1 files changed, 67 insertions, 23 deletions

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
index 7cf1fa7074ab..f429bce24c20 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
@@ -364,7 +364,6 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
         struct ipoib_dev_priv *priv = netdev_priv(dev);
         unsigned int wr_id = wc->wr_id;
         struct ipoib_tx_buf *tx_req;
-        unsigned long flags;
 
         ipoib_dbg_data(priv, "send completion: id %d, status: %d\n",
                        wr_id, wc->status);
@@ -384,13 +383,11 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
 
         dev_kfree_skb_any(tx_req->skb);
 
-        spin_lock_irqsave(&priv->tx_lock, flags);
         ++priv->tx_tail;
         if (unlikely(--priv->tx_outstanding == ipoib_sendq_size >> 1) &&
             netif_queue_stopped(dev) &&
             test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
                 netif_wake_queue(dev);
-        spin_unlock_irqrestore(&priv->tx_lock, flags);
 
         if (wc->status != IB_WC_SUCCESS &&
             wc->status != IB_WC_WR_FLUSH_ERR)
@@ -399,6 +396,17 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
                            wc->status, wr_id, wc->vendor_err);
 }
 
+static int poll_tx(struct ipoib_dev_priv *priv)
+{
+        int n, i;
+
+        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);
+
+        return n == MAX_SEND_CQE;
+}
+
 int ipoib_poll(struct napi_struct *napi, int budget)
 {
         struct ipoib_dev_priv *priv = container_of(napi, struct ipoib_dev_priv, napi);
@@ -414,7 +422,7 @@ poll_more:
                 int max = (budget - done);
 
                 t = min(IPOIB_NUM_WC, max);
-                n = ib_poll_cq(priv->cq, t, priv->ibwc);
+                n = ib_poll_cq(priv->recv_cq, t, priv->ibwc);
 
                 for (i = 0; i < n; i++) {
                         struct ib_wc *wc = priv->ibwc + i;
@@ -425,12 +433,8 @@ poll_more:
                                         ipoib_cm_handle_rx_wc(dev, wc);
                                 else
                                         ipoib_ib_handle_rx_wc(dev, wc);
-                        } else {
-                                if (wc->wr_id & IPOIB_OP_CM)
-                                        ipoib_cm_handle_tx_wc(dev, wc);
-                                else
-                                        ipoib_ib_handle_tx_wc(dev, wc);
-                        }
+                        } else
+                                ipoib_cm_handle_tx_wc(priv->dev, wc);
                 }
 
                 if (n != t)
@@ -439,7 +443,7 @@ poll_more:
 
         if (done < budget) {
                 netif_rx_complete(dev, napi);
-                if (unlikely(ib_req_notify_cq(priv->cq,
+                if (unlikely(ib_req_notify_cq(priv->recv_cq,
                                               IB_CQ_NEXT_COMP |
                                               IB_CQ_REPORT_MISSED_EVENTS)) &&
                     netif_rx_reschedule(dev, napi))
@@ -457,6 +461,26 @@ void ipoib_ib_completion(struct ib_cq *cq, void *dev_ptr)
         netif_rx_schedule(dev, &priv->napi);
 }
 
+static void drain_tx_cq(struct net_device *dev)
+{
+        struct ipoib_dev_priv *priv = netdev_priv(dev);
+        unsigned long flags;
+
+        spin_lock_irqsave(&priv->tx_lock, flags);
+        while (poll_tx(priv))
+                ; /* nothing */
+
+        if (netif_queue_stopped(dev))
+                mod_timer(&priv->poll_timer, jiffies + 1);
+
+        spin_unlock_irqrestore(&priv->tx_lock, flags);
+}
+
+void ipoib_send_comp_handler(struct ib_cq *cq, void *dev_ptr)
+{
+        drain_tx_cq((struct net_device *)dev_ptr);
+}
+
 static inline int post_send(struct ipoib_dev_priv *priv,
                             unsigned int wr_id,
                             struct ib_ah *address, u32 qpn,
@@ -551,23 +575,34 @@ void ipoib_send(struct net_device *dev, struct sk_buff *skb,
         else
                 priv->tx_wr.send_flags &= ~IB_SEND_IP_CSUM;
 
