tcp_ipv4.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

			   as state holder.

			   If TW bucket has been already destroyed we
			   fall back to VJ's scheme and use initial
			   timestamp retrieved from peer table.
			 */
			if (tw->ts_recent_stamp) {
				if ((tp->write_seq = tw->snd_nxt+65535+2) == 0)
					tp->write_seq = 1;
				tp->ts_recent = tw->ts_recent;
				tp->ts_recent_stamp = tw->ts_recent_stamp;
				sock_hold(sk2);
				skp = &head->chain;
				goto unique;
			} else
				goto not_unique;
		}
	}
	tw = NULL;

	/* And established part... */
	for(skp = &head->chain; (sk2=*skp)!=NULL; skp = &sk2->next) {
		if(TCP_IPV4_MATCH(sk2, acookie, saddr, daddr, ports, dif))
			goto not_unique;
	}

unique:
	BUG_TRAP(sk->pprev==NULL);
	if ((sk->next = *skp) != NULL)
		(*skp)->pprev = &sk->next;

	*skp = sk;
	sk->pprev = skp;
	sk->hashent = hash;
	sock_prot_inc_use(sk->prot);
	write_unlock_bh(&head->lock);

	if (tw) {
		/* Silly. Should hash-dance instead... */
		local_bh_disable();
		tcp_tw_deschedule(tw);
		tcp_timewait_kill(tw);
		NET_INC_STATS_BH(TimeWaitRecycled);
		local_bh_enable();

		tcp_tw_put(tw);
	}

	return 0;

not_unique:
	write_unlock_bh(&head->lock);
	return -EADDRNOTAVAIL;
}

/* Hash SYN-SENT socket to established hash table after
 * checking that it is unique. Note, that without kernel lock
 * we MUST make these two operations atomically.
 *
 * Optimization: if it is bound and tcp_bind_bucket has the only
 * owner (us), we need not to scan established bucket.
 */

int tcp_v4_hash_connecting(struct sock *sk)
{
	unsigned short snum = sk->num;
	struct tcp_bind_hashbucket *head = &tcp_bhash[tcp_bhashfn(snum)];
	struct tcp_bind_bucket *tb = (struct tcp_bind_bucket *)sk->prev;

	spin_lock_bh(&head->lock);
	if (tb->owners == sk && sk->bind_next == NULL) {
		__tcp_v4_hash(sk);
		spin_unlock_bh(&head->lock);
		return 0;
	} else {
		spin_unlock_bh(&head->lock);

		/* No definite answer... Walk to established hash table */
		return tcp_v4_check_established(sk);
	}
}

/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct sockaddr_in *usin = (struct sockaddr_in *) uaddr;
	struct sk_buff *buff;
	struct rtable *rt;
	u32 daddr, nexthop;
	int tmp;
	int err;

	if (addr_len < sizeof(struct sockaddr_in))
		return(-EINVAL);

	if (usin->sin_family != AF_INET)
		return(-EAFNOSUPPORT);

	nexthop = daddr = usin->sin_addr.s_addr;
	if (sk->protinfo.af_inet.opt && sk->protinfo.af_inet.opt->srr) {
		if (daddr == 0)
			return -EINVAL;
		nexthop = sk->protinfo.af_inet.opt->faddr;
	}

	tmp = ip_route_connect(&rt, nexthop, sk->saddr,
			       RT_TOS(sk->protinfo.af_inet.tos)|RTO_CONN|sk->localroute, sk->bound_dev_if);
	if (tmp < 0)
		return tmp;

	if (rt->rt_flags&(RTCF_MULTICAST|RTCF_BROADCAST)) {
		ip_rt_put(rt);
		return -ENETUNREACH;
	}

	__sk_dst_set(sk, &rt->u.dst);

	if (!sk->protinfo.af_inet.opt || !sk->protinfo.af_inet.opt->srr)
		daddr = rt->rt_dst;

	err = -ENOBUFS;
	buff = alloc_skb(MAX_TCP_HEADER + 15, GFP_KERNEL);

	if (buff == NULL)
		goto failure;

	if (!sk->saddr)
		sk->saddr = rt->rt_src;
	sk->rcv_saddr = sk->saddr;

	if (tp->ts_recent_stamp && sk->daddr != daddr) {
		/* Reset inherited state */
		tp->ts_recent = 0;
		tp->ts_recent_stamp = 0;
		tp->write_seq = 0;
	}

	if (sysctl_tcp_tw_recycle &&
	    !tp->ts_recent_stamp &&
	    rt->rt_dst == daddr) {
		struct inet_peer *peer = rt_get_peer(rt);

