io.c 28.9 KB
Newer Older
1 2 3
/*
 *	BIRD Internet Routing Daemon -- Unix I/O
 *
4
 *	(c) 1998--2004 Martin Mares <mj@ucw.cz>
5
 *      (c) 2004       Ondrej Filip <feela@network.cz>
6 7 8 9 10 11
 *
 *	Can be freely distributed and used under the terms of the GNU GPL.
 */

#include <stdio.h>
#include <stdlib.h>
12
#include <time.h>
13 14 15 16
#include <sys/time.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/fcntl.h>
17
#include <sys/un.h>
18 19 20 21 22 23 24 25
#include <unistd.h>
#include <errno.h>

#include "nest/bird.h"
#include "lib/lists.h"
#include "lib/resource.h"
#include "lib/timer.h"
#include "lib/socket.h"
26
#include "lib/event.h"
27
#include "lib/string.h"
28 29 30
#include "nest/iface.h"

#include "lib/unix.h"
31
#include "lib/sysio.h"
32

33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
/*
 *	Tracked Files
 */

struct rfile {
  resource r;
  FILE *f;
};

static void
rf_free(resource *r)
{
  struct rfile *a = (struct rfile *) r;

  fclose(a->f);
}

static void
rf_dump(resource *r)
{
  struct rfile *a = (struct rfile *) r;

  debug("(FILE *%p)\n", a->f);
}

static struct resclass rf_class = {
  "FILE",
  sizeof(struct rfile),
  rf_free,
  rf_dump
};

void *
66
tracked_fopen(pool *p, char *name, char *mode)
67 68 69 70 71 72 73 74 75 76 77
{
  FILE *f = fopen(name, mode);

  if (f)
    {
      struct rfile *r = ralloc(p, &rf_class);
      r->f = f;
    }
  return f;
}

78 79 80 81 82
/**
 * DOC: Timers
 *
 * Timers are resources which represent a wish of a module to call
 * a function at the specified time. The platform dependent code
Martin Mareš's avatar
Martin Mareš committed
83
 * doesn't guarantee exact timing, only that a timer function
84 85
 * won't be called before the requested time.
 *
86 87 88 89 90
 * In BIRD, time is represented by values of the &bird_clock_t type
 * which are integral numbers interpreted as a relative number of seconds since
 * some fixed time point in past. The current time can be read
 * from variable @now with reasonable accuracy and is monotonic. There is also
 * a current 'absolute' time in variable @now_real reported by OS.
91 92 93 94 95
 *
 * Each timer is described by a &timer structure containing a pointer
 * to the handler function (@hook), data private to this function (@data),
 * time the function should be called at (@expires, 0 for inactive timers),
 * for the other fields see |timer.h|.
96 97 98 99 100 101 102
 */

#define NEAR_TIMER_LIMIT 4

static list near_timers, far_timers;
static bird_clock_t first_far_timer = TIME_INFINITY;

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
bird_clock_t now, now_real;

static void
update_times_plain(void)
{
  bird_clock_t new_time = time(NULL);
  int delta = new_time - now_real;

  if ((delta >= 0) && (delta < 60))
    now += delta;
  else if (now_real != 0)
   log(L_WARN "Time jump, delta %d s", delta);

  now_real = new_time;
}

static void
update_times_gettime(void)
{
  struct timespec ts;
  int rv;

  rv = clock_gettime(CLOCK_MONOTONIC, &ts);
  if (rv != 0)
    die("clock_gettime: %m");

  if (ts.tv_sec != now) {
    if (ts.tv_sec < now)
      log(L_ERR "Monotonic timer is broken");

    now = ts.tv_sec;
    now_real = time(NULL);
  }
}

static int clock_monotonic_available;

static inline void
update_times(void)
{
  if (clock_monotonic_available)
    update_times_gettime();
  else
    update_times_plain();
}

static inline void
init_times(void)
{
 struct timespec ts;
 clock_monotonic_available = (clock_gettime(CLOCK_MONOTONIC, &ts) == 0);
 if (!clock_monotonic_available)
   log(L_WARN "Monotonic timer is missing");
}

158 159 160 161 162 163 164 165 166 167 168 169 170 171

static void
tm_free(resource *r)
{
  timer *t = (timer *) r;

  tm_stop(t);
}

static void
tm_dump(resource *r)
{
  timer *t = (timer *) r;

172
  debug("(code %p, data %p, ", t->hook, t->data);
173 174 175 176
  if (t->randomize)
    debug("rand %d, ", t->randomize);
  if (t->recurrent)
    debug("recur %d, ", t->recurrent);
177 178 179 180 181 182 183 184 185 186 187 188 189
  if (t->expires)
    debug("expires in %d sec)\n", t->expires - now);
  else
    debug("inactive)\n");
}

static struct resclass tm_class = {
  "Timer",
  sizeof(timer),
  tm_free,
  tm_dump
};

190 191 192 193 194 195 196 197
/**
 * tm_new - create a timer
 * @p: pool
 *
 * This function creates a new timer resource and returns
 * a pointer to it. To use the timer, you need to fill in
 * the structure fields and call tm_start() to start timing.
 */
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218
timer *
tm_new(pool *p)
{
  timer *t = ralloc(p, &tm_class);
  t->hook = NULL;
  t->data = NULL;
  t->randomize = 0;
  t->expires = 0;
  return t;
}

static inline void
tm_insert_near(timer *t)
{
  node *n = HEAD(near_timers);

  while (n->next && (SKIP_BACK(timer, n, n)->expires < t->expires))
    n = n->next;
  insert_node(&t->n, n->prev);
}

