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 "repository.h"
15 #include "config.h"
16 #include "tempfile.h"
17 #include "lockfile.h"
18 #include "parse-options.h"
19 #include "run-command.h"
20 #include "sigchain.h"
21 #include "strvec.h"
22 #include "commit.h"
23 #include "commit-graph.h"
24 #include "packfile.h"
25 #include "object-store.h"
26 #include "pack.h"
27 #include "pack-objects.h"
28 #include "blob.h"
29 #include "tree.h"
30 #include "promisor-remote.h"
31 #include "refs.h"
32 #include "remote.h"
33 #include "object-store.h"
34 #include "exec-cmd.h"
35
36 #define FAILED_RUN "failed to run %s"
37
38 static const char * const builtin_gc_usage[] = {
39 N_("git gc [<options>]"),
40 NULL
41 };
42
43 static int pack_refs = 1;
44 static int prune_reflogs = 1;
45 static int aggressive_depth = 50;
46 static int aggressive_window = 250;
47 static int gc_auto_threshold = 6700;
48 static int gc_auto_pack_limit = 50;
49 static int detach_auto = 1;
50 static timestamp_t gc_log_expire_time;
51 static const char *gc_log_expire = "1.day.ago";
52 static const char *prune_expire = "2.weeks.ago";
53 static const char *prune_worktrees_expire = "3.months.ago";
54 static unsigned long big_pack_threshold;
55 static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
56
57 static struct strvec reflog = STRVEC_INIT;
58 static struct strvec repack = STRVEC_INIT;
59 static struct strvec prune = STRVEC_INIT;
60 static struct strvec prune_worktrees = STRVEC_INIT;
61 static struct strvec rerere = STRVEC_INIT;
62
63 static struct tempfile *pidfile;
64 static struct lock_file log_lock;
65
66 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
67
clean_pack_garbage(void)68 static void clean_pack_garbage(void)
69 {
70 int i;
71 for (i = 0; i < pack_garbage.nr; i++)
72 unlink_or_warn(pack_garbage.items[i].string);
73 string_list_clear(&pack_garbage, 0);
74 }
75
report_pack_garbage(unsigned seen_bits,const char * path)76 static void report_pack_garbage(unsigned seen_bits, const char *path)
77 {
78 if (seen_bits == PACKDIR_FILE_IDX)
79 string_list_append(&pack_garbage, path);
80 }
81
process_log_file(void)82 static void process_log_file(void)
83 {
84 struct stat st;
85 if (fstat(get_lock_file_fd(&log_lock), &st)) {
86 /*
87 * Perhaps there was an i/o error or another
88 * unlikely situation. Try to make a note of
89 * this in gc.log along with any existing
90 * messages.
91 */
92 int saved_errno = errno;
93 fprintf(stderr, _("Failed to fstat %s: %s"),
94 get_lock_file_path(&log_lock),
95 strerror(saved_errno));
96 fflush(stderr);
97 commit_lock_file(&log_lock);
98 errno = saved_errno;
99 } else if (st.st_size) {
100 /* There was some error recorded in the lock file */
101 commit_lock_file(&log_lock);
102 } else {
103 /* No error, clean up any old gc.log */
104 unlink(git_path("gc.log"));
105 rollback_lock_file(&log_lock);
106 }
107 }
108
process_log_file_at_exit(void)109 static void process_log_file_at_exit(void)
110 {
111 fflush(stderr);
112 process_log_file();
113 }
114
process_log_file_on_signal(int signo)115 static void process_log_file_on_signal(int signo)
116 {
117 process_log_file();
118 sigchain_pop(signo);
119 raise(signo);
120 }
121
gc_config_is_timestamp_never(const char * var)122 static int gc_config_is_timestamp_never(const char *var)
123 {
124 const char *value;
125 timestamp_t expire;
126
127 if (!git_config_get_value(var, &value) && value) {
128 if (parse_expiry_date(value, &expire))
129 die(_("failed to parse '%s' value '%s'"), var, value);
130 return expire == 0;
131 }
132 return 0;
133 }
134
gc_config(void)135 static void gc_config(void)
136 {
137 const char *value;
138
139 if (!git_config_get_value("gc.packrefs", &value)) {
140 if (value && !strcmp(value, "notbare"))
141 pack_refs = -1;
142 else
143 pack_refs = git_config_bool("gc.packrefs", value);
144 }
145
146 if (gc_config_is_timestamp_never("gc.reflogexpire") &&
147 gc_config_is_timestamp_never("gc.reflogexpireunreachable"))
148 prune_reflogs = 0;
149
150 git_config_get_int("gc.aggressivewindow", &aggressive_window);
151 git_config_get_int("gc.aggressivedepth", &aggressive_depth);
152 git_config_get_int("gc.auto", &gc_auto_threshold);
153 git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
154 git_config_get_bool("gc.autodetach", &detach_auto);
155 git_config_get_expiry("gc.pruneexpire", &prune_expire);
156 git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
157 git_config_get_expiry("gc.logexpiry", &gc_log_expire);
158
159 git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold);
160 git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size);
161
162 git_config(git_default_config, NULL);
163 }
164
165 struct maintenance_run_opts;
maintenance_task_pack_refs(MAYBE_UNUSED struct maintenance_run_opts * opts)166 static int maintenance_task_pack_refs(MAYBE_UNUSED struct maintenance_run_opts *opts)
167 {
168 struct strvec pack_refs_cmd = STRVEC_INIT;
169 strvec_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
170
171 return run_command_v_opt(pack_refs_cmd.v, RUN_GIT_CMD);
172 }
173
too_many_loose_objects(void)174 static int too_many_loose_objects(void)
175 {
176 /*
177 * Quickly check if a "gc" is needed, by estimating how
178 * many loose objects there are. Because SHA-1 is evenly
179 * distributed, we can check only one and get a reasonable
180 * estimate.
181 */
182 DIR *dir;
183 struct dirent *ent;
184 int auto_threshold;
185 int num_loose = 0;
186 int needed = 0;
187 const unsigned hexsz_loose = the_hash_algo->hexsz - 2;
188
189 dir = opendir(git_path("objects/17"));
190 if (!dir)
191 return 0;
192
193 auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
194 while ((ent = readdir(dir)) != NULL) {
195 if (strspn(ent->d_name, "0123456789abcdef") != hexsz_loose ||
196 ent->d_name[hexsz_loose] != '\0')
197 continue;
198 if (++num_loose > auto_threshold) {
199 needed = 1;
200 break;
201 }
202 }
203 closedir(dir);
204 return needed;
205 }
206
find_base_packs(struct string_list * packs,unsigned long limit)207 static struct packed_git *find_base_packs(struct string_list *packs,
208 unsigned long limit)
209 {
210 struct packed_git *p, *base = NULL;
211
212 for (p = get_all_packs(the_repository); p; p = p->next) {
213 if (!p->pack_local)
214 continue;
215 if (limit) {
216 if (p->pack_size >= limit)
217 string_list_append(packs, p->pack_name);
218 } else if (!base || base->pack_size < p->pack_size) {
219 base = p;
220 }
221 }
222
223 if (base)
224 string_list_append(packs, base->pack_name);
225
226 return base;
227 }
228
too_many_packs(void)229 static int too_many_packs(void)
230 {
231 struct packed_git *p;
232 int cnt;
233
234 if (gc_auto_pack_limit <= 0)
235 return 0;
236
237 for (cnt = 0, p = get_all_packs(the_repository); p; p = p->next) {
238 if (!p->pack_local)
239 continue;
240 if (p->pack_keep)
241 continue;
242 /*
243 * Perhaps check the size of the pack and count only
244 * very small ones here?
