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
  33 #define FAILED_RUN "failed to run %s"
  34
  35 static const char * const builtin_gc_usage[] = {
  36         N_("git gc [<options>]"),
  37         NULL
  38 };
  39
  40 static int pack_refs = 1;
  41 static int prune_reflogs = 1;
  42 static int aggressive_depth = 50;
  43 static int aggressive_window = 250;
  44 static int gc_auto_threshold = 6700;
  45 static int gc_auto_pack_limit = 50;
  46 static int detach_auto = 1;
  47 static timestamp_t gc_log_expire_time;
  48 static const char *gc_log_expire = "1.day.ago";
  49 static const char *prune_expire = "2.weeks.ago";
  50 static const char *prune_worktrees_expire = "3.months.ago";
  51 static unsigned long big_pack_threshold;
  52 static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
  53
  54 static struct strvec pack_refs_cmd = STRVEC_INIT;
  55 static struct strvec reflog = STRVEC_INIT;
  56 static struct strvec repack = STRVEC_INIT;
  57 static struct strvec prune = STRVEC_INIT;
  58 static struct strvec prune_worktrees = STRVEC_INIT;
  59 static struct strvec rerere = STRVEC_INIT;
  60
  61 static struct tempfile *pidfile;
  62 static struct lock_file log_lock;
  63
  64 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  65
  66 static void clean_pack_garbage(void)
  67 {
  68         int i;
  69         for (i = 0; i < pack_garbage.nr; i++)
  70                 unlink_or_warn(pack_garbage.items[i].string);
  71         string_list_clear(&pack_garbage, 0);
  72 }
  73
  74 static void report_pack_garbage(unsigned seen_bits, const char *path)
  75 {
  76         if (seen_bits == PACKDIR_FILE_IDX)
  77                 string_list_append(&pack_garbage, path);
  78 }
  79
  80 static void process_log_file(void)
  81 {
  82         struct stat st;
  83         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  84                 /*
  85                  * Perhaps there was an i/o error or another
  86                  * unlikely situation.  Try to make a note of
  87                  * this in gc.log along with any existing
  88                  * messages.
  89                  */
  90                 int saved_errno = errno;
  91                 fprintf(stderr, _("Failed to fstat %s: %s"),
  92                         get_tempfile_path(log_lock.tempfile),
  93                         strerror(saved_errno));
  94                 fflush(stderr);
  95                 commit_lock_file(&log_lock);
  96                 errno = saved_errno;
  97         } else if (st.st_size) {
  98                 /* There was some error recorded in the lock file */
  99                 commit_lock_file(&log_lock);
 100         } else {
 101                 /* No error, clean up any old gc.log */
 102                 unlink(git_path("gc.log"));
 103                 rollback_lock_file(&log_lock);
 104         }
 105 }
 106
 107 static void process_log_file_at_exit(void)
 108 {
 109         fflush(stderr);
 110         process_log_file();
 111 }
 112
 113 static void process_log_file_on_signal(int signo)
 114 {
 115         process_log_file();
 116         sigchain_pop(signo);
 117         raise(signo);
 118 }
 119
 120 static int gc_config_is_timestamp_never(const char *var)
 121 {
 122         const char *value;
 123         timestamp_t expire;
 124
 125         if (!git_config_get_value(var, &value) && value) {
 126                 if (parse_expiry_date(value, &expire))
 127                         die(_("failed to parse '%s' value '%s'"), var, value);
 128                 return expire == 0;
 129         }
 130         return 0;
 131 }
 132
 133 static void gc_config(void)
 134 {
 135         const char *value;
 136
 137         if (!git_config_get_value("gc.packrefs", &value)) {
 138                 if (value && !strcmp(value, "notbare"))
 139                         pack_refs = -1;
 140                 else
 141                         pack_refs = git_config_bool("gc.packrefs", value);
 142         }
 143
 144         if (gc_config_is_timestamp_never("gc.reflogexpire") &&
 145             gc_config_is_timestamp_never("gc.reflogexpireunreachable"))
 146                 prune_reflogs = 0;
 147
 148         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 149         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 150         git_config_get_int("gc.auto", &gc_auto_threshold);
 151         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 152         git_config_get_bool("gc.autodetach", &detach_auto);
 153         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 154         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 155         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 156
 157         git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold);
 158         git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size);
 159
 160         git_config(git_default_config, NULL);
 161 }
 162
 163 static int too_many_loose_objects(void)
 164 {
 165         /*
 166          * Quickly check if a "gc" is needed, by estimating how
 167          * many loose objects there are.  Because SHA-1 is evenly
 168          * distributed, we can check only one and get a reasonable
 169          * estimate.
