socket.c 20.1 KB
Newer Older
B
bellard 已提交
1 2
/*
 * Copyright (c) 1995 Danny Gasparovski.
3 4
 *
 * Please read the file COPYRIGHT for the
B
bellard 已提交
5 6 7
 * terms and conditions of the copyright.
 */

P
Peter Maydell 已提交
8
#include "qemu/osdep.h"
9
#include "qemu-common.h"
B
bellard 已提交
10 11
#include <slirp.h>
#include "ip_icmp.h"
B
bellard 已提交
12 13 14
#ifdef __sun__
#include <sys/filio.h>
#endif
B
bellard 已提交
15

16 17 18
static void sofcantrcvmore(struct socket *so);
static void sofcantsendmore(struct socket *so);

19 20
struct socket *solookup(struct socket **last, struct socket *head,
        struct sockaddr_storage *lhost, struct sockaddr_storage *fhost)
B
bellard 已提交
21
{
22
    struct socket *so = *last;
23

24
    /* Optimisation */
25 26
    if (so != head && sockaddr_equal(&(so->lhost.ss), lhost)
            && (!fhost || sockaddr_equal(&so->fhost.ss, fhost))) {
27 28
        return so;
    }
29

30
    for (so = head->so_next; so != head; so = so->so_next) {
31 32
        if (sockaddr_equal(&(so->lhost.ss), lhost)
                && (!fhost || sockaddr_equal(&so->fhost.ss, fhost))) {
33 34 35 36
            *last = so;
            return so;
        }
    }
37

38
    return (struct socket *)NULL;
B
bellard 已提交
39 40 41 42 43 44 45 46
}

/*
 * Create a new socket, initialise the fields
 * It is the responsibility of the caller to
 * insque() it into the correct linked-list
 */
struct socket *
47
socreate(Slirp *slirp)
B
bellard 已提交
48 49
{
  struct socket *so;
50

B
bellard 已提交
51 52 53 54 55
  so = (struct socket *)malloc(sizeof(struct socket));
  if(so) {
    memset(so, 0, sizeof(struct socket));
    so->so_state = SS_NOFDREF;
    so->s = -1;
56
    so->slirp = slirp;
57
    so->pollfds_idx = -1;
B
bellard 已提交
58 59 60 61 62 63 64 65
  }
  return(so);
}

/*
 * remque and free a socket, clobber cache
 */
void
66
sofree(struct socket *so)
B
bellard 已提交
67
{
68 69
  Slirp *slirp = so->slirp;

B
bellard 已提交
70 71 72 73
  if (so->so_emu==EMU_RSH && so->extra) {
	sofree(so->extra);
	so->extra=NULL;
  }
74 75 76 77
  if (so == slirp->tcp_last_so) {
      slirp->tcp_last_so = &slirp->tcb;
  } else if (so == slirp->udp_last_so) {
      slirp->udp_last_so = &slirp->udb;
78 79
  } else if (so == slirp->icmp_last_so) {
      slirp->icmp_last_so = &slirp->icmp;
80
  }
B
bellard 已提交
81
  m_free(so->so_m);
82 83

  if(so->so_next && so->so_prev)
B
bellard 已提交
84 85 86 87 88
    remque(so);  /* crashes if so is not in a queue */

  free(so);
}

89
size_t sopreprbuf(struct socket *so, struct iovec *iov, int *np)
B
bellard 已提交
90
{
91
	int n, lss, total;
B
bellard 已提交
92 93 94
	struct sbuf *sb = &so->so_snd;
	int len = sb->sb_datalen - sb->sb_cc;
	int mss = so->so_tcpcb->t_maxseg;
95

96
	DEBUG_CALL("sopreprbuf");
97
	DEBUG_ARG("so = %p", so);
98

99 100 101
	if (len <= 0)
		return 0;

