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         prepare_packed_git(the_repository);
 178         for (cnt = 0, p = get_packed_git(the_repository); p; p = p->next) {
 179                 if (!p->pack_local)
 180                         continue;
 181                 if (p->pack_keep)
 182                         continue;
 183                 /*
 184                  * Perhaps check the size of the pack and count only
 185                  * very small ones here?
 186                  */
 187                 cnt++;
 188         }
 189         return gc_auto_pack_limit < cnt;
 190 }
 191
 192 static void add_repack_all_option(void)
 193 {
 194         if (prune_expire && !strcmp(prune_expire, "now"))
 195                 argv_array_push(&repack, "-a");
 196         else {
 197                 argv_array_push(&repack, "-A");
 198                 if (prune_expire)
 199                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 200         }
 201 }
 202
 203 static void add_repack_incremental_option(void)
 204 {
 205        argv_array_push(&repack, "--no-write-bitmap-index");
 206 }
 207
 208 static int need_to_gc(void)
 209 {
 210         /*
 211          * Setting gc.auto to 0 or negative can disable the
 212          * automatic gc.
 213          */
 214         if (gc_auto_threshold <= 0)
 215                 return 0;
 216
 217         /*
 218          * If there are too many loose objects, but not too many
 219          * packs, we run "repack -d -l".  If there are too many packs,
 220          * we run "repack -A -d -l".  Otherwise we tell the caller
 221          * there is no need.
 222          */
 223         if (too_many_packs())
 224                 add_repack_all_option();
 225         else if (too_many_loose_objects())
 226                 add_repack_incremental_option();
 227         else
 228                 return 0;
 229
 230         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 231                 return 0;
 232         return 1;
 233 }
 234
 235 /* return NULL on success, else hostname running the gc */
 236 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 237 {
 238         static struct lock_file lock;
 239         char my_host[HOST_NAME_MAX + 1];
 240         struct strbuf sb = STRBUF_INIT;
 241         struct stat st;
 242         uintmax_t pid;
 243         FILE *fp;
 244         int fd;
 245         char *pidfile_path;
 246
 247         if (is_tempfile_active(pidfile))
 248                 /* already locked */
 249                 return NULL;
 250
 251         if (xgethostname(my_host, sizeof(my_host)))
 252                 xsnprintf(my_host, sizeof(my_host), "unknown");
 253
 254         pidfile_path = git_pathdup("gc.pid");
 255         fd = hold_lock_file_for_update(&lock, pidfile_path,
 256                                        LOCK_DIE_ON_ERROR);
 257         if (!force) {
 258                 static char locking_host[HOST_NAME_MAX + 1];
 259                 static char *scan_fmt;
 260                 int should_exit;
 261
 262                 if (!scan_fmt)
 263                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
 264                 fp = fopen(pidfile_path, "r");
 265                 memset(locking_host, 0, sizeof(locking_host));
 266                 should_exit =
 267                         fp != NULL &&
 268                         !fstat(fileno(fp), &st) &&
 269                         /*
 270                          * 12 hour limit is very generous as gc should
 271                          * never take that long. On the other hand we
 272                          * don't really need a strict limit here,
 273                          * running gc --auto one day late is not a big
 274                          * problem. --force can be used in manual gc
 275                          * after the user verifies that no gc is
 276                          * running.
 277                          */
 278                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 279                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
 280                         /* be gentle to concurrent "gc" on remote hosts */
 281                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 282                 if (fp != NULL)
 283                         fclose(fp);
 284                 if (should_exit) {
 285                         if (fd >= 0)
 286                                 rollback_lock_file(&lock);
 287                         *ret_pid = pid;
 288                         free(pidfile_path);
 289                         return locking_host;
 290                 }
 291         }
 292
 293         strbuf_addf(&sb, "%"PRIuMAX" %s",
 294                     (uintmax_t) getpid(), my_host);
 295         write_in_full(fd, sb.buf, sb.len);
 296         strbuf_release(&sb);
 297         commit_lock_file(&lock);
 298         pidfile = register_tempfile(pidfile_path);
 299         free(pidfile_path);
 300         return NULL;
 301 }
 302
 303 static int report_last_gc_error(void)
 304 {
 305         struct strbuf sb = STRBUF_INIT;
 306         int ret = 0;
 307         struct stat st;
 308         char *gc_log_path = git_pathdup("gc.log");
 309
 310         if (stat(gc_log_path, &st)) {
 311                 if (errno == ENOENT)
 312                         goto done;
 313
 314                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 315                 goto done;
 316         }
 317
 318         if (st.st_mtime < gc_log_expire_time)
 319                 goto done;
 320
 321         ret = strbuf_read_file(&sb, gc_log_path, 0);
 322         if (ret > 0)
 323                 ret = error(_("The last gc run reported the following. "
 324                                "Please correct the root cause\n"
 325                                "and remove %s.\n"
 326                                "Automatic cleanup will not be performed "
 327                                "until the file is removed.\n\n"
 328                                "%s"),
 329                             gc_log_path, sb.buf);
 330         strbuf_release(&sb);
 331 done:
 332         free(gc_log_path);
 333         return ret;
 334 }
 335
 336 static int gc_before_repack(void)
 337 {
 338         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 339                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 340
 341         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 342                 return error(FAILED_RUN, reflog.argv[0]);
