summaryrefslogtreecommitdiff
path: root/net/ipv6
diff options
context:
space:
mode:
authorTom Herbert <therbert@google.com>2013-01-22 09:50:39 (GMT)
committerDavid S. Miller <davem@davemloft.net>2013-01-23 18:44:01 (GMT)
commit5ba24953e9707387cce87b07f0d5fbdd03c5c11b (patch)
treec98e56f8a06f07ff585f85cbe6af8cd9c19f2ca6 /net/ipv6
parentba418fa357a7b3c9d477f4706c6c7c96ddbd1360 (diff)
downloadlinux-5ba24953e9707387cce87b07f0d5fbdd03c5c11b.tar.xz
soreuseport: TCP/IPv6 implementation
Motivation for soreuseport would be something like a web server binding to port 80 running with multiple threads, where each thread might have it's own listener socket. This could be done as an alternative to other models: 1) have one listener thread which dispatches completed connections to workers. 2) accept on a single listener socket from multiple threads. In case #1 the listener thread can easily become the bottleneck with high connection turn-over rate. In case #2, the proportion of connections accepted per thread tends to be uneven under high connection load (assuming simple event loop: while (1) { accept(); process() }, wakeup does not promote fairness among the sockets. We have seen the disproportion to be as high as 3:1 ratio between thread accepting most connections and the one accepting the fewest. With so_reusport the distribution is uniform. Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv6')
-rw-r--r--net/ipv6/inet6_connection_sock.c19
-rw-r--r--net/ipv6/inet6_hashtables.c19
-rw-r--r--net/ipv6/tcp_ipv6.c4
3 files changed, 33 insertions, 9 deletions
diff --git a/net/ipv6/inet6_connection_sock.c b/net/ipv6/inet6_connection_sock.c
index 3064785..e4297a3 100644
--- a/net/ipv6/inet6_connection_sock.c
+++ b/net/ipv6/inet6_connection_sock.c
@@ -32,6 +32,9 @@ int inet6_csk_bind_conflict(const struct sock *sk,
{
const struct sock *sk2;
const struct hlist_node *node;
+ int reuse = sk->sk_reuse;
+ int reuseport = sk->sk_reuseport;
+ int uid = sock_i_uid((struct sock *)sk);
/* We must walk the whole port owner list in this case. -DaveM */
/*
@@ -42,11 +45,17 @@ int inet6_csk_bind_conflict(const struct sock *sk,
if (sk != sk2 &&
(!sk->sk_bound_dev_if ||
!sk2->sk_bound_dev_if ||
- sk->sk_bound_dev_if == sk2->sk_bound_dev_if) &&
- (!sk->sk_reuse || !sk2->sk_reuse ||
- sk2->sk_state == TCP_LISTEN) &&
- ipv6_rcv_saddr_equal(sk, sk2))
- break;
+ sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
+ if ((!reuse || !sk2->sk_reuse ||
+ sk2->sk_state == TCP_LISTEN) &&
+ (!reuseport || !sk2->sk_reuseport ||
+ (sk2->sk_state != TCP_TIME_WAIT &&
+ !uid_eq(uid,
+ sock_i_uid((struct sock *)sk2))))) {
+ if (ipv6_rcv_saddr_equal(sk, sk2))
+ break;
+ }
+ }
}
return node != NULL;
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index dea17fd..32b4a16 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -158,25 +158,38 @@ static inline int compute_score(struct sock *sk, struct net *net,
}
struct sock *inet6_lookup_listener(struct net *net,
- struct inet_hashinfo *hashinfo, const struct in6_addr *daddr,
+ struct inet_hashinfo *hashinfo, const struct in6_addr *saddr,
+ const __be16 sport, const struct in6_addr *daddr,
const unsigned short hnum, const int dif)
{
struct sock *sk;
const struct hlist_nulls_node *node;
struct sock *result;
- int score, hiscore;
+ int score, hiscore, matches = 0, reuseport = 0;
+ u32 phash = 0;
unsigned int hash = inet_lhashfn(net, hnum);
struct inet_listen_hashbucket *ilb = &hashinfo->listening_hash[hash];
rcu_read_lock();
begin:
result = NULL;
- hiscore = -1;
+ hiscore = 0;
sk_nulls_for_each(sk, node, &ilb->head) {
score = compute_score(sk, net, hnum, daddr, dif);
if (score > hiscore) {
hiscore = score;
result = sk;
+ reuseport = sk->sk_reuseport;
+ if (reuseport) {
+ phash = inet6_ehashfn(net, daddr, hnum,
+ saddr, sport);
+ matches = 1;
+ }
+ } else if (score == hiscore && reuseport) {
+ matches++;
+ if (((u64)phash * matches) >> 32 == 0)
+ result = sk;
+ phash = next_pseudo_random32(phash);
}
}
/*
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 3701c3c..06087e5 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -834,7 +834,8 @@ static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
* no RST generated if md5 hash doesn't match.
*/
sk1 = inet6_lookup_listener(dev_net(skb_dst(skb)->dev),
- &tcp_hashinfo, &ipv6h->daddr,
+ &tcp_hashinfo, &ipv6h->saddr,
+ th->source, &ipv6h->daddr,
ntohs(th->source), inet6_iif(skb));
if (!sk1)
return;
@@ -1598,6 +1599,7 @@ do_time_wait:
struct sock *sk2;
sk2 = inet6_lookup_listener(dev_net(skb->dev), &tcp_hashinfo,
+ &ipv6_hdr(skb)->saddr, th->source,
&ipv6_hdr(skb)->daddr,
ntohs(th->dest), inet6_iif(skb));
if (sk2 != NULL) {