[INET]: Generalise tcp_tw_bucket, aka TIME_WAIT sockets
This paves the way to generalise the rest of the sock ID lookup
routines and saves some bytes in TCPv4 TIME_WAIT sockets on distro
kernels (where IPv6 is always built as a module):
[root@qemu ~]# grep tw_sock /proc/slabinfo
tw_sock_TCPv6 0 0 128 31 1
tw_sock_TCP 0 0 96 41 1
[root@qemu ~]#
Now if a protocol wants to use the TIME_WAIT generic infrastructure it
only has to set the sk_prot->twsk_obj_size field with the size of its
inet_timewait_sock derived sock and proto_register will create
sk_prot->twsk_slab, for now its only for INET sockets, but we can
introduce timewait_sock later if some non INET transport protocolo
wants to use this stuff.
Next changesets will take advantage of this new infrastructure to
generalise even more TCP code.
[acme@toy net-2.6.14]$ grep built-in /tmp/before.size /tmp/after.size
/tmp/before.size: 188646 11764 5068 205478 322a6 net/ipv4/built-in.o
/tmp/after.size: 188144 11764 5068 204976 320b0 net/ipv4/built-in.o
[acme@toy net-2.6.14]$
Tested with both IPv4 & IPv6 (::1 (localhost) & ::ffff:172.20.0.1
(qemu host)).
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index 1c4fa00..c38c637 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -14,6 +14,8 @@
#ifndef _INET_HASHTABLES_H
#define _INET_HASHTABLES_H
+#include <linux/config.h>
+
#include <linux/interrupt.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
@@ -310,4 +312,43 @@
read_unlock(&hashinfo->lhash_lock);
return sk;
}
+
+/* Socket demux engine toys. */
+#ifdef __BIG_ENDIAN
+#define INET_COMBINED_PORTS(__sport, __dport) \
+ (((__u32)(__sport) << 16) | (__u32)(__dport))
+#else /* __LITTLE_ENDIAN */
+#define INET_COMBINED_PORTS(__sport, __dport) \
+ (((__u32)(__dport) << 16) | (__u32)(__sport))
+#endif
+
+#if (BITS_PER_LONG == 64)
+#ifdef __BIG_ENDIAN
+#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
+ const __u64 __name = (((__u64)(__saddr)) << 32) | ((__u64)(__daddr));
+#else /* __LITTLE_ENDIAN */
+#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
+ const __u64 __name = (((__u64)(__daddr)) << 32) | ((__u64)(__saddr));
+#endif /* __BIG_ENDIAN */
+#define INET_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
+ (((*((__u64 *)&(inet_sk(__sk)->daddr))) == (__cookie)) && \
+ ((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports)) && \
+ (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
+#define INET_TW_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
+ (((*((__u64 *)&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) && \
+ ((*((__u32 *)&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
+ (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
+#else /* 32-bit arch */
+#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
+#define INET_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif) \
+ ((inet_sk(__sk)->daddr == (__saddr)) && \
+ (inet_sk(__sk)->rcv_saddr == (__daddr)) && \
+ ((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports)) && \
+ (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
+#define INET_TW_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif) \
+ ((inet_twsk(__sk)->tw_daddr == (__saddr)) && \
+ (inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
+ ((*((__u32 *)&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
+ (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
+#endif /* 64-bit arch */
#endif /* _INET_HASHTABLES_H */
diff --git a/include/net/inet_timewait_sock.h b/include/net/inet_timewait_sock.h
new file mode 100644
index 0000000..ce11704
--- /dev/null
+++ b/include/net/inet_timewait_sock.h
@@ -0,0 +1,142 @@
+/*
+ * 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.
+ *
+ * Definitions for a generic INET TIMEWAIT sock
+ *
+ * From code originally in net/tcp.h
+ *
+ * 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.
+ */
+#ifndef _INET_TIMEWAIT_SOCK_
+#define _INET_TIMEWAIT_SOCK_
+
+#include <linux/config.h>
+
+#include <linux/list.h>
+#include <linux/types.h>
+
+#include <net/sock.h>
+#include <net/tcp_states.h>
+
+#include <asm/atomic.h>
+
+#if (BITS_PER_LONG == 64)
+#define INET_TIMEWAIT_ADDRCMP_ALIGN_BYTES 8
+#else
+#define INET_TIMEWAIT_ADDRCMP_ALIGN_BYTES 4
+#endif
+
+struct inet_bind_bucket;
+
+/*
+ * This is a TIME_WAIT sock. It works around the memory consumption
+ * problems of sockets in such a state on heavily loaded servers, but
+ * without violating the protocol specification.
