Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (81 commits)
  RDMA/cxgb3: Fix the T3A workaround checks
  IB/ipath: Remove unnecessary cast
  IPoIB: Constify seq_operations function pointer tables
  RDMA/cxgb3: Mark QP as privileged based on user capabilities
  RDMA/cxgb3: Fix page shift calculation in build_phys_page_list()
  RDMA/cxgb3: Flush the receive queue when closing
  IB/ipath: Trivial simplification of ipath_make_ud_req()
  IB/mthca: Update latest "native Arbel" firmware revision
  IPoIB: Remove redundant check of netif_queue_stopped() in xmit handler
  IB/ipath: Add mappings from HW register to PortInfo port physical state
  IB/ipath: Changes to support PIO bandwidth check on IBA7220
  IB/ipath: Minor cleanup of unused fields and chip-specific errors
  IB/ipath: New sysfs entries to control 7220 features
  IB/ipath: Add new chip-specific functions to older chips, consistent init
  IB/ipath: Remove unused MDIO interface code
  IB/ehca: Prevent RDMA-related connection failures on some eHCA2 hardware
  IB/ehca: Add "port connection autodetect mode"
  IB/ehca: Define array to store SMI/GSI QPs
  IB/ehca: Remove CQ-QP-link before destroying QP in error path of create_qp()
  IB/iser: Add change_queue_depth method
  ...

Documentation/feature-removal-schedule.txt

@@ -295,16 +295,6 @@ Who:  linuxppc-dev@ozlabs.org
 
 ---------------------------
 
-What:	mthca driver's MSI support
-When:	January 2008
-Files:	drivers/infiniband/hw/mthca/*.[ch]
-Why:	All mthca hardware also supports MSI-X, which provides
-	strictly more functionality than MSI.  So there is no point in
-	having both MSI-X and MSI support in the driver.
-Who:	Roland Dreier <rolandd@cisco.com>
-
----------------------------
-
 What:   sk98lin network driver
 When:   Feburary 2008
 Why:    In kernel tree version of driver is unmaintained. Sk98lin driver

drivers/infiniband/core/cma.c

@@ -488,7 +488,8 @@ void rdma_destroy_qp(struct rdma_cm_id *id)
 }
 EXPORT_SYMBOL(rdma_destroy_qp);
 
-static int cma_modify_qp_rtr(struct rdma_id_private *id_priv)
+static int cma_modify_qp_rtr(struct rdma_id_private *id_priv,
+			     struct rdma_conn_param *conn_param)
 {
 	struct ib_qp_attr qp_attr;
 	int qp_attr_mask, ret;
@@ -514,13 +515,16 @@ static int cma_modify_qp_rtr(struct rdma_id_private *id_priv)
 	if (ret)
 		goto out;
 
+	if (conn_param)
+		qp_attr.max_dest_rd_atomic = conn_param->responder_resources;
 	ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
 out:
 	mutex_unlock(&id_priv->qp_mutex);
 	return ret;
 }
 
-static int cma_modify_qp_rts(struct rdma_id_private *id_priv)
+static int cma_modify_qp_rts(struct rdma_id_private *id_priv,
+			     struct rdma_conn_param *conn_param)
 {
 	struct ib_qp_attr qp_attr;
 	int qp_attr_mask, ret;
@@ -536,6 +540,8 @@ static int cma_modify_qp_rts(struct rdma_id_private *id_priv)
 	if (ret)
 		goto out;
 
+	if (conn_param)
+		qp_attr.max_rd_atomic = conn_param->initiator_depth;
 	ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
 out:
 	mutex_unlock(&id_priv->qp_mutex);
@@ -866,11 +872,11 @@ static int cma_rep_recv(struct rdma_id_private *id_priv)
 {
 	int ret;
 
-	ret = cma_modify_qp_rtr(id_priv);
+	ret = cma_modify_qp_rtr(id_priv, NULL);
 	if (ret)
 		goto reject;
 
-	ret = cma_modify_qp_rts(id_priv);
+	ret = cma_modify_qp_rts(id_priv, NULL);
 	if (ret)
 		goto reject;
 
@@ -1122,8 +1128,10 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	cm_id->cm_handler = cma_ib_handler;
 
 	ret = conn_id->id.event_handler(&conn_id->id, &event);
-	if (!ret)
+	if (!ret) {
+		cma_enable_remove(conn_id);
 		goto out;
+	}
 
 	/* Destroy the CM ID by returning a non-zero value. */
 	conn_id->cm_id.ib = NULL;
@@ -1262,6 +1270,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	struct net_device *dev = NULL;
 	struct rdma_cm_event event;
 	int ret;
+	struct ib_device_attr attr;
 
 	listen_id = cm_id->context;
 	if (cma_disable_remove(listen_id, CMA_LISTEN))
@@ -1311,10 +1320,19 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	sin = (struct sockaddr_in *) &new_cm_id->route.addr.dst_addr;
 	*sin = iw_event->remote_addr;
 
+	ret = ib_query_device(conn_id->id.device, &attr);
+	if (ret) {
+		cma_enable_remove(conn_id);
+		rdma_destroy_id(new_cm_id);
+		goto out;
+	}
+
 	memset(&event, 0, sizeof event);
 	event.event = RDMA_CM_EVENT_CONNECT_REQUEST;
 	event.param.conn.private_data = iw_event->private_data;
 	event.param.conn.private_data_len = iw_event->private_data_len;
+	event.param.conn.initiator_depth = attr.max_qp_init_rd_atom;
+	event.param.conn.responder_resources = attr.max_qp_rd_atom;
 	ret = conn_id->id.event_handler(&conn_id->id, &event);
 	if (ret) {
 		/* User wants to destroy the CM ID */
@@ -2272,7 +2290,7 @@ static int cma_connect_iw(struct rdma_id_private *id_priv,
 	sin = (struct sockaddr_in*) &id_priv->id.route.addr.dst_addr;
 	cm_id->remote_addr = *sin;
 
-	ret = cma_modify_qp_rtr(id_priv);
+	ret = cma_modify_qp_rtr(id_priv, conn_param);
 	if (ret)
 		goto out;
 
@@ -2335,25 +2353,15 @@ static int cma_accept_ib(struct rdma_id_private *id_priv,
 			 struct rdma_conn_param *conn_param)
 {
 	struct ib_cm_rep_param rep;
-	struct ib_qp_attr qp_attr;
-	int qp_attr_mask, ret;
-
-	if (id_priv->id.qp) {
-		ret = cma_modify_qp_rtr(id_priv);
-		if (ret)
-			goto out;
+	int ret;
 
-		qp_attr.qp_state = IB_QPS_RTS;
-		ret = ib_cm_init_qp_attr(id_priv->cm_id.ib, &qp_attr,
-					 &qp_attr_mask);
-		if (ret)
-			goto out;
+	ret = cma_modify_qp_rtr(id_priv, conn_param);
+	if (ret)
+		goto out;
 
-		qp_attr.max_rd_atomic = conn_param->initiator_depth;
-		ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
-		if (ret)
-			goto out;
-	}
+	ret = cma_modify_qp_rts(id_priv, conn_param);
+	if (ret)
+		goto out;
 
 	memset(&rep, 0, sizeof rep);
 	rep.qp_num = id_priv->qp_num;
@@ -2378,7 +2386,7 @@ static int cma_accept_iw(struct rdma_id_private *id_priv,
 	struct iw_cm_conn_param iw_param;
 	int ret;
 
-	ret = cma_modify_qp_rtr(id_priv);
+	ret = cma_modify_qp_rtr(id_priv, conn_param);
 	if (ret)
 		return ret;
 
@@ -2598,11 +2606,9 @@ static void cma_set_mgid(struct rdma_id_private *id_priv,
 		/* IPv6 address is an SA assigned MGID. */
 		memcpy(mgid, &sin6->sin6_addr, sizeof *mgid);
 	} else {
-		ip_ib_mc_map(sin->sin_addr.s_addr, mc_map);
+		ip_ib_mc_map(sin->sin_addr.s_addr, dev_addr->broadcast, mc_map);
 		if (id_priv->id.ps == RDMA_PS_UDP)
 			mc_map[7] = 0x01;	/* Use RDMA CM signature */
-		mc_map[8] = ib_addr_get_pkey(dev_addr) >> 8;
-		mc_map[9] = (unsigned char) ib_addr_get_pkey(dev_addr);
 		*mgid = *(union ib_gid *) (mc_map + 4);
 	}
 }

drivers/infiniband/core/fmr_pool.c

@@ -139,7 +139,7 @@ static inline struct ib_pool_fmr *ib_fmr_cache_lookup(struct ib_fmr_pool *pool,
 static void ib_fmr_batch_release(struct ib_fmr_pool *pool)
 {
 	int                 ret;
-	struct ib_pool_fmr *fmr;
+	struct ib_pool_fmr *fmr, *next;
 	LIST_HEAD(unmap_list);
 	LIST_HEAD(fmr_list);
 
@@ -158,6 +158,20 @@ static void ib_fmr_batch_release(struct ib_fmr_pool *pool)
 #endif
 	}
 
+	/*
+	 * The free_list may hold FMRs that have been put there
+	 * because they haven't reached the max_remap count.
+	 * Invalidate their mapping as well.
+	 */
+	list_for_each_entry_safe(fmr, next, &pool->free_list, list) {
+		if (fmr->remap_count == 0)
+			continue;
+		hlist_del_init(&fmr->cache_node);
+		fmr->remap_count = 0;
+		list_add_tail(&fmr->fmr->list, &fmr_list);
+		list_move(&fmr->list, &unmap_list);
+	}
+
 	list_splice(&pool->dirty_list, &unmap_list);
 	INIT_LIST_HEAD(&pool->dirty_list);
 	pool->dirty_len = 0;
@@ -182,8 +196,7 @@ static int ib_fmr_cleanup_thread(void *pool_ptr)
 	struct ib_fmr_pool *pool = pool_ptr;
 
 	do {
-		if (pool->dirty_len >= pool->dirty_watermark ||
-		    atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) < 0) {
+		if (atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) < 0) {
 			ib_fmr_batch_release(pool);
 
 			atomic_inc(&pool->flush_ser);
@@ -194,8 +207,7 @@ static int ib_fmr_cleanup_thread(void *pool_ptr)
 		}
 
 		set_current_state(TASK_INTERRUPTIBLE);
-		if (pool->dirty_len < pool->dirty_watermark &&
-		    atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) >= 0 &&
+		if (atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) >= 0 &&
 		    !kthread_should_stop())
 			schedule();
 		__set_current_state(TASK_RUNNING);
@@ -369,11 +381,6 @@ void ib_destroy_fmr_pool(struct ib_fmr_pool *pool)
 
 	i = 0;
 	list_for_each_entry_safe(fmr, tmp, &pool->free_list, list) {
-		if (fmr->remap_count) {
-			INIT_LIST_HEAD(&fmr_list);
-			list_add_tail(&fmr->fmr->list, &fmr_list);
-			ib_unmap_fmr(&fmr_list);
-		}
 		ib_dealloc_fmr(fmr->fmr);
 		list_del(&fmr->list);
 		kfree(fmr);
@@ -511,8 +518,10 @@ int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr)
 			list_add_tail(&fmr->list, &pool->free_list);
 		} else {
 			list_add_tail(&fmr->list, &pool->dirty_list);
-			++pool->dirty_len;
-			wake_up_process(pool->thread);
+			if (++pool->dirty_len >= pool->dirty_watermark) {
+				atomic_inc(&pool->req_ser);
+				wake_up_process(pool->thread);
+			}
 		}
 	}
 

drivers/infiniband/core/mad.c

@@ -701,7 +701,8 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv,
 	}
 
 	/* Check to post send on QP or process locally */
-	if (smi_check_local_smp(smp, device) == IB_SMI_DISCARD)
+	if (smi_check_local_smp(smp, device) == IB_SMI_DISCARD &&
+	    smi_check_local_returning_smp(smp, device) == IB_SMI_DISCARD)
 		goto out;
 
 	local = kmalloc(sizeof *local, GFP_ATOMIC);
@@ -752,8 +753,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv,
 		port_priv = ib_get_mad_port(mad_agent_priv->agent.device,
 					    mad_agent_priv->agent.port_num);
 		if (port_priv) {
-			mad_priv->mad.mad.mad_hdr.tid =
-				((struct ib_mad *)smp)->mad_hdr.tid;
+			memcpy(&mad_priv->mad.mad, smp, sizeof(struct ib_mad));
 			recv_mad_agent = find_mad_agent(port_priv,
 						        &mad_priv->mad.mad);
 		}
@@ -1100,7 +1100,9 @@ int ib_post_send_mad(struct ib_mad_send_buf *send_buf,
 		mad_send_wr->tid = ((struct ib_mad_hdr *) send_buf->mad)->tid;
 		/* Timeout will be updated after send completes */
 		mad_send_wr->timeout = msecs_to_jiffies(send_buf->timeout_ms);
-		mad_send_wr->retries = send_buf->retries;
+		mad_send_wr->max_retries = send_buf->retries;
+		mad_send_wr->retries_left = send_buf->retries;
+		send_buf->retries = 0;
 		/* Reference for work request to QP + response */
 		mad_send_wr->refcount = 1 + (mad_send_wr->timeout > 0);
 		mad_send_wr->status = IB_WC_SUCCESS;
@@ -1931,15 +1933,6 @@ static void ib_mad_recv_done_handler(struct ib_mad_port_private *port_priv,
 	if (port_priv->device->process_mad) {
 		int ret;
 