 170          */
 171         DIR *dir;
 172         struct dirent *ent;
 173         int auto_threshold;
 174         int num_loose = 0;
 175         int needed = 0;
 176         const unsigned hexsz_loose = the_hash_algo->hexsz - 2;
 177
 178         dir = opendir(git_path("objects/17"));
 179         if (!dir)
 180                 return 0;
 181
 182         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
 183         while ((ent = readdir(dir)) != NULL) {
 184                 if (strspn(ent->d_name, "0123456789abcdef") != hexsz_loose ||
 185                     ent->d_name[hexsz_loose] != '\0')
 186                         continue;
 187                 if (++num_loose > auto_threshold) {
 188                         needed = 1;
 189                         break;
 190                 }
 191         }
 192         closedir(dir);
 193         return needed;
 194 }
 195
 196 static struct packed_git *find_base_packs(struct string_list *packs,
 197                                           unsigned long limit)
 198 {
 199         struct packed_git *p, *base = NULL;
 200
 201         for (p = get_all_packs(the_repository); p; p = p->next) {
 202                 if (!p->pack_local)
 203                         continue;
 204                 if (limit) {
 205                         if (p->pack_size >= limit)
 206                                 string_list_append(packs, p->pack_name);
 207                 } else if (!base || base->pack_size < p->pack_size) {
 208                         base = p;
 209                 }
 210         }
 211
 212         if (base)
 213                 string_list_append(packs, base->pack_name);
 214
 215         return base;
 216 }
 217
 218 static int too_many_packs(void)
 219 {
 220         struct packed_git *p;
 221         int cnt;
 222
 223         if (gc_auto_pack_limit <= 0)
 224                 return 0;
 225
 226         for (cnt = 0, p = get_all_packs(the_repository); p; p = p->next) {
 227                 if (!p->pack_local)
 228                         continue;
 229                 if (p->pack_keep)
 230                         continue;
 231                 /*
 232                  * Perhaps check the size of the pack and count only
 233                  * very small ones here?
 234                  */
 235                 cnt++;
 236         }
 237         return gc_auto_pack_limit < cnt;
 238 }
 239
 240 static uint64_t total_ram(void)
 241 {
 242 #if defined(HAVE_SYSINFO)
 243         struct sysinfo si;
 244
 245         if (!sysinfo(&si))
 246                 return si.totalram;
 247 #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM))
 248         int64_t physical_memory;
 249         int mib[2];
 250         size_t length;
 251
 252         mib[0] = CTL_HW;
 253 # if defined(HW_MEMSIZE)
 254         mib[1] = HW_MEMSIZE;
 255 # else
 256         mib[1] = HW_PHYSMEM;
 257 # endif
 258         length = sizeof(int64_t);
 259         if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
 260                 return physical_memory;
 261 #elif defined(GIT_WINDOWS_NATIVE)
 262         MEMORYSTATUSEX memInfo;
 263
 264         memInfo.dwLength = sizeof(MEMORYSTATUSEX);
 265         if (GlobalMemoryStatusEx(&memInfo))
 266                 return memInfo.ullTotalPhys;
 267 #endif
 268         return 0;
 269 }
 270
 271 static uint64_t estimate_repack_memory(struct packed_git *pack)
 272 {
 273         unsigned long nr_objects = approximate_object_count();
 274         size_t os_cache, heap;
 275
 276         if (!pack || !nr_objects)
 277                 return 0;
 278
 279         /*
 280          * First we have to scan through at least one pack.
 281          * Assume enough room in OS file cache to keep the entire pack
 282          * or we may accidentally evict data of other processes from
 283          * the cache.
