/* Must be called with locally disabled BHs. */
static void tcp_timewait_kill(struct tcp_tw_bucket *tw)
{
- struct tcp_ehash_bucket *ehead;
- struct tcp_bind_hashbucket *bhead;
- struct tcp_bind_bucket *tb;
-
+ struct inet_bind_hashbucket *bhead;
+ struct inet_bind_bucket *tb;
/* Unlink from established hashes. */
- ehead = &tcp_ehash[tw->tw_hashent];
+ struct inet_ehash_bucket *ehead = &tcp_hashinfo.ehash[tw->tw_hashent];
+
write_lock(&ehead->lock);
if (hlist_unhashed(&tw->tw_node)) {
write_unlock(&ehead->lock);
write_unlock(&ehead->lock);
/* Disassociate with bind bucket. */
- bhead = &tcp_bhash[tcp_bhashfn(tw->tw_num)];
+ bhead = &tcp_hashinfo.bhash[inet_bhashfn(tw->tw_num, tcp_hashinfo.bhash_size)];
spin_lock(&bhead->lock);
tb = tw->tw_tb;
__hlist_del(&tw->tw_bind_node);
tw->tw_tb = NULL;
- tcp_bucket_destroy(tb);
+ inet_bind_bucket_destroy(tcp_hashinfo.bind_bucket_cachep, tb);
spin_unlock(&bhead->lock);
-#ifdef INET_REFCNT_DEBUG
+#ifdef SOCK_REFCNT_DEBUG
if (atomic_read(&tw->tw_refcnt) != 1) {
printk(KERN_DEBUG "tw_bucket %p refcnt=%d\n", tw,
atomic_read(&tw->tw_refcnt));
*/
static void __tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *tw)
{
- struct tcp_ehash_bucket *ehead = &tcp_ehash[sk->sk_hashent];
- struct tcp_bind_hashbucket *bhead;
-
+ const struct inet_sock *inet = inet_sk(sk);
+ struct inet_ehash_bucket *ehead = &tcp_hashinfo.ehash[sk->sk_hashent];
+ struct inet_bind_hashbucket *bhead;
/* Step 1: Put TW into bind hash. Original socket stays there too.
- Note, that any socket with inet_sk(sk)->num != 0 MUST be bound in
+ Note, that any socket with inet->num != 0 MUST be bound in
binding cache, even if it is closed.
*/
- bhead = &tcp_bhash[tcp_bhashfn(inet_sk(sk)->num)];
+ bhead = &tcp_hashinfo.bhash[inet_bhashfn(inet->num, tcp_hashinfo.bhash_size)];
spin_lock(&bhead->lock);
- tw->tw_tb = tcp_sk(sk)->bind_hash;
- BUG_TRAP(tcp_sk(sk)->bind_hash);
+ tw->tw_tb = inet->bind_hash;
+ BUG_TRAP(inet->bind_hash);
tw_add_bind_node(tw, &tw->tw_tb->owners);
spin_unlock(&bhead->lock);
sock_prot_dec_use(sk->sk_prot);
/* Step 3: Hash TW into TIMEWAIT half of established hash table. */
- tw_add_node(tw, &(ehead + tcp_ehash_size)->chain);
+ tw_add_node(tw, &(ehead + tcp_hashinfo.ehash_size)->chain);
atomic_inc(&tw->tw_refcnt);
write_unlock(&ehead->lock);
* Actually, we could lots of memory writes here. tp of listening
* socket contains all necessary default parameters.
