![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to tcp_subr.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Apple Darwin 0.x] / kernel / bsd / netinet|
Return to tcp_subr.c CVS log | Up to [Apple Darwin 0.x] / kernel / bsd / netinet |
1.1 root 1: /*
2: * Copyright (c) 1999 Apple Computer, Inc. All rights reserved.
3: *
4: * @APPLE_LICENSE_HEADER_START@
5: *
6: * Portions Copyright (c) 1999 Apple Computer, Inc. All Rights
7: * Reserved. This file contains Original Code and/or Modifications of
8: * Original Code as defined in and that are subject to the Apple Public
9: * Source License Version 1.1 (the "License"). You may not use this file
10: * except in compliance with the License. Please obtain a copy of the
11: * License at http://www.apple.com/publicsource and read it before using
12: * this file.
13: *
14: * The Original Code and all software distributed under the License are
15: * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
16: * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
17: * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
18: * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
19: * License for the specific language governing rights and limitations
20: * under the License.
21: *
22: * @APPLE_LICENSE_HEADER_END@
23: */
24:
25: /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
26: /*
27: * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
28: * The Regents of the University of California. All rights reserved.
29: *
30: * Redistribution and use in source and binary forms, with or without
31: * modification, are permitted provided that the following conditions
32: * are met:
33: * 1. Redistributions of source code must retain the above copyright
34: * notice, this list of conditions and the following disclaimer.
35: * 2. Redistributions in binary form must reproduce the above copyright
36: * notice, this list of conditions and the following disclaimer in the
37: * documentation and/or other materials provided with the distribution.
38: * 3. All advertising materials mentioning features or use of this software
39: * must display the following acknowledgement:
40: * This product includes software developed by the University of
41: * California, Berkeley and its contributors.
42: * 4. Neither the name of the University nor the names of its contributors
43: * may be used to endorse or promote products derived from this software
44: * without specific prior written permission.
45: *
46: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
47: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
48: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
49: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
50: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
51: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
52: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
53: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
54: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
55: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
56: * SUCH DAMAGE.
57: *
58: * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95
59: */
60:
61: #include <sys/param.h>
62: #include <sys/proc.h>
63: #include <sys/systm.h>
64: #include <sys/malloc.h>
65: #include <sys/mbuf.h>
66: #include <sys/socket.h>
67: #include <sys/socketvar.h>
68: #include <sys/protosw.h>
69: #include <sys/errno.h>
70:
71: #include <net/route.h>
72: #include <net/if.h>
73:
74: #include <netinet/in.h>
75: #include <netinet/in_systm.h>
76: #include <netinet/ip.h>
77: #include <netinet/in_pcb.h>
78: #include <netinet/ip_var.h>
79: #include <netinet/ip_icmp.h>
80: #include <netinet/tcp.h>
81: #include <netinet/tcp_fsm.h>
82: #include <netinet/tcp_seq.h>
83: #include <netinet/tcp_timer.h>
84: #include <netinet/tcp_var.h>
85: #include <netinet/tcpip.h>
86:
87: #include <kern/kdebug.h>
88:
89: extern struct inpcb *tcp_hash_array[];
90: extern struct inpcb *tcp_lport_hash_array[];
91:
92:
93: #if KDEBUG
94:
95: #define DBG_FNC_TCP_CLOSE NETDBG_CODE(DBG_NETTCP, ((5 << 8) | 2))
96:
97: #endif
98:
99: #ifndef SOCKET_CACHE_ON
100: struct zone *tcpcb_zone;
101: #endif
102:
103: extern struct zone *inpcb_zone;
104:
105: struct inpcb tcb; /* head of queue of active tcpcb's */
106:
107:
108: struct tcpstat tcpstat; /* tcp statistics */
109: u_long tcp_now; /* for RFC 1323 timestamps */
110: tcp_seq tcp_iss; /* tcp initial send seq # */
111:
112: /* patchable/settable parameters for tcp */
113: int tcp_mssdflt = TCP_MSS;
114: int tcp_rttdflt = TCPTV_SRTTDFLT / PR_SLOWHZ;
115: int tcp_do_rfc1323 = 1;
116: int memzone_init_done = 0;
117:
118:
119: extern struct inpcb time_wait_slots[];
120: extern int cur_tw_slot;
121:
122: #if DELACK_BITMASK_ON
123: extern u_long delack_bitmask[];
124: #endif
125:
126: void tcpip_memzone_init()
127: {
128: vm_size_t str_size;
129:
130: if (memzone_init_done == 0) {
131: memzone_init_done = 1;
132:
133: #ifndef SOCKET_CACHE_ON
134: str_size = (vm_size_t) sizeof(struct tcpcb);
135: tcpcb_zone = zinit (str_size, 120000*str_size, 8192, FALSE, "tcpcb zone");
136: #endif
137: str_size = (vm_size_t) sizeof(struct inpcb);
138: inpcb_zone = zinit (str_size, 120000*str_size, 8192, FALSE, "inpcb zone");
139: }
140: }
141:
142: int get_inpcb_str_size()
143: {
144: return sizeof(struct inpcb);
145: }
146:
147:
148: int get_tcp_str_size()
149: {
150: return sizeof(struct tcpcb);
151: }
152:
153:
154: /*
155: * Tcp initialization
156: */
157: void
158: tcp_init()
159: {
160: int i;
161:
162: tcp_iss = random(); /* wrong, but better than a constant */
163: for (i=0; i < N_TCP_HASH_ELEMENTS; i++)
164: tcp_hash_array[i] = 0;
165:
166: #if DELACK_BITMASK_ON
167: for (i=0; i < (N_TCP_HASH_ELEMENTS / 32); i++)
168: delack_bitmask[i] = 0;
169: #endif
170: for (i=0; i < N_TCP_LPORT_HASH_ELEMENTS; i++)
171: tcp_lport_hash_array[i] = 0;
172:
173: for (i=0; i < N_TIME_WAIT_SLOTS; i++) {
174: time_wait_slots[i].inp_next = &time_wait_slots[i];
175: time_wait_slots[i].inp_prev = &time_wait_slots[i];
176: }
177:
178: tcpip_memzone_init();
179: tcb.inp_next = &tcb;
180: tcb.inp_prev = &tcb;
181:
182: if (max_protohdr < sizeof(struct tcpiphdr))
183: max_protohdr = sizeof(struct tcpiphdr);
184: if (max_linkhdr + sizeof(struct tcpiphdr) > MHLEN)
185: panic("tcp_init");
186:
187:
188: }
189:
190: /*
191: * Create template to be used to send tcp packets on a connection.
192: * Call after host entry created, allocates an mbuf and fills
193: * in a skeletal tcp/ip header, minimizing the amount of work
194: * necessary when the connection is used.
195: */
196: struct tcpiphdr *
197: tcp_template(tp)
198: struct tcpcb *tp;
199: {
200: register struct inpcb *inp = tp->t_inpcb;
201: register struct mbuf *m;
202: register struct tcpiphdr *n;
203:
204: if ((n = tp->t_template) == 0) {
205: m = m_get(M_DONTWAIT, MT_HEADER);
206: if (m == NULL)
207: return (0);
208: m->m_len = sizeof (struct tcpiphdr);
209: n = mtod(m, struct tcpiphdr *);
210: }
211: n->ti_next = n->ti_prev = 0;
212: n->ti_x1 = 0;
213: n->ti_pr = IPPROTO_TCP;
214: n->ti_len = htons(sizeof (struct tcpiphdr) - sizeof (struct ip));
215: n->ti_src = inp->inp_laddr;
216: n->ti_dst = inp->inp_faddr;
217: n->ti_sport = inp->inp_lport;
218: n->ti_dport = inp->inp_fport;
219: n->ti_seq = 0;
220: n->ti_ack = 0;
221: n->ti_x2 = 0;
222: n->ti_off = 5;
223: n->ti_flags = 0;
224: n->ti_win = 0;
225: n->ti_sum = 0;
226: n->ti_urp = 0;
227: return (n);
228: }
229:
230:
231: /*
232: * Send a single message to the TCP at address specified by
233: * the given TCP/IP header. If m == 0, then we make a copy
234: * of the tcpiphdr at ti and send directly to the addressed host.
235: * This is used to force keep alive messages out using the TCP
236: * template for a connection tp->t_template. If flags are given
237: * then we send a message back to the TCP which originated the
238: * segment ti, and discard the mbuf containing it and any other
239: * attached mbufs.