219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
/**
 * tm_start - start a timer
 * @t: timer
 * @after: number of seconds the timer should be run after
 *
 * This function schedules the hook function of the timer to
 * be called after @after seconds. If the timer has been already
 * started, it's @expire time is replaced by the new value.
 *
 * You can have set the @randomize field of @t, the timeout
 * will be increased by a random number of seconds chosen
 * uniformly from range 0 .. @randomize.
 *
 * You can call tm_start() from the handler function of the timer
 * to request another run of the timer. Also, you can set the @recurrent
 * field to have the timer re-added automatically with the same timeout.
 */
236 237 238 239 240 241
void
tm_start(timer *t, unsigned after)
{
  bird_clock_t when;

  if (t->randomize)
242
    after += random() % (t->randomize + 1);
243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258
  when = now + after;
  if (t->expires == when)
    return;
  if (t->expires)
    rem_node(&t->n);
  t->expires = when;
  if (after <= NEAR_TIMER_LIMIT)
    tm_insert_near(t);
  else
    {
      if (!first_far_timer || first_far_timer > when)
	first_far_timer = when;
      add_tail(&far_timers, &t->n);
    }
}

259 260 261 262 263 264 265
/**
 * tm_stop - stop a timer
 * @t: timer
 *
 * This function stops a timer. If the timer is already stopped,
 * nothing happens.
 */
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320
void
tm_stop(timer *t)
{
  if (t->expires)
    {
      rem_node(&t->n);
      t->expires = 0;
    }
}

static void
tm_dump_them(char *name, list *l)
{
  node *n;
  timer *t;

  debug("%s timers:\n", name);
  WALK_LIST(n, *l)
    {
      t = SKIP_BACK(timer, n, n);
      debug("%p ", t);
      tm_dump(&t->r);
    }
  debug("\n");
}

void
tm_dump_all(void)
{
  tm_dump_them("Near", &near_timers);
  tm_dump_them("Far", &far_timers);
}

static inline time_t
tm_first_shot(void)
{
  time_t x = first_far_timer;

  if (!EMPTY_LIST(near_timers))
    {
      timer *t = SKIP_BACK(timer, n, HEAD(near_timers));
      if (t->expires < x)
	x = t->expires;
    }
  return x;
}

static void
tm_shot(void)
{
  timer *t;
  node *n, *m;

  if (first_far_timer <= now)
    {
321
      bird_clock_t limit = now + NEAR_TIMER_LIMIT;
322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338
      first_far_timer = TIME_INFINITY;
      n = HEAD(far_timers);
      while (m = n->next)
	{
	  t = SKIP_BACK(timer, n, n);
	  if (t->expires <= limit)
	    {
	      rem_node(n);
	      tm_insert_near(t);
	    }
	  else if (t->expires < first_far_timer)
	    first_far_timer = t->expires;
	  n = m;
	}
    }
  while ((n = HEAD(near_timers)) -> next)
    {
339
      int delay;
340 341 342 343
      t = SKIP_BACK(timer, n, n);
      if (t->expires > now)
	break;
      rem_node(n);
344
      delay = t->expires - now;
345
      t->expires = 0;
346 347 348 349 350 351 352
      if (t->recurrent)
	{
	  int i = t->recurrent - delay;
	  if (i < 0)
	    i = 0;
	  tm_start(t, i);
	}
353 354 355 356
      t->hook(t);
    }
}

357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
/**
 * tm_parse_datetime - parse a date and time
 * @x: datetime string
 *
 * tm_parse_datetime() takes a textual representation of
 * a date and time (dd-mm-yyyy hh:mm:ss)
 * and converts it to the corresponding value of type &bird_clock_t.
 */
bird_clock_t
tm_parse_datetime(char *x)
{
  struct tm tm;
  int n;
  time_t t;

  if (sscanf(x, "%d-%d-%d %d:%d:%d%n", &tm.tm_mday, &tm.tm_mon, &tm.tm_year, &tm.tm_hour, &tm.tm_min, &tm.tm_sec, &n) != 6 || x[n])
    return tm_parse_date(x);
  tm.tm_mon--;
  tm.tm_year -= 1900;
  t = mktime(&tm);
  if (t == (time_t) -1)
    return 0;
  return t;
}
381 382 383 384 385 386 387
/**
 * tm_parse_date - parse a date
 * @x: date string
 *
 * tm_parse_date() takes a textual representation of a date (dd-mm-yyyy)
 * and converts it to the corresponding value of type &bird_clock_t.
 */
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
bird_clock_t
tm_parse_date(char *x)
{
  struct tm tm;
  int n;
  time_t t;

  if (sscanf(x, "%d-%d-%d%n", &tm.tm_mday, &tm.tm_mon, &tm.tm_year, &n) != 3 || x[n])
    return 0;
  tm.tm_mon--;
  tm.tm_year -= 1900;
  tm.tm_hour = tm.tm_min = tm.tm_sec = 0;
  t = mktime(&tm);
  if (t == (time_t) -1)
    return 0;
  return t;
}

406 407 408 409 410
/**
 * tm_format_date - convert date to textual representation
 * @x: destination buffer of size %TM_DATE_BUFFER_SIZE
 * @t: time
 *
411 412
 * This function formats the given relative time value @t to a textual
 * date representation (dd-mm-yyyy) in real time..
413
 */
414 415 416 417 418 419
void
tm_format_date(char *x, bird_clock_t t)
{
  struct tm *tm;

  tm = localtime(&t);
420
  bsprintf(x, "%02d-%02d-%04d", tm->tm_mday, tm->tm_mon+1, tm->tm_year+1900);
421 422
}

423 424 425 426 427
/**
 * tm_format_datetime - convert date and time to textual representation
 * @x: destination buffer of size %TM_DATETIME_BUFFER_SIZE
 * @t: time
 *
428 429
 * This function formats the given relative time value @t to a textual
 * date/time representation (dd-mm-yyyy hh:mm:ss) in real time.
430
 */
431 432 433 434
void
tm_format_datetime(char *x, bird_clock_t t)
{
  struct tm *tm;
435 436
  bird_clock_t delta = now - t;
  t = now_real - delta;
437 438 439 440 441
  tm = localtime(&t);
  if (strftime(x, TM_DATETIME_BUFFER_SIZE, "%d-%m-%Y %H:%M:%S", tm) == TM_DATETIME_BUFFER_SIZE)
    strcpy(x, "<too-long>");
}

442 443 444 445 446
/**
 * tm_format_reltime - convert date and time to relative textual representation
 * @x: destination buffer of size %TM_RELTIME_BUFFER_SIZE
 * @t: time
 *
447 448
 * This function formats the given relative time value @t to a short
 * textual representation in real time, relative to the current time.
449
 */
450 451 452 453 454 455
void
tm_format_reltime(char *x, bird_clock_t t)
{
  struct tm *tm;
  static char *month_names[12] = { "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };

456 457
  bird_clock_t delta = now - t;
  t = now_real - delta;
458
  tm = localtime(&t);
459
  if (delta < 20*3600)
460 461 462 463 464 465 466
    bsprintf(x, "%02d:%02d", tm->tm_hour, tm->tm_min);
  else if (delta < 360*86400)
    bsprintf(x, "%s%02d", month_names[tm->tm_mon], tm->tm_mday);
  else
    bsprintf(x, "%d", tm->tm_year+1900);
}

467 468 469 470 471 472 473 474 475 476
/**
 * DOC: Sockets
 *
 * Socket resources represent network connections. Their data structure (&socket)
 * contains a lot of fields defining the exact type of the socket, the local and
 * remote addresses and ports, pointers to socket buffers and finally pointers to
 * hook functions to be called when new data have arrived to the receive buffer
 * (@rx_hook), when the contents of the transmit buffer have been transmitted
 * (@tx_hook) and when an error or connection close occurs (@err_hook).
 *
477
 * Freeing of sockets from inside socket hooks is perfectly safe.
478 479
 */

480 481 482 483
#ifndef SOL_IP
#define SOL_IP IPPROTO_IP
#endif

484 485 486 487 488 489 490 491
#ifndef SOL_IPV6
#define SOL_IPV6 IPPROTO_IPV6
#endif

#ifndef IPV6_ADD_MEMBERSHIP
#define IPV6_ADD_MEMBERSHIP IP_ADD_MEMBERSHIP
#endif

492
static list sock_list;
493 494 495 496 497 498 499 500 501 502 503
static struct birdsock *current_sock;
static int sock_recalc_fdsets_p;

static inline sock *
sk_next(sock *s)
{
  if (!s->n.next->next)
    return NULL;
  else
    return SKIP_BACK(sock, n, s->n.next);
}
504 505

static void
506
sk_alloc_bufs(sock *s)
507
{
508 509 510 511 512 513 514
  if (!s->rbuf && s->rbsize)
    s->rbuf = s->rbuf_alloc = xmalloc(s->rbsize);
  s->rpos = s->rbuf;
  if (!s->tbuf && s->tbsize)
    s->tbuf = s->tbuf_alloc = xmalloc(s->tbsize);
  s->tpos = s->ttx = s->tbuf;
}
515

516 517 518
static void
sk_free_bufs(sock *s)
{
519
  if (s->rbuf_alloc)
520 521 522 523
    {
      xfree(s->rbuf_alloc);
      s->rbuf = s->rbuf_alloc = NULL;
    }
524
  if (s->tbuf_alloc)
525 526 527 528 529 530 531 532 533 534 535 536
    {
      xfree(s->tbuf_alloc);
      s->tbuf = s->tbuf_alloc = NULL;
    }
}

static void
sk_free(resource *r)
{
  sock *s = (sock *) r;

  sk_free_bufs(s);
537
  if (s->fd >= 0)
538 539
    {
      close(s->fd);
540 541
      if (s == current_sock)
	current_sock = sk_next(s);
542
      rem_node(&s->n);
543
      sock_recalc_fdsets_p = 1;
544
    }
545 546
}

547 548 549 550 551 552 553
void
sk_reallocate(sock *s)
{
  sk_free_bufs(s);
  sk_alloc_bufs(s);
}

554 555 556 557
static void
sk_dump(resource *r)
{
  sock *s = (sock *) r;
558
  static char *sk_type_names[] = { "TCP<", "TCP>", "TCP", "UDP", "UDP/MC", "IP", "IP/MC", "MAGIC", "UNIX<", "UNIX", "DEL!" };
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578

  debug("(%s, ud=%p, sa=%08x, sp=%d, da=%08x, dp=%d, tos=%d, ttl=%d, if=%s)\n",
	sk_type_names[s->type],
	s->data,
	s->saddr,
	s->sport,
	s->daddr,
	s->dport,
	s->tos,
	s->ttl,
	s->iface ? s->iface->name : "none");
}

static struct resclass sk_class = {
  "Socket",
  sizeof(sock),
  sk_free,
  sk_dump
};

