builtin/gc.c

   1 /*
   2  * git gc builtin command
   3  *
   4  * Cleanup unreachable files and optimize the repository.
   5  *
   6  * Copyright (c) 2007 James Bowes
   7  *
   8  * Based on git-gc.sh, which is
   9  *
  10  * Copyright (c) 2006 Shawn O. Pearce
  11  */
  12
  13 #include "builtin.h"
  14 #include "repository.h"
  15 #include "config.h"
  16 #include "tempfile.h"
  17 #include "lockfile.h"
  18 #include "parse-options.h"
  19 #include "run-command.h"
  20 #include "sigchain.h"
  21 #include "strvec.h"
  22 #include "commit.h"
  23 #include "commit-graph.h"
  24 #include "packfile.h"
  25 #include "object-store.h"
  26 #include "pack.h"
  27 #include "pack-objects.h"
  28 #include "blob.h"
  29 #include "tree.h"
  30 #include "promisor-remote.h"
  31 #include "refs.h"
  32 #include "remote.h"
  33 #include "object-store.h"
  34 #include "exec-cmd.h"
  35
  36 #define FAILED_RUN "failed to run %s"
  37
  38 static const char * const builtin_gc_usage[] = {
  39         N_("git gc [<options>]"),
  40         NULL
  41 };
  42
  43 static int pack_refs = 1;
  44 static int prune_reflogs = 1;
  45 static int aggressive_depth = 50;
  46 static int aggressive_window = 250;
  47 static int gc_auto_threshold = 6700;
  48 static int gc_auto_pack_limit = 50;
  49 static int detach_auto = 1;
  50 static timestamp_t gc_log_expire_time;
  51 static const char *gc_log_expire = "1.day.ago";
  52 static const char *prune_expire = "2.weeks.ago";
  53 static const char *prune_worktrees_expire = "3.months.ago";
  54 static unsigned long big_pack_threshold;
  55 static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
  56
  57 static struct strvec reflog = STRVEC_INIT;
  58 static struct strvec repack = STRVEC_INIT;
  59 static struct strvec prune = STRVEC_INIT;
  60 static struct strvec prune_worktrees = STRVEC_INIT;
  61 static struct strvec rerere = STRVEC_INIT;
  62
  63 static struct tempfile *pidfile;
  64 static struct lock_file log_lock;
  65
  66 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  67
  68 static void clean_pack_garbage(void)
  69 {
  70         int i;
  71         for (i = 0; i < pack_garbage.nr; i++)
  72                 unlink_or_warn(pack_garbage.items[i].string);
  73         string_list_clear(&pack_garbage, 0);
  74 }
  75
  76 static void report_pack_garbage(unsigned seen_bits, const char *path)
  77 {
  78         if (seen_bits == PACKDIR_FILE_IDX)
  79                 string_list_append(&pack_garbage, path);
  80 }
  81
  82 static void process_log_file(void)
  83 {
  84         struct stat st;
  85         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  86                 /*
  87                  * Perhaps there was an i/o error or another
  88                  * unlikely situation.  Try to make a note of
  89                  * this in gc.log along with any existing
  90                  * messages.
  91                  */
  92                 int saved_errno = errno;
  93                 fprintf(stderr, _("Failed to fstat %s: %s"),
  94                         get_lock_file_path(&log_lock),
  95                         strerror(saved_errno));
  96                 fflush(stderr);
  97                 commit_lock_file(&log_lock);
  98                 errno = saved_errno;
  99         } else if (st.st_size) {
 100                 /* There was some error recorded in the lock file */
 101                 commit_lock_file(&log_lock);
 102         } else {
 103                 /* No error, clean up any old gc.log */
 104                 unlink(git_path("gc.log"));
 105                 rollback_lock_file(&log_lock);
 106         }
 107 }
 108
 109 static void process_log_file_at_exit(void)
 110 {
 111         fflush(stderr);
 112         process_log_file();
 113 }
 114
 115 static void process_log_file_on_signal(int signo)
 116 {
 117         process_log_file();
 118         sigchain_pop(signo);
 119         raise(signo);
 120 }
 121
 122 static int gc_config_is_timestamp_never(const char *var)
 123 {
 124         const char *value;
 125         timestamp_t expire;
 126
 127         if (!git_config_get_value(var, &value) && value) {
 128                 if (parse_expiry_date(value, &expire))
 129                         die(_("failed to parse '%s' value '%s'"), var, value);
 130                 return expire == 0;
 131         }
 132         return 0;
 133 }
 134
 135 static void gc_config(void)
 136 {
 137         const char *value;
 138
 139         if (!git_config_get_value("gc.packrefs", &value)) {
 140                 if (value && !strcmp(value, "notbare"))
 141                         pack_refs = -1;
 142                 else
 143                         pack_refs = git_config_bool("gc.packrefs", value);
 144         }
 145
 146         if (gc_config_is_timestamp_never("gc.reflogexpire") &&
 147             gc_config_is_timestamp_never("gc.reflogexpireunreachable"))
 148                 prune_reflogs = 0;
 149
 150         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 151         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 152         git_config_get_int("gc.auto", &gc_auto_threshold);
 153         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 154         git_config_get_bool("gc.autodetach", &detach_auto);
 155         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 156         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 157         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 158
 159         git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold);
 160         git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size);
 161
 162         git_config(git_default_config, NULL);
 163 }
 164
 165 struct maintenance_run_opts;
 166 static int maintenance_task_pack_refs(MAYBE_UNUSED struct maintenance_run_opts *opts)
 167 {
 168         struct strvec pack_refs_cmd = STRVEC_INIT;
 169         strvec_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 170
 171         return run_command_v_opt(pack_refs_cmd.v, RUN_GIT_CMD);
 172 }
 173
 174 static int too_many_loose_objects(void)
 175 {
 176         /*
 177          * Quickly check if a "gc" is needed, by estimating how
 178          * many loose objects there are.  Because SHA-1 is evenly
 179          * distributed, we can check only one and get a reasonable
 180          * estimate.
 181          */
 182         DIR *dir;
 183         struct dirent *ent;
 184         int auto_threshold;
 185         int num_loose = 0;
 186         int needed = 0;
 187         const unsigned hexsz_loose = the_hash_algo->hexsz - 2;
 188
 189         dir = opendir(git_path("objects/17"));
 190         if (!dir)
 191                 return 0;
 192
 193         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
 194         while ((ent = readdir(dir)) != NULL) {
 195                 if (strspn(ent->d_name, "0123456789abcdef") != hexsz_loose ||
 196                     ent->d_name[hexsz_loose] != '\0')
 197                         continue;
 198                 if (++num_loose > auto_threshold) {
 199                         needed = 1;
 200                         break;
 201                 }
 202         }
 203         closedir(dir);
 204         return needed;
 205 }
 206
 207 static struct packed_git *find_base_packs(struct string_list *packs,
 208                                           unsigned long limit)
 209 {
 210         struct packed_git *p, *base = NULL;
 211
 212         for (p = get_all_packs(the_repository); p; p = p->next) {
 213                 if (!p->pack_local)
 214                         continue;
 215                 if (limit) {
 216                         if (p->pack_size >= limit)
 217                                 string_list_append(packs, p->pack_name);
 218                 } else if (!base || base->pack_size < p->pack_size) {
 219                         base = p;
 220                 }
 221         }
 222
 223         if (base)
 224                 string_list_append(packs, base->pack_name);
 225
 226         return base;
 227 }
 228
 229 static int too_many_packs(void)
 230 {
 231         struct packed_git *p;
 232         int cnt;
 233
 234         if (gc_auto_pack_limit <= 0)
 235                 return 0;
 236
 237         for (cnt = 0, p = get_all_packs(the_repository); p; p = p->next) {
 238                 if (!p->pack_local)
 239                         continue;
 240                 if (p->pack_keep)
 241                         continue;
 242                 /*
 243                  * Perhaps check the size of the pack and count only
 244                  * very small ones here?
 245                  */
 246                 cnt++;
 247         }
 248         return gc_auto_pack_limit < cnt;
 249 }
 250
 251 static uint64_t total_ram(void)
 252 {
 253 #if defined(HAVE_SYSINFO)
 254         struct sysinfo si;
 255
 256         if (!sysinfo(&si))
 257                 return si.totalram;
 258 #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM))
 259         int64_t physical_memory;
 260         int mib[2];
 261         size_t length;
 262
 263         mib[0] = CTL_HW;
 264 # if defined(HW_MEMSIZE)
 265         mib[1] = HW_MEMSIZE;
 266 # else
 267         mib[1] = HW_PHYSMEM;
 268 # endif
 269         length = sizeof(int64_t);
 270         if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
 271                 return physical_memory;
 272 #elif defined(GIT_WINDOWS_NATIVE)
 273         MEMORYSTATUSEX memInfo;
 274
 275         memInfo.dwLength = sizeof(MEMORYSTATUSEX);
 276         if (GlobalMemoryStatusEx(&memInfo))
 277                 return memInfo.ullTotalPhys;
 278 #endif
 279         return 0;
 280 }
 281
 282 static uint64_t estimate_repack_memory(struct packed_git *pack)
 283 {
 284         unsigned long nr_objects = approximate_object_count();
 285         size_t os_cache, heap;
 286
 287         if (!pack || !nr_objects)
 288                 return 0;
 289
 290         /*
 291          * First we have to scan through at least one pack.
