[NET]: Detect hardware rx checksum faults correctly

Here is the patch that introduces the generic skb_checksum_complete
which also checks for hardware RX checksum faults.  If that happens,
it'll call netdev_rx_csum_fault which currently prints out a stack
trace with the device name.  In future it can turn off RX checksum.

I've converted every spot under net/ that does RX checksum checks to
use skb_checksum_complete or __skb_checksum_complete with the
exceptions of:

* Those places where checksums are done bit by bit.  These will call
netdev_rx_csum_fault directly.

* The following have not been completely checked/converted:

ipmr
ip_vs
netfilter
dccp

This patch is based on patches and suggestions from Stephen Hemminger
and David S. Miller.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 44 insertions, 64 deletions

diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index 23e5403..1bdf0fb 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -585,17 +585,16 @@ static int icmpv6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
 	daddr = &skb->nh.ipv6h->daddr;
 
 	/* Perform checksum. */
-	if (skb->ip_summed == CHECKSUM_HW) {
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
-		if (csum_ipv6_magic(saddr, daddr, skb->len, IPPROTO_ICMPV6,
-				    skb->csum)) {
-			LIMIT_NETDEBUG(KERN_DEBUG "ICMPv6 hw checksum failed\n");
-			skb->ip_summed = CHECKSUM_NONE;
-		}
-	}
-	if (skb->ip_summed == CHECKSUM_NONE) {
-		if (csum_ipv6_magic(saddr, daddr, skb->len, IPPROTO_ICMPV6,
-				    skb_checksum(skb, 0, skb->len, 0))) {
+	switch (skb->ip_summed) {
+	case CHECKSUM_HW:
+		if (!csum_ipv6_magic(saddr, daddr, skb->len, IPPROTO_ICMPV6,
+				     skb->csum))
+			break;
+		/* fall through */
+	case CHECKSUM_NONE:
+		skb->csum = ~csum_ipv6_magic(saddr, daddr, skb->len,
+					     IPPROTO_ICMPV6, 0);
+		if (__skb_checksum_complete(skb)) {
 			LIMIT_NETDEBUG(KERN_DEBUG "ICMPv6 checksum failed [%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x > %04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x]\n",
 				       NIP6(*saddr), NIP6(*daddr));
 			goto discard_it;
diff --git a/net/ipv6/raw.c b/net/ipv6/raw.c
index 651c79b..8e9628f 100644
--- a/net/ipv6/raw.c
+++ b/net/ipv6/raw.c
@@ -298,13 +298,10 @@ void rawv6_err(struct sock *sk, struct sk_buff *skb,
 static inline int rawv6_rcv_skb(struct sock * sk, struct sk_buff * skb)
 {
 	if ((raw6_sk(sk)->checksum || sk->sk_filter) && 
-	    skb->ip_summed != CHECKSUM_UNNECESSARY) {
-		if ((unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum))) {
-			/* FIXME: increment a raw6 drops counter here */
-			kfree_skb(skb);
-			return 0;
-		}
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
+	    skb_checksum_complete(skb)) {
+		/* FIXME: increment a raw6 drops counter here */
+		kfree_skb(skb);
+		return 0;
 	}
 
 	/* Charge it to the socket. */
@@ -337,32 +334,25 @@ int rawv6_rcv(struct sock *sk, struct sk_buff *skb)
 	if (!rp->checksum)
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 
-	if (skb->ip_summed != CHECKSUM_UNNECESSARY) {
-		if (skb->ip_summed == CHECKSUM_HW) {
-			skb_postpull_rcsum(skb, skb->nh.raw,
-			                   skb->h.raw - skb->nh.raw);
+	if (skb->ip_summed == CHECKSUM_HW) {
+		skb_postpull_rcsum(skb, skb->nh.raw,
+		                   skb->h.raw - skb->nh.raw);
+		if (!csum_ipv6_magic(&skb->nh.ipv6h->saddr,
+				     &skb->nh.ipv6h->daddr,
+				     skb->len, inet->num, skb->csum))
 			skb->ip_summed = CHECKSUM_UNNECESSARY;
-			if (csum_ipv6_magic(&skb->nh.ipv6h->saddr,
-					    &skb->nh.ipv6h->daddr,
-					    skb->len, inet->num, skb->csum)) {
-				LIMIT_NETDEBUG(KERN_DEBUG "raw v6 hw csum failure.\n");
-				skb->ip_summed = CHECKSUM_NONE;
-			}
-		}
-		if (skb->ip_summed == CHECKSUM_NONE)
-			skb->csum = ~csum_ipv6_magic(&skb->nh.ipv6h->saddr,
-						     &skb->nh.ipv6h->daddr,
-						     skb->len, inet->num, 0);
 	}
+	if (skb->ip_summed != CHECKSUM_UNNECESSARY)
+		skb->csum = ~csum_ipv6_magic(&skb->nh.ipv6h->saddr,
+					     &skb->nh.ipv6h->daddr,
+					     skb->len, inet->num, 0);
 