579 580 581 582 583 584 585 586
/**
 * sk_new - create a socket
 * @p: pool
 *
 * This function creates a new socket resource. If you want to use it,
 * you need to fill in all the required fields of the structure and
 * call sk_open() to do the actual opening of the socket.
 */
587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
sock *
sk_new(pool *p)
{
  sock *s = ralloc(p, &sk_class);
  s->pool = p;
  s->data = NULL;
  s->saddr = s->daddr = IPA_NONE;
  s->sport = s->dport = 0;
  s->tos = s->ttl = -1;
  s->iface = NULL;
  s->rbuf = NULL;
  s->rx_hook = NULL;
  s->rbsize = 0;
  s->tbuf = NULL;
  s->tx_hook = NULL;
  s->tbsize = 0;
  s->err_hook = NULL;
  s->fd = -1;
605
  s->rbuf_alloc = s->tbuf_alloc = NULL;
606
  s->password = NULL;
607 608 609
  return s;
}

610 611 612 613 614 615
static void
sk_insert(sock *s)
{
  add_tail(&sock_list, &s->n);
  sock_recalc_fdsets_p = 1;
}
616

617 618 619 620 621
#ifdef IPV6

void
fill_in_sockaddr(sockaddr *sa, ip_addr a, unsigned port)
{
622
  memset (sa, 0, sizeof (struct sockaddr_in6));
623 624 625
  sa->sin6_family = AF_INET6;
  sa->sin6_port = htons(port);
  sa->sin6_flowinfo = 0;
626 627 628
#ifdef HAVE_SIN_LEN
  sa->sin6_len = sizeof(struct sockaddr_in6);
#endif
629 630 631 632
  set_inaddr(&sa->sin6_addr, a);
}

void
633
get_sockaddr(struct sockaddr_in6 *sa, ip_addr *a, unsigned *port, int check)
634
{
635 636
  if (check && sa->sin6_family != AF_INET6)
    bug("get_sockaddr called for wrong address family (%d)", sa->sin6_family);
637 638 639 640 641 642 643 644
  if (port)
    *port = ntohs(sa->sin6_port);
  memcpy(a, &sa->sin6_addr, sizeof(*a));
  ipa_ntoh(*a);
}

#else

645
void
646
fill_in_sockaddr(sockaddr *sa, ip_addr a, unsigned port)
647
{
648
  memset (sa, 0, sizeof (struct sockaddr_in));
649 650
  sa->sin_family = AF_INET;
  sa->sin_port = htons(port);
651 652 653
#ifdef HAVE_SIN_LEN
  sa->sin_len = sizeof(struct sockaddr_in);
#endif
654 655 656
  set_inaddr(&sa->sin_addr, a);
}

657
void
658
get_sockaddr(struct sockaddr_in *sa, ip_addr *a, unsigned *port, int check)
659
{
660 661
  if (check && sa->sin_family != AF_INET)
    bug("get_sockaddr called for wrong address family (%d)", sa->sin_family);
662 663
  if (port)
    *port = ntohs(sa->sin_port);
664
  memcpy(a, &sa->sin_addr.s_addr, sizeof(*a));
665
  ipa_ntoh(*a);
666 667
}

668 669
#endif

670 671 672 673 674 675
static char *
sk_set_ttl_int(sock *s)
{
  int one = 1;
#ifdef IPV6
  if (s->type != SK_UDP_MC && s->type != SK_IP_MC &&
676
      setsockopt(s->fd, SOL_IPV6, IPV6_UNICAST_HOPS, &s->ttl, sizeof(s->ttl)) < 0)
677 678 679 680 681 682 683 684 685 686 687 688
    return "IPV6_UNICAST_HOPS";
#else
  if (setsockopt(s->fd, SOL_IP, IP_TTL, &s->ttl, sizeof(s->ttl)) < 0)
    return "IP_TTL";
#ifdef CONFIG_UNIX_DONTROUTE
  if (s->ttl == 1 && setsockopt(s->fd, SOL_SOCKET, SO_DONTROUTE, &one, sizeof(one)) < 0)
    return "SO_DONTROUTE";
#endif 
#endif
  return NULL;
}

689 690 691
#define ERR(x) do { err = x; goto bad; } while(0)
#define WARN(x) log(L_WARN "sk_setup: %s: %m", x)

692 693 694 695 696 697 698 699
static char *
sk_setup(sock *s)
{
  int fd = s->fd;
  char *err;

  if (fcntl(fd, F_SETFL, O_NONBLOCK) < 0)
    ERR("fcntl(O_NONBLOCK)");
700 701
  if (s->type == SK_UNIX)
    return NULL;
702
#ifndef IPV6
703
  if ((s->tos >= 0) && setsockopt(fd, SOL_IP, IP_TOS, &s->tos, sizeof(s->tos)) < 0)
704
    WARN("IP_TOS");
705
#endif
706 707 708 709 710 711
  
  if (s->ttl >= 0)
    err = sk_set_ttl_int(s);
  else
    err = NULL;

