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