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(struct remote *remote, void *cbdata)
 877 {
 878         struct maintenance_run_opts *opts = cbdata;
 879         struct child_process child = CHILD_PROCESS_INIT;
 880
 881         if (remote->skip_default_update)
 882                 return 0;
 883
 884         child.git_cmd = 1;
 885         strvec_pushl(&child.args, "fetch", remote->name,
 886                      "--prefetch", "--prune", "--no-tags",
 887                      "--no-write-fetch-head", "--recurse-submodules=no",
 888                      NULL);
 889
 890         if (opts->quiet)
 891                 strvec_push(&child.args, "--quiet");
 892
 893         return !!run_command(&child);
 894 }
 895
 896 static int maintenance_task_prefetch(struct maintenance_run_opts *opts)
 897 {
 898         git_config_set_multivar_gently("log.excludedecoration",
 899                                         "refs/prefetch/",
 900                                         "refs/prefetch/",
 901                                         CONFIG_FLAGS_FIXED_VALUE |
 902                                         CONFIG_FLAGS_MULTI_REPLACE);
 903
 904         if (for_each_remote(fetch_remote, opts)) {
 905                 error(_("failed to prefetch remotes"));
 906                 return 1;
 907         }
 908
 909         return 0;
 910 }
 911
 912 static int maintenance_task_gc(struct maintenance_run_opts *opts)
 913 {
 914         struct child_process child = CHILD_PROCESS_INIT;
 915
 916         child.git_cmd = 1;
 917         strvec_push(&child.args, "gc");
 918
 919         if (opts->auto_flag)
 920                 strvec_push(&child.args, "--auto");
 921         if (opts->quiet)
 922                 strvec_push(&child.args, "--quiet");
 923         else
 924                 strvec_push(&child.args, "--no-quiet");
 925
 926         close_object_store(the_repository->objects);
 927         return run_command(&child);
 928 }
 929
 930 static int prune_packed(struct maintenance_run_opts *opts)
 931 {
 932         struct child_process child = CHILD_PROCESS_INIT;
 933
 934         child.git_cmd = 1;
 935         strvec_push(&child.args, "prune-packed");
 936
 937         if (opts->quiet)
 938                 strvec_push(&child.args, "--quiet");
 939
 940         return !!run_command(&child);
 941 }
 942
 943 struct write_loose_object_data {
 944         FILE *in;
 945         int count;
 946         int batch_size;
 947 };
 948
 949 static int loose_object_auto_limit = 100;
 950
 951 static int loose_object_count(const struct object_id *oid,
 952                                const char *path,
 953                                void *data)
 954 {
 955         int *count = (int*)data;
 956         if (++(*count) >= loose_object_auto_limit)
 957                 return 1;
 958         return 0;
 959 }
 960
 961 static int loose_object_auto_condition(void)
 962 {
 963         int count = 0;
 964
 965         git_config_get_int("maintenance.loose-objects.auto",
 966                            &loose_object_auto_limit);
 967
 968         if (!loose_object_auto_limit)
 969                 return 0;
 970         if (loose_object_auto_limit < 0)
 971                 return 1;
 972
 973         return for_each_loose_file_in_objdir(the_repository->objects->odb->path,
 974                                              loose_object_count,
 975                                              NULL, NULL, &count);
 976 }
 977
 978 static int bail_on_loose(const struct object_id *oid,
 979                          const char *path,
 980                          void *data)
 981 {
 982         return 1;
 983 }
 984
 985 static int write_loose_object_to_stdin(const struct object_id *oid,
 986                                        const char *path,
 987                                        void *data)
 988 {
 989         struct write_loose_object_data *d = (struct write_loose_object_data *)data;
 990
 991         fprintf(d->in, "%s\n", oid_to_hex(oid));
 992
 993         return ++(d->count) > d->batch_size;
 994 }
 995
 996 static int pack_loose(struct maintenance_run_opts *opts)
 997 {
 998         struct repository *r = the_repository;
 999         int result = 0;
1000         struct write_loose_object_data data;
1001         struct child_process pack_proc = CHILD_PROCESS_INIT;
1002
1003         /*
1004          * Do not start pack-objects process
1005          * if there are no loose objects.
1006          */
1007         if (!for_each_loose_file_in_objdir(r->objects->odb->path,
1008                                            bail_on_loose,
1009                                            NULL, NULL, NULL))
1010                 return 0;
1011
1012         pack_proc.git_cmd = 1;
1013
1014         strvec_push(&pack_proc.args, "pack-objects");
1015         if (opts->quiet)
1016                 strvec_push(&pack_proc.args, "--quiet");
1017         strvec_pushf(&pack_proc.args, "%s/pack/loose", r->objects->odb->path);
1018
1019         pack_proc.in = -1;
1020
1021         if (start_command(&pack_proc)) {
1022                 error(_("failed to start 'git pack-objects' process"));
1023                 return 1;
1024         }
1025
1026         data.in = xfdopen(pack_proc.in, "w");
1027         data.count = 0;
1028         data.batch_size = 50000;
1029
1030         for_each_loose_file_in_objdir(r->objects->odb->path,
1031                                       write_loose_object_to_stdin,
1032                                       NULL,
1033                                       NULL,
1034                                       &data);
1035
1036         fclose(data.in);
1037
1038         if (finish_command(&pack_proc)) {
1039                 error(_("failed to finish 'git pack-objects' process"));
1040                 result = 1;
1041         }
1042
1043         return result;
1044 }
1045
1046 static int maintenance_task_loose_objects(struct maintenance_run_opts *opts)
1047 {
1048         return prune_packed(opts) || pack_loose(opts);
1049 }
1050
1051 static int incremental_repack_auto_condition(void)
1052 {
1053         struct packed_git *p;
1054         int enabled;
1055         int incremental_repack_auto_limit = 10;
1056         int count = 0;
1057
1058         if (git_config_get_bool("core.multiPackIndex", &enabled) ||
1059             !enabled)
1060                 return 0;
1061
1062         git_config_get_int("maintenance.incremental-repack.auto",
1063                            &incremental_repack_auto_limit);
1064
1065         if (!incremental_repack_auto_limit)
1066                 return 0;
1067         if (incremental_repack_auto_limit < 0)
1068                 return 1;
1069
1070         for (p = get_packed_git(the_repository);
1071              count < incremental_repack_auto_limit && p;
1072              p = p->next) {
1073                 if (!p->multi_pack_index)
1074                         count++;
1075         }
1076
1077         return count >= incremental_repack_auto_limit;
1078 }
1079
1080 static int multi_pack_index_write(struct maintenance_run_opts *opts)
1081 {
1082         struct child_process child = CHILD_PROCESS_INIT;
1083
1084         child.git_cmd = 1;
1085         strvec_pushl(&child.args, "multi-pack-index", "write", NULL);
1086
1087         if (opts->quiet)
1088                 strvec_push(&child.args, "--no-progress");
1089
1090         if (run_command(&child))
1091                 return error(_("failed to write multi-pack-index"));
1092
1093         return 0;
1094 }
1095
1096 static int multi_pack_index_expire(struct maintenance_run_opts *opts)
1097 {
1098         struct child_process child = CHILD_PROCESS_INIT;
1099
1100         child.git_cmd = 1;
1101         strvec_pushl(&child.args, "multi-pack-index", "expire", NULL);
1102
1103         if (opts->quiet)
1104                 strvec_push(&child.args, "--no-progress");
1105
1106         close_object_store(the_repository->objects);
1107
1108         if (run_command(&child))
1109                 return error(_("'git multi-pack-index expire' failed"));
1110
1111         return 0;
1112 }
1113
1114 #define TWO_GIGABYTES (INT32_MAX)
1115
1116 static off_t get_auto_pack_size(void)
1117 {
1118         /*
1119          * The "auto" value is special: we optimize for
1120          * one large pack-file (i.e. from a clone) and
1121          * expect the rest to be small and they can be
1122          * repacked quickly.
1123          *
1124          * The strategy we select here is to select a
1125          * size that is one more than the second largest
1126          * pack-file. This ensures that we will repack
1127          * at least two packs if there are three or more
1128          * packs.
1129          */
1130         off_t max_size = 0;
1131         off_t second_largest_size = 0;
1132         off_t result_size;
1133         struct packed_git *p;
1134         struct repository *r = the_repository;
1135
1136         reprepare_packed_git(r);
1137         for (p = get_all_packs(r); p; p = p->next) {
1138                 if (p->pack_size > max_size) {
1139                         second_largest_size = max_size;
1140                         max_size = p->pack_size;
1141                 } else if (p->pack_size > second_largest_size)
1142                         second_largest_size = p->pack_size;
1143         }
1144
1145         result_size = second_largest_size + 1;
1146
1147         /* But limit ourselves to a batch size of 2g */
1148         if (result_size > TWO_GIGABYTES)
1149                 result_size = TWO_GIGABYTES;
1150
1151         return result_size;
1152 }
1153
1154 static int multi_pack_index_repack(struct maintenance_run_opts *opts)
1155 {
1156         struct child_process child = CHILD_PROCESS_INIT;
1157
1158         child.git_cmd = 1;
1159         strvec_pushl(&child.args, "multi-pack-index", "repack", NULL);
1160
1161         if (opts->quiet)
1162                 strvec_push(&child.args, "--no-progress");
1163
1164         strvec_pushf(&child.args, "--batch-size=%"PRIuMAX,
1165                                   (uintmax_t)get_auto_pack_size());
1166
1167         close_object_store(the_repository->objects);
1168
1169         if (run_command(&child))
1170                 return error(_("'git multi-pack-index repack' failed"));
1171
1172         return 0;
1173 }
1174
1175 static int maintenance_task_incremental_repack(struct maintenance_run_opts *opts)
1176 {
1177         prepare_repo_settings(the_repository);
1178         if (!the_repository->settings.core_multi_pack_index) {
1179                 warning(_("skipping incremental-repack task because core.multiPackIndex is disabled"));
1180                 return 0;
1181         }
1182
1183         if (multi_pack_index_write(opts))
1184                 return 1;
1185         if (multi_pack_index_expire(opts))
1186                 return 1;
1187         if (multi_pack_index_repack(opts))
1188                 return 1;
1189         return 0;
1190 }
1191
1192 typedef int maintenance_task_fn(struct maintenance_run_opts *opts);
1193
1194 /*
1195  * An auto condition function returns 1 if the task should run
1196  * and 0 if the task should NOT run. See needs_to_gc() for an
1197  * example.
1198  */
1199 typedef int maintenance_auto_fn(void);
1200
1201 struct maintenance_task {
1202         const char *name;
1203         maintenance_task_fn *fn;
1204         maintenance_auto_fn *auto_condition;
1205         unsigned enabled:1;
1206
1207         enum schedule_priority schedule;
1208
1209         /* -1 if not selected. */
1210         int selected_order;
1211 };
1212
1213 enum maintenance_task_label {
1214         TASK_PREFETCH,
1215         TASK_LOOSE_OBJECTS,
1216         TASK_INCREMENTAL_REPACK,
1217         TASK_GC,
1218         TASK_COMMIT_GRAPH,
1219         TASK_PACK_REFS,
1220
1221         /* Leave as final value */
1222         TASK__COUNT
1223 };
1224
1225 static struct maintenance_task tasks[] = {
1226         [TASK_PREFETCH] = {
1227                 "prefetch",
1228                 maintenance_task_prefetch,
1229         },
1230         [TASK_LOOSE_OBJECTS] = {
1231                 "loose-objects",
1232                 maintenance_task_loose_objects,
1233                 loose_object_auto_condition,
1234         },
1235         [TASK_INCREMENTAL_REPACK] = {
1236                 "incremental-repack",
1237                 maintenance_task_incremental_repack,
1238                 incremental_repack_auto_condition,
1239         },
1240         [TASK_GC] = {
1241                 "gc",
1242                 maintenance_task_gc,
1243                 need_to_gc,
1244                 1,
1245         },
1246         [TASK_COMMIT_GRAPH] = {
1247                 "commit-graph",
1248                 maintenance_task_commit_graph,
1249                 should_write_commit_graph,
1250         },
1251         [TASK_PACK_REFS] = {
1252                 "pack-refs",
1253                 maintenance_task_pack_refs,
1254                 NULL,
1255         },
1256 };
1257
1258 static int compare_tasks_by_selection(const void *a_, const void *b_)
1259 {
1260         const struct maintenance_task *a = a_;
1261         const struct maintenance_task *b = b_;
1262
1263         return b->selected_order - a->selected_order;
1264 }
1265
1266 static int maintenance_run_tasks(struct maintenance_run_opts *opts)
1267 {
1268         int i, found_selected = 0;
1269         int result = 0;
1270         struct lock_file lk;
1271         struct repository *r = the_repository;
1272         char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path);
1273
1274         if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
1275                 /*
1276                  * Another maintenance command is running.
1277                  *
1278                  * If --auto was provided, then it is likely due to a
1279                  * recursive process stack. Do not report an error in
1280                  * that case.
