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         char path[PATH_MAX];
 139         const char *objdir = get_object_directory();
 140         DIR *dir;
 141         struct dirent *ent;
 142         int auto_threshold;
 143         int num_loose = 0;
 144         int needed = 0;
 145
 146         if (gc_auto_threshold <= 0)
 147                 return 0;
 148
 149         if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
 150                 warning(_("insanely long object directory %.*s"), 50, objdir);
 151                 return 0;
 152         }
 153         dir = opendir(path);
 154         if (!dir)
 155                 return 0;
 156
 157         auto_threshold = (gc_auto_threshold + 255) / 256;
 158         while ((ent = readdir(dir)) != NULL) {
 159                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 160                     ent->d_name[38] != '\0')
 161                         continue;
 162                 if (++num_loose > auto_threshold) {
 163                         needed = 1;
 164                         break;
 165                 }
 166         }
 167         closedir(dir);
 168         return needed;
 169 }
 170
 171 static int too_many_packs(void)
 172 {
 173         struct packed_git *p;
 174         int cnt;
 175
 176         if (gc_auto_pack_limit <= 0)
 177                 return 0;
 178
 179         prepare_packed_git();
 180         for (cnt = 0, p = packed_git; p; p = p->next) {
 181                 if (!p->pack_local)
 182                         continue;
 183                 if (p->pack_keep)
 184                         continue;
 185                 /*
 186                  * Perhaps check the size of the pack and count only
 187                  * very small ones here?
 188                  */
 189                 cnt++;
 190         }
 191         return gc_auto_pack_limit < cnt;
 192 }
 193
 194 static void add_repack_all_option(void)
 195 {
 196         if (prune_expire && !strcmp(prune_expire, "now"))
 197                 argv_array_push(&repack, "-a");
 198         else {
 199                 argv_array_push(&repack, "-A");
 200                 if (prune_expire)
 201                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 202         }
 203 }
 204
 205 static void add_repack_incremental_option(void)
 206 {
 207        argv_array_push(&repack, "--no-write-bitmap-index");
 208 }
 209
 210 static int need_to_gc(void)
 211 {
 212         /*
 213          * Setting gc.auto to 0 or negative can disable the
 214          * automatic gc.
 215          */
 216         if (gc_auto_threshold <= 0)
 217                 return 0;
 218
 219         /*
 220          * If there are too many loose objects, but not too many
 221          * packs, we run "repack -d -l".  If there are too many packs,
 222          * we run "repack -A -d -l".  Otherwise we tell the caller
 223          * there is no need.
 224          */
 225         if (too_many_packs())
 226                 add_repack_all_option();
 227         else if (too_many_loose_objects())
 228                 add_repack_incremental_option();
 229         else
 230                 return 0;
 231
 232         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 233                 return 0;
 234         return 1;
 235 }
 236
 237 /* return NULL on success, else hostname running the gc */
 238 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 239 {
 240         static struct lock_file lock;
 241         char my_host[128];
 242         struct strbuf sb = STRBUF_INIT;
 243         struct stat st;
 244         uintmax_t pid;
 245         FILE *fp;
 246         int fd;
 247         char *pidfile_path;
 248
 249         if (is_tempfile_active(&pidfile))
 250                 /* already locked */
 251                 return NULL;
 252
 253         if (gethostname(my_host, sizeof(my_host)))
 254                 xsnprintf(my_host, sizeof(my_host), "unknown");
 255
 256         pidfile_path = git_pathdup("gc.pid");
 257         fd = hold_lock_file_for_update(&lock, pidfile_path,
 258                                        LOCK_DIE_ON_ERROR);
 259         if (!force) {
 260                 static char locking_host[128];
 261                 int should_exit;
 262                 fp = fopen(pidfile_path, "r");
 263                 memset(locking_host, 0, sizeof(locking_host));
 264                 should_exit =
 265                         fp != NULL &&
 266                         !fstat(fileno(fp), &st) &&
 267                         /*
 268                          * 12 hour limit is very generous as gc should
 269                          * never take that long. On the other hand we
 270                          * don't really need a strict limit here,
 271                          * running gc --auto one day late is not a big
 272                          * problem. --force can be used in manual gc
 273                          * after the user verifies that no gc is
 274                          * running.
