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 "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
  23 static const char * const builtin_gc_usage[] = {
  24         N_("git gc [options]"),
  25         NULL
  26 };
  27
  28 static int pack_refs = 1;
  29 static int prune_reflogs = 1;
  30 static int aggressive_depth = 250;
  31 static int aggressive_window = 250;
  32 static int gc_auto_threshold = 6700;
  33 static int gc_auto_pack_limit = 50;
  34 static int detach_auto = 1;
  35 static const char *prune_expire = "2.weeks.ago";
  36 static const char *prune_repos_expire = "3.months.ago";
  37
  38 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  39 static struct argv_array reflog = ARGV_ARRAY_INIT;
  40 static struct argv_array repack = ARGV_ARRAY_INIT;
  41 static struct argv_array prune = ARGV_ARRAY_INIT;
  42 static struct argv_array prune_repos = ARGV_ARRAY_INIT;
  43 static struct argv_array rerere = ARGV_ARRAY_INIT;
  44
  45 static char *pidfile;
  46
  47 static void remove_pidfile(void)
  48 {
  49         if (pidfile)
  50                 unlink(pidfile);
  51 }
  52
  53 static void remove_pidfile_on_signal(int signo)
  54 {
  55         remove_pidfile();
  56         sigchain_pop(signo);
  57         raise(signo);
  58 }
  59
  60 static void git_config_get_date_string_const(const char *var, const char **val)
  61 {
  62         if (!git_config_get_string_const(var, val) && *val && strcmp(*val, "now")) {
  63                 unsigned long now = approxidate("now");
  64                 if (approxidate(*val) >= now)
  65                         git_die_config(var, _("Invalid %s: '%s'"), var, *val);
  66         }
  67 }
  68
  69 static void gc_config(void)
  70 {
  71         const char *value;
  72
  73         if (!git_config_get_value("gc.packrefs", &value)) {
  74                 if (value && !strcmp(value, "notbare"))
  75                         pack_refs = -1;
  76                 else
  77                         pack_refs = git_config_bool("gc.packrefs", value);
  78         }
  79
  80         git_config_get_int("gc.aggressivewindow", &aggressive_window);
  81         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
  82         git_config_get_int("gc.auto", &gc_auto_threshold);
  83         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
  84         git_config_get_bool("gc.autodetach", &detach_auto);
  85         git_config_get_date_string_const("gc.pruneexpire", &prune_expire);
  86         git_config_get_date_string_const("gc.prunereposexpire", &prune_repos_expire);
  87         git_config(git_default_config, NULL);
  88 }
  89
  90 static int too_many_loose_objects(void)
  91 {
  92         /*
  93          * Quickly check if a "gc" is needed, by estimating how
  94          * many loose objects there are.  Because SHA-1 is evenly
  95          * distributed, we can check only one and get a reasonable
  96          * estimate.
  97          */
  98         char path[PATH_MAX];
  99         const char *objdir = get_object_directory();
 100         DIR *dir;
 101         struct dirent *ent;
 102         int auto_threshold;
 103         int num_loose = 0;
 104         int needed = 0;
 105
 106         if (gc_auto_threshold <= 0)
 107                 return 0;
 108
 109         if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
 110                 warning(_("insanely long object directory %.*s"), 50, objdir);
 111                 return 0;
 112         }
 113         dir = opendir(path);
 114         if (!dir)
 115                 return 0;
 116
 117         auto_threshold = (gc_auto_threshold + 255) / 256;
 118         while ((ent = readdir(dir)) != NULL) {
 119                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 120                     ent->d_name[38] != '\0')
 121                         continue;
 122                 if (++num_loose > auto_threshold) {
 123                         needed = 1;
 124                         break;
 125                 }
 126         }
 127         closedir(dir);
 128         return needed;
 129 }
 130
 131 static int too_many_packs(void)
 132 {
 133         struct packed_git *p;
 134         int cnt;
 135
 136         if (gc_auto_pack_limit <= 0)
 137                 return 0;
 138
 139         prepare_packed_git();
 140         for (cnt = 0, p = packed_git; p; p = p->next) {
 141                 if (!p->pack_local)
 142                         continue;
 143                 if (p->pack_keep)
 144                         continue;
 145                 /*
 146                  * Perhaps check the size of the pack and count only
 147                  * very small ones here?
 148                  */
 149                 cnt++;
 150         }
 151         return gc_auto_pack_limit <= cnt;
 152 }
 153
 154 static void add_repack_all_option(void)
 155 {
 156         if (prune_expire && !strcmp(prune_expire, "now"))
 157                 argv_array_push(&repack, "-a");
 158         else {
 159                 argv_array_push(&repack, "-A");
 160                 if (prune_expire)
 161                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 162         }
 163 }
 164
 165 static int need_to_gc(void)
 166 {
 167         /*
 168          * Setting gc.auto to 0 or negative can disable the
 169          * automatic gc.
 170          */
 171         if (gc_auto_threshold <= 0)
 172                 return 0;
 173
 174         /*
 175          * If there are too many loose objects, but not too many
 176          * packs, we run "repack -d -l".  If there are too many packs,
 177          * we run "repack -A -d -l".  Otherwise we tell the caller
 178          * there is no need.
 179          */
 180         if (too_many_packs())
 181                 add_repack_all_option();
 182         else if (!too_many_loose_objects())
 183                 return 0;
 184
 185         if (run_hook_le(NULL, "pre-auto-gc", NULL))
 186                 return 0;
 187         return 1;
 188 }
 189
 190 /* return NULL on success, else hostname running the gc */
 191 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 192 {
 193         static struct lock_file lock;
 194         char my_host[128];
 195         struct strbuf sb = STRBUF_INIT;
 196         struct stat st;
 197         uintmax_t pid;
 198         FILE *fp;
 199         int fd;
 200
 201         if (pidfile)
 202                 /* already locked */
 203                 return NULL;
 204
 205         if (gethostname(my_host, sizeof(my_host)))
 206                 strcpy(my_host, "unknown");
 207
 208         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
 209                                        LOCK_DIE_ON_ERROR);
 210         if (!force) {
 211                 static char locking_host[128];
 212                 int should_exit;
 213                 fp = fopen(git_path("gc.pid"), "r");
 214                 memset(locking_host, 0, sizeof(locking_host));
 215                 should_exit =
 216                         fp != NULL &&
 217                         !fstat(fileno(fp), &st) &&
 218                         /*
 219                          * 12 hour limit is very generous as gc should
 220                          * never take that long. On the other hand we
 221                          * don't really need a strict limit here,
 222                          * running gc --auto one day late is not a big
 223                          * problem. --force can be used in manual gc
 224                          * after the user verifies that no gc is
 225                          * running.
 226                          */
 227                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 228                         fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
 229                         /* be gentle to concurrent "gc" on remote hosts */
 230                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 231                 if (fp != NULL)
 232                         fclose(fp);
 233                 if (should_exit) {
 234                         if (fd >= 0)
 235                                 rollback_lock_file(&lock);
 236                         *ret_pid = pid;
 237                         return locking_host;
 238                 }
 239         }
 240
 241         strbuf_addf(&sb, "%"PRIuMAX" %s",
 242                     (uintmax_t) getpid(), my_host);
 243         write_in_full(fd, sb.buf, sb.len);
 244         strbuf_release(&sb);
 245         commit_lock_file(&lock);
 246
 247         pidfile = git_pathdup("gc.pid");
 248         sigchain_push_common(remove_pidfile_on_signal);
 249         atexit(remove_pidfile);
 250
 251         return NULL;
 252 }
 253
 254 static int gc_before_repack(void)
 255 {
 256         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 257                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 258
 259         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 260                 return error(FAILED_RUN, reflog.argv[0]);
 261
 262         pack_refs = 0;
 263         prune_reflogs = 0;
 264         return 0;
 265 }
 266
 267 int cmd_gc(int argc, const char **argv, const char *prefix)
 268 {
 269         int aggressive = 0;
 270         int auto_gc = 0;
 271         int quiet = 0;
 272         int force = 0;
 273         const char *name;
 274         pid_t pid;
 275
 276         struct option builtin_gc_options[] = {
 277                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 278                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 279                         N_("prune unreferenced objects"),
 280                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 281                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 282                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 283                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 284                 OPT_END()
 285         };
 286
 287         if (argc == 2 && !strcmp(argv[1], "-h"))
 288                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 289
 290         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 291         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 292         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 293         argv_array_pushl(&prune, "prune", "--expire", NULL);
 294         argv_array_pushl(&prune_repos, "prune", "--repos", "--expire", NULL);
 295         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 296
 297         gc_config();
 298
 299         if (pack_refs < 0)
 300                 pack_refs = !is_bare_repository();
 301
 302         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 303                              builtin_gc_usage, 0);
 304         if (argc > 0)
 305                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 306
 307         if (aggressive) {
 308                 argv_array_push(&repack, "-f");
 309                 if (aggressive_depth > 0)
 310                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 311                 if (aggressive_window > 0)
 312                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 313         }
 314         if (quiet)
 315                 argv_array_push(&repack, "-q");
 316
 317         if (auto_gc) {
 318                 /*
 319                  * Auto-gc should be least intrusive as possible.
 320                  */
 321                 if (!need_to_gc())
 322                         return 0;
 323                 if (!quiet) {
 324                         if (detach_auto)
 325                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 326                         else
 327                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 328                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 329                 }
 330                 if (detach_auto) {
 331                         if (gc_before_repack())
 332                                 return -1;
 333                         /*
 334                          * failure to daemonize is ok, we'll continue
 335                          * in foreground
 336                          */
 337                         daemonize();
 338                 }
 339         } else
 340                 add_repack_all_option();
 341
 342         name = lock_repo_for_gc(force, &pid);
 343         if (name) {
 344                 if (auto_gc)
 345                         return 0; /* be quiet on --auto */
 346                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 347                     name, (uintmax_t)pid);
 348         }
 349
 350         if (gc_before_repack())
 351                 return -1;
 352
 353         if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 354                 return error(FAILED_RUN, repack.argv[0]);
 355
 356         if (prune_expire) {
 357                 argv_array_push(&prune, prune_expire);
 358                 if (quiet)
 359                         argv_array_push(&prune, "--no-progress");
 360                 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 361                         return error(FAILED_RUN, prune.argv[0]);
 362         }
 363
 364         if (prune_repos_expire) {
 365                 argv_array_push(&prune_repos, prune_repos_expire);
 366                 if (run_command_v_opt(prune_repos.argv, RUN_GIT_CMD))
 367                         return error(FAILED_RUN, prune_repos.argv[0]);
 368         }
 369
 370         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 371                 return error(FAILED_RUN, rerere.argv[0]);
 372
 373         if (auto_gc && too_many_loose_objects())
 374                 warning(_("There are too many unreachable loose objects; "
 375                         "run 'git prune' to remove them."));
 376
 377         return 0;
 378 }