-		if (!response) {
-			printk(KERN_ERR PFX "No memory for response MAD\n");
-			/*
-			 * Is it better to assume that
-			 * it wouldn't be processed ?
-			 */
-			goto out;
-		}
-
 		ret = port_priv->device->process_mad(port_priv->device, 0,
 						     port_priv->port_num,
 						     wc, &recv->grh,
@@ -2282,8 +2275,6 @@ static void cancel_mads(struct ib_mad_agent_private *mad_agent_priv)
 
 	/* Empty wait list to prevent receives from finding a request */
 	list_splice_init(&mad_agent_priv->wait_list, &cancel_list);
-	/* Empty local completion list as well */
-	list_splice_init(&mad_agent_priv->local_list, &cancel_list);
 	spin_unlock_irqrestore(&mad_agent_priv->lock, flags);
 
 	/* Report all cancelled requests */
@@ -2445,9 +2436,12 @@ static int retry_send(struct ib_mad_send_wr_private *mad_send_wr)
 {
 	int ret;
 
-	if (!mad_send_wr->retries--)
+	if (!mad_send_wr->retries_left)
 		return -ETIMEDOUT;
 
+	mad_send_wr->retries_left--;
+	mad_send_wr->send_buf.retries++;
+
 	mad_send_wr->timeout = msecs_to_jiffies(mad_send_wr->send_buf.timeout_ms);
 
 	if (mad_send_wr->mad_agent_priv->agent.rmpp_version) {

drivers/infiniband/core/mad_priv.h

@@ -131,7 +131,8 @@ struct ib_mad_send_wr_private {
 	struct ib_sge sg_list[IB_MAD_SEND_REQ_MAX_SG];
 	__be64 tid;
 	unsigned long timeout;
-	int retries;
+	int max_retries;
+	int retries_left;
 	int retry;
 	int refcount;
 	enum ib_wc_status status;

drivers/infiniband/core/mad_rmpp.c

@@ -684,7 +684,7 @@ static void process_rmpp_ack(struct ib_mad_agent_private *agent,
 
 	if (seg_num > mad_send_wr->last_ack) {
 		adjust_last_ack(mad_send_wr, seg_num);
-		mad_send_wr->retries = mad_send_wr->send_buf.retries;
+		mad_send_wr->retries_left = mad_send_wr->max_retries;
 	}
 	mad_send_wr->newwin = newwin;
 	if (mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) {

drivers/infiniband/core/multicast.c

@@ -73,11 +73,20 @@ struct mcast_device {
 };
 
 enum mcast_state {
-	MCAST_IDLE,
 	MCAST_JOINING,
 	MCAST_MEMBER,
+	MCAST_ERROR,
+};
+
+enum mcast_group_state {
+	MCAST_IDLE,
 	MCAST_BUSY,
-	MCAST_ERROR
+	MCAST_GROUP_ERROR,
+	MCAST_PKEY_EVENT
+};
+
+enum {
+	MCAST_INVALID_PKEY_INDEX = 0xFFFF
 };
 
 struct mcast_member;
@@ -93,9 +102,10 @@ struct mcast_group {
 	struct mcast_member	*last_join;
 	int			members[3];
 	atomic_t		refcount;
-	enum mcast_state	state;
+	enum mcast_group_state	state;
 	struct ib_sa_query	*query;
 	int			query_id;
+	u16			pkey_index;
 };
 
 struct mcast_member {
@@ -378,9 +388,19 @@ static int fail_join(struct mcast_group *group, struct mcast_member *member,
 static void process_group_error(struct mcast_group *group)
 {
 	struct mcast_member *member;
-	int ret;
+	int ret = 0;
+	u16 pkey_index;
+
+	if (group->state == MCAST_PKEY_EVENT)
+		ret = ib_find_pkey(group->port->dev->device,
+				   group->port->port_num,
+				   be16_to_cpu(group->rec.pkey), &pkey_index);
 
 	spin_lock_irq(&group->lock);
+	if (group->state == MCAST_PKEY_EVENT && !ret &&
+	    group->pkey_index == pkey_index)
+		goto out;
+
 	while (!list_empty(&group->active_list)) {
 		member = list_entry(group->active_list.next,
 				    struct mcast_member, list);
@@ -399,6 +419,7 @@ static void process_group_error(struct mcast_group *group)
 	}
 
 	group->rec.join_state = 0;
+out:
 	group->state = MCAST_BUSY;
 	spin_unlock_irq(&group->lock);
 }
@@ -415,9 +436,9 @@ static void mcast_work_handler(struct work_struct *work)
 retest:
 	spin_lock_irq(&group->lock);
 	while (!list_empty(&group->pending_list) ||
-	       (group->state == MCAST_ERROR)) {
+	       (group->state != MCAST_BUSY)) {
 
-		if (group->state == MCAST_ERROR) {
+		if (group->state != MCAST_BUSY) {
 			spin_unlock_irq(&group->lock);
 			process_group_error(group);
 			goto retest;
@@ -494,12 +515,19 @@ static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
 			 void *context)
 {
 	struct mcast_group *group = context;
+	u16 pkey_index = MCAST_INVALID_PKEY_INDEX;
 
 	if (status)
 		process_join_error(group, status);
 	else {
+		ib_find_pkey(group->port->dev->device, group->port->port_num,
+			     be16_to_cpu(rec->pkey), &pkey_index);
+
 		spin_lock_irq(&group->port->lock);
 		group->rec = *rec;
+		if (group->state == MCAST_BUSY &&
+		    group->pkey_index == MCAST_INVALID_PKEY_INDEX)
+			group->pkey_index = pkey_index;
 		if (!memcmp(&mgid0, &group->rec.mgid, sizeof mgid0)) {
 			rb_erase(&group->node, &group->port->table);
 			mcast_insert(group->port, group, 1);
@@ -539,6 +567,7 @@ static struct mcast_group *acquire_group(struct mcast_port *port,
 
 	group->port = port;
 	group->rec.mgid = *mgid;
+	group->pkey_index = MCAST_INVALID_PKEY_INDEX;
 	INIT_LIST_HEAD(&group->pending_list);
 	INIT_LIST_HEAD(&group->active_list);
 	INIT_WORK(&group->work, mcast_work_handler);
@@ -707,7 +736,8 @@ int ib_init_ah_from_mcmember(struct ib_device *device, u8 port_num,
 }
 EXPORT_SYMBOL(ib_init_ah_from_mcmember);
 
-static void mcast_groups_lost(struct mcast_port *port)
+static void mcast_groups_event(struct mcast_port *port,
+			       enum mcast_group_state state)
 {
 	struct mcast_group *group;
 	struct rb_node *node;
@@ -721,7 +751,8 @@ static void mcast_groups_lost(struct mcast_port *port)
 			atomic_inc(&group->refcount);
 			queue_work(mcast_wq, &group->work);
 		}
-		group->state = MCAST_ERROR;
+		if (group->state != MCAST_GROUP_ERROR)
+			group->state = state;
 		spin_unlock(&group->lock);
 	}
 	spin_unlock_irqrestore(&port->lock, flags);
@@ -731,16 +762,20 @@ static void mcast_event_handler(struct ib_event_handler *handler,
 				struct ib_event *event)
 {
 	struct mcast_device *dev;
+	int index;
 
 	dev = container_of(handler, struct mcast_device, event_handler);
+	index = event->element.port_num - dev->start_port;
 
 	switch (event->event) {
 	case IB_EVENT_PORT_ERR:
 	case IB_EVENT_LID_CHANGE:
 	case IB_EVENT_SM_CHANGE:
 	case IB_EVENT_CLIENT_REREGISTER:
-		mcast_groups_lost(&dev->port[event->element.port_num -
-					     dev->start_port]);
+		mcast_groups_event(&dev->port[index], MCAST_GROUP_ERROR);
+		break;
+	case IB_EVENT_PKEY_CHANGE:
+		mcast_groups_event(&dev->port[index], MCAST_PKEY_EVENT);
 		break;
 	default:
 		break;

drivers/infiniband/core/smi.h

@@ -59,7 +59,8 @@ extern enum smi_action smi_handle_dr_smp_send(struct ib_smp *smp,
 					      u8 node_type, int port_num);
 
 /*
- * Return 1 if the SMP should be handled by the local SMA/SM via process_mad
+ * Return IB_SMI_HANDLE if the SMP should be handled by the local SMA/SM
+ * via process_mad
  */
 static inline enum smi_action smi_check_local_smp(struct ib_smp *smp,
 						  struct ib_device *device)
@@ -71,4 +72,19 @@ static inline enum smi_action smi_check_local_smp(struct ib_smp *smp,
 		(smp->hop_ptr == smp->hop_cnt + 1)) ?
 		IB_SMI_HANDLE : IB_SMI_DISCARD);
 }
+
+/*
+ * Return IB_SMI_HANDLE if the SMP should be handled by the local SMA/SM
+ * via process_mad
+ */
+static inline enum smi_action smi_check_local_returning_smp(struct ib_smp *smp,
+						   struct ib_device *device)
+{
+	/* C14-13:3 -- We're at the end of the DR segment of path */
+	/* C14-13:4 -- Hop Pointer == 0 -> give to SM */
+	return ((device->process_mad &&
+		ib_get_smp_direction(smp) &&
+		!smp->hop_ptr) ? IB_SMI_HANDLE : IB_SMI_DISCARD);
+}
+
 #endif	/* __SMI_H_ */

drivers/infiniband/core/ucm.c

@@ -106,6 +106,9 @@ enum {
 	IB_UCM_MAX_DEVICES = 32
 };
 
+/* ib_cm and ib_user_cm modules share /sys/class/infiniband_cm */
+extern struct class cm_class;
+
 #define IB_UCM_BASE_DEV MKDEV(IB_UCM_MAJOR, IB_UCM_BASE_MINOR)
 
 static void ib_ucm_add_one(struct ib_device *device);
@@ -1199,7 +1202,7 @@ static int ib_ucm_close(struct inode *inode, struct file *filp)
 	return 0;
 }
 
-static void ib_ucm_release_class_dev(struct class_device *class_dev)
+static void ucm_release_class_dev(struct class_device *class_dev)
 {
 	struct ib_ucm_device *dev;
 
@@ -1217,11 +1220,6 @@ static const struct file_operations ucm_fops = {
 	.poll    = ib_ucm_poll,
 };
 
-static struct class ucm_class = {
-	.name    = "infiniband_cm",
-	.release = ib_ucm_release_class_dev
-};
-
 static ssize_t show_ibdev(struct class_device *class_dev, char *buf)
 {
 	struct ib_ucm_device *dev;
@@ -1257,9 +1255,10 @@ static void ib_ucm_add_one(struct ib_device *device)
 	if (cdev_add(&ucm_dev->dev, IB_UCM_BASE_DEV + ucm_dev->devnum, 1))
 		goto err;
 
-	ucm_dev->class_dev.class = &ucm_class;
+	ucm_dev->class_dev.class = &cm_class;
 	ucm_dev->class_dev.dev = device->dma_device;
 	ucm_dev->class_dev.devt = ucm_dev->dev.dev;
+	ucm_dev->class_dev.release = ucm_release_class_dev;
 	snprintf(ucm_dev->class_dev.class_id, BUS_ID_SIZE, "ucm%d",
 		 ucm_dev->devnum);
 	if (class_device_register(&ucm_dev->class_dev))
@@ -1306,40 +1305,34 @@ static int __init ib_ucm_init(void)
 				     "infiniband_cm");
 	if (ret) {
 		printk(KERN_ERR "ucm: couldn't register device number\n");
-		goto err;
+		goto error1;
 	}
 
-	ret = class_register(&ucm_class);
-	if (ret) {
-		printk(KERN_ERR "ucm: couldn't create class infiniband_cm\n");
-		goto err_chrdev;
-	}
-
-	ret = class_create_file(&ucm_class, &class_attr_abi_version);
+	ret = class_create_file(&cm_class, &class_attr_abi_version);
 	if (ret) {
 		printk(KERN_ERR "ucm: couldn't create abi_version attribute\n");
-		goto err_class;
+		goto error2;
 	}
 
 	ret = ib_register_client(&ucm_client);
 	if (ret) {
 		printk(KERN_ERR "ucm: couldn't register client\n");
-		goto err_class;
+		goto error3;
 	}
 	return 0;
 
