int sysctl_tcp_syncookies = SYNC_INIT;
int sysctl_tcp_abort_on_overflow;
-static void tcp_tw_schedule(struct tcp_tw_bucket *tw, int timeo);
+static void tcp_tw_schedule(struct inet_timewait_sock *tw, int timeo);
static __inline__ int tcp_in_window(u32 seq, u32 end_seq, u32 s_win, u32 e_win)
{
int tcp_tw_count;
-
-/* 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;
-
- /* Unlink from established hashes. */
- ehead = &tcp_ehash[tw->tw_hashent];
- write_lock(&ehead->lock);
- if (hlist_unhashed(&tw->tw_node)) {
- write_unlock(&ehead->lock);
- return;
- }
- __hlist_del(&tw->tw_node);
- sk_node_init(&tw->tw_node);
- write_unlock(&ehead->lock);
-
- /* Disassociate with bind bucket. */
- bhead = &tcp_bhash[tcp_bhashfn(tw->tw_num)];
- spin_lock(&bhead->lock);
- tb = tw->tw_tb;
- __hlist_del(&tw->tw_bind_node);
- tw->tw_tb = NULL;
- tcp_bucket_destroy(tb);
- spin_unlock(&bhead->lock);
-
-#ifdef INET_REFCNT_DEBUG
- if (atomic_read(&tw->tw_refcnt) != 1) {
- printk(KERN_DEBUG "tw_bucket %p refcnt=%d\n", tw,
- atomic_read(&tw->tw_refcnt));
- }
-#endif
- tcp_tw_put(tw);
-}
-
/*
* * Main purpose of TIME-WAIT state is to close connection gracefully,
* when one of ends sits in LAST-ACK or CLOSING retransmitting FIN
* to avoid misread sequence numbers, states etc. --ANK
*/
enum tcp_tw_status
-tcp_timewait_state_process(struct tcp_tw_bucket *tw, struct sk_buff *skb,
- struct tcphdr *th, unsigned len)
+tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
+ const struct tcphdr *th)
{
+ struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
struct tcp_options_received tmp_opt;
int paws_reject = 0;
tmp_opt.saw_tstamp = 0;
- if (th->doff > (sizeof(struct tcphdr) >> 2) && tw->tw_ts_recent_stamp) {
+ if (th->doff > (sizeof(*th) >> 2) && tcptw->tw_ts_recent_stamp) {
tcp_parse_options(skb, &tmp_opt, 0);
if (tmp_opt.saw_tstamp) {
- tmp_opt.ts_recent = tw->tw_ts_recent;
- tmp_opt.ts_recent_stamp = tw->tw_ts_recent_stamp;
+ tmp_opt.ts_recent = tcptw->tw_ts_recent;
+ tmp_opt.ts_recent_stamp = tcptw->tw_ts_recent_stamp;
paws_reject = tcp_paws_check(&tmp_opt, th->rst);
}
}
/* Out of window, send ACK */
if (paws_reject ||
!tcp_in_window(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq,
- tw->tw_rcv_nxt,
- tw->tw_rcv_nxt + tw->tw_rcv_wnd))
+ tcptw->tw_rcv_nxt,
+ tcptw->tw_rcv_nxt + tcptw->tw_rcv_wnd))
return TCP_TW_ACK;
if (th->rst)
goto kill;
- if (th->syn && !before(TCP_SKB_CB(skb)->seq, tw->tw_rcv_nxt))
+ if (th->syn && !before(TCP_SKB_CB(skb)->seq, tcptw->tw_rcv_nxt))
goto kill_with_rst;
/* Dup ACK? */
- if (!after(TCP_SKB_CB(skb)->end_seq, tw->tw_rcv_nxt) ||
+ if (!after(TCP_SKB_CB(skb)->end_seq, tcptw->tw_rcv_nxt) ||
TCP_SKB_CB(skb)->end_seq == TCP_SKB_CB(skb)->seq) {
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
return TCP_TW_SUCCESS;
}
* reset.
