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:
  IPoIB: Fix crash in path_rec_completion()
  IPoIB: Fix hang in ipoib_flush_paths()
  IPoIB: Don't enable NAPI when it's already enabled
  RDMA/cxgb3: Fix deadlock in iw_cxgb3 (hang when configuring interface)
  IB/ehca: Remove reference to special QP in case of port activation failure
  IB/mlx4: Set umem field to NULL in mlx4_ib_alloc_fast_reg_mr()
  mlx4_core: Fix unused variable warning
  RDMA/nes: Mitigate compatibility issue regarding PCIe write credits
  RDMA/nes: Fix CQ allocation scheme for multicast receive queue apps
  RDMA/nes: Correct handling of PBL resources
  RDMA/nes: Reindent mis-indented spinlocks
  RDMA/cxgb3: Fix too-big reserved field zeroing in iwch_post_zb_read()
  IB/ipath: Fix RDMA write with immediate copy of last packet

drivers/infiniband/hw/cxgb3/iwch_provider.c

@@ -1102,9 +1102,7 @@ static u64 fw_vers_string_to_u64(struct iwch_dev *iwch_dev)
 	char *cp, *next;
 	unsigned fw_maj, fw_min, fw_mic;
 
-	rtnl_lock();
 	lldev->ethtool_ops->get_drvinfo(lldev, &info);
-	rtnl_unlock();
 
 	next = info.fw_version + 1;
 	cp = strsep(&next, ".");
@@ -1192,9 +1190,7 @@ static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, ch
 	struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
 
 	PDBG("%s dev 0x%p\n", __func__, dev);
-	rtnl_lock();
 	lldev->ethtool_ops->get_drvinfo(lldev, &info);
-	rtnl_unlock();
 	return sprintf(buf, "%s\n", info.fw_version);
 }
 
@@ -1207,9 +1203,7 @@ static ssize_t show_hca(struct device *dev, struct device_attribute *attr,
 	struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
 
 	PDBG("%s dev 0x%p\n", __func__, dev);
-	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

@@ -745,7 +745,6 @@ int iwch_post_zb_read(struct iwch_qp *qhp)
 	wqe->read.rdmaop = T3_READ_REQ;
 	wqe->read.reserved[0] = 0;
 	wqe->read.reserved[1] = 0;
-	wqe->read.reserved[2] = 0;
 	wqe->read.rem_stag = cpu_to_be32(1);
 	wqe->read.rem_to = cpu_to_be64(1);
 	wqe->read.local_stag = cpu_to_be32(1);

drivers/infiniband/hw/ehca/ehca_irq.c

@@ -359,36 +359,48 @@ static void notify_port_conf_change(struct ehca_shca *shca, int port_num)
 	*old_attr = new_attr;
 }
 
+/* replay modify_qp for sqps -- return 0 if all is well, 1 if AQP1 destroyed */
+static int replay_modify_qp(struct ehca_sport *sport)
+{
+	int aqp1_destroyed;
+	unsigned long flags;
+
+	spin_lock_irqsave(&sport->mod_sqp_lock, flags);
+
+	aqp1_destroyed = !sport->ibqp_sqp[IB_QPT_GSI];
+
+	if (sport->ibqp_sqp[IB_QPT_SMI])
+		ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_SMI]);
+	if (!aqp1_destroyed)
+		ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_GSI]);
+
+	spin_unlock_irqrestore(&sport->mod_sqp_lock, flags);
+
+	return aqp1_destroyed;
+}
+
 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)) {
-			int suppress_event;
+			/* only replay modify_qp calls in autodetect mode;
-			/* replay modify_qp for sqps */
+			 * if AQP1 was destroyed, the port is already down
-			spin_lock_irqsave(&sport->mod_sqp_lock, flags);
+			 * again and we can drop the event.
-			suppress_event = !sport->ibqp_sqp[IB_QPT_GSI];
+			 */
-			if (sport->ibqp_sqp[IB_QPT_SMI])
+			if (ehca_nr_ports < 0)
-				ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_SMI]);
+				if (replay_modify_qp(sport))
-			if (!suppress_event)
+					break;
-				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,
+			ehca_query_sma_attr(shca, port, &sport->saved_attr);
-					    &sport->saved_attr);
 		} else {
 			sport->port_state = IB_PORT_DOWN;
 			dispatch_port_event(shca, port, IB_EVENT_PORT_ERR,