-err_class:
-	class_unregister(&ucm_class);
-err_chrdev:
+error3:
+	class_remove_file(&cm_class, &class_attr_abi_version);
+error2:
 	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES);
-err:
+error1:
 	return ret;
 }
 
 static void __exit ib_ucm_cleanup(void)
 {
 	ib_unregister_client(&ucm_client);
-	class_unregister(&ucm_class);
+	class_remove_file(&cm_class, &class_attr_abi_version);
 	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES);
 	idr_destroy(&ctx_id_table);
 }

drivers/infiniband/core/ucma.c

@@ -31,6 +31,7 @@
  */
 
 #include <linux/completion.h>
+#include <linux/file.h>
 #include <linux/mutex.h>
 #include <linux/poll.h>
 #include <linux/idr.h>
@@ -991,6 +992,96 @@ static ssize_t ucma_leave_multicast(struct ucma_file *file,
 	return ret;
 }
 
+static void ucma_lock_files(struct ucma_file *file1, struct ucma_file *file2)
+{
+	/* Acquire mutex's based on pointer comparison to prevent deadlock. */
+	if (file1 < file2) {
+		mutex_lock(&file1->mut);
+		mutex_lock(&file2->mut);
+	} else {
+		mutex_lock(&file2->mut);
+		mutex_lock(&file1->mut);
+	}
+}
+
+static void ucma_unlock_files(struct ucma_file *file1, struct ucma_file *file2)
+{
+	if (file1 < file2) {
+		mutex_unlock(&file2->mut);
+		mutex_unlock(&file1->mut);
+	} else {
+		mutex_unlock(&file1->mut);
+		mutex_unlock(&file2->mut);
+	}
+}
+
+static void ucma_move_events(struct ucma_context *ctx, struct ucma_file *file)
+{
+	struct ucma_event *uevent, *tmp;
+
+	list_for_each_entry_safe(uevent, tmp, &ctx->file->event_list, list)
+		if (uevent->ctx == ctx)
+			list_move_tail(&uevent->list, &file->event_list);
+}
+
+static ssize_t ucma_migrate_id(struct ucma_file *new_file,
+			       const char __user *inbuf,
+			       int in_len, int out_len)
+{
+	struct rdma_ucm_migrate_id cmd;
+	struct rdma_ucm_migrate_resp resp;
+	struct ucma_context *ctx;
+	struct file *filp;
+	struct ucma_file *cur_file;
+	int ret = 0;
+
+	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+		return -EFAULT;
+
+	/* Get current fd to protect against it being closed */
+	filp = fget(cmd.fd);
+	if (!filp)
+		return -ENOENT;
+
+	/* Validate current fd and prevent destruction of id. */
+	ctx = ucma_get_ctx(filp->private_data, cmd.id);
+	if (IS_ERR(ctx)) {
+		ret = PTR_ERR(ctx);
+		goto file_put;
+	}
+
+	cur_file = ctx->file;
+	if (cur_file == new_file) {
+		resp.events_reported = ctx->events_reported;
+		goto response;
+	}
+
+	/*
+	 * Migrate events between fd's, maintaining order, and avoiding new
+	 * events being added before existing events.
+	 */
+	ucma_lock_files(cur_file, new_file);
+	mutex_lock(&mut);
+
+	list_move_tail(&ctx->list, &new_file->ctx_list);
+	ucma_move_events(ctx, new_file);
+	ctx->file = new_file;
+	resp.events_reported = ctx->events_reported;
+
+	mutex_unlock(&mut);
+	ucma_unlock_files(cur_file, new_file);
+
+response:
+	if (copy_to_user((void __user *)(unsigned long)cmd.response,
+			 &resp, sizeof(resp)))
+		ret = -EFAULT;
+
+	ucma_put_ctx(ctx);
+file_put:
+	fput(filp);
+	return ret;
+}
+
 static ssize_t (*ucma_cmd_table[])(struct ucma_file *file,
 				   const char __user *inbuf,
 				   int in_len, int out_len) = {
@@ -1012,6 +1103,7 @@ static ssize_t (*ucma_cmd_table[])(struct ucma_file *file,
 	[RDMA_USER_CM_CMD_NOTIFY]	= ucma_notify,
 	[RDMA_USER_CM_CMD_JOIN_MCAST]	= ucma_join_multicast,
 	[RDMA_USER_CM_CMD_LEAVE_MCAST]	= ucma_leave_multicast,
+	[RDMA_USER_CM_CMD_MIGRATE_ID]	= ucma_migrate_id
 };
 
 static ssize_t ucma_write(struct file *filp, const char __user *buf,

drivers/infiniband/hw/cxgb3/cxio_hal.c

@@ -179,7 +179,7 @@ int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
 	setup.size = 1UL << cq->size_log2;
 	setup.credits = 65535;
 	setup.credit_thres = 1;
-	if (rdev_p->t3cdev_p->type == T3B)
+	if (rdev_p->t3cdev_p->type != T3A)
 		setup.ovfl_mode = 0;
 	else
 		setup.ovfl_mode = 1;
@@ -584,7 +584,7 @@ static int cxio_hal_ctrl_qp_write_mem(struct cxio_rdev *rdev_p, u32 addr,
 {
 	u32 i, nr_wqe, copy_len;
 	u8 *copy_data;
-	u8 wr_len, utx_len;	/* lenght in 8 byte flit */
+	u8 wr_len, utx_len;	/* length in 8 byte flit */
 	enum t3_wr_flags flag;
 	__be64 *wqe;
 	u64 utx_cmd;

drivers/infiniband/hw/cxgb3/cxio_wr.h

@@ -315,7 +315,7 @@ struct t3_rdma_init_wr {
 	__be32 ird;
 	__be64 qp_dma_addr;	/* 7 */
 	__be32 qp_dma_size;	/* 8 */
-	u32 irs;
+	__be32 irs;
 };
 
 struct t3_genbit {
@@ -324,7 +324,8 @@ struct t3_genbit {
 };
 
 enum rdma_init_wr_flags {
-	RECVS_POSTED = 1,
+	RECVS_POSTED = (1<<0),
+	PRIV_QP = (1<<1),
 };
 
 union t3_wr {

drivers/infiniband/hw/cxgb3/iwch_cm.c

@@ -1118,7 +1118,7 @@ static int act_open_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
 	     status2errno(rpl->status));
 	connect_reply_upcall(ep, status2errno(rpl->status));
 	state_set(&ep->com, DEAD);
-	if (ep->com.tdev->type == T3B && act_open_has_tid(rpl->status))
+	if (ep->com.tdev->type != T3A && act_open_has_tid(rpl->status))
 		release_tid(ep->com.tdev, GET_TID(rpl), NULL);
 	cxgb3_free_atid(ep->com.tdev, ep->atid);
 	dst_release(ep->dst);
@@ -1249,7 +1249,7 @@ static void reject_cr(struct t3cdev *tdev, u32 hwtid, __be32 peer_ip,
 	skb_trim(skb, sizeof(struct cpl_tid_release));
 	skb_get(skb);
 
-	if (tdev->type == T3B)
+	if (tdev->type != T3A)
 		release_tid(tdev, hwtid, skb);
 	else {
 		struct cpl_pass_accept_rpl *rpl;

drivers/infiniband/hw/cxgb3/iwch_mem.c

@@ -122,6 +122,13 @@ int build_phys_page_list(struct ib_phys_buf *buffer_list,
 		*total_size += buffer_list[i].size;
 		if (i > 0)
 			mask |= buffer_list[i].addr;
+		else
+			mask |= buffer_list[i].addr & PAGE_MASK;
+		if (i != num_phys_buf - 1)
+			mask |= buffer_list[i].addr + buffer_list[i].size;
+		else
+			mask |= (buffer_list[i].addr + buffer_list[i].size +
+				PAGE_SIZE - 1) & PAGE_MASK;
 	}
 
 	if (*total_size > 0xFFFFFFFFULL)

drivers/infiniband/hw/cxgb3/iwch_provider.c

@@ -39,6 +39,7 @@
 #include <linux/list.h>
 #include <linux/spinlock.h>
 #include <linux/ethtool.h>
+#include <linux/rtnetlink.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -645,7 +646,7 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	if (err)
 		goto err;
 
-	if (udata && t3b_device(rhp)) {
+	if (udata && !t3a_device(rhp)) {
 		uresp.pbl_addr = (mhp->attr.pbl_addr -
 	                         rhp->rdev.rnic_info.pbl_base) >> 3;
 		PDBG("%s user resp pbl_addr 0x%x\n", __FUNCTION__,
@@ -1053,7 +1054,9 @@ static ssize_t show_fw_ver(struct class_device *cdev, char *buf)
 	struct net_device *lldev = dev->rdev.t3cdev_p->lldev;
 
 	PDBG("%s class dev 0x%p\n", __FUNCTION__, cdev);
+	rtnl_lock();
 	lldev->ethtool_ops->get_drvinfo(lldev, &info);
+	rtnl_unlock();
 	return sprintf(buf, "%s\n", info.fw_version);
 }
 
@@ -1065,7 +1068,9 @@ static ssize_t show_hca(struct class_device *cdev, char *buf)
 	struct net_device *lldev = dev->rdev.t3cdev_p->lldev;
 
 	PDBG("%s class dev 0x%p\n", __FUNCTION__, cdev);
+	rtnl_lock();
 	lldev->ethtool_ops->get_drvinfo(lldev, &info);
+	rtnl_unlock();
 	return sprintf(buf, "%s\n", info.driver);
 }
 

drivers/infiniband/hw/cxgb3/iwch_qp.c

@@ -208,36 +208,19 @@ static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
 static int iwch_build_rdma_recv(struct iwch_dev *rhp, union t3_wr *wqe,
 				struct ib_recv_wr *wr)
 {
-	int i, err = 0;
-	u32 pbl_addr[4];
-	u8 page_size[4];
+	int i;
 	if (wr->num_sge > T3_MAX_SGE)
 		return -EINVAL;
-	err = iwch_sgl2pbl_map(rhp, wr->sg_list, wr->num_sge, pbl_addr,
-			       page_size);
-	if (err)
-		return err;
-	wqe->recv.pagesz[0] = page_size[0];
-	wqe->recv.pagesz[1] = page_size[1];
-	wqe->recv.pagesz[2] = page_size[2];
-	wqe->recv.pagesz[3] = page_size[3];
 	wqe->recv.num_sgle = cpu_to_be32(wr->num_sge);
 	for (i = 0; i < wr->num_sge; i++) {
 		wqe->recv.sgl[i].stag = cpu_to_be32(wr->sg_list[i].lkey);
 		wqe->recv.sgl[i].len = cpu_to_be32(wr->sg_list[i].length);
-
-		/* to in the WQE == the offset into the page */
-		wqe->recv.sgl[i].to = cpu_to_be64(((u32) wr->sg_list[i].addr) %
-				(1UL << (12 + page_size[i])));
-
-		/* pbl_addr is the adapters address in the PBL */
-		wqe->recv.pbl_addr[i] = cpu_to_be32(pbl_addr[i]);
+		wqe->recv.sgl[i].to = cpu_to_be64(wr->sg_list[i].addr);
 	}
 	for (; i < T3_MAX_SGE; i++) {
 		wqe->recv.sgl[i].stag = 0;
 		wqe->recv.sgl[i].len = 0;
 		wqe->recv.sgl[i].to = 0;
-		wqe->recv.pbl_addr[i] = 0;
 	}
 	return 0;
 }
@@ -659,6 +642,7 @@ static void __flush_qp(struct iwch_qp *qhp, unsigned long *flag)
 	cxio_flush_rq(&qhp->wq, &rchp->cq, count);
 	spin_unlock(&qhp->lock);
 	spin_unlock_irqrestore(&rchp->lock, *flag);
+	(*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context);
 
 	/* locking heirarchy: cq lock first, then qp lock. */
 	spin_lock_irqsave(&schp->lock, *flag);
@@ -668,6 +652,7 @@ static void __flush_qp(struct iwch_qp *qhp, unsigned long *flag)
 	cxio_flush_sq(&qhp->wq, &schp->cq, count);
 	spin_unlock(&qhp->lock);
 	spin_unlock_irqrestore(&schp->lock, *flag);
+	(*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context);
 