B
bellard 已提交
102
	iov[0].iov_base = sb->sb_wptr;
103 104
        iov[1].iov_base = NULL;
        iov[1].iov_len = 0;
B
bellard 已提交
105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141
	if (sb->sb_wptr < sb->sb_rptr) {
		iov[0].iov_len = sb->sb_rptr - sb->sb_wptr;
		/* Should never succeed, but... */
		if (iov[0].iov_len > len)
		   iov[0].iov_len = len;
		if (iov[0].iov_len > mss)
		   iov[0].iov_len -= iov[0].iov_len%mss;
		n = 1;
	} else {
		iov[0].iov_len = (sb->sb_data + sb->sb_datalen) - sb->sb_wptr;
		/* Should never succeed, but... */
		if (iov[0].iov_len > len) iov[0].iov_len = len;
		len -= iov[0].iov_len;
		if (len) {
			iov[1].iov_base = sb->sb_data;
			iov[1].iov_len = sb->sb_rptr - sb->sb_data;
			if(iov[1].iov_len > len)
			   iov[1].iov_len = len;
			total = iov[0].iov_len + iov[1].iov_len;
			if (total > mss) {
				lss = total%mss;
				if (iov[1].iov_len > lss) {
					iov[1].iov_len -= lss;
					n = 2;
				} else {
					lss -= iov[1].iov_len;
					iov[0].iov_len -= lss;
					n = 1;
				}
			} else
				n = 2;
		} else {
			if (iov[0].iov_len > mss)
			   iov[0].iov_len -= iov[0].iov_len%mss;
			n = 1;
		}
	}
142 143 144 145 146 147 148 149 150 151 152 153
	if (np)
		*np = n;

	return iov[0].iov_len + (n - 1) * iov[1].iov_len;
}

/*
 * Read from so's socket into sb_snd, updating all relevant sbuf fields
 * NOTE: This will only be called if it is select()ed for reading, so
 * a read() of 0 (or less) means it's disconnected
 */
int
154
soread(struct socket *so)
155 156 157 158 159 160
{
	int n, nn;
	struct sbuf *sb = &so->so_snd;
	struct iovec iov[2];

	DEBUG_CALL("soread");
161
	DEBUG_ARG("so = %p", so);
162 163 164 165 166 167

	/*
	 * No need to check if there's enough room to read.
	 * soread wouldn't have been called if there weren't
	 */
	sopreprbuf(so, iov, &n);
168

B
bellard 已提交
169 170 171 172
#ifdef HAVE_READV
	nn = readv(so->s, (struct iovec *)iov, n);
	DEBUG_MISC((dfd, " ... read nn = %d bytes\n", nn));
#else
B
Blue Swirl 已提交
173
	nn = qemu_recv(so->s, iov[0].iov_base, iov[0].iov_len,0);
174
#endif
B
bellard 已提交
175 176 177 178 179 180 181 182 183 184
	if (nn <= 0) {
		if (nn < 0 && (errno == EINTR || errno == EAGAIN))
			return 0;
		else {
			DEBUG_MISC((dfd, " --- soread() disconnected, nn = %d, errno = %d-%s\n", nn, errno,strerror(errno)));
			sofcantrcvmore(so);
			tcp_sockclosed(sototcpcb(so));
			return -1;
		}
	}
185

B
bellard 已提交
186 187 188 189 190 191 192 193 194 195
#ifndef HAVE_READV
	/*
	 * If there was no error, try and read the second time round
	 * We read again if n = 2 (ie, there's another part of the buffer)
	 * and we read as much as we could in the first read
	 * We don't test for <= 0 this time, because there legitimately
	 * might not be any more data (since the socket is non-blocking),
	 * a close will be detected on next iteration.
	 * A return of -1 wont (shouldn't) happen, since it didn't happen above
	 */
B
bellard 已提交
196 197
	if (n == 2 && nn == iov[0].iov_len) {
            int ret;
B
Blue Swirl 已提交
198
            ret = qemu_recv(so->s, iov[1].iov_base, iov[1].iov_len,0);
B
bellard 已提交
199 200 201
            if (ret > 0)
                nn += ret;
        }
202

B
bellard 已提交
203 204
	DEBUG_MISC((dfd, " ... read nn = %d bytes\n", nn));
#endif
205

B
bellard 已提交
206 207 208 209 210 211 212
	/* Update fields */
	sb->sb_cc += nn;
	sb->sb_wptr += nn;
	if (sb->sb_wptr >= (sb->sb_data + sb->sb_datalen))
		sb->sb_wptr -= sb->sb_datalen;
	return nn;
}
213

