Merge branches 'ehca', 'ipoib' and 'mlx4' into for-linus

drivers/infiniband/hw/mlx4/qp.c

@@ -1462,7 +1462,8 @@ static void __set_data_seg(struct mlx4_wqe_data_seg *dseg, struct ib_sge *sg)
 }
 
 static int build_lso_seg(struct mlx4_wqe_lso_seg *wqe, struct ib_send_wr *wr,
-			 struct mlx4_ib_qp *qp, unsigned *lso_seg_len)
+			 struct mlx4_ib_qp *qp, unsigned *lso_seg_len,
+			 __be32 *lso_hdr_sz)
 {
 	unsigned halign = ALIGN(sizeof *wqe + wr->wr.ud.hlen, 16);
 
@@ -1479,12 +1480,8 @@ static int build_lso_seg(struct mlx4_wqe_lso_seg *wqe, struct ib_send_wr *wr,
 
 	memcpy(wqe->header, wr->wr.ud.header, wr->wr.ud.hlen);
 
-	/* make sure LSO header is written before overwriting stamping */
-	wmb();
-
-	wqe->mss_hdr_size = cpu_to_be32((wr->wr.ud.mss - wr->wr.ud.hlen) << 16 |
-					wr->wr.ud.hlen);
-
+	*lso_hdr_sz  = cpu_to_be32((wr->wr.ud.mss - wr->wr.ud.hlen) << 16 |
+				   wr->wr.ud.hlen);
 	*lso_seg_len = halign;
 	return 0;
 }
@@ -1518,6 +1515,9 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 	int uninitialized_var(stamp);
 	int uninitialized_var(size);
 	unsigned uninitialized_var(seglen);
+	__be32 dummy;
+	__be32 *lso_wqe;
+	__be32 uninitialized_var(lso_hdr_sz);
 	int i;
 
 	spin_lock_irqsave(&qp->sq.lock, flags);
@@ -1525,6 +1525,8 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 	ind = qp->sq_next_wqe;
 
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		lso_wqe = &dummy;
+
 		if (mlx4_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq)) {
 			err = -ENOMEM;
 			*bad_wr = wr;
@@ -1606,11 +1608,12 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			size += sizeof (struct mlx4_wqe_datagram_seg) / 16;
 
 			if (wr->opcode == IB_WR_LSO) {
-				err = build_lso_seg(wqe, wr, qp, &seglen);
+				err = build_lso_seg(wqe, wr, qp, &seglen, &lso_hdr_sz);
 				if (unlikely(err)) {
 					*bad_wr = wr;
 					goto out;
 				}
+				lso_wqe = (__be32 *) wqe;
 				wqe  += seglen;
 				size += seglen / 16;
 			}
@@ -1652,6 +1655,14 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 		for (i = wr->num_sge - 1; i >= 0; --i, --dseg)
 			set_data_seg(dseg, wr->sg_list + i);
 
+		/*
+		 * Possibly overwrite stamping in cacheline with LSO
+		 * segment only after making sure all data segments
+		 * are written.
+		 */
+		wmb();
+		*lso_wqe = lso_hdr_sz;
+
 		ctrl->fence_size = (wr->send_flags & IB_SEND_FENCE ?
 				    MLX4_WQE_CTRL_FENCE : 0) | size;
 
@@ -1686,7 +1697,6 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			stamp_send_wqe(qp, stamp, size * 16);
 			ind = pad_wraparound(qp, ind);
 		}
-
 	}
 
 out:

drivers/infiniband/ulp/ipoib/ipoib_main.c

@@ -106,23 +106,17 @@ int ipoib_open(struct net_device *dev)
 
 	ipoib_dbg(priv, "bringing up interface\n");
 
-	set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags);
+	if (!test_and_set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
+		napi_enable(&priv->napi);
 
 	if (ipoib_pkey_dev_delay_open(dev))
 		return 0;
 
-	napi_enable(&priv->napi);
+	if (ipoib_ib_dev_open(dev))
+		goto err_disable;
 
