git.meshlink.io Git - utcp/blob - utcp.c

   1 /*
   2     utcp.c -- Userspace TCP
   3     Copyright (C) 2014-2017 Guus Sliepen <guus@tinc-vpn.org>
   4
   5     This program is free software; you can redistribute it and/or modify
   6     it under the terms of the GNU General Public License as published by
   7     the Free Software Foundation; either version 2 of the License, or
   8     (at your option) any later version.
   9
  10     This program is distributed in the hope that it will be useful,
  11     but WITHOUT ANY WARRANTY; without even the implied warranty of
  12     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13     GNU General Public License for more details.
  14
  15     You should have received a copy of the GNU General Public License along
  16     with this program; if not, write to the Free Software Foundation, Inc.,
  17     51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18 */
  19
  20 #define _GNU_SOURCE
  21
  22 #include <assert.h>
  23 #include <errno.h>
  24 #include <stdio.h>
  25 #include <stdlib.h>
  26 #include <stdint.h>
  27 #include <stdbool.h>
  28 #include <string.h>
  29 #include <unistd.h>
  30 #include <time.h>
  31
  32 #include "utcp_priv.h"
  33
  34 #ifndef EBADMSG
  35 #define EBADMSG         104
  36 #endif
  37
  38 #ifndef SHUT_RDWR
  39 #define SHUT_RDWR 2
  40 #endif
  41
  42 #ifdef poll
  43 #undef poll
  44 #endif
  45
  46 #ifndef UTCP_CLOCK
  47 #if defined(CLOCK_MONOTONIC_RAW) && defined(__x86_64__)
  48 #define UTCP_CLOCK CLOCK_MONOTONIC_RAW
  49 #else
  50 #define UTCP_CLOCK CLOCK_MONOTONIC
  51 #endif
  52 #endif
  53
  54 static void timespec_sub(const struct timespec *a, const struct timespec *b, struct timespec *r) {
  55         r->tv_sec = a->tv_sec - b->tv_sec;
  56         r->tv_nsec = a->tv_nsec - b->tv_nsec;
  57
  58         if(r->tv_nsec < 0) {
  59                 r->tv_sec--, r->tv_nsec += NSEC_PER_SEC;
  60         }
  61 }
  62
  63 static int32_t timespec_diff_usec(const struct timespec *a, const struct timespec *b) {
  64         int64_t diff = (a->tv_sec - b->tv_sec) * 1000000000 + a->tv_sec - b->tv_sec;
  65         return diff / 1000;
  66 }
  67
  68 static bool timespec_lt(const struct timespec *a, const struct timespec *b) {
  69         if(a->tv_sec == b->tv_sec) {
  70                 return a->tv_nsec < b->tv_nsec;
  71         } else {
  72                 return a->tv_sec < b->tv_sec;
  73         }
  74 }
  75
  76 static void timespec_clear(struct timespec *a) {
  77         a->tv_sec = 0;
  78 }
  79
  80 static bool timespec_isset(const struct timespec *a) {
  81         return a->tv_sec;
  82 }
  83
  84 static long CLOCK_GRANULARITY; // usec
  85
  86 static inline size_t min(size_t a, size_t b) {
  87         return a < b ? a : b;
  88 }
  89
  90 static inline size_t max(size_t a, size_t b) {
  91         return a > b ? a : b;
  92 }
  93
  94 #ifdef UTCP_DEBUG
  95 #include <stdarg.h>
  96
  97 #ifndef UTCP_DEBUG_DATALEN
  98 #define UTCP_DEBUG_DATALEN 20
  99 #endif
 100
 101 static void debug(struct utcp_connection *c, const char *format, ...) {
 102         struct timespec tv;
 103         char buf[1024];
 104         int len;
 105
 106         clock_gettime(CLOCK_REALTIME, &tv);
 107         len = snprintf(buf, sizeof(buf), "%ld.%06lu %u:%u ", (long)tv.tv_sec, tv.tv_nsec / 1000, c ? c->src : 0, c ? c->dst : 0);
 108         va_list ap;
 109         va_start(ap, format);
 110         len += vsnprintf(buf + len, sizeof(buf) - len, format, ap);
 111         va_end(ap);
 112
 113         if(len > 0 && (size_t)len < sizeof(buf)) {
 114                 fwrite(buf, len, 1, stderr);
 115         }
 116 }
 117
 118 static void print_packet(struct utcp_connection *c, const char *dir, const void *pkt, size_t len) {
 119         struct hdr hdr;
 120
 121         if(len < sizeof(hdr)) {
 122                 debug(c, "%s: short packet (%lu bytes)\n", dir, (unsigned long)len);
 123                 return;
 124         }
 125
 126         memcpy(&hdr, pkt, sizeof(hdr));
 127
 128         uint32_t datalen;
 129
 130         if(len > sizeof(hdr)) {
 131                 datalen = min(len - sizeof(hdr), UTCP_DEBUG_DATALEN);
 132         } else {
 133                 datalen = 0;
 134         }
 135
 136
 137         const uint8_t *data = (uint8_t *)pkt + sizeof(hdr);
 138         char str[datalen * 2 + 1];
 139         char *p = str;
 140
 141         for(uint32_t i = 0; i < datalen; i++) {
 142                 *p++ = "0123456789ABCDEF"[data[i] >> 4];
 143                 *p++ = "0123456789ABCDEF"[data[i] & 15];
 144         }
 145
 146         *p = 0;
 147
 148         debug(c, "%s: len %lu src %u dst %u seq %u ack %u wnd %u aux %x ctl %s%s%s%s data %s\n",
 149               dir, (unsigned long)len, hdr.src, hdr.dst, hdr.seq, hdr.ack, hdr.wnd, hdr.aux,
 150               hdr.ctl & SYN ? "SYN" : "",
 151               hdr.ctl & RST ? "RST" : "",
 152               hdr.ctl & FIN ? "FIN" : "",
 153               hdr.ctl & ACK ? "ACK" : "",
 154               str
 155              );
 156 }
 157
 158 static void debug_cwnd(struct utcp_connection *c) {
 159         debug(c, "snd.cwnd %u snd.ssthresh %u\n", c->snd.cwnd, ~c->snd.ssthresh ? c->snd.ssthresh : 0);
 160 }
 161 #else
 162 #define debug(...) do {} while(0)
 163 #define print_packet(...) do {} while(0)
 164 #define debug_cwnd(...) do {} while(0)
 165 #endif
 166
 167 static void set_state(struct utcp_connection *c, enum state state) {
 168         c->state = state;
 169
 170         if(state == ESTABLISHED) {
 171                 timespec_clear(&c->conn_timeout);
 172         }
 173
 174         debug(c, "state %s\n", strstate[state]);
 175 }
 176
 177 static bool fin_wanted(struct utcp_connection *c, uint32_t seq) {
 178         if(seq != c->snd.last) {
 179                 return false;
 180         }
 181
 182         switch(c->state) {
 183         case FIN_WAIT_1:
 184         case CLOSING:
 185         case LAST_ACK:
 186                 return true;
 187
 188         default:
 189                 return false;
 190         }
 191 }
 192
 193 static bool is_reliable(struct utcp_connection *c) {
 194         return c->flags & UTCP_RELIABLE;
 195 }
 196
 197 static int32_t seqdiff(uint32_t a, uint32_t b) {
 198         return a - b;
 199 }
 200
 201 // Buffer functions
 202 static bool buffer_wraps(struct buffer *buf) {
 203         return buf->size - buf->offset < buf->used;
 204 }
 205
 206 static bool buffer_resize(struct buffer *buf, uint32_t newsize) {
 207         char *newdata = realloc(buf->data, newsize);
 208
 209         if(!newdata) {
 210                 return false;
 211         }
 212
 213         buf->data = newdata;
 214
 215         if(buffer_wraps(buf)) {
 216                 // Shift the right part of the buffer until it hits the end of the new buffer.
 217                 // Old situation:
 218                 // [345......012]
 219                 // New situation:
 220                 // [345.........|........012]
 221                 uint32_t tailsize = buf->size - buf->offset;
 222                 uint32_t newoffset = newsize - tailsize;
 223                 memmove(buf->data + newoffset, buf->data + buf->offset, tailsize);
 224                 buf->offset = newoffset;
 225         }
 226
 227         buf->size = newsize;
 228         return true;
 229 }
 230
 231 // Store data into the buffer
 232 static ssize_t buffer_put_at(struct buffer *buf, size_t offset, const void *data, size_t len) {
 233         debug(NULL, "buffer_put_at %lu %lu %lu\n", (unsigned long)buf->used, (unsigned long)offset, (unsigned long)len);
 234
 235         // Ensure we don't store more than maxsize bytes in total
 236         size_t required = offset + len;
 237
 238         if(required > buf->maxsize) {
 239                 if(offset >= buf->maxsize) {
 240                         return 0;
 241                 }
 242
 243                 len = buf->maxsize - offset;
 244                 required = buf->maxsize;
 245         }
 246
 247         // Check if we need to resize the buffer
 248         if(required > buf->size) {
 249                 size_t newsize = buf->size;
 250
 251                 if(!newsize) {
 252                         newsize = 4096;
 253                 }
 254
 255                 do {
 256                         newsize *= 2;
 257                 } while(newsize < required);
 258
 259                 if(newsize > buf->maxsize) {
 260                         newsize = buf->maxsize;
 261                 }
 262
 263                 if(!buffer_resize(buf, newsize)) {
 264                         return -1;
 265                 }
 266         }
 267
 268         uint32_t realoffset = buf->offset + offset;
 269
 270         if(buf->size - buf->offset < offset) {
 271                 // The offset wrapped
 272                 realoffset -= buf->size;
 273         }
 274
 275         if(buf->size - realoffset < len) {
 276                 // The new chunk of data must be wrapped
 277                 memcpy(buf->data + realoffset, data, buf->size - realoffset);
 278                 memcpy(buf->data, (char *)data + buf->size - realoffset, len - (buf->size - realoffset));
 279         } else {
 280                 memcpy(buf->data + realoffset, data, len);
 281         }
 282
 283         if(required > buf->used) {
 284                 buf->used = required;
 285         }
 286
 287         return len;
 288 }
 289
 290 static ssize_t buffer_put(struct buffer *buf, const void *data, size_t len) {
 291         return buffer_put_at(buf, buf->used, data, len);
 292 }
 293
 294 // Copy data from the buffer without removing it.