		/* VJ's idea. We save last timestamp seen from
		 * the destination in peer table, when entering state TIME-WAIT
		 * and initialize ts_recent from it, when trying new connection.
		 */

		if (peer && peer->tcp_ts_stamp + TCP_PAWS_MSL >= xtime.tv_sec) {
			tp->ts_recent_stamp = peer->tcp_ts_stamp;
			tp->ts_recent = peer->tcp_ts;
		}
	}

	sk->dport = usin->sin_port;
	sk->daddr = daddr;

	if (!tp->write_seq)
		tp->write_seq = secure_tcp_sequence_number(sk->saddr, sk->daddr,
							   sk->sport, usin->sin_port);

	tp->ext_header_len = 0;
	if (sk->protinfo.af_inet.opt)
		tp->ext_header_len = sk->protinfo.af_inet.opt->optlen;

	tp->mss_clamp = 536;

	err = tcp_connect(sk, buff);
	if (err == 0)
		return 0;

failure:
	__sk_dst_reset(sk);
	sk->dport = 0;
	return err;
}

static __inline__ int tcp_v4_iif(struct sk_buff *skb)
{
	return ((struct rtable*)skb->dst)->rt_iif;
}

static __inline__ unsigned tcp_v4_synq_hash(u32 raddr, u16 rport)
{
	unsigned h = raddr ^ rport;
	h ^= h>>16;
	h ^= h>>8;
	return h&(TCP_SYNQ_HSIZE-1);
}

static struct open_request *tcp_v4_search_req(struct tcp_opt *tp, 
					      struct iphdr *iph,
					      struct tcphdr *th,
					      struct open_request ***prevp)
{
	struct tcp_listen_opt *lopt = tp->listen_opt;
	struct open_request *req, **prev;  
	__u16 rport = th->source;
	__u32 raddr = iph->saddr;

	for (prev = &lopt->syn_table[tcp_v4_synq_hash(raddr, rport)];
	     (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 == iph->daddr &&
		    TCP_INET_FAMILY(req->class->family)) {
			BUG_TRAP(req->sk == NULL);
			*prevp = prev;
			return req; 
		}
	}

	return NULL;
}

static void tcp_v4_synq_add(struct sock *sk, struct open_request *req)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
	struct tcp_listen_opt *lopt = tp->listen_opt;
	unsigned h = tcp_v4_synq_hash(req->af.v4_req.rmt_addr, req->rmt_port);

	req->expires = jiffies + TCP_TIMEOUT_INIT;
	req->retrans = 0;
	req->sk = NULL;
	req->index = h;
	req->dl_next = lopt->syn_table[h];

	write_lock(&tp->syn_wait_lock);
	lopt->syn_table[h] = req;
	write_unlock(&tp->syn_wait_lock);

	tcp_synq_added(sk);
}


/* 
 * This routine does path mtu discovery as defined in RFC1191.
 */
static inline void do_pmtu_discovery(struct sock *sk, struct iphdr *ip, unsigned mtu)
{
	struct dst_entry *dst;
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;

	/* We are not interested in TCP_LISTEN and open_requests (SYN-ACKs
	 * send out by Linux are always <576bytes so they should go through
	 * unfragmented).
	 */
	if (sk->state == TCP_LISTEN)
		return; 

	/* We don't check in the destentry if pmtu discovery is forbidden
	 * on this route. We just assume that no packet_to_big packets
	 * are send back when pmtu discovery is not active.
     	 * There is a small race when the user changes this flag in the
	 * route, but I think that's acceptable.
	 */
	if ((dst = __sk_dst_check(sk, 0)) == NULL)
		return;

	ip_rt_update_pmtu(dst, mtu);

	/* Something is about to be wrong... Remember soft error
	 * for the case, if this connection will not able to recover.
	 */
	if (mtu < dst->pmtu && ip_dont_fragment(sk, dst))
		sk->err_soft = EMSGSIZE;

	if (sk->protinfo.af_inet.pmtudisc != IP_PMTUDISC_DONT &&
	    tp->pmtu_cookie > dst->pmtu) {
		tcp_sync_mss(sk, dst->pmtu);