214 215 216 217 218 219 220
int soreadbuf(struct socket *so, const char *buf, int size)
{
    int n, nn, copy = size;
	struct sbuf *sb = &so->so_snd;
	struct iovec iov[2];

	DEBUG_CALL("soreadbuf");
221
	DEBUG_ARG("so = %p", so);
222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255

	/*
	 * No need to check if there's enough room to read.
	 * soread wouldn't have been called if there weren't
	 */
	if (sopreprbuf(so, iov, &n) < size)
        goto err;

    nn = MIN(iov[0].iov_len, copy);
    memcpy(iov[0].iov_base, buf, nn);

    copy -= nn;
    buf += nn;

    if (copy == 0)
        goto done;

    memcpy(iov[1].iov_base, buf, copy);

done:
    /* Update fields */
	sb->sb_cc += size;
	sb->sb_wptr += size;
	if (sb->sb_wptr >= (sb->sb_data + sb->sb_datalen))
		sb->sb_wptr -= sb->sb_datalen;
    return size;
err:

    sofcantrcvmore(so);
    tcp_sockclosed(sototcpcb(so));
    fprintf(stderr, "soreadbuf buffer to small");
    return -1;
}

B
bellard 已提交
256 257
/*
 * Get urgent data
258
 *
B
bellard 已提交
259 260 261 262 263
 * When the socket is created, we set it SO_OOBINLINE,
 * so when OOB data arrives, we soread() it and everything
 * in the send buffer is sent as urgent data
 */
void
264
sorecvoob(struct socket *so)
B
bellard 已提交
265 266 267 268
{
	struct tcpcb *tp = sototcpcb(so);

	DEBUG_CALL("sorecvoob");
269
	DEBUG_ARG("so = %p", so);
270

B
bellard 已提交
271 272 273 274 275
	/*
	 * We take a guess at how much urgent data has arrived.
	 * In most situations, when urgent data arrives, the next
	 * read() should get all the urgent data.  This guess will
	 * be wrong however if more data arrives just after the
276
	 * urgent data, or the read() doesn't return all the
B
bellard 已提交
277 278 279 280 281 282 283 284 285 286 287 288 289 290
	 * urgent data.
	 */
	soread(so);
	tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
	tp->t_force = 1;
	tcp_output(tp);
	tp->t_force = 0;
}

/*
 * Send urgent data
 * There's a lot duplicated code here, but...
 */
int
291
sosendoob(struct socket *so)
B
bellard 已提交
292 293 294
{
	struct sbuf *sb = &so->so_rcv;
	char buff[2048]; /* XXX Shouldn't be sending more oob data than this */
295

B
bellard 已提交
296
	int n, len;
297

B
bellard 已提交
298
	DEBUG_CALL("sosendoob");
299
	DEBUG_ARG("so = %p", so);
B
bellard 已提交
300
	DEBUG_ARG("sb->sb_cc = %d", sb->sb_cc);
301

B
bellard 已提交
302 303
	if (so->so_urgc > 2048)
	   so->so_urgc = 2048; /* XXXX */
304

B
bellard 已提交
305 306
	if (sb->sb_rptr < sb->sb_wptr) {
		/* We can send it directly */
307
		n = slirp_send(so, sb->sb_rptr, so->so_urgc, (MSG_OOB)); /* |MSG_DONTWAIT)); */
B
bellard 已提交
308
		so->so_urgc -= n;
309

