/* BSDI tcp_input.c,v 2.1 1995/02/03 08:19:57 polk Exp */ /* * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_input.c 8.5 (Berkeley) 4/10/94 */ #ifndef TUBA_INCLUDE #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include int tcprexmtthresh = 3; struct tcpiphdr tcp_saveti; struct inpcb *tcp_last_inpcb = &tcb; #endif /* TUBA_INCLUDE */ #ifdef SACK extern int sack_disable; #endif #define TCP_PAWS_IDLE (24 * 24 * 60 * 60 * PR_SLOWHZ) /* for modulo comparisons of timestamps */ #define TSTMP_LT(a,b) ((int)((a)-(b)) < 0) #define TSTMP_GEQ(a,b) ((int)((a)-(b)) >= 0) /* * Insert segment ti into reassembly queue of tcp with * control block tp. Return TH_FIN if reassembly now includes * a segment with FIN. The macro form does the common case inline * (segment is the next to be received on an established connection, * and the queue is empty), avoiding linkage into and removal * from the queue and repetition of various conversions. * Set DELACK for segments received in order, but ack immediately * when segments are out of order (so fast retransmit can work). */ #define TCP_REASS(tp, ti, m, so, flags) { \ if ((ti)->ti_seq == (tp)->rcv_nxt && \ (tp)->seg_next == (struct tcpiphdr *)(tp) && \ (tp)->t_state == TCPS_ESTABLISHED) { \ tp->t_flags |= TF_DELACK; \ (tp)->rcv_nxt += (ti)->ti_len; \ flags = (ti)->ti_flags & TH_FIN; \ tcpstat.tcps_rcvpack++;\ tcpstat.tcps_rcvbyte += (ti)->ti_len;\ sbappend(&(so)->so_rcv, (m)); \ sorwakeup(so); \ } else { \ (flags) = tcp_reass((tp), (ti), (m)); \ tp->t_flags |= TF_ACKNOW; \ } \ } #ifndef TUBA_INCLUDE int tcp_reass(tp, ti, m) register struct tcpcb *tp; register struct tcpiphdr *ti; struct mbuf *m; { register struct tcpiphdr *q; struct socket *so = tp->t_inpcb->inp_socket; int flags; /* * Call with ti==0 after become established to * force pre-ESTABLISHED data up to user socket. */ if (ti == 0) goto present; /* * Find a segment which begins after this one does. */ for (q = tp->seg_next; q != (struct tcpiphdr *)tp; q = (struct tcpiphdr *)q->ti_next) if (SEQ_GT(q->ti_seq, ti->ti_seq)) break; /* * If there is a preceding segment, it may provide some of * our data already. If so, drop the data from the incoming * segment. If it provides all of our data, drop us. */ if ((struct tcpiphdr *)q->ti_prev != (struct tcpiphdr *)tp) { register int i; q = (struct tcpiphdr *)q->ti_prev; /* conversion to int (in i) handles seq wraparound */ i = q->ti_seq + q->ti_len - ti->ti_seq; if (i > 0) { if (i >= ti->ti_len) { tcpstat.tcps_rcvduppack++; tcpstat.tcps_rcvdupbyte += ti->ti_len; m_freem(m); return (0); } m_adj(m, i); ti->ti_len -= i; ti->ti_seq += i; } q = (struct tcpiphdr *)(q->ti_next); } tcpstat.tcps_rcvoopack++; tcpstat.tcps_rcvoobyte += ti->ti_len; REASS_MBUF(ti) = m; /* XXX */ /* * While we overlap succeeding segments trim them or, * if they are completely covered, dequeue them. */ while (q != (struct tcpiphdr *)tp) { register int i = (ti->ti_seq + ti->ti_len) - q->ti_seq; if (i <= 0) break; if (i < q->ti_len) { q->ti_seq += i; q->ti_len -= i; m_adj(REASS_MBUF(q), i); break; } q = (struct tcpiphdr *)q->ti_next; m = REASS_MBUF((struct tcpiphdr *)q->ti_prev); remque(q->ti_prev); m_freem(m); } /* * Stick new segment in its place. */ insque(ti, q->ti_prev); present: /* * Present data to user, advancing rcv_nxt through * completed sequence space. */ if (TCPS_HAVERCVDSYN(tp->t_state) == 0) return (0); ti = tp->seg_next; if (ti == (struct tcpiphdr *)tp || ti->ti_seq != tp->rcv_nxt) return (0); if (tp->t_state == TCPS_SYN_RECEIVED && ti->ti_len) return (0); do { tp->rcv_nxt += ti->ti_len; flags = ti->ti_flags & TH_FIN; remque(ti); m = REASS_MBUF(ti); ti = (struct tcpiphdr *)ti->ti_next; if (so->so_state & SS_CANTRCVMORE) m_freem(m); else sbappend(&so->so_rcv, m); } while (ti != (struct tcpiphdr *)tp && ti->ti_seq == tp->rcv_nxt); sorwakeup(so); return (flags); } /* * TCP input routine, follows pages 65-76 of the * protocol specification dated September, 1981 very closely. */ void tcp_input(m, iphlen) register struct mbuf *m; int iphlen; { register struct tcpiphdr *ti; register struct inpcb *inp; u_char *optp = NULL; int optlen; int len, tlen, off; register struct tcpcb *tp = 0; register int tiflags; struct socket *so; int todrop, acked, ourfinisacked, needoutput = 0; short ostate; struct in_addr laddr; int dropsocket = 0; int iss = 0; u_long tiwin, ts_val, ts_ecr; int ts_present = 0; #ifdef SACK int i; #endif tcpstat.tcps_rcvtotal++; /* * Get IP and TCP header together in first mbuf. * Note: IP leaves IP header in first mbuf. */ ti = mtod(m, struct tcpiphdr *); if (iphlen > sizeof (struct ip)) ip_stripoptions(m, (struct mbuf *)0); if (m->m_len < sizeof (struct tcpiphdr)) { if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == 0) { tcpstat.tcps_rcvshort++; return; } ti = mtod(m, struct tcpiphdr *); } /* * Checksum extended TCP header and data. */ tlen = ((struct ip *)ti)->ip_len; len = sizeof (struct ip) + tlen; ti->ti_next = ti->ti_prev = 0; ti->ti_x1 = 0; ti->ti_len = (u_short)tlen; HTONS(ti->ti_len); if (ti->ti_sum = in_cksum(m, len)) { tcpstat.tcps_rcvbadsum++; goto drop; } #endif /* TUBA_INCLUDE */ /* * Check that TCP offset makes sense, * pull out TCP options and adjust length. XXX */ off = ti->ti_off << 2; if (off < sizeof (struct tcphdr) || off > tlen) { tcpstat.tcps_rcvbadoff++; goto drop; } tlen -= off; ti->ti_len = tlen; if (off > sizeof (struct tcphdr)) { if (m->m_len < sizeof(struct ip) + off) { if ((m = m_pullup(m, sizeof (struct ip) + off)) == 0) { tcpstat.tcps_rcvshort++; return; } ti = mtod(m, struct tcpiphdr *); } optlen = off - sizeof (struct tcphdr); optp = mtod(m, u_char *) + sizeof (struct tcpiphdr); /* * Do quick retrieval of timestamp options ("options * prediction?"). If timestamp is the only option and it's * formatted as recommended in RFC 1323 appendix A, we * quickly get the values now and not bother calling * tcp_dooptions(), etc. */ if ((optlen == TCPOLEN_TSTAMP_APPA || (optlen > TCPOLEN_TSTAMP_APPA && optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) && *(u_long *)optp == htonl(TCPOPT_TSTAMP_HDR) && (ti->ti_flags & TH_SYN) == 0) { ts_present = 1; ts_val = ntohl(*(u_long *)(optp + 4)); ts_ecr = ntohl(*(u_long *)(optp + 8)); optp = NULL; /* we've parsed the options */ } } tiflags = ti->ti_flags; /* * Convert TCP protocol specific fields to host format. */ NTOHL(ti->ti_seq); NTOHL(ti->ti_ack); NTOHS(ti->ti_win); NTOHS(ti->ti_urp); /* * Locate pcb for segment. */ findpcb: inp = tcp_last_inpcb; if (inp->inp_lport != ti->ti_dport || inp->inp_fport != ti->ti_sport || inp->inp_faddr.s_addr != ti->ti_src.s_addr || inp->inp_laddr.s_addr != ti->ti_dst.s_addr) { inp = in_pcblookup(&tcb, ti->ti_src, ti->ti_sport, ti->ti_dst, ti->ti_dport, INPLOOKUP_WILDCARD); if (inp) tcp_last_inpcb = inp; ++tcpstat.tcps_pcbcachemiss; } /* * If the state is CLOSED (i.e., TCB does not exist) then * all data in the incoming segment is discarded. * If the TCB exists but is in CLOSED state, it is embryonic, * but should either do a listen or a connect soon. */ if (inp == 0) goto dropwithreset; tp = intotcpcb(inp); if (tp == 0) goto dropwithreset; if (tp->t_state == TCPS_CLOSED) goto drop; /* Unscale the window into a 32-bit value. */ if ((tiflags & TH_SYN) == 0) tiwin = ti->ti_win << tp->snd_scale; else tiwin = ti->ti_win; so = inp->inp_socket; if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) { if (so->so_options & SO_DEBUG) { ostate = tp->t_state; tcp_saveti = *ti; } if (so->so_options & SO_ACCEPTCONN) { if (tiflags & TH_ACK) { tcpstat.tcps_badsyn++; goto dropwithreset; } so = sonewconn(so, 0); if (so == 0) goto drop; /* * This is ugly, but .... * * Mark socket as temporary until we're * committed to keeping it. The code at * ``drop'' and ``dropwithreset'' check the * flag dropsocket to see if the temporary * socket created here should be discarded. * We mark the socket as discardable until * we're committed to it below in TCPS_LISTEN. */ dropsocket++; inp = (struct inpcb *)so->so_pcb; inp->inp_laddr = ti->ti_dst; inp->inp_lport = ti->ti_dport; #if BSD>=43 inp->inp_options = ip_srcroute(); #endif tp = intotcpcb(inp); tp->t_state = TCPS_LISTEN; /* Compute proper scaling value from buffer space */ while (tp->request_r_scale < TCP_MAX_WINSHIFT && TCP_MAXWIN << tp->request_r_scale < so->so_rcv.sb_hiwat) tp->request_r_scale++; } } /* * Segment received on connection. * Reset idle time and keep-alive timer. */ tp->t_idle = 0; tp->t_timer[TCPT_KEEP] = tcp_keepidle; #ifdef SACK /* * Delete stale (i.e, cumulatively ack'd) holes. */ if (!sack_disable && tp->t_state != TCPS_LISTEN) { tcp_seq lastack = max(ti->ti_ack, tp->snd_una); struct sackhole *cur = tp->snd_holes; struct sackhole *prev = cur; while (cur) if (SEQ_LEQ(cur->end, lastack)) { free(prev, M_PCB); prev = cur; cur = cur->next; tp->snd_numholes--; } else if (SEQ_LT(cur->start, lastack)) { cur->start = lastack; break; } else break; tp->snd_holes = cur; } #endif /* * Process options if not in LISTEN state, * else do it below (after getting remote address). */ if (optp && tp->t_state != TCPS_LISTEN) tcp_dooptions(tp, optp, optlen, ti, &ts_present, &ts_val, &ts_ecr); #ifdef SACK if (sack_disable) { tp->rcv_laststart = ti->ti_seq; /* last received segment */ tp->rcv_lastend = ti->ti_seq + ti->ti_len; } #endif /* * Header prediction: check for the two common cases * of a uni-directional data xfer. If the packet has * no control flags, is in-sequence, the window didn't * change and we're not retransmitting, it's a * candidate. If the length is zero and the ack moved * forward, we're the sender side of the xfer. Just * free the data acked & wake any higher level process * that was blocked waiting for space. If the length * is non-zero and the ack didn't move, we're the * receiver side. If we're getting packets in-order * (the reassembly queue is empty), add the data to * the socket buffer and note that we need a delayed ack. */ if (tp->t_state == TCPS_ESTABLISHED && (tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK && (!ts_present || TSTMP_GEQ(ts_val, tp->ts_recent)) && ti->ti_seq == tp->rcv_nxt && tiwin && tiwin == tp->snd_wnd && tp->snd_nxt == tp->snd_max) { /* * If last ACK falls within this segment's sequence numbers, * record the timestamp. */ if (ts_present && SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) && SEQ_LT(tp->last_ack_sent, ti->ti_seq + ti->ti_len)) { tp->ts_recent_age = tcp_now; tp->ts_recent = ts_val; } if (ti->ti_len == 0) { if (SEQ_GT(ti->ti_ack, tp->snd_una) && SEQ_LEQ(ti->ti_ack, tp->snd_max) && tp->snd_cwnd >= tp->snd_wnd) { /* * this is a pure ack for outstanding data. */ ++tcpstat.tcps_predack; if (ts_present) tcp_xmit_timer(tp, tcp_now - ts_ecr + 1); else if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq)) tcp_xmit_timer(tp, tp->t_rtt); acked = ti->ti_ack - tp->snd_una; tcpstat.tcps_rcvackpack++; tcpstat.tcps_rcvackbyte += acked; sbdrop(&so->so_snd, acked); tp->snd_una = ti->ti_ack; m_freem(m); /* * If all outstanding data are acked, stop * retransmit timer, otherwise restart timer * using current (possibly backed-off) value. * If process is waiting for space, * wakeup/selwakeup/signal. If data * are ready to send, let tcp_output * decide between more output or persist. */ if (tp->snd_una == tp->snd_max) tp->t_timer[TCPT_REXMT] = 0; else if (tp->t_timer[TCPT_PERSIST] == 0) tp->t_timer[TCPT_REXMT] = tp->t_rxtcur; if (so->so_snd.sb_flags & SB_NOTIFY) sowwakeup(so); if (so->so_snd.sb_cc) (void) tcp_output(tp); return; } } else if (ti->ti_ack == tp->snd_una && tp->seg_next == (struct tcpiphdr *)tp && ti->ti_len <= sbspace(&so->so_rcv)) { /* * this is a pure, in-sequence data packet * with nothing on the reassembly queue and * we have enough buffer space to take it. */ #ifdef SACK if (!sack_disable) { tp->rcv_numsacks = 0; for (i = 0; i < MAX_SACK_BLKS; i++) tp->sackblks[i].start= tp->sackblks[i].end=0; } #endif ++tcpstat.tcps_preddat; tp->rcv_nxt += ti->ti_len; tcpstat.tcps_rcvpack++; tcpstat.tcps_rcvbyte += ti->ti_len; /* * Drop TCP, IP headers and TCP options then add data * to socket buffer. */ m->m_data += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr); m->m_len -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr); sbappend(&so->so_rcv, m); sorwakeup(so); tp->t_flags |= TF_DELACK; return; } } /* * Drop TCP, IP headers and TCP options. */ m->m_data += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr); m->m_len -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr); /* * Calculate amount of space in receive window, * and then do TCP input processing. * Receive window is amount of space in rcv queue, * but not less than advertised window. */ { int win; win = sbspace(&so->so_rcv); if (win < 0) win = 0; tp->rcv_wnd = max(win, (int)(tp->rcv_adv - tp->rcv_nxt)); } switch (tp->t_state) { /* * If the state is LISTEN then ignore segment if it contains an RST. * If the segment contains an ACK then it is bad and send a RST. * If it does not contain a SYN then it is not interesting; drop it. * Don't bother responding if the destination was a broadcast. * Otherwise initialize tp->rcv_nxt, and tp->irs, select an initial * tp->iss, and send a segment: * * Also initialize tp->snd_nxt to tp->iss+1 and tp->snd_una to tp->iss. * Fill in remote peer address fields if not previously specified. * Enter SYN_RECEIVED state, and process any other fields of this * segment in this state. */ case TCPS_LISTEN: { struct mbuf *am; register struct sockaddr_in *sin; if (tiflags & TH_RST) goto drop; if (tiflags & TH_ACK) goto dropwithreset; if ((tiflags & TH_SYN) == 0) goto drop; /* * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN * in_broadcast() should never return true on a received * packet with M_BCAST not set. */ if (m->m_flags & (M_BCAST|M_MCAST) || IN_MULTICAST(ntohl(ti->ti_dst.s_addr))) goto drop; am = m_get(M_DONTWAIT, MT_SONAME); /* XXX */ if (am == NULL) goto drop; am->m_len = sizeof (struct sockaddr_in); sin = mtod(am, struct sockaddr_in *); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); sin->sin_addr = ti->ti_src; sin->sin_port = ti->ti_sport; bzero((caddr_t)sin->sin_zero, sizeof(sin->sin_zero)); laddr = inp->inp_laddr; if (inp->inp_laddr.s_addr == INADDR_ANY) inp->inp_laddr = ti->ti_dst; if (in_pcbconnect(inp, am)) { inp->inp_laddr = laddr; (void) m_free(am); goto drop; } (void) m_free(am); tp->t_template = tcp_template(tp); if (tp->t_template == 0) { tp = tcp_drop(tp, ENOBUFS); dropsocket = 0; /* socket is already gone */ goto drop; } if (optp) tcp_dooptions(tp, optp, optlen, ti, &ts_present, &ts_val, &ts_ecr); else tp->t_flags &= ~(TF_SEND_TSTMP | TF_USE_SCALE); if (iss) tp->iss = iss; else tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/4; tp->irs = ti->ti_seq; tcp_sendseqinit(tp); tcp_rcvseqinit(tp); tp->t_flags |= TF_ACKNOW; tp->t_state = TCPS_SYN_RECEIVED; tp->t_timer[TCPT_KEEP] = tcp_conntimeo; dropsocket = 0; /* committed to socket */ tcpstat.tcps_accepts++; goto trimthenstep6; } /* * If the state is SYN_SENT: * if seg contains an ACK, but not for our SYN, drop the input. * if seg contains a RST, then drop the connection. * if seg does not contain SYN, then drop it. * Otherwise this is an acceptable SYN segment * initialize tp->rcv_nxt and tp->irs * if seg contains ack then advance tp->snd_una * if SYN has been acked change to ESTABLISHED else SYN_RCVD state * arrange for segment to be acked (eventually) * continue processing rest of data/controls, beginning with URG */ case TCPS_SYN_SENT: if ((tiflags & TH_ACK) && (SEQ_LEQ(ti->ti_ack, tp->iss) || SEQ_GT(ti->ti_ack, tp->snd_max))) goto dropwithreset; if (tiflags & TH_RST) { if (tiflags & TH_ACK) tp = tcp_drop(tp, ECONNREFUSED); goto drop; } if ((tiflags & TH_SYN) == 0) goto drop; if (tiflags & TH_ACK) { tp->snd_una = ti->ti_ack; if (SEQ_LT(tp->snd_nxt, tp->snd_una)) tp->snd_nxt = tp->snd_una; } tp->t_timer[TCPT_REXMT] = 0; tp->irs = ti->ti_seq; tcp_rcvseqinit(tp); tp->t_flags |= TF_ACKNOW; if (tiflags & TH_ACK && SEQ_GT(tp->snd_una, tp->iss)) { tcpstat.tcps_connects++; soisconnected(so); tp->t_state = TCPS_ESTABLISHED; /* Do window scaling on this connection? */ if (tp->t_flags & TF_USE_SCALE) { tp->snd_scale = tp->requested_s_scale; tp->rcv_scale = tp->request_r_scale; } (void) tcp_reass(tp, (struct tcpiphdr *)0, (struct mbuf *)0); /* * if we didn't have to retransmit the SYN, * use its rtt as our initial srtt & rtt var. */ if (tp->t_rtt) tcp_xmit_timer(tp, tp->t_rtt); } else tp->t_state = TCPS_SYN_RECEIVED; trimthenstep6: /* * Advance ti->ti_seq to correspond to first data byte. * If data, trim to stay within window, * dropping FIN if necessary. */ ti->ti_seq++; if (ti->ti_len > tp->rcv_wnd) { todrop = ti->ti_len - tp->rcv_wnd; m_adj(m, -todrop); ti->ti_len = tp->rcv_wnd; tiflags &= ~TH_FIN; tcpstat.tcps_rcvpackafterwin++; tcpstat.tcps_rcvbyteafterwin += todrop; } tp->snd_wl1 = ti->ti_seq - 1; tp->rcv_up = ti->ti_seq; goto step6; } /* * States other than LISTEN or SYN_SENT. * First check timestamp, if present. * Then check that at least some bytes of segment are within * receive window. If segment begins before rcv_nxt, * drop leading data (and SYN); if nothing left, just ack. * * RFC 1323 PAWS: If we have a timestamp reply on this segment * and it's less than ts_recent, drop it. */ if (ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent && TSTMP_LT(ts_val, tp->ts_recent)) { /* Check to see if ts_recent is over 24 days old. */ if ((int)(tcp_now - tp->ts_recent_age) > TCP_PAWS_IDLE) { /* * Invalidate ts_recent. If this segment updates * ts_recent, the age will be reset later and ts_recent * will get a valid value. If it does not, setting * ts_recent to zero will at least satisfy the * requirement that zero be placed in the timestamp * echo reply when ts_recent isn't valid. The * age isn't reset until we get a valid ts_recent * because we don't want out-of-order segments to be * dropped when ts_recent is old. */ tp->ts_recent = 0; } else { tcpstat.tcps_rcvduppack++; tcpstat.tcps_rcvdupbyte += ti->ti_len; tcpstat.tcps_pawsdrop++; goto dropafterack; } } todrop = tp->rcv_nxt - ti->ti_seq; if (todrop > 0) { if (tiflags & TH_SYN) { tiflags &= ~TH_SYN; ti->ti_seq++; if (ti->ti_urp > 1) ti->ti_urp--; else tiflags &= ~TH_URG; todrop--; } if (todrop >= ti->ti_len) { tcpstat.tcps_rcvduppack++; tcpstat.tcps_rcvdupbyte += ti->ti_len; /* * If segment is just one to the left of the window, * check three special cases: * 1. Don't toss RST in response to 4.2-style keepalive. * 2. If the only thing to drop is a FIN, we can drop * it, but check the ACK or we will get into FIN * wars if our FINs crossed (both CLOSING). * 3. If we have sent a window probe, it may or may not * have been accepted. If window probes crossed, * we must accept ACK on segments one to the left * of the window, or we can get ACK wars after * exchanging probes. (After sending a probe, * ACK-only packets are sent with the pre-probe * sequence number.) * In any of these cases, send ACK to resynchronize, * but keep on processing for RST or ACK. */ if (((tiflags & TH_FIN || todrop == 1) && todrop == ti->ti_len + 1) #ifdef TCP_COMPAT_42 || (tiflags & TH_RST && ti->ti_seq == tp->rcv_nxt - 1) #endif ) { todrop = ti->ti_len; tiflags &= ~TH_FIN; tp->t_flags |= TF_ACKNOW; } else { /* * Handle the case when a bound socket connects * to itself. Allow packets with a SYN and * an ACK to continue with the processing. */ if (todrop != 0 || (tiflags & TH_ACK) == 0) goto dropafterack; } } else { tcpstat.tcps_rcvpartduppack++; tcpstat.tcps_rcvpartdupbyte += todrop; } m_adj(m, todrop); ti->ti_seq += todrop; ti->ti_len -= todrop; if (ti->ti_urp > todrop) ti->ti_urp -= todrop; else { tiflags &= ~TH_URG; ti->ti_urp = 0; } } /* * If new data are received on a connection after the * user processes are gone, then RST the other end. */ if ((so->so_state & SS_NOFDREF) && tp->t_state > TCPS_CLOSE_WAIT && ti->ti_len) { tp = tcp_close(tp); tcpstat.tcps_rcvafterclose++; goto dropwithreset; } /* * If segment ends after window, drop trailing data * (and PUSH and FIN); if nothing left, just ACK. */ todrop = (ti->ti_seq+ti->ti_len) - (tp->rcv_nxt+tp->rcv_wnd); if (todrop > 0) { tcpstat.tcps_rcvpackafterwin++; if (todrop >= ti->ti_len) { tcpstat.tcps_rcvbyteafterwin += ti->ti_len; /* * If a new connection request is received * while in TIME_WAIT, drop the old connection * and start over if the sequence numbers * are above the previous ones. */ if (tiflags & TH_SYN && tp->t_state == TCPS_TIME_WAIT && SEQ_GT(ti->ti_seq, tp->rcv_nxt)) { iss = tp->snd_nxt + TCP_ISSINCR; tp = tcp_close(tp); goto findpcb; } /* * If window is closed can only take segments at * window edge, and have to drop data and PUSH from * incoming segments. Continue processing, but * remember to ack. Otherwise, drop segment * and ack. */ if (tp->rcv_wnd == 0 && ti->ti_seq == tp->rcv_nxt) { tp->t_flags |= TF_ACKNOW; tcpstat.tcps_rcvwinprobe++; } else goto dropafterack; } else tcpstat.tcps_rcvbyteafterwin += todrop; m_adj(m, -todrop); ti->ti_len -= todrop; tiflags &= ~(TH_PUSH|TH_FIN); } /* * If last ACK falls within this segment's sequence numbers, * record its timestamp. */ if (ts_present && SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) && SEQ_LT(tp->last_ack_sent, ti->ti_seq + ti->ti_len + ((tiflags & (TH_SYN|TH_FIN)) != 0))) { tp->ts_recent_age = tcp_now; tp->ts_recent = ts_val; } /* * If the RST bit is set examine the state: * SYN_RECEIVED STATE: * If passive open, return to LISTEN state. * If active open, inform user that connection was refused. * ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES: * Inform user that connection was reset, and close tcb. * CLOSING, LAST_ACK, TIME_WAIT STATES * Close the tcb. */ if (tiflags&TH_RST) switch (tp->t_state) { case TCPS_SYN_RECEIVED: so->so_error = ECONNREFUSED; goto close; case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: so->so_error = ECONNRESET; close: tp->t_state = TCPS_CLOSED; tcpstat.tcps_drops++; tp = tcp_close(tp); goto drop; case TCPS_CLOSING: case TCPS_LAST_ACK: case TCPS_TIME_WAIT: tp = tcp_close(tp); goto drop; } /* * If a SYN is in the window, then this is an * error and we send an RST and drop the connection. */ if (tiflags & TH_SYN) { tp = tcp_drop(tp, ECONNRESET); goto dropwithreset; } /* * If the ACK bit is off we drop the segment and return. */ if ((tiflags & TH_ACK) == 0) goto drop; /* * Ack processing. */ switch (tp->t_state) { /* * In SYN_RECEIVED state if the ack ACKs our SYN then enter * ESTABLISHED state and continue processing, otherwise * send an RST. */ case TCPS_SYN_RECEIVED: if (SEQ_GT(tp->snd_una, ti->ti_ack) || SEQ_GT(ti->ti_ack, tp->snd_max)) goto dropwithreset; tcpstat.tcps_connects++; soisconnected(so); tp->t_state = TCPS_ESTABLISHED; /* Do window scaling? */ if (tp->t_flags & TF_USE_SCALE) { tp->snd_scale = tp->requested_s_scale; tp->rcv_scale = tp->request_r_scale; } (void) tcp_reass(tp, (struct tcpiphdr *)0, (struct mbuf *)0); tp->snd_wl1 = ti->ti_seq - 1; /* fall into ... */ /* * In ESTABLISHED state: drop duplicate ACKs; ACK out of range * ACKs. If the ack is in the range * tp->snd_una < ti->ti_ack <= tp->snd_max * then advance tp->snd_una to ti->ti_ack and drop * data from the retransmission queue. If this ACK reflects * more up to date window information we update our window information. */ case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: case TCPS_CLOSING: case TCPS_LAST_ACK: case TCPS_TIME_WAIT: if (SEQ_LEQ(ti->ti_ack, tp->snd_una)) { if (ti->ti_len == 0 && tiwin == tp->snd_wnd) { tcpstat.tcps_rcvdupack++; /* * If we have outstanding data (other than * a window probe), this is a completely * duplicate ack (ie, window info didn't * change), the ack is the biggest we've * seen and we've seen exactly our rexmt * threshhold of them, assume a packet * has been dropped and retransmit it. * Kludge snd_nxt & the congestion * window so we send only this one * packet. * * We know we're losing at the current * window size so do congestion avoidance * (set ssthresh to half the current window * and pull our congestion window back to * the new ssthresh). * * Dup acks mean that packets have left the * network (they're now cached at the receiver) * so bump cwnd by the amount in the receiver * to keep a constant cwnd packets in the * network. */ if (tp->t_timer[TCPT_REXMT] == 0 || ti->ti_ack != tp->snd_una) { tp->t_dupacks = 0; } else if (++tp->t_dupacks == tcprexmtthresh) { tcp_seq onxt = tp->snd_nxt; u_int win = min(tp->snd_wnd, tp->snd_cwnd)/ 