1 files changed, 34 insertions, 119 deletions

diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
index 7dc2d64db3cb..997cf0901ac2 100644
--- a/drivers/net/xen-netback/netback.c
+++ b/drivers/net/xen-netback/netback.c
@@ -96,6 +96,7 @@ static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
 static void make_tx_response(struct xenvif_queue *queue,
 			     struct xen_netif_tx_request *txp,
 			     s8       st);
+static void push_tx_responses(struct xenvif_queue *queue);
 
 static inline int tx_work_todo(struct xenvif_queue *queue);
 
@@ -233,51 +234,6 @@ static void xenvif_rx_queue_drop_expired(struct xenvif_queue *queue)
 	}
 }
 
-/*
- * Returns true if we should start a new receive buffer instead of
- * adding 'size' bytes to a buffer which currently contains 'offset'
- * bytes.
- */
-static bool start_new_rx_buffer(int offset, unsigned long size, int head,
-				bool full_coalesce)
-{
-	/* simple case: we have completely filled the current buffer. */
-	if (offset == MAX_BUFFER_OFFSET)
-		return true;
-
-	/*
-	 * complex case: start a fresh buffer if the current frag
-	 * would overflow the current buffer but only if:
-	 *     (i)   this frag would fit completely in the next buffer
-	 * and (ii)  there is already some data in the current buffer
-	 * and (iii) this is not the head buffer.
-	 * and (iv)  there is no need to fully utilize the buffers
-	 *
-	 * Where:
-	 * - (i) stops us splitting a frag into two copies
-	 *   unless the frag is too large for a single buffer.
-	 * - (ii) stops us from leaving a buffer pointlessly empty.
-	 * - (iii) stops us leaving the first buffer
-	 *   empty. Strictly speaking this is already covered
-	 *   by (ii) but is explicitly checked because
-	 *   netfront relies on the first buffer being
-	 *   non-empty and can crash otherwise.
-	 * - (iv) is needed for skbs which can use up more than MAX_SKB_FRAGS
-	 *   slot
-	 *
-	 * This means we will effectively linearise small
-	 * frags but do not needlessly split large buffers
-	 * into multiple copies tend to give large frags their
-	 * own buffers as before.
-	 */
-	BUG_ON(size > MAX_BUFFER_OFFSET);
-	if ((offset + size > MAX_BUFFER_OFFSET) && offset && !head &&
-	    !full_coalesce)
-		return true;
-
-	return false;
-}
-
 struct netrx_pending_operations {
 	unsigned copy_prod, copy_cons;
 	unsigned meta_prod, meta_cons;
@@ -336,24 +292,13 @@ static void xenvif_gop_frag_copy(struct xenvif_queue *queue, struct sk_buff *skb
 		BUG_ON(offset >= PAGE_SIZE);
 		BUG_ON(npo->copy_off > MAX_BUFFER_OFFSET);
 
-		bytes = PAGE_SIZE - offset;
+		if (npo->copy_off == MAX_BUFFER_OFFSET)
+			meta = get_next_rx_buffer(queue, npo);
 
+		bytes = PAGE_SIZE - offset;
 		if (bytes > size)
 			bytes = size;
 
-		if (start_new_rx_buffer(npo->copy_off,
-					bytes,
-					*head,
-					XENVIF_RX_CB(skb)->full_coalesce)) {
-			/*
-			 * Netfront requires there to be some data in the head
-			 * buffer.
-			 */
-			BUG_ON(*head);
-
-			meta = get_next_rx_buffer(queue, npo);
-		}
-
 		if (npo->copy_off + bytes > MAX_BUFFER_OFFSET)
 			bytes = MAX_BUFFER_OFFSET - npo->copy_off;
 
@@ -570,60 +515,15 @@ static void xenvif_rx_action(struct xenvif_queue *queue)
 
 	while (xenvif_rx_ring_slots_available(queue, XEN_NETBK_RX_SLOTS_MAX)
 	       && (skb = xenvif_rx_dequeue(queue)) != NULL) {
-		RING_IDX max_slots_needed;
 		RING_IDX old_req_cons;
 		RING_IDX ring_slots_used;
-		int i;
 
 		queue->last_rx_time = jiffies;
 
