[PATCH net-next v3 0/2] ila: Precompute checksums

[PATCH net-next v3 0/2] ila: Precompute checksums

[Tom Herbert]

This patch set:
 - Adds argument ot LWT build_state that holds a pointer to the fib
   configuration being applied to the new route
 - Adds support in ILA to precompute checksum difference for
   performance optimization

v2:
 - Move return argument in build_state to end of arguments

v3:
 - Update the signature for ip6_tun_build_state()


Tom Herbert (2):
  lwt: Add cfg argument to build_state
  ila: Precompute checksum difference for translations

 include/net/lwtunnel.h    |  3 +++
 net/core/lwtunnel.c       |  5 +++--
 net/ipv4/fib_semantics.c  | 17 ++++++++++-------
 net/ipv4/ip_tunnel_core.c |  2 ++
 net/ipv6/ila.c            | 19 +++++++++++++++++++
 net/ipv6/route.c          |  3 ++-
 net/mpls/mpls_iptunnel.c  |  1 +
 7 files changed, 40 insertions(+), 10 deletions(-)

-- 
1.8.1

[PATCH net-next v3 1/2] lwt: Add cfg argument to build_state

[Tom Herbert]

Add cfg and family arguments to lwt build state functions. cfg is a void
pointer and will either be a pointer to a fib_config or fib6_config
structure. The family parameter indicates which one (either AF_INET
or AF_INET6).

LWT encpasulation implementation may use the fib configuration to build
the LWT state.

Signed-off-by: Tom Herbert <tom@herbertland.com>
---
 include/net/lwtunnel.h    |  3 +++
 net/core/lwtunnel.c       |  5 +++--
 net/ipv4/fib_semantics.c  | 17 ++++++++++-------
 net/ipv4/ip_tunnel_core.c |  2 ++
 net/ipv6/ila.c            |  1 +
 net/ipv6/route.c          |  3 ++-
 net/mpls/mpls_iptunnel.c  |  1 +
 7 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/include/net/lwtunnel.h b/include/net/lwtunnel.h
index 8434898..fce0e35 100644
--- a/include/net/lwtunnel.h
+++ b/include/net/lwtunnel.h
@@ -26,6 +26,7 @@ struct lwtunnel_state {
 
 struct lwtunnel_encap_ops {
 	int (*build_state)(struct net_device *dev, struct nlattr *encap,
+			   unsigned int family, const void *cfg,
 			   struct lwtunnel_state **ts);
 	int (*output)(struct sock *sk, struct sk_buff *skb);
 	int (*input)(struct sk_buff *skb);
@@ -80,6 +81,7 @@ int lwtunnel_encap_del_ops(const struct lwtunnel_encap_ops *op,
 			   unsigned int num);
 int lwtunnel_build_state(struct net_device *dev, u16 encap_type,
 			 struct nlattr *encap,
+			 unsigned int family, const void *cfg,
 			 struct lwtunnel_state **lws);
 int lwtunnel_fill_encap(struct sk_buff *skb,
 			struct lwtunnel_state *lwtstate);
@@ -130,6 +132,7 @@ static inline int lwtunnel_encap_del_ops(const struct lwtunnel_encap_ops *op,
 
 static inline int lwtunnel_build_state(struct net_device *dev, u16 encap_type,
 				       struct nlattr *encap,
+				       unsigned int family, const void *cfg,
 				       struct lwtunnel_state **lws)
 {
 	return -EOPNOTSUPP;
diff --git a/net/core/lwtunnel.c b/net/core/lwtunnel.c
index e924c2e..dfb1a9c 100644
--- a/net/core/lwtunnel.c
+++ b/net/core/lwtunnel.c
@@ -72,7 +72,8 @@ int lwtunnel_encap_del_ops(const struct lwtunnel_encap_ops *ops,
 EXPORT_SYMBOL(lwtunnel_encap_del_ops);
 
 int lwtunnel_build_state(struct net_device *dev, u16 encap_type,
-			 struct nlattr *encap, struct lwtunnel_state **lws)
+			 struct nlattr *encap, unsigned int family,
+			 const void *cfg, struct lwtunnel_state **lws)
 {
 	const struct lwtunnel_encap_ops *ops;
 	int ret = -EINVAL;
@@ -85,7 +86,7 @@ int lwtunnel_build_state(struct net_device *dev, u16 encap_type,
 	rcu_read_lock();
 	ops = rcu_dereference(lwtun_encaps[encap_type]);
 	if (likely(ops && ops->build_state))
-		ret = ops->build_state(dev, encap, lws);
+		ret = ops->build_state(dev, encap, family, cfg, lws);
 	rcu_read_unlock();
 