 	/* deref */
 	if (atomic_dec_and_test(&qhp->refcnt))
@@ -678,7 +663,7 @@ static void __flush_qp(struct iwch_qp *qhp, unsigned long *flag)
 
 static void flush_qp(struct iwch_qp *qhp, unsigned long *flag)
 {
-	if (t3b_device(qhp->rhp))
+	if (qhp->ibqp.uobject)
 		cxio_set_wq_in_error(&qhp->wq);
 	else
 		__flush_qp(qhp, flag);
@@ -732,6 +717,7 @@ static int rdma_init(struct iwch_dev *rhp, struct iwch_qp *qhp,
 	init_attr.qp_dma_addr = qhp->wq.dma_addr;
 	init_attr.qp_dma_size = (1UL << qhp->wq.size_log2);
 	init_attr.flags = rqes_posted(qhp) ? RECVS_POSTED : 0;
+	init_attr.flags |= capable(CAP_NET_BIND_SERVICE) ? PRIV_QP : 0;
 	init_attr.irs = qhp->ep->rcv_seq;
 	PDBG("%s init_attr.rq_addr 0x%x init_attr.rq_size = %d "
 	     "flags 0x%x qpcaps 0x%x\n", __FUNCTION__,
@@ -847,10 +833,11 @@ int iwch_modify_qp(struct iwch_dev *rhp, struct iwch_qp *qhp,
 				disconnect = 1;
 				ep = qhp->ep;
 			}
+			flush_qp(qhp, &flag);
 			break;
 		case IWCH_QP_STATE_TERMINATE:
 			qhp->attr.state = IWCH_QP_STATE_TERMINATE;
-			if (t3b_device(qhp->rhp))
+			if (qhp->ibqp.uobject)
 				cxio_set_wq_in_error(&qhp->wq);
 			if (!internal)
 				terminate = 1;

drivers/infiniband/hw/ehca/ehca_av.c

 /*
  *  IBM eServer eHCA Infiniband device driver for Linux on POWER
  *
- *  adress vector functions
+ *  address vector functions
  *
  *  Authors: Hoang-Nam Nguyen <hnguyen@de.ibm.com>
  *           Khadija Souissi <souissik@de.ibm.com>

drivers/infiniband/hw/ehca/ehca_classes.h

@@ -94,7 +94,11 @@ struct ehca_sma_attr {
 
 struct ehca_sport {
 	struct ib_cq *ibcq_aqp1;
-	struct ib_qp *ibqp_aqp1;
+	struct ib_qp *ibqp_sqp[2];
+	/* lock to serialze modify_qp() calls for sqp in normal
+	 * and irq path (when event PORT_ACTIVE is received first time)
+	 */
+	spinlock_t mod_sqp_lock;
 	enum ib_port_state port_state;
 	struct ehca_sma_attr saved_attr;
 };
@@ -141,6 +145,14 @@ enum ehca_ext_qp_type {
 	EQPT_SRQ       = 3,
 };
 
+/* struct to cache modify_qp()'s parms for GSI/SMI qp */
+struct ehca_mod_qp_parm {
+	int mask;
+	struct ib_qp_attr attr;
+};
+
+#define EHCA_MOD_QP_PARM_MAX 4
+
 struct ehca_qp {
 	union {
 		struct ib_qp ib_qp;
@@ -164,10 +176,18 @@ struct ehca_qp {
 	struct ehca_cq *recv_cq;
 	unsigned int sqerr_purgeflag;
 	struct hlist_node list_entries;
+	/* array to cache modify_qp()'s parms for GSI/SMI qp */
+	struct ehca_mod_qp_parm *mod_qp_parm;
+	int mod_qp_parm_idx;
 	/* mmap counter for resources mapped into user space */
 	u32 mm_count_squeue;
 	u32 mm_count_rqueue;
 	u32 mm_count_galpa;
+	/* unsolicited ack circumvention */
+	int unsol_ack_circ;
+	int mtu_shift;
+	u32 message_count;
+	u32 packet_count;
 };
 
 #define IS_SRQ(qp) (qp->ext_type == EQPT_SRQ)
@@ -323,6 +343,7 @@ extern int ehca_port_act_time;
 extern int ehca_use_hp_mr;
 extern int ehca_scaling_code;
 extern int ehca_lock_hcalls;
+extern int ehca_nr_ports;
 
 struct ipzu_queue_resp {
 	u32 qe_size;      /* queue entry size */

drivers/infiniband/hw/ehca/ehca_cq.c

@@ -246,7 +246,7 @@ struct ib_cq *ehca_create_cq(struct ib_device *device, int cqe, int comp_vector,
 		} else {
 			if (h_ret != H_PAGE_REGISTERED) {
 				ehca_err(device, "Registration of page failed "
-					 "ehca_cq=%p cq_num=%x h_ret=%li"
+					 "ehca_cq=%p cq_num=%x h_ret=%li "
 					 "counter=%i act_pages=%i",
 					 my_cq, my_cq->cq_number,
 					 h_ret, counter, param.act_pages);

drivers/infiniband/hw/ehca/ehca_irq.c

@@ -62,6 +62,7 @@
 #define NEQE_PORT_NUMBER       EHCA_BMASK_IBM( 8, 15)
 #define NEQE_PORT_AVAILABILITY EHCA_BMASK_IBM(16, 16)
 #define NEQE_DISRUPTIVE        EHCA_BMASK_IBM(16, 16)
+#define NEQE_SPECIFIC_EVENT    EHCA_BMASK_IBM(16, 23)
 
 #define ERROR_DATA_LENGTH      EHCA_BMASK_IBM(52, 63)
 #define ERROR_DATA_TYPE        EHCA_BMASK_IBM( 0,  7)
@@ -354,17 +355,34 @@ static void parse_ec(struct ehca_shca *shca, u64 eqe)
 {
 	u8 ec   = EHCA_BMASK_GET(NEQE_EVENT_CODE, eqe);
 	u8 port = EHCA_BMASK_GET(NEQE_PORT_NUMBER, eqe);
+	u8 spec_event;
+	struct ehca_sport *sport = &shca->sport[port - 1];
+	unsigned long flags;
 
 	switch (ec) {
 	case 0x30: /* port availability change */
 		if (EHCA_BMASK_GET(NEQE_PORT_AVAILABILITY, eqe)) {
-			shca->sport[port - 1].port_state = IB_PORT_ACTIVE;
+			int suppress_event;
+			/* replay modify_qp for sqps */
+			spin_lock_irqsave(&sport->mod_sqp_lock, flags);
+			suppress_event = !sport->ibqp_sqp[IB_QPT_GSI];
+			if (sport->ibqp_sqp[IB_QPT_SMI])
+				ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_SMI]);
+			if (!suppress_event)
+				ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_GSI]);
+			spin_unlock_irqrestore(&sport->mod_sqp_lock, flags);
+
+			/* AQP1 was destroyed, ignore this event */
+			if (suppress_event)
+				break;
+
+			sport->port_state = IB_PORT_ACTIVE;
 			dispatch_port_event(shca, port, IB_EVENT_PORT_ACTIVE,
 					    "is active");
 			ehca_query_sma_attr(shca, port,
-					    &shca->sport[port - 1].saved_attr);
+					    &sport->saved_attr);
 		} else {
-			shca->sport[port - 1].port_state = IB_PORT_DOWN;
+			sport->port_state = IB_PORT_DOWN;
 			dispatch_port_event(shca, port, IB_EVENT_PORT_ERR,
 					    "is inactive");
 		}
@@ -378,11 +396,11 @@ static void parse_ec(struct ehca_shca *shca, u64 eqe)
 			ehca_warn(&shca->ib_device, "disruptive port "
 				  "%d configuration change", port);
 
-			shca->sport[port - 1].port_state = IB_PORT_DOWN;
+			sport->port_state = IB_PORT_DOWN;
 			dispatch_port_event(shca, port, IB_EVENT_PORT_ERR,
 					    "is inactive");
 
-			shca->sport[port - 1].port_state = IB_PORT_ACTIVE;
+			sport->port_state = IB_PORT_ACTIVE;
 			dispatch_port_event(shca, port, IB_EVENT_PORT_ACTIVE,
 					    "is active");
 		} else
@@ -394,6 +412,16 @@ static void parse_ec(struct ehca_shca *shca, u64 eqe)
 	case 0x33:  /* trace stopped */
 		ehca_err(&shca->ib_device, "Traced stopped.");
 		break;
+	case 0x34: /* util async event */
+		spec_event = EHCA_BMASK_GET(NEQE_SPECIFIC_EVENT, eqe);
+		if (spec_event == 0x80) /* client reregister required */
+			dispatch_port_event(shca, port,
+					    IB_EVENT_CLIENT_REREGISTER,
+					    "client reregister req.");
+		else
+			ehca_warn(&shca->ib_device, "Unknown util async "
+				  "event %x on port %x", spec_event, port);
+		break;
 	default:
 		ehca_err(&shca->ib_device, "Unknown event code: %x on %s.",
 			 ec, shca->ib_device.name);

drivers/infiniband/hw/ehca/ehca_iverbs.h

@@ -200,4 +200,6 @@ void ehca_free_fw_ctrlblock(void *ptr);
 #define ehca_free_fw_ctrlblock(ptr) free_page((unsigned long)(ptr))
 #endif
 
+void ehca_recover_sqp(struct ib_qp *sqp);
+
 #endif

drivers/infiniband/hw/ehca/ehca_main.c

@@ -90,7 +90,8 @@ MODULE_PARM_DESC(hw_level,
 		 "hardware level"
 		 " (0: autosensing (default), 1: v. 0.20, 2: v. 0.21)");
 MODULE_PARM_DESC(nr_ports,
-		 "number of connected ports (default: 2)");
+		 "number of connected ports (-1: autodetect, 1: port one only, "
+		 "2: two ports (default)");
 MODULE_PARM_DESC(use_hp_mr,
 		 "high performance MRs (0: no (default), 1: yes)");
 MODULE_PARM_DESC(port_act_time,
@@ -511,7 +512,7 @@ static int ehca_create_aqp1(struct ehca_shca *shca, u32 port)
 	}
 	sport->ibcq_aqp1 = ibcq;
 
-	if (sport->ibqp_aqp1) {
+	if (sport->ibqp_sqp[IB_QPT_GSI]) {
 		ehca_err(&shca->ib_device, "AQP1 QP is already created.");
 		ret = -EPERM;
 		goto create_aqp1;
@@ -537,7 +538,7 @@ static int ehca_create_aqp1(struct ehca_shca *shca, u32 port)
 		ret = PTR_ERR(ibqp);
 		goto create_aqp1;
 	}
-	sport->ibqp_aqp1 = ibqp;
+	sport->ibqp_sqp[IB_QPT_GSI] = ibqp;
 
 	return 0;
 
@@ -550,7 +551,7 @@ static int ehca_destroy_aqp1(struct ehca_sport *sport)
 {
 	int ret;
 
-	ret = ib_destroy_qp(sport->ibqp_aqp1);
+	ret = ib_destroy_qp(sport->ibqp_sqp[IB_QPT_GSI]);
 	if (ret) {
 		ehca_gen_err("Cannot destroy AQP1 QP. ret=%i", ret);
 		return ret;
@@ -693,7 +694,7 @@ static int __devinit ehca_probe(struct of_device *dev,
 	struct ehca_shca *shca;
 	const u64 *handle;
 	struct ib_pd *ibpd;
-	int ret;
+	int ret, i;
 
 	handle = of_get_property(dev->node, "ibm,hca-handle", NULL);
 	if (!handle) {
@@ -714,6 +715,8 @@ static int __devinit ehca_probe(struct of_device *dev,
 		return -ENOMEM;
 	}
 	mutex_init(&shca->modify_mutex);
+	for (i = 0; i < ARRAY_SIZE(shca->sport); i++)
+		spin_lock_init(&shca->sport[i].mod_sqp_lock);
 
 	shca->ofdev = dev;
 	shca->ipz_hca_handle.handle = *handle;
@@ -934,7 +937,7 @@ void ehca_poll_eqs(unsigned long data)
 				ehca_process_eq(shca, 0);
 		}
 	}
-	mod_timer(&poll_eqs_timer, jiffies + HZ);
+	mod_timer(&poll_eqs_timer, round_jiffies(jiffies + HZ));
 	spin_unlock(&shca_list_lock);
 }
 

drivers/infiniband/hw/ehca/ehca_reqs.c

@@ -50,6 +50,9 @@
 #include "hcp_if.h"
 #include "hipz_fns.h"
 
+/* in RC traffic, insert an empty RDMA READ every this many packets */
+#define ACK_CIRC_THRESHOLD 2000000
+
 static inline int ehca_write_rwqe(struct ipz_queue *ipz_rqueue,
 				  struct ehca_wqe *wqe_p,
 				  struct ib_recv_wr *recv_wr)
@@ -81,7 +84,7 @@ static inline int ehca_write_rwqe(struct ipz_queue *ipz_rqueue,
 	if (ehca_debug_level) {
 		ehca_gen_dbg("RECEIVE WQE written into ipz_rqueue=%p",
 			     ipz_rqueue);
-		ehca_dmp( wqe_p, 16*(6 + wqe_p->nr_of_data_seg), "recv wqe");
+		ehca_dmp(wqe_p, 16*(6 + wqe_p->nr_of_data_seg), "recv wqe");
 	}
 
 	return 0;
@@ -135,7 +138,8 @@ static void trace_send_wr_ud(const struct ib_send_wr *send_wr)
 
 static inline int ehca_write_swqe(struct ehca_qp *qp,
 				  struct ehca_wqe *wqe_p,
-				  const struct ib_send_wr *send_wr)
+				  const struct ib_send_wr *send_wr,
+				  int hidden)
 {
 	u32 idx;
 	u64 dma_length;
@@ -176,7 +180,9 @@ static inline int ehca_write_swqe(struct ehca_qp *qp,
 
 	wqe_p->wr_flag = 0;
 
-	if (send_wr->send_flags & IB_SEND_SIGNALED)
+	if ((send_wr->send_flags & IB_SEND_SIGNALED ||
+	    qp->init_attr.sq_sig_type == IB_SIGNAL_ALL_WR)
+	    && !hidden)
 		wqe_p->wr_flag |= WQE_WRFLAG_REQ_SIGNAL_COM;
 
 	if (send_wr->opcode == IB_WR_SEND_WITH_IMM ||
@@ -199,7 +205,7 @@ static inline int ehca_write_swqe(struct ehca_qp *qp,
 
 		wqe_p->destination_qp_number = send_wr->wr.ud.remote_qpn << 8;
 		wqe_p->local_ee_context_qkey = remote_qkey;
-		if (!send_wr->wr.ud.ah) {
+		if (unlikely(!send_wr->wr.ud.ah)) {
 			ehca_gen_err("wr.ud.ah is NULL. qp=%p", qp);
 			return -EINVAL;
 		}
@@ -255,6 +261,15 @@ static inline int ehca_write_swqe(struct ehca_qp *qp,
 		} /* eof idx */
 		wqe_p->u.nud.atomic_1st_op_dma_len = dma_length;
 
+		/* unsolicited ack circumvention */
+		if (send_wr->opcode == IB_WR_RDMA_READ) {
+			/* on RDMA read, switch on and reset counters */
+			qp->message_count = qp->packet_count = 0;
+			qp->unsol_ack_circ = 1;
+		} else
+			/* else estimate #packets */
+			qp->packet_count += (dma_length >> qp->mtu_shift) + 1;
+
 		break;
 
 	default:
@@ -355,13 +370,49 @@ static inline void map_ib_wc_status(u32 cqe_status,
 		*wc_status = IB_WC_SUCCESS;
 }
 
+static inline int post_one_send(struct ehca_qp *my_qp,
+			 struct ib_send_wr *cur_send_wr,
+			 struct ib_send_wr **bad_send_wr,
+			 int hidden)
+{
+	struct ehca_wqe *wqe_p;
+	int ret;
+	u64 start_offset = my_qp->ipz_squeue.current_q_offset;
+
+	/* get pointer next to free WQE */
+	wqe_p = ipz_qeit_get_inc(&my_qp->ipz_squeue);
+	if (unlikely(!wqe_p)) {
+		/* too many posted work requests: queue overflow */
+		if (bad_send_wr)
+			*bad_send_wr = cur_send_wr;
+		ehca_err(my_qp->ib_qp.device, "Too many posted WQEs "
+			 "qp_num=%x", my_qp->ib_qp.qp_num);
+		return -ENOMEM;
+	}
+	/* write a SEND WQE into the QUEUE */
+	ret = ehca_write_swqe(my_qp, wqe_p, cur_send_wr, hidden);
+	/*
+	 * if something failed,
+	 * reset the free entry pointer to the start value
+	 */
+	if (unlikely(ret)) {
+		my_qp->ipz_squeue.current_q_offset = start_offset;
+		if (bad_send_wr)
+			*bad_send_wr = cur_send_wr;
+		ehca_err(my_qp->ib_qp.device, "Could not write WQE "
+			 "qp_num=%x", my_qp->ib_qp.qp_num);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 int ehca_post_send(struct ib_qp *qp,
 		   struct ib_send_wr *send_wr,
 		   struct ib_send_wr **bad_send_wr)
 {
 	struct ehca_qp *my_qp = container_of(qp, struct ehca_qp, ib_qp);
 	struct ib_send_wr *cur_send_wr;
-	struct ehca_wqe *wqe_p;
 	int wqe_cnt = 0;
 	int ret = 0;
 	unsigned long flags;
@@ -369,37 +420,33 @@ int ehca_post_send(struct ib_qp *qp,
 	/* LOCK the QUEUE */
 	spin_lock_irqsave(&my_qp->spinlock_s, flags);
 
+	/* Send an empty extra RDMA read if:
+	 *  1) there has been an RDMA read on this connection before
+	 *  2) no RDMA read occurred for ACK_CIRC_THRESHOLD link packets
+	 *  3) we can be sure that any previous extra RDMA read has been
+	 *     processed so we don't overflow the SQ
+	 */
+	if (unlikely(my_qp->unsol_ack_circ &&
+		     my_qp->packet_count > ACK_CIRC_THRESHOLD &&
+		     my_qp->message_count > my_qp->init_attr.cap.max_send_wr)) {
+		/* insert an empty RDMA READ to fix up the remote QP state */
+		struct ib_send_wr circ_wr;
+		memset(&circ_wr, 0, sizeof(circ_wr));
+		circ_wr.opcode = IB_WR_RDMA_READ;
+		post_one_send(my_qp, &circ_wr, NULL, 1); /* ignore retcode */
+		wqe_cnt++;
+		ehca_dbg(qp->device, "posted circ wr  qp_num=%x", qp->qp_num);
+		my_qp->message_count = my_qp->packet_count = 0;
+	}
+
 	/* loop processes list of send reqs */
 	for (cur_send_wr = send_wr; cur_send_wr != NULL;
 	     cur_send_wr = cur_send_wr->next) {
-		u64 start_offset = my_qp->ipz_squeue.current_q_offset;
-		/* get pointer next to free WQE */
-		wqe_p = ipz_qeit_get_inc(&my_qp->ipz_squeue);
-		if (unlikely(!wqe_p)) {
-			/* too many posted work requests: queue overflow */
-			if (bad_send_wr)
-				*bad_send_wr = cur_send_wr;
-			if (wqe_cnt == 0) {
-				ret = -ENOMEM;
-				ehca_err(qp->device, "Too many posted WQEs "
-					 "qp_num=%x", qp->qp_num);
-			}
-			goto post_send_exit0;
-		}
-		/* write a SEND WQE into the QUEUE */
-		ret = ehca_write_swqe(my_qp, wqe_p, cur_send_wr);
-		/*
-		 * if something failed,
-		 * reset the free entry pointer to the start value
-		 */
+		ret = post_one_send(my_qp, cur_send_wr, bad_send_wr, 0);
 		if (unlikely(ret)) {
-			my_qp->ipz_squeue.current_q_offset = start_offset;
-			*bad_send_wr = cur_send_wr;
-			if (wqe_cnt == 0) {
-				ret = -EINVAL;
-				ehca_err(qp->device, "Could not write WQE "
-					 "qp_num=%x", qp->qp_num);
-			}
+			/* if one or more WQEs were successful, don't fail */
+			if (wqe_cnt)
+				ret = 0;
 			goto post_send_exit0;
 		}
 		wqe_cnt++;
@@ -410,6 +457,7 @@ int ehca_post_send(struct ib_qp *qp,
 post_send_exit0:
 	iosync(); /* serialize GAL register access */
 	hipz_update_sqa(my_qp, wqe_cnt);
+	my_qp->message_count += wqe_cnt;
 	spin_unlock_irqrestore(&my_qp->spinlock_s, flags);
 	return ret;
 }

drivers/infiniband/hw/ehca/ehca_sqp.c

@@ -40,11 +40,8 @@
  */
 
 
-#include <linux/module.h>
-#include <linux/err.h>
 #include "ehca_classes.h"
 #include "ehca_tools.h"
-#include "ehca_qes.h"
 #include "ehca_iverbs.h"
 #include "hcp_if.h"
 
@@ -93,6 +90,9 @@ u64 ehca_define_sqp(struct ehca_shca *shca,
 		return H_PARAMETER;
 	}
 
+	if (ehca_nr_ports < 0) /* autodetect mode */
+		return H_SUCCESS;
+
 	for (counter = 0;
 	     shca->sport[port - 1].port_state != IB_PORT_ACTIVE &&
 		     counter < ehca_port_act_time;

drivers/infiniband/hw/ipath/ipath_common.h

@@ -81,6 +81,16 @@
 #define IPATH_IB_LINK_LOOPBACK	6 /* enable local loopback */
 #define IPATH_IB_LINK_EXTERNAL	7 /* normal, disable local loopback */
 
+/*
+ * These 3 values (SDR and DDR may be ORed for auto-speed
+ * negotiation) are used for the 3rd argument to path_f_set_ib_cfg
+ * with cmd IPATH_IB_CFG_SPD_ENB, by direct calls or via sysfs.  They
+ * are also the the possible values for ipath_link_speed_enabled and active
+ * The values were chosen to match values used within the IB spec.
+ */
+#define IPATH_IB_SDR 1
+#define IPATH_IB_DDR 2
+
 /*
  * stats maintained by the driver.  For now, at least, this is global
  * to all minor devices.
@@ -433,8 +443,9 @@ struct ipath_user_info {
 #define IPATH_CMD_UNUSED_2	26
 #define IPATH_CMD_PIOAVAILUPD	27	/* force an update of PIOAvail reg */
 #define IPATH_CMD_POLL_TYPE	28	/* set the kind of polling we want */
+#define IPATH_CMD_ARMLAUNCH_CTRL	29 /* armlaunch detection control */
 
-#define IPATH_CMD_MAX		28
+#define IPATH_CMD_MAX		29
 
 /*
  * Poll types
@@ -477,6 +488,8 @@ struct ipath_cmd {
 		__u64 port_info;
 		/* enable/disable receipt of packets */
 		__u32 recv_ctrl;
+		/* enable/disable armlaunch errors (non-zero to enable) */
+		__u32 armlaunch_ctrl;
 		/* partition key to set */
 		__u16 part_key;
 		/* user address of __u32 bitmask of active slaves */
@@ -579,7 +592,7 @@ struct ipath_flash {
 struct infinipath_counters {
 	__u64 LBIntCnt;
 	__u64 LBFlowStallCnt;
-	__u64 Reserved1;
+	__u64 TxSDmaDescCnt;	/* was Reserved1 */
 	__u64 TxUnsupVLErrCnt;
 	__u64 TxDataPktCnt;
 	__u64 TxFlowPktCnt;
@@ -615,12 +628,26 @@ struct infinipath_counters {
 	__u64 RxP6HdrEgrOvflCnt;
 	__u64 RxP7HdrEgrOvflCnt;
 	__u64 RxP8HdrEgrOvflCnt;
-	__u64 Reserved6;
-	__u64 Reserved7;
+	__u64 RxP9HdrEgrOvflCnt;	/* was Reserved6 */
+	__u64 RxP10HdrEgrOvflCnt;	/* was Reserved7 */
+	__u64 RxP11HdrEgrOvflCnt;	/* new for IBA7220 */
+	__u64 RxP12HdrEgrOvflCnt;	/* new for IBA7220 */
+	__u64 RxP13HdrEgrOvflCnt;	/* new for IBA7220 */
+	__u64 RxP14HdrEgrOvflCnt;	/* new for IBA7220 */
+	__u64 RxP15HdrEgrOvflCnt;	/* new for IBA7220 */
+	__u64 RxP16HdrEgrOvflCnt;	/* new for IBA7220 */
 	__u64 IBStatusChangeCnt;
 	__u64 IBLinkErrRecoveryCnt;
 	__u64 IBLinkDownedCnt;
 	__u64 IBSymbolErrCnt;
+	/* The following are new for IBA7220 */
+	__u64 RxVL15DroppedPktCnt;
+	__u64 RxOtherLocalPhyErrCnt;
+	__u64 PcieRetryBufDiagQwordCnt;
+	__u64 ExcessBufferOvflCnt;
+	__u64 LocalLinkIntegrityErrCnt;
+	__u64 RxVlErrCnt;
+	__u64 RxDlidFltrCnt;
 };
 
 /*

drivers/infiniband/hw/ipath/ipath_cq.c

@@ -421,7 +421,7 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
 	else
 		n = head - tail;
 	if (unlikely((u32)cqe < n)) {
-		ret = -EOVERFLOW;
+		ret = -EINVAL;
 		goto bail_unlock;
 	}
 	for (n = 0; tail != head; n++) {

drivers/infiniband/hw/ipath/ipath_debug.h

@@ -55,7 +55,7 @@
 #define __IPATH_PKTDBG      0x80	/* print packet data */
 /* print process startup (init)/exit messages */
 #define __IPATH_PROCDBG     0x100
-/* print mmap/nopage stuff, not using VDBG any more */
+/* print mmap/fault stuff, not using VDBG any more */
 #define __IPATH_MMDBG       0x200
 #define __IPATH_ERRPKTDBG   0x400
 #define __IPATH_USER_SEND   0x1000	/* use user mode send */
@@ -81,7 +81,7 @@
 #define __IPATH_VERBDBG   0x0	/* very verbose debug */
 #define __IPATH_PKTDBG    0x0	/* print packet data */
 #define __IPATH_PROCDBG   0x0	/* process startup (init)/exit messages */
-/* print mmap/nopage stuff, not using VDBG any more */
+/* print mmap/fault stuff, not using VDBG any more */
 #define __IPATH_MMDBG     0x0
 #define __IPATH_EPKTDBG   0x0	/* print ethernet packet data */
 #define __IPATH_IPATHDBG  0x0	/* Ethernet (IPATH) table dump on */

drivers/infiniband/hw/ipath/ipath_eeprom.c

@@ -510,10 +510,10 @@ int ipath_eeprom_read(struct ipath_devdata *dd, u8 eeprom_offset,
 {
 	int ret;
 