240: *
241: * In any case the ack and sequence number of the transmitted
242: * segment are as specified by the parameters.
243: */
244: void
245: tcp_respond(tp, ti, m, ack, seq, flags)
246: struct tcpcb *tp;
247: register struct tcpiphdr *ti;
248: register struct mbuf *m;
249: tcp_seq ack, seq;
250: int flags;
251: {
252: register int tlen;
253: int win = 0;
254: struct route *ro = 0;
255:
256: if (tp) {
257: win = sbspace(&tp->t_inpcb->inp_socket->so_rcv);
258: ro = &tp->t_inpcb->inp_route;
259: }
260: if (m == 0) {
261: m = m_gethdr(M_DONTWAIT, MT_HEADER);
262: if (m == NULL)
263: return;
264: #ifdef TCP_COMPAT_42
265: tlen = 1;
266: #else
267: tlen = 0;
268: #endif
269: m->m_data += max_linkhdr;
270: *mtod(m, struct tcpiphdr *) = *ti;
271: ti = mtod(m, struct tcpiphdr *);
272: flags = TH_ACK;
273: } else {
274: m_freem(m->m_next);
275: m->m_next = 0;
276: m->m_data = (caddr_t)ti;
277: m->m_len = sizeof (struct tcpiphdr);
278: tlen = 0;
279: #define xchg(a,b,type) { type t; t=a; a=b; b=t; }
280: xchg(ti->ti_dst.s_addr, ti->ti_src.s_addr, u_long);
281: xchg(ti->ti_dport, ti->ti_sport, u_short);
282: #undef xchg
283: }
284: ti->ti_len = htons((u_short)(sizeof (struct tcphdr) + tlen));
285: tlen += sizeof (struct tcpiphdr);
286: m->m_len = tlen;
287: m->m_pkthdr.len = tlen;
288: m->m_pkthdr.rcvif = (struct ifnet *) 0;
289: ti->ti_next = ti->ti_prev = 0;
290: ti->ti_x1 = 0;
291: ti->ti_seq = htonl(seq);
292: ti->ti_ack = htonl(ack);
293: ti->ti_x2 = 0;
294: ti->ti_off = sizeof (struct tcphdr) >> 2;
295: ti->ti_flags = flags;
296: if (tp)
297: ti->ti_win = htons((u_short) (win >> tp->rcv_scale));
298: else
299: ti->ti_win = htons((u_short)win);
300: ti->ti_urp = 0;
301: ti->ti_sum = 0;
302: ti->ti_sum = in_cksum(m, tlen);
303: ((struct ip *)ti)->ip_len = tlen;
304: ((struct ip *)ti)->ip_ttl = ip_defttl;
305: (void) ip_output(m, NULL, ro, 0, NULL);
306: }
307:
308: /*
309: * Create a new TCP control block, making an
310: * empty reassembly queue and hooking it to the argument
311: * protocol control block.
312: */
313: struct tcpcb *
314: tcp_newtcpcb(inp)
315: struct inpcb *inp;
316: {
317: register struct tcpcb *tp;
318: register struct socket *so = inp->inp_socket;
319:
320: #ifndef SOCKET_CACHE_ON
321: if (so->cached_in_sock_layer == 0) {
322: tp = (struct tcpcb *) zalloc_noblock(tcpcb_zone);
323: if (tp == NULL)
324: return ((struct tcpcb *)0);
325: }
326: else
327: #endif
328: tp = inp->inp_saved_ppcb;
329:
330: bzero((char *) tp, sizeof(struct tcpcb));
331: tp->seg_next = tp->seg_prev = (struct tcpiphdr *)tp;
332: tp->t_maxseg = tcp_mssdflt;
333:
334: tp->t_flags = tcp_do_rfc1323 ? (TF_REQ_SCALE|TF_REQ_TSTMP) : 0;
335: tp->t_inpcb = inp;
336: /*
337: * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no
338: * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives
339: * reasonable initial retransmit time.