drivers/infiniband/hw/ehca/ehca_qp.c

@@ -860,6 +860,11 @@ static struct ehca_qp *internal_create_qp(
 	if (qp_type == IB_QPT_GSI) {
 		h_ret = ehca_define_sqp(shca, my_qp, init_attr);
 		if (h_ret != H_SUCCESS) {
+			kfree(my_qp->mod_qp_parm);
+			my_qp->mod_qp_parm = NULL;
+			/* the QP pointer is no longer valid */
+			shca->sport[init_attr->port_num - 1].ibqp_sqp[qp_type] =
+				NULL;
 			ret = ehca2ib_return_code(h_ret);
 			goto create_qp_exit6;
 		}

drivers/infiniband/hw/ipath/ipath_ruc.c

@@ -156,7 +156,7 @@ bail:
 /**
  * ipath_get_rwqe - copy the next RWQE into the QP's RWQE
  * @qp: the QP
- * @wr_id_only: update wr_id only, not SGEs
+ * @wr_id_only: update qp->r_wr_id only, not qp->r_sge
  *
  * Return 0 if no RWQE is available, otherwise return 1.
  *
@@ -173,8 +173,6 @@ int ipath_get_rwqe(struct ipath_qp *qp, int wr_id_only)
 	u32 tail;
 	int ret;
 
-	qp->r_sge.sg_list = qp->r_sg_list;
-
 	if (qp->ibqp.srq) {
 		srq = to_isrq(qp->ibqp.srq);
 		handler = srq->ibsrq.event_handler;
@@ -206,8 +204,10 @@ int ipath_get_rwqe(struct ipath_qp *qp, int wr_id_only)
 		wqe = get_rwqe_ptr(rq, tail);
 		if (++tail >= rq->size)
 			tail = 0;
-	} while (!wr_id_only && !ipath_init_sge(qp, wqe, &qp->r_len,
+		if (wr_id_only)
-						&qp->r_sge));
+			break;
+		qp->r_sge.sg_list = qp->r_sg_list;
+	} while (!ipath_init_sge(qp, wqe, &qp->r_len, &qp->r_sge));
 	qp->r_wr_id = wqe->wr_id;
 	wq->tail = tail;
 

drivers/infiniband/hw/mlx4/mr.c

@@ -205,6 +205,7 @@ struct ib_mr *mlx4_ib_alloc_fast_reg_mr(struct ib_pd *pd,
 		goto err_mr;
 
 	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+	mr->umem = NULL;
 
 	return &mr->ibmr;
 

drivers/infiniband/hw/nes/nes.c

@@ -95,6 +95,10 @@ unsigned int wqm_quanta = 0x10000;
 module_param(wqm_quanta, int, 0644);
 MODULE_PARM_DESC(wqm_quanta, "WQM quanta");
 
+static unsigned int limit_maxrdreqsz;
+module_param(limit_maxrdreqsz, bool, 0644);
+MODULE_PARM_DESC(limit_maxrdreqsz, "Limit max read request size to 256 Bytes");
+
 LIST_HEAD(nes_adapter_list);
 static LIST_HEAD(nes_dev_list);
 
@@ -588,6 +592,18 @@ static int __devinit nes_probe(struct pci_dev *pcidev, const struct pci_device_i
 						nesdev->nesadapter->port_count;
 	}
 
+	if ((limit_maxrdreqsz ||
+	     ((nesdev->nesadapter->phy_type[0] == NES_PHY_TYPE_GLADIUS) &&
+	      (hw_rev == NE020_REV1))) &&
+	    (pcie_get_readrq(pcidev) > 256)) {
+		if (pcie_set_readrq(pcidev, 256))
+			printk(KERN_ERR PFX "Unable to set max read request"
+				" to 256 bytes\n");
+		else
+			nes_debug(NES_DBG_INIT, "Max read request size set"
+				" to 256 bytes\n");
+	}
+
 	tasklet_init(&nesdev->dpc_tasklet, nes_dpc, (unsigned long)nesdev);
 