 	return ret;
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 01f1c7d..1b2d011 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -511,7 +511,8 @@ static int fib_get_nhs(struct fib_info *fi, struct rtnexthop *rtnh,
 					dev = __dev_get_by_index(net, cfg->fc_oif);
 				ret = lwtunnel_build_state(dev, nla_get_u16(
 							   nla_entype),
-							   nla, &lwtstate);
+							   nla,  AF_INET, cfg,
+							   &lwtstate);
 				if (ret)
 					goto errout;
 				nexthop_nh->nh_lwtstate =
@@ -535,7 +536,8 @@ errout:
 
 static int fib_encap_match(struct net *net, u16 encap_type,
 			   struct nlattr *encap,
-			   int oif, const struct fib_nh *nh)
+			   int oif, const struct fib_nh *nh,
+			   const struct fib_config *cfg)
 {
 	struct lwtunnel_state *lwtstate;
 	struct net_device *dev = NULL;
@@ -546,8 +548,8 @@ static int fib_encap_match(struct net *net, u16 encap_type,
 
 	if (oif)
 		dev = __dev_get_by_index(net, oif);
-	ret = lwtunnel_build_state(dev, encap_type,
-				   encap, &lwtstate);
+	ret = lwtunnel_build_state(dev, encap_type, encap,
+				   AF_INET, cfg, &lwtstate);
 	if (!ret) {
 		result = lwtunnel_cmp_encap(lwtstate, nh->nh_lwtstate);
 		lwtstate_free(lwtstate);
@@ -571,7 +573,7 @@ int fib_nh_match(struct fib_config *cfg, struct fib_info *fi)
 		if (cfg->fc_encap) {
 			if (fib_encap_match(net, cfg->fc_encap_type,
 					    cfg->fc_encap, cfg->fc_oif,
-					    fi->fib_nh))
+					    fi->fib_nh, cfg))
 			    return 1;
 		}
 		if ((!cfg->fc_oif || cfg->fc_oif == fi->fib_nh->nh_oif) &&
@@ -663,7 +665,7 @@ int fib_nh_match(struct fib_config *cfg, struct fib_info *fi)
 static int fib_check_nh(struct fib_config *cfg, struct fib_info *fi,
 			struct fib_nh *nh)
 {
-	int err;
+	int err = 0;
 	struct net *net;
 	struct net_device *dev;
 
@@ -1005,7 +1007,8 @@ struct fib_info *fib_create_info(struct fib_config *cfg)
 			if (cfg->fc_oif)
 				dev = __dev_get_by_index(net, cfg->fc_oif);
 			err = lwtunnel_build_state(dev, cfg->fc_encap_type,
-						   cfg->fc_encap, &lwtstate);
+						   cfg->fc_encap, AF_INET, cfg,
+						   &lwtstate);
 			if (err)
 				goto failure;
 
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index 289b6c2..934f2ac 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -204,6 +204,7 @@ static const struct nla_policy ip_tun_policy[LWTUNNEL_IP_MAX + 1] = {
 };
 
 static int ip_tun_build_state(struct net_device *dev, struct nlattr *attr,
+			      unsigned int family, const void *cfg,
 			      struct lwtunnel_state **ts)
 {
 	struct ip_tunnel_info *tun_info;
@@ -311,6 +312,7 @@ static const struct nla_policy ip6_tun_policy[LWTUNNEL_IP6_MAX + 1] = {
 };
 
 static int ip6_tun_build_state(struct net_device *dev, struct nlattr *attr,
+			       unsigned int family, const void *cfg,
 			       struct lwtunnel_state **ts)
 {
 	struct ip_tunnel_info *tun_info;
diff --git a/net/ipv6/ila.c b/net/ipv6/ila.c
index f011c3d..ffe4dca 100644
--- a/net/ipv6/ila.c
+++ b/net/ipv6/ila.c
@@ -123,6 +123,7 @@ static struct nla_policy ila_nl_policy[ILA_ATTR_MAX + 1] = {
 };
 
 static int ila_build_state(struct net_device *dev, struct nlattr *nla,
+			   unsigned int family, const void *cfg,
 			   struct lwtunnel_state **ts)
 {
 	struct ila_params *p;
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index e476f01..df3e353 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -1819,7 +1819,8 @@ int ip6_route_add(struct fib6_config *cfg)
 		struct lwtunnel_state *lwtstate;
 
 		err = lwtunnel_build_state(dev, cfg->fc_encap_type,
-					   cfg->fc_encap, &lwtstate);
+					   cfg->fc_encap, AF_INET6, cfg,
+					   &lwtstate);
 		if (err)
 			goto out;
 		rt->dst.lwtstate = lwtstate_get(lwtstate);
diff --git a/net/mpls/mpls_iptunnel.c b/net/mpls/mpls_iptunnel.c
index 3da5ca3..21e70bc 100644
--- a/net/mpls/mpls_iptunnel.c
+++ b/net/mpls/mpls_iptunnel.c
@@ -123,6 +123,7 @@ drop:
 }
 
 static int mpls_build_state(struct net_device *dev, struct nlattr *nla,
+			    unsigned int family, const void *cfg,
 			    struct lwtunnel_state **ts)
 {
 	struct mpls_iptunnel_encap *tun_encap_info;
-- 
1.8.1

[PATCH net-next v3 2/2] ila: Precompute checksum difference for translations

[Tom Herbert]

In the ILA build state for LWT compute the checksum difference to apply
to transport checksums that include the IPv6 pseudo header. The
difference is between the route destination (from fib6_config) and the
locator to write.

Signed-off-by: Tom Herbert <tom@herbertland.com>
---
 net/ipv6/ila.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/net/ipv6/ila.c b/net/ipv6/ila.c
index ffe4dca..678d2df 100644
--- a/net/ipv6/ila.c
+++ b/net/ipv6/ila.c
@@ -14,6 +14,8 @@
 
 struct ila_params {
 	__be64 locator;
+	__be64 locator_match;
+	__wsum csum_diff;
 };
 
 static inline struct ila_params *ila_params_lwtunnel(
@@ -33,6 +35,9 @@ static inline __wsum compute_csum_diff8(const __be32 *from, const __be32 *to)
 
 static inline __wsum get_csum_diff(struct ipv6hdr *ip6h, struct ila_params *p)
 {
+	if (*(__be64 *)&ip6h->daddr == p->locator_match)
+		return p->csum_diff;
+	else
 		return compute_csum_diff8((__be32 *)&ip6h->daddr,
 					  (__be32 *)&p->locator);
 }
@@ -130,8 +135,12 @@ static int ila_build_state(struct net_device *dev, struct nlattr *nla,
 	struct nlattr *tb[ILA_ATTR_MAX + 1];
 	size_t encap_len = sizeof(*p);
 	struct lwtunnel_state *newts;
+	const struct fib6_config *cfg6 = cfg;
 	int ret;
 
+	if (family != AF_INET6)
+		return -EINVAL;
+
 	ret = nla_parse_nested(tb, ILA_ATTR_MAX, nla,
 			       ila_nl_policy);
 	if (ret < 0)
@@ -149,6 +158,15 @@ static int ila_build_state(struct net_device *dev, struct nlattr *nla,
 
 	p->locator = (__force __be64)nla_get_u64(tb[ILA_ATTR_LOCATOR]);
 
+	if (cfg6->fc_dst_len > sizeof(__be64)) {
+		/* Precompute checksum difference for translation since we
+		 * know both the old locator and the new one.
+		 */
+		p->locator_match = *(__be64 *)&cfg6->fc_dst;
+		p->csum_diff = compute_csum_diff8(
+			(__be32 *)&p->locator_match, (__be32 *)&p->locator);
+	}
+
 	newts->type = LWTUNNEL_ENCAP_ILA;
 	newts->flags |= LWTUNNEL_STATE_OUTPUT_REDIRECT |
 			LWTUNNEL_STATE_INPUT_REDIRECT;
-- 
1.8.1

Re: [PATCH net-next v3 0/2] ila: Precompute checksums

[David Miller]

From: Tom Herbert <tom@herbertland.com>
Date: Mon, 24 Aug 2015 09:45:40 -0700

> This patch set:
>  - Adds argument ot LWT build_state that holds a pointer to the fib
>    configuration being applied to the new route
>  - Adds support in ILA to precompute checksum difference for
>    performance optimization
> 
> v2:
>  - Move return argument in build_state to end of arguments
> 
> v3:
>  - Update the signature for ip6_tun_build_state()

Applied, thanks.