Commit | Line | Data |
---|---|---|
6757ada4 JB |
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 | ||
baffc0e7 | 13 | #include "builtin.h" |
697cc8ef | 14 | #include "lockfile.h" |
44c637c8 | 15 | #include "parse-options.h" |
6757ada4 | 16 | #include "run-command.h" |
4c5baf02 | 17 | #include "sigchain.h" |
234587fc | 18 | #include "argv-array.h" |
eab3296c | 19 | #include "commit.h" |
6757ada4 JB |
20 | |
21 | #define FAILED_RUN "failed to run %s" | |
22 | ||
44c637c8 | 23 | static const char * const builtin_gc_usage[] = { |
6705c162 | 24 | N_("git gc [options]"), |
44c637c8 JB |
25 | NULL |
26 | }; | |
6757ada4 | 27 | |
56752391 | 28 | static int pack_refs = 1; |
62aad184 | 29 | static int prune_reflogs = 1; |
125f8146 | 30 | static int aggressive_depth = 250; |
1c192f34 | 31 | static int aggressive_window = 250; |
2c3c4399 | 32 | static int gc_auto_threshold = 6700; |
97063974 | 33 | static int gc_auto_pack_limit = 50; |
9f673f94 | 34 | static int detach_auto = 1; |
d3154b44 | 35 | static const char *prune_expire = "2.weeks.ago"; |
6757ada4 | 36 | |
234587fc JK |
37 | static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT; |
38 | static struct argv_array reflog = ARGV_ARRAY_INIT; | |
39 | static struct argv_array repack = ARGV_ARRAY_INIT; | |
40 | static struct argv_array prune = ARGV_ARRAY_INIT; | |
41 | static struct argv_array rerere = ARGV_ARRAY_INIT; | |
6757ada4 | 42 | |
4c5baf02 JN |
43 | static char *pidfile; |
44 | ||
45 | static void remove_pidfile(void) | |
46 | { | |
47 | if (pidfile) | |
48 | unlink(pidfile); | |
49 | } | |
50 | ||
51 | static void remove_pidfile_on_signal(int signo) | |
52 | { | |
53 | remove_pidfile(); | |
54 | sigchain_pop(signo); | |
55 | raise(signo); | |
56 | } | |
57 | ||
09dbb90b NTND |
58 | static void git_config_date_string(const char *key, const char **output) |
59 | { | |
60 | if (git_config_get_string_const(key, output)) | |
61 | return; | |
62 | if (strcmp(*output, "now")) { | |
63 | unsigned long now = approxidate("now"); | |
64 | if (approxidate(*output) >= now) | |
65 | git_die_config(key, _("Invalid %s: '%s'"), key, *output); | |
66 | } | |
67 | } | |
68 | ||
5801d3b4 | 69 | static void gc_config(void) |
6757ada4 | 70 | { |
5801d3b4 TA |
71 | const char *value; |
72 | ||
73 | if (!git_config_get_value("gc.packrefs", &value)) { | |
c5e5a2c0 | 74 | if (value && !strcmp(value, "notbare")) |
6757ada4 JB |
75 | pack_refs = -1; |
76 | else | |
5801d3b4 | 77 | pack_refs = git_config_bool("gc.packrefs", value); |
17815501 | 78 | } |
5801d3b4 TA |
79 | |
80 | git_config_get_int("gc.aggressivewindow", &aggressive_window); | |
81 | git_config_get_int("gc.aggressivedepth", &aggressive_depth); | |
82 | git_config_get_int("gc.auto", &gc_auto_threshold); | |
83 | git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit); | |
84 | git_config_get_bool("gc.autodetach", &detach_auto); | |
09dbb90b | 85 | git_config_date_string("gc.pruneexpire", &prune_expire); |
5801d3b4 | 86 | git_config(git_default_config, NULL); |
6757ada4 JB |
87 | } |
88 | ||
a087cc98 | 89 | static int too_many_loose_objects(void) |
2c3c4399 JH |
90 | { |
91 | /* | |
92 | * Quickly check if a "gc" is needed, by estimating how | |
93 | * many loose objects there are. Because SHA-1 is evenly | |
94 | * distributed, we can check only one and get a reasonable | |
