tcp_output.c

MIPS处理器的bootloader,龙芯就是用的修改过的PMON2

		sigoff = optlen + 2;

		{
			unsigned int i;

			for (i = 0; i < 16; i++)
				*(bp++) = 0;
		}

		optlen += TCPOLEN_SIGNATURE;

		/* Pad options list to the next 32 bit boundary and 
		 * terminate it.
		 */
		*bp++ = TCPOPT_NOP;
		*bp++ = TCPOPT_EOL;
		optlen += 2;
	}
#endif /* TCP_SIGNATURE */

#ifdef TCP_SACK
	/*
	 * Send SACKs if necessary.  This should be the last option processed.
	 * Only as many SACKs are sent as are permitted by the maximum options
	 * size.  No more than three SACKs are sent.
	 */
	if (!tp->sack_disable && tp->t_state == TCPS_ESTABLISHED &&
	    (tp->t_flags & (TF_SACK_PERMIT|TF_NOOPT)) == TF_SACK_PERMIT &&
	    tp->rcv_numsacks) {
		u_int32_t *lp = (u_int32_t *)(opt + optlen);
		u_int32_t *olp = lp++;
		int count = 0;  /* actual number of SACKs inserted */
		int maxsack = (MAX_TCPOPTLEN - (optlen + 4))/TCPOLEN_SACK;

		maxsack = min(maxsack, TCP_MAX_SACK);
		for (i = 0; (i < tp->rcv_numsacks && count < maxsack); i++) {
			struct sackblk sack = tp->sackblks[i];
			if (sack.start == 0 && sack.end == 0)
				continue;
			*lp++ = htonl(sack.start);
			*lp++ = htonl(sack.end);
			count++;
		}
		*olp = htonl(TCPOPT_SACK_HDR|(TCPOLEN_SACK*count+2));
		optlen += TCPOLEN_SACK*count + 4; /* including leading NOPs */
	}
#endif /* TCP_SACK */

#ifdef DIAGNOSTIC
	if (optlen > MAX_TCPOPTLEN)
		panic("tcp_output: options too long");
#endif /* DIAGNOSTIC */

	hdrlen += optlen;
 
	/*
	 * Adjust data length if insertion of options will
	 * bump the packet length beyond the t_maxopd length.
	 */
	if (len > tp->t_maxopd - optlen) {
		len = tp->t_maxopd - optlen;
		sendalot = 1;
		flags &= ~TH_FIN;
	 }

#ifdef DIAGNOSTIC
	if (max_linkhdr + hdrlen > MCLBYTES)
		panic("tcphdr too big");
#endif

	/*
	 * Grab a header mbuf, attaching a copy of data to
	 * be transmitted, and initialize the header from
	 * the template for sends on this connection.
	 */
	if (len) {
		if (tp->t_force && len == 1)
			tcpstat.tcps_sndprobe++;
		else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) {
			tcpstat.tcps_sndrexmitpack++;
			tcpstat.tcps_sndrexmitbyte += len;
		} else {
			tcpstat.tcps_sndpack++;
			tcpstat.tcps_sndbyte += len;
		}
#ifdef notyet
		if ((m = m_copypack(so->so_snd.sb_mb, off,
		    (int)len, max_linkhdr + hdrlen)) == 0) {
			error = ENOBUFS;
			goto out;
		}
		/*
		 * m_copypack left space for our hdr; use it.
		 */
		m->m_len += hdrlen;
		m->m_data -= hdrlen;
#else
		MGETHDR(m, M_DONTWAIT, MT_HEADER);
		if (m != NULL) {
			MCLGET(m, M_DONTWAIT);
			if ((m->m_flags & M_EXT) == 0) {
				m_freem(m);
				m = NULL;
			}
		}
		if (m == NULL) {
			error = ENOBUFS;
			goto out;
		}
		m->m_data += max_linkhdr;
		m->m_len = hdrlen;
		if (len <= MCLBYTES - hdrlen - max_linkhdr) {
			m_copydata(so->so_snd.sb_mb, off, (int) len,
			    mtod(m, caddr_t) + hdrlen);
			m->m_len += len;
		} else {
			m->m_next = m_copy(so->so_snd.sb_mb, off, (int) len);
			if (m->m_next == 0) {
				(void) m_free(m);
				error = ENOBUFS;
				goto out;
			}
		}
#endif
		/*
		 * If we're sending everything we've got, set PUSH.
		 * (This will keep happy those implementations which only
		 * give data to the user when a buffer fills or
		 * a PUSH comes in.)
		 */
		if (off + len == so->so_snd.sb_cc)
			flags |= TH_PUSH;
	} else {
		if (tp->t_flags & TF_ACKNOW)
			tcpstat.tcps_sndacks++;
		else if (flags & (TH_SYN|TH_FIN|TH_RST))
			tcpstat.tcps_sndctrl++;
		else if (SEQ_GT(tp->snd_up, tp->snd_una))
			tcpstat.tcps_sndurg++;
		else
			tcpstat.tcps_sndwinup++;

