
 {
 	struct nvme_ns *ns = disk->private_data;
-	u8 lbaf, pi_type;
-	u16 old_ms;
-	unsigned short bs;
-
-	old_ms = ns->ms;
-	lbaf = id->flbas & NVME_NS_FLBAS_LBA_MASK;
-	ns->lba_shift = id->lbaf[lbaf].ds;
-	ns->ms = le16_to_cpu(id->lbaf[lbaf].ms);
-	ns->ext = ns->ms && (id->flbas & NVME_NS_FLBAS_META_EXT);
+	u16 bs;
 
 	/*
 	 * If identify namespace failed, use default 512 byte block size so
 	 * block layer can use before failing read/write for 0 capacity.
 	 */
+	ns->lba_shift = id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ds;
 	if (ns->lba_shift == 0)
 		ns->lba_shift = 9;
 	bs = 1 << ns->lba_shift;
-	/* XXX: PI implementation requires metadata equal t10 pi tuple size */
-	pi_type = ns->ms == sizeof(struct t10_pi_tuple) ?
-					id->dps & NVME_NS_DPS_PI_MASK : 0;
 
 	blk_mq_freeze_queue(disk->queue);
-	if (blk_get_integrity(disk) && (ns->pi_type != pi_type ||
-				ns->ms != old_ms ||
-				bs != queue_logical_block_size(disk->queue) ||
-				(ns->ms && ns->ext)))
-		blk_integrity_unregister(disk);
 
-	ns->pi_type = pi_type;
+	if (ns->ctrl->ops->flags & NVME_F_METADATA_SUPPORTED)
+		nvme_prep_integrity(disk, id, bs);
 	blk_queue_logical_block_size(ns->queue, bs);
-
 	if (ns->ms && !blk_get_integrity(disk) && !ns->ext)
 		nvme_init_integrity(ns);
 	if (ns->ms && !(ns->ms == 8 && ns->pi_type) && !blk_get_integrity(disk))
@@ -1330,7 +1342,7 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 	 * transitioning between power states.  Therefore, when running
 	 * in any given state, we will enter the next lower-power
 	 * non-operational state after waiting 50 * (enlat + exlat)
-	 * microseconds, as long as that state's total latency is under
+	 * microseconds, as long as that state's exit latency is under
 	 * the requested maximum latency.
 	 *
 	 * We will not autonomously enter any non-operational state for
@@ -1375,7 +1387,7 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 		 * lowest-power state, not the number of states.
 		 */
 		for (state = (int)ctrl->npss; state >= 0; state--) {
-			u64 total_latency_us, transition_ms;
+			u64 total_latency_us, exit_latency_us, transition_ms;
 
 			if (target)
 				table->entries[state] = target;
@@ -1396,12 +1408,15 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 			      NVME_PS_FLAGS_NON_OP_STATE))
 				continue;
 
-			total_latency_us =
-				(u64)le32_to_cpu(ctrl->psd[state].entry_lat) +
-				+ le32_to_cpu(ctrl->psd[state].exit_lat);
-			if (total_latency_us > ctrl->ps_max_latency_us)
+			exit_latency_us =
+				(u64)le32_to_cpu(ctrl->psd[state].exit_lat);
+			if (exit_latency_us > ctrl->ps_max_latency_us)
 				continue;
 
+			total_latency_us =
+				exit_latency_us +
+				le32_to_cpu(ctrl->psd[state].entry_lat);
+
 			/*
 			 * This state is good.  Use it as the APST idle
 			 * target for higher power states.
@@ -1605,7 +1620,7 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 	}
 	memcpy(ctrl->psd, id->psd, sizeof(ctrl->psd));
 
-	if (ctrl->ops->is_fabrics) {
+	if (ctrl->ops->flags & NVME_F_FABRICS) {
 		ctrl->icdoff = le16_to_cpu(id->icdoff);
 		ctrl->ioccsz = le32_to_cpu(id->ioccsz);
 		ctrl->iorcsz = le32_to_cpu(id->iorcsz);
@@ -2098,7 +2113,6 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 		if (ns->ndev)
 			nvme_nvm_unregister_sysfs(ns);
 		del_gendisk(ns->disk);
-		blk_mq_abort_requeue_list(ns->queue);
 		blk_cleanup_queue(ns->queue);
 	}
 