*/
if (!th->fin ||
- TCP_SKB_CB(skb)->end_seq != tw->tw_rcv_nxt + 1) {
+ TCP_SKB_CB(skb)->end_seq != tcptw->tw_rcv_nxt + 1) {
kill_with_rst:
tcp_tw_deschedule(tw);
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
return TCP_TW_RST;
}
/* FIN arrived, enter true time-wait state. */
- tw->tw_substate = TCP_TIME_WAIT;
- tw->tw_rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+ tw->tw_substate = TCP_TIME_WAIT;
+ tcptw->tw_rcv_nxt = TCP_SKB_CB(skb)->end_seq;
if (tmp_opt.saw_tstamp) {
- tw->tw_ts_recent_stamp = xtime.tv_sec;
- tw->tw_ts_recent = tmp_opt.rcv_tsval;
+ tcptw->tw_ts_recent_stamp = xtime.tv_sec;
+ tcptw->tw_ts_recent = tmp_opt.rcv_tsval;
}
/* I am shamed, but failed to make it more elegant.
* do not undertsnad recycling in any case, it not
* a big problem in practice. --ANK */
if (tw->tw_family == AF_INET &&
- sysctl_tcp_tw_recycle && tw->tw_ts_recent_stamp &&
+ sysctl_tcp_tw_recycle && tcptw->tw_ts_recent_stamp &&
tcp_v4_tw_remember_stamp(tw))
tcp_tw_schedule(tw, tw->tw_timeout);
else
*/
if (!paws_reject &&
- (TCP_SKB_CB(skb)->seq == tw->tw_rcv_nxt &&
+ (TCP_SKB_CB(skb)->seq == tcptw->tw_rcv_nxt &&
(TCP_SKB_CB(skb)->seq == TCP_SKB_CB(skb)->end_seq || th->rst))) {
/* In window segment, it may be only reset or bare ack. */
if (sysctl_tcp_rfc1337 == 0) {
kill:
tcp_tw_deschedule(tw);
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
return TCP_TW_SUCCESS;
}
}
tcp_tw_schedule(tw, TCP_TIMEWAIT_LEN);
if (tmp_opt.saw_tstamp) {
- tw->tw_ts_recent = tmp_opt.rcv_tsval;
- tw->tw_ts_recent_stamp = xtime.tv_sec;
+ tcptw->tw_ts_recent = tmp_opt.rcv_tsval;
+ tcptw->tw_ts_recent_stamp = xtime.tv_sec;
}
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
return TCP_TW_SUCCESS;
}
*/
if (th->syn && !th->rst && !th->ack && !paws_reject &&
- (after(TCP_SKB_CB(skb)->seq, tw->tw_rcv_nxt) ||
- (tmp_opt.saw_tstamp && (s32)(tw->tw_ts_recent - tmp_opt.rcv_tsval) < 0))) {
- u32 isn = tw->tw_snd_nxt + 65535 + 2;
+ (after(TCP_SKB_CB(skb)->seq, tcptw->tw_rcv_nxt) ||
+ (tmp_opt.saw_tstamp &&
+ (s32)(tcptw->tw_ts_recent - tmp_opt.rcv_tsval) < 0))) {
+ u32 isn = tcptw->tw_snd_nxt + 65535 + 2;
if (isn == 0)
isn++;
TCP_SKB_CB(skb)->when = isn;
*/
return TCP_TW_ACK;
}
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
return TCP_TW_SUCCESS;
}
-/* Enter the time wait state. This is called with locally disabled BH.
- * Essentially we whip up a timewait bucket, copy the
- * relevant info into it from the SK, and mess with hash chains
- * and list linkage.
- */
-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;
-
- /* 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
- binding cache, even if it is closed.
- */
- bhead = &tcp_bhash[tcp_bhashfn(inet_sk(sk)->num)];
- spin_lock(&bhead->lock);
- tw->tw_tb = tcp_sk(sk)->bind_hash;
- BUG_TRAP(tcp_sk(sk)->bind_hash);
- tw_add_bind_node(tw, &tw->tw_tb->owners);
- spin_unlock(&bhead->lock);
-
- write_lock(&ehead->lock);
-
- /* Step 2: Remove SK from established hash. */
- if (__sk_del_node_init(sk))
- 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);
- atomic_inc(&tw->tw_refcnt);
-
- write_unlock(&ehead->lock);
-}
-
/*
* Move a socket to time-wait or dead fin-wait-2 state.