245 */
246 cnt++;
247 }
248 return gc_auto_pack_limit < cnt;
249 }
250
total_ram(void)251 static uint64_t total_ram(void)
252 {
253 #if defined(HAVE_SYSINFO)
254 struct sysinfo si;
255
256 if (!sysinfo(&si))
257 return si.totalram;
258 #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM))
259 int64_t physical_memory;
260 int mib[2];
261 size_t length;
262
263 mib[0] = CTL_HW;
264 # if defined(HW_MEMSIZE)
265 mib[1] = HW_MEMSIZE;
266 # else
267 mib[1] = HW_PHYSMEM;
268 # endif
269 length = sizeof(int64_t);
270 if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
271 return physical_memory;
272 #elif defined(GIT_WINDOWS_NATIVE)
273 MEMORYSTATUSEX memInfo;
274
275 memInfo.dwLength = sizeof(MEMORYSTATUSEX);
276 if (GlobalMemoryStatusEx(&memInfo))
277 return memInfo.ullTotalPhys;
278 #endif
279 return 0;
280 }
281
estimate_repack_memory(struct packed_git * pack)282 static uint64_t estimate_repack_memory(struct packed_git *pack)
283 {
284 unsigned long nr_objects = approximate_object_count();
285 size_t os_cache, heap;
286
287 if (!pack || !nr_objects)
288 return 0;
289
290 /*
291 * First we have to scan through at least one pack.
292 * Assume enough room in OS file cache to keep the entire pack
293 * or we may accidentally evict data of other processes from
294 * the cache.
295 */
296 os_cache = pack->pack_size + pack->index_size;
297 /* then pack-objects needs lots more for book keeping */
298 heap = sizeof(struct object_entry) * nr_objects;
299 /*
300 * internal rev-list --all --objects takes up some memory too,
301 * let's say half of it is for blobs
302 */
303 heap += sizeof(struct blob) * nr_objects / 2;
304 /*
305 * and the other half is for trees (commits and tags are
306 * usually insignificant)
307 */
308 heap += sizeof(struct tree) * nr_objects / 2;
309 /* and then obj_hash[], underestimated in fact */
310 heap += sizeof(struct object *) * nr_objects;
311 /* revindex is used also */
312 heap += (sizeof(off_t) + sizeof(uint32_t)) * nr_objects;
313 /*
314 * read_sha1_file() (either at delta calculation phase, or
315 * writing phase) also fills up the delta base cache
316 */
317 heap += delta_base_cache_limit;
318 /* and of course pack-objects has its own delta cache */
319 heap += max_delta_cache_size;
320
321 return os_cache + heap;
322 }
323
keep_one_pack(struct string_list_item * item,void * data)324 static int keep_one_pack(struct string_list_item *item, void *data)
325 {
326 strvec_pushf(&repack, "--keep-pack=%s", basename(item->string));
327 return 0;
328 }
329
add_repack_all_option(struct string_list * keep_pack)330 static void add_repack_all_option(struct string_list *keep_pack)
331 {
332 if (prune_expire && !strcmp(prune_expire, "now"))
333 strvec_push(&repack, "-a");
334 else {
335 strvec_push(&repack, "-A");
336 if (prune_expire)
337 strvec_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
338 }
339
340 if (keep_pack)
341 for_each_string_list(keep_pack, keep_one_pack, NULL);
342 }
343
add_repack_incremental_option(void)344 static void add_repack_incremental_option(void)
345 {
346 strvec_push(&repack, "--no-write-bitmap-index");
347 }
348
need_to_gc(void)349 static int need_to_gc(void)
350 {
351 /*
352 * Setting gc.auto to 0 or negative can disable the
353 * automatic gc.
354 */
355 if (gc_auto_threshold <= 0)
356 return 0;
357
358 /*
359 * If there are too many loose objects, but not too many
360 * packs, we run "repack -d -l". If there are too many packs,
361 * we run "repack -A -d -l". Otherwise we tell the caller
362 * there is no need.
363 */
364 if (too_many_packs()) {
365 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
366
367 if (big_pack_threshold) {
368 find_base_packs(&keep_pack, big_pack_threshold);
369 if (keep_pack.nr >= gc_auto_pack_limit) {
370 big_pack_threshold = 0;
371 string_list_clear(&keep_pack, 0);
372 find_base_packs(&keep_pack, 0);
373 }
374 } else {
375 struct packed_git *p = find_base_packs(&keep_pack, 0);
376 uint64_t mem_have, mem_want;
377
378 mem_have = total_ram();
379 mem_want = estimate_repack_memory(p);
380
381 /*
382 * Only allow 1/2 of memory for pack-objects, leave
383 * the rest for the OS and other processes in the
384 * system.
385 */
386 if (!mem_have || mem_want < mem_have / 2)
387 string_list_clear(&keep_pack, 0);
388 }
389
390 add_repack_all_option(&keep_pack);
391 string_list_clear(&keep_pack, 0);
392 } else if (too_many_loose_objects())
393 add_repack_incremental_option();
394 else
395 return 0;
396
397 if (run_hook_le(NULL, "pre-auto-gc", NULL))
398 return 0;
399 return 1;
400 }
401
402 /* return NULL on success, else hostname running the gc */
lock_repo_for_gc(int force,pid_t * ret_pid)403 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
404 {
405 struct lock_file lock = LOCK_INIT;
406 char my_host[HOST_NAME_MAX + 1];
407 struct strbuf sb = STRBUF_INIT;
408 struct stat st;
409 uintmax_t pid;
410 FILE *fp;
411 int fd;
412 char *pidfile_path;
413
414 if (is_tempfile_active(pidfile))
415 /* already locked */
416 return NULL;
417
418 if (xgethostname(my_host, sizeof(my_host)))
419 xsnprintf(my_host, sizeof(my_host), "unknown");
420
421 pidfile_path = git_pathdup("gc.pid");
422 fd = hold_lock_file_for_update(&lock, pidfile_path,
423 LOCK_DIE_ON_ERROR);
424 if (!force) {
425 static char locking_host[HOST_NAME_MAX + 1];
426 static char *scan_fmt;
427 int should_exit;
428
429 if (!scan_fmt)
430 scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
431 fp = fopen(pidfile_path, "r");
432 memset(locking_host, 0, sizeof(locking_host));
433 should_exit =
434 fp != NULL &&
435 !fstat(fileno(fp), &st) &&
436 /*
437 * 12 hour limit is very generous as gc should
438 * never take that long. On the other hand we
439 * don't really need a strict limit here,
440 * running gc --auto one day late is not a big
441 * problem. --force can be used in manual gc
442 * after the user verifies that no gc is
443 * running.
