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 "tempfile.h"
  15 #include "lockfile.h"
  16 #include "parse-options.h"
  17 #include "run-command.h"
  18 #include "sigchain.h"
  19 #include "argv-array.h"
  20 #include "commit.h"
  21
  22 #define FAILED_RUN "failed to run %s"
  23
  24 static const char * const builtin_gc_usage[] = {
  25         N_("git gc [<options>]"),
  26         NULL
  27 };
  28
  29 static int pack_refs = 1;
  30 static int prune_reflogs = 1;
  31 static int aggressive_depth = 50;
  32 static int aggressive_window = 250;
  33 static int gc_auto_threshold = 6700;
  34 static int gc_auto_pack_limit = 50;
  35 static int detach_auto = 1;
  36 static unsigned long gc_log_expire_time;
  37 static const char *gc_log_expire = "1.day.ago";
  38 static const char *prune_expire = "2.weeks.ago";
  39 static const char *prune_worktrees_expire = "3.months.ago";
  40
  41 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  42 static struct argv_array reflog = ARGV_ARRAY_INIT;
  43 static struct argv_array repack = ARGV_ARRAY_INIT;
  44 static struct argv_array prune = ARGV_ARRAY_INIT;
  45 static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
  46 static struct argv_array rerere = ARGV_ARRAY_INIT;
  47
  48 static struct tempfile pidfile;
  49 static struct lock_file log_lock;
  50
  51 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  52
  53 static void clean_pack_garbage(void)
  54 {
  55         int i;
  56         for (i = 0; i < pack_garbage.nr; i++)
  57                 unlink_or_warn(pack_garbage.items[i].string);
  58         string_list_clear(&pack_garbage, 0);
  59 }
  60
  61 static void report_pack_garbage(unsigned seen_bits, const char *path)
  62 {
  63         if (seen_bits == PACKDIR_FILE_IDX)
  64                 string_list_append(&pack_garbage, path);
  65 }
  66
  67 static void process_log_file(void)
  68 {
  69         struct stat st;
  70         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  71                 /*
  72                  * Perhaps there was an i/o error or another
  73                  * unlikely situation.  Try to make a note of
  74                  * this in gc.log along with any existing
  75                  * messages.
  76                  */
  77                 int saved_errno = errno;
  78                 fprintf(stderr, _("Failed to fstat %s: %s"),
  79                         get_tempfile_path(&log_lock.tempfile),
  80                         strerror(saved_errno));
  81                 fflush(stderr);
  82                 commit_lock_file(&log_lock);
  83                 errno = saved_errno;
  84         } else if (st.st_size) {
  85                 /* There was some error recorded in the lock file */
  86                 commit_lock_file(&log_lock);
  87         } else {
  88                 /* No error, clean up any old gc.log */
  89                 unlink(git_path("gc.log"));
  90                 rollback_lock_file(&log_lock);
  91         }
  92 }
  93
  94 static void process_log_file_at_exit(void)
  95 {
  96         fflush(stderr);
  97         process_log_file();
  98 }
  99
 100 static void process_log_file_on_signal(int signo)
 101 {
 102         process_log_file();
 103         sigchain_pop(signo);
 104         raise(signo);
 105 }
 106
 107 static void gc_config(void)
 108 {
 109         const char *value;
 110
 111         if (!git_config_get_value("gc.packrefs", &value)) {
 112                 if (value && !strcmp(value, "notbare"))
 113                         pack_refs = -1;
 114                 else
 115                         pack_refs = git_config_bool("gc.packrefs", value);
 116         }
 117
 118         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 119         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 120         git_config_get_int("gc.auto", &gc_auto_threshold);
 121         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 122         git_config_get_bool("gc.autodetach", &detach_auto);
 123         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 124         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 125         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 126
 127         git_config(git_default_config, NULL);
 128 }
 129
 130 static int too_many_loose_objects(void)
 131 {
 132         /*
 133          * Quickly check if a "gc" is needed, by estimating how
 134          * many loose objects there are.  Because SHA-1 is evenly
 135          * distributed, we can check only one and get a reasonable
 136          * estimate.
 137          */
 138         DIR *dir;
 139         struct dirent *ent;
 140         int auto_threshold;
 141         int num_loose = 0;
 142         int needed = 0;
 143
 144         if (gc_auto_threshold <= 0)
 145                 return 0;
 146
 147         dir = opendir(git_path("objects/17"));
 148         if (!dir)
 149                 return 0;
 150
 151         auto_threshold = (gc_auto_threshold + 255) / 256;
 152         while ((ent = readdir(dir)) != NULL) {
 153                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 154                     ent->d_name[38] != '\0')
 155                         continue;
 156                 if (++num_loose > auto_threshold) {
 157                         needed = 1;
 158                         break;
 159                 }
 160         }
 161         closedir(dir);
 162         return needed;
 163 }
 164
 165 static int too_many_packs(void)
 166 {
 167         struct packed_git *p;
 168         int cnt;
 169
 170         if (gc_auto_pack_limit <= 0)
 171                 return 0;
 172
 173         prepare_packed_git();
 174         for (cnt = 0, p = packed_git; p; p = p->next) {
 175                 if (!p->pack_local)
 176                         continue;
 177                 if (p->pack_keep)
 178                         continue;
 179                 /*
 180                  * Perhaps check the size of the pack and count only
 181                  * very small ones here?