-	ret = down_interruptible(&dd->ipath_eep_sem);
+	ret = mutex_lock_interruptible(&dd->ipath_eep_lock);
 	if (!ret) {
 		ret = ipath_eeprom_internal_read(dd, eeprom_offset, buff, len);
-		up(&dd->ipath_eep_sem);
+		mutex_unlock(&dd->ipath_eep_lock);
 	}
 
 	return ret;
@@ -524,10 +524,10 @@ int ipath_eeprom_write(struct ipath_devdata *dd, u8 eeprom_offset,
 {
 	int ret;
 
-	ret = down_interruptible(&dd->ipath_eep_sem);
+	ret = mutex_lock_interruptible(&dd->ipath_eep_lock);
 	if (!ret) {
 		ret = ipath_eeprom_internal_write(dd, eeprom_offset, buff, len);
-		up(&dd->ipath_eep_sem);
+		mutex_unlock(&dd->ipath_eep_lock);
 	}
 
 	return ret;
@@ -574,7 +574,7 @@ void ipath_get_eeprom_info(struct ipath_devdata *dd)
 	struct ipath_devdata *dd0 = ipath_lookup(0);
 
 	if (t && dd0->ipath_nguid > 1 && t <= dd0->ipath_nguid) {
-		u8 *bguid, oguid;
+		u8 oguid;
 		dd->ipath_guid = dd0->ipath_guid;
 		bguid = (u8 *) & dd->ipath_guid;
 
@@ -616,9 +616,9 @@ void ipath_get_eeprom_info(struct ipath_devdata *dd)
 		goto bail;
 	}
 
-	down(&dd->ipath_eep_sem);
+	mutex_lock(&dd->ipath_eep_lock);
 	eep_stat = ipath_eeprom_internal_read(dd, 0, buf, len);
-	up(&dd->ipath_eep_sem);
+	mutex_unlock(&dd->ipath_eep_lock);
 
 	if (eep_stat) {
 		ipath_dev_err(dd, "Failed reading GUID from eeprom\n");
@@ -674,7 +674,6 @@ void ipath_get_eeprom_info(struct ipath_devdata *dd)
 		 * elsewhere for backward-compatibility.
 		 */
 		char *snp = dd->ipath_serial;
-		int len;
 		memcpy(snp, ifp->if_sprefix, sizeof ifp->if_sprefix);
 		snp[sizeof ifp->if_sprefix] = '\0';
 		len = strlen(snp);
@@ -764,14 +763,14 @@ int ipath_update_eeprom_log(struct ipath_devdata *dd)
 	/* Grab semaphore and read current EEPROM. If we get an
 	 * error, let go, but if not, keep it until we finish write.
 	 */
-	ret = down_interruptible(&dd->ipath_eep_sem);
+	ret = mutex_lock_interruptible(&dd->ipath_eep_lock);
 	if (ret) {
 		ipath_dev_err(dd, "Unable to acquire EEPROM for logging\n");
 		goto free_bail;
 	}
 	ret = ipath_eeprom_internal_read(dd, 0, buf, len);
 	if (ret) {
-		up(&dd->ipath_eep_sem);
+		mutex_unlock(&dd->ipath_eep_lock);
 		ipath_dev_err(dd, "Unable read EEPROM for logging\n");
 		goto free_bail;
 	}
@@ -779,7 +778,7 @@ int ipath_update_eeprom_log(struct ipath_devdata *dd)
 
 	csum = flash_csum(ifp, 0);
 	if (csum != ifp->if_csum) {
-		up(&dd->ipath_eep_sem);
+		mutex_unlock(&dd->ipath_eep_lock);
 		ipath_dev_err(dd, "EEPROM cks err (0x%02X, S/B 0x%02X)\n",
 				csum, ifp->if_csum);
 		ret = 1;
@@ -849,7 +848,7 @@ int ipath_update_eeprom_log(struct ipath_devdata *dd)
 		csum = flash_csum(ifp, 1);
 		ret = ipath_eeprom_internal_write(dd, 0, buf, hi_water + 1);
 	}
-	up(&dd->ipath_eep_sem);
+	mutex_unlock(&dd->ipath_eep_lock);
 	if (ret)
 		ipath_dev_err(dd, "Failed updating EEPROM\n");
 

drivers/infiniband/hw/ipath/ipath_file_ops.c

@@ -169,7 +169,7 @@ static int ipath_get_base_info(struct file *fp,
 		kinfo->spi_piocnt = dd->ipath_pbufsport;
 		kinfo->spi_piobufbase = (u64) pd->port_piobufs;
 		kinfo->__spi_uregbase = (u64) dd->ipath_uregbase +
-			dd->ipath_palign * pd->port_port;
+			dd->ipath_ureg_align * pd->port_port;
 	} else if (master) {
 		kinfo->spi_piocnt = (dd->ipath_pbufsport / subport_cnt) +
 				    (dd->ipath_pbufsport % subport_cnt);
@@ -186,7 +186,7 @@ static int ipath_get_base_info(struct file *fp,
 	}
 	if (shared) {
 		kinfo->spi_port_uregbase = (u64) dd->ipath_uregbase +
-			dd->ipath_palign * pd->port_port;
+			dd->ipath_ureg_align * pd->port_port;
 		kinfo->spi_port_rcvegrbuf = kinfo->spi_rcv_egrbufs;
 		kinfo->spi_port_rcvhdr_base = kinfo->spi_rcvhdr_base;
 		kinfo->spi_port_rcvhdr_tailaddr = kinfo->spi_rcvhdr_tailaddr;
@@ -742,11 +742,12 @@ static int ipath_manage_rcvq(struct ipath_portdata *pd, unsigned subport,
 		 * updated and correct itself, even in the face of software
 		 * bugs.
 		 */
-		*(volatile u64 *)pd->port_rcvhdrtail_kvaddr = 0;
-		set_bit(INFINIPATH_R_PORTENABLE_SHIFT + pd->port_port,
+		if (pd->port_rcvhdrtail_kvaddr)
+			ipath_clear_rcvhdrtail(pd);
+		set_bit(dd->ipath_r_portenable_shift + pd->port_port,
 			&dd->ipath_rcvctrl);
 	} else
-		clear_bit(INFINIPATH_R_PORTENABLE_SHIFT + pd->port_port,
+		clear_bit(dd->ipath_r_portenable_shift + pd->port_port,
 			  &dd->ipath_rcvctrl);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
 			 dd->ipath_rcvctrl);
@@ -881,7 +882,7 @@ static int ipath_create_user_egr(struct ipath_portdata *pd)
 
 	egrcnt = dd->ipath_rcvegrcnt;
 	/* TID number offset for this port */
-	egroff = pd->port_port * egrcnt;
+	egroff = (pd->port_port - 1) * egrcnt + dd->ipath_p0_rcvegrcnt;
 	egrsize = dd->ipath_rcvegrbufsize;
 	ipath_cdbg(VERBOSE, "Allocating %d egr buffers, at egrtid "
 		   "offset %x, egrsize %u\n", egrcnt, egroff, egrsize);
@@ -1049,11 +1050,6 @@ static int mmap_piobufs(struct vm_area_struct *vma,
 
 	phys = dd->ipath_physaddr + piobufs;
 
-	/*
-	 * Don't mark this as non-cached, or we don't get the
-	 * write combining behavior we want on the PIO buffers!
-	 */
-
 #if defined(__powerpc__)
 	/* There isn't a generic way to specify writethrough mappings */
 	pgprot_val(vma->vm_page_prot) |= _PAGE_NO_CACHE;
@@ -1120,33 +1116,24 @@ static int mmap_rcvegrbufs(struct vm_area_struct *vma,
 }
 
 /*
- * ipath_file_vma_nopage - handle a VMA page fault.
+ * ipath_file_vma_fault - handle a VMA page fault.
  */
-static struct page *ipath_file_vma_nopage(struct vm_area_struct *vma,
-					  unsigned long address, int *type)
+static int ipath_file_vma_fault(struct vm_area_struct *vma,
+					struct vm_fault *vmf)
 {
-	unsigned long offset = address - vma->vm_start;
-	struct page *page = NOPAGE_SIGBUS;
-	void *pageptr;
+	struct page *page;
 
-	/*
-	 * Convert the vmalloc address into a struct page.
-	 */
-	pageptr = (void *)(offset + (vma->vm_pgoff << PAGE_SHIFT));
-	page = vmalloc_to_page(pageptr);
+	page = vmalloc_to_page((void *)(vmf->pgoff << PAGE_SHIFT));
 	if (!page)
-		goto out;
-
-	/* Increment the reference count. */
+		return VM_FAULT_SIGBUS;
 	get_page(page);
-	if (type)
-		*type = VM_FAULT_MINOR;
-out:
-	return page;
+	vmf->page = page;
+
+	return 0;
 }
 
 static struct vm_operations_struct ipath_file_vm_ops = {
-	.nopage = ipath_file_vma_nopage,
+	.fault = ipath_file_vma_fault,
 };
 
 static int mmap_kvaddr(struct vm_area_struct *vma, u64 pgaddr,
@@ -1284,7 +1271,7 @@ static int ipath_mmap(struct file *fp, struct vm_area_struct *vma)
 		goto bail;
 	}
 
-	ureg = dd->ipath_uregbase + dd->ipath_palign * pd->port_port;
+	ureg = dd->ipath_uregbase + dd->ipath_ureg_align * pd->port_port;
 	if (!pd->port_subport_cnt) {
 		/* port is not shared */
 		piocnt = dd->ipath_pbufsport;
@@ -1400,7 +1387,10 @@ static unsigned int ipath_poll_next(struct ipath_portdata *pd,
 	pollflag = ipath_poll_hdrqfull(pd);
 
 	head = ipath_read_ureg32(dd, ur_rcvhdrhead, pd->port_port);
-	tail = *(volatile u64 *)pd->port_rcvhdrtail_kvaddr;
+	if (pd->port_rcvhdrtail_kvaddr)
+		tail = ipath_get_rcvhdrtail(pd);
+	else
+		tail = ipath_read_ureg32(dd, ur_rcvhdrtail, pd->port_port);
 
 	if (head != tail)
 		pollflag |= POLLIN | POLLRDNORM;
@@ -1410,7 +1400,7 @@ static unsigned int ipath_poll_next(struct ipath_portdata *pd,
 		/* flush waiting flag so we don't miss an event */
 		wmb();
 
-		set_bit(pd->port_port + INFINIPATH_R_INTRAVAIL_SHIFT,
+		set_bit(pd->port_port + dd->ipath_r_intravail_shift,
 			&dd->ipath_rcvctrl);
 
 		ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
@@ -1790,6 +1780,7 @@ static int find_shared_port(struct file *fp,
 			}
 			port_fp(fp) = pd;
 			subport_fp(fp) = pd->port_cnt++;
+			pd->port_subpid[subport_fp(fp)] = current->pid;
 			tidcursor_fp(fp) = 0;
 			pd->active_slaves |= 1 << subport_fp(fp);
 			ipath_cdbg(PROC,
@@ -1920,8 +1911,7 @@ static int ipath_do_user_init(struct file *fp,
 	 */
 	head32 = ipath_read_ureg32(dd, ur_rcvegrindextail, pd->port_port);
 	ipath_write_ureg(dd, ur_rcvegrindexhead, head32, pd->port_port);
-	dd->ipath_lastegrheads[pd->port_port] = -1;
-	dd->ipath_lastrcvhdrqtails[pd->port_port] = -1;
+	pd->port_lastrcvhdrqtail = -1;
 	ipath_cdbg(VERBOSE, "Wrote port%d egrhead %x from tail regs\n",
 		pd->port_port, head32);
 	pd->port_tidcursor = 0;	/* start at beginning after open */
@@ -1941,11 +1931,13 @@ static int ipath_do_user_init(struct file *fp,
 	 * We explictly set the in-memory copy to 0 beforehand, so we don't
 	 * have to wait to be sure the DMA update has happened.
 	 */
-	*(volatile u64 *)pd->port_rcvhdrtail_kvaddr = 0ULL;
-	set_bit(INFINIPATH_R_PORTENABLE_SHIFT + pd->port_port,
+	if (pd->port_rcvhdrtail_kvaddr)
+		ipath_clear_rcvhdrtail(pd);
+	set_bit(dd->ipath_r_portenable_shift + pd->port_port,
 		&dd->ipath_rcvctrl);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
-			 dd->ipath_rcvctrl & ~INFINIPATH_R_TAILUPD);
+			dd->ipath_rcvctrl &
+			~(1ULL << dd->ipath_r_tailupd_shift));
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
 			 dd->ipath_rcvctrl);
 	/* Notify any waiting slaves */
@@ -2022,6 +2014,7 @@ static int ipath_close(struct inode *in, struct file *fp)
 		 * the slave(s) don't wait for receive data forever.
 		 */
 		pd->active_slaves &= ~(1 << fd->subport);
+		pd->port_subpid[fd->subport] = 0;
 		mutex_unlock(&ipath_mutex);
 		goto bail;
 	}
@@ -2054,9 +2047,9 @@ static int ipath_close(struct inode *in, struct file *fp)
 	if (dd->ipath_kregbase) {
 		int i;
 		/* atomically clear receive enable port and intr avail. */
-		clear_bit(INFINIPATH_R_PORTENABLE_SHIFT + port,
+		clear_bit(dd->ipath_r_portenable_shift + port,
 			  &dd->ipath_rcvctrl);
-		clear_bit(pd->port_port + INFINIPATH_R_INTRAVAIL_SHIFT,
+		clear_bit(pd->port_port + dd->ipath_r_intravail_shift,
 			  &dd->ipath_rcvctrl);
 		ipath_write_kreg( dd, dd->ipath_kregs->kr_rcvctrl,
 			dd->ipath_rcvctrl);
@@ -2149,11 +2142,15 @@ static int ipath_get_slave_info(struct ipath_portdata *pd,
 
 static int ipath_force_pio_avail_update(struct ipath_devdata *dd)
 {
-	u64 reg = dd->ipath_sendctrl;
+	unsigned long flags;
 