444 */
445 time(NULL) - st.st_mtime <= 12 * 3600 &&
446 fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
447 /* be gentle to concurrent "gc" on remote hosts */
448 (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
449 if (fp != NULL)
450 fclose(fp);
451 if (should_exit) {
452 if (fd >= 0)
453 rollback_lock_file(&lock);
454 *ret_pid = pid;
455 free(pidfile_path);
456 return locking_host;
457 }
458 }
459
460 strbuf_addf(&sb, "%"PRIuMAX" %s",
461 (uintmax_t) getpid(), my_host);
462 write_in_full(fd, sb.buf, sb.len);
463 strbuf_release(&sb);
464 commit_lock_file(&lock);
465 pidfile = register_tempfile(pidfile_path);
466 free(pidfile_path);
467 return NULL;
468 }
469
470 /*
471 * Returns 0 if there was no previous error and gc can proceed, 1 if
472 * gc should not proceed due to an error in the last run. Prints a
473 * message and returns -1 if an error occurred while reading gc.log
474 */
report_last_gc_error(void)475 static int report_last_gc_error(void)
476 {
477 struct strbuf sb = STRBUF_INIT;
478 int ret = 0;
479 ssize_t len;
480 struct stat st;
481 char *gc_log_path = git_pathdup("gc.log");
482
483 if (stat(gc_log_path, &st)) {
484 if (errno == ENOENT)
485 goto done;
486
487 ret = error_errno(_("cannot stat '%s'"), gc_log_path);
488 goto done;
489 }
490
491 if (st.st_mtime < gc_log_expire_time)
492 goto done;
493
494 len = strbuf_read_file(&sb, gc_log_path, 0);
495 if (len < 0)
496 ret = error_errno(_("cannot read '%s'"), gc_log_path);
497 else if (len > 0) {
498 /*
499 * A previous gc failed. Report the error, and don't
500 * bother with an automatic gc run since it is likely
501 * to fail in the same way.
502 */
503 warning(_("The last gc run reported the following. "
504 "Please correct the root cause\n"
505 "and remove %s\n"
506 "Automatic cleanup will not be performed "
507 "until the file is removed.\n\n"
508 "%s"),
509 gc_log_path, sb.buf);
510 ret = 1;
511 }
512 strbuf_release(&sb);
513 done:
514 free(gc_log_path);
515 return ret;
516 }
517
gc_before_repack(void)518 static void gc_before_repack(void)
519 {
520 /*
521 * We may be called twice, as both the pre- and
522 * post-daemonized phases will call us, but running these
523 * commands more than once is pointless and wasteful.
524 */
525 static int done = 0;
526 if (done++)
527 return;
528
529 if (pack_refs && maintenance_task_pack_refs(NULL))
530 die(FAILED_RUN, "pack-refs");
531
532 if (prune_reflogs && run_command_v_opt(reflog.v, RUN_GIT_CMD))
533 die(FAILED_RUN, reflog.v[0]);
534 }
535
cmd_gc(int argc,const char ** argv,const char * prefix)536 int cmd_gc(int argc, const char **argv, const char *prefix)
537 {
538 int aggressive = 0;
539 int auto_gc = 0;
540 int quiet = 0;
541 int force = 0;
542 const char *name;
543 pid_t pid;
544 int daemonized = 0;
545 int keep_largest_pack = -1;
546 timestamp_t dummy;
547
548 struct option builtin_gc_options[] = {
549 OPT__QUIET(&quiet, N_("suppress progress reporting")),
550 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
551 N_("prune unreferenced objects"),
552 PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
553 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
554 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
555 PARSE_OPT_NOCOMPLETE),
556 OPT_BOOL_F(0, "force", &force,
557 N_("force running gc even if there may be another gc running"),
558 PARSE_OPT_NOCOMPLETE),
559 OPT_BOOL(0, "keep-largest-pack", &keep_largest_pack,
560 N_("repack all other packs except the largest pack")),
561 OPT_END()
562 };
563
564 if (argc == 2 && !strcmp(argv[1], "-h"))
565 usage_with_options(builtin_gc_usage, builtin_gc_options);
566
567 strvec_pushl(&reflog, "reflog", "expire", "--all", NULL);
568 strvec_pushl(&repack, "repack", "-d", "-l", NULL);
569 strvec_pushl(&prune, "prune", "--expire", NULL);
570 strvec_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
571 strvec_pushl(&rerere, "rerere", "gc", NULL);
572
573 /* default expiry time, overwritten in gc_config */
574 gc_config();
575 if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
576 die(_("failed to parse gc.logexpiry value %s"), gc_log_expire);
577
578 if (pack_refs < 0)
579 pack_refs = !is_bare_repository();
580
581 argc = parse_options(argc, argv, prefix, builtin_gc_options,
582 builtin_gc_usage, 0);
583 if (argc > 0)
584 usage_with_options(builtin_gc_usage, builtin_gc_options);
585
586 if (prune_expire && parse_expiry_date(prune_expire, &dummy))
587 die(_("failed to parse prune expiry value %s"), prune_expire);
588
589 if (aggressive) {
590 strvec_push(&repack, "-f");
591 if (aggressive_depth > 0)
592 strvec_pushf(&repack, "--depth=%d", aggressive_depth);
593 if (aggressive_window > 0)
594 strvec_pushf(&repack, "--window=%d", aggressive_window);
595 }
596 if (quiet)
597 strvec_push(&repack, "-q");
598
599 if (auto_gc) {
600 /*
601 * Auto-gc should be least intrusive as possible.
602 */
603 if (!need_to_gc())
604 return 0;
605 if (!quiet) {
606 if (detach_auto)
607 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
608 else
609 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
610 fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
611 }
612 if (detach_auto) {
613 int ret = report_last_gc_error();
614 if (ret < 0)
615 /* an I/O error occurred, already reported */
616 exit(128);
617 if (ret == 1)
618 /* Last gc --auto failed. Skip this one. */
619 return 0;
620
621 if (lock_repo_for_gc(force, &pid))
622 return 0;
623 gc_before_repack(); /* dies on failure */
624 delete_tempfile(&pidfile);
625
626 /*
627 * failure to daemonize is ok, we'll continue
628 * in foreground
629 */
630 daemonized = !daemonize();
631 }
632 } else {
633 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
634
635 if (keep_largest_pack != -1) {
636 if (keep_largest_pack)
637 find_base_packs(&keep_pack, 0);
638 } else if (big_pack_threshold) {
639 find_base_packs(&keep_pack, big_pack_threshold);
640 }
641
642 add_repack_all_option(&keep_pack);
643 string_list_clear(&keep_pack, 0);
644 }
645
646 name = lock_repo_for_gc(force, &pid);
647 if (name) {
648 if (auto_gc)
649 return 0; /* be quiet on --auto */
650 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
651 name, (uintmax_t)pid);
652 }
653
654 if (daemonized) {
655 hold_lock_file_for_update(&log_lock,
656 git_path("gc.log"),
657 LOCK_DIE_ON_ERROR);
658 dup2(get_lock_file_fd(&log_lock), 2);
659 sigchain_push_common(process_log_file_on_signal);
660 atexit(process_log_file_at_exit);
661 }
662
663 gc_before_repack();