712 713 714 715
bad:
  return err;
}

716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739
/**
 * sk_set_ttl - set TTL for given socket.
 * @s: socket
 * @ttl: TTL value
 *
 * Set TTL for already opened connections when TTL was not set before.
 * Useful for accepted connections when different ones should have 
 * different TTL.
 *
 * Result: 0 for success, -1 for an error.
 */

int
sk_set_ttl(sock *s, int ttl)
{
  char *err;

  s->ttl = ttl;
  if (err = sk_set_ttl_int(s))
    log(L_ERR "sk_set_ttl: %s: %m", err);

  return (err ? -1 : 0);
}

740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804

/* FIXME: check portability  */

static int
sk_set_md5_auth_int(sock *s, sockaddr *sa, char *passwd)
{
  struct tcp_md5sig md5;

  memset(&md5, 0, sizeof(md5));
  memcpy(&md5.tcpm_addr, (struct sockaddr *) sa, sizeof(*sa));

  if (passwd)
    {
      int len = strlen(passwd);

      if (len > TCP_MD5SIG_MAXKEYLEN)
	{
	  log(L_ERR "MD5 password too long");
	  return -1;
	}

      md5.tcpm_keylen = len;
      memcpy(&md5.tcpm_key, passwd, len);
    }

  int rv = setsockopt(s->fd, IPPROTO_TCP, TCP_MD5SIG, &md5, sizeof(md5));

  if (rv < 0) 
    {
      if (errno == ENOPROTOOPT)
	log(L_ERR "Kernel does not support TCP MD5 signatures");
      else
	log(L_ERR "sk_set_md5_auth_int: setsockopt: %m");
    }

  return rv;
}

/**
 * sk_set_md5_auth - add / remove MD5 security association for given socket.
 * @s: socket
 * @a: IP address of the other side
 * @passwd: password used for MD5 authentication
 *
 * In TCP MD5 handling code in kernel, there is a set of pairs
 * (address, password) used to choose password according to
 * address of the other side. This function is useful for
 * listening socket, for active sockets it is enough to set
 * s->password field.
 *
 * When called with passwd != NULL, the new pair is added,
 * When called with passwd == NULL, the existing pair is removed.
 *
 * Result: 0 for success, -1 for an error.
 */

int
sk_set_md5_auth(sock *s, ip_addr a, char *passwd)
{
  sockaddr sa;
  fill_in_sockaddr(&sa, a, 0);
  return sk_set_md5_auth_int(s, &sa, passwd);
}


805
static void
806 807 808 809
sk_tcp_connected(sock *s)
{
  s->type = SK_TCP;
  sk_alloc_bufs(s);
810
  s->tx_hook(s);
811 812
}

813 814 815 816 817 818 819 820 821 822
static int
sk_passive_connected(sock *s, struct sockaddr *sa, int al, int type)
{
  int fd = accept(s->fd, sa, &al);
  if (fd >= 0)
    {
      sock *t = sk_new(s->pool);
      char *err;
      t->type = type;
      t->fd = fd;
823 824 825 826 827
      t->ttl = s->ttl;
      t->tos = s->tos;
      t->rbsize = s->rbsize;
      t->tbsize = s->tbsize;
      if (type == SK_TCP)
828
	get_sockaddr((sockaddr *) sa, &t->daddr, &t->dport, 1);
829
      sk_insert(t);
830 831 832
      if (err = sk_setup(t))
	{
	  log(L_ERR "Incoming connection: %s: %m", err);
833 834
	  rfree(t);
	  return 1;
835 836
	}
      sk_alloc_bufs(t);
837
      s->rx_hook(t, 0);
838 839 840 841 842
      return 1;
    }
  else if (errno != EINTR && errno != EAGAIN)
    {
      log(L_ERR "accept: %m");
843
      s->err_hook(s, errno);
844 845 846 847
    }
  return 0;
}

848 849 850 851 852 853 854 855 856 857
/**
 * sk_open - open a socket
 * @s: socket
 *
 * This function takes a socket resource created by sk_new() and
 * initialized by the user and binds a corresponding network connection
 * to it.
 *
 * Result: 0 for success, -1 for an error.
 */
858 859 860
int
sk_open(sock *s)
{
861
  int fd;
862
  sockaddr sa;
863 864 865 866 867 868 869 870
  int one = 1;
  int type = s->type;
  int has_src = ipa_nonzero(s->saddr) || s->sport;
  char *err;

  switch (type)
    {
    case SK_TCP_ACTIVE:
871 872
      s->ttx = "";			/* Force s->ttx != s->tpos */
      /* Fall thru */
873
    case SK_TCP_PASSIVE:
874
      fd = socket(BIRD_PF, SOCK_STREAM, IPPROTO_TCP);
875 876 877
      break;
    case SK_UDP:
    case SK_UDP_MC:
878
      fd = socket(BIRD_PF, SOCK_DGRAM, IPPROTO_UDP);
879 880 881
      break;
    case SK_IP:
    case SK_IP_MC:
882
      fd = socket(BIRD_PF, SOCK_RAW, s->dport);
883
      break;
884 885 886
    case SK_MAGIC:
      fd = s->fd;
      break;
887
    default:
888
      bug("sk_open() called for invalid sock type %d", type);
889 890 891 892 893 894 895
    }
  if (fd < 0)
    die("sk_open: socket: %m");
  s->fd = fd;

