3 files changed, 29 insertions, 14 deletions

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index a3c5af1d7ec..de5263beab4 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -367,7 +367,7 @@ int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
 		if (err)
 			goto out;
 	} else {
-		/* Can't be smaller then the number of outstanding CQEs */
+		/* Can't be smaller than the number of outstanding CQEs */
 		outst_cqe = mlx4_ib_get_outstanding_cqes(cq);
 		if (entries < outst_cqe + 1) {
 			err = 0;
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index dcefe1fceb5..61588bd273b 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -543,14 +543,21 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 {
 	static int mlx4_ib_version_printed;
 	struct mlx4_ib_dev *ibdev;
+	int num_ports = 0;
 	int i;
 
-
 	if (!mlx4_ib_version_printed) {
 		printk(KERN_INFO "%s", mlx4_ib_version);
 		++mlx4_ib_version_printed;
 	}
 
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
+		num_ports++;
+
+	/* No point in registering a device with no ports... */
+	if (num_ports == 0)
+		return NULL;
+
 	ibdev = (struct mlx4_ib_dev *) ib_alloc_device(sizeof *ibdev);
 	if (!ibdev) {
 		dev_err(&dev->pdev->dev, "Device struct alloc failed\n");
@@ -574,9 +581,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.owner		= THIS_MODULE;
 	ibdev->ib_dev.node_type		= RDMA_NODE_IB_CA;
 	ibdev->ib_dev.local_dma_lkey	= dev->caps.reserved_lkey;
-	ibdev->num_ports = 0;
-	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
-		ibdev->num_ports++;
+	ibdev->num_ports		= num_ports;
 	ibdev->ib_dev.phys_port_cnt     = ibdev->num_ports;
 	ibdev->ib_dev.num_comp_vectors	= dev->caps.num_comp_vectors;
 	ibdev->ib_dev.dma_device	= &dev->pdev->dev;
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index 39167a797f9..a91cb4c3fa5 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/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: