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