 284          */
 285         os_cache = pack->pack_size + pack->index_size;
 286         /* then pack-objects needs lots more for book keeping */
 287         heap = sizeof(struct object_entry) * nr_objects;
 288         /*
 289          * internal rev-list --all --objects takes up some memory too,
 290          * let's say half of it is for blobs
 291          */
 292         heap += sizeof(struct blob) * nr_objects / 2;
 293         /*
 294          * and the other half is for trees (commits and tags are
 295          * usually insignificant)
 296          */
 297         heap += sizeof(struct tree) * nr_objects / 2;
 298         /* and then obj_hash[], underestimated in fact */
 299         heap += sizeof(struct object *) * nr_objects;
 300         /* revindex is used also */
 301         heap += sizeof(struct revindex_entry) * nr_objects;
 302         /*
 303          * read_sha1_file() (either at delta calculation phase, or
 304          * writing phase) also fills up the delta base cache
 305          */
 306         heap += delta_base_cache_limit;
 307         /* and of course pack-objects has its own delta cache */
 308         heap += max_delta_cache_size;
 309
 310         return os_cache + heap;
 311 }
 312
 313 static int keep_one_pack(struct string_list_item *item, void *data)
 314 {
 315         strvec_pushf(&repack, "--keep-pack=%s", basename(item->string));
 316         return 0;
 317 }
 318
 319 static void add_repack_all_option(struct string_list *keep_pack)
 320 {
 321         if (prune_expire && !strcmp(prune_expire, "now"))
 322                 strvec_push(&repack, "-a");
 323         else {
 324                 strvec_push(&repack, "-A");
 325                 if (prune_expire)
 326                         strvec_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 327         }
 328
 329         if (keep_pack)
 330                 for_each_string_list(keep_pack, keep_one_pack, NULL);
 331 }
 332
 333 static void add_repack_incremental_option(void)
 334 {
 335         strvec_push(&repack, "--no-write-bitmap-index");
 336 }
 337
 338 static int need_to_gc(void)
 339 {
 340         /*
 341          * Setting gc.auto to 0 or negative can disable the
 342          * automatic gc.
 343          */
 344         if (gc_auto_threshold <= 0)
 345                 return 0;
 346
 347         /*
 348          * If there are too many loose objects, but not too many
 349          * packs, we run "repack -d -l".  If there are too many packs,
 350          * we run "repack -A -d -l".  Otherwise we tell the caller
 351          * there is no need.
 352          */
 353         if (too_many_packs()) {
 354                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 355
 356                 if (big_pack_threshold) {
 357                         find_base_packs(&keep_pack, big_pack_threshold);
 358                         if (keep_pack.nr >= gc_auto_pack_limit) {
 359                                 big_pack_threshold = 0;
 360                                 string_list_clear(&keep_pack, 0);
 361                                 find_base_packs(&keep_pack, 0);
 362                         }
 363                 } else {
 364                         struct packed_git *p = find_base_packs(&keep_pack, 0);
 365                         uint64_t mem_have, mem_want;
 366
 367                         mem_have = total_ram();
 368                         mem_want = estimate_repack_memory(p);
 369
 370                         /*
 371                          * Only allow 1/2 of memory for pack-objects, leave
 372                          * the rest for the OS and other processes in the
 373                          * system.
 374                          */
 375                         if (!mem_have || mem_want < mem_have / 2)
 376                                 string_list_clear(&keep_pack, 0);
 377                 }
 378
 379                 add_repack_all_option(&keep_pack);
 380                 string_list_clear(&keep_pack, 0);
 381         } else if (too_many_loose_objects())
 382                 add_repack_incremental_option();
 383         else
 384                 return 0;
 385
 386         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 387                 return 0;
 388         return 1;
 389 }
 390
 391 /* return NULL on success, else hostname running the gc */
 392 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 393 {
 394         struct lock_file lock = LOCK_INIT;
 395         char my_host[HOST_NAME_MAX + 1];
 396         struct strbuf sb = STRBUF_INIT;
 397         struct stat st;
 398         uintmax_t pid;
 399         FILE *fp;
 400         int fd;
 401         char *pidfile_path;
 402
 403         if (is_tempfile_active(pidfile))
 404                 /* already locked */
 405                 return NULL;
 406
 407         if (xgethostname(my_host, sizeof(my_host)))
 408                 xsnprintf(my_host, sizeof(my_host), "unknown");
 409
 410         pidfile_path = git_pathdup("gc.pid");
 411         fd = hold_lock_file_for_update(&lock, pidfile_path,
 412                                        LOCK_DIE_ON_ERROR);
 413         if (!force) {
 414                 static char locking_host[HOST_NAME_MAX + 1];
 415                 static char *scan_fmt;
 416                 int should_exit;
 417
 418                 if (!scan_fmt)
 419                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
 420                 fp = fopen(pidfile_path, "r");
 421                 memset(locking_host, 0, sizeof(locking_host));
 422                 should_exit =
 423                         fp != NULL &&
 424                         !fstat(fileno(fp), &st) &&
 425                         /*
 426                          * 12 hour limit is very generous as gc should
 427                          * never take that long. On the other hand we
 428                          * don't really need a strict limit here,
 429                          * running gc --auto one day late is not a big
 430                          * problem. --force can be used in manual gc
 431                          * after the user verifies that no gc is
 432                          * running.