 	if (inet->hdrincl) {
-		if (skb->ip_summed != CHECKSUM_UNNECESSARY &&
-		    (unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum))) {
+		if (skb_checksum_complete(skb)) {
 			/* FIXME: increment a raw6 drops counter here */
 			kfree_skb(skb);
 			return 0;
 		}
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
 	}
 
 	rawv6_rcv_skb(sk, skb);
@@ -407,7 +397,7 @@ static int rawv6_recvmsg(struct kiocb *iocb, struct sock *sk,
 	if (skb->ip_summed==CHECKSUM_UNNECESSARY) {
 		err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 	} else if (msg->msg_flags&MSG_TRUNC) {
-		if ((unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum)))
+		if (__skb_checksum_complete(skb))
 			goto csum_copy_err;
 		err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 	} else {
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index d746d3b..62c0e5b 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1401,20 +1401,18 @@ out:
 static int tcp_v6_checksum_init(struct sk_buff *skb)
 {
 	if (skb->ip_summed == CHECKSUM_HW) {
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
 		if (!tcp_v6_check(skb->h.th,skb->len,&skb->nh.ipv6h->saddr,
-				  &skb->nh.ipv6h->daddr,skb->csum))
+				  &skb->nh.ipv6h->daddr,skb->csum)) {
+			skb->ip_summed = CHECKSUM_UNNECESSARY;
 			return 0;
-		LIMIT_NETDEBUG(KERN_DEBUG "hw tcp v6 csum failed\n");
+		}
 	}
+
+	skb->csum = ~tcp_v6_check(skb->h.th,skb->len,&skb->nh.ipv6h->saddr,
+				  &skb->nh.ipv6h->daddr, 0);
+
 	if (skb->len <= 76) {
-		if (tcp_v6_check(skb->h.th,skb->len,&skb->nh.ipv6h->saddr,
-				 &skb->nh.ipv6h->daddr,skb_checksum(skb, 0, skb->len, 0)))
-			return -1;
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
-	} else {
-		skb->csum = ~tcp_v6_check(skb->h.th,skb->len,&skb->nh.ipv6h->saddr,
-					  &skb->nh.ipv6h->daddr,0);
+		return __skb_checksum_complete(skb);
 	}
 	return 0;
 }
@@ -1575,7 +1573,7 @@ static int tcp_v6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
 		goto discard_it;
 
 	if ((skb->ip_summed != CHECKSUM_UNNECESSARY &&
-	     tcp_v6_checksum_init(skb) < 0))
+	     tcp_v6_checksum_init(skb)))
 		goto bad_packet;
 
 	th = skb->h.th;
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index bf95193..e671153 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -248,7 +248,7 @@ try_again:
 		err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov,
 					      copied);
 	} else if (msg->msg_flags&MSG_TRUNC) {
-		if ((unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum)))
+		if (__skb_checksum_complete(skb))
 			goto csum_copy_err;
 		err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov,
 					      copied);
@@ -363,13 +363,10 @@ static inline int udpv6_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
 		return -1;
 	}
 
-	if (skb->ip_summed != CHECKSUM_UNNECESSARY) {
-		if ((unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum))) {
-			UDP6_INC_STATS_BH(UDP_MIB_INERRORS);
-			kfree_skb(skb);
-			return 0;
-		}
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
+	if (skb_checksum_complete(skb)) {
+		UDP6_INC_STATS_BH(UDP_MIB_INERRORS);
+		kfree_skb(skb);
+		return 0;
 	}
 
 	if (sock_queue_rcv_skb(sk,skb)<0) {
@@ -491,13 +488,10 @@ static int udpv6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
 		uh = skb->h.uh;
 	}
 
-	if (skb->ip_summed==CHECKSUM_HW) {
+	if (skb->ip_summed == CHECKSUM_HW &&
+	    !csum_ipv6_magic(saddr, daddr, ulen, IPPROTO_UDP, skb->csum))
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
-		if (csum_ipv6_magic(saddr, daddr, ulen, IPPROTO_UDP, skb->csum)) {
-			LIMIT_NETDEBUG(KERN_DEBUG "udp v6 hw csum failure.\n");
-			skb->ip_summed = CHECKSUM_NONE;
-		}
-	}
+
 	if (skb->ip_summed != CHECKSUM_UNNECESSARY)
 		skb->csum = ~csum_ipv6_magic(saddr, daddr, ulen, IPPROTO_UDP, 0);
 
@@ -521,8 +515,7 @@ static int udpv6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
 		if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
 			goto discard;
 
-		if (skb->ip_summed != CHECKSUM_UNNECESSARY &&
-		    (unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum)))
+		if (skb_checksum_complete(skb))
 			goto discard;
 		UDP6_INC_STATS_BH(UDP_MIB_NOPORTS);