]>
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" |
a80d72db | 14 | #include "repository.h" |
b2141fc1 | 15 | #include "config.h" |
ebebeaea | 16 | #include "tempfile.h" |
697cc8ef | 17 | #include "lockfile.h" |
44c637c8 | 18 | #include "parse-options.h" |
6757ada4 | 19 | #include "run-command.h" |
4c5baf02 | 20 | #include "sigchain.h" |
dbbcd44f | 21 | #include "strvec.h" |
eab3296c | 22 | #include "commit.h" |
d5d5d7b6 | 23 | #include "commit-graph.h" |
0abe14f6 | 24 | #include "packfile.h" |
a80d72db | 25 | #include "object-store.h" |
9806f5a7 NTND |
26 | #include "pack.h" |
27 | #include "pack-objects.h" | |
28 | #include "blob.h" | |
29 | #include "tree.h" | |
b14ed5ad | 30 | #include "promisor-remote.h" |
4ddc79b2 | 31 | #include "refs.h" |
6757ada4 JB |
32 | |
33 | #define FAILED_RUN "failed to run %s" | |
34 | ||
44c637c8 | 35 | static const char * const builtin_gc_usage[] = { |
9c9b4f2f | 36 | N_("git gc [<options>]"), |
44c637c8 JB |
37 | NULL |
38 | }; | |
6757ada4 | 39 | |
56752391 | 40 | static int pack_refs = 1; |
62aad184 | 41 | static int prune_reflogs = 1; |
07e7dbf0 | 42 | static int aggressive_depth = 50; |
1c192f34 | 43 | static int aggressive_window = 250; |
2c3c4399 | 44 | static int gc_auto_threshold = 6700; |
97063974 | 45 | static int gc_auto_pack_limit = 50; |
9f673f94 | 46 | static int detach_auto = 1; |
dddbad72 | 47 | static timestamp_t gc_log_expire_time; |
a831c06a | 48 | static const char *gc_log_expire = "1.day.ago"; |
d3154b44 | 49 | static const char *prune_expire = "2.weeks.ago"; |
e3df33bb | 50 | static const char *prune_worktrees_expire = "3.months.ago"; |
55dfe13d | 51 | static unsigned long big_pack_threshold; |
9806f5a7 | 52 | static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE; |
6757ada4 | 53 | |
22f9b7f3 JK |
54 | static struct strvec pack_refs_cmd = STRVEC_INIT; |
55 | static struct strvec reflog = STRVEC_INIT; | |
56 | static struct strvec repack = STRVEC_INIT; | |
57 | static struct strvec prune = STRVEC_INIT; | |
58 | static struct strvec prune_worktrees = STRVEC_INIT; | |
59 | static struct strvec rerere = STRVEC_INIT; | |
6757ada4 | 60 | |
076aa2cb | 61 | static struct tempfile *pidfile; |
329e6e87 | 62 | static struct lock_file log_lock; |
4c5baf02 | 63 | |
478f34d2 DK |
64 | static struct string_list pack_garbage = STRING_LIST_INIT_DUP; |
65 | ||
66 | static void clean_pack_garbage(void) | |
67 | { | |
68 | int i; | |
69 | for (i = 0; i < pack_garbage.nr; i++) | |
70 | unlink_or_warn(pack_garbage.items[i].string); | |
71 | string_list_clear(&pack_garbage, 0); | |
72 | } | |
73 | ||
74 | static void report_pack_garbage(unsigned seen_bits, const char *path) | |
75 | { | |
76 | if (seen_bits == PACKDIR_FILE_IDX) | |
77 | string_list_append(&pack_garbage, path); | |
78 | } | |
79 | ||
329e6e87 NTND |
80 | static void process_log_file(void) |
81 | { | |
82 | struct stat st; | |
a831c06a DT |
83 | if (fstat(get_lock_file_fd(&log_lock), &st)) { |
84 | /* | |
85 | * Perhaps there was an i/o error or another | |
86 | * unlikely situation. Try to make a note of | |
87 | * this in gc.log along with any existing | |
88 | * messages. | |
89 | */ | |
90 | int saved_errno = errno; | |
91 | fprintf(stderr, _("Failed to fstat %s: %s"), | |
076aa2cb | 92 | get_tempfile_path(log_lock.tempfile), |
a831c06a DT |
93 | strerror(saved_errno)); |
94 | fflush(stderr); | |
329e6e87 | 95 | commit_lock_file(&log_lock); |
a831c06a DT |
96 | errno = saved_errno; |
97 | } else if (st.st_size) { | |
98 | /* There was some error recorded in the lock file */ | |
99 | commit_lock_file(&log_lock); | |
100 | } else { | |
101 | /* No error, clean up any old gc.log */ | |
102 | unlink(git_path("gc.log")); | |
329e6e87 | 103 | rollback_lock_file(&log_lock); |
a831c06a | 104 | } |
329e6e87 NTND |
105 | } |
106 | ||
107 | static void process_log_file_at_exit(void) | |
108 | { | |
109 | fflush(stderr); | |
110 | process_log_file(); | |
111 | } | |
112 | ||
113 | static void process_log_file_on_signal(int signo) | |
114 | { | |
115 | process_log_file(); | |
116 | sigchain_pop(signo); | |
117 | raise(signo); | |
118 | } | |
119 | ||
bf3d70fe ÆAB |
120 | static int gc_config_is_timestamp_never(const char *var) |
121 | { | |
122 | const char *value; | |
123 | timestamp_t expire; | |
124 | ||
125 | if (!git_config_get_value(var, &value) && value) { | |
126 | if (parse_expiry_date(value, &expire)) | |
127 | die(_("failed to parse '%s' value '%s'"), var, value); | |
128 | return expire == 0; | |
129 | } | |
130 | return 0; | |
131 | } | |
132 | ||
5801d3b4 | 133 | static void gc_config(void) |
6757ada4 | 134 | { |
5801d3b4 TA |
135 | const char *value; |
136 | ||
137 | if (!git_config_get_value("gc.packrefs", &value)) { | |