-	if (ipoib_ib_dev_open(dev)) {
-		napi_disable(&priv->napi);
-		return -EINVAL;
-	}
-
-	if (ipoib_ib_dev_up(dev)) {
-		ipoib_ib_dev_stop(dev, 1);
-		napi_disable(&priv->napi);
-		return -EINVAL;
-	}
+	if (ipoib_ib_dev_up(dev))
+		goto err_stop;
 
 	if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) {
 		struct ipoib_dev_priv *cpriv;
@@ -144,6 +138,15 @@ int ipoib_open(struct net_device *dev)
 	netif_start_queue(dev);
 
 	return 0;
+
+err_stop:
+	ipoib_ib_dev_stop(dev, 1);
+
+err_disable:
+	napi_disable(&priv->napi);
+	clear_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags);
+
+	return -EINVAL;
 }
 
 static int ipoib_stop(struct net_device *dev)

drivers/infiniband/ulp/ipoib/ipoib_multicast.c

@@ -409,7 +409,7 @@ static int ipoib_mcast_join_complete(int status,
 	}
 
 	if (mcast->logcount++ < 20) {
-		if (status == -ETIMEDOUT) {
+		if (status == -ETIMEDOUT || status == -EAGAIN) {
 			ipoib_dbg_mcast(priv, "multicast join failed for %pI6, status %d\n",
 					mcast->mcmember.mgid.raw, status);
 		} else {

drivers/infiniband/ulp/ipoib/ipoib_vlan.c

@@ -61,6 +61,7 @@ int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
 
 	ppriv = netdev_priv(pdev);
 
+	rtnl_lock();
 	mutex_lock(&ppriv->vlan_mutex);
 
 	/*
@@ -111,7 +112,7 @@ int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
 		goto device_init_failed;
 	}
 
-	result = register_netdev(priv->dev);
+	result = register_netdevice(priv->dev);
 	if (result) {
 		ipoib_warn(priv, "failed to initialize; error %i", result);
 		goto register_failed;
@@ -134,12 +135,13 @@ int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
 	list_add_tail(&priv->list, &ppriv->child_intfs);
 
 	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
 
 	return 0;
 
 sysfs_failed:
 	ipoib_delete_debug_files(priv->dev);
-	unregister_netdev(priv->dev);
+	unregister_netdevice(priv->dev);
 
 register_failed:
 	ipoib_dev_cleanup(priv->dev);
@@ -149,6 +151,7 @@ int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
 
 err:
 	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
 	return result;
 }
 
@@ -162,10 +165,11 @@ int ipoib_vlan_delete(struct net_device *pdev, unsigned short pkey)
 
 	ppriv = netdev_priv(pdev);
 
+	rtnl_lock();
 	mutex_lock(&ppriv->vlan_mutex);
 	list_for_each_entry_safe(priv, tpriv, &ppriv->child_intfs, list) {
 		if (priv->pkey == pkey) {
-			unregister_netdev(priv->dev);
+			unregister_netdevice(priv->dev);
 			ipoib_dev_cleanup(priv->dev);
 			list_del(&priv->list);
 			free_netdev(priv->dev);
@@ -175,6 +179,7 @@ int ipoib_vlan_delete(struct net_device *pdev, unsigned short pkey)
 		}
 	}
 	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
 
 	return ret;
 }

drivers/net/mlx4/profile.c

@@ -107,9 +107,9 @@ u64 mlx4_make_profile(struct mlx4_dev *dev,
 	profile[MLX4_RES_AUXC].num    = request->num_qp;
 	profile[MLX4_RES_SRQ].num     = request->num_srq;
 	profile[MLX4_RES_CQ].num      = request->num_cq;
-	profile[MLX4_RES_EQ].num      = min(dev_cap->max_eqs,
-					    dev_cap->reserved_eqs +
-					    num_possible_cpus() + 1);
+	profile[MLX4_RES_EQ].num      = min_t(unsigned, dev_cap->max_eqs,
+					      dev_cap->reserved_eqs +
+					      num_possible_cpus() + 1);
 	profile[MLX4_RES_DMPT].num    = request->num_mpt;
 	profile[MLX4_RES_CMPT].num    = MLX4_NUM_CMPTS;
 	profile[MLX4_RES_MTT].num     = request->num_mtt;