* ACK bit.
* Andi Kleen : Implemented fast path mtu discovery.
* Fixed many serious bugs in the
- * open_request handling and moved
+ * request_sock handling and moved
* most of it into the af independent code.
* Added tail drop and some other bugfixes.
* Added new listen sematics.
#include <linux/times.h>
#include <net/icmp.h>
+#include <net/inet_hashtables.h>
#include <net/tcp.h>
#include <net/ipv6.h>
#include <net/inet_common.h>
void tcp_v4_send_check(struct sock *sk, struct tcphdr *th, int len,
struct sk_buff *skb);
-struct tcp_hashinfo __cacheline_aligned tcp_hashinfo = {
- .__tcp_lhash_lock = RW_LOCK_UNLOCKED,
- .__tcp_lhash_users = ATOMIC_INIT(0),
- .__tcp_lhash_wait
- = __WAIT_QUEUE_HEAD_INITIALIZER(tcp_hashinfo.__tcp_lhash_wait),
- .__tcp_portalloc_lock = SPIN_LOCK_UNLOCKED
+struct inet_hashinfo __cacheline_aligned tcp_hashinfo = {
+ .lhash_lock = RW_LOCK_UNLOCKED,
+ .lhash_users = ATOMIC_INIT(0),
+ .lhash_wait = __WAIT_QUEUE_HEAD_INITIALIZER(tcp_hashinfo.lhash_wait),
+ .portalloc_lock = SPIN_LOCK_UNLOCKED,
+ .port_rover = 1024 - 1,
};
/*
* 32768-61000
*/
int sysctl_local_port_range[2] = { 1024, 4999 };
-int tcp_port_rover = 1024 - 1;
-static __inline__ int tcp_hashfn(__u32 laddr, __u16 lport,
- __u32 faddr, __u16 fport)
-{
- int h = (laddr ^ lport) ^ (faddr ^ fport);
- h ^= h >> 16;
- h ^= h >> 8;
- return h & (tcp_ehash_size - 1);
-}
-
-static __inline__ int tcp_sk_hashfn(struct sock *sk)
-{
- struct inet_sock *inet = inet_sk(sk);
- __u32 laddr = inet->rcv_saddr;
- __u16 lport = inet->num;
- __u32 faddr = inet->daddr;
- __u16 fport = inet->dport;
-
- return tcp_hashfn(laddr, lport, faddr, fport);
-}
-
-/* Allocate and initialize a new TCP local port bind bucket.
- * The bindhash mutex for snum's hash chain must be held here.
- */
-struct tcp_bind_bucket *tcp_bucket_create(struct tcp_bind_hashbucket *head,
- unsigned short snum)
-{
- struct tcp_bind_bucket *tb = kmem_cache_alloc(tcp_bucket_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 tcp_bucket_destroy(struct tcp_bind_bucket *tb)
-{
- if (hlist_empty(&tb->owners)) {
- __hlist_del(&tb->node);
- kmem_cache_free(tcp_bucket_cachep, tb);
- }
-}
-
-/* Caller must disable local BH processing. */
-static __inline__ void __tcp_inherit_port(struct sock *sk, struct sock *child)
-{
- struct tcp_bind_hashbucket *head =
- &tcp_bhash[tcp_bhashfn(inet_sk(child)->num)];
- struct tcp_bind_bucket *tb;
-
- spin_lock(&head->lock);
- tb = tcp_sk(sk)->bind_hash;
- sk_add_bind_node(child, &tb->owners);
- tcp_sk(child)->bind_hash = tb;
- spin_unlock(&head->lock);
-}
-
-inline void tcp_inherit_port(struct sock *sk, struct sock *child)
-{
- local_bh_disable();
- __tcp_inherit_port(sk, child);
- local_bh_enable();
-}
-
-void tcp_bind_hash(struct sock *sk, struct tcp_bind_bucket *tb,
- unsigned short snum)
-{
- inet_sk(sk)->num = snum;
- sk_add_bind_node(sk, &tb->owners);
- tcp_sk(sk)->bind_hash = tb;
-}
-
-static inline int tcp_bind_conflict(struct sock *sk, struct tcp_bind_bucket *tb)
+static inline int tcp_bind_conflict(struct sock *sk, struct inet_bind_bucket *tb)
{
const u32 sk_rcv_saddr = tcp_v4_rcv_saddr(sk);
struct sock *sk2;
*/
static int tcp_v4_get_port(struct sock *sk, unsigned short snum)
{
- struct tcp_bind_hashbucket *head;
+ struct inet_bind_hashbucket *head;
struct hlist_node *node;
- struct tcp_bind_bucket *tb;
+ struct inet_bind_bucket *tb;
int ret;
local_bh_disable();
int remaining = (high - low) + 1;
int rover;
- spin_lock(&tcp_portalloc_lock);
- if (tcp_port_rover < low)
+ spin_lock(&tcp_hashinfo.portalloc_lock);
+ if (tcp_hashinfo.port_rover < low)
rover = low;
else
- rover = tcp_port_rover;
+ rover = tcp_hashinfo.port_rover;
do {
rover++;
if (rover > high)
rover = low;
- head = &tcp_bhash[tcp_bhashfn(rover)];
+ head = &tcp_hashinfo.bhash[inet_bhashfn(rover, tcp_hashinfo.bhash_size)];
spin_lock(&head->lock);
- tb_for_each(tb, node, &head->chain)
+ inet_bind_bucket_for_each(tb, node, &head->chain)
if (tb->port == rover)
goto next;
break;
next:
spin_unlock(&head->lock);
} while (--remaining > 0);
- tcp_port_rover = rover;
- spin_unlock(&tcp_portalloc_lock);
-
- /* Exhausted local port range during search? */
+ tcp_hashinfo.port_rover = rover;
+ spin_unlock(&tcp_hashinfo.portalloc_lock);
+
+ /* Exhausted local port range during search? It is not
+ * possible for us to be holding one of the bind hash
+ * locks if this test triggers, because if 'remaining'
+ * drops to zero, we broke out of the do/while loop at
+ * the top level, not from the 'break;' statement.
