Merge branch 'csums-next'

Tom Herbert says:

====================
net: Checksum offload changes - Part V

I am working on overhauling RX checksum offload. Goals of this effort
are:

- Specify what exactly it means when driver returns CHECKSUM_UNNECESSARY
- Preserve CHECKSUM_COMPLETE through encapsulation layers
- Don't do skb_checksum more than once per packet
- Unify GRO and non-GRO csum verification as much as possible
- Unify the checksum functions (checksum_init)
- Simplify code

What is in this fifth patch set:

- Added GRO checksum validation functions
- Call the GRO validations functions from TCP and GRE gro_receive
- Perform checksum verification in the UDP gro_receive path using
  GRO functions and add support for gro_receive in UDP6

Changes in V2:

- Change ip_summed to CHECKSUM_UNNECESSARY instead of moving it
  to CHECKSUM_COMPLETE from GRO checksum validation. This avoids
  performance penalty in checksumming bytes which are before the header
  GRO is at.

Please review carefully and test if possible, mucking with basic
checksum functions is always a little precarious :-)

----

Test results with this patch set are below. I did not notice any
performace regression.

Tests run:
   TCP_STREAM: super_netperf with 200 streams
   TCP_RR: super_netperf with 200 streams and -r 1,1

Device bnx2x (10Gbps):
   No GRE RSS hash (RX interrupts occur on one core)
   UDP RSS port hashing enabled.

* GRE with checksum with IPv4 encapsulated packets
  With fix:
    TCP_STREAM
        9.91% CPU utilization
        5163.78 Mbps
    TCP_RR
        50.64% CPU utilization
        219/347/502 90/95/99% latencies
        834103 tps
  Without fix:
    TCP_STREAM
        10.05% CPU utilization
        5186.22 tps
    TCP_RR
        49.70% CPU utilization
        227/338/486 90/95/99% latencies
        813450 tps

* GRE without checksum with IPv4 encapsulated packets
  With fix:
    TCP_STREAM
        10.18% CPU utilization
        5159 Mbps
    TCP_RR
        51.86% CPU utilization
        214/325/471 90/95/99% latencies
        865943 tps
  Without fix:
    TCP_STREAM
        10.26% CPU utilization
        5307.87 Mbps
    TCP_RR
        50.59% CPU utilization
        224/325/476 90/95/99% latencies
        846429 tps

*** Simulate device returns CHECKSUM_COMPLETE

* VXLAN with checksum
  With fix:
    TCP_STREAM
        13.03% CPU utilization
        9093.9 Mbps
    TCP_RR
        95.96% CPU utilization
        161/259/474 90/95/99% latencies
        1.14806e+06 tps
  Without fix:
    TCP_STREAM
        13.59% CPU utilization
        9093.97 Mbps
    TCP_RR
        93.95% CPU utilization
        160/259/484 90/95/99% latencies
        1.10262e+06 tps

* VXLAN without checksum
  With fix:
    TCP_STREAM
        13.28% CPU utilization
        9093.87 Mbps
    TCP_RR
        95.04% CPU utilization
        155/246/439 90/95/99% latencies
        1.15e+06 tps
  Without fix:
    TCP_STREAM
        13.37% CPU utilization
        9178.45 Mbps
    TCP_RR
        93.74% CPU utilization
        161/257/469 90/95/99% latencies
        1.1068e+06 Mbps
====================

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

4 files changed, 108 insertions, 2 deletions

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 7e2b0b8b5cd7..eb73444e1bd0 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1883,7 +1883,13 @@ struct napi_gro_cb {
 	u16	proto;
 
 	/* Used in udp_gro_receive */
-	u16	udp_mark;
+	u8	udp_mark:1;
+
+	/* GRO checksum is valid */
+	u8	csum_valid:1;
+
+	/* Number encapsulation layers crossed */
+	u8	encapsulation;
 
 	/* used to support CHECKSUM_COMPLETE for tunneling protocols */
 	__wsum	csum;
@@ -2154,11 +2160,77 @@ static inline void *skb_gro_network_header(struct sk_buff *skb)
 static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
 					const void *start, unsigned int len)
 {
-	if (skb->ip_summed == CHECKSUM_COMPLETE)
+	if (NAPI_GRO_CB(skb)->csum_valid)
 		NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
 						  csum_partial(start, len, 0));
 }
 