 343
 344         pack_refs = 0;
 345         prune_reflogs = 0;
 346         return 0;
 347 }
 348
 349 int cmd_gc(int argc, const char **argv, const char *prefix)
 350 {
 351         int aggressive = 0;
 352         int auto_gc = 0;
 353         int quiet = 0;
 354         int force = 0;
 355         const char *name;
 356         pid_t pid;
 357         int daemonized = 0;
 358
 359         struct option builtin_gc_options[] = {
 360                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 361                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 362                         N_("prune unreferenced objects"),
 363                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 364                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 365                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 366                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 367                 OPT_END()
 368         };
 369
 370         if (argc == 2 && !strcmp(argv[1], "-h"))
 371                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 372
 373         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 374         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 375         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 376         argv_array_pushl(&prune, "prune", "--expire", NULL);
 377         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 378         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 379
 380         /* default expiry time, overwritten in gc_config */
 381         gc_config();
 382         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 383                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
 384
 385         if (pack_refs < 0)
 386                 pack_refs = !is_bare_repository();
 387
 388         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 389                              builtin_gc_usage, 0);
 390         if (argc > 0)
 391                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 392
 393         if (aggressive) {
 394                 argv_array_push(&repack, "-f");
 395                 if (aggressive_depth > 0)
 396                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 397                 if (aggressive_window > 0)
 398                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 399         }
 400         if (quiet)
 401                 argv_array_push(&repack, "-q");
 402
 403         if (auto_gc) {
 404                 /*
 405                  * Auto-gc should be least intrusive as possible.
 406                  */
 407                 if (!need_to_gc())
 408                         return 0;
 409                 if (!quiet) {
 410                         if (detach_auto)
 411                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 412                         else
 413                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 414                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 415                 }
 416                 if (detach_auto) {
 417                         if (report_last_gc_error())
 418                                 return -1;
 419
 420                         if (lock_repo_for_gc(force, &pid))
 421                                 return 0;
 422                         if (gc_before_repack())
 423                                 return -1;
 424                         delete_tempfile(&pidfile);
 425
 426                         /*
 427                          * failure to daemonize is ok, we'll continue
 428                          * in foreground
 429                          */
 430                         daemonized = !daemonize();
 431                 }
 432         } else
 433                 add_repack_all_option();
 434
 435         name = lock_repo_for_gc(force, &pid);
 436         if (name) {
 437                 if (auto_gc)
 438                         return 0; /* be quiet on --auto */
 439                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 440                     name, (uintmax_t)pid);
 441         }
 442
 443         if (daemonized) {
 444                 hold_lock_file_for_update(&log_lock,
 445                                           git_path("gc.log"),
 446                                           LOCK_DIE_ON_ERROR);
 447                 dup2(get_lock_file_fd(&log_lock), 2);
 448                 sigchain_push_common(process_log_file_on_signal);
 449                 atexit(process_log_file_at_exit);
 450         }
 451
 452         if (gc_before_repack())
 453                 return -1;
 454
 455         if (!repository_format_precious_objects) {
 456                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 457                         return error(FAILED_RUN, repack.argv[0]);
 458
 459                 if (prune_expire) {
 460                         argv_array_push(&prune, prune_expire);
 461                         if (quiet)
 462                                 argv_array_push(&prune, "--no-progress");
 463                         if (repository_format_partial_clone)
 464                                 argv_array_push(&prune,
 465                                                 "--exclude-promisor-objects");
 466                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 467                                 return error(FAILED_RUN, prune.argv[0]);
 468                 }
 469         }
 470
 471         if (prune_worktrees_expire) {
 472                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 473                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 474                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 475         }
 476
 477         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 478                 return error(FAILED_RUN, rerere.argv[0]);
 479
 480         report_garbage = report_pack_garbage;
 481         reprepare_packed_git(the_repository);
 482         if (pack_garbage.nr > 0)
 483                 clean_pack_garbage();
 484
 485         if (auto_gc && too_many_loose_objects())
 486                 warning(_("There are too many unreachable loose objects; "
 487                         "run 'git prune' to remove them."));
 488
 489         if (!daemonized)
 490                 unlink(git_path("gc.log"));
 491
 492         return 0;
 493 }