2 / tp->t_maxseg; if (win < 2) win = 2; tp->snd_ssthresh = win * tp->t_maxseg; tp->t_timer[TCPT_REXMT] = 0; tp->t_rtt = 0; tp->snd_nxt = ti->ti_ack; tp->snd_cwnd = tp->t_maxseg; (void) tcp_output(tp); #ifdef SACK if (SEQ_GT(tp->snd_max, tp->snd_una) && tp->t_timer[TCPT_REXMT] == 0 && tp->t_timer[TCPT_PERSIST] == 0) tp->t_timer[TCPT_REXMT] = tp->t_rxtcur; #endif tp->snd_cwnd = tp->snd_ssthresh + tp->t_maxseg * tp->t_dupacks; if (SEQ_GT(onxt, tp->snd_nxt)) tp->snd_nxt = onxt; goto drop; } else if (tp->t_dupacks > tcprexmtthresh) { /* XXX shd use SACK info for this */ tp->snd_cwnd += tp->t_maxseg; /* * Rather than call tcp_output() on * every dup ack, call it only on * every other (i.e. odd) one here. */ #ifdef SACK if (sack_disable || tp->t_dupacks%2 == 1) #endif (void) tcp_output(tp); goto drop; } } else tp->t_dupacks = 0; break; } /* * If the congestion window was inflated to account * for the other side's cached packets, retract it. */ if (tp->t_dupacks > tcprexmtthresh && tp->snd_cwnd > tp->snd_ssthresh) { tp->snd_cwnd = tp->snd_ssthresh; } tp->t_dupacks = 0; if (SEQ_GT(ti->ti_ack, tp->snd_max)) { tcpstat.tcps_rcvacktoomuch++; goto dropafterack; } acked = ti->ti_ack - tp->snd_una; tcpstat.tcps_rcvackpack++; tcpstat.tcps_rcvackbyte += acked; /* * If we have a timestamp reply, update smoothed * round trip time. If no timestamp is present but * transmit timer is running and timed sequence * number was acked, update smoothed round trip time. * Since we now have an rtt measurement, cancel the * timer backoff (cf., Phil Karn's retransmit alg.). * Recompute the initial retransmit timer. */ if (ts_present) tcp_xmit_timer(tp, tcp_now - ts_ecr + 1); else if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq)) tcp_xmit_timer(tp, tp->t_rtt); /* * If all outstanding data is acked, stop retransmit * timer and remember to restart (more output or persist). * If there is more data to be acked, restart retransmit * timer, using current (possibly backed-off) value. */ if (ti->ti_ack == tp->snd_max) { tp->t_timer[TCPT_REXMT] = 0; needoutput = 1; } else if (tp->t_timer[TCPT_PERSIST] == 0) tp->t_timer[TCPT_REXMT] = tp->t_rxtcur; /* * When new data is acked, open the congestion window. * If the window gives us less than ssthresh packets * in flight, open exponentially (maxseg per packet). * Otherwise open linearly: maxseg per window * (maxseg * (maxseg / cwnd) per packet). */ { register u_int cw = tp->snd_cwnd; register u_int incr = tp->t_maxseg; if (cw > tp->snd_ssthresh) incr = incr * incr / cw; tp->snd_cwnd = min(cw + incr, TCP_MAXWIN<snd_scale); } if (acked > so->so_snd.sb_cc) { tp->snd_wnd -= so->so_snd.sb_cc; sbdrop(&so->so_snd, (int)so->so_snd.sb_cc); ourfinisacked = 1; } else { sbdrop(&so->so_snd, acked); tp->snd_wnd -= acked; ourfinisacked = 0; } if (so->so_snd.sb_flags & SB_NOTIFY) sowwakeup(so); tp->snd_una = ti->ti_ack; if (SEQ_LT(tp->snd_nxt, tp->snd_una)) tp->snd_nxt = tp->snd_una; switch (tp->t_state) { /* * In FIN_WAIT_1 STATE in addition to the processing * for the ESTABLISHED state if our FIN is now acknowledged * then enter FIN_WAIT_2. */ case TCPS_FIN_WAIT_1: if (ourfinisacked) { /* * If we can't receive any more * data, then closing user can proceed. * Starting the timer is contrary to the * specification, but if we don't get a FIN * we'll hang forever. */ if (so->so_state & SS_CANTRCVMORE) { soisdisconnected(so); tp->t_timer[TCPT_2MSL] = tcp_maxidle; } tp->t_state = TCPS_FIN_WAIT_2; } break; /* * In CLOSING STATE in addition to the processing for * the ESTABLISHED state if the ACK acknowledges our FIN * then enter the TIME-WAIT state, otherwise ignore * the segment. */ case TCPS_CLOSING: if (ourfinisacked) { tp->t_state = TCPS_TIME_WAIT; tcp_canceltimers(tp); tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL; soisdisconnected(so); } break; /* * In LAST_ACK, we may still be waiting for data to drain * and/or to be acked, as well as for the ack of our FIN. * If our FIN is now acknowledged, delete the TCB, * enter the closed state and return. */ case TCPS_LAST_ACK: if (ourfinisacked) { tp = tcp_close(tp); goto drop; } break; /* * In TIME_WAIT state the only thing that should arrive * is a retransmission of the remote FIN. Acknowledge * it and restart the finack timer. */ case TCPS_TIME_WAIT: tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL; goto dropafterack; } } step6: /* * Update window information. * Don't look at window if no ACK: TAC's send garbage on first SYN. */ if ((tiflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, ti->ti_seq) || tp->snd_wl1 == ti->ti_seq && (SEQ_LT(tp->snd_wl2, ti->ti_ack) || tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd))) { /* keep track of pure window updates */ if (ti->ti_len == 0 && tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd) tcpstat.tcps_rcvwinupd++; tp->snd_wnd = tiwin; tp->snd_wl1 = ti->ti_seq; tp->snd_wl2 = ti->ti_ack; if (tp->snd_wnd > tp->max_sndwnd) tp->max_sndwnd = tp->snd_wnd; needoutput = 1; } /* * Process segments with URG. */ if ((tiflags & TH_URG) && ti->ti_urp && TCPS_HAVERCVDFIN(tp->t_state) == 0) { /* * This is a kludge, but if we receive and accept * random urgent pointers, we'll crash in * soreceive. It's hard to imagine someone * actually wanting to send this much urgent data. */ if (ti->ti_urp + so->so_rcv.sb_cc > sb_max) { ti->ti_urp = 0; /* XXX */ tiflags &= ~TH_URG; /* XXX */ goto dodata; /* XXX */ } /* * If this segment advances the known urgent pointer, * then mark the data stream. This should not happen * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since * a FIN has been received from the remote side. * In these states we ignore the URG. * * According to RFC961 (Assigned Protocols), * the urgent pointer points to the last octet * of urgent data. We continue, however, * to consider it to indicate the first octet * of data past the urgent section as the original * spec states (in one of two places). */ if (SEQ_GT(ti->ti_seq+ti->ti_urp, tp->rcv_up)) { tp->rcv_up = ti->ti_seq + ti->ti_urp; so->so_oobmark = so->so_rcv.sb_cc + (tp->rcv_up - tp->rcv_nxt) - 1; if (so->so_oobmark == 0) so->so_state |= SS_RCVATMARK; sohasoutofband(so); tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA); } /* * Remove out of band data so doesn't get presented to user. * This can happen independent of advancing the URG pointer, * but if two URG's are pending at once, some out-of-band * data may creep in... ick. */ if (ti->ti_urp <= ti->ti_len #ifdef SO_OOBINLINE && (so->so_options & SO_OOBINLINE) == 0 #endif ) tcp_pulloutofband(so, ti, m); } else /* * If no out of band data is expected, * pull receive urgent pointer along * with the receive window. */ if (SEQ_GT(tp->rcv_nxt, tp->rcv_up)) tp->rcv_up = tp->rcv_nxt; dodata: /* XXX */ /* * Process the segment text, merging it into the TCP sequencing queue, * and arranging for acknowledgment of receipt if necessary. * This process logically involves adjusting tp->rcv_wnd as data * is presented to the user (this happens in tcp_usrreq.c, * case PRU_RCVD). If a FIN has already been received on this * connection then we just ignore the text. */ if ((ti->ti_len || (tiflags&TH_FIN)) && TCPS_HAVERCVDFIN(tp->t_state) == 0) { TCP_REASS(tp, ti, m, so, tiflags); #ifdef SACK if (!sack_disable) tcp_generate_sacks(tp); #endif /* * Note the amount of data that peer has sent into * our window, in order to estimate the sender's * buffer size. */ len = so->so_rcv.sb_hiwat - (tp->rcv_adv - tp->rcv_nxt); } else { m_freem(m); tiflags &= ~TH_FIN; } /* * If FIN is received ACK the FIN and let the user know * that the connection is closing. */ if (tiflags & TH_FIN) { if (TCPS_HAVERCVDFIN(tp->t_state) == 0) { socantrcvmore(so); tp->t_flags |= TF_ACKNOW; tp->rcv_nxt++; } switch (tp->t_state) { /* * In SYN_RECEIVED and ESTABLISHED STATES * enter the CLOSE_WAIT state. */ case TCPS_SYN_RECEIVED: case TCPS_ESTABLISHED: tp->t_state = TCPS_CLOSE_WAIT; break; /* * If still in FIN_WAIT_1 STATE FIN has not been acked so * enter the CLOSING state. */ case TCPS_FIN_WAIT_1: tp->t_state = TCPS_CLOSING; break; /* * In FIN_WAIT_2 state enter the TIME_WAIT state, * starting the time-wait timer, turning off the other * standard timers. */ case TCPS_FIN_WAIT_2: tp->t_state = TCPS_TIME_WAIT; tcp_canceltimers(tp); tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL; soisdisconnected(so); break; /* * In TIME_WAIT state restart the 2 MSL time_wait timer. */ case TCPS_TIME_WAIT: tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL; break; } } if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, &tcp_saveti, 0); /* * Return any desired output. */ if (needoutput || (tp->t_flags & TF_ACKNOW)) (void) tcp_output(tp); return; dropafterack: /* * Generate an ACK dropping incoming segment if it occupies * sequence space, where the ACK reflects our state. */ if (tiflags & TH_RST) goto drop; m_freem(m); tp->t_flags |= TF_ACKNOW; (void) tcp_output(tp); return; dropwithreset: /* * Generate a RST, dropping incoming segment. * Make ACK acceptable to originator of segment. * Don't bother to respond if destination was broadcast/multicast. */ if ((tiflags & TH_RST) || m->m_flags & (M_BCAST|M_MCAST) || IN_MULTICAST(ntohl(ti->ti_dst.s_addr))) goto drop; if (tiflags & TH_ACK) tcp_respond(tp, ti, m, (tcp_seq)0, ti->ti_ack, TH_RST); else { if (tiflags & TH_SYN) ti->ti_len++; tcp_respond(tp, ti, m, ti->ti_seq+ti->ti_len, (tcp_seq)0, TH_RST|TH_ACK); } /* destroy temporarily created socket */ if (dropsocket) (void) soabort(so); return; drop: /* * Drop space held by incoming segment and return. */ if (tp && (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_DROP, ostate, tp, &tcp_saveti, 0); m_freem(m); /* destroy temporarily created socket */ if (dropsocket) (void) soabort(so); return; #ifndef TUBA_INCLUDE } #ifdef SACK void tcp_generate_sacks(struct tcpcb *tp) { /* * First reported block MUST be the most recent one. Subsequent * blocks SHOULD be in the order in which they arrived at the * receiver. These two conditions make the implementation fully * compliant with the Internet Draft (Jan. 1996). */ int i, j = 0, count = 0, lastpos = -1; struct sackblk sack, firstsack, temp[MAX_SACK_BLKS]; /* First clean up current list of sacks */ for (i = 0; i < tp->rcv_numsacks; i++) { sack = tp->sackblks[i]; if (sack.start == 0 && sack.end == 0) count++; /* count = number of blocks to be discarded */ if (SEQ_LEQ(sack.end, tp->rcv_nxt)) { tp->sackblks[i].start = tp->sackblks[i].end = 0; count++; } else { temp[j].start = tp->sackblks[i].start; temp[j++].end = tp->sackblks[i].end; } } tp->rcv_numsacks -= count; if (tp->rcv_numsacks == 0) { /* no sack blocks currently (fast path) */ if (SEQ_LT(tp->rcv_nxt, tp->rcv_laststart)) { /* ==> need first sack block */ tp->sackblks[0].start = tp->rcv_laststart; tp->sackblks[0].end = tp->rcv_lastend; tp->rcv_numsacks = 1; } return; } /* Otherwise, sack blocks are already present. */ for (i = 0; i < tp->rcv_numsacks; i++) tp->sackblks[i] = temp[i]; /* first copy back sack list */ if (SEQ_GEQ(tp->rcv_nxt, tp->rcv_lastend)) return; /* sack list remains unchanged */ /* * From here, segment just received should be (part of) the 1st sack. * Go through list possibly coalescing sack block entries. */ firstsack.start = tp->rcv_laststart; firstsack.end = tp->rcv_lastend; for (i = 0; i < tp->rcv_numsacks; i++) { sack = tp->sackblks[i]; if (SEQ_LT(sack.end, firstsack.start) || SEQ_GT(sack.start, firstsack.end) || (sack.start == firstsack.start && sack.end == firstsack.end)) continue; /* either no overlap or identical block */ if (SEQ_LEQ(sack.start, firstsack.start)) firstsack.start = sack.start; /* merge blocks */ if (SEQ_GEQ(sack.end, firstsack.end)) firstsack.end = sack.end; /* merge blocks */ tp->sackblks[i].start = tp->sackblks[i].end = 0; lastpos = i; /* last posn with a zero entry */ } if (lastpos != -1) { /* at least one merge */ for (i = 0, j = 1; i < tp->rcv_numsacks; i++) { sack = tp->sackblks[i]; if (sack.start == 0 && sack.end == 0) continue; temp[j++] = sack; } tp->rcv_numsacks = j; /* including first blk (added later) */ for (i = 1; i < tp->rcv_numsacks; i++) /* now copy back */ tp->sackblks[i] = temp[i]; } else { /* no merges -- shift sacks by 1 */ if (tp->rcv_numsacks < MAX_SACK_BLKS) tp->rcv_numsacks++; for (i = tp->rcv_numsacks-1; i > 0; i--) tp->sackblks[i] = tp->sackblks[i-1]; } tp->sackblks[0] = firstsack; return; } #endif SACK void tcp_dooptions(tp, cp, cnt, ti, ts_present, ts_val, ts_ecr) struct tcpcb *tp; u_char *cp; int cnt; struct tcpiphdr *ti; int *ts_present; u_long *ts_val, *ts_ecr; { u_short mss; int opt, optlen; int scale_present = 0; for (; cnt > 0; cnt -= optlen, cp += optlen) { opt = cp[0]; if (opt == TCPOPT_EOL) break; if (opt == TCPOPT_NOP) optlen = 1; else { optlen = cp[1]; if (optlen <= 0) break; } switch (opt) { default: continue; case TCPOPT_MAXSEG: if (optlen != TCPOLEN_MAXSEG) continue; if (!