*/
-struct sock *tcp_create_openreq_child(struct sock *sk, struct open_request *req, struct sk_buff *skb)
+struct sock *tcp_create_openreq_child(struct sock *sk, struct request_sock *req, struct sk_buff *skb)
{
/* allocate the newsk from the same slab of the master sock,
* if not, at sk_free time we'll try to free it from the wrong
struct sock *newsk = sk_alloc(PF_INET, GFP_ATOMIC, sk->sk_prot, 0);
if(newsk != NULL) {
+ struct inet_request_sock *ireq = inet_rsk(req);
+ struct tcp_request_sock *treq = tcp_rsk(req);
+ struct inet_sock *newinet = inet_sk(newsk);
struct tcp_sock *newtp;
struct sk_filter *filter;
/* SANITY */
sk_node_init(&newsk->sk_node);
- tcp_sk(newsk)->bind_hash = NULL;
+ newinet->bind_hash = NULL;
/* Clone the TCP header template */
- inet_sk(newsk)->dport = req->rmt_port;
+ newinet->dport = ireq->rmt_port;
sock_lock_init(newsk);
bh_lock_sock(newsk);
rwlock_init(&newsk->sk_dst_lock);
+ newsk->sk_dst_cache = NULL;
atomic_set(&newsk->sk_rmem_alloc, 0);
skb_queue_head_init(&newsk->sk_receive_queue);
atomic_set(&newsk->sk_wmem_alloc, 0);
/* Now setup tcp_sock */
newtp = tcp_sk(newsk);
newtp->pred_flags = 0;
- newtp->rcv_nxt = req->rcv_isn + 1;
- newtp->snd_nxt = req->snt_isn + 1;
- newtp->snd_una = req->snt_isn + 1;
- newtp->snd_sml = req->snt_isn + 1;
+ newtp->rcv_nxt = treq->rcv_isn + 1;
+ newtp->snd_nxt = treq->snt_isn + 1;
+ newtp->snd_una = treq->snt_isn + 1;
+ newtp->snd_sml = treq->snt_isn + 1;
tcp_prequeue_init(newtp);
- tcp_init_wl(newtp, req->snt_isn, req->rcv_isn);
+ tcp_init_wl(newtp, treq->snt_isn, treq->rcv_isn);
newtp->retransmits = 0;
newtp->backoff = 0;
newtp->frto_counter = 0;
newtp->frto_highmark = 0;
+ newtp->ca_ops = &tcp_reno;
+
tcp_set_ca_state(newtp, TCP_CA_Open);
tcp_init_xmit_timers(newsk);
skb_queue_head_init(&newtp->out_of_order_queue);
- newtp->rcv_wup = req->rcv_isn + 1;
- newtp->write_seq = req->snt_isn + 1;
+ newtp->rcv_wup = treq->rcv_isn + 1;
+ newtp->write_seq = treq->snt_isn + 1;
newtp->pushed_seq = newtp->write_seq;
- newtp->copied_seq = req->rcv_isn + 1;
+ newtp->copied_seq = treq->rcv_isn + 1;
newtp->rx_opt.saw_tstamp = 0;
newtp->probes_out = 0;
newtp->rx_opt.num_sacks = 0;
newtp->urg_data = 0;
- newtp->listen_opt = NULL;
- newtp->accept_queue = newtp->accept_queue_tail = NULL;
- /* Deinitialize syn_wait_lock to trap illegal accesses. */
- memset(&newtp->syn_wait_lock, 0, sizeof(newtp->syn_wait_lock));
+ /* Deinitialize accept_queue to trap illegal accesses. */
+ memset(&newtp->accept_queue, 0, sizeof(newtp->accept_queue));
/* Back to base struct sock members. */
newsk->sk_err = 0;
newsk->sk_priority = 0;
atomic_set(&newsk->sk_refcnt, 2);
-#ifdef INET_REFCNT_DEBUG
- atomic_inc(&inet_sock_nr);
-#endif
+
+ /*
+ * Increment the counter in the same struct proto as the master
+ * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
+ * is the same as sk->sk_prot->socks, as this field was copied
+ * with memcpy), same rationale as the first comment in this
+ * function.
+ *
+ * This _changes_ the previous behaviour, where
+ * tcp_create_openreq_child always was incrementing the
+ * equivalent to tcp_prot->socks (inet_sock_nr), so this have
+ * to be taken into account in all callers. -acme
+ */
+ sk_refcnt_debug_inc(newsk);
+
atomic_inc(&tcp_sockets_allocated);
if (sock_flag(newsk, SOCK_KEEPOPEN))
newsk->sk_socket = NULL;
newsk->sk_sleep = NULL;
- newtp->rx_opt.tstamp_ok = req->tstamp_ok;
- if((newtp->rx_opt.sack_ok = req->sack_ok) != 0) {
+ newtp->rx_opt.tstamp_ok = ireq->tstamp_ok;
+ if((newtp->rx_opt.sack_ok = ireq->sack_ok) != 0) {
if (sysctl_tcp_fack)
newtp->rx_opt.sack_ok |= 2;
}
newtp->window_clamp = req->window_clamp;
newtp->rcv_ssthresh = req->rcv_wnd;
newtp->rcv_wnd = req->rcv_wnd;
- newtp->rx_opt.wscale_ok = req->wscale_ok;
+ newtp->rx_opt.wscale_ok = ireq->wscale_ok;
if (newtp->rx_opt.wscale_ok) {
- newtp->rx_opt.snd_wscale = req->snd_wscale;
- newtp->rx_opt.rcv_wscale = req->rcv_wscale;
+ newtp->rx_opt.snd_wscale = ireq->snd_wscale;
+ newtp->rx_opt.rcv_wscale = ireq->rcv_wscale;
} else {
newtp->rx_opt.snd_wscale = newtp->rx_opt.rcv_wscale = 0;
newtp->window_clamp = min(newtp->window_clamp, 65535U);
if (newtp->ecn_flags&TCP_ECN_OK)
sock_set_flag(newsk, SOCK_NO_LARGESEND);
- tcp_ca_init(newtp);
-
TCP_INC_STATS_BH(TCP_MIB_PASSIVEOPENS);
}
return newsk;
/*
* Process an incoming packet for SYN_RECV sockets represented
- * as an open_request.
