7 files changed, 364 insertions, 117 deletions

diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c
index fb7df7b7581..a65598b1e53 100644
--- a/drivers/scsi/libsas/sas_discover.c
+++ b/drivers/scsi/libsas/sas_discover.c
@@ -548,7 +548,7 @@ int sas_discover_sata(struct domain_device *dev)
 
 	res = sas_notify_lldd_dev_found(dev);
 	if (res)
-		return res;
+		goto out_err2;
 
 	switch (dev->dev_type) {
 	case SATA_DEV:
@@ -560,11 +560,23 @@ int sas_discover_sata(struct domain_device *dev)
 	default:
 		break;
 	}
+	if (res)
+		goto out_err;
 
 	sas_notify_lldd_dev_gone(dev);
-	if (!res) {
-		sas_notify_lldd_dev_found(dev);
-	}
+	res = sas_notify_lldd_dev_found(dev);
+	if (res)
+		goto out_err2;
+
+	res = sas_rphy_add(dev->rphy);
+	if (res)
+		goto out_err;
+
+	return res;
+
+out_err:
+	sas_notify_lldd_dev_gone(dev);
+out_err2:
 	return res;
 }
 
@@ -580,21 +592,17 @@ int sas_discover_end_dev(struct domain_device *dev)
 
 	res = sas_notify_lldd_dev_found(dev);
 	if (res)
-		return res;
+		goto out_err2;
 
 	res = sas_rphy_add(dev->rphy);
 	if (res)
 		goto out_err;
 
-	/* do this to get the end device port attributes which will have
-	 * been scanned in sas_rphy_add */
-	sas_notify_lldd_dev_gone(dev);
-	sas_notify_lldd_dev_found(dev);
-
 	return 0;
 
 out_err:
 	sas_notify_lldd_dev_gone(dev);
+out_err2:
 	return res;
 }
 
@@ -649,6 +657,7 @@ void sas_unregister_domain_devices(struct asd_sas_port *port)
  */
 static void sas_discover_domain(struct work_struct *work)
 {
+	struct domain_device *dev;
 	int error = 0;
 	struct sas_discovery_event *ev =
 		container_of(work, struct sas_discovery_event, work);
@@ -658,35 +667,42 @@ static void sas_discover_domain(struct work_struct *work)
 			&port->disc.pending);
 
 	if (port->port_dev)
-		return ;
-	else {
-		error = sas_get_port_device(port);
-		if (error)
-			return;
-	}
+		return;
+
+	error = sas_get_port_device(port);
+	if (error)
+		return;
+	dev = port->port_dev;
 
 	SAS_DPRINTK("DOING DISCOVERY on port %d, pid:%d\n", port->id,
 		    current->pid);
 
-	switch (port->port_dev->dev_type) {
+	switch (dev->dev_type) {
 	case SAS_END_DEV:
-		error = sas_discover_end_dev(port->port_dev);
+		error = sas_discover_end_dev(dev);
 		break;
 	case EDGE_DEV:
 	case FANOUT_DEV:
-		error = sas_discover_root_expander(port->port_dev);
+		error = sas_discover_root_expander(dev);
 		break;
 	case SATA_DEV:
 	case SATA_PM:
-		error = sas_discover_sata(port->port_dev);
+		error = sas_discover_sata(dev);
 		break;
 	default:
-		SAS_DPRINTK("unhandled device %d\n", port->port_dev->dev_type);
+		SAS_DPRINTK("unhandled device %d\n", dev->dev_type);
 		break;
 	}
 
 	if (error) {
-		kfree(port->port_dev); /* not kobject_register-ed yet */
+		sas_rphy_free(dev->rphy);
+		dev->rphy = NULL;
+
+		spin_lock(&port->dev_list_lock);
+		list_del_init(&dev->dev_list_node);
+		spin_unlock(&port->dev_list_lock);
+
+		kfree(dev); /* not kobject_register-ed yet */
 		port->port_dev = NULL;
 	}
 
@@ -726,7 +742,7 @@ int sas_discover_event(struct asd_sas_port *port, enum discover_event ev)
 	BUG_ON(ev >= DISC_NUM_EVENTS);
 
 	sas_queue_event(ev, &disc->disc_event_lock, &disc->pending,
-			&disc->disc_work[ev].work, port->ha->core.shost);
+			&disc->disc_work[ev].work, port->ha);
 
 	return 0;
 }
diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c
index d83392ee682..9db30fb5caf 100644
--- a/drivers/scsi/libsas/sas_event.c
+++ b/drivers/scsi/libsas/sas_event.c
@@ -31,7 +31,7 @@ static void notify_ha_event(struct sas_ha_struct *sas_ha, enum ha_event event)
 	BUG_ON(event >= HA_NUM_EVENTS);
 
 	sas_queue_event(event, &sas_ha->event_lock, &sas_ha->pending,
-			&sas_ha->ha_events[event].work, sas_ha->core.shost);
+			&sas_ha->ha_events[event].work, sas_ha);
 }
 
