From: Herbert Xu <herbert@gondor.apana.org.au> Date: Wed, 27 May 2009 15:31:17 +1000 Subject: [net] ipv4: add GRO infrastructure Message-id: E1M9BjR-0002Bp-BG@gondolin.me.apana.org.au O-Subject: [PATCH 12/17] ipv4: Add GRO infrastructure Bugzilla: 499347 RH-Acked-by: David Miller <davem@redhat.com> RH-Acked-by: Neil Horman <nhorman@redhat.com> RH-Acked-by: Thomas Graf <tgraf@redhat.com> RHEL5 bugzilla #499347 ipv4: Add GRO infrastructure This patch adds GRO support for IPv4. The criteria for merging is more stringent than LRO, in particular, we require all fields in the IP header to be identical except for the length, ID and checksum. In addition, the ID must form an arithmetic sequence with a difference of one. The ID requirement might seem overly strict, however, most hardware TSO solutions already obey this rule. Linux itself also obeys this whether GSO is in use or not. In future we could relax this rule by storing the IDs (or rather making sure that we don't drop them when pulling the aggregate skb's tail). Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net> diff --git a/include/net/protocol.h b/include/net/protocol.h index c643bce..38d4bf6 100644 --- a/include/net/protocol.h +++ b/include/net/protocol.h @@ -42,6 +42,12 @@ struct net_protocol { int no_policy; }; +struct net_gro_protocol { + struct sk_buff **(*gro_receive)(struct sk_buff **head, + struct sk_buff *skb); + int (*gro_complete)(struct sk_buff *skb); +}; + #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) struct inet6_protocol { @@ -89,12 +95,15 @@ struct inet_protosw { extern struct net_protocol *inet_protocol_base; extern struct net_protocol *inet_protos[MAX_INET_PROTOS]; +extern struct net_gro_protocol *inet_gro_protos[MAX_INET_PROTOS]; #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) extern struct inet6_protocol *inet6_protos[MAX_INET_PROTOS]; #endif extern int inet_add_protocol(struct net_protocol *prot, unsigned char num); +extern int inet_add_gro_protocol(struct net_gro_protocol *prot, + unsigned char protocol); extern int inet_del_protocol(struct net_protocol *prot, unsigned char num); extern void inet_register_protosw(struct inet_protosw *p); extern void inet_unregister_protosw(struct inet_protosw *p); diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index af39956..ce3ac46 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -1189,6 +1189,110 @@ out: return segs; } +static struct sk_buff **inet_gro_receive(struct sk_buff **head, + struct sk_buff *skb) +{ + struct net_gro_protocol *ops; + struct sk_buff **pp = NULL; + struct sk_buff *p; + struct iphdr *iph; + unsigned int hlen; + unsigned int off; + int flush = 1; + int proto; + int id; + + off = skb_gro_offset(skb); + hlen = off + sizeof(*iph); + iph = skb_gro_header_fast(skb, off); + if (skb_gro_header_hard(skb, hlen)) { + iph = skb_gro_header_slow(skb, hlen, off); + if (unlikely(!iph)) + goto out; + } + + proto = iph->protocol & (MAX_INET_PROTOS - 1); + + rcu_read_lock(); + ops = rcu_dereference(inet_gro_protos[proto]); + if (!ops || !ops->gro_receive) + goto out_unlock; + + if (*(u8 *)iph != 0x45) + goto out_unlock; + + if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl))) + goto out_unlock; + + flush = ntohs(iph->tot_len) != skb_gro_len(skb) || + iph->frag_off != htons(IP_DF); + id = ntohs(iph->id); + + for (p = *head; p; p = p->next) { + struct iphdr *iph2; + + if (!NAPI_GRO_CB(p)->same_flow) + continue; + + iph2 = ip_hdr(p); + + if ((iph->protocol ^ iph2->protocol) | + (iph->tos ^ iph2->tos) | + (iph->saddr ^ iph2->saddr) | + (iph->daddr ^ iph2->daddr)) { + NAPI_GRO_CB(p)->same_flow = 0; + continue; + } + + /* All fields must match except length and checksum. */ + NAPI_GRO_CB(p)->flush |= + (iph->ttl ^ iph2->ttl) | + ((u16)(ntohs(iph2->id) + NAPI_GRO_CB(p)->count) ^ id); + + NAPI_GRO_CB(p)->flush |= flush; + } + + NAPI_GRO_CB(skb)->flush |= flush; + skb_gro_pull(skb, sizeof(*iph)); + skb_set_transport_header(skb, skb_gro_offset(skb)); + + pp = ops->gro_receive(head, skb); + +out_unlock: + rcu_read_unlock(); + +out: + NAPI_GRO_CB(skb)->flush |= flush; + + return pp; +} + +static int inet_gro_complete(struct sk_buff *skb) +{ + struct net_gro_protocol *ops; + struct iphdr *iph = ip_hdr(skb); + int proto = iph->protocol & (MAX_INET_PROTOS - 1); + int err = -ENOSYS; + __be16 newlen = htons(skb->len - skb_network_offset(skb)); + + csum_replace2(&iph->check, iph->tot_len, newlen); + iph->tot_len = newlen; + + rcu_read_lock(); + ops = rcu_dereference(inet_gro_protos[proto]); + if (unlikely(!ops || !ops->gro_complete)) { + WARN_ON(1); + goto out_unlock; + } + + err = ops->gro_complete(skb); + +out_unlock: + rcu_read_unlock(); + + return err; +} + #ifdef CONFIG_IP_MULTICAST static struct net_protocol igmp_protocol = { .handler = igmp_rcv, @@ -1253,6 +1357,12 @@ static struct packet_type ip_packet_type = { .gso_segment = inet_gso_segment, }; +static struct gro_packet_type ip_gro_packet_type = { + .type = __constant_htons(ETH_P_IP), + .gro_receive = inet_gro_receive, + .gro_complete = inet_gro_complete, +}; + static int __init inet_init(void) { struct sk_buff *dummy_skb; @@ -1349,6 +1459,7 @@ static int __init inet_init(void) ipfrag_init(); dev_add_pack(&ip_packet_type); + dev_add_pack_gro(&ip_gro_packet_type); rc = 0; out: diff --git a/net/ipv4/protocol.c b/net/ipv4/protocol.c index 05f5114..f79baab 100644 --- a/net/ipv4/protocol.c +++ b/net/ipv4/protocol.c @@ -47,6 +47,7 @@ #include <linux/igmp.h> struct net_protocol *inet_protos[MAX_INET_PROTOS]; +struct net_gro_protocol *inet_gro_protos[MAX_INET_PROTOS]; static DEFINE_SPINLOCK(inet_proto_lock); /* @@ -71,6 +72,24 @@ int inet_add_protocol(struct net_protocol *prot, unsigned char protocol) return ret; } +int inet_add_gro_protocol(struct net_gro_protocol *prot, unsigned char protocol) +{ + int hash, ret; + + hash = protocol & (MAX_INET_PROTOS - 1); + + spin_lock_bh(&inet_proto_lock); + if (inet_gro_protos[hash]) { + ret = -1; + } else { + inet_gro_protos[hash] = prot; + ret = 0; + } + spin_unlock_bh(&inet_proto_lock); + + return ret; +} + /* * Remove a protocol from the hash tables. */