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 static const char *prune_worktrees_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_worktrees = 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 struct string_list pack_garbage = STRING_LIST_INIT_DUP;
  61
  62 static void clean_pack_garbage(void)
  63 {
  64         int i;
  65         for (i = 0; i < pack_garbage.nr; i++)
  66                 unlink_or_warn(pack_garbage.items[i].string);
  67         string_list_clear(&pack_garbage, 0);
  68 }
  69
  70 static void report_pack_garbage(unsigned seen_bits, const char *path)
  71 {
  72         if (seen_bits == PACKDIR_FILE_IDX)
  73                 string_list_append(&pack_garbage, path);
  74 }
  75
  76 static void git_config_date_string(const char *key, const char **output)
  77 {
  78         if (git_config_get_string_const(key, output))
  79                 return;
  80         if (strcmp(*output, "now")) {
  81                 unsigned long now = approxidate("now");
  82                 if (approxidate(*output) >= now)
  83                         git_die_config(key, _("Invalid %s: '%s'"), key, *output);
  84         }
  85 }
  86
  87 static 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_date_string("gc.pruneexpire", &prune_expire);
 104         git_config_date_string("gc.pruneworktreesexpire", &prune_worktrees_expire);
 105         git_config(git_default_config, NULL);
 106 }
 107
 108 static 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
 149 static 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
 172 static 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
 183 static 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 */
 209 static 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
 219         if (pidfile)
 220                 /* already locked */
 221                 return NULL;
 222
 223         if (gethostname(my_host, sizeof(my_host)))
 224                 strcpy(my_host, "unknown");
 225
 226         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
 227                                        LOCK_DIE_ON_ERROR);
 228         if (!force) {
 229                 static char locking_host[128];
 230                 int should_exit;
 231                 fp = fopen(git_path("gc.pid"), "r");
 232                 memset(locking_host, 0, sizeof(locking_host));
 233                 should_exit =
 234                         fp != NULL &&
 235                         !fstat(fileno(fp), &st) &&
 236                         /*
 237                          * 12 hour limit is very generous as gc should
 238                          * never take that long. On the other hand we
 239                          * don't really need a strict limit here,
 240                          * running gc --auto one day late is not a big
 241                          * problem. --force can be used in manual gc
 242                          * after the user verifies that no gc is
 243                          * running.
 244                          */
 245                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 246                         fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
 247                         /* be gentle to concurrent "gc" on remote hosts */
 248                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 249                 if (fp != NULL)
 250                         fclose(fp);
 251                 if (should_exit) {
 252                         if (fd >= 0)
 253                                 rollback_lock_file(&lock);
 254                         *ret_pid = pid;
 255                         return locking_host;
 256                 }
 257         }
 258
 259         strbuf_addf(&sb, "%"PRIuMAX" %s",
 260                     (uintmax_t) getpid(), my_host);
 261         write_in_full(fd, sb.buf, sb.len);
 262         strbuf_release(&sb);
 263         commit_lock_file(&lock);
 264
 265         pidfile = git_pathdup("gc.pid");
 266         sigchain_push_common(remove_pidfile_on_signal);
 267         atexit(remove_pidfile);
 268
 269         return NULL;
 270 }
 271
 272 static int gc_before_repack(void)
 273 {
 274         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 275                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 276
 277         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 278                 return error(FAILED_RUN, reflog.argv[0]);
 279
 280         pack_refs = 0;
 281         prune_reflogs = 0;
 282         return 0;
 283 }
 284
 285 int cmd_gc(int argc, const char **argv, const char *prefix)
 286 {
 287         int aggressive = 0;
 288         int auto_gc = 0;
 289         int quiet = 0;
 290         int force = 0;
 291         const char *name;
 292         pid_t pid;
 293
 294         struct option builtin_gc_options[] = {
 295                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 296                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 297                         N_("prune unreferenced objects"),
 298                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 299                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 300                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 301                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 302                 OPT_END()
 303         };
 304
 305         if (argc == 2 && !strcmp(argv[1], "-h"))
 306                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 307
 308         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 309         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 310         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 311         argv_array_pushl(&prune, "prune", "--expire", NULL);
 312         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
 313         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 314
 315         gc_config();
 316
 317         if (pack_refs < 0)
 318                 pack_refs = !is_bare_repository();
 319
 320         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 321                              builtin_gc_usage, 0);
 322         if (argc > 0)
 323                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 324
 325         if (aggressive) {
 326                 argv_array_push(&repack, "-f");
 327                 if (aggressive_depth > 0)
 328                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 329                 if (aggressive_window > 0)
 330                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 331         }
 332         if (quiet)
 333                 argv_array_push(&repack, "-q");
 334
 335         if (auto_gc) {
 336                 /*
 337                  * Auto-gc should be least intrusive as possible.
 338                  */
 339                 if (!need_to_gc())
 340                         return 0;
 341                 if (!quiet) {
 342                         if (detach_auto)
 343                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 344                         else
 345                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 346                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 347                 }
 348                 if (detach_auto) {
 349                         if (gc_before_repack())
 350                                 return -1;
 351                         /*
 352                          * failure to daemonize is ok, we'll continue
 353                          * in foreground
 354                          */
 355                         daemonize();
 356                 }
 357         } else
 358                 add_repack_all_option();
 359
 360         name = lock_repo_for_gc(force, &pid);
 361         if (name) {
 362                 if (auto_gc)
 363                         return 0; /* be quiet on --auto */
 364                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 365                     name, (uintmax_t)pid);
 366         }
 367
 368         if (gc_before_repack())
 369                 return -1;
 370
 371         if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 372                 return error(FAILED_RUN, repack.argv[0]);
 373
 374         if (prune_expire) {
 375                 argv_array_push(&prune, prune_expire);
 376                 if (quiet)
 377                         argv_array_push(&prune, "--no-progress");
 378                 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 379                         return error(FAILED_RUN, prune.argv[0]);
 380         }
 381
 382         if (prune_worktrees_expire) {
 383                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 384                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 385                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 386         }
 387
 388         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 389                 return error(FAILED_RUN, rerere.argv[0]);
 390
 391         report_garbage = report_pack_garbage;
 392         reprepare_packed_git();
 393         if (pack_garbage.nr > 0)
 394                 clean_pack_garbage();
 395
 396         if (auto_gc && too_many_loose_objects())
 397                 warning(_("There are too many unreachable loose objects; "
 398                         "run 'git prune' to remove them."));
 399
 400         return 0;
 401 }