c5e5a2c0 | 138 | if (value && !strcmp(value, "notbare")) |
6757ada4 JB |
139 | pack_refs = -1; |
140 | else | |
5801d3b4 | 141 | pack_refs = git_config_bool("gc.packrefs", value); |
17815501 | 142 | } |
5801d3b4 | 143 | |
bf3d70fe ÆAB |
144 | if (gc_config_is_timestamp_never("gc.reflogexpire") && |
145 | gc_config_is_timestamp_never("gc.reflogexpireunreachable")) | |
146 | prune_reflogs = 0; | |
147 | ||
5801d3b4 TA |
148 | git_config_get_int("gc.aggressivewindow", &aggressive_window); |
149 | git_config_get_int("gc.aggressivedepth", &aggressive_depth); | |
150 | git_config_get_int("gc.auto", &gc_auto_threshold); | |
151 | git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit); | |
152 | git_config_get_bool("gc.autodetach", &detach_auto); | |
77d67977 CC |
153 | git_config_get_expiry("gc.pruneexpire", &prune_expire); |
154 | git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire); | |
94c9b5af | 155 | git_config_get_expiry("gc.logexpiry", &gc_log_expire); |
a831c06a | 156 | |
55dfe13d | 157 | git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold); |
9806f5a7 | 158 | git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size); |
55dfe13d | 159 | |
5801d3b4 | 160 | git_config(git_default_config, NULL); |
6757ada4 JB |
161 | } |
162 | ||
a087cc98 | 163 | static int too_many_loose_objects(void) |
2c3c4399 JH |
164 | { |
165 | /* | |
166 | * Quickly check if a "gc" is needed, by estimating how | |
167 | * many loose objects there are. Because SHA-1 is evenly | |
168 | * distributed, we can check only one and get a reasonable | |
169 | * estimate. | |
170 | */ | |
2c3c4399 JH |
171 | DIR *dir; |
172 | struct dirent *ent; | |
173 | int auto_threshold; | |
174 | int num_loose = 0; | |
175 | int needed = 0; | |
e5cdbd5f | 176 | const unsigned hexsz_loose = the_hash_algo->hexsz - 2; |
2c3c4399 | 177 | |
07af8891 | 178 | dir = opendir(git_path("objects/17")); |
2c3c4399 JH |
179 | if (!dir) |
180 | return 0; | |
181 | ||
42c78a21 | 182 | auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256); |
2c3c4399 | 183 | while ((ent = readdir(dir)) != NULL) { |
e5cdbd5f ÆAB |
184 | if (strspn(ent->d_name, "0123456789abcdef") != hexsz_loose || |
185 | ent->d_name[hexsz_loose] != '\0') | |
2c3c4399 JH |
186 | continue; |
187 | if (++num_loose > auto_threshold) { | |
188 | needed = 1; | |
189 | break; | |
190 | } | |
191 | } | |
192 | closedir(dir); | |
193 | return needed; | |
194 | } | |
195 | ||
9806f5a7 NTND |
196 | static struct packed_git *find_base_packs(struct string_list *packs, |
197 | unsigned long limit) | |
ae4e89e5 NTND |
198 | { |
199 | struct packed_git *p, *base = NULL; | |
200 | ||
454ea2e4 | 201 | for (p = get_all_packs(the_repository); p; p = p->next) { |
ae4e89e5 NTND |
202 | if (!p->pack_local) |
203 | continue; | |
55dfe13d NTND |
204 | if (limit) { |
205 | if (p->pack_size >= limit) | |
206 | string_list_append(packs, p->pack_name); | |
207 | } else if (!base || base->pack_size < p->pack_size) { | |
ae4e89e5 NTND |
208 | base = p; |
209 | } | |
210 | } | |
211 | ||
212 | if (base) | |
213 | string_list_append(packs, base->pack_name); | |
9806f5a7 NTND |
214 | |
215 | return base; | |
ae4e89e5 NTND |
216 | } |
217 | ||
17815501 JH |
218 | static int too_many_packs(void) |
219 | { | |
220 | struct packed_git *p; | |
221 | int cnt; | |
222 | ||
223 | if (gc_auto_pack_limit <= 0) | |
224 | return 0; | |
225 | ||
454ea2e4 | 226 | for (cnt = 0, p = get_all_packs(the_repository); p; p = p->next) { |
17815501 JH |
227 | if (!p->pack_local) |
228 | continue; | |
01af249f | 229 | if (p->pack_keep) |
17815501 JH |
230 | continue; |
231 | /* | |
232 | * Perhaps check the size of the pack and count only | |
233 | * very small ones here? | |
234 | */ | |
235 | cnt++; | |
236 | } | |
5f4e3bf5 | 237 | return gc_auto_pack_limit < cnt; |
17815501 JH |
238 | } |
239 | ||
9806f5a7 NTND |
240 | static uint64_t total_ram(void) |
241 | { | |
242 | #if defined(HAVE_SYSINFO) | |
243 | struct sysinfo si; | |
244 | ||
245 | if (!sysinfo(&si)) | |
246 | return si.totalram; | |
247 | #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM)) | |
248 | int64_t physical_memory; | |
249 | int mib[2]; | |
250 | size_t length; | |
251 | ||
252 | mib[0] = CTL_HW; | |
253 | # if defined(HW_MEMSIZE) | |
254 | mib[1] = HW_MEMSIZE; | |
255 | # else | |
256 | mib[1] = HW_PHYSMEM; | |
257 | # endif | |
258 | length = sizeof(int64_t); | |
259 | if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0)) | |
260 | return physical_memory; | |
261 | #elif defined(GIT_WINDOWS_NATIVE) | |
262 | MEMORYSTATUSEX memInfo; | |
263 | ||
264 | memInfo.dwLength = sizeof(MEMORYSTATUSEX); | |
265 | if (GlobalMemoryStatusEx(&memInfo)) | |
266 | return memInfo.ullTotalPhys; | |
267 | #endif | |
