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
  26 #define FAILED_RUN "failed to run %s"
  27
  28 static const char * const builtin_gc_usage[] = {
  29         N_("git gc [<options>]"),
  30         NULL
  31 };
  32
  33 static int pack_refs = 1;
  34 static int prune_reflogs = 1;
  35 static int aggressive_depth = 50;
  36 static int aggressive_window = 250;
  37 static int gc_auto_threshold = 6700;
  38 static int gc_auto_pack_limit = 50;
  39 static int detach_auto = 1;
  40 static timestamp_t gc_log_expire_time;
  41 static const char *gc_log_expire = "1.day.ago";
  42 static const char *prune_expire = "2.weeks.ago";
  43 static const char *prune_worktrees_expire = "3.months.ago";
  44
  45 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  46 static struct argv_array reflog = ARGV_ARRAY_INIT;
  47 static struct argv_array repack = ARGV_ARRAY_INIT;
  48 static struct argv_array prune = ARGV_ARRAY_INIT;
  49 static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
  50 static struct argv_array rerere = ARGV_ARRAY_INIT;
  51
  52 static struct tempfile *pidfile;
  53 static struct lock_file log_lock;
  54
  55 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  56
  57 static void clean_pack_garbage(void)
  58 {
  59         int i;
  60         for (i = 0; i < pack_garbage.nr; i++)
  61                 unlink_or_warn(pack_garbage.items[i].string);
  62         string_list_clear(&pack_garbage, 0);
  63 }
  64
  65 static void report_pack_garbage(unsigned seen_bits, const char *path)
  66 {
  67         if (seen_bits == PACKDIR_FILE_IDX)
  68                 string_list_append(&pack_garbage, path);
  69 }
  70
  71 static void process_log_file(void)
  72 {
  73         struct stat st;
  74         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  75                 /*
  76                  * Perhaps there was an i/o error or another
  77                  * unlikely situation.  Try to make a note of
  78                  * this in gc.log along with any existing
  79                  * messages.
  80                  */
  81                 int saved_errno = errno;
  82                 fprintf(stderr, _("Failed to fstat %s: %s"),
  83                         get_tempfile_path(log_lock.tempfile),
  84                         strerror(saved_errno));
  85                 fflush(stderr);
  86                 commit_lock_file(&log_lock);
  87                 errno = saved_errno;
  88         } else if (st.st_size) {
  89                 /* There was some error recorded in the lock file */
  90                 commit_lock_file(&log_lock);
  91         } else {
  92                 /* No error, clean up any old gc.log */
  93                 unlink(git_path("gc.log"));
  94                 rollback_lock_file(&log_lock);
  95         }
  96 }
  97
  98 static void process_log_file_at_exit(void)
  99 {
 100         fflush(stderr);
 101         process_log_file();
 102 }
 103
 104 static void process_log_file_on_signal(int signo)
 105 {
 106         process_log_file();
 107         sigchain_pop(signo);
 108         raise(signo);
 109 }
 110
 111 static void gc_config(void)
 112 {
 113         const char *value;
 114
 115         if (!git_config_get_value("gc.packrefs", &value)) {
 116                 if (value && !strcmp(value, "notbare"))
 117                         pack_refs = -1;
 118                 else
 119                         pack_refs = git_config_bool("gc.packrefs", value);
 120         }
 121
 122         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 123         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 124         git_config_get_int("gc.auto", &gc_auto_threshold);
 125         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 126         git_config_get_bool("gc.autodetach", &detach_auto);
 127         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 128         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 129         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 130
 131         git_config(git_default_config, NULL);
 132 }
 133
 134 static int too_many_loose_objects(void)
 135 {
 136         /*
 137          * Quickly check if a "gc" is needed, by estimating how
 138          * many loose objects there are.  Because SHA-1 is evenly
 139          * distributed, we can check only one and get a reasonable
 140          * estimate.
 141          */
 142         DIR *dir;
 143         struct dirent *ent;
 144         int auto_threshold;
 145         int num_loose = 0;
 146         int needed = 0;
 147
 148         if (gc_auto_threshold <= 0)
 149                 return 0;
 150
 151         dir = opendir(git_path("objects/17"));
 152         if (!dir)
 153                 return 0;
 154
 155         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
 156         while ((ent = readdir(dir)) != NULL) {
 157                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 158                     ent->d_name[38] != '\0')
 159                         continue;
 160                 if (++num_loose > auto_threshold) {
 161                         needed = 1;
 162                         break;
 163                 }
 164         }
 165         closedir(dir);
 166         return needed;
 167 }
 168
 169 static int too_many_packs(void)
 170 {
 171         struct packed_git *p;
 172         int cnt;
 173
 174         if (gc_auto_pack_limit <= 0)
 175                 return 0;
 176
 177         for (cnt = 0, p = get_packed_git(the_repository); p; p = p->next) {
 178                 if (!p->pack_local)
 179                         continue;
 180                 if (p->pack_keep)
 181                         continue;
 182                 /*
 183                  * Perhaps check the size of the pack and count only
 184                  * very small ones here?