664
665 if (!repository_format_precious_objects) {
666 if (run_command_v_opt(repack.v,
667 RUN_GIT_CMD | RUN_CLOSE_OBJECT_STORE))
668 die(FAILED_RUN, repack.v[0]);
669
670 if (prune_expire) {
671 strvec_push(&prune, prune_expire);
672 if (quiet)
673 strvec_push(&prune, "--no-progress");
674 if (has_promisor_remote())
675 strvec_push(&prune,
676 "--exclude-promisor-objects");
677 if (run_command_v_opt(prune.v, RUN_GIT_CMD))
678 die(FAILED_RUN, prune.v[0]);
679 }
680 }
681
682 if (prune_worktrees_expire) {
683 strvec_push(&prune_worktrees, prune_worktrees_expire);
684 if (run_command_v_opt(prune_worktrees.v, RUN_GIT_CMD))
685 die(FAILED_RUN, prune_worktrees.v[0]);
686 }
687
688 if (run_command_v_opt(rerere.v, RUN_GIT_CMD))
689 die(FAILED_RUN, rerere.v[0]);
690
691 report_garbage = report_pack_garbage;
692 reprepare_packed_git(the_repository);
693 if (pack_garbage.nr > 0) {
694 close_object_store(the_repository->objects);
695 clean_pack_garbage();
696 }
697
698 prepare_repo_settings(the_repository);
699 if (the_repository->settings.gc_write_commit_graph == 1)
700 write_commit_graph_reachable(the_repository->objects->odb,
701 !quiet && !daemonized ? COMMIT_GRAPH_WRITE_PROGRESS : 0,
702 NULL);
703
704 if (auto_gc && too_many_loose_objects())
705 warning(_("There are too many unreachable loose objects; "
706 "run 'git prune' to remove them."));
707
708 if (!daemonized)
709 unlink(git_path("gc.log"));
710
711 return 0;
712 }
713
714 static const char *const builtin_maintenance_run_usage[] = {
715 N_("git maintenance run [--auto] [--[no-]quiet] [--task=<task>] [--schedule]"),
716 NULL
717 };
718
719 enum schedule_priority {
720 SCHEDULE_NONE = 0,
721 SCHEDULE_WEEKLY = 1,
722 SCHEDULE_DAILY = 2,
723 SCHEDULE_HOURLY = 3,
724 };
725
parse_schedule(const char * value)726 static enum schedule_priority parse_schedule(const char *value)
727 {
728 if (!value)
729 return SCHEDULE_NONE;
730 if (!strcasecmp(value, "hourly"))
731 return SCHEDULE_HOURLY;
732 if (!strcasecmp(value, "daily"))
733 return SCHEDULE_DAILY;
734 if (!strcasecmp(value, "weekly"))
735 return SCHEDULE_WEEKLY;
736 return SCHEDULE_NONE;
737 }
738
maintenance_opt_schedule(const struct option * opt,const char * arg,int unset)739 static int maintenance_opt_schedule(const struct option *opt, const char *arg,
740 int unset)
741 {
742 enum schedule_priority *priority = opt->value;
743
744 if (unset)
745 die(_("--no-schedule is not allowed"));
746
747 *priority = parse_schedule(arg);
748
749 if (!*priority)
750 die(_("unrecognized --schedule argument '%s'"), arg);
751
752 return 0;
753 }
754
755 struct maintenance_run_opts {
756 int auto_flag;
757 int quiet;
758 enum schedule_priority schedule;
759 };
760
761 /* Remember to update object flag allocation in object.h */
762 #define SEEN (1u<<0)
763
764 struct cg_auto_data {
765 int num_not_in_graph;
766 int limit;
767 };
768
dfs_on_ref(const char * refname,const struct object_id * oid,int flags,void * cb_data)769 static int dfs_on_ref(const char *refname,
770 const struct object_id *oid, int flags,
771 void *cb_data)
772 {
773 struct cg_auto_data *data = (struct cg_auto_data *)cb_data;
774 int result = 0;
775 struct object_id peeled;
776 struct commit_list *stack = NULL;
777 struct commit *commit;
778
779 if (!peel_iterated_oid(oid, &peeled))
780 oid = &peeled;
781 if (oid_object_info(the_repository, oid, NULL) != OBJ_COMMIT)
782 return 0;
783
784 commit = lookup_commit(the_repository, oid);
785 if (!commit)
786 return 0;
787 if (parse_commit(commit) ||
788 commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
789 return 0;
790
791 data->num_not_in_graph++;
792
793 if (data->num_not_in_graph >= data->limit)
794 return 1;
795
796 commit_list_append(commit, &stack);
797
798 while (!result && stack) {
799 struct commit_list *parent;
800
801 commit = pop_commit(&stack);
802
803 for (parent = commit->parents; parent; parent = parent->next) {
804 if (parse_commit(parent->item) ||
805 commit_graph_position(parent->item) != COMMIT_NOT_FROM_GRAPH ||
806 parent->item->object.flags & SEEN)
807 continue;
808
809 parent->item->object.flags |= SEEN;
810 data->num_not_in_graph++;
811
812 if (data->num_not_in_graph >= data->limit) {
813 result = 1;
814 break;
815 }
816
817 commit_list_append(parent->item, &stack);
818 }
819 }
820
821 free_commit_list(stack);
822 return result;
823 }
824
should_write_commit_graph(void)825 static int should_write_commit_graph(void)
826 {
827 int result;
828 struct cg_auto_data data;
829
830 data.num_not_in_graph = 0;
831 data.limit = 100;
832 git_config_get_int("maintenance.commit-graph.auto",
833 &data.limit);
834
835 if (!data.limit)
836 return 0;
837 if (data.limit < 0)
838 return 1;
839
840 result = for_each_ref(dfs_on_ref, &data);
841
842 repo_clear_commit_marks(the_repository, SEEN);
843
844 return result;
845 }
846
run_write_commit_graph(struct maintenance_run_opts * opts)847 static int run_write_commit_graph(struct maintenance_run_opts *opts)
848 {
849 struct child_process child = CHILD_PROCESS_INIT;
850
851 child.git_cmd = child.close_object_store = 1;
852 strvec_pushl(&child.args, "commit-graph", "write",
853 "--split", "--reachable", NULL);
854
855 if (opts->quiet)
856 strvec_push(&child.args, "--no-progress");
857
858 return !!run_command(&child);
859 }
860
maintenance_task_commit_graph(struct maintenance_run_opts * opts)861 static int maintenance_task_commit_graph(struct maintenance_run_opts *opts)
862 {
863 prepare_repo_settings(the_repository);
864 if (!the_repository->settings.core_commit_graph)
865 return 0;
866
867 if (run_write_commit_graph(opts)) {
868 error(_("failed to write commit-graph"));
869 return 1;
870 }
871
872 return 0;
873 }
874
fetch_remote(struct remote * remote,void * cbdata)875 static int fetch_remote(struct remote *remote, void *cbdata)
876 {
877 struct maintenance_run_opts *opts = cbdata;
878 struct child_process child = CHILD_PROCESS_INIT;
879
880 if (remote->skip_default_update)
881 return 0;
882
883 child.git_cmd = 1;
884 strvec_pushl(&child.args, "fetch", remote->name,
885 "--prefetch", "--prune", "--no-tags",
886 "--no-write-fetch-head", "--recurse-submodules=no",
887 NULL);
888
889 if (opts->quiet)
890 strvec_push(&child.args, "--quiet");
891
892 return !!run_command(&child);
893 }
894
maintenance_task_prefetch(struct maintenance_run_opts * opts)895 static int maintenance_task_prefetch(struct maintenance_run_opts *opts)
896 {
897 git_config_set_multivar_gently("log.excludedecoration",