*/
void tcp_time_wait(struct sock *sk, int state, int timeo)
{
- struct tcp_tw_bucket *tw = NULL;
- struct tcp_sock *tp = tcp_sk(sk);
+ struct inet_timewait_sock *tw = NULL;
+ const struct tcp_sock *tp = tcp_sk(sk);
int recycle_ok = 0;
if (sysctl_tcp_tw_recycle && tp->rx_opt.ts_recent_stamp)
recycle_ok = tp->af_specific->remember_stamp(sk);
if (tcp_tw_count < sysctl_tcp_max_tw_buckets)
- tw = kmem_cache_alloc(tcp_timewait_cachep, SLAB_ATOMIC);
-
- if(tw != NULL) {
- struct inet_sock *inet = inet_sk(sk);
- int rto = (tp->rto<<2) - (tp->rto>>1);
-
- /* Give us an identity. */
- tw->tw_daddr = inet->daddr;
- tw->tw_rcv_saddr = inet->rcv_saddr;
- tw->tw_bound_dev_if = sk->sk_bound_dev_if;
- tw->tw_num = inet->num;
- tw->tw_state = TCP_TIME_WAIT;
- tw->tw_substate = state;
- tw->tw_sport = inet->sport;
- tw->tw_dport = inet->dport;
- tw->tw_family = sk->sk_family;
- tw->tw_reuse = sk->sk_reuse;
- tw->tw_rcv_wscale = tp->rx_opt.rcv_wscale;
- atomic_set(&tw->tw_refcnt, 1);
+ tw = inet_twsk_alloc(sk, state);
- tw->tw_hashent = sk->sk_hashent;
- tw->tw_rcv_nxt = tp->rcv_nxt;
- tw->tw_snd_nxt = tp->snd_nxt;
- tw->tw_rcv_wnd = tcp_receive_window(tp);
- tw->tw_ts_recent = tp->rx_opt.ts_recent;
- tw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
- tw_dead_node_init(tw);
+ if (tw != NULL) {
+ struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
+ const struct inet_connection_sock *icsk = inet_csk(sk);
+ const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);
+
+ tw->tw_rcv_wscale = tp->rx_opt.rcv_wscale;
+ tcptw->tw_rcv_nxt = tp->rcv_nxt;
+ tcptw->tw_snd_nxt = tp->snd_nxt;
+ tcptw->tw_rcv_wnd = tcp_receive_window(tp);
+ tcptw->tw_ts_recent = tp->rx_opt.ts_recent;
+ tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
if (tw->tw_family == PF_INET6) {
struct ipv6_pinfo *np = inet6_sk(sk);
+ struct tcp6_timewait_sock *tcp6tw = tcp6_twsk((struct sock *)tw);
- ipv6_addr_copy(&tw->tw_v6_daddr, &np->daddr);
- ipv6_addr_copy(&tw->tw_v6_rcv_saddr, &np->rcv_saddr);
- tw->tw_v6_ipv6only = np->ipv6only;
- } else {
- memset(&tw->tw_v6_daddr, 0, sizeof(tw->tw_v6_daddr));
- memset(&tw->tw_v6_rcv_saddr, 0, sizeof(tw->tw_v6_rcv_saddr));
- tw->tw_v6_ipv6only = 0;
+ ipv6_addr_copy(&tcp6tw->tw_v6_daddr, &np->daddr);
+ ipv6_addr_copy(&tcp6tw->tw_v6_rcv_saddr, &np->rcv_saddr);
+ tw->tw_ipv6only = np->ipv6only;
}
#endif
/* Linkage updates. */
- __tcp_tw_hashdance(sk, tw);
+ __inet_twsk_hashdance(tw, sk, &tcp_hashinfo);
/* Get the TIME_WAIT timeout firing. */
if (timeo < rto)
}
tcp_tw_schedule(tw, timeo);
- tcp_tw_put(tw);
+ inet_twsk_put(tw);
} else {
/* Sorry, if we're out of memory, just CLOSE this
* socket up. We've got bigger problems than
/* Returns non-zero if quota exceeded. */
static int tcp_do_twkill_work(int slot, unsigned int quota)
{
- struct tcp_tw_bucket *tw;
