[NET]: IPV6 checksum offloading in network devices

The existing model for checksum offload does not correctly handle
devices that can offload IPV4 and IPV6 only. The NETIF_F_HW_CSUM flag
implies device can do any arbitrary protocol.

This patch:
 * adds NETIF_F_IPV6_CSUM for those devices
 * fixes bnx2 and tg3 devices that need it
 * add NETIF_F_IPV6_CSUM to ipv6 output (incl GSO)
 * fixes assumptions about NETIF_F_ALL_CSUM in nat
 * adjusts bridge union of checksumming computation

Signed-off-by: David S. Miller <davem@davemloft.net>

9 files changed, 49 insertions, 17 deletions

diff --git a/drivers/net/bnx2.c b/drivers/net/bnx2.c
index ce3ed67..0f4f76f 100644
--- a/drivers/net/bnx2.c
+++ b/drivers/net/bnx2.c
@@ -6490,10 +6490,10 @@ bnx2_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	memcpy(dev->perm_addr, bp->mac_addr, 6);
 	bp->name = board_info[ent->driver_data].name;
 
+	dev->features |= NETIF_F_IP_CSUM | NETIF_F_SG;
 	if (CHIP_NUM(bp) == CHIP_NUM_5709)
-		dev->features |= NETIF_F_HW_CSUM | NETIF_F_SG;
-	else
-		dev->features |= NETIF_F_IP_CSUM | NETIF_F_SG;
+		dev->features |= NETIF_F_IPV6_CSUM;
+
 #ifdef BCM_VLAN
 	dev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX;
 #endif
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 2f31841..3a43426 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -11944,12 +11944,11 @@ static int __devinit tg3_init_one(struct pci_dev *pdev,
 	 * checksumming.
 	 */
 	if ((tp->tg3_flags & TG3_FLAG_BROKEN_CHECKSUMS) == 0) {
+		dev->features |= NETIF_F_IP_CSUM | NETIF_F_SG;
 		if (GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5755 ||
 		    GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5787)
-			dev->features |= NETIF_F_HW_CSUM;
-		else
-			dev->features |= NETIF_F_IP_CSUM;
-		dev->features |= NETIF_F_SG;
+			dev->features |= NETIF_F_IPV6_CSUM;
+
 		tp->tg3_flags |= TG3_FLAG_RX_CHECKSUMS;
 	} else
 		tp->tg3_flags &= ~TG3_FLAG_RX_CHECKSUMS;
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index e7913ee..7a8f22f 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -314,9 +314,10 @@ struct net_device
 	/* Net device features */
 	unsigned long		features;
 #define NETIF_F_SG		1	/* Scatter/gather IO. */
-#define NETIF_F_IP_CSUM		2	/* Can checksum only TCP/UDP over IPv4. */
+#define NETIF_F_IP_CSUM		2	/* Can checksum TCP/UDP over IPv4. */
 #define NETIF_F_NO_CSUM		4	/* Does not require checksum. F.e. loopack. */
 #define NETIF_F_HW_CSUM		8	/* Can checksum all the packets. */
+#define NETIF_F_IPV6_CSUM	16	/* Can checksum TCP/UDP over IPV6 */
 #define NETIF_F_HIGHDMA		32	/* Can DMA to high memory. */
 #define NETIF_F_FRAGLIST	64	/* Scatter/gather IO. */
 #define NETIF_F_HW_VLAN_TX	128	/* Transmit VLAN hw acceleration */
@@ -338,8 +339,11 @@ struct net_device
 	/* List of features with software fallbacks. */
 #define NETIF_F_GSO_SOFTWARE	(NETIF_F_TSO | NETIF_F_TSO_ECN | NETIF_F_TSO6)
 
+
 #define NETIF_F_GEN_CSUM	(NETIF_F_NO_CSUM | NETIF_F_HW_CSUM)
-#define NETIF_F_ALL_CSUM	(NETIF_F_IP_CSUM | NETIF_F_GEN_CSUM)
+#define NETIF_F_V4_CSUM		(NETIF_F_GEN_CSUM | NETIF_F_IP_CSUM)
+#define NETIF_F_V6_CSUM		(NETIF_F_GEN_CSUM | NETIF_F_IPV6_CSUM)
+#define NETIF_F_ALL_CSUM	(NETIF_F_V4_CSUM | NETIF_F_V6_CSUM)
 
 	struct net_device	*next_sched;
 
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 849deaf..7b4ce91 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -368,10 +368,18 @@ void br_features_recompute(struct net_bridge *br)
 	list_for_each_entry(p, &br->port_list, list) {
 		unsigned long feature = p->dev->features;
 
