IB/ipoib: move back IB LL address into the hard header

After the commit 9207f9d45b0a ("net: preserve IP control block
during GSO segmentation"), the GSO CB and the IPoIB CB conflict.
That destroy the IPoIB address information cached there,
causing a severe performance regression, as better described here:

http://marc.info/?l=linux-kernel&m=146787279825501&w=2

This change moves the data cached by the IPoIB driver from the
skb control lock into the IPoIB hard header, as done before
the commit 936d7de3d736 ("IPoIB: Stop lying about hard_header_len
and use skb->cb to stash LL addresses").
In order to avoid GRO issue, on packet reception, the IPoIB driver
stash into the skb a dummy pseudo header, so that the received
packets have actually a hard header matching the declared length.
To avoid changing the connected mode maximum mtu, the allocated
head buffer size is increased by the pseudo header length.

After this commit, IPoIB performances are back to pre-regression
value.

v2 -> v3: rebased
v1 -> v2: avoid changing the max mtu, increasing the head buf size

Fixes: 9207f9d45b0a ("net: preserve IP control block during GSO segmentation")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 33 insertions, 21 deletions

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 5636fc3da6b8..b58d9dca5c93 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -925,9 +925,12 @@ static void neigh_add_path(struct sk_buff *skb, u8 *daddr,
 				ipoib_neigh_free(neigh);
 				goto err_drop;
 			}
-			if (skb_queue_len(&neigh->queue) < IPOIB_MAX_PATH_REC_QUEUE)
+			if (skb_queue_len(&neigh->queue) <
+			    IPOIB_MAX_PATH_REC_QUEUE) {
+				/* put pseudoheader back on for next time */
+				skb_push(skb, IPOIB_PSEUDO_LEN);
 				__skb_queue_tail(&neigh->queue, skb);
-			else {
+			} else {
 				ipoib_warn(priv, "queue length limit %d. Packet drop.\n",
 					   skb_queue_len(&neigh->queue));
 				goto err_drop;
@@ -964,7 +967,7 @@ err_drop:
 }
 
 static void unicast_arp_send(struct sk_buff *skb, struct net_device *dev,
-			     struct ipoib_cb *cb)
+			     struct ipoib_pseudo_header *phdr)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
 	struct ipoib_path *path;
@@ -972,16 +975,18 @@ static void unicast_arp_send(struct sk_buff *skb, struct net_device *dev,
 
 	spin_lock_irqsave(&priv->lock, flags);
 
-	path = __path_find(dev, cb->hwaddr + 4);
+	path = __path_find(dev, phdr->hwaddr + 4);
 	if (!path || !path->valid) {
 		int new_path = 0;
 
 		if (!path) {
-			path = path_rec_create(dev, cb->hwaddr + 4);
+			path = path_rec_create(dev, phdr->hwaddr + 4);
 			new_path = 1;
 		}
 		if (path) {
 			if (skb_queue_len(&path->queue) < IPOIB_MAX_PATH_REC_QUEUE) {
+				/* put pseudoheader back on for next time */
+				skb_push(skb, IPOIB_PSEUDO_LEN);
 				__skb_queue_tail(&path->queue, skb);
 			} else {
 				++dev->stats.tx_dropped;
@@ -1009,10 +1014,12 @@ static void unicast_arp_send(struct sk_buff *skb, struct net_device *dev,
 			  be16_to_cpu(path->pathrec.dlid));
 
 		spin_unlock_irqrestore(&priv->lock, flags);
-		ipoib_send(dev, skb, path->ah, IPOIB_QPN(cb->hwaddr));
+		ipoib_send(dev, skb, path->ah, IPOIB_QPN(phdr->hwaddr));
 		return;
 	} else if ((path->query || !path_rec_start(dev, path)) &&
 		   skb_queue_len(&path->queue) < IPOIB_MAX_PATH_REC_QUEUE) {
+		/* put pseudoheader back on for next time */
+		skb_push(skb, IPOIB_PSEUDO_LEN);
 		__skb_queue_tail(&path->queue, skb);
 	} else {
 		++dev->stats.tx_dropped;
@@ -1026,13 +1033,15 @@ static int ipoib_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
 	struct ipoib_neigh *neigh;
-	struct ipoib_cb *cb = ipoib_skb_cb(skb);
+	struct ipoib_pseudo_header *phdr;
 	struct ipoib_header *header;
 	unsigned long flags;
 
+	phdr = (struct ipoib_pseudo_header *) skb->data;
+	skb_pull(skb, sizeof(*phdr));
 	header = (struct ipoib_header *) skb->data;
 