@@ -2427,6 +2441,10 @@ void nvme_kill_queues(struct nvme_ctrl *ctrl)
 	struct nvme_ns *ns;
 
 	mutex_lock(&ctrl->namespaces_mutex);
+
+	/* Forcibly start all queues to avoid having stuck requests */
+	blk_mq_start_hw_queues(ctrl->admin_q);
+
 	list_for_each_entry(ns, &ctrl->namespaces, list) {
 		/*
 		 * Revalidating a dead namespace sets capacity to 0. This will
@@ -2436,8 +2454,16 @@ void nvme_kill_queues(struct nvme_ctrl *ctrl)
 			continue;
 		revalidate_disk(ns->disk);
 		blk_set_queue_dying(ns->queue);
-		blk_mq_abort_requeue_list(ns->queue);
-		blk_mq_start_stopped_hw_queues(ns->queue, true);
+
+		/*
+		 * Forcibly start all queues to avoid having stuck requests.
+		 * Note that we must ensure the queues are not stopped
+		 * when the final removal happens.
+		 */
+		blk_mq_start_hw_queues(ns->queue);
+
+		/* draining requests in requeue list */
+		blk_mq_kick_requeue_list(ns->queue);
 	}
 	mutex_unlock(&ctrl->namespaces_mutex);
 }
diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c
index dca7165..92964ce 100644
--- a/drivers/nvme/host/fc.c
+++ b/drivers/nvme/host/fc.c
@@ -45,8 +45,6 @@ enum nvme_fc_queue_flags {
 
 #define NVMEFC_QUEUE_DELAY	3		/* ms units */
 
-#define NVME_FC_MAX_CONNECT_ATTEMPTS	1
-
 struct nvme_fc_queue {
 	struct nvme_fc_ctrl	*ctrl;
 	struct device		*dev;
@@ -165,8 +163,6 @@ struct nvme_fc_ctrl {
 	struct work_struct	delete_work;
 	struct work_struct	reset_work;
 	struct delayed_work	connect_work;
-	int			reconnect_delay;
-	int			connect_attempts;
 
 	struct kref		ref;
 	u32			flags;
@@ -1143,6 +1139,7 @@ nvme_fc_xmt_disconnect_assoc(struct nvme_fc_ctrl *ctrl)
 /* *********************** NVME Ctrl Routines **************************** */
 
 static void __nvme_fc_final_op_cleanup(struct request *rq);
+static void nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg);
 
 static int
 nvme_fc_reinit_request(void *data, struct request *rq)
@@ -1269,7 +1266,7 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	struct nvme_command *sqe = &op->cmd_iu.sqe;
 	__le16 status = cpu_to_le16(NVME_SC_SUCCESS << 1);
 	union nvme_result result;
-	bool complete_rq;
+	bool complete_rq, terminate_assoc = true;
 
 	/*
 	 * WARNING:
@@ -1298,6 +1295,14 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	 * fabricate a CQE, the following fields will not be set as they
 	 * are not referenced:
 	 *      cqe.sqid,  cqe.sqhd,  cqe.command_id
+	 *
+	 * Failure or error of an individual i/o, in a transport
+	 * detected fashion unrelated to the nvme completion status,
+	 * potentially cause the initiator and target sides to get out
+	 * of sync on SQ head/tail (aka outstanding io count allowed).
+	 * Per FC-NVME spec, failure of an individual command requires
+	 * the connection to be terminated, which in turn requires the
+	 * association to be terminated.
 	 */
 
 	fc_dma_sync_single_for_cpu(ctrl->lport->dev, op->fcp_req.rspdma,
@@ -1363,6 +1368,8 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 		goto done;
 	}
 