 433                          */
 434                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 435                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 436                         /* be gentle to concurrent "gc" on remote hosts */
 437                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 438                 if (fp != NULL)
 439                         fclose(fp);
 440                 if (should_exit) {
 441                         if (fd >= 0)
 442                                 rollback_lock_file(&lock);
 443                         *ret_pid = pid;
 444                         free(pidfile_path);
 445                         return locking_host;
 446                 }
 447         }
 448
 449         strbuf_addf(&sb, "%"PRIuMAX" %s",
 450                     (uintmax_t) getpid(), my_host);
 451         write_in_full(fd, sb.buf, sb.len);
 452         strbuf_release(&sb);
 453         commit_lock_file(&lock);
 454         pidfile = register_tempfile(pidfile_path);
 455         free(pidfile_path);
 456         return NULL;
 457 }
 458
 459 /*
 460  * Returns 0 if there was no previous error and gc can proceed, 1 if
 461  * gc should not proceed due to an error in the last run. Prints a
 462  * message and returns -1 if an error occurred while reading gc.log
 463  */
 464 static int report_last_gc_error(void)
 465 {
 466         struct strbuf sb = STRBUF_INIT;
 467         int ret = 0;
 468         ssize_t len;
 469         struct stat st;
 470         char *gc_log_path = git_pathdup("gc.log");
 471
 472         if (stat(gc_log_path, &st)) {
 473                 if (errno == ENOENT)
 474                         goto done;
 475
 476                 ret = error_errno(_("cannot stat '%s'"), gc_log_path);
 477                 goto done;
 478         }
 479
 480         if (st.st_mtime < gc_log_expire_time)
 481                 goto done;
 482
 483         len = strbuf_read_file(&sb, gc_log_path, 0);
 484         if (len < 0)
 485                 ret = error_errno(_("cannot read '%s'"), gc_log_path);
 486         else if (len > 0) {
 487                 /*
 488                  * A previous gc failed.  Report the error, and don't
 489                  * bother with an automatic gc run since it is likely
 490                  * to fail in the same way.
 491                  */
 492                 warning(_("The last gc run reported the following. "
 493                                "Please correct the root cause\n"
 494                                "and remove %s.\n"
 495                                "Automatic cleanup will not be performed "
 496                                "until the file is removed.\n\n"
 497                                "%s"),
 498                             gc_log_path, sb.buf);
 499                 ret = 1;
 500         }
 501         strbuf_release(&sb);
 502 done:
 503         free(gc_log_path);
 504         return ret;
 505 }
 506
 507 static void gc_before_repack(void)
 508 {
 509         /*
 510          * We may be called twice, as both the pre- and
 511          * post-daemonized phases will call us, but running these
 512          * commands more than once is pointless and wasteful.