340: */
341: tp->t_srtt = TCPTV_SRTTBASE;
342: tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << 2;
343: tp->t_rttmin = TCPTV_MIN;
344: TCPT_RANGESET(tp->t_rxtcur,
345: ((TCPTV_SRTTBASE >> 2) + (TCPTV_SRTTDFLT << 2)) >> 1,
346: TCPTV_MIN, TCPTV_REXMTMAX);
347: tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT;
348: tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT;
349: inp->inp_ip.ip_ttl = ip_defttl;
350: inp->inp_ppcb = (caddr_t)tp;
351: return (tp);
352: }
353:
354: /*
355: * Drop a TCP connection, reporting
356: * the specified error. If connection is synchronized,
357: * then send a RST to peer.
358: */
359: struct tcpcb *
360: tcp_drop(tp, errno)
361: register struct tcpcb *tp;
362: int errno;
363: {
364: struct socket *so = tp->t_inpcb->inp_socket;
365:
366: if (TCPS_HAVERCVDSYN(tp->t_state)) {
367: tp->t_state = TCPS_CLOSED;
368: (void) tcp_output(tp);
369: tcpstat.tcps_drops++;
370: } else
371: tcpstat.tcps_conndrops++;
372: if (errno == ETIMEDOUT && tp->t_softerror)
373: errno = tp->t_softerror;
374: so->so_error = errno;
375: return (tcp_close(tp));
376: }
377:
378: /*
379: * Close a TCP control block:
380: * discard all space held by the tcp
381: * discard internet protocol block
382: * wake up any sleepers
383: */
384: struct tcpcb *
385: tcp_close(tp)
386: register struct tcpcb *tp;
387: {
388: register struct tcpiphdr *t;
389: struct inpcb *inp = tp->t_inpcb;
390: struct socket *so = inp->inp_socket;
391: register struct mbuf *m;
392: #ifdef RTV_RTT
393: register struct rtentry *rt;
394:
395: KERNEL_DEBUG(DBG_FNC_TCP_CLOSE | DBG_FUNC_START, tp,0,0,0,0);
396: /*
397: * If we sent enough data to get some meaningful characteristics,
398: * save them in the routing entry. 'Enough' is arbitrarily
399: * defined as the sendpipesize (default 4K) * 16. This would
400: * give us 16 rtt samples assuming we only get one sample per
401: * window (the usual case on a long haul net). 16 samples is
402: * enough for the srtt filter to converge to within 5% of the correct
403: * value; fewer samples and we could save a very bogus rtt.
404: *
405: * Don't update the default route's characteristics and don't
406: * update anything that the user "locked".
407: */
408: if (SEQ_LT(tp->iss + so->so_snd.sb_hiwat * 16, tp->snd_max) &&
409: (rt = inp->inp_route.ro_rt) &&
410: ((struct sockaddr_in *)rt_key(rt))->sin_addr.s_addr != INADDR_ANY) {
411: register u_long i;
412:
413: if ((rt->rt_rmx.rmx_locks & RTV_RTT) == 0) {
414: i = tp->t_srtt *
415: (RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTT_SCALE));
416: if (rt->rt_rmx.rmx_rtt && i)
417: /*
418: * filter this update to half the old & half
419: * the new values, converting scale.
420: * See route.h and tcp_var.h for a
421: * description of the scaling constants.
422: */
423: rt->rt_rmx.rmx_rtt =
424: (rt->rt_rmx.rmx_rtt + i) / 2;
425: else
426: rt->rt_rmx.rmx_rtt = i;
427: }
428: if ((rt->rt_rmx.rmx_locks & RTV_RTTVAR) == 0) {
429: i = tp->t_rttvar *
430: (RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTTVAR_SCALE));
431: if (rt->rt_rmx.rmx_rttvar && i)
432: rt->rt_rmx.rmx_rttvar =
433: (rt->rt_rmx.rmx_rttvar + i) / 2;
434: else
435: rt->rt_rmx.rmx_rttvar = i;
436: }
437: /*
438: * update the pipelimit (ssthresh) if it has been updated
439: * already or if a pipesize was specified & the threshhold
440: * got below half the pipesize. I.e., wait for bad news
441: * before we start updating, then update on both good
442: * and bad news.
443: */
444: if ((rt->rt_rmx.rmx_locks & RTV_SSTHRESH) == 0 &&
445: (i = tp->snd_ssthresh) && rt->rt_rmx.rmx_ssthresh ||
446: i < (rt->rt_rmx.rmx_sendpipe / 2)) {
447: /*
448: * convert the limit from user data bytes to
449: * packets then to packet data bytes.