		MGETHDR(m, M_DONTWAIT, MT_HEADER);
		if (m != NULL) {
			MCLGET(m, M_DONTWAIT);
			if ((m->m_flags & M_EXT) == 0) {
				m_freem(m);
				m = NULL;
			}
		}
		if (m == NULL) {
			error = ENOBUFS;
			goto out;
		}
		m->m_data += max_linkhdr;
		m->m_len = hdrlen;
	}
	m->m_pkthdr.rcvif = (struct ifnet *)0;

	if (!tp->t_template)
		panic("tcp_output");
#ifdef DIAGNOSTIC
	if (tp->t_template->m_len != hdrlen - optlen)
		panic("tcp_output: template len != hdrlen - optlen");
#endif /* DIAGNOSTIC */
	bcopy(mtod(tp->t_template, caddr_t), mtod(m, caddr_t),
		tp->t_template->m_len);
	th = (struct tcphdr *)(mtod(m, caddr_t) + tp->t_template->m_len -
		sizeof(struct tcphdr));

	/*
	 * Fill in fields, remembering maximum advertised
	 * window for use in delaying messages about window sizes.
	 * If resending a FIN, be sure not to use a new sequence number.
	 */
	if ((flags & TH_FIN) && (tp->t_flags & TF_SENTFIN) && 
	    (tp->snd_nxt == tp->snd_max))
		tp->snd_nxt--;
	/*
	 * If we are doing retransmissions, then snd_nxt will
	 * not reflect the first unsent octet.  For ACK only
	 * packets, we do not want the sequence number of the
	 * retransmitted packet, we want the sequence number
	 * of the next unsent octet.  So, if there is no data
	 * (and no SYN or FIN), use snd_max instead of snd_nxt
	 * when filling in ti_seq.  But if we are in persist
	 * state, snd_max might reflect one byte beyond the
	 * right edge of the window, so use snd_nxt in that
	 * case, since we know we aren't doing a retransmission.
	 * (retransmit and persist are mutually exclusive...)
	 */
	if (len || (flags & (TH_SYN|TH_FIN)) || tp->t_timer[TCPT_PERSIST])
		th->th_seq = htonl(tp->snd_nxt);
	else
		th->th_seq = htonl(tp->snd_max);

#ifdef TCP_SACK
	if (sack_rxmit) {
		/* 
		 * If sendalot was turned on (due to option stuffing), turn it 
		 * off. Properly set th_seq field.  Advance the ret'x pointer 
		 * by len.  
		 */
		if (sendalot)
			sendalot = 0;
		th->th_seq = htonl(p->rxmit);
		p->rxmit += len;
#if defined(TCP_SACK) && defined(TCP_FACK)
		tp->retran_data += len;
#endif /* TCP_FACK */
	}
#endif /* TCP_SACK */

	th->th_ack = htonl(tp->rcv_nxt);
	if (optlen) {
		bcopy((caddr_t)opt, (caddr_t)(th + 1), optlen);
		th->th_off = (sizeof (struct tcphdr) + optlen) >> 2;
	}
	th->th_flags = flags;

	/*
	 * Calculate receive window.  Don't shrink window,
	 * but avoid silly window syndrome.
	 */
	if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)tp->t_maxseg)
		win = 0;
	if (win > (long)TCP_MAXWIN << tp->rcv_scale)
		win = (long)TCP_MAXWIN << tp->rcv_scale;
	if (win < (long)(tp->rcv_adv - tp->rcv_nxt))
		win = (long)(tp->rcv_adv - tp->rcv_nxt);
	if (flags & TH_RST)
		win = 0;
	th->th_win = htons((u_int16_t) (win>>tp->rcv_scale));
	if (SEQ_GT(tp->snd_up, tp->snd_nxt)) {
		u_int32_t urp = tp->snd_up - tp->snd_nxt;
		if (urp > IP_MAXPACKET)
			urp = IP_MAXPACKET;
		th->th_urp = htons((u_int16_t)urp);
		th->th_flags |= TH_URG;
	} else
		/*
		 * If no urgent pointer to send, then we pull
		 * the urgent pointer to the left edge of the send window
		 * so that it doesn't drift into the send window on sequence
		 * number wraparound.
		 */
		tp->snd_up = tp->snd_una;		/* drag it along */

	/* Put TCP length in pseudo-header */
	switch (tp->pf) {
	case 0:	/*default to PF_INET*/
#ifdef INET
	case AF_INET:
		if (len + optlen)
			mtod(m, struct ipovly *)->ih_len = htons((u_int16_t)(
				sizeof (struct tcphdr) + optlen + len));
		break;
#endif /* INET */
#ifdef INET6
	case AF_INET6:
		break;
#endif /* INET6 */
	}