+ struct inet_timewait_sock *tw;
struct hlist_node *node;
unsigned int killed;
int ret;
killed = 0;
ret = 0;
rescan:
- tw_for_each_inmate(tw, node, &tcp_tw_death_row[slot]) {
- __tw_del_dead_node(tw);
+ inet_twsk_for_each_inmate(tw, node, &tcp_tw_death_row[slot]) {
+ __inet_twsk_del_dead_node(tw);
spin_unlock(&tw_death_lock);
- tcp_timewait_kill(tw);
- tcp_tw_put(tw);
+ __inet_twsk_kill(tw, &tcp_hashinfo);
+ inet_twsk_put(tw);
killed++;
spin_lock(&tw_death_lock);
if (killed > quota) {
*/
/* This is for handling early-kills of TIME_WAIT sockets. */
-void tcp_tw_deschedule(struct tcp_tw_bucket *tw)
+void tcp_tw_deschedule(struct inet_timewait_sock *tw)
{
spin_lock(&tw_death_lock);
- if (tw_del_dead_node(tw)) {
- tcp_tw_put(tw);
+ if (inet_twsk_del_dead_node(tw)) {
+ inet_twsk_put(tw);
if (--tcp_tw_count == 0)
del_timer(&tcp_tw_timer);
}
spin_unlock(&tw_death_lock);
- tcp_timewait_kill(tw);
+ __inet_twsk_kill(tw, &tcp_hashinfo);
}
/* Short-time timewait calendar */
TIMER_INITIALIZER(tcp_twcal_tick, 0, 0);
static struct hlist_head tcp_twcal_row[TCP_TW_RECYCLE_SLOTS];
-static void tcp_tw_schedule(struct tcp_tw_bucket *tw, int timeo)
+static void tcp_tw_schedule(struct inet_timewait_sock *tw, const int timeo)
{
struct hlist_head *list;
int slot;
spin_lock(&tw_death_lock);
/* Unlink it, if it was scheduled */
- if (tw_del_dead_node(tw))
+ if (inet_twsk_del_dead_node(tw))
tcp_tw_count--;
else
atomic_inc(&tw->tw_refcnt);
for (n=0; n<TCP_TW_RECYCLE_SLOTS; n++) {
if (time_before_eq(j, now)) {
struct hlist_node *node, *safe;
- struct tcp_tw_bucket *tw;
+ struct inet_timewait_sock *tw;
- tw_for_each_inmate_safe(tw, node, safe,
- &tcp_twcal_row[slot]) {
- __tw_del_dead_node(tw);
- tcp_timewait_kill(tw);
- tcp_tw_put(tw);
+ inet_twsk_for_each_inmate_safe(tw, node, safe,
+ &tcp_twcal_row[slot]) {
+ __inet_twsk_del_dead_node(tw);
+ __inet_twsk_kill(tw, &tcp_hashinfo);
+ inet_twsk_put(tw);
killed++;
}
} else {
*/
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
- * slabcache (i.e. is it TCPv4 or v6?), this is handled thru sk->sk_prot -acme */
- struct sock *newsk = sk_alloc(PF_INET, GFP_ATOMIC, sk->sk_prot, 0);
+ struct sock *newsk = sk_clone(sk, GFP_ATOMIC);
- if(newsk != NULL) {
+ 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 inet_connection_sock *newicsk = inet_csk(newsk);
struct tcp_sock *newtp;
- struct sk_filter *filter;
- memcpy(newsk, sk, sizeof(struct tcp_sock));
newsk->sk_state = TCP_SYN_RECV;
-
- /* SANITY */
- sk_node_init(&newsk->sk_node);
- tcp_sk(newsk)->bind_hash = NULL;
+ newicsk->icsk_bind_hash = NULL;
/* Clone the TCP header template */
- inet_sk(newsk)->dport = ireq->rmt_port;
-
- sock_lock_init(newsk);
- bh_lock_sock(newsk);
-
- rwlock_init(&newsk->sk_dst_lock);
- atomic_set(&newsk->sk_rmem_alloc, 0);
- skb_queue_head_init(&newsk->sk_receive_queue);
