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_window = 250;
  31 static int gc_auto_threshold = 6700;
  32 static int gc_auto_pack_limit = 50;
  33 static int detach_auto = 1;
  34 static const char *prune_expire = "2.weeks.ago";
  35
  36 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
  37 static struct argv_array reflog = ARGV_ARRAY_INIT;
  38 static struct argv_array repack = ARGV_ARRAY_INIT;
  39 static struct argv_array prune = ARGV_ARRAY_INIT;
  40 static struct argv_array rerere = ARGV_ARRAY_INIT;
  41
  42 static char *pidfile;
  43
  44 static void remove_pidfile(void)
  45 {
  46         if (pidfile)
  47                 unlink(pidfile);
  48 }
  49
  50 static void remove_pidfile_on_signal(int signo)
  51 {
  52         remove_pidfile();
  53         sigchain_pop(signo);
  54         raise(signo);
  55 }
  56
  57 static int gc_config(const char *var, const char *value, void *cb)
  58 {
  59         if (!strcmp(var, "gc.packrefs")) {
  60                 if (value && !strcmp(value, "notbare"))
  61                         pack_refs = -1;
  62                 else
  63                         pack_refs = git_config_bool(var, value);
  64                 return 0;
  65         }
  66         if (!strcmp(var, "gc.aggressivewindow")) {
  67                 aggressive_window = git_config_int(var, value);
  68                 return 0;
  69         }
  70         if (!strcmp(var, "gc.auto")) {
  71                 gc_auto_threshold = git_config_int(var, value);
  72                 return 0;
  73         }
  74         if (!strcmp(var, "gc.autopacklimit")) {
  75                 gc_auto_pack_limit = git_config_int(var, value);
  76                 return 0;
  77         }
  78         if (!strcmp(var, "gc.autodetach")) {
  79                 detach_auto = git_config_bool(var, value);
  80                 return 0;
  81         }
  82         if (!strcmp(var, "gc.pruneexpire")) {
  83                 if (value && strcmp(value, "now")) {
  84                         unsigned long now = approxidate("now");
  85                         if (approxidate(value) >= now)
  86                                 return error(_("Invalid %s: '%s'"), var, value);
  87                 }
  88                 return git_config_string(&prune_expire, var, value);
  89         }
  90         return git_default_config(var, value, cb);
  91 }
  92
  93 static int too_many_loose_objects(void)
  94 {
  95         /*
  96          * Quickly check if a "gc" is needed, by estimating how
  97          * many loose objects there are.  Because SHA-1 is evenly
  98          * distributed, we can check only one and get a reasonable
  99          * estimate.
 100          */
 101         char path[PATH_MAX];
 102         const char *objdir = get_object_directory();
 103         DIR *dir;
 104         struct dirent *ent;
 105         int auto_threshold;
 106         int num_loose = 0;
 107         int needed = 0;
 108
 109         if (gc_auto_threshold <= 0)
 110                 return 0;
 111
 112         if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
 113                 warning(_("insanely long object directory %.*s"), 50, objdir);
 114                 return 0;
 115         }
 116         dir = opendir(path);
 117         if (!dir)
 118                 return 0;
 119
 120         auto_threshold = (gc_auto_threshold + 255) / 256;
 121         while ((ent = readdir(dir)) != NULL) {
 122                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 123                     ent->d_name[38] != '\0')
 124                         continue;
 125                 if (++num_loose > auto_threshold) {
 126                         needed = 1;
 127                         break;
 128                 }
 129         }
 130         closedir(dir);
 131         return needed;
 132 }
 133
 134 static int too_many_packs(void)
 135 {
 136         struct packed_git *p;
 137         int cnt;
 138
 139         if (gc_auto_pack_limit <= 0)
 140                 return 0;
 141
 142         prepare_packed_git();
 143         for (cnt = 0, p = packed_git; p; p = p->next) {
 144                 if (!p->pack_local)
 145                         continue;
 146                 if (p->pack_keep)
 147                         continue;
 148                 /*
 149                  * Perhaps check the size of the pack and count only
 150                  * very small ones here?
 151                  */
 152                 cnt++;
 153         }
 154         return gc_auto_pack_limit <= cnt;
 155 }
 156
 157 static void add_repack_all_option(void)
 158 {
 159         if (prune_expire && !strcmp(prune_expire, "now"))
 160                 argv_array_push(&repack, "-a");
 161         else {
 162                 argv_array_push(&repack, "-A");
 163                 if (prune_expire)
 164                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
 165         }
 166 }
 167
 168 static int need_to_gc(void)
 169 {
 170         /*
 171          * Setting gc.auto to 0 or negative can disable the
 172          * automatic gc.
 173          */
 174         if (gc_auto_threshold <= 0)
 175                 return 0;
 176
 177         /*
 178          * If there are too many loose objects, but not too many
 179          * packs, we run "repack -d -l".  If there are too many packs,
 180          * we run "repack -A -d -l".  Otherwise we tell the caller
 181          * there is no need.
 182          */
 183         if (too_many_packs())
 184                 add_repack_all_option();
 185         else if (!too_many_loose_objects())
 186                 return 0;
 187
 188         if (run_hook(NULL, "pre-auto-gc", NULL))
 189                 return 0;
 190         return 1;
 191 }
 192
 193 /* return NULL on success, else hostname running the gc */
 194 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
 195 {
 196         static struct lock_file lock;
 197         static char locking_host[128];
 198         char my_host[128];
 199         struct strbuf sb = STRBUF_INIT;
 200         struct stat st;
 201         uintmax_t pid;
 202         FILE *fp;
 203         int fd, should_exit;
 204
 205         if (pidfile)
 206                 /* already locked */
 207                 return NULL;
 208
 209         if (gethostname(my_host, sizeof(my_host)))
 210                 strcpy(my_host, "unknown");
 211
 212         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
 213                                        LOCK_DIE_ON_ERROR);
 214         if (!force) {
 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(&rerere, "rerere", "gc", NULL);
 297
 298         git_config(gc_config, NULL);
 299
 300         if (pack_refs < 0)
 301                 pack_refs = !is_bare_repository();
 302
 303         argc = parse_options(argc, argv, prefix, builtin_gc_options,
 304                              builtin_gc_usage, 0);
 305         if (argc > 0)
 306                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 307
 308         if (aggressive) {
 309                 argv_array_push(&repack, "-f");
 310                 argv_array_push(&repack, "--depth=250");
 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 (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
 365                 return error(FAILED_RUN, rerere.argv[0]);
 366
 367         if (auto_gc && too_many_loose_objects())
 368                 warning(_("There are too many unreachable loose objects; "
 369                         "run 'git prune' to remove them."));
 370
 371         return 0;
 372 }