 static void notify_port_event(struct asd_sas_phy *phy, enum port_event event)
@@ -41,7 +41,7 @@ static void notify_port_event(struct asd_sas_phy *phy, enum port_event event)
 	BUG_ON(event >= PORT_NUM_EVENTS);
 
 	sas_queue_event(event, &ha->event_lock, &phy->port_events_pending,
-			&phy->port_events[event].work, ha->core.shost);
+			&phy->port_events[event].work, ha);
 }
 
 static void notify_phy_event(struct asd_sas_phy *phy, enum phy_event event)
@@ -51,7 +51,7 @@ static void notify_phy_event(struct asd_sas_phy *phy, enum phy_event event)
 	BUG_ON(event >= PHY_NUM_EVENTS);
 
 	sas_queue_event(event, &ha->event_lock, &phy->phy_events_pending,
-			&phy->phy_events[event].work, ha->core.shost);
+			&phy->phy_events[event].work, ha);
 }
 
 int sas_init_events(struct sas_ha_struct *sas_ha)
diff --git a/drivers/scsi/libsas/sas_expander.c b/drivers/scsi/libsas/sas_expander.c
index d31e6fa466f..dc70c180e11 100644
--- a/drivers/scsi/libsas/sas_expander.c
+++ b/drivers/scsi/libsas/sas_expander.c
@@ -667,8 +667,9 @@ static struct domain_device *sas_ex_discover_end_dev(
 	return child;
 
  out_list_del:
+	sas_rphy_free(child->rphy);
+	child->rphy = NULL;
 	list_del(&child->dev_list_node);
-	sas_rphy_free(rphy);
  out_free:
 	sas_port_delete(phy->port);
  out_err:
@@ -677,6 +678,29 @@ static struct domain_device *sas_ex_discover_end_dev(
 	return NULL;
 }
 
+/* See if this phy is part of a wide port */
+static int sas_ex_join_wide_port(struct domain_device *parent, int phy_id)
+{
+	struct ex_phy *phy = &parent->ex_dev.ex_phy[phy_id];
+	int i;
+
+	for (i = 0; i < parent->ex_dev.num_phys; i++) {
+		struct ex_phy *ephy = &parent->ex_dev.ex_phy[i];
+
+		if (ephy == phy)
+			continue;
+
+		if (!memcmp(phy->attached_sas_addr, ephy->attached_sas_addr,
+			    SAS_ADDR_SIZE) && ephy->port) {
+			sas_port_add_phy(ephy->port, phy->phy);
+			phy->phy_state = PHY_DEVICE_DISCOVERED;
+			return 0;
+		}
+	}
+
+	return -ENODEV;
+}
+
 static struct domain_device *sas_ex_discover_expander(
 	struct domain_device *parent, int phy_id)
 {
@@ -809,6 +833,13 @@ static int sas_ex_discover_dev(struct domain_device *dev, int phy_id)
 		return res;
 	}
 
+	res = sas_ex_join_wide_port(dev, phy_id);
+	if (!res) {
+		SAS_DPRINTK("Attaching ex phy%d to wide port %016llx\n",
+			    phy_id, SAS_ADDR(ex_phy->attached_sas_addr));
+		return res;
+	}
+
 	switch (ex_phy->attached_dev_type) {
 	case SAS_END_DEV:
 		child = sas_ex_discover_end_dev(dev, phy_id);
@@ -1431,14 +1462,23 @@ int sas_discover_root_expander(struct domain_device *dev)
 	int res;
 	struct sas_expander_device *ex = rphy_to_expander_device(dev->rphy);
 
-	sas_rphy_add(dev->rphy);
+	res = sas_rphy_add(dev->rphy);
+	if (res)
+		goto out_err;
 
 	ex->level = dev->port->disc.max_level; /* 0 */
 	res = sas_discover_expander(dev);
-	if (!res)
-		sas_ex_bfs_disc(dev->port);
+	if (res)
+		goto out_err2;
+
+	sas_ex_bfs_disc(dev->port);
 
 	return res;
+
+out_err2:
+	sas_rphy_remove(dev->rphy);
+out_err:
+	return res;
 }
 
 /* ---------- Domain revalidation ---------- */
diff --git a/drivers/scsi/libsas/sas_init.c b/drivers/scsi/libsas/sas_init.c
index 2f0c07fc3f4..965698c8b7b 100644
--- a/drivers/scsi/libsas/sas_init.c
+++ b/drivers/scsi/libsas/sas_init.c
@@ -87,6 +87,9 @@ int sas_register_ha(struct sas_ha_struct *sas_ha)
 	else if (sas_ha->lldd_queue_size == -1)
 		sas_ha->lldd_queue_size = 128; /* Sanity */
 