 292          * Assume enough room in OS file cache to keep the entire pack
 293          * or we may accidentally evict data of other processes from
 294          * the cache.
 295          */
 296         os_cache = pack->pack_size + pack->index_size;
 297         /* then pack-objects needs lots more for book keeping */
 298         heap = sizeof(struct object_entry) * nr_objects;
 299         /*
 300          * internal rev-list --all --objects takes up some memory too,
 301          * let's say half of it is for blobs
 302          */
 303         heap += sizeof(struct blob) * nr_objects / 2;
 304         /*
 305          * and the other half is for trees (commits and tags are
 306          * usually insignificant)
 307          */
 308         heap += sizeof(struct tree) * nr_objects / 2;
 309         /* and then obj_hash[], underestimated in fact */
 310         heap += sizeof(struct object *) * nr_objects;
 311         /* revindex is used also */
 312         heap += (sizeof(off_t) + sizeof(uint32_t)) * nr_objects;
 313         /*
 314          * read_sha1_file() (either at delta calculation phase, or
 315          * writing phase) also fills up the delta base cache
 316          */
 317         heap += delta_base_cache_limit;
 318         /* and of course pack-objects has its own delta cache */
 319         heap += max_delta_cache_size;
 320
 321         return os_cache + heap;
 322 }
 323
 324 static int keep_one_pack(struct string_list_item *item, void *data)
 325 {
 326         strvec_pushf(&repack, "--keep-pack=%s", basename(item->string));
 327         return 0;
 328 }
 329
 330 static void add_repack_all_option(struct string_list *keep_pack)
 331 {
 332         if (prune_expire && !strcmp(prune_expire, "now"))
 333                 strvec_push(&repack, "-a");
 334         else {
 335                 strvec_push(&repack, "-A");
 336                 if (prune_expire)
 337                         strvec_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 338         }
 339
 340         if (keep_pack)
 341                 for_each_string_list(keep_pack, keep_one_pack, NULL);
 342 }
 343
 344 static void add_repack_incremental_option(void)
 345 {
 346         strvec_push(&repack, "--no-write-bitmap-index");
 347 }
 348
 349 static int need_to_gc(void)
 350 {
 351         /*
 352          * Setting gc.auto to 0 or negative can disable the
 353          * automatic gc.
 354          */
 355         if (gc_auto_threshold <= 0)
 356                 return 0;
 357
 358         /*
 359          * If there are too many loose objects, but not too many
 360          * packs, we run "repack -d -l".  If there are too many packs,
 361          * we run "repack -A -d -l".  Otherwise we tell the caller
 362          * there is no need.
 363          */
 364         if (too_many_packs()) {
 365                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 366
 367                 if (big_pack_threshold) {
 368                         find_base_packs(&keep_pack, big_pack_threshold);
 369                         if (keep_pack.nr >= gc_auto_pack_limit) {
 370                                 big_pack_threshold = 0;
 371                                 string_list_clear(&keep_pack, 0);
 372                                 find_base_packs(&keep_pack, 0);
 373                         }
 374                 } else {
 375                         struct packed_git *p = find_base_packs(&keep_pack, 0);
 376                         uint64_t mem_have, mem_want;
 377
 378                         mem_have = total_ram();
 379                         mem_want = estimate_repack_memory(p);
 380
 381                         /*
 382                          * Only allow 1/2 of memory for pack-objects, leave
 383                          * the rest for the OS and other processes in the
 384                          * system.
 385                          */
 386                         if (!mem_have || mem_want < mem_have / 2)
 387                                 string_list_clear(&keep_pack, 0);
 388                 }
 389
 390                 add_repack_all_option(&keep_pack);
 391                 string_list_clear(&keep_pack, 0);
 392         } else if (too_many_loose_objects())
 393                 add_repack_incremental_option();
 394         else
 395                 return 0;
 396
 397         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 398                 return 0;
 399         return 1;
 400 }
 401
 402 /* return NULL on success, else hostname running the gc */
 403 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 404 {
 405         struct lock_file lock = LOCK_INIT;
 406         char my_host[HOST_NAME_MAX + 1];
 407         struct strbuf sb = STRBUF_INIT;
 408         struct stat st;
 409         uintmax_t pid;
 410         FILE *fp;
 411         int fd;
 412         char *pidfile_path;
 413
 414         if (is_tempfile_active(pidfile))
 415                 /* already locked */
 416                 return NULL;
 417
 418         if (xgethostname(my_host, sizeof(my_host)))
 419                 xsnprintf(my_host, sizeof(my_host), "unknown");
 420
 421         pidfile_path = git_pathdup("gc.pid");
 422         fd = hold_lock_file_for_update(&lock, pidfile_path,
 423                                        LOCK_DIE_ON_ERROR);
 424         if (!force) {
 425                 static char locking_host[HOST_NAME_MAX + 1];
 426                 static char *scan_fmt;
 427                 int should_exit;
 428
 429                 if (!scan_fmt)
 430                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
 431                 fp = fopen(pidfile_path, "r");
 432                 memset(locking_host, 0, sizeof(locking_host));
 433                 should_exit =
 434                         fp != NULL &&
 435                         !fstat(fileno(fp), &st) &&
 436                         /*
 437                          * 12 hour limit is very generous as gc should
 438                          * never take that long. On the other hand we
 439                          * don't really need a strict limit here,
 440                          * running gc --auto one day late is not a big
 441                          * problem. --force can be used in manual gc
 442                          * after the user verifies that no gc is
 443                          * running.
 444                          */
 445                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 446                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 447                         /* be gentle to concurrent "gc" on remote hosts */
 448                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 449                 if (fp != NULL)
 450                         fclose(fp);
 451                 if (should_exit) {
 452                         if (fd >= 0)
 453                                 rollback_lock_file(&lock);
 454                         *ret_pid = pid;
 455                         free(pidfile_path);
 456                         return locking_host;
 457                 }
 458         }
 459
 460         strbuf_addf(&sb, "%"PRIuMAX" %s",
 461                     (uintmax_t) getpid(), my_host);
 462         write_in_full(fd, sb.buf, sb.len);
 463         strbuf_release(&sb);
 464         commit_lock_file(&lock);
 465         pidfile = register_tempfile(pidfile_path);
 466         free(pidfile_path);
 467         return NULL;
 468 }
 469
 470 /*
 471  * Returns 0 if there was no previous error and gc can proceed, 1 if
 472  * gc should not proceed due to an error in the last run. Prints a
 473  * message and returns -1 if an error occurred while reading gc.log
 474  */
 475 static int report_last_gc_error(void)
 476 {
 477         struct strbuf sb = STRBUF_INIT;
 478         int ret = 0;
 479         ssize_t len;
 480         struct stat st;
 481         char *gc_log_path = git_pathdup("gc.log");
 482
 483         if (stat(gc_log_path, &st)) {
 484                 if (errno == ENOENT)
 485                         goto done;
 486
 487                 ret = error_errno(_("cannot stat '%s'"), gc_log_path);
 488                 goto done;
 489         }
 490
 491         if (st.st_mtime < gc_log_expire_time)
 492                 goto done;
 493
 494         len = strbuf_read_file(&sb, gc_log_path, 0);
 495         if (len < 0)
 496                 ret = error_errno(_("cannot read '%s'"), gc_log_path);
 497         else if (len > 0) {
 498                 /*
 499                  * A previous gc failed.  Report the error, and don't
 500                  * bother with an automatic gc run since it is likely
 501                  * to fail in the same way.
 502                  */
 503                 warning(_("The last gc run reported the following. "
 504                                "Please correct the root cause\n"
 505                                "and remove %s.\n"
 506                                "Automatic cleanup will not be performed "
 507                                "until the file is removed.\n\n"
 508                                "%s"),
 509                             gc_log_path, sb.buf);
 510                 ret = 1;
 511         }
 512         strbuf_release(&sb);
 513 done:
 514         free(gc_log_path);
 515         return ret;
 516 }
 517
 518 static void gc_before_repack(void)
 519 {
 520         /*
 521          * We may be called twice, as both the pre- and
 522          * post-daemonized phases will call us, but running these
 523          * commands more than once is pointless and wasteful.
 524          */
 525         static int done = 0;
 526         if (done++)
 527                 return;
 528
 529         if (pack_refs && maintenance_task_pack_refs(NULL))
 530                 die(FAILED_RUN, "pack-refs");
 531
 532         if (prune_reflogs && run_command_v_opt(reflog.v, RUN_GIT_CMD))
 533                 die(FAILED_RUN, reflog.v[0]);
 534 }
 535
 536 int cmd_gc(int argc, const char **argv, const char *prefix)
 537 {
 538         int aggressive = 0;
 539         int auto_gc = 0;
 540         int quiet = 0;
 541         int force = 0;
 542         const char *name;
 543         pid_t pid;
 544         int daemonized = 0;
 545         int keep_largest_pack = -1;
 546         timestamp_t dummy;
 547
 548         struct option builtin_gc_options[] = {
 549                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 550                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 551                         N_("prune unreferenced objects"),
 552                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 553                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 554                 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
 555                            PARSE_OPT_NOCOMPLETE),
 556                 OPT_BOOL_F(0, "force", &force,
 557                            N_("force running gc even if there may be another gc running"),
 558                            PARSE_OPT_NOCOMPLETE),
 559                 OPT_BOOL(0, "keep-largest-pack", &keep_largest_pack,
 560                          N_("repack all other packs except the largest pack")),
 561                 OPT_END()
 562         };
 563
 564         if (argc == 2 && !strcmp(argv[1], "-h"))
 565                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 566
 567         strvec_pushl(&reflog, "reflog", "expire", "--all", NULL);
 568         strvec_pushl(&repack, "repack", "-d", "-l", NULL);
 569         strvec_pushl(&prune, "prune", "--expire", NULL);
 570         strvec_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 571         strvec_pushl(&rerere, "rerere", "gc", NULL);
 572
 573         /* default expiry time, overwritten in gc_config */
 574         gc_config();
 575         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 576                 die(_("failed to parse gc.logexpiry value %s"), gc_log_expire);
 577
 578         if (pack_refs < 0)
 579                 pack_refs = !is_bare_repository();
 580
 581         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 582                              builtin_gc_usage, 0);
 583         if (argc > 0)
 584                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 585
 586         if (prune_expire && parse_expiry_date(prune_expire, &dummy))
 587                 die(_("failed to parse prune expiry value %s"), prune_expire);
 588
 589         if (aggressive) {
 590                 strvec_push(&repack, "-f");
 591                 if (aggressive_depth > 0)
 592                         strvec_pushf(&repack, "--depth=%d", aggressive_depth);
 593                 if (aggressive_window > 0)
 594                         strvec_pushf(&repack, "--window=%d", aggressive_window);
 595         }
 596         if (quiet)
 597                 strvec_push(&repack, "-q");
 598
 599         if (auto_gc) {
 600                 /*
 601                  * Auto-gc should be least intrusive as possible.
 602                  */
 603                 if (!need_to_gc())
 604                         return 0;
 605                 if (!quiet) {
 606                         if (detach_auto)
 607                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 608                         else
 609                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 610                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 611                 }
 612                 if (detach_auto) {
 613                         int ret = report_last_gc_error();
 614                         if (ret < 0)
 615                                 /* an I/O error occurred, already reported */
 616                                 exit(128);
 617                         if (ret == 1)
 618                                 /* Last gc --auto failed. Skip this one. */
 619                                 return 0;
 620
 621                         if (lock_repo_for_gc(force, &pid))
 622                                 return 0;
 623                         gc_before_repack(); /* dies on failure */
 624                         delete_tempfile(&pidfile);
 625
 626                         /*
 627                          * failure to daemonize is ok, we'll continue
 628                          * in foreground
 629                          */
 630                         daemonized = !daemonize();
 631                 }
 632         } else {
 633                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 634
 635                 if (keep_largest_pack != -1) {
 636                         if (keep_largest_pack)
 637                                 find_base_packs(&keep_pack, 0);
 638                 } else if (big_pack_threshold) {
 639                         find_base_packs(&keep_pack, big_pack_threshold);
 640                 }
 641
 642                 add_repack_all_option(&keep_pack);
 643                 string_list_clear(&keep_pack, 0);
 644         }
 645
 646         name = lock_repo_for_gc(force, &pid);
 647         if (name) {
 648                 if (auto_gc)
 649                         return 0; /* be quiet on --auto */
 650                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 651                     name, (uintmax_t)pid);
 652         }
 653
 654         if (daemonized) {
 655                 hold_lock_file_for_update(&log_lock,
 656                                           git_path("gc.log"),
 657                                           LOCK_DIE_ON_ERROR);
 658                 dup2(get_lock_file_fd(&log_lock), 2);
 659                 sigchain_push_common(process_log_file_on_signal);
 660                 atexit(process_log_file_at_exit);
 661         }
 662
 663         gc_before_repack();
 664
 665         if (!repository_format_precious_objects) {
 666                 close_object_store(the_repository->objects);
 667                 if (run_command_v_opt(repack.v, RUN_GIT_CMD))
 668                         die(FAILED_RUN, repack.v[0]);
 669
 670                 if (prune_expire) {
 671                         strvec_push(&prune, prune_expire);
 672                         if (quiet)
 673                                 strvec_push(&prune, "--no-progress");
 674                         if (has_promisor_remote())
 675                                 strvec_push(&prune,
 676                                             "--exclude-promisor-objects");
 677                         if (run_command_v_opt(prune.v, RUN_GIT_CMD))
 678                                 die(FAILED_RUN, prune.v[0]);
 679                 }
 680         }
 681
 682         if (prune_worktrees_expire) {
 683                 strvec_push(&prune_worktrees, prune_worktrees_expire);
 684                 if (run_command_v_opt(prune_worktrees.v, RUN_GIT_CMD))
 685                         die(FAILED_RUN, prune_worktrees.v[0]);
 686         }
 687
 688         if (run_command_v_opt(rerere.v, RUN_GIT_CMD))
 689                 die(FAILED_RUN, rerere.v[0]);
 690
 691         report_garbage = report_pack_garbage;
 692         reprepare_packed_git(the_repository);
 693         if (pack_garbage.nr > 0) {
 694                 close_object_store(the_repository->objects);
 695                 clean_pack_garbage();
 696         }
 697
 698         prepare_repo_settings(the_repository);
 699         if (the_repository->settings.gc_write_commit_graph == 1)
 700                 write_commit_graph_reachable(the_repository->objects->odb,
 701                                              !quiet && !daemonized ? COMMIT_GRAPH_WRITE_PROGRESS : 0,
 702                                              NULL);
 703
 704         if (auto_gc && too_many_loose_objects())
 705                 warning(_("There are too many unreachable loose objects; "
 706                         "run 'git prune' to remove them."));
 707
 708         if (!daemonized)
 709                 unlink(git_path("gc.log"));
 710
 711         return 0;
 712 }
 713
 714 static const char *const builtin_maintenance_run_usage[] = {
 715         N_("git maintenance run [--auto] [--[no-]quiet] [--task=<task>] [--schedule]"),
 716         NULL
 717 };
 718
 719 enum schedule_priority {
 720         SCHEDULE_NONE = 0,
 721         SCHEDULE_WEEKLY = 1,
 722         SCHEDULE_DAILY = 2,
 723         SCHEDULE_HOURLY = 3,
 724 };
 725
 726 static enum schedule_priority parse_schedule(const char *value)
 727 {
 728         if (!value)
 729                 return SCHEDULE_NONE;
 730         if (!strcasecmp(value, "hourly"))
 731                 return SCHEDULE_HOURLY;
 732         if (!strcasecmp(value, "daily"))
 733                 return SCHEDULE_DAILY;
 734         if (!strcasecmp(value, "weekly"))
 735                 return SCHEDULE_WEEKLY;
 736         return SCHEDULE_NONE;
 737 }
 738
 739 static int maintenance_opt_schedule(const struct option *opt, const char *arg,
 740                                     int unset)
 741 {
 742         enum schedule_priority *priority = opt->value;
 743
 744         if (unset)
 745                 die(_("--no-schedule is not allowed"));
 746
 747         *priority = parse_schedule(arg);
 748
 749         if (!*priority)
 750                 die(_("unrecognized --schedule argument '%s'"), arg);
 751
 752         return 0;
 753 }
 754
 755 struct maintenance_run_opts {
 756         int auto_flag;
 757         int quiet;
 758         enum schedule_priority schedule;
 759 };
 760
 761 /* Remember to update object flag allocation in object.h */
 762 #define SEEN            (1u<<0)
 763
 764 struct cg_auto_data {
 765         int num_not_in_graph;
 766         int limit;
 767 };
 768
 769 static int dfs_on_ref(const char *refname,
 770                       const struct object_id *oid, int flags,
 771                       void *cb_data)
 772 {
 773         struct cg_auto_data *data = (struct cg_auto_data *)cb_data;
 774         int result = 0;
 775         struct object_id peeled;
 776         struct commit_list *stack = NULL;
 777         struct commit *commit;
 778
 779         if (!peel_iterated_oid(oid, &peeled))
 780                 oid = &peeled;
 781         if (oid_object_info(the_repository, oid, NULL) != OBJ_COMMIT)
 782                 return 0;
 783
 784         commit = lookup_commit(the_repository, oid);
 785         if (!commit)
 786                 return 0;
 787         if (parse_commit(commit) ||
 788             commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 789                 return 0;
 790
 791         data->num_not_in_graph++;
 792
 793         if (data->num_not_in_graph >= data->limit)
 794                 return 1;
 795
 796         commit_list_append(commit, &stack);
 797
 798         while (!result && stack) {
 799                 struct commit_list *parent;
 800
 801                 commit = pop_commit(&stack);
 802
 803                 for (parent = commit->parents; parent; parent = parent->next) {
 804                         if (parse_commit(parent->item) ||
 805                             commit_graph_position(parent->item) != COMMIT_NOT_FROM_GRAPH ||
 806                             parent->item->object.flags & SEEN)
 807                                 continue;
 808
 809                         parent->item->object.flags |= SEEN;
 810                         data->num_not_in_graph++;
 811
 812                         if (data->num_not_in_graph >= data->limit) {
 813                                 result = 1;
 814                                 break;
 815                         }
 816
 817                         commit_list_append(parent->item, &stack);
 818                 }
 819         }
 820
 821         free_commit_list(stack);
 822         return result;
 823 }
 824
 825 static int should_write_commit_graph(void)
 826 {
 827         int result;
 828         struct cg_auto_data data;
 829
 830         data.num_not_in_graph = 0;
 831         data.limit = 100;
 832         git_config_get_int("maintenance.commit-graph.auto",
 833                            &data.limit);
 834
 835         if (!data.limit)
 836                 return 0;
 837         if (data.limit < 0)
 838                 return 1;
 839
 840         result = for_each_ref(dfs_on_ref, &data);
 841
 842         repo_clear_commit_marks(the_repository, SEEN);
 843
 844         return result;
 845 }
 846
 847 static int run_write_commit_graph(struct maintenance_run_opts *opts)
 848 {
 849         struct child_process child = CHILD_PROCESS_INIT;
 850
 851         child.git_cmd = 1;
 852         strvec_pushl(&child.args, "commit-graph", "write",
 853                      "--split", "--reachable", NULL);
 854
 855         if (opts->quiet)
 856                 strvec_push(&child.args, "--no-progress");
 857
 858         return !!run_command(&child);
 859 }
 860
 861 static int maintenance_task_commit_graph(struct maintenance_run_opts *opts)
 862 {
 863         prepare_repo_settings(the_repository);
 864         if (!the_repository->settings.core_commit_graph)
 865                 return 0;
 866
 867         close_object_store(the_repository->objects);
 868         if (run_write_commit_graph(opts)) {
 869                 error(_("failed to write commit-graph"));
 870                 return 1;
 871         }
 872
 873         return 0;
 874 }
 875
 876 static int fetch_remote(const char *remote, struct maintenance_run_opts *opts)
 877 {
 878         struct child_process child = CHILD_PROCESS_INIT;
 879
 880         child.git_cmd = 1;
 881         strvec_pushl(&child.args, "fetch", remote, "--prune", "--no-tags",
 882                      "--no-write-fetch-head", "--recurse-submodules=no",
 883                      "--refmap=", NULL);
 884
 885         if (opts->quiet)
 886                 strvec_push(&child.args, "--quiet");
 887
 888         strvec_pushf(&child.args, "+refs/heads/*:refs/prefetch/%s/*", remote);
 889
 890         return !!run_command(&child);
 891 }
 892
 893 static int append_remote(struct remote *remote, void *cbdata)
 894 {
 895         struct string_list *remotes = (struct string_list *)cbdata;
 896
 897         string_list_append(remotes, remote->name);
 898         return 0;
 899 }
 900
 901 static int maintenance_task_prefetch(struct maintenance_run_opts *opts)
 902 {
 903         int result = 0;
 904         struct string_list_item *item;
 905         struct string_list remotes = STRING_LIST_INIT_DUP;
 906
 907         git_config_set_multivar_gently("log.excludedecoration",
 908                                         "refs/prefetch/",
 909                                         "refs/prefetch/",
 910                                         CONFIG_FLAGS_FIXED_VALUE |
 911                                         CONFIG_FLAGS_MULTI_REPLACE);
 912
 913         if (for_each_remote(append_remote, &remotes)) {
 914                 error(_("failed to fill remotes"));
 915                 result = 1;
 916                 goto cleanup;
 917         }
 918
 919         for_each_string_list_item(item, &remotes)
 920                 result |= fetch_remote(item->string, opts);
 921
 922 cleanup:
 923         string_list_clear(&remotes, 0);
 924         return result;
 925 }
 926
 927 static int maintenance_task_gc(struct maintenance_run_opts *opts)
 928 {
 929         struct child_process child = CHILD_PROCESS_INIT;
 930
 931         child.git_cmd = 1;
 932         strvec_push(&child.args, "gc");
 933
 934         if (opts->auto_flag)
 935                 strvec_push(&child.args, "--auto");
 936         if (opts->quiet)
 937                 strvec_push(&child.args, "--quiet");
 938         else
 939                 strvec_push(&child.args, "--no-quiet");
 940
 941         close_object_store(the_repository->objects);
 942         return run_command(&child);
 943 }
 944
 945 static int prune_packed(struct maintenance_run_opts *opts)
 946 {
 947         struct child_process child = CHILD_PROCESS_INIT;
 948
 949         child.git_cmd = 1;
 950         strvec_push(&child.args, "prune-packed");
 951
 952         if (opts->quiet)
 953                 strvec_push(&child.args, "--quiet");
 954
 955         return !!run_command(&child);
 956 }
 957
 958 struct write_loose_object_data {
 959         FILE *in;
 960         int count;
 961         int batch_size;
 962 };
 963
 964 static int loose_object_auto_limit = 100;
 965
 966 static int loose_object_count(const struct object_id *oid,
 967                                const char *path,
 968                                void *data)
 969 {
 970         int *count = (int*)data;
 971         if (++(*count) >= loose_object_auto_limit)
 972                 return 1;
 973         return 0;
 974 }
 975
 976 static int loose_object_auto_condition(void)
 977 {
 978         int count = 0;
 979
 980         git_config_get_int("maintenance.loose-objects.auto",
 981                            &loose_object_auto_limit);
 982
 983         if (!loose_object_auto_limit)
 984                 return 0;
 985         if (loose_object_auto_limit < 0)
 986                 return 1;
 987
 988         return for_each_loose_file_in_objdir(the_repository->objects->odb->path,
 989                                              loose_object_count,
 990                                              NULL, NULL, &count);
 991 }
 992
 993 static int bail_on_loose(const struct object_id *oid,
 994                          const char *path,
 995                          void *data)
 996 {
 997         return 1;
 998 }
 999
1000 static int write_loose_object_to_stdin(const struct object_id *oid,
1001                                        const char *path,
1002                                        void *data)
1003 {
1004         struct write_loose_object_data *d = (struct write_loose_object_data *)data;
1005
1006         fprintf(d->in, "%s\n", oid_to_hex(oid));
1007
1008         return ++(d->count) > d->batch_size;
1009 }
1010
1011 static int pack_loose(struct maintenance_run_opts *opts)
1012 {
1013         struct repository *r = the_repository;
1014         int result = 0;
1015         struct write_loose_object_data data;
1016         struct child_process pack_proc = CHILD_PROCESS_INIT;
1017
1018         /*
1019          * Do not start pack-objects process
1020          * if there are no loose objects.