-	if (unlikely(cb->hwaddr[4] == 0xff)) {
+	if (unlikely(phdr->hwaddr[4] == 0xff)) {
 		/* multicast, arrange "if" according to probability */
 		if ((header->proto != htons(ETH_P_IP)) &&
 		    (header->proto != htons(ETH_P_IPV6)) &&
@@ -1045,13 +1054,13 @@ static int ipoib_start_xmit(struct sk_buff *skb, struct net_device *dev)
 			return NETDEV_TX_OK;
 		}
 		/* Add in the P_Key for multicast*/
-		cb->hwaddr[8] = (priv->pkey >> 8) & 0xff;
-		cb->hwaddr[9] = priv->pkey & 0xff;
+		phdr->hwaddr[8] = (priv->pkey >> 8) & 0xff;
+		phdr->hwaddr[9] = priv->pkey & 0xff;
 
-		neigh = ipoib_neigh_get(dev, cb->hwaddr);
+		neigh = ipoib_neigh_get(dev, phdr->hwaddr);
 		if (likely(neigh))
 			goto send_using_neigh;
-		ipoib_mcast_send(dev, cb->hwaddr, skb);
+		ipoib_mcast_send(dev, phdr->hwaddr, skb);
 		return NETDEV_TX_OK;
 	}
 
@@ -1060,16 +1069,16 @@ static int ipoib_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	case htons(ETH_P_IP):
 	case htons(ETH_P_IPV6):
 	case htons(ETH_P_TIPC):
-		neigh = ipoib_neigh_get(dev, cb->hwaddr);
+		neigh = ipoib_neigh_get(dev, phdr->hwaddr);
 		if (unlikely(!neigh)) {
-			neigh_add_path(skb, cb->hwaddr, dev);
+			neigh_add_path(skb, phdr->hwaddr, dev);
 			return NETDEV_TX_OK;
 		}
 		break;
 	case htons(ETH_P_ARP):
 	case htons(ETH_P_RARP):
 		/* for unicast ARP and RARP should always perform path find */
-		unicast_arp_send(skb, dev, cb);
+		unicast_arp_send(skb, dev, phdr);
 		return NETDEV_TX_OK;
 	default:
 		/* ethertype not supported by IPoIB */
@@ -1086,11 +1095,13 @@ send_using_neigh:
 			goto unref;
 		}
 	} else if (neigh->ah) {
-		ipoib_send(dev, skb, neigh->ah, IPOIB_QPN(cb->hwaddr));
+		ipoib_send(dev, skb, neigh->ah, IPOIB_QPN(phdr->hwaddr));
 		goto unref;
 	}
 
 	if (skb_queue_len(&neigh->queue) < IPOIB_MAX_PATH_REC_QUEUE) {
+		/* put pseudoheader back on for next time */
+		skb_push(skb, sizeof(*phdr));
 		spin_lock_irqsave(&priv->lock, flags);
 		__skb_queue_tail(&neigh->queue, skb);
 		spin_unlock_irqrestore(&priv->lock, flags);
@@ -1122,8 +1133,8 @@ static int ipoib_hard_header(struct sk_buff *skb,
 			     unsigned short type,
 			     const void *daddr, const void *saddr, unsigned len)
 {
+	struct ipoib_pseudo_header *phdr;
 	struct ipoib_header *header;
-	struct ipoib_cb *cb = ipoib_skb_cb(skb);
 
 	header = (struct ipoib_header *) skb_push(skb, sizeof *header);
 
@@ -1132,12 +1143,13 @@ static int ipoib_hard_header(struct sk_buff *skb,
 
 	/*
 	 * we don't rely on dst_entry structure,  always stuff the
-	 * destination address into skb->cb so we can figure out where
+	 * destination address into skb hard header so we can figure out where
 	 * to send the packet later.
 	 */
-	memcpy(cb->hwaddr, daddr, INFINIBAND_ALEN);
+	phdr = (struct ipoib_pseudo_header *) skb_push(skb, sizeof(*phdr));
+	memcpy(phdr->hwaddr, daddr, INFINIBAND_ALEN);
 
-	return sizeof *header;
+	return IPOIB_HARD_LEN;
 }
 
 static void ipoib_set_mcast_list(struct net_device *dev)
@@ -1759,7 +1771,7 @@ void ipoib_setup(struct net_device *dev)
 
 	dev->flags		|= IFF_BROADCAST | IFF_MULTICAST;
 
-	dev->hard_header_len	 = IPOIB_ENCAP_LEN;
+	dev->hard_header_len	 = IPOIB_HARD_LEN;
 	dev->addr_len		 = INFINIBAND_ALEN;
 	dev->type		 = ARPHRD_INFINIBAND;
 	dev->tx_queue_len	 = ipoib_sendq_size * 2;