+/* GRO checksum functions. These are logical equivalents of the normal
+ * checksum functions (in skbuff.h) except that they operate on the GRO
+ * offsets and fields in sk_buff.
+ */
+
+__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
+
+static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
+						      bool zero_okay,
+						      __sum16 check)
+{
+	return (skb->ip_summed != CHECKSUM_PARTIAL &&
+		(skb->ip_summed != CHECKSUM_UNNECESSARY ||
+		 (NAPI_GRO_CB(skb)->encapsulation > skb->encapsulation)) &&
+		(!zero_okay || check));
+}
+
+static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
+							   __wsum psum)
+{
+	if (NAPI_GRO_CB(skb)->csum_valid &&
+	    !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
+		return 0;
+
+	NAPI_GRO_CB(skb)->csum = psum;
+
+	return __skb_gro_checksum_complete(skb);
+}
+
+/* Update skb for CHECKSUM_UNNECESSARY when we verified a top level
+ * checksum or an encapsulated one during GRO. This saves work
+ * if we fallback to normal path with the packet.
+ */
+static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
+{
+	if (skb->ip_summed == CHECKSUM_UNNECESSARY) {
+		if (NAPI_GRO_CB(skb)->encapsulation)
+			skb->encapsulation = 1;
+	} else if (skb->ip_summed != CHECKSUM_PARTIAL) {
+		skb->ip_summed = CHECKSUM_UNNECESSARY;
+		skb->encapsulation = 0;
+	}
+}
+
+#define __skb_gro_checksum_validate(skb, proto, zero_okay, check,	\
+				    compute_pseudo)			\
+({									\
+	__sum16 __ret = 0;						\
+	if (__skb_gro_checksum_validate_needed(skb, zero_okay, check))	\
+		__ret = __skb_gro_checksum_validate_complete(skb,	\
+				compute_pseudo(skb, proto));		\
+	if (!__ret)							\
+		skb_gro_incr_csum_unnecessary(skb);			\
+	__ret;								\
+})
+
+#define skb_gro_checksum_validate(skb, proto, compute_pseudo)		\
+	__skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
+
+#define skb_gro_checksum_validate_zero_check(skb, proto, check,		\
+					     compute_pseudo)		\
+	__skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
+
+#define skb_gro_checksum_simple_validate(skb)				\
+	__skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
+
 static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
 				  unsigned short type,
 				  const void *daddr, const void *saddr,
diff --git a/include/net/ip.h b/include/net/ip.h
index db4a771b9ef3..c8fd6112bd0b 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -364,6 +364,14 @@ static inline void inet_set_txhash(struct sock *sk)
 	sk->sk_txhash = flow_hash_from_keys(&keys);
 }
 
+static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
+{
+	const struct iphdr *iph = skb_gro_network_header(skb);
+
+	return csum_tcpudp_nofold(iph->saddr, iph->daddr,
+				  skb_gro_len(skb), proto, 0);
+}
+
 /*
  *	Map a multicast IP onto multicast MAC for type ethernet.
  */
diff --git a/include/net/ip6_checksum.h b/include/net/ip6_checksum.h
index 55236cb71174..1a49b73f7f6e 100644
--- a/include/net/ip6_checksum.h
+++ b/include/net/ip6_checksum.h
@@ -48,6 +48,14 @@ static inline __wsum ip6_compute_pseudo(struct sk_buff *skb, int proto)
 					    skb->len, proto, 0));
 }
 
+static inline __wsum ip6_gro_compute_pseudo(struct sk_buff *skb, int proto)
+{
+	const struct ipv6hdr *iph = skb_gro_network_header(skb);
+
+	return ~csum_unfold(csum_ipv6_magic(&iph->saddr, &iph->daddr,
+					    skb_gro_len(skb), proto, 0));
+}
+
 static __inline__ __sum16 tcp_v6_check(int len,
 				   const struct in6_addr *saddr,
 				   const struct in6_addr *daddr,
diff --git a/include/net/udp.h b/include/net/udp.h
index 70f941368ace..16f4e80f0519 100644
--- a/include/net/udp.h
+++ b/include/net/udp.h
@@ -158,6 +158,24 @@ static inline __sum16 udp_v4_check(int len, __be32 saddr,
 void udp_set_csum(bool nocheck, struct sk_buff *skb,
 		  __be32 saddr, __be32 daddr, int len);
 
+struct sk_buff **udp_gro_receive(struct sk_buff **head, struct sk_buff *skb,
+				 struct udphdr *uh);
+int udp_gro_complete(struct sk_buff *skb, int nhoff);
+
+static inline struct udphdr *udp_gro_udphdr(struct sk_buff *skb)
+{
+	struct udphdr *uh;
+	unsigned int hlen, off;
+
+	off  = skb_gro_offset(skb);
+	hlen = off + sizeof(*uh);
+	uh   = skb_gro_header_fast(skb, off);
+	if (skb_gro_header_hard(skb, hlen))
+		uh = skb_gro_header_slow(skb, hlen, off);
+
+	return uh;
+}
+
 /* hash routines shared between UDPv4/6 and UDP-Litev4/6 */
 static inline void udp_lib_hash(struct sock *sk)
 {