268 | return 0; | |
269 | } | |
270 | ||
271 | static uint64_t estimate_repack_memory(struct packed_git *pack) | |
272 | { | |
273 | unsigned long nr_objects = approximate_object_count(); | |
274 | size_t os_cache, heap; | |
275 | ||
276 | if (!pack || !nr_objects) | |
277 | return 0; | |
278 | ||
279 | /* | |
280 | * First we have to scan through at least one pack. | |
281 | * Assume enough room in OS file cache to keep the entire pack | |
282 | * or we may accidentally evict data of other processes from | |
283 | * the cache. | |
284 | */ | |
285 | os_cache = pack->pack_size + pack->index_size; | |
286 | /* then pack-objects needs lots more for book keeping */ | |
287 | heap = sizeof(struct object_entry) * nr_objects; | |
288 | /* | |
289 | * internal rev-list --all --objects takes up some memory too, | |
290 | * let's say half of it is for blobs | |
291 | */ | |
292 | heap += sizeof(struct blob) * nr_objects / 2; | |
293 | /* | |
294 | * and the other half is for trees (commits and tags are | |
295 | * usually insignificant) | |
296 | */ | |
297 | heap += sizeof(struct tree) * nr_objects / 2; | |
298 | /* and then obj_hash[], underestimated in fact */ | |
299 | heap += sizeof(struct object *) * nr_objects; | |
300 | /* revindex is used also */ | |
301 | heap += sizeof(struct revindex_entry) * nr_objects; | |
302 | /* | |
303 | * read_sha1_file() (either at delta calculation phase, or | |
304 | * writing phase) also fills up the delta base cache | |
305 | */ | |
306 | heap += delta_base_cache_limit; | |
307 | /* and of course pack-objects has its own delta cache */ | |
308 | heap += max_delta_cache_size; | |
309 | ||
310 | return os_cache + heap; | |
311 | } | |
312 | ||
ae4e89e5 NTND |
313 | static int keep_one_pack(struct string_list_item *item, void *data) |
314 | { | |
22f9b7f3 | 315 | strvec_pushf(&repack, "--keep-pack=%s", basename(item->string)); |
ae4e89e5 NTND |
316 | return 0; |
317 | } | |
318 | ||
319 | static void add_repack_all_option(struct string_list *keep_pack) | |
7e52f566 JK |
320 | { |
321 | if (prune_expire && !strcmp(prune_expire, "now")) | |
22f9b7f3 | 322 | strvec_push(&repack, "-a"); |
7e52f566 | 323 | else { |
22f9b7f3 | 324 | strvec_push(&repack, "-A"); |
234587fc | 325 | if (prune_expire) |
22f9b7f3 | 326 | strvec_pushf(&repack, "--unpack-unreachable=%s", prune_expire); |
7e52f566 | 327 | } |
ae4e89e5 NTND |
328 | |
329 | if (keep_pack) | |
330 | for_each_string_list(keep_pack, keep_one_pack, NULL); | |
7e52f566 JK |
331 | } |
332 | ||
bdf56de8 DT |
333 | static void add_repack_incremental_option(void) |
334 | { | |
22f9b7f3 | 335 | strvec_push(&repack, "--no-write-bitmap-index"); |
bdf56de8 DT |
336 | } |
337 | ||
a087cc98 JH |
338 | static int need_to_gc(void) |
339 | { | |
340 | /* | |
b14d255b BC |
341 | * Setting gc.auto to 0 or negative can disable the |
342 | * automatic gc. | |
a087cc98 | 343 | */ |
b14d255b | 344 | if (gc_auto_threshold <= 0) |
95143f9e JH |
345 | return 0; |
346 | ||
17815501 JH |
347 | /* |
348 | * If there are too many loose objects, but not too many | |
349 | * packs, we run "repack -d -l". If there are too many packs, | |
350 | * we run "repack -A -d -l". Otherwise we tell the caller | |
351 | * there is no need. | |
352 | */ | |
55dfe13d NTND |
353 | if (too_many_packs()) { |
354 | struct string_list keep_pack = STRING_LIST_INIT_NODUP; | |
355 | ||
8fc67762 | 356 | if (big_pack_threshold) { |
55dfe13d | 357 | find_base_packs(&keep_pack, big_pack_threshold); |
8fc67762 NTND |
358 | if (keep_pack.nr >= gc_auto_pack_limit) { |
359 | big_pack_threshold = 0; | |
360 | string_list_clear(&keep_pack, 0); | |
361 | find_base_packs(&keep_pack, 0); | |
362 | } | |
9806f5a7 NTND |
363 | } else { |
364 | struct packed_git *p = find_base_packs(&keep_pack, 0); | |
365 | uint64_t mem_have, mem_want; | |
366 | ||
367 | mem_have = total_ram(); | |
368 | mem_want = estimate_repack_memory(p); | |
369 | ||
370 | /* | |
371 | * Only allow 1/2 of memory for pack-objects, leave | |
372 | * the rest for the OS and other processes in the | |
373 | * system. | |
374 | */ | |
375 | if (!mem_have || mem_want < mem_have / 2) | |
376 | string_list_clear(&keep_pack, 0); | |
8fc67762 | 377 | } |
55dfe13d NTND |
378 | |
379 | add_repack_all_option(&keep_pack); | |
380 | string_list_clear(&keep_pack, 0); | |
381 | } else if (too_many_loose_objects()) | |
bdf56de8 DT |
382 | add_repack_incremental_option(); |
383 | else | |
17815501 | 384 | return 0; |
bde30540 | 385 | |
15048f8a | 386 | if (run_hook_le(NULL, "pre-auto-gc", NULL)) |
bde30540 | 387 | return 0; |
95143f9e | 388 | return 1; |
a087cc98 JH |
389 | } |
390 | ||
64a99eb4 NTND |
391 | /* return NULL on success, else hostname running the gc */ |