450: */
451: i = (i + tp->t_maxseg / 2) / tp->t_maxseg;
452: if (i < 2)
453: i = 2;
454: i *= (u_long)(tp->t_maxseg + sizeof (struct tcpiphdr));
455: if (rt->rt_rmx.rmx_ssthresh)
456: rt->rt_rmx.rmx_ssthresh =
457: (rt->rt_rmx.rmx_ssthresh + i) / 2;
458: else
459: rt->rt_rmx.rmx_ssthresh = i;
460: }
461: }
462: #endif /* RTV_RTT */
463: /* free the reassembly queue, if any */
464: t = tp->seg_next;
465: while (t != (struct tcpiphdr *)tp) {
466: t = (struct tcpiphdr *)t->ti_next;
467: m = REASS_MBUF((struct tcpiphdr *)t->ti_prev);
468: remque((queue_t) t->ti_prev);
469: m_freem(m);
470: }
471: if (tp->t_template)
472: (void) m_free(dtom(tp->t_template));
473:
474: #ifdef SOCKET_CACHE_ON
475: inp->inp_saved_ppcb = tp;
476: #else
477: if (so->cached_in_sock_layer)
478: inp->inp_saved_ppcb = tp;
479: else
480: zfree(tcpcb_zone, (vm_offset_t) tp);
481: #endif
482:
483: inp->inp_ppcb = 0;
484:
485: soisdisconnected(so);
486:
487: in_pcbdetach(inp);
488: tcpstat.tcps_closed++;
489:
490: KERNEL_DEBUG(DBG_FNC_TCP_CLOSE | DBG_FUNC_END, tcpstat.tcps_closed,0,0,0,0);
491:
492: return ((struct tcpcb *)0);
493: }
494:
495: void
496: tcp_drain()
497: {
498:
499: }
500:
501: /*
502: * Notify a tcp user of an asynchronous error;
503: * store error as soft error, but wake up user
504: * (for now, won't do anything until can select for soft error).
505: */
506: void
507: tcp_notify(inp, error)
508: struct inpcb *inp;
509: int error;
510: {
511: register struct tcpcb *tp = (struct tcpcb *)inp->inp_ppcb;
512: register struct socket *so = inp->inp_socket;
513:
514: /*
515: * Ignore some errors if we are hooked up.
516: * If connection hasn't completed, has retransmitted several times,
517: * and receives a second error, give up now. This is better
518: * than waiting a long time to establish a connection that
519: * can never complete.
520: */
521: if (tp->t_state == TCPS_ESTABLISHED &&
522: (error == EHOSTUNREACH || error == ENETUNREACH ||
523: error == EHOSTDOWN)) {
524: return;
525: } else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 &&
526: tp->t_softerror)
527: so->so_error = error;
528: else
529: tp->t_softerror = error;
530: wakeup((caddr_t) &so->so_timeo);
531: sorwakeup(so);
532: sowwakeup(so);
533: }
534:
535: void
536: tcp_ctlinput(cmd, sa, ip)
537: int cmd;
538: struct sockaddr *sa;
539: register struct ip *ip;
540: {
541: register struct tcphdr *th;
542: extern struct in_addr zeroin_addr;
543: extern u_char inetctlerrmap[];
544: void (*notify) __P((struct inpcb *, int)) = tcp_notify;
545:
546: if (cmd == PRC_QUENCH)
547: notify = tcp_quench;
548: else if (!PRC_IS_REDIRECT(cmd) &&
549: ((unsigned)cmd > PRC_NCMDS || inetctlerrmap[cmd] == 0))
550: return;
551: if (ip) {
552: th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2));
553: in_pcbnotify(&tcb, sa, th->th_dport, ip->ip_src, th->th_sport,
554: cmd, notify);
555: } else
556: in_pcbnotify(&tcb, sa, 0, zeroin_addr, 0, cmd, notify);
557: }
558:
559: /*
560: * When a source quench is received, close congestion window
561: * to one segment. We will gradually open it again as we proceed.
562: */
563: void
564: tcp_quench(inp, errno)
565: struct inpcb *inp;
566: int errno;
567: {
568: struct tcpcb *tp = intotcpcb(inp);
569:
570: if (tp)
571: tp->snd_cwnd = tp->t_maxseg;
572: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.