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 void git_config_date_string(const char *key, const char **output)
  61 {
  62         if (git_config_get_string_const(key, output))
  63                 return;
  64         if (strcmp(*output, "now")) {
  65                 unsigned long now = approxidate("now");
  66                 if (approxidate(*output) >= now)
  67                         git_die_config(key, _("Invalid %s: '%s'"), key, *output);
  68         }
  69 }
  70
  71 static void gc_config(void)
  72 {
  73         const char *value;
  74
  75         if (!git_config_get_value("gc.packrefs", &value)) {
  76                 if (value && !strcmp(value, "notbare"))
  77                         pack_refs = -1;
  78                 else
  79                         pack_refs = git_config_bool("gc.packrefs", value);
  80         }
  81
  82         git_config_get_int("gc.aggressivewindow", &aggressive_window);
  83         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
  84         git_config_get_int("gc.auto", &gc_auto_threshold);
  85         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
  86         git_config_get_bool("gc.autodetach", &detach_auto);
  87         git_config_date_string("gc.pruneexpire", &prune_expire);
  88         git_config_date_string("gc.pruneworktreesexpire", &prune_worktrees_expire);
  89         git_config(git_default_config, NULL);
  90 }
  91
  92 static 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
 133 static 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
 156 static 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
 167 static 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_le(NULL, "pre-auto-gc", NULL))
 188                 return 0;
 189         return 1;
 190 }
 191
 192 /* return NULL on success, else hostname running the gc */
 193 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 194 {
 195         static struct lock_file lock;
 196         char my_host[128];
 197         struct strbuf sb = STRBUF_INIT;
 198         struct stat st;
 199         uintmax_t pid;
 200         FILE *fp;
 201         int fd;
 202
 203         if (pidfile)
 204                 /* already locked */
 205                 return NULL;
 206
 207         if (gethostname(my_host, sizeof(my_host)))
 208                 strcpy(my_host, "unknown");
 209
 210         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
 211                                        LOCK_DIE_ON_ERROR);
 212         if (!force) {
 213                 static char locking_host[128];
 214                 int should_exit;
 215                 fp = fopen(git_path("gc.pid"), "r");
 216                 memset(locking_host, 0, sizeof(locking_host));
 217                 should_exit =
 218                         fp != NULL &&
 219                         !fstat(fileno(fp), &st) &&
 220                         /*
 221                          * 12 hour limit is very generous as gc should
 222                          * never take that long. On the other hand we
 223                          * don't really need a strict limit here,
 224                          * running gc --auto one day late is not a big
 225                          * problem. --force can be used in manual gc
 226                          * after the user verifies that no gc is
 227                          * running.
 228                          */
 229                         time(NULL) - st.st_mtime <= 12 * 3600 &&
 230                         fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
 231                         /* be gentle to concurrent "gc" on remote hosts */
 232                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
 233                 if (fp != NULL)
 234                         fclose(fp);
 235                 if (should_exit) {
 236                         if (fd >= 0)
 237                                 rollback_lock_file(&lock);
 238                         *ret_pid = pid;
 239                         return locking_host;
 240                 }
 241         }
 242
 243         strbuf_addf(&sb, "%"PRIuMAX" %s",
 244                     (uintmax_t) getpid(), my_host);
 245         write_in_full(fd, sb.buf, sb.len);
 246         strbuf_release(&sb);
 247         commit_lock_file(&lock);
 248
 249         pidfile = git_pathdup("gc.pid");
 250         sigchain_push_common(remove_pidfile_on_signal);
 251         atexit(remove_pidfile);
 252
 253         return NULL;
 254 }
 255
 256 static int gc_before_repack(void)
 257 {
 258         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
 259                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
 260
 261         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
 262                 return error(FAILED_RUN, reflog.argv[0]);
 263
 264         pack_refs = 0;
 265         prune_reflogs = 0;
 266         return 0;
 267 }
 268
 269 int cmd_gc(int argc, const char **argv, const char *prefix)
 270 {
 271         int aggressive = 0;
 272         int auto_gc = 0;
 273         int quiet = 0;
 274         int force = 0;
 275         const char *name;
 276         pid_t pid;
 277
 278         struct option builtin_gc_options[] = {
 279                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
 280                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
 281                         N_("prune unreferenced objects"),
 282                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
 283                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
 284                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
 285                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
 286                 OPT_END()
 287         };
 288
 289         if (argc == 2 && !strcmp(argv[1], "-h"))
 290                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 291
 292         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
 293         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
 294         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
 295         argv_array_pushl(&prune, "prune", "--expire", NULL);
 296         argv_array_pushl(&prune_worktrees, "prune", "--worktrees", "--expire", NULL);
 297         argv_array_pushl(&rerere, "rerere", "gc", NULL);
 298
 299         gc_config();
 300
 301         if (pack_refs < 0)
 302                 pack_refs = !is_bare_repository();
 303
 304         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 305                              builtin_gc_usage, 0);
 306         if (argc > 0)
 307                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 308
 309         if (aggressive) {
 310                 argv_array_push(&repack, "-f");
 311                 if (aggressive_depth > 0)
 312                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
 313                 if (aggressive_window > 0)
 314                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
 315         }
 316         if (quiet)
 317                 argv_array_push(&repack, "-q");
 318
 319         if (auto_gc) {
 320                 /*
 321                  * Auto-gc should be least intrusive as possible.
 322                  */
 323                 if (!need_to_gc())
 324                         return 0;
 325                 if (!quiet) {
 326                         if (detach_auto)
 327                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
 328                         else
 329                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
 330                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
 331                 }
 332                 if (detach_auto) {
 333                         if (gc_before_repack())
 334                                 return -1;
 335                         /*
 336                          * failure to daemonize is ok, we'll continue
 337                          * in foreground
 338                          */
 339                         daemonize();
 340                 }
 341         } else
 342                 add_repack_all_option();
 343
 344         name = lock_repo_for_gc(force, &pid);
 345         if (name) {
 346                 if (auto_gc)
 347                         return 0; /* be quiet on --auto */
 348                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
 349                     name, (uintmax_t)pid);
 350         }
 351
 352         if (gc_before_repack())
 353                 return -1;
 354
 355         if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
 356                 return error(FAILED_RUN, repack.argv[0]);
 357
 358         if (prune_expire) {
 359                 argv_array_push(&prune, prune_expire);
 360                 if (quiet)
 361                         argv_array_push(&prune, "--no-progress");
 362                 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
 363                         return error(FAILED_RUN, prune.argv[0]);
 364         }
 365
 366         if (prune_worktrees_expire) {
 367                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
 368                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
 369                         return error(FAILED_RUN, prune_worktrees.argv[0]);
 370         }
 371
 372         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 373                 return error(FAILED_RUN, rerere.argv[0]);
 374
 375         if (auto_gc && too_many_loose_objects())
 376                 warning(_("There are too many unreachable loose objects; "
 377                         "run 'git prune' to remove them."));
 378
 379         return 0;
 380 }