+ */
ret = 1;
- if (remaining <= 0)
+ if (unlikely(remaining <= 0))
goto fail;
/* OK, here is the one we will use. HEAD is
*/
snum = rover;
} else {
- head = &tcp_bhash[tcp_bhashfn(snum)];
+ head = &tcp_hashinfo.bhash[inet_bhashfn(snum, tcp_hashinfo.bhash_size)];
spin_lock(&head->lock);
- tb_for_each(tb, node, &head->chain)
+ inet_bind_bucket_for_each(tb, node, &head->chain)
if (tb->port == snum)
goto tb_found;
}
}
tb_not_found:
ret = 1;
- if (!tb && (tb = tcp_bucket_create(head, snum)) == NULL)
+ if (!tb && (tb = inet_bind_bucket_create(tcp_hashinfo.bind_bucket_cachep, head, snum)) == NULL)
goto fail_unlock;
if (hlist_empty(&tb->owners)) {
if (sk->sk_reuse && sk->sk_state != TCP_LISTEN)
(!sk->sk_reuse || sk->sk_state == TCP_LISTEN))
tb->fastreuse = 0;
success:
- if (!tcp_sk(sk)->bind_hash)
- tcp_bind_hash(sk, tb, snum);
- BUG_TRAP(tcp_sk(sk)->bind_hash == tb);
+ if (!inet_sk(sk)->bind_hash)
+ inet_bind_hash(sk, tb, snum);
+ BUG_TRAP(inet_sk(sk)->bind_hash == tb);
ret = 0;
fail_unlock:
return ret;
}
-/* Get rid of any references to a local port held by the
- * given sock.
- */
-static void __tcp_put_port(struct sock *sk)
-{
- struct inet_sock *inet = inet_sk(sk);
- struct tcp_bind_hashbucket *head = &tcp_bhash[tcp_bhashfn(inet->num)];
- struct tcp_bind_bucket *tb;
-
- spin_lock(&head->lock);
- tb = tcp_sk(sk)->bind_hash;
- __sk_del_bind_node(sk);
- tcp_sk(sk)->bind_hash = NULL;
- inet->num = 0;
- tcp_bucket_destroy(tb);
- spin_unlock(&head->lock);
-}
-
-void tcp_put_port(struct sock *sk)
-{
- local_bh_disable();
- __tcp_put_port(sk);
- local_bh_enable();
-}
-
-/* This lock without WQ_FLAG_EXCLUSIVE is good on UP and it can be very bad on SMP.
- * Look, when several writers sleep and reader wakes them up, all but one
- * immediately hit write lock and grab all the cpus. Exclusive sleep solves
- * this, _but_ remember, it adds useless work on UP machines (wake up each
- * exclusive lock release). It should be ifdefed really.
- */
-
-void tcp_listen_wlock(void)
-{
- write_lock(&tcp_lhash_lock);
-
- if (atomic_read(&tcp_lhash_users)) {
- DEFINE_WAIT(wait);
-
- for (;;) {
- prepare_to_wait_exclusive(&tcp_lhash_wait,
- &wait, TASK_UNINTERRUPTIBLE);
- if (!atomic_read(&tcp_lhash_users))
- break;
- write_unlock_bh(&tcp_lhash_lock);
- schedule();
- write_lock_bh(&tcp_lhash_lock);
- }
-
- finish_wait(&tcp_lhash_wait, &wait);
- }
-}
-
-static __inline__ void __tcp_v4_hash(struct sock *sk, const int listen_possible)
-{
- struct hlist_head *list;
- rwlock_t *lock;
-
- BUG_TRAP(sk_unhashed(sk));
- if (listen_possible && sk->sk_state == TCP_LISTEN) {
- list = &tcp_listening_hash[tcp_sk_listen_hashfn(sk)];
- lock = &tcp_lhash_lock;
- tcp_listen_wlock();
- } else {
- list = &tcp_ehash[(sk->sk_hashent = tcp_sk_hashfn(sk))].chain;
- lock = &tcp_ehash[sk->sk_hashent].lock;
- write_lock(lock);
- }
- __sk_add_node(sk, list);
- sock_prot_inc_use(sk->sk_prot);
- write_unlock(lock);
- if (listen_possible && sk->sk_state == TCP_LISTEN)
- wake_up(&tcp_lhash_wait);
-}
-
static void tcp_v4_hash(struct sock *sk)
{
if (sk->sk_state != TCP_CLOSE) {
local_bh_disable();
- __tcp_v4_hash(sk, 1);
+ __inet_hash(&tcp_hashinfo, sk, 1);
local_bh_enable();
}
}
if (sk->sk_state == TCP_LISTEN) {
local_bh_disable();
- tcp_listen_wlock();
- lock = &tcp_lhash_lock;
+ inet_listen_wlock(&tcp_hashinfo);
+ lock = &tcp_hashinfo.lhash_lock;
} else {
- struct tcp_ehash_bucket *head = &tcp_ehash[sk->sk_hashent];
+ struct inet_ehash_bucket *head = &tcp_hashinfo.ehash[sk->sk_hashent];
lock = &head->lock;
write_lock_bh(&head->lock);
}
ende:
if (sk->sk_state == TCP_LISTEN)
- wake_up(&tcp_lhash_wait);
+ wake_up(&tcp_hashinfo.lhash_wait);
}
/* Don't inline this cruft. Here are some nice properties to
* connection. So always assume those are both wildcarded
* during the search since they can never be otherwise.