95 | * estimate. | |
96 | */ | |
97 | char path[PATH_MAX]; | |
98 | const char *objdir = get_object_directory(); | |
99 | DIR *dir; | |
100 | struct dirent *ent; | |
101 | int auto_threshold; | |
102 | int num_loose = 0; | |
103 | int needed = 0; | |
104 | ||
17815501 JH |
105 | if (gc_auto_threshold <= 0) |
106 | return 0; | |
107 | ||
2c3c4399 | 108 | if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) { |
fea6128b | 109 | warning(_("insanely long object directory %.*s"), 50, objdir); |
2c3c4399 JH |
110 | return 0; |
111 | } | |
112 | dir = opendir(path); | |
113 | if (!dir) | |
114 | return 0; | |
115 | ||
116 | auto_threshold = (gc_auto_threshold + 255) / 256; | |
117 | while ((ent = readdir(dir)) != NULL) { | |
118 | if (strspn(ent->d_name, "0123456789abcdef") != 38 || | |
119 | ent->d_name[38] != '\0') | |
120 | continue; | |
121 | if (++num_loose > auto_threshold) { | |
122 | needed = 1; | |
123 | break; | |
124 | } | |
125 | } | |
126 | closedir(dir); | |
127 | return needed; | |
128 | } | |
129 | ||
17815501 JH |
130 | static int too_many_packs(void) |
131 | { | |
132 | struct packed_git *p; | |
133 | int cnt; | |
134 | ||
135 | if (gc_auto_pack_limit <= 0) | |
136 | return 0; | |
137 | ||
138 | prepare_packed_git(); | |
139 | for (cnt = 0, p = packed_git; p; p = p->next) { | |
17815501 JH |
140 | if (!p->pack_local) |
141 | continue; | |
01af249f | 142 | if (p->pack_keep) |
17815501 JH |
143 | continue; |
144 | /* | |
145 | * Perhaps check the size of the pack and count only | |
146 | * very small ones here? | |
147 | */ | |
148 | cnt++; | |
149 | } | |
150 | return gc_auto_pack_limit <= cnt; | |
151 | } | |
152 | ||
7e52f566 JK |
153 | static void add_repack_all_option(void) |
154 | { | |
155 | if (prune_expire && !strcmp(prune_expire, "now")) | |
234587fc | 156 | argv_array_push(&repack, "-a"); |
7e52f566 | 157 | else { |
234587fc JK |
158 | argv_array_push(&repack, "-A"); |
159 | if (prune_expire) | |
160 | argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire); | |
7e52f566 JK |
161 | } |
162 | } | |
163 | ||
a087cc98 JH |
164 | static int need_to_gc(void) |
165 | { | |
166 | /* | |
b14d255b BC |
167 | * Setting gc.auto to 0 or negative can disable the |
168 | * automatic gc. | |
a087cc98 | 169 | */ |
b14d255b | 170 | if (gc_auto_threshold <= 0) |
95143f9e JH |
171 | return 0; |
172 | ||
17815501 JH |
173 | /* |
174 | * If there are too many loose objects, but not too many | |
175 | * packs, we run "repack -d -l". If there are too many packs, | |
176 | * we run "repack -A -d -l". Otherwise we tell the caller | |
177 | * there is no need. | |
178 | */ | |
17815501 | 179 | if (too_many_packs()) |
7e52f566 | 180 | add_repack_all_option(); |
17815501 JH |
181 | else if (!too_many_loose_objects()) |
182 | return 0; | |
bde30540 | 183 | |
15048f8a | 184 | if (run_hook_le(NULL, "pre-auto-gc", NULL)) |
bde30540 | 185 | return 0; |
95143f9e | 186 | return 1; |
a087cc98 JH |
187 | } |
188 | ||
64a99eb4 NTND |
189 | /* return NULL on success, else hostname running the gc */ |
190 | static const char *lock_repo_for_gc(int force, pid_t* ret_pid) | |
191 | { | |
192 | static struct lock_file lock; | |
64a99eb4 NTND |
193 | char my_host[128]; |
194 | struct strbuf sb = STRBUF_INIT; | |
195 | struct stat st; | |
196 | uintmax_t pid; | |
197 | FILE *fp; | |
4f1c0b21 | 198 | int fd; |