 	/* bring up the Control QP */

drivers/infiniband/hw/nes/nes_hw.h

@@ -40,6 +40,7 @@
 #define NES_PHY_TYPE_ARGUS     4
 #define NES_PHY_TYPE_PUMA_1G   5
 #define NES_PHY_TYPE_PUMA_10G  6
+#define NES_PHY_TYPE_GLADIUS   7
 
 #define NES_MULTICAST_PF_MAX 8
 

drivers/infiniband/hw/nes/nes_verbs.c

@@ -220,14 +220,14 @@ static int nes_bind_mw(struct ib_qp *ibqp, struct ib_mw *ibmw,
 	if (nesqp->ibqp_state > IB_QPS_RTS)
 		return -EINVAL;
 
-		spin_lock_irqsave(&nesqp->lock, flags);
+	spin_lock_irqsave(&nesqp->lock, flags);
 
 	head = nesqp->hwqp.sq_head;
 	qsize = nesqp->hwqp.sq_tail;
 
 	/* Check for SQ overflow */
 	if (((head + (2 * qsize) - nesqp->hwqp.sq_tail) % qsize) == (qsize - 1)) {
-			spin_unlock_irqrestore(&nesqp->lock, flags);
+		spin_unlock_irqrestore(&nesqp->lock, flags);
 		return -EINVAL;
 	}
 
@@ -269,7 +269,7 @@ static int nes_bind_mw(struct ib_qp *ibqp, struct ib_mw *ibmw,
 	nes_write32(nesdev->regs+NES_WQE_ALLOC,
 			(1 << 24) | 0x00800000 | nesqp->hwqp.qp_id);
 
-		spin_unlock_irqrestore(&nesqp->lock, flags);
+	spin_unlock_irqrestore(&nesqp->lock, flags);
 
 	return 0;
 }
@@ -349,7 +349,7 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd,
 			if (nesfmr->nesmr.pbls_used > nesadapter->free_4kpbl) {
 				spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 				ret = -ENOMEM;
-				goto failed_vpbl_alloc;
+				goto failed_vpbl_avail;
 			} else {
 				nesadapter->free_4kpbl -= nesfmr->nesmr.pbls_used;
 			}
@@ -357,7 +357,7 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd,
 			if (nesfmr->nesmr.pbls_used > nesadapter->free_256pbl) {
 				spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 				ret = -ENOMEM;
-				goto failed_vpbl_alloc;
+				goto failed_vpbl_avail;
 			} else {
 				nesadapter->free_256pbl -= nesfmr->nesmr.pbls_used;
 			}
@@ -391,14 +391,14 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd,
 			goto failed_vpbl_alloc;
 		}
 
-		nesfmr->root_vpbl.leaf_vpbl = kzalloc(sizeof(*nesfmr->root_vpbl.leaf_vpbl)*1024, GFP_KERNEL);
+		nesfmr->leaf_pbl_cnt = nesfmr->nesmr.pbls_used-1;
+		nesfmr->root_vpbl.leaf_vpbl = kzalloc(sizeof(*nesfmr->root_vpbl.leaf_vpbl)*1024, GFP_ATOMIC);
 		if (!nesfmr->root_vpbl.leaf_vpbl) {
 			spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
 			ret = -ENOMEM;
 			goto failed_leaf_vpbl_alloc;
 		}
 
-		nesfmr->leaf_pbl_cnt = nesfmr->nesmr.pbls_used-1;
 		nes_debug(NES_DBG_MR, "two level pbl, root_vpbl.pbl_vbase=%p"
 				" leaf_pbl_cnt=%d root_vpbl.leaf_vpbl=%p\n",
 				nesfmr->root_vpbl.pbl_vbase, nesfmr->leaf_pbl_cnt, nesfmr->root_vpbl.leaf_vpbl);
@@ -519,6 +519,16 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd,
 				nesfmr->root_vpbl.pbl_pbase);
 
 	failed_vpbl_alloc:
+	if (nesfmr->nesmr.pbls_used != 0) {
+		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+		if (nesfmr->nesmr.pbl_4k)
+			nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used;
+		else
+			nesadapter->free_256pbl += nesfmr->nesmr.pbls_used;
+		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+	}
+
+failed_vpbl_avail:
 	kfree(nesfmr);
 