		/* Resend the TCP packet because it's  
		 * clear that the old packet has been
		 * dropped. This is the new "fast" path mtu
		 * discovery.
		 */
		tcp_simple_retransmit(sk);
	} /* else let the usual retransmit timer handle it */
}

/*
 * This routine is called by the ICMP module when it gets some
 * sort of error condition.  If err < 0 then the socket should
 * be closed and the error returned to the user.  If err > 0
 * it's just the icmp type << 8 | icmp code.  After adjustment
 * header points to the first 8 bytes of the tcp header.  We need
 * to find the appropriate port.
 *
 * The locking strategy used here is very "optimistic". When
 * someone else accesses the socket the ICMP is just dropped
 * and for some paths there is no check at all.
 * A more general error queue to queue errors for later handling
 * is probably better.
 *
 */

void tcp_v4_err(struct sk_buff *skb, unsigned char *dp, int len)
{
	struct iphdr *iph = (struct iphdr*)dp;
	struct tcphdr *th; 
	struct tcp_opt *tp;
	int type = skb->h.icmph->type;
	int code = skb->h.icmph->code;
#if ICMP_MIN_LENGTH < 14
	int no_flags = 0;
#else
#define no_flags 0
#endif
	struct sock *sk;
	__u32 seq;
	int err;

	if (len < (iph->ihl << 2) + ICMP_MIN_LENGTH) { 
		ICMP_INC_STATS_BH(IcmpInErrors); 
		return;
	}
#if ICMP_MIN_LENGTH < 14
	if (len < (iph->ihl << 2) + 14)
		no_flags = 1;
#endif

	th = (struct tcphdr*)(dp+(iph->ihl<<2));

	sk = tcp_v4_lookup(iph->daddr, th->dest, iph->saddr, th->source, tcp_v4_iif(skb));
	if (sk == NULL) {
		ICMP_INC_STATS_BH(IcmpInErrors);
		return;
	}
	if (sk->state == TCP_TIME_WAIT) {
		tcp_tw_put((struct tcp_tw_bucket*)sk);
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
	 */
	if (sk->lock.users != 0)
		NET_INC_STATS_BH(LockDroppedIcmps);

	if (sk->state == TCP_CLOSE)
		goto out;

	tp = &sk->tp_pinfo.af_tcp;
	seq = ntohl(th->seq);
	if (sk->state != TCP_LISTEN && !between(seq, tp->snd_una, tp->snd_nxt)) {
		NET_INC_STATS(OutOfWindowIcmps);
		goto out;
	}

	switch (type) {
	case ICMP_SOURCE_QUENCH:
		/* This is deprecated, but if someone generated it,
		 * we have no reasons to ignore it.
		 */
		if (sk->lock.users == 0)
			tcp_enter_cwr(tp);
		goto out;
	case ICMP_PARAMETERPROB:
		err = EPROTO;
		break; 
	case ICMP_DEST_UNREACH:
		if (code > NR_ICMP_UNREACH)
			goto out;

		if (code == ICMP_FRAG_NEEDED) { /* PMTU discovery (RFC1191) */
			if (sk->lock.users == 0)
				do_pmtu_discovery(sk, iph, ntohs(skb->h.icmph->un.frag.mtu));
			goto out;
		}

		err = icmp_err_convert[code].errno;
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->state) {
		struct open_request *req, **prev;
	case TCP_LISTEN:
		if (sk->lock.users != 0)
			goto out;

		/* The final ACK of the handshake should be already 
		 * handled in the new socket context, not here.
		 * Strictly speaking - an ICMP error for the final
		 * ACK should set the opening flag, but that is too
		 * complicated right now. 
		 */ 
		if (!no_flags && !th->syn && !th->ack)
			goto out;

		req = tcp_v4_search_req(tp, iph, th, &prev); 
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
		BUG_TRAP(req->sk == NULL);

		if (seq != req->snt_isn) {
			NET_INC_STATS_BH(OutOfWindowIcmps);
			goto out;
		}