 295 static ssize_t buffer_copy(struct buffer *buf, void *data, size_t offset, size_t len) {
 296         // Ensure we don't copy more than is actually stored in the buffer
 297         if(offset >= buf->used) {
 298                 return 0;
 299         }
 300
 301         if(buf->used - offset < len) {
 302                 len = buf->used - offset;
 303         }
 304
 305         uint32_t realoffset = buf->offset + offset;
 306
 307         if(buf->size - buf->offset < offset) {
 308                 // The offset wrapped
 309                 realoffset -= buf->size;
 310         }
 311
 312         if(buf->size - realoffset < len) {
 313                 // The data is wrapped
 314                 memcpy(data, buf->data + realoffset, buf->size - realoffset);
 315                 memcpy((char *)data + buf->size - realoffset, buf->data, len - (buf->size - realoffset));
 316         } else {
 317                 memcpy(data, buf->data + realoffset, len);
 318         }
 319
 320         return len;
 321 }
 322
 323 // Copy data from the buffer without removing it.
 324 static ssize_t buffer_call(struct buffer *buf, utcp_recv_t cb, void *arg, size_t offset, size_t len) {
 325         // Ensure we don't copy more than is actually stored in the buffer
 326         if(offset >= buf->used) {
 327                 return 0;
 328         }
 329
 330         if(buf->used - offset < len) {
 331                 len = buf->used - offset;
 332         }
 333
 334         uint32_t realoffset = buf->offset + offset;
 335
 336         if(buf->size - buf->offset < offset) {
 337                 // The offset wrapped
 338                 realoffset -= buf->size;
 339         }
 340
 341         if(buf->size - realoffset < len) {
 342                 // The data is wrapped
 343                 ssize_t rx1 = cb(arg, buf->data + realoffset, buf->size - realoffset);
 344
 345                 if(rx1 < buf->size - realoffset) {
 346                         return rx1;
 347                 }
 348
 349                 ssize_t rx2 = cb(arg, buf->data, len - (buf->size - realoffset));
 350
 351                 if(rx2 < 0) {
 352                         return rx2;
 353                 } else {
 354                         return rx1 + rx2;
 355                 }
 356         } else {
 357                 return cb(arg, buf->data + realoffset, len);
 358         }
 359 }
 360
 361 // Discard data from the buffer.
 362 static ssize_t buffer_discard(struct buffer *buf, size_t len) {
 363         if(buf->used < len) {
 364                 len = buf->used;
 365         }
 366
 367         if(buf->size - buf->offset < len) {
 368                 buf->offset -= buf->size;
 369         }
 370
 371         buf->offset += len;
 372         buf->used -= len;
 373
 374         return len;
 375 }
 376
 377 static bool buffer_set_size(struct buffer *buf, uint32_t minsize, uint32_t maxsize) {
 378         if(maxsize < minsize) {
 379                 maxsize = minsize;
 380         }
 381
 382         buf->maxsize = maxsize;
 383
 384         return buf->size >= minsize || buffer_resize(buf, minsize);
 385 }
 386
 387 static void buffer_exit(struct buffer *buf) {
 388         free(buf->data);
 389         memset(buf, 0, sizeof(*buf));
 390 }
 391
 392 static uint32_t buffer_free(const struct buffer *buf) {
 393         return buf->maxsize - buf->used;
 394 }
 395
 396 // Connections are stored in a sorted list.
 397 // This gives O(log(N)) lookup time, O(N log(N)) insertion time and O(N) deletion time.
 398
 399 static int compare(const void *va, const void *vb) {
 400         assert(va && vb);
 401
 402         const struct utcp_connection *a = *(struct utcp_connection **)va;
 403         const struct utcp_connection *b = *(struct utcp_connection **)vb;
 404
 405         assert(a && b);
 406         assert(a->src && b->src);
 407
 408         int c = (int)a->src - (int)b->src;
 409
 410         if(c) {
 411                 return c;
 412         }
 413
 414         c = (int)a->dst - (int)b->dst;
 415         return c;
 416 }
 417
 418 static struct utcp_connection *find_connection(const struct utcp *utcp, uint16_t src, uint16_t dst) {
 419         if(!utcp->nconnections) {
 420                 return NULL;
 421         }
 422
 423         struct utcp_connection key = {
 424                 .src = src,
 425                 .dst = dst,
 426         }, *keyp = &key;
 427         struct utcp_connection **match = bsearch(&keyp, utcp->connections, utcp->nconnections, sizeof(*utcp->connections), compare);
 428         return match ? *match : NULL;
 429 }
 430
 431 static void free_connection(struct utcp_connection *c) {
 432         struct utcp *utcp = c->utcp;
 433         struct utcp_connection **cp = bsearch(&c, utcp->connections, utcp->nconnections, sizeof(*utcp->connections), compare);
 434
 435         assert(cp);
 436
 437         int i = cp - utcp->connections;
 438         memmove(cp, cp + 1, (utcp->nconnections - i - 1) * sizeof(*cp));
 439         utcp->nconnections--;
 440
 441         buffer_exit(&c->rcvbuf);
 442         buffer_exit(&c->sndbuf);
 443         free(c);
 444 }
 445
 446 static struct utcp_connection *allocate_connection(struct utcp *utcp, uint16_t src, uint16_t dst) {
 447         // Check whether this combination of src and dst is free
 448
 449         if(src) {
 450                 if(find_connection(utcp, src, dst)) {
 451                         errno = EADDRINUSE;
 452                         return NULL;
 453                 }
 454         } else { // If src == 0, generate a random port number with the high bit set
 455                 if(utcp->nconnections >= 32767) {
 456                         errno = ENOMEM;
 457                         return NULL;
 458                 }
 459
 460                 src = rand() | 0x8000;
 461
 462                 while(find_connection(utcp, src, dst)) {
 463                         src++;
 464                 }
 465         }
 466
 467         // Allocate memory for the new connection
 468
 469         if(utcp->nconnections >= utcp->nallocated) {
 470                 if(!utcp->nallocated) {
 471                         utcp->nallocated = 4;
 472                 } else {
 473                         utcp->nallocated *= 2;
 474                 }
 475
 476                 struct utcp_connection **new_array = realloc(utcp->connections, utcp->nallocated * sizeof(*utcp->connections));
 477
 478                 if(!new_array) {
 479                         return NULL;
 480                 }
 481
 482                 utcp->connections = new_array;
 483         }
 484
 485         struct utcp_connection *c = calloc(1, sizeof(*c));
 486
 487         if(!c) {
 488                 return NULL;
 489         }
 490
 491         if(!buffer_set_size(&c->sndbuf, DEFAULT_SNDBUFSIZE, DEFAULT_MAXSNDBUFSIZE)) {
 492                 free(c);
 493                 return NULL;
 494         }
 495
 496         if(!buffer_set_size(&c->rcvbuf, DEFAULT_RCVBUFSIZE, DEFAULT_MAXRCVBUFSIZE)) {
 497                 buffer_exit(&c->sndbuf);
 498                 free(c);
 499                 return NULL;
 500         }
 501
 502         // Fill in the details
 503
 504         c->src = src;
 505         c->dst = dst;
 506 #ifdef UTCP_DEBUG
 507         c->snd.iss = 0;
 508 #else
 509         c->snd.iss = rand();
 510 #endif
 511         c->snd.una = c->snd.iss;
 512         c->snd.nxt = c->snd.iss + 1;
 513         c->snd.last = c->snd.nxt;
 514         c->snd.cwnd = (utcp->mss > 2190 ? 2 : utcp->mss > 1095 ? 3 : 4) * utcp->mss;
 515         c->snd.ssthresh = ~0;
 516         debug_cwnd(c);
 517         c->utcp = utcp;
 518
 519         // Add it to the sorted list of connections
 520
 521         utcp->connections[utcp->nconnections++] = c;
 522         qsort(utcp->connections, utcp->nconnections, sizeof(*utcp->connections), compare);
 523
 524         return c;
 525 }
 526
 527 static inline uint32_t absdiff(uint32_t a, uint32_t b) {
 528         if(a > b) {
 529                 return a - b;
 530         } else {
 531                 return b - a;
 532         }
 533 }
 534
 535 // Update RTT variables. See RFC 6298.
 536 static void update_rtt(struct utcp_connection *c, uint32_t rtt) {
 537         if(!rtt) {
 538                 debug(c, "invalid rtt\n");
 539                 return;
 540         }
 541
 542         struct utcp *utcp = c->utcp;
 543
 544         if(!utcp->srtt) {
 545                 utcp->srtt = rtt;
 546                 utcp->rttvar = rtt / 2;
 547         } else {
 548                 utcp->rttvar = (utcp->rttvar * 3 + absdiff(utcp->srtt, rtt)) / 4;
 549                 utcp->srtt = (utcp->srtt * 7 + rtt) / 8;
 550         }
 551
 552         utcp->rto = utcp->srtt + max(4 * utcp->rttvar, CLOCK_GRANULARITY);
 553
 554         if(utcp->rto > MAX_RTO) {
 555                 utcp->rto = MAX_RTO;
 556         }
 557
 558         debug(c, "rtt %u srtt %u rttvar %u rto %u\n", rtt, utcp->srtt, utcp->rttvar, utcp->rto);
 559 }
 560
 561 static void start_retransmit_timer(struct utcp_connection *c) {
 562         clock_gettime(UTCP_CLOCK, &c->rtrx_timeout);
 563
 564         uint32_t rto = c->utcp->rto;
 565
 566         while(rto > USEC_PER_SEC) {
 567                 c->rtrx_timeout.tv_sec++;
 568                 rto -= USEC_PER_SEC;
 569         }
 570
 571         c->rtrx_timeout.tv_nsec += c->utcp->rto * 1000;
 572
 573         if(c->rtrx_timeout.tv_nsec >= NSEC_PER_SEC) {
 574                 c->rtrx_timeout.tv_nsec -= NSEC_PER_SEC;
 575                 c->rtrx_timeout.tv_sec++;
 576         }
 577
 578         debug(c, "rtrx_timeout %ld.%06lu\n", c->rtrx_timeout.tv_sec, c->rtrx_timeout.tv_nsec);
 579 }
 580
 581 static void stop_retransmit_timer(struct utcp_connection *c) {
 582         timespec_clear(&c->rtrx_timeout);
 583         debug(c, "rtrx_timeout cleared\n");
 584 }
 585
 586 struct utcp_connection *utcp_connect_ex(struct utcp *utcp, uint16_t dst, utcp_recv_t recv, void *priv, uint32_t flags) {
 587         struct utcp_connection *c = allocate_connection(utcp, 0, dst);
 588
 589         if(!c) {
 590                 return NULL;
 591         }
 592
 593         assert((flags & ~0x1f) == 0);
 594
 595         c->flags = flags;
 596         c->recv = recv;
 597         c->priv = priv;
 598
 599         struct {
 600                 struct hdr hdr;
 601                 uint8_t init[4];
 602         } pkt;
 603
 604         pkt.hdr.src = c->src;
 605         pkt.hdr.dst = c->dst;
 606         pkt.hdr.seq = c->snd.iss;
 607         pkt.hdr.ack = 0;
 608         pkt.hdr.wnd = c->rcvbuf.maxsize;
 609         pkt.hdr.ctl = SYN;
 610         pkt.hdr.aux = 0x0101;
 611         pkt.init[0] = 1;
 612         pkt.init[1] = 0;
 613         pkt.init[2] = 0;
 614         pkt.init[3] = flags & 0x7;
 615
 616         set_state(c, SYN_SENT);
 617
 618         print_packet(c, "send", &pkt, sizeof(pkt));
 619         utcp->send(utcp, &pkt, sizeof(pkt));
 620
 621         clock_gettime(UTCP_CLOCK, &c->conn_timeout);
 622         c->conn_timeout.tv_sec += utcp->timeout;
 623
 624         start_retransmit_timer(c);
 625
 626         return c;
 627 }
 628
 629 struct utcp_connection *utcp_connect(struct utcp *utcp, uint16_t dst, utcp_recv_t recv, void *priv) {
 630         return utcp_connect_ex(utcp, dst, recv, priv, UTCP_TCP);
 631 }
 632
 633 void utcp_accept(struct utcp_connection *c, utcp_recv_t recv, void *priv) {
 634         if(c->reapable || c->state != SYN_RECEIVED) {
 635                 debug(c, "accept() called on invalid connection in state %s\n", c, strstate[c->state]);
 636                 return;
 637         }
 638
 639         debug(c, "accepted %p %p\n", c, recv, priv);
 640         c->recv = recv;
 641         c->priv = priv;
 642         set_state(c, ESTABLISHED);
 643 }
 644
 645 static void ack(struct utcp_connection *c, bool sendatleastone) {
 646         int32_t left = seqdiff(c->snd.last, c->snd.nxt);
 647         int32_t cwndleft = min(c->snd.cwnd, c->snd.wnd) - seqdiff(c->snd.nxt, c->snd.una);
 648
 649         assert(left >= 0);
 650
 651         if(cwndleft <= 0) {
 652                 left = 0;
 653         } else if(cwndleft < left) {
 654                 left = cwndleft;
 655
 656                 if(!sendatleastone || cwndleft > c->utcp->mss) {
 657                         left -= left % c->utcp->mss;
 658                 }
 659         }
 660
 661         debug(c, "cwndleft %d left %d\n", cwndleft, left);
 662
 663         if(!left && !sendatleastone) {
 664                 return;
 665         }
 666
 667         struct {
 668                 struct hdr hdr;
 669                 uint8_t data[];
 670         } *pkt = c->utcp->pkt;
 671
 672         pkt->hdr.src = c->src;
 673         pkt->hdr.dst = c->dst;
 674         pkt->hdr.ack = c->rcv.nxt;
 675         pkt->hdr.wnd = c->rcvbuf.maxsize;
 676         pkt->hdr.ctl = ACK;
 677         pkt->hdr.aux = 0;
 678
 679         do {
 680                 uint32_t seglen = left > c->utcp->mss ? c->utcp->mss : left;
 681                 pkt->hdr.seq = c->snd.nxt;
 682
 683                 buffer_copy(&c->sndbuf, pkt->data, seqdiff(c->snd.nxt, c->snd.una), seglen);
 684
 685                 c->snd.nxt += seglen;
 686                 left -= seglen;
 687
 688                 if(seglen && fin_wanted(c, c->snd.nxt)) {
 689                         seglen--;
 690                         pkt->hdr.ctl |= FIN;
 691                 }
 692
 693                 if(!c->rtt_start.tv_sec) {
 694                         // Start RTT measurement
 695                         clock_gettime(UTCP_CLOCK, &c->rtt_start);
 696                         c->rtt_seq = pkt->hdr.seq + seglen;
 697                         debug(c, "starting RTT measurement, expecting ack %u\n", c->rtt_seq);
 698                 }
 699
 700                 print_packet(c, "send", pkt, sizeof(pkt->hdr) + seglen);
 701                 c->utcp->send(c->utcp, pkt, sizeof(pkt->hdr) + seglen);
 702         } while(left);
 703 }
 704
 705 ssize_t utcp_send(struct utcp_connection *c, const void *data, size_t len) {
 706         if(c->reapable) {
 707                 debug(c, "send() called on closed connection\n");
 708                 errno = EBADF;
 709                 return -1;
 710         }
 711
 712         switch(c->state) {
 713         case CLOSED:
 714         case LISTEN:
 715                 debug(c, "send() called on unconnected connection\n");
 716                 errno = ENOTCONN;
 717                 return -1;
 718
 719         case SYN_SENT:
 720         case SYN_RECEIVED:
 721         case ESTABLISHED:
 722         case CLOSE_WAIT:
 723                 break;
 724
 725         case FIN_WAIT_1:
 726         case FIN_WAIT_2:
 727         case CLOSING:
 728         case LAST_ACK:
 729         case TIME_WAIT:
 730                 debug(c, "send() called on closed connection\n");
 731                 errno = EPIPE;
 732                 return -1;
 733         }
 734
 735         // Exit early if we have nothing to send.
 736
 737         if(!len) {
 738                 return 0;
 739         }
 740
 741         if(!data) {
 742                 errno = EFAULT;
 743                 return -1;
 744         }
 745
 746         // Check if we need to be able to buffer all data
 747
 748         if(c->flags & UTCP_NO_PARTIAL) {
 749                 if(len > buffer_free(&c->sndbuf)) {
 750                         if(len > c->sndbuf.maxsize) {
 751                                 errno = EMSGSIZE;
 752                                 return -1;
 753                         } else {
 754                                 errno = EWOULDBLOCK;
 755                                 return 0;
 756                         }
 757                 }
 758         }
 759
 760         // Add data to send buffer.
 761
 762         if(is_reliable(c) || (c->state != SYN_SENT && c->state != SYN_RECEIVED)) {
 763                 len = buffer_put(&c->sndbuf, data, len);
 764         } else {
 765                 return 0;
 766         }
 767
 768         if(len <= 0) {
 769                 if(is_reliable(c)) {
 770                         errno = EWOULDBLOCK;
 771                         return 0;
 772                 } else {
 773                         return len;
 774                 }
 775         }
 776
 777         c->snd.last += len;
 778
 779         // Don't send anything yet if the connection has not fully established yet
 780
 781         if(c->state == SYN_SENT || c->state == SYN_RECEIVED) {
 782                 return len;
 783         }
 784
 785         ack(c, false);
 786
 787         if(!is_reliable(c)) {
 788                 c->snd.una = c->snd.nxt = c->snd.last;
 789                 buffer_discard(&c->sndbuf, c->sndbuf.used);
 790                 c->do_poll = true;
 791         }
 792
 793         if(is_reliable(c) && !timespec_isset(&c->rtrx_timeout)) {
 794                 start_retransmit_timer(c);
 795         }
 796
 797         if(is_reliable(c) && !timespec_isset(&c->conn_timeout)) {
 798                 clock_gettime(UTCP_CLOCK, &c->conn_timeout);
 799                 c->conn_timeout.tv_sec += c->utcp->timeout;
 800         }
 801
 802         return len;
 803 }
 804
 805 static void swap_ports(struct hdr *hdr) {
 806         uint16_t tmp = hdr->src;
 807         hdr->src = hdr->dst;
 808         hdr->dst = tmp;
 809 }
 810
 811 static void fast_retransmit(struct utcp_connection *c) {
 812         if(c->state == CLOSED || c->snd.last == c->snd.una) {
 813                 debug(c, "fast_retransmit() called but nothing to retransmit!\n");
 814                 return;
 815         }
 816
 817         struct utcp *utcp = c->utcp;
 818
 819         struct {
 820                 struct hdr hdr;
 821                 uint8_t data[];
 822         } *pkt;
 823
 824         pkt = malloc(c->utcp->mtu);
 825
 826         if(!pkt) {
 827                 return;
 828         }
 829
 830         pkt->hdr.src = c->src;
 831         pkt->hdr.dst = c->dst;
 832         pkt->hdr.wnd = c->rcvbuf.maxsize;
 833         pkt->hdr.aux = 0;
 834
 835         switch(c->state) {
 836         case ESTABLISHED:
 837         case FIN_WAIT_1:
 838         case CLOSE_WAIT:
 839         case CLOSING:
 840         case LAST_ACK:
 841                 // Send unacked data again.
