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
  18 #define FAILED_RUN "failed to run %s"
  19
  20 static const char * const builtin_gc_usage[] = {
  21         "git-gc [options]",
  22         NULL
  23 };
  24
  25 static int pack_refs = 1;
  26 static int aggressive_window = -1;
  27 static int gc_auto_threshold = 6700;
  28 static int gc_auto_pack_limit = 50;
  29 static char *prune_expire = "2.weeks.ago";
  30
  31 #define MAX_ADD 10
  32 static const char *argv_pack_refs[] = {"pack-refs", "--all", "--prune", NULL};
  33 static const char *argv_reflog[] = {"reflog", "expire", "--all", NULL};
  34 static const char *argv_repack[MAX_ADD] = {"repack", "-d", "-l", NULL};
  35 static const char *argv_prune[] = {"prune", "--expire", NULL, NULL};
  36 static const char *argv_rerere[] = {"rerere", "gc", NULL};
  37
  38 static int gc_config(const char *var, const char *value)
  39 {
  40         if (!strcmp(var, "gc.packrefs")) {
  41                 if (value && !strcmp(value, "notbare"))
  42                         pack_refs = -1;
  43                 else
  44                         pack_refs = git_config_bool(var, value);
  45                 return 0;
  46         }
  47         if (!strcmp(var, "gc.aggressivewindow")) {
  48                 aggressive_window = git_config_int(var, value);
  49                 return 0;
  50         }
  51         if (!strcmp(var, "gc.auto")) {
  52                 gc_auto_threshold = git_config_int(var, value);
  53                 return 0;
  54         }
  55         if (!strcmp(var, "gc.autopacklimit")) {
  56                 gc_auto_pack_limit = git_config_int(var, value);
  57                 return 0;
  58         }
  59         if (!strcmp(var, "gc.pruneexpire")) {
  60                 if (!value)
  61                         return config_error_nonbool(var);
  62                 if (strcmp(value, "now")) {
  63                         unsigned long now = approxidate("now");
  64                         if (approxidate(value) >= now)
  65                                 return error("Invalid %s: '%s'", var, value);
  66                 }
  67                 prune_expire = xstrdup(value);
  68                 return 0;
  69         }
  70         return git_default_config(var, value);
  71 }
  72
  73 static void append_option(const char **cmd, const char *opt, int max_length)
  74 {
  75         int i;
  76
  77         for (i = 0; cmd[i]; i++)
  78                 ;
  79
  80         if (i + 2 >= max_length)
  81                 die("Too many options specified");
  82         cmd[i++] = opt;
  83         cmd[i] = NULL;
  84 }
  85
  86 static int too_many_loose_objects(void)
  87 {
  88         /*
  89          * Quickly check if a "gc" is needed, by estimating how
  90          * many loose objects there are.  Because SHA-1 is evenly
  91          * distributed, we can check only one and get a reasonable
  92          * estimate.
  93          */
  94         char path[PATH_MAX];
  95         const char *objdir = get_object_directory();
  96         DIR *dir;
  97         struct dirent *ent;
  98         int auto_threshold;
  99         int num_loose = 0;
 100         int needed = 0;
 101
 102         if (gc_auto_threshold <= 0)
 103                 return 0;
 104
 105         if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
 106                 warning("insanely long object directory %.*s", 50, objdir);
 107                 return 0;
 108         }
 109         dir = opendir(path);
 110         if (!dir)
 111                 return 0;
 112
 113         auto_threshold = (gc_auto_threshold + 255) / 256;
 114         while ((ent = readdir(dir)) != NULL) {
 115                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
 116                     ent->d_name[38] != '\0')
 117                         continue;
 118                 if (++num_loose > auto_threshold) {
 119                         needed = 1;
 120                         break;
 121                 }
 122         }
 123         closedir(dir);
 124         return needed;
 125 }
 126
 127 static int too_many_packs(void)
 128 {
 129         struct packed_git *p;
 130         int cnt;
 131
 132         if (gc_auto_pack_limit <= 0)
 133                 return 0;
 134
 135         prepare_packed_git();
 136         for (cnt = 0, p = packed_git; p; p = p->next) {
 137                 char path[PATH_MAX];
 138                 size_t len;
 139                 int keep;
 140
 141                 if (!p->pack_local)
 142                         continue;
 143                 len = strlen(p->pack_name);
 144                 if (PATH_MAX <= len + 1)
 145                         continue; /* oops, give up */
 146                 memcpy(path, p->pack_name, len-5);
 147                 memcpy(path + len - 5, ".keep", 6);
 148                 keep = access(p->pack_name, F_OK) && (errno == ENOENT);
 149                 if (keep)
 150                         continue;
 151                 /*
 152                  * Perhaps check the size of the pack and count only
 153                  * very small ones here?