		/* 
		 * Still in SYN_RECV, just remove it silently.
		 * There is no good way to pass the error to the newly
		 * created socket, and POSIX does not want network
		 * errors returned from accept(). 
		 */ 
		tcp_synq_drop(sk, req, prev);
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
			       It can f.e. if SYNs crossed.
			     */ 
		if (!no_flags && !th->syn)
			goto out;
		if (sk->lock.users == 0) {
			TCP_INC_STATS_BH(TcpAttemptFails);
			sk->err = err;

			sk->error_report(sk);

			tcp_done(sk);
		} else {
			sk->err_soft = err;
		}
		goto out;
	}

	/* If we've already connected we will keep trying
	 * until we time out, or the user gives up.
	 *
	 * rfc1122 4.2.3.9 allows to consider as hard errors
	 * only PROTO_UNREACH and PORT_UNREACH (well, FRAG_FAILED too,
	 * but it is obsoleted by pmtu discovery).
	 *
	 * Note, that in modern internet, where routing is unreliable
	 * and in each dark corner broken firewalls sit, sending random
	 * errors ordered by their masters even this two messages finally lose
	 * their original sense (even Linux sends invalid PORT_UNREACHs)
	 *
	 * Now we are in compliance with RFCs.
	 *							--ANK (980905)
	 */

	if (sk->lock.users == 0 && sk->protinfo.af_inet.recverr) {
		sk->err = err;
		sk->error_report(sk);
	} else	{ /* Only an error on timeout */
		sk->err_soft = err;
	}

out:
	bh_unlock_sock(sk);
	sock_put(sk);
}

/* This routine computes an IPv4 TCP checksum. */
void tcp_v4_send_check(struct sock *sk, struct tcphdr *th, int len, 
		       struct sk_buff *skb)
{
	th->check = tcp_v4_check(th, len, sk->saddr, sk->daddr,
				 csum_partial((char *)th, th->doff<<2, skb->csum));
}

/*
 *	This routine will send an RST to the other tcp.
 *
 *	Someone asks: why I NEVER use socket parameters (TOS, TTL etc.)
 *		      for reset.
 *	Answer: if a packet caused RST, it is not for a socket
 *		existing in our system, if it is matched to a socket,
 *		it is just duplicate segment or bug in other side's TCP.
 *		So that we build reply only basing on parameters
 *		arrived with segment.
 *	Exception: precedence violation. We do not implement it in any case.
 */

static void tcp_v4_send_reset(struct sk_buff *skb)
{
	struct tcphdr *th = skb->h.th;
	struct tcphdr rth;
	struct ip_reply_arg arg;

	/* Never send a reset in response to a reset. */
	if (th->rst)
		return;

	if (((struct rtable*)skb->dst)->rt_type != RTN_LOCAL)
		return;

	/* Swap the send and the receive. */
	memset(&rth, 0, sizeof(struct tcphdr)); 
	rth.dest = th->source;
	rth.source = th->dest; 
	rth.doff = sizeof(struct tcphdr)/4;
	rth.rst = 1;

	if (th->ack) {
		rth.seq = th->ack_seq;
	} else {
		rth.ack = 1;
		rth.ack_seq = htonl(ntohl(th->seq) + th->syn + th->fin
				    + skb->len - (th->doff<<2));
	}

	memset(&arg, 0, sizeof arg); 
	arg.iov[0].iov_base = (unsigned char *)&rth; 
	arg.iov[0].iov_len  = sizeof rth;
	arg.csum = csum_tcpudp_nofold(skb->nh.iph->daddr, 
				      skb->nh.iph->saddr, /*XXX*/
				      sizeof(struct tcphdr),
				      IPPROTO_TCP,
				      0); 
	arg.n_iov = 1;
	arg.csumoffset = offsetof(struct tcphdr, check) / 2; 

	ip_send_reply(tcp_socket->sk, skb, &arg, sizeof rth);

	TCP_INC_STATS_BH(TcpOutSegs);
	TCP_INC_STATS_BH(TcpOutRsts);
}

/* The code following below sending ACKs in SYN-RECV and TIME-WAIT states
   outside socket context is ugly, certainly. What can I do?
 */

static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 ts)
{
	struct tcphdr *th = skb->h.th;
	struct {
		struct tcphdr th;
		u32 tsopt[3];
	} rep;
	struct ip_reply_arg arg;

	memset(&rep.th, 0, sizeof(struct tcphdr));
	memset(&arg, 0, sizeof arg);

	arg.iov[0].iov_base = (unsigned char *)&rep; 
	arg.iov[0].iov_len  = sizeof(rep.th);
	arg.n_iov = 1;
	if (ts) {
		rep.tsopt[0] = __constant_htonl((TCPOPT_NOP << 24) |
						(TCPOPT_NOP << 16) |
						(TCPOPT_TIMESTAMP << 8) |