64a99eb4 | 199 | |
4c5baf02 JN |
200 | if (pidfile) |
201 | /* already locked */ | |
202 | return NULL; | |
203 | ||
64a99eb4 NTND |
204 | if (gethostname(my_host, sizeof(my_host))) |
205 | strcpy(my_host, "unknown"); | |
206 | ||
207 | fd = hold_lock_file_for_update(&lock, git_path("gc.pid"), | |
208 | LOCK_DIE_ON_ERROR); | |
209 | if (!force) { | |
4f1c0b21 EP |
210 | static char locking_host[128]; |
211 | int should_exit; | |
64a99eb4 NTND |
212 | fp = fopen(git_path("gc.pid"), "r"); |
213 | memset(locking_host, 0, sizeof(locking_host)); | |
214 | should_exit = | |
215 | fp != NULL && | |
216 | !fstat(fileno(fp), &st) && | |
217 | /* | |
218 | * 12 hour limit is very generous as gc should | |
219 | * never take that long. On the other hand we | |
220 | * don't really need a strict limit here, | |
221 | * running gc --auto one day late is not a big | |
222 | * problem. --force can be used in manual gc | |
223 | * after the user verifies that no gc is | |
224 | * running. | |
225 | */ | |
226 | time(NULL) - st.st_mtime <= 12 * 3600 && | |
227 | fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 && | |
228 | /* be gentle to concurrent "gc" on remote hosts */ | |
ed7eda8b | 229 | (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM); |
64a99eb4 NTND |
230 | if (fp != NULL) |
231 | fclose(fp); | |
232 | if (should_exit) { | |
233 | if (fd >= 0) | |
234 | rollback_lock_file(&lock); | |
235 | *ret_pid = pid; | |
236 | return locking_host; | |
237 | } | |
238 | } | |
239 | ||
240 | strbuf_addf(&sb, "%"PRIuMAX" %s", | |
241 | (uintmax_t) getpid(), my_host); | |
242 | write_in_full(fd, sb.buf, sb.len); | |
243 | strbuf_release(&sb); | |
244 | commit_lock_file(&lock); | |
245 | ||
4c5baf02 JN |
246 | pidfile = git_pathdup("gc.pid"); |
247 | sigchain_push_common(remove_pidfile_on_signal); | |
248 | atexit(remove_pidfile); | |
249 | ||
64a99eb4 NTND |
250 | return NULL; |
251 | } | |
252 | ||
62aad184 NTND |
253 | static int gc_before_repack(void) |
254 | { | |
255 | if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD)) | |
256 | return error(FAILED_RUN, pack_refs_cmd.argv[0]); | |
257 | ||
258 | if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD)) | |
259 | return error(FAILED_RUN, reflog.argv[0]); | |
260 | ||
261 | pack_refs = 0; | |
262 | prune_reflogs = 0; | |
263 | return 0; | |
264 | } | |
265 | ||
6757ada4 JB |
266 | int cmd_gc(int argc, const char **argv, const char *prefix) |
267 | { | |
44c637c8 | 268 | int aggressive = 0; |
2c3c4399 | 269 | int auto_gc = 0; |
a0c14cbb | 270 | int quiet = 0; |
64a99eb4 NTND |
271 | int force = 0; |
272 | const char *name; | |
273 | pid_t pid; | |
6757ada4 | 274 | |
44c637c8 | 275 | struct option builtin_gc_options[] = { |
6705c162 NTND |
276 | OPT__QUIET(&quiet, N_("suppress progress reporting")), |
277 | { OPTION_STRING, 0, "prune", &prune_expire, N_("date"), | |
278 | N_("prune unreferenced objects"), | |
58e9d9d4 | 279 | PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire }, |
d5d09d47 SB |
280 | OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")), |
281 | OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")), | |
64a99eb4 | 282 | OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")), |
44c637c8 JB |
283 | OPT_END() |
284 | }; | |
285 | ||
0c8151b6 NTND |
286 | if (argc == 2 && !strcmp(argv[1], "-h")) |
