Andrew's git - gitweb.git/blob

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