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 "argv-array.h"
  22 #include "commit.h"
  23 #include "packfile.h"
  24 #include "object-store.h"
  25 #include "pack.h"
  26 #include "pack-objects.h"
  27 #include "blob.h"
  28 #include "tree.h"
  29
  30 #define FAILED_RUN "failed to run %s"
  31
  32 static const char * const builtin_gc_usage[] = {
  33         N_("git gc [<options>]"),
  34         NULL
  35 };
  36
  37 static int pack_refs = 1;
  38 static int prune_reflogs = 1;
  39 static int aggressive_depth = 50;
  40 static int aggressive_window = 250;
  41 static int gc_auto_threshold = 6700;
  42 static int gc_auto_pack_limit = 50;
  43 static int detach_auto = 1;
  44 static timestamp_t gc_log_expire_time;
  45 static const char *gc_log_expire = "1.day.ago";
  46 static const char *prune_expire = "2.weeks.ago";
  47 static const char *prune_worktrees_expire = "3.months.ago";
  48 static unsigned long big_pack_threshold;
  49 static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
  50
  51 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  52 static struct argv_array reflog = ARGV_ARRAY_INIT;
  53 static struct argv_array repack = ARGV_ARRAY_INIT;
  54 static struct argv_array prune = ARGV_ARRAY_INIT;
  55 static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
  56 static struct argv_array rerere = ARGV_ARRAY_INIT;
  57
  58 static struct tempfile *pidfile;
  59 static struct lock_file log_lock;
  60
  61 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  62
  63 static void clean_pack_garbage(void)
  64 {
  65         int i;
  66         for (i = 0; i < pack_garbage.nr; i++)
  67                 unlink_or_warn(pack_garbage.items[i].string);
  68         string_list_clear(&pack_garbage, 0);
  69 }
  70
  71 static void report_pack_garbage(unsigned seen_bits, const char *path)
  72 {
  73         if (seen_bits == PACKDIR_FILE_IDX)
  74                 string_list_append(&pack_garbage, path);
  75 }
  76
  77 static void process_log_file(void)
  78 {
  79         struct stat st;
  80         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  81                 /*
  82                  * Perhaps there was an i/o error or another
  83                  * unlikely situation.  Try to make a note of
  84                  * this in gc.log along with any existing
  85                  * messages.
  86                  */
  87                 int saved_errno = errno;
  88                 fprintf(stderr, _("Failed to fstat %s: %s"),
  89                         get_tempfile_path(log_lock.tempfile),
  90                         strerror(saved_errno));
  91                 fflush(stderr);
  92                 commit_lock_file(&log_lock);
  93                 errno = saved_errno;
  94         } else if (st.st_size) {
  95                 /* There was some error recorded in the lock file */
  96                 commit_lock_file(&log_lock);
  97         } else {
  98                 /* No error, clean up any old gc.log */
  99                 unlink(git_path("gc.log"));
 100                 rollback_lock_file(&log_lock);
 101         }
 102 }
 103
 104 static void process_log_file_at_exit(void)
 105 {
 106         fflush(stderr);
 107         process_log_file();
 108 }
 109
 110 static void process_log_file_on_signal(int signo)
 111 {
 112         process_log_file();
 113         sigchain_pop(signo);
 114         raise(signo);
 115 }
 116
 117 static void gc_config(void)
 118 {
 119         const char *value;
 120
 121         if (!git_config_get_value("gc.packrefs", &value)) {
 122                 if (value && !strcmp(value, "notbare"))
 123                         pack_refs = -1;
 124                 else
 125                         pack_refs = git_config_bool("gc.packrefs", value);
 126         }
 127
 128         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 129         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 130         git_config_get_int("gc.auto", &gc_auto_threshold);
 131         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 132         git_config_get_bool("gc.autodetach", &detach_auto);
 133         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 134         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 135         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 136
 137         git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold);
 138         git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size);
 139
 140         git_config(git_default_config, NULL);
 141 }
 142
 143 static int too_many_loose_objects(void)
 144 {
 145         /*
 146          * Quickly check if a "gc" is needed, by estimating how
 147          * many loose objects there are.  Because SHA-1 is evenly
 148          * distributed, we can check only one and get a reasonable
 149          * estimate.
 150          */
 151         DIR *dir;
 152         struct dirent *ent;
 153         int auto_threshold;
 154         int num_loose = 0;
 155         int needed = 0;
 156
 157         if (gc_auto_threshold <= 0)
 158                 return 0;
 159
 160         dir = opendir(git_path("objects/17"));
 161         if (!dir)
 162                 return 0;
 163
 164         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
 165         while ((ent = readdir(dir)) != NULL) {
 166                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 167                     ent->d_name[38] != '\0')
 168                         continue;
 169                 if (++num_loose > auto_threshold) {
 170                         needed = 1;
 171                         break;
 172                 }
 173         }
 174         closedir(dir);
 175         return needed;
 176 }
 177
 178 static struct packed_git *find_base_packs(struct string_list *packs,
 179                                           unsigned long limit)
 180 {
 181         struct packed_git *p, *base = NULL;
 182
 183         for (p = get_packed_git(the_repository); p; p = p->next) {
 184                 if (!p->pack_local)
 185                         continue;
 186                 if (limit) {
 187                         if (p->pack_size >= limit)
 188                                 string_list_append(packs, p->pack_name);
 189                 } else if (!base || base->pack_size < p->pack_size) {
 190                         base = p;
 191                 }
 192         }
 193
 194         if (base)
 195                 string_list_append(packs, base->pack_name);
 196
 197         return base;
 198 }
 199
 200 static int too_many_packs(void)
 201 {
 202         struct packed_git *p;
 203         int cnt;
 204
 205         if (gc_auto_pack_limit <= 0)
 206                 return 0;
 207
 208         for (cnt = 0, p = get_packed_git(the_repository); p; p = p->next) {
 209                 if (!p->pack_local)
 210                         continue;
 211                 if (p->pack_keep)
 212                         continue;
 213                 /*
 214                  * Perhaps check the size of the pack and count only
 215                  * very small ones here?
 216                  */
 217                 cnt++;
 218         }
 219         return gc_auto_pack_limit < cnt;
 220 }
 221
 222 static uint64_t total_ram(void)
 223 {
 224 #if defined(HAVE_SYSINFO)
 225         struct sysinfo si;
 226
 227         if (!sysinfo(&si))
 228                 return si.totalram;
 229 #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM))
 230         int64_t physical_memory;
 231         int mib[2];
 232         size_t length;
 233
 234         mib[0] = CTL_HW;
 235 # if defined(HW_MEMSIZE)
 236         mib[1] = HW_MEMSIZE;
 237 # else
 238         mib[1] = HW_PHYSMEM;
 239 # endif
 240         length = sizeof(int64_t);
 241         if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
 242                 return physical_memory;
 243 #elif defined(GIT_WINDOWS_NATIVE)
 244         MEMORYSTATUSEX memInfo;
 245
 246         memInfo.dwLength = sizeof(MEMORYSTATUSEX);
 247         if (GlobalMemoryStatusEx(&memInfo))
 248                 return memInfo.ullTotalPhys;
 249 #endif
 250         return 0;
 251 }
 252
 253 static uint64_t estimate_repack_memory(struct packed_git *pack)
 254 {
 255         unsigned long nr_objects = approximate_object_count();
 256         size_t os_cache, heap;
 257
 258         if (!pack || !nr_objects)
 259                 return 0;
 260
 261         /*
 262          * First we have to scan through at least one pack.
 263          * Assume enough room in OS file cache to keep the entire pack
 264          * or we may accidentally evict data of other processes from
 265          * the cache.