-	clear_bit(IPATH_S_PIOBUFAVAILUPD, &reg);
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, reg);
+	spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
+		dd->ipath_sendctrl & ~INFINIPATH_S_PIOBUFAVAILUPD);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, dd->ipath_sendctrl);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+	spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 
 	return 0;
 }
@@ -2227,6 +2224,11 @@ static ssize_t ipath_write(struct file *fp, const char __user *data,
 		dest = &cmd.cmd.poll_type;
 		src = &ucmd->cmd.poll_type;
 		break;
+	case IPATH_CMD_ARMLAUNCH_CTRL:
+		copy = sizeof(cmd.cmd.armlaunch_ctrl);
+		dest = &cmd.cmd.armlaunch_ctrl;
+		src = &ucmd->cmd.armlaunch_ctrl;
+		break;
 	default:
 		ret = -EINVAL;
 		goto bail;
@@ -2302,6 +2304,12 @@ static ssize_t ipath_write(struct file *fp, const char __user *data,
 	case IPATH_CMD_POLL_TYPE:
 		pd->poll_type = cmd.cmd.poll_type;
 		break;
+	case IPATH_CMD_ARMLAUNCH_CTRL:
+		if (cmd.cmd.armlaunch_ctrl)
+			ipath_enable_armlaunch(pd->port_dd);
+		else
+			ipath_disable_armlaunch(pd->port_dd);
+		break;
 	}
 
 	if (ret >= 0)

drivers/infiniband/hw/ipath/ipath_fs.c

@@ -108,21 +108,16 @@ static const struct file_operations atomic_stats_ops = {
 	.read = atomic_stats_read,
 };
 
-#define NUM_COUNTERS sizeof(struct infinipath_counters) / sizeof(u64)
-
 static ssize_t atomic_counters_read(struct file *file, char __user *buf,
 				    size_t count, loff_t *ppos)
 {
-	u64 counters[NUM_COUNTERS];
-	u16 i;
+	struct infinipath_counters counters;
 	struct ipath_devdata *dd;
 
 	dd = file->f_path.dentry->d_inode->i_private;
+	dd->ipath_f_read_counters(dd, &counters);
 
-	for (i = 0; i < NUM_COUNTERS; i++)
-		counters[i] = ipath_snap_cntr(dd, i);
-
-	return simple_read_from_buffer(buf, count, ppos, counters,
+	return simple_read_from_buffer(buf, count, ppos, &counters,
 				       sizeof counters);
 }
 
@@ -243,8 +238,7 @@ static int create_device_files(struct super_block *sb,
 
 	snprintf(unit, sizeof unit, "%02d", dd->ipath_unit);
 	ret = create_file(unit, S_IFDIR|S_IRUGO|S_IXUGO, sb->s_root, &dir,
-			  (struct file_operations *) &simple_dir_operations,
-			  dd);
+			  &simple_dir_operations, dd);
 	if (ret) {
 		printk(KERN_ERR "create_file(%s) failed: %d\n", unit, ret);
 		goto bail;

drivers/infiniband/hw/ipath/ipath_init_chip.c

@@ -91,7 +91,7 @@ static int create_port0_egr(struct ipath_devdata *dd)
 	struct ipath_skbinfo *skbinfo;
 	int ret;
 
-	egrcnt = dd->ipath_rcvegrcnt;
+	egrcnt = dd->ipath_p0_rcvegrcnt;
 
 	skbinfo = vmalloc(sizeof(*dd->ipath_port0_skbinfo) * egrcnt);
 	if (skbinfo == NULL) {
@@ -244,8 +244,7 @@ static int init_chip_first(struct ipath_devdata *dd,
 	 * cfgports.  We do still check and report a difference, if
 	 * not same (should be impossible).
 	 */
-	dd->ipath_portcnt =
-		ipath_read_kreg32(dd, dd->ipath_kregs->kr_portcnt);
+	dd->ipath_f_config_ports(dd, ipath_cfgports);
 	if (!ipath_cfgports)
 		dd->ipath_cfgports = dd->ipath_portcnt;
 	else if (ipath_cfgports <= dd->ipath_portcnt) {
@@ -272,22 +271,7 @@ static int init_chip_first(struct ipath_devdata *dd,
 		goto done;
 	}
 
-	dd->ipath_lastegrheads = kzalloc(sizeof(*dd->ipath_lastegrheads)
-					 * dd->ipath_cfgports,
-					 GFP_KERNEL);
-	dd->ipath_lastrcvhdrqtails =
-		kzalloc(sizeof(*dd->ipath_lastrcvhdrqtails)
-			* dd->ipath_cfgports, GFP_KERNEL);
-
-	if (!dd->ipath_lastegrheads || !dd->ipath_lastrcvhdrqtails) {
-		ipath_dev_err(dd, "Unable to allocate head arrays, "
-			      "failing\n");
-		ret = -ENOMEM;
-		goto done;
-	}
-
 	pd = create_portdata0(dd);
-
 	if (!pd) {
 		ipath_dev_err(dd, "Unable to allocate portdata for port "
 			      "0, failing\n");
@@ -345,10 +329,10 @@ static int init_chip_first(struct ipath_devdata *dd,
 		       dd->ipath_piobcnt2k, dd->ipath_pio2kbase);
 
 	spin_lock_init(&dd->ipath_tid_lock);
-
+	spin_lock_init(&dd->ipath_sendctrl_lock);
 	spin_lock_init(&dd->ipath_gpio_lock);
 	spin_lock_init(&dd->ipath_eep_st_lock);
-	sema_init(&dd->ipath_eep_sem, 1);
+	mutex_init(&dd->ipath_eep_lock);
 
 done:
 	*pdp = pd;
@@ -372,9 +356,9 @@ static int init_chip_reset(struct ipath_devdata *dd,
 	*pdp = dd->ipath_pd[0];
 	/* ensure chip does no sends or receives while we re-initialize */
 	dd->ipath_control = dd->ipath_sendctrl = dd->ipath_rcvctrl = 0U;
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl, 0);
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, 0);
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_control, 0);
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl, dd->ipath_rcvctrl);
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, dd->ipath_sendctrl);
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_control, dd->ipath_control);
 
 	rtmp = ipath_read_kreg32(dd, dd->ipath_kregs->kr_portcnt);
 	if (dd->ipath_portcnt != rtmp)
@@ -487,6 +471,7 @@ static void enable_chip(struct ipath_devdata *dd,
 			struct ipath_portdata *pd, int reinit)
 {
 	u32 val;
+	unsigned long flags;
 	int i;
 
 	if (!reinit)
@@ -495,19 +480,21 @@ static void enable_chip(struct ipath_devdata *dd,
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
 			 dd->ipath_rcvctrl);
 
+	spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
 	/* Enable PIO send, and update of PIOavail regs to memory. */
 	dd->ipath_sendctrl = INFINIPATH_S_PIOENABLE |
 		INFINIPATH_S_PIOBUFAVAILUPD;
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
-			 dd->ipath_sendctrl);
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, dd->ipath_sendctrl);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+	spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 
 	/*
 	 * enable port 0 receive, and receive interrupt.  other ports
 	 * done as user opens and inits them.
 	 */
-	dd->ipath_rcvctrl = INFINIPATH_R_TAILUPD |
-		(1ULL << INFINIPATH_R_PORTENABLE_SHIFT) |
-		(1ULL << INFINIPATH_R_INTRAVAIL_SHIFT);
+	dd->ipath_rcvctrl = (1ULL << dd->ipath_r_tailupd_shift) |
+		(1ULL << dd->ipath_r_portenable_shift) |
+		(1ULL << dd->ipath_r_intravail_shift);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
 			 dd->ipath_rcvctrl);
 
@@ -523,12 +510,11 @@ static void enable_chip(struct ipath_devdata *dd,
 	 */
 	val = ipath_read_ureg32(dd, ur_rcvegrindextail, 0);
 	(void)ipath_write_ureg(dd, ur_rcvegrindexhead, val, 0);
-	dd->ipath_port0head = ipath_read_ureg32(dd, ur_rcvhdrtail, 0);
 
 	/* Initialize so we interrupt on next packet received */
 	(void)ipath_write_ureg(dd, ur_rcvhdrhead,
 			       dd->ipath_rhdrhead_intr_off |
-			       dd->ipath_port0head, 0);
+			       dd->ipath_pd[0]->port_head, 0);
 
 	/*
 	 * by now pioavail updates to memory should have occurred, so
@@ -542,12 +528,8 @@ static void enable_chip(struct ipath_devdata *dd,
 		/*
 		 * Chip Errata bug 6641; even and odd qwords>3 are swapped.
 		 */
-		if (i > 3) {
-			if (i & 1)
-				val = dd->ipath_pioavailregs_dma[i - 1];
-			else
-				val = dd->ipath_pioavailregs_dma[i + 1];
-		}
+		if (i > 3 && (dd->ipath_flags & IPATH_SWAP_PIOBUFS))
+			val = dd->ipath_pioavailregs_dma[i ^ 1];
 		else
 			val = dd->ipath_pioavailregs_dma[i];
 		dd->ipath_pioavailshadow[i] = le64_to_cpu(val);
@@ -690,12 +672,13 @@ static int init_housekeeping(struct ipath_devdata *dd,
  */
 int ipath_init_chip(struct ipath_devdata *dd, int reinit)
 {
-	int ret = 0, i;
+	int ret = 0;
 	u32 val32, kpiobufs;
 	u32 piobufs, uports;
 	u64 val;
 	struct ipath_portdata *pd = NULL; /* keep gcc4 happy */
 	gfp_t gfp_flags = GFP_USER | __GFP_COMP;
+	unsigned long flags;
 
 	ret = init_housekeeping(dd, &pd, reinit);
 	if (ret)
@@ -746,7 +729,7 @@ int ipath_init_chip(struct ipath_devdata *dd, int reinit)
 		kpiobufs = ipath_kpiobufs;
 
 	if (kpiobufs + (uports * IPATH_MIN_USER_PORT_BUFCNT) > piobufs) {
-		i = (int) piobufs -
+		int i = (int) piobufs -
 			(int) (uports * IPATH_MIN_USER_PORT_BUFCNT);
 		if (i < 0)
 			i = 0;
@@ -827,8 +810,12 @@ int ipath_init_chip(struct ipath_devdata *dd, int reinit)
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_hwerrclear,
 			 ~0ULL&~INFINIPATH_HWE_MEMBISTFAILED);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_control, 0ULL);
-	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
-			 INFINIPATH_S_PIOENABLE);
+
+	spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
+	dd->ipath_sendctrl = INFINIPATH_S_PIOENABLE;
+	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl, dd->ipath_sendctrl);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+	spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 
 	/*
 	 * before error clears, since we expect serdes pll errors during

drivers/infiniband/hw/ipath/ipath_intr.c

@@ -683,7 +683,7 @@ static int handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
 		for (i = 0; i < dd->ipath_cfgports; i++) {
 			struct ipath_portdata *pd = dd->ipath_pd[i];
 			if (i == 0) {
-				hd = dd->ipath_port0head;
+				hd = pd->port_head;
 				tl = (u32) le64_to_cpu(
 					*dd->ipath_hdrqtailptr);
 			} else if (pd && pd->port_cnt &&
@@ -693,7 +693,7 @@ static int handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
 				 * except kernel
 				 */
 				tl = *(u64 *) pd->port_rcvhdrtail_kvaddr;
-				if (tl == dd->ipath_lastrcvhdrqtails[i])
+				if (tl == pd->port_lastrcvhdrqtail)
 					continue;
 				hd = ipath_read_ureg32(dd, ur_rcvhdrhead,
 						       i);
@@ -703,7 +703,7 @@ static int handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
 			    (!hd && tl == dd->ipath_hdrqlast)) {
 				if (i == 0)
 					chkerrpkts = 1;
-				dd->ipath_lastrcvhdrqtails[i] = tl;
+				pd->port_lastrcvhdrqtail = tl;
 				pd->port_hdrqfull++;
 				/* flush hdrqfull so that poll() sees it */
 				wmb();
@@ -712,6 +712,8 @@ static int handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
 		}
 	}
 	if (errs & INFINIPATH_E_RRCVEGRFULL) {
+		struct ipath_portdata *pd = dd->ipath_pd[0];
+
 		/*
 		 * since this is of less importance and not likely to
 		 * happen without also getting hdrfull, only count
@@ -719,7 +721,7 @@ static int handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
 		 * vs user)
 		 */
 		ipath_stats.sps_etidfull++;
-		if (dd->ipath_port0head !=
+		if (pd->port_head !=
 		    (u32) le64_to_cpu(*dd->ipath_hdrqtailptr))
 			chkerrpkts = 1;
 	}
@@ -795,6 +797,7 @@ void ipath_clear_freeze(struct ipath_devdata *dd)
 {
 	int i, im;
 	__le64 val;
+	unsigned long flags;
 