+ * as a request_sock.
*/
struct sock *tcp_check_req(struct sock *sk,struct sk_buff *skb,
- struct open_request *req,
- struct open_request **prev)
+ struct request_sock *req,
+ struct request_sock **prev)
{
struct tcphdr *th = skb->h.th;
struct tcp_sock *tp = tcp_sk(sk);
}
/* Check for pure retransmitted SYN. */
- if (TCP_SKB_CB(skb)->seq == req->rcv_isn &&
+ if (TCP_SKB_CB(skb)->seq == tcp_rsk(req)->rcv_isn &&
flg == TCP_FLAG_SYN &&
!paws_reject) {
/*
* Enforce "SYN-ACK" according to figure 8, figure 6
* of RFC793, fixed by RFC1122.
*/
- req->class->rtx_syn_ack(sk, req, NULL);
+ req->rsk_ops->rtx_syn_ack(sk, req, NULL);
return NULL;
}
* Invalid ACK: reset will be sent by listening socket
*/
if ((flg & TCP_FLAG_ACK) &&
- (TCP_SKB_CB(skb)->ack_seq != req->snt_isn+1))
+ (TCP_SKB_CB(skb)->ack_seq != tcp_rsk(req)->snt_isn + 1))
return sk;
/* Also, it would be not so bad idea to check rcv_tsecr, which
/* RFC793: "first check sequence number". */
if (paws_reject || !tcp_in_window(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq,
- req->rcv_isn+1, req->rcv_isn+1+req->rcv_wnd)) {
+ tcp_rsk(req)->rcv_isn + 1, tcp_rsk(req)->rcv_isn + 1 + req->rcv_wnd)) {
/* Out of window: send ACK and drop. */
if (!(flg & TCP_FLAG_RST))
- req->class->send_ack(skb, req);
+ req->rsk_ops->send_ack(skb, req);
if (paws_reject)
NET_INC_STATS_BH(LINUX_MIB_PAWSESTABREJECTED);
return NULL;
/* In sequence, PAWS is OK. */
- if (tmp_opt.saw_tstamp && !after(TCP_SKB_CB(skb)->seq, req->rcv_isn+1))
+ if (tmp_opt.saw_tstamp && !after(TCP_SKB_CB(skb)->seq, tcp_rsk(req)->rcv_isn + 1))
req->ts_recent = tmp_opt.rcv_tsval;
- if (TCP_SKB_CB(skb)->seq == req->rcv_isn) {
+ if (TCP_SKB_CB(skb)->seq == tcp_rsk(req)->rcv_isn) {
/* Truncate SYN, it is out of window starting
- at req->rcv_isn+1. */
+ at tcp_rsk(req)->rcv_isn + 1. */
flg &= ~TCP_FLAG_SYN;
}
return NULL;
/* If TCP_DEFER_ACCEPT is set, drop bare ACK. */
- if (tp->defer_accept && TCP_SKB_CB(skb)->end_seq == req->rcv_isn+1) {
- req->acked = 1;
+ if (tp->defer_accept && TCP_SKB_CB(skb)->end_seq == tcp_rsk(req)->rcv_isn + 1) {
+ inet_rsk(req)->acked = 1;
return NULL;
}
listen_overflow:
if (!sysctl_tcp_abort_on_overflow) {
- req->acked = 1;
+ inet_rsk(req)->acked = 1;
return NULL;
}
embryonic_reset:
NET_INC_STATS_BH(LINUX_MIB_EMBRYONICRSTS);
if (!(flg & TCP_FLAG_RST))
- req->class->send_reset(skb);
+ req->rsk_ops->send_reset(skb);
tcp_synq_drop(sk, req, prev);
return NULL;