*/
-static struct sock *__tcp_v4_lookup_listener(struct hlist_head *head, u32 daddr,
- unsigned short hnum, int dif)
+static struct sock *__tcp_v4_lookup_listener(struct hlist_head *head,
+ const u32 daddr,
+ const unsigned short hnum,
+ const int dif)
{
struct sock *result = NULL, *sk;
struct hlist_node *node;
}
/* Optimize the common listener case. */
-static inline struct sock *tcp_v4_lookup_listener(u32 daddr,
- unsigned short hnum, int dif)
+static inline struct sock *tcp_v4_lookup_listener(const u32 daddr,
+ const unsigned short hnum,
+ const int dif)
{
struct sock *sk = NULL;
struct hlist_head *head;
- read_lock(&tcp_lhash_lock);
- head = &tcp_listening_hash[tcp_lhashfn(hnum)];
+ read_lock(&tcp_hashinfo.lhash_lock);
+ head = &tcp_hashinfo.listening_hash[inet_lhashfn(hnum)];
if (!hlist_empty(head)) {
struct inet_sock *inet = inet_sk((sk = __sk_head(head)));
sherry_cache:
sock_hold(sk);
}
- read_unlock(&tcp_lhash_lock);
+ read_unlock(&tcp_hashinfo.lhash_lock);
return sk;
}
* Local BH must be disabled here.
*/
-static inline struct sock *__tcp_v4_lookup_established(u32 saddr, u16 sport,
- u32 daddr, u16 hnum,
- int dif)
+static inline struct sock *__tcp_v4_lookup_established(const u32 saddr,
+ const u16 sport,
+ const u32 daddr,
+ const u16 hnum,
+ const int dif)
{
- struct tcp_ehash_bucket *head;
+ struct inet_ehash_bucket *head;
TCP_V4_ADDR_COOKIE(acookie, saddr, daddr)
__u32 ports = TCP_COMBINED_PORTS(sport, hnum);
struct sock *sk;
/* Optimize here for direct hit, only listening connections can
* have wildcards anyways.
*/
- int hash = tcp_hashfn(daddr, hnum, saddr, sport);
- head = &tcp_ehash[hash];
+ const int hash = inet_ehashfn(daddr, hnum, saddr, sport, tcp_hashinfo.ehash_size);
+ head = &tcp_hashinfo.ehash[hash];
read_lock(&head->lock);
sk_for_each(sk, node, &head->chain) {
if (TCP_IPV4_MATCH(sk, acookie, saddr, daddr, ports, dif))
}
/* Must check for a TIME_WAIT'er before going to listener hash. */
- sk_for_each(sk, node, &(head + tcp_ehash_size)->chain) {
+ sk_for_each(sk, node, &(head + tcp_hashinfo.ehash_size)->chain) {
if (TCP_IPV4_TW_MATCH(sk, acookie, saddr, daddr, ports, dif))
goto hit;
}
int dif = sk->sk_bound_dev_if;
TCP_V4_ADDR_COOKIE(acookie, saddr, daddr)
__u32 ports = TCP_COMBINED_PORTS(inet->dport, lport);
- int hash = tcp_hashfn(daddr, lport, saddr, inet->dport);
- struct tcp_ehash_bucket *head = &tcp_ehash[hash];
+ const int hash = inet_ehashfn(daddr, lport, saddr, inet->dport, tcp_hashinfo.ehash_size);
+ struct inet_ehash_bucket *head = &tcp_hashinfo.ehash[hash];
struct sock *sk2;
struct hlist_node *node;
struct tcp_tw_bucket *tw;
write_lock(&head->lock);
/* Check TIME-WAIT sockets first. */
- sk_for_each(sk2, node, &(head + tcp_ehash_size)->chain) {
+ sk_for_each(sk2, node, &(head + tcp_hashinfo.ehash_size)->chain) {
tw = (struct tcp_tw_bucket *)sk2;
if (TCP_IPV4_TW_MATCH(sk2, acookie, saddr, daddr, ports, dif)) {
*/
static inline int tcp_v4_hash_connect(struct sock *sk)
{
- unsigned short snum = inet_sk(sk)->num;
- struct tcp_bind_hashbucket *head;
- struct tcp_bind_bucket *tb;
+ const unsigned short snum = inet_sk(sk)->num;
+ struct inet_bind_hashbucket *head;
+ struct inet_bind_bucket *tb;
int ret;
if (!snum) {
local_bh_disable();
for (i = 1; i <= range; i++) {
port = low + (i + offset) % range;
- head = &tcp_bhash[tcp_bhashfn(port)];
+ head = &tcp_hashinfo.bhash[inet_bhashfn(port, tcp_hashinfo.bhash_size)];
spin_lock(&head->lock);
/* Does not bother with rcv_saddr checks,
* because the established check is already
* unique enough.