 513          */
 514         static int done = 0;
 515         if (done++)
 516                 return;
 517
 518         if (pack_refs && run_command_v_opt(pack_refs_cmd.v, RUN_GIT_CMD))
 519                 die(FAILED_RUN, pack_refs_cmd.v[0]);
 520
 521         if (prune_reflogs && run_command_v_opt(reflog.v, RUN_GIT_CMD))
 522                 die(FAILED_RUN, reflog.v[0]);
 523 }
 524
 525 int cmd_gc(int argc, const char **argv, const char *prefix)
 526 {
 527         int aggressive = 0;
 528         int auto_gc = 0;
 529         int quiet = 0;
 530         int force = 0;
 531         const char *name;
 532         pid_t pid;
 533         int daemonized = 0;
 534         int keep_base_pack = -1;
 535         timestamp_t dummy;
 536
 537         struct option builtin_gc_options[] = {
 538                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 539                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 540                         N_("prune unreferenced objects"),
 541                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 542                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 543                 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
 544                            PARSE_OPT_NOCOMPLETE),
 545                 OPT_BOOL_F(0, "force", &force,
 546                            N_("force running gc even if there may be another gc running"),
 547                            PARSE_OPT_NOCOMPLETE),
 548                 OPT_BOOL(0, "keep-largest-pack", &keep_base_pack,
 549                          N_("repack all other packs except the largest pack")),
 550                 OPT_END()
 551         };
 552
 553         if (argc == 2 && !strcmp(argv[1], "-h"))
 554                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 555
 556         strvec_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 557         strvec_pushl(&reflog, "reflog", "expire", "--all", NULL);
 558         strvec_pushl(&repack, "repack", "-d", "-l", NULL);
 559         strvec_pushl(&prune, "prune", "--expire", NULL);
 560         strvec_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 561         strvec_pushl(&rerere, "rerere", "gc", NULL);
 562
 563         /* default expiry time, overwritten in gc_config */
 564         gc_config();
 565         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 566                 die(_("failed to parse gc.logexpiry value %s"), gc_log_expire);
 567
 568         if (pack_refs < 0)
 569                 pack_refs = !is_bare_repository();
 570
 571         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 572                              builtin_gc_usage, 0);
 573         if (argc > 0)
 574                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 575
 576         if (prune_expire && parse_expiry_date(prune_expire, &dummy))
 577                 die(_("failed to parse prune expiry value %s"), prune_expire);
 578
 579         if (aggressive) {
 580                 strvec_push(&repack, "-f");
 581                 if (aggressive_depth > 0)
 582                         strvec_pushf(&repack, "--depth=%d", aggressive_depth);
 583                 if (aggressive_window > 0)
 584                         strvec_pushf(&repack, "--window=%d", aggressive_window);
 585         }
 586         if (quiet)
 587                 strvec_push(&repack, "-q");
 588
 589         if (auto_gc) {
 590                 /*
 591                  * Auto-gc should be least intrusive as possible.
 592                  */
 593                 if (!need_to_gc())
 594                         return 0;
 595                 if (!quiet) {
 596                         if (detach_auto)
 597                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 598                         else
 599                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 600                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 601                 }
 602                 if (detach_auto) {
 603                         int ret = report_last_gc_error();
 604                         if (ret < 0)
 605                                 /* an I/O error occurred, already reported */
 606                                 exit(128);
 607                         if (ret == 1)
 608                                 /* Last gc --auto failed. Skip this one. */
 609                                 return 0;
 610
 611                         if (lock_repo_for_gc(force, &pid))
 612                                 return 0;
 613                         gc_before_repack(); /* dies on failure */
 614                         delete_tempfile(&pidfile);
 615
 616                         /*
 617                          * failure to daemonize is ok, we'll continue
 618                          * in foreground
 619                          */
 620                         daemonized = !daemonize();
 621                 }
 622         } else {