1281                  */
1282                 if (!opts->auto_flag && !opts->quiet)
1283                         warning(_("lock file '%s' exists, skipping maintenance"),
1284                                 lock_path);
1285                 free(lock_path);
1286                 return 0;
1287         }
1288         free(lock_path);
1289
1290         for (i = 0; !found_selected && i < TASK__COUNT; i++)
1291                 found_selected = tasks[i].selected_order >= 0;
1292
1293         if (found_selected)
1294                 QSORT(tasks, TASK__COUNT, compare_tasks_by_selection);
1295
1296         for (i = 0; i < TASK__COUNT; i++) {
1297                 if (found_selected && tasks[i].selected_order < 0)
1298                         continue;
1299
1300                 if (!found_selected && !tasks[i].enabled)
1301                         continue;
1302
1303                 if (opts->auto_flag &&
1304                     (!tasks[i].auto_condition ||
1305                      !tasks[i].auto_condition()))
1306                         continue;
1307
1308                 if (opts->schedule && tasks[i].schedule < opts->schedule)
1309                         continue;
1310
1311                 trace2_region_enter("maintenance", tasks[i].name, r);
1312                 if (tasks[i].fn(opts)) {
1313                         error(_("task '%s' failed"), tasks[i].name);
1314                         result = 1;
1315                 }
1316                 trace2_region_leave("maintenance", tasks[i].name, r);
1317         }
1318
1319         rollback_lock_file(&lk);
1320         return result;
1321 }
1322
1323 static void initialize_maintenance_strategy(void)
1324 {
1325         char *config_str;
1326
1327         if (git_config_get_string("maintenance.strategy", &config_str))
1328                 return;
1329
1330         if (!strcasecmp(config_str, "incremental")) {
1331                 tasks[TASK_GC].schedule = SCHEDULE_NONE;
1332                 tasks[TASK_COMMIT_GRAPH].enabled = 1;
1333                 tasks[TASK_COMMIT_GRAPH].schedule = SCHEDULE_HOURLY;
1334                 tasks[TASK_PREFETCH].enabled = 1;
1335                 tasks[TASK_PREFETCH].schedule = SCHEDULE_HOURLY;
1336                 tasks[TASK_INCREMENTAL_REPACK].enabled = 1;
1337                 tasks[TASK_INCREMENTAL_REPACK].schedule = SCHEDULE_DAILY;
1338                 tasks[TASK_LOOSE_OBJECTS].enabled = 1;
1339                 tasks[TASK_LOOSE_OBJECTS].schedule = SCHEDULE_DAILY;
1340                 tasks[TASK_PACK_REFS].enabled = 1;
1341                 tasks[TASK_PACK_REFS].schedule = SCHEDULE_WEEKLY;
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 char *get_maintpath(void)
1450 {
1451         struct strbuf sb = STRBUF_INIT;
1452         const char *p = the_repository->worktree ?
1453                 the_repository->worktree : the_repository->gitdir;
1454
1455         strbuf_realpath(&sb, p, 1);
1456         return strbuf_detach(&sb, NULL);
1457 }
1458
1459 static int maintenance_register(void)
1460 {
1461         int rc;
1462         char *config_value;
1463         struct child_process config_set = CHILD_PROCESS_INIT;
1464         struct child_process config_get = CHILD_PROCESS_INIT;
1465         char *maintpath = get_maintpath();
1466
1467         /* Disable foreground maintenance */
1468         git_config_set("maintenance.auto", "false");
1469
1470         /* Set maintenance strategy, if unset */
1471         if (!git_config_get_string("maintenance.strategy", &config_value))
1472                 free(config_value);
1473         else
1474                 git_config_set("maintenance.strategy", "incremental");
1475
1476         config_get.git_cmd = 1;
1477         strvec_pushl(&config_get.args, "config", "--global", "--get",
1478                      "--fixed-value", "maintenance.repo", maintpath, NULL);
1479         config_get.out = -1;
1480
1481         if (start_command(&config_get)) {
1482                 rc = error(_("failed to run 'git config'"));
1483                 goto done;
1484         }
1485
1486         /* We already have this value in our config! */
1487         if (!finish_command(&config_get)) {
1488                 rc = 0;
1489                 goto done;
1490         }
1491
1492         config_set.git_cmd = 1;
1493         strvec_pushl(&config_set.args, "config", "--add", "--global", "maintenance.repo",
1494                      maintpath, NULL);
1495
1496         rc = run_command(&config_set);
1497
1498 done:
1499         free(maintpath);
1500         return rc;
1501 }
1502
1503 static int maintenance_unregister(void)
1504 {
1505         int rc;
1506         struct child_process config_unset = CHILD_PROCESS_INIT;
1507         char *maintpath = get_maintpath();
1508
1509         config_unset.git_cmd = 1;
1510         strvec_pushl(&config_unset.args, "config", "--global", "--unset",
1511                      "--fixed-value", "maintenance.repo", maintpath, NULL);
1512
1513         rc = run_command(&config_unset);
1514         free(maintpath);
1515         return rc;
1516 }
1517
1518 static const char *get_frequency(enum schedule_priority schedule)
1519 {
1520         switch (schedule) {
1521         case SCHEDULE_HOURLY:
1522                 return "hourly";
1523         case SCHEDULE_DAILY:
1524                 return "daily";
1525         case SCHEDULE_WEEKLY:
1526                 return "weekly";
1527         default:
1528                 BUG("invalid schedule %d", schedule);
1529         }
1530 }
1531
1532 static char *launchctl_service_name(const char *frequency)
1533 {
1534         struct strbuf label = STRBUF_INIT;
1535         strbuf_addf(&label, "org.git-scm.git.%s", frequency);
1536         return strbuf_detach(&label, NULL);
1537 }
1538
1539 static char *launchctl_service_filename(const char *name)
1540 {
1541         char *expanded;
1542         struct strbuf filename = STRBUF_INIT;
1543         strbuf_addf(&filename, "~/Library/LaunchAgents/%s.plist", name);
1544
1545         expanded = expand_user_path(filename.buf, 1);
1546         if (!expanded)
1547                 die(_("failed to expand path '%s'"), filename.buf);
1548
1549         strbuf_release(&filename);
1550         return expanded;
1551 }
1552
1553 static char *launchctl_get_uid(void)
1554 {
1555         return xstrfmt("gui/%d", getuid());
1556 }
1557
1558 static int launchctl_boot_plist(int enable, const char *filename, const char *cmd)
1559 {
1560         int result;
1561         struct child_process child = CHILD_PROCESS_INIT;
1562         char *uid = launchctl_get_uid();
1563
1564         strvec_split(&child.args, cmd);
1565         if (enable)
1566                 strvec_push(&child.args, "bootstrap");
1567         else
1568                 strvec_push(&child.args, "bootout");
1569         strvec_push(&child.args, uid);
1570         strvec_push(&child.args, filename);
1571
1572         child.no_stderr = 1;
1573         child.no_stdout = 1;
1574
1575         if (start_command(&child))
1576                 die(_("failed to start launchctl"));
1577
1578         result = finish_command(&child);
1579
1580         free(uid);
1581         return result;
1582 }
1583
1584 static int launchctl_remove_plist(enum schedule_priority schedule, const char *cmd)
1585 {
1586         const char *frequency = get_frequency(schedule);
1587         char *name = launchctl_service_name(frequency);
1588         char *filename = launchctl_service_filename(name);
1589         int result = launchctl_boot_plist(0, filename, cmd);
1590         unlink(filename);
1591         free(filename);
1592         free(name);
1593         return result;
1594 }
1595
1596 static int launchctl_remove_plists(const char *cmd)
1597 {
1598         return launchctl_remove_plist(SCHEDULE_HOURLY, cmd) ||
1599                 launchctl_remove_plist(SCHEDULE_DAILY, cmd) ||
1600                 launchctl_remove_plist(SCHEDULE_WEEKLY, cmd);
1601 }
1602
1603 static int launchctl_schedule_plist(const char *exec_path, enum schedule_priority schedule, const char *cmd)
1604 {
1605         FILE *plist;
1606         int i;
1607         const char *preamble, *repeat;
1608         const char *frequency = get_frequency(schedule);
1609         char *name = launchctl_service_name(frequency);
1610         char *filename = launchctl_service_filename(name);
1611
1612         if (safe_create_leading_directories(filename))
1613                 die(_("failed to create directories for '%s'"), filename);
1614         plist = xfopen(filename, "w");
1615
1616         preamble = "<?xml version=\"1.0\"?>\n"
1617                    "<!DOCTYPE plist PUBLIC \"-//Apple//DTD PLIST 1.0//EN\" \"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
1618                    "<plist version=\"1.0\">"
1619                    "<dict>\n"
1620                    "<key>Label</key><string>%s</string>\n"
1621                    "<key>ProgramArguments</key>\n"
1622                    "<array>\n"
1623                    "<string>%s/git</string>\n"
1624                    "<string>--exec-path=%s</string>\n"
1625                    "<string>for-each-repo</string>\n"
1626                    "<string>--config=maintenance.repo</string>\n"
1627                    "<string>maintenance</string>\n"
1628                    "<string>run</string>\n"
1629                    "<string>--schedule=%s</string>\n"
1630                    "</array>\n"
1631                    "<key>StartCalendarInterval</key>\n"
1632                    "<array>\n";
1633         fprintf(plist, preamble, name, exec_path, exec_path, frequency);
1634
1635         switch (schedule) {
1636         case SCHEDULE_HOURLY:
1637                 repeat = "<dict>\n"
1638                          "<key>Hour</key><integer>%d</integer>\n"
1639                          "<key>Minute</key><integer>0</integer>\n"
1640                          "</dict>\n";
1641                 for (i = 1; i <= 23; i++)
1642                         fprintf(plist, repeat, i);
1643                 break;
