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         if (prepare_revision_walk(&revs))
 133                 die("revision walk setup failed");
 134         mark_edges_uninteresting(revs.commits, &revs, show_edge);
 135         traverse_commit_list(&revs, show_commit, show_object);
 136         return 0;
 137 }
 138
 139 static void create_pack_file(void)
 140 {
 141         struct async rev_list;
 142         struct child_process pack_objects;
 143         int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
 144         char data[8193], progress[128];
 145         char abort_msg[] = "aborting due to possible repository "
 146                 "corruption on the remote side.";
 147         int buffered = -1;
 148         ssize_t sz;
 149         const char *argv[10];
 150         int arg = 0;
 151
 152         rev_list.proc = do_rev_list;
 153         /* .data is just a boolean: any non-NULL value will do */
 154         rev_list.data = create_full_pack ? &rev_list : NULL;
 155         if (start_async(&rev_list))
 156                 die("git-upload-pack: unable to fork git-rev-list");
 157
 158         argv[arg++] = "pack-objects";
 159         argv[arg++] = "--stdout";
 160         if (!no_progress)
 161                 argv[arg++] = "--progress";
 162         if (use_ofs_delta)
 163                 argv[arg++] = "--delta-base-offset";
 164         argv[arg++] = NULL;
 165
 166         memset(&pack_objects, 0, sizeof(pack_objects));
 167         pack_objects.in = rev_list.out; /* start_command closes it */
 168         pack_objects.out = -1;
 169         pack_objects.err = -1;
 170         pack_objects.git_cmd = 1;
 171         pack_objects.argv = argv;
 172
 173         if (start_command(&pack_objects))
 174                 die("git-upload-pack: unable to fork git-pack-objects");
 175
 176         /* We read from pack_objects.err to capture stderr output for
 177          * progress bar, and pack_objects.out to capture the pack data.
 178          */
 179
 180         while (1) {
 181                 struct pollfd pfd[2];
 182                 int pe, pu, pollsize;
 183
 184                 reset_timeout();
 185
 186                 pollsize = 0;
 187                 pe = pu = -1;
 188
 189                 if (0 <= pack_objects.out) {
 190                         pfd[pollsize].fd = pack_objects.out;
 191                         pfd[pollsize].events = POLLIN;
 192                         pu = pollsize;
 193                         pollsize++;
 194                 }
 195                 if (0 <= pack_objects.err) {
 196                         pfd[pollsize].fd = pack_objects.err;
 197                         pfd[pollsize].events = POLLIN;
 198                         pe = pollsize;
 199                         pollsize++;
 200                 }
 201
 202                 if (!pollsize)
 203                         break;
 204
 205                 if (poll(pfd, pollsize, -1) < 0) {
 206                         if (errno != EINTR) {
 207                                 error("poll failed, resuming: %s",
 208                                       strerror(errno));
 209                                 sleep(1);
 210                         }
 211                         continue;
 212                 }
 213                 if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
 214                         /* Data ready; we keep the last byte to ourselves
 215                          * in case we detect broken rev-list, so that we
 216                          * can leave the stream corrupted.  This is
 217                          * unfortunate -- unpack-objects would happily
 218                          * accept a valid packdata with trailing garbage,
 219                          * so appending garbage after we pass all the
 220                          * pack data is not good enough to signal
 221                          * breakage to downstream.
 222                          */
 223                         char *cp = data;
 224                         ssize_t outsz = 0;
 225                         if (0 <= buffered) {
 226                                 *cp++ = buffered;
 227                                 outsz++;
 228                         }
 229                         sz = xread(pack_objects.out, cp,
 230                                   sizeof(data) - outsz);
 231                         if (0 < sz)
 232                                         ;
 233                         else if (sz == 0) {
 234                                 close(pack_objects.out);
 235                                 pack_objects.out = -1;
 236                         }
 237                         else
 238                                 goto fail;
 239                         sz += outsz;
 240                         if (1 < sz) {
 241                                 buffered = data[sz-1] & 0xFF;
 242                                 sz--;
 243                         }
 244                         else
 245                                 buffered = -1;
 246                         sz = send_client_data(1, data, sz);
 247                         if (sz < 0)
 248                                 goto fail;
 249                 }
 250                 if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
 251                         /* Status ready; we ship that in the side-band
 252                          * or dump to the standard error.
