diff options
| -rw-r--r-- | include/net/inet_common.h | 5 | ||||
| -rw-r--r-- | include/net/inet_hashtables.h | 122 | ||||
| -rw-r--r-- | include/net/tcp.h | 120 | ||||
| -rw-r--r-- | net/ipv4/Makefile | 2 | ||||
| -rw-r--r-- | net/ipv4/inet_hashtables.c | 51 | ||||
| -rw-r--r-- | net/ipv4/tcp_ipv4.c | 26 |
6 files changed, 181 insertions, 145 deletions
diff --git a/include/net/inet_common.h b/include/net/inet_common.h index 1fbd94d..f943306 100644 --- a/include/net/inet_common.h +++ b/include/net/inet_common.h @@ -8,6 +8,11 @@ extern struct proto_ops inet_dgram_ops; * INET4 prototypes used by INET6 */ +struct msghdr; +struct sock; +struct sockaddr; +struct socket; + extern void inet_remove_sock(struct sock *sk1); extern void inet_put_sock(unsigned short num, struct sock *sk); diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h index c4c9e39..3a6c11c 100644 --- a/include/net/inet_hashtables.h +++ b/include/net/inet_hashtables.h @@ -14,8 +14,107 @@ #ifndef _INET_HASHTABLES_H #define _INET_HASHTABLES_H +#include <linux/ip.h> +#include <linux/list.h> +#include <linux/slab.h> +#include <linux/spinlock.h> #include <linux/types.h> +/* This is for all connections with a full identity, no wildcards. + * New scheme, half the table is for TIME_WAIT, the other half is + * for the rest. I'll experiment with dynamic table growth later. + */ +struct inet_ehash_bucket { + rwlock_t lock; + struct hlist_head chain; +} __attribute__((__aligned__(8))); + +/* There are a few simple rules, which allow for local port reuse by + * an application. In essence: + * + * 1) Sockets bound to different interfaces may share a local port. + * Failing that, goto test 2. + * 2) If all sockets have sk->sk_reuse set, and none of them are in + * TCP_LISTEN state, the port may be shared. + * Failing that, goto test 3. + * 3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local + * address, and none of them are the same, the port may be + * shared. + * Failing this, the port cannot be shared. + * + * The interesting point, is test #2. This is what an FTP server does + * all day. To optimize this case we use a specific flag bit defined + * below. As we add sockets to a bind bucket list, we perform a + * check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN)) + * As long as all sockets added to a bind bucket pass this test, + * the flag bit will be set. + * The resulting situation is that tcp_v[46]_verify_bind() can just check + * for this flag bit, if it is set and the socket trying to bind has + * sk->sk_reuse set, we don't even have to walk the owners list at all, + * we return that it is ok to bind this socket to the requested local port. + * + * Sounds like a lot of work, but it is worth it. In a more naive + * implementation (ie. current FreeBSD etc.) the entire list of ports + * must be walked for each data port opened by an ftp server. Needless + * to say, this does not scale at all. With a couple thousand FTP + * users logged onto your box, isn't it nice to know that new data + * ports are created in O(1) time? I thought so. ;-) -DaveM + */ +struct inet_bind_bucket { + unsigned short port; + signed short fastreuse; + struct hlist_node node; + struct hlist_head owners; +}; + +#define inet_bind_bucket_for_each(tb, node, head) \ + hlist_for_each_entry(tb, node, head, node) + +struct inet_bind_hashbucket { + spinlock_t lock; + struct hlist_head chain; +}; + +/* This is for listening sockets, thus all sockets which possess wildcards. */ +#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */ + +struct inet_hashinfo { + /* This is for sockets with full identity only. Sockets here will + * always be without wildcards and will have the following invariant: + * + * TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE + * + * First half of the table is for sockets not in TIME_WAIT, second half + * is for TIME_WAIT sockets only. + */ + struct inet_ehash_bucket *ehash; + + /* Ok, let's try this, I give up, we do need a local binding + * TCP hash as well as the others for fast bind/connect. + */ + struct inet_bind_hashbucket *bhash; + + int bhash_size; + int ehash_size; + + /* All sockets in TCP_LISTEN state will be in here. This is the only + * table where wildcard'd TCP sockets can exist. Hash function here + * is just local port number. + */ + struct hlist_head listening_hash[INET_LHTABLE_SIZE]; + + /* All the above members are written once at bootup and + * never written again _or_ are predominantly read-access. + * + * Now align to a new cache line as all the following members + * are often dirty. + */ + rwlock_t lhash_lock ____cacheline_aligned; + atomic_t lhash_users; + wait_queue_head_t lhash_wait; + spinlock_t portalloc_lock; +}; + static inline int inet_ehashfn(const __u32 laddr, const __u16 lport, const __u32 faddr, const __u16 fport, const int ehash_size) @@ -37,4 +136,27 @@ static inline int inet_sk_ehashfn(const struct sock *sk, const int ehash_size) return inet_ehashfn(laddr, lport, faddr, fport, ehash_size); } +extern struct inet_bind_bucket * + inet_bind_bucket_create(kmem_cache_t *cachep, + struct inet_bind_hashbucket *head, + const unsigned short snum); +extern void inet_bind_bucket_destroy(kmem_cache_t *cachep, + struct inet_bind_bucket *tb); + +static inline int inet_bhashfn(const __u16 lport, const int bhash_size) +{ + return lport & (bhash_size - 1); +} + +/* These can have wildcards, don't try too hard. */ +static inline int inet_lhashfn(const unsigned short num) +{ + return num & (INET_LHTABLE_SIZE - 1); +} + +static inline int inet_sk_listen_hashfn(const struct sock *sk) +{ + return inet_lhashfn(inet_sk(sk)->num); +} + #endif /* _INET_HASHTABLES_H */ diff --git a/include/net/tcp.h b/include/net/tcp.h index 6c9f6f7..ff5d30a 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -30,6 +30,7 @@ #include <linux/slab.h> #include <linux/cache.h> #include <linux/percpu.h> +#include <net/inet_hashtables.h> #include <net/checksum.h> #include <net/request_sock.h> #include <net/sock.h> @@ -40,101 +41,6 @@ #endif #include <linux/seq_file.h> -/* This is for all connections with a full identity, no wildcards. - * New scheme, half the table is for TIME_WAIT, the other half is - * for the rest. I'll experiment with dynamic table growth later. - */ -struct inet_ehash_bucket { - rwlock_t lock; - struct hlist_head chain; -} __attribute__((__aligned__(8))); - -/* This is for listening sockets, thus all sockets which possess wildcards. */ -#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */ - -/* There are a few simple rules, which allow for local port reuse by - * an application. In essence: - * - * 1) Sockets bound to different interfaces may share a local port. - * Failing that, goto test 2. - * 2) If all sockets have sk->sk_reuse set, and none of them are in - * TCP_LISTEN state, the port may be shared. - * Failing that, goto test 3. - * 3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local - * address, and none of them are the same, the port may be - * shared. - * Failing this, the port cannot be shared. - * - * The interesting point, is test #2. This is what an FTP server does - * all day. To optimize this case we use a specific flag bit defined - * below. As we add sockets to a bind bucket list, we perform a - * check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN)) - * As long as all sockets added to a bind bucket pass this test, - * the flag bit will be set. - * The resulting situation is that tcp_v[46]_verify_bind() can just check - * for this flag bit, if it is set and the socket trying to bind has - * sk->sk_reuse set, we don't even have to walk the owners list at all, - * we return that it is ok to bind this socket to the requested local port. - * - * Sounds like a lot of work, but it is worth it. In a more naive - * implementation (ie. current FreeBSD etc.) the entire list of ports - * must be walked for each data port opened by an ftp server. Needless - * to say, this does not scale at all. With a couple thousand FTP - * users logged onto your box, isn't it nice to know that new data - * ports are created in O(1) time? I thought so. ;-) -DaveM - */ -struct inet_bind_bucket { - unsigned short port; - signed short fastreuse; - struct hlist_node node; - struct hlist_head owners; -}; - -#define inet_bind_bucket_for_each(tb, node, head) \ - hlist_for_each_entry(tb, node, head, node) - -struct inet_bind_hashbucket { - spinlock_t lock; - struct hlist_head chain; -}; - -struct inet_hashinfo { - /* This is for sockets with full identity only. Sockets here will - * always be without wildcards and will have the following invariant: - * - * TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE - * - * First half of the table is for sockets not in TIME_WAIT, second half - * is for TIME_WAIT sockets only. - */ - struct inet_ehash_bucket *ehash; - - /* Ok, let's try this, I give up, we do need a local binding - * TCP hash as well as the others for fast bind/connect. - */ - struct inet_bind_hashbucket *bhash; - - int bhash_size; - int ehash_size; - - /* All sockets in TCP_LISTEN state will be in here. This is the only - * table where wildcard'd TCP sockets can exist. Hash function here - * is just local port number. - */ - struct hlist_head