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 "config.h"
  15 #include "tempfile.h"
  16 #include "lockfile.h"
  17 #include "parse-options.h"
  18 #include "run-command.h"
  19 #include "sigchain.h"
  20 #include "argv-array.h"
  21 #include "commit.h"
  22
  23 #define FAILED_RUN "failed to run %s"
  24
  25 static const char * const builtin_gc_usage[] = {
  26         N_("git gc [<options>]"),
  27         NULL
  28 };
  29
  30 static int pack_refs = 1;
  31 static int prune_reflogs = 1;
  32 static int aggressive_depth = 50;
  33 static int aggressive_window = 250;
  34 static int gc_auto_threshold = 6700;
  35 static int gc_auto_pack_limit = 50;
  36 static int detach_auto = 1;
  37 static timestamp_t gc_log_expire_time;
  38 static const char *gc_log_expire = "1.day.ago";
  39 static const char *prune_expire = "2.weeks.ago";
  40 static const char *prune_worktrees_expire = "3.months.ago";
  41
  42 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  43 static struct argv_array reflog = ARGV_ARRAY_INIT;
  44 static struct argv_array repack = ARGV_ARRAY_INIT;
  45 static struct argv_array prune = ARGV_ARRAY_INIT;
  46 static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
  47 static struct argv_array rerere = ARGV_ARRAY_INIT;
  48
  49 static struct tempfile pidfile;
  50 static struct lock_file log_lock;
  51
  52 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  53
  54 static void clean_pack_garbage(void)
  55 {
  56         int i;
  57         for (i = 0; i < pack_garbage.nr; i++)
  58                 unlink_or_warn(pack_garbage.items[i].string);
  59         string_list_clear(&pack_garbage, 0);
  60 }
  61
  62 static void report_pack_garbage(unsigned seen_bits, const char *path)
  63 {
  64         if (seen_bits == PACKDIR_FILE_IDX)
  65                 string_list_append(&pack_garbage, path);
  66 }
  67
  68 static void process_log_file(void)
  69 {
  70         struct stat st;
  71         if (fstat(get_lock_file_fd(&log_lock), &st)) {
  72                 /*
  73                  * Perhaps there was an i/o error or another
  74                  * unlikely situation.  Try to make a note of
  75                  * this in gc.log along with any existing
  76                  * messages.
  77                  */
  78                 int saved_errno = errno;
  79                 fprintf(stderr, _("Failed to fstat %s: %s"),
  80                         get_tempfile_path(&log_lock.tempfile),
  81                         strerror(saved_errno));
  82                 fflush(stderr);
  83                 commit_lock_file(&log_lock);
  84                 errno = saved_errno;
  85         } else if (st.st_size) {
  86                 /* There was some error recorded in the lock file */
  87                 commit_lock_file(&log_lock);
  88         } else {
  89                 /* No error, clean up any old gc.log */
  90                 unlink(git_path("gc.log"));
  91                 rollback_lock_file(&log_lock);
  92         }
  93 }
  94
  95 static void process_log_file_at_exit(void)
  96 {
  97         fflush(stderr);
  98         process_log_file();
  99 }
 100
 101 static void process_log_file_on_signal(int signo)
 102 {
 103         process_log_file();
 104         sigchain_pop(signo);
 105         raise(signo);
 106 }
 107
 108 static void gc_config(void)
 109 {
 110         const char *value;
 111
 112         if (!git_config_get_value("gc.packrefs", &value)) {
 113                 if (value && !strcmp(value, "notbare"))
 114                         pack_refs = -1;
 115                 else
 116                         pack_refs = git_config_bool("gc.packrefs", value);
 117         }
 118
 119         git_config_get_int("gc.aggressivewindow", &aggressive_window);
 120         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 121         git_config_get_int("gc.auto", &gc_auto_threshold);
 122         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
 123         git_config_get_bool("gc.autodetach", &detach_auto);
 124         git_config_get_expiry("gc.pruneexpire", &prune_expire);
 125         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
 126         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
 127
 128         git_config(git_default_config, NULL);
 129 }
 130
 131 static int too_many_loose_objects(void)
 132 {
 133         /*
 134          * Quickly check if a "gc" is needed, by estimating how
 135          * many loose objects there are.  Because SHA-1 is evenly
 136          * distributed, we can check only one and get a reasonable
 137          * estimate.