B
bellard 已提交
310 311
		DEBUG_MISC((dfd, " --- sent %d bytes urgent data, %d urgent bytes left\n", n, so->so_urgc));
	} else {
312
		/*
B
bellard 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
		 * Since there's no sendv or sendtov like writev,
		 * we must copy all data to a linear buffer then
		 * send it all
		 */
		len = (sb->sb_data + sb->sb_datalen) - sb->sb_rptr;
		if (len > so->so_urgc) len = so->so_urgc;
		memcpy(buff, sb->sb_rptr, len);
		so->so_urgc -= len;
		if (so->so_urgc) {
			n = sb->sb_wptr - sb->sb_data;
			if (n > so->so_urgc) n = so->so_urgc;
			memcpy((buff + len), sb->sb_data, n);
			so->so_urgc -= n;
			len += n;
		}
328
		n = slirp_send(so, buff, len, (MSG_OOB)); /* |MSG_DONTWAIT)); */
B
bellard 已提交
329 330 331
#ifdef DEBUG
		if (n != len)
		   DEBUG_ERROR((dfd, "Didn't send all data urgently XXXXX\n"));
332
#endif
B
bellard 已提交
333 334
		DEBUG_MISC((dfd, " ---2 sent %d bytes urgent data, %d urgent bytes left\n", n, so->so_urgc));
	}
335

B
bellard 已提交
336 337 338 339
	sb->sb_cc -= n;
	sb->sb_rptr += n;
	if (sb->sb_rptr >= (sb->sb_data + sb->sb_datalen))
		sb->sb_rptr -= sb->sb_datalen;
340

B
bellard 已提交
341 342 343 344
	return n;
}

/*
345
 * Write data from so_rcv to so's socket,
B
bellard 已提交
346 347 348
 * updating all sbuf field as necessary
 */
int
349
sowrite(struct socket *so)
B
bellard 已提交
350 351 352 353 354
{
	int  n,nn;
	struct sbuf *sb = &so->so_rcv;
	int len = sb->sb_cc;
	struct iovec iov[2];
355

B
bellard 已提交
356
	DEBUG_CALL("sowrite");
357
	DEBUG_ARG("so = %p", so);
358

B
bellard 已提交
359 360 361 362 363 364 365 366 367 368
	if (so->so_urgc) {
		sosendoob(so);
		if (sb->sb_cc == 0)
			return 0;
	}

	/*
	 * No need to check if there's something to write,
	 * sowrite wouldn't have been called otherwise
	 */
369

B
bellard 已提交
370
	iov[0].iov_base = sb->sb_rptr;
371 372
        iov[1].iov_base = NULL;
        iov[1].iov_len = 0;
B
bellard 已提交
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
	if (sb->sb_rptr < sb->sb_wptr) {
		iov[0].iov_len = sb->sb_wptr - sb->sb_rptr;
		/* Should never succeed, but... */
		if (iov[0].iov_len > len) iov[0].iov_len = len;
		n = 1;
	} else {
		iov[0].iov_len = (sb->sb_data + sb->sb_datalen) - sb->sb_rptr;
		if (iov[0].iov_len > len) iov[0].iov_len = len;
		len -= iov[0].iov_len;
		if (len) {
			iov[1].iov_base = sb->sb_data;
			iov[1].iov_len = sb->sb_wptr - sb->sb_data;
			if (iov[1].iov_len > len) iov[1].iov_len = len;
			n = 2;
		} else
			n = 1;
	}
	/* Check if there's urgent data to send, and if so, send it */

#ifdef HAVE_READV
	nn = writev(so->s, (const struct iovec *)iov, n);
394

B
bellard 已提交
395 396
	DEBUG_MISC((dfd, "  ... wrote nn = %d bytes\n", nn));
#else
397
	nn = slirp_send(so, iov[0].iov_base, iov[0].iov_len,0);
B
bellard 已提交
398 399 400 401
#endif
	/* This should never happen, but people tell me it does *shrug* */
	if (nn < 0 && (errno == EAGAIN || errno == EINTR))
		return 0;
402

B
bellard 已提交
403 404 405 406 407 408 409
	if (nn <= 0) {
		DEBUG_MISC((dfd, " --- sowrite disconnected, so->so_state = %x, errno = %d\n",
			so->so_state, errno));
		sofcantsendmore(so);
		tcp_sockclosed(sototcpcb(so));
		return -1;
	}
410

B
bellard 已提交
411
#ifndef HAVE_READV
B
bellard 已提交
412 413
	if (n == 2 && nn == iov[0].iov_len) {
            int ret;
414
            ret = slirp_send(so, iov[1].iov_base, iov[1].iov_len,0);
B
bellard 已提交
415 416 417
            if (ret > 0)
                nn += ret;
        }
B
bellard 已提交
418 419
        DEBUG_MISC((dfd, "  ... wrote nn = %d bytes\n", nn));
#endif
420