 	failed_fmr_alloc:
@@ -534,18 +544,14 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd,
  */
 static int nes_dealloc_fmr(struct ib_fmr *ibfmr)
 {
+	unsigned long flags;
 	struct nes_mr *nesmr = to_nesmr_from_ibfmr(ibfmr);
 	struct nes_fmr *nesfmr = to_nesfmr(nesmr);
 	struct nes_vnic *nesvnic = to_nesvnic(ibfmr->device);
 	struct nes_device *nesdev = nesvnic->nesdev;
-	struct nes_mr temp_nesmr = *nesmr;
+	struct nes_adapter *nesadapter = nesdev->nesadapter;
 	int i = 0;
 
-	temp_nesmr.ibmw.device = ibfmr->device;
-	temp_nesmr.ibmw.pd = ibfmr->pd;
-	temp_nesmr.ibmw.rkey = ibfmr->rkey;
-	temp_nesmr.ibmw.uobject = NULL;
-
 	/* free the resources */
 	if (nesfmr->leaf_pbl_cnt == 0) {
 		/* single PBL case */
@@ -561,8 +567,24 @@ static int nes_dealloc_fmr(struct ib_fmr *ibfmr)
 		pci_free_consistent(nesdev->pcidev, 8192, nesfmr->root_vpbl.pbl_vbase,
 				nesfmr->root_vpbl.pbl_pbase);
 	}
+	nesmr->ibmw.device = ibfmr->device;
+	nesmr->ibmw.pd = ibfmr->pd;
+	nesmr->ibmw.rkey = ibfmr->rkey;
+	nesmr->ibmw.uobject = NULL;
 
-	return nes_dealloc_mw(&temp_nesmr.ibmw);
+	if (nesfmr->nesmr.pbls_used != 0) {
+		spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+		if (nesfmr->nesmr.pbl_4k) {
+			nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used;
+			WARN_ON(nesadapter->free_4kpbl > nesadapter->max_4kpbl);
+		} else {
+			nesadapter->free_256pbl += nesfmr->nesmr.pbls_used;
+			WARN_ON(nesadapter->free_256pbl > nesadapter->max_256pbl);
+		}
+		spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+	}
+
+	return nes_dealloc_mw(&nesmr->ibmw);
 }
 
 
@@ -1595,7 +1617,7 @@ static struct ib_cq *nes_create_cq(struct ib_device *ibdev, int entries,
 		nes_ucontext->mcrqf = req.mcrqf;
 		if (nes_ucontext->mcrqf) {
 			if (nes_ucontext->mcrqf & 0x80000000)
-				nescq->hw_cq.cq_number = nesvnic->nic.qp_id + 12 + (nes_ucontext->mcrqf & 0xf) - 1;
+				nescq->hw_cq.cq_number = nesvnic->nic.qp_id + 28 + 2 * ((nes_ucontext->mcrqf & 0xf) - 1);
 			else if (nes_ucontext->mcrqf & 0x40000000)
 				nescq->hw_cq.cq_number = nes_ucontext->mcrqf & 0xffff;
 			else
@@ -3212,7 +3234,7 @@ static int nes_post_send(struct ib_qp *ibqp, struct ib_send_wr *ib_wr,
 	if (nesqp->ibqp_state > IB_QPS_RTS)
 		return -EINVAL;
 
-		spin_lock_irqsave(&nesqp->lock, flags);
+	spin_lock_irqsave(&nesqp->lock, flags);
 
 	head = nesqp->hwqp.sq_head;
 
@@ -3337,7 +3359,7 @@ static int nes_post_send(struct ib_qp *ibqp, struct ib_send_wr *ib_wr,
 				(counter << 24) | 0x00800000 | nesqp->hwqp.qp_id);
 	}
 
-		spin_unlock_irqrestore(&nesqp->lock, flags);
+	spin_unlock_irqrestore(&nesqp->lock, flags);
 
 	if (err)
 		*bad_wr = ib_wr;
@@ -3368,7 +3390,7 @@ static int nes_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *ib_wr,
 	if (nesqp->ibqp_state > IB_QPS_RTS)
 		return -EINVAL;
 