 266          */
 267         os_cache = pack->pack_size + pack->index_size;
 268         /* then pack-objects needs lots more for book keeping */
 269         heap = sizeof(struct object_entry) * nr_objects;
 270         /*
 271          * internal rev-list --all --objects takes up some memory too,
 272          * let's say half of it is for blobs
 273          */
 274         heap += sizeof(struct blob) * nr_objects / 2;
 275         /*
 276          * and the other half is for trees (commits and tags are
 277          * usually insignificant)
 278          */
 279         heap += sizeof(struct tree) * nr_objects / 2;
 280         /* and then obj_hash[], underestimated in fact */
 281         heap += sizeof(struct object *) * nr_objects;
 282         /* revindex is used also */
 283         heap += sizeof(struct revindex_entry) * nr_objects;
 284         /*
 285          * read_sha1_file() (either at delta calculation phase, or
 286          * writing phase) also fills up the delta base cache
 287          */
 288         heap += delta_base_cache_limit;
 289         /* and of course pack-objects has its own delta cache */
 290         heap += max_delta_cache_size;
 291
 292         return os_cache + heap;
 293 }
 294
 295 static int keep_one_pack(struct string_list_item *item, void *data)
 296 {
 297         argv_array_pushf(&repack, "--keep-pack=%s", basename(item->string));
 298         return 0;
 299 }
 300
 301 static void add_repack_all_option(struct string_list *keep_pack)
 302 {
 303         if (prune_expire && !strcmp(prune_expire, "now"))
 304                 argv_array_push(&repack, "-a");
 305         else {
 306                 argv_array_push(&repack, "-A");
 307                 if (prune_expire)
 308                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 309         }
 310
 311         if (keep_pack)
 312                 for_each_string_list(keep_pack, keep_one_pack, NULL);
 313 }
 314
 315 static void add_repack_incremental_option(void)
 316 {
 317        argv_array_push(&repack, "--no-write-bitmap-index");
 318 }
 319
 320 static int need_to_gc(void)
 321 {
 322         /*
 323          * Setting gc.auto to 0 or negative can disable the
 324          * automatic gc.
 325          */
 326         if (gc_auto_threshold <= 0)
 327                 return 0;
 328
 329         /*
 330          * If there are too many loose objects, but not too many
 331          * packs, we run "repack -d -l".  If there are too many packs,
 332          * we run "repack -A -d -l".  Otherwise we tell the caller
 333          * there is no need.
 334          */
 335         if (too_many_packs()) {
 336                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 337
 338                 if (big_pack_threshold) {
 339                         find_base_packs(&keep_pack, big_pack_threshold);
 340                         if (keep_pack.nr >= gc_auto_pack_limit) {
 341                                 big_pack_threshold = 0;
 342                                 string_list_clear(&keep_pack, 0);
 343                                 find_base_packs(&keep_pack, 0);
 344                         }
 345                 } else {
 346                         struct packed_git *p = find_base_packs(&keep_pack, 0);
 347                         uint64_t mem_have, mem_want;
 348
 349                         mem_have = total_ram();
 350                         mem_want = estimate_repack_memory(p);
 351
 352                         /*
 353                          * Only allow 1/2 of memory for pack-objects, leave
 354                          * the rest for the OS and other processes in the
 355                          * system.
 356                          */
 357                         if (!mem_have || mem_want < mem_have / 2)
 358                                 string_list_clear(&keep_pack, 0);
 359                 }
 360
 361                 add_repack_all_option(&keep_pack);
 362                 string_list_clear(&keep_pack, 0);
 363         } else if (too_many_loose_objects())
 364                 add_repack_incremental_option();
 365         else
 366                 return 0;
 367
 368         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 369                 return 0;
 370         return 1;
 371 }
 372
 373 /* return NULL on success, else hostname running the gc */
 374 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 375 {
 376         struct lock_file lock = LOCK_INIT;
 377         char my_host[HOST_NAME_MAX + 1];
 378         struct strbuf sb = STRBUF_INIT;
 379         struct stat st;
 380         uintmax_t pid;
 381         FILE *fp;
 382         int fd;
 383         char *pidfile_path;
 384
 385         if (is_tempfile_active(pidfile))
 386                 /* already locked */
 387                 return NULL;
 388
 389         if (xgethostname(my_host, sizeof(my_host)))
 390                 xsnprintf(my_host, sizeof(my_host), "unknown");
 391
 392         pidfile_path = git_pathdup("gc.pid");
 393         fd = hold_lock_file_for_update(&lock, pidfile_path,
 394                                        LOCK_DIE_ON_ERROR);
 395         if (!force) {
 396                 static char locking_host[HOST_NAME_MAX + 1];
 397                 static char *scan_fmt;
 398                 int should_exit;
 399
 400                 if (!scan_fmt)
 401                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
 402                 fp = fopen(pidfile_path, "r");
 403                 memset(locking_host, 0, sizeof(locking_host));
 404                 should_exit =
 405                         fp != NULL &&
 406                         !fstat(fileno(fp), &st) &&
 407                         /*
 408                          * 12 hour limit is very generous as gc should
 409                          * never take that long. On the other hand we
 410                          * don't really need a strict limit here,
 411                          * running gc --auto one day late is not a big
 412                          * problem. --force can be used in manual gc
 413                          * after the user verifies that no gc is
 414                          * running.