 275                          */
 276                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 277                         fscanf(fp, "%"SCNuMAX" %127c", &pid, locking_host) == 2 &&
 278                         /* be gentle to concurrent "gc" on remote hosts */
 279                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 280                 if (fp != NULL)
 281                         fclose(fp);
 282                 if (should_exit) {
 283                         if (fd >= 0)
 284                                 rollback_lock_file(&lock);
 285                         *ret_pid = pid;
 286                         free(pidfile_path);
 287                         return locking_host;
 288                 }
 289         }
 290
 291         strbuf_addf(&sb, "%"PRIuMAX" %s",
 292                     (uintmax_t) getpid(), my_host);
 293         write_in_full(fd, sb.buf, sb.len);
 294         strbuf_release(&sb);
 295         commit_lock_file(&lock);
 296         register_tempfile(&pidfile, pidfile_path);
 297         free(pidfile_path);
 298         return NULL;
 299 }
 300
 301 static int report_last_gc_error(void)
 302 {
 303         struct strbuf sb = STRBUF_INIT;
 304         int ret = 0;
 305         struct stat st;
 306         char *gc_log_path = git_pathdup("gc.log");
 307
 308         if (stat(gc_log_path, &st)) {
 309                 if (errno == ENOENT)
 310                         goto done;
 311
 312                 ret = error_errno(_("Can't stat %s"), gc_log_path);
 313                 goto done;
 314         }
 315
 316         if (st.st_mtime < gc_log_expire_time)
 317                 goto done;
 318
 319         ret = strbuf_read_file(&sb, gc_log_path, 0);
 320         if (ret > 0)
 321                 ret = error(_("The last gc run reported the following. "
 322                                "Please correct the root cause\n"
 323                                "and remove %s.\n"
 324                                "Automatic cleanup will not be performed "
 325                                "until the file is removed.\n\n"
 326                                "%s"),
 327                             gc_log_path, sb.buf);
 328         strbuf_release(&sb);
 329 done:
 330         free(gc_log_path);
 331         return ret;
 332 }
 333
 334 static int gc_before_repack(void)
 335 {
 336         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 337                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 338
 339         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 340                 return error(FAILED_RUN, reflog.argv[0]);
 341
 342         pack_refs = 0;
 343         prune_reflogs = 0;
 344         return 0;
 345 }
 346
 347 int cmd_gc(int argc, const char **argv, const char *prefix)
 348 {
 349         int aggressive = 0;
 350         int auto_gc = 0;
 351         int quiet = 0;
 352         int force = 0;
 353         const char *name;
 354         pid_t pid;
 355         int daemonized = 0;
 356
 357         struct option builtin_gc_options[] = {
 358                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 359                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 360                         N_("prune unreferenced objects"),
 361                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 362                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 363                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 364                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 365                 OPT_END()
 366         };
 367
 368         if (argc == 2 && !strcmp(argv[1], "-h"))
 369                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 370
 371         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 372         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 373         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 374         argv_array_pushl(&prune, "prune", "--expire", NULL);
 375         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 376         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 377
 378         /* default expiry time, overwritten in gc_config */
 379         gc_config();
 380         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
 381                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
 382
 383         if (pack_refs < 0)
 384                 pack_refs = !is_bare_repository();
 385
 386         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 387                              builtin_gc_usage, 0);
 388         if (argc > 0)
 389                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 390
 391         if (aggressive) {
 392                 argv_array_push(&repack, "-f");
 393                 if (aggressive_depth > 0)
 394                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 395                 if (aggressive_window > 0)
 396                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 397         }
 398         if (quiet)
 399                 argv_array_push(&repack, "-q");
 400
 401         if (auto_gc) {
 402                 /*
 403                  * Auto-gc should be least intrusive as possible.
 404                  */
 405                 if (!need_to_gc())
 406                         return 0;
 407                 if (!quiet) {
 408                         if (detach_auto)
 409                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 410                         else
 411                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 412                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 413                 }
 414                 if (detach_auto) {
 415                         if (report_last_gc_error())
 416                                 return -1;
 417
 418                         if (gc_before_repack())
 419                                 return -1;
 420                         /*
 421                          * failure to daemonize is ok, we'll continue
 422                          * in foreground
 423                          */
 424                         daemonized = !daemonize();
 425                 }
 426         } else
 427                 add_repack_all_option();
 428
 429         name = lock_repo_for_gc(force, &pid);
 430         if (name) {
 431                 if (auto_gc)
 432                         return 0; /* be quiet on --auto */
 433                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 434                     name, (uintmax_t)pid);
 435         }
 436
 437         if (daemonized) {
 438                 hold_lock_file_for_update(&log_lock,
 439                                           git_path("gc.log"),
 440                                           LOCK_DIE_ON_ERROR);
 441                 dup2(get_lock_file_fd(&log_lock), 2);
 442                 sigchain_push_common(process_log_file_on_signal);
 443                 atexit(process_log_file_at_exit);
 444         }
 445
 446         if (gc_before_repack())
 447                 return -1;
 448
 449         if (!repository_format_precious_objects) {
 450                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 451                         return error(FAILED_RUN, repack.argv[0]);
 452
 453                 if (prune_expire) {
 454                         argv_array_push(&prune, prune_expire);
 455                         if (quiet)
 456                                 argv_array_push(&prune, "--no-progress");
 457                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 458                                 return error(FAILED_RUN, prune.argv[0]);
 459                 }
 460         }
 461
 462         if (prune_worktrees_expire) {
 463                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 464                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 465                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 466         }
 467
 468         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 469                 return error(FAILED_RUN, rerere.argv[0]);
 470
 471         report_garbage = report_pack_garbage;
 472         reprepare_packed_git();
 473         if (pack_garbage.nr > 0)
 474                 clean_pack_garbage();
 475
 476         if (auto_gc && too_many_loose_objects())
 477                 warning(_("There are too many unreachable loose objects; "
 478                         "run 'git prune' to remove them."));
 479
 480         if (!daemonized)
 481                 unlink(git_path("gc.log"));
 482
 483         return 0;
 484 }