*/
- tb_for_each(tb, node, &head->chain) {
+ inet_bind_bucket_for_each(tb, node, &head->chain) {
if (tb->port == port) {
BUG_TRAP(!hlist_empty(&tb->owners));
if (tb->fastreuse >= 0)
}
}
- tb = tcp_bucket_create(head, port);
+ tb = inet_bind_bucket_create(tcp_hashinfo.bind_bucket_cachep, head, port);
if (!tb) {
spin_unlock(&head->lock);
break;
hint += i;
/* Head lock still held and bh's disabled */
- tcp_bind_hash(sk, tb, port);
+ inet_bind_hash(sk, tb, port);
if (sk_unhashed(sk)) {
inet_sk(sk)->sport = htons(port);
- __tcp_v4_hash(sk, 0);
+ __inet_hash(&tcp_hashinfo, sk, 0);
}
spin_unlock(&head->lock);
goto out;
}
- head = &tcp_bhash[tcp_bhashfn(snum)];
- tb = tcp_sk(sk)->bind_hash;
+ head = &tcp_hashinfo.bhash[inet_bhashfn(snum, tcp_hashinfo.bhash_size)];
+ tb = inet_sk(sk)->bind_hash;
spin_lock_bh(&head->lock);
if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
- __tcp_v4_hash(sk, 0);
+ __inet_hash(&tcp_hashinfo, sk, 0);
spin_unlock_bh(&head->lock);
return 0;
} else {
goto failure;
/* OK, now commit destination to socket. */
- __sk_dst_set(sk, &rt->u.dst);
- tcp_v4_setup_caps(sk, &rt->u.dst);
+ sk_setup_caps(sk, &rt->u.dst);
if (!tp->write_seq)
tp->write_seq = secure_tcp_sequence_number(inet->saddr,
return (jhash_2words(raddr, (u32) rport, rnd) & (TCP_SYNQ_HSIZE - 1));
}
-static struct open_request *tcp_v4_search_req(struct tcp_sock *tp,
- struct open_request ***prevp,
+static struct request_sock *tcp_v4_search_req(struct tcp_sock *tp,
+ struct request_sock ***prevp,
__u16 rport,
__u32 raddr, __u32 laddr)
{
- struct tcp_listen_opt *lopt = tp->listen_opt;
- struct open_request *req, **prev;
+ struct listen_sock *lopt = tp->accept_queue.listen_opt;
+ struct request_sock *req, **prev;
for (prev = &lopt->syn_table[tcp_v4_synq_hash(raddr, rport, lopt->hash_rnd)];
(req = *prev) != NULL;
prev = &req->dl_next) {
- if (req->rmt_port == rport &&
- req->af.v4_req.rmt_addr == raddr &&
- req->af.v4_req.loc_addr == laddr &&
- TCP_INET_FAMILY(req->class->family)) {
+ const struct inet_request_sock *ireq = inet_rsk(req);
+
+ if (ireq->rmt_port == rport &&
+ ireq->rmt_addr == raddr &&
+ ireq->loc_addr == laddr &&
+ TCP_INET_FAMILY(req->rsk_ops->family)) {
BUG_TRAP(!req->sk);
*prevp = prev;
break;
return req;
}
-static void tcp_v4_synq_add(struct sock *sk, struct open_request *req)
+static void tcp_v4_synq_add(struct sock *sk, struct request_sock *req)
{
struct tcp_sock *tp = tcp_sk(sk);
- struct tcp_listen_opt *lopt = tp->listen_opt;
- u32 h = tcp_v4_synq_hash(req->af.v4_req.rmt_addr, req->rmt_port, lopt->hash_rnd);
-
- req->expires = jiffies + TCP_TIMEOUT_INIT;
- req->retrans = 0;
- req->sk = NULL;
- req->dl_next = lopt->syn_table[h];
-
- write_lock(&tp->syn_wait_lock);
- lopt->syn_table[h] = req;
- write_unlock(&tp->syn_wait_lock);
+ struct listen_sock *lopt = tp->accept_queue.listen_opt;
+ u32 h = tcp_v4_synq_hash(inet_rsk(req)->rmt_addr, inet_rsk(req)->rmt_port, lopt->hash_rnd);
+ reqsk_queue_hash_req(&tp->accept_queue, h, req, TCP_TIMEOUT_INIT);
tcp_synq_added(sk);
}
}
switch (sk->sk_state) {
- struct open_request *req, **prev;
+ struct request_sock *req, **prev;
case TCP_LISTEN:
if (sock_owned_by_user(sk))
goto out;
*/
BUG_TRAP(!req->sk);
- if (seq != req->snt_isn) {
+ if (seq != tcp_rsk(req)->snt_isn) {
NET_INC_STATS_BH(LINUX_MIB_OUTOFWINDOWICMPS);
goto out;
}
tcp_tw_put(tw);
}
-static void tcp_v4_or_send_ack(struct sk_buff *skb, struct open_request *req)
+static void tcp_v4_reqsk_send_ack(struct sk_buff *skb, struct request_sock *req)
{
- tcp_v4_send_ack(skb, req->snt_isn + 1, req->rcv_isn + 1, req->rcv_wnd,
+ tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1, tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
req->ts_recent);
}
static struct dst_entry* tcp_v4_route_req(struct sock *sk,
- struct open_request *req)
+ struct request_sock *req)
{
struct rtable *rt;
- struct ip_options *opt = req->af.v4_req.opt;
+ const struct inet_request_sock *ireq = inet_rsk(req);
+ struct ip_options *opt = inet_rsk(req)->opt;
struct flowi fl = { .oif = sk->sk_bound_dev_if,
.nl_u = { .ip4_u =
{ .daddr = ((opt && opt->srr) ?
opt->faddr :
- req->af.v4_req.rmt_addr),
- .saddr = req->af.v4_req.loc_addr,
+ ireq->rmt_addr),
+ .saddr = ireq->loc_addr,
.tos = RT_CONN_FLAGS(sk) } },
.proto = IPPROTO_TCP,
.uli_u = { .ports =
{ .sport = inet_sk(sk)->sport,
- .dport = req->rmt_port } } };
+ .dport = ireq->rmt_port } } };
if (ip_route_output_flow(&rt, &fl, sk, 0)) {
IP_INC_STATS_BH(IPSTATS_MIB_OUTNOROUTES);
/*
* Send a SYN-ACK after having received an ACK.
