upload-pack.c

   1 #include "cache.h"
   2 #include "refs.h"
   3 #include "pkt-line.h"
   4 #include "sideband.h"
   5 #include "tag.h"
   6 #include "object.h"
   7 #include "commit.h"
   8 #include "exec_cmd.h"
   9 #include "diff.h"
  10 #include "revision.h"
  11 #include "list-objects.h"
  12 #include "run-command.h"
  13
  14 static const char upload_pack_usage[] = "git-upload-pack [--strict] [--timeout=nn] <dir>";
  15
  16 /* bits #0..7 in revision.h, #8..10 in commit.c */
  17 #define THEY_HAVE       (1u << 11)
  18 #define OUR_REF         (1u << 12)
  19 #define WANTED          (1u << 13)
  20 #define COMMON_KNOWN    (1u << 14)
  21 #define REACHABLE       (1u << 15)
  22
  23 #define SHALLOW         (1u << 16)
  24 #define NOT_SHALLOW     (1u << 17)
  25 #define CLIENT_SHALLOW  (1u << 18)
  26
  27 static unsigned long oldest_have;
  28
  29 static int multi_ack, nr_our_refs;
  30 static int use_thin_pack, use_ofs_delta, no_progress;
  31 static struct object_array have_obj;
  32 static struct object_array want_obj;
  33 static unsigned int timeout;
  34 /* 0 for no sideband,
  35  * otherwise maximum packet size (up to 65520 bytes).
  36  */
  37 static int use_sideband;
  38
  39 static void reset_timeout(void)
  40 {
  41         alarm(timeout);
  42 }
  43
  44 static int strip(char *line, int len)
  45 {
  46         if (len && line[len-1] == '\n')
  47                 line[--len] = 0;
  48         return len;
  49 }
  50
  51 static ssize_t send_client_data(int fd, const char *data, ssize_t sz)
  52 {
  53         if (use_sideband)
  54                 return send_sideband(1, fd, data, sz, use_sideband);
  55         if (fd == 3)
  56                 /* emergency quit */
  57                 fd = 2;
  58         if (fd == 2) {
  59                 /* XXX: are we happy to lose stuff here? */
  60                 xwrite(fd, data, sz);
  61                 return sz;
  62         }
  63         return safe_write(fd, data, sz);
  64 }
  65
  66 static FILE *pack_pipe = NULL;
  67 static void show_commit(struct commit *commit)
  68 {
  69         if (commit->object.flags & BOUNDARY)
  70                 fputc('-', pack_pipe);
  71         if (fputs(sha1_to_hex(commit->object.sha1), pack_pipe) < 0)
  72                 die("broken output pipe");
  73         fputc('\n', pack_pipe);
  74         fflush(pack_pipe);
  75         free(commit->buffer);
  76         commit->buffer = NULL;
  77 }
  78
  79 static void show_object(struct object_array_entry *p)
  80 {
  81         /* An object with name "foo\n0000000..." can be used to
  82          * confuse downstream git-pack-objects very badly.
  83          */
  84         const char *ep = strchr(p->name, '\n');
  85         if (ep) {
  86                 fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(p->item->sha1),
  87                        (int) (ep - p->name),
  88                        p->name);
  89         }
  90         else
  91                 fprintf(pack_pipe, "%s %s\n",
  92                                 sha1_to_hex(p->item->sha1), p->name);
  93 }
  94
  95 static void show_edge(struct commit *commit)
  96 {
  97         fprintf(pack_pipe, "-%s\n", sha1_to_hex(commit->object.sha1));
  98 }
  99
 100 static int do_rev_list(int fd, void *create_full_pack)
 101 {
 102         int i;
 103         struct rev_info revs;
 104
 105         pack_pipe = fdopen(fd, "w");
 106         if (create_full_pack)
 107                 use_thin_pack = 0; /* no point doing it */
 108         init_revisions(&revs, NULL);
 109         revs.tag_objects = 1;
 110         revs.tree_objects = 1;
 111         revs.blob_objects = 1;
 112         if (use_thin_pack)
 113                 revs.edge_hint = 1;
 114
 115         if (create_full_pack) {
 116                 const char *args[] = {"rev-list", "--all", NULL};
 117                 setup_revisions(2, args, &revs, NULL);
 118         } else {
 119                 for (i = 0; i < want_obj.nr; i++) {
 120                         struct object *o = want_obj.objects[i].item;
 121                         /* why??? */
 122                         o->flags &= ~UNINTERESTING;
 123                         add_pending_object(&revs, o, NULL);
 124                 }
 125                 for (i = 0; i < have_obj.nr; i++) {
 126                         struct object *o = have_obj.objects[i].item;
 127                         o->flags |= UNINTERESTING;
 128                         add_pending_object(&revs, o, NULL);
 129                 }
 130                 setup_revisions(0, NULL, &revs, NULL);
 131         }
 132         prepare_revision_walk(&revs);
 133         mark_edges_uninteresting(revs.commits, &revs, show_edge);
 134         traverse_commit_list(&revs, show_commit, show_object);
 135         return 0;
 136 }
 137
 138 static void create_pack_file(void)
 139 {
 140         struct async rev_list;
 141         struct child_process pack_objects;
 142         int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
 143         char data[8193], progress[128];
 144         char abort_msg[] = "aborting due to possible repository "
 145                 "corruption on the remote side.";
 146         int buffered = -1;
 147         ssize_t sz;
 148         const char *argv[10];
 149         int arg = 0;
 150
 151         rev_list.proc = do_rev_list;
 152         /* .data is just a boolean: any non-NULL value will do */
 153         rev_list.data = create_full_pack ? &rev_list : NULL;
 154         if (start_async(&rev_list))
 155                 die("git-upload-pack: unable to fork git-rev-list");
 156
 157         argv[arg++] = "pack-objects";
 158         argv[arg++] = "--stdout";
 159         if (!no_progress)
 160                 argv[arg++] = "--progress";
 161         if (use_ofs_delta)
 162                 argv[arg++] = "--delta-base-offset";
 163         argv[arg++] = NULL;
 164
 165         memset(&pack_objects, 0, sizeof(pack_objects));
 166         pack_objects.in = rev_list.out; /* start_command closes it */
 167         pack_objects.out = -1;
 168         pack_objects.err = -1;
 169         pack_objects.git_cmd = 1;
 170         pack_objects.argv = argv;
 171
 172         if (start_command(&pack_objects))
 173                 die("git-upload-pack: unable to fork git-pack-objects");
 174
 175         /* We read from pack_objects.err to capture stderr output for
 176          * progress bar, and pack_objects.out to capture the pack data.
 177          */
 178
 179         while (1) {
 180                 struct pollfd pfd[2];
 181                 int pe, pu, pollsize;
 182
 183                 reset_timeout();
 184
 185                 pollsize = 0;
 186                 pe = pu = -1;
 187
 188                 if (0 <= pack_objects.out) {
 189                         pfd[pollsize].fd = pack_objects.out;
 190                         pfd[pollsize].events = POLLIN;
 191                         pu = pollsize;
 192                         pollsize++;
 193                 }
 194                 if (0 <= pack_objects.err) {
 195                         pfd[pollsize].fd = pack_objects.err;
 196                         pfd[pollsize].events = POLLIN;
 197                         pe = pollsize;
 198                         pollsize++;
 199                 }
 200
 201                 if (!pollsize)
 202                         break;
 203
 204                 if (poll(pfd, pollsize, -1) < 0) {
 205                         if (errno != EINTR) {
 206                                 error("poll failed, resuming: %s",
 207                                       strerror(errno));
 208                                 sleep(1);
 209                         }
 210                         continue;
 211                 }
 212                 if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
 213                         /* Data ready; we keep the last byte to ourselves
 214                          * in case we detect broken rev-list, so that we
 215                          * can leave the stream corrupted.  This is
 216                          * unfortunate -- unpack-objects would happily
 217                          * accept a valid packdata with trailing garbage,
 218                          * so appending garbage after we pass all the
 219                          * pack data is not good enough to signal
 220                          * breakage to downstream.
 221                          */
 222                         char *cp = data;
 223                         ssize_t outsz = 0;
 224                         if (0 <= buffered) {
 225                                 *cp++ = buffered;
 226                                 outsz++;
 227                         }
 228                         sz = xread(pack_objects.out, cp,
 229                                   sizeof(data) - outsz);
 230                         if (0 < sz)
 231                                         ;
 232                         else if (sz == 0) {
 233                                 close(pack_objects.out);
 234                                 pack_objects.out = -1;
 235                         }
 236                         else
 237                                 goto fail;
 238                         sz += outsz;
 239                         if (1 < sz) {
 240                                 buffered = data[sz-1] & 0xFF;
 241                                 sz--;
 242                         }
 243                         else
 244                                 buffered = -1;
 245                         sz = send_client_data(1, data, sz);
 246                         if (sz < 0)
 247                                 goto fail;
 248                 }
 249                 if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
 250                         /* Status ready; we ship that in the side-band
 251                          * or dump to the standard error.
 252                          */
 253                         sz = xread(pack_objects.err, progress,
 254                                   sizeof(progress));
 255                         if (0 < sz)
 256                                 send_client_data(2, progress, sz);
 257                         else if (sz == 0) {
 258                                 close(pack_objects.err);
 259                                 pack_objects.err = -1;
 260                         }
 261                         else
 262                                 goto fail;
 263                 }
 264         }
 265
 266         if (finish_command(&pack_objects)) {
 267                 error("git-upload-pack: git-pack-objects died with error.");
 268                 goto fail;
 269         }
 270         if (finish_async(&rev_list))
 271                 goto fail;      /* error was already reported */
 272
 273         /* flush the data */
 274         if (0 <= buffered) {
 275                 data[0] = buffered;
 276                 sz = send_client_data(1, data, 1);
 277                 if (sz < 0)
 278                         goto fail;
 279                 fprintf(stderr, "flushed.\n");
 280         }
 281         if (use_sideband)
 282                 packet_flush(1);
 283         return;
 284
 285  fail:
 286         send_client_data(3, abort_msg, sizeof(abort_msg));
 287         die("git-upload-pack: %s", abort_msg);
 288 }
 289
 290 static int got_sha1(char *hex, unsigned char *sha1)
 291 {
 292         struct object *o;
 293         int we_knew_they_have = 0;
 294
 295         if (get_sha1_hex(hex, sha1))
 296                 die("git-upload-pack: expected SHA1 object, got '%s'", hex);
 297         if (!has_sha1_file(sha1))
 298                 return -1;
 299
 300         o = lookup_object(sha1);
 301         if (!(o && o->parsed))
 302                 o = parse_object(sha1);
 303         if (!o)
 304                 die("oops (%s)", sha1_to_hex(sha1));
 305         if (o->type == OBJ_COMMIT) {
 306                 struct commit_list *parents;
 307                 struct commit *commit = (struct commit *)o;
 308                 if (o->flags & THEY_HAVE)
 309                         we_knew_they_have = 1;
 310                 else
 311                         o->flags |= THEY_HAVE;
 312                 if (!oldest_have || (commit->date < oldest_have))
 313                         oldest_have = commit->date;
 314                 for (parents = commit->parents;
 315                      parents;
 316                      parents = parents->next)
 317                         parents->item->object.flags |= THEY_HAVE;
 318         }
 319         if (!we_knew_they_have) {
 320                 add_object_array(o, NULL, &have_obj);
 321                 return 1;
 322         }
 323         return 0;
 324 }
 325
 326 static int reachable(struct commit *want)
 327 {
 328         struct commit_list *work = NULL;
 329
 330         insert_by_date(want, &work);
 331         while (work) {
 332                 struct commit_list *list = work->next;
 333                 struct commit *commit = work->item;
 334                 free(work);
 335                 work = list;
 336
 337                 if (commit->object.flags & THEY_HAVE) {
 338                         want->object.flags |= COMMON_KNOWN;
 339                         break;
 340                 }
 341                 if (!commit->object.parsed)
 342                         parse_object(commit->object.sha1);
 343                 if (commit->object.flags & REACHABLE)
 344                         continue;
 345                 commit->object.flags |= REACHABLE;
 346                 if (commit->date < oldest_have)
 347                         continue;
 348                 for (list = commit->parents; list; list = list->next) {
 349                         struct commit *parent = list->item;
 350                         if (!(parent->object.flags & REACHABLE))
 351                                 insert_by_date(parent, &work);
 352                 }
 353         }
 354         want->object.flags |= REACHABLE;
 355         clear_commit_marks(want, REACHABLE);
 356         free_commit_list(work);
 357         return (want->object.flags & COMMON_KNOWN);
 358 }
 359
 360 static int ok_to_give_up(void)
 361 {
 362         int i;
 363
 364         if (!have_obj.nr)
 365                 return 0;
 366
 367         for (i = 0; i < want_obj.nr; i++) {
 368                 struct object *want = want_obj.objects[i].item;
 369
 370                 if (want->flags & COMMON_KNOWN)
 371                         continue;
 372                 want = deref_tag(want, "a want line", 0);
 373                 if (!want || want->type != OBJ_COMMIT) {
 374                         /* no way to tell if this is reachable by
 375                          * looking at the ancestry chain alone, so
 376                          * leave a note to ourselves not to worry about
 377                          * this object anymore.
 378                          */
 379                         want_obj.objects[i].item->flags |= COMMON_KNOWN;
 380                         continue;
 381                 }
 382                 if (!reachable((struct commit *)want))
 383                         return 0;
 384         }
 385         return 1;
 386 }
 387
 388 static int get_common_commits(void)
 389 {
 390         static char line[1000];
 391         unsigned char sha1[20];
 392         char hex[41], last_hex[41];
 393         int len;
 394
 395         track_object_refs = 0;
 396         save_commit_buffer = 0;
 397
 398         for(;;) {
 399                 len = packet_read_line(0, line, sizeof(line));
 400                 reset_timeout();
 401
 402                 if (!len) {
 403                         if (have_obj.nr == 0 || multi_ack)
 404                                 packet_write(1, "NAK\n");
 405                         continue;
 406                 }
 407                 len = strip(line, len);
 408                 if (!prefixcmp(line, "have ")) {
 409                         switch (got_sha1(line+5, sha1)) {
 410                         case -1: /* they have what we do not */
 411                                 if (multi_ack && ok_to_give_up())
 412                                         packet_write(1, "ACK %s continue\n",
 413                                                      sha1_to_hex(sha1));
 414                                 break;
 415                         default:
 416                                 memcpy(hex, sha1_to_hex(sha1), 41);
 417                                 if (multi_ack) {
 418                                         const char *msg = "ACK %s continue\n";
 419                                         packet_write(1, msg, hex);
 420                                         memcpy(last_hex, hex, 41);
 421                                 }
 422                                 else if (have_obj.nr == 1)
 423                                         packet_write(1, "ACK %s\n", hex);
 424                                 break;
 425                         }
 426                         continue;
 427                 }
 428                 if (!strcmp(line, "done")) {
 429                         if (have_obj.nr > 0) {
 430                                 if (multi_ack)
 431                                         packet_write(1, "ACK %s\n", last_hex);
 432                                 return 0;
 433                         }
 434                         packet_write(1, "NAK\n");
 435                         return -1;
 436                 }
 437                 die("git-upload-pack: expected SHA1 list, got '%s'", line);
 438         }
 439 }
 440
 441 static void receive_needs(void)
 442 {
 443         struct object_array shallows = {0, 0, NULL};
 444         static char line[1000];
 445         int len, depth = 0;
 446
 447         for (;;) {
 448                 struct object *o;
 449                 unsigned char sha1_buf[20];
 450                 len = packet_read_line(0, line, sizeof(line));
 451                 reset_timeout();
 452                 if (!len)
 453                         break;
 454
 455                 if (!prefixcmp(line, "shallow ")) {
 456                         unsigned char sha1[20];
 457                         struct object *object;
 458                         use_thin_pack = 0;
 459                         if (get_sha1(line + 8, sha1))
 460                                 die("invalid shallow line: %s", line);
 461                         object = parse_object(sha1);
 462                         if (!object)
 463                                 die("did not find object for %s", line);
 464                         object->flags |= CLIENT_SHALLOW;
 465                         add_object_array(object, NULL, &shallows);
 466                         continue;
 467                 }
 468                 if (!prefixcmp(line, "deepen ")) {
 469                         char *end;
 470                         use_thin_pack = 0;
 471                         depth = strtol(line + 7, &end, 0);
 472                         if (end == line + 7 || depth <= 0)
 473                                 die("Invalid deepen: %s", line);
 474                         continue;
 475                 }
 476                 if (prefixcmp(line, "want ") ||
 477                     get_sha1_hex(line+5, sha1_buf))
 478                         die("git-upload-pack: protocol error, "
 479                             "expected to get sha, not '%s'", line);
 480                 if (strstr(line+45, "multi_ack"))
 481                         multi_ack = 1;
 482                 if (strstr(line+45, "thin-pack"))
 483                         use_thin_pack = 1;
 484                 if (strstr(line+45, "ofs-delta"))
 485                         use_ofs_delta = 1;
 486                 if (strstr(line+45, "side-band-64k"))
 487                         use_sideband = LARGE_PACKET_MAX;
 488                 else if (strstr(line+45, "side-band"))
 489                         use_sideband = DEFAULT_PACKET_MAX;
 490                 if (strstr(line+45, "no-progress"))
 491                         no_progress = 1;
 492
 493                 /* We have sent all our refs already, and the other end
 494                  * should have chosen out of them; otherwise they are
 495                  * asking for nonsense.
 496                  *
 497                  * Hmph.  We may later want to allow "want" line that
 498                  * asks for something like "master~10" (symbolic)...
 499                  * would it make sense?  I don't know.
 500                  */
 501                 o = lookup_object(sha1_buf);
 502                 if (!o || !(o->flags & OUR_REF))
 503                         die("git-upload-pack: not our ref %s", line+5);
 504                 if (!(o->flags & WANTED)) {
 505                         o->flags |= WANTED;
 506                         add_object_array(o, NULL, &want_obj);
 507                 }
 508         }
 509         if (depth == 0 && shallows.nr == 0)
 510                 return;
 511         if (depth > 0) {
 512                 struct commit_list *result, *backup;
 513                 int i;
 514                 backup = result = get_shallow_commits(&want_obj, depth,
 515                         SHALLOW, NOT_SHALLOW);
 516                 while (result) {
 517                         struct object *object = &result->item->object;
 518                         if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
 519                                 packet_write(1, "shallow %s",
 520                                                 sha1_to_hex(object->sha1));
 521                                 register_shallow(object->sha1);
 522                         }
 523                         result = result->next;
 524                 }
 525                 free_commit_list(backup);
 526                 for (i = 0; i < shallows.nr; i++) {
 527                         struct object *object = shallows.objects[i].item;
 528                         if (object->flags & NOT_SHALLOW) {
 529                                 struct commit_list *parents;
 530                                 packet_write(1, "unshallow %s",
 531                                         sha1_to_hex(object->sha1));
 532                                 object->flags &= ~CLIENT_SHALLOW;
 533                                 /* make sure the real parents are parsed */
 534                                 unregister_shallow(object->sha1);
 535                                 object->parsed = 0;
 536                                 parse_commit((struct commit *)object);
 537                                 parents = ((struct commit *)object)->parents;
 538                                 while (parents) {
 539                                         add_object_array(&parents->item->object,
 540                                                         NULL, &want_obj);
 541                                         parents = parents->next;
 542                                 }
 543                         }
 544                         /* make sure commit traversal conforms to client */
 545                         register_shallow(object->sha1);
 546                 }
 547                 packet_flush(1);
 548         } else
 549                 if (shallows.nr > 0) {
 550                         int i;
 551                         for (i = 0; i < shallows.nr; i++)
 552                                 register_shallow(shallows.objects[i].item->sha1);
 553                 }
 554         free(shallows.objects);
 555 }
 556
 557 static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 558 {
 559         static const char *capabilities = "multi_ack thin-pack side-band"
 560                 " side-band-64k ofs-delta shallow no-progress";
 561         struct object *o = parse_object(sha1);
 562
 563         if (!o)
 564                 die("git-upload-pack: cannot find object %s:", sha1_to_hex(sha1));
 565
 566         if (capabilities)
 567                 packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
 568                         0, capabilities);
 569         else
 570                 packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
 571         capabilities = NULL;
 572         if (!(o->flags & OUR_REF)) {
 573                 o->flags |= OUR_REF;
 574                 nr_our_refs++;
 575         }
 576         if (o->type == OBJ_TAG) {
 577                 o = deref_tag(o, refname, 0);
 578                 packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
 579         }
 580         return 0;
 581 }
 582
 583 static void upload_pack(void)
 584 {
 585         reset_timeout();
 586         head_ref(send_ref, NULL);
 587         for_each_ref(send_ref, NULL);
 588         packet_flush(1);
 589         receive_needs();
 590         if (want_obj.nr) {
 591                 get_common_commits();
 592                 create_pack_file();
 593         }
 594 }
 595
 596 int main(int argc, char **argv)
 597 {
 598         char *dir;
 599         int i;
 600         int strict = 0;
 601
 602         for (i = 1; i < argc; i++) {
 603                 char *arg = argv[i];
 604
 605                 if (arg[0] != '-')
 606                         break;
 607                 if (!strcmp(arg, "--strict")) {
 608                         strict = 1;
 609                         continue;
 610                 }
 611                 if (!prefixcmp(arg, "--timeout=")) {
 612                         timeout = atoi(arg+10);
 613                         continue;
 614                 }
 615                 if (!strcmp(arg, "--")) {
 616                         i++;
 617                         break;
 618                 }
 619         }
 620
 621         if (i != argc-1)
 622                 usage(upload_pack_usage);
 623
 624         setup_path(NULL);
 625
 626         dir = argv[i];
 627
 628         if (!enter_repo(dir, strict))
 629                 die("'%s': unable to chdir or not a git archive", dir);
 630         if (is_repository_shallow())
 631                 die("attempt to fetch/clone from a shallow repository");
 632         upload_pack();
 633         return 0;
 634 }