+	sas_ha->state = SAS_HA_REGISTERED;
+	spin_lock_init(&sas_ha->state_lock);
+
 	error = sas_register_phys(sas_ha);
 	if (error) {
 		printk(KERN_NOTICE "couldn't register sas phys:%d\n", error);
@@ -127,12 +130,22 @@ Undo_phys:
 
 int sas_unregister_ha(struct sas_ha_struct *sas_ha)
 {
+	unsigned long flags;
+
+	/* Set the state to unregistered to avoid further
+	 * events to be queued */
+	spin_lock_irqsave(&sas_ha->state_lock, flags);
+	sas_ha->state = SAS_HA_UNREGISTERED;
+	spin_unlock_irqrestore(&sas_ha->state_lock, flags);
+	scsi_flush_work(sas_ha->core.shost);
+
+	sas_unregister_ports(sas_ha);
+
 	if (sas_ha->lldd_max_execute_num > 1) {
 		sas_shutdown_queue(sas_ha);
+		sas_ha->lldd_max_execute_num = 1;
 	}
 
-	sas_unregister_ports(sas_ha);
-
 	return 0;
 }
 
@@ -146,6 +159,36 @@ static int sas_get_linkerrors(struct sas_phy *phy)
 	return sas_smp_get_phy_events(phy);
 }
 
+int sas_phy_enable(struct sas_phy *phy, int enable)
+{
+	int ret;
+	enum phy_func command;
+
+	if (enable)
+		command = PHY_FUNC_LINK_RESET;
+	else
+		command = PHY_FUNC_DISABLE;
+
+	if (scsi_is_sas_phy_local(phy)) {
+		struct Scsi_Host *shost = dev_to_shost(phy->dev.parent);
+		struct sas_ha_struct *sas_ha = SHOST_TO_SAS_HA(shost);
+		struct asd_sas_phy *asd_phy = sas_ha->sas_phy[phy->number];
+		struct sas_internal *i =
+			to_sas_internal(sas_ha->core.shost->transportt);
+
+		if (!enable) {
+			sas_phy_disconnected(asd_phy);
+			sas_ha->notify_phy_event(asd_phy, PHYE_LOSS_OF_SIGNAL);
+		}
+		ret = i->dft->lldd_control_phy(asd_phy, command, NULL);
+	} else {
+		struct sas_rphy *rphy = dev_to_rphy(phy->dev.parent);
+		struct domain_device *ddev = sas_find_dev_by_rphy(rphy);
+		ret = sas_smp_phy_control(ddev, phy->number, command, NULL);
+	}
+	return ret;
+}
+
 int sas_phy_reset(struct sas_phy *phy, int hard_reset)
 {
 	int ret;
@@ -172,8 +215,8 @@ int sas_phy_reset(struct sas_phy *phy, int hard_reset)
 	return ret;
 }
 
-static int sas_set_phy_speed(struct sas_phy *phy,
-			     struct sas_phy_linkrates *rates)
+int sas_set_phy_speed(struct sas_phy *phy,
+		      struct sas_phy_linkrates *rates)
 {
 	int ret;
 
@@ -212,6 +255,7 @@ static int sas_set_phy_speed(struct sas_phy *phy,
 }
 
 static struct sas_function_template sft = {
+	.phy_enable = sas_phy_enable,
 	.phy_reset = sas_phy_reset,
 	.set_phy_speed = sas_set_phy_speed,
 	.get_linkerrors = sas_get_linkerrors,
diff --git a/drivers/scsi/libsas/sas_internal.h b/drivers/scsi/libsas/sas_internal.h
index 137d7e496b6..a78638df201 100644
--- a/drivers/scsi/libsas/sas_internal.h
+++ b/drivers/scsi/libsas/sas_internal.h
@@ -80,7 +80,7 @@ void sas_hae_reset(struct work_struct *work);
 static inline void sas_queue_event(int event, spinlock_t *lock,
 				   unsigned long *pending,
 				   struct work_struct *work,
-				   struct Scsi_Host *shost)
+				   struct sas_ha_struct *sas_ha)
 {
 	unsigned long flags;
 
@@ -91,7 +91,12 @@ static inline void sas_queue_event(int event, spinlock_t *lock,
 	}
 	__set_bit(event, pending);
 	spin_unlock_irqrestore(lock, flags);
-	scsi_queue_work(shost, work);
+
+	spin_lock_irqsave(&sas_ha->state_lock, flags);
+	if (sas_ha->state != SAS_HA_UNREGISTERED) {
+		scsi_queue_work(sas_ha->core.shost, work);
+	}
+	spin_unlock_irqrestore(&sas_ha->state_lock, flags);
 }
 
 static inline void sas_begin_event(int event, spinlock_t *lock,
diff --git a/drivers/scsi/libsas/sas_port.c b/drivers/scsi/libsas/sas_port.c
index 971c37ceecb..e1e2d085c92 100644
--- a/drivers/scsi/libsas/sas_port.c
+++ b/drivers/scsi/libsas/sas_port.c
@@ -42,10 +42,11 @@ static void sas_form_port(struct asd_sas_phy *phy)
 	struct asd_sas_port *port = phy->port;
 	struct sas_internal *si =
 		to_sas_internal(sas_ha->core.shost->transportt);
+	unsigned long flags;
 
 	if (port) {
 		if (memcmp(port->attached_sas_addr, phy->attached_sas_addr,
-			   SAS_ADDR_SIZE) == 0)
+			   SAS_ADDR_SIZE) != 0)
 			sas_deform_port(phy);
 		else {
 			SAS_DPRINTK("%s: phy%d belongs to port%d already(%d)!\n",
@@ -56,7 +57,7 @@ static void sas_form_port(struct asd_sas_phy *phy)
 	}
 