B
bellard 已提交
421 422 423 424 425
	/* Update sbuf */
	sb->sb_cc -= nn;
	sb->sb_rptr += nn;
	if (sb->sb_rptr >= (sb->sb_data + sb->sb_datalen))
		sb->sb_rptr -= sb->sb_datalen;
426

B
bellard 已提交
427 428 429 430 431 432
	/*
	 * If in DRAIN mode, and there's no more data, set
	 * it CANTSENDMORE
	 */
	if ((so->so_state & SS_FWDRAIN) && sb->sb_cc == 0)
		sofcantsendmore(so);
433

B
bellard 已提交
434 435 436 437 438 439 440
	return nn;
}

/*
 * recvfrom() a UDP socket
 */
void
441
sorecvfrom(struct socket *so)
B
bellard 已提交
442
{
443
	struct sockaddr_storage addr;
444
	struct sockaddr_storage saddr, daddr;
445
	socklen_t addrlen = sizeof(struct sockaddr_storage);
446

B
bellard 已提交
447
	DEBUG_CALL("sorecvfrom");
448
	DEBUG_ARG("so = %p", so);
449

B
bellard 已提交
450 451 452
	if (so->so_type == IPPROTO_ICMP) {   /* This is a "ping" reply */
	  char buff[256];
	  int len;
453

454
	  len = recvfrom(so->s, buff, 256, 0,
B
bellard 已提交
455 456
			 (struct sockaddr *)&addr, &addrlen);
	  /* XXX Check if reply is "correct"? */
457

B
bellard 已提交
458 459 460 461 462
	  if(len == -1 || len == 0) {
	    u_char code=ICMP_UNREACH_PORT;

	    if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
	    else if(errno == ENETUNREACH) code=ICMP_UNREACH_NET;
463

B
bellard 已提交
464 465
	    DEBUG_MISC((dfd," udp icmp rx errno = %d-%s\n",
			errno,strerror(errno)));
Y
Yann Bordenave 已提交
466
	    icmp_send_error(so->so_m, ICMP_UNREACH, code, 0, strerror(errno));
B
bellard 已提交
467 468
	  } else {
	    icmp_reflect(so->so_m);
469
            so->so_m = NULL; /* Don't m_free() it again! */
B
bellard 已提交
470 471 472 473 474
	  }
	  /* No need for this socket anymore, udp_detach it */
	  udp_detach(so);
	} else {                            	/* A "normal" UDP packet */
	  struct mbuf *m;
B
Blue Swirl 已提交
475 476 477 478 479 480
          int len;
#ifdef _WIN32
          unsigned long n;
#else
          int n;
#endif
B
bellard 已提交
481

482 483 484 485
	  m = m_get(so->slirp);
	  if (!m) {
	      return;
	  }
486
	  m->m_data += IF_MAXLINKHDR;
487

488
	  /*
B
bellard 已提交
489 490 491 492 493
	   * XXX Shouldn't FIONREAD packets destined for port 53,
	   * but I don't know the max packet size for DNS lookups
	   */
	  len = M_FREEROOM(m);
	  /* if (so->so_fport != htons(53)) { */
B
bellard 已提交
494
	  ioctlsocket(so->s, FIONREAD, &n);
495

B
bellard 已提交
496 497 498 499 500 501
	  if (n > len) {
	    n = (m->m_data - m->m_dat) + m->m_len + n + 1;
	    m_inc(m, n);
	    len = M_FREEROOM(m);
	  }
	  /* } */
502

B
bellard 已提交
503 504
	  m->m_len = recvfrom(so->s, m->m_data, len, 0,
			      (struct sockaddr *)&addr, &addrlen);
505
	  DEBUG_MISC((dfd, " did recvfrom %d, errno = %d-%s\n",
B
bellard 已提交
506 507
		      m->m_len, errno,strerror(errno)));
	  if(m->m_len<0) {
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
	    /* Report error as ICMP */
	    switch (so->so_lfamily) {
	    uint8_t code;
	    case AF_INET:
	      code = ICMP_UNREACH_PORT;

	      if (errno == EHOSTUNREACH) {
		code = ICMP_UNREACH_HOST;
	      } else if (errno == ENETUNREACH) {
		code = ICMP_UNREACH_NET;
	      }