 623                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 624
 625                 if (keep_base_pack != -1) {
 626                         if (keep_base_pack)
 627                                 find_base_packs(&keep_pack, 0);
 628                 } else if (big_pack_threshold) {
 629                         find_base_packs(&keep_pack, big_pack_threshold);
 630                 }
 631
 632                 add_repack_all_option(&keep_pack);
 633                 string_list_clear(&keep_pack, 0);
 634         }
 635
 636         name = lock_repo_for_gc(force, &pid);
 637         if (name) {
 638                 if (auto_gc)
 639                         return 0; /* be quiet on --auto */
 640                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 641                     name, (uintmax_t)pid);
 642         }
 643
 644         if (daemonized) {
 645                 hold_lock_file_for_update(&log_lock,
 646                                           git_path("gc.log"),
 647                                           LOCK_DIE_ON_ERROR);
 648                 dup2(get_lock_file_fd(&log_lock), 2);
 649                 sigchain_push_common(process_log_file_on_signal);
 650                 atexit(process_log_file_at_exit);
 651         }
 652
 653         gc_before_repack();
 654
 655         if (!repository_format_precious_objects) {
 656                 close_object_store(the_repository->objects);
 657                 if (run_command_v_opt(repack.v, RUN_GIT_CMD))
 658                         die(FAILED_RUN, repack.v[0]);
 659
 660                 if (prune_expire) {
 661                         strvec_push(&prune, prune_expire);
 662                         if (quiet)
 663                                 strvec_push(&prune, "--no-progress");
 664                         if (has_promisor_remote())
 665                                 strvec_push(&prune,
 666                                             "--exclude-promisor-objects");
 667                         if (run_command_v_opt(prune.v, RUN_GIT_CMD))
 668                                 die(FAILED_RUN, prune.v[0]);
 669                 }
 670         }
 671
 672         if (prune_worktrees_expire) {
 673                 strvec_push(&prune_worktrees, prune_worktrees_expire);
 674                 if (run_command_v_opt(prune_worktrees.v, RUN_GIT_CMD))
 675                         die(FAILED_RUN, prune_worktrees.v[0]);
 676         }
 677
 678         if (run_command_v_opt(rerere.v, RUN_GIT_CMD))
 679                 die(FAILED_RUN, rerere.v[0]);
 680
 681         report_garbage = report_pack_garbage;
 682         reprepare_packed_git(the_repository);
 683         if (pack_garbage.nr > 0) {
 684                 close_object_store(the_repository->objects);
 685                 clean_pack_garbage();
 686         }
 687
 688         prepare_repo_settings(the_repository);
 689         if (the_repository->settings.gc_write_commit_graph == 1)
 690                 write_commit_graph_reachable(the_repository->objects->odb,
 691                                              !quiet && !daemonized ? COMMIT_GRAPH_WRITE_PROGRESS : 0,
 692                                              NULL);
 693
 694         if (auto_gc && too_many_loose_objects())
 695                 warning(_("There are too many unreachable loose objects; "
 696                         "run 'git prune' to remove them."));
 697
 698         if (!daemonized)
 699                 unlink(git_path("gc.log"));
 700
 701         return 0;
 702 }
 703
 704 static const char * const builtin_maintenance_run_usage[] = {
 705         N_("git maintenance run [--auto] [--[no-]quiet] [--task=<task>]"),
 706         NULL
 707 };
 708
 709 struct maintenance_run_opts {
 710         int auto_flag;
 711         int quiet;
 712 };
 713
 714 /* Remember to update object flag allocation in object.h */
 715 #define SEEN            (1u<<0)
 716
 717 struct cg_auto_data {
 718         int num_not_in_graph;
 719         int limit;
 720 };
 721
 722 static int dfs_on_ref(const char *refname,
 723                       const struct object_id *oid, int flags,
 724                       void *cb_data)
 725 {
 726         struct cg_auto_data *data = (struct cg_auto_data *)cb_data;
 727         int result = 0;
 728         struct object_id peeled;
 729         struct commit_list *stack = NULL;
 730         struct commit *commit;
 731
 732         if (!peel_ref(refname, &peeled))
 733                 oid = &peeled;
 734         if (oid_object_info(the_repository, oid, NULL) != OBJ_COMMIT)
 735                 return 0;
 736
 737         commit = lookup_commit(the_repository, oid);
 738         if (!commit)
 739                 return 0;
 740         if (parse_commit(commit))
 741                 return 0;
 742
 743         commit_list_append(commit, &stack);
 744
 745         while (!result && stack) {
 746                 struct commit_list *parent;
 747
 748                 commit = pop_commit(&stack);
 749
 750                 for (parent = commit->parents; parent; parent = parent->next) {
