Using a timer wheel for timewait sockets was nice ~15 years ago when memory was expensive and machines had a single processor. This does not scale, code is ugly and source of huge latencies (Typically 30 ms have been seen, cpus spinning on death_lock spinlock.) We can afford to use an extra 64 bytes per timewait sock and spread timewait load to all cpus to have better behavior. Tested: On following test, /proc/sys/net/ipv4/tcp_tw_recycle is set to 1 on the target (lpaa24) Before patch : lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0 419594 lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0 437171 While test is running, we can observe 25 or even 33 ms latencies. lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23 ... 1000 packets transmitted, 1000 received, 0% packet loss, time 20601ms rtt min/avg/max/mdev = 0.020/0.217/25.771/1.535 ms, pipe 2 lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23 ... 1000 packets transmitted, 1000 received, 0% packet loss, time 20702ms rtt min/avg/max/mdev = 0.019/0.183/33.761/1.441 ms, pipe 2 After patch : About 90% increase of throughput : lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0 810442 lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0 800992 And latencies are kept to minimal values during this load, even if network utilization is 90% higher : lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23 ... 1000 packets transmitted, 1000 received, 0% packet loss, time 19991ms rtt min/avg/max/mdev = 0.023/0.064/0.360/0.042 ms Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
		
			
				
	
	
		
			275 lines
		
	
	
	
		
			7.4 KiB
			
		
	
	
	
		
			C
		
	
	
	
	
	
			
		
		
	
	
			275 lines
		
	
	
	
		
			7.4 KiB
			
		
	
	
	
		
			C
		
	
	
	
	