898 "refs/prefetch/",
899 "refs/prefetch/",
900 CONFIG_FLAGS_FIXED_VALUE |
901 CONFIG_FLAGS_MULTI_REPLACE);
902
903 if (for_each_remote(fetch_remote, opts)) {
904 error(_("failed to prefetch remotes"));
905 return 1;
906 }
907
908 return 0;
909 }
910
maintenance_task_gc(struct maintenance_run_opts * opts)911 static int maintenance_task_gc(struct maintenance_run_opts *opts)
912 {
913 struct child_process child = CHILD_PROCESS_INIT;
914
915 child.git_cmd = child.close_object_store = 1;
916 strvec_push(&child.args, "gc");
917
918 if (opts->auto_flag)
919 strvec_push(&child.args, "--auto");
920 if (opts->quiet)
921 strvec_push(&child.args, "--quiet");
922 else
923 strvec_push(&child.args, "--no-quiet");
924
925 return run_command(&child);
926 }
927
prune_packed(struct maintenance_run_opts * opts)928 static int prune_packed(struct maintenance_run_opts *opts)
929 {
930 struct child_process child = CHILD_PROCESS_INIT;
931
932 child.git_cmd = 1;
933 strvec_push(&child.args, "prune-packed");
934
935 if (opts->quiet)
936 strvec_push(&child.args, "--quiet");
937
938 return !!run_command(&child);
939 }
940
941 struct write_loose_object_data {
942 FILE *in;
943 int count;
944 int batch_size;
945 };
946
947 static int loose_object_auto_limit = 100;
948
loose_object_count(const struct object_id * oid,const char * path,void * data)949 static int loose_object_count(const struct object_id *oid,
950 const char *path,
951 void *data)
952 {
953 int *count = (int*)data;
954 if (++(*count) >= loose_object_auto_limit)
955 return 1;
956 return 0;
957 }
958
loose_object_auto_condition(void)959 static int loose_object_auto_condition(void)
960 {
961 int count = 0;
962
963 git_config_get_int("maintenance.loose-objects.auto",
964 &loose_object_auto_limit);
965
966 if (!loose_object_auto_limit)
967 return 0;
968 if (loose_object_auto_limit < 0)
969 return 1;
970
971 return for_each_loose_file_in_objdir(the_repository->objects->odb->path,
972 loose_object_count,
973 NULL, NULL, &count);
974 }
975
bail_on_loose(const struct object_id * oid,const char * path,void * data)976 static int bail_on_loose(const struct object_id *oid,
977 const char *path,
978 void *data)
979 {
980 return 1;
981 }
982
write_loose_object_to_stdin(const struct object_id * oid,const char * path,void * data)983 static int write_loose_object_to_stdin(const struct object_id *oid,
984 const char *path,
985 void *data)
986 {
987 struct write_loose_object_data *d = (struct write_loose_object_data *)data;
988
989 fprintf(d->in, "%s\n", oid_to_hex(oid));
990
991 return ++(d->count) > d->batch_size;
992 }
993
pack_loose(struct maintenance_run_opts * opts)994 static int pack_loose(struct maintenance_run_opts *opts)
995 {
996 struct repository *r = the_repository;
997 int result = 0;
998 struct write_loose_object_data data;
999 struct child_process pack_proc = CHILD_PROCESS_INIT;
1000
1001 /*
1002 * Do not start pack-objects process
1003 * if there are no loose objects.
1004 */
1005 if (!for_each_loose_file_in_objdir(r->objects->odb->path,
1006 bail_on_loose,
1007 NULL, NULL, NULL))
1008 return 0;
1009
1010 pack_proc.git_cmd = 1;
1011
1012 strvec_push(&pack_proc.args, "pack-objects");
1013 if (opts->quiet)
1014 strvec_push(&pack_proc.args, "--quiet");
1015 strvec_pushf(&pack_proc.args, "%s/pack/loose", r->objects->odb->path);
1016
1017 pack_proc.in = -1;
1018
1019 if (start_command(&pack_proc)) {
1020 error(_("failed to start 'git pack-objects' process"));
1021 return 1;
1022 }
1023
1024 data.in = xfdopen(pack_proc.in, "w");
1025 data.count = 0;
1026 data.batch_size = 50000;
1027
1028 for_each_loose_file_in_objdir(r->objects->odb->path,
1029 write_loose_object_to_stdin,
1030 NULL,
1031 NULL,
1032 &data);
1033
1034 fclose(data.in);
1035
1036 if (finish_command(&pack_proc)) {
1037 error(_("failed to finish 'git pack-objects' process"));
1038 result = 1;
1039 }
1040
1041 return result;
1042 }
1043
maintenance_task_loose_objects(struct maintenance_run_opts * opts)1044 static int maintenance_task_loose_objects(struct maintenance_run_opts *opts)
1045 {
1046 return prune_packed(opts) || pack_loose(opts);
1047 }
1048
incremental_repack_auto_condition(void)1049 static int incremental_repack_auto_condition(void)
1050 {
1051 struct packed_git *p;
1052 int incremental_repack_auto_limit = 10;
1053 int count = 0;
1054
1055 prepare_repo_settings(the_repository);
1056 if (!the_repository->settings.core_multi_pack_index)
1057 return 0;
1058
1059 git_config_get_int("maintenance.incremental-repack.auto",
1060 &incremental_repack_auto_limit);
1061
1062 if (!incremental_repack_auto_limit)
1063 return 0;
1064 if (incremental_repack_auto_limit < 0)
1065 return 1;
1066
1067 for (p = get_packed_git(the_repository);
1068 count < incremental_repack_auto_limit && p;
1069 p = p->next) {
1070 if (!p->multi_pack_index)
1071 count++;
1072 }
1073
1074 return count >= incremental_repack_auto_limit;
1075 }
1076
multi_pack_index_write(struct maintenance_run_opts * opts)1077 static int multi_pack_index_write(struct maintenance_run_opts *opts)
1078 {
1079 struct child_process child = CHILD_PROCESS_INIT;
1080
1081 child.git_cmd = 1;
1082 strvec_pushl(&child.args, "multi-pack-index", "write", NULL);
1083
1084 if (opts->quiet)
1085 strvec_push(&child.args, "--no-progress");
1086
1087 if (run_command(&child))
1088 return error(_("failed to write multi-pack-index"));
1089
1090 return 0;
1091 }
1092
multi_pack_index_expire(struct maintenance_run_opts * opts)1093 static int multi_pack_index_expire(struct maintenance_run_opts *opts)
1094 {
1095 struct child_process child = CHILD_PROCESS_INIT;
1096
1097 child.git_cmd = child.close_object_store = 1;
1098 strvec_pushl(&child.args, "multi-pack-index", "expire", NULL);
1099
1100 if (opts->quiet)
1101 strvec_push(&child.args, "--no-progress");
1102
1103 if (run_command(&child))
1104 return error(_("'git multi-pack-index expire' failed"));
1105
1106 return 0;
1107 }
1108
1109 #define TWO_GIGABYTES (INT32_MAX)
1110
get_auto_pack_size(void)1111 static off_t get_auto_pack_size(void)
1112 {
1113 /*
1114 * The "auto" value is special: we optimize for
1115 * one large pack-file (i.e. from a clone) and
1116 * expect the rest to be small and they can be
1117 * repacked quickly.
1118 *
1119 * The strategy we select here is to select a
1120 * size that is one more than the second largest
1121 * pack-file. This ensures that we will repack
1122 * at least two packs if there are three or more
1123 * packs.