392 | static const char *lock_repo_for_gc(int force, pid_t* ret_pid) | |
393 | { | |
b2275868 | 394 | struct lock_file lock = LOCK_INIT; |
da25bdb7 | 395 | char my_host[HOST_NAME_MAX + 1]; |
64a99eb4 NTND |
396 | struct strbuf sb = STRBUF_INIT; |
397 | struct stat st; | |
398 | uintmax_t pid; | |
399 | FILE *fp; | |
4f1c0b21 | 400 | int fd; |
00539cef | 401 | char *pidfile_path; |
64a99eb4 | 402 | |
076aa2cb | 403 | if (is_tempfile_active(pidfile)) |
4c5baf02 JN |
404 | /* already locked */ |
405 | return NULL; | |
406 | ||
5781a9a2 | 407 | if (xgethostname(my_host, sizeof(my_host))) |
5096d490 | 408 | xsnprintf(my_host, sizeof(my_host), "unknown"); |
64a99eb4 | 409 | |
00539cef MH |
410 | pidfile_path = git_pathdup("gc.pid"); |
411 | fd = hold_lock_file_for_update(&lock, pidfile_path, | |
64a99eb4 NTND |
412 | LOCK_DIE_ON_ERROR); |
413 | if (!force) { | |
da25bdb7 RS |
414 | static char locking_host[HOST_NAME_MAX + 1]; |
415 | static char *scan_fmt; | |
4f1c0b21 | 416 | int should_exit; |
da25bdb7 RS |
417 | |
418 | if (!scan_fmt) | |
afe2fab7 | 419 | scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX); |
00539cef | 420 | fp = fopen(pidfile_path, "r"); |
64a99eb4 NTND |
421 | memset(locking_host, 0, sizeof(locking_host)); |
422 | should_exit = | |
423 | fp != NULL && | |
424 | !fstat(fileno(fp), &st) && | |
425 | /* | |
426 | * 12 hour limit is very generous as gc should | |
427 | * never take that long. On the other hand we | |
428 | * don't really need a strict limit here, | |
429 | * running gc --auto one day late is not a big | |
430 | * problem. --force can be used in manual gc | |
431 | * after the user verifies that no gc is | |
432 | * running. | |
433 | */ | |
434 | time(NULL) - st.st_mtime <= 12 * 3600 && | |
da25bdb7 | 435 | fscanf(fp, scan_fmt, &pid, locking_host) == 2 && |
64a99eb4 | 436 | /* be gentle to concurrent "gc" on remote hosts */ |
ed7eda8b | 437 | (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM); |
64a99eb4 NTND |
438 | if (fp != NULL) |
439 | fclose(fp); | |
440 | if (should_exit) { | |
441 | if (fd >= 0) | |
442 | rollback_lock_file(&lock); | |
443 | *ret_pid = pid; | |
00539cef | 444 | free(pidfile_path); |
64a99eb4 NTND |
445 | return locking_host; |
446 | } | |
447 | } | |
448 | ||
449 | strbuf_addf(&sb, "%"PRIuMAX" %s", | |
450 | (uintmax_t) getpid(), my_host); | |
451 | write_in_full(fd, sb.buf, sb.len); | |
452 | strbuf_release(&sb); | |
453 | commit_lock_file(&lock); | |
076aa2cb | 454 | pidfile = register_tempfile(pidfile_path); |
ebebeaea | 455 | free(pidfile_path); |
64a99eb4 NTND |
456 | return NULL; |
457 | } | |
458 | ||
30299702 JN |
459 | /* |
460 | * Returns 0 if there was no previous error and gc can proceed, 1 if | |
461 | * gc should not proceed due to an error in the last run. Prints a | |
15beaaa3 | 462 | * message and returns -1 if an error occurred while reading gc.log |
30299702 JN |
463 | */ |
464 | static int report_last_gc_error(void) | |
329e6e87 NTND |
465 | { |
466 | struct strbuf sb = STRBUF_INIT; | |
30299702 | 467 | int ret = 0; |
3c426ecc | 468 | ssize_t len; |
a831c06a DT |
469 | struct stat st; |
470 | char *gc_log_path = git_pathdup("gc.log"); | |
329e6e87 | 471 | |
a831c06a DT |
472 | if (stat(gc_log_path, &st)) { |
473 | if (errno == ENOENT) | |
474 | goto done; | |
475 | ||
30299702 JN |
476 | ret = error_errno(_("cannot stat '%s'"), gc_log_path); |
477 | goto done; | |
a831c06a DT |
478 | } |
479 | ||
480 | if (st.st_mtime < gc_log_expire_time) | |
481 | goto done; | |
482 | ||
3c426ecc JN |
483 | len = strbuf_read_file(&sb, gc_log_path, 0); |
484 | if (len < 0) | |
30299702 JN |
485 | ret = error_errno(_("cannot read '%s'"), gc_log_path); |
486 | else if (len > 0) { | |
487 | /* | |
488 | * A previous gc failed. Report the error, and don't | |
489 | * bother with an automatic gc run since it is likely | |
490 | * to fail in the same way. | |
491 | */ | |
492 | warning(_("The last gc run reported the following. " | |
329e6e87 NTND |
493 | "Please correct the root cause\n" |
494 | "and remove %s.\n" | |
495 | "Automatic cleanup will not be performed " | |
496 | "until the file is removed.\n\n" | |
497 | "%s"), | |
a831c06a | 498 | gc_log_path, sb.buf); |
30299702 JN |
499 | ret = 1; |
500 | } | |
329e6e87 | 501 | strbuf_release(&sb); |
a831c06a DT |
502 | done: |
503 | free(gc_log_path); | |
30299702 | 504 | return ret; |
329e6e87 NTND |
505 | } |
506 | ||
fec2ed21 | 507 | static void gc_before_repack(void) |
62aad184 | 508 | { |
cd8eb3a0 ÆAB |
509 | /* |
510 | * We may be called twice, as both the pre- and | |
511 | * post-daemonized phases will call us, but running these | |
512 | * commands more than once is pointless and wasteful. | |
513 | */ | |
514 | static int done = 0; | |
515 | if (done++) | |