1021          */
1022         if (!for_each_loose_file_in_objdir(r->objects->odb->path,
1023                                            bail_on_loose,
1024                                            NULL, NULL, NULL))
1025                 return 0;
1026
1027         pack_proc.git_cmd = 1;
1028
1029         strvec_push(&pack_proc.args, "pack-objects");
1030         if (opts->quiet)
1031                 strvec_push(&pack_proc.args, "--quiet");
1032         strvec_pushf(&pack_proc.args, "%s/pack/loose", r->objects->odb->path);
1033
1034         pack_proc.in = -1;
1035
1036         if (start_command(&pack_proc)) {
1037                 error(_("failed to start 'git pack-objects' process"));
1038                 return 1;
1039         }
1040
1041         data.in = xfdopen(pack_proc.in, "w");
1042         data.count = 0;
1043         data.batch_size = 50000;
1044
1045         for_each_loose_file_in_objdir(r->objects->odb->path,
1046                                       write_loose_object_to_stdin,
1047                                       NULL,
1048                                       NULL,
1049                                       &data);
1050
1051         fclose(data.in);
1052
1053         if (finish_command(&pack_proc)) {
1054                 error(_("failed to finish 'git pack-objects' process"));
1055                 result = 1;
1056         }
1057
1058         return result;
1059 }
1060
1061 static int maintenance_task_loose_objects(struct maintenance_run_opts *opts)
1062 {
1063         return prune_packed(opts) || pack_loose(opts);
1064 }
1065
1066 static int incremental_repack_auto_condition(void)
1067 {
1068         struct packed_git *p;
1069         int enabled;
1070         int incremental_repack_auto_limit = 10;
1071         int count = 0;
1072
1073         if (git_config_get_bool("core.multiPackIndex", &enabled) ||
1074             !enabled)
1075                 return 0;
1076
1077         git_config_get_int("maintenance.incremental-repack.auto",
1078                            &incremental_repack_auto_limit);
1079
1080         if (!incremental_repack_auto_limit)
1081                 return 0;
1082         if (incremental_repack_auto_limit < 0)
1083                 return 1;
1084
1085         for (p = get_packed_git(the_repository);
1086              count < incremental_repack_auto_limit && p;
1087              p = p->next) {
1088                 if (!p->multi_pack_index)
1089                         count++;
1090         }
1091
1092         return count >= incremental_repack_auto_limit;
1093 }
1094
1095 static int multi_pack_index_write(struct maintenance_run_opts *opts)
1096 {
1097         struct child_process child = CHILD_PROCESS_INIT;
1098
1099         child.git_cmd = 1;
1100         strvec_pushl(&child.args, "multi-pack-index", "write", NULL);
1101
1102         if (opts->quiet)
1103                 strvec_push(&child.args, "--no-progress");
1104
1105         if (run_command(&child))
1106                 return error(_("failed to write multi-pack-index"));
1107
1108         return 0;
1109 }
1110
1111 static int multi_pack_index_expire(struct maintenance_run_opts *opts)
1112 {
1113         struct child_process child = CHILD_PROCESS_INIT;
1114
1115         child.git_cmd = 1;
1116         strvec_pushl(&child.args, "multi-pack-index", "expire", NULL);
1117
1118         if (opts->quiet)
1119                 strvec_push(&child.args, "--no-progress");
1120
1121         close_object_store(the_repository->objects);
1122
1123         if (run_command(&child))
1124                 return error(_("'git multi-pack-index expire' failed"));
1125
1126         return 0;
1127 }
1128
1129 #define TWO_GIGABYTES (INT32_MAX)
1130
1131 static off_t get_auto_pack_size(void)
1132 {
1133         /*
1134          * The "auto" value is special: we optimize for
1135          * one large pack-file (i.e. from a clone) and
1136          * expect the rest to be small and they can be
1137          * repacked quickly.
1138          *
1139          * The strategy we select here is to select a
1140          * size that is one more than the second largest
1141          * pack-file. This ensures that we will repack
1142          * at least two packs if there are three or more
1143          * packs.
1144          */
1145         off_t max_size = 0;
1146         off_t second_largest_size = 0;
1147         off_t result_size;
1148         struct packed_git *p;
1149         struct repository *r = the_repository;
1150
1151         reprepare_packed_git(r);
1152         for (p = get_all_packs(r); p; p = p->next) {
1153                 if (p->pack_size > max_size) {
1154                         second_largest_size = max_size;
1155                         max_size = p->pack_size;
1156                 } else if (p->pack_size > second_largest_size)
1157                         second_largest_size = p->pack_size;
1158         }
1159
1160         result_size = second_largest_size + 1;
1161
1162         /* But limit ourselves to a batch size of 2g */
1163         if (result_size > TWO_GIGABYTES)
1164                 result_size = TWO_GIGABYTES;
1165
1166         return result_size;
1167 }
1168
1169 static int multi_pack_index_repack(struct maintenance_run_opts *opts)
1170 {
1171         struct child_process child = CHILD_PROCESS_INIT;
1172
1173         child.git_cmd = 1;
1174         strvec_pushl(&child.args, "multi-pack-index", "repack", NULL);
1175
1176         if (opts->quiet)
1177                 strvec_push(&child.args, "--no-progress");
1178
1179         strvec_pushf(&child.args, "--batch-size=%"PRIuMAX,
1180                                   (uintmax_t)get_auto_pack_size());
1181
1182         close_object_store(the_repository->objects);
1183
1184         if (run_command(&child))
1185                 return error(_("'git multi-pack-index repack' failed"));
1186
1187         return 0;
1188 }
1189
1190 static int maintenance_task_incremental_repack(struct maintenance_run_opts *opts)
1191 {
1192         prepare_repo_settings(the_repository);
1193         if (!the_repository->settings.core_multi_pack_index) {
1194                 warning(_("skipping incremental-repack task because core.multiPackIndex is disabled"));
1195                 return 0;
1196         }
1197
1198         if (multi_pack_index_write(opts))
1199                 return 1;
1200         if (multi_pack_index_expire(opts))
1201                 return 1;
1202         if (multi_pack_index_repack(opts))
1203                 return 1;
1204         return 0;
1205 }
1206
1207 typedef int maintenance_task_fn(struct maintenance_run_opts *opts);
1208
1209 /*
1210  * An auto condition function returns 1 if the task should run
1211  * and 0 if the task should NOT run. See needs_to_gc() for an
1212  * example.
1213  */
1214 typedef int maintenance_auto_fn(void);
1215
1216 struct maintenance_task {
1217         const char *name;
1218         maintenance_task_fn *fn;
1219         maintenance_auto_fn *auto_condition;
1220         unsigned enabled:1;
1221
1222         enum schedule_priority schedule;
1223
1224         /* -1 if not selected. */
1225         int selected_order;
1226 };
1227
1228 enum maintenance_task_label {
1229         TASK_PREFETCH,
1230         TASK_LOOSE_OBJECTS,
1231         TASK_INCREMENTAL_REPACK,
1232         TASK_GC,
1233         TASK_COMMIT_GRAPH,
1234         TASK_PACK_REFS,
1235
1236         /* Leave as final value */
1237         TASK__COUNT
1238 };
1239
1240 static struct maintenance_task tasks[] = {
1241         [TASK_PREFETCH] = {
1242                 "prefetch",
1243                 maintenance_task_prefetch,
1244         },
1245         [TASK_LOOSE_OBJECTS] = {
1246                 "loose-objects",
1247                 maintenance_task_loose_objects,
1248                 loose_object_auto_condition,
1249         },
1250         [TASK_INCREMENTAL_REPACK] = {
1251                 "incremental-repack",
1252                 maintenance_task_incremental_repack,
1253                 incremental_repack_auto_condition,
1254         },
1255         [TASK_GC] = {
1256                 "gc",
1257                 maintenance_task_gc,
1258                 need_to_gc,
1259                 1,
1260         },
1261         [TASK_COMMIT_GRAPH] = {
1262                 "commit-graph",
1263                 maintenance_task_commit_graph,
1264                 should_write_commit_graph,
1265         },
1266         [TASK_PACK_REFS] = {
1267                 "pack-refs",
1268                 maintenance_task_pack_refs,
1269                 NULL,
1270         },
1271 };
1272
1273 static int compare_tasks_by_selection(const void *a_, const void *b_)
1274 {
1275         const struct maintenance_task *a = a_;
1276         const struct maintenance_task *b = b_;
1277
1278         return b->selected_order - a->selected_order;
1279 }
1280
1281 static int maintenance_run_tasks(struct maintenance_run_opts *opts)
1282 {
1283         int i, found_selected = 0;
1284         int result = 0;
1285         struct lock_file lk;
1286         struct repository *r = the_repository;
1287         char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path);
1288
1289         if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
1290                 /*
1291                  * Another maintenance command is running.
1292                  *
1293                  * If --auto was provided, then it is likely due to a
1294                  * recursive process stack. Do not report an error in
1295                  * that case.