(ti->ti_flags & TH_SYN)) continue; bcopy((char *) cp + 2, (char *) &mss, sizeof(mss)); NTOHS(mss); (void) tcp_mss(tp, mss); /* sets t_maxseg */ break; case TCPOPT_WINDOW: if (optlen != TCPOLEN_WINDOW) continue; if (!(ti->ti_flags & TH_SYN)) continue; scale_present = 1; tp->requested_s_scale = min(cp[2], TCP_MAX_WINSHIFT); break; case TCPOPT_TIMESTAMP: if (optlen != TCPOLEN_TIMESTAMP) continue; *ts_present = 1; bcopy((char *)cp + 2, (char *)ts_val, sizeof(*ts_val)); NTOHL(*ts_val); bcopy((char *)cp + 6, (char *)ts_ecr, sizeof(*ts_ecr)); NTOHL(*ts_ecr); /* * A timestamp received in a SYN makes * it ok to send timestamp requests and replies. */ if (ti->ti_flags & TH_SYN) { tp->ts_recent = *ts_val; tp->ts_recent_age = tcp_now; } break; #ifdef SACK case TCPOPT_SACK_PERMITTED: if (sack_disable || optlen != TCPOLEN_SACK_PERMITTED) continue; if (ti->ti_flags & TH_SYN) { /* MUST only be set on SYN */ tp->t_flags |= TF_SACK_PERMIT; } break; case TCPOPT_SACK: { int i, tmp_olen, off, len; tcp_seq lastack = max(ti->ti_ack, tp->snd_una); u_char *tmp_cp; struct sackblk sack; struct sackhole *cur, *p, *temp; if (sack_disable) continue; /* Note: TCPOLEN_SACK must be 2*sizeof(tcp_seq) */ if (optlen <= 2 || (optlen - 2) % TCPOLEN_SACK != 0) continue; tmp_cp = cp + 2; tmp_olen = optlen - 2; if (tp->snd_numholes < 0) tp->snd_numholes = 0; while (tmp_olen > 0) { struct sackblk sack; bcopy((char *) tmp_cp, (char *) &(sack.start), sizeof(tcp_seq)); NTOHL(sack.start); bcopy((char *) tmp_cp + sizeof(tcp_seq), (char *) &(sack.end), sizeof(tcp_seq)); NTOHL(sack.end); tmp_olen -= TCPOLEN_SACK; tmp_cp += TCPOLEN_SACK; if (SEQ_LT(sack.end, sack.start)) continue; /* bad SACK fields */ if (SEQ_LEQ(sack.end, tp->snd_una)) continue; if (tp->snd_holes == 0) { /* first hole */ tp->snd_holes = (struct sackhole *) malloc(sizeof(struct sackhole), M_PCB, M_NOWAIT); cur = tp->snd_holes; cur->start = ti->ti_ack; cur->end = sack.start; cur->dups = 1; cur->rxmit = cur->start; cur->next = 0; tp->snd_numholes = 1; tp->rcv_lastsack = sack.end; continue; /* with next sack block */ } /* Go thru list of holes */ p = cur = tp->snd_holes; while (cur) { /* * First 2 cases can be used to * do different things (e.g, failed * rexmissions?) */ if (SEQ_LEQ(sack.end, cur->start)) { if (ti->ti_ack==tp->snd_una) /*&& ti->ti_win==tp->snd_wnd)*/ cur->dups++; p = cur; cur = cur->next; continue; } if (SEQ_GEQ(sack.start, cur->end)) { if (ti->ti_ack==tp->snd_una) /* && ti->ti_win==tp->snd_wnd)*/ cur->dups++; p = cur; cur = cur->next; continue; } if (SEQ_LEQ(sack.start, cur->start)) { if(SEQ_GEQ(sack.end,cur->end)){ /* delete hole */ if (p != cur) { p->next=cur->next; free(cur, M_PCB); cur = p->next; } else { cur=cur->next; free(p, M_PCB); p = cur; tp->snd_holes=p; } tp->snd_numholes--; continue; } /* otherwise */ cur->start = sack.end; cur->rxmit = max (cur->rxmit, cur->start); p = cur; cur = cur->next; continue; } if (SEQ_GEQ(sack.end, cur->end)) { cur->end = sack.start; cur->rxmit = min (cur->rxmit, cur->end); p = cur; cur = cur->next; continue; } if (SEQ_LT(cur->start, sack.start) && SEQ_GT(cur->end, sack.end)) { /* need to split cur */ temp = (struct sackhole *) malloc(sizeof(*temp), M_PCB,M_NOWAIT); temp->next = cur->next; temp->start = sack.end; temp->end = cur->end; temp->dups = cur->dups; /*+1?*/ temp->rxmit = max (cur->rxmit, temp->start); cur->end = sack.start; cur->rxmit = min (cur->rxmit, cur->end); cur->dups++; cur->next = temp; p = temp; cur = p->next; tp->snd_numholes++; } } if (SEQ_LT(tp->rcv_lastsack, sack.start)) { /* * Need to append new hole at end. * Last hole is p (and it's not NULL). */ temp = (struct sackhole *) malloc(sizeof(*temp), M_PCB, M_NOWAIT); temp->start = tp->rcv_lastsack; temp->end = sack.start; temp->dups = 1; temp->rxmit = temp->start; temp->next = 0; p->next = temp; tp->rcv_lastsack = sack.end; tp->snd_numholes++; } } break; } #endif SACK } } if (ti->ti_flags & TH_SYN) { if (*ts_present == 0) tp->t_flags &= ~TF_SEND_TSTMP; if (scale_present == 0) tp->t_flags &= ~TF_USE_SCALE; } } /* * Pull out of band byte out of a segment so * it doesn't appear in the user's data queue. * It is still reflected in the segment length for * sequencing purposes. */ void tcp_pulloutofband(so, ti, m) struct socket *so; struct tcpiphdr *ti; register struct mbuf *m; { int cnt = ti->ti_urp - 1; while (cnt >= 0) { if (m->m_len > cnt) { char *cp = mtod(m, caddr_t) + cnt; struct tcpcb *tp = sototcpcb(so); tp->t_iobc = *cp; tp->t_oobflags |= TCPOOB_HAVEDATA; bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1)); m->m_len--; return; } cnt -= m->m_len; m = m->m_next; if (m == 0) break; } panic("tcp_pulloutofband"); } /* * Collect new round-trip time estimate * and update averages and current timeout. */ void tcp_xmit_timer(tp, rtt) register struct tcpcb *tp; short rtt; { register short delta; tcpstat.tcps_rttupdated++; if (tp->t_srtt != 0) { /* * srtt is stored as fixed point with 3 bits after the * binary point (i.e., scaled by 8). The following magic * is equivalent to the smoothing algorithm in rfc793 with * an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed * point). Adjust rtt to origin 0. */ delta = rtt - 1 - (tp->t_srtt >> TCP_RTT_SHIFT); if ((tp->t_srtt += delta) <= 0) tp->t_srtt = 1; /* * We accumulate a smoothed rtt variance (actually, a * smoothed mean difference), then set the retransmit * timer to smoothed rtt + 4 times the smoothed variance. * rttvar is