 	/* find a port */
-	spin_lock(&sas_ha->phy_port_lock);
+	spin_lock_irqsave(&sas_ha->phy_port_lock, flags);
 	for (i = 0; i < sas_ha->num_phys; i++) {
 		port = sas_ha->sas_port[i];
 		spin_lock(&port->phy_list_lock);
@@ -78,7 +79,7 @@ static void sas_form_port(struct asd_sas_phy *phy)
 	if (i >= sas_ha->num_phys) {
 		printk(KERN_NOTICE "%s: couldn't find a free port, bug?\n",
 		       __FUNCTION__);
-		spin_unlock(&sas_ha->phy_port_lock);
+		spin_unlock_irqrestore(&sas_ha->phy_port_lock, flags);
 		return;
 	}
 
@@ -105,7 +106,7 @@ static void sas_form_port(struct asd_sas_phy *phy)
 	} else
 		port->linkrate = max(port->linkrate, phy->linkrate);
 	spin_unlock(&port->phy_list_lock);
-	spin_unlock(&sas_ha->phy_port_lock);
+	spin_unlock_irqrestore(&sas_ha->phy_port_lock, flags);
 
 	if (!port->port) {
 		port->port = sas_port_alloc(phy->phy->dev.parent, port->id);
@@ -137,6 +138,7 @@ void sas_deform_port(struct asd_sas_phy *phy)
 	struct asd_sas_port *port = phy->port;
 	struct sas_internal *si =
 		to_sas_internal(sas_ha->core.shost->transportt);
+	unsigned long flags;
 
 	if (!port)
 		return;		  /* done by a phy event */
@@ -155,7 +157,7 @@ void sas_deform_port(struct asd_sas_phy *phy)
 	if (si->dft->lldd_port_deformed)
 		si->dft->lldd_port_deformed(phy);
 
-	spin_lock(&sas_ha->phy_port_lock);
+	spin_lock_irqsave(&sas_ha->phy_port_lock, flags);
 	spin_lock(&port->phy_list_lock);
 
 	list_del_init(&phy->port_phy_el);
@@ -174,7 +176,7 @@ void sas_deform_port(struct asd_sas_phy *phy)
 		port->phy_mask = 0;
 	}
 	spin_unlock(&port->phy_list_lock);
-	spin_unlock(&sas_ha->phy_port_lock);
+	spin_unlock_irqrestore(&sas_ha->phy_port_lock, flags);
 
 	return;
 }
diff --git a/drivers/scsi/libsas/sas_scsi_host.c b/drivers/scsi/libsas/sas_scsi_host.c
index 22672d54aa2..897a5e2c55e 100644
--- a/drivers/scsi/libsas/sas_scsi_host.c
+++ b/drivers/scsi/libsas/sas_scsi_host.c
@@ -34,6 +34,7 @@
 #include <scsi/scsi_transport_sas.h>
 #include "../scsi_sas_internal.h"
 #include "../scsi_transport_api.h"
+#include "../scsi_priv.h"
 
 #include <linux/err.h>
 #include <linux/blkdev.h>
@@ -130,7 +131,7 @@ static enum task_attribute sas_scsi_get_task_attr(struct scsi_cmnd *cmd)
 	if (cmd->request && blk_rq_tagged(cmd->request)) {
 		if (cmd->device->ordered_tags &&
 		    (cmd->request->cmd_flags & REQ_HARDBARRIER))
-			ta = TASK_ATTR_HOQ;
+			ta = TASK_ATTR_ORDERED;
 	}
 	return ta;
 }
@@ -281,6 +282,7 @@ enum task_disposition {
 	TASK_IS_ABORTED,
 	TASK_IS_AT_LU,
 	TASK_IS_NOT_AT_LU,
+	TASK_ABORT_FAILED,
 };
 
 static enum task_disposition sas_scsi_find_task(struct sas_task *task)
@@ -310,15 +312,6 @@ static enum task_disposition sas_scsi_find_task(struct sas_task *task)
 		spin_unlock_irqrestore(&core->task_queue_lock, flags);
 	}
 
