Merge branch 'net-more-bulk-free-users'

Jesper Dangaard Brouer says:

====================
net: bulk free adjustment and two driver use-cases

I've split out the bulk free adjustments, from the bulk alloc patches,
as I want the adjustment to napi_consume_skb be in same kernel cycle
the API was introduced.

Adjustments based on discussion:
 Subj: "mlx4: use napi_consume_skb API to get bulk free operations"
 http://thread.gmane.org/gmane.linux.network/402503/focus=403386

Patchset based on net-next at commit 3ebeac1d

V4: more nitpicks from Sergei
V3: spelling fixes from Sergei
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlx4/en_tx.c

@@ -276,7 +276,8 @@ static void mlx4_en_stamp_wqe(struct mlx4_en_priv *priv,
 
 static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 				struct mlx4_en_tx_ring *ring,
-				int index, u8 owner, u64 timestamp)
+				int index, u8 owner, u64 timestamp,
+				int napi_mode)
 {
 	struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
 	struct mlx4_en_tx_desc *tx_desc = ring->buf + index * TXBB_SIZE;
@@ -347,7 +348,8 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 			}
 		}
 	}
-	dev_consume_skb_any(skb);
+	napi_consume_skb(skb, napi_mode);
+
 	return tx_info->nr_txbb;
 }
 
@@ -371,7 +373,8 @@ int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring)
 	while (ring->cons != ring->prod) {
 		ring->last_nr_txbb = mlx4_en_free_tx_desc(priv, ring,
 						ring->cons & ring->size_mask,
-						!!(ring->cons & ring->size), 0);
+						!!(ring->cons & ring->size), 0,
+						0 /* Non-NAPI caller */);
 		ring->cons += ring->last_nr_txbb;
 		cnt++;
 	}
@@ -385,7 +388,7 @@ int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring)
 }
 
 static bool mlx4_en_process_tx_cq(struct net_device *dev,
-				 struct mlx4_en_cq *cq)
+				  struct mlx4_en_cq *cq, int napi_budget)
 {
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	struct mlx4_cq *mcq = &cq->mcq;
@@ -451,7 +454,7 @@ static bool mlx4_en_process_tx_cq(struct net_device *dev,
 			last_nr_txbb = mlx4_en_free_tx_desc(
 					priv, ring, ring_index,
 					!!((ring_cons + txbbs_skipped) &
-					ring->size), timestamp);
+					ring->size), timestamp, napi_budget);
 
 			mlx4_en_stamp_wqe(priv, ring, stamp_index,
 					  !!((ring_cons + txbbs_stamp) &
@@ -511,7 +514,7 @@ int mlx4_en_poll_tx_cq(struct napi_struct *napi, int budget)
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	int clean_complete;
 
-	clean_complete = mlx4_en_process_tx_cq(dev, cq);
+	clean_complete = mlx4_en_process_tx_cq(dev, cq, budget);
 	if (!clean_complete)
 		return budget;
 

drivers/net/ethernet/mellanox/mlx5/core/en.h

@@ -629,7 +629,7 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev);
 void mlx5e_completion_event(struct mlx5_core_cq *mcq);
 void mlx5e_cq_error_event(struct mlx5_core_cq *mcq, enum mlx5_event event);
 int mlx5e_napi_poll(struct napi_struct *napi, int budget);
-bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq);
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget);
 int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget);
 bool mlx5e_post_rx_wqes(struct mlx5e_rq *rq);
 struct mlx5_cqe64 *mlx5e_get_cqe(struct mlx5e_cq *cq);

drivers/net/ethernet/mellanox/mlx5/core/en_tx.c

@@ -339,7 +339,7 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
 	return mlx5e_sq_xmit(sq, skb);
 }
 
-bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq)
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget)
 {
 	struct mlx5e_sq *sq;
 	u32 dma_fifo_cc;
@@ -411,7 +411,7 @@ bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq)
 			npkts++;
 			nbytes += wi->num_bytes;
 			sqcc += wi->num_wqebbs;
-			dev_kfree_skb(skb);
+			napi_consume_skb(skb, napi_budget);
 		} while (!last_wqe);
 	}
 

drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c

@@ -60,7 +60,7 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
 	clear_bit(MLX5E_CHANNEL_NAPI_SCHED, &c->flags);
 
 	for (i = 0; i < c->num_tc; i++)
-		busy |= mlx5e_poll_tx_cq(&c->sq[i].cq);
+		busy |= mlx5e_poll_tx_cq(&c->sq[i].cq, budget);
 
 	work_done = mlx5e_poll_rx_cq(&c->rq.cq, budget);
 	busy |= work_done == budget;

net/core/skbuff.c

@@ -801,9 +801,9 @@ void napi_consume_skb(struct sk_buff *skb, int budget)
 	if (unlikely(!skb))
 		return;
 
-	/* if budget is 0 assume netpoll w/ IRQs disabled */
+	/* Zero budget indicate non-NAPI context called us, like netpoll */
 	if (unlikely(!budget)) {
-		dev_consume_skb_irq(skb);
+		dev_consume_skb_any(skb);
 		return;
 	}