virtio_net: Fix skb->csum_start computation
authorMark McLoughlin <markmc@redhat.com>
Sun, 8 Jun 2008 10:49:00 +0000 (20:49 +1000)
committerJeff Garzik <jgarzik@redhat.com>
Tue, 10 Jun 2008 22:20:29 +0000 (18:20 -0400)
hdr->csum_start is the offset from the start of the ethernet
header to the transport layer checksum field. skb->csum_start
is the offset from skb->head.

skb_partial_csum_set() assumes that skb->data points to the
ethernet header - i.e. it computes skb->csum_start by adding
the headroom to hdr->csum_start.

Since eth_type_trans() skb_pull()s the ethernet header,
skb_partial_csum_set() should be called before
eth_type_trans().

(Without this patch, GSO packets from a guest to the world outside the
host are corrupted).

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
drivers/net/virtio_net.c

index 5450eac..9a3b85e 100644 (file)
@@ -94,9 +94,7 @@ static void receive_skb(struct net_device *dev, struct sk_buff *skb,
        BUG_ON(len > MAX_PACKET_LEN);
 
        skb_trim(skb, len);
-       skb->protocol = eth_type_trans(skb, dev);
-       pr_debug("Receiving skb proto 0x%04x len %i type %i\n",
-                ntohs(skb->protocol), skb->len, skb->pkt_type);
+
        dev->stats.rx_bytes += skb->len;
        dev->stats.rx_packets++;
 
@@ -106,6 +104,10 @@ static void receive_skb(struct net_device *dev, struct sk_buff *skb,
                        goto frame_err;
        }
 
+       skb->protocol = eth_type_trans(skb, dev);
+       pr_debug("Receiving skb proto 0x%04x len %i type %i\n",
+                ntohs(skb->protocol), skb->len, skb->pkt_type);
+
        if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
                pr_debug("GSO!\n");
                switch (hdr->gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {