diff options
author | Eric Dumazet <dada1@cosmosbay.com> | 2005-10-03 17:13:38 -0400 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2005-10-03 17:13:38 -0400 |
commit | 81c3d5470ecc70564eb9209946730fe2be93ad06 (patch) | |
tree | 1efa553c305a6453769dacfaf580bc6ccf146d82 /net/ipv4 | |
parent | 399de50bbbb2501a6db43daaa8a2dafbc9bcfe0c (diff) |
[INET]: speedup inet (tcp/dccp) lookups
Arnaldo and I agreed it could be applied now, because I have other
pending patches depending on this one (Thank you Arnaldo)
(The other important patch moves skc_refcnt in a separate cache line,
so that the SMP/NUMA performance doesnt suffer from cache line ping pongs)
1) First some performance data :
--------------------------------
tcp_v4_rcv() wastes a *lot* of time in __inet_lookup_established()
The most time critical code is :
sk_for_each(sk, node, &head->chain) {
if (INET_MATCH(sk, acookie, saddr, daddr, ports, dif))
goto hit; /* You sunk my battleship! */
}
The sk_for_each() does use prefetch() hints but only the begining of
"struct sock" is prefetched.
As INET_MATCH first comparison uses inet_sk(__sk)->daddr, wich is far
away from the begining of "struct sock", it has to bring into CPU
cache cold cache line. Each iteration has to use at least 2 cache
lines.
This can be problematic if some chains are very long.
2) The goal
-----------
The idea I had is to change things so that INET_MATCH() may return
FALSE in 99% of cases only using the data already in the CPU cache,
using one cache line per iteration.
3) Description of the patch
---------------------------
Adds a new 'unsigned int skc_hash' field in 'struct sock_common',
filling a 32 bits hole on 64 bits platform.
struct sock_common {
unsigned short skc_family;
volatile unsigned char skc_state;
unsigned char skc_reuse;
int skc_bound_dev_if;
struct hlist_node skc_node;
struct hlist_node skc_bind_node;
atomic_t skc_refcnt;
+ unsigned int skc_hash;
struct proto *skc_prot;
};
Store in this 32 bits field the full hash, not masked by (ehash_size -
1) Using this full hash as the first comparison done in INET_MATCH
permits us immediatly skip the element without touching a second cache
line in case of a miss.
Suppress the sk_hashent/tw_hashent fields since skc_hash (aliased to
sk_hash and tw_hash) already contains the slot number if we mask with
(ehash_size - 1)
File include/net/inet_hashtables.h
64 bits platforms :
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash))
((*((__u64 *)&(inet_sk(__sk)->daddr)))== (__cookie)) && \
((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
32bits platforms:
#define TCP_IPV4_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash)) && \
(inet_sk(__sk)->daddr == (__saddr)) && \
(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
- Adds a prefetch(head->chain.first) in
__inet_lookup_established()/__tcp_v4_check_established() and
__inet6_lookup_established()/__tcp_v6_check_established() and
__dccp_v4_check_established() to bring into cache the first element of the
list, before the {read|write}_lock(&head->lock);
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4')
-rw-r--r-- | net/ipv4/inet_timewait_sock.c | 6 | ||||
-rw-r--r-- | net/ipv4/tcp_ipv4.c | 11 |
2 files changed, 9 insertions, 8 deletions
diff --git a/net/ipv4/inet_timewait_sock.c b/net/ipv4/inet_timewait_sock.c index 4d1502a49852..f9076ef3a1a8 100644 --- a/net/ipv4/inet_timewait_sock.c +++ b/net/ipv4/inet_timewait_sock.c | |||
@@ -20,7 +20,7 @@ void __inet_twsk_kill(struct inet_timewait_sock *tw, struct inet_hashinfo *hashi | |||
20 | struct inet_bind_hashbucket *bhead; | 20 | struct inet_bind_hashbucket *bhead; |
21 | struct inet_bind_bucket *tb; | 21 | struct inet_bind_bucket *tb; |
22 | /* Unlink from established hashes. */ | 22 | /* Unlink from established hashes. */ |
23 | struct inet_ehash_bucket *ehead = &hashinfo->ehash[tw->tw_hashent]; | 23 | struct inet_ehash_bucket *ehead = inet_ehash_bucket(hashinfo, tw->tw_hash); |
24 | 24 | ||
25 | write_lock(&ehead->lock); | 25 | write_lock(&ehead->lock); |
26 | if (hlist_unhashed(&tw->tw_node)) { | 26 | if (hlist_unhashed(&tw->tw_node)) { |
@@ -60,7 +60,7 @@ void __inet_twsk_hashdance(struct inet_timewait_sock *tw, struct sock *sk, | |||
60 | { | 60 | { |
61 | const struct inet_sock *inet = inet_sk(sk); | 61 | const struct inet_sock *inet = inet_sk(sk); |
62 | const struct inet_connection_sock *icsk = inet_csk(sk); | 62 | const struct inet_connection_sock *icsk = inet_csk(sk); |
63 | struct inet_ehash_bucket *ehead = &hashinfo->ehash[sk->sk_hashent]; | 63 | struct inet_ehash_bucket *ehead = inet_ehash_bucket(hashinfo, sk->sk_hash); |
64 | struct inet_bind_hashbucket *bhead; | 64 | struct inet_bind_hashbucket *bhead; |
65 | /* Step 1: Put TW into bind hash. Original socket stays there too. | 65 | /* Step 1: Put TW into bind hash. Original socket stays there too. |
66 | Note, that any socket with inet->num != 0 MUST be bound in | 66 | Note, that any socket with inet->num != 0 MUST be bound in |
@@ -106,7 +106,7 @@ struct inet_timewait_sock *inet_twsk_alloc(const struct sock *sk, const int stat | |||
106 | tw->tw_dport = inet->dport; | 106 | tw->tw_dport = inet->dport; |
107 | tw->tw_family = sk->sk_family; | 107 | tw->tw_family = sk->sk_family; |
108 | tw->tw_reuse = sk->sk_reuse; | 108 | tw->tw_reuse = sk->sk_reuse; |
109 | tw->tw_hashent = sk->sk_hashent; | 109 | tw->tw_hash = sk->sk_hash; |
110 | tw->tw_ipv6only = 0; | 110 | tw->tw_ipv6only = 0; |
111 | tw->tw_prot = sk->sk_prot_creator; | 111 | tw->tw_prot = sk->sk_prot_creator; |
112 | atomic_set(&tw->tw_refcnt, 1); | 112 | atomic_set(&tw->tw_refcnt, 1); |
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 13dfb391cdf1..c85819d8474b 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c | |||
@@ -130,19 +130,20 @@ static int __tcp_v4_check_established(struct sock *sk, __u16 lport, | |||
130 | int dif = sk->sk_bound_dev_if; | 130 | int dif = sk->sk_bound_dev_if; |
131 | INET_ADDR_COOKIE(acookie, saddr, daddr) | 131 | INET_ADDR_COOKIE(acookie, saddr, daddr) |
132 | const __u32 ports = INET_COMBINED_PORTS(inet->dport, lport); | 132 | const __u32 ports = INET_COMBINED_PORTS(inet->dport, lport); |
133 | const int hash = inet_ehashfn(daddr, lport, saddr, inet->dport, tcp_hashinfo.ehash_size); | 133 | unsigned int hash = inet_ehashfn(daddr, lport, saddr, inet->dport); |
134 | struct inet_ehash_bucket *head = &tcp_hashinfo.ehash[hash]; | 134 | struct inet_ehash_bucket *head = inet_ehash_bucket(&tcp_hashinfo, hash); |
135 | struct sock *sk2; | 135 | struct sock *sk2; |
136 | const struct hlist_node *node; | 136 | const struct hlist_node *node; |
137 | struct inet_timewait_sock *tw; | 137 | struct inet_timewait_sock *tw; |
138 | 138 | ||
139 | prefetch(head->chain.first); | ||
139 | write_lock(&head->lock); | 140 | write_lock(&head->lock); |
140 | 141 | ||
141 | /* Check TIME-WAIT sockets first. */ | 142 | /* Check TIME-WAIT sockets first. */ |
142 | sk_for_each(sk2, node, &(head + tcp_hashinfo.ehash_size)->chain) { | 143 | sk_for_each(sk2, node, &(head + tcp_hashinfo.ehash_size)->chain) { |
143 | tw = inet_twsk(sk2); | 144 | tw = inet_twsk(sk2); |
144 | 145 | ||
145 | if (INET_TW_MATCH(sk2, acookie, saddr, daddr, ports, dif)) { | 146 | if (INET_TW_MATCH(sk2, hash, acookie, saddr, daddr, ports, dif)) { |
146 | const struct tcp_timewait_sock *tcptw = tcp_twsk(sk2); | 147 | const struct tcp_timewait_sock *tcptw = tcp_twsk(sk2); |
147 | struct tcp_sock *tp = tcp_sk(sk); | 148 | struct tcp_sock *tp = tcp_sk(sk); |
148 | 149 | ||
@@ -179,7 +180,7 @@ static int __tcp_v4_check_established(struct sock *sk, __u16 lport, | |||
179 | 180 | ||
180 | /* And established part... */ | 181 | /* And established part... */ |
181 | sk_for_each(sk2, node, &head->chain) { | 182 | sk_for_each(sk2, node, &head->chain) { |
182 | if (INET_MATCH(sk2, acookie, saddr, daddr, ports, dif)) | 183 | if (INET_MATCH(sk2, hash, acookie, saddr, daddr, ports, dif)) |
183 | goto not_unique; | 184 | goto not_unique; |
184 | } | 185 | } |
185 | 186 | ||
@@ -188,7 +189,7 @@ unique: | |||
188 | * in hash table socket with a funny identity. */ | 189 | * in hash table socket with a funny identity. */ |
189 | inet->num = lport; | 190 | inet->num = lport; |
190 | inet->sport = htons(lport); | 191 | inet->sport = htons(lport); |
191 | sk->sk_hashent = hash; | 192 | sk->sk_hash = hash; |
192 | BUG_TRAP(sk_unhashed(sk)); | 193 | BUG_TRAP(sk_unhashed(sk)); |
193 | __sk_add_node(sk, &head->chain); | 194 | __sk_add_node(sk, &head->chain); |
194 | sock_prot_inc_use(sk->sk_prot); | 195 | sock_prot_inc_use(sk->sk_prot); |