stored as fixed point with 2 bits after the * binary point (scaled by 4). The following is * equivalent to rfc793 smoothing with an alpha of .75 * (rttvar = rttvar*3/4 + |delta| / 4). This replaces * rfc793's wired-in beta. */ if (delta < 0) delta = -delta; delta -= (tp->t_rttvar >> TCP_RTTVAR_SHIFT); if ((tp->t_rttvar += delta) <= 0) tp->t_rttvar = 1; } else { /* * No rtt measurement yet - use the unsmoothed rtt. * Set the variance to half the rtt (so our first * retransmit happens at 3*rtt). */ tp->t_srtt = rtt << TCP_RTT_SHIFT; tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1); } tp->t_rtt = 0; tp->t_rxtshift = 0; /* * the retransmit should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). */ TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), tp->t_rttmin, TCPTV_REXMTMAX); /* * We received an ack for a packet that wasn't retransmitted; * it is probably safe to discard any error indications we've * received recently. This isn't quite right, but close enough * for now (a route might have failed after we sent a segment, * and the return path might not be symmetrical). */ tp->t_softerror = 0; } /* * Determine a reasonable value for maxseg size. * If the route is known, check route for mtu. * If none, use an mss that can be handled on the outgoing * interface without forcing IP to fragment; if bigger than * an mbuf cluster (MCLBYTES), round down to nearest multiple of MCLBYTES * to utilize large mbufs. If no route is found, route has no mtu, * or the destination isn't local, use a default, hopefully conservative * size (usually 512 or the default IP max size, but no more than the mtu * of the interface), as we can't discover anything about intervening * gateways or networks. We also initialize the congestion/slow start * window to be a single segment if the destination isn't local. * While looking at the routing entry, we also initialize other path-dependent * parameters from pre-set or cached values in the routing entry. * * If (potentially) using the timestamp option, the mss value that * we send to the peer is not reduced for the option. However, we * set t_maxseg to a value small enough for the option if we will * be sending it. */ int tcp_mss(tp, offer) register struct tcpcb *tp; u_int offer; { struct route *ro; register struct rtentry *rt; struct ifnet *ifp; register int rtt, mss; /* mss is size to offer */ int maxseg; /* maxseg is size of t_maxseg */ u_long bufsize; struct inpcb *inp; struct socket *so; extern int tcp_mssdflt; inp = tp->t_inpcb; ro = &inp->inp_route; if ((rt = ro->ro_rt) == (struct rtentry *)0) { /* No route yet, so try to acquire one */ if (inp->inp_faddr.s_addr != INADDR_ANY) { ro->ro_dst.sa_family = AF_INET; ro->ro_dst.sa_len = sizeof(ro->ro_dst); ((struct sockaddr_in *) &ro->ro_dst)->sin_addr = inp->inp_faddr; rtalloc(ro); } if ((rt = ro->ro_rt) == (struct rtentry *)0) return (tcp_mssdflt); } ifp = rt->rt_ifp; so = inp->inp_socket; #ifdef RTV_MTU /* if route characteristics exist ... */ /* * While we're here, check if there's an initial rtt * or rttvar. Convert from the route-table units * to scaled multiples of the slow timeout timer. */ if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt)) { /* * XXX the lock bit for RTT indicates that the value * is also a minimum value; this is subject to time. */ if (rt->rt_rmx.rmx_locks & RTV_RTT) tp->t_rttmin = rtt / (RTM_RTTUNIT / PR_SLOWHZ); tp->t_srtt = rtt / (RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTT_SCALE)); if (rt->rt_rmx.rmx_rttvar) tp->t_rttvar = rt->rt_rmx.rmx_rttvar / (RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTTVAR_SCALE)); else /* default variation is +- 1 rtt */ tp->t_rttvar = tp->t_srtt * TCP_RTTVAR_SCALE / TCP_RTT_SCALE; TCPT_RANGESET(tp->t_rxtcur, ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1, tp->t_rttmin, TCPTV_REXMTMAX); } /* * if there's an mtu associated with the route, use it */ if (rt->rt_rmx.rmx_mtu) mss = rt->rt_rmx.rmx_mtu - sizeof(struct tcpiphdr); else #endif /* RTV_MTU */ { mss = ifp->if_mtu - sizeof(struct tcpiphdr); #ifdef IPINIP mss -= sizeof(struct ip); #endif if (!in_localaddr(inp->inp_faddr)) mss = min(mss, tcp_mssdflt); } /* * The current mss, t_maxseg, is initialized to the default value. * If we compute a smaller value, reduce the current mss. * If we compute a larger value, return it for use in sending * a max seg size option, but don't store it for use * unless we received an offer at least that large from peer. * However, do not accept offers under 32 bytes. */ if (offer) mss = min(mss, offer); mss = max(mss, 32); /* sanity */ if (mss < tp->t_maxseg || offer != 0) { /* * See whether we (may) need to save space * for timestamp options. */ maxseg = mss; if (tp->t_flags & TF_USE_SCALE) maxseg -= TCPOLEN_TSTAMP_APPA; maxseg = tcp_mss_round(maxseg); /* * If there's a pipesize, change the socket buffer * to that size. Make the socket buffers an integral * number of mss units; if the mss is larger than * the socket buffer, decrease the mss. */ #ifdef RTV_SPIPE if ((bufsize = rt->rt_rmx.rmx_sendpipe) == 0) #endif bufsize = so->so_snd.sb_hiwat; if (bufsize < maxseg) maxseg = bufsize; else { bufsize = roundup(bufsize, maxseg); if (bufsize > sb_max) bufsize = sb_max; (void)sbreserve(&so->so_snd, bufsize); } tp->t_maxseg = maxseg; #ifdef RTV_RPIPE if ((bufsize = rt->rt_rmx.rmx_recvpipe) == 0) #endif bufsize = so->so_rcv.sb_hiwat; if (bufsize < mss) mss = bufsize; /* * The following is problematical. * If using timestamp options, the peer may send * packets smaller than mss, and may round down * the maxseg to a "nice" value (as we do). */ if (bufsize > mss) { bufsize = roundup(bufsize, mss); if (bufsize > sb_max) bufsize = sb_max; (void)sbreserve(&so->so_rcv, bufsize); } } tp->snd_cwnd = maxseg; #ifdef RTV_SSTHRESH if (rt->rt_rmx.rmx_ssthresh) { /* * There's some sort of gateway or interface * buffer limit on the path. Use this to set * the slow start threshhold, but set the * threshold to no less than 2*maxseg. */ tp->snd_ssthresh = max(2 * maxseg, rt->rt_rmx.rmx_ssthresh); } #endif /* RTV_MTU */ return (mss); } #endif /* TUBA_INCLUDE */