+ */
+struct inet_timewait_sock {
+ /*
+ * Now struct sock also uses sock_common, so please just
+ * don't add nothing before this first member (__tw_common) --acme
+ */
+ struct sock_common __tw_common;
+#define tw_family __tw_common.skc_family
+#define tw_state __tw_common.skc_state
+#define tw_reuse __tw_common.skc_reuse
+#define tw_bound_dev_if __tw_common.skc_bound_dev_if
+#define tw_node __tw_common.skc_node
+#define tw_bind_node __tw_common.skc_bind_node
+#define tw_refcnt __tw_common.skc_refcnt
+#define tw_prot __tw_common.skc_prot
+ volatile unsigned char tw_substate;
+ /* 3 bits hole, try to pack */
+ unsigned char tw_rcv_wscale;
+ /* Socket demultiplex comparisons on incoming packets. */
+ /* these five are in inet_sock */
+ __u16 tw_sport;
+ __u32 tw_daddr __attribute__((aligned(INET_TIMEWAIT_ADDRCMP_ALIGN_BYTES)));
+ __u32 tw_rcv_saddr;
+ __u16 tw_dport;
+ __u16 tw_num;
+ /* And these are ours. */
+ __u8 tw_ipv6only:1;
+ /* 31 bits hole, try to pack */
+ int tw_hashent;
+ int tw_timeout;
+ unsigned long tw_ttd;
+ struct inet_bind_bucket *tw_tb;
+ struct hlist_node tw_death_node;
+};
+
+static inline void inet_twsk_add_node(struct inet_timewait_sock *tw,
+ struct hlist_head *list)
+{
+ hlist_add_head(&tw->tw_node, list);
+}
+
+static inline void inet_twsk_add_bind_node(struct inet_timewait_sock *tw,
+ struct hlist_head *list)
+{
+ hlist_add_head(&tw->tw_bind_node, list);
+}
+
+static inline int inet_twsk_dead_hashed(const struct inet_timewait_sock *tw)
+{
+ return tw->tw_death_node.pprev != NULL;
+}
+
+static inline void inet_twsk_dead_node_init(struct inet_timewait_sock *tw)
+{
+ tw->tw_death_node.pprev = NULL;
+}
+
+static inline void __inet_twsk_del_dead_node(struct inet_timewait_sock *tw)
+{
+ __hlist_del(&tw->tw_death_node);
+ inet_twsk_dead_node_init(tw);
+}
+
+static inline int inet_twsk_del_dead_node(struct inet_timewait_sock *tw)
+{
+ if (inet_twsk_dead_hashed(tw)) {
+ __inet_twsk_del_dead_node(tw);
+ return 1;
+ }
+ return 0;
+}
+
+#define inet_twsk_for_each(tw, node, head) \
+ hlist_for_each_entry(tw, node, head, tw_node)
+
+#define inet_twsk_for_each_inmate(tw, node, jail) \
+ hlist_for_each_entry(tw, node, jail, tw_death_node)
+
+#define inet_twsk_for_each_inmate_safe(tw, node, safe, jail) \
+ hlist_for_each_entry_safe(tw, node, safe, jail, tw_death_node)
+
+static inline struct inet_timewait_sock *inet_twsk(const struct sock *sk)
+{
+ return (struct inet_timewait_sock *)sk;
+}
+
+static inline u32 inet_rcv_saddr(const struct sock *sk)
+{
+ return likely(sk->sk_state != TCP_TIME_WAIT) ?
