IB/ipoib: Use NAPI in UD/TX flows

Instead of explicit call to poll_cq of the tx ring, use the NAPI mechanism
to handle the completions of each packet that has been sent to the HW.

The next major changes were taken:
 * The driver init completion function in the creation of the send CQ,
   that function triggers the napi scheduling.
 * The driver uses CQ for RX for both modes UD and CM, and CQ for TX
   for CM and UD.

Cc: Kamal Heib <kamalh@mellanox.com>
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

1 files changed, 76 insertions, 47 deletions

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
index c978f8ffd2bb..3b96cdaf9a83 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
@@ -264,7 +264,7 @@ static void ipoib_ib_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
 			likely(wc->wc_flags & IB_WC_IP_CSUM_OK))
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 
-	napi_gro_receive(&priv->napi, skb);
+	napi_gro_receive(&priv->recv_napi, skb);
 
 repost:
 	if (unlikely(ipoib_ib_post_receive(dev, wr_id)))
@@ -406,9 +406,10 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
 	dev_kfree_skb_any(tx_req->skb);
 
 	++priv->tx_tail;
-	if (unlikely((priv->tx_head - priv->tx_tail) == ipoib_sendq_size >> 1) &&
-	    netif_queue_stopped(dev) &&
-	    test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
+
+	if (unlikely(netif_queue_stopped(dev) &&
+		     ((priv->tx_head - priv->tx_tail) <= ipoib_sendq_size >> 1) &&
+		     test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)))
 		netif_wake_queue(dev);
 
 	if (wc->status != IB_WC_SUCCESS &&
@@ -430,17 +431,23 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
 static int poll_tx(struct ipoib_dev_priv *priv)
 {
 	int n, i;
+	struct ib_wc *wc;
 
 	n = ib_poll_cq(priv->send_cq, MAX_SEND_CQE, priv->send_wc);
-	for (i = 0; i < n; ++i)
-		ipoib_ib_handle_tx_wc(priv->dev, priv->send_wc + i);
-
+	for (i = 0; i < n; ++i) {
+		wc = priv->send_wc + i;
+		if (wc->wr_id & IPOIB_OP_CM)
+			ipoib_cm_handle_tx_wc(priv->dev, priv->send_wc + i);
+		else
+			ipoib_ib_handle_tx_wc(priv->dev, priv->send_wc + i);
+	}
 	return n == MAX_SEND_CQE;
 }
 
-int ipoib_poll(struct napi_struct *napi, int budget)
+int ipoib_rx_poll(struct napi_struct *napi, int budget)
 {
-	struct ipoib_dev_priv *priv = container_of(napi, struct ipoib_dev_priv, napi);
+	struct ipoib_dev_priv *priv =
+		container_of(napi, struct ipoib_dev_priv, recv_napi);
 	struct net_device *dev = priv->dev;
 	int done;
 	int t;
@@ -464,8 +471,9 @@ poll_more:
 					ipoib_cm_handle_rx_wc(dev, wc);
 				else
 					ipoib_ib_handle_rx_wc(dev, wc);
-			} else
-				ipoib_cm_handle_tx_wc(priv->dev, wc);
+			} else {
+				pr_warn("%s: Got unexpected wqe id\n", __func__);
+			}
 		}
 
 		if (n != t)
@@ -484,33 +492,47 @@ poll_more:
 	return done;
 }
 
-void ipoib_ib_completion(struct ib_cq *cq, void *dev_ptr)
+int ipoib_tx_poll(struct napi_struct *napi, int budget)
 {
-	struct net_device *dev = dev_ptr;
-	struct ipoib_dev_priv *priv = ipoib_priv(dev);
+	struct ipoib_dev_priv *priv = container_of(napi, struct ipoib_dev_priv,
+						   send_napi);
+	struct net_device *dev = priv->dev;
+	int n, i;
+	struct ib_wc *wc;
 
-	napi_schedule(&priv->napi);
-}
+poll_more:
+	n = ib_poll_cq(priv->send_cq, MAX_SEND_CQE, priv->send_wc);
 
-static void drain_tx_cq(struct net_device *dev)
-{
-	struct ipoib_dev_priv *priv = ipoib_priv(dev);
+	for (i = 0; i < n; i++) {
+		wc = priv->send_wc + i;
+		if (wc->wr_id & IPOIB_OP_CM)
+			ipoib_cm_handle_tx_wc(dev, wc);
+		else
+			ipoib_ib_handle_tx_wc(dev, wc);
+	}
 
-	netif_tx_lock(dev);
-	while (poll_tx(priv))
-		; /* nothing */
+	if (n < budget) {
+		napi_complete(napi);
+		if (unlikely(ib_req_notify_cq(priv->send_cq, IB_CQ_NEXT_COMP |
+					      IB_CQ_REPORT_MISSED_EVENTS)) &&
+		    napi_reschedule(napi))
+			goto poll_more;
+	}
+	return n < 0 ? 0 : n;
+}
 