 253                          */
 254                         sz = xread(pack_objects.err, progress,
 255                                   sizeof(progress));
 256                         if (0 < sz)
 257                                 send_client_data(2, progress, sz);
 258                         else if (sz == 0) {
 259                                 close(pack_objects.err);
 260                                 pack_objects.err = -1;
 261                         }
 262                         else
 263                                 goto fail;
 264                 }
 265         }
 266
 267         if (finish_command(&pack_objects)) {
 268                 error("git-upload-pack: git-pack-objects died with error.");
 269                 goto fail;
 270         }
 271         if (finish_async(&rev_list))
 272                 goto fail;      /* error was already reported */
 273
 274         /* flush the data */
 275         if (0 <= buffered) {
 276                 data[0] = buffered;
 277                 sz = send_client_data(1, data, 1);
 278                 if (sz < 0)
 279                         goto fail;
 280                 fprintf(stderr, "flushed.\n");
 281         }
 282         if (use_sideband)
 283                 packet_flush(1);
 284         return;
 285
 286  fail:
 287         send_client_data(3, abort_msg, sizeof(abort_msg));
 288         die("git-upload-pack: %s", abort_msg);
 289 }
 290
 291 static int got_sha1(char *hex, unsigned char *sha1)
 292 {
 293         struct object *o;
 294         int we_knew_they_have = 0;
 295
 296         if (get_sha1_hex(hex, sha1))
 297                 die("git-upload-pack: expected SHA1 object, got '%s'", hex);
 298         if (!has_sha1_file(sha1))
 299                 return -1;
 300
 301         o = lookup_object(sha1);
 302         if (!(o && o->parsed))
 303                 o = parse_object(sha1);
 304         if (!o)
 305                 die("oops (%s)", sha1_to_hex(sha1));
 306         if (o->type == OBJ_COMMIT) {
 307                 struct commit_list *parents;
 308                 struct commit *commit = (struct commit *)o;
 309                 if (o->flags & THEY_HAVE)
 310                         we_knew_they_have = 1;
 311                 else
 312                         o->flags |= THEY_HAVE;
 313                 if (!oldest_have || (commit->date < oldest_have))
 314                         oldest_have = commit->date;
 315                 for (parents = commit->parents;
 316                      parents;
 317                      parents = parents->next)
 318                         parents->item->object.flags |= THEY_HAVE;
 319         }
 320         if (!we_knew_they_have) {
 321                 add_object_array(o, NULL, &have_obj);
 322                 return 1;
 323         }
 324         return 0;
 325 }
 326
 327 static int reachable(struct commit *want)
 328 {
 329         struct commit_list *work = NULL;
 330
 331         insert_by_date(want, &work);
 332         while (work) {
 333                 struct commit_list *list = work->next;
 334                 struct commit *commit = work->item;
 335                 free(work);
 336                 work = list;
 337
 338                 if (commit->object.flags & THEY_HAVE) {
 339                         want->object.flags |= COMMON_KNOWN;
 340                         break;
 341                 }
 342                 if (!commit->object.parsed)
 343                         parse_object(commit->object.sha1);
 344                 if (commit->object.flags & REACHABLE)
 345                         continue;
 346                 commit->object.flags |= REACHABLE;
 347                 if (commit->date < oldest_have)
 348                         continue;
 349                 for (list = commit->parents; list; list = list->next) {
 350                         struct commit *parent = list->item;
 351                         if (!(parent->object.flags & REACHABLE))
 352                                 insert_by_date(parent, &work);
 353                 }
 354         }
 355         want->object.flags |= REACHABLE;
 356         clear_commit_marks(want, REACHABLE);
 357         free_commit_list(work);
 358         return (want->object.flags & COMMON_KNOWN);
 359 }
 360
 361 static int ok_to_give_up(void)
 362 {
 363         int i;
 364
 365         if (!have_obj.nr)
 366                 return 0;
 367
 368         for (i = 0; i < want_obj.nr; i++) {
 369                 struct object *want = want_obj.objects[i].item;
 370
 371                 if (want->flags & COMMON_KNOWN)
 372                         continue;
 373                 want = deref_tag(want, "a want line", 0);
 374                 if (!want || want->type != OBJ_COMMIT) {
 375                         /* no way to tell if this is reachable by
 376                          * looking at the ancestry chain alone, so
 377                          * leave a note to ourselves not to worry about
 378                          * this object anymore.
 379                          */
 380                         want_obj.objects[i].item->flags |= COMMON_KNOWN;
 381                         continue;
 382                 }
 383                 if (!reachable((struct commit *)want))
 384                         return 0;
 385         }
 386         return 1;
 387 }
 388
 389 static int get_common_commits(void)
 390 {
 391         static char line[1000];
 392         unsigned char sha1[20];
 393         char hex[41], last_hex[41];
 394         int len;
 395
 396         track_object_refs = 0;
 397         save_commit_buffer = 0;
 398
 399         for(;;) {
 400                 len = packet_read_line(0, line, sizeof(line));
 401                 reset_timeout();
 402
 403                 if (!len) {
 404                         if (have_obj.nr == 0 || multi_ack)
 405                                 packet_write(1, "NAK\n");
 406                         continue;
 407                 }
 408                 len = strip(line, len);
 409                 if (!prefixcmp(line, "have ")) {
 410                         switch (got_sha1(line+5, sha1)) {
 411                         case -1: /* they have what we do not */
 412                                 if (multi_ack && ok_to_give_up())
 413                                         packet_write(1, "ACK %s continue\n",
 414                                                      sha1_to_hex(sha1));
 415                                 break;
 416                         default:
 417                                 memcpy(hex, sha1_to_hex(sha1), 41);
 418                                 if (multi_ack) {
 419                                         const char *msg = "ACK %s continue\n";
 420                                         packet_write(1, msg, hex);
 421                                         memcpy(last_hex, hex, 41);
 422                                 }
 423                                 else if (have_obj.nr == 1)
 424                                         packet_write(1, "ACK %s\n", hex);
 425                                 break;
 426                         }
 427                         continue;
 428                 }
 429                 if (!strcmp(line, "done")) {
 430                         if (have_obj.nr > 0) {
 431                                 if (multi_ack)
 432                                         packet_write(1, "ACK %s\n", last_hex);
 433                                 return 0;
 434                         }
 435                         packet_write(1, "NAK\n");
 436                         return -1;
 437                 }
 438                 die("git-upload-pack: expected SHA1 list, got '%s'", line);
 439         }
 440 }
 441
 442 static void receive_needs(void)
 443 {
 444         struct object_array shallows = {0, 0, NULL};
 445         static char line[1000];
 446         int len, depth = 0;
 447
 448         for (;;) {
 449                 struct object *o;
 450                 unsigned char sha1_buf[20];
 451                 len = packet_read_line(0, line, sizeof(line));
 452                 reset_timeout();
 453                 if (!len)
 454                         break;
 455
 456                 if (!prefixcmp(line, "shallow ")) {
 457                         unsigned char sha1[20];
 458                         struct object *object;
 459                         use_thin_pack = 0;
 460                         if (get_sha1(line + 8, sha1))
 461                                 die("invalid shallow line: %s", line);
 462                         object = parse_object(sha1);
 463                         if (!object)
 464                                 die("did not find object for %s", line);
 465                         object->flags |= CLIENT_SHALLOW;
 466                         add_object_array(object, NULL, &shallows);
 467                         continue;
 468                 }
 469                 if (!prefixcmp(line, "deepen ")) {
 470                         char *end;
 471                         use_thin_pack = 0;
 472                         depth = strtol(line + 7, &end, 0);
 473                         if (end == line + 7 || depth <= 0)
 474                                 die("Invalid deepen: %s", line);
 475                         continue;
 476                 }
 477                 if (prefixcmp(line, "want ") ||
 478                     get_sha1_hex(line+5, sha1_buf))
 479                         die("git-upload-pack: protocol error, "
 480                             "expected to get sha, not '%s'", line);
 481                 if (strstr(line+45, "multi_ack"))
 482                         multi_ack = 1;
 483                 if (strstr(line+45, "thin-pack"))
 484                         use_thin_pack = 1;
 485                 if (strstr(line+45, "ofs-delta"))
 486                         use_ofs_delta = 1;
 487                 if (strstr(line+45, "side-band-64k"))
 488                         use_sideband = LARGE_PACKET_MAX;
 489                 else if (strstr(line+45, "side-band"))
 490                         use_sideband = DEFAULT_PACKET_MAX;
 491                 if (strstr(line+45, "no-progress"))
 492                         no_progress = 1;
 493
 494                 /* We have sent all our refs already, and the other end
 495                  * should have chosen out of them; otherwise they are
 496                  * asking for nonsense.
 497                  *
 498                  * Hmph.  We may later want to allow "want" line that
 499                  * asks for something like "master~10" (symbolic)...
 500                  * would it make sense?  I don't know.