+	terminate_assoc = false;
+
 done:
 	if (op->flags & FCOP_FLAGS_AEN) {
 		nvme_complete_async_event(&queue->ctrl->ctrl, status, &result);
@@ -1370,19 +1377,23 @@ done:
 		atomic_set(&op->state, FCPOP_STATE_IDLE);
 		op->flags = FCOP_FLAGS_AEN;	/* clear other flags */
 		nvme_fc_ctrl_put(ctrl);
-		return;
+		goto check_error;
 	}
 
 	complete_rq = __nvme_fc_fcpop_chk_teardowns(ctrl, op);
 	if (!complete_rq) {
 		if (unlikely(op->flags & FCOP_FLAGS_TERMIO)) {
-			status = cpu_to_le16(NVME_SC_ABORT_REQ);
+			status = cpu_to_le16(NVME_SC_ABORT_REQ << 1);
 			if (blk_queue_dying(rq->q))
-				status |= cpu_to_le16(NVME_SC_DNR);
+				status |= cpu_to_le16(NVME_SC_DNR << 1);
 		}
 		nvme_end_request(rq, status, result);
 	} else
 		__nvme_fc_final_op_cleanup(rq);
+
+check_error:
+	if (terminate_assoc)
+		nvme_fc_error_recovery(ctrl, "transport detected io error");
 }
 
 static int
@@ -1751,7 +1762,7 @@ nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg)
 	dev_warn(ctrl->ctrl.device,
 		"NVME-FC{%d}: transport association error detected: %s\n",
 		ctrl->cnum, errmsg);
-	dev_info(ctrl->ctrl.device,
+	dev_warn(ctrl->ctrl.device,
 		"NVME-FC{%d}: resetting controller\n", ctrl->cnum);
 
 	/* stop the queues on error, cleanup is in reset thread */
@@ -2195,9 +2206,6 @@ nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
 	if (!opts->nr_io_queues)
 		return 0;
 
-	dev_info(ctrl->ctrl.device, "creating %d I/O queues.\n",
-			opts->nr_io_queues);
-
 	nvme_fc_init_io_queues(ctrl);
 
 	memset(&ctrl->tag_set, 0, sizeof(ctrl->tag_set));
@@ -2268,9 +2276,6 @@ nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 	if (ctrl->queue_count == 1)
 		return 0;
 
-	dev_info(ctrl->ctrl.device, "Recreating %d I/O queues.\n",
-			opts->nr_io_queues);
-
 	nvme_fc_init_io_queues(ctrl);
 
 	ret = blk_mq_reinit_tagset(&ctrl->tag_set);
@@ -2306,7 +2311,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	int ret;
 	bool changed;
 
-	ctrl->connect_attempts++;
+	++ctrl->ctrl.opts->nr_reconnects;
 
 	/*
 	 * Create the admin queue
@@ -2403,9 +2408,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 	WARN_ON_ONCE(!changed);
 
-	ctrl->connect_attempts = 0;
-
-	kref_get(&ctrl->ctrl.kref);
+	ctrl->ctrl.opts->nr_reconnects = 0;
 
 	if (ctrl->queue_count > 1) {
 		nvme_start_queues(&ctrl->ctrl);
@@ -2536,26 +2539,32 @@ nvme_fc_delete_ctrl_work(struct work_struct *work)
 
 	/*
 	 * tear down the controller
-	 * This will result in the last reference on the nvme ctrl to
-	 * expire, calling the transport nvme_fc_nvme_ctrl_freed() callback.
-	 * From there, the transport will tear down it's logical queues and
-	 * association.
+	 * After the last reference on the nvme ctrl is removed,
+	 * the transport nvme_fc_nvme_ctrl_freed() callback will be
+	 * invoked. From there, the transport will tear down it's
+	 * logical queues and association.
 	 */
 	nvme_uninit_ctrl(&ctrl->ctrl);
 
 	nvme_put_ctrl(&ctrl->ctrl);
 }
 
-static int
-__nvme_fc_del_ctrl(struct nvme_fc_ctrl *ctrl)
+static bool
+__nvme_fc_schedule_delete_work(struct nvme_fc_ctrl *ctrl)
 {
 	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING))
-		return -EBUSY;
+		return true;
 
 	if (!queue_work(nvme_fc_wq, &ctrl->delete_work))
-		return -EBUSY;
+		return true;
 
-	return 0;
+	return false;
+}
+
+static int
+__nvme_fc_del_ctrl(struct nvme_fc_ctrl *ctrl)
+{
+	return __nvme_fc_schedule_delete_work(ctrl) ? -EBUSY : 0;
 }
 
 /*
@@ -2581,6 +2590,35 @@ nvme_fc_del_nvme_ctrl(struct nvme_ctrl *nctrl)
 }
 
 static void
+nvme_fc_reconnect_or_delete(struct nvme_fc_ctrl *ctrl, int status)
+{
+	/* If we are resetting/deleting then do nothing */
+	if (ctrl->ctrl.state != NVME_CTRL_RECONNECTING) {
+		WARN_ON_ONCE(ctrl->ctrl.state == NVME_CTRL_NEW ||
+			ctrl->ctrl.state == NVME_CTRL_LIVE);
+		return;
+	}
+
+	dev_info(ctrl->ctrl.device,
+		"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
+		ctrl->cnum, status);
+
+	if (nvmf_should_reconnect(&ctrl->ctrl)) {
+		dev_info(ctrl->ctrl.device,
+			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
+			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
+		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
+				ctrl->ctrl.opts->reconnect_delay * HZ);
+	} else {
+		dev_warn(ctrl->ctrl.device,
+				"NVME-FC{%d}: Max reconnect attempts (%d) "
+				"reached. Removing controller\n",
+				ctrl->cnum, ctrl->ctrl.opts->nr_reconnects);
+		WARN_ON(__nvme_fc_schedule_delete_work(ctrl));
+	}
+}
+
+static void
 nvme_fc_reset_ctrl_work(struct work_struct *work)
 {
 	struct nvme_fc_ctrl *ctrl =
@@ -2591,34 +2629,9 @@ nvme_fc_reset_ctrl_work(struct work_struct *work)
 	nvme_fc_delete_association(ctrl);
 
 	ret = nvme_fc_create_association(ctrl);
-	if (ret) {
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
-			ctrl->cnum, ret);
-		if (ctrl->connect_attempts >= NVME_FC_MAX_CONNECT_ATTEMPTS) {
-			dev_warn(ctrl->ctrl.device,
-				"NVME-FC{%d}: Max reconnect attempts (%d) "
-				"reached. Removing controller\n",
-				ctrl->cnum, ctrl->connect_attempts);
-
-			if (!nvme_change_ctrl_state(&ctrl->ctrl,
-				NVME_CTRL_DELETING)) {
-				dev_err(ctrl->ctrl.device,
-					"NVME-FC{%d}: failed to change state "
-					"to DELETING\n", ctrl->cnum);
-				return;
-			}
-
-			WARN_ON(!queue_work(nvme_fc_wq, &ctrl->delete_work));
-			return;
-		}
-
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->reconnect_delay);
-		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->reconnect_delay * HZ);
-	} else
+	if (ret)
+		nvme_fc_reconnect_or_delete(ctrl, ret);
+	else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reset complete\n", ctrl->cnum);
 }
@@ -2632,7 +2645,7 @@ nvme_fc_reset_nvme_ctrl(struct nvme_ctrl *nctrl)
 {
 	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);
 
-	dev_warn(ctrl->ctrl.device,
+	dev_info(ctrl->ctrl.device,
 		"NVME-FC{%d}: admin requested controller reset\n", ctrl->cnum);
 