 182                  */
 183                 cnt++;
 184         }
 185         return gc_auto_pack_limit < cnt;
 186 }
 187
 188 static void add_repack_all_option(void)
 189 {
 190         if (prune_expire && !strcmp(prune_expire, "now"))
 191                 argv_array_push(&repack, "-a");
 192         else {
 193                 argv_array_push(&repack, "-A");
 194                 if (prune_expire)
 195                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 196         }
 197 }
 198
 199 static void add_repack_incremental_option(void)
 200 {
 201        argv_array_push(&repack, "--no-write-bitmap-index");
 202 }
 203
 204 static int need_to_gc(void)
 205 {
 206         /*
 207          * Setting gc.auto to 0 or negative can disable the
 208          * automatic gc.
 209          */
 210         if (gc_auto_threshold <= 0)
 211                 return 0;
 212
 213         /*
 214          * If there are too many loose objects, but not too many
 215          * packs, we run "repack -d -l".  If there are too many packs,
 216          * we run "repack -A -d -l".  Otherwise we tell the caller
 217          * there is no need.
 218          */
 219         if (too_many_packs())
 220                 add_repack_all_option();
 221         else if (too_many_loose_objects())
 222                 add_repack_incremental_option();
 223         else
 224                 return 0;
 225
 226         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 227                 return 0;
 228         return 1;
 229 }
 230
 231 /* return NULL on success, else hostname running the gc */
 232 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 233 {
 234         static struct lock_file lock;
 235         char my_host[HOST_NAME_MAX + 1];
 236         struct strbuf sb = STRBUF_INIT;
 237         struct stat st;
 238         uintmax_t pid;
 239         FILE *fp;
 240         int fd;
 241         char *pidfile_path;
 242
 243         if (is_tempfile_active(&pidfile))
 244                 /* already locked */
 245                 return NULL;
 246
 247         if (xgethostname(my_host, sizeof(my_host)))
 248                 xsnprintf(my_host, sizeof(my_host), "unknown");
 249
 250         pidfile_path = git_pathdup("gc.pid");
 251         fd = hold_lock_file_for_update(&lock, pidfile_path,
 252                                        LOCK_DIE_ON_ERROR);
 253         if (!force) {
 254                 static char locking_host[HOST_NAME_MAX + 1];
 255                 static char *scan_fmt;
 256                 int should_exit;
 257
 258                 if (!scan_fmt)
 259                         scan_fmt = xstrfmt("%s %%%dc", "%"SCNuMAX, HOST_NAME_MAX);
 260                 fp = fopen(pidfile_path, "r");
 261                 memset(locking_host, 0, sizeof(locking_host));
 262                 should_exit =
 263                         fp != NULL &&
 264                         !fstat(fileno(fp), &st) &&
 265                         /*
 266                          * 12 hour limit is very generous as gc should
 267                          * never take that long. On the other hand we
 268                          * don't really need a strict limit here,
 269                          * running gc --auto one day late is not a big
 270                          * problem. --force can be used in manual gc
 271                          * after the user verifies that no gc is
 272                          * running.