 138          */
 139         DIR *dir;
 140         struct dirent *ent;
 141         int auto_threshold;
 142         int num_loose = 0;
 143         int needed = 0;
 144
 145         if (gc_auto_threshold <= 0)
 146                 return 0;
 147
 148         dir = opendir(git_path("objects/17"));
 149         if (!dir)
 150                 return 0;
 151
 152         auto_threshold = (gc_auto_threshold + 255) / 256;
 153         while ((ent = readdir(dir)) != NULL) {
 154                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 155                     ent->d_name[38] != '\0')
 156                         continue;
 157                 if (++num_loose > auto_threshold) {
 158                         needed = 1;
 159                         break;
 160                 }
 161         }
 162         closedir(dir);
 163         return needed;
 164 }
 165
 166 static int too_many_packs(void)
 167 {
 168         struct packed_git *p;
 169         int cnt;
 170
 171         if (gc_auto_pack_limit <= 0)
 172                 return 0;
 173
 174         prepare_packed_git();
 175         for (cnt = 0, p = packed_git; p; p = p->next) {
 176                 if (!p->pack_local)
 177                         continue;
 178                 if (p->pack_keep)
 179                         continue;
 180                 /*
 181                  * Perhaps check the size of the pack and count only
 182                  * very small ones here?
 183                  */
 184                 cnt++;
 185         }
 186         return gc_auto_pack_limit < cnt;
 187 }
 188
 189 static void add_repack_all_option(void)
 190 {
 191         if (prune_expire && !strcmp(prune_expire, "now"))
 192                 argv_array_push(&repack, "-a");
 193         else {
 194                 argv_array_push(&repack, "-A");
 195                 if (prune_expire)
 196                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 197         }
 198 }
 199
 200 static void add_repack_incremental_option(void)
 201 {
 202        argv_array_push(&repack, "--no-write-bitmap-index");
 203 }
 204
 205 static int need_to_gc(void)
 206 {
 207         /*
 208          * Setting gc.auto to 0 or negative can disable the
 209          * automatic gc.
 210          */
 211         if (gc_auto_threshold <= 0)
 212                 return 0;
 213
 214         /*
 215          * If there are too many loose objects, but not too many
 216          * packs, we run "repack -d -l".  If there are too many packs,
 217          * we run "repack -A -d -l".  Otherwise we tell the caller
 218          * there is no need.
 219          */
 220         if (too_many_packs())
 221                 add_repack_all_option();
 222         else if (too_many_loose_objects())
 223                 add_repack_incremental_option();
 224         else
 225                 return 0;
 226
 227         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 228                 return 0;
 229         return 1;
 230 }
 231
 232 /* return NULL on success, else hostname running the gc */
 233 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 234 {
 235         static struct lock_file lock;
 236         char my_host[HOST_NAME_MAX + 1];
 237         struct strbuf sb = STRBUF_INIT;
 238         struct stat st;
 239         uintmax_t pid;
 240         FILE *fp;
 241         int fd;
 242         char *pidfile_path;
 243
 244         if (is_tempfile_active(&pidfile))
 245                 /* already locked */
 246                 return NULL;
 247
 248         if (xgethostname(my_host, sizeof(my_host)))
 249                 xsnprintf(my_host, sizeof(my_host), "unknown");
 250
 251         pidfile_path = git_pathdup("gc.pid");
 252         fd = hold_lock_file_for_update(&lock, pidfile_path,
 253                                        LOCK_DIE_ON_ERROR);
 254         if (!force) {
 255                 static char locking_host[HOST_NAME_MAX + 1];
 256                 static char *scan_fmt;
 257                 int should_exit;
 258
 259                 if (!scan_fmt)
 260                         scan_fmt = xstrfmt("%s %%%dc", "%"SCNuMAX, HOST_NAME_MAX);
 261                 fp = fopen(pidfile_path, "r");
 262                 memset(locking_host, 0, sizeof(locking_host));
 263                 should_exit =
 264                         fp != NULL &&
 265                         !fstat(fileno(fp), &st) &&
 266                         /*
 267                          * 12 hour limit is very generous as gc should
 268                          * never take that long. On the other hand we
 269                          * don't really need a strict limit here,
 270                          * running gc --auto one day late is not a big
 271                          * problem. --force can be used in manual gc
 272                          * after the user verifies that no gc is
 273                          * running.