 501                  */
 502                 o = lookup_object(sha1_buf);
 503                 if (!o || !(o->flags & OUR_REF))
 504                         die("git-upload-pack: not our ref %s", line+5);
 505                 if (!(o->flags & WANTED)) {
 506                         o->flags |= WANTED;
 507                         add_object_array(o, NULL, &want_obj);
 508                 }
 509         }
 510         if (depth == 0 && shallows.nr == 0)
 511                 return;
 512         if (depth > 0) {
 513                 struct commit_list *result, *backup;
 514                 int i;
 515                 backup = result = get_shallow_commits(&want_obj, depth,
 516                         SHALLOW, NOT_SHALLOW);
 517                 while (result) {
 518                         struct object *object = &result->item->object;
 519                         if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
 520                                 packet_write(1, "shallow %s",
 521                                                 sha1_to_hex(object->sha1));
 522                                 register_shallow(object->sha1);
 523                         }
 524                         result = result->next;
 525                 }
 526                 free_commit_list(backup);
 527                 for (i = 0; i < shallows.nr; i++) {
 528                         struct object *object = shallows.objects[i].item;
 529                         if (object->flags & NOT_SHALLOW) {
 530                                 struct commit_list *parents;
 531                                 packet_write(1, "unshallow %s",
 532                                         sha1_to_hex(object->sha1));
 533                                 object->flags &= ~CLIENT_SHALLOW;
 534                                 /* make sure the real parents are parsed */
 535                                 unregister_shallow(object->sha1);
 536                                 object->parsed = 0;
 537                                 if (parse_commit((struct commit *)object))
 538                                         die("invalid commit");
 539                                 parents = ((struct commit *)object)->parents;
 540                                 while (parents) {
 541                                         add_object_array(&parents->item->object,
 542                                                         NULL, &want_obj);
 543                                         parents = parents->next;
 544                                 }
 545                         }
 546                         /* make sure commit traversal conforms to client */
 547                         register_shallow(object->sha1);
 548                 }
 549                 packet_flush(1);
 550         } else
 551                 if (shallows.nr > 0) {
 552                         int i;
 553                         for (i = 0; i < shallows.nr; i++)
 554                                 register_shallow(shallows.objects[i].item->sha1);
 555                 }
 556         free(shallows.objects);
 557 }
 558
 559 static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 560 {
 561         static const char *capabilities = "multi_ack thin-pack side-band"
 562                 " side-band-64k ofs-delta shallow no-progress";
 563         struct object *o = parse_object(sha1);
 564
 565         if (!o)
 566                 die("git-upload-pack: cannot find object %s:", sha1_to_hex(sha1));
 567
 568         if (capabilities)
 569                 packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
 570                         0, capabilities);
 571         else
 572                 packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
 573         capabilities = NULL;
 574         if (!(o->flags & OUR_REF)) {
 575                 o->flags |= OUR_REF;
 576                 nr_our_refs++;
 577         }
 578         if (o->type == OBJ_TAG) {
 579                 o = deref_tag(o, refname, 0);
 580                 if (o)
 581                         packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
 582         }
 583         return 0;
 584 }
 585
 586 static void upload_pack(void)
 587 {
 588         reset_timeout();
 589         head_ref(send_ref, NULL);
 590         for_each_ref(send_ref, NULL);
 591         packet_flush(1);
 592         receive_needs();
 593         if (want_obj.nr) {
 594                 get_common_commits();
 595                 create_pack_file();
 596         }
 597 }
 598
 599 int main(int argc, char **argv)
 600 {
 601         char *dir;
 602         int i;
 603         int strict = 0;
 604
 605         for (i = 1; i < argc; i++) {
 606                 char *arg = argv[i];
 607
 608                 if (arg[0] != '-')
 609                         break;
 610                 if (!strcmp(arg, "--strict")) {
 611                         strict = 1;
 612                         continue;
 613                 }
 614                 if (!prefixcmp(arg, "--timeout=")) {
 615                         timeout = atoi(arg+10);
 616                         continue;
 617                 }
 618                 if (!strcmp(arg, "--")) {
 619                         i++;
 620                         break;
 621                 }
 622         }
 623
 624         if (i != argc-1)
 625                 usage(upload_pack_usage);
 626
 627         setup_path(NULL);
 628
 629         dir = argv[i];
 630
 631         if (!enter_repo(dir, strict))
 632                 die("'%s': unable to chdir or not a git archive", dir);
 633         if (is_repository_shallow())
 634                 die("attempt to fetch/clone from a shallow repository");
 635         upload_pack();
 636         return 0;
 637 }