+		/* if device needs checksumming, downgrade to hw checksumming */
 		if (checksum & NETIF_F_NO_CSUM && !(feature & NETIF_F_NO_CSUM))
 			checksum ^= NETIF_F_NO_CSUM | NETIF_F_HW_CSUM;
+
+		/* if device can't do all checksum, downgrade to ipv4/ipv6 */
 		if (checksum & NETIF_F_HW_CSUM && !(feature & NETIF_F_HW_CSUM))
-			checksum ^= NETIF_F_HW_CSUM | NETIF_F_IP_CSUM;
+			checksum ^= NETIF_F_HW_CSUM
+				| NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM;
+
+		if (checksum & NETIF_F_IPV6_CSUM && !(feature & NETIF_F_IPV6_CSUM))
+			checksum &= ~NETIF_F_IPV6_CSUM;
+
 		if (!(feature & NETIF_F_IP_CSUM))
 			checksum = 0;
 
diff --git a/net/core/dev.c b/net/core/dev.c
index ee051bb..a0a46e7 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1509,9 +1509,11 @@ int dev_queue_xmit(struct sk_buff *skb)
 		skb_set_transport_header(skb, skb->csum_start -
 					      skb_headroom(skb));
 
-		if (!(dev->features & NETIF_F_GEN_CSUM) &&
-		    (!(dev->features & NETIF_F_IP_CSUM) ||
-		     skb->protocol != htons(ETH_P_IP)))
+		if (!(dev->features & NETIF_F_GEN_CSUM)
+		    || ((dev->features & NETIF_F_IP_CSUM)
+			&& skb->protocol == htons(ETH_P_IP))
+		    || ((dev->features & NETIF_F_IPV6_CSUM)
+			&& skb->protocol == htons(ETH_P_IPV6)))
 			if (skb_checksum_help(skb))
 				goto out_kfree_skb;
 	}
@@ -3107,6 +3109,22 @@ int register_netdevice(struct net_device *dev)
 		}
 	}
 
+	/* Fix illegal checksum combinations */
+	if ((dev->features & NETIF_F_HW_CSUM) &&
+	    (dev->features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
+		printk(KERN_NOTICE "%s: mixed HW and IP checksum settings.\n",
+		       dev->name);
+		dev->features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
+	}
+
+	if ((dev->features & NETIF_F_NO_CSUM) &&
+	    (dev->features & (NETIF_F_HW_CSUM|NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
+		printk(KERN_NOTICE "%s: mixed no checksumming and other settings.\n",
+		       dev->name);
+		dev->features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM|NETIF_F_HW_CSUM);
+	}
+
+
 	/* Fix illegal SG+CSUM combinations. */
 	if ((dev->features & NETIF_F_SG) &&
 	    !(dev->features & NETIF_F_ALL_CSUM)) {
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 041fba3..06c08e5 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1170,6 +1170,9 @@ static struct sk_buff *inet_gso_segment(struct sk_buff *skb, int features)
 	int ihl;
 	int id;
 
+	if (!(features & NETIF_F_V4_CSUM))
+		features &= ~NETIF_F_SG;
+
 	if (unlikely(skb_shinfo(skb)->gso_type &
 		     ~(SKB_GSO_TCPV4 |
 		       SKB_GSO_UDP |
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 34ea454..a7dd343 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -837,7 +837,7 @@ int ip_append_data(struct sock *sk,
 	 */
 	if (transhdrlen &&
 	    length + fragheaderlen <= mtu &&
-	    rt->u.dst.dev->features & NETIF_F_ALL_CSUM &&
+	    rt->u.dst.dev->features & NETIF_F_V4_CSUM &&
 	    !exthdrlen)
 		csummode = CHECKSUM_PARTIAL;
 
diff --git a/net/ipv4/netfilter/nf_nat_helper.c b/net/ipv4/netfilter/nf_nat_helper.c
index 15b6e5c..b1aa598 100644
--- a/net/ipv4/netfilter/nf_nat_helper.c
+++ b/net/ipv4/netfilter/nf_nat_helper.c
@@ -178,7 +178,7 @@ nf_nat_mangle_tcp_packet(struct sk_buff **pskb,
 	datalen = (*pskb)->len - iph->ihl*4;
 	if ((*pskb)->ip_summed != CHECKSUM_PARTIAL) {
 		if (!(rt->rt_flags & RTCF_LOCAL) &&
-		    (*pskb)->dev->features & NETIF_F_ALL_CSUM) {
+		    (*pskb)->dev->features & NETIF_F_V4_CSUM) {
 			(*pskb)->ip_summed = CHECKSUM_PARTIAL;
 			(*pskb)->csum_start = skb_headroom(*pskb) +
 					      skb_network_offset(*pskb) +
@@ -265,7 +265,7 @@ nf_nat_mangle_udp_packet(struct sk_buff **pskb,
 
 	if ((*pskb)->ip_summed != CHECKSUM_PARTIAL) {
 		if (!(rt->rt_flags & RTCF_LOCAL) &&
-		    (*pskb)->dev->features & NETIF_F_ALL_CSUM) {
+		    (*pskb)->dev->features & NETIF_F_V4_CSUM) {
 			(*pskb)->ip_summed = CHECKSUM_PARTIAL;
 			(*pskb)->csum_start = skb_headroom(*pskb) +
 					      skb_network_offset(*pskb) +
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index b636c38..1c35066 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -123,7 +123,7 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb, int features)
 	struct ipv6hdr *ipv6h;
 	struct inet6_protocol *ops;
 
-	if (!(features & NETIF_F_HW_CSUM))
+	if (!(features & NETIF_F_V6_CSUM))
 		features &= ~NETIF_F_SG;
 
 	if (unlikely(skb_shinfo(skb)->gso_type &