1296                  */
1297                 if (!opts->auto_flag && !opts->quiet)
1298                         warning(_("lock file '%s' exists, skipping maintenance"),
1299                                 lock_path);
1300                 free(lock_path);
1301                 return 0;
1302         }
1303         free(lock_path);
1304
1305         for (i = 0; !found_selected && i < TASK__COUNT; i++)
1306                 found_selected = tasks[i].selected_order >= 0;
1307
1308         if (found_selected)
1309                 QSORT(tasks, TASK__COUNT, compare_tasks_by_selection);
1310
1311         for (i = 0; i < TASK__COUNT; i++) {
1312                 if (found_selected && tasks[i].selected_order < 0)
1313                         continue;
1314
1315                 if (!found_selected && !tasks[i].enabled)
1316                         continue;
1317
1318                 if (opts->auto_flag &&
1319                     (!tasks[i].auto_condition ||
1320                      !tasks[i].auto_condition()))
1321                         continue;
1322
1323                 if (opts->schedule && tasks[i].schedule < opts->schedule)
1324                         continue;
1325
1326                 trace2_region_enter("maintenance", tasks[i].name, r);
1327                 if (tasks[i].fn(opts)) {
1328                         error(_("task '%s' failed"), tasks[i].name);
1329                         result = 1;
1330                 }
1331                 trace2_region_leave("maintenance", tasks[i].name, r);
1332         }
1333
1334         rollback_lock_file(&lk);
1335         return result;
1336 }
1337
1338 static void initialize_maintenance_strategy(void)
1339 {
1340         char *config_str;
1341
1342         if (git_config_get_string("maintenance.strategy", &config_str))
1343                 return;
1344
1345         if (!strcasecmp(config_str, "incremental")) {
1346                 tasks[TASK_GC].schedule = SCHEDULE_NONE;
1347                 tasks[TASK_COMMIT_GRAPH].enabled = 1;
1348                 tasks[TASK_COMMIT_GRAPH].schedule = SCHEDULE_HOURLY;
1349                 tasks[TASK_PREFETCH].enabled = 1;
1350                 tasks[TASK_PREFETCH].schedule = SCHEDULE_HOURLY;
1351                 tasks[TASK_INCREMENTAL_REPACK].enabled = 1;
1352                 tasks[TASK_INCREMENTAL_REPACK].schedule = SCHEDULE_DAILY;
1353                 tasks[TASK_LOOSE_OBJECTS].enabled = 1;
1354                 tasks[TASK_LOOSE_OBJECTS].schedule = SCHEDULE_DAILY;
1355                 tasks[TASK_PACK_REFS].enabled = 1;
1356                 tasks[TASK_PACK_REFS].schedule = SCHEDULE_WEEKLY;
1357         }
1358 }
1359
1360 static void initialize_task_config(int schedule)
1361 {
1362         int i;
1363         struct strbuf config_name = STRBUF_INIT;
1364         gc_config();
1365
1366         if (schedule)
1367                 initialize_maintenance_strategy();
1368
1369         for (i = 0; i < TASK__COUNT; i++) {
1370                 int config_value;
1371                 char *config_str;
1372
1373                 strbuf_reset(&config_name);
1374                 strbuf_addf(&config_name, "maintenance.%s.enabled",
1375                             tasks[i].name);
1376
1377                 if (!git_config_get_bool(config_name.buf, &config_value))
1378                         tasks[i].enabled = config_value;
1379
1380                 strbuf_reset(&config_name);
1381                 strbuf_addf(&config_name, "maintenance.%s.schedule",
1382                             tasks[i].name);
1383
1384                 if (!git_config_get_string(config_name.buf, &config_str)) {
1385                         tasks[i].schedule = parse_schedule(config_str);
1386                         free(config_str);
1387                 }
1388         }
1389
1390         strbuf_release(&config_name);
1391 }
1392
1393 static int task_option_parse(const struct option *opt,
1394                              const char *arg, int unset)
1395 {
1396         int i, num_selected = 0;
1397         struct maintenance_task *task = NULL;
1398
1399         BUG_ON_OPT_NEG(unset);
1400
1401         for (i = 0; i < TASK__COUNT; i++) {
1402                 if (tasks[i].selected_order >= 0)
1403                         num_selected++;
1404                 if (!strcasecmp(tasks[i].name, arg)) {
1405                         task = &tasks[i];
1406                 }
1407         }
1408
1409         if (!task) {
1410                 error(_("'%s' is not a valid task"), arg);
1411                 return 1;
1412         }
1413
1414         if (task->selected_order >= 0) {
1415                 error(_("task '%s' cannot be selected multiple times"), arg);
1416                 return 1;
1417         }
1418
1419         task->selected_order = num_selected + 1;
1420
1421         return 0;
1422 }
1423
1424 static int maintenance_run(int argc, const char **argv, const char *prefix)
1425 {
1426         int i;
1427         struct maintenance_run_opts opts;
1428         struct option builtin_maintenance_run_options[] = {
1429                 OPT_BOOL(0, "auto", &opts.auto_flag,
1430                          N_("run tasks based on the state of the repository")),
1431                 OPT_CALLBACK(0, "schedule", &opts.schedule, N_("frequency"),
1432                              N_("run tasks based on frequency"),
1433                              maintenance_opt_schedule),
1434                 OPT_BOOL(0, "quiet", &opts.quiet,
1435                          N_("do not report progress or other information over stderr")),
1436                 OPT_CALLBACK_F(0, "task", NULL, N_("task"),
1437                         N_("run a specific task"),
1438                         PARSE_OPT_NONEG, task_option_parse),
1439                 OPT_END()
1440         };
1441         memset(&opts, 0, sizeof(opts));
1442
1443         opts.quiet = !isatty(2);
1444
1445         for (i = 0; i < TASK__COUNT; i++)
1446                 tasks[i].selected_order = -1;
1447
1448         argc = parse_options(argc, argv, prefix,
1449                              builtin_maintenance_run_options,
1450                              builtin_maintenance_run_usage,
1451                              PARSE_OPT_STOP_AT_NON_OPTION);
1452
1453         if (opts.auto_flag && opts.schedule)
1454                 die(_("use at most one of --auto and --schedule=<frequency>"));
1455
1456         initialize_task_config(opts.schedule);
1457
1458         if (argc != 0)
1459                 usage_with_options(builtin_maintenance_run_usage,
1460                                    builtin_maintenance_run_options);
1461         return maintenance_run_tasks(&opts);
1462 }
1463
1464 static char *get_maintpath(void)
1465 {
1466         struct strbuf sb = STRBUF_INIT;
1467         const char *p = the_repository->worktree ?
1468                 the_repository->worktree : the_repository->gitdir;
1469
1470         strbuf_realpath(&sb, p, 1);
1471         return strbuf_detach(&sb, NULL);
1472 }
1473
1474 static int maintenance_register(void)
1475 {
1476         int rc;
1477         char *config_value;
1478         struct child_process config_set = CHILD_PROCESS_INIT;
1479         struct child_process config_get = CHILD_PROCESS_INIT;
1480         char *maintpath = get_maintpath();
1481
1482         /* Disable foreground maintenance */
1483         git_config_set("maintenance.auto", "false");
1484
1485         /* Set maintenance strategy, if unset */
1486         if (!git_config_get_string("maintenance.strategy", &config_value))
1487                 free(config_value);
1488         else
1489                 git_config_set("maintenance.strategy", "incremental");
1490
1491         config_get.git_cmd = 1;
1492         strvec_pushl(&config_get.args, "config", "--global", "--get",
1493                      "--fixed-value", "maintenance.repo", maintpath, NULL);
1494         config_get.out = -1;
1495
1496         if (start_command(&config_get)) {
1497                 rc = error(_("failed to run 'git config'"));
1498                 goto done;
1499         }
1500
1501         /* We already have this value in our config! */
1502         if (!finish_command(&config_get)) {
1503                 rc = 0;
1504                 goto done;
1505         }
1506
1507         config_set.git_cmd = 1;
1508         strvec_pushl(&config_set.args, "config", "--add", "--global", "maintenance.repo",
1509                      maintpath, NULL);
1510
1511         rc = run_command(&config_set);
1512
1513 done:
1514         free(maintpath);
1515         return rc;
1516 }
1517
1518 static int maintenance_unregister(void)
1519 {
1520         int rc;
1521         struct child_process config_unset = CHILD_PROCESS_INIT;
1522         char *maintpath = get_maintpath();
1523
1524         config_unset.git_cmd = 1;
1525         strvec_pushl(&config_unset.args, "config", "--global", "--unset",
1526                      "--fixed-value", "maintenance.repo", maintpath, NULL);
1527
1528         rc = run_command(&config_unset);
1529         free(maintpath);
1530         return rc;
1531 }
1532
1533 static const char *get_frequency(enum schedule_priority schedule)
1534 {
1535         switch (schedule) {
1536         case SCHEDULE_HOURLY:
1537                 return "hourly";
1538         case SCHEDULE_DAILY:
1539                 return "daily";
1540         case SCHEDULE_WEEKLY:
1541                 return "weekly";
1542         default:
1543                 BUG("invalid schedule %d", schedule);
1544         }
1545 }
1546
1547 static char *launchctl_service_name(const char *frequency)
1548 {
1549         struct strbuf label = STRBUF_INIT;
1550         strbuf_addf(&label, "org.git-scm.git.%s", frequency);
1551         return strbuf_detach(&label, NULL);
1552 }
1553
1554 static char *launchctl_service_filename(const char *name)
1555 {
1556         char *expanded;
