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