1124 */
1125 off_t max_size = 0;
1126 off_t second_largest_size = 0;
1127 off_t result_size;
1128 struct packed_git *p;
1129 struct repository *r = the_repository;
1130
1131 reprepare_packed_git(r);
1132 for (p = get_all_packs(r); p; p = p->next) {
1133 if (p->pack_size > max_size) {
1134 second_largest_size = max_size;
1135 max_size = p->pack_size;
1136 } else if (p->pack_size > second_largest_size)
1137 second_largest_size = p->pack_size;
1138 }
1139
1140 result_size = second_largest_size + 1;
1141
1142 /* But limit ourselves to a batch size of 2g */
1143 if (result_size > TWO_GIGABYTES)
1144 result_size = TWO_GIGABYTES;
1145
1146 return result_size;
1147 }
1148
multi_pack_index_repack(struct maintenance_run_opts * opts)1149 static int multi_pack_index_repack(struct maintenance_run_opts *opts)
1150 {
1151 struct child_process child = CHILD_PROCESS_INIT;
1152
1153 child.git_cmd = child.close_object_store = 1;
1154 strvec_pushl(&child.args, "multi-pack-index", "repack", NULL);
1155
1156 if (opts->quiet)
1157 strvec_push(&child.args, "--no-progress");
1158
1159 strvec_pushf(&child.args, "--batch-size=%"PRIuMAX,
1160 (uintmax_t)get_auto_pack_size());
1161
1162 if (run_command(&child))
1163 return error(_("'git multi-pack-index repack' failed"));
1164
1165 return 0;
1166 }
1167
maintenance_task_incremental_repack(struct maintenance_run_opts * opts)1168 static int maintenance_task_incremental_repack(struct maintenance_run_opts *opts)
1169 {
1170 prepare_repo_settings(the_repository);
1171 if (!the_repository->settings.core_multi_pack_index) {
1172 warning(_("skipping incremental-repack task because core.multiPackIndex is disabled"));
1173 return 0;
1174 }
1175
1176 if (multi_pack_index_write(opts))
1177 return 1;
1178 if (multi_pack_index_expire(opts))
1179 return 1;
1180 if (multi_pack_index_repack(opts))
1181 return 1;
1182 return 0;
1183 }
1184
1185 typedef int maintenance_task_fn(struct maintenance_run_opts *opts);
1186
1187 /*
1188 * An auto condition function returns 1 if the task should run
1189 * and 0 if the task should NOT run. See needs_to_gc() for an
1190 * example.
1191 */
1192 typedef int maintenance_auto_fn(void);
1193
1194 struct maintenance_task {
1195 const char *name;
1196 maintenance_task_fn *fn;
1197 maintenance_auto_fn *auto_condition;
1198 unsigned enabled:1;
1199
1200 enum schedule_priority schedule;
1201
1202 /* -1 if not selected. */
1203 int selected_order;
1204 };
1205
1206 enum maintenance_task_label {
1207 TASK_PREFETCH,
1208 TASK_LOOSE_OBJECTS,
1209 TASK_INCREMENTAL_REPACK,
1210 TASK_GC,
1211 TASK_COMMIT_GRAPH,
1212 TASK_PACK_REFS,
1213
1214 /* Leave as final value */
1215 TASK__COUNT
1216 };
1217
1218 static struct maintenance_task tasks[] = {
1219 [TASK_PREFETCH] = {
1220 "prefetch",
1221 maintenance_task_prefetch,
1222 },
1223 [TASK_LOOSE_OBJECTS] = {
1224 "loose-objects",
1225 maintenance_task_loose_objects,
1226 loose_object_auto_condition,
1227 },
1228 [TASK_INCREMENTAL_REPACK] = {
1229 "incremental-repack",
1230 maintenance_task_incremental_repack,
1231 incremental_repack_auto_condition,
1232 },
1233 [TASK_GC] = {
1234 "gc",
1235 maintenance_task_gc,
1236 need_to_gc,
1237 1,
1238 },
1239 [TASK_COMMIT_GRAPH] = {
1240 "commit-graph",
1241 maintenance_task_commit_graph,
1242 should_write_commit_graph,
1243 },
1244 [TASK_PACK_REFS] = {
1245 "pack-refs",
1246 maintenance_task_pack_refs,
1247 NULL,
1248 },
1249 };
1250
compare_tasks_by_selection(const void * a_,const void * b_)1251 static int compare_tasks_by_selection(const void *a_, const void *b_)
1252 {
1253 const struct maintenance_task *a = a_;
1254 const struct maintenance_task *b = b_;
1255
1256 return b->selected_order - a->selected_order;
1257 }
1258
maintenance_run_tasks(struct maintenance_run_opts * opts)1259 static int maintenance_run_tasks(struct maintenance_run_opts *opts)
1260 {
1261 int i, found_selected = 0;
1262 int result = 0;
1263 struct lock_file lk;
1264 struct repository *r = the_repository;
1265 char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path);
1266
1267 if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
1268 /*
1269 * Another maintenance command is running.
1270 *
1271 * If --auto was provided, then it is likely due to a
1272 * recursive process stack. Do not report an error in
1273 * that case.
1274 */
1275 if (!opts->auto_flag && !opts->quiet)
1276 warning(_("lock file '%s' exists, skipping maintenance"),
1277 lock_path);
1278 free(lock_path);
1279 return 0;
1280 }
1281 free(lock_path);
1282
1283 for (i = 0; !found_selected && i < TASK__COUNT; i++)
1284 found_selected = tasks[i].selected_order >= 0;
1285
1286 if (found_selected)
1287 QSORT(tasks, TASK__COUNT, compare_tasks_by_selection);
1288
1289 for (i = 0; i < TASK__COUNT; i++) {
1290 if (found_selected && tasks[i].selected_order < 0)
1291 continue;
1292
1293 if (!found_selected && !tasks[i].enabled)
1294 continue;
1295
1296 if (opts->auto_flag &&
1297 (!tasks[i].auto_condition ||
1298 !tasks[i].auto_condition()))
1299 continue;
1300
1301 if (opts->schedule && tasks[i].schedule < opts->schedule)
1302 continue;
1303
1304 trace2_region_enter("maintenance", tasks[i].name, r);
1305 if (tasks[i].fn(opts)) {
1306 error(_("task '%s' failed"), tasks[i].name);
1307 result = 1;
1308 }
1309 trace2_region_leave("maintenance", tasks[i].name, r);
1310 }
1311
1312 rollback_lock_file(&lk);
1313 return result;
1314 }
1315
initialize_maintenance_strategy(void)1316 static void initialize_maintenance_strategy(void)
1317 {
1318 char *config_str;
1319
1320 if (git_config_get_string("maintenance.strategy", &config_str))
1321 return;
1322
1323 if (!strcasecmp(config_str, "incremental")) {
1324 tasks[TASK_GC].schedule = SCHEDULE_NONE;
1325 tasks[TASK_COMMIT_GRAPH].enabled = 1;
1326 tasks[TASK_COMMIT_GRAPH].schedule = SCHEDULE_HOURLY;
1327 tasks[TASK_PREFETCH].enabled = 1;
1328 tasks[TASK_PREFETCH].schedule = SCHEDULE_HOURLY;
1329 tasks[TASK_INCREMENTAL_REPACK].enabled = 1;
1330 tasks[TASK_INCREMENTAL_REPACK].schedule = SCHEDULE_DAILY;