+ inet_sk(sk)->rcv_saddr : inet_twsk(sk)->tw_rcv_saddr;
+}
+
+static inline void inet_twsk_put(struct inet_timewait_sock *tw)
+{
+ if (atomic_dec_and_test(&tw->tw_refcnt)) {
+#ifdef SOCK_REFCNT_DEBUG
+ printk(KERN_DEBUG "%s timewait_sock %p released\n",
+ tw->tw_prot->name, tw);
+#endif
+ kmem_cache_free(tw->tw_prot->twsk_slab, tw);
+ }
+}
+#endif /* _INET_TIMEWAIT_SOCK_ */
diff --git a/include/net/sock.h b/include/net/sock.h
index 391d00b..c902c57 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -88,6 +88,7 @@
} while(0)
struct sock;
+struct proto;
/**
* struct sock_common - minimal network layer representation of sockets
@@ -98,10 +99,11 @@
* @skc_node: main hash linkage for various protocol lookup tables
* @skc_bind_node: bind hash linkage for various protocol lookup tables
* @skc_refcnt: reference count
+ * @skc_prot: protocol handlers inside a network family
*
* This is the minimal network layer representation of sockets, the header
- * for struct sock and struct tcp_tw_bucket.
- */
+ * for struct sock and struct inet_timewait_sock.
+ */
struct sock_common {
unsigned short skc_family;
volatile unsigned char skc_state;
@@ -110,11 +112,12 @@
struct hlist_node skc_node;
struct hlist_node skc_bind_node;
atomic_t skc_refcnt;
+ struct proto *skc_prot;
};
/**
* struct sock - network layer representation of sockets
- * @__sk_common: shared layout with tcp_tw_bucket
+ * @__sk_common: shared layout with inet_timewait_sock
* @sk_shutdown: mask of %SEND_SHUTDOWN and/or %RCV_SHUTDOWN
* @sk_userlocks: %SO_SNDBUF and %SO_RCVBUF settings
* @sk_lock: synchronizer
@@ -140,7 +143,6 @@
* @sk_backlog: always used with the per-socket spinlock held
* @sk_callback_lock: used with the callbacks in the end of this struct
* @sk_error_queue: rarely used
- * @sk_prot: protocol handlers inside a network family
* @sk_prot_creator: sk_prot of original sock creator (see ipv6_setsockopt, IPV6_ADDRFORM for instance)
* @sk_err: last error
* @sk_err_soft: errors that don't cause failure but are the cause of a persistent failure not just 'timed out'
@@ -173,7 +175,7 @@
*/
struct sock {
/*
- * Now struct tcp_tw_bucket also uses sock_common, so please just
+ * Now struct inet_timewait_sock also uses sock_common, so please just
* don't add nothing before this first member (__sk_common) --acme
*/
struct sock_common __sk_common;
@@ -184,6 +186,7 @@
#define sk_node __sk_common.skc_node
#define sk_bind_node __sk_common.skc_bind_node
#define sk_refcnt __sk_common.skc_refcnt
+#define sk_prot __sk_common.skc_prot
unsigned char sk_shutdown : 2,
sk_no_check : 2,
sk_userlocks : 4;
@@ -218,7 +221,6 @@
struct sk_buff *tail;
} sk_backlog;
struct sk_buff_head sk_error_queue;
- struct proto *sk_prot;
struct proto *sk_prot_creator;
rwlock_t sk_callback_lock;
int sk_err,
@@ -557,6 +559,9 @@
kmem_cache_t *slab;
unsigned int obj_size;
+ kmem_cache_t *twsk_slab;
+ unsigned int twsk_obj_size;
+
struct request_sock_ops *rsk_prot;
struct module *owner;
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 9d026d8..cf8e664 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -38,207 +38,14 @@
#include <net/ip.h>
#include <net/tcp_states.h>
-#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
-#include <linux/ipv6.h>
-#endif
#include <linux/seq_file.h>
extern struct inet_hashinfo tcp_hashinfo;
-#if (BITS_PER_LONG == 64)
-#define TCP_ADDRCMP_ALIGN_BYTES 8
-#else
-#define TCP_ADDRCMP_ALIGN_BYTES 4
-#endif
-
-/* This is a TIME_WAIT bucket. It works around the memory consumption
- * problems of sockets in such a state on heavily loaded servers, but
- * without violating the protocol specification.