287 | usage_with_options(builtin_gc_usage, builtin_gc_options); | |
288 | ||
234587fc JK |
289 | argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL); |
290 | argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL); | |
291 | argv_array_pushl(&repack, "repack", "-d", "-l", NULL); | |
2cfe2a78 | 292 | argv_array_pushl(&prune, "prune", "--expire", NULL); |
234587fc JK |
293 | argv_array_pushl(&rerere, "rerere", "gc", NULL); |
294 | ||
5801d3b4 | 295 | gc_config(); |
6757ada4 JB |
296 | |
297 | if (pack_refs < 0) | |
298 | pack_refs = !is_bare_repository(); | |
299 | ||
37782920 SB |
300 | argc = parse_options(argc, argv, prefix, builtin_gc_options, |
301 | builtin_gc_usage, 0); | |
44c637c8 JB |
302 | if (argc > 0) |
303 | usage_with_options(builtin_gc_usage, builtin_gc_options); | |
304 | ||
305 | if (aggressive) { | |
234587fc | 306 | argv_array_push(&repack, "-f"); |
125f8146 NTND |
307 | if (aggressive_depth > 0) |
308 | argv_array_pushf(&repack, "--depth=%d", aggressive_depth); | |
234587fc JK |
309 | if (aggressive_window > 0) |
310 | argv_array_pushf(&repack, "--window=%d", aggressive_window); | |
6757ada4 | 311 | } |
a0c14cbb | 312 | if (quiet) |
234587fc | 313 | argv_array_push(&repack, "-q"); |
6757ada4 | 314 | |
2c3c4399 JH |
315 | if (auto_gc) { |
316 | /* | |
317 | * Auto-gc should be least intrusive as possible. | |
318 | */ | |
2c3c4399 JH |
319 | if (!need_to_gc()) |
320 | return 0; | |
9f673f94 NTND |
321 | if (!quiet) { |
322 | if (detach_auto) | |
323 | fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n")); | |
324 | else | |
325 | fprintf(stderr, _("Auto packing the repository for optimum performance.\n")); | |
326 | fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n")); | |
327 | } | |
62aad184 NTND |
328 | if (detach_auto) { |
329 | if (gc_before_repack()) | |
330 | return -1; | |
9f673f94 NTND |
331 | /* |
332 | * failure to daemonize is ok, we'll continue | |
333 | * in foreground | |
334 | */ | |
335 | daemonize(); | |
62aad184 | 336 | } |
a37cce3b | 337 | } else |
7e52f566 | 338 | add_repack_all_option(); |
2c3c4399 | 339 | |
64a99eb4 NTND |
340 | name = lock_repo_for_gc(force, &pid); |
341 | if (name) { | |
342 | if (auto_gc) | |
343 | return 0; /* be quiet on --auto */ | |
344 | die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"), | |
345 | name, (uintmax_t)pid); | |
346 | } | |
347 | ||
62aad184 NTND |
348 | if (gc_before_repack()) |
349 | return -1; | |
6757ada4 | 350 | |
234587fc JK |
351 | if (run_command_v_opt(repack.argv, RUN_GIT_CMD)) |
352 | return error(FAILED_RUN, repack.argv[0]); | |
6757ada4 | 353 | |
58e9d9d4 | 354 | if (prune_expire) { |
234587fc | 355 | argv_array_push(&prune, prune_expire); |
bf0a59b3 | 356 | if (quiet) |
234587fc JK |
357 | argv_array_push(&prune, "--no-progress"); |
358 | if (run_command_v_opt(prune.argv, RUN_GIT_CMD)) | |
359 | return error(FAILED_RUN, prune.argv[0]); | |
58e9d9d4 | 360 | } |
6757ada4 | 361 | |
234587fc JK |
362 | if (run_command_v_opt(rerere.argv, RUN_GIT_CMD)) |
363 | return error(FAILED_RUN, rerere.argv[0]); | |
6757ada4 | 364 | |
a087cc98 | 365 | if (auto_gc && too_many_loose_objects()) |
fea6128b ÆAB |
366 | warning(_("There are too many unreachable loose objects; " |
367 | "run 'git prune' to remove them.")); | |
a087cc98 | 368 | |
6757ada4 JB |
369 | return 0; |
370 | } |