+        if (++priv->tx_outstanding == ipoib_sendq_size) {
+                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);
+        }
+
         if (unlikely(post_send(priv, priv->tx_head & (ipoib_sendq_size - 1),
                                address->ah, qpn, tx_req, phead, hlen))) {
                 ipoib_warn(priv, "post_send failed\n");
                 ++dev->stats.tx_errors;
+                --priv->tx_outstanding;
                 ipoib_dma_unmap_tx(priv->ca, tx_req);
                 dev_kfree_skb_any(skb);
+                if (netif_queue_stopped(dev))
+                        netif_wake_queue(dev);
         } else {
                 dev->trans_start = jiffies;
 
                 address->last_send = priv->tx_head;
                 ++priv->tx_head;
+                skb_orphan(skb);
 
-                if (++priv->tx_outstanding == ipoib_sendq_size) {
-                        ipoib_dbg(priv, "TX ring full, stopping kernel net queue\n");
-                        netif_stop_queue(dev);
-                }
         }
+
+        if (unlikely(priv->tx_outstanding > MAX_SEND_CQE))
+                while (poll_tx(priv))
+                        ; /* nothing */
 }
 
 static void __ipoib_reap_ah(struct net_device *dev)
@@ -601,6 +636,11 @@ void ipoib_reap_ah(struct work_struct *work)
                                    round_jiffies_relative(HZ));
 }
 
+static void ipoib_ib_tx_timer_func(unsigned long ctx)
+{
+        drain_tx_cq((struct net_device *)ctx);
+}
+
 int ipoib_ib_dev_open(struct net_device *dev)
 {
         struct ipoib_dev_priv *priv = netdev_priv(dev);
@@ -637,6 +677,10 @@ int ipoib_ib_dev_open(struct net_device *dev)
         queue_delayed_work(ipoib_workqueue, &priv->ah_reap_task,
                            round_jiffies_relative(HZ));
 
+        init_timer(&priv->poll_timer);
+        priv->poll_timer.function = ipoib_ib_tx_timer_func;
+        priv->poll_timer.data = (unsigned long)dev;
+
         set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags);
 
         return 0;
@@ -714,7 +758,7 @@ void ipoib_drain_cq(struct net_device *dev)
         struct ipoib_dev_priv *priv = netdev_priv(dev);
         int i, n;
         do {
-                n = ib_poll_cq(priv->cq, IPOIB_NUM_WC, priv->ibwc);
+                n = ib_poll_cq(priv->recv_cq, IPOIB_NUM_WC, priv->ibwc);
                 for (i = 0; i < n; ++i) {
                         /*
                          * Convert any successful completions to flush
@@ -729,14 +773,13 @@ 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 {
-                                if (priv->ibwc[i].wr_id & IPOIB_OP_CM)
-                                        ipoib_cm_handle_tx_wc(dev, priv->ibwc + i);
-                                else
-                                        ipoib_ib_handle_tx_wc(dev, priv->ibwc + i);
-                        }
+                        } else
+                                ipoib_cm_handle_tx_wc(dev, priv->ibwc + i);
                 }
         } while (n == IPOIB_NUM_WC);
+
+        while (poll_tx(priv))
+                ; /* nothing */
 }
 
 int ipoib_ib_dev_stop(struct net_device *dev, int flush)
@@ -803,6 +846,7 @@ int ipoib_ib_dev_stop(struct net_device *dev, int flush)
         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");
@@ -826,7 +870,7 @@ timeout:
                 msleep(1);
         }
 
-        ib_req_notify_cq(priv->cq, IB_CQ_NEXT_COMP);
+        ib_req_notify_cq(priv->recv_cq, IB_CQ_NEXT_COMP);
 
         return 0;
 }