 185                  */
 186                 cnt++;
 187         }
 188         return gc_auto_pack_limit < cnt;
 189 }
 190
 191 static void add_repack_all_option(void)
 192 {
 193         if (prune_expire && !strcmp(prune_expire, "now"))
 194                 argv_array_push(&repack, "-a");
 195         else {
 196                 argv_array_push(&repack, "-A");
 197                 if (prune_expire)
 198                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 199         }
 200 }
 201
 202 static void add_repack_incremental_option(void)
 203 {
 204        argv_array_push(&repack, "--no-write-bitmap-index");
 205 }
 206
 207 static int need_to_gc(void)
 208 {
 209         /*
 210          * Setting gc.auto to 0 or negative can disable the
 211          * automatic gc.
 212          */
 213         if (gc_auto_threshold <= 0)
 214                 return 0;
 215
 216         /*
 217          * If there are too many loose objects, but not too many
 218          * packs, we run "repack -d -l".  If there are too many packs,
 219          * we run "repack -A -d -l".  Otherwise we tell the caller
 220          * there is no need.
 221          */
 222         if (too_many_packs())
 223                 add_repack_all_option();
 224         else if (too_many_loose_objects())
 225                 add_repack_incremental_option();
 226         else
 227                 return 0;
 228
 229         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 230                 return 0;
 231         return 1;
 232 }
 233
 234 /* return NULL on success, else hostname running the gc */
 235 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 236 {
 237         static struct lock_file lock;
 238         char my_host[HOST_NAME_MAX + 1];
 239         struct strbuf sb = STRBUF_INIT;
 240         struct stat st;
 241         uintmax_t pid;
 242         FILE *fp;
 243         int fd;
 244         char *pidfile_path;
 245
 246         if (is_tempfile_active(pidfile))
 247                 /* already locked */
 248                 return NULL;
 249
 250         if (xgethostname(my_host, sizeof(my_host)))
 251                 xsnprintf(my_host, sizeof(my_host), "unknown");
 252
 253         pidfile_path = git_pathdup("gc.pid");
 254         fd = hold_lock_file_for_update(&lock, pidfile_path,
 255                                        LOCK_DIE_ON_ERROR);
 256         if (!force) {
 257                 static char locking_host[HOST_NAME_MAX + 1];
 258                 static char *scan_fmt;
 259                 int should_exit;
 260
 261                 if (!scan_fmt)
 262                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
 263                 fp = fopen(pidfile_path, "r");
 264                 memset(locking_host, 0, sizeof(locking_host));
 265                 should_exit =
 266                         fp != NULL &&
 267                         !fstat(fileno(fp), &st) &&
 268                         /*
 269                          * 12 hour limit is very generous as gc should
 270                          * never take that long. On the other hand we
 271                          * don't really need a strict limit here,
 272                          * running gc --auto one day late is not a big
 273                          * problem. --force can be used in manual gc
 274                          * after the user verifies that no gc is
 275                          * running.
 276                          */
 277                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 278                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 279                         /* be gentle to concurrent "gc" on remote hosts */
 280                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 281                 if (fp != NULL)
 282                         fclose(fp);
 283                 if (should_exit) {
 284                         if (fd >= 0)
 285                                 rollback_lock_file(&lock);
 286                         *ret_pid = pid;
 287                         free(pidfile_path);
 288                         return locking_host;
 289                 }
 290         }
 291
 292         strbuf_addf(&sb, "%"PRIuMAX" %s",
 293                     (uintmax_t) getpid(), my_host);
 294         write_in_full(fd, sb.buf, sb.len);
 295         strbuf_release(&sb);
 296         commit_lock_file(&lock);
 297         pidfile = register_tempfile(pidfile_path);
 298         free(pidfile_path);
 299         return NULL;
 300 }
 301
 302 static int report_last_gc_error(void)
 303 {
 304         struct strbuf sb = STRBUF_INIT;
 305         int ret = 0;
 306         struct stat st;
 307         char *gc_log_path = git_pathdup("gc.log");
 308
 309         if (stat(gc_log_path, &st)) {
 310                 if (errno == ENOENT)
 311                         goto done;
 312
 313                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 314                 goto done;
 315         }
 316
 317         if (st.st_mtime < gc_log_expire_time)
 318                 goto done;
 319
 320         ret = strbuf_read_file(&sb, gc_log_path, 0);
 321         if (ret > 0)
 322                 ret = error(_("The last gc run reported the following. "
 323                                "Please correct the root cause\n"
 324                                "and remove %s.\n"
 325                                "Automatic cleanup will not be performed "
 326                                "until the file is removed.\n\n"
 327                                "%s"),
 328                             gc_log_path, sb.buf);
 329         strbuf_release(&sb);
 330 done:
 331         free(gc_log_path);
 332         return ret;
 333 }
 334
 335 static int gc_before_repack(void)
 336 {
 337         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 338                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 339
 340         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 341                 return error(FAILED_RUN, reflog.argv[0]);
 342
 343         pack_refs = 0;
 344         prune_reflogs = 0;
 345         return 0;
 346 }
 347
 348 int cmd_gc(int argc, const char **argv, const char *prefix)
 349 {
 350         int aggressive = 0;
 351         int auto_gc = 0;
 352         int quiet = 0;