1644
1645         case SCHEDULE_DAILY:
1646                 repeat = "<dict>\n"
1647                          "<key>Day</key><integer>%d</integer>\n"
1648                          "<key>Hour</key><integer>0</integer>\n"
1649                          "<key>Minute</key><integer>0</integer>\n"
1650                          "</dict>\n";
1651                 for (i = 1; i <= 6; i++)
1652                         fprintf(plist, repeat, i);
1653                 break;
1654
1655         case SCHEDULE_WEEKLY:
1656                 fprintf(plist,
1657                         "<dict>\n"
1658                         "<key>Day</key><integer>0</integer>\n"
1659                         "<key>Hour</key><integer>0</integer>\n"
1660                         "<key>Minute</key><integer>0</integer>\n"
1661                         "</dict>\n");
1662                 break;
1663
1664         default:
1665                 /* unreachable */
1666                 break;
1667         }
1668         fprintf(plist, "</array>\n</dict>\n</plist>\n");
1669         fclose(plist);
1670
1671         /* bootout might fail if not already running, so ignore */
1672         launchctl_boot_plist(0, filename, cmd);
1673         if (launchctl_boot_plist(1, filename, cmd))
1674                 die(_("failed to bootstrap service %s"), filename);
1675
1676         free(filename);
1677         free(name);
1678         return 0;
1679 }
1680
1681 static int launchctl_add_plists(const char *cmd)
1682 {
1683         const char *exec_path = git_exec_path();
1684
1685         return launchctl_schedule_plist(exec_path, SCHEDULE_HOURLY, cmd) ||
1686                 launchctl_schedule_plist(exec_path, SCHEDULE_DAILY, cmd) ||
1687                 launchctl_schedule_plist(exec_path, SCHEDULE_WEEKLY, cmd);
1688 }
1689
1690 static int launchctl_update_schedule(int run_maintenance, int fd, const char *cmd)
1691 {
1692         if (run_maintenance)
1693                 return launchctl_add_plists(cmd);
1694         else
1695                 return launchctl_remove_plists(cmd);
1696 }
1697
1698 static char *schtasks_task_name(const char *frequency)
1699 {
1700         struct strbuf label = STRBUF_INIT;
1701         strbuf_addf(&label, "Git Maintenance (%s)", frequency);
1702         return strbuf_detach(&label, NULL);
1703 }
1704
1705 static int schtasks_remove_task(enum schedule_priority schedule, const char *cmd)
1706 {
1707         int result;
1708         struct strvec args = STRVEC_INIT;
1709         const char *frequency = get_frequency(schedule);
1710         char *name = schtasks_task_name(frequency);
1711
1712         strvec_split(&args, cmd);
1713         strvec_pushl(&args, "/delete", "/tn", name, "/f", NULL);
1714
1715         result = run_command_v_opt(args.v, 0);
1716
1717         strvec_clear(&args);
1718         free(name);
1719         return result;
1720 }
1721
1722 static int schtasks_remove_tasks(const char *cmd)
1723 {
1724         return schtasks_remove_task(SCHEDULE_HOURLY, cmd) ||
1725                 schtasks_remove_task(SCHEDULE_DAILY, cmd) ||
1726                 schtasks_remove_task(SCHEDULE_WEEKLY, cmd);
1727 }
1728
1729 static int schtasks_schedule_task(const char *exec_path, enum schedule_priority schedule, const char *cmd)
1730 {
1731         int result;
1732         struct child_process child = CHILD_PROCESS_INIT;
1733         const char *xml;
1734         struct tempfile *tfile;
1735         const char *frequency = get_frequency(schedule);
1736         char *name = schtasks_task_name(frequency);
1737         struct strbuf tfilename = STRBUF_INIT;
1738
1739         strbuf_addf(&tfilename, "%s/schedule_%s_XXXXXX",
1740                     get_git_common_dir(), frequency);
1741         tfile = xmks_tempfile(tfilename.buf);
1742         strbuf_release(&tfilename);
1743
1744         if (!fdopen_tempfile(tfile, "w"))
1745                 die(_("failed to create temp xml file"));
1746
1747         xml = "<?xml version=\"1.0\" ?>\n"
1748               "<Task version=\"1.4\" xmlns=\"http://schemas.microsoft.com/windows/2004/02/mit/task\">\n"
1749               "<Triggers>\n"
1750               "<CalendarTrigger>\n";
1751         fputs(xml, tfile->fp);
1752
1753         switch (schedule) {
1754         case SCHEDULE_HOURLY:
1755                 fprintf(tfile->fp,
1756                         "<StartBoundary>2020-01-01T01:00:00</StartBoundary>\n"
1757                         "<Enabled>true</Enabled>\n"
1758                         "<ScheduleByDay>\n"
1759                         "<DaysInterval>1</DaysInterval>\n"
1760                         "</ScheduleByDay>\n"
1761                         "<Repetition>\n"
1762                         "<Interval>PT1H</Interval>\n"
1763                         "<Duration>PT23H</Duration>\n"
1764                         "<StopAtDurationEnd>false</StopAtDurationEnd>\n"
1765                         "</Repetition>\n");
1766                 break;
1767
1768         case SCHEDULE_DAILY:
1769                 fprintf(tfile->fp,
1770                         "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1771                         "<Enabled>true</Enabled>\n"
1772                         "<ScheduleByWeek>\n"
1773                         "<DaysOfWeek>\n"
1774                         "<Monday />\n"
1775                         "<Tuesday />\n"
1776                         "<Wednesday />\n"
1777                         "<Thursday />\n"
1778                         "<Friday />\n"
1779                         "<Saturday />\n"
1780                         "</DaysOfWeek>\n"
1781                         "<WeeksInterval>1</WeeksInterval>\n"
1782                         "</ScheduleByWeek>\n");
1783                 break;
1784
1785         case SCHEDULE_WEEKLY:
1786                 fprintf(tfile->fp,
1787                         "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1788                         "<Enabled>true</Enabled>\n"
1789                         "<ScheduleByWeek>\n"
1790                         "<DaysOfWeek>\n"
1791                         "<Sunday />\n"
1792                         "</DaysOfWeek>\n"
1793                         "<WeeksInterval>1</WeeksInterval>\n"
1794                         "</ScheduleByWeek>\n");
1795                 break;
1796
1797         default:
1798                 break;
1799         }
1800
1801         xml = "</CalendarTrigger>\n"
1802               "</Triggers>\n"
1803               "<Principals>\n"
1804               "<Principal id=\"Author\">\n"
1805               "<LogonType>InteractiveToken</LogonType>\n"
1806               "<RunLevel>LeastPrivilege</RunLevel>\n"
1807               "</Principal>\n"
1808               "</Principals>\n"
1809               "<Settings>\n"
1810               "<MultipleInstancesPolicy>IgnoreNew</MultipleInstancesPolicy>\n"
1811               "<Enabled>true</Enabled>\n"
1812               "<Hidden>true</Hidden>\n"
1813               "<UseUnifiedSchedulingEngine>true</UseUnifiedSchedulingEngine>\n"
1814               "<WakeToRun>false</WakeToRun>\n"
1815               "<ExecutionTimeLimit>PT72H</ExecutionTimeLimit>\n"
1816               "<Priority>7</Priority>\n"
1817               "</Settings>\n"
1818               "<Actions Context=\"Author\">\n"
1819               "<Exec>\n"
1820               "<Command>\"%s\\git.exe\"</Command>\n"
1821               "<Arguments>--exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%s</Arguments>\n"
1822               "</Exec>\n"
1823               "</Actions>\n"
1824               "</Task>\n";
1825         fprintf(tfile->fp, xml, exec_path, exec_path, frequency);
1826         strvec_split(&child.args, cmd);
1827         strvec_pushl(&child.args, "/create", "/tn", name, "/f", "/xml",
1828                                   get_tempfile_path(tfile), NULL);
1829         close_tempfile_gently(tfile);
1830
1831         child.no_stdout = 1;
1832         child.no_stderr = 1;
1833
1834         if (start_command(&child))
1835                 die(_("failed to start schtasks"));
1836         result = finish_command(&child);
1837
1838         delete_tempfile(&tfile);
1839         free(name);
1840         return result;
1841 }
1842
1843 static int schtasks_schedule_tasks(const char *cmd)
1844 {
1845         const char *exec_path = git_exec_path();
1846
1847         return schtasks_schedule_task(exec_path, SCHEDULE_HOURLY, cmd) ||
1848                 schtasks_schedule_task(exec_path, SCHEDULE_DAILY, cmd) ||
1849                 schtasks_schedule_task(exec_path, SCHEDULE_WEEKLY, cmd);
1850 }
1851
1852 static int schtasks_update_schedule(int run_maintenance, int fd, const char *cmd)
1853 {
1854         if (run_maintenance)
1855                 return schtasks_schedule_tasks(cmd);
1856         else
1857                 return schtasks_remove_tasks(cmd);
1858 }
1859
1860 #define BEGIN_LINE "# BEGIN GIT MAINTENANCE SCHEDULE"
1861 #define END_LINE "# END GIT MAINTENANCE SCHEDULE"
1862
1863 static int crontab_update_schedule(int run_maintenance, int fd, const char *cmd)
1864 {
1865         int result = 0;
1866         int in_old_region = 0;
1867         struct child_process crontab_list = CHILD_PROCESS_INIT;
1868         struct child_process crontab_edit = CHILD_PROCESS_INIT;
1869         FILE *cron_list, *cron_in;
1870         struct strbuf line = STRBUF_INIT;
1871
1872         strvec_split(&crontab_list.args, cmd);
1873         strvec_push(&crontab_list.args, "-l");
1874         crontab_list.in = -1;
1875         crontab_list.out = dup(fd);
1876         crontab_list.git_cmd = 0;
1877
1878         if (start_command(&crontab_list))
1879                 return error(_("failed to run 'crontab -l'; your system might not support 'cron'"));
1880
1881         /* Ignore exit code, as an empty crontab will return error. */
1882         finish_command(&crontab_list);
1883
1884         /*
1885          * Read from the .lock file, filtering out the old
1886          * schedule while appending the new schedule.
1887          */
1888         cron_list = fdopen(fd, "r");
1889         rewind(cron_list);
1890
1891         strvec_split(&crontab_edit.args, cmd);
1892         crontab_edit.in = -1;
1893         crontab_edit.git_cmd = 0;
1894
1895         if (start_command(&crontab_edit))
1896                 return error(_("failed to run 'crontab'; your system might not support 'cron'"));
1897
1898         cron_in = fdopen(crontab_edit.in, "w");
1899         if (!cron_in) {
1900                 result = error(_("failed to open stdin of 'crontab'"));
1901                 goto done_editing;
1902         }
1903
1904         while (!strbuf_getline_lf(&line, cron_list)) {
1905                 if (!in_old_region && !strcmp(line.buf, BEGIN_LINE))
1906                         in_old_region = 1;
1907                 else if (in_old_region && !strcmp(line.buf, END_LINE))
1908                         in_old_region = 0;
1909                 else if (!in_old_region)
1910                         fprintf(cron_in, "%s\n", line.buf);
1911         }
1912         strbuf_release(&line);
1913
1914         if (run_maintenance) {
1915                 struct strbuf line_format = STRBUF_INIT;
1916                 const char *exec_path = git_exec_path();
1917
1918                 fprintf(cron_in, "%s\n", BEGIN_LINE);
1919                 fprintf(cron_in,
1920                         "# The following schedule was created by Git\n");
1921                 fprintf(cron_in, "# Any edits made in this region might be\n");
1922                 fprintf(cron_in,
1923                         "# replaced in the future by a Git command.\n\n");
1924
1925                 strbuf_addf(&line_format,
1926                             "%%s %%s * * %%s \"%s/git\" --exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%%s\n",
1927                             exec_path, exec_path);
1928                 fprintf(cron_in, line_format.buf, "0", "1-23", "*", "hourly");
1929                 fprintf(cron_in, line_format.buf, "0", "0", "1-6", "daily");
1930                 fprintf(cron_in, line_format.buf, "0", "0", "0", "weekly");
1931                 strbuf_release(&line_format);
1932
1933                 fprintf(cron_in, "\n%s\n", END_LINE);
1934         }
1935
1936         fflush(cron_in);
1937         fclose(cron_in);
1938         close(crontab_edit.in);
1939
1940 done_editing:
1941         if (finish_command(&crontab_edit))
1942                 result = error(_("'crontab' died"));
1943         else
1944                 fclose(cron_list);
1945         return result;
1946 }
1947
1948 #if defined(__APPLE__)
1949 static const char platform_scheduler[] = "launchctl";
1950 #elif defined(GIT_WINDOWS_NATIVE)
1951 static const char platform_scheduler[] = "schtasks";
1952 #else
1953 static const char platform_scheduler[] = "crontab";
1954 #endif
1955
1956 static int update_background_schedule(int enable)
1957 {
1958         int result;
1959         const char *scheduler = platform_scheduler;
1960         const char *cmd = scheduler;
1961         char *testing;
1962         struct lock_file lk;
1963         char *lock_path = xstrfmt("%s/schedule", the_repository->objects->odb->path);
1964
1965         testing = xstrdup_or_null(getenv("GIT_TEST_MAINT_SCHEDULER"));
1966         if (testing) {
1967                 char *sep = strchr(testing, ':');
1968                 if (!sep)
1969                         die("GIT_TEST_MAINT_SCHEDULER unparseable: %s", testing);
1970                 *sep = '\0';
1971                 scheduler = testing;
1972                 cmd = sep + 1;
1973         }
1974
1975         if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
1976                 result = error(_("another process is scheduling background maintenance"));
1977                 goto cleanup;
1978         }
1979
1980         if (!strcmp(scheduler, "launchctl"))
1981                 result = launchctl_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1982         else if (!strcmp(scheduler, "schtasks"))
1983                 result = schtasks_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1984         else if (!strcmp(scheduler, "crontab"))
1985                 result = crontab_update_schedule(enable, get_lock_file_fd(&lk), cmd);
1986         else
1987                 die("unknown background scheduler: %s", scheduler);
1988
1989         rollback_lock_file(&lk);
1990
1991 cleanup:
1992         free(lock_path);
1993         free(testing);
1994         return result;
1995 }
1996
1997 static int maintenance_start(void)
1998 {
1999         if (maintenance_register())
2000                 warning(_("failed to add repo to global config"));
2001
2002         return update_background_schedule(1);
2003 }
2004
2005 static int maintenance_stop(void)
2006 {
2007         return update_background_schedule(0);
2008 }
2009
2010 static const char builtin_maintenance_usage[] = N_("git maintenance <subcommand> [<options>]");
2011
2012 int cmd_maintenance(int argc, const char **argv, const char *prefix)
2013 {
2014         if (argc < 2 ||
2015             (argc == 2 && !strcmp(argv[1], "-h")))
2016                 usage(builtin_maintenance_usage);
2017
2018         if (!strcmp(argv[1], "run"))
2019                 return maintenance_run(argc - 1, argv + 1, prefix);
2020         if (!strcmp(argv[1], "start"))
2021                 return maintenance_start();
2022         if (!strcmp(argv[1], "stop"))
2023                 return maintenance_stop();
2024         if (!strcmp(argv[1], "register"))
2025                 return maintenance_register();
2026         if (!strcmp(argv[1], "unregister"))
2027                 return maintenance_unregister();
2028
2029         die(_("invalid subcommand: %s"), argv[1]);
2030 }