516 | return; | |
517 | ||
d70a9eb6 JK |
518 | if (pack_refs && run_command_v_opt(pack_refs_cmd.v, RUN_GIT_CMD)) |
519 | die(FAILED_RUN, pack_refs_cmd.v[0]); | |
62aad184 | 520 | |
d70a9eb6 JK |
521 | if (prune_reflogs && run_command_v_opt(reflog.v, RUN_GIT_CMD)) |
522 | die(FAILED_RUN, reflog.v[0]); | |
62aad184 NTND |
523 | } |
524 | ||
6757ada4 JB |
525 | int cmd_gc(int argc, const char **argv, const char *prefix) |
526 | { | |
44c637c8 | 527 | int aggressive = 0; |
2c3c4399 | 528 | int auto_gc = 0; |
a0c14cbb | 529 | int quiet = 0; |
64a99eb4 NTND |
530 | int force = 0; |
531 | const char *name; | |
532 | pid_t pid; | |
329e6e87 | 533 | int daemonized = 0; |
ae4e89e5 | 534 | int keep_base_pack = -1; |
8ab5aa4b | 535 | timestamp_t dummy; |
6757ada4 | 536 | |
44c637c8 | 537 | struct option builtin_gc_options[] = { |
6705c162 NTND |
538 | OPT__QUIET(&quiet, N_("suppress progress reporting")), |
539 | { OPTION_STRING, 0, "prune", &prune_expire, N_("date"), | |
540 | N_("prune unreferenced objects"), | |
58e9d9d4 | 541 | PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire }, |
d5d09d47 | 542 | OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")), |
7e1eeaa4 NTND |
543 | OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"), |
544 | PARSE_OPT_NOCOMPLETE), | |
545 | OPT_BOOL_F(0, "force", &force, | |
546 | N_("force running gc even if there may be another gc running"), | |
547 | PARSE_OPT_NOCOMPLETE), | |
ae4e89e5 NTND |
548 | OPT_BOOL(0, "keep-largest-pack", &keep_base_pack, |
549 | N_("repack all other packs except the largest pack")), | |
44c637c8 JB |
550 | OPT_END() |
551 | }; | |
552 | ||
0c8151b6 NTND |
553 | if (argc == 2 && !strcmp(argv[1], "-h")) |
554 | usage_with_options(builtin_gc_usage, builtin_gc_options); | |
555 | ||
22f9b7f3 JK |
556 | strvec_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL); |
557 | strvec_pushl(&reflog, "reflog", "expire", "--all", NULL); | |
558 | strvec_pushl(&repack, "repack", "-d", "-l", NULL); | |
559 | strvec_pushl(&prune, "prune", "--expire", NULL); | |
560 | strvec_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL); | |
561 | strvec_pushl(&rerere, "rerere", "gc", NULL); | |
234587fc | 562 | |
a831c06a | 563 | /* default expiry time, overwritten in gc_config */ |
5801d3b4 | 564 | gc_config(); |
a831c06a | 565 | if (parse_expiry_date(gc_log_expire, &gc_log_expire_time)) |
96913c9d | 566 | die(_("failed to parse gc.logexpiry value %s"), gc_log_expire); |
6757ada4 JB |
567 | |
568 | if (pack_refs < 0) | |
569 | pack_refs = !is_bare_repository(); | |
570 | ||
37782920 SB |
571 | argc = parse_options(argc, argv, prefix, builtin_gc_options, |
572 | builtin_gc_usage, 0); | |
44c637c8 JB |
573 | if (argc > 0) |
574 | usage_with_options(builtin_gc_usage, builtin_gc_options); | |
575 | ||
8ab5aa4b JH |
576 | if (prune_expire && parse_expiry_date(prune_expire, &dummy)) |
577 | die(_("failed to parse prune expiry value %s"), prune_expire); | |
578 | ||
44c637c8 | 579 | if (aggressive) { |
22f9b7f3 | 580 | strvec_push(&repack, "-f"); |
125f8146 | 581 | if (aggressive_depth > 0) |
22f9b7f3 | 582 | strvec_pushf(&repack, "--depth=%d", aggressive_depth); |
234587fc | 583 | if (aggressive_window > 0) |
22f9b7f3 | 584 | strvec_pushf(&repack, "--window=%d", aggressive_window); |
6757ada4 | 585 | } |
a0c14cbb | 586 | if (quiet) |
22f9b7f3 | 587 | strvec_push(&repack, "-q"); |
6757ada4 | 588 | |
2c3c4399 JH |
589 | if (auto_gc) { |
590 | /* | |
591 | * Auto-gc should be least intrusive as possible. | |
592 | */ | |
2c3c4399 JH |
593 | if (!need_to_gc()) |
594 | return 0; | |
9f673f94 NTND |
595 | if (!quiet) { |
596 | if (detach_auto) | |
597 | fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n")); | |
598 | else | |
599 | fprintf(stderr, _("Auto packing the repository for optimum performance.\n")); | |
600 | fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n")); | |
601 | } | |
62aad184 | 602 | if (detach_auto) { |
30299702 JN |
603 | int ret = report_last_gc_error(); |
604 | if (ret < 0) | |
15beaaa3 | 605 | /* an I/O error occurred, already reported */ |
30299702 JN |
606 | exit(128); |
607 | if (ret == 1) | |
608 | /* Last gc --auto failed. Skip this one. */ | |
609 | return 0; | |
329e6e87 | 610 | |
c45af94d JK |
611 | if (lock_repo_for_gc(force, &pid)) |
612 | return 0; | |
fec2ed21 | 613 | gc_before_repack(); /* dies on failure */ |
c45af94d JK |
614 | delete_tempfile(&pidfile); |
615 | ||
9f673f94 NTND |
616 | /* |
617 | * failure to daemonize is ok, we'll continue | |
618 | * in foreground | |
619 | */ | |
329e6e87 | 620 | daemonized = !daemonize(); |
62aad184 | 621 | } |
ae4e89e5 NTND |
622 | } else { |
623 | struct string_list keep_pack = STRING_LIST_INIT_NODUP; | |
624 | ||