 353         int force = 0;
 354         const char *name;
 355         pid_t pid;
 356         int daemonized = 0;
 357
 358         struct option builtin_gc_options[] = {
 359                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 360                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 361                         N_("prune unreferenced objects"),
 362                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 363                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 364                 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
 365                            PARSE_OPT_NOCOMPLETE),
 366                 OPT_BOOL_F(0, "force", &force,
 367                            N_("force running gc even if there may be another gc running"),
 368                            PARSE_OPT_NOCOMPLETE),
 369                 OPT_END()
 370         };
 371
 372         if (argc == 2 && !strcmp(argv[1], "-h"))
 373                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 374
 375         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 376         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 377         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 378         argv_array_pushl(&prune, "prune", "--expire", NULL);
 379         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 380         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 381
 382         /* default expiry time, overwritten in gc_config */
 383         gc_config();
 384         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 385                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
 386
 387         if (pack_refs < 0)
 388                 pack_refs = !is_bare_repository();
 389
 390         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 391                              builtin_gc_usage, 0);
 392         if (argc > 0)
 393                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 394
 395         if (aggressive) {
 396                 argv_array_push(&repack, "-f");
 397                 if (aggressive_depth > 0)
 398                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 399                 if (aggressive_window > 0)
 400                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 401         }
 402         if (quiet)
 403                 argv_array_push(&repack, "-q");
 404
 405         if (auto_gc) {
 406                 /*
 407                  * Auto-gc should be least intrusive as possible.
 408                  */
 409                 if (!need_to_gc())
 410                         return 0;
 411                 if (!quiet) {
 412                         if (detach_auto)
 413                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 414                         else
 415                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 416                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 417                 }
 418                 if (detach_auto) {
 419                         if (report_last_gc_error())
 420                                 return -1;
 421
 422                         if (lock_repo_for_gc(force, &pid))
 423                                 return 0;
 424                         if (gc_before_repack())
 425                                 return -1;
 426                         delete_tempfile(&pidfile);
 427
 428                         /*
 429                          * failure to daemonize is ok, we'll continue
 430                          * in foreground
 431                          */
 432                         daemonized = !daemonize();
 433                 }
 434         } else
 435                 add_repack_all_option();
 436
 437         name = lock_repo_for_gc(force, &pid);
 438         if (name) {
 439                 if (auto_gc)
 440                         return 0; /* be quiet on --auto */
 441                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 442                     name, (uintmax_t)pid);
 443         }
 444
 445         if (daemonized) {
 446                 hold_lock_file_for_update(&log_lock,
 447                                           git_path("gc.log"),
 448                                           LOCK_DIE_ON_ERROR);
 449                 dup2(get_lock_file_fd(&log_lock), 2);
 450                 sigchain_push_common(process_log_file_on_signal);
 451                 atexit(process_log_file_at_exit);
 452         }
 453
 454         if (gc_before_repack())
 455                 return -1;
 456
 457         if (!repository_format_precious_objects) {
 458                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 459                         return error(FAILED_RUN, repack.argv[0]);
 460
 461                 if (prune_expire) {
 462                         argv_array_push(&prune, prune_expire);
 463                         if (quiet)
 464                                 argv_array_push(&prune, "--no-progress");
 465                         if (repository_format_partial_clone)
 466                                 argv_array_push(&prune,
 467                                                 "--exclude-promisor-objects");
 468                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 469                                 return error(FAILED_RUN, prune.argv[0]);
 470                 }
 471         }
 472
 473         if (prune_worktrees_expire) {
 474                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 475                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 476                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 477         }
 478
 479         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 480                 return error(FAILED_RUN, rerere.argv[0]);
 481
 482         report_garbage = report_pack_garbage;
 483         reprepare_packed_git(the_repository);
 484         if (pack_garbage.nr > 0)
 485                 clean_pack_garbage();
 486
 487         if (auto_gc && too_many_loose_objects())
 488                 warning(_("There are too many unreachable loose objects; "
 489                         "run 'git prune' to remove them."));
 490
 491         if (!daemonized)
 492                 unlink(git_path("gc.log"));
 493
 494         return 0;
 495 }