	      DEBUG_MISC((dfd, " rx error, tx icmp ICMP_UNREACH:%i\n", code));
	      icmp_send_error(so->so_m, ICMP_UNREACH, code, 0, strerror(errno));
	      break;
	    case AF_INET6:
	      code = ICMP6_UNREACH_PORT;

	      if (errno == EHOSTUNREACH) {
		code = ICMP6_UNREACH_ADDRESS;
	      } else if (errno == ENETUNREACH) {
		code = ICMP6_UNREACH_NO_ROUTE;
	      }

	      DEBUG_MISC((dfd, " rx error, tx icmp6 ICMP_UNREACH:%i\n", code));
	      icmp6_send_error(so->so_m, ICMP6_UNREACH, code);
	      break;
	    default:
	      g_assert_not_reached();
	      break;
	    }
B
bellard 已提交
539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
	    m_free(m);
	  } else {
	  /*
	   * Hack: domain name lookup will be used the most for UDP,
	   * and since they'll only be used once there's no need
	   * for the 4 minute (or whatever) timeout... So we time them
	   * out much quicker (10 seconds  for now...)
	   */
	    if (so->so_expire) {
	      if (so->so_fport == htons(53))
		so->so_expire = curtime + SO_EXPIREFAST;
	      else
		so->so_expire = curtime + SO_EXPIRE;
	    }

554
	    /*
B
bellard 已提交
555
	     * If this packet was destined for CTL_ADDR,
556
	     * make it look like that's where it came from
B
bellard 已提交
557
	     */
558 559 560 561
	    saddr = addr;
	    sotranslate_in(so, &saddr);
	    daddr = so->lhost.ss;

562 563
	    switch (so->so_ffamily) {
	    case AF_INET:
564 565 566
	        udp_output(so, m, (struct sockaddr_in *) &saddr,
	                   (struct sockaddr_in *) &daddr,
	                   so->so_iptos);
567
	        break;
568 569 570 571
	    case AF_INET6:
	        udp6_output(so, m, (struct sockaddr_in6 *) &saddr,
	                    (struct sockaddr_in6 *) &daddr);
	        break;
572
	    default:
573
	        g_assert_not_reached();
574 575
	        break;
	    }
B
bellard 已提交
576 577 578 579 580 581 582 583
	  } /* rx error */
	} /* if ping packet */
}

/*
 * sendto() a socket
 */
int
584
sosendto(struct socket *so, struct mbuf *m)
B
bellard 已提交
585 586
{
	int ret;
587
	struct sockaddr_storage addr;
B
bellard 已提交
588 589

	DEBUG_CALL("sosendto");
590 591
	DEBUG_ARG("so = %p", so);
	DEBUG_ARG("m = %p", m);
592

593 594 595
	addr = so->fhost.ss;
	DEBUG_CALL(" sendto()ing)");
	sotranslate_out(so, &addr);
596

B
bellard 已提交
597 598
	/* Don't care what port we get */
	ret = sendto(so->s, m->m_data, m->m_len, 0,
599
		     (struct sockaddr *)&addr, sizeof(addr));
B
bellard 已提交
600 601
	if (ret < 0)
		return -1;
602

B
bellard 已提交
603 604 605 606 607 608
	/*
	 * Kill the socket if there's no reply in 4 minutes,
	 * but only if it's an expirable socket
	 */
	if (so->so_expire)
		so->so_expire = curtime + SO_EXPIRE;
609 610
	so->so_state &= SS_PERSISTENT_MASK;
	so->so_state |= SS_ISFCONNECTED; /* So that it gets select()ed */
B
bellard 已提交
611 612 613 614
	return 0;
}

/*
615
 * Listen for incoming TCP connections
B
bellard 已提交
616 617
 */
struct socket *
618
tcp_listen(Slirp *slirp, uint32_t haddr, u_int hport, uint32_t laddr,
619
           u_int lport, int flags)
B
bellard 已提交
620 621 622
{
	struct sockaddr_in addr;
	struct socket *so;
623 624
	int s, opt = 1;
	socklen_t addrlen = sizeof(addr);
625
	memset(&addr, 0, addrlen);
B
bellard 已提交
626