625 | if (keep_base_pack != -1) { | |
626 | if (keep_base_pack) | |
55dfe13d NTND |
627 | find_base_packs(&keep_pack, 0); |
628 | } else if (big_pack_threshold) { | |
629 | find_base_packs(&keep_pack, big_pack_threshold); | |
ae4e89e5 NTND |
630 | } |
631 | ||
632 | add_repack_all_option(&keep_pack); | |
633 | string_list_clear(&keep_pack, 0); | |
634 | } | |
2c3c4399 | 635 | |
64a99eb4 NTND |
636 | name = lock_repo_for_gc(force, &pid); |
637 | if (name) { | |
638 | if (auto_gc) | |
639 | return 0; /* be quiet on --auto */ | |
640 | die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"), | |
641 | name, (uintmax_t)pid); | |
642 | } | |
643 | ||
329e6e87 NTND |
644 | if (daemonized) { |
645 | hold_lock_file_for_update(&log_lock, | |
646 | git_path("gc.log"), | |
647 | LOCK_DIE_ON_ERROR); | |
076c8278 | 648 | dup2(get_lock_file_fd(&log_lock), 2); |
329e6e87 NTND |
649 | sigchain_push_common(process_log_file_on_signal); |
650 | atexit(process_log_file_at_exit); | |
651 | } | |
652 | ||
fec2ed21 | 653 | gc_before_repack(); |
6757ada4 | 654 | |
067fbd41 | 655 | if (!repository_format_precious_objects) { |
2d511cfc | 656 | close_object_store(the_repository->objects); |
d70a9eb6 JK |
657 | if (run_command_v_opt(repack.v, RUN_GIT_CMD)) |
658 | die(FAILED_RUN, repack.v[0]); | |
067fbd41 JK |
659 | |
660 | if (prune_expire) { | |
22f9b7f3 | 661 | strvec_push(&prune, prune_expire); |
067fbd41 | 662 | if (quiet) |
22f9b7f3 | 663 | strvec_push(&prune, "--no-progress"); |
b14ed5ad | 664 | if (has_promisor_remote()) |
22f9b7f3 | 665 | strvec_push(&prune, |
f6d8942b | 666 | "--exclude-promisor-objects"); |
d70a9eb6 JK |
667 | if (run_command_v_opt(prune.v, RUN_GIT_CMD)) |
668 | die(FAILED_RUN, prune.v[0]); | |
067fbd41 | 669 | } |
58e9d9d4 | 670 | } |
6757ada4 | 671 | |
e3df33bb | 672 | if (prune_worktrees_expire) { |
22f9b7f3 | 673 | strvec_push(&prune_worktrees, prune_worktrees_expire); |
d70a9eb6 JK |
674 | if (run_command_v_opt(prune_worktrees.v, RUN_GIT_CMD)) |
675 | die(FAILED_RUN, prune_worktrees.v[0]); | |
e3df33bb NTND |
676 | } |
677 | ||
d70a9eb6 JK |
678 | if (run_command_v_opt(rerere.v, RUN_GIT_CMD)) |
679 | die(FAILED_RUN, rerere.v[0]); | |
6757ada4 | 680 | |
478f34d2 | 681 | report_garbage = report_pack_garbage; |
a49d2834 | 682 | reprepare_packed_git(the_repository); |
5bdece0d | 683 | if (pack_garbage.nr > 0) { |
2d511cfc | 684 | close_object_store(the_repository->objects); |
478f34d2 | 685 | clean_pack_garbage(); |
5bdece0d | 686 | } |
478f34d2 | 687 | |
7211b9e7 DS |
688 | prepare_repo_settings(the_repository); |
689 | if (the_repository->settings.gc_write_commit_graph == 1) | |
0bd52e27 | 690 | write_commit_graph_reachable(the_repository->objects->odb, |
f4f8dfe1 | 691 | !quiet && !daemonized ? COMMIT_GRAPH_WRITE_PROGRESS : 0, |
7211b9e7 | 692 | NULL); |
d5d5d7b6 | 693 | |
a087cc98 | 694 | if (auto_gc && too_many_loose_objects()) |
fea6128b ÆAB |
695 | warning(_("There are too many unreachable loose objects; " |
696 | "run 'git prune' to remove them.")); | |
a087cc98 | 697 | |
a831c06a DT |
698 | if (!daemonized) |
699 | unlink(git_path("gc.log")); | |
700 | ||
6757ada4 JB |
701 | return 0; |
702 | } | |
2057d750 DS |
703 | |
704 | static const char * const builtin_maintenance_run_usage[] = { | |
090511bc | 705 | N_("git maintenance run [--auto] [--[no-]quiet] [--task=<task>]"), |
2057d750 DS |
706 | NULL |
707 | }; | |
708 | ||
709 | struct maintenance_run_opts { | |
710 | int auto_flag; | |
3ddaad0e | 711 | int quiet; |
2057d750 DS |
712 | }; |
713 | ||
4ddc79b2 DS |
714 | /* Remember to update object flag allocation in object.h */ |
715 | #define SEEN (1u<<0) | |
716 | ||
717 | struct cg_auto_data { | |
718 | int num_not_in_graph; | |
719 | int limit; | |
720 | }; | |
721 | ||
722 | static int dfs_on_ref(const char *refname, | |
723 | const struct object_id *oid, int flags, | |
724 | void *cb_data) | |
725 | { | |
726 | struct cg_auto_data *data = (struct cg_auto_data *)cb_data; | |
727 | int result = 0; | |
728 | struct object_id peeled; | |
729 | struct commit_list *stack = NULL; | |
730 | struct commit *commit; | |
731 | ||
732 | if (!peel_ref(refname, &peeled)) | |
733 | oid = &peeled; | |
734 | if (oid_object_info(the_repository, oid, NULL) != OBJ_COMMIT) | |
735 | return 0; | |
736 | ||
737 | commit = lookup_commit(the_repository, oid); | |
738 | if (!commit) | |
739 | return 0; | |
740 | if (parse_commit(commit)) | |
741 | return 0; | |
742 | ||
743 | commit_list_append(commit, &stack); | |
744 | ||
745 | while (!result && stack) { | |
746 | struct commit_list *parent; | |
747 | ||
748 | commit = pop_commit(&stack); | |
749 | ||
750 | for (parent = commit->parents; parent; parent = parent->next) { | |
751 | if (parse_commit(parent->item) || | |
