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