1331 tasks[TASK_LOOSE_OBJECTS].enabled = 1;
1332 tasks[TASK_LOOSE_OBJECTS].schedule = SCHEDULE_DAILY;
1333 tasks[TASK_PACK_REFS].enabled = 1;
1334 tasks[TASK_PACK_REFS].schedule = SCHEDULE_WEEKLY;
1335 }
1336 }
1337
initialize_task_config(int schedule)1338 static void initialize_task_config(int schedule)
1339 {
1340 int i;
1341 struct strbuf config_name = STRBUF_INIT;
1342 gc_config();
1343
1344 if (schedule)
1345 initialize_maintenance_strategy();
1346
1347 for (i = 0; i < TASK__COUNT; i++) {
1348 int config_value;
1349 char *config_str;
1350
1351 strbuf_reset(&config_name);
1352 strbuf_addf(&config_name, "maintenance.%s.enabled",
1353 tasks[i].name);
1354
1355 if (!git_config_get_bool(config_name.buf, &config_value))
1356 tasks[i].enabled = config_value;
1357
1358 strbuf_reset(&config_name);
1359 strbuf_addf(&config_name, "maintenance.%s.schedule",
1360 tasks[i].name);
1361
1362 if (!git_config_get_string(config_name.buf, &config_str)) {
1363 tasks[i].schedule = parse_schedule(config_str);
1364 free(config_str);
1365 }
1366 }
1367
1368 strbuf_release(&config_name);
1369 }
1370
task_option_parse(const struct option * opt,const char * arg,int unset)1371 static int task_option_parse(const struct option *opt,
1372 const char *arg, int unset)
1373 {
1374 int i, num_selected = 0;
1375 struct maintenance_task *task = NULL;
1376
1377 BUG_ON_OPT_NEG(unset);
1378
1379 for (i = 0; i < TASK__COUNT; i++) {
1380 if (tasks[i].selected_order >= 0)
1381 num_selected++;
1382 if (!strcasecmp(tasks[i].name, arg)) {
1383 task = &tasks[i];
1384 }
1385 }
1386
1387 if (!task) {
1388 error(_("'%s' is not a valid task"), arg);
1389 return 1;
1390 }
1391
1392 if (task->selected_order >= 0) {
1393 error(_("task '%s' cannot be selected multiple times"), arg);
1394 return 1;
1395 }
1396
1397 task->selected_order = num_selected + 1;
1398
1399 return 0;
1400 }
1401
maintenance_run(int argc,const char ** argv,const char * prefix)1402 static int maintenance_run(int argc, const char **argv, const char *prefix)
1403 {
1404 int i;
1405 struct maintenance_run_opts opts;
1406 struct option builtin_maintenance_run_options[] = {
1407 OPT_BOOL(0, "auto", &opts.auto_flag,
1408 N_("run tasks based on the state of the repository")),
1409 OPT_CALLBACK(0, "schedule", &opts.schedule, N_("frequency"),
1410 N_("run tasks based on frequency"),
1411 maintenance_opt_schedule),
1412 OPT_BOOL(0, "quiet", &opts.quiet,
1413 N_("do not report progress or other information over stderr")),
1414 OPT_CALLBACK_F(0, "task", NULL, N_("task"),
1415 N_("run a specific task"),
1416 PARSE_OPT_NONEG, task_option_parse),
1417 OPT_END()
1418 };
1419 memset(&opts, 0, sizeof(opts));
1420
1421 opts.quiet = !isatty(2);
1422
1423 for (i = 0; i < TASK__COUNT; i++)
1424 tasks[i].selected_order = -1;
1425
1426 argc = parse_options(argc, argv, prefix,
1427 builtin_maintenance_run_options,
1428 builtin_maintenance_run_usage,
1429 PARSE_OPT_STOP_AT_NON_OPTION);
1430
1431 if (opts.auto_flag && opts.schedule)
1432 die(_("use at most one of --auto and --schedule=<frequency>"));
1433
1434 initialize_task_config(opts.schedule);
1435
1436 if (argc != 0)
1437 usage_with_options(builtin_maintenance_run_usage,
1438 builtin_maintenance_run_options);
1439 return maintenance_run_tasks(&opts);
1440 }
1441
get_maintpath(void)1442 static char *get_maintpath(void)
1443 {
1444 struct strbuf sb = STRBUF_INIT;
1445 const char *p = the_repository->worktree ?
1446 the_repository->worktree : the_repository->gitdir;
1447
1448 strbuf_realpath(&sb, p, 1);
1449 return strbuf_detach(&sb, NULL);
1450 }
1451
maintenance_register(void)1452 static int maintenance_register(void)
1453 {
1454 int rc;
1455 char *config_value;
1456 struct child_process config_set = CHILD_PROCESS_INIT;
1457 struct child_process config_get = CHILD_PROCESS_INIT;
1458 char *maintpath = get_maintpath();
1459
1460 /* Disable foreground maintenance */
1461 git_config_set("maintenance.auto", "false");
1462
1463 /* Set maintenance strategy, if unset */
1464 if (!git_config_get_string("maintenance.strategy", &config_value))
1465 free(config_value);
1466 else
1467 git_config_set("maintenance.strategy", "incremental");
1468
1469 config_get.git_cmd = 1;
1470 strvec_pushl(&config_get.args, "config", "--global", "--get",
1471 "--fixed-value", "maintenance.repo", maintpath, NULL);
1472 config_get.out = -1;
1473
1474 if (start_command(&config_get)) {
1475 rc = error(_("failed to run 'git config'"));
1476 goto done;
1477 }
1478
1479 /* We already have this value in our config! */
1480 if (!finish_command(&config_get)) {
1481 rc = 0;
1482 goto done;
1483 }
1484
1485 config_set.git_cmd = 1;
1486 strvec_pushl(&config_set.args, "config", "--add", "--global", "maintenance.repo",
1487 maintpath, NULL);
1488
1489 rc = run_command(&config_set);
1490
1491 done:
1492 free(maintpath);
1493 return rc;
1494 }
1495
maintenance_unregister(void)1496 static int maintenance_unregister(void)
1497 {
1498 int rc;
1499 struct child_process config_unset = CHILD_PROCESS_INIT;
1500 char *maintpath = get_maintpath();
1501
1502 config_unset.git_cmd = 1;
1503 strvec_pushl(&config_unset.args, "config", "--global", "--unset",
1504 "--fixed-value", "maintenance.repo", maintpath, NULL);
1505
1506 rc = run_command(&config_unset);
1507 free(maintpath);
1508 return rc;
1509 }
1510
get_frequency(enum schedule_priority schedule)1511 static const char *get_frequency(enum schedule_priority schedule)
1512 {
1513 switch (schedule) {
1514 case SCHEDULE_HOURLY:
1515 return "hourly";
1516 case SCHEDULE_DAILY:
1517 return "daily";
1518 case SCHEDULE_WEEKLY:
1519 return "weekly";
1520 default:
1521 BUG("invalid schedule %d", schedule);
1522 }
1523 }
1524
1525 /*
1526 * get_schedule_cmd` reads the GIT_TEST_MAINT_SCHEDULER environment variable
1527 * to mock the schedulers that `git maintenance start` rely on.
1528 *
1529 * For test purpose, GIT_TEST_MAINT_SCHEDULER can be set to a comma-separated
1530 * list of colon-separated key/value pairs where each pair contains a scheduler
1531 * and its corresponding mock.
