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