 842                 pkt->hdr.seq = c->snd.una;
 843                 pkt->hdr.ack = c->rcv.nxt;
 844                 pkt->hdr.ctl = ACK;
 845                 uint32_t len = min(seqdiff(c->snd.last, c->snd.una), utcp->mss);
 846
 847                 if(fin_wanted(c, c->snd.una + len)) {
 848                         len--;
 849                         pkt->hdr.ctl |= FIN;
 850                 }
 851
 852                 buffer_copy(&c->sndbuf, pkt->data, 0, len);
 853                 print_packet(c, "rtrx", pkt, sizeof(pkt->hdr) + len);
 854                 utcp->send(utcp, pkt, sizeof(pkt->hdr) + len);
 855                 break;
 856
 857         default:
 858                 break;
 859         }
 860
 861         free(pkt);
 862 }
 863
 864 static void retransmit(struct utcp_connection *c) {
 865         if(c->state == CLOSED || c->snd.last == c->snd.una) {
 866                 debug(c, "retransmit() called but nothing to retransmit!\n");
 867                 stop_retransmit_timer(c);
 868                 return;
 869         }
 870
 871         struct utcp *utcp = c->utcp;
 872
 873         struct {
 874                 struct hdr hdr;
 875                 uint8_t data[];
 876         } *pkt = c->utcp->pkt;
 877
 878         pkt->hdr.src = c->src;
 879         pkt->hdr.dst = c->dst;
 880         pkt->hdr.wnd = c->rcvbuf.maxsize;
 881         pkt->hdr.aux = 0;
 882
 883         switch(c->state) {
 884         case SYN_SENT:
 885                 // Send our SYN again
 886                 pkt->hdr.seq = c->snd.iss;
 887                 pkt->hdr.ack = 0;
 888                 pkt->hdr.ctl = SYN;
 889                 pkt->hdr.aux = 0x0101;
 890                 pkt->data[0] = 1;
 891                 pkt->data[1] = 0;
 892                 pkt->data[2] = 0;
 893                 pkt->data[3] = c->flags & 0x7;
 894                 print_packet(c, "rtrx", pkt, sizeof(pkt->hdr) + 4);
 895                 utcp->send(utcp, pkt, sizeof(pkt->hdr) + 4);
 896                 break;
 897
 898         case SYN_RECEIVED:
 899                 // Send SYNACK again
 900                 pkt->hdr.seq = c->snd.nxt;
 901                 pkt->hdr.ack = c->rcv.nxt;
 902                 pkt->hdr.ctl = SYN | ACK;
 903                 print_packet(c, "rtrx", pkt, sizeof(pkt->hdr));
 904                 utcp->send(utcp, pkt, sizeof(pkt->hdr));
 905                 break;
 906
 907         case ESTABLISHED:
 908         case FIN_WAIT_1:
 909         case CLOSE_WAIT:
 910         case CLOSING:
 911         case LAST_ACK:
 912                 // Send unacked data again.
 913                 pkt->hdr.seq = c->snd.una;
 914                 pkt->hdr.ack = c->rcv.nxt;
 915                 pkt->hdr.ctl = ACK;
 916                 uint32_t len = min(seqdiff(c->snd.last, c->snd.una), utcp->mss);
 917
 918                 if(fin_wanted(c, c->snd.una + len)) {
 919                         len--;
 920                         pkt->hdr.ctl |= FIN;
 921                 }
 922
 923                 // RFC 5681 slow start after timeout
 924                 uint32_t flightsize = seqdiff(c->snd.nxt, c->snd.una);
 925                 c->snd.ssthresh = max(flightsize / 2, utcp->mss * 2); // eq. 4
 926                 c->snd.cwnd = utcp->mss;
 927                 debug_cwnd(c);
 928
 929                 buffer_copy(&c->sndbuf, pkt->data, 0, len);
 930                 print_packet(c, "rtrx", pkt, sizeof(pkt->hdr) + len);
 931                 utcp->send(utcp, pkt, sizeof(pkt->hdr) + len);
 932
 933                 c->snd.nxt = c->snd.una + len;
 934                 break;
 935
 936         case CLOSED:
 937         case LISTEN:
 938         case TIME_WAIT:
 939         case FIN_WAIT_2:
 940                 // We shouldn't need to retransmit anything in this state.
 941 #ifdef UTCP_DEBUG
 942                 abort();
 943 #endif
 944                 stop_retransmit_timer(c);
 945                 goto cleanup;
 946         }
 947
 948         start_retransmit_timer(c);
 949         utcp->rto *= 2;
 950
 951         if(utcp->rto > MAX_RTO) {
 952                 utcp->rto = MAX_RTO;
 953         }
 954
 955         c->rtt_start.tv_sec = 0; // invalidate RTT timer
 956         c->dupack = 0; // cancel any ongoing fast recovery
 957
 958 cleanup:
 959         return;
 960 }
 961
 962 /* Update receive buffer and SACK entries after consuming data.
 963  *
 964  * Situation:
 965  *
 966  * |.....0000..1111111111.....22222......3333|
 967  * |---------------^
 968  *
 969  * 0..3 represent the SACK entries. The ^ indicates up to which point we want
 970  * to remove data from the receive buffer. The idea is to substract "len"
 971  * from the offset of all the SACK entries, and then remove/cut down entries
 972  * that are shifted to before the start of the receive buffer.
 973  *
 974  * There are three cases:
 975  * - the SACK entry is after ^, in that case just change the offset.
 976  * - the SACK entry starts before and ends after ^, so we have to
 977  *   change both its offset and size.
 978  * - the SACK entry is completely before ^, in that case delete it.
 979  */
 980 static void sack_consume(struct utcp_connection *c, size_t len) {
 981         debug(c, "sack_consume %lu\n", (unsigned long)len);
 982
 983         if(len > c->rcvbuf.used) {
 984                 debug(c, "all SACK entries consumed\n");
 985                 c->sacks[0].len = 0;
 986                 return;
 987         }
 988
 989         buffer_discard(&c->rcvbuf, len);
 990
 991         for(int i = 0; i < NSACKS && c->sacks[i].len;) {
 992                 if(len < c->sacks[i].offset) {
 993                         c->sacks[i].offset -= len;
 994                         i++;
 995                 } else if(len < c->sacks[i].offset + c->sacks[i].len) {
 996                         c->sacks[i].len -= len - c->sacks[i].offset;
 997                         c->sacks[i].offset = 0;
 998                         i++;
 999                 } else {
1000                         if(i < NSACKS - 1) {
1001                                 memmove(&c->sacks[i], &c->sacks[i + 1], (NSACKS - 1 - i) * sizeof(c->sacks)[i]);
1002                                 c->sacks[NSACKS - 1].len = 0;
1003                         } else {
1004                                 c->sacks[i].len = 0;
1005                                 break;
1006                         }
1007                 }
1008         }
1009
1010         for(int i = 0; i < NSACKS && c->sacks[i].len; i++) {
1011                 debug(c, "SACK[%d] offset %u len %u\n", i, c->sacks[i].offset, c->sacks[i].len);
1012         }
1013 }
1014
1015 static void handle_out_of_order(struct utcp_connection *c, uint32_t offset, const void *data, size_t len) {
1016         debug(c, "out of order packet, offset %u\n", offset);
1017         // Packet loss or reordering occured. Store the data in the buffer.
1018         ssize_t rxd = buffer_put_at(&c->rcvbuf, offset, data, len);
1019
1020         if(rxd < 0 || (size_t)rxd < len) {
1021                 abort();
1022         }
1023
1024         // Make note of where we put it.
1025         for(int i = 0; i < NSACKS; i++) {
1026                 if(!c->sacks[i].len) { // nothing to merge, add new entry
1027                         debug(c, "new SACK entry %d\n", i);
1028                         c->sacks[i].offset = offset;
1029                         c->sacks[i].len = rxd;
1030                         break;
1031                 } else if(offset < c->sacks[i].offset) {
1032                         if(offset + rxd < c->sacks[i].offset) { // insert before
1033                                 if(!c->sacks[NSACKS - 1].len) { // only if room left
1034                                         debug(c, "insert SACK entry at %d\n", i);
1035                                         memmove(&c->sacks[i + 1], &c->sacks[i], (NSACKS - i - 1) * sizeof(c->sacks)[i]);
1036                                         c->sacks[i].offset = offset;
1037                                         c->sacks[i].len = rxd;
1038                                 } else {
1039                                         debug(c, "SACK entries full, dropping packet\n");
1040                                 }
1041
1042                                 break;
1043                         } else { // merge
1044                                 debug(c, "merge with start of SACK entry at %d\n", i);
1045                                 c->sacks[i].offset = offset;
1046                                 break;
1047                         }
1048                 } else if(offset <= c->sacks[i].offset + c->sacks[i].len) {
1049                         if(offset + rxd > c->sacks[i].offset + c->sacks[i].len) { // merge
1050                                 debug(c, "merge with end of SACK entry at %d\n", i);
1051                                 c->sacks[i].len = offset + rxd - c->sacks[i].offset;
1052                                 // TODO: handle potential merge with next entry
1053                         }
1054
1055                         break;
1056                 }
1057         }
1058
1059         for(int i = 0; i < NSACKS && c->sacks[i].len; i++) {
1060                 debug(c, "SACK[%d] offset %u len %u\n", i, c->sacks[i].offset, c->sacks[i].len);
1061         }
1062 }
1063
1064 static void handle_in_order(struct utcp_connection *c, const void *data, size_t len) {
1065         if(c->recv) {
1066                 ssize_t rxd = c->recv(c, data, len);
1067
1068                 if(rxd != (ssize_t)len) {
1069                         // TODO: handle the application not accepting all data.
1070                         abort();
1071                 }
1072         }
1073
1074         // Check if we can process out-of-order data now.
1075         if(c->sacks[0].len && len >= c->sacks[0].offset) {
1076                 debug(c, "incoming packet len %lu connected with SACK at %u\n", (unsigned long)len, c->sacks[0].offset);
1077
1078                 if(len < c->sacks[0].offset + c->sacks[0].len) {
1079                         size_t offset = len;
1080                         len = c->sacks[0].offset + c->sacks[0].len;
1081                         size_t remainder = len - offset;
1082                         ssize_t rxd = buffer_call(&c->rcvbuf, c->recv, c, offset, remainder);
1083
1084                         if(rxd != (ssize_t)remainder) {
1085                                 // TODO: handle the application not accepting all data.