  if (err = sk_setup(s))
    goto bad;
896

897 898 899 900 901
  switch (type)
    {
    case SK_UDP:
    case SK_IP:
      if (s->iface)			/* It's a broadcast socket */
902 903 904
#ifdef IPV6
	bug("IPv6 has no broadcasts");
#else
905 906
	if (setsockopt(fd, SOL_SOCKET, SO_BROADCAST, &one, sizeof(one)) < 0)
	  ERR("SO_BROADCAST");
907
#endif
908 909 910 911
      break;
    case SK_UDP_MC:
    case SK_IP_MC:
      {
912 913 914
#ifdef IPV6
	/* Fortunately, IPv6 socket interface is recent enough and therefore standardized */
	ASSERT(s->iface && s->iface->addr);
915
	if (ipa_nonzero(s->daddr))
916 917
	  {
	    int t = s->iface->index;
918
	    int zero = 0;
919 920 921 922 923 924 925 926 927 928 929
	    if (setsockopt(fd, SOL_IPV6, IPV6_MULTICAST_HOPS, &s->ttl, sizeof(s->ttl)) < 0)
	      ERR("IPV6_MULTICAST_HOPS");
	    if (setsockopt(fd, SOL_IPV6, IPV6_MULTICAST_LOOP, &zero, sizeof(zero)) < 0)
	      ERR("IPV6_MULTICAST_LOOP");
	    if (setsockopt(fd, SOL_IPV6, IPV6_MULTICAST_IF, &t, sizeof(t)) < 0)
	      ERR("IPV6_MULTICAST_IF");
	  }
	if (has_src)
	  {
	    struct ipv6_mreq mreq;
	    set_inaddr(&mreq.ipv6mr_multiaddr, s->daddr);
930
#ifdef CONFIG_IPV6_GLIBC_20
931
	    mreq.ipv6mr_ifindex = s->iface->index;
932 933
#else
	    mreq.ipv6mr_interface = s->iface->index;
934
#endif /* CONFIG_IPV6_GLIBC_20 */
935 936 937 938 939
	    if (setsockopt(fd, SOL_IPV6, IPV6_ADD_MEMBERSHIP, &mreq, sizeof(mreq)) < 0)
	      ERR("IPV6_ADD_MEMBERSHIP");
	  }
#else
	/* With IPv4 there are zillions of different socket interface variants. Ugh. */
940
	ASSERT(s->iface && s->iface->addr);
941 942
	if (err = sysio_mcast_join(s))
	  goto bad;
943
#endif /* IPV6 */
944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959
      break;
      }
    }
  if (has_src)
    {
      int port;

      if (type == SK_IP || type == SK_IP_MC)
	port = 0;
      else
	{
	  port = s->sport;
	  if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &one, sizeof(one)) < 0)
	    ERR("SO_REUSEADDR");
	}
      fill_in_sockaddr(&sa, s->saddr, port);
960 961 962
#ifdef CONFIG_SKIP_MC_BIND
      if (type == SK_IP && bind(fd, (struct sockaddr *) &sa, sizeof(sa)) < 0)
#else
963
      if (bind(fd, (struct sockaddr *) &sa, sizeof(sa)) < 0)
964
#endif
965 966 967
	ERR("bind");
    }
  fill_in_sockaddr(&sa, s->daddr, s->dport);
968 969 970 971 972 973 974 975

  if (s->password)
    {
      int rv = sk_set_md5_auth_int(s, &sa, s->password);
      if (rv < 0)
	goto bad_no_log;
    }

976 977 978 979 980
  switch (type)
    {
    case SK_TCP_ACTIVE:
      if (connect(fd, (struct sockaddr *) &sa, sizeof(sa)) >= 0)
	sk_tcp_connected(s);
981 982
      else if (errno != EINTR && errno != EAGAIN && errno != EINPROGRESS &&
	       errno != ECONNREFUSED && errno != EHOSTUNREACH)
983 984 985 986 987 988
	ERR("connect");
      break;
    case SK_TCP_PASSIVE:
      if (listen(fd, 8))
	ERR("listen");
      break;
989 990 991
    case SK_MAGIC:
      break;
    default:
992
      sk_alloc_bufs(s);
993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009
#ifdef IPV6
#ifdef IPV6_MTU_DISCOVER
      {
	int dont = IPV6_PMTUDISC_DONT;
	if (setsockopt(fd, SOL_IPV6, IPV6_MTU_DISCOVER, &dont, sizeof(dont)) < 0)
	  ERR("IPV6_MTU_DISCOVER");
      }
#endif
#else
#ifdef IP_PMTUDISC
      {
	int dont = IP_PMTUDISC_DONT;
	if (setsockopt(fd, SOL_IP, IP_PMTUDISC, &dont, sizeof(dont)) < 0)
	  ERR("IP_PMTUDISC");
      }
#endif
#endif
1010 1011
    }

1012
  sk_insert(s);
1013 1014 1015 1016
  return 0;

bad:
  log(L_ERR "sk_open: %s: %m", err);
1017
bad_no_log:
1018 1019 1020 1021 1022
  close(fd);
  s->fd = -1;
  return -1;
}

1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
int
sk_open_unix(sock *s, char *name)
{
  int fd;
  struct sockaddr_un sa;
  char *err;

  fd = socket(AF_UNIX, SOCK_STREAM, 0);
  if (fd < 0)
    die("sk_open_unix: socket: %m");
  s->fd = fd;
  if (err = sk_setup(s))
    goto bad;
  unlink(name);
1037 1038 1039 1040
 
  if (strlen(name) >= sizeof(sa.sun_path))
    die("sk_open_unix: path too long");

1041
  sa.sun_family = AF_UNIX;
1042
  strcpy(sa.sun_path, name);
1043
  if (bind(fd, (struct sockaddr *) &sa, SUN_LEN(&sa)) < 0)
1044 1045 1046
    ERR("bind");
  if (listen(fd, 8))
    ERR("listen");
1047
  sk_insert(s);
1048 1049 1050 1051 1052 1053 1054 1055 1056
  return 0;

bad:
  log(L_ERR "sk_open_unix: %s: %m", err);
  close(fd);
  s->fd = -1;
  return -1;
}

