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_object.h"
  47 #include "got_lib_object_idset.h"
  48 #include "got_lib_object_cache.h"
  49 #include "got_lib_pack.h"
  50 #include "got_lib_privsep.h"
  51 #include "got_lib_repository.h"
  52 #include "got_lib_ratelimit.h"
  53 #include "got_lib_pack_create.h"
  54 #include "got_lib_hash.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 const struct got_error *
 143 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
 144     struct got_reflist_head *include_refs,
 145     struct got_reflist_head *exclude_refs, struct got_repository *repo,
 146     int loose_obj_only, int force_refdelta,
 147     got_pack_progress_cb progress_cb, void *progress_arg,
 148     got_cancel_cb cancel_cb, void *cancel_arg)
 149 {
 150         const struct got_error *err = NULL;
 151         struct got_object_id **ours = NULL, **theirs = NULL;
 152         int nours = 0, ntheirs = 0, packfd = -1, i;
 153         char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
 154         char *sha1_str = NULL;
 155         FILE *delta_cache = NULL;
 156         struct got_ratelimit rl;
 157
 158         *packfile = NULL;
 159         *pack_hash = NULL;
 160
 161         got_ratelimit_init(&rl, 0, 500);
 162
 163         if (asprintf(&path, "%s/%s/packing.pack",
 164             got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
 165                 err = got_error_from_errno("asprintf");
 166                 goto done;
 167         }
 168         err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
 169         if (err)
 170                 goto done;
 171
 172         if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
 173                 err = got_error_from_errno2("fchmod", tmpfile_path);
 174                 goto done;
 175         }
 176
 177         delta_cache = got_opentemp();
 178         if (delta_cache == NULL) {
 179                 err = got_error_from_errno("got_opentemp");
 180                 goto done;
 181         }
 182
 183         err = get_reflist_object_ids(&ours, &nours,
 184             (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
 185             include_refs, repo, cancel_cb, cancel_arg);
 186         if (err)
 187                 goto done;
 188
 189         if (nours == 0) {
 190                 err = got_error(GOT_ERR_CANNOT_PACK);
 191                 goto done;
 192         }
 193
 194         if (!TAILQ_EMPTY(exclude_refs)) {
 195                 err = get_reflist_object_ids(&theirs, &ntheirs,
 196                     (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
 197                     exclude_refs, repo,
 198                     cancel_cb, cancel_arg);
 199                 if (err)
 200                         goto done;
 201         }
 202
 203         *pack_hash = calloc(1, sizeof(**pack_hash));
 204         if (*pack_hash == NULL) {
 205                 err = got_error_from_errno("calloc");
 206                 goto done;
 207         }
 208
 209         err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
 210             theirs, ntheirs, ours, nours, repo, loose_obj_only,
 211             0, force_refdelta, progress_cb, progress_arg, &rl,
 212             cancel_cb, cancel_arg);
 213         if (err)
 214                 goto done;
 215
 216         err = got_object_id_str(&sha1_str, *pack_hash);
 217         if (err)
 218                 goto done;
 219         if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
 220             got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
 221             sha1_str) == -1) {
 222                 err = got_error_from_errno("asprintf");
 223                 goto done;
 224         }
 225
 226         if (lseek(packfd, 0L, SEEK_SET) == -1) {
 227                 err = got_error_from_errno("lseek");
 228                 goto done;
 229         }
 230         if (rename(tmpfile_path, packfile_path) == -1) {
 231                 err = got_error_from_errno3("rename", tmpfile_path,
 232                     packfile_path);
 233                 goto done;
 234         }
 235         free(tmpfile_path);
 236         tmpfile_path = NULL;
 237
 238         *packfile = fdopen(packfd, "w");
 239         if (*packfile == NULL) {
 240                 err = got_error_from_errno2("fdopen", tmpfile_path);
 241                 goto done;
 242         }
 243         packfd = -1;
 244 done:
 245         for (i = 0; i < nours; i++)
 246                 free(ours[i]);
 247         free(ours);
 248         for (i = 0; i < ntheirs; i++)
 249                 free(theirs[i]);
 250         free(theirs);
 251         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 252                 err = got_error_from_errno2("close", packfile_path);
 253         if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
 254                 err = got_error_from_errno("fclose");
 255         if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
 256                 err = got_error_from_errno2("unlink", tmpfile_path);
 257         free(tmpfile_path);
 258         free(packfile_path);
 259         free(sha1_str);
 260         free(path);
 261         if (err) {
 262                 free(*pack_hash);
 263                 *pack_hash = NULL;
 264                 if (*packfile)
 265                         fclose(*packfile);
 266                 *packfile = NULL;
 267         }
 268         return err;
 269 }
 270
 271 const struct got_error *
 272 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
 273     struct got_repository *repo,
 274     got_pack_index_progress_cb progress_cb, void *progress_arg,
 275     got_cancel_cb cancel_cb, void *cancel_arg)
 276 {
 277         size_t i;
 278         char *path;
 279         int imsg_idxfds[2];
 280         int npackfd = -1, idxfd = -1, nidxfd = -1;
 281         int tmpfds[3];
 282         int idxstatus, done = 0;
 283         const struct got_error *err;
 284         struct imsgbuf idxibuf;
 285         pid_t idxpid;
 286         char *tmpidxpath = NULL;
 287         char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
 288         const char *repo_path = got_repo_get_path_git_dir(repo);
 289         struct stat sb;
 290
 291         for (i = 0; i < nitems(tmpfds); i++)
 292                 tmpfds[i] = -1;
 293
 294         if (asprintf(&path, "%s/%s/indexing.idx",
 295             repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 296                 err = got_error_from_errno("asprintf");
 297                 goto done;
 298         }
 299         err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
 300         free(path);
 301         if (err)
 302                 goto done;
 303         if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
 304                 err = got_error_from_errno2("fchmod", tmpidxpath);
 305                 goto done;
 306         }
 307
 308         nidxfd = dup(idxfd);
 309         if (nidxfd == -1) {
 310                 err = got_error_from_errno("dup");
 311                 goto done;
 312         }
 313
 314         for (i = 0; i < nitems(tmpfds); i++) {
 315                 tmpfds[i] = got_opentempfd();
 316                 if (tmpfds[i] == -1) {
 317                         err = got_error_from_errno("got_opentempfd");
 318                         goto done;
 319                 }
 320         }
 321
 322         err = got_object_id_str(&id_str, pack_hash);
 323         if (err)
 324                 goto done;
 325
 326         if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
 327             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 328                 err = got_error_from_errno("asprintf");
 329                 goto done;
 330         }
 331
 332         if (fstat(fileno(packfile), &sb) == -1) {
 333                 err = got_error_from_errno2("fstat", packfile_path);
 334                 goto done;
 335         }
 336
 337         if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
 338             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 339                 err = got_error_from_errno("asprintf");
 340                 goto done;
 341         }
 342
 343         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
 344                 err = got_error_from_errno("socketpair");
 345                 goto done;
 346         }
 347         idxpid = fork();
 348         if (idxpid == -1) {
 349                 err= got_error_from_errno("fork");
 350                 goto done;
 351         } else if (idxpid == 0)
 352                 got_privsep_exec_child(imsg_idxfds,
 353                     GOT_PATH_PROG_INDEX_PACK, packfile_path);
 354         if (close(imsg_idxfds[1]) == -1) {
 355                 err = got_error_from_errno("close");
 356                 goto done;
 357         }
 358         imsg_init(&idxibuf, imsg_idxfds[0]);
 359
 360         npackfd = dup(fileno(packfile));
 361         if (npackfd == -1) {
 362                 err = got_error_from_errno("dup");
 363                 goto done;
 364         }
 365         err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
 366             npackfd);
 367         if (err != NULL)
 368                 goto done;
 369         npackfd = -1;
 370         err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
 371         if (err != NULL)
 372                 goto done;
 373         nidxfd = -1;
 374         for (i = 0; i < nitems(tmpfds); i++) {
 375                 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
 376                 if (err != NULL)
 377                         goto done;
 378                 tmpfds[i] = -1;
 379         }
 380         done = 0;
 381         while (!done) {
 382                 int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
 383
 384                 if (cancel_cb) {
 385                         err = cancel_cb(cancel_arg);
 386                         if (err)
 387                                 goto done;
 388                 }
 389
 390                 err = got_privsep_recv_index_progress(&done, &nobj_total,
 391                     &nobj_indexed, &nobj_loose, &nobj_resolved,
 392                     &idxibuf);
 393                 if (err != NULL)
 394                         goto done;
 395                 if (nobj_indexed != 0) {
 396                         err = progress_cb(progress_arg, sb.st_size,
 397                             nobj_total, nobj_indexed, nobj_loose,
 398                             nobj_resolved);
 399                         if (err)
 400                                 break;
 401                 }
 402         }
 403         if (close(imsg_idxfds[0]) == -1) {
 404                 err = got_error_from_errno("close");
 405                 goto done;
 406         }
 407         if (waitpid(idxpid, &idxstatus, 0) == -1) {
 408                 err = got_error_from_errno("waitpid");
 409                 goto done;
 410         }
 411
 412         if (rename(tmpidxpath, idxpath) == -1) {
 413                 err = got_error_from_errno3("rename", tmpidxpath, idxpath);
 414                 goto done;
 415         }
 416         free(tmpidxpath);
 417         tmpidxpath = NULL;
 418
 419 done:
 420         if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
 421                 err = got_error_from_errno2("unlink", tmpidxpath);
 422         if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
 423                 err = got_error_from_errno("close");
 424         if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
 425                 err = got_error_from_errno("close");
 426         for (i = 0; i < nitems(tmpfds); i++) {
 427                 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
 428                         err = got_error_from_errno("close");
 429         }
 430         free(tmpidxpath);
 431         free(idxpath);
 432         free(packfile_path);
 433         return err;
 434 }
 435
 436 const struct got_error *
 437 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
 438     struct got_repository *repo, const char *packfile_path)
 439 {
 440         const struct got_error *err = NULL;
 441         const char *packdir_path = NULL;
 442         char *packfile_name = NULL, *p, *dot;
 443         struct got_object_id id;
 444         int packfd = -1;
 445
 446         *packfile = NULL;
 447         *pack_hash = NULL;
 448
 449         packdir_path = got_repo_get_path_objects_pack(repo);
 450         if (packdir_path == NULL)
 451                 return got_error_from_errno("got_repo_get_path_objects_pack");
 452
 453         if (!got_path_is_child(packfile_path, packdir_path,
 454             strlen(packdir_path))) {
 455                 err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
 456                 goto done;
 457
 458         }
 459
 460         err = got_path_basename(&packfile_name, packfile_path);
 461         if (err)
 462                 goto done;
 463         p = packfile_name;
 464
 465         if (strncmp(p, "pack-", 5) != 0) {
 466                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 467                     "'%s' is not a valid pack file name",
 468                     packfile_name);
 469                 goto done;
 470         }
 471         p += 5;
 472         dot = strchr(p, '.');
 473         if (dot == NULL) {
 474                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 475                     "'%s' is not a valid pack file name",
 476                     packfile_name);
 477                 goto done;
 478         }
 479         if (strcmp(dot + 1, "pack") != 0) {
 480                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 481                     "'%s' is not a valid pack file name",
 482                     packfile_name);
 483                 goto done;
 484         }
 485         *dot = '\0';
 486         if (!got_parse_object_id(&id, p, repo->algo)) {
 487                 err = got_error_fmt(GOT_ERR_BAD_PATH,
 488                     "'%s' is not a valid pack file name",
 489                     packfile_name);
 490                 goto done;
 491         }
 492
 493         *pack_hash = got_object_id_dup(&id);
 494         if (*pack_hash == NULL) {
 495                 err = got_error_from_errno("got_object_id_dup");
 496                 goto done;
 497         }
 498
 499         packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
 500         if (packfd == -1) {
 501                 err = got_error_from_errno2("open", packfile_path);
 502                 goto done;
 503         }
 504
 505         *packfile = fdopen(packfd, "r");
 506         if (*packfile == NULL) {
 507                 err = got_error_from_errno2("fdopen", packfile_path);
 508                 goto done;
 509         }
 510         packfd = -1;
 511 done:
 512         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 513                 err = got_error_from_errno2("close", packfile_path);
 514         free(packfile_name);
 515         if (err) {
 516                 free(*pack_hash);
 517                 *pack_hash = NULL;
 518         }
 519         return err;
 520 }
 521
 522 const struct got_error *
 523 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
 524     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
 525     got_cancel_cb cancel_cb, void *cancel_arg)
 526 {
 527         const struct got_error *err = NULL;
 528         char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
 529         struct got_packidx *packidx = NULL;
 530         struct got_pack *pack = NULL;
 531         uint32_t nobj, i;
 532
 533         err = got_object_id_str(&id_str, pack_hash);
 534         if (err)
 535                 goto done;
 536
 537         if (asprintf(&packpath, "%s/pack-%s.pack",
 538             GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 539                 err = got_error_from_errno("asprintf");
 540                 goto done;
 541         }
 542         if (asprintf(&idxpath, "%s/pack-%s.idx",
 543             GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 544                 err = got_error_from_errno("asprintf");
 545                 goto done;
 546         }
 547
 548         err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
 549         if (err)
 550                 goto done;
 551
 552         err = got_repo_cache_pack(&pack, repo, packpath, packidx);
 553         if (err)
 554                 goto done;
 555
 556         nobj = be32toh(packidx->hdr.fanout_table[0xff]);
 557         for (i = 0; i < nobj; i++) {
 558                 struct got_packidx_object_id *oid;
 559                 struct got_object_id id, base_id;
 560                 off_t offset, base_offset = 0;
 561                 uint8_t type;
 562                 uint64_t size;
 563                 size_t tslen, len;
 564
 565                 if (cancel_cb) {
 566                         err = cancel_cb(cancel_arg);
 567                         if (err)
 568                                 break;
 569                 }
 570                 oid = &packidx->hdr.sorted_ids[i];
 571                 memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
 572
 573                 offset = got_packidx_get_object_offset(packidx, i);
 574                 if (offset == -1) {
 575                         err = got_error(GOT_ERR_BAD_PACKIDX);
 576                         goto done;
 577                 }
 578
 579                 err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
 580                     pack, offset);
 581                 if (err)
 582                         goto done;
 583
 584                 switch (type) {
 585                 case GOT_OBJ_TYPE_OFFSET_DELTA:
 586                         err = got_pack_parse_offset_delta(&base_offset, &len,
 587                             pack, offset, tslen);
 588                         if (err)
 589                                 goto done;
 590                         break;
 591                 case GOT_OBJ_TYPE_REF_DELTA:
 592                         err = got_pack_parse_ref_delta(&base_id,
 593                             pack, offset, tslen);
 594                         if (err)
 595                                 goto done;
 596                         break;
 597                 }
 598                 err = (*list_cb)(list_arg, &id, type, offset, size,
 599                     base_offset, &base_id);
 600                 if (err)
 601                         goto done;
 602         }
 603
 604 done:
 605         free(id_str);
 606         free(idxpath);
 607         free(packpath);
 608         if (packidx)
 609                 got_packidx_close(packidx);
 610         return err;
 611 }
 612
 613 static const struct got_error *
 614 repo_cleanup_lock(struct got_repository *repo, struct got_lockfile **lk)
 615 {
 616         const struct got_error *err;
 617         char myname[_POSIX_HOST_NAME_MAX + 1];
 618
 619         if (gethostname(myname, sizeof(myname)) == -1)
 620                 return got_error_from_errno("gethostname");
 621
 622         err = got_lockfile_lock(lk, "gc.pid", got_repo_get_fd(repo));
 623         if (err)
 624                 return err;
 625
 626         /*
 627          * Git uses these info to provide some verbiage when finds a
 628          * lock during `git gc --force' so don't try too hard to avoid
 629          * short writes and don't care if a race happens between the
 630          * lockfile creation and the write itself.
 631          */
 632         if (dprintf((*lk)->fd, "%d %s", getpid(), myname) < 0)
 633                 return got_error_from_errno("dprintf");
 634
 635         return NULL;
 636 }
 637
 638 static const struct got_error *
 639 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
 640     void *progress_arg, struct got_ratelimit *rl,
 641     int ncommits, int nloose, int npurged, int nredundant)
 642 {
 643         const struct got_error *err;
 644         int elapsed;
 645
 646         if (progress_cb == NULL)
 647                 return NULL;
 648
 649         err = got_ratelimit_check(&elapsed, rl);
 650         if (err || !elapsed)
 651                 return err;
 652
 653         return progress_cb(progress_arg, ncommits, nloose, npurged,
 654             nredundant);
 655 }
 656
 657 static const struct got_error *
 658 get_loose_object_ids(struct got_object_idset **loose_ids,
 659     off_t *ondisk_size, int ncommits,
 660     got_cleanup_progress_cb progress_cb, void *progress_arg,
 661     struct got_ratelimit *rl, struct got_repository *repo)
 662 {
 663         const struct got_error *err = NULL;
 664         char *path_objects = NULL, *path = NULL;
 665         DIR *dir = NULL;
 666         struct got_object *obj = NULL;
 667         struct got_object_id id;
 668         int i, fd = -1;
 669         struct stat sb;
 670
 671         *ondisk_size = 0;
 672         *loose_ids = got_object_idset_alloc();
 673         if (*loose_ids == NULL)
 674                 return got_error_from_errno("got_object_idset_alloc");
 675
 676         path_objects = got_repo_get_path_objects(repo);
 677         if (path_objects == NULL) {
 678                 err = got_error_from_errno("got_repo_get_path_objects");
 679                 goto done;
 680         }
 681
 682         for (i = 0; i <= 0xff; i++) {
 683                 struct dirent *dent;
 684
 685                 if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
 686                         err = got_error_from_errno("asprintf");
 687                         break;
 688                 }
 689
 690                 dir = opendir(path);
 691                 if (dir == NULL) {
 692                         if (errno == ENOENT) {
 693                                 err = NULL;
 694                                 continue;
 695                         }
 696                         err = got_error_from_errno2("opendir", path);
 697                         break;
 698                 }
 699
 700                 while ((dent = readdir(dir)) != NULL) {
 701                         char *id_str;
 702
 703                         if (strcmp(dent->d_name, ".") == 0 ||
 704                             strcmp(dent->d_name, "..") == 0)
 705                                 continue;
 706
 707                         if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
 708                                 err = got_error_from_errno("asprintf");
 709                                 goto done;
 710                         }
 711
 712                         if (!got_parse_object_id(&id, id_str, repo->algo)) {
 713                                 free(id_str);
 714                                 continue;
 715                         }
 716                         free(id_str);
 717
 718                         err = got_object_open_loose_fd(&fd, &id, repo);
 719                         if (err)
 720                                 goto done;
 721                         if (fstat(fd, &sb) == -1) {
 722                                 err = got_error_from_errno("fstat");
 723                                 goto done;
 724                         }
 725                         err = got_object_read_header_privsep(&obj, &id, repo,
 726                             fd);
 727                         if (err)
 728                                 goto done;
 729                         fd = -1; /* already closed */
 730
 731                         switch (obj->type) {
 732                         case GOT_OBJ_TYPE_COMMIT:
 733                         case GOT_OBJ_TYPE_TREE:
 734                         case GOT_OBJ_TYPE_BLOB:
 735                         case GOT_OBJ_TYPE_TAG:
 736                                 break;
 737                         default:
 738                                 err = got_error_fmt(GOT_ERR_OBJ_TYPE,
 739                                     "%d", obj->type);
 740                                 goto done;
 741                         }
 742                         got_object_close(obj);
 743                         obj = NULL;
 744                         (*ondisk_size) += sb.st_size;
 745                         err = got_object_idset_add(*loose_ids, &id, NULL);
 746                         if (err)
 747                                 goto done;
 748                         err = report_cleanup_progress(progress_cb,
 749                             progress_arg, rl, ncommits,
 750                             got_object_idset_num_elements(*loose_ids),
 751                             -1, -1);
 752                         if (err)
 753                                 goto done;
 754                 }
 755
 756                 if (closedir(dir) != 0) {
 757                         err = got_error_from_errno("closedir");
 758                         goto done;
 759                 }
 760                 dir = NULL;
 761
 762                 free(path);
 763                 path = NULL;
 764         }
 765 done:
 766         if (dir && closedir(dir) != 0 && err == NULL)
 767                 err = got_error_from_errno("closedir");
 768         if (fd != -1 && close(fd) == -1 && err == NULL)
 769                 err = got_error_from_errno("close");
 770         if (err) {
 771                 got_object_idset_free(*loose_ids);
 772                 *loose_ids = NULL;
 773         }
 774         if (obj)
 775                 got_object_close(obj);
 776         free(path_objects);
 777         free(path);
 778         return err;
 779 }
 780
 781 static const struct got_error *
 782 load_tree_entries(struct got_object_id_queue *ids,
 783     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
 784     const char *dpath, struct got_repository *repo,
 785     got_cancel_cb cancel_cb, void *cancel_arg)
 786 {
 787         const struct got_error *err;
 788         struct got_tree_object *tree;
 789         char *p = NULL;
 790         int i;
 791
 792         err = got_object_open_as_tree(&tree, repo, tree_id);
 793         if (err)
 794                 return err;
 795
 796         for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
 797                 struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
 798                 struct got_object_id *id = got_tree_entry_get_id(e);
 799                 mode_t mode = got_tree_entry_get_mode(e);
 800
 801                 if (cancel_cb) {
 802                         err = (*cancel_cb)(cancel_arg);
 803                         if (err)
 804                                 break;
 805                 }
 806
 807                 if (got_object_tree_entry_is_symlink(e) ||
 808                     got_object_tree_entry_is_submodule(e) ||
 809                     got_object_idset_contains(traversed_ids, id))
 810                         continue;
 811
 812                 if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
 813                     got_tree_entry_get_name(e)) == -1) {
 814                         err = got_error_from_errno("asprintf");
 815                         break;
 816                 }
 817
 818                 if (S_ISDIR(mode)) {
 819                         struct got_object_qid *qid;
 820                         err = got_object_qid_alloc(&qid, id);
 821                         if (err)
 822                                 break;
 823                         STAILQ_INSERT_TAIL(ids, qid, entry);
 824                 } else if (S_ISREG(mode)) {
 825                         /* This blob is referenced. */
 826                         err = got_object_idset_add(traversed_ids, id, NULL);
 827                         if (err)
 828                                 break;
 829                 }
 830                 free(p);
 831                 p = NULL;
 832         }
 833
 834         got_object_tree_close(tree);
 835         free(p);
 836         return err;
 837 }
 838
 839 static const struct got_error *
 840 load_tree(struct got_object_idset *traversed_ids,
 841     struct got_object_id *tree_id,
 842     const char *dpath, struct got_repository *repo,
 843     got_cancel_cb cancel_cb, void *cancel_arg)
 844 {
 845         const struct got_error *err = NULL;
 846         struct got_object_id_queue tree_ids;
 847         struct got_object_qid *qid;
 848
 849         err = got_object_qid_alloc(&qid, tree_id);
 850         if (err)
 851                 return err;
 852
 853         STAILQ_INIT(&tree_ids);
 854         STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
 855
 856         while (!STAILQ_EMPTY(&tree_ids)) {
 857                 if (cancel_cb) {
 858                         err = (*cancel_cb)(cancel_arg);
 859                         if (err)
 860                                 break;
 861                 }
 862
 863                 qid = STAILQ_FIRST(&tree_ids);
 864                 STAILQ_REMOVE_HEAD(&tree_ids, entry);
 865
 866                 if (got_object_idset_contains(traversed_ids, &qid->id)) {
 867                         got_object_qid_free(qid);
 868                         continue;
 869                 }
 870
 871                 err = got_object_idset_add(traversed_ids, &qid->id, NULL);
 872                 if (err) {
 873                         got_object_qid_free(qid);
 874                         break;
 875                 }
 876
 877                 err = load_tree_entries(&tree_ids, traversed_ids,
 878                     &qid->id, dpath, repo, cancel_cb, cancel_arg);
 879                 got_object_qid_free(qid);
 880                 if (err)
 881                         break;
 882         }
 883
 884         got_object_id_queue_free(&tree_ids);
 885         return err;
 886 }
 887
 888 static const struct got_error *
 889 load_commit_or_tag(int *ncommits, struct got_object_idset *traversed_ids,
 890     struct got_object_id *id, struct got_repository *repo,
 891     got_cleanup_progress_cb progress_cb, void *progress_arg,
 892     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 893 {
 894         const struct got_error *err;
 895         struct got_commit_object *commit = NULL;
 896         struct got_tag_object *tag = NULL;
 897         struct got_object_id *tree_id = NULL;
 898         struct got_object_id_queue ids;
 899         struct got_object_qid *qid;
 900         int obj_type;
 901
 902         err = got_object_qid_alloc(&qid, id);
 903         if (err)
 904                 return err;
 905
 906         STAILQ_INIT(&ids);
 907         STAILQ_INSERT_TAIL(&ids, qid, entry);
 908
 909         while (!STAILQ_EMPTY(&ids)) {
 910                 if (cancel_cb) {
 911                         err = (*cancel_cb)(cancel_arg);
 912                         if (err)
 913                                 break;
 914                 }
 915
 916                 qid = STAILQ_FIRST(&ids);
 917                 STAILQ_REMOVE_HEAD(&ids, entry);
 918
 919                 if (got_object_idset_contains(traversed_ids, &qid->id)) {
 920                         got_object_qid_free(qid);
 921                         qid = NULL;
 922                         continue;
 923                 }
 924
 925                 err = got_object_idset_add(traversed_ids, &qid->id, NULL);
 926                 if (err)
 927                         break;
 928
 929                 err = got_object_get_type(&obj_type, repo, &qid->id);
 930                 if (err)
 931                         break;
 932                 switch (obj_type) {
 933                 case GOT_OBJ_TYPE_COMMIT:
 934                         err = got_object_open_as_commit(&commit, repo,
 935                             &qid->id);
 936                         if (err)
 937                                 goto done;
 938                         break;
 939                 case GOT_OBJ_TYPE_TAG:
 940                         err = got_object_open_as_tag(&tag, repo, &qid->id);
 941                         if (err)
 942                                 goto done;
 943                         break;
 944                 default:
 945                         /* should not happen */
 946                         err = got_error(GOT_ERR_OBJ_TYPE);
 947                         goto done;
 948                 }
 949
 950                 /* Find a tree object to scan. */
 951                 if (commit) {
 952                         tree_id = got_object_commit_get_tree_id(commit);
 953                 } else if (tag) {
 954                         obj_type = got_object_tag_get_object_type(tag);
 955                         switch (obj_type) {
 956                         case GOT_OBJ_TYPE_COMMIT:
 957                                 err = got_object_open_as_commit(&commit, repo,
 958                                     got_object_tag_get_object_id(tag));
 959                                 if (err)
 960                                         goto done;
 961                                 tree_id = got_object_commit_get_tree_id(commit);
 962                                 break;
 963                         case GOT_OBJ_TYPE_TREE:
 964                                 tree_id = got_object_tag_get_object_id(tag);
 965                                 break;
 966                         default:
 967                                 /*
 968                                  * Tag points at something other than a
 969                                  * commit or tree. Leave this weird tag object
 970                                  * and the object it points to.
 971                                  */
 972                                 if (got_object_idset_contains(traversed_ids,
 973                                     got_object_tag_get_object_id(tag)))
 974                                         break;
 975                                 err = got_object_idset_add(traversed_ids,
 976                                     got_object_tag_get_object_id(tag), NULL);
 977                                 if (err)
 978                                         goto done;
 979                                 break;
 980                         }
 981                 }
 982
 983                 if (tree_id) {
 984                         err = load_tree(traversed_ids, tree_id, "",
 985                             repo, cancel_cb, cancel_arg);
 986                         if (err)
 987                                 break;
 988                 }
 989
 990                 if (commit || tag)
 991                         (*ncommits)++; /* scanned tags are counted as commits */
 992
 993                 err = report_cleanup_progress(progress_cb, progress_arg, rl,
 994                     *ncommits, -1, -1, -1);
 995                 if (err)
 996                         break;
 997
 998                 if (commit) {
 999                         /* Find parent commits to scan. */
1000                         const struct got_object_id_queue *parent_ids;
1001                         parent_ids = got_object_commit_get_parent_ids(commit);
1002                         err = got_object_id_queue_copy(parent_ids, &ids);
1003                         if (err)
1004                                 break;
1005                         got_object_commit_close(commit);
1006                         commit = NULL;
1007                 }
1008                 if (tag) {
1009                         got_object_tag_close(tag);
1010                         tag = NULL;
1011                 }
1012                 got_object_qid_free(qid);
1013                 qid = NULL;
1014         }
1015 done:
1016         if (qid)
1017                 got_object_qid_free(qid);
1018         if (commit)
1019                 got_object_commit_close(commit);
1020         if (tag)
1021                 got_object_tag_close(tag);
1022         got_object_id_queue_free(&ids);
1023         return err;
1024 }
1025
1026 static const struct got_error *
1027 is_object_packed(int *packed, struct got_repository *repo,
1028     struct got_object_id *id)
1029 {
1030         const struct got_error *err;
1031         struct got_object *obj;
1032
1033         *packed = 0;
1034
1035         err = got_object_open_packed(&obj, id, repo);
1036         if (err) {
1037                 if (err->code == GOT_ERR_NO_OBJ)
1038                         err = NULL;
1039                 return err;
1040         }
1041         got_object_close(obj);
1042         *packed = 1;
1043         return NULL;
1044 }
1045
1046 struct purge_loose_object_arg {
1047         struct got_repository *repo;
1048         got_cleanup_progress_cb progress_cb;
1049         void *progress_arg;
1050         struct got_ratelimit *rl;
1051         struct got_object_idset *traversed_ids;
1052         int nloose;
1053         int ncommits;
1054         int npacked;
1055         int npurged;
1056         off_t size_purged;
1057         int dry_run;
1058         time_t max_mtime;
1059         int ignore_mtime;
1060 };
1061
1062 static const struct got_error *
1063 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1064 {
1065         struct purge_loose_object_arg *a = arg;
1066         const struct got_error *err, *unlock_err = NULL;
1067         char *path = NULL;
1068         int packed, fd = -1;
1069         struct stat sb;
1070         struct got_lockfile *lf = NULL;
1071
1072         err = is_object_packed(&packed, a->repo, id);
1073         if (err)
1074                 return err;
1075
1076         if (!packed && got_object_idset_contains(a->traversed_ids, id))
1077                 return NULL;
1078
1079         if (packed)
1080                 a->npacked++;
1081
1082         err = got_object_get_path(&path, id, a->repo);
1083         if (err)
1084                 return err;
1085
1086         err = got_object_open_loose_fd(&fd, id, a->repo);
1087         if (err)
1088                 goto done;
1089
1090         if (fstat(fd, &sb) == -1) {
1091                 err = got_error_from_errno("fstat");
1092                 goto done;
1093         }
1094
1095         /*
1096          * Do not delete objects which are younger than our maximum
1097          * modification time threshold. This prevents a race where
1098          * new objects which are being added to the repository
1099          * concurrently would be deleted.
1100          */
1101         if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1102                 if (!a->dry_run) {
1103                         err = got_lockfile_lock(&lf, path, -1);
1104                         if (err)
1105                                 goto done;
1106                         if (unlink(path) == -1) {
1107                                 err = got_error_from_errno2("unlink", path);
1108                                 goto done;
1109                         }
1110                 }
1111
1112                 a->npurged++;
1113                 a->size_purged += sb.st_size;
1114                 err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1115                     a->rl, a->ncommits, a->nloose, a->npurged, -1);
1116                 if (err)
1117                         goto done;
1118         }
1119 done:
1120         if (fd != -1 && close(fd) == -1 && err == NULL)
1121                 err = got_error_from_errno("close");
1122         free(path);
1123         if (lf)
1124                 unlock_err = got_lockfile_unlock(lf, -1);
1125         return err ? err : unlock_err;
1126 }
1127
1128 static const struct got_error *
1129 repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1130     struct got_object_idset *traversed_ids,
1131     off_t *size_before, off_t *size_after, int ncommits, int *nloose,
1132     int *npacked, int *npurged, int dry_run, int ignore_mtime,
1133     time_t max_mtime, struct got_ratelimit *rl,
1134     got_cleanup_progress_cb progress_cb, void *progress_arg,
1135     got_cancel_cb cancel_cb, void *cancel_arg)
1136 {
1137         const struct got_error *err;
1138         struct got_object_idset *loose_ids;
1139         struct purge_loose_object_arg arg;
1140
1141         err = get_loose_object_ids(&loose_ids, size_before, ncommits,
1142             progress_cb, progress_arg, rl, repo);
1143         if (err)
1144                 return err;
1145         *nloose = got_object_idset_num_elements(loose_ids);
1146         if (*nloose == 0) {
1147                 got_object_idset_free(loose_ids);
1148                 if (progress_cb) {
1149                         err = progress_cb(progress_arg, 0, 0, 0, -1);
1150                         if (err)
1151                                 return err;
1152                 }
1153                 return NULL;
1154         }
1155
1156         memset(&arg, 0, sizeof(arg));
1157         arg.repo = repo;
1158         arg.progress_arg = progress_arg;
1159         arg.progress_cb = progress_cb;
1160         arg.rl = rl;
1161         arg.traversed_ids = traversed_ids;
1162         arg.nloose = *nloose;
1163         arg.npacked = 0;
1164         arg.npurged = 0;
1165         arg.size_purged = 0;
1166         arg.dry_run = dry_run;
1167         arg.max_mtime = max_mtime;
1168         arg.ignore_mtime = ignore_mtime;
1169         err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1170         if (err)
1171                 goto done;
1172
1173         *size_after = *size_before - arg.size_purged;
1174         *npacked = arg.npacked;
1175         *npurged = arg.npurged;
1176
1177         /* Produce a final progress report. */
1178         if (progress_cb) {
1179                 err = progress_cb(progress_arg, ncommits, *nloose,
1180                     arg.npurged, -1);
1181                 if (err)
1182                         goto done;
1183         }
1184 done:
1185         got_object_idset_free(loose_ids);
1186         return err;
1187 }
1188
1189 static const struct got_error *
1190 purge_redundant_pack(struct got_repository *repo, const char *packidx_path,
1191     int dry_run, int ignore_mtime, time_t max_mtime,
1192     int *remove, off_t *size_before, off_t *size_after)
1193 {
1194         static const char *ext[] = {".idx", ".pack", ".rev", ".bitmap",
1195             ".promisor", ".mtimes"};
1196         struct stat sb;
1197         char *dot, path[PATH_MAX];
1198         size_t i;
1199
1200         if (strlcpy(path, packidx_path, sizeof(path)) >= sizeof(path))
1201                 return got_error(GOT_ERR_NO_SPACE);
1202
1203         /*
1204          * Do not delete pack files which are younger than our maximum
1205          * modification time threshold.  This prevents a race where a
1206          * new pack file which is being added to the repository
1207          * concurrently would be deleted.
1208          */
1209         if (fstatat(got_repo_get_fd(repo), path, &sb, 0) == -1) {
1210                 if (errno == ENOENT)
1211                         return NULL;
1212                 return got_error_from_errno2("fstatat", path);
1213         }
1214         if (!ignore_mtime && sb.st_mtime > max_mtime)
1215                 *remove = 0;
1216
1217         /*
1218          * For compatibility with Git, if a matching .keep file exist
1219          * don't delete the packfile.
1220          */
1221         dot = strrchr(path, '.');
1222         *dot = '\0';
1223         if (strlcat(path, ".keep", sizeof(path)) >= sizeof(path))
1224                 return got_error(GOT_ERR_NO_SPACE);
1225         if (faccessat(got_repo_get_fd(repo), path, F_OK, 0) == 0)
1226                 *remove = 0;
1227
1228         for (i = 0; i < nitems(ext); ++i) {
1229                 *dot = '\0';
1230
1231                 if (strlcat(path, ext[i], sizeof(path)) >=
1232                     sizeof(path))
1233                         return got_error(GOT_ERR_NO_SPACE);
1234
1235                 if (fstatat(got_repo_get_fd(repo), path, &sb, 0) ==
1236                     -1) {
1237                         if (errno == ENOENT)
1238                                 continue;
1239                         return got_error_from_errno2("fstatat", path);
1240                 }
1241
1242                 *size_before += sb.st_size;
1243                 if (!*remove) {
1244                         *size_after += sb.st_size;
1245                         continue;
1246                 }
1247
1248                 if (dry_run)
1249                         continue;
1250
1251                 if (unlinkat(got_repo_get_fd(repo), path, 0) == -1) {
1252                         if (errno == ENOENT)
1253                                 continue;
1254                         return got_error_from_errno2("unlinkat",
1255                             path);
1256                 }
1257         }
1258
1259         return NULL;
1260 }
1261
1262 static const struct got_error *
1263 pack_is_redundant(int *redundant, struct got_repository *repo,
1264     struct got_object_idset *traversed_ids,
1265     const char *packidx_path, struct got_object_idset *idset)
1266 {
1267         const struct got_error *err;
1268         struct got_packidx *packidx;
1269         struct got_packidx_object_id *pid;
1270         struct got_object_id id;
1271         size_t i, nobjects;
1272
1273         *redundant = 1;
1274
1275         err = got_repo_get_packidx(&packidx, packidx_path, repo);
1276         if (err)
1277                 return err;
1278
1279         nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1280         for (i = 0; i < nobjects; ++i) {
1281                 pid = &packidx->hdr.sorted_ids[i];
1282
1283                 memset(&id, 0, sizeof(id));
1284                 memcpy(&id.sha1, pid->sha1, sizeof(id.sha1));
1285
1286                 if (got_object_idset_contains(idset, &id))
1287                         continue;
1288
1289                 if (!got_object_idset_contains(traversed_ids, &id))
1290                         continue;
1291
1292                 *redundant = 0;
1293                 err = got_object_idset_add(idset, &id, NULL);
1294                 if (err)
1295                         return err;
1296         }
1297
1298         return NULL;
1299 }
1300
1301 struct pack_info {
1302         const char      *path;
1303         size_t           nobjects;
1304 };
1305
1306 static int
1307 pack_info_cmp(const void *a, const void *b)
1308 {
1309         const struct pack_info  *pa, *pb;
1310
1311         pa = a;
1312         pb = b;
1313         if (pa->nobjects == pb->nobjects)
1314                 return strcmp(pa->path, pb->path);
1315         if (pa->nobjects > pb->nobjects)
1316                 return -1;
1317         return 1;
1318 }
1319
1320 static const struct got_error *
1321 repo_purge_redundant_packfiles(struct got_repository *repo,
1322     struct got_object_idset *traversed_ids,
1323     off_t *size_before, off_t *size_after, int dry_run, int ignore_mtime,
1324     time_t max_mtime, int nloose, int ncommits, int npurged,
1325     struct got_ratelimit *rl,
1326     got_cleanup_progress_cb progress_cb, void *progress_arg,
1327     got_cancel_cb cancel_cb, void *cancel_arg)
1328 {
1329         const struct got_error *err;
1330         struct pack_info *pinfo, *sorted = NULL;
1331         struct got_packidx *packidx;
1332         struct got_object_idset *idset = NULL;
1333         struct got_pathlist_entry *pe;
1334         size_t i, npacks;
1335         int remove, redundant_packs = 0;
1336
1337         npacks = 0;
1338         TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
1339                 npacks++;
1340
1341         if (npacks == 0)
1342                 return NULL;
1343
1344         sorted = calloc(npacks, sizeof(*sorted));
1345         if (sorted == NULL)
1346                 return got_error_from_errno("calloc");
1347
1348         i = 0;
1349         TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1350                 err = got_repo_get_packidx(&packidx, pe->path, repo);
1351                 if (err)
1352                         goto done;
1353
1354                 pinfo = &sorted[i++];
1355                 pinfo->path = pe->path;
1356                 pinfo->nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1357         }
1358         qsort(sorted, npacks, sizeof(*sorted), pack_info_cmp);
1359
1360         idset = got_object_idset_alloc();
1361         if (idset == NULL) {
1362                 err = got_error_from_errno("got_object_idset_alloc");
1363                 goto done;
1364         }
1365
1366         for (i = 0; i < npacks; ++i) {
1367                 if (cancel_cb) {
1368                         err = (*cancel_cb)(cancel_arg);
1369                         if (err)
1370                                 break;
1371                 }
1372
1373                 err = pack_is_redundant(&remove, repo, traversed_ids,
1374                     sorted[i].path, idset);
1375                 if (err)
1376                         goto done;
1377                 err = purge_redundant_pack(repo, sorted[i].path, dry_run,
1378                     ignore_mtime, max_mtime, &remove, size_before, size_after);
1379                 if (err)
1380                         goto done;
1381                 if (!remove)
1382                         continue;
1383                 err = report_cleanup_progress(progress_cb, progress_arg,
1384                     rl, ncommits, nloose, npurged, ++redundant_packs);
1385                 if (err)
1386                         goto done;
1387         }
1388
1389         /* Produce a final progress report. */
1390         if (progress_cb) {
1391                 err = progress_cb(progress_arg, ncommits, nloose, npurged,
1392                     redundant_packs);
1393                 if (err)
1394                         goto done;
1395         }
1396  done:
1397         free(sorted);
1398         if (idset)
1399                 got_object_idset_free(idset);
1400         return err;
1401 }
1402
1403 const struct got_error *
1404 got_repo_cleanup(struct got_repository *repo,
1405     off_t *loose_before, off_t *loose_after,
1406     off_t *pack_before, off_t *pack_after,
1407     int *ncommits, int *nloose, int *npacked, int dry_run, int ignore_mtime,
1408     got_cleanup_progress_cb progress_cb, void *progress_arg,
1409     got_cancel_cb cancel_cb, void *cancel_arg)
1410 {
1411         const struct got_error *unlock_err, *err = NULL;
1412         struct got_lockfile *lk = NULL;
1413         struct got_ratelimit rl;
1414         struct got_reflist_head refs;
1415         struct got_object_idset *traversed_ids = NULL;
1416         struct got_reflist_entry *re;
1417         struct got_object_id **referenced_ids;
1418         int i, nreferenced;
1419         int npurged = 0;
1420         time_t max_mtime = 0;
1421
1422         TAILQ_INIT(&refs);
1423         got_ratelimit_init(&rl, 0, 500);
1424
1425         *loose_before = 0;
1426         *loose_after = 0;
1427         *pack_before = 0;
1428         *pack_after = 0;
1429         *ncommits = 0;
1430         *nloose = 0;
1431         *npacked = 0;
1432
1433         err = repo_cleanup_lock(repo, &lk);
1434         if (err)
1435                 return err;
1436
1437         traversed_ids = got_object_idset_alloc();
1438         if (traversed_ids == NULL) {
1439                 err = got_error_from_errno("got_object_idset_alloc");
1440                 goto done;
1441         }
1442
1443         err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1444         if (err)
1445                 goto done;
1446         if (!ignore_mtime) {
1447                 TAILQ_FOREACH(re, &refs, entry) {
1448                         time_t mtime = got_ref_get_mtime(re->ref);
1449                         if (mtime > max_mtime)
1450                                 max_mtime = mtime;
1451                 }
1452                 /*
1453                  * For safety, keep objects created within 10 minutes
1454                  * before the youngest reference was created.
1455                  */
1456                 if (max_mtime >= 600)
1457                         max_mtime -= 600;
1458         }
1459
1460         err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1461             (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1462             &refs, repo, cancel_cb, cancel_arg);
1463         if (err)
1464                 goto done;
1465
1466         for (i = 0; i < nreferenced; i++) {
1467                 struct got_object_id *id = referenced_ids[i];
1468                 err = load_commit_or_tag(ncommits, traversed_ids,
1469                     id, repo, progress_cb, progress_arg, &rl,
1470                     cancel_cb, cancel_arg);
1471                 if (err)
1472                         goto done;
1473         }
1474
1475         err = repo_purge_unreferenced_loose_objects(repo, traversed_ids,
1476             loose_before, loose_after, *ncommits, nloose, npacked, &npurged,
1477             dry_run, ignore_mtime, max_mtime, &rl, progress_cb, progress_arg,
1478             cancel_cb, cancel_arg);
1479         if (err)
1480                 goto done;
1481
1482         err = repo_purge_redundant_packfiles(repo, traversed_ids,
1483             pack_before, pack_after, dry_run, ignore_mtime, max_mtime,
1484             *nloose, *ncommits, npurged, &rl, progress_cb, progress_arg,
1485             cancel_cb, cancel_arg);
1486         if (err)
1487                 goto done;
1488
1489  done:
1490         if (lk) {
1491                 unlock_err = got_lockfile_unlock(lk, got_repo_get_fd(repo));
1492                 if (err == NULL)
1493                         err = unlock_err;
1494         }
1495         if (traversed_ids)
1496                 got_object_idset_free(traversed_ids);
1497         return err;
1498 }
1499
1500 const struct got_error *
1501 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1502     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1503     got_cancel_cb cancel_cb, void *cancel_arg)
1504 {
1505         const struct got_error *err = NULL;
1506         DIR *packdir = NULL;
1507         struct dirent *dent;
1508         char *pack_relpath = NULL;
1509         int packdir_fd;
1510         struct stat sb;
1511
1512         packdir_fd = openat(got_repo_get_fd(repo),
1513             GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1514         if (packdir_fd == -1) {
1515                 if (errno == ENOENT)
1516                         return NULL;
1517                 return got_error_from_errno_fmt("openat: %s/%s",
1518                     got_repo_get_path_git_dir(repo),
1519                     GOT_OBJECTS_PACK_DIR);
1520         }
1521
1522         packdir = fdopendir(packdir_fd);
1523         if (packdir == NULL) {
1524                 err = got_error_from_errno("fdopendir");
1525                 goto done;
1526         }
1527
1528         while ((dent = readdir(packdir)) != NULL) {
1529                 if (cancel_cb) {
1530                         err = cancel_cb(cancel_arg);
1531                         if (err)
1532                                 goto done;
1533                 }
1534
1535                 if (!got_repo_is_packidx_filename(dent->d_name,
1536                     strlen(dent->d_name)))
1537                         continue;
1538
1539                 err = got_packidx_get_packfile_path(&pack_relpath,
1540                     dent->d_name);
1541                 if (err)
1542                         goto done;
1543
1544                 if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1545                         free(pack_relpath);
1546                         pack_relpath = NULL;
1547                         continue;
1548                 }
1549                 if (errno != ENOENT) {
1550                         err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1551                             got_repo_get_path_git_dir(repo),
1552                             GOT_OBJECTS_PACK_DIR,
1553                             pack_relpath);
1554                         goto done;
1555                 }
1556
1557                 if (!dry_run) {
1558                         if (unlinkat(packdir_fd, dent->d_name, 0) == -1) {
1559                                 err = got_error_from_errno("unlinkat");
1560                                 goto done;
1561                         }
1562                 }
1563                 if (progress_cb) {
1564                         char *path;
1565                         if (asprintf(&path, "%s/%s/%s",
1566                             got_repo_get_path_git_dir(repo),
1567                             GOT_OBJECTS_PACK_DIR,
1568                             dent->d_name) == -1) {
1569                                 err = got_error_from_errno("asprintf");
1570                                 goto done;
1571                         }
1572                         err = progress_cb(progress_arg, path);
1573                         free(path);
1574                         if (err)
1575                                 goto done;
1576                 }
1577                 free(pack_relpath);
1578                 pack_relpath = NULL;
1579         }
1580 done:
1581         if (packdir && closedir(packdir) != 0 && err == NULL)
1582                 err = got_error_from_errno("closedir");
1583         free(pack_relpath);
1584         return err;
1585 }