 	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_RESETTING))
@@ -2649,7 +2662,7 @@ nvme_fc_reset_nvme_ctrl(struct nvme_ctrl *nctrl)
 static const struct nvme_ctrl_ops nvme_fc_ctrl_ops = {
 	.name			= "fc",
 	.module			= THIS_MODULE,
-	.is_fabrics		= true,
+	.flags			= NVME_F_FABRICS,
 	.reg_read32		= nvmf_reg_read32,
 	.reg_read64		= nvmf_reg_read64,
 	.reg_write32		= nvmf_reg_write32,
@@ -2671,34 +2684,9 @@ nvme_fc_connect_ctrl_work(struct work_struct *work)
 				struct nvme_fc_ctrl, connect_work);
 
 	ret = nvme_fc_create_association(ctrl);
-	if (ret) {
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt failed (%d)\n",
-			ctrl->cnum, ret);
-		if (ctrl->connect_attempts >= NVME_FC_MAX_CONNECT_ATTEMPTS) {
-			dev_warn(ctrl->ctrl.device,
-				"NVME-FC{%d}: Max reconnect attempts (%d) "
-				"reached. Removing controller\n",
-				ctrl->cnum, ctrl->connect_attempts);
-
-			if (!nvme_change_ctrl_state(&ctrl->ctrl,
-				NVME_CTRL_DELETING)) {
-				dev_err(ctrl->ctrl.device,
-					"NVME-FC{%d}: failed to change state "
-					"to DELETING\n", ctrl->cnum);
-				return;
-			}
-
-			WARN_ON(!queue_work(nvme_fc_wq, &ctrl->delete_work));
-			return;
-		}
-
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->reconnect_delay);
-		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->reconnect_delay * HZ);
-	} else
+	if (ret)
+		nvme_fc_reconnect_or_delete(ctrl, ret);
+	else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reconnect complete\n",
 			ctrl->cnum);
@@ -2755,7 +2743,6 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	INIT_WORK(&ctrl->delete_work, nvme_fc_delete_ctrl_work);
 	INIT_WORK(&ctrl->reset_work, nvme_fc_reset_ctrl_work);
 	INIT_DELAYED_WORK(&ctrl->connect_work, nvme_fc_connect_ctrl_work);
-	ctrl->reconnect_delay = opts->reconnect_delay;
 	spin_lock_init(&ctrl->lock);
 
 	/* io queue count */
@@ -2835,6 +2822,8 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 		return ERR_PTR(ret);
 	}
 
+	kref_get(&ctrl->ctrl.kref);
+
 	dev_info(ctrl->ctrl.device,
 		"NVME-FC{%d}: new ctrl: NQN \"%s\"\n",
 		ctrl->cnum, ctrl->ctrl.opts->subsysnqn);
@@ -2971,7 +2960,7 @@ nvme_fc_create_ctrl(struct device *dev, struct nvmf_ctrl_options *opts)
 static struct nvmf_transport_ops nvme_fc_transport = {
 	.name		= "fc",
 	.required_opts	= NVMF_OPT_TRADDR | NVMF_OPT_HOST_TRADDR,
-	.allowed_opts	= NVMF_OPT_RECONNECT_DELAY,
+	.allowed_opts	= NVMF_OPT_RECONNECT_DELAY | NVMF_OPT_CTRL_LOSS_TMO,
 	.create_ctrl	= nvme_fc_create_ctrl,
 };
 
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 29c708c..9d6a070 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -208,7 +208,9 @@ struct nvme_ns {
 struct nvme_ctrl_ops {
 	const char *name;
 	struct module *module;
-	bool is_fabrics;
+	unsigned int flags;
+#define NVME_F_FABRICS			(1 << 0)
+#define NVME_F_METADATA_SUPPORTED	(1 << 1)
 	int (*reg_read32)(struct nvme_ctrl *ctrl, u32 off, u32 *val);
 	int (*reg_write32)(struct nvme_ctrl *ctrl, u32 off, u32 val);
 	int (*reg_read64)(struct nvme_ctrl *ctrl, u32 off, u64 *val);
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 4c2ff2b..951042a 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -263,7 +263,7 @@ static void nvme_dbbuf_set(struct nvme_dev *dev)
 	c.dbbuf.prp2 = cpu_to_le64(dev->dbbuf_eis_dma_addr);
 