-	if (netif_queue_stopped(dev))
-		mod_timer(&priv->poll_timer, jiffies + 1);
+void ipoib_ib_rx_completion(struct ib_cq *cq, void *ctx_ptr)
+{
+	struct ipoib_dev_priv *priv = ctx_ptr;
 
-	netif_tx_unlock(dev);
+	napi_schedule(&priv->recv_napi);
 }
 
-void ipoib_send_comp_handler(struct ib_cq *cq, void *dev_ptr)
+void ipoib_ib_tx_completion(struct ib_cq *cq, void *ctx_ptr)
 {
-	struct ipoib_dev_priv *priv = ipoib_priv(dev_ptr);
+	struct ipoib_dev_priv *priv = ctx_ptr;
 
-	mod_timer(&priv->poll_timer, jiffies);
+	napi_schedule(&priv->send_napi);
 }
 
 static inline int post_send(struct ipoib_dev_priv *priv,
@@ -614,14 +636,17 @@ int ipoib_send(struct net_device *dev, struct sk_buff *skb,
 	/* increase the tx_head after send success, but use it for queue state */
 	if (priv->tx_head - priv->tx_tail == ipoib_sendq_size - 1) {
 		ipoib_dbg(priv, "TX ring full, stopping kernel net queue\n");
-		if (ib_req_notify_cq(priv->send_cq, IB_CQ_NEXT_COMP))
-			ipoib_warn(priv, "request notify on send CQ failed\n");
 		netif_stop_queue(dev);
 	}
 
 	skb_orphan(skb);
 	skb_dst_drop(skb);
 
+	if (netif_queue_stopped(dev))
+		if (ib_req_notify_cq(priv->send_cq, IB_CQ_NEXT_COMP |
+				     IB_CQ_REPORT_MISSED_EVENTS))
+			ipoib_warn(priv, "request notify on send CQ failed\n");
+
 	rc = post_send(priv, priv->tx_head & (ipoib_sendq_size - 1),
 		       address, dqpn, tx_req, phead, hlen);
 	if (unlikely(rc)) {
@@ -638,11 +663,6 @@ int ipoib_send(struct net_device *dev, struct sk_buff *skb,
 		rc = priv->tx_head;
 		++priv->tx_head;
 	}
-
-	if (unlikely(priv->tx_head - priv->tx_tail > MAX_SEND_CQE))
-		while (poll_tx(priv))
-			; /* nothing */
-
 	return rc;
 }
 
@@ -731,6 +751,22 @@ static void check_qp_movement_and_print(struct ipoib_dev_priv *priv,
 			   new_state, qp_attr.qp_state);
 }
 
+static void ipoib_napi_enable(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = ipoib_priv(dev);
+
+	napi_enable(&priv->recv_napi);
+	napi_enable(&priv->send_napi);
+}
+
+static void ipoib_napi_disable(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = ipoib_priv(dev);
+
+	napi_disable(&priv->recv_napi);
+	napi_disable(&priv->send_napi);
+}
+
 int ipoib_ib_dev_stop_default(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = ipoib_priv(dev);
@@ -740,7 +776,7 @@ int ipoib_ib_dev_stop_default(struct net_device *dev)
 	int i;
 
 	if (test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
-		napi_disable(&priv->napi);
+		ipoib_napi_disable(dev);
 
 	ipoib_cm_dev_stop(dev);
 
@@ -797,7 +833,6 @@ int ipoib_ib_dev_stop_default(struct net_device *dev)
 	ipoib_dbg(priv, "All sends and receives done.\n");
 
 timeout:
-	del_timer_sync(&priv->poll_timer);
 	qp_attr.qp_state = IB_QPS_RESET;
 	if (ib_modify_qp(priv->qp, &qp_attr, IB_QP_STATE))
 		ipoib_warn(priv, "Failed to modify QP to RESET state\n");
@@ -819,13 +854,6 @@ int ipoib_ib_dev_stop(struct net_device *dev)
 	return 0;
 }
 
-void ipoib_ib_tx_timer_func(struct timer_list *t)
-{
-	struct ipoib_dev_priv *priv = from_timer(priv, t, poll_timer);
-
-	drain_tx_cq(priv->dev);
-}
-
 int ipoib_ib_dev_open_default(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = ipoib_priv(dev);
@@ -850,7 +878,7 @@ int ipoib_ib_dev_open_default(struct net_device *dev)
 	}
 
 	if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
-		napi_enable(&priv->napi);
+		ipoib_napi_enable(dev);
 
 	return 0;
 out:
@@ -965,8 +993,9 @@ void ipoib_drain_cq(struct net_device *dev)
 					ipoib_cm_handle_rx_wc(dev, priv->ibwc + i);
 				else
 					ipoib_ib_handle_rx_wc(dev, priv->ibwc + i);
-			} else
-				ipoib_cm_handle_tx_wc(dev, priv->ibwc + i);
+			} else {
+				pr_warn("%s: Got unexpected wqe id\n", __func__);
+			}
 		}
 	} while (n == IPOIB_NUM_WC);