 751                         if (parse_commit(parent->item) ||
 752                             commit_graph_position(parent->item) != COMMIT_NOT_FROM_GRAPH ||
 753                             parent->item->object.flags & SEEN)
 754                                 continue;
 755
 756                         parent->item->object.flags |= SEEN;
 757                         data->num_not_in_graph++;
 758
 759                         if (data->num_not_in_graph >= data->limit) {
 760                                 result = 1;
 761                                 break;
 762                         }
 763
 764                         commit_list_append(parent->item, &stack);
 765                 }
 766         }
 767
 768         free_commit_list(stack);
 769         return result;
 770 }
 771
 772 static int should_write_commit_graph(void)
 773 {
 774         int result;
 775         struct cg_auto_data data;
 776
 777         data.num_not_in_graph = 0;
 778         data.limit = 100;
 779         git_config_get_int("maintenance.commit-graph.auto",
 780                            &data.limit);
 781
 782         if (!data.limit)
 783                 return 0;
 784         if (data.limit < 0)
 785                 return 1;
 786
 787         result = for_each_ref(dfs_on_ref, &data);
 788
 789         clear_commit_marks_all(SEEN);
 790
 791         return result;
 792 }
 793
 794 static int run_write_commit_graph(struct maintenance_run_opts *opts)
 795 {
 796         struct child_process child = CHILD_PROCESS_INIT;
 797
 798         child.git_cmd = 1;
 799         strvec_pushl(&child.args, "commit-graph", "write",
 800                      "--split", "--reachable", NULL);
 801
 802         if (opts->quiet)
 803                 strvec_push(&child.args, "--no-progress");
 804
 805         return !!run_command(&child);
 806 }
 807
 808 static int maintenance_task_commit_graph(struct maintenance_run_opts *opts)
 809 {
 810         close_object_store(the_repository->objects);
 811         if (run_write_commit_graph(opts)) {
 812                 error(_("failed to write commit-graph"));
 813                 return 1;
 814         }
 815
 816         return 0;
 817 }
 818
 819 static int maintenance_task_gc(struct maintenance_run_opts *opts)
 820 {
 821         struct child_process child = CHILD_PROCESS_INIT;
 822
 823         child.git_cmd = 1;
 824         strvec_push(&child.args, "gc");
 825
 826         if (opts->auto_flag)
 827                 strvec_push(&child.args, "--auto");
 828         if (opts->quiet)
 829                 strvec_push(&child.args, "--quiet");
 830         else
 831                 strvec_push(&child.args, "--no-quiet");
 832
 833         close_object_store(the_repository->objects);
 834         return run_command(&child);
 835 }
 836
 837 typedef int maintenance_task_fn(struct maintenance_run_opts *opts);
 838
 839 /*
 840  * An auto condition function returns 1 if the task should run
 841  * and 0 if the task should NOT run. See needs_to_gc() for an
 842  * example.
 843  */
 844 typedef int maintenance_auto_fn(void);
 845
 846 struct maintenance_task {
 847         const char *name;
 848         maintenance_task_fn *fn;
 849         maintenance_auto_fn *auto_condition;
 850         unsigned enabled:1;
 851
 852         /* -1 if not selected. */
 853         int selected_order;
 854 };
 855
 856 enum maintenance_task_label {
 857         TASK_GC,
 858         TASK_COMMIT_GRAPH,
 859
 860         /* Leave as final value */
 861         TASK__COUNT
 862 };
 863
 864 static struct maintenance_task tasks[] = {
 865         [TASK_GC] = {
 866                 "gc",
 867                 maintenance_task_gc,
 868                 need_to_gc,
 869                 1,
 870         },
 871         [TASK_COMMIT_GRAPH] = {
 872                 "commit-graph",
 873                 maintenance_task_commit_graph,
 874                 should_write_commit_graph,
 875         },
 876 };
 877
 878 static int compare_tasks_by_selection(const void *a_, const void *b_)
 879 {
 880         const struct maintenance_task *a, *b;
 881
 882         a = (const struct maintenance_task *)&a_;
 883         b = (const struct maintenance_task *)&b_;
 884
 885         return b->selected_order - a->selected_order;
 886 }
 887
 888 static int maintenance_run_tasks(struct maintenance_run_opts *opts)
 889 {
 890         int i, found_selected = 0;
 891         int result = 0;
 892         struct lock_file lk;
 893         struct repository *r = the_repository;
 894         char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path);
 895
 896         if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
 897                 /*
 898                  * Another maintenance command is running.
 899                  *
 900                  * If --auto was provided, then it is likely due to a
 901                  * recursive process stack. Do not report an error in
 902                  * that case.