 	/* disable error interrupts, to avoid confusion */
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_errormask, 0ULL);
@@ -813,11 +816,14 @@ void ipath_clear_freeze(struct ipath_devdata *dd)
 			 dd->ipath_control);
 
 	/* ensure pio avail updates continue */
+	spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
 		 dd->ipath_sendctrl & ~INFINIPATH_S_PIOBUFAVAILUPD);
 	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
-		 dd->ipath_sendctrl);
+			 dd->ipath_sendctrl);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+	spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 
 	/*
 	 * We just enabled pioavailupdate, so dma copy is almost certainly
@@ -825,8 +831,8 @@ void ipath_clear_freeze(struct ipath_devdata *dd)
 	 */
 	for (i = 0; i < dd->ipath_pioavregs; i++) {
 		/* deal with 6110 chip bug */
-		im = i > 3 ? ((i&1) ? i-1 : i+1) : i;
-		val = ipath_read_kreg64(dd, (0x1000/sizeof(u64))+im);
+		im = i > 3 ? i ^ 1 : i;
+		val = ipath_read_kreg64(dd, (0x1000 / sizeof(u64)) + im);
 		dd->ipath_pioavailregs_dma[i] = dd->ipath_pioavailshadow[i]
 			= le64_to_cpu(val);
 	}
@@ -849,7 +855,7 @@ void ipath_clear_freeze(struct ipath_devdata *dd)
 
 /* this is separate to allow for better optimization of ipath_intr() */
 
-static void ipath_bad_intr(struct ipath_devdata *dd, u32 * unexpectp)
+static noinline void ipath_bad_intr(struct ipath_devdata *dd, u32 *unexpectp)
 {
 	/*
 	 * sometimes happen during driver init and unload, don't want
@@ -877,7 +883,7 @@ static void ipath_bad_intr(struct ipath_devdata *dd, u32 * unexpectp)
 				dd->ipath_f_free_irq(dd);
 			}
 		}
-		if (ipath_read_kreg32(dd, dd->ipath_kregs->kr_intmask)) {
+		if (ipath_read_ireg(dd, dd->ipath_kregs->kr_intmask)) {
 			ipath_dev_err(dd, "%u unexpected interrupts, "
 				      "disabling interrupts completely\n",
 				      *unexpectp);
@@ -892,7 +898,7 @@ static void ipath_bad_intr(struct ipath_devdata *dd, u32 * unexpectp)
 			  "ignoring\n");
 }
 
-static void ipath_bad_regread(struct ipath_devdata *dd)
+static noinline void ipath_bad_regread(struct ipath_devdata *dd)
 {
 	static int allbits;
 
@@ -920,31 +926,9 @@ static void ipath_bad_regread(struct ipath_devdata *dd)
 	}
 }
 
-static void handle_port_pioavail(struct ipath_devdata *dd)
-{
-	u32 i;
-	/*
-	 * start from port 1, since for now port 0  is never using
-	 * wait_event for PIO
-	 */
-	for (i = 1; dd->ipath_portpiowait && i < dd->ipath_cfgports; i++) {
-		struct ipath_portdata *pd = dd->ipath_pd[i];
-
-		if (pd && pd->port_cnt &&
-		    dd->ipath_portpiowait & (1U << i)) {
-			clear_bit(i, &dd->ipath_portpiowait);
-			if (test_bit(IPATH_PORT_WAITING_PIO,
-				     &pd->port_flag)) {
-				clear_bit(IPATH_PORT_WAITING_PIO,
-					  &pd->port_flag);
-				wake_up_interruptible(&pd->port_wait);
-			}
-		}
-	}
-}
-
 static void handle_layer_pioavail(struct ipath_devdata *dd)
 {
+	unsigned long flags;
 	int ret;
 
 	ret = ipath_ib_piobufavail(dd->verbs_dev);
@@ -953,9 +937,12 @@ static void handle_layer_pioavail(struct ipath_devdata *dd)
 
 	return;
 set:
-	set_bit(IPATH_S_PIOINTBUFAVAIL, &dd->ipath_sendctrl);
+	spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
+	dd->ipath_sendctrl |= INFINIPATH_S_PIOINTBUFAVAIL;
 	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
 			 dd->ipath_sendctrl);
+	ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+	spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 }
 
 /*
@@ -969,7 +956,15 @@ static void handle_urcv(struct ipath_devdata *dd, u32 istat)
 	int i;
 	int rcvdint = 0;
 
-	/* test_bit below needs this... */
+	/*
+	 * test_and_clear_bit(IPATH_PORT_WAITING_RCV) and
+	 * test_and_clear_bit(IPATH_PORT_WAITING_URG) below
+	 * would both like timely updates of the bits so that
+	 * we don't pass them by unnecessarily.  the rmb()
+	 * here ensures that we see them promptly -- the
+	 * corresponding wmb()'s are in ipath_poll_urgent()
+	 * and ipath_poll_next()...
+	 */
 	rmb();
 	portr = ((istat >> INFINIPATH_I_RCVAVAIL_SHIFT) &
 		 dd->ipath_i_rcvavail_mask)
@@ -980,7 +975,7 @@ static void handle_urcv(struct ipath_devdata *dd, u32 istat)
 		if (portr & (1 << i) && pd && pd->port_cnt) {
 			if (test_and_clear_bit(IPATH_PORT_WAITING_RCV,
 					       &pd->port_flag)) {
-				clear_bit(i + INFINIPATH_R_INTRAVAIL_SHIFT,
+				clear_bit(i + dd->ipath_r_intravail_shift,
 					  &dd->ipath_rcvctrl);
 				wake_up_interruptible(&pd->port_wait);
 				rcvdint = 1;
@@ -1039,7 +1034,7 @@ irqreturn_t ipath_intr(int irq, void *data)
 		goto bail;
 	}
 
-	istat = ipath_read_kreg32(dd, dd->ipath_kregs->kr_intstatus);
+	istat = ipath_read_ireg(dd, dd->ipath_kregs->kr_intstatus);
 
 	if (unlikely(!istat)) {
 		ipath_stats.sps_nullintr++;
@@ -1180,7 +1175,7 @@ irqreturn_t ipath_intr(int irq, void *data)
 	 * for receive are at the bottom.
 	 */
 	if (chk0rcv) {
-		ipath_kreceive(dd);
+		ipath_kreceive(dd->ipath_pd[0]);
 		istat &= ~port0rbits;
 	}
 
@@ -1191,12 +1186,14 @@ irqreturn_t ipath_intr(int irq, void *data)
 		handle_urcv(dd, istat);
 
 	if (istat & INFINIPATH_I_SPIOBUFAVAIL) {
-		clear_bit(IPATH_S_PIOINTBUFAVAIL, &dd->ipath_sendctrl);
+		unsigned long flags;
+
+		spin_lock_irqsave(&dd->ipath_sendctrl_lock, flags);
+		dd->ipath_sendctrl &= ~INFINIPATH_S_PIOINTBUFAVAIL;
 		ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
 				 dd->ipath_sendctrl);
-
-		if (dd->ipath_portpiowait)
-			handle_port_pioavail(dd);
+		ipath_read_kreg64(dd, dd->ipath_kregs->kr_scratch);
+		spin_unlock_irqrestore(&dd->ipath_sendctrl_lock, flags);
 
 		handle_layer_pioavail(dd);
 	}

drivers/infiniband/hw/ipath/ipath_keys.c

@@ -128,9 +128,8 @@ int ipath_lkey_ok(struct ipath_qp *qp, struct ipath_sge *isge,
 	int ret;
 
 	/*
-	 * We use LKEY == zero to mean a physical kmalloc() address.
-	 * This is a bit of a hack since we rely on dma_map_single()
-	 * being reversible by calling bus_to_virt().
+	 * We use LKEY == zero for kernel virtual addresses
+	 * (see ipath_get_dma_mr and ipath_dma.c).
 	 */
 	if (sge->lkey == 0) {
 		struct ipath_pd *pd = to_ipd(qp->ibqp.pd);

drivers/infiniband/hw/ipath/ipath_qp.c

@@ -387,8 +387,8 @@ int ipath_error_qp(struct ipath_qp *qp, enum ib_wc_status err)
 	struct ib_wc wc;
 	int ret = 0;
 
-	ipath_dbg("QP%d/%d in error state\n",
-		  qp->ibqp.qp_num, qp->remote_qpn);
+	ipath_dbg("QP%d/%d in error state (%d)\n",
+		  qp->ibqp.qp_num, qp->remote_qpn, err);
 
 	spin_lock(&dev->pending_lock);
 	/* XXX What if its already removed by the timeout code? */
@@ -855,8 +855,6 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd,
 	 * See ipath_mmap() for details.
 	 */
 	if (udata && udata->outlen >= sizeof(__u64)) {
-		int err;
-
 		if (!qp->r_rq.wq) {
 			__u64 offset = 0;
 

drivers/infiniband/hw/ipath/ipath_rc.c

@@ -647,6 +647,7 @@ static void send_rc_ack(struct ipath_qp *qp)
 
 queue_ack:
 	spin_lock_irqsave(&qp->s_lock, flags);
+	dev->n_rc_qacks++;
 	qp->s_flags |= IPATH_S_ACK_PENDING;
 	qp->s_nak_state = qp->r_nak_state;
 	qp->s_ack_psn = qp->r_ack_psn;
@@ -798,11 +799,13 @@ void ipath_restart_rc(struct ipath_qp *qp, u32 psn, struct ib_wc *wc)
 
 static inline void update_last_psn(struct ipath_qp *qp, u32 psn)
 {
-	if (qp->s_wait_credit) {
-		qp->s_wait_credit = 0;
-		tasklet_hi_schedule(&qp->s_task);
+	if (qp->s_last_psn != psn) {
+		qp->s_last_psn = psn;
+		if (qp->s_wait_credit) {
+			qp->s_wait_credit = 0;
+			tasklet_hi_schedule(&qp->s_task);
+		}
 	}
-	qp->s_last_psn = psn;
 }
 
 /**
@@ -1653,13 +1656,6 @@ void ipath_rc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
 	case OP(SEND_FIRST):
 		if (!ipath_get_rwqe(qp, 0)) {
 		rnr_nak:
-			/*
-			 * A RNR NAK will ACK earlier sends and RDMA writes.
-			 * Don't queue the NAK if a RDMA read or atomic
-			 * is pending though.
-			 */
-			if (qp->r_nak_state)
-				goto done;
 			qp->r_nak_state = IB_RNR_NAK | qp->r_min_rnr_timer;
 			qp->r_ack_psn = qp->r_psn;
 			goto send_ack;

drivers/infiniband/hw/ipath/ipath_srq.c

@@ -94,8 +94,8 @@ int ipath_post_srq_receive(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
 /**
  * ipath_create_srq - create a shared receive queue
  * @ibpd: the protection domain of the SRQ to create
- * @attr: the attributes of the SRQ
- * @udata: not used by the InfiniPath verbs driver
+ * @srq_init_attr: the attributes of the SRQ
+ * @udata: data from libipathverbs when creating a user SRQ
  */
 struct ib_srq *ipath_create_srq(struct ib_pd *ibpd,
 				struct ib_srq_init_attr *srq_init_attr,

drivers/infiniband/ulp/iser/Kconfig

@@ -8,5 +8,5 @@ config INFINIBAND_ISER
           that speak iSCSI over iSER over InfiniBand.
 
 	  The iSER protocol is defined by IETF.
-	  See <http://www.ietf.org/internet-drafts/draft-ietf-ips-iser-05.txt>
-	  and <http://www.infinibandta.org/members/spec/iser_annex_060418.pdf>
+	  See <http://www.ietf.org/rfc/rfc5046.txt>
+	  and <http://www.infinibandta.org/members/spec/Annex_iSER.PDF>