752 | commit_graph_position(parent->item) != COMMIT_NOT_FROM_GRAPH || | |
753 | parent->item->object.flags & SEEN) | |
754 | continue; | |
755 | ||
756 | parent->item->object.flags |= SEEN; | |
757 | data->num_not_in_graph++; | |
758 | ||
759 | if (data->num_not_in_graph >= data->limit) { | |
760 | result = 1; | |
761 | break; | |
762 | } | |
763 | ||
764 | commit_list_append(parent->item, &stack); | |
765 | } | |
766 | } | |
767 | ||
768 | free_commit_list(stack); | |
769 | return result; | |
770 | } | |
771 | ||
772 | static int should_write_commit_graph(void) | |
773 | { | |
774 | int result; | |
775 | struct cg_auto_data data; | |
776 | ||
777 | data.num_not_in_graph = 0; | |
778 | data.limit = 100; | |
779 | git_config_get_int("maintenance.commit-graph.auto", | |
780 | &data.limit); | |
781 | ||
782 | if (!data.limit) | |
783 | return 0; | |
784 | if (data.limit < 0) | |
785 | return 1; | |
786 | ||
787 | result = for_each_ref(dfs_on_ref, &data); | |
788 | ||
789 | clear_commit_marks_all(SEEN); | |
790 | ||
791 | return result; | |
792 | } | |
793 | ||
663b2b1b DS |
794 | static int run_write_commit_graph(struct maintenance_run_opts *opts) |
795 | { | |
796 | struct child_process child = CHILD_PROCESS_INIT; | |
797 | ||
798 | child.git_cmd = 1; | |
799 | strvec_pushl(&child.args, "commit-graph", "write", | |
800 | "--split", "--reachable", NULL); | |
801 | ||
802 | if (opts->quiet) | |
803 | strvec_push(&child.args, "--no-progress"); | |
804 | ||
805 | return !!run_command(&child); | |
806 | } | |
807 | ||
808 | static int maintenance_task_commit_graph(struct maintenance_run_opts *opts) | |
809 | { | |
810 | close_object_store(the_repository->objects); | |
811 | if (run_write_commit_graph(opts)) { | |
812 | error(_("failed to write commit-graph")); | |
813 | return 1; | |
814 | } | |
815 | ||
816 | return 0; | |
817 | } | |
818 | ||
2057d750 DS |
819 | static int maintenance_task_gc(struct maintenance_run_opts *opts) |
820 | { | |
821 | struct child_process child = CHILD_PROCESS_INIT; | |
822 | ||
823 | child.git_cmd = 1; | |
824 | strvec_push(&child.args, "gc"); | |
825 | ||
826 | if (opts->auto_flag) | |
827 | strvec_push(&child.args, "--auto"); | |
3ddaad0e DS |
828 | if (opts->quiet) |
829 | strvec_push(&child.args, "--quiet"); | |
830 | else | |
831 | strvec_push(&child.args, "--no-quiet"); | |
2057d750 DS |
832 | |
833 | close_object_store(the_repository->objects); | |
834 | return run_command(&child); | |
835 | } | |
836 | ||
3103e984 DS |
837 | typedef int maintenance_task_fn(struct maintenance_run_opts *opts); |
838 | ||
916d0626 DS |
839 | /* |
840 | * An auto condition function returns 1 if the task should run | |
841 | * and 0 if the task should NOT run. See needs_to_gc() for an | |
842 | * example. | |
843 | */ | |
844 | typedef int maintenance_auto_fn(void); | |
845 | ||
3103e984 DS |
846 | struct maintenance_task { |
847 | const char *name; | |
848 | maintenance_task_fn *fn; | |
916d0626 | 849 | maintenance_auto_fn *auto_condition; |
3103e984 | 850 | unsigned enabled:1; |
090511bc DS |
851 | |
852 | /* -1 if not selected. */ | |
853 | int selected_order; | |
3103e984 DS |
854 | }; |
855 | ||
856 | enum maintenance_task_label { | |
857 | TASK_GC, | |
663b2b1b | 858 | TASK_COMMIT_GRAPH, |
3103e984 DS |
859 | |
860 | /* Leave as final value */ | |
861 | TASK__COUNT | |
862 | }; | |
863 | ||
864 | static struct maintenance_task tasks[] = { | |
865 | [TASK_GC] = { | |
866 | "gc", | |
867 | maintenance_task_gc, | |
916d0626 | 868 | need_to_gc, |
3103e984 DS |
869 | 1, |
870 | }, | |
663b2b1b DS |
871 | [TASK_COMMIT_GRAPH] = { |
872 | "commit-graph", | |
873 | maintenance_task_commit_graph, | |
4ddc79b2 | 874 | should_write_commit_graph, |
663b2b1b | 875 | }, |
3103e984 DS |
876 | }; |
877 | ||
090511bc DS |
878 | static int compare_tasks_by_selection(const void *a_, const void *b_) |
879 | { | |
880 | const struct maintenance_task *a, *b; | |
881 | ||
882 | a = (const struct maintenance_task *)&a_; | |
883 | b = (const struct maintenance_task *)&b_; | |
884 | ||
885 | return b->selected_order - a->selected_order; | |
886 | } | |
887 | ||
3103e984 DS |
888 | static int maintenance_run_tasks(struct maintenance_run_opts *opts) |
889 | { | |
090511bc | 890 | int i, found_selected = 0; |
3103e984 | 891 | int result = 0; |
d7514f6e DS |
892 | struct lock_file lk; |
893 | struct repository *r = the_repository; | |
894 | char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path); | |
895 | ||
896 | if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) { | |
897 | /* | |
898 | * Another maintenance command is running. | |
899 | * | |
900 | * If --auto was provided, then it is likely due to a | |
901 | * recursive process stack. Do not report an error in | |