#ifdef TCP_SIGNATURE
	if (tp->t_flags & TF_SIGNATURE) {
		MD5_CTX ctx;
		union sockaddr_union sa;
		struct tdb *tdb;

		bzero(&sa, sizeof(union sockaddr_union));

		switch (tp->pf) {
		case 0:	/*default to PF_INET*/
#ifdef INET
		case AF_INET:
			sa.sa.sa_len = sizeof(struct sockaddr_in);
			sa.sa.sa_family = AF_INET;
			sa.sin.sin_addr = mtod(m, struct ip *)->ip_dst;
			break;
#endif /* INET */
#ifdef INET6
		case AF_INET6:
			sa.sa.sa_len = sizeof(struct sockaddr_in6);
			sa.sa.sa_family = AF_INET6;
			sa.sin6.sin6_addr = mtod(m, struct ip6_hdr *)->ip6_dst;
			break;
#endif /* INET6 */
		}

		/* XXX gettdb() should really be called at spltdb().      */
		/* XXX this is splsoftnet(), currently they are the same. */
		tdb = gettdb(0, &sa, IPPROTO_TCP);
		if (tdb == NULL)
			return (EPERM);

		MD5Init(&ctx);

		switch (tp->pf) {
		case 0:	/*default to PF_INET*/
#ifdef INET
		case AF_INET:
			{
				struct ippseudo ippseudo;
				struct ipovly *ipovly;

				ipovly = mtod(m, struct ipovly *);

				ippseudo.ippseudo_src = ipovly->ih_src;
				ippseudo.ippseudo_dst = ipovly->ih_dst;
				ippseudo.ippseudo_pad = 0;
				ippseudo.ippseudo_p   = IPPROTO_TCP;
				ippseudo.ippseudo_len = ipovly->ih_len;
				MD5Update(&ctx, (char *)&ippseudo,
					sizeof(struct ippseudo));
				MD5Update(&ctx, mtod(m, caddr_t) +
					sizeof(struct ip),
					sizeof(struct tcphdr));
			}
			break;
#endif /* INET */
#ifdef INET6
		case AF_INET6:
			{
				static int printed = 0;

				if (!printed) {
					printf("error: TCP MD5 support for "
						"IPv6 not yet implemented.\n");
					printed = 1;
				}
			}
			break;
#endif /* INET6 */
		}

		if (len && m_apply(m, hdrlen, len, tcp_signature_apply,
				(caddr_t)&ctx))
			return (EINVAL);

		MD5Update(&ctx, tdb->tdb_amxkey, tdb->tdb_amxkeylen);
		MD5Final(mtod(m, caddr_t) + hdrlen - optlen + sigoff, &ctx);
	}
#endif /* TCP_SIGNATURE */

	/*
	 * Put TCP length in extended header, and then
	 * checksum extended header and data.
	 */
	switch (tp->pf) {
	case 0:	/*default to PF_INET*/
#ifdef INET
	case AF_INET:
		th->th_sum = in_cksum(m, (int)(hdrlen + len));
		break;
#endif /* INET */
#ifdef INET6
	case AF_INET6:
		m->m_pkthdr.len = hdrlen + len;
  		th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr),
			hdrlen - sizeof(struct ip6_hdr) + len);
		break;
#endif /* INET6 */
	}

	/*
	 * In transmit state, time the transmission and arrange for
	 * the retransmit.  In persist state, just set snd_max.
	 */
	if (tp->t_force == 0 || tp->t_timer[TCPT_PERSIST] == 0) {
		tcp_seq startseq = tp->snd_nxt;

		/*
		 * Advance snd_nxt over sequence space of this segment.
		 */
		if (flags & (TH_SYN|TH_FIN)) {
			if (flags & TH_SYN)
				tp->snd_nxt++;
			if (flags & TH_FIN) {
				tp->snd_nxt++;
				tp->t_flags |= TF_SENTFIN;
			}
		}
#ifdef TCP_SACK
		if (!tp->sack_disable) {
			if (sack_rxmit && (p->rxmit != tp->snd_nxt)) {
				goto timer;
			}
		}
#endif
		tp->snd_nxt += len;
		if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
			tp->snd_max = tp->snd_nxt;
			/*
			 * Time this transmission if not a retransmission and
			 * not currently timing anything.
			 */
			if (tp->t_rtt == 0) {
				tp->t_rtt = 1;
				tp->t_rtseq = startseq;
				tcpstat.tcps_segstimed++;
			}
		}