 415                          */
 416                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 417                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 418                         /* be gentle to concurrent "gc" on remote hosts */
 419                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 420                 if (fp != NULL)
 421                         fclose(fp);
 422                 if (should_exit) {
 423                         if (fd >= 0)
 424                                 rollback_lock_file(&lock);
 425                         *ret_pid = pid;
 426                         free(pidfile_path);
 427                         return locking_host;
 428                 }
 429         }
 430
 431         strbuf_addf(&sb, "%"PRIuMAX" %s",
 432                     (uintmax_t) getpid(), my_host);
 433         write_in_full(fd, sb.buf, sb.len);
 434         strbuf_release(&sb);
 435         commit_lock_file(&lock);
 436         pidfile = register_tempfile(pidfile_path);
 437         free(pidfile_path);
 438         return NULL;
 439 }
 440
 441 static int report_last_gc_error(void)
 442 {
 443         struct strbuf sb = STRBUF_INIT;
 444         int ret = 0;
 445         struct stat st;
 446         char *gc_log_path = git_pathdup("gc.log");
 447
 448         if (stat(gc_log_path, &st)) {
 449                 if (errno == ENOENT)
 450                         goto done;
 451
 452                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 453                 goto done;
 454         }
 455
 456         if (st.st_mtime < gc_log_expire_time)
 457                 goto done;
 458
 459         ret = strbuf_read_file(&sb, gc_log_path, 0);
 460         if (ret > 0)
 461                 ret = error(_("The last gc run reported the following. "
 462                                "Please correct the root cause\n"
 463                                "and remove %s.\n"
 464                                "Automatic cleanup will not be performed "
 465                                "until the file is removed.\n\n"
 466                                "%s"),
 467                             gc_log_path, sb.buf);
 468         strbuf_release(&sb);
 469 done:
 470         free(gc_log_path);
 471         return ret;
 472 }
 473
 474 static int gc_before_repack(void)
 475 {
 476         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 477                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 478
 479         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 480                 return error(FAILED_RUN, reflog.argv[0]);
 481
 482         pack_refs = 0;
 483         prune_reflogs = 0;
 484         return 0;
 485 }
 486
 487 int cmd_gc(int argc, const char **argv, const char *prefix)
 488 {
 489         int aggressive = 0;
 490         int auto_gc = 0;
 491         int quiet = 0;
 492         int force = 0;
 493         const char *name;
 494         pid_t pid;
 495         int daemonized = 0;
 496         int keep_base_pack = -1;
 497         timestamp_t dummy;
 498
 499         struct option builtin_gc_options[] = {
 500                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 501                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 502                         N_("prune unreferenced objects"),
 503                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 504                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 505                 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
 506                            PARSE_OPT_NOCOMPLETE),
 507                 OPT_BOOL_F(0, "force", &force,
 508                            N_("force running gc even if there may be another gc running"),
 509                            PARSE_OPT_NOCOMPLETE),
 510                 OPT_BOOL(0, "keep-largest-pack", &keep_base_pack,
 511                          N_("repack all other packs except the largest pack")),
 512                 OPT_END()
 513         };
 514
 515         if (argc == 2 && !strcmp(argv[1], "-h"))
 516                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 517
 518         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 519         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 520         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 521         argv_array_pushl(&prune, "prune", "--expire", NULL);
 522         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 523         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 524
 525         /* default expiry time, overwritten in gc_config */
 526         gc_config();
 527         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 528                 die(_("failed to parse gc.logexpiry value %s"), gc_log_expire);
 529
 530         if (pack_refs < 0)
 531                 pack_refs = !is_bare_repository();
 532
 533         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 534                              builtin_gc_usage, 0);
 535         if (argc > 0)
 536                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 537
 538         if (prune_expire && parse_expiry_date(prune_expire, &dummy))
 539                 die(_("failed to parse prune expiry value %s"), prune_expire);
 540
 541         if (aggressive) {
 542                 argv_array_push(&repack, "-f");
 543                 if (aggressive_depth > 0)
 544                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 545                 if (aggressive_window > 0)
 546                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 547         }
 548         if (quiet)
 549                 argv_array_push(&repack, "-q");
 550
 551         if (auto_gc) {
 552                 /*
 553                  * Auto-gc should be least intrusive as possible.