 274                          */
 275                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 276                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 277                         /* be gentle to concurrent "gc" on remote hosts */
 278                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 279                 if (fp != NULL)
 280                         fclose(fp);
 281                 if (should_exit) {
 282                         if (fd >= 0)
 283                                 rollback_lock_file(&lock);
 284                         *ret_pid = pid;
 285                         free(pidfile_path);
 286                         return locking_host;
 287                 }
 288         }
 289
 290         strbuf_addf(&sb, "%"PRIuMAX" %s",
 291                     (uintmax_t) getpid(), my_host);
 292         write_in_full(fd, sb.buf, sb.len);
 293         strbuf_release(&sb);
 294         commit_lock_file(&lock);
 295         register_tempfile(&pidfile, pidfile_path);
 296         free(pidfile_path);
 297         return NULL;
 298 }
 299
 300 static int report_last_gc_error(void)
 301 {
 302         struct strbuf sb = STRBUF_INIT;
 303         int ret = 0;
 304         struct stat st;
 305         char *gc_log_path = git_pathdup("gc.log");
 306
 307         if (stat(gc_log_path, &st)) {
 308                 if (errno == ENOENT)
 309                         goto done;
 310
 311                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 312                 goto done;
 313         }
 314
 315         if (st.st_mtime < gc_log_expire_time)
 316                 goto done;
 317
 318         ret = strbuf_read_file(&sb, gc_log_path, 0);
 319         if (ret > 0)
 320                 ret = error(_("The last gc run reported the following. "
 321                                "Please correct the root cause\n"
 322                                "and remove %s.\n"
 323                                "Automatic cleanup will not be performed "
 324                                "until the file is removed.\n\n"
 325                                "%s"),
 326                             gc_log_path, sb.buf);
 327         strbuf_release(&sb);
 328 done:
 329         free(gc_log_path);
 330         return ret;
 331 }
 332
 333 static int gc_before_repack(void)
 334 {
 335         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 336                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 337
 338         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 339                 return error(FAILED_RUN, reflog.argv[0]);
 340
 341         pack_refs = 0;
 342         prune_reflogs = 0;
 343         return 0;
 344 }
 345
 346 int cmd_gc(int argc, const char **argv, const char *prefix)
 347 {
 348         int aggressive = 0;
 349         int auto_gc = 0;
 350         int quiet = 0;
 351         int force = 0;
 352         const char *name;
 353         pid_t pid;
 354         int daemonized = 0;
 355
 356         struct option builtin_gc_options[] = {
 357                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 358                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 359                         N_("prune unreferenced objects"),
 360                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 361                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 362                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 363                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 364                 OPT_END()
 365         };
 366
 367         if (argc == 2 && !strcmp(argv[1], "-h"))
 368                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 369
 370         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 371         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 372         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 373         argv_array_pushl(&prune, "prune", "--expire", NULL);
 374         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 375         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 376
 377         /* default expiry time, overwritten in gc_config */
 378         gc_config();
 379         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 380                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
 381
 382         if (pack_refs < 0)
 383                 pack_refs = !is_bare_repository();
 384
 385         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 386                              builtin_gc_usage, 0);
 387         if (argc > 0)
 388                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 389
 390         if (aggressive) {
 391                 argv_array_push(&repack, "-f");
 392                 if (aggressive_depth > 0)
 393                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 394                 if (aggressive_window > 0)
 395                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 396         }
 397         if (quiet)
 398                 argv_array_push(&repack, "-q");
 399
 400         if (auto_gc) {
 401                 /*
 402                  * Auto-gc should be least intrusive as possible.
 403                  */
 404                 if (!need_to_gc())
 405                         return 0;
 406                 if (!quiet) {
 407                         if (detach_auto)
 408                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 409                         else
 410                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 411                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 412                 }
 413                 if (detach_auto) {
 414                         if (report_last_gc_error())
 415                                 return -1;
 416
 417                         if (gc_before_repack())
 418                                 return -1;
 419                         /*
 420                          * failure to daemonize is ok, we'll continue
 421                          * in foreground
 422                          */
 423                         daemonized = !daemonize();
 424                 }
 425         } else
 426                 add_repack_all_option();
 427
 428         name = lock_repo_for_gc(force, &pid);
 429         if (name) {
 430                 if (auto_gc)
 431                         return 0; /* be quiet on --auto */
 432                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 433                     name, (uintmax_t)pid);
 434         }
 435
 436         if (daemonized) {
 437                 hold_lock_file_for_update(&log_lock,
 438                                           git_path("gc.log"),
 439                                           LOCK_DIE_ON_ERROR);
 440                 dup2(get_lock_file_fd(&log_lock), 2);
 441                 sigchain_push_common(process_log_file_on_signal);
 442                 atexit(process_log_file_at_exit);
 443         }
 444
 445         if (gc_before_repack())
 446                 return -1;
 447
 448         if (!repository_format_precious_objects) {
 449                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 450                         return error(FAILED_RUN, repack.argv[0]);
 451
 452                 if (prune_expire) {
 453                         argv_array_push(&prune, prune_expire);
 454                         if (quiet)
 455                                 argv_array_push(&prune, "--no-progress");
 456                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 457                                 return error(FAILED_RUN, prune.argv[0]);
 458                 }
 459         }
 460
 461         if (prune_worktrees_expire) {
 462                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 463                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 464                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 465         }
 466
 467         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 468                 return error(FAILED_RUN, rerere.argv[0]);
 469
 470         report_garbage = report_pack_garbage;
 471         reprepare_packed_git();
 472         if (pack_garbage.nr > 0)
 473                 clean_pack_garbage();
 474
 475         if (auto_gc && too_many_loose_objects())
 476                 warning(_("There are too many unreachable loose objects; "
 477                         "run 'git prune' to remove them."));
 478
 479         if (!daemonized)
 480                 unlink(git_path("gc.log"));
 481
 482         return 0;
 483 }