-	spin_lock_irqsave(&task->task_state_lock, flags);
-	if (task->task_state_flags & SAS_TASK_INITIATOR_ABORTED) {
-		spin_unlock_irqrestore(&task->task_state_lock, flags);
-		SAS_DPRINTK("%s: task 0x%p already aborted\n",
-			    __FUNCTION__, task);
-		return TASK_IS_ABORTED;
-	}
-	spin_unlock_irqrestore(&task->task_state_lock, flags);
-
 	for (i = 0; i < 5; i++) {
 		SAS_DPRINTK("%s: aborting task 0x%p\n", __FUNCTION__, task);
 		res = si->dft->lldd_abort_task(task);
@@ -340,15 +333,21 @@ static enum task_disposition sas_scsi_find_task(struct sas_task *task)
 			SAS_DPRINTK("%s: querying task 0x%p\n",
 				    __FUNCTION__, task);
 			res = si->dft->lldd_query_task(task);
-			if (res == TMF_RESP_FUNC_SUCC) {
+			switch (res) {
+			case TMF_RESP_FUNC_SUCC:
 				SAS_DPRINTK("%s: task 0x%p at LU\n",
 					    __FUNCTION__, task);
 				return TASK_IS_AT_LU;
-			} else if (res == TMF_RESP_FUNC_COMPLETE) {
+			case TMF_RESP_FUNC_COMPLETE:
 				SAS_DPRINTK("%s: task 0x%p not at LU\n",
 					    __FUNCTION__, task);
 				return TASK_IS_NOT_AT_LU;
-			}
+			case TMF_RESP_FUNC_FAILED:
+                                SAS_DPRINTK("%s: task 0x%p failed to abort\n",
+                                                __FUNCTION__, task);
+                                return TASK_ABORT_FAILED;
+                        }
+
 		}
 	}
 	return res;
@@ -398,35 +397,113 @@ static int sas_recover_I_T(struct domain_device *dev)
 	return res;
 }
 
-void sas_scsi_recover_host(struct Scsi_Host *shost)
+/* Find the sas_phy that's attached to this device */
+struct sas_phy *find_local_sas_phy(struct domain_device *dev)
+{
+	struct domain_device *pdev = dev->parent;
+	struct ex_phy *exphy = NULL;
+	int i;
+
+	/* Directly attached device */
+	if (!pdev)
+		return dev->port->phy;
+
+	/* Otherwise look in the expander */
+	for (i = 0; i < pdev->ex_dev.num_phys; i++)
+		if (!memcmp(dev->sas_addr,
+			    pdev->ex_dev.ex_phy[i].attached_sas_addr,
+			    SAS_ADDR_SIZE)) {
+			exphy = &pdev->ex_dev.ex_phy[i];
+			break;
+		}
+
+	BUG_ON(!exphy);
+	return exphy->phy;
+}
+
+/* Attempt to send a LUN reset message to a device */
+int sas_eh_device_reset_handler(struct scsi_cmnd *cmd)
+{
+	struct domain_device *dev = cmd_to_domain_dev(cmd);
+	struct sas_internal *i =
+		to_sas_internal(dev->port->ha->core.shost->transportt);
+	struct scsi_lun lun;
+	int res;
+
+	int_to_scsilun(cmd->device->lun, &lun);
+
+	if (!i->dft->lldd_lu_reset)
+		return FAILED;
+
+	res = i->dft->lldd_lu_reset(dev, lun.scsi_lun);
+	if (res == TMF_RESP_FUNC_SUCC || res == TMF_RESP_FUNC_COMPLETE)
+		return SUCCESS;
+
+	return FAILED;
+}
+
+/* Attempt to send a phy (bus) reset */
+int sas_eh_bus_reset_handler(struct scsi_cmnd *cmd)
+{
+	struct domain_device *dev = cmd_to_domain_dev(cmd);
+	struct sas_phy *phy = find_local_sas_phy(dev);
+	int res;
+
+	res = sas_phy_reset(phy, 1);
+	if (res)
+		SAS_DPRINTK("Bus reset of %s failed 0x%x\n",
+			    phy->dev.kobj.k_name,
+			    res);
+	if (res == TMF_RESP_FUNC_SUCC || res == TMF_RESP_FUNC_COMPLETE)
+		return SUCCESS;
+
+	return FAILED;
+}
+
+/* Try to reset a device */
+static int try_to_reset_cmd_device(struct Scsi_Host *shost,
+				   struct scsi_cmnd *cmd)
+{
+	int res;
+
+	if (!shost->hostt->eh_device_reset_handler)
+		goto try_bus_reset;
+
+	res = shost->hostt->eh_device_reset_handler(cmd);
+	if (res == SUCCESS)
+		return res;
+
+try_bus_reset:
+	if (shost->hostt->eh_bus_reset_handler)
+		return shost->hostt->eh_bus_reset_handler(cmd);
+
+	return FAILED;
+}
+
+static int sas_eh_handle_sas_errors(struct Scsi_Host *shost,
+				    struct list_head *work_q,
+				    struct list_head *done_q)
 {
-	struct sas_ha_struct *ha = SHOST_TO_SAS_HA(shost);
-	unsigned long flags;
-	LIST_HEAD(error_q);
 	struct scsi_cmnd *cmd, *n;
 	enum task_disposition res = TASK_IS_DONE;
-	int tmf_resp;
+	int tmf_resp, need_reset;
 	struct sas_internal *i = to_sas_internal(shost->transportt);
+	unsigned long flags;
+	struct sas_ha_struct *ha = SHOST_TO_SAS_HA(shost);
 