1086                                 abort();
1087                         }
1088                 }
1089         }
1090
1091         if(c->rcvbuf.used) {
1092                 sack_consume(c, len);
1093         }
1094
1095         c->rcv.nxt += len;
1096 }
1097
1098
1099 static void handle_incoming_data(struct utcp_connection *c, uint32_t seq, const void *data, size_t len) {
1100         if(!is_reliable(c)) {
1101                 c->recv(c, data, len);
1102                 c->rcv.nxt = seq + len;
1103                 return;
1104         }
1105
1106         uint32_t offset = seqdiff(seq, c->rcv.nxt);
1107
1108         if(offset + len > c->rcvbuf.maxsize) {
1109                 abort();
1110         }
1111
1112         if(offset) {
1113                 handle_out_of_order(c, offset, data, len);
1114         } else {
1115                 handle_in_order(c, data, len);
1116         }
1117 }
1118
1119
1120 ssize_t utcp_recv(struct utcp *utcp, const void *data, size_t len) {
1121         const uint8_t *ptr = data;
1122
1123         if(!utcp) {
1124                 errno = EFAULT;
1125                 return -1;
1126         }
1127
1128         if(!len) {
1129                 return 0;
1130         }
1131
1132         if(!data) {
1133                 errno = EFAULT;
1134                 return -1;
1135         }
1136
1137         // Drop packets smaller than the header
1138
1139         struct hdr hdr;
1140
1141         if(len < sizeof(hdr)) {
1142                 print_packet(NULL, "recv", data, len);
1143                 errno = EBADMSG;
1144                 return -1;
1145         }
1146
1147         // Make a copy from the potentially unaligned data to a struct hdr
1148
1149         memcpy(&hdr, ptr, sizeof(hdr));
1150
1151         // Try to match the packet to an existing connection
1152
1153         struct utcp_connection *c = find_connection(utcp, hdr.dst, hdr.src);
1154         print_packet(c, "recv", data, len);
1155
1156         // Process the header
1157
1158         ptr += sizeof(hdr);
1159         len -= sizeof(hdr);
1160
1161         // Drop packets with an unknown CTL flag
1162
1163         if(hdr.ctl & ~(SYN | ACK | RST | FIN)) {
1164                 print_packet(NULL, "recv", data, len);
1165                 errno = EBADMSG;
1166                 return -1;
1167         }
1168
1169         // Check for auxiliary headers
1170
1171         const uint8_t *init = NULL;
1172
1173         uint16_t aux = hdr.aux;
1174
1175         while(aux) {
1176                 size_t auxlen = 4 * (aux >> 8) & 0xf;
1177                 uint8_t auxtype = aux & 0xff;
1178
1179                 if(len < auxlen) {
1180                         errno = EBADMSG;
1181                         return -1;
1182                 }
1183
1184                 switch(auxtype) {
1185                 case AUX_INIT:
1186                         if(!(hdr.ctl & SYN) || auxlen != 4) {
1187                                 errno = EBADMSG;
1188                                 return -1;
1189                         }
1190
1191                         init = ptr;
1192                         break;
1193
1194                 default:
1195                         errno = EBADMSG;
1196                         return -1;
1197                 }
1198
1199                 len -= auxlen;
1200                 ptr += auxlen;
1201
1202                 if(!(aux & 0x800)) {
1203                         break;
1204                 }
1205
1206                 if(len < 2) {
1207                         errno = EBADMSG;
1208                         return -1;
1209                 }
1210
1211                 memcpy(&aux, ptr, 2);
1212                 len -= 2;
1213                 ptr += 2;
1214         }
1215
1216         bool has_data = len || (hdr.ctl & (SYN | FIN));
1217
1218         // Is it for a new connection?
1219
1220         if(!c) {
1221                 // Ignore RST packets
1222
1223                 if(hdr.ctl & RST) {
1224                         return 0;
1225                 }
1226
1227                 // Is it a SYN packet and are we LISTENing?
1228
1229                 if(hdr.ctl & SYN && !(hdr.ctl & ACK) && utcp->accept) {
1230                         // If we don't want to accept it, send a RST back
1231                         if((utcp->pre_accept && !utcp->pre_accept(utcp, hdr.dst))) {
1232                                 len = 1;
1233                                 goto reset;
1234                         }
1235
1236                         // Try to allocate memory, otherwise send a RST back
1237                         c = allocate_connection(utcp, hdr.dst, hdr.src);
1238
1239                         if(!c) {
1240                                 len = 1;
1241                                 goto reset;
1242                         }
1243
1244                         // Parse auxilliary information
1245                         if(init) {
1246                                 if(init[0] < 1) {
1247                                         len = 1;
1248                                         goto reset;
1249                                 }
1250
1251                                 c->flags = init[3] & 0x7;
1252                         } else {
1253                                 c->flags = UTCP_TCP;
1254                         }
1255
1256 synack:
1257                         // Return SYN+ACK, go to SYN_RECEIVED state
1258                         c->snd.wnd = hdr.wnd;
1259                         c->rcv.irs = hdr.seq;
1260                         c->rcv.nxt = c->rcv.irs + 1;
1261                         set_state(c, SYN_RECEIVED);
1262
1263                         struct {
1264                                 struct hdr hdr;
1265                                 uint8_t data[4];
1266                         } pkt;
1267
1268                         pkt.hdr.src = c->src;
1269                         pkt.hdr.dst = c->dst;
1270                         pkt.hdr.ack = c->rcv.irs + 1;
1271                         pkt.hdr.seq = c->snd.iss;
1272                         pkt.hdr.wnd = c->rcvbuf.maxsize;
1273                         pkt.hdr.ctl = SYN | ACK;
1274
1275                         if(init) {
1276                                 pkt.hdr.aux = 0x0101;
1277                                 pkt.data[0] = 1;
1278                                 pkt.data[1] = 0;
1279                                 pkt.data[2] = 0;
1280                                 pkt.data[3] = c->flags & 0x7;
1281                                 print_packet(c, "send", &pkt, sizeof(hdr) + 4);
1282                                 utcp->send(utcp, &pkt, sizeof(hdr) + 4);
1283                         } else {
1284                                 pkt.hdr.aux = 0;
1285                                 print_packet(c, "send", &pkt, sizeof(hdr));
1286                                 utcp->send(utcp, &pkt, sizeof(hdr));
1287                         }
1288                 } else {
1289                         // No, we don't want your packets, send a RST back
1290                         len = 1;
1291                         goto reset;
1292                 }
1293
1294                 return 0;
1295         }
1296
1297         debug(c, "state %s\n", strstate[c->state]);
1298
1299         // In case this is for a CLOSED connection, ignore the packet.
1300         // TODO: make it so incoming packets can never match a CLOSED connection.
1301
1302         if(c->state == CLOSED) {
1303                 debug(c, "got packet for closed connection\n");
1304                 return 0;
1305         }
1306
1307         // It is for an existing connection.
1308
1309         // 1. Drop invalid packets.
1310
1311         // 1a. Drop packets that should not happen in our current state.
1312
1313         switch(c->state) {
1314         case SYN_SENT:
1315         case SYN_RECEIVED:
1316         case ESTABLISHED:
1317         case FIN_WAIT_1:
1318         case FIN_WAIT_2:
1319         case CLOSE_WAIT:
1320         case CLOSING:
1321         case LAST_ACK:
1322         case TIME_WAIT:
1323                 break;
1324
1325         default:
1326 #ifdef UTCP_DEBUG
1327                 abort();
1328 #endif
1329                 break;
1330         }
1331
1332         // 1b. Discard data that is not in our receive window.
1333
1334         if(is_reliable(c)) {
1335                 bool acceptable;
1336
1337                 if(c->state == SYN_SENT) {
1338                         acceptable = true;
1339                 } else if(len == 0) {
1340                         acceptable = seqdiff(hdr.seq, c->rcv.nxt) >= 0;
1341                 } else {
1342                         int32_t rcv_offset = seqdiff(hdr.seq, c->rcv.nxt);
1343
1344                         // cut already accepted front overlapping
1345                         if(rcv_offset < 0) {
1346                                 acceptable = len > (size_t) - rcv_offset;
1347
1348                                 if(acceptable) {
1349                                         ptr -= rcv_offset;
1350                                         len += rcv_offset;
1351                                         hdr.seq -= rcv_offset;
1352                                 }
1353                         } else {
1354                                 acceptable = seqdiff(hdr.seq, c->rcv.nxt) >= 0 && seqdiff(hdr.seq, c->rcv.nxt) + len <= c->rcvbuf.maxsize;
1355                         }
1356                 }
1357
1358                 if(!acceptable) {
1359                         debug(c, "packet not acceptable, %u <= %u + %lu < %u\n", c->rcv.nxt, hdr.seq, (unsigned long)len, c->rcv.nxt + c->rcvbuf.maxsize);
1360
1361                         // Ignore unacceptable RST packets.
1362                         if(hdr.ctl & RST) {
1363                                 return 0;
1364                         }
1365
1366                         // Otherwise, continue processing.
1367                         len = 0;
1368                 }
1369         } else {
1370 #if UTCP_DEBUG
1371                 int32_t rcv_offset = seqdiff(hdr.seq, c->rcv.nxt);
1372
1373                 if(rcv_offset) {
1374                         debug(c, "packet out of order, offset %u bytes", rcv_offset);
1375                 }
1376
1377                 if(rcv_offset >= 0) {
1378                         c->rcv.nxt = hdr.seq + len;
1379                 }
1380
1381 #endif
1382         }
1383
1384         c->snd.wnd = hdr.wnd; // TODO: move below
1385
1386         // 1c. Drop packets with an invalid ACK.
1387         // ackno should not roll back, and it should also not be bigger than what we ever could have sent
1388         // (= snd.una + c->sndbuf.used).
1389
1390         if(!is_reliable(c)) {
1391                 if(hdr.ack != c->snd.last && c->state >= ESTABLISHED) {
1392                         hdr.ack = c->snd.una;
1393                 }
1394         }
1395
1396         if(hdr.ctl & ACK && (seqdiff(hdr.ack, c->snd.last) > 0 || seqdiff(hdr.ack, c->snd.una) < 0)) {
1397                 debug(c, "packet ack seqno out of range, %u <= %u < %u\n", c->snd.una, hdr.ack, c->snd.una + c->sndbuf.used);
1398
1399                 // Ignore unacceptable RST packets.
1400                 if(hdr.ctl & RST) {
1401                         return 0;
1402                 }
1403
1404                 goto reset;
1405         }
1406
1407         // 2. Handle RST packets
1408
1409         if(hdr.ctl & RST) {
1410                 switch(c->state) {
1411                 case SYN_SENT:
1412                         if(!(hdr.ctl & ACK)) {
1413                                 return 0;
1414                         }
1415
1416                         // The peer has refused our connection.
