ehea: Fix TSO and hw checksums with non-accelerated vlan packets.
authorVlad Yasevich <vyasevich@gmail.com>
Mon, 25 Aug 2014 14:34:51 +0000 (10:34 -0400)
committerDavid S. Miller <davem@davemloft.net>
Tue, 26 Aug 2014 00:27:09 +0000 (17:27 -0700)
The driver claims that it can do TSO and IP checksums on vlan
devices and also allows user to control vlan acceleration offloading.
This makes it possible to push traffic to this driver that has TSO or
partial checksums set, but also have a non-accelearted vlan
header.  In this case, the driver will fail to correctly
identify such traffic and will not correctly perform
segmentation and checksum calculation.

Fix this by using vlan_get_protocol() helper instead of
assuming skb->protocol always has this information.

CC: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/ibm/ehea/ehea_main.c

index a0b418e007a0d09303a32838d8acf1a06ca9d061..566b17db135a306a2bb8eea0ce167a70e6e165f4 100644 (file)
@@ -1994,7 +1994,7 @@ static void xmit_common(struct sk_buff *skb, struct ehea_swqe *swqe)
 {
        swqe->tx_control |= EHEA_SWQE_IMM_DATA_PRESENT | EHEA_SWQE_CRC;
 
-       if (skb->protocol != htons(ETH_P_IP))
+       if (vlan_get_protocol(skb) != htons(ETH_P_IP))
                return;
 
        if (skb->ip_summed == CHECKSUM_PARTIAL)