-	spin_lock_irqsave(shost->host_lock, flags);
-	list_splice_init(&shost->eh_cmd_q, &error_q);
-	spin_unlock_irqrestore(shost->host_lock, flags);
-
-	SAS_DPRINTK("Enter %s\n", __FUNCTION__);
-
-	/* All tasks on this list were marked SAS_TASK_STATE_ABORTED
-	 * by sas_scsi_timed_out() callback.
-	 */
 Again:
-	SAS_DPRINTK("going over list...\n");
-	list_for_each_entry_safe(cmd, n, &error_q, eh_entry) {
+	list_for_each_entry_safe(cmd, n, work_q, eh_entry) {
 		struct sas_task *task = TO_SAS_TASK(cmd);
-		list_del_init(&cmd->eh_entry);
 
-		if (!task) {
-			SAS_DPRINTK("%s: taskless cmd?!\n", __FUNCTION__);
+		if (!task)
 			continue;
-		}
+
+		list_del_init(&cmd->eh_entry);
+
+		spin_lock_irqsave(&task->task_state_lock, flags);
+		need_reset = task->task_state_flags & SAS_TASK_NEED_DEV_RESET;
+		spin_unlock_irqrestore(&task->task_state_lock, flags);
+
 		SAS_DPRINTK("trying to find task 0x%p\n", task);
 		res = sas_scsi_find_task(task);
 
@@ -437,11 +514,15 @@ Again:
 			SAS_DPRINTK("%s: task 0x%p is done\n", __FUNCTION__,
 				    task);
 			task->task_done(task);
+			if (need_reset)
+				try_to_reset_cmd_device(shost, cmd);
 			continue;
 		case TASK_IS_ABORTED:
 			SAS_DPRINTK("%s: task 0x%p is aborted\n",
 				    __FUNCTION__, task);
 			task->task_done(task);
+			if (need_reset)
+				try_to_reset_cmd_device(shost, cmd);
 			continue;
 		case TASK_IS_AT_LU:
 			SAS_DPRINTK("task 0x%p is at LU: lu recover\n", task);
@@ -452,11 +533,14 @@ Again:
 					    SAS_ADDR(task->dev),
 					    cmd->device->lun);
 				task->task_done(task);
-				sas_scsi_clear_queue_lu(&error_q, cmd);
+				if (need_reset)
+					try_to_reset_cmd_device(shost, cmd);
+				sas_scsi_clear_queue_lu(work_q, cmd);
 				goto Again;
 			}
 			/* fallthrough */
 		case TASK_IS_NOT_AT_LU:
+		case TASK_ABORT_FAILED:
 			SAS_DPRINTK("task 0x%p is not at LU: I_T recover\n",
 				    task);
 			tmf_resp = sas_recover_I_T(task->dev);
@@ -464,7 +548,9 @@ Again:
 				SAS_DPRINTK("I_T %016llx recovered\n",
 					    SAS_ADDR(task->dev->sas_addr));
 				task->task_done(task);
-				sas_scsi_clear_queue_I_T(&error_q, task->dev);
+				if (need_reset)
+					try_to_reset_cmd_device(shost, cmd);
+				sas_scsi_clear_queue_I_T(work_q, task->dev);
 				goto Again;
 			}
 			/* Hammer time :-) */
@@ -477,7 +563,9 @@ Again:
 					SAS_DPRINTK("clear nexus port:%d "
 						    "succeeded\n", port->id);
 					task->task_done(task);
-					sas_scsi_clear_queue_port(&error_q,
+					if (need_reset)
+						try_to_reset_cmd_device(shost, cmd);
+					sas_scsi_clear_queue_port(work_q,
 								  port);
 					goto Again;
 				}
@@ -489,6 +577,8 @@ Again:
 					SAS_DPRINTK("clear nexus ha "
 						    "succeeded\n");
 					task->task_done(task);
+					if (need_reset)
+						try_to_reset_cmd_device(shost, cmd);
 					goto out;
 				}
 			}
@@ -502,20 +592,54 @@ Again:
 				    cmd->device->lun);
 
 			task->task_done(task);
+			if (need_reset)
+				try_to_reset_cmd_device(shost, cmd);
 			goto clear_q;
 		}
 	}
 out:
-	scsi_eh_flush_done_q(&ha->eh_done_q);
-	SAS_DPRINTK("--- Exit %s\n", __FUNCTION__);
-	return;
+	return list_empty(work_q);
 clear_q:
 	SAS_DPRINTK("--- Exit %s -- clear_q\n", __FUNCTION__);
-	list_for_each_entry_safe(cmd, n, &error_q, eh_entry) {
+	list_for_each_entry_safe(cmd, n, work_q, eh_entry) {
 		struct sas_task *task = TO_SAS_TASK(cmd);
 		list_del_init(&cmd->eh_entry);
 		task->task_done(task);
 	}
+	return list_empty(work_q);
+}
+
+void sas_scsi_recover_host(struct Scsi_Host *shost)
+{
+	struct sas_ha_struct *ha = SHOST_TO_SAS_HA(shost);
+	unsigned long flags;
+	LIST_HEAD(eh_work_q);
+
+	spin_lock_irqsave(shost->host_lock, flags);
+	list_splice_init(&shost->eh_cmd_q, &eh_work_q);
+	spin_unlock_irqrestore(shost->host_lock, flags);
+
+	SAS_DPRINTK("Enter %s\n", __FUNCTION__);
+	/*
+	 * Deal with commands that still have SAS tasks (i.e. they didn't
+	 * complete via the normal sas_task completion mechanism)
+	 */
+	if (sas_eh_handle_sas_errors(shost, &eh_work_q, &ha->eh_done_q))
+		goto out;
+
+	/*
+	 * Now deal with SCSI commands that completed ok but have a an error
+	 * code (and hopefully sense data) attached.  This is roughly what
+	 * scsi_unjam_host does, but we skip scsi_eh_abort_cmds because any
+	 * command we see here has no sas_task and is thus unknown to the HA.
+	 */
+	if (!scsi_eh_get_sense(&eh_work_q, &ha->eh_done_q))
+		scsi_eh_ready_devs(shost, &eh_work_q, &ha->eh_done_q);
+
+out:
+	scsi_eh_flush_done_q(&ha->eh_done_q);
+	SAS_DPRINTK("--- Exit %s\n", __FUNCTION__);
+	return;
 }
 
 enum scsi_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *cmd)
@@ -524,24 +648,30 @@ enum scsi_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *cmd)
 	unsigned long flags;
 
 	if (!task) {
-		SAS_DPRINTK("command 0x%p, task 0x%p, gone: EH_HANDLED\n",
-			    cmd, task);
-		return EH_HANDLED;
+		cmd->timeout_per_command /= 2;
+		SAS_DPRINTK("command 0x%p, task 0x%p, gone: %s\n",
+			    cmd, task, (cmd->timeout_per_command ?
+			    "EH_RESET_TIMER" : "EH_NOT_HANDLED"));
+		if (!cmd->timeout_per_command)
+			return EH_NOT_HANDLED;
+		return EH_RESET_TIMER;
 	}
 
 	spin_lock_irqsave(&task->task_state_lock, flags);
-	if (task->task_state_flags & SAS_TASK_INITIATOR_ABORTED) {
-		spin_unlock_irqrestore(&task->task_state_lock, flags);
-		SAS_DPRINTK("command 0x%p, task 0x%p, aborted by initiator: "
-			    "EH_NOT_HANDLED\n", cmd, task);
-		return EH_NOT_HANDLED;
-	}
+	BUG_ON(task->task_state_flags & SAS_TASK_STATE_ABORTED);
 	if (task->task_state_flags & SAS_TASK_STATE_DONE) {
 		spin_unlock_irqrestore(&task->task_state_lock, flags);
 		SAS_DPRINTK("command 0x%p, task 0x%p, timed out: EH_HANDLED\n",
 			    cmd, task);
 		return EH_HANDLED;
 	}
+	if (!(task->task_state_flags & SAS_TASK_AT_INITIATOR)) {
+		spin_unlock_irqrestore(&task->task_state_lock, flags);
+		SAS_DPRINTK("command 0x%p, task 0x%p, not at initiator: "
+			    "EH_RESET_TIMER\n",
+			    cmd, task);
+		return EH_RESET_TIMER;
+	}
 	task->task_state_flags |= SAS_TASK_STATE_ABORTED;
 	spin_unlock_irqrestore(&task->task_state_lock, flags);
 
@@ -557,8 +687,9 @@ struct domain_device *sas_find_dev_by_rphy(struct sas_rphy *rphy)
 	struct sas_ha_struct *ha = SHOST_TO_SAS_HA(shost);
 	struct domain_device *found_dev = NULL;
 	int i;
+	unsigned long flags;
 
-	spin_lock(&ha->phy_port_lock);
+	spin_lock_irqsave(&ha->phy_port_lock, flags);
 	for (i = 0; i < ha->num_phys; i++) {
 		struct asd_sas_port *port = ha->sas_port[i];
 		struct domain_device *dev;
@@ -574,7 +705,7 @@ struct domain_device *sas_find_dev_by_rphy(struct sas_rphy *rphy)
 		spin_unlock(&port->dev_list_lock);
 	}
  found:
-	spin_unlock(&ha->phy_port_lock);
+	spin_unlock_irqrestore(&ha->phy_port_lock, flags);
 
 	return found_dev;
 }
@@ -623,6 +754,8 @@ int sas_slave_configure(struct scsi_device *scsi_dev)
 		scsi_deactivate_tcq(scsi_dev, 1);
 	}
 