1417                         set_state(c, CLOSED);
1418                         errno = ECONNREFUSED;
1419
1420                         if(c->recv) {
1421                                 c->recv(c, NULL, 0);
1422                         }
1423
1424                         if(c->poll && !c->reapable) {
1425                                 c->poll(c, 0);
1426                         }
1427
1428                         return 0;
1429
1430                 case SYN_RECEIVED:
1431                         if(hdr.ctl & ACK) {
1432                                 return 0;
1433                         }
1434
1435                         // We haven't told the application about this connection yet. Silently delete.
1436                         free_connection(c);
1437                         return 0;
1438
1439                 case ESTABLISHED:
1440                 case FIN_WAIT_1:
1441                 case FIN_WAIT_2:
1442                 case CLOSE_WAIT:
1443                         if(hdr.ctl & ACK) {
1444                                 return 0;
1445                         }
1446
1447                         // The peer has aborted our connection.
1448                         set_state(c, CLOSED);
1449                         errno = ECONNRESET;
1450
1451                         if(c->recv) {
1452                                 c->recv(c, NULL, 0);
1453                         }
1454
1455                         if(c->poll && !c->reapable) {
1456                                 c->poll(c, 0);
1457                         }
1458
1459                         return 0;
1460
1461                 case CLOSING:
1462                 case LAST_ACK:
1463                 case TIME_WAIT:
1464                         if(hdr.ctl & ACK) {
1465                                 return 0;
1466                         }
1467
1468                         // As far as the application is concerned, the connection has already been closed.
1469                         // If it has called utcp_close() already, we can immediately free this connection.
1470                         if(c->reapable) {
1471                                 free_connection(c);
1472                                 return 0;
1473                         }
1474
1475                         // Otherwise, immediately move to the CLOSED state.
1476                         set_state(c, CLOSED);
1477                         return 0;
1478
1479                 default:
1480 #ifdef UTCP_DEBUG
1481                         abort();
1482 #endif
1483                         break;
1484                 }
1485         }
1486
1487         uint32_t advanced;
1488
1489         if(!(hdr.ctl & ACK)) {
1490                 advanced = 0;
1491                 goto skip_ack;
1492         }
1493
1494         // 3. Advance snd.una
1495
1496         advanced = seqdiff(hdr.ack, c->snd.una);
1497
1498         if(advanced) {
1499                 // RTT measurement
1500                 if(c->rtt_start.tv_sec) {
1501                         if(c->rtt_seq == hdr.ack) {
1502                                 struct timespec now;
1503                                 clock_gettime(UTCP_CLOCK, &now);
1504                                 int32_t diff = timespec_diff_usec(&now, &c->rtt_start);
1505                                 update_rtt(c, diff);
1506                                 c->rtt_start.tv_sec = 0;
1507                         } else if(c->rtt_seq < hdr.ack) {
1508                                 debug(c, "cancelling RTT measurement: %u < %u\n", c->rtt_seq, hdr.ack);
1509                                 c->rtt_start.tv_sec = 0;
1510                         }
1511                 }
1512
1513                 int32_t data_acked = advanced;
1514
1515                 switch(c->state) {
1516                 case SYN_SENT:
1517                 case SYN_RECEIVED:
1518                         data_acked--;
1519                         break;
1520
1521                 // TODO: handle FIN as well.
1522                 default:
1523                         break;
1524                 }
1525
1526                 assert(data_acked >= 0);
1527
1528 #ifndef NDEBUG
1529                 int32_t bufused = seqdiff(c->snd.last, c->snd.una);
1530                 assert(data_acked <= bufused);
1531 #endif
1532
1533                 if(data_acked) {
1534                         buffer_discard(&c->sndbuf, data_acked);
1535                         c->do_poll = true;
1536                 }
1537
1538                 // Also advance snd.nxt if possible
1539                 if(seqdiff(c->snd.nxt, hdr.ack) < 0) {
1540                         c->snd.nxt = hdr.ack;
1541                 }
1542
1543                 c->snd.una = hdr.ack;
1544
1545                 if(c->dupack) {
1546                         if(c->dupack >= 3) {
1547                                 debug(c, "fast recovery ended\n");
1548                                 c->snd.cwnd = c->snd.ssthresh;
1549                         }
1550
1551                         c->dupack = 0;
1552                 }
1553
1554                 // Increase the congestion window according to RFC 5681
1555                 if(c->snd.cwnd < c->snd.ssthresh) {
1556                         c->snd.cwnd += min(advanced, utcp->mss); // eq. 2
1557                 } else {
1558                         c->snd.cwnd += max(1, (utcp->mss * utcp->mss) / c->snd.cwnd); // eq. 3
1559                 }
1560
1561                 if(c->snd.cwnd > c->sndbuf.maxsize) {
1562                         c->snd.cwnd = c->sndbuf.maxsize;
1563                 }
1564
1565                 debug_cwnd(c);
1566
1567                 // Check if we have sent a FIN that is now ACKed.
1568                 switch(c->state) {
1569                 case FIN_WAIT_1:
1570                         if(c->snd.una == c->snd.last) {
1571                                 set_state(c, FIN_WAIT_2);
1572                         }
1573
1574                         break;
1575
1576                 case CLOSING:
1577                         if(c->snd.una == c->snd.last) {
1578                                 clock_gettime(UTCP_CLOCK, &c->conn_timeout);
1579                                 c->conn_timeout.tv_sec += utcp->timeout;
1580                                 set_state(c, TIME_WAIT);
1581                         }
1582
1583                         break;
1584
1585                 default:
1586                         break;
1587                 }
1588         } else {
1589                 if(!len && is_reliable(c) && c->snd.una != c->snd.last) {
1590                         c->dupack++;
1591                         debug(c, "duplicate ACK %d\n", c->dupack);
1592
1593                         if(c->dupack == 3) {
1594                                 // RFC 5681 fast recovery
1595                                 debug(c, "fast recovery started\n", c->dupack);
1596                                 uint32_t flightsize = seqdiff(c->snd.nxt, c->snd.una);
1597                                 c->snd.ssthresh = max(flightsize / 2, utcp->mss * 2); // eq. 4
1598                                 c->snd.cwnd = min(c->snd.ssthresh + 3 * utcp->mss, c->sndbuf.maxsize);
1599
1600                                 if(c->snd.cwnd > c->sndbuf.maxsize) {
1601                                         c->snd.cwnd = c->sndbuf.maxsize;
1602                                 }
1603
1604                                 debug_cwnd(c);
1605
1606                                 fast_retransmit(c);
1607                         } else if(c->dupack > 3) {
1608                                 c->snd.cwnd += utcp->mss;
1609
1610                                 if(c->snd.cwnd > c->sndbuf.maxsize) {
1611                                         c->snd.cwnd = c->sndbuf.maxsize;
1612                                 }
1613
1614                                 debug_cwnd(c);
1615                         }
1616
1617                         // We got an ACK which indicates the other side did get one of our packets.
1618                         // Reset the retransmission timer to avoid going to slow start,
1619                         // but don't touch the connection timeout.
1620                         start_retransmit_timer(c);
1621                 }
1622         }
1623
1624         // 4. Update timers
1625
1626         if(advanced) {
1627                 if(c->snd.una == c->snd.last) {
1628                         stop_retransmit_timer(c);
1629                         timespec_clear(&c->conn_timeout);
1630                 } else if(is_reliable(c)) {
1631                         start_retransmit_timer(c);
1632                         clock_gettime(UTCP_CLOCK, &c->conn_timeout);
1633                         c->conn_timeout.tv_sec += utcp->timeout;
1634                 }
1635         }
1636
1637 skip_ack:
1638         // 5. Process SYN stuff
1639
1640         if(hdr.ctl & SYN) {
1641                 switch(c->state) {
1642                 case SYN_SENT:
1643
1644                         // This is a SYNACK. It should always have ACKed the SYN.
1645                         if(!advanced) {
1646                                 goto reset;
1647                         }
1648
1649                         c->rcv.irs = hdr.seq;
1650                         c->rcv.nxt = hdr.seq;
1651
1652                         if(c->shut_wr) {
1653                                 c->snd.last++;
1654                                 set_state(c, FIN_WAIT_1);
1655                         } else {
1656                                 set_state(c, ESTABLISHED);
1657                         }
1658
1659                         // TODO: notify application of this somehow.
1660                         break;
1661
1662                 case SYN_RECEIVED:
1663                         // This is a retransmit of a SYN, send back the SYNACK.
1664                         goto synack;
1665
1666                 case ESTABLISHED:
1667                 case FIN_WAIT_1:
1668                 case FIN_WAIT_2:
1669                 case CLOSE_WAIT:
1670                 case CLOSING:
1671                 case LAST_ACK:
1672                 case TIME_WAIT:
1673                         // Ehm, no. We should never receive a second SYN.
1674                         return 0;
1675
1676                 default:
1677 #ifdef UTCP_DEBUG
1678                         abort();
1679 #endif
1680                         return 0;
1681                 }
1682
1683                 // SYN counts as one sequence number
1684                 c->rcv.nxt++;
1685         }
1686
1687         // 6. Process new data
1688
1689         if(c->state == SYN_RECEIVED) {
1690                 // This is the ACK after the SYNACK. It should always have ACKed the SYNACK.
1691                 if(!advanced) {
1692                         goto reset;
1693                 }
1694
1695                 // Are we still LISTENing?
1696                 if(utcp->accept) {
1697                         utcp->accept(c, c->src);
1698                 }
1699
1700                 if(c->state != ESTABLISHED) {
1701                         set_state(c, CLOSED);
1702                         c->reapable = true;
1703                         goto reset;
1704                 }
1705         }
1706
1707         if(len) {
1708                 switch(c->state) {
1709                 case SYN_SENT:
1710                 case SYN_RECEIVED:
1711                         // This should never happen.
1712 #ifdef UTCP_DEBUG
1713                         abort();
1714 #endif
1715                         return 0;
1716
1717                 case ESTABLISHED:
1718                 case FIN_WAIT_1:
1719                 case FIN_WAIT_2:
1720                         break;
1721
1722                 case CLOSE_WAIT:
1723                 case CLOSING:
1724                 case LAST_ACK:
1725                 case TIME_WAIT:
1726                         // Ehm no, We should never receive more data after a FIN.