1057 1058 1059 1060 1061 1062 1063 1064
static int
sk_maybe_write(sock *s)
{
  int e;

  switch (s->type)
    {
    case SK_TCP:
1065
    case SK_MAGIC:
1066
    case SK_UNIX:
1067 1068 1069 1070 1071 1072 1073
      while (s->ttx != s->tpos)
	{
	  e = write(s->fd, s->ttx, s->tpos - s->ttx);
	  if (e < 0)
	    {
	      if (errno != EINTR && errno != EAGAIN)
		{
1074 1075
                  s->ttx = s->tpos;	/* empty tx buffer */
		  s->err_hook(s, errno);
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
		  return -1;
		}
	      return 0;
	    }
	  s->ttx += e;
	}
      s->ttx = s->tpos = s->tbuf;
      return 1;
    case SK_UDP:
    case SK_UDP_MC:
    case SK_IP:
    case SK_IP_MC:
      {
1089
	sockaddr sa;
1090 1091 1092 1093

	if (s->tbuf == s->tpos)
	  return 1;
	fill_in_sockaddr(&sa, s->faddr, s->fport);
1094

1095 1096 1097 1098 1099
	e = sendto(s->fd, s->tbuf, s->tpos - s->tbuf, 0, (struct sockaddr *) &sa, sizeof(sa));
	if (e < 0)
	  {
	    if (errno != EINTR && errno != EAGAIN)
	      {
1100 1101
                s->ttx = s->tpos;	/* empty tx buffer */
		s->err_hook(s, errno);
1102 1103 1104 1105 1106 1107 1108 1109
		return -1;
	      }
	    return 0;
	  }
	s->tpos = s->tbuf;
	return 1;
      }
    default:
1110
      bug("sk_maybe_write: unknown socket type %d", s->type);
1111 1112 1113
    }
}

1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
/**
 * sk_send - send data to a socket
 * @s: socket
 * @len: number of bytes to send
 *
 * This function sends @len bytes of data prepared in the
 * transmit buffer of the socket @s to the network connection.
 * If the packet can be sent immediately, it does so and returns
 * 1, else it queues the packet for later processing, returns 0
 * and calls the @tx_hook of the socket when the tranmission
 * takes place.
 */
1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
int
sk_send(sock *s, unsigned len)
{
  s->faddr = s->daddr;
  s->fport = s->dport;
  s->ttx = s->tbuf;
  s->tpos = s->tbuf + len;
  return sk_maybe_write(s);
}

1136 1137 1138 1139 1140 1141 1142
/**
 * sk_send_to - send data to a specific destination
 * @s: socket
 * @len: number of bytes to send
 * @addr: IP address to send the packet to
 * @port: port to send the packet to
 *
1143
 * This is a sk_send() replacement for connection-less packet sockets
1144 1145
 * which allows destination of the packet to be chosen dynamically.
 */
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
int
sk_send_to(sock *s, unsigned len, ip_addr addr, unsigned port)
{
  s->faddr = addr;
  s->fport = port;
  s->ttx = s->tbuf;
  s->tpos = s->tbuf + len;
  return sk_maybe_write(s);
}

static int
sk_read(sock *s)
{
  switch (s->type)
    {
    case SK_TCP_PASSIVE:
      {
1163
	sockaddr sa;
1164 1165 1166 1167 1168 1169
	return sk_passive_connected(s, (struct sockaddr *) &sa, sizeof(sa), SK_TCP);
      }
    case SK_UNIX_PASSIVE:
      {
	struct sockaddr_un sa;
	return sk_passive_connected(s, (struct sockaddr *) &sa, sizeof(sa), SK_UNIX);
1170 1171
      }
    case SK_TCP:
1172
    case SK_UNIX:
1173 1174 1175 1176 1177 1178
      {
	int c = read(s->fd, s->rpos, s->rbuf + s->rbsize - s->rpos);

	if (c < 0)
	  {
	    if (errno != EINTR && errno != EAGAIN)
1179
	      s->err_hook(s, errno);
1180 1181
	  }
	else if (!c)
1182
	  s->err_hook(s, 0);
1183 1184 1185 1186
	else
	  {
	    s->rpos += c;
	    if (s->rx_hook(s, s->rpos - s->rbuf))
1187 1188 1189 1190 1191
	      {
		/* We need to be careful since the socket could have been deleted by the hook */
		if (current_sock == s)
		  s->rpos = s->rbuf;
	      }
1192 1193 1194 1195
	    return 1;
	  }
	return 0;
      }
1196 1197
    case SK_MAGIC:
      return s->rx_hook(s, 0);
1198 1199
    default:
      {
1200
	sockaddr sa;
1201 1202 1203 1204 1205 1206
	int al = sizeof(sa);
	int e = recvfrom(s->fd, s->rbuf, s->rbsize, 0, (struct sockaddr *) &sa, &al);

	if (e < 0)
	  {
	    if (errno != EINTR && errno != EAGAIN)
1207
	      s->err_hook(s, errno);
1208 1209 1210
	    return 0;
	  }
	s->rpos = s->rbuf + e;
1211
	get_sockaddr(&sa, &s->faddr, &s->fport, 1);
1212 1213 1214 1215 1216 1217
	s->rx_hook(s, e);
	return 1;
      }
    }
}