- * This still operates on a open_request only, not on a big
+ * This still operates on a request_sock only, not on a big
* socket.
*/
-static int tcp_v4_send_synack(struct sock *sk, struct open_request *req,
+static int tcp_v4_send_synack(struct sock *sk, struct request_sock *req,
struct dst_entry *dst)
{
+ const struct inet_request_sock *ireq = inet_rsk(req);
int err = -1;
struct sk_buff * skb;
struct tcphdr *th = skb->h.th;
th->check = tcp_v4_check(th, skb->len,
- req->af.v4_req.loc_addr,
- req->af.v4_req.rmt_addr,
+ ireq->loc_addr,
+ ireq->rmt_addr,
csum_partial((char *)th, skb->len,
skb->csum));
- err = ip_build_and_send_pkt(skb, sk, req->af.v4_req.loc_addr,
- req->af.v4_req.rmt_addr,
- req->af.v4_req.opt);
+ err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
+ ireq->rmt_addr,
+ ireq->opt);
if (err == NET_XMIT_CN)
err = 0;
}
}
/*
- * IPv4 open_request destructor.
+ * IPv4 request_sock destructor.
*/
-static void tcp_v4_or_free(struct open_request *req)
+static void tcp_v4_reqsk_destructor(struct request_sock *req)
{
- if (req->af.v4_req.opt)
- kfree(req->af.v4_req.opt);
+ if (inet_rsk(req)->opt)
+ kfree(inet_rsk(req)->opt);
}
static inline void syn_flood_warning(struct sk_buff *skb)
}
/*
- * Save and compile IPv4 options into the open_request if needed.
+ * Save and compile IPv4 options into the request_sock if needed.
*/
static inline struct ip_options *tcp_v4_save_options(struct sock *sk,
struct sk_buff *skb)
return dopt;
}
-/*
- * Maximum number of SYN_RECV sockets in queue per LISTEN socket.
- * One SYN_RECV socket costs about 80bytes on a 32bit machine.
- * It would be better to replace it with a global counter for all sockets
- * but then some measure against one socket starving all other sockets
- * would be needed.
- *
- * It was 128 by default. Experiments with real servers show, that
- * it is absolutely not enough even at 100conn/sec. 256 cures most
- * of problems. This value is adjusted to 128 for very small machines
- * (<=32Mb of memory) and to 1024 on normal or better ones (>=256Mb).
- * Further increasing requires to change hash table size.
- */
-int sysctl_max_syn_backlog = 256;
-
-struct or_calltable or_ipv4 = {
+struct request_sock_ops tcp_request_sock_ops = {
.family = PF_INET,
+ .obj_size = sizeof(struct tcp_request_sock),
.rtx_syn_ack = tcp_v4_send_synack,
- .send_ack = tcp_v4_or_send_ack,
- .destructor = tcp_v4_or_free,
+ .send_ack = tcp_v4_reqsk_send_ack,
+ .destructor = tcp_v4_reqsk_destructor,
.send_reset = tcp_v4_send_reset,
};
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
+ struct inet_request_sock *ireq;
struct tcp_options_received tmp_opt;
- struct open_request *req;
+ struct request_sock *req;
__u32 saddr = skb->nh.iph->saddr;
__u32 daddr = skb->nh.iph->daddr;
__u32 isn = TCP_SKB_CB(skb)->when;
if (sk_acceptq_is_full(sk) && tcp_synq_young(sk) > 1)
goto drop;
- req = tcp_openreq_alloc();
+ req = reqsk_alloc(&tcp_request_sock_ops);
if (!req)
goto drop;
tcp_openreq_init(req, &tmp_opt, skb);
- req->af.v4_req.loc_addr = daddr;
- req->af.v4_req.rmt_addr = saddr;
- req->af.v4_req.opt = tcp_v4_save_options(sk, skb);
- req->class = &or_ipv4;
+ ireq = inet_rsk(req);
+ ireq->loc_addr = daddr;
+ ireq->rmt_addr = saddr;
+ ireq->opt = tcp_v4_save_options(sk, skb);
if (!want_cookie)
TCP_ECN_create_request(req, skb->h.th);
* to destinations, already remembered
* to the moment of synflood.
*/
- NETDEBUG(if (net_ratelimit()) \
- printk(KERN_DEBUG "TCP: drop open "
- "request from %u.%u."
- "%u.%u/%u\n", \
- NIPQUAD(saddr),
- ntohs(skb->h.th->source)));
+ LIMIT_NETDEBUG(printk(KERN_DEBUG "TCP: drop open "
+ "request from %u.%u."
+ "%u.%u/%u\n",
+ NIPQUAD(saddr),
+ ntohs(skb->h.th->source)));
dst_release(dst);
goto drop_and_free;
}
isn = tcp_v4_init_sequence(sk, skb);
}
- req->snt_isn = isn;
+ tcp_rsk(req)->snt_isn = isn;
if (tcp_v4_send_synack(sk, req, dst))
goto drop_and_free;
if (want_cookie) {
- tcp_openreq_free(req);
+ reqsk_free(req);
} else {
tcp_v4_synq_add(sk, req);
}
return 0;
drop_and_free:
- tcp_openreq_free(req);
+ reqsk_free(req);
drop:
TCP_INC_STATS_BH(TCP_MIB_ATTEMPTFAILS);
return 0;
* now create the new socket.