+	scsi_dev->allow_restart = 1;
+
 	return 0;
 }
 
@@ -799,46 +932,42 @@ void sas_shutdown_queue(struct sas_ha_struct *sas_ha)
 	spin_unlock_irqrestore(&core->task_queue_lock, flags);
 }
 
-static int do_sas_task_abort(struct sas_task *task)
+/*
+ * Call the LLDD task abort routine directly.  This function is intended for
+ * use by upper layers that need to tell the LLDD to abort a task.
+ */
+int __sas_task_abort(struct sas_task *task)
 {
-	struct scsi_cmnd *sc = task->uldd_task;
 	struct sas_internal *si =
 		to_sas_internal(task->dev->port->ha->core.shost->transportt);
 	unsigned long flags;
 	int res;
 
 	spin_lock_irqsave(&task->task_state_lock, flags);
-	if (task->task_state_flags & SAS_TASK_STATE_ABORTED) {
+	if (task->task_state_flags & SAS_TASK_STATE_ABORTED ||
+	    task->task_state_flags & SAS_TASK_STATE_DONE) {
 		spin_unlock_irqrestore(&task->task_state_lock, flags);
-		SAS_DPRINTK("%s: Task %p already aborted.\n", __FUNCTION__,
+		SAS_DPRINTK("%s: Task %p already finished.\n", __FUNCTION__,
 			    task);
 		return 0;
 	}
-
-	task->task_state_flags |= SAS_TASK_INITIATOR_ABORTED;
-	if (!(task->task_state_flags & SAS_TASK_STATE_DONE))
-		task->task_state_flags |= SAS_TASK_STATE_ABORTED;
+	task->task_state_flags |= SAS_TASK_STATE_ABORTED;
 	spin_unlock_irqrestore(&task->task_state_lock, flags);
 
 	if (!si->dft->lldd_abort_task)
 		return -ENODEV;
 
 	res = si->dft->lldd_abort_task(task);
+
+	spin_lock_irqsave(&task->task_state_lock, flags);
 	if ((task->task_state_flags & SAS_TASK_STATE_DONE) ||
 	    (res == TMF_RESP_FUNC_COMPLETE))
 	{
-		/* SMP commands don't have scsi_cmds(?) */
-		if (!sc) {
-			task->task_done(task);
-			return 0;
-		}
-		scsi_req_abort_cmd(sc);
-		scsi_schedule_eh(sc->device->host);
+		spin_unlock_irqrestore(&task->task_state_lock, flags);
+		task->task_done(task);
 		return 0;
 	}
 
-	spin_lock_irqsave(&task->task_state_lock, flags);
-	task->task_state_flags &= ~SAS_TASK_INITIATOR_ABORTED;
 	if (!(task->task_state_flags & SAS_TASK_STATE_DONE))
 		task->task_state_flags &= ~SAS_TASK_STATE_ABORTED;
 	spin_unlock_irqrestore(&task->task_state_lock, flags);
@@ -846,17 +975,24 @@ static int do_sas_task_abort(struct sas_task *task)
 	return -EAGAIN;
 }
 
-void sas_task_abort(struct work_struct *work)
+/*
+ * Tell an upper layer that it needs to initiate an abort for a given task.
+ * This should only ever be called by an LLDD.
+ */
+void sas_task_abort(struct sas_task *task)
 {
-	struct sas_task *task =
-		container_of(work, struct sas_task, abort_work);
-	int i;
+	struct scsi_cmnd *sc = task->uldd_task;
 
-	for (i = 0; i < 5; i++)
-		if (!do_sas_task_abort(task))
+	/* Escape for libsas internal commands */
+	if (!sc) {
+		if (!del_timer(&task->timer))
 			return;
+		task->timer.function(task->timer.data);
+		return;
+	}
 
-	SAS_DPRINTK("%s: Could not kill task!\n", __FUNCTION__);
+	scsi_req_abort_cmd(sc);
+	scsi_schedule_eh(sc->device->host);
 }
 
 EXPORT_SYMBOL_GPL(sas_queuecommand);
@@ -866,5 +1002,9 @@ EXPORT_SYMBOL_GPL(sas_slave_destroy);
 EXPORT_SYMBOL_GPL(sas_change_queue_depth);
 EXPORT_SYMBOL_GPL(sas_change_queue_type);
 EXPORT_SYMBOL_GPL(sas_bios_param);
+EXPORT_SYMBOL_GPL(__sas_task_abort);
 EXPORT_SYMBOL_GPL(sas_task_abort);
 EXPORT_SYMBOL_GPL(sas_phy_reset);
+EXPORT_SYMBOL_GPL(sas_phy_enable);
+EXPORT_SYMBOL_GPL(sas_eh_device_reset_handler);
+EXPORT_SYMBOL_GPL(sas_eh_bus_reset_handler);