1532 *
1533 * * If $GIT_TEST_MAINT_SCHEDULER is not set, return false and leave the
1534 * arguments unmodified.
1535 *
1536 * * If $GIT_TEST_MAINT_SCHEDULER is set, return true.
1537 * In this case, the *cmd value is read as input.
1538 *
1539 * * if the input value *cmd is the key of one of the comma-separated list
1540 * item, then *is_available is set to true and *cmd is modified and becomes
1541 * the mock command.
1542 *
1543 * * if the input value *cmd isn’t the key of any of the comma-separated list
1544 * item, then *is_available is set to false.
1545 *
1546 * Ex.:
1547 * GIT_TEST_MAINT_SCHEDULER not set
1548 * +-------+-------------------------------------------------+
1549 * | Input | Output |
1550 * | *cmd | return code | *cmd | *is_available |
1551 * +-------+-------------+-------------------+---------------+
1552 * | "foo" | false | "foo" (unchanged) | (unchanged) |
1553 * +-------+-------------+-------------------+---------------+
1554 *
1555 * GIT_TEST_MAINT_SCHEDULER set to “foo:./mock_foo.sh,bar:./mock_bar.sh”
1556 * +-------+-------------------------------------------------+
1557 * | Input | Output |
1558 * | *cmd | return code | *cmd | *is_available |
1559 * +-------+-------------+-------------------+---------------+
1560 * | "foo" | true | "./mock.foo.sh" | true |
1561 * | "qux" | true | "qux" (unchanged) | false |
1562 * +-------+-------------+-------------------+---------------+
1563 */
get_schedule_cmd(const char ** cmd,int * is_available)1564 static int get_schedule_cmd(const char **cmd, int *is_available)
1565 {
1566 char *testing = xstrdup_or_null(getenv("GIT_TEST_MAINT_SCHEDULER"));
1567 struct string_list_item *item;
1568 struct string_list list = STRING_LIST_INIT_NODUP;
1569
1570 if (!testing)
1571 return 0;
1572
1573 if (is_available)
1574 *is_available = 0;
1575
1576 string_list_split_in_place(&list, testing, ',', -1);
1577 for_each_string_list_item(item, &list) {
1578 struct string_list pair = STRING_LIST_INIT_NODUP;
1579
1580 if (string_list_split_in_place(&pair, item->string, ':', 2) != 2)
1581 continue;
1582
1583 if (!strcmp(*cmd, pair.items[0].string)) {
1584 *cmd = pair.items[1].string;
1585 if (is_available)
1586 *is_available = 1;
1587 string_list_clear(&list, 0);
1588 UNLEAK(testing);
1589 return 1;
1590 }
1591 }
1592
1593 string_list_clear(&list, 0);
1594 free(testing);
1595 return 1;
1596 }
1597
is_launchctl_available(void)1598 static int is_launchctl_available(void)
1599 {
1600 const char *cmd = "launchctl";
1601 int is_available;
1602 if (get_schedule_cmd(&cmd, &is_available))
1603 return is_available;
1604
1605 #ifdef __APPLE__
1606 return 1;
1607 #else
1608 return 0;
1609 #endif
1610 }
1611
launchctl_service_name(const char * frequency)1612 static char *launchctl_service_name(const char *frequency)
1613 {
1614 struct strbuf label = STRBUF_INIT;
1615 strbuf_addf(&label, "org.git-scm.git.%s", frequency);
1616 return strbuf_detach(&label, NULL);
1617 }
1618
launchctl_service_filename(const char * name)1619 static char *launchctl_service_filename(const char *name)
1620 {
1621 char *expanded;
1622 struct strbuf filename = STRBUF_INIT;
1623 strbuf_addf(&filename, "~/Library/LaunchAgents/%s.plist", name);
1624
1625 expanded = interpolate_path(filename.buf, 1);
1626 if (!expanded)
1627 die(_("failed to expand path '%s'"), filename.buf);
1628
1629 strbuf_release(&filename);
1630 return expanded;
1631 }
1632
launchctl_get_uid(void)1633 static char *launchctl_get_uid(void)
1634 {
1635 return xstrfmt("gui/%d", getuid());
1636 }
1637
launchctl_boot_plist(int enable,const char * filename)1638 static int launchctl_boot_plist(int enable, const char *filename)
1639 {
1640 const char *cmd = "launchctl";
1641 int result;
1642 struct child_process child = CHILD_PROCESS_INIT;
1643 char *uid = launchctl_get_uid();
1644
1645 get_schedule_cmd(&cmd, NULL);
1646 strvec_split(&child.args, cmd);
1647 strvec_pushl(&child.args, enable ? "bootstrap" : "bootout", uid,
1648 filename, NULL);
1649
1650 child.no_stderr = 1;
1651 child.no_stdout = 1;
1652
1653 if (start_command(&child))
1654 die(_("failed to start launchctl"));
1655
1656 result = finish_command(&child);
1657
1658 free(uid);
1659 return result;
1660 }
1661
launchctl_remove_plist(enum schedule_priority schedule)1662 static int launchctl_remove_plist(enum schedule_priority schedule)
1663 {
1664 const char *frequency = get_frequency(schedule);
1665 char *name = launchctl_service_name(frequency);
1666 char *filename = launchctl_service_filename(name);
1667 int result = launchctl_boot_plist(0, filename);
1668 unlink(filename);
1669 free(filename);
1670 free(name);
1671 return result;
1672 }
1673
launchctl_remove_plists(void)1674 static int launchctl_remove_plists(void)
1675 {
1676 return launchctl_remove_plist(SCHEDULE_HOURLY) ||
1677 launchctl_remove_plist(SCHEDULE_DAILY) ||
1678 launchctl_remove_plist(SCHEDULE_WEEKLY);
1679 }
1680
launchctl_list_contains_plist(const char * name,const char * cmd)1681 static int launchctl_list_contains_plist(const char *name, const char *cmd)
1682 {
1683 struct child_process child = CHILD_PROCESS_INIT;
1684
1685 strvec_split(&child.args, cmd);
1686 strvec_pushl(&child.args, "list", name, NULL);
1687
1688 child.no_stderr = 1;
1689 child.no_stdout = 1;
1690
1691 if (start_command(&child))
1692 die(_("failed to start launchctl"));
1693
1694 /* Returns failure if 'name' doesn't exist. */
1695 return !finish_command(&child);
1696 }
1697
launchctl_schedule_plist(const char * exec_path,enum schedule_priority schedule)1698 static int launchctl_schedule_plist(const char *exec_path, enum schedule_priority schedule)
1699 {
1700 int i, fd;
1701 const char *preamble, *repeat;
1702 const char *frequency = get_frequency(schedule);
1703 char *name = launchctl_service_name(frequency);
1704 char *filename = launchctl_service_filename(name);
1705 struct lock_file lk = LOCK_INIT;
1706 static unsigned long lock_file_timeout_ms = ULONG_MAX;
1707 struct strbuf plist = STRBUF_INIT, plist2 = STRBUF_INIT;
1708 struct stat st;
1709 const char *cmd = "launchctl";
1710
1711 get_schedule_cmd(&cmd, NULL);
1712 preamble = "<?xml version=\"1.0\"?>\n"
1713 "<!DOCTYPE plist PUBLIC \"-//Apple//DTD PLIST 1.0//EN\" \"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
1714 "<plist version=\"1.0\">"
1715 "<dict>\