 	if (nvme_submit_sync_cmd(dev->ctrl.admin_q, &c, NULL, 0)) {
-		dev_warn(dev->dev, "unable to set dbbuf\n");
+		dev_warn(dev->ctrl.device, "unable to set dbbuf\n");
 		/* Free memory and continue on */
 		nvme_dbbuf_dma_free(dev);
 	}
@@ -1367,7 +1367,7 @@ static bool nvme_should_reset(struct nvme_dev *dev, u32 csts)
 	bool nssro = dev->subsystem && (csts & NVME_CSTS_NSSRO);
 
 	/* If there is a reset ongoing, we shouldn't reset again. */
-	if (work_busy(&dev->reset_work))
+	if (dev->ctrl.state == NVME_CTRL_RESETTING)
 		return false;
 
 	/* We shouldn't reset unless the controller is on fatal error state
@@ -1394,11 +1394,11 @@ static void nvme_warn_reset(struct nvme_dev *dev, u32 csts)
 	result = pci_read_config_word(to_pci_dev(dev->dev), PCI_STATUS,
 				      &pci_status);
 	if (result == PCIBIOS_SUCCESSFUL)
-		dev_warn(dev->dev,
+		dev_warn(dev->ctrl.device,
 			 "controller is down; will reset: CSTS=0x%x, PCI_STATUS=0x%hx\n",
 			 csts, pci_status);
 	else
-		dev_warn(dev->dev,
+		dev_warn(dev->ctrl.device,
 			 "controller is down; will reset: CSTS=0x%x, PCI_STATUS read failed (%d)\n",
 			 csts, result);
 }
@@ -1740,8 +1740,8 @@ static int nvme_pci_enable(struct nvme_dev *dev)
 	 */
 	if (pdev->vendor == PCI_VENDOR_ID_APPLE && pdev->device == 0x2001) {
 		dev->q_depth = 2;
-		dev_warn(dev->dev, "detected Apple NVMe controller, set "
-			"queue depth=%u to work around controller resets\n",
+		dev_warn(dev->ctrl.device, "detected Apple NVMe controller, "
+			"set queue depth=%u to work around controller resets\n",
 			dev->q_depth);
 	}
 
@@ -1759,7 +1759,7 @@ static int nvme_pci_enable(struct nvme_dev *dev)
 		if (dev->cmbsz) {
 			if (sysfs_add_file_to_group(&dev->ctrl.device->kobj,
 						    &dev_attr_cmb.attr, NULL))
-				dev_warn(dev->dev,
+				dev_warn(dev->ctrl.device,
 					 "failed to add sysfs attribute for CMB\n");
 		}
 	}
@@ -1903,7 +1903,7 @@ static void nvme_reset_work(struct work_struct *work)
 	bool was_suspend = !!(dev->ctrl.ctrl_config & NVME_CC_SHN_NORMAL);
 	int result = -ENODEV;
 
-	if (WARN_ON(dev->ctrl.state == NVME_CTRL_RESETTING))
+	if (WARN_ON(dev->ctrl.state != NVME_CTRL_RESETTING))
 		goto out;
 
 	/*
@@ -1913,9 +1913,6 @@ static void nvme_reset_work(struct work_struct *work)
 	if (dev->ctrl.ctrl_config & NVME_CC_ENABLE)
 		nvme_dev_disable(dev, false);
 