 154                  */
 155                 cnt++;
 156         }
 157         return gc_auto_pack_limit <= cnt;
 158 }
 159
 160 static int need_to_gc(void)
 161 {
 162         /*
 163          * Setting gc.auto to 0 or negative can disable the
 164          * automatic gc.
 165          */
 166         if (gc_auto_threshold <= 0)
 167                 return 0;
 168
 169         /*
 170          * If there are too many loose objects, but not too many
 171          * packs, we run "repack -d -l".  If there are too many packs,
 172          * we run "repack -A -d -l".  Otherwise we tell the caller
 173          * there is no need.
 174          */
 175         if (too_many_packs())
 176                 append_option(argv_repack, "-A", MAX_ADD);
 177         else if (!too_many_loose_objects())
 178                 return 0;
 179         return 1;
 180 }
 181
 182 int cmd_gc(int argc, const char **argv, const char *prefix)
 183 {
 184         int prune = 0;
 185         int aggressive = 0;
 186         int auto_gc = 0;
 187         int quiet = 0;
 188         char buf[80];
 189
 190         struct option builtin_gc_options[] = {
 191                 OPT_BOOLEAN(0, "prune", &prune, "prune unreferenced objects"),
 192                 OPT_BOOLEAN(0, "aggressive", &aggressive, "be more thorough (increased runtime)"),
 193                 OPT_BOOLEAN(0, "auto", &auto_gc, "enable auto-gc mode"),
 194                 OPT_BOOLEAN('q', "quiet", &quiet, "suppress progress reports"),
 195                 OPT_END()
 196         };
 197
 198         git_config(gc_config);
 199
 200         if (pack_refs < 0)
 201                 pack_refs = !is_bare_repository();
 202
 203         argc = parse_options(argc, argv, builtin_gc_options, builtin_gc_usage, 0);
 204         if (argc > 0)
 205                 usage_with_options(builtin_gc_usage, builtin_gc_options);
 206
 207         if (aggressive) {
 208                 append_option(argv_repack, "-f", MAX_ADD);
 209                 if (aggressive_window > 0) {
 210                         sprintf(buf, "--window=%d", aggressive_window);
 211                         append_option(argv_repack, buf, MAX_ADD);
 212                 }
 213         }
 214         if (quiet)
 215                 append_option(argv_repack, "-q", MAX_ADD);
 216
 217         if (auto_gc) {
 218                 /*
 219                  * Auto-gc should be least intrusive as possible.
 220                  */
 221                 prune = 0;
 222                 if (!need_to_gc())
 223                         return 0;
 224                 fprintf(stderr, "Auto packing your repository for optimum "
 225                         "performance. You may also\n"
 226                         "run \"git gc\" manually. See "
 227                         "\"git help gc\" for more information.\n");
 228         } else {
 229                 /*
 230                  * Use safer (for shared repos) "-A" option to
 231                  * repack when not pruning. Auto-gc makes its
 232                  * own decision.
 233                  */
 234                 if (prune)
 235                         append_option(argv_repack, "-a", MAX_ADD);
 236                 else
 237                         append_option(argv_repack, "-A", MAX_ADD);
 238         }
 239
 240         if (pack_refs && run_command_v_opt(argv_pack_refs, RUN_GIT_CMD))
 241                 return error(FAILED_RUN, argv_pack_refs[0]);
 242
 243         if (run_command_v_opt(argv_reflog, RUN_GIT_CMD))
 244                 return error(FAILED_RUN, argv_reflog[0]);
 245
 246         if (run_command_v_opt(argv_repack, RUN_GIT_CMD))
 247                 return error(FAILED_RUN, argv_repack[0]);
 248
 249         argv_prune[2] = prune_expire;
 250         if (run_command_v_opt(argv_prune, RUN_GIT_CMD))
 251                 return error(FAILED_RUN, argv_prune[0]);
 252
 253         if (run_command_v_opt(argv_rerere, RUN_GIT_CMD))
 254                 return error(FAILED_RUN, argv_rerere[0]);
 255
 256         if (auto_gc && too_many_loose_objects())
 257                 warning("There are too many unreachable loose objects; "
 258                         "run 'git prune' to remove them.");
 259
 260         return 0;
 261 }