listening_hash[INET_LHTABLE_SIZE]; - - /* All the above members are written once at bootup and - * never written again _or_ are predominantly read-access. - * - * Now align to a new cache line as all the following members - * are often dirty. - */ - rwlock_t lhash_lock ____cacheline_aligned; - atomic_t lhash_users; - wait_queue_head_t lhash_wait; - spinlock_t portalloc_lock; -}; - extern struct inet_hashinfo tcp_hashinfo; #define tcp_ehash (tcp_hashinfo.ehash) #define tcp_bhash (tcp_hashinfo.bhash) @@ -147,19 +53,8 @@ extern struct inet_hashinfo tcp_hashinfo; #define tcp_portalloc_lock (tcp_hashinfo.portalloc_lock) extern kmem_cache_t *tcp_bucket_cachep; -extern struct inet_bind_bucket * - inet_bind_bucket_create(kmem_cache_t *cachep, - struct inet_bind_hashbucket *head, - const unsigned short snum); -extern void inet_bind_bucket_destroy(kmem_cache_t *cachep, - struct inet_bind_bucket *tb); -extern int tcp_port_rover; -/* These are AF independent. */ -static inline int inet_bhashfn(const __u16 lport, const int bhash_size) -{ - return lport & (bhash_size - 1); -} +extern int tcp_port_rover; extern void tcp_bind_hash(struct sock *sk, struct inet_bind_bucket *tb, unsigned short snum); @@ -359,17 +254,6 @@ extern void tcp_tw_deschedule(struct tcp_tw_bucket *tw); ipv6_addr_equal(&inet6_sk(__sk)->rcv_saddr, (__daddr)) && \ (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif)))) -/* These can have wildcards, don't try too hard. */ -static inline int inet_lhashfn(const unsigned short num) -{ - return num & (INET_LHTABLE_SIZE - 1); -} - -static inline int inet_sk_listen_hashfn(const struct sock *sk) -{ - return inet_lhashfn(inet_sk(sk)->num); -} - #define MAX_TCP_HEADER (128 + MAX_HEADER) /* diff --git a/net/ipv4/Makefile b/net/ipv4/Makefile index 61c7386..2d8d30e 100644 --- a/net/ipv4/Makefile +++ b/net/ipv4/Makefile @@ -4,7 +4,7 @@ obj-y := route.o inetpeer.o protocol.o \ ip_input.o ip_fragment.o ip_forward.o ip_options.o \ - ip_output.o ip_sockglue.o \ + ip_output.o ip_sockglue.o inet_hashtables.o \ tcp.o tcp_input.o tcp_output.o tcp_timer.o tcp_ipv4.o \ tcp_minisocks.o tcp_cong.o \ datagram.o raw.o udp.o arp.o icmp.o devinet.o af_inet.o igmp.o \ diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c new file mode 100644 index 0000000..343a890 --- /dev/null +++ b/net/ipv4/inet_hashtables.c @@ -0,0 +1,51 @@ +/* + * 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 INET transport hashtables + * + * Authors: Lotsa people, from code originally in tcp + * + * 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/config.h> +#include <linux/slab.h> + +#include <net/inet_hashtables.h> + +/* + * Allocate and initialize a new local port bind bucket. + * The bindhash mutex for snum's hash chain must be held here. + */ +struct inet_bind_bucket *inet_bind_bucket_create(kmem_cache_t *cachep, + struct inet_bind_hashbucket *head, + const unsigned short snum) +{ + struct inet_bind_bucket *tb = kmem_cache_alloc(cachep, SLAB_ATOMIC); + + if (tb != NULL) { + tb->port = snum; + tb->fastreuse = 0; + INIT_HLIST_HEAD(&tb->owners); + hlist_add_head(&tb->node, &head->chain); + } + return tb; +} + +EXPORT_SYMBOL(inet_bind_bucket_create); + +/* + * Caller must hold hashbucket lock for this tb with local BH disabled + */ +void inet_bind_bucket_destroy(kmem_cache_t *cachep, struct inet_bind_bucket *tb) +{ + if (hlist_empty(&tb->owners)) { + __hlist_del(&tb->node); + kmem_cache_free(cachep, tb); + } +} diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 4138630..58e36ed 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -104,32 +104,6 @@ struct inet_hashinfo __cacheline_aligned tcp_hashinfo = { int sysctl_local_port_range[2] = { 1024, 4999 }; int tcp_port_rover = 1024 - 1; -/* Allocate and initialize a new local port bind bucket. - * The bindhash mutex for snum's hash chain must be held here. - */ -struct inet_bind_bucket *inet_bind_bucket_create(kmem_cache_t *cachep, - struct inet_bind_hashbucket *head, - const unsigned short snum) -{ - struct inet_bind_bucket *tb = kmem_cache_alloc(cachep, SLAB_ATOMIC); - if (tb) { - tb->port = snum; - tb->fastreuse = 0; - INIT_HLIST_HEAD(&tb->owners); - hlist_add_head(&tb->node, &head->chain); - } - return tb; -} - -/* Caller must hold hashbucket lock for this tb with local BH disabled */ -void inet_bind_bucket_destroy(kmem_cache_t *cachep, struct inet_bind_bucket *tb) -{ - if (hlist_empty(&tb->owners)) { - __hlist_del(&tb->node); - kmem_cache_free(cachep, tb); - } -} - /* Caller must disable local BH processing. */ static __inline__ void __tcp_inherit_port(struct sock *sk, struct sock *child) { |