lib/repository_admin.c

   1 /*
   2  * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
   3  *
   4  * Permission to use, copy, modify, and distribute this software for any
   5  * purpose with or without fee is hereby granted, provided that the above
   6  * copyright notice and this permission notice appear in all copies.
   7  *
   8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
   9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  15  */
  16
  17 #include "got_compat.h"
  18
  19 #include <sys/types.h>
  20 #include <sys/queue.h>
  21 #include <sys/uio.h>
  22 #include <sys/stat.h>
  23 #include <sys/socket.h>
  24 #include <sys/wait.h>
  25
  26 #include <dirent.h>
  27 #include <errno.h>
  28 #include <fcntl.h>
  29 #include <stdint.h>
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <limits.h>
  34 #include <unistd.h>
  35
  36 #include "got_error.h"
  37 #include "got_cancel.h"
  38 #include "got_object.h"
  39 #include "got_reference.h"
  40 #include "got_repository.h"
  41 #include "got_repository_admin.h"
  42 #include "got_opentemp.h"
  43 #include "got_path.h"
  44
  45 #include "got_lib_delta.h"
  46 #include "got_lib_hash.h"
  47 #include "got_lib_object.h"
  48 #include "got_lib_object_idset.h"
  49 #include "got_lib_object_cache.h"
  50 #include "got_lib_pack.h"
  51 #include "got_lib_privsep.h"
  52 #include "got_lib_repository.h"
  53 #include "got_lib_ratelimit.h"
  54 #include "got_lib_pack_create.h"
  55 #include "got_lib_lockfile.h"
  56
  57 #ifndef nitems
  58 #define nitems(_a)      (sizeof((_a)) / sizeof((_a)[0]))
  59 #endif
  60
  61 static const struct got_error *
  62 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
  63     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
  64     struct got_repository *repo,
  65     got_cancel_cb cancel_cb, void *cancel_arg)
  66 {
  67         const struct got_error *err = NULL;
  68         const size_t alloc_chunksz = 256;
  69         size_t nalloc;
  70         struct got_reflist_entry *re;
  71         int i;
  72
  73         *ids = NULL;
  74         *nobjects = 0;
  75
  76         err = got_reflist_sort(refs,
  77             got_ref_cmp_by_commit_timestamp_descending, repo);
  78         if (err)
  79                 return err;
  80
  81         *ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
  82         if (*ids == NULL)
  83                 return got_error_from_errno("reallocarray");
  84         nalloc = alloc_chunksz;
  85
  86         TAILQ_FOREACH(re, refs, entry) {
  87                 struct got_object_id *id;
  88
  89                 if (cancel_cb) {
  90                         err = cancel_cb(cancel_arg);
  91                         if (err)
  92                                 goto done;
  93                 }
  94
  95                 err = got_ref_resolve(&id, repo, re->ref);
  96                 if (err)
  97                         goto done;
  98
  99                 if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
 100                         int obj_type;
 101                         err = got_object_get_type(&obj_type, repo, id);
 102                         if (err)
 103                                 goto done;
 104                         if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
 105                                 free(id);
 106                                 id = NULL;
 107                                 continue;
 108                         }
 109                 }
 110
 111                 if (nalloc <= *nobjects) {
 112                         struct got_object_id **new;
 113                         new = recallocarray(*ids, nalloc,
 114                             nalloc + alloc_chunksz,
 115                             sizeof(struct got_object_id *));
 116                         if (new == NULL) {
 117                                 err = got_error_from_errno(
 118                                     "recallocarray");
 119                                 goto done;
 120                         }
 121                         *ids = new;
 122                         nalloc += alloc_chunksz;
 123                 }
 124                 (*ids)[*nobjects] = id;
 125                 if ((*ids)[*nobjects] == NULL) {
 126                         err = got_error_from_errno("got_object_id_dup");
 127                         goto done;
 128                 }
 129                 (*nobjects)++;
 130         }
 131 done:
 132         if (err) {
 133                 for (i = 0; i < *nobjects; i++)
 134                         free((*ids)[i]);
 135                 free(*ids);
 136                 *ids = NULL;
 137                 *nobjects = 0;
 138         }
 139         return err;
 140 }
 141
 142 static const struct got_error *
 143 create_temp_packfile(int *packfd, char **tmpfile_path,
 144     struct got_repository *repo)
 145 {
 146         const struct got_error *err = NULL;
 147         char *path;
 148
 149         *packfd = -1;
 150
 151         if (asprintf(&path, "%s/%s/packing.pack",
 152             got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1)
 153                 return got_error_from_errno("asprintf");
 154
 155         err = got_opentemp_named_fd(tmpfile_path, packfd, path, "");
 156         if (err)
 157                 goto done;
 158
 159         if (fchmod(*packfd, GOT_DEFAULT_PACK_MODE) == -1)
 160                 err = got_error_from_errno2("fchmod", *tmpfile_path);
 161 done:
 162         free(path);
 163         if (err) {
 164                 if (*packfd != -1)
 165                         close(*packfd);
 166                 *packfd = -1;
 167                 free(*tmpfile_path);
 168                 *tmpfile_path = NULL;
 169         }
 170         return err;
 171 }
 172
 173 static const struct got_error *
 174 install_packfile(FILE **packfile, int *packfd, char **packfile_path,
 175     char **tmpfile_path, struct got_object_id *pack_hash,
 176     struct got_repository *repo)
 177 {
 178         const struct got_error *err;
 179         char *hash_str;
 180
 181         err = got_object_id_str(&hash_str, pack_hash);
 182         if (err)
 183                 return err;
 184
 185         if (asprintf(packfile_path, "%s/%s/pack-%s.pack",
 186             got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
 187             hash_str) == -1) {
 188                 err = got_error_from_errno("asprintf");
 189                 goto done;
 190         }
 191
 192         if (lseek(*packfd, 0L, SEEK_SET) == -1) {
 193                 err = got_error_from_errno("lseek");
 194                 goto done;
 195         }
 196
 197         if (rename(*tmpfile_path, *packfile_path) == -1) {
 198                 err = got_error_from_errno3("rename", *tmpfile_path,
 199                     *packfile_path);
 200                 goto done;
 201         }
 202
 203         free(*tmpfile_path);
 204         *tmpfile_path = NULL;
 205
 206         *packfile = fdopen(*packfd, "w");
 207         if (*packfile == NULL) {
 208                 err = got_error_from_errno2("fdopen", *packfile_path);
 209                 goto done;
 210         }
 211         *packfd = -1;
 212 done:
 213         free(hash_str);
 214         return err;
 215 }
 216
 217 const struct got_error *
 218 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
 219     struct got_reflist_head *include_refs,
 220     struct got_reflist_head *exclude_refs, struct got_repository *repo,
 221     int loose_obj_only, int force_refdelta,
 222     got_pack_progress_cb progress_cb, void *progress_arg,
 223     got_cancel_cb cancel_cb, void *cancel_arg)
 224 {
 225         const struct got_error *err = NULL;
 226         struct got_object_id **ours = NULL, **theirs = NULL;
 227         int nours = 0, ntheirs = 0, packfd = -1, i;
 228         char *tmpfile_path = NULL, *packfile_path = NULL;
 229         FILE *delta_cache = NULL;
 230         struct got_ratelimit rl;
 231
 232         *packfile = NULL;
 233         *pack_hash = NULL;
 234
 235         got_ratelimit_init(&rl, 0, 500);
 236
 237         err = create_temp_packfile(&packfd, &tmpfile_path, repo);
 238         if (err)
 239                 return err;
 240
 241         delta_cache = got_opentemp();
 242         if (delta_cache == NULL) {
 243                 err = got_error_from_errno("got_opentemp");
 244                 goto done;
 245         }
 246
 247         err = get_reflist_object_ids(&ours, &nours,
 248             (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
 249             include_refs, repo, cancel_cb, cancel_arg);
 250         if (err)
 251                 goto done;
 252
 253         if (nours == 0) {
 254                 err = got_error(GOT_ERR_CANNOT_PACK);
 255                 goto done;
 256         }
 257
 258         if (!TAILQ_EMPTY(exclude_refs)) {
 259                 err = get_reflist_object_ids(&theirs, &ntheirs,
 260                     (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
 261                     exclude_refs, repo,
 262                     cancel_cb, cancel_arg);
 263                 if (err)
 264                         goto done;
 265         }
 266
 267         *pack_hash = calloc(1, sizeof(**pack_hash));
 268         if (*pack_hash == NULL) {
 269                 err = got_error_from_errno("calloc");
 270                 goto done;
 271         }
 272         err = got_pack_create(*pack_hash, packfd, delta_cache,
 273             theirs, ntheirs, ours, nours, repo, loose_obj_only,
 274             0, force_refdelta, progress_cb, progress_arg, &rl,
 275             cancel_cb, cancel_arg);
 276         if (err)
 277                 goto done;
 278
 279         err = install_packfile(packfile, &packfd, &packfile_path,
 280             &tmpfile_path, *pack_hash, repo);
 281 done:
 282         for (i = 0; i < nours; i++)
 283                 free(ours[i]);
 284         free(ours);
 285         for (i = 0; i < ntheirs; i++)
 286                 free(theirs[i]);
 287         free(theirs);
 288         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 289                 err = got_error_from_errno2("close",
 290                     packfile_path ? packfile_path : tmpfile_path);
 291         if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
 292                 err = got_error_from_errno("fclose");
 293         if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
 294                 err = got_error_from_errno2("unlink", tmpfile_path);
 295         free(tmpfile_path);
 296         free(packfile_path);
 297         if (err) {
 298                 free(*pack_hash);
 299                 *pack_hash = NULL;
 300                 if (*packfile)
 301                         fclose(*packfile);
 302                 *packfile = NULL;
 303         }
 304         return err;
 305 }
 306
 307 const struct got_error *
 308 got_repo_index_pack(char **idxpath, FILE *packfile,
 309     struct got_object_id *pack_hash, struct got_repository *repo,
 310     got_pack_index_progress_cb progress_cb, void *progress_arg,
 311     got_cancel_cb cancel_cb, void *cancel_arg)
 312 {
 313         size_t i;
 314         char *path;
 315         int imsg_idxfds[2];
 316         int npackfd = -1, idxfd = -1, nidxfd = -1;
 317         int tmpfds[3];
 318         int idxstatus, done = 0;
 319         int nobj_total = 0, nobj_indexed = 0, nobj_loose = 0, nobj_resolved = 0;
 320         const struct got_error *err;
 321         struct imsgbuf idxibuf;
 322         pid_t idxpid;
 323         char *tmpidxpath = NULL;
 324         char *packfile_path = NULL, *id_str = NULL;
 325         const char *repo_path = got_repo_get_path_git_dir(repo);
 326         struct stat sb;
 327
 328         *idxpath = NULL;
 329         memset(&idxibuf, 0, sizeof(idxibuf));
 330
 331         for (i = 0; i < nitems(tmpfds); i++)
 332                 tmpfds[i] = -1;
 333
 334         if (asprintf(&path, "%s/%s/indexing.idx",
 335             repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 336                 err = got_error_from_errno("asprintf");
 337                 goto done;
 338         }
 339         err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
 340         free(path);
 341         if (err)
 342                 goto done;
 343         if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
 344                 err = got_error_from_errno2("fchmod", tmpidxpath);
 345                 goto done;
 346         }
 347
 348         nidxfd = dup(idxfd);
 349         if (nidxfd == -1) {
 350                 err = got_error_from_errno("dup");
 351                 goto done;
 352         }
 353
 354         for (i = 0; i < nitems(tmpfds); i++) {
 355                 tmpfds[i] = got_opentempfd();
 356                 if (tmpfds[i] == -1) {
 357                         err = got_error_from_errno("got_opentempfd");
 358                         goto done;
 359                 }
 360         }
 361
 362         err = got_object_id_str(&id_str, pack_hash);
 363         if (err)
 364                 goto done;
 365
 366         if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
 367             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 368                 err = got_error_from_errno("asprintf");
 369                 goto done;
 370         }
 371
 372         if (fstat(fileno(packfile), &sb) == -1) {
 373                 err = got_error_from_errno2("fstat", packfile_path);
 374                 goto done;
 375         }
 376
 377         if (asprintf(idxpath, "%s/%s/pack-%s.idx",
 378             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 379                 err = got_error_from_errno("asprintf");
 380                 goto done;
 381         }
 382
 383         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
 384                 err = got_error_from_errno("socketpair");
 385                 goto done;
 386         }
 387         idxpid = fork();
 388         if (idxpid == -1) {
 389                 err= got_error_from_errno("fork");
 390                 goto done;
 391         } else if (idxpid == 0)
 392                 got_privsep_exec_child(imsg_idxfds,
 393                     GOT_PATH_PROG_INDEX_PACK, packfile_path);
 394         if (close(imsg_idxfds[1]) == -1) {
 395                 err = got_error_from_errno("close");
 396                 goto done;
 397         }
 398         if (imsgbuf_init(&idxibuf, imsg_idxfds[0]) == -1) {
 399                 err = got_error_from_errno("imsgbuf_init");
 400                 goto done;
 401         }
 402         imsgbuf_allow_fdpass(&idxibuf);
 403
 404         npackfd = dup(fileno(packfile));
 405         if (npackfd == -1) {
 406                 err = got_error_from_errno("dup");
 407                 goto done;
 408         }
 409         err = got_privsep_send_index_pack_req(&idxibuf, pack_hash, npackfd);
 410         if (err != NULL)
 411                 goto done;
 412         npackfd = -1;
 413         err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
 414         if (err != NULL)
 415                 goto done;
 416         nidxfd = -1;
 417         for (i = 0; i < nitems(tmpfds); i++) {
 418                 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
 419                 if (err != NULL)
 420                         goto done;
 421                 tmpfds[i] = -1;
 422         }
 423         done = 0;
 424         while (!done) {
 425                 if (cancel_cb) {
 426                         err = cancel_cb(cancel_arg);
 427                         if (err)
 428                                 goto done;
 429                 }
 430
 431                 err = got_privsep_recv_index_progress(&done, &nobj_total,
 432                     &nobj_indexed, &nobj_loose, &nobj_resolved,
 433                     &idxibuf);
 434                 if (err != NULL)
 435                         goto done;
 436                 if (nobj_indexed != 0) {
 437                         err = progress_cb(progress_arg, sb.st_size,
 438                             nobj_total, nobj_indexed, nobj_loose,
 439                             nobj_resolved, 0);
 440                         if (err)
 441                                 break;
 442                 }
 443         }
 444         if (done) {
 445                 err = progress_cb(progress_arg, sb.st_size,
 446                     nobj_total, nobj_indexed, nobj_loose,
 447                     nobj_resolved, done);
 448                 if (err)
 449                         goto done;
 450         }
 451         if (close(imsg_idxfds[0]) == -1) {
 452                 err = got_error_from_errno("close");
 453                 goto done;
 454         }
 455         if (waitpid(idxpid, &idxstatus, 0) == -1) {
 456                 err = got_error_from_errno("waitpid");
 457                 goto done;
 458         }
 459
 460         if (rename(tmpidxpath, *idxpath) == -1) {
 461                 err = got_error_from_errno3("rename", tmpidxpath, *idxpath);
 462                 goto done;
 463         }
 464         free(tmpidxpath);
 465         tmpidxpath = NULL;
 466
 467 done:
 468         if (idxibuf.w)
 469                 imsgbuf_clear(&idxibuf);
 470         if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
 471                 err = got_error_from_errno2("unlink", tmpidxpath);
 472         if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
 473                 err = got_error_from_errno("close");
 474         if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
 475                 err = got_error_from_errno("close");
 476         for (i = 0; i < nitems(tmpfds); i++) {
 477                 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
 478                         err = got_error_from_errno("close");
 479         }
 480         free(tmpidxpath);
 481         free(packfile_path);
 482         return err;
 483 }
 484
 485 const struct got_error *
 486 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
 487     struct got_repository *repo, const char *packfile_path)
 488 {
 489         const struct got_error *err = NULL;
 490         const char *packdir_path = NULL;
 491         char *packfile_name = NULL, *p, *dot;
 492         struct got_object_id id;
 493         int packfd = -1;
 494
 495         *packfile = NULL;
 496         *pack_hash = NULL;
 497
 498         packdir_path = got_repo_get_path_objects_pack(repo);
 499         if (packdir_path == NULL)
 500                 return got_error_from_errno("got_repo_get_path_objects_pack");
 501
 502         if (!got_path_is_child(packfile_path, packdir_path,
 503             strlen(packdir_path))) {
 504                 err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
 505                 goto done;
 506
 507         }
 508
 509         err = got_path_basename(&packfile_name, packfile_path);
 510         if (err)
 511                 goto done;
 512         p = packfile_name;
 513
 514         if (strncmp(p, "pack-", 5) != 0) {
 515                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 516                     "'%s' is not a valid pack file name",
 517                     packfile_name);
 518                 goto done;
 519         }
 520         p += 5;
 521         dot = strchr(p, '.');
 522         if (dot == NULL) {
 523                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 524                     "'%s' is not a valid pack file name",
 525                     packfile_name);
 526                 goto done;
 527         }
 528         if (strcmp(dot + 1, "pack") != 0) {
 529                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 530                     "'%s' is not a valid pack file name",
 531                     packfile_name);
 532                 goto done;
 533         }
 534         *dot = '\0';
 535         if (!got_parse_object_id(&id, p, repo->algo)) {
 536                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 537                     "'%s' is not a valid pack file name",
 538                     packfile_name);
 539                 goto done;
 540         }
 541
 542         *pack_hash = got_object_id_dup(&id);
 543         if (*pack_hash == NULL) {
 544                 err = got_error_from_errno("got_object_id_dup");
 545                 goto done;
 546         }
 547
 548         packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
 549         if (packfd == -1) {
 550                 err = got_error_from_errno2("open", packfile_path);
 551                 goto done;
 552         }
 553
 554         *packfile = fdopen(packfd, "r");
 555         if (*packfile == NULL) {
 556                 err = got_error_from_errno2("fdopen", packfile_path);
 557                 goto done;
 558         }
 559         packfd = -1;
 560 done:
 561         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 562                 err = got_error_from_errno2("close", packfile_path);
 563         free(packfile_name);
 564         if (err) {
 565                 free(*pack_hash);
 566                 *pack_hash = NULL;
 567         }
 568         return err;
 569 }
 570
 571 const struct got_error *
 572 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
 573     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
 574     got_cancel_cb cancel_cb, void *cancel_arg)
 575 {
 576         const struct got_error *err = NULL;
 577         char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
 578         struct got_packidx *packidx = NULL;
 579         struct got_pack *pack = NULL;
 580         uint32_t nobj, i;
 581         size_t digest_len = got_hash_digest_length(repo->algo);
 582
 583         err = got_object_id_str(&id_str, pack_hash);
 584         if (err)
 585                 goto done;
 586
 587         if (asprintf(&packpath, "%s/pack-%s.pack",
 588             GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 589                 err = got_error_from_errno("asprintf");
 590                 goto done;
 591         }
 592         if (asprintf(&idxpath, "%s/pack-%s.idx",
 593             GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 594                 err = got_error_from_errno("asprintf");
 595                 goto done;
 596         }
 597
 598         err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1,
 599             repo->algo);
 600         if (err)
 601                 goto done;
 602
 603         err = got_repo_cache_pack(&pack, repo, packpath, packidx);
 604         if (err)
 605                 goto done;
 606
 607         nobj = be32toh(packidx->hdr.fanout_table[0xff]);
 608         for (i = 0; i < nobj; i++) {
 609                 uint8_t *oid;
 610                 struct got_object_id id, base_id;
 611                 off_t offset, base_offset = 0;
 612                 uint8_t type;
 613                 uint64_t size;
 614                 size_t tslen, len;
 615
 616                 if (cancel_cb) {
 617                         err = cancel_cb(cancel_arg);
 618                         if (err)
 619                                 break;
 620                 }
 621                 oid = packidx->hdr.sorted_ids + i * digest_len;
 622                 id.algo = repo->algo;
 623                 memcpy(id.hash, oid, digest_len);
 624
 625                 offset = got_packidx_get_object_offset(packidx, i);
 626                 if (offset == -1) {
 627                         err = got_error(GOT_ERR_BAD_PACKIDX);
 628                         goto done;
 629                 }
 630
 631                 err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
 632                     pack, offset);
 633                 if (err)
 634                         goto done;
 635
 636                 switch (type) {
 637                 case GOT_OBJ_TYPE_OFFSET_DELTA:
 638                         err = got_pack_parse_offset_delta(&base_offset, &len,
 639                             pack, offset, tslen);
 640                         if (err)
 641                                 goto done;
 642                         break;
 643                 case GOT_OBJ_TYPE_REF_DELTA:
 644                         err = got_pack_parse_ref_delta(&base_id,
 645                             pack, offset, tslen);
 646                         if (err)
 647                                 goto done;
 648                         break;
 649                 }
 650                 err = (*list_cb)(list_arg, &id, type, offset, size,
 651                     base_offset, &base_id);
 652                 if (err)
 653                         goto done;
 654         }
 655
 656 done:
 657         free(id_str);
 658         free(idxpath);
 659         free(packpath);
 660         if (packidx)
 661                 got_packidx_close(packidx);
 662         return err;
 663 }
 664
 665 static const struct got_error *
 666 repo_cleanup_lock(struct got_repository *repo, struct got_lockfile **lk)
 667 {
 668         const struct got_error *err;
 669         char myname[_POSIX_HOST_NAME_MAX + 1];
 670
 671         if (gethostname(myname, sizeof(myname)) == -1)
 672                 return got_error_from_errno("gethostname");
 673
 674         err = got_lockfile_lock(lk, "gc.pid", got_repo_get_fd(repo));
 675         if (err)
 676                 return err;
 677
 678         /*
 679          * Git uses these info to provide some verbiage when finds a
 680          * lock during `git gc --force' so don't try too hard to avoid
 681          * short writes and don't care if a race happens between the
 682          * lockfile creation and the write itself.
 683          */
 684         if (dprintf((*lk)->fd, "%d %s", getpid(), myname) < 0)
 685                 return got_error_from_errno("dprintf");
 686
 687         return NULL;
 688 }
 689
 690 static const struct got_error *
 691 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
 692     void *progress_arg, struct got_ratelimit *rl,
 693     int ncommits, int nloose, int npurged, int nredundant)
 694 {
 695         const struct got_error *err;
 696         int elapsed;
 697
 698         if (progress_cb == NULL)
 699                 return NULL;
 700
 701         err = got_ratelimit_check(&elapsed, rl);
 702         if (err || !elapsed)
 703                 return err;
 704
 705         return progress_cb(progress_arg, ncommits, nloose, npurged,
 706             nredundant);
 707 }
 708
 709 static const struct got_error *
 710 get_loose_object_ids(struct got_object_idset **loose_ids,
 711     off_t *ondisk_size, int ncommits,
 712     got_cleanup_progress_cb progress_cb, void *progress_arg,
 713     struct got_ratelimit *rl, struct got_repository *repo)
 714 {
 715         const struct got_error *err = NULL;
 716         char *path_objects = NULL, *path = NULL;
 717         DIR *dir = NULL;
 718         struct got_object *obj = NULL;
 719         struct got_object_id id;
 720         int i, fd = -1;
 721         struct stat sb;
 722
 723         *ondisk_size = 0;
 724         *loose_ids = got_object_idset_alloc();
 725         if (*loose_ids == NULL)
 726                 return got_error_from_errno("got_object_idset_alloc");
 727
 728         path_objects = got_repo_get_path_objects(repo);
 729         if (path_objects == NULL) {
 730                 err = got_error_from_errno("got_repo_get_path_objects");
 731                 goto done;
 732         }
 733
 734         for (i = 0; i <= 0xff; i++) {
 735                 struct dirent *dent;
 736
 737                 if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
 738                         err = got_error_from_errno("asprintf");
 739                         break;
 740                 }
 741
 742                 dir = opendir(path);
 743                 if (dir == NULL) {
 744                         if (errno == ENOENT) {
 745                                 err = NULL;
 746                                 continue;
 747                         }
 748                         err = got_error_from_errno2("opendir", path);
 749                         break;
 750                 }
 751
 752                 while ((dent = readdir(dir)) != NULL) {
 753                         char *id_str;
 754
 755                         if (strcmp(dent->d_name, ".") == 0 ||
 756                             strcmp(dent->d_name, "..") == 0)
 757                                 continue;
 758
 759                         if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
 760                                 err = got_error_from_errno("asprintf");
 761                                 goto done;
 762                         }
 763
 764                         if (!got_parse_object_id(&id, id_str, repo->algo)) {
 765                                 free(id_str);
 766                                 continue;
 767                         }
 768                         free(id_str);
 769
 770                         err = got_object_open_loose_fd(&fd, &id, repo);
 771                         if (err)
 772                                 goto done;
 773                         if (fstat(fd, &sb) == -1) {
 774                                 err = got_error_from_errno("fstat");
 775                                 goto done;
 776                         }
 777                         err = got_object_read_header_privsep(&obj, &id, repo,
 778                             fd);
 779                         if (err)
 780                                 goto done;
 781                         fd = -1; /* already closed */
 782
 783                         switch (obj->type) {
 784                         case GOT_OBJ_TYPE_COMMIT:
 785                         case GOT_OBJ_TYPE_TREE:
 786                         case GOT_OBJ_TYPE_BLOB:
 787                         case GOT_OBJ_TYPE_TAG:
 788                                 break;
 789                         default:
 790                                 err = got_error_fmt(GOT_ERR_OBJ_TYPE,
 791                                     "%d", obj->type);
 792                                 goto done;
 793                         }
 794                         got_object_close(obj);
 795                         obj = NULL;
 796                         (*ondisk_size) += sb.st_size;
 797                         err = got_object_idset_add(*loose_ids, &id, NULL);
 798                         if (err)
 799                                 goto done;
 800                         err = report_cleanup_progress(progress_cb,
 801                             progress_arg, rl, ncommits,
 802                             got_object_idset_num_elements(*loose_ids),
 803                             -1, -1);
 804                         if (err)
 805                                 goto done;
 806                 }
 807
 808                 if (closedir(dir) != 0) {
 809                         err = got_error_from_errno("closedir");
 810                         goto done;
 811                 }
 812                 dir = NULL;
 813
 814                 free(path);
 815                 path = NULL;
 816         }
 817 done:
 818         if (dir && closedir(dir) != 0 && err == NULL)
 819                 err = got_error_from_errno("closedir");
 820         if (fd != -1 && close(fd) == -1 && err == NULL)
 821                 err = got_error_from_errno("close");
 822         if (err) {
 823                 got_object_idset_free(*loose_ids);
 824                 *loose_ids = NULL;
 825         }
 826         if (obj)
 827                 got_object_close(obj);
 828         free(path_objects);
 829         free(path);
 830         return err;
 831 }
 832
 833 static const struct got_error *
 834 load_tree_entries(struct got_object_id_queue *ids,
 835     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
 836     const char *dpath, struct got_repository *repo,
 837     got_cancel_cb cancel_cb, void *cancel_arg)
 838 {
 839         const struct got_error *err;
 840         struct got_tree_object *tree;
 841         char *p = NULL;
 842         int i;
 843
 844         err = got_object_open_as_tree(&tree, repo, tree_id);
 845         if (err)
 846                 return err;
 847
 848         for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
 849                 struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
 850                 struct got_object_id *id = got_tree_entry_get_id(e);
 851                 mode_t mode = got_tree_entry_get_mode(e);
 852
 853                 if (cancel_cb) {
 854                         err = (*cancel_cb)(cancel_arg);
 855                         if (err)
 856                                 break;
 857                 }
 858
 859                 if (got_object_tree_entry_is_symlink(e) ||
 860                     got_object_tree_entry_is_submodule(e) ||
 861                     got_object_idset_contains(traversed_ids, id))
 862                         continue;
 863
 864                 if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
 865                     got_tree_entry_get_name(e)) == -1) {
 866                         err = got_error_from_errno("asprintf");
 867                         break;
 868                 }
 869
 870                 if (S_ISDIR(mode)) {
 871                         struct got_object_qid *qid;
 872                         err = got_object_qid_alloc(&qid, id);
 873                         if (err)
 874                                 break;
 875                         STAILQ_INSERT_TAIL(ids, qid, entry);
 876                 } else if (S_ISREG(mode)) {
 877                         /* This blob is referenced. */
 878                         err = got_object_idset_add(traversed_ids, id, NULL);
 879                         if (err)
 880                                 break;
 881                 }
 882                 free(p);
 883                 p = NULL;
 884         }
 885
 886         got_object_tree_close(tree);
 887         free(p);
 888         return err;
 889 }
 890
 891 static const struct got_error *
 892 load_tree(struct got_object_idset *traversed_ids,
 893     struct got_object_id *tree_id,
 894     const char *dpath, struct got_repository *repo,
 895     got_cancel_cb cancel_cb, void *cancel_arg)
 896 {
 897         const struct got_error *err = NULL;
 898         struct got_object_id_queue tree_ids;
 899         struct got_object_qid *qid;
 900
 901         err = got_object_qid_alloc(&qid, tree_id);
 902         if (err)
 903                 return err;
 904
 905         STAILQ_INIT(&tree_ids);
 906         STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
 907
 908         while (!STAILQ_EMPTY(&tree_ids)) {
 909                 if (cancel_cb) {
 910                         err = (*cancel_cb)(cancel_arg);
 911                         if (err)
 912                                 break;
 913                 }
 914
 915                 qid = STAILQ_FIRST(&tree_ids);
 916                 STAILQ_REMOVE_HEAD(&tree_ids, entry);
 917
 918                 if (got_object_idset_contains(traversed_ids, &qid->id)) {
 919                         got_object_qid_free(qid);
 920                         continue;
 921                 }
 922
 923                 err = got_object_idset_add(traversed_ids, &qid->id, NULL);
 924                 if (err) {
 925                         got_object_qid_free(qid);
 926                         break;
 927                 }
 928
 929                 err = load_tree_entries(&tree_ids, traversed_ids,
 930                     &qid->id, dpath, repo, cancel_cb, cancel_arg);
 931                 got_object_qid_free(qid);
 932                 if (err)
 933                         break;
 934         }
 935
 936         got_object_id_queue_free(&tree_ids);
 937         return err;
 938 }
 939
 940 static const struct got_error *
 941 load_commit_or_tag(int *ncommits, struct got_object_idset *traversed_ids,
 942     struct got_object_id *id, struct got_repository *repo,
 943     got_cleanup_progress_cb progress_cb, void *progress_arg,
 944     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 945 {
 946         const struct got_error *err;
 947         struct got_commit_object *commit = NULL;
 948         struct got_tag_object *tag = NULL;
 949         struct got_object_id *tree_id = NULL;
 950         struct got_object_id_queue ids;
 951         struct got_object_qid *qid;
 952         int obj_type;
 953
 954         err = got_object_qid_alloc(&qid, id);
 955         if (err)
 956                 return err;
 957
 958         STAILQ_INIT(&ids);
 959         STAILQ_INSERT_TAIL(&ids, qid, entry);
 960
 961         while (!STAILQ_EMPTY(&ids)) {
 962                 if (cancel_cb) {
 963                         err = (*cancel_cb)(cancel_arg);
 964                         if (err)
 965                                 break;
 966                 }
 967
 968                 qid = STAILQ_FIRST(&ids);
 969                 STAILQ_REMOVE_HEAD(&ids, entry);
 970
 971                 if (got_object_idset_contains(traversed_ids, &qid->id)) {
 972                         got_object_qid_free(qid);
 973                         qid = NULL;
 974                         continue;
 975                 }
 976
 977                 err = got_object_idset_add(traversed_ids, &qid->id, NULL);
 978                 if (err)
 979                         break;
 980
 981                 err = got_object_get_type(&obj_type, repo, &qid->id);
 982                 if (err)
 983                         break;
 984                 switch (obj_type) {
 985                 case GOT_OBJ_TYPE_COMMIT:
 986                         err = got_object_open_as_commit(&commit, repo,
 987                             &qid->id);
 988                         if (err)
 989                                 goto done;
 990                         tree_id = got_object_commit_get_tree_id(commit);
 991                         break;
 992                 case GOT_OBJ_TYPE_TAG:
 993                         err = got_object_open_as_tag(&tag, repo, &qid->id);
 994                         if (err)
 995                                 goto done;
 996                         /* tree_id will be set below */
 997                         break;
 998                 case GOT_OBJ_TYPE_TREE:
 999                         tree_id = &qid->id;
1000                         break;
1001                 case GOT_OBJ_TYPE_BLOB:
1002                         tree_id = NULL;
1003                         break;
1004                 default:
1005                         /* should not happen */
1006                         err = got_error(GOT_ERR_OBJ_TYPE);
1007                         goto done;
1008                 }
1009
1010                 if (tag) {
1011                         struct got_object_id *id;
1012
1013                         obj_type = got_object_tag_get_object_type(tag);
1014                         while (obj_type == GOT_OBJ_TYPE_TAG) {
1015                                 struct got_tag_object *next_tag;
1016
1017                                 id = got_object_tag_get_object_id(tag);
1018                                 if (!got_object_idset_contains(traversed_ids,
1019                                     id)) {
1020                                         err = got_object_idset_add(
1021                                             traversed_ids, id, NULL);
1022                                         if (err)
1023                                                 goto done;
1024                                 }
1025
1026                                 err = got_object_open_as_tag(&next_tag, repo,
1027                                     id);
1028                                 if (err)
1029                                         goto done;
1030
1031                                 got_object_tag_close(tag);
1032                                 tag = next_tag;
1033                                 obj_type = got_object_tag_get_object_type(tag);
1034                         }
1035                         id = got_object_tag_get_object_id(tag);
1036                         switch (obj_type) {
1037                         case GOT_OBJ_TYPE_COMMIT:
1038                                 err = got_object_open_as_commit(&commit, repo,
1039                                     id);
1040                                 if (err)
1041                                         goto done;
1042                                 tree_id = got_object_commit_get_tree_id(commit);
1043                                 break;
1044                         case GOT_OBJ_TYPE_TREE:
1045                                 tree_id = id;
1046                                 break;
1047                         case GOT_OBJ_TYPE_BLOB:
1048                                 if (got_object_idset_contains(traversed_ids,
1049                                     id))
1050                                         break;
1051                                 err = got_object_idset_add(traversed_ids, id,
1052                                     NULL);
1053                                 if (err)
1054                                         goto done;
1055                                 break;
1056                         default:
1057                                 /* should not happen */
1058                                 err = got_error(GOT_ERR_OBJ_TYPE);
1059                                 goto done;
1060                         }
1061                 } else if (tree_id == NULL) {
1062                         /* Blob which has already been marked as traversed. */
1063                         continue;
1064                 }
1065
1066                 if (tree_id) {
1067                         err = load_tree(traversed_ids, tree_id, "",
1068                             repo, cancel_cb, cancel_arg);
1069                         if (err)
1070                                 break;
1071                         tree_id = NULL;
1072                 }
1073
1074                 if (commit || tag)
1075                         (*ncommits)++; /* scanned tags are counted as commits */
1076
1077                 err = report_cleanup_progress(progress_cb, progress_arg, rl,
1078                     *ncommits, -1, -1, -1);
1079                 if (err)
1080                         break;
1081
1082                 if (commit) {
1083                         /* Find parent commits to scan. */
1084                         const struct got_object_id_queue *parent_ids;
1085                         parent_ids = got_object_commit_get_parent_ids(commit);
1086                         err = got_object_id_queue_copy(parent_ids, &ids);
1087                         if (err)
1088                                 break;
1089                         got_object_commit_close(commit);
1090                         commit = NULL;
1091                 }
1092                 if (tag) {
1093                         got_object_tag_close(tag);
1094                         tag = NULL;
1095                 }
1096                 got_object_qid_free(qid);
1097                 qid = NULL;
1098         }
1099 done:
1100         if (qid)
1101                 got_object_qid_free(qid);
1102         if (commit)
1103                 got_object_commit_close(commit);
1104         if (tag)
1105                 got_object_tag_close(tag);
1106         got_object_id_queue_free(&ids);
1107         return err;
1108 }
1109
1110 static const struct got_error *
1111 is_object_packed(int *packed, struct got_repository *repo,
1112     struct got_object_id *id)
1113 {
1114         const struct got_error *err;
1115         struct got_object *obj;
1116
1117         *packed = 0;
1118
1119         err = got_object_open_packed(&obj, id, repo);
1120         if (err) {
1121                 if (err->code == GOT_ERR_NO_OBJ)
1122                         err = NULL;
1123                 return err;
1124         }
1125         got_object_close(obj);
1126         *packed = 1;
1127         return NULL;
1128 }
1129
1130 struct purge_loose_object_arg {
1131         struct got_repository *repo;
1132         got_cleanup_progress_cb progress_cb;
1133         void *progress_arg;
1134         struct got_ratelimit *rl;
1135         struct got_object_idset *traversed_ids;
1136         int nloose;
1137         int ncommits;
1138         int npacked;
1139         int npurged;
1140         off_t size_purged;
1141         int dry_run;
1142         time_t max_mtime;
1143         int ignore_mtime;
1144 };
1145
1146 static const struct got_error *
1147 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1148 {
1149         struct purge_loose_object_arg *a = arg;
1150         const struct got_error *err, *unlock_err = NULL;
1151         char *path = NULL;
1152         int packed, fd = -1;
1153         struct stat sb;
1154         struct got_lockfile *lf = NULL;
1155
1156         err = is_object_packed(&packed, a->repo, id);
1157         if (err)
1158                 return err;
1159
1160         if (!packed && got_object_idset_contains(a->traversed_ids, id))
1161                 return NULL;
1162
1163         if (packed)
1164                 a->npacked++;
1165
1166         err = got_object_get_path(&path, id, a->repo);
1167         if (err)
1168                 return err;
1169
1170         err = got_object_open_loose_fd(&fd, id, a->repo);
1171         if (err)
1172                 goto done;
1173
1174         if (fstat(fd, &sb) == -1) {
1175                 err = got_error_from_errno("fstat");
1176                 goto done;
1177         }
1178
1179         /*
1180          * Do not delete objects which are younger than our maximum
1181          * modification time threshold. This prevents a race where
1182          * new objects which are being added to the repository
1183          * concurrently would be deleted.
1184          */
1185         if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1186                 if (!a->dry_run) {
1187                         err = got_lockfile_lock(&lf, path, -1);
1188                         if (err)
1189                                 goto done;
1190                         if (unlink(path) == -1) {
1191                                 err = got_error_from_errno2("unlink", path);
1192                                 goto done;
1193                         }
1194                 }
1195
1196                 a->npurged++;
1197                 a->size_purged += sb.st_size;
1198                 err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1199                     a->rl, a->ncommits, a->nloose, a->npurged, -1);
1200                 if (err)
1201                         goto done;
1202         }
1203 done:
1204         if (fd != -1 && close(fd) == -1 && err == NULL)
1205                 err = got_error_from_errno("close");
1206         free(path);
1207         if (lf)
1208                 unlock_err = got_lockfile_unlock(lf, -1);
1209         return err ? err : unlock_err;
1210 }
1211
1212 static const struct got_error *
1213 repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1214     struct got_object_idset *traversed_ids,
1215     off_t *size_before, off_t *size_after, int ncommits, int *nloose,
1216     int *npacked, int *npurged, int dry_run, int ignore_mtime,
1217     time_t max_mtime, struct got_ratelimit *rl,
1218     got_cleanup_progress_cb progress_cb, void *progress_arg,
1219     got_cancel_cb cancel_cb, void *cancel_arg)
1220 {
1221         const struct got_error *err;
1222         struct got_object_idset *loose_ids;
1223         struct purge_loose_object_arg arg;
1224
1225         err = get_loose_object_ids(&loose_ids, size_before, ncommits,
1226             progress_cb, progress_arg, rl, repo);
1227         if (err)
1228                 return err;
1229         *nloose = got_object_idset_num_elements(loose_ids);
1230         if (*nloose == 0) {
1231                 got_object_idset_free(loose_ids);
1232                 if (progress_cb) {
1233                         err = progress_cb(progress_arg, 0, 0, 0, -1);
1234                         if (err)
1235                                 return err;
1236                 }
1237                 return NULL;
1238         }
1239
1240         memset(&arg, 0, sizeof(arg));
1241         arg.repo = repo;
1242         arg.progress_arg = progress_arg;
1243         arg.progress_cb = progress_cb;
1244         arg.rl = rl;
1245         arg.traversed_ids = traversed_ids;
1246         arg.nloose = *nloose;
1247         arg.npacked = 0;
1248         arg.npurged = 0;
1249         arg.size_purged = 0;
1250         arg.dry_run = dry_run;
1251         arg.max_mtime = max_mtime;
1252         arg.ignore_mtime = ignore_mtime;
1253         err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1254         if (err)
1255                 goto done;
1256
1257         *size_after = *size_before - arg.size_purged;
1258         *npacked = arg.npacked;
1259         *npurged = arg.npurged;
1260
1261         /* Produce a final progress report. */
1262         if (progress_cb) {
1263                 err = progress_cb(progress_arg, ncommits, *nloose,
1264                     arg.npurged, -1);
1265                 if (err)
1266                         goto done;
1267         }
1268 done:
1269         got_object_idset_free(loose_ids);
1270         return err;
1271 }
1272
1273 static const struct got_error *
1274 purge_redundant_pack(struct got_repository *repo, const char *packidx_path,
1275     int dry_run, int ignore_mtime, time_t max_mtime,
1276     int *remove, off_t *size_before, off_t *size_after)
1277 {
1278         static const char *ext[] = {".idx", ".pack", ".rev", ".bitmap",
1279             ".promisor", ".mtimes"};
1280         struct stat sb;
1281         char *dot, path[PATH_MAX];
1282         size_t i;
1283
1284         if (strlcpy(path, packidx_path, sizeof(path)) >= sizeof(path))
1285                 return got_error(GOT_ERR_NO_SPACE);
1286
1287         /*
1288          * Do not delete pack files which are younger than our maximum
1289          * modification time threshold.  This prevents a race where a
1290          * new pack file which is being added to the repository
1291          * concurrently would be deleted.
1292          */
1293         if (fstatat(got_repo_get_fd(repo), path, &sb, 0) == -1) {
1294                 if (errno == ENOENT)
1295                         return NULL;
1296                 return got_error_from_errno2("fstatat", path);
1297         }
1298         if (!ignore_mtime && sb.st_mtime > max_mtime)
1299                 *remove = 0;
1300
1301         /*
1302          * For compatibility with Git, if a matching .keep file exist
1303          * don't delete the packfile.
1304          */
1305         dot = strrchr(path, '.');
1306         *dot = '\0';
1307         if (strlcat(path, ".keep", sizeof(path)) >= sizeof(path))
1308                 return got_error(GOT_ERR_NO_SPACE);
1309         if (faccessat(got_repo_get_fd(repo), path, F_OK, 0) == 0)
1310                 *remove = 0;
1311
1312         for (i = 0; i < nitems(ext); ++i) {
1313                 *dot = '\0';
1314
1315                 if (strlcat(path, ext[i], sizeof(path)) >=
1316                     sizeof(path))
1317                         return got_error(GOT_ERR_NO_SPACE);
1318
1319                 if (fstatat(got_repo_get_fd(repo), path, &sb, 0) ==
1320                     -1) {
1321                         if (errno == ENOENT)
1322                                 continue;
1323                         return got_error_from_errno2("fstatat", path);
1324                 }
1325
1326                 *size_before += sb.st_size;
1327                 if (!*remove) {
1328                         *size_after += sb.st_size;
1329                         continue;
1330                 }
1331
1332                 if (dry_run)
1333                         continue;
1334
1335                 if (unlinkat(got_repo_get_fd(repo), path, 0) == -1) {
1336                         if (errno == ENOENT)
1337                                 continue;
1338                         return got_error_from_errno2("unlinkat",
1339                             path);
1340                 }
1341         }
1342
1343         return NULL;
1344 }
1345
1346 static const struct got_error *
1347 pack_is_redundant(int *redundant, struct got_repository *repo,
1348     struct got_object_idset *traversed_ids,
1349     const char *packidx_path, struct got_object_idset *idset)
1350 {
1351         const struct got_error *err;
1352         struct got_packidx *packidx;
1353         uint8_t *pid;
1354         struct got_object_id id;
1355         size_t i, nobjects;
1356         size_t digest_len = got_hash_digest_length(repo->algo);
1357
1358         *redundant = 1;
1359
1360         err = got_repo_get_packidx(&packidx, packidx_path, repo);
1361         if (err)
1362                 return err;
1363
1364         nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1365         for (i = 0; i < nobjects; ++i) {
1366                 pid = packidx->hdr.sorted_ids + i * digest_len;
1367
1368                 memset(&id, 0, sizeof(id));
1369                 memcpy(&id.hash, pid, digest_len);
1370                 id.algo = repo->algo;
1371
1372                 if (got_object_idset_contains(idset, &id))
1373                         continue;
1374
1375                 if (!got_object_idset_contains(traversed_ids, &id))
1376                         continue;
1377
1378                 *redundant = 0;
1379                 err = got_object_idset_add(idset, &id, NULL);
1380                 if (err)
1381                         return err;
1382         }
1383
1384         return NULL;
1385 }
1386
1387 struct pack_info {
1388         const char      *path;
1389         size_t           nobjects;
1390 };
1391
1392 static int
1393 pack_info_cmp(const void *a, const void *b)
1394 {
1395         const struct pack_info  *pa, *pb;
1396
1397         pa = a;
1398         pb = b;
1399         if (pa->nobjects == pb->nobjects)
1400                 return strcmp(pa->path, pb->path);
1401         if (pa->nobjects > pb->nobjects)
1402                 return -1;
1403         return 1;
1404 }
1405
1406 static const struct got_error *
1407 repo_purge_redundant_packfiles(struct got_repository *repo,
1408     struct got_object_idset *traversed_ids,
1409     off_t *size_before, off_t *size_after, int dry_run, int ignore_mtime,
1410     time_t max_mtime, int nloose, int ncommits, int npurged,
1411     struct got_ratelimit *rl,
1412     got_cleanup_progress_cb progress_cb, void *progress_arg,
1413     got_cancel_cb cancel_cb, void *cancel_arg)
1414 {
1415         const struct got_error *err;
1416         struct pack_info *pinfo, *sorted = NULL;
1417         struct got_packidx *packidx;
1418         struct got_object_idset *idset = NULL;
1419         struct got_pathlist_entry *pe;
1420         size_t i, npacks;
1421         int remove, redundant_packs = 0;
1422
1423         npacks = 0;
1424         RB_FOREACH(pe, got_pathlist_head, &repo->packidx_paths)
1425                 npacks++;
1426
1427         if (npacks == 0)
1428                 return NULL;
1429
1430         sorted = calloc(npacks, sizeof(*sorted));
1431         if (sorted == NULL)
1432                 return got_error_from_errno("calloc");
1433
1434         i = 0;
1435         RB_FOREACH(pe, got_pathlist_head, &repo->packidx_paths) {
1436                 err = got_repo_get_packidx(&packidx, pe->path, repo);
1437                 if (err)
1438                         goto done;
1439
1440                 pinfo = &sorted[i++];
1441                 pinfo->path = pe->path;
1442                 pinfo->nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1443         }
1444         qsort(sorted, npacks, sizeof(*sorted), pack_info_cmp);
1445
1446         idset = got_object_idset_alloc();
1447         if (idset == NULL) {
1448                 err = got_error_from_errno("got_object_idset_alloc");
1449                 goto done;
1450         }
1451
1452         for (i = 0; i < npacks; ++i) {
1453                 if (cancel_cb) {
1454                         err = (*cancel_cb)(cancel_arg);
1455                         if (err)
1456                                 break;
1457                 }
1458
1459                 err = pack_is_redundant(&remove, repo, traversed_ids,
1460                     sorted[i].path, idset);
1461                 if (err)
1462                         goto done;
1463                 err = purge_redundant_pack(repo, sorted[i].path, dry_run,
1464                     ignore_mtime, max_mtime, &remove, size_before, size_after);
1465                 if (err)
1466                         goto done;
1467                 if (!remove)
1468                         continue;
1469                 err = report_cleanup_progress(progress_cb, progress_arg,
1470                     rl, ncommits, nloose, npurged, ++redundant_packs);
1471                 if (err)
1472                         goto done;
1473         }
1474
1475         /* Produce a final progress report. */
1476         if (progress_cb) {
1477                 err = progress_cb(progress_arg, ncommits, nloose, npurged,
1478                     redundant_packs);
1479                 if (err)
1480                         goto done;
1481         }
1482  done:
1483         free(sorted);
1484         if (idset)
1485                 got_object_idset_free(idset);
1486         return err;
1487 }
1488
1489 const struct got_error *
1490 got_repo_cleanup(struct got_repository *repo,
1491     off_t *loose_before, off_t *loose_after,
1492     off_t *pack_before, off_t *pack_after,
1493     int *ncommits, int *nloose, int *npacked, int dry_run, int ignore_mtime,
1494     got_cleanup_progress_cb progress_cb, void *progress_arg,
1495     got_pack_progress_cb pack_progress_cb, void *pack_progress_arg,
1496     got_pack_index_progress_cb index_progress_cb, void *index_progress_arg,
1497     got_cancel_cb cancel_cb, void *cancel_arg)
1498 {
1499         const struct got_error *unlock_err, *err = NULL;
1500         struct got_lockfile *lk = NULL;
1501         struct got_ratelimit rl;
1502         struct got_reflist_head refs;
1503         struct got_object_idset *traversed_ids = NULL;
1504         struct got_reflist_entry *re;
1505         struct got_object_id **referenced_ids;
1506         int i, nreferenced;
1507         int npurged = 0, packfd = -1;
1508         char *tmpfile_path = NULL, *packfile_path = NULL, *idxpath = NULL;
1509         FILE *delta_cache = NULL, *packfile = NULL;
1510         struct got_object_id pack_hash;
1511         time_t max_mtime = 0;
1512
1513         TAILQ_INIT(&refs);
1514         got_ratelimit_init(&rl, 0, 500);
1515         memset(&pack_hash, 0, sizeof(pack_hash));
1516
1517         *loose_before = 0;
1518         *loose_after = 0;
1519         *pack_before = 0;
1520         *pack_after = 0;
1521         *ncommits = 0;
1522         *nloose = 0;
1523         *npacked = 0;
1524
1525         err = repo_cleanup_lock(repo, &lk);
1526         if (err)
1527                 return err;
1528
1529         err = create_temp_packfile(&packfd, &tmpfile_path, repo);
1530         if (err)
1531                 goto done;
1532
1533         delta_cache = got_opentemp();
1534         if (delta_cache == NULL) {
1535                 err = got_error_from_errno("got_opentemp");
1536                 goto done;
1537         }
1538
1539         traversed_ids = got_object_idset_alloc();
1540         if (traversed_ids == NULL) {
1541                 err = got_error_from_errno("got_object_idset_alloc");
1542                 goto done;
1543         }
1544
1545         err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1546         if (err)
1547                 goto done;
1548         if (!ignore_mtime) {
1549                 TAILQ_FOREACH(re, &refs, entry) {
1550                         time_t mtime = got_ref_get_mtime(re->ref);
1551                         if (mtime > max_mtime)
1552                                 max_mtime = mtime;
1553                 }
1554                 /*
1555                  * For safety, keep objects created within 10 minutes
1556                  * before the youngest reference was created.
1557                  */
1558                 if (max_mtime >= 600)
1559                         max_mtime -= 600;
1560         }
1561
1562         err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1563             GOT_OBJ_TYPE_ANY, &refs, repo, cancel_cb, cancel_arg);
1564         if (err)
1565                 goto done;
1566
1567         for (i = 0; i < nreferenced; i++) {
1568                 struct got_object_id *id = referenced_ids[i];
1569                 err = load_commit_or_tag(ncommits, traversed_ids,
1570                     id, repo, progress_cb, progress_arg, &rl,
1571                     cancel_cb, cancel_arg);
1572                 if (err)
1573                         goto done;
1574         }
1575
1576         err = got_pack_create(&pack_hash, packfd, delta_cache,
1577             NULL, 0, referenced_ids, nreferenced, repo, 0,
1578             0, 0, pack_progress_cb, pack_progress_arg,
1579             &rl, cancel_cb, cancel_arg);
1580         if (err)
1581                 goto done;
1582
1583         err = install_packfile(&packfile, &packfd, &packfile_path,
1584             &tmpfile_path, &pack_hash, repo);
1585         if (err)
1586                 goto done;
1587
1588         err = got_repo_index_pack(&idxpath, packfile, &pack_hash, repo,
1589             index_progress_cb, index_progress_arg,
1590             cancel_cb, cancel_arg);
1591         if (err)
1592                 goto done;
1593
1594         err = got_repo_list_packidx(&repo->packidx_paths, repo);
1595         if (err)
1596                 goto done;
1597
1598         err = repo_purge_unreferenced_loose_objects(repo, traversed_ids,
1599             loose_before, loose_after, *ncommits, nloose, npacked, &npurged,
1600             dry_run, ignore_mtime, max_mtime, &rl, progress_cb, progress_arg,
1601             cancel_cb, cancel_arg);
1602         if (err)
1603                 goto done;
1604
1605         err = repo_purge_redundant_packfiles(repo, traversed_ids,
1606             pack_before, pack_after, dry_run, ignore_mtime, max_mtime,
1607             *nloose, *ncommits, npurged, &rl, progress_cb, progress_arg,
1608             cancel_cb, cancel_arg);
1609         if (err)
1610                 goto done;
1611
1612         if (dry_run) {
1613                 if (idxpath && unlink(idxpath) == -1)
1614                         err = got_error_from_errno2("unlink", idxpath);
1615                 if (packfile_path && unlink(packfile_path) == -1 && err == NULL)
1616                         err = got_error_from_errno2("unlink", packfile_path);
1617         }
1618  done:
1619         if (lk) {
1620                 unlock_err = got_lockfile_unlock(lk, got_repo_get_fd(repo));
1621                 if (err == NULL)
1622                         err = unlock_err;
1623         }
1624         if (traversed_ids)
1625                 got_object_idset_free(traversed_ids);
1626         if (packfd != -1 && close(packfd) == -1 && err == NULL)
1627                 err = got_error_from_errno2("close",
1628                     packfile_path ? packfile_path : tmpfile_path);
1629         if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1630                 err = got_error_from_errno("fclose");
1631         if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
1632                 err = got_error_from_errno2("unlink", tmpfile_path);
1633         free(tmpfile_path);
1634         free(packfile_path);
1635         free(idxpath);
1636         return err;
1637 }
1638
1639 const struct got_error *
1640 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1641     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1642     got_cancel_cb cancel_cb, void *cancel_arg)
1643 {
1644         const struct got_error *err = NULL;
1645         DIR *packdir = NULL;
1646         struct dirent *dent;
1647         char *pack_relpath = NULL;
1648         int packdir_fd;
1649         struct stat sb;
1650
1651         packdir_fd = openat(got_repo_get_fd(repo),
1652             GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1653         if (packdir_fd == -1) {
1654                 if (errno == ENOENT)
1655                         return NULL;
1656                 return got_error_from_errno_fmt("openat: %s/%s",
1657                     got_repo_get_path_git_dir(repo),
1658                     GOT_OBJECTS_PACK_DIR);
1659         }
1660
1661         packdir = fdopendir(packdir_fd);
1662         if (packdir == NULL) {
1663                 err = got_error_from_errno("fdopendir");
1664                 close(packdir_fd);
1665                 goto done;
1666         }
1667
1668         while ((dent = readdir(packdir)) != NULL) {
1669                 if (cancel_cb) {
1670                         err = cancel_cb(cancel_arg);
1671                         if (err)
1672                                 goto done;
1673                 }
1674
1675                 if (!got_repo_is_packidx_filename(dent->d_name,
1676                     strlen(dent->d_name),
1677                     got_repo_get_object_format(repo)))
1678                         continue;
1679
1680                 err = got_packidx_get_packfile_path(&pack_relpath,
1681                     dent->d_name);
1682                 if (err)
1683                         goto done;
1684
1685                 if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1686                         free(pack_relpath);
1687                         pack_relpath = NULL;
1688                         continue;
1689                 }
1690                 if (errno != ENOENT) {
1691                         err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1692                             got_repo_get_path_git_dir(repo),
1693                             GOT_OBJECTS_PACK_DIR,
1694                             pack_relpath);
1695                         goto done;
1696                 }
1697
1698                 if (!dry_run) {
1699                         if (unlinkat(packdir_fd, dent->d_name, 0) == -1) {
1700                                 err = got_error_from_errno("unlinkat");
1701                                 goto done;
1702                         }
1703                 }
1704                 if (progress_cb) {
1705                         char *path;
1706                         if (asprintf(&path, "%s/%s/%s",
1707                             got_repo_get_path_git_dir(repo),
1708                             GOT_OBJECTS_PACK_DIR,
1709                             dent->d_name) == -1) {
1710                                 err = got_error_from_errno("asprintf");
1711                                 goto done;
1712                         }
1713                         err = progress_cb(progress_arg, path);
1714                         free(path);
1715                         if (err)
1716                                 goto done;
1717                 }
1718                 free(pack_relpath);
1719                 pack_relpath = NULL;
1720         }
1721 done:
1722         if (packdir && closedir(packdir) != 0 && err == NULL)
1723                 err = got_error_from_errno("closedir");
1724         free(pack_relpath);
1725         return err;
1726 }