-	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING))
-		goto out;
-
 	result = nvme_pci_enable(dev);
 	if (result)
 		goto out;
@@ -2009,8 +2006,8 @@ static int nvme_reset(struct nvme_dev *dev)
 {
 	if (!dev->ctrl.admin_q || blk_queue_dying(dev->ctrl.admin_q))
 		return -ENODEV;
-	if (work_busy(&dev->reset_work))
-		return -ENODEV;
+	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING))
+		return -EBUSY;
 	if (!queue_work(nvme_workq, &dev->reset_work))
 		return -EBUSY;
 	return 0;
@@ -2047,6 +2044,7 @@ static int nvme_pci_reset_ctrl(struct nvme_ctrl *ctrl)
 static const struct nvme_ctrl_ops nvme_pci_ctrl_ops = {
 	.name			= "pcie",
 	.module			= THIS_MODULE,
+	.flags			= NVME_F_METADATA_SUPPORTED,
 	.reg_read32		= nvme_pci_reg_read32,
 	.reg_write32		= nvme_pci_reg_write32,
 	.reg_read64		= nvme_pci_reg_read64,
@@ -2135,6 +2133,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	if (result)
 		goto release_pools;
 
+	nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING);
 	dev_info(dev->ctrl.device, "pci function %s\n", dev_name(&pdev->dev));
 
 	queue_work(nvme_workq, &dev->reset_work);
@@ -2178,6 +2177,7 @@ static void nvme_remove(struct pci_dev *pdev)
 
 	nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_DELETING);
 
+	cancel_work_sync(&dev->reset_work);
 	pci_set_drvdata(pdev, NULL);
 
 	if (!pci_device_is_present(pdev)) {
@@ -2293,6 +2293,8 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_VDEVICE(INTEL, 0x0a54),
 		.driver_data = NVME_QUIRK_STRIPE_SIZE |
 				NVME_QUIRK_DEALLOCATE_ZEROES, },
+	{ PCI_VDEVICE(INTEL, 0xf1a5),	/* Intel 600P/P3100 */
+		.driver_data = NVME_QUIRK_NO_DEEPEST_PS },
 	{ PCI_VDEVICE(INTEL, 0x5845),	/* Qemu emulated controller */
 		.driver_data = NVME_QUIRK_IDENTIFY_CNS, },
 	{ PCI_DEVICE(0x1c58, 0x0003),	/* HGST adapter */
diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index dd1c6de..24397d3 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -753,28 +753,26 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	if (ret)
 		goto requeue;
 
-	blk_mq_start_stopped_hw_queues(ctrl->ctrl.admin_q, true);
-
 	ret = nvmf_connect_admin_queue(&ctrl->ctrl);
 	if (ret)
-		goto stop_admin_q;
+		goto requeue;
 
 	set_bit(NVME_RDMA_Q_LIVE, &ctrl->queues[0].flags);
 
 	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
 	if (ret)
-		goto stop_admin_q;
+		goto requeue;
 
 	nvme_start_keep_alive(&ctrl->ctrl);
 
 	if (ctrl->queue_count > 1) {
 		ret = nvme_rdma_init_io_queues(ctrl);
 		if (ret)
-			goto stop_admin_q;
+			goto requeue;
 
 		ret = nvme_rdma_connect_io_queues(ctrl);
 		if (ret)
-			goto stop_admin_q;
+			goto requeue;
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
@@ -782,7 +780,6 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	ctrl->ctrl.opts->nr_reconnects = 0;
 
 	if (ctrl->queue_count > 1) {
-		nvme_start_queues(&ctrl->ctrl);
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);
 	}
@@ -791,8 +788,6 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	return;
 
-stop_admin_q:
-	blk_mq_stop_hw_queues(ctrl->ctrl.admin_q);
 requeue:
 	dev_info(ctrl->ctrl.device, "Failed reconnect attempt %d\n",
 			ctrl->ctrl.opts->nr_reconnects);
@@ -823,6 +818,13 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
 				nvme_cancel_request, &ctrl->ctrl);
 
+	/*
+	 * queues are not a live anymore, so restart the queues to fail fast
+	 * new IO
+	 */
+	blk_mq_start_stopped_hw_queues(ctrl->ctrl.admin_q, true);
+	nvme_start_queues(&ctrl->ctrl);
+
 	nvme_rdma_reconnect_or_remove(ctrl);
 }
 
@@ -1038,6 +1040,19 @@ static void nvme_rdma_send_done(struct ib_cq *cq, struct ib_wc *wc)
 		nvme_rdma_wr_error(cq, wc, "SEND");
 }
 