 903                  */
 904                 if (!opts->auto_flag && !opts->quiet)
 905                         warning(_("lock file '%s' exists, skipping maintenance"),
 906                                 lock_path);
 907                 free(lock_path);
 908                 return 0;
 909         }
 910         free(lock_path);
 911
 912         for (i = 0; !found_selected && i < TASK__COUNT; i++)
 913                 found_selected = tasks[i].selected_order >= 0;
 914
 915         if (found_selected)
 916                 QSORT(tasks, TASK__COUNT, compare_tasks_by_selection);
 917
 918         for (i = 0; i < TASK__COUNT; i++) {
 919                 if (found_selected && tasks[i].selected_order < 0)
 920                         continue;
 921
 922                 if (!found_selected && !tasks[i].enabled)
 923                         continue;
 924
 925                 if (opts->auto_flag &&
 926                     (!tasks[i].auto_condition ||
 927                      !tasks[i].auto_condition()))
 928                         continue;
 929
 930                 trace2_region_enter("maintenance", tasks[i].name, r);
 931                 if (tasks[i].fn(opts)) {
 932                         error(_("task '%s' failed"), tasks[i].name);
 933                         result = 1;
 934                 }
 935                 trace2_region_leave("maintenance", tasks[i].name, r);
 936         }
 937
 938         rollback_lock_file(&lk);
 939         return result;
 940 }
 941
 942 static void initialize_task_config(void)
 943 {
 944         int i;
 945         struct strbuf config_name = STRBUF_INIT;
 946         gc_config();
 947
 948         for (i = 0; i < TASK__COUNT; i++) {
 949                 int config_value;
 950
 951                 strbuf_setlen(&config_name, 0);
 952                 strbuf_addf(&config_name, "maintenance.%s.enabled",
 953                             tasks[i].name);
 954
 955                 if (!git_config_get_bool(config_name.buf, &config_value))
 956                         tasks[i].enabled = config_value;
 957         }
 958
 959         strbuf_release(&config_name);
 960 }
 961
 962 static int task_option_parse(const struct option *opt,
 963                              const char *arg, int unset)
 964 {
 965         int i, num_selected = 0;
 966         struct maintenance_task *task = NULL;
 967
 968         BUG_ON_OPT_NEG(unset);
 969
 970         for (i = 0; i < TASK__COUNT; i++) {
 971                 if (tasks[i].selected_order >= 0)
 972                         num_selected++;
 973                 if (!strcasecmp(tasks[i].name, arg)) {
 974                         task = &tasks[i];
 975                 }
 976         }
 977
 978         if (!task) {
 979                 error(_("'%s' is not a valid task"), arg);
 980                 return 1;
 981         }
 982
 983         if (task->selected_order >= 0) {
 984                 error(_("task '%s' cannot be selected multiple times"), arg);
 985                 return 1;
 986         }
 987
 988         task->selected_order = num_selected + 1;
 989
 990         return 0;
 991 }
 992
 993 static int maintenance_run(int argc, const char **argv, const char *prefix)
 994 {
 995         int i;
 996         struct maintenance_run_opts opts;
 997         struct option builtin_maintenance_run_options[] = {
 998                 OPT_BOOL(0, "auto", &opts.auto_flag,
 999                          N_("run tasks based on the state of the repository")),
1000                 OPT_BOOL(0, "quiet", &opts.quiet,
1001                          N_("do not report progress or other information over stderr")),
1002                 OPT_CALLBACK_F(0, "task", NULL, N_("task"),
1003                         N_("run a specific task"),
1004                         PARSE_OPT_NONEG, task_option_parse),
1005                 OPT_END()
1006         };
1007         memset(&opts, 0, sizeof(opts));
1008
1009         opts.quiet = !isatty(2);
1010         initialize_task_config();
1011
1012         for (i = 0; i < TASK__COUNT; i++)
1013                 tasks[i].selected_order = -1;
1014
1015         argc = parse_options(argc, argv, prefix,
1016                              builtin_maintenance_run_options,
1017                              builtin_maintenance_run_usage,
1018                              PARSE_OPT_STOP_AT_NON_OPTION);
1019
1020         if (argc != 0)
1021                 usage_with_options(builtin_maintenance_run_usage,
1022                                    builtin_maintenance_run_options);
1023         return maintenance_run_tasks(&opts);
1024 }
1025
1026 static const char builtin_maintenance_usage[] = N_("git maintenance run [<options>]");
1027
1028 int cmd_maintenance(int argc, const char **argv, const char *prefix)
1029 {
1030         if (argc < 2 ||
1031             (argc == 2 && !strcmp(argv[1], "-h")))
1032                 usage(builtin_maintenance_usage);
1033
1034         if (!strcmp(argv[1], "run"))
1035                 return maintenance_run(argc - 1, argv + 1, prefix);
1036
1037         die(_("invalid subcommand: %s"), argv[1]);
1038 }