Andrew's git - gitweb.git/blob

ed5cc3c0318f87abd8674f7b98936a2265fff577
   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 "cache.h"
  15#include "parse-options.h"
  16#include "run-command.h"
  17#include "sigchain.h"
  18#include "argv-array.h"
  19#include "commit.h"
  20
  21#define FAILED_RUN "failed to run %s"
  22
  23static const char * const builtin_gc_usage[] = {
  24        N_("git gc [options]"),
  25        NULL
  26};
  27
  28static int pack_refs = 1;
  29static int aggressive_window = 250;
  30static int gc_auto_threshold = 6700;
  31static int gc_auto_pack_limit = 50;
  32static int detach_auto = 1;
  33static const char *prune_expire = "2.weeks.ago";
  34
  35static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  36static struct argv_array reflog = ARGV_ARRAY_INIT;
  37static struct argv_array repack = ARGV_ARRAY_INIT;
  38static struct argv_array prune = ARGV_ARRAY_INIT;
  39static struct argv_array rerere = ARGV_ARRAY_INIT;
  40
  41static char *pidfile;
  42
  43static void remove_pidfile(void)
  44{
  45        if (pidfile)
  46                unlink(pidfile);
  47}
  48
  49static void remove_pidfile_on_signal(int signo)
  50{
  51        remove_pidfile();
  52        sigchain_pop(signo);
  53        raise(signo);
  54}
  55
  56static int gc_config(const char *var, const char *value, void *cb)
  57{
  58        if (!strcmp(var, "gc.packrefs")) {
  59                if (value && !strcmp(value, "notbare"))
  60                        pack_refs = -1;
  61                else
  62                        pack_refs = git_config_bool(var, value);
  63                return 0;
  64        }
  65        if (!strcmp(var, "gc.aggressivewindow")) {
  66                aggressive_window = git_config_int(var, value);
  67                return 0;
  68        }
  69        if (!strcmp(var, "gc.auto")) {
  70                gc_auto_threshold = git_config_int(var, value);
  71                return 0;
  72        }
  73        if (!strcmp(var, "gc.autopacklimit")) {
  74                gc_auto_pack_limit = git_config_int(var, value);
  75                return 0;
  76        }
  77        if (!strcmp(var, "gc.autodetach")) {
  78                detach_auto = git_config_bool(var, value);
  79                return 0;
  80        }
  81        if (!strcmp(var, "gc.pruneexpire")) {
  82                if (value && strcmp(value, "now")) {
  83                        unsigned long now = approxidate("now");
  84                        if (approxidate(value) >= now)
  85                                return error(_("Invalid %s: '%s'"), var, value);
  86                }
  87                return git_config_string(&prune_expire, var, value);
  88        }
  89        return git_default_config(var, value, cb);
  90}
  91
  92static int too_many_loose_objects(void)
  93{
  94        /*
  95         * Quickly check if a "gc" is needed, by estimating how
  96         * many loose objects there are.  Because SHA-1 is evenly
  97         * distributed, we can check only one and get a reasonable
  98         * estimate.
  99         */
 100        char path[PATH_MAX];
 101        const char *objdir = get_object_directory();
 102        DIR *dir;
 103        struct dirent *ent;
 104        int auto_threshold;
 105        int num_loose = 0;
 106        int needed = 0;
 107
 108        if (gc_auto_threshold <= 0)
 109                return 0;
 110
 111        if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
 112                warning(_("insanely long object directory %.*s"), 50, objdir);
 113                return 0;
 114        }
 115        dir = opendir(path);
 116        if (!dir)
 117                return 0;
 118
 119        auto_threshold = (gc_auto_threshold + 255) / 256;
 120        while ((ent = readdir(dir)) != NULL) {
 121                if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 122                    ent->d_name[38] != '\0')
 123                        continue;
 124                if (++num_loose > auto_threshold) {
 125                        needed = 1;
 126                        break;
 127                }
 128        }
 129        closedir(dir);
 130        return needed;
 131}
 132
 133static int too_many_packs(void)
 134{
 135        struct packed_git *p;
 136        int cnt;
 137
 138        if (gc_auto_pack_limit <= 0)
 139                return 0;
 140
 141        prepare_packed_git();
 142        for (cnt = 0, p = packed_git; p; p = p->next) {
 143                if (!p->pack_local)
 144                        continue;
 145                if (p->pack_keep)
 146                        continue;
 147                /*
 148                 * Perhaps check the size of the pack and count only
 149                 * very small ones here?
 150                 */
 151                cnt++;
 152        }
 153        return gc_auto_pack_limit <= cnt;
 154}
 155
 156static void add_repack_all_option(void)
 157{
 158        if (prune_expire && !strcmp(prune_expire, "now"))
 159                argv_array_push(&repack, "-a");
 160        else {
 161                argv_array_push(&repack, "-A");
 162                if (prune_expire)
 163                        argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 164        }
 165}
 166
 167static int need_to_gc(void)
 168{
 169        /*
 170         * Setting gc.auto to 0 or negative can disable the
 171         * automatic gc.
 172         */
 173        if (gc_auto_threshold <= 0)
 174                return 0;
 175
 176        /*
 177         * If there are too many loose objects, but not too many
 178         * packs, we run "repack -d -l".  If there are too many packs,
 179         * we run "repack -A -d -l".  Otherwise we tell the caller
 180         * there is no need.
 181         */
 182        if (too_many_packs())
 183                add_repack_all_option();
 184        else if (!too_many_loose_objects())
 185                return 0;
 186
 187        if (run_hook(NULL, "pre-auto-gc", NULL))
 188                return 0;
 189        return 1;
 190}
 191
 192/* return NULL on success, else hostname running the gc */
 193static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 194{
 195        static struct lock_file lock;
 196        static char locking_host[128];
 197        char my_host[128];
 198        struct strbuf sb = STRBUF_INIT;
 199        struct stat st;
 200        uintmax_t pid;
 201        FILE *fp;
 202        int fd, should_exit;
 203
 204        if (pidfile)
 205                /* already locked */
 206                return NULL;
 207
 208        if (gethostname(my_host, sizeof(my_host)))
 209                strcpy(my_host, "unknown");
 210
 211        fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
 212                                       LOCK_DIE_ON_ERROR);
 213        if (!force) {
 214                fp = fopen(git_path("gc.pid"), "r");
 215                memset(locking_host, 0, sizeof(locking_host));
 216                should_exit =
 217                        fp != NULL &&
 218                        !fstat(fileno(fp), &st) &&
 219                        /*
 220                         * 12 hour limit is very generous as gc should
 221                         * never take that long. On the other hand we
 222                         * don't really need a strict limit here,
 223                         * running gc --auto one day late is not a big
 224                         * problem. --force can be used in manual gc
 225                         * after the user verifies that no gc is
 226                         * running.
 227                         */
 228                        time(NULL) - st.st_mtime <= 12 * 3600 &&
 229                        fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
 230                        /* be gentle to concurrent "gc" on remote hosts */
 231                        (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 232                if (fp != NULL)
 233                        fclose(fp);
 234                if (should_exit) {
 235                        if (fd >= 0)
 236                                rollback_lock_file(&lock);
 237                        *ret_pid = pid;
 238                        return locking_host;
 239                }
 240        }
 241
 242        strbuf_addf(&sb, "%"PRIuMAX" %s",
 243                    (uintmax_t) getpid(), my_host);
 244        write_in_full(fd, sb.buf, sb.len);
 245        strbuf_release(&sb);
 246        commit_lock_file(&lock);
 247
 248        pidfile = git_pathdup("gc.pid");
 249        sigchain_push_common(remove_pidfile_on_signal);
 250        atexit(remove_pidfile);
 251
 252        return NULL;
 253}
 254
 255int cmd_gc(int argc, const char **argv, const char *prefix)
 256{
 257        int aggressive = 0;
 258        int auto_gc = 0;
 259        int quiet = 0;
 260        int force = 0;
 261        const char *name;
 262        pid_t pid;
 263
 264        struct option builtin_gc_options[] = {
 265                OPT__QUIET(&quiet, N_("suppress progress reporting")),
 266                { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 267                        N_("prune unreferenced objects"),
 268                        PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 269                OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 270                OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 271                OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 272                OPT_END()
 273        };
 274
 275        if (argc == 2 && !strcmp(argv[1], "-h"))
 276                usage_with_options(builtin_gc_usage, builtin_gc_options);
 277
 278        argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 279        argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 280        argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 281        argv_array_pushl(&prune, "prune", "--expire", NULL );
 282        argv_array_pushl(&rerere, "rerere", "gc", NULL);
 283
 284        git_config(gc_config, NULL);
 285
 286        if (pack_refs < 0)
 287                pack_refs = !is_bare_repository();
 288
 289        argc = parse_options(argc, argv, prefix, builtin_gc_options,
 290                             builtin_gc_usage, 0);
 291        if (argc > 0)
 292                usage_with_options(builtin_gc_usage, builtin_gc_options);
 293
 294        if (aggressive) {
 295                argv_array_push(&repack, "-f");
 296                argv_array_push(&repack, "--depth=250");
 297                if (aggressive_window > 0)
 298                        argv_array_pushf(&repack, "--window=%d", aggressive_window);
 299        }
 300        if (quiet)
 301                argv_array_push(&repack, "-q");
 302
 303        if (auto_gc) {
 304                /*
 305                 * Auto-gc should be least intrusive as possible.
 306                 */
 307                if (!need_to_gc())
 308                        return 0;
 309                if (!quiet) {
 310                        if (detach_auto)
 311                                fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 312                        else
 313                                fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 314                        fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 315                }
 316                if (detach_auto)
 317                        /*
 318                         * failure to daemonize is ok, we'll continue
 319                         * in foreground
 320                         */
 321                        daemonize();
 322        } else
 323                add_repack_all_option();
 324
 325        name = lock_repo_for_gc(force, &pid);
 326        if (name) {
 327                if (auto_gc)
 328                        return 0; /* be quiet on --auto */
 329                die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 330                    name, (uintmax_t)pid);
 331        }
 332
 333        if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 334                return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 335
 336        if (run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 337                return error(FAILED_RUN, reflog.argv[0]);
 338
 339        if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 340                return error(FAILED_RUN, repack.argv[0]);
 341
 342        if (prune_expire) {
 343                argv_array_push(&prune, prune_expire);
 344                if (quiet)
 345                        argv_array_push(&prune, "--no-progress");
 346                if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 347                        return error(FAILED_RUN, prune.argv[0]);
 348        }
 349
 350        if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 351                return error(FAILED_RUN, rerere.argv[0]);
 352
 353        if (auto_gc && too_many_loose_objects())
 354                warning(_("There are too many unreachable loose objects; "
 355                        "run 'git prune' to remove them."));
 356
 357        return 0;
 358}