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