627
	DEBUG_CALL("tcp_listen");
628 629
	DEBUG_ARG("haddr = %x", haddr);
	DEBUG_ARG("hport = %d", hport);
B
bellard 已提交
630 631 632
	DEBUG_ARG("laddr = %x", laddr);
	DEBUG_ARG("lport = %d", lport);
	DEBUG_ARG("flags = %x", flags);
633

634 635
	so = socreate(slirp);
	if (!so) {
B
bellard 已提交
636 637
	  return NULL;
	}
638

B
bellard 已提交
639 640 641 642 643
	/* Don't tcp_attach... we don't need so_snd nor so_rcv */
	if ((so->so_tcpcb = tcp_newtcpcb(so)) == NULL) {
		free(so);
		return NULL;
	}
644
	insque(so, &slirp->tcb);
645 646

	/*
B
bellard 已提交
647 648 649 650
	 * SS_FACCEPTONCE sockets must time out.
	 */
	if (flags & SS_FACCEPTONCE)
	   so->so_tcpcb->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT*2;
651

652 653
	so->so_state &= SS_PERSISTENT_MASK;
	so->so_state |= (SS_FACCEPTCONN | flags);
654
	so->so_lfamily = AF_INET;
B
bellard 已提交
655 656
	so->so_lport = lport; /* Kept in network format */
	so->so_laddr.s_addr = laddr; /* Ditto */
657

B
bellard 已提交
658
	addr.sin_family = AF_INET;
659 660
	addr.sin_addr.s_addr = haddr;
	addr.sin_port = hport;
661

K
Kevin Wolf 已提交
662
	if (((s = qemu_socket(AF_INET,SOCK_STREAM,0)) < 0) ||
663
	    (socket_set_fast_reuse(s) < 0) ||
B
bellard 已提交
664 665 666
	    (bind(s,(struct sockaddr *)&addr, sizeof(addr)) < 0) ||
	    (listen(s,1) < 0)) {
		int tmperrno = errno; /* Don't clobber the real reason we failed */
667

B
bellard 已提交
668 669 670
		close(s);
		sofree(so);
		/* Restore the real errno */
B
bellard 已提交
671 672 673
#ifdef _WIN32
		WSASetLastError(tmperrno);
#else
B
bellard 已提交
674
		errno = tmperrno;
B
bellard 已提交
675
#endif
B
bellard 已提交
676 677
		return NULL;
	}
678
	qemu_setsockopt(s, SOL_SOCKET, SO_OOBINLINE, &opt, sizeof(int));
679

B
bellard 已提交
680
	getsockname(s,(struct sockaddr *)&addr,&addrlen);
681
	so->so_ffamily = AF_INET;
B
bellard 已提交
682 683
	so->so_fport = addr.sin_port;
	if (addr.sin_addr.s_addr == 0 || addr.sin_addr.s_addr == loopback_addr.s_addr)
684
	   so->so_faddr = slirp->vhost_addr;
B
bellard 已提交
685 686 687 688 689 690 691 692 693 694 695 696 697 698
	else
	   so->so_faddr = addr.sin_addr;

	so->s = s;
	return so;
}

/*
 * Various session state calls
 * XXX Should be #define's
 * The socket state stuff needs work, these often get call 2 or 3
 * times each when only 1 was needed
 */
void
699
soisfconnecting(struct socket *so)
B
bellard 已提交
700 701 702 703 704 705 706
{
	so->so_state &= ~(SS_NOFDREF|SS_ISFCONNECTED|SS_FCANTRCVMORE|
			  SS_FCANTSENDMORE|SS_FWDRAIN);
	so->so_state |= SS_ISFCONNECTING; /* Clobber other states */
}

void
707
soisfconnected(struct socket *so)
B
bellard 已提交
708 709 710 711 712
{
	so->so_state &= ~(SS_ISFCONNECTING|SS_FWDRAIN|SS_NOFDREF);
	so->so_state |= SS_ISFCONNECTED; /* Clobber other states */
}

