IB/ipath: Add barrier before updating WC head in shared memory

Add a barrier to make sure the CPU doesn't reorder writes to memory,
since user programs can be polling on the head index update and the
entry should be written before that.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

drivers/infiniband/hw/ipath/ipath_cq.c

@@ -90,6 +90,8 @@ void ipath_cq_enter(struct ipath_cq *cq, struct ib_wc *entry, int solicited)
 	wc->queue[head].sl = entry->sl;
 	wc->queue[head].dlid_path_bits = entry->dlid_path_bits;
 	wc->queue[head].port_num = entry->port_num;
+	/* Make sure queue entry is written before the head index. */
+	smp_wmb();
 	wc->head = next;
 
 	if (cq->notify == IB_CQ_NEXT_COMP ||
@@ -139,7 +141,8 @@ int ipath_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
 
 		if (tail == wc->head)
 			break;
-
+		/* Make sure entry is read after head index is read. */
+		smp_rmb();
 		qp = ipath_lookup_qpn(&to_idev(cq->ibcq.device)->qp_table,
 				      wc->queue[tail].qp_num);
 		entry->qp = &qp->ibqp;

drivers/infiniband/hw/ipath/ipath_ruc.c

@@ -194,6 +194,8 @@ int ipath_get_rwqe(struct ipath_qp *qp, int wr_id_only)
 			ret = 0;
 			goto bail;
 		}
+		/* Make sure entry is read after head index is read. */
+		smp_rmb();
 		wqe = get_rwqe_ptr(rq, tail);
 		if (++tail >= rq->size)
 			tail = 0;

drivers/infiniband/hw/ipath/ipath_srq.c

@@ -80,6 +80,8 @@ int ipath_post_srq_receive(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
 		wqe->num_sge = wr->num_sge;
 		for (i = 0; i < wr->num_sge; i++)
 			wqe->sg_list[i] = wr->sg_list[i];
+		/* Make sure queue entry is written before the head index. */
+		smp_wmb();
 		wq->head = next;
 		spin_unlock_irqrestore(&srq->rq.lock, flags);
 	}

drivers/infiniband/hw/ipath/ipath_ud.c

@@ -176,6 +176,8 @@ static void ipath_ud_loopback(struct ipath_qp *sqp,
 			dev->n_pkt_drops++;
 			goto bail_sge;
 		}
+		/* Make sure entry is read after head index is read. */
+		smp_rmb();
 		wqe = get_rwqe_ptr(rq, tail);
 		if (++tail >= rq->size)
 			tail = 0;

drivers/infiniband/hw/ipath/ipath_verbs.c

@@ -327,6 +327,8 @@ static int ipath_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 		wqe->num_sge = wr->num_sge;
 		for (i = 0; i < wr->num_sge; i++)
 			wqe->sg_list[i] = wr->sg_list[i];
+		/* Make sure queue entry is written before the head index. */
+		smp_wmb();
 		wq->head = next;
 		spin_unlock_irqrestore(&qp->r_rq.lock, flags);
 	}