 554                  */
 555                 if (!need_to_gc())
 556                         return 0;
 557                 if (!quiet) {
 558                         if (detach_auto)
 559                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 560                         else
 561                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 562                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 563                 }
 564                 if (detach_auto) {
 565                         if (report_last_gc_error())
 566                                 return -1;
 567
 568                         if (lock_repo_for_gc(force, &pid))
 569                                 return 0;
 570                         if (gc_before_repack())
 571                                 return -1;
 572                         delete_tempfile(&pidfile);
 573
 574                         /*
 575                          * failure to daemonize is ok, we'll continue
 576                          * in foreground
 577                          */
 578                         daemonized = !daemonize();
 579                 }
 580         } else {
 581                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
 582
 583                 if (keep_base_pack != -1) {
 584                         if (keep_base_pack)
 585                                 find_base_packs(&keep_pack, 0);
 586                 } else if (big_pack_threshold) {
 587                         find_base_packs(&keep_pack, big_pack_threshold);
 588                 }
 589
 590                 add_repack_all_option(&keep_pack);
 591                 string_list_clear(&keep_pack, 0);
 592         }
 593
 594         name = lock_repo_for_gc(force, &pid);
 595         if (name) {
 596                 if (auto_gc)
 597                         return 0; /* be quiet on --auto */
 598                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 599                     name, (uintmax_t)pid);
 600         }
 601
 602         if (daemonized) {
 603                 hold_lock_file_for_update(&log_lock,
 604                                           git_path("gc.log"),
 605                                           LOCK_DIE_ON_ERROR);
 606                 dup2(get_lock_file_fd(&log_lock), 2);
 607                 sigchain_push_common(process_log_file_on_signal);
 608                 atexit(process_log_file_at_exit);
 609         }
 610
 611         if (gc_before_repack())
 612                 return -1;
 613
 614         if (!repository_format_precious_objects) {
 615                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 616                         return error(FAILED_RUN, repack.argv[0]);
 617
 618                 if (prune_expire) {
 619                         argv_array_push(&prune, prune_expire);
 620                         if (quiet)
 621                                 argv_array_push(&prune, "--no-progress");
 622                         if (repository_format_partial_clone)
 623                                 argv_array_push(&prune,
 624                                                 "--exclude-promisor-objects");
 625                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 626                                 return error(FAILED_RUN, prune.argv[0]);
 627                 }
 628         }
 629
 630         if (prune_worktrees_expire) {
 631                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 632                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 633                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 634         }
 635
 636         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 637                 return error(FAILED_RUN, rerere.argv[0]);
 638
 639         report_garbage = report_pack_garbage;
 640         reprepare_packed_git(the_repository);
 641         if (pack_garbage.nr > 0)
 642                 clean_pack_garbage();
 643
 644         if (auto_gc && too_many_loose_objects())
 645                 warning(_("There are too many unreachable loose objects; "
 646                         "run 'git prune' to remove them."));
 647
 648         if (!daemonized)
 649                 unlink(git_path("gc.log"));
 650
 651         return 0;
 652 }