- atomic_set(&newsk->sk_wmem_alloc, 0);
- skb_queue_head_init(&newsk->sk_write_queue);
- atomic_set(&newsk->sk_omem_alloc, 0);
- newsk->sk_wmem_queued = 0;
- newsk->sk_forward_alloc = 0;
-
- sock_reset_flag(newsk, SOCK_DONE);
- newsk->sk_userlocks = sk->sk_userlocks & ~SOCK_BINDPORT_LOCK;
- newsk->sk_backlog.head = newsk->sk_backlog.tail = NULL;
- newsk->sk_send_head = NULL;
- rwlock_init(&newsk->sk_callback_lock);
- skb_queue_head_init(&newsk->sk_error_queue);
+ newinet->dport = ireq->rmt_port;
newsk->sk_write_space = sk_stream_write_space;
- if ((filter = newsk->sk_filter) != NULL)
- sk_filter_charge(newsk, filter);
-
- if (unlikely(xfrm_sk_clone_policy(newsk))) {
- /* It is still raw copy of parent, so invalidate
- * destructor and make plain sk_free() */
- newsk->sk_destruct = NULL;
- sk_free(newsk);
- return NULL;
- }
-
/* Now setup tcp_sock */
newtp = tcp_sk(newsk);
newtp->pred_flags = 0;
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;
+ newtp->snd_nxt = newtp->snd_una = newtp->snd_sml = treq->snt_isn + 1;
tcp_prequeue_init(newtp);
tcp_init_wl(newtp, treq->snt_isn, treq->rcv_isn);
- newtp->retransmits = 0;
- newtp->backoff = 0;
+ newicsk->icsk_retransmits = 0;
+ newicsk->icsk_backoff = 0;
newtp->srtt = 0;
newtp->mdev = TCP_TIMEOUT_INIT;
- newtp->rto = TCP_TIMEOUT_INIT;
+ newicsk->icsk_rto = TCP_TIMEOUT_INIT;
newtp->packets_out = 0;
newtp->left_out = 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->rx_opt.num_sacks = 0;
newtp->urg_data = 0;
/* 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
- atomic_inc(&tcp_sockets_allocated);
+ memset(&newicsk->icsk_accept_queue, 0, sizeof(newicsk->icsk_accept_queue));
if (sock_flag(newsk, SOCK_KEEPOPEN))
- tcp_reset_keepalive_timer(newsk,
- keepalive_time_when(newtp));
- newsk->sk_socket = NULL;
- newsk->sk_sleep = NULL;
+ inet_csk_reset_keepalive_timer(newsk,
+ keepalive_time_when(newtp));
newtp->rx_opt.tstamp_ok = ireq->tstamp_ok;
if((newtp->rx_opt.sack_ok = ireq->sack_ok) != 0) {
newtp->tcp_header_len = sizeof(struct tcphdr);
}
if (skb->len >= TCP_MIN_RCVMSS+newtp->tcp_header_len)
- newtp->ack.last_seg_size = skb->len-newtp->tcp_header_len;
+ newicsk->icsk_ack.last_seg_size = skb->len - newtp->tcp_header_len;
newtp->rx_opt.mss_clamp = req->mss;
TCP_ECN_openreq_child(newtp, req);
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;
if (child == NULL)
goto listen_overflow;
- tcp_synq_unlink(tp, req, prev);
- tcp_synq_removed(sk, req);
+ inet_csk_reqsk_queue_unlink(sk, req, prev);
+ inet_csk_reqsk_queue_removed(sk, req);
- tcp_acceptq_queue(sk, req, child);
+ inet_csk_reqsk_queue_add(sk, req, child);
return child;
listen_overflow:
if (!(flg & TCP_FLAG_RST))
req->rsk_ops->send_reset(skb);
- tcp_synq_drop(sk, req, prev);
+ inet_csk_reqsk_queue_drop(sk, req, prev);
return NULL;
}