1727                         goto reset;
1728
1729                 default:
1730 #ifdef UTCP_DEBUG
1731                         abort();
1732 #endif
1733                         return 0;
1734                 }
1735
1736                 handle_incoming_data(c, hdr.seq, ptr, len);
1737         }
1738
1739         // 7. Process FIN stuff
1740
1741         if((hdr.ctl & FIN) && (!is_reliable(c) || hdr.seq + len == c->rcv.nxt)) {
1742                 switch(c->state) {
1743                 case SYN_SENT:
1744                 case SYN_RECEIVED:
1745                         // This should never happen.
1746 #ifdef UTCP_DEBUG
1747                         abort();
1748 #endif
1749                         break;
1750
1751                 case ESTABLISHED:
1752                         set_state(c, CLOSE_WAIT);
1753                         break;
1754
1755                 case FIN_WAIT_1:
1756                         set_state(c, CLOSING);
1757                         break;
1758
1759                 case FIN_WAIT_2:
1760                         clock_gettime(UTCP_CLOCK, &c->conn_timeout);
1761                         c->conn_timeout.tv_sec += utcp->timeout;
1762                         set_state(c, TIME_WAIT);
1763                         break;
1764
1765                 case CLOSE_WAIT:
1766                 case CLOSING:
1767                 case LAST_ACK:
1768                 case TIME_WAIT:
1769                         // Ehm, no. We should never receive a second FIN.
1770                         goto reset;
1771
1772                 default:
1773 #ifdef UTCP_DEBUG
1774                         abort();
1775 #endif
1776                         break;
1777                 }
1778
1779                 // FIN counts as one sequence number
1780                 c->rcv.nxt++;
1781                 len++;
1782
1783                 // Inform the application that the peer closed its end of the connection.
1784                 if(c->recv) {
1785                         errno = 0;
1786                         c->recv(c, NULL, 0);
1787                 }
1788         }
1789
1790         // Now we send something back if:
1791         // - we received data, so we have to send back an ACK
1792         //   -> sendatleastone = true
1793         // - or we got an ack, so we should maybe send a bit more data
1794         //   -> sendatleastone = false
1795
1796         if(is_reliable(c) || hdr.ctl & SYN || hdr.ctl & FIN) {
1797                 ack(c, has_data);
1798         }
1799
1800         return 0;
1801
1802 reset:
1803         swap_ports(&hdr);
1804         hdr.wnd = 0;
1805         hdr.aux = 0;
1806
1807         if(hdr.ctl & ACK) {
1808                 hdr.seq = hdr.ack;
1809                 hdr.ctl = RST;
1810         } else {
1811                 hdr.ack = hdr.seq + len;
1812                 hdr.seq = 0;
1813                 hdr.ctl = RST | ACK;
1814         }
1815
1816         print_packet(c, "send", &hdr, sizeof(hdr));
1817         utcp->send(utcp, &hdr, sizeof(hdr));
1818         return 0;
1819
1820 }
1821
1822 int utcp_shutdown(struct utcp_connection *c, int dir) {
1823         debug(c, "shutdown %d at %u\n", dir, c ? c->snd.last : 0);
1824
1825         if(!c) {
1826                 errno = EFAULT;
1827                 return -1;
1828         }
1829
1830         if(c->reapable) {
1831                 debug(c, "shutdown() called on closed connection\n");
1832                 errno = EBADF;
1833                 return -1;
1834         }
1835
1836         if(!(dir == UTCP_SHUT_RD || dir == UTCP_SHUT_WR || dir == UTCP_SHUT_RDWR)) {
1837                 errno = EINVAL;
1838                 return -1;
1839         }
1840
1841         // TCP does not have a provision for stopping incoming packets.
1842         // The best we can do is to just ignore them.
1843         if(dir == UTCP_SHUT_RD || dir == UTCP_SHUT_RDWR) {
1844                 c->recv = NULL;
1845         }
1846
1847         // The rest of the code deals with shutting down writes.
1848         if(dir == UTCP_SHUT_RD) {
1849                 return 0;
1850         }
1851
1852         // Only process shutting down writes once.
1853         if(c->shut_wr) {
1854                 return 0;
1855         }
1856
1857         c->shut_wr = true;
1858
1859         switch(c->state) {
1860         case CLOSED:
1861         case LISTEN:
1862                 errno = ENOTCONN;
1863                 return -1;
1864
1865         case SYN_SENT:
1866                 return 0;
1867
1868         case SYN_RECEIVED:
1869         case ESTABLISHED:
1870                 set_state(c, FIN_WAIT_1);
1871                 break;
1872
1873         case FIN_WAIT_1:
1874         case FIN_WAIT_2:
1875                 return 0;
1876
1877         case CLOSE_WAIT:
1878                 set_state(c, CLOSING);
1879                 break;
1880
1881         case CLOSING:
1882         case LAST_ACK:
1883         case TIME_WAIT:
1884                 return 0;
1885         }
1886
1887         c->snd.last++;
1888
1889         ack(c, false);
1890
1891         if(!timespec_isset(&c->rtrx_timeout)) {
1892                 start_retransmit_timer(c);
1893         }
1894
1895         return 0;
1896 }
1897
1898 static bool reset_connection(struct utcp_connection *c) {
1899         if(!c) {
1900                 errno = EFAULT;
1901                 return false;
1902         }
1903
1904         if(c->reapable) {
1905                 debug(c, "abort() called on closed connection\n");
1906                 errno = EBADF;
1907                 return false;
1908         }
1909
1910         c->recv = NULL;
1911         c->poll = NULL;
1912
1913         switch(c->state) {
1914         case CLOSED:
1915                 return true;
1916
1917         case LISTEN:
1918         case SYN_SENT:
1919         case CLOSING:
1920         case LAST_ACK:
1921         case TIME_WAIT:
1922                 set_state(c, CLOSED);
1923                 return true;
1924
1925         case SYN_RECEIVED:
1926         case ESTABLISHED:
1927         case FIN_WAIT_1:
1928         case FIN_WAIT_2:
1929         case CLOSE_WAIT:
1930                 set_state(c, CLOSED);
1931                 break;
1932         }
1933
1934         // Send RST
1935
1936         struct hdr hdr;
1937
1938         hdr.src = c->src;
1939         hdr.dst = c->dst;
1940         hdr.seq = c->snd.nxt;
1941         hdr.ack = 0;
1942         hdr.wnd = 0;
1943         hdr.ctl = RST;
1944
1945         print_packet(c, "send", &hdr, sizeof(hdr));
1946         c->utcp->send(c->utcp, &hdr, sizeof(hdr));
1947         return true;
1948 }
1949
1950 // Closes all the opened connections
1951 void utcp_abort_all_connections(struct utcp *utcp) {
1952         if(!utcp) {
1953                 errno = EINVAL;
1954                 return;
1955         }
1956
1957         for(int i = 0; i < utcp->nconnections; i++) {
1958                 struct utcp_connection *c = utcp->connections[i];
1959
1960                 if(c->reapable || c->state == CLOSED) {
1961                         continue;
1962                 }
1963
1964                 utcp_recv_t old_recv = c->recv;
1965                 utcp_poll_t old_poll = c->poll;
1966
1967                 reset_connection(c);
1968
1969                 if(old_recv) {
1970                         errno = 0;
1971                         old_recv(c, NULL, 0);
1972                 }
1973
1974                 if(old_poll && !c->reapable) {
1975                         errno = 0;
1976                         old_poll(c, 0);
1977                 }
1978         }
1979
1980         return;
1981 }
1982
1983 int utcp_close(struct utcp_connection *c) {
1984         if(utcp_shutdown(c, SHUT_RDWR) && errno != ENOTCONN) {
1985                 return -1;
1986         }
1987
1988         c->recv = NULL;
1989         c->poll = NULL;
1990         c->reapable = true;
1991         return 0;
1992 }
1993
1994 int utcp_abort(struct utcp_connection *c) {
1995         if(!reset_connection(c)) {
1996                 return -1;
1997         }
1998
1999         c->reapable = true;
2000         return 0;
2001 }
2002
2003 /* Handle timeouts.
2004  * One call to this function will loop through all connections,
2005  * checking if something needs to be resent or not.
2006  * The return value is the time to the next timeout in milliseconds,
2007  * or maybe a negative value if the timeout is infinite.
2008  */
2009 struct timespec utcp_timeout(struct utcp *utcp) {
2010         struct timespec now;
2011         clock_gettime(UTCP_CLOCK, &now);
2012         struct timespec next = {now.tv_sec + 3600, now.tv_nsec};
2013
2014         for(int i = 0; i < utcp->nconnections; i++) {
2015                 struct utcp_connection *c = utcp->connections[i];
2016
2017                 if(!c) {
2018                         continue;
2019                 }
2020
2021                 // delete connections that have been utcp_close()d.
