vxge: fix pktgen hangs (don't abuse skb->cb[])

This patch fixes a case in the transmit completion code which was resulting
in pktgen hanging at the end of a run.  The cause is due to the fact that
the ->cb[] area of an skb cannot be used in a network driver's transmit
path, as that area belongs to the network protocol.  Pktgen hangs, as it
sends out the same packet multiple times, and vxge's use of this area of
the skb for a temporary list can only add the packet to the temporary list
once (while it may be on the queue many times).  The fix is to remove this
abuse of skb->cb[].  Instead, skb pointers are placed into a temporary
stack array, and then free outside of the tx lock.  This retains the smp
optimization of doing dev_kfree_skb() outside of the tx lock.

Signed-off-by: Benjamin LaHaise <ben.lahaise@neterion.com>
Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: Ramkrishna Vepa <ram.vepa@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/drivers/net/vxge/vxge-main.c b/drivers/net/vxge/vxge-main.c
index beee4ab..4b22513 100644
--- a/drivers/net/vxge/vxge-main.c
+++ b/drivers/net/vxge/vxge-main.c
@@ -87,22 +87,25 @@
 static inline void VXGE_COMPLETE_VPATH_TX(struct vxge_fifo *fifo)
 {
 	unsigned long flags = 0;
-	struct sk_buff *skb_ptr = NULL;
-	struct sk_buff **temp, *head, *skb;
+	struct sk_buff **skb_ptr = NULL;
+	struct sk_buff **temp;
+#define NR_SKB_COMPLETED 128
+	struct sk_buff *completed[NR_SKB_COMPLETED];
+	int more;
 
-	if (spin_trylock_irqsave(&fifo->tx_lock, flags)) {
-		vxge_hw_vpath_poll_tx(fifo->handle, (void **)&skb_ptr);
-		spin_unlock_irqrestore(&fifo->tx_lock, flags);
-	}
-	/* free SKBs */
-	head = skb_ptr;
-	while (head) {
-		skb = head;
-		temp = (struct sk_buff **)&skb->cb;
-		head = *temp;
-		*temp = NULL;
-		dev_kfree_skb_irq(skb);
-	}
+	do {
+		more = 0;
+		skb_ptr = completed;
+
+		if (spin_trylock_irqsave(&fifo->tx_lock, flags)) {
+			vxge_hw_vpath_poll_tx(fifo->handle, &skb_ptr,
+						NR_SKB_COMPLETED, &more);
+			spin_unlock_irqrestore(&fifo->tx_lock, flags);
+		}
+		/* free SKBs */
+		for (temp = completed; temp != skb_ptr; temp++)
+			dev_kfree_skb_irq(*temp);
+	} while (more) ;
 }
 
 static inline void VXGE_COMPLETE_ALL_TX(struct vxgedev *vdev)
@@ -600,11 +603,10 @@
 enum vxge_hw_status
 vxge_xmit_compl(struct __vxge_hw_fifo *fifo_hw, void *dtr,
 		enum vxge_hw_fifo_tcode t_code, void *userdata,
-		void **skb_ptr)
+		struct sk_buff ***skb_ptr, int nr_skb, int *more)
 {
 	struct vxge_fifo *fifo = (struct vxge_fifo *)userdata;
-	struct sk_buff *skb, *head = NULL;
-	struct sk_buff **temp;
+	struct sk_buff *skb, **done_skb = *skb_ptr;
 	int pkt_cnt = 0;
 
 	vxge_debug_entryexit(VXGE_TRACE,
@@ -657,9 +659,12 @@
 		fifo->stats.tx_frms++;
 		fifo->stats.tx_bytes += skb->len;
 
-		temp = (struct sk_buff **)&skb->cb;
-		*temp = head;
-		head = skb;
+		*done_skb++ = skb;
+
+		if (--nr_skb <= 0) {
+			*more = 1;
+			break;
+		}
 
 		pkt_cnt++;
 		if (pkt_cnt > fifo->indicate_max_pkts)
@@ -668,11 +673,9 @@
 	} while (vxge_hw_fifo_txdl_next_completed(fifo_hw,
 				&dtr, &t_code) == VXGE_HW_OK);
 
+	*skb_ptr = done_skb;
 	vxge_wake_tx_queue(fifo, skb);
 
-	if (skb_ptr)
-		*skb_ptr = (void *) head;
-
 	vxge_debug_entryexit(VXGE_TRACE,
 				"%s: %s:%d  Exiting...",
 				fifo->ndev->name, __func__, __LINE__);