+static inline int nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue)
+{
+	int sig_limit;
+
+	/*
+	 * We signal completion every queue depth/2 and also handle the
+	 * degenerated case of a  device with queue_depth=1, where we
+	 * would need to signal every message.
+	 */
+	sig_limit = max(queue->queue_size / 2, 1);
+	return (++queue->sig_count % sig_limit) == 0;
+}
+
 static int nvme_rdma_post_send(struct nvme_rdma_queue *queue,
 		struct nvme_rdma_qe *qe, struct ib_sge *sge, u32 num_sge,
 		struct ib_send_wr *first, bool flush)
@@ -1065,9 +1080,6 @@ static int nvme_rdma_post_send(struct nvme_rdma_queue *queue,
 	 * Would have been way to obvious to handle this in hardware or
 	 * at least the RDMA stack..
 	 *
-	 * This messy and racy code sniplet is copy and pasted from the iSER
-	 * initiator, and the magic '32' comes from there as well.
-	 *
 	 * Always signal the flushes. The magic request used for the flush
 	 * sequencer is not allocated in our driver's tagset and it's
 	 * triggered to be freed by blk_cleanup_queue(). So we need to
@@ -1075,7 +1087,7 @@ static int nvme_rdma_post_send(struct nvme_rdma_queue *queue,
 	 * embedded in request's payload, is not freed when __ib_process_cq()
 	 * calls wr_cqe->done().
 	 */
-	if ((++queue->sig_count % 32) == 0 || flush)
+	if (nvme_rdma_queue_sig_limit(queue) || flush)
 		wr.send_flags |= IB_SEND_SIGNALED;
 
 	if (first)
@@ -1423,7 +1435,7 @@ nvme_rdma_timeout(struct request *rq, bool reserved)
 /*
  * We cannot accept any other command until the Connect command has completed.
  */
-static inline bool nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
+static inline int nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
 		struct request *rq)
 {
 	if (unlikely(!test_bit(NVME_RDMA_Q_LIVE, &queue->flags))) {
@@ -1431,11 +1443,22 @@ static inline bool nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
 
 		if (!blk_rq_is_passthrough(rq) ||
 		    cmd->common.opcode != nvme_fabrics_command ||
-		    cmd->fabrics.fctype != nvme_fabrics_type_connect)
-			return false;
+		    cmd->fabrics.fctype != nvme_fabrics_type_connect) {
+			/*
+			 * reconnecting state means transport disruption, which
+			 * can take a long time and even might fail permanently,
+			 * so we can't let incoming I/O be requeued forever.
+			 * fail it fast to allow upper layers a chance to
+			 * failover.
+			 */
+			if (queue->ctrl->ctrl.state == NVME_CTRL_RECONNECTING)
+				return -EIO;
+			else
+				return -EAGAIN;
+		}
 	}
 
-	return true;
+	return 0;
 }
 
 static int nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx,
@@ -1453,8 +1476,9 @@ static int nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx,
 
 	WARN_ON_ONCE(rq->tag < 0);
 
-	if (!nvme_rdma_queue_is_ready(queue, rq))
-		return BLK_MQ_RQ_QUEUE_BUSY;
+	ret = nvme_rdma_queue_is_ready(queue, rq);
+	if (unlikely(ret))
+		goto err;
 
 	dev = queue->device->dev;
 	ib_dma_sync_single_for_cpu(dev, sqe->dma,
@@ -1782,7 +1806,7 @@ static int nvme_rdma_reset_ctrl(struct nvme_ctrl *nctrl)
 static const struct nvme_ctrl_ops nvme_rdma_ctrl_ops = {
 	.name			= "rdma",
 	.module			= THIS_MODULE,
-	.is_fabrics		= true,
+	.flags			= NVME_F_FABRICS,
 	.reg_read32		= nvmf_reg_read32,
 	.reg_read64		= nvmf_reg_read64,
 	.reg_write32		= nvmf_reg_write32,