*/
struct sock *tcp_v4_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
- struct open_request *req,
+ struct request_sock *req,
struct dst_entry *dst)
{
+ struct inet_request_sock *ireq;
struct inet_sock *newinet;
struct tcp_sock *newtp;
struct sock *newsk;
if (!newsk)
goto exit;
- newsk->sk_dst_cache = dst;
- tcp_v4_setup_caps(newsk, dst);
+ sk_setup_caps(newsk, dst);
newtp = tcp_sk(newsk);
newinet = inet_sk(newsk);
- newinet->daddr = req->af.v4_req.rmt_addr;
- newinet->rcv_saddr = req->af.v4_req.loc_addr;
- newinet->saddr = req->af.v4_req.loc_addr;
- newinet->opt = req->af.v4_req.opt;
- req->af.v4_req.opt = NULL;
+ ireq = inet_rsk(req);
+ newinet->daddr = ireq->rmt_addr;
+ newinet->rcv_saddr = ireq->loc_addr;
+ newinet->saddr = ireq->loc_addr;
+ newinet->opt = ireq->opt;
+ ireq->opt = NULL;
newinet->mc_index = tcp_v4_iif(skb);
newinet->mc_ttl = skb->nh.iph->ttl;
newtp->ext_header_len = 0;
newtp->advmss = dst_metric(dst, RTAX_ADVMSS);
tcp_initialize_rcv_mss(newsk);
- __tcp_v4_hash(newsk, 0);
- __tcp_inherit_port(sk, newsk);
+ __inet_hash(&tcp_hashinfo, newsk, 0);
+ __inet_inherit_port(&tcp_hashinfo, sk, newsk);
return newsk;
struct iphdr *iph = skb->nh.iph;
struct tcp_sock *tp = tcp_sk(sk);
struct sock *nsk;
- struct open_request **prev;
+ struct request_sock **prev;
/* Find possible connection requests. */
- struct open_request *req = tcp_v4_search_req(tp, &prev, th->source,
+ struct request_sock *req = tcp_v4_search_req(tp, &prev, th->source,
iph->saddr, iph->daddr);
if (req)
return tcp_check_req(sk, skb, req, prev);
skb->nh.iph->daddr, skb->csum))
return 0;
- NETDEBUG(if (net_ratelimit())
- printk(KERN_DEBUG "hw tcp v4 csum failed\n"));
+ LIMIT_NETDEBUG(printk(KERN_DEBUG "hw tcp v4 csum failed\n"));
skb->ip_summed = CHECKSUM_NONE;
}
if (skb->len <= 76) {
goto discard_it;
}
-/* With per-bucket locks this operation is not-atomic, so that
- * this version is not worse.
- */
-static void __tcp_v4_rehash(struct sock *sk)
-{
- sk->sk_prot->unhash(sk);
- sk->sk_prot->hash(sk);
-}
-
-static int tcp_v4_reselect_saddr(struct sock *sk)
-{
- struct inet_sock *inet = inet_sk(sk);
- int err;
- struct rtable *rt;
- __u32 old_saddr = inet->saddr;
- __u32 new_saddr;
- __u32 daddr = inet->daddr;
-
- if (inet->opt && inet->opt->srr)
- daddr = inet->opt->faddr;
-
- /* Query new route. */
- err = ip_route_connect(&rt, daddr, 0,
- RT_CONN_FLAGS(sk),
- sk->sk_bound_dev_if,
- IPPROTO_TCP,
- inet->sport, inet->dport, sk);
- if (err)
- return err;
-
- __sk_dst_set(sk, &rt->u.dst);
- tcp_v4_setup_caps(sk, &rt->u.dst);
-
- new_saddr = rt->rt_src;
-
- if (new_saddr == old_saddr)
- return 0;
-
- if (sysctl_ip_dynaddr > 1) {
- printk(KERN_INFO "tcp_v4_rebuild_header(): shifting inet->"
- "saddr from %d.%d.%d.%d to %d.%d.%d.%d\n",
- NIPQUAD(old_saddr),
- NIPQUAD(new_saddr));
- }
-
- inet->saddr = new_saddr;
- inet->rcv_saddr = new_saddr;
-
- /* XXX The only one ugly spot where we need to
- * XXX really change the sockets identity after
- * XXX it has entered the hashes. -DaveM
- *
- * Besides that, it does not check for connection
- * uniqueness. Wait for troubles.