1557         struct strbuf filename = STRBUF_INIT;
1558         strbuf_addf(&filename, "~/Library/LaunchAgents/%s.plist", name);
1559
1560         expanded = expand_user_path(filename.buf, 1);
1561         if (!expanded)
1562                 die(_("failed to expand path '%s'"), filename.buf);
1563
1564         strbuf_release(&filename);
1565         return expanded;
1566 }
1567
1568 static char *launchctl_get_uid(void)
1569 {
1570         return xstrfmt("gui/%d", getuid());
1571 }
1572
1573 static int launchctl_boot_plist(int enable, const char *filename, const char *cmd)
1574 {
1575         int result;
1576         struct child_process child = CHILD_PROCESS_INIT;
1577         char *uid = launchctl_get_uid();
1578
1579         strvec_split(&child.args, cmd);
1580         if (enable)
1581                 strvec_push(&child.args, "bootstrap");
1582         else
1583                 strvec_push(&child.args, "bootout");
1584         strvec_push(&child.args, uid);
1585         strvec_push(&child.args, filename);
1586
1587         child.no_stderr = 1;
1588         child.no_stdout = 1;
1589
1590         if (start_command(&child))
1591                 die(_("failed to start launchctl"));
1592
1593         result = finish_command(&child);
1594
1595         free(uid);
1596         return result;
1597 }
1598
1599 static int launchctl_remove_plist(enum schedule_priority schedule, const char *cmd)
1600 {
1601         const char *frequency = get_frequency(schedule);
1602         char *name = launchctl_service_name(frequency);
1603         char *filename = launchctl_service_filename(name);
1604         int result = launchctl_boot_plist(0, filename, cmd);
1605         unlink(filename);
1606         free(filename);
1607         free(name);
1608         return result;
1609 }
1610
1611 static int launchctl_remove_plists(const char *cmd)
1612 {
1613         return launchctl_remove_plist(SCHEDULE_HOURLY, cmd) ||
1614                 launchctl_remove_plist(SCHEDULE_DAILY, cmd) ||
1615                 launchctl_remove_plist(SCHEDULE_WEEKLY, cmd);
1616 }
1617
1618 static int launchctl_schedule_plist(const char *exec_path, enum schedule_priority schedule, const char *cmd)
1619 {
1620         FILE *plist;
1621         int i;
1622         const char *preamble, *repeat;
1623         const char *frequency = get_frequency(schedule);
1624         char *name = launchctl_service_name(frequency);
1625         char *filename = launchctl_service_filename(name);
1626
1627         if (safe_create_leading_directories(filename))
1628                 die(_("failed to create directories for '%s'"), filename);
1629         plist = xfopen(filename, "w");
1630
1631         preamble = "<?xml version=\"1.0\"?>\n"
1632                    "<!DOCTYPE plist PUBLIC \"-//Apple//DTD PLIST 1.0//EN\" \"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
1633                    "<plist version=\"1.0\">"
1634                    "<dict>\n"
1635                    "<key>Label</key><string>%s</string>\n"
1636                    "<key>ProgramArguments</key>\n"
1637                    "<array>\n"
1638                    "<string>%s/git</string>\n"
1639                    "<string>--exec-path=%s</string>\n"
1640                    "<string>for-each-repo</string>\n"
1641                    "<string>--config=maintenance.repo</string>\n"
1642                    "<string>maintenance</string>\n"
1643                    "<string>run</string>\n"
1644                    "<string>--schedule=%s</string>\n"
1645                    "</array>\n"
1646                    "<key>StartCalendarInterval</key>\n"
1647                    "<array>\n";
1648         fprintf(plist, preamble, name, exec_path, exec_path, frequency);
1649
1650         switch (schedule) {
1651         case SCHEDULE_HOURLY:
1652                 repeat = "<dict>\n"
1653                          "<key>Hour</key><integer>%d</integer>\n"
1654                          "<key>Minute</key><integer>0</integer>\n"
1655                          "</dict>\n";
1656                 for (i = 1; i <= 23; i++)
1657                         fprintf(plist, repeat, i);
1658                 break;
1659
1660         case SCHEDULE_DAILY:
1661                 repeat = "<dict>\n"
1662                          "<key>Day</key><integer>%d</integer>\n"
1663                          "<key>Hour</key><integer>0</integer>\n"
1664                          "<key>Minute</key><integer>0</integer>\n"
1665                          "</dict>\n";
1666                 for (i = 1; i <= 6; i++)
1667                         fprintf(plist, repeat, i);
1668                 break;
1669
1670         case SCHEDULE_WEEKLY:
1671                 fprintf(plist,
1672                         "<dict>\n"
1673                         "<key>Day</key><integer>0</integer>\n"
1674                         "<key>Hour</key><integer>0</integer>\n"
1675                         "<key>Minute</key><integer>0</integer>\n"
1676                         "</dict>\n");
1677                 break;
1678
1679         default:
1680                 /* unreachable */
1681                 break;
1682         }
1683         fprintf(plist, "</array>\n</dict>\n</plist>\n");
1684         fclose(plist);
1685
1686         /* bootout might fail if not already running, so ignore */
1687         launchctl_boot_plist(0, filename, cmd);
1688         if (launchctl_boot_plist(1, filename, cmd))
1689                 die(_("failed to bootstrap service %s"), filename);
1690
1691         free(filename);
1692         free(name);
1693         return 0;
1694 }
1695
1696 static int launchctl_add_plists(const char *cmd)
1697 {
1698         const char *exec_path = git_exec_path();
1699
1700         return launchctl_schedule_plist(exec_path, SCHEDULE_HOURLY, cmd) ||
1701                 launchctl_schedule_plist(exec_path, SCHEDULE_DAILY, cmd) ||
1702                 launchctl_schedule_plist(exec_path, SCHEDULE_WEEKLY, cmd);
1703 }
1704
1705 static int launchctl_update_schedule(int run_maintenance, int fd, const char *cmd)
1706 {
1707         if (run_maintenance)
1708                 return launchctl_add_plists(cmd);
1709         else
1710                 return launchctl_remove_plists(cmd);
1711 }
1712
1713 static char *schtasks_task_name(const char *frequency)
1714 {
1715         struct strbuf label = STRBUF_INIT;
1716         strbuf_addf(&label, "Git Maintenance (%s)", frequency);
1717         return strbuf_detach(&label, NULL);
1718 }
1719
1720 static int schtasks_remove_task(enum schedule_priority schedule, const char *cmd)
1721 {
1722         int result;
1723         struct strvec args = STRVEC_INIT;
1724         const char *frequency = get_frequency(schedule);
1725         char *name = schtasks_task_name(frequency);
1726
1727         strvec_split(&args, cmd);
1728         strvec_pushl(&args, "/delete", "/tn", name, "/f", NULL);
1729
1730         result = run_command_v_opt(args.v, 0);
1731
1732         strvec_clear(&args);
1733         free(name);
1734         return result;
1735 }
1736
1737 static int schtasks_remove_tasks(const char *cmd)
1738 {
1739         return schtasks_remove_task(SCHEDULE_HOURLY, cmd) ||
1740                 schtasks_remove_task(SCHEDULE_DAILY, cmd) ||
1741                 schtasks_remove_task(SCHEDULE_WEEKLY, cmd);
1742 }
1743
1744 static int schtasks_schedule_task(const char *exec_path, enum schedule_priority schedule, const char *cmd)
1745 {
1746         int result;
1747         struct child_process child = CHILD_PROCESS_INIT;
1748         const char *xml;
1749         struct tempfile *tfile;
1750         const char *frequency = get_frequency(schedule);
1751         char *name = schtasks_task_name(frequency);
1752         struct strbuf tfilename = STRBUF_INIT;
1753
1754         strbuf_addf(&tfilename, "%s/schedule_%s_XXXXXX",
1755                     get_git_common_dir(), frequency);
1756         tfile = xmks_tempfile(tfilename.buf);
1757         strbuf_release(&tfilename);
1758
1759         if (!fdopen_tempfile(tfile, "w"))
1760                 die(_("failed to create temp xml file"));
1761
1762         xml = "<?xml version=\"1.0\" ?>\n"
1763               "<Task version=\"1.4\" xmlns=\"http://schemas.microsoft.com/windows/2004/02/mit/task\">\n"
1764               "<Triggers>\n"
1765               "<CalendarTrigger>\n";
1766         fputs(xml, tfile->fp);
1767
1768         switch (schedule) {
1769         case SCHEDULE_HOURLY:
1770                 fprintf(tfile->fp,
1771                         "<StartBoundary>2020-01-01T01:00:00</StartBoundary>\n"
1772                         "<Enabled>true</Enabled>\n"
1773                         "<ScheduleByDay>\n"
1774                         "<DaysInterval>1</DaysInterval>\n"
1775                         "</ScheduleByDay>\n"
1776                         "<Repetition>\n"
1777                         "<Interval>PT1H</Interval>\n"
1778                         "<Duration>PT23H</Duration>\n"
1779                         "<StopAtDurationEnd>false</StopAtDurationEnd>\n"
1780                         "</Repetition>\n");
1781                 break;
1782
1783         case SCHEDULE_DAILY:
1784                 fprintf(tfile->fp,
1785                         "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1786                         "<Enabled>true</Enabled>\n"
1787                         "<ScheduleByWeek>\n"
1788                         "<DaysOfWeek>\n"
1789                         "<Monday />\n"
1790                         "<Tuesday />\n"
1791                         "<Wednesday />\n"
1792                         "<Thursday />\n"
1793                         "<Friday />\n"
1794                         "<Saturday />\n"
1795                         "</DaysOfWeek>\n"
1796                         "<WeeksInterval>1</WeeksInterval>\n"
1797                         "</ScheduleByWeek>\n");
1798                 break;
1799
1800         case SCHEDULE_WEEKLY:
1801                 fprintf(tfile->fp,
1802                         "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1803                         "<Enabled>true</Enabled>\n"
1804                         "<ScheduleByWeek>\n"
1805                         "<DaysOfWeek>\n"
1806                         "<Sunday />\n"
1807                         "</DaysOfWeek>\n"
1808                         "<WeeksInterval>1</WeeksInterval>\n"
1809                         "</ScheduleByWeek>\n");
1810                 break;
1811
1812         default:
1813                 break;
1814         }
1815
1816         xml = "</CalendarTrigger>\n"
1817               "</Triggers>\n"
1818               "<Principals>\n"
1819               "<Principal id=\"Author\">\n"
1820               "<LogonType>InteractiveToken</LogonType>\n"
1821               "<RunLevel>LeastPrivilege</RunLevel>\n"
1822               "</Principal>\n"
1823               "</Principals>\n"
1824               "<Settings>\n"
1825               "<MultipleInstancesPolicy>IgnoreNew</MultipleInstancesPolicy>\n"
1826               "<Enabled>true</Enabled>\n"
1827               "<Hidden>true</Hidden>\n"
1828               "<UseUnifiedSchedulingEngine>true</UseUnifiedSchedulingEngine>\n"
1829               "<WakeToRun>false</WakeToRun>\n"
1830               "<ExecutionTimeLimit>PT72H</ExecutionTimeLimit>\n"
1831               "<Priority>7</Priority>\n"
1832               "</Settings>\n"
1833               "<Actions Context=\"Author\">\n"
1834               "<Exec>\n"
1835               "<Command>\"%s\\git.exe\"</Command>\n"
1836               "<Arguments>--exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%s</Arguments>\n"
1837               "</Exec>\n"
1838               "</Actions>\n"
1839               "</Task>\n";
1840         fprintf(tfile->fp, xml, exec_path, exec_path, frequency);
1841         strvec_split(&child.args, cmd);
1842         strvec_pushl(&child.args, "/create", "/tn", name, "/f", "/xml",
1843                                   get_tempfile_path(tfile), NULL);
1844         close_tempfile_gently(tfile);
1845
1846         child.no_stdout = 1;
1847         child.no_stderr = 1;
1848
1849         if (start_command(&child))
1850                 die(_("failed to start schtasks"));
1851         result = finish_command(&child);
1852
1853         delete_tempfile(&tfile);
1854         free(name);
1855         return result;
1856 }
1857
1858 static int schtasks_schedule_tasks(const char *cmd)
1859 {
1860         const char *exec_path = git_exec_path();
1861
1862         return schtasks_schedule_task(exec_path, SCHEDULE_HOURLY, cmd) ||
1863                 schtasks_schedule_task(exec_path, SCHEDULE_DAILY, cmd) ||
1864                 schtasks_schedule_task(exec_path, SCHEDULE_WEEKLY, cmd);
1865 }
1866
1867 static int schtasks_update_schedule(int run_maintenance, int fd, const char *cmd)
1868 {
1869         if (run_maintenance)
1870                 return schtasks_schedule_tasks(cmd);
1871         else
1872                 return schtasks_remove_tasks(cmd);
1873 }
1874
1875 #define BEGIN_LINE "# BEGIN GIT MAINTENANCE SCHEDULE"
1876 #define END_LINE "# END GIT MAINTENANCE SCHEDULE"
1877
1878 static int crontab_update_schedule(int run_maintenance, int fd, const char *cmd)
1879 {
1880         int result = 0;
1881         int in_old_region = 0;
1882         struct child_process crontab_list = CHILD_PROCESS_INIT;
1883         struct child_process crontab_edit = CHILD_PROCESS_INIT;
1884         FILE *cron_list, *cron_in;
1885         struct strbuf line = STRBUF_INIT;
1886
1887         strvec_split(&crontab_list.args, cmd);
1888         strvec_push(&crontab_list.args, "-l");
1889         crontab_list.in = -1;
1890         crontab_list.out = dup(fd);
1891         crontab_list.git_cmd = 0;
1892
1893         if (start_command(&crontab_list))
1894                 return error(_("failed to run 'crontab -l'; your system might not support 'cron'"));
1895
1896         /* Ignore exit code, as an empty crontab will return error. */
1897         finish_command(&crontab_list);
1898
1899         /*
1900          * Read from the .lock file, filtering out the old
1901          * schedule while appending the new schedule.
1902          */
1903         cron_list = fdopen(fd, "r");
1904         rewind(cron_list);
1905
1906         strvec_split(&crontab_edit.args, cmd);
1907         crontab_edit.in = -1;
1908         crontab_edit.git_cmd = 0;
1909
1910         if (start_command(&crontab_edit))
1911                 return error(_("failed to run 'crontab'; your system might not support 'cron'"));
1912
1913         cron_in = fdopen(crontab_edit.in, "w");
1914         if (!cron_in) {
1915                 result = error(_("failed to open stdin of 'crontab'"));
1916                 goto done_editing;
1917         }
1918
1919         while (!strbuf_getline_lf(&line, cron_list)) {
1920                 if (!in_old_region && !strcmp(line.buf, BEGIN_LINE))
1921                         in_old_region = 1;
1922                 else if (in_old_region && !strcmp(line.buf, END_LINE))
1923                         in_old_region = 0;
1924                 else if (!in_old_region)
1925                         fprintf(cron_in, "%s\n", line.buf);
1926         }
1927
1928         if (run_maintenance) {
1929                 struct strbuf line_format = STRBUF_INIT;
1930                 const char *exec_path = git_exec_path();
1931
1932                 fprintf(cron_in, "%s\n", BEGIN_LINE);
1933                 fprintf(cron_in,
1934                         "# The following schedule was created by Git\n");
1935                 fprintf(cron_in, "# Any edits made in this region might be\n");
1936                 fprintf(cron_in,
1937                         "# replaced in the future by a Git command.\n\n");
1938
1939                 strbuf_addf(&line_format,
1940                             "%%s %%s * * %%s \"%s/git\" --exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%%s\n",
1941                             exec_path, exec_path);
1942                 fprintf(cron_in, line_format.buf, "0", "1-23", "*", "hourly");
1943                 fprintf(cron_in, line_format.buf, "0", "0", "1-6", "daily");
1944                 fprintf(cron_in, line_format.buf, "0", "0", "0", "weekly");
1945                 strbuf_release(&line_format);
1946
1947                 fprintf(cron_in, "\n%s\n", END_LINE);
1948         }
1949
1950         fflush(cron_in);
1951         fclose(cron_in);
1952         close(crontab_edit.in);
1953
1954 done_editing:
1955         if (finish_command(&crontab_edit))
1956                 result = error(_("'crontab' died"));
1957         else
1958                 fclose(cron_list);
1959         return result;
1960 }
1961
1962 #if defined(__APPLE__)
1963 static const char platform_scheduler[] = "launchctl";
1964 #elif defined(GIT_WINDOWS_NATIVE)
1965 static const char platform_scheduler[] = "schtasks";
1966 #else
1967 static const char platform_scheduler[] = "crontab";
1968 #endif
1969
1970 static int update_background_schedule(int enable)
1971 {
1972         int result;
1973         const char *scheduler = platform_scheduler;
1974         const char *cmd = scheduler;
1975         char *testing;
1976         struct lock_file lk;
1977         char *lock_path = xstrfmt("%s/schedule", the_repository->objects->odb->path);
1978
1979         testing = xstrdup_or_null(getenv("GIT_TEST_MAINT_SCHEDULER"));
1980         if (testing) {
1981                 char *sep = strchr(testing, ':');
1982                 if (!sep)
1983                         die("GIT_TEST_MAINT_SCHEDULER unparseable: %s", testing);
1984                 *sep = '\0';
1985                 scheduler = testing;
1986                 cmd = sep + 1;
1987         }
1988
1989         if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0)
1990                 return error(_("another process is scheduling background maintenance"));
1991
1992         if (!strcmp(scheduler, "launchctl"))
1993                 result = launchctl_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1994         else if (!strcmp(scheduler, "schtasks"))
1995                 result = schtasks_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1996         else if (!strcmp(scheduler, "crontab"))
1997                 result = crontab_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1998         else
1999                 die("unknown background scheduler: %s", scheduler);
2000
2001         rollback_lock_file(&lk);
2002         free(testing);
2003         return result;
2004 }
2005
2006 static int maintenance_start(void)
2007 {
2008         if (maintenance_register())
2009                 warning(_("failed to add repo to global config"));
2010
2011         return update_background_schedule(1);
2012 }
2013
2014 static int maintenance_stop(void)
2015 {
2016         return update_background_schedule(0);
2017 }
2018
2019 static const char builtin_maintenance_usage[] = N_("git maintenance <subcommand> [<options>]");
2020
2021 int cmd_maintenance(int argc, const char **argv, const char *prefix)
2022 {
2023         if (argc < 2 ||
2024             (argc == 2 && !strcmp(argv[1], "-h")))
2025                 usage(builtin_maintenance_usage);
2026
2027         if (!strcmp(argv[1], "run"))
2028                 return maintenance_run(argc - 1, argv + 1, prefix);
2029         if (!strcmp(argv[1], "start"))
2030                 return maintenance_start();
2031         if (!strcmp(argv[1], "stop"))
2032                 return maintenance_stop();
2033         if (!strcmp(argv[1], "register"))
2034                 return maintenance_register();
2035         if (!strcmp(argv[1], "unregister"))
2036                 return maintenance_unregister();
2037
2038         die(_("invalid subcommand: %s"), argv[1]);
2039 }