713 714
static void
sofcantrcvmore(struct socket *so)
B
bellard 已提交
715 716 717 718 719
{
	if ((so->so_state & SS_NOFDREF) == 0) {
		shutdown(so->s,0);
	}
	so->so_state &= ~(SS_ISFCONNECTING);
720 721 722 723
	if (so->so_state & SS_FCANTSENDMORE) {
	   so->so_state &= SS_PERSISTENT_MASK;
	   so->so_state |= SS_NOFDREF; /* Don't select it */
	} else {
B
bellard 已提交
724
	   so->so_state |= SS_FCANTRCVMORE;
725
	}
B
bellard 已提交
726 727
}

728 729
static void
sofcantsendmore(struct socket *so)
B
bellard 已提交
730 731
{
	if ((so->so_state & SS_NOFDREF) == 0) {
B
bellard 已提交
732
            shutdown(so->s,1);           /* send FIN to fhost */
B
bellard 已提交
733 734
	}
	so->so_state &= ~(SS_ISFCONNECTING);
735 736 737 738
	if (so->so_state & SS_FCANTRCVMORE) {
	   so->so_state &= SS_PERSISTENT_MASK;
	   so->so_state |= SS_NOFDREF; /* as above */
	} else {
B
bellard 已提交
739
	   so->so_state |= SS_FCANTSENDMORE;
740
	}
B
bellard 已提交
741 742 743 744 745 746 747
}

/*
 * Set write drain mode
 * Set CANTSENDMORE once all data has been write()n
 */
void
748
sofwdrain(struct socket *so)
B
bellard 已提交
749 750 751 752 753 754
{
	if (so->so_rcv.sb_cc)
		so->so_state |= SS_FWDRAIN;
	else
		sofcantsendmore(so);
}
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832

/*
 * Translate addr in host addr when it is a virtual address
 */
void sotranslate_out(struct socket *so, struct sockaddr_storage *addr)
{
    Slirp *slirp = so->slirp;
    struct sockaddr_in *sin = (struct sockaddr_in *)addr;

    switch (addr->ss_family) {
    case AF_INET:
        if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) ==
                slirp->vnetwork_addr.s_addr) {
            /* It's an alias */
            if (so->so_faddr.s_addr == slirp->vnameserver_addr.s_addr) {
                if (get_dns_addr(&sin->sin_addr) < 0) {
                    sin->sin_addr = loopback_addr;
                }
            } else {
                sin->sin_addr = loopback_addr;
            }
        }

        DEBUG_MISC((dfd, " addr.sin_port=%d, "
            "addr.sin_addr.s_addr=%.16s\n",
            ntohs(sin->sin_port), inet_ntoa(sin->sin_addr)));
        break;

    default:
        break;
    }
}

void sotranslate_in(struct socket *so, struct sockaddr_storage *addr)
{
    Slirp *slirp = so->slirp;
    struct sockaddr_in *sin = (struct sockaddr_in *)addr;

    switch (addr->ss_family) {
    case AF_INET:
        if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) ==
            slirp->vnetwork_addr.s_addr) {
            uint32_t inv_mask = ~slirp->vnetwork_mask.s_addr;

            if ((so->so_faddr.s_addr & inv_mask) == inv_mask) {
                sin->sin_addr = slirp->vhost_addr;
            } else if (sin->sin_addr.s_addr == loopback_addr.s_addr ||
                       so->so_faddr.s_addr != slirp->vhost_addr.s_addr) {
                sin->sin_addr = so->so_faddr;
            }
        }
        break;

    default:
        break;
    }
}

/*
 * Translate connections from localhost to the real hostname
 */
void sotranslate_accept(struct socket *so)
{
    Slirp *slirp = so->slirp;

    switch (so->so_ffamily) {
    case AF_INET:
        if (so->so_faddr.s_addr == INADDR_ANY ||
            (so->so_faddr.s_addr & loopback_mask) ==
            (loopback_addr.s_addr & loopback_mask)) {
           so->so_faddr = slirp->vhost_addr;
        }
        break;

    default:
        break;
    }
}