-		spin_lock_irqsave(&nesqp->lock, flags);
+	spin_lock_irqsave(&nesqp->lock, flags);
 
 	head = nesqp->hwqp.rq_head;
 
@@ -3421,7 +3443,7 @@ static int nes_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *ib_wr,
 		nes_write32(nesdev->regs+NES_WQE_ALLOC, (counter<<24) | nesqp->hwqp.qp_id);
 	}
 
-		spin_unlock_irqrestore(&nesqp->lock, flags);
+	spin_unlock_irqrestore(&nesqp->lock, flags);
 
 	if (err)
 		*bad_wr = ib_wr;
@@ -3453,7 +3475,7 @@ static int nes_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
 
 	nes_debug(NES_DBG_CQ, "\n");
 
-		spin_lock_irqsave(&nescq->lock, flags);
+	spin_lock_irqsave(&nescq->lock, flags);
 
 	head = nescq->hw_cq.cq_head;
 	cq_size = nescq->hw_cq.cq_size;
@@ -3562,7 +3584,7 @@ static int nes_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
 	nes_debug(NES_DBG_CQ, "Reporting %u completions for CQ%u.\n",
 			cqe_count, nescq->hw_cq.cq_number);
 
-		spin_unlock_irqrestore(&nescq->lock, flags);
+	spin_unlock_irqrestore(&nescq->lock, flags);
 
 	return cqe_count;
 }

drivers/infiniband/ulp/ipoib/ipoib_main.c

@@ -106,12 +106,13 @@ int ipoib_open(struct net_device *dev)
 
 	ipoib_dbg(priv, "bringing up interface\n");
 
-	napi_enable(&priv->napi);
 	set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags);
 
 	if (ipoib_pkey_dev_delay_open(dev))
 		return 0;
 
+	napi_enable(&priv->napi);
+
 	if (ipoib_ib_dev_open(dev)) {
 		napi_disable(&priv->napi);
 		return -EINVAL;
@@ -546,6 +547,7 @@ static int path_rec_start(struct net_device *dev,
 	if (path->query_id < 0) {
 		ipoib_warn(priv, "ib_sa_path_rec_get failed: %d\n", path->query_id);
 		path->query = NULL;
+		complete(&path->done);
 		return path->query_id;
 	}
 
@@ -662,7 +664,7 @@ static void unicast_arp_send(struct sk_buff *skb, struct net_device *dev,
 			skb_push(skb, sizeof *phdr);
 			__skb_queue_tail(&path->queue, skb);
 
-			if (path_rec_start(dev, path)) {
+			if (!path->query && path_rec_start(dev, path)) {
 				spin_unlock_irqrestore(&priv->lock, flags);
 				path_free(dev, path);
 				return;

drivers/net/cxgb3/cxgb3_main.c

@@ -1307,8 +1307,10 @@ static void get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info)
 	u32 fw_vers = 0;
 	u32 tp_vers = 0;
 
+	spin_lock(&adapter->stats_lock);
 	t3_get_fw_version(adapter, &fw_vers);
 	t3_get_tp_version(adapter, &tp_vers);
+	spin_unlock(&adapter->stats_lock);
 
 	strcpy(info->driver, DRV_NAME);
 	strcpy(info->version, DRV_VERSION);

drivers/net/mlx4/mlx4.h

@@ -87,6 +87,9 @@ enum {
 
 #ifdef CONFIG_MLX4_DEBUG
 extern int mlx4_debug_level;
+#else /* CONFIG_MLX4_DEBUG */
+#define mlx4_debug_level	(0)
+#endif /* CONFIG_MLX4_DEBUG */
 
 #define mlx4_dbg(mdev, format, arg...)					\
 	do {								\
@@ -94,12 +97,6 @@ extern int mlx4_debug_level;
 			dev_printk(KERN_DEBUG, &mdev->pdev->dev, format, ## arg); \
 	} while (0)
 
-#else /* CONFIG_MLX4_DEBUG */
-
-#define mlx4_dbg(mdev, format, arg...) do { (void) mdev; } while (0)
-
-#endif /* CONFIG_MLX4_DEBUG */
-
 #define mlx4_err(mdev, format, arg...) \
 	dev_err(&mdev->pdev->dev, format, ## arg)
 #define mlx4_info(mdev, format, arg...) \