-		/* We need a cheap worse case estimate for the number of
-		 * slots we'll use.
-		 */
-
-		max_slots_needed = DIV_ROUND_UP(offset_in_page(skb->data) +
-						skb_headlen(skb),
-						PAGE_SIZE);
-		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
-			unsigned int size;
-			unsigned int offset;
-
-			size = skb_frag_size(&skb_shinfo(skb)->frags[i]);
-			offset = skb_shinfo(skb)->frags[i].page_offset;
-
-			/* For a worse-case estimate we need to factor in
-			 * the fragment page offset as this will affect the
-			 * number of times xenvif_gop_frag_copy() will
-			 * call start_new_rx_buffer().
-			 */
-			max_slots_needed += DIV_ROUND_UP(offset + size,
-							 PAGE_SIZE);
-		}
-
-		/* To avoid the estimate becoming too pessimal for some
-		 * frontends that limit posted rx requests, cap the estimate
-		 * at MAX_SKB_FRAGS. In this case netback will fully coalesce
-		 * the skb into the provided slots.
-		 */
-		if (max_slots_needed > MAX_SKB_FRAGS) {
-			max_slots_needed = MAX_SKB_FRAGS;
-			XENVIF_RX_CB(skb)->full_coalesce = true;
-		} else {
-			XENVIF_RX_CB(skb)->full_coalesce = false;
-		}
-
-		/* We may need one more slot for GSO metadata */
-		if (skb_is_gso(skb) &&
-		   (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4 ||
-		    skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6))
-			max_slots_needed++;
-
 		old_req_cons = queue->rx.req_cons;
 		XENVIF_RX_CB(skb)->meta_slots_used = xenvif_gop_skb(skb, &npo, queue);
 		ring_slots_used = queue->rx.req_cons - old_req_cons;
 
-		BUG_ON(ring_slots_used > max_slots_needed);
-
 		__skb_queue_tail(&rxq, skb);
 	}
 
@@ -758,6 +658,7 @@ static void xenvif_tx_err(struct xenvif_queue *queue,
 	do {
 		spin_lock_irqsave(&queue->response_lock, flags);
 		make_tx_response(queue, txp, XEN_NETIF_RSP_ERROR);
+		push_tx_responses(queue);
 		spin_unlock_irqrestore(&queue->response_lock, flags);
 		if (cons == end)
 			break;
@@ -1444,7 +1345,7 @@ static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *s
 {
 	unsigned int offset = skb_headlen(skb);
 	skb_frag_t frags[MAX_SKB_FRAGS];
-	int i;
+	int i, f;
 	struct ubuf_info *uarg;
 	struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
 
@@ -1484,23 +1385,25 @@ static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *s
 		frags[i].page_offset = 0;
 		skb_frag_size_set(&frags[i], len);
 	}
-	/* swap out with old one */
-	memcpy(skb_shinfo(skb)->frags,
-	       frags,
-	       i * sizeof(skb_frag_t));
-	skb_shinfo(skb)->nr_frags = i;
-	skb->truesize += i * PAGE_SIZE;
 
-	/* remove traces of mapped pages and frag_list */
+	/* Copied all the bits from the frag list -- free it. */
 	skb_frag_list_init(skb);
+	xenvif_skb_zerocopy_prepare(queue, nskb);
+	kfree_skb(nskb);
+
+	/* Release all the original (foreign) frags. */
+	for (f = 0; f < skb_shinfo(skb)->nr_frags; f++)
+		skb_frag_unref(skb, f);
 	uarg = skb_shinfo(skb)->destructor_arg;
 	/* increase inflight counter to offset decrement in callback */
 	atomic_inc(&queue->inflight_packets);
 	uarg->callback(uarg, true);
 	skb_shinfo(skb)->destructor_arg = NULL;
 
-	xenvif_skb_zerocopy_prepare(queue, nskb);
-	kfree_skb(nskb);
+	/* Fill the skb with the new (local) frags. */
+	memcpy(skb_shinfo(skb)->frags, frags, i * sizeof(skb_frag_t));
+	skb_shinfo(skb)->nr_frags = i;
+	skb->truesize += i * PAGE_SIZE;
 
 	return 0;
 }
@@ -1753,13 +1656,20 @@ static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
 	unsigned long flags;
 
 	pending_tx_info = &queue->pending_tx_info[pending_idx];
+
 	spin_lock_irqsave(&queue->response_lock, flags);
+
 	make_tx_response(queue, &pending_tx_info->req, status);
-	index = pending_index(queue->pending_prod);
+
+	/* Release the pending index before pusing the Tx response so
+	 * its available before a new Tx request is pushed by the
+	 * frontend.
+	 */
+	index = pending_index(queue->pending_prod++);
 	queue->pending_ring[index] = pending_idx;
-	/* TX shouldn't use the index before we give it back here */
-	mb();
-	queue->pending_prod++;
+
+	push_tx_responses(queue);
+
 	spin_unlock_irqrestore(&queue->response_lock, flags);
 }
 
@@ -1770,7 +1680,6 @@ static void make_tx_response(struct xenvif_queue *queue,
 {
 	RING_IDX i = queue->tx.rsp_prod_pvt;
 	struct xen_netif_tx_response *resp;
-	int notify;
 
 	resp = RING_GET_RESPONSE(&queue->tx, i);
 	resp->id     = txp->id;
@@ -1780,6 +1689,12 @@ static void make_tx_response(struct xenvif_queue *queue,
 		RING_GET_RESPONSE(&queue->tx, ++i)->status = XEN_NETIF_RSP_NULL;
 
 	queue->tx.rsp_prod_pvt = ++i;
+}
+
+static void push_tx_responses(struct xenvif_queue *queue)
+{
+	int notify;
+
 	RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&queue->tx, notify);
 	if (notify)
 		notify_remote_via_irq(queue->tx_irq);