		/*
		 * Set retransmit timer if not currently set,
		 * and not doing an ack or a keep-alive probe.
		 * Initial value for retransmit timer is smoothed
		 * round-trip time + 2 * round-trip time variance.
		 * Initialize shift counter which is used for backoff
		 * of retransmit time.
		 */
#ifdef TCP_SACK
 timer:
		if (!tp->sack_disable && sack_rxmit &&
		    tp->t_timer[TCPT_REXMT] == 0 &&
		    tp->snd_nxt != tp->snd_max) {
			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
			if (tp->t_timer[TCPT_PERSIST]) {
				tp->t_timer[TCPT_PERSIST] = 0;
				tp->t_rxtshift = 0;
			}
		}
#endif

		if (tp->t_timer[TCPT_REXMT] == 0 &&
		    tp->snd_nxt != tp->snd_una) {
			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
			if (tp->t_timer[TCPT_PERSIST]) {
				tp->t_timer[TCPT_PERSIST] = 0;
				tp->t_rxtshift = 0;
			}
		}
	} else
		if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
			tp->snd_max = tp->snd_nxt + len;

	/*
	 * Trace.
	 */
	if (so->so_options & SO_DEBUG)
		tcp_trace(TA_OUTPUT, tp->t_state, tp, mtod(m, caddr_t), 0,
			len);

	/*
	 * Fill in IP length and desired time to live and
	 * send to IP level.  There should be a better way
	 * to handle ttl and tos; we could keep them in
	 * the template, but need a way to checksum without them.
	 */
	m->m_pkthdr.len = hdrlen + len;

	switch (tp->pf) {
	case 0:	/*default to PF_INET*/
#ifdef INET
	case AF_INET:
		{
			struct ip *ip;

			ip = mtod(m, struct ip *);
			ip->ip_len = m->m_pkthdr.len;
			ip->ip_ttl = tp->t_inpcb->inp_ip.ip_ttl;
			ip->ip_tos = tp->t_inpcb->inp_ip.ip_tos;
		}
		error = ip_output(m, tp->t_inpcb->inp_options,
			&tp->t_inpcb->inp_route, so->so_options & SO_DONTROUTE,
			0, tp->t_inpcb);
		break;
#endif /* INET */
#ifdef INET6
	case AF_INET6:
		{
			struct ip6_hdr *ipv6;
			
			ipv6 = mtod(m, struct ip6_hdr *);
			ipv6->ip6_plen = m->m_pkthdr.len -
				sizeof(struct ip6_hdr);
			ipv6->ip6_nxt = IPPROTO_TCP;
			ipv6->ip6_hlim = in6_selecthlim(tp->t_inpcb, NULL);
		}
		error = ip6_output(m, tp->t_inpcb->inp_outputopts6,
			  &tp->t_inpcb->inp_route6,
			  (so->so_options & SO_DONTROUTE), NULL, NULL);
		break;
#endif /* INET6 */
#ifdef TUBA
	case AF_ISO:
		if (tp->t_tuba_pcb)
			error = tuba_output(m, tp);
		break;
#endif /* TUBA */
	}

#if defined(TCP_SACK) && defined(TCP_FACK)
	/* Update snd_awnd to reflect the new data that was sent.  */
	tp->snd_awnd = tcp_seq_subtract(tp->snd_max, tp->snd_fack) +
		tp->retran_data;                
#endif /* defined(TCP_SACK) && defined(TCP_FACK) */

	if (error) {
out:
		if (error == ENOBUFS) {
			tcp_quench(tp->t_inpcb, 0);
			return (0);
		}
		if ((error == EHOSTUNREACH || error == ENETDOWN)
		    && TCPS_HAVERCVDSYN(tp->t_state)) {
			tp->t_softerror = error;
			return (0);
		}
		return (error);
	}
	tcpstat.tcps_sndtotal++;

	/*
	 * Data sent (as far as we can tell).
	 * If this advertises a larger window than any other segment,
	 * then remember the size of the advertised window.
	 * Any pending ACK has now been sent.
	 */
	if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv))
		tp->rcv_adv = tp->rcv_nxt + win;
	tp->last_ack_sent = tp->rcv_nxt;
	tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);
#if defined(TCP_SACK)
	if (sendalot && --maxburst)
#else
	if (sendalot)
#endif
		goto again;
	return (0);
}

void
tcp_setpersist(tp)
	register struct tcpcb *tp;
{
	register int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;

	if (tp->t_timer[TCPT_REXMT])
		panic("tcp_output REXMT");
	/*
	 * Start/restart persistance timer.
	 */
	if (t < tp->t_rttmin)
		t = tp->t_rttmin;
	TCPT_RANGESET(tp->t_timer[TCPT_PERSIST],
	    t * tcp_backoff[tp->t_rxtshift],
	    TCPTV_PERSMIN, TCPTV_PERSMAX);
	if (tp->t_rxtshift < TCP_MAXRXTSHIFT)
		tp->t_rxtshift++;
}