 273                          */
 274                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 275                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 276                         /* be gentle to concurrent "gc" on remote hosts */
 277                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 278                 if (fp != NULL)
 279                         fclose(fp);
 280                 if (should_exit) {
 281                         if (fd >= 0)
 282                                 rollback_lock_file(&lock);
 283                         *ret_pid = pid;
 284                         free(pidfile_path);
 285                         return locking_host;
 286                 }
 287         }
 288
 289         strbuf_addf(&sb, "%"PRIuMAX" %s",
 290                     (uintmax_t) getpid(), my_host);
 291         write_in_full(fd, sb.buf, sb.len);
 292         strbuf_release(&sb);
 293         commit_lock_file(&lock);
 294         register_tempfile(&pidfile, pidfile_path);
 295         free(pidfile_path);
 296         return NULL;
 297 }
 298
 299 static int report_last_gc_error(void)
 300 {
 301         struct strbuf sb = STRBUF_INIT;
 302         int ret = 0;
 303         struct stat st;
 304         char *gc_log_path = git_pathdup("gc.log");
 305
 306         if (stat(gc_log_path, &st)) {
 307                 if (errno == ENOENT)
 308                         goto done;
 309
 310                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 311                 goto done;
 312         }
 313
 314         if (st.st_mtime < gc_log_expire_time)
 315                 goto done;
 316
 317         ret = strbuf_read_file(&sb, gc_log_path, 0);
 318         if (ret > 0)
 319                 ret = error(_("The last gc run reported the following. "
 320                                "Please correct the root cause\n"
 321                                "and remove %s.\n"
 322                                "Automatic cleanup will not be performed "
 323                                "until the file is removed.\n\n"
 324                                "%s"),
 325                             gc_log_path, sb.buf);
 326         strbuf_release(&sb);
 327 done:
 328         free(gc_log_path);
 329         return ret;
 330 }
 331
 332 static int gc_before_repack(void)
 333 {
 334         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 335                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 336
 337         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 338                 return error(FAILED_RUN, reflog.argv[0]);
 339
 340         pack_refs = 0;
 341         prune_reflogs = 0;
 342         return 0;
 343 }
 344
 345 int cmd_gc(int argc, const char **argv, const char *prefix)
 346 {
 347         int aggressive = 0;
 348         int auto_gc = 0;
 349         int quiet = 0;
 350         int force = 0;
 351         const char *name;
 352         pid_t pid;
 353         int daemonized = 0;
 354
 355         struct option builtin_gc_options[] = {
 356                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 357                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 358                         N_("prune unreferenced objects"),
 359                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 360                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 361                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 362                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 363                 OPT_END()
 364         };
 365
 366         if (argc == 2 && !strcmp(argv[1], "-h"))
 367                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 368
 369         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 370         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 371         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 372         argv_array_pushl(&prune, "prune", "--expire", NULL);
 373         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 374         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 375
 376         /* default expiry time, overwritten in gc_config */
 377         gc_config();
 378         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 379                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
 380
 381         if (pack_refs < 0)
 382                 pack_refs = !is_bare_repository();
 383
 384         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 385                              builtin_gc_usage, 0);
 386         if (argc > 0)
 387                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 388
 389         if (aggressive) {
 390                 argv_array_push(&repack, "-f");
 391                 if (aggressive_depth > 0)
 392                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 393                 if (aggressive_window > 0)
 394                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 395         }
 396         if (quiet)
 397                 argv_array_push(&repack, "-q");
 398
 399         if (auto_gc) {
 400                 /*
 401                  * Auto-gc should be least intrusive as possible.
 402                  */
 403                 if (!need_to_gc())
 404                         return 0;
 405                 if (!quiet) {
 406                         if (detach_auto)
 407                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 408                         else
 409                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 410                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 411                 }
 412                 if (detach_auto) {
 413                         if (report_last_gc_error())
 414                                 return -1;
 415
 416                         if (gc_before_repack())
 417                                 return -1;
 418                         /*
 419                          * failure to daemonize is ok, we'll continue
 420                          * in foreground
 421                          */
 422                         daemonized = !daemonize();
 423                 }
 424         } else
 425                 add_repack_all_option();
 426
 427         name = lock_repo_for_gc(force, &pid);
 428         if (name) {
 429                 if (auto_gc)
 430                         return 0; /* be quiet on --auto */
 431                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 432                     name, (uintmax_t)pid);
 433         }
 434
 435         if (daemonized) {
 436                 hold_lock_file_for_update(&log_lock,
 437                                           git_path("gc.log"),
 438                                           LOCK_DIE_ON_ERROR);
 439                 dup2(get_lock_file_fd(&log_lock), 2);
 440                 sigchain_push_common(process_log_file_on_signal);
 441                 atexit(process_log_file_at_exit);
 442         }
 443
 444         if (gc_before_repack())
 445                 return -1;
 446
 447         if (!repository_format_precious_objects) {
 448                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 449                         return error(FAILED_RUN, repack.argv[0]);
 450
 451                 if (prune_expire) {
 452                         argv_array_push(&prune, prune_expire);
 453                         if (quiet)
 454                                 argv_array_push(&prune, "--no-progress");
 455                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 456                                 return error(FAILED_RUN, prune.argv[0]);
 457                 }
 458         }
 459
 460         if (prune_worktrees_expire) {
 461                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 462                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 463                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 464         }
 465
 466         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 467                 return error(FAILED_RUN, rerere.argv[0]);
 468
 469         report_garbage = report_pack_garbage;
 470         reprepare_packed_git();
 471         if (pack_garbage.nr > 0)
 472                 clean_pack_garbage();
 473
 474         if (auto_gc && too_many_loose_objects())
 475                 warning(_("There are too many unreachable loose objects; "
 476                         "run 'git prune' to remove them."));
 477
 478         if (!daemonized)
 479                 unlink(git_path("gc.log"));
 480
 481         return 0;
 482 }