902 | * that case. | |
903 | */ | |
904 | if (!opts->auto_flag && !opts->quiet) | |
905 | warning(_("lock file '%s' exists, skipping maintenance"), | |
906 | lock_path); | |
907 | free(lock_path); | |
908 | return 0; | |
909 | } | |
910 | free(lock_path); | |
3103e984 | 911 | |
090511bc DS |
912 | for (i = 0; !found_selected && i < TASK__COUNT; i++) |
913 | found_selected = tasks[i].selected_order >= 0; | |
914 | ||
915 | if (found_selected) | |
916 | QSORT(tasks, TASK__COUNT, compare_tasks_by_selection); | |
917 | ||
3103e984 | 918 | for (i = 0; i < TASK__COUNT; i++) { |
090511bc DS |
919 | if (found_selected && tasks[i].selected_order < 0) |
920 | continue; | |
921 | ||
922 | if (!found_selected && !tasks[i].enabled) | |
3103e984 DS |
923 | continue; |
924 | ||
916d0626 DS |
925 | if (opts->auto_flag && |
926 | (!tasks[i].auto_condition || | |
927 | !tasks[i].auto_condition())) | |
928 | continue; | |
929 | ||
25914c4f | 930 | trace2_region_enter("maintenance", tasks[i].name, r); |
3103e984 DS |
931 | if (tasks[i].fn(opts)) { |
932 | error(_("task '%s' failed"), tasks[i].name); | |
933 | result = 1; | |
934 | } | |
25914c4f | 935 | trace2_region_leave("maintenance", tasks[i].name, r); |
3103e984 DS |
936 | } |
937 | ||
d7514f6e | 938 | rollback_lock_file(&lk); |
3103e984 DS |
939 | return result; |
940 | } | |
941 | ||
65d655b5 DS |
942 | static void initialize_task_config(void) |
943 | { | |
944 | int i; | |
945 | struct strbuf config_name = STRBUF_INIT; | |
916d0626 DS |
946 | gc_config(); |
947 | ||
65d655b5 DS |
948 | for (i = 0; i < TASK__COUNT; i++) { |
949 | int config_value; | |
950 | ||
951 | strbuf_setlen(&config_name, 0); | |
952 | strbuf_addf(&config_name, "maintenance.%s.enabled", | |
953 | tasks[i].name); | |
954 | ||
955 | if (!git_config_get_bool(config_name.buf, &config_value)) | |
956 | tasks[i].enabled = config_value; | |
957 | } | |
958 | ||
959 | strbuf_release(&config_name); | |
960 | } | |
961 | ||
090511bc DS |
962 | static int task_option_parse(const struct option *opt, |
963 | const char *arg, int unset) | |
964 | { | |
965 | int i, num_selected = 0; | |
966 | struct maintenance_task *task = NULL; | |
967 | ||
968 | BUG_ON_OPT_NEG(unset); | |
969 | ||
970 | for (i = 0; i < TASK__COUNT; i++) { | |
971 | if (tasks[i].selected_order >= 0) | |
972 | num_selected++; | |
973 | if (!strcasecmp(tasks[i].name, arg)) { | |
974 | task = &tasks[i]; | |
975 | } | |
976 | } | |
977 | ||
978 | if (!task) { | |
979 | error(_("'%s' is not a valid task"), arg); | |
980 | return 1; | |
981 | } | |
982 | ||
983 | if (task->selected_order >= 0) { | |
984 | error(_("task '%s' cannot be selected multiple times"), arg); | |
985 | return 1; | |
986 | } | |
987 | ||
988 | task->selected_order = num_selected + 1; | |
989 | ||
990 | return 0; | |
991 | } | |
992 | ||
2057d750 DS |
993 | static int maintenance_run(int argc, const char **argv, const char *prefix) |
994 | { | |
090511bc | 995 | int i; |
2057d750 DS |
996 | struct maintenance_run_opts opts; |
997 | struct option builtin_maintenance_run_options[] = { | |
998 | OPT_BOOL(0, "auto", &opts.auto_flag, | |
999 | N_("run tasks based on the state of the repository")), | |
3ddaad0e DS |
1000 | OPT_BOOL(0, "quiet", &opts.quiet, |
1001 | N_("do not report progress or other information over stderr")), | |
090511bc DS |
1002 | OPT_CALLBACK_F(0, "task", NULL, N_("task"), |
1003 | N_("run a specific task"), | |
1004 | PARSE_OPT_NONEG, task_option_parse), | |
2057d750 DS |
1005 | OPT_END() |
1006 | }; | |
1007 | memset(&opts, 0, sizeof(opts)); | |
1008 | ||
3ddaad0e | 1009 | opts.quiet = !isatty(2); |
65d655b5 | 1010 | initialize_task_config(); |
3ddaad0e | 1011 | |
090511bc DS |
1012 | for (i = 0; i < TASK__COUNT; i++) |
1013 | tasks[i].selected_order = -1; | |
1014 | ||
2057d750 DS |
1015 | argc = parse_options(argc, argv, prefix, |
1016 | builtin_maintenance_run_options, | |
1017 | builtin_maintenance_run_usage, | |
1018 | PARSE_OPT_STOP_AT_NON_OPTION); | |
1019 | ||
1020 | if (argc != 0) | |
1021 | usage_with_options(builtin_maintenance_run_usage, | |
1022 | builtin_maintenance_run_options); | |
3103e984 | 1023 | return maintenance_run_tasks(&opts); |
2057d750 DS |
1024 | } |
1025 | ||
1026 | static const char builtin_maintenance_usage[] = N_("git maintenance run [<options>]"); | |
1027 | ||
1028 | int cmd_maintenance(int argc, const char **argv, const char *prefix) | |
1029 | { | |
1030 | if (argc < 2 || | |
1031 | (argc == 2 && !strcmp(argv[1], "-h"))) | |
1032 | usage(builtin_maintenance_usage); | |
1033 | ||
1034 | if (!strcmp(argv[1], "run")) | |
1035 | return maintenance_run(argc - 1, argv + 1, prefix); | |
1036 | ||
1037 | die(_("invalid subcommand: %s"), argv[1]); | |
1038 | } |