upload-pack.con commit Merge branch 'jc/maint-unpack-objects-strict' (ab36d06)
   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
  14static 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
  27static unsigned long oldest_have;
  28
  29static int multi_ack, nr_our_refs;
  30static int use_thin_pack, use_ofs_delta, use_include_tag;
  31static int no_progress, daemon_mode;
  32static int shallow_nr;
  33static struct object_array have_obj;
  34static struct object_array want_obj;
  35static unsigned int timeout;
  36/* 0 for no sideband,
  37 * otherwise maximum packet size (up to 65520 bytes).
  38 */
  39static int use_sideband;
  40static int debug_fd;
  41
  42static void reset_timeout(void)
  43{
  44        alarm(timeout);
  45}
  46
  47static int strip(char *line, int len)
  48{
  49        if (len && line[len-1] == '\n')
  50                line[--len] = 0;
  51        return len;
  52}
  53
  54static ssize_t send_client_data(int fd, const char *data, ssize_t sz)
  55{
  56        if (use_sideband)
  57                return send_sideband(1, fd, data, sz, use_sideband);
  58        if (fd == 3)
  59                /* emergency quit */
  60                fd = 2;
  61        if (fd == 2) {
  62                /* XXX: are we happy to lose stuff here? */
  63                xwrite(fd, data, sz);
  64                return sz;
  65        }
  66        return safe_write(fd, data, sz);
  67}
  68
  69static FILE *pack_pipe = NULL;
  70static void show_commit(struct commit *commit, void *data)
  71{
  72        if (commit->object.flags & BOUNDARY)
  73                fputc('-', pack_pipe);
  74        if (fputs(sha1_to_hex(commit->object.sha1), pack_pipe) < 0)
  75                die("broken output pipe");
  76        fputc('\n', pack_pipe);
  77        fflush(pack_pipe);
  78        free(commit->buffer);
  79        commit->buffer = NULL;
  80}
  81
  82static void show_object(struct object *obj, const struct name_path *path, const char *component)
  83{
  84        /* An object with name "foo\n0000000..." can be used to
  85         * confuse downstream git-pack-objects very badly.
  86         */
  87        const char *name = path_name(path, component);
  88        const char *ep = strchr(name, '\n');
  89        if (ep) {
  90                fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(obj->sha1),
  91                       (int) (ep - name),
  92                       name);
  93        }
  94        else
  95                fprintf(pack_pipe, "%s %s\n",
  96                                sha1_to_hex(obj->sha1), name);
  97        free((char *)name);
  98}
  99
 100static void show_edge(struct commit *commit)
 101{
 102        fprintf(pack_pipe, "-%s\n", sha1_to_hex(commit->object.sha1));
 103}
 104
 105static int do_rev_list(int fd, void *create_full_pack)
 106{
 107        int i;
 108        struct rev_info revs;
 109
 110        pack_pipe = fdopen(fd, "w");
 111        init_revisions(&revs, NULL);
 112        revs.tag_objects = 1;
 113        revs.tree_objects = 1;
 114        revs.blob_objects = 1;
 115        if (use_thin_pack)
 116                revs.edge_hint = 1;
 117
 118        if (create_full_pack) {
 119                const char *args[] = {"rev-list", "--all", NULL};
 120                setup_revisions(2, args, &revs, NULL);
 121        } else {
 122                for (i = 0; i < want_obj.nr; i++) {
 123                        struct object *o = want_obj.objects[i].item;
 124                        /* why??? */
 125                        o->flags &= ~UNINTERESTING;
 126                        add_pending_object(&revs, o, NULL);
 127                }
 128                for (i = 0; i < have_obj.nr; i++) {
 129                        struct object *o = have_obj.objects[i].item;
 130                        o->flags |= UNINTERESTING;
 131                        add_pending_object(&revs, o, NULL);
 132                }
 133                setup_revisions(0, NULL, &revs, NULL);
 134        }
 135        if (prepare_revision_walk(&revs))
 136                die("revision walk setup failed");
 137        mark_edges_uninteresting(revs.commits, &revs, show_edge);
 138        traverse_commit_list(&revs, show_commit, show_object, NULL);
 139        fflush(pack_pipe);
 140        fclose(pack_pipe);
 141        return 0;
 142}
 143
 144static void create_pack_file(void)
 145{
 146        struct async rev_list;
 147        struct child_process pack_objects;
 148        int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
 149        char data[8193], progress[128];
 150        char abort_msg[] = "aborting due to possible repository "
 151                "corruption on the remote side.";
 152        int buffered = -1;
 153        ssize_t sz;
 154        const char *argv[10];
 155        int arg = 0;
 156
 157        if (shallow_nr) {
 158                rev_list.proc = do_rev_list;
 159                rev_list.data = 0;
 160                if (start_async(&rev_list))
 161                        die("git upload-pack: unable to fork git-rev-list");
 162                argv[arg++] = "pack-objects";
 163        } else {
 164                argv[arg++] = "pack-objects";
 165                argv[arg++] = "--revs";
 166                if (create_full_pack)
 167                        argv[arg++] = "--all";
 168                else if (use_thin_pack)
 169                        argv[arg++] = "--thin";
 170        }
 171
 172        argv[arg++] = "--stdout";
 173        if (!no_progress)
 174                argv[arg++] = "--progress";
 175        if (use_ofs_delta)
 176                argv[arg++] = "--delta-base-offset";
 177        if (use_include_tag)
 178                argv[arg++] = "--include-tag";
 179        argv[arg++] = NULL;
 180
 181        memset(&pack_objects, 0, sizeof(pack_objects));
 182        pack_objects.in = shallow_nr ? rev_list.out : -1;
 183        pack_objects.out = -1;
 184        pack_objects.err = -1;
 185        pack_objects.git_cmd = 1;
 186        pack_objects.argv = argv;
 187
 188        if (start_command(&pack_objects))
 189                die("git upload-pack: unable to fork git-pack-objects");
 190
 191        /* pass on revisions we (don't) want */
 192        if (!shallow_nr) {
 193                FILE *pipe_fd = fdopen(pack_objects.in, "w");
 194                if (!create_full_pack) {
 195                        int i;
 196                        for (i = 0; i < want_obj.nr; i++)
 197                                fprintf(pipe_fd, "%s\n", sha1_to_hex(want_obj.objects[i].item->sha1));
 198                        fprintf(pipe_fd, "--not\n");
 199                        for (i = 0; i < have_obj.nr; i++)
 200                                fprintf(pipe_fd, "%s\n", sha1_to_hex(have_obj.objects[i].item->sha1));
 201                }
 202
 203                fprintf(pipe_fd, "\n");
 204                fflush(pipe_fd);
 205                fclose(pipe_fd);
 206        }
 207
 208
 209        /* We read from pack_objects.err to capture stderr output for
 210         * progress bar, and pack_objects.out to capture the pack data.
 211         */
 212
 213        while (1) {
 214                struct pollfd pfd[2];
 215                int pe, pu, pollsize;
 216
 217                reset_timeout();
 218
 219                pollsize = 0;
 220                pe = pu = -1;
 221
 222                if (0 <= pack_objects.out) {
 223                        pfd[pollsize].fd = pack_objects.out;
 224                        pfd[pollsize].events = POLLIN;
 225                        pu = pollsize;
 226                        pollsize++;
 227                }
 228                if (0 <= pack_objects.err) {
 229                        pfd[pollsize].fd = pack_objects.err;
 230                        pfd[pollsize].events = POLLIN;
 231                        pe = pollsize;
 232                        pollsize++;
 233                }
 234
 235                if (!pollsize)
 236                        break;
 237
 238                if (poll(pfd, pollsize, -1) < 0) {
 239                        if (errno != EINTR) {
 240                                error("poll failed, resuming: %s",
 241                                      strerror(errno));
 242                                sleep(1);
 243                        }
 244                        continue;
 245                }
 246                if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
 247                        /* Data ready; we keep the last byte to ourselves
 248                         * in case we detect broken rev-list, so that we
 249                         * can leave the stream corrupted.  This is
 250                         * unfortunate -- unpack-objects would happily
 251                         * accept a valid packdata with trailing garbage,
 252                         * so appending garbage after we pass all the
 253                         * pack data is not good enough to signal
 254                         * breakage to downstream.
 255                         */
 256                        char *cp = data;
 257                        ssize_t outsz = 0;
 258                        if (0 <= buffered) {
 259                                *cp++ = buffered;
 260                                outsz++;
 261                        }
 262                        sz = xread(pack_objects.out, cp,
 263                                  sizeof(data) - outsz);
 264                        if (0 < sz)
 265                                        ;
 266                        else if (sz == 0) {
 267                                close(pack_objects.out);
 268                                pack_objects.out = -1;
 269                        }
 270                        else
 271                                goto fail;
 272                        sz += outsz;
 273                        if (1 < sz) {
 274                                buffered = data[sz-1] & 0xFF;
 275                                sz--;
 276                        }
 277                        else
 278                                buffered = -1;
 279                        sz = send_client_data(1, data, sz);
 280                        if (sz < 0)
 281                                goto fail;
 282                }
 283                if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
 284                        /* Status ready; we ship that in the side-band
 285                         * or dump to the standard error.
 286                         */
 287                        sz = xread(pack_objects.err, progress,
 288                                  sizeof(progress));
 289                        if (0 < sz)
 290                                send_client_data(2, progress, sz);
 291                        else if (sz == 0) {
 292                                close(pack_objects.err);
 293                                pack_objects.err = -1;
 294                        }
 295                        else
 296                                goto fail;
 297                }
 298        }
 299
 300        if (finish_command(&pack_objects)) {
 301                error("git upload-pack: git-pack-objects died with error.");
 302                goto fail;
 303        }
 304        if (shallow_nr && finish_async(&rev_list))
 305                goto fail;      /* error was already reported */
 306
 307        /* flush the data */
 308        if (0 <= buffered) {
 309                data[0] = buffered;
 310                sz = send_client_data(1, data, 1);
 311                if (sz < 0)
 312                        goto fail;
 313                fprintf(stderr, "flushed.\n");
 314        }
 315        if (use_sideband)
 316                packet_flush(1);
 317        return;
 318
 319 fail:
 320        send_client_data(3, abort_msg, sizeof(abort_msg));
 321        die("git upload-pack: %s", abort_msg);
 322}
 323
 324static int got_sha1(char *hex, unsigned char *sha1)
 325{
 326        struct object *o;
 327        int we_knew_they_have = 0;
 328
 329        if (get_sha1_hex(hex, sha1))
 330                die("git upload-pack: expected SHA1 object, got '%s'", hex);
 331        if (!has_sha1_file(sha1))
 332                return -1;
 333
 334        o = lookup_object(sha1);
 335        if (!(o && o->parsed))
 336                o = parse_object(sha1);
 337        if (!o)
 338                die("oops (%s)", sha1_to_hex(sha1));
 339        if (o->type == OBJ_COMMIT) {
 340                struct commit_list *parents;
 341                struct commit *commit = (struct commit *)o;
 342                if (o->flags & THEY_HAVE)
 343                        we_knew_they_have = 1;
 344                else
 345                        o->flags |= THEY_HAVE;
 346                if (!oldest_have || (commit->date < oldest_have))
 347                        oldest_have = commit->date;
 348                for (parents = commit->parents;
 349                     parents;
 350                     parents = parents->next)
 351                        parents->item->object.flags |= THEY_HAVE;
 352        }
 353        if (!we_knew_they_have) {
 354                add_object_array(o, NULL, &have_obj);
 355                return 1;
 356        }
 357        return 0;
 358}
 359
 360static int reachable(struct commit *want)
 361{
 362        struct commit_list *work = NULL;
 363
 364        insert_by_date(want, &work);
 365        while (work) {
 366                struct commit_list *list = work->next;
 367                struct commit *commit = work->item;
 368                free(work);
 369                work = list;
 370
 371                if (commit->object.flags & THEY_HAVE) {
 372                        want->object.flags |= COMMON_KNOWN;
 373                        break;
 374                }
 375                if (!commit->object.parsed)
 376                        parse_object(commit->object.sha1);
 377                if (commit->object.flags & REACHABLE)
 378                        continue;
 379                commit->object.flags |= REACHABLE;
 380                if (commit->date < oldest_have)
 381                        continue;
 382                for (list = commit->parents; list; list = list->next) {
 383                        struct commit *parent = list->item;
 384                        if (!(parent->object.flags & REACHABLE))
 385                                insert_by_date(parent, &work);
 386                }
 387        }
 388        want->object.flags |= REACHABLE;
 389        clear_commit_marks(want, REACHABLE);
 390        free_commit_list(work);
 391        return (want->object.flags & COMMON_KNOWN);
 392}
 393
 394static int ok_to_give_up(void)
 395{
 396        int i;
 397
 398        if (!have_obj.nr)
 399                return 0;
 400
 401        for (i = 0; i < want_obj.nr; i++) {
 402                struct object *want = want_obj.objects[i].item;
 403
 404                if (want->flags & COMMON_KNOWN)
 405                        continue;
 406                want = deref_tag(want, "a want line", 0);
 407                if (!want || want->type != OBJ_COMMIT) {
 408                        /* no way to tell if this is reachable by
 409                         * looking at the ancestry chain alone, so
 410                         * leave a note to ourselves not to worry about
 411                         * this object anymore.
 412                         */
 413                        want_obj.objects[i].item->flags |= COMMON_KNOWN;
 414                        continue;
 415                }
 416                if (!reachable((struct commit *)want))
 417                        return 0;
 418        }
 419        return 1;
 420}
 421
 422static int get_common_commits(void)
 423{
 424        static char line[1000];
 425        unsigned char sha1[20];
 426        char hex[41], last_hex[41];
 427
 428        save_commit_buffer = 0;
 429
 430        for(;;) {
 431                int len = packet_read_line(0, line, sizeof(line));
 432                reset_timeout();
 433
 434                if (!len) {
 435                        if (have_obj.nr == 0 || multi_ack)
 436                                packet_write(1, "NAK\n");
 437                        continue;
 438                }
 439                strip(line, len);
 440                if (!prefixcmp(line, "have ")) {
 441                        switch (got_sha1(line+5, sha1)) {
 442                        case -1: /* they have what we do not */
 443                                if (multi_ack && ok_to_give_up())
 444                                        packet_write(1, "ACK %s continue\n",
 445                                                     sha1_to_hex(sha1));
 446                                break;
 447                        default:
 448                                memcpy(hex, sha1_to_hex(sha1), 41);
 449                                if (multi_ack) {
 450                                        const char *msg = "ACK %s continue\n";
 451                                        packet_write(1, msg, hex);
 452                                        memcpy(last_hex, hex, 41);
 453                                }
 454                                else if (have_obj.nr == 1)
 455                                        packet_write(1, "ACK %s\n", hex);
 456                                break;
 457                        }
 458                        continue;
 459                }
 460                if (!strcmp(line, "done")) {
 461                        if (have_obj.nr > 0) {
 462                                if (multi_ack)
 463                                        packet_write(1, "ACK %s\n", last_hex);
 464                                return 0;
 465                        }
 466                        packet_write(1, "NAK\n");
 467                        return -1;
 468                }
 469                die("git upload-pack: expected SHA1 list, got '%s'", line);
 470        }
 471}
 472
 473static void receive_needs(void)
 474{
 475        struct object_array shallows = {0, 0, NULL};
 476        static char line[1000];
 477        int len, depth = 0;
 478
 479        shallow_nr = 0;
 480        if (debug_fd)
 481                write_in_full(debug_fd, "#S\n", 3);
 482        for (;;) {
 483                struct object *o;
 484                unsigned char sha1_buf[20];
 485                len = packet_read_line(0, line, sizeof(line));
 486                reset_timeout();
 487                if (!len)
 488                        break;
 489                if (debug_fd)
 490                        write_in_full(debug_fd, line, len);
 491
 492                if (!prefixcmp(line, "shallow ")) {
 493                        unsigned char sha1[20];
 494                        struct object *object;
 495                        use_thin_pack = 0;
 496                        if (get_sha1(line + 8, sha1))
 497                                die("invalid shallow line: %s", line);
 498                        object = parse_object(sha1);
 499                        if (!object)
 500                                die("did not find object for %s", line);
 501                        object->flags |= CLIENT_SHALLOW;
 502                        add_object_array(object, NULL, &shallows);
 503                        continue;
 504                }
 505                if (!prefixcmp(line, "deepen ")) {
 506                        char *end;
 507                        use_thin_pack = 0;
 508                        depth = strtol(line + 7, &end, 0);
 509                        if (end == line + 7 || depth <= 0)
 510                                die("Invalid deepen: %s", line);
 511                        continue;
 512                }
 513                if (prefixcmp(line, "want ") ||
 514                    get_sha1_hex(line+5, sha1_buf))
 515                        die("git upload-pack: protocol error, "
 516                            "expected to get sha, not '%s'", line);
 517                if (strstr(line+45, "multi_ack"))
 518                        multi_ack = 1;
 519                if (strstr(line+45, "thin-pack"))
 520                        use_thin_pack = 1;
 521                if (strstr(line+45, "ofs-delta"))
 522                        use_ofs_delta = 1;
 523                if (strstr(line+45, "side-band-64k"))
 524                        use_sideband = LARGE_PACKET_MAX;
 525                else if (strstr(line+45, "side-band"))
 526                        use_sideband = DEFAULT_PACKET_MAX;
 527                if (strstr(line+45, "no-progress"))
 528                        no_progress = 1;
 529                if (strstr(line+45, "include-tag"))
 530                        use_include_tag = 1;
 531
 532                /* We have sent all our refs already, and the other end
 533                 * should have chosen out of them; otherwise they are
 534                 * asking for nonsense.
 535                 *
 536                 * Hmph.  We may later want to allow "want" line that
 537                 * asks for something like "master~10" (symbolic)...
 538                 * would it make sense?  I don't know.
 539                 */
 540                o = lookup_object(sha1_buf);
 541                if (!o || !(o->flags & OUR_REF))
 542                        die("git upload-pack: not our ref %s", line+5);
 543                if (!(o->flags & WANTED)) {
 544                        o->flags |= WANTED;
 545                        add_object_array(o, NULL, &want_obj);
 546                }
 547        }
 548        if (debug_fd)
 549                write_in_full(debug_fd, "#E\n", 3);
 550
 551        if (!use_sideband && daemon_mode)
 552                no_progress = 1;
 553
 554        if (depth == 0 && shallows.nr == 0)
 555                return;
 556        if (depth > 0) {
 557                struct commit_list *result, *backup;
 558                int i;
 559                backup = result = get_shallow_commits(&want_obj, depth,
 560                        SHALLOW, NOT_SHALLOW);
 561                while (result) {
 562                        struct object *object = &result->item->object;
 563                        if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
 564                                packet_write(1, "shallow %s",
 565                                                sha1_to_hex(object->sha1));
 566                                register_shallow(object->sha1);
 567                                shallow_nr++;
 568                        }
 569                        result = result->next;
 570                }
 571                free_commit_list(backup);
 572                for (i = 0; i < shallows.nr; i++) {
 573                        struct object *object = shallows.objects[i].item;
 574                        if (object->flags & NOT_SHALLOW) {
 575                                struct commit_list *parents;
 576                                packet_write(1, "unshallow %s",
 577                                        sha1_to_hex(object->sha1));
 578                                object->flags &= ~CLIENT_SHALLOW;
 579                                /* make sure the real parents are parsed */
 580                                unregister_shallow(object->sha1);
 581                                object->parsed = 0;
 582                                if (parse_commit((struct commit *)object))
 583                                        die("invalid commit");
 584                                parents = ((struct commit *)object)->parents;
 585                                while (parents) {
 586                                        add_object_array(&parents->item->object,
 587                                                        NULL, &want_obj);
 588                                        parents = parents->next;
 589                                }
 590                        }
 591                        /* make sure commit traversal conforms to client */
 592                        register_shallow(object->sha1);
 593                }
 594                packet_flush(1);
 595        } else
 596                if (shallows.nr > 0) {
 597                        int i;
 598                        for (i = 0; i < shallows.nr; i++)
 599                                register_shallow(shallows.objects[i].item->sha1);
 600                }
 601
 602        shallow_nr += shallows.nr;
 603        free(shallows.objects);
 604}
 605
 606static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 607{
 608        static const char *capabilities = "multi_ack thin-pack side-band"
 609                " side-band-64k ofs-delta shallow no-progress"
 610                " include-tag";
 611        struct object *o = parse_object(sha1);
 612
 613        if (!o)
 614                die("git upload-pack: cannot find object %s:", sha1_to_hex(sha1));
 615
 616        if (capabilities)
 617                packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
 618                        0, capabilities);
 619        else
 620                packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
 621        capabilities = NULL;
 622        if (!(o->flags & OUR_REF)) {
 623                o->flags |= OUR_REF;
 624                nr_our_refs++;
 625        }
 626        if (o->type == OBJ_TAG) {
 627                o = deref_tag(o, refname, 0);
 628                if (o)
 629                        packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
 630        }
 631        return 0;
 632}
 633
 634static void upload_pack(void)
 635{
 636        reset_timeout();
 637        head_ref(send_ref, NULL);
 638        for_each_ref(send_ref, NULL);
 639        packet_flush(1);
 640        receive_needs();
 641        if (want_obj.nr) {
 642                get_common_commits();
 643                create_pack_file();
 644        }
 645}
 646
 647int main(int argc, char **argv)
 648{
 649        char *dir;
 650        int i;
 651        int strict = 0;
 652
 653        git_extract_argv0_path(argv[0]);
 654        read_replace_refs = 0;
 655
 656        for (i = 1; i < argc; i++) {
 657                char *arg = argv[i];
 658
 659                if (arg[0] != '-')
 660                        break;
 661                if (!strcmp(arg, "--strict")) {
 662                        strict = 1;
 663                        continue;
 664                }
 665                if (!prefixcmp(arg, "--timeout=")) {
 666                        timeout = atoi(arg+10);
 667                        daemon_mode = 1;
 668                        continue;
 669                }
 670                if (!strcmp(arg, "--")) {
 671                        i++;
 672                        break;
 673                }
 674        }
 675
 676        if (i != argc-1)
 677                usage(upload_pack_usage);
 678
 679        setup_path();
 680
 681        dir = argv[i];
 682
 683        if (!enter_repo(dir, strict))
 684                die("'%s' does not appear to be a git repository", dir);
 685        if (is_repository_shallow())
 686                die("attempt to fetch/clone from a shallow repository");
 687        if (getenv("GIT_DEBUG_SEND_PACK"))
 688                debug_fd = atoi(getenv("GIT_DEBUG_SEND_PACK"));
 689        upload_pack();
 690        return 0;
 691}