- */
-struct tcp_tw_bucket {
- /*
- * Now struct sock also uses sock_common, so please just
- * don't add nothing before this first member (__tw_common) --acme
- */
- struct sock_common __tw_common;
-#define tw_family __tw_common.skc_family
-#define tw_state __tw_common.skc_state
-#define tw_reuse __tw_common.skc_reuse
-#define tw_bound_dev_if __tw_common.skc_bound_dev_if
-#define tw_node __tw_common.skc_node
-#define tw_bind_node __tw_common.skc_bind_node
-#define tw_refcnt __tw_common.skc_refcnt
- volatile unsigned char tw_substate;
- unsigned char tw_rcv_wscale;
- __u16 tw_sport;
- /* Socket demultiplex comparisons on incoming packets. */
- /* these five are in inet_sock */
- __u32 tw_daddr
- __attribute__((aligned(TCP_ADDRCMP_ALIGN_BYTES)));
- __u32 tw_rcv_saddr;
- __u16 tw_dport;
- __u16 tw_num;
- /* And these are ours. */
- int tw_hashent;
- int tw_timeout;
- __u32 tw_rcv_nxt;
- __u32 tw_snd_nxt;
- __u32 tw_rcv_wnd;
- __u32 tw_ts_recent;
- long tw_ts_recent_stamp;
- unsigned long tw_ttd;
- struct inet_bind_bucket *tw_tb;
- struct hlist_node tw_death_node;
-#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
- struct in6_addr tw_v6_daddr;
- struct in6_addr tw_v6_rcv_saddr;
- int tw_v6_ipv6only;
-#endif
-};
-
-static __inline__ void tw_add_node(struct tcp_tw_bucket *tw,
- struct hlist_head *list)
-{
- hlist_add_head(&tw->tw_node, list);
-}
-
-static __inline__ void tw_add_bind_node(struct tcp_tw_bucket *tw,
- struct hlist_head *list)
-{
- hlist_add_head(&tw->tw_bind_node, list);
-}
-
-static inline int tw_dead_hashed(struct tcp_tw_bucket *tw)
-{
- return tw->tw_death_node.pprev != NULL;
-}
-
-static __inline__ void tw_dead_node_init(struct tcp_tw_bucket *tw)
-{
- tw->tw_death_node.pprev = NULL;
-}
-
-static __inline__ void __tw_del_dead_node(struct tcp_tw_bucket *tw)
-{
- __hlist_del(&tw->tw_death_node);
- tw_dead_node_init(tw);
-}
-
-static __inline__ int tw_del_dead_node(struct tcp_tw_bucket *tw)
-{
- if (tw_dead_hashed(tw)) {
- __tw_del_dead_node(tw);
- return 1;
- }
- return 0;
-}
-
-#define tw_for_each(tw, node, head) \
- hlist_for_each_entry(tw, node, head, tw_node)
-
-#define tw_for_each_inmate(tw, node, jail) \
- hlist_for_each_entry(tw, node, jail, tw_death_node)
-
-#define tw_for_each_inmate_safe(tw, node, safe, jail) \
- hlist_for_each_entry_safe(tw, node, safe, jail, tw_death_node)
-
-#define tcptw_sk(__sk) ((struct tcp_tw_bucket *)(__sk))
-
-static inline u32 tcp_v4_rcv_saddr(const struct sock *sk)
-{
- return likely(sk->sk_state != TCP_TIME_WAIT) ?
- inet_sk(sk)->rcv_saddr : tcptw_sk(sk)->tw_rcv_saddr;
-}
-
-#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
-static inline struct in6_addr *__tcp_v6_rcv_saddr(const struct sock *sk)
-{
- return likely(sk->sk_state != TCP_TIME_WAIT) ?
- &inet6_sk(sk)->rcv_saddr : &tcptw_sk(sk)->tw_v6_rcv_saddr;
-}
-
-static inline struct in6_addr *tcp_v6_rcv_saddr(const struct sock *sk)
-{
- return sk->sk_family == AF_INET6 ? __tcp_v6_rcv_saddr(sk) : NULL;
-}
-
-#define tcptw_sk_ipv6only(__sk) (tcptw_sk(__sk)->tw_v6_ipv6only)
-
-static inline int tcp_v6_ipv6only(const struct sock *sk)
-{
- return likely(sk->sk_state != TCP_TIME_WAIT) ?