- */
- __tcp_v4_rehash(sk);
- return 0;
-}
-
-int tcp_v4_rebuild_header(struct sock *sk)
-{
- struct inet_sock *inet = inet_sk(sk);
- struct rtable *rt = (struct rtable *)__sk_dst_check(sk, 0);
- u32 daddr;
- int err;
-
- /* Route is OK, nothing to do. */
- if (rt)
- return 0;
-
- /* Reroute. */
- daddr = inet->daddr;
- if (inet->opt && inet->opt->srr)
- daddr = inet->opt->faddr;
-
- {
- struct flowi fl = { .oif = sk->sk_bound_dev_if,
- .nl_u = { .ip4_u =
- { .daddr = daddr,
- .saddr = inet->saddr,
- .tos = RT_CONN_FLAGS(sk) } },
- .proto = IPPROTO_TCP,
- .uli_u = { .ports =
- { .sport = inet->sport,
- .dport = inet->dport } } };
-
- err = ip_route_output_flow(&rt, &fl, sk, 0);
- }
- if (!err) {
- __sk_dst_set(sk, &rt->u.dst);
- tcp_v4_setup_caps(sk, &rt->u.dst);
- return 0;
- }
-
- /* Routing failed... */
- sk->sk_route_caps = 0;
-
- if (!sysctl_ip_dynaddr ||
- sk->sk_state != TCP_SYN_SENT ||
- (sk->sk_userlocks & SOCK_BINDADDR_LOCK) ||
- (err = tcp_v4_reselect_saddr(sk)) != 0)
- sk->sk_err_soft = -err;
-
- return err;
-}
-
static void v4_addr2sockaddr(struct sock *sk, struct sockaddr * uaddr)
{
struct sockaddr_in *sin = (struct sockaddr_in *) uaddr;
struct tcp_func ipv4_specific = {
.queue_xmit = ip_queue_xmit,
.send_check = tcp_v4_send_check,
- .rebuild_header = tcp_v4_rebuild_header,
+ .rebuild_header = inet_sk_rebuild_header,
.conn_request = tcp_v4_conn_request,
.syn_recv_sock = tcp_v4_syn_recv_sock,
.remember_stamp = tcp_v4_remember_stamp,
*/
tp->snd_ssthresh = 0x7fffffff; /* Infinity */
tp->snd_cwnd_clamp = ~0;
- tp->mss_cache_std = tp->mss_cache = 536;
+ tp->mss_cache = 536;
tp->reordering = sysctl_tcp_reordering;
+ tp->ca_ops = &tcp_init_congestion_ops;
sk->sk_state = TCP_CLOSE;
tcp_clear_xmit_timers(sk);
+ tcp_cleanup_congestion_control(tp);
+
/* Cleanup up the write buffer. */
sk_stream_writequeue_purge(sk);
__skb_queue_purge(&tp->ucopy.prequeue);
/* Clean up a referenced TCP bind bucket. */
- if (tp->bind_hash)
- tcp_put_port(sk);
+ if (inet_sk(sk)->bind_hash)
+ inet_put_port(&tcp_hashinfo, sk);
/*
* If sendmsg cached page exists, toss it.
if (!sk) {
st->bucket = 0;
- sk = sk_head(&tcp_listening_hash[0]);
+ sk = sk_head(&tcp_hashinfo.listening_hash[0]);
goto get_sk;
}
++st->num;
if (st->state == TCP_SEQ_STATE_OPENREQ) {
- struct open_request *req = cur;
+ struct request_sock *req = cur;
tp = tcp_sk(st->syn_wait_sk);
req = req->dl_next;
while (1) {
while (req) {
- if (req->class->family == st->family) {
+ if (req->rsk_ops->family == st->family) {
cur = req;
goto out;
}
if (++st->sbucket >= TCP_SYNQ_HSIZE)
break;
get_req:
- req = tp->listen_opt->syn_table[st->sbucket];
+ req = tp->accept_queue.listen_opt->syn_table[st->sbucket];
}
sk = sk_next(st->syn_wait_sk);
st->state = TCP_SEQ_STATE_LISTENING;
- read_unlock_bh(&tp->syn_wait_lock);
+ read_unlock_bh(&tp->accept_queue.syn_wait_lock);
} else {
tp = tcp_sk(sk);
- read_lock_bh(&tp->syn_wait_lock);
- if (tp->listen_opt && tp->listen_opt->qlen)
+ read_lock_bh(&tp->accept_queue.syn_wait_lock);
+ if (reqsk_queue_len(&tp->accept_queue))
goto start_req;
- read_unlock_bh(&tp->syn_wait_lock);
+ read_unlock_bh(&tp->accept_queue.syn_wait_lock);
sk = sk_next(sk);
}
get_sk:
goto out;
}
tp = tcp_sk(sk);
- read_lock_bh(&tp->syn_wait_lock);
- if (tp->listen_opt && tp->listen_opt->qlen) {
+ read_lock_bh(&tp->accept_queue.syn_wait_lock);
+ if (reqsk_queue_len(&tp->accept_queue)) {
start_req:
st->uid = sock_i_uid(sk);
st->syn_wait_sk = sk;
st->sbucket = 0;
goto get_req;
}
- read_unlock_bh(&tp->syn_wait_lock);
+ read_unlock_bh(&tp->accept_queue.syn_wait_lock);
}
- if (++st->bucket < TCP_LHTABLE_SIZE) {
- sk = sk_head(&tcp_listening_hash[st->bucket]);
+ if (++st->bucket < INET_LHTABLE_SIZE) {
+ sk = sk_head(&tcp_hashinfo.listening_hash[st->bucket]);
goto get_sk;
}
cur = NULL;
struct tcp_iter_state* st = seq->private;
void *rc = NULL;
- for (st->bucket = 0; st->bucket < tcp_ehash_size; ++st->bucket) {
+ for (st->bucket = 0; st->bucket < tcp_hashinfo.ehash_size; ++st->bucket) {
struct sock *sk;
struct hlist_node *node;
struct tcp_tw_bucket *tw;
/* We can reschedule _before_ having picked the target: */
cond_resched_softirq();