2022                 if(c->state == CLOSED) {
2023                         if(c->reapable) {
2024                                 debug(c, "reaping\n");
2025                                 free_connection(c);
2026                                 i--;
2027                         }
2028
2029                         continue;
2030                 }
2031
2032                 if(timespec_isset(&c->conn_timeout) && timespec_lt(&c->conn_timeout, &now)) {
2033                         errno = ETIMEDOUT;
2034                         c->state = CLOSED;
2035
2036                         if(c->recv) {
2037                                 c->recv(c, NULL, 0);
2038                         }
2039
2040                         if(c->poll && !c->reapable) {
2041                                 c->poll(c, 0);
2042                         }
2043
2044                         continue;
2045                 }
2046
2047                 if(timespec_isset(&c->rtrx_timeout) && timespec_lt(&c->rtrx_timeout, &now)) {
2048                         debug(c, "retransmitting after timeout\n");
2049                         retransmit(c);
2050                 }
2051
2052                 if(c->poll) {
2053                         if((c->state == ESTABLISHED || c->state == CLOSE_WAIT) && c->do_poll) {
2054                                 c->do_poll = false;
2055                                 uint32_t len = buffer_free(&c->sndbuf);
2056
2057                                 if(len) {
2058                                         c->poll(c, len);
2059                                 }
2060                         } else if(c->state == CLOSED) {
2061                                 c->poll(c, 0);
2062                         }
2063                 }
2064
2065                 if(timespec_isset(&c->conn_timeout) && timespec_lt(&c->conn_timeout, &next)) {
2066                         next = c->conn_timeout;
2067                 }
2068
2069                 if(timespec_isset(&c->rtrx_timeout) && timespec_lt(&c->rtrx_timeout, &next)) {
2070                         next = c->rtrx_timeout;
2071                 }
2072         }
2073
2074         struct timespec diff;
2075
2076         timespec_sub(&next, &now, &diff);
2077
2078         return diff;
2079 }
2080
2081 bool utcp_is_active(struct utcp *utcp) {
2082         if(!utcp) {
2083                 return false;
2084         }
2085
2086         for(int i = 0; i < utcp->nconnections; i++)
2087                 if(utcp->connections[i]->state != CLOSED && utcp->connections[i]->state != TIME_WAIT) {
2088                         return true;
2089                 }
2090
2091         return false;
2092 }
2093
2094 struct utcp *utcp_init(utcp_accept_t accept, utcp_pre_accept_t pre_accept, utcp_send_t send, void *priv) {
2095         if(!send) {
2096                 errno = EFAULT;
2097                 return NULL;
2098         }
2099
2100         struct utcp *utcp = calloc(1, sizeof(*utcp));
2101
2102         if(!utcp) {
2103                 return NULL;
2104         }
2105
2106         if(!CLOCK_GRANULARITY) {
2107                 struct timespec res;
2108                 clock_getres(UTCP_CLOCK, &res);
2109                 CLOCK_GRANULARITY = res.tv_sec * USEC_PER_SEC + res.tv_nsec / 1000;
2110         }
2111
2112         utcp->accept = accept;
2113         utcp->pre_accept = pre_accept;
2114         utcp->send = send;
2115         utcp->priv = priv;
2116         utcp_set_mtu(utcp, DEFAULT_MTU);
2117         utcp->timeout = DEFAULT_USER_TIMEOUT; // sec
2118         utcp->rto = START_RTO; // usec
2119
2120         return utcp;
2121 }
2122
2123 void utcp_exit(struct utcp *utcp) {
2124         if(!utcp) {
2125                 return;
2126         }
2127
2128         for(int i = 0; i < utcp->nconnections; i++) {
2129                 struct utcp_connection *c = utcp->connections[i];
2130
2131                 if(!c->reapable) {
2132                         if(c->recv) {
2133                                 c->recv(c, NULL, 0);
2134                         }
2135
2136                         if(c->poll && !c->reapable) {
2137                                 c->poll(c, 0);
2138                         }
2139                 }
2140
2141                 buffer_exit(&c->rcvbuf);
2142                 buffer_exit(&c->sndbuf);
2143                 free(c);
2144         }
2145
2146         free(utcp->connections);
2147         free(utcp->pkt);
2148         free(utcp);
2149 }
2150
2151 uint16_t utcp_get_mtu(struct utcp *utcp) {
2152         return utcp ? utcp->mtu : 0;
2153 }
2154
2155 uint16_t utcp_get_mss(struct utcp *utcp) {
2156         return utcp ? utcp->mss : 0;
2157 }
2158
2159 void utcp_set_mtu(struct utcp *utcp, uint16_t mtu) {
2160         if(!utcp) {
2161                 return;
2162         }
2163
2164         if(mtu <= sizeof(struct hdr)) {
2165                 return;
2166         }
2167
2168         if(mtu > utcp->mtu) {
2169                 char *new = realloc(utcp->pkt, mtu + sizeof(struct hdr));
2170
2171                 if(!new) {
2172                         return;
2173                 }
2174
2175                 utcp->pkt = new;
2176         }
2177
2178         utcp->mtu = mtu;
2179         utcp->mss = mtu - sizeof(struct hdr);
2180 }
2181
2182 void utcp_reset_timers(struct utcp *utcp) {
2183         if(!utcp) {
2184                 return;
2185         }
2186
2187         struct timespec now, then;
2188
2189         clock_gettime(UTCP_CLOCK, &now);
2190
2191         then = now;
2192
2193         then.tv_sec += utcp->timeout;
2194
2195         for(int i = 0; i < utcp->nconnections; i++) {
2196                 struct utcp_connection *c = utcp->connections[i];
2197
2198                 if(c->reapable) {
2199                         continue;
2200                 }
2201
2202                 if(timespec_isset(&c->rtrx_timeout)) {
2203                         c->rtrx_timeout = now;
2204                 }
2205
2206                 if(timespec_isset(&c->conn_timeout)) {
2207                         c->conn_timeout = then;
2208                 }
2209
2210                 c->rtt_start.tv_sec = 0;
2211         }
2212
2213         if(utcp->rto > START_RTO) {
2214                 utcp->rto = START_RTO;
2215         }
2216 }
2217
2218 int utcp_get_user_timeout(struct utcp *u) {
2219         return u ? u->timeout : 0;
2220 }
2221
2222 void utcp_set_user_timeout(struct utcp *u, int timeout) {
2223         if(u) {
2224                 u->timeout = timeout;
2225         }
2226 }
2227
2228 size_t utcp_get_sndbuf(struct utcp_connection *c) {
2229         return c ? c->sndbuf.maxsize : 0;
2230 }
2231
2232 size_t utcp_get_sndbuf_free(struct utcp_connection *c) {
2233         if(!c) {
2234                 return 0;
2235         }
2236
2237         switch(c->state) {
2238         case SYN_SENT:
2239         case SYN_RECEIVED:
2240         case ESTABLISHED:
2241         case CLOSE_WAIT:
2242                 return buffer_free(&c->sndbuf);
2243
2244         default:
2245                 return 0;
2246         }
2247 }
2248
2249 void utcp_set_sndbuf(struct utcp_connection *c, size_t size) {
2250         if(!c) {
2251                 return;
2252         }
2253
2254         c->sndbuf.maxsize = size;
2255
2256         if(c->sndbuf.maxsize != size) {
2257                 c->sndbuf.maxsize = -1;
2258         }
2259
2260         c->do_poll = buffer_free(&c->sndbuf);
2261 }
2262
2263 size_t utcp_get_rcvbuf(struct utcp_connection *c) {
2264         return c ? c->rcvbuf.maxsize : 0;
2265 }
2266
2267 size_t utcp_get_rcvbuf_free(struct utcp_connection *c) {
2268         if(c && (c->state == ESTABLISHED || c->state == CLOSE_WAIT)) {
2269                 return buffer_free(&c->rcvbuf);
2270         } else {
2271                 return 0;
2272         }
2273 }
2274
2275 void utcp_set_rcvbuf(struct utcp_connection *c, size_t size) {
2276         if(!c) {
2277                 return;
2278         }
2279
2280         c->rcvbuf.maxsize = size;
2281
2282         if(c->rcvbuf.maxsize != size) {
2283                 c->rcvbuf.maxsize = -1;
2284         }
2285 }
2286
2287 size_t utcp_get_sendq(struct utcp_connection *c) {
2288         return c->sndbuf.used;
2289 }
2290
2291 size_t utcp_get_recvq(struct utcp_connection *c) {
2292         return c->rcvbuf.used;
2293 }
2294
2295 bool utcp_get_nodelay(struct utcp_connection *c) {
2296         return c ? c->nodelay : false;
2297 }
2298
2299 void utcp_set_nodelay(struct utcp_connection *c, bool nodelay) {
2300         if(c) {
2301                 c->nodelay = nodelay;
2302         }
2303 }
2304
2305 bool utcp_get_keepalive(struct utcp_connection *c) {
2306         return c ? c->keepalive : false;
2307 }
2308
2309 void utcp_set_keepalive(struct utcp_connection *c, bool keepalive) {
2310         if(c) {
2311                 c->keepalive = keepalive;
2312         }
2313 }
2314
2315 size_t utcp_get_outq(struct utcp_connection *c) {
2316         return c ? seqdiff(c->snd.nxt, c->snd.una) : 0;
2317 }
2318
2319 void utcp_set_recv_cb(struct utcp_connection *c, utcp_recv_t recv) {
2320         if(c) {
2321                 c->recv = recv;
2322         }
2323 }
2324
2325 void utcp_set_poll_cb(struct utcp_connection *c, utcp_poll_t poll) {
2326         if(c) {
2327                 c->poll = poll;
2328                 c->do_poll = buffer_free(&c->sndbuf);
2329         }
2330 }
2331
2332 void utcp_set_accept_cb(struct utcp *utcp, utcp_accept_t accept, utcp_pre_accept_t pre_accept) {
2333         if(utcp) {
2334                 utcp->accept = accept;
2335                 utcp->pre_accept = pre_accept;
2336         }
2337 }
2338
2339 void utcp_expect_data(struct utcp_connection *c, bool expect) {
2340         if(!c || c->reapable) {
2341                 return;
2342         }
2343
2344         if(!(c->state == ESTABLISHED || c->state == FIN_WAIT_1 || c->state == FIN_WAIT_2)) {
2345                 return;
2346         }
2347
2348         if(expect) {
2349                 // If we expect data, start the connection timer.
2350                 if(!timespec_isset(&c->conn_timeout)) {
2351                         clock_gettime(UTCP_CLOCK, &c->conn_timeout);
2352                         c->conn_timeout.tv_sec += c->utcp->timeout;
2353                 }
2354         } else {
2355                 // If we want to cancel expecting data, only clear the timer when there is no unACKed data.
2356                 if(c->snd.una == c->snd.last) {
2357                         timespec_clear(&c->conn_timeout);
2358                 }
2359         }
2360 }
2361
2362 void utcp_offline(struct utcp *utcp, bool offline) {
2363         struct timespec now;
2364         clock_gettime(UTCP_CLOCK, &now);
2365
2366         for(int i = 0; i < utcp->nconnections; i++) {
2367                 struct utcp_connection *c = utcp->connections[i];
2368
2369                 if(c->reapable) {
2370                         continue;
2371                 }
2372
2373                 utcp_expect_data(c, offline);
2374
2375                 if(!offline) {
2376                         if(timespec_isset(&c->rtrx_timeout)) {
2377                                 c->rtrx_timeout = now;
2378                         }
2379
2380                         utcp->connections[i]->rtt_start.tv_sec = 0;
2381                 }
2382         }
2383
2384         if(!offline && utcp->rto > START_RTO) {
2385                 utcp->rto = START_RTO;
2386         }
2387 }
2388
2389 void utcp_set_clock_granularity(long granularity) {
2390         CLOCK_GRANULARITY = granularity;
2391 }