- ipv6_only_sock(sk) : tcptw_sk_ipv6only(sk);
-}
-#else
-# define __tcp_v6_rcv_saddr(__sk) NULL
-# define tcp_v6_rcv_saddr(__sk) NULL
-# define tcptw_sk_ipv6only(__sk) 0
-# define tcp_v6_ipv6only(__sk) 0
-#endif
-
-extern kmem_cache_t *tcp_timewait_cachep;
-
-static inline void tcp_tw_put(struct tcp_tw_bucket *tw)
-{
- if (atomic_dec_and_test(&tw->tw_refcnt)) {
-#ifdef SOCK_REFCNT_DEBUG
- printk(KERN_DEBUG "tw_bucket %p released\n", tw);
-#endif
- kmem_cache_free(tcp_timewait_cachep, tw);
- }
-}
-
extern atomic_t tcp_orphan_count;
extern int tcp_tw_count;
extern void tcp_time_wait(struct sock *sk, int state, int timeo);
-extern void tcp_tw_deschedule(struct tcp_tw_bucket *tw);
-
-
-/* Socket demux engine toys. */
-#ifdef __BIG_ENDIAN
-#define TCP_COMBINED_PORTS(__sport, __dport) \
- (((__u32)(__sport)<<16) | (__u32)(__dport))
-#else /* __LITTLE_ENDIAN */
-#define TCP_COMBINED_PORTS(__sport, __dport) \
- (((__u32)(__dport)<<16) | (__u32)(__sport))
-#endif
-
-#if (BITS_PER_LONG == 64)
-#ifdef __BIG_ENDIAN
-#define TCP_V4_ADDR_COOKIE(__name, __saddr, __daddr) \
- __u64 __name = (((__u64)(__saddr))<<32)|((__u64)(__daddr));
-#else /* __LITTLE_ENDIAN */
-#define TCP_V4_ADDR_COOKIE(__name, __saddr, __daddr) \
- __u64 __name = (((__u64)(__daddr))<<32)|((__u64)(__saddr));
-#endif /* __BIG_ENDIAN */
-#define TCP_IPV4_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
- (((*((__u64 *)&(inet_sk(__sk)->daddr)))== (__cookie)) && \
- ((*((__u32 *)&(inet_sk(__sk)->dport)))== (__ports)) && \
- (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
-#define TCP_IPV4_TW_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
- (((*((__u64 *)&(tcptw_sk(__sk)->tw_daddr))) == (__cookie)) && \
- ((*((__u32 *)&(tcptw_sk(__sk)->tw_dport))) == (__ports)) && \
- (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
-#else /* 32-bit arch */
-#define TCP_V4_ADDR_COOKIE(__name, __saddr, __daddr)
-#define TCP_IPV4_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
- ((inet_sk(__sk)->daddr == (__saddr)) && \
- (inet_sk(__sk)->rcv_saddr == (__daddr)) && \
- ((*((__u32 *)&(inet_sk(__sk)->dport)))== (__ports)) && \
- (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
-#define TCP_IPV4_TW_MATCH(__sk, __cookie, __saddr, __daddr, __ports, __dif)\
- ((tcptw_sk(__sk)->tw_daddr == (__saddr)) && \
- (tcptw_sk(__sk)->tw_rcv_saddr == (__daddr)) && \
- ((*((__u32 *)&(tcptw_sk(__sk)->tw_dport))) == (__ports)) && \
- (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
-#endif /* 64-bit arch */
-
-#define TCP_IPV6_MATCH(__sk, __saddr, __daddr, __ports, __dif) \
- (((*((__u32 *)&(inet_sk(__sk)->dport)))== (__ports)) && \
- ((__sk)->sk_family == AF_INET6) && \
- ipv6_addr_equal(&inet6_sk(__sk)->daddr, (__saddr)) && \
- ipv6_addr_equal(&inet6_sk(__sk)->rcv_saddr, (__daddr)) && \
- (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
+extern void tcp_tw_deschedule(struct inet_timewait_sock *tw);
#define MAX_TCP_HEADER (128 + MAX_HEADER)
@@ -543,7 +350,7 @@
extern int tcp_v4_remember_stamp(struct sock *sk);
-extern int tcp_v4_tw_remember_stamp(struct tcp_tw_bucket *tw);
+extern int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw);
extern int tcp_sendmsg(struct kiocb *iocb, struct sock *sk,
struct msghdr *msg, size_t size);
@@ -616,10 +423,9 @@
};
-extern enum tcp_tw_status tcp_timewait_state_process(struct tcp_tw_bucket *tw,
+extern enum tcp_tw_status tcp_timewait_state_process(struct inet_timewait_sock *tw,
struct sk_buff *skb,
- struct tcphdr *th,
- unsigned len);
+ const struct tcphdr *th);
extern struct sock * tcp_check_req(struct sock *sk,struct sk_buff *skb,
struct request_sock *req,