- read_lock(&tcp_ehash[st->bucket].lock);
- sk_for_each(sk, node, &tcp_ehash[st->bucket].chain) {
+ read_lock(&tcp_hashinfo.ehash[st->bucket].lock);
+ sk_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
if (sk->sk_family != st->family) {
continue;
}
}
st->state = TCP_SEQ_STATE_TIME_WAIT;
tw_for_each(tw, node,
- &tcp_ehash[st->bucket + tcp_ehash_size].chain) {
+ &tcp_hashinfo.ehash[st->bucket + tcp_hashinfo.ehash_size].chain) {
if (tw->tw_family != st->family) {
continue;
}
rc = tw;
goto out;
}
- read_unlock(&tcp_ehash[st->bucket].lock);
+ read_unlock(&tcp_hashinfo.ehash[st->bucket].lock);
st->state = TCP_SEQ_STATE_ESTABLISHED;
}
out:
cur = tw;
goto out;
}
- read_unlock(&tcp_ehash[st->bucket].lock);
+ read_unlock(&tcp_hashinfo.ehash[st->bucket].lock);
st->state = TCP_SEQ_STATE_ESTABLISHED;
/* We can reschedule between buckets: */
cond_resched_softirq();
- if (++st->bucket < tcp_ehash_size) {
- read_lock(&tcp_ehash[st->bucket].lock);
- sk = sk_head(&tcp_ehash[st->bucket].chain);
+ if (++st->bucket < tcp_hashinfo.ehash_size) {
+ read_lock(&tcp_hashinfo.ehash[st->bucket].lock);
+ sk = sk_head(&tcp_hashinfo.ehash[st->bucket].chain);
} else {
cur = NULL;
goto out;
}
st->state = TCP_SEQ_STATE_TIME_WAIT;
- tw = tw_head(&tcp_ehash[st->bucket + tcp_ehash_size].chain);
+ tw = tw_head(&tcp_hashinfo.ehash[st->bucket + tcp_hashinfo.ehash_size].chain);
goto get_tw;
found:
cur = sk;
void *rc;
struct tcp_iter_state* st = seq->private;
- tcp_listen_lock();
+ inet_listen_lock(&tcp_hashinfo);
st->state = TCP_SEQ_STATE_LISTENING;
rc = listening_get_idx(seq, &pos);
if (!rc) {
- tcp_listen_unlock();
+ inet_listen_unlock(&tcp_hashinfo);
local_bh_disable();
st->state = TCP_SEQ_STATE_ESTABLISHED;
rc = established_get_idx(seq, pos);
case TCP_SEQ_STATE_LISTENING:
rc = listening_get_next(seq, v);
if (!rc) {
- tcp_listen_unlock();
+ inet_listen_unlock(&tcp_hashinfo);
local_bh_disable();
st->state = TCP_SEQ_STATE_ESTABLISHED;
rc = established_get_first(seq);
case TCP_SEQ_STATE_OPENREQ:
if (v) {
struct tcp_sock *tp = tcp_sk(st->syn_wait_sk);
- read_unlock_bh(&tp->syn_wait_lock);
+ read_unlock_bh(&tp->accept_queue.syn_wait_lock);
}
case TCP_SEQ_STATE_LISTENING:
if (v != SEQ_START_TOKEN)
- tcp_listen_unlock();
+ inet_listen_unlock(&tcp_hashinfo);
break;
case TCP_SEQ_STATE_TIME_WAIT:
case TCP_SEQ_STATE_ESTABLISHED:
if (v)
- read_unlock(&tcp_ehash[st->bucket].lock);
+ read_unlock(&tcp_hashinfo.ehash[st->bucket].lock);
local_bh_enable();
break;
}
memset(afinfo->seq_fops, 0, sizeof(*afinfo->seq_fops));
}
-static void get_openreq4(struct sock *sk, struct open_request *req,
+static void get_openreq4(struct sock *sk, struct request_sock *req,
char *tmpbuf, int i, int uid)
{
+ const struct inet_request_sock *ireq = inet_rsk(req);
int ttd = req->expires - jiffies;
sprintf(tmpbuf, "%4d: %08X:%04X %08X:%04X"
" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p",
i,
- req->af.v4_req.loc_addr,
+ ireq->loc_addr,
ntohs(inet_sk(sk)->sport),
- req->af.v4_req.rmt_addr,
- ntohs(req->rmt_port),
+ ireq->rmt_addr,
+ ntohs(ireq->rmt_port),
TCP_SYN_RECV,
0, 0, /* could print option size, but that is af dependent. */
1, /* timers active (only the expire timer) */
.sysctl_rmem = sysctl_tcp_rmem,
.max_header = MAX_TCP_HEADER,
.obj_size = sizeof(struct tcp_sock),
+ .rsk_prot = &tcp_request_sock_ops,
};
}
EXPORT_SYMBOL(ipv4_specific);
-EXPORT_SYMBOL(tcp_bind_hash);
-EXPORT_SYMBOL(tcp_bucket_create);
+EXPORT_SYMBOL(inet_bind_bucket_create);
EXPORT_SYMBOL(tcp_hashinfo);
-EXPORT_SYMBOL(tcp_inherit_port);
-EXPORT_SYMBOL(tcp_listen_wlock);
-EXPORT_SYMBOL(tcp_port_rover);
EXPORT_SYMBOL(tcp_prot);
-EXPORT_SYMBOL(tcp_put_port);
EXPORT_SYMBOL(tcp_unhash);
EXPORT_SYMBOL(tcp_v4_conn_request);
EXPORT_SYMBOL(tcp_v4_connect);
EXPORT_SYMBOL(tcp_v4_do_rcv);
-EXPORT_SYMBOL(tcp_v4_rebuild_header);
EXPORT_SYMBOL(tcp_v4_remember_stamp);
EXPORT_SYMBOL(tcp_v4_send_check);
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
EXPORT_SYMBOL(tcp_proc_unregister);
#endif
EXPORT_SYMBOL(sysctl_local_port_range);
-EXPORT_SYMBOL(sysctl_max_syn_backlog);
EXPORT_SYMBOL(sysctl_tcp_low_latency);
EXPORT_SYMBOL(sysctl_tcp_tw_reuse);