1218
static int
1219 1220
sk_write(sock *s)
{
1221 1222 1223 1224 1225 1226
  switch (s->type)
    {
    case SK_TCP_ACTIVE:
      {
	sockaddr sa;
	fill_in_sockaddr(&sa, s->daddr, s->dport);
Ondřej Filip's avatar
Ondřej Filip committed
1227
	if (connect(s->fd, (struct sockaddr *) &sa, sizeof(sa)) >= 0 || errno == EISCONN)
1228 1229
	  sk_tcp_connected(s);
	else if (errno != EINTR && errno != EAGAIN && errno != EINPROGRESS)
1230
	  s->err_hook(s, errno);
1231
	return 0;
1232 1233
      }
    default:
1234 1235 1236 1237 1238 1239
      if (s->ttx != s->tpos && sk_maybe_write(s) > 0)
	{
	  s->tx_hook(s);
	  return 1;
	}
      return 0;
1240
    }
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259
}

void
sk_dump_all(void)
{
  node *n;
  sock *s;

  debug("Open sockets:\n");
  WALK_LIST(n, sock_list)
    {
      s = SKIP_BACK(sock, n, n);
      debug("%p ", s);
      sk_dump(&s->r);
    }
  debug("\n");
}

#undef ERR
1260
#undef WARN
1261 1262 1263 1264 1265

/*
 *	Main I/O Loop
 */

1266 1267 1268
volatile int async_config_flag;		/* Asynchronous reconfiguration/dump scheduled */
volatile int async_dump_flag;

1269 1270 1271 1272 1273 1274
void
io_init(void)
{
  init_list(&near_timers);
  init_list(&far_timers);
  init_list(&sock_list);
1275
  init_list(&global_event_list);
1276
  krt_io_init();
1277 1278 1279
  init_times();
  update_times();
  srandom((int) now_real);
1280 1281 1282 1283 1284 1285 1286 1287
}

void
io_loop(void)
{
  fd_set rd, wr;
  struct timeval timo;
  time_t tout;
1288
  int hi, events;
1289
  sock *s;
1290
  node *n;
1291

1292
  sock_recalc_fdsets_p = 1;
1293 1294
  for(;;)
    {
1295
      events = ev_run_list(&global_event_list);
1296
      update_times();
1297 1298 1299 1300 1301 1302
      tout = tm_first_shot();
      if (tout <= now)
	{
	  tm_shot();
	  continue;
	}
1303 1304
      timo.tv_sec = events ? 0 : tout - now;
      timo.tv_usec = 0;
1305

1306 1307 1308 1309 1310 1311 1312
      if (sock_recalc_fdsets_p)
	{
	  sock_recalc_fdsets_p = 0;
	  FD_ZERO(&rd);
	  FD_ZERO(&wr);
	}

1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
      hi = 0;
      WALK_LIST(n, sock_list)
	{
	  s = SKIP_BACK(sock, n, n);
	  if (s->rx_hook)
	    {
	      FD_SET(s->fd, &rd);
	      if (s->fd > hi)
		hi = s->fd;
	    }
1323 1324
	  else
	    FD_CLR(s->fd, &rd);
1325 1326 1327 1328 1329 1330
	  if (s->tx_hook && s->ttx != s->tpos)
	    {
	      FD_SET(s->fd, &wr);
	      if (s->fd > hi)
		hi = s->fd;
	    }
1331 1332
	  else
	    FD_CLR(s->fd, &wr);
1333 1334
	}

1335 1336 1337 1338 1339 1340 1341 1342 1343
      /*
       * Yes, this is racy. But even if the signal comes before this test
       * and entering select(), it gets caught on the next timer tick.
       */

      if (async_config_flag)
	{
	  async_config();
	  async_config_flag = 0;
1344
	  continue;
1345 1346 1347 1348 1349
	}
      if (async_dump_flag)
	{
	  async_dump();
	  async_dump_flag = 0;
1350 1351 1352 1353 1354 1355 1356
	  continue;
	}
      if (async_shutdown_flag)
	{
	  async_shutdown();
	  async_shutdown_flag = 0;
	  continue;
1357 1358 1359 1360
	}

      /* And finally enter select() to find active sockets */

1361 1362 1363 1364 1365 1366 1367 1368 1369
      hi = select(hi+1, &rd, &wr, NULL, &timo);
      if (hi < 0)
	{
	  if (errno == EINTR || errno == EAGAIN)
	    continue;
	  die("select: %m");
	}
      if (hi)
	{
1370 1371
	  current_sock = SKIP_BACK(sock, n, HEAD(sock_list));	/* guaranteed to be non-empty */
	  while (current_sock)
1372
	    {
1373 1374
	      sock *s = current_sock;
	      int e;
1375
	      if (FD_ISSET(s->fd, &rd))
1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392
		do
		  {
		    e = sk_read(s);
		    if (s != current_sock)
		      goto next;
		  }
		while (e);
	      if (FD_ISSET(s->fd, &wr))
		do
		  {
		    e = sk_write(s);
		    if (s != current_sock)
		      goto next;
		  }
		while (e);
	      current_sock = sk_next(s);
	    next: ;
1393 1394 1395 1396
	    }
	}
    }
}
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416

void
test_old_bird(char *path)
{
  int fd;
  struct sockaddr_un sa;

  fd = socket(AF_UNIX, SOCK_STREAM, 0);

  if (fd < 0)
    die("Cannot create socket: %m");
  bzero(&sa, sizeof(sa));
  sa.sun_family = AF_UNIX;
  strcpy(sa.sun_path, path);
  if (connect(fd, (struct sockaddr *) &sa, SUN_LEN(&sa)) == 0)
    die("I found another BIRD running.");
  close(fd);
}