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, 8 insertions, 7 deletions

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
index 4ad297d3de89..339a1eecdfe3 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
@@ -63,6 +63,8 @@ MODULE_PARM_DESC(cm_data_debug_level,
 #define IPOIB_CM_RX_DELAY       (3 * 256 * HZ)
 #define IPOIB_CM_RX_UPDATE_MASK (0x3)
 
+#define IPOIB_CM_RX_RESERVE     (ALIGN(IPOIB_HARD_LEN, 16) - IPOIB_ENCAP_LEN)
+
 static struct ib_qp_attr ipoib_cm_err_attr = {
 	.qp_state = IB_QPS_ERR
 };
@@ -146,15 +148,15 @@ static struct sk_buff *ipoib_cm_alloc_rx_skb(struct net_device *dev,
 	struct sk_buff *skb;
 	int i;
 
-	skb = dev_alloc_skb(IPOIB_CM_HEAD_SIZE + 12);
+	skb = dev_alloc_skb(ALIGN(IPOIB_CM_HEAD_SIZE + IPOIB_PSEUDO_LEN, 16));
 	if (unlikely(!skb))
 		return NULL;
 
 	/*
-	 * IPoIB adds a 4 byte header. So we need 12 more bytes to align the
+	 * IPoIB adds a IPOIB_ENCAP_LEN byte header, this will align the
 	 * IP header to a multiple of 16.
 	 */
-	skb_reserve(skb, 12);
+	skb_reserve(skb, IPOIB_CM_RX_RESERVE);
 
 	mapping[0] = ib_dma_map_single(priv->ca, skb->data, IPOIB_CM_HEAD_SIZE,
 				       DMA_FROM_DEVICE);
@@ -624,9 +626,9 @@ void ipoib_cm_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
 	if (wc->byte_len < IPOIB_CM_COPYBREAK) {
 		int dlen = wc->byte_len;
 
-		small_skb = dev_alloc_skb(dlen + 12);
+		small_skb = dev_alloc_skb(dlen + IPOIB_CM_RX_RESERVE);
 		if (small_skb) {
-			skb_reserve(small_skb, 12);
+			skb_reserve(small_skb, IPOIB_CM_RX_RESERVE);
 			ib_dma_sync_single_for_cpu(priv->ca, rx_ring[wr_id].mapping[0],
 						   dlen, DMA_FROM_DEVICE);
 			skb_copy_from_linear_data(skb, small_skb->data, dlen);
@@ -663,8 +665,7 @@ void ipoib_cm_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
 
 copied:
 	skb->protocol = ((struct ipoib_header *) skb->data)->proto;
-	skb_reset_mac_header(skb);
-	skb_pull(skb, IPOIB_ENCAP_LEN);
+	skb_add_pseudo_hdr(skb);
 
 	++dev->stats.rx_packets;
 	dev->stats.rx_bytes += skb->len;