	
/*
 | 
						|
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 | 
						|
 *		operating system.  INET is implemented using the BSD Socket
 | 
						|
 *		interface as the means of communication with the user level.
 | 
						|
 *
 | 
						|
 *		Generic INET6 transport hashtables
 | 
						|
 *
 | 
						|
 * Authors:	Lotsa people, from code originally in tcp, generalised here
 | 
						|
 *		by Arnaldo Carvalho de Melo <acme@mandriva.com>
 | 
						|
 *
 | 
						|
 *	This program is free software; you can redistribute it and/or
 | 
						|
 *      modify it under the terms of the GNU General Public License
 | 
						|
 *      as published by the Free Software Foundation; either version
 | 
						|
 *      2 of the License, or (at your option) any later version.
 | 
						|
 */
 | 
						|
 | 
						|
#include <linux/module.h>
 | 
						|
#include <linux/random.h>
 | 
						|
 | 
						|
#include <net/inet_connection_sock.h>
 | 
						|
#include <net/inet_hashtables.h>
 | 
						|
#include <net/inet6_hashtables.h>
 | 
						|
#include <net/secure_seq.h>
 | 
						|
#include <net/ip.h>
 | 
						|
 | 
						|
u32 inet6_ehashfn(const struct net *net,
 | 
						|
		  const struct in6_addr *laddr, const u16 lport,
 | 
						|
		  const struct in6_addr *faddr, const __be16 fport)
 | 
						|
{
 | 
						|
	static u32 inet6_ehash_secret __read_mostly;
 | 
						|
	static u32 ipv6_hash_secret __read_mostly;
 | 
						|
 | 
						|
	u32 lhash, fhash;
 | 
						|
 | 
						|
	net_get_random_once(&inet6_ehash_secret, sizeof(inet6_ehash_secret));
 | 
						|
	net_get_random_once(&ipv6_hash_secret, sizeof(ipv6_hash_secret));
 | 
						|
 | 
						|
	lhash = (__force u32)laddr->s6_addr32[3];
 | 
						|
	fhash = __ipv6_addr_jhash(faddr, ipv6_hash_secret);
 | 
						|
 | 
						|
	return __inet6_ehashfn(lhash, lport, fhash, fport,
 | 
						|
			       inet6_ehash_secret + net_hash_mix(net));
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
 | 
						|
 * we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
 | 
						|
 *
 | 
						|
 * The sockhash lock must be held as a reader here.
 | 
						|
 */
 | 
						|
struct sock *__inet6_lookup_established(struct net *net,
 | 
						|
					struct inet_hashinfo *hashinfo,
 | 
						|
					   const struct in6_addr *saddr,
 | 
						|
					   const __be16 sport,
 | 
						|
					   const struct in6_addr *daddr,
 | 
						|
					   const u16 hnum,
 | 
						|
					   const int dif)
 | 
						|
{
 | 
						|
	struct sock *sk;
 | 
						|
	const struct hlist_nulls_node *node;
 | 
						|
	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
 | 
						|
	/* Optimize here for direct hit, only listening connections can
 | 
						|
	 * have wildcards anyways.
 | 
						|
	 */
 | 
						|
	unsigned int hash = inet6_ehashfn(net, daddr, hnum, saddr, sport);
 | 
						|
	unsigned int slot = hash & hashinfo->ehash_mask;
 | 
						|
	struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
 | 
						|
 | 
						|
 | 
						|
	rcu_read_lock();
 | 
						|
begin:
 | 
						|
	sk_nulls_for_each_rcu(sk, node, &head->chain) {
 | 
						|
		if (sk->sk_hash != hash)
 | 
						|
			continue;
 | 
						|
		if (!INET6_MATCH(sk, net, saddr, daddr, ports, dif))
 | 
						|
			continue;
 | 
						|
		if (unlikely(!atomic_inc_not_zero(&sk->sk_refcnt)))
 | 
						|
			goto out;
 | 
						|
 | 
						|
		if (unlikely(!INET6_MATCH(sk, net, saddr, daddr, ports, dif))) {
 | 
						|
			sock_gen_put(sk);
 | 
						|
			goto begin;
 | 
						|
		}
 | 
						|
		goto found;
 | 
						|
	}
 | 
						|
	if (get_nulls_value(node) != slot)
 | 
						|
		goto begin;
 | 
						|
out:
 | 
						|
	sk = NULL;
 | 
						|
found:
 | 
						|
	rcu_read_unlock();
 | 
						|
	return sk;
 | 
						|
}
 | 
						|
EXPORT_SYMBOL(__inet6_lookup_established);
 | 
						|
 | 
						|
static inline int compute_score(struct sock *sk, struct net *net,
 | 
						|
				const unsigned short hnum,
 | 
						|
				const struct in6_addr *daddr,
 | 
						|
				const int dif)
 | 
						|
{
 | 
						|
	int score = -1;
 | 
						|
 | 
						|
	if (net_eq(sock_net(sk), net) && inet_sk(sk)->inet_num == hnum &&
 | 
						|
	    sk->sk_family == PF_INET6) {
 | 
						|
 | 
						|
		score = 1;
 | 
						|
		if (!ipv6_addr_any(&sk->sk_v6_rcv_saddr)) {
 | 
						|
			if (!ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr))
 | 
						|
				return -1;
 | 
						|
			score++;
 | 
						|
		}
 | 
						|
		if (sk->sk_bound_dev_if) {
 | 
						|
			if (sk->sk_bound_dev_if != dif)
 | 
						|
				return -1;
 | 
						|
			score++;
 | 
						|
		}
 | 
						|
	}
 | 
						|
	return score;
 | 
						|
}
 | 
						|
 | 
						|
struct sock *inet6_lookup_listener(struct net *net,
 | 
						|
		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, 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 = 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 (reciprocal_scale(phash, matches) == 0)
 | 
						|
				result = sk;
 | 
						|
			phash = next_pseudo_random32(phash);
 | 
						|
		}
 | 
						|
	}
 | 
						|
	/*
 | 
						|
	 * if the nulls value we got at the end of this lookup is
 | 
						|
	 * not the expected one, we must restart lookup.
 | 
						|
	 * We probably met an item that was moved to another chain.
 | 
						|
	 */
 | 
						|
	if (get_nulls_value(node) != hash + LISTENING_NULLS_BASE)
 | 
						|
		goto begin;
 | 
						|
	if (result) {
 | 
						|
		if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
 | 
						|
			result = NULL;
 | 
						|
		else if (unlikely(compute_score(result, net, hnum, daddr,
 | 
						|
				  dif) < hiscore)) {
 | 
						|
			sock_put(result);
 | 
						|
			goto begin;
 | 
						|
		}
 | 
						|
	}
 | 
						|
	rcu_read_unlock();
 | 
						|
	return result;
 | 
						|
}
 | 
						|
EXPORT_SYMBOL_GPL(inet6_lookup_listener);
 | 
						|
 | 
						|
struct sock *inet6_lookup(struct net *net, struct inet_hashinfo *hashinfo,
 | 
						|
			  const struct in6_addr *saddr, const __be16 sport,
 | 
						|
			  const struct in6_addr *daddr, const __be16 dport,
 | 
						|
			  const int dif)
 | 
						|
{
 | 
						|
	struct sock *sk;
 | 
						|
 | 
						|
	local_bh_disable();
 | 
						|
	sk = __inet6_lookup(net, hashinfo, saddr, sport, daddr, ntohs(dport), dif);
 | 
						|
	local_bh_enable();
 | 
						|
 | 
						|
	return sk;
 | 
						|
}
 | 
						|
EXPORT_SYMBOL_GPL(inet6_lookup);
 | 
						|
 | 
						|
static int __inet6_check_established(struct inet_timewait_death_row *death_row,
 | 
						|
				     struct sock *sk, const __u16 lport,
 | 
						|
				     struct inet_timewait_sock **twp)
 | 
						|
{
 | 
						|
	struct inet_hashinfo *hinfo = death_row->hashinfo;
 | 
						|
	struct inet_sock *inet = inet_sk(sk);
 | 
						|
	const struct in6_addr *daddr = &sk->sk_v6_rcv_saddr;
 | 
						|
	const struct in6_addr *saddr = &sk->sk_v6_daddr;
 | 
						|
	const int dif = sk->sk_bound_dev_if;
 | 
						|
	const __portpair ports = INET_COMBINED_PORTS(inet->inet_dport, lport);
 | 
						|
	struct net *net = sock_net(sk);
 | 
						|
	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
 | 
						|
						inet->inet_dport);
 | 
						|
	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
 | 
						|
	spinlock_t *lock = inet_ehash_lockp(hinfo, hash);
 | 
						|
	struct sock *sk2;
 | 
						|
	const struct hlist_nulls_node *node;
 | 
						|
	struct inet_timewait_sock *tw = NULL;
 | 
						|
	int twrefcnt = 0;
 | 
						|
 | 
						|
	spin_lock(lock);
 | 
						|
 | 
						|
	sk_nulls_for_each(sk2, node, &head->chain) {
 | 
						|
		if (sk2->sk_hash != hash)
 | 
						|
			continue;
 | 
						|
 | 
						|
		if (likely(INET6_MATCH(sk2, net, saddr, daddr, ports, dif))) {
 | 
						|
			if (sk2->sk_state == TCP_TIME_WAIT) {
 | 
						|
				tw = inet_twsk(sk2);
 | 
						|
				if (twsk_unique(sk, sk2, twp))
 | 
						|
					break;
 | 
						|
			}
 | 
						|
			goto not_unique;
 | 
						|
		}
 | 
						|
	}
 | 
						|
 | 
						|
	/* Must record num and sport now. Otherwise we will see
 | 
						|
	 * in hash table socket with a funny identity.
 | 
						|
	 */
 | 
						|
	inet->inet_num = lport;
 | 
						|
	inet->inet_sport = htons(lport);
 | 
						|
	sk->sk_hash = hash;
 | 
						|
	WARN_ON(!sk_unhashed(sk));
 | 
						|
	__sk_nulls_add_node_rcu(sk, &head->chain);
 | 
						|
	if (tw) {
 | 
						|
		twrefcnt = inet_twsk_unhash(tw);
 | 
						|
		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
 | 
						|
	}
 | 
						|
	spin_unlock(lock);
 | 
						|
	if (twrefcnt)
 | 
						|
		inet_twsk_put(tw);
 | 
						|
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 | 
						|
 | 
						|
	if (twp) {
 | 
						|
		*twp = tw;
 | 
						|
	} else if (tw) {
 | 
						|
		/* Silly. Should hash-dance instead... */
 | 
						|
		inet_twsk_deschedule(tw);
 | 
						|
 | 
						|
		inet_twsk_put(tw);
 | 
						|
	}
 | 
						|
	return 0;
 | 
						|
 | 
						|
not_unique:
 | 
						|
	spin_unlock(lock);
 | 
						|
	return -EADDRNOTAVAIL;
 | 
						|
}
 | 
						|
 | 
						|
static inline u32 inet6_sk_port_offset(const struct sock *sk)
 | 
						|
{
 | 
						|
	const struct inet_sock *inet = inet_sk(sk);
 | 
						|
 | 
						|
	return secure_ipv6_port_ephemeral(sk->sk_v6_rcv_saddr.s6_addr32,
 | 
						|
					  sk->sk_v6_daddr.s6_addr32,
 | 
						|
					  inet->inet_dport);
 | 
						|
}
 | 
						|
 | 
						|
int inet6_hash_connect(struct inet_timewait_death_row *death_row,
 | 
						|
		       struct sock *sk)
 | 
						|
{
 | 
						|
	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
 | 
						|
				   __inet6_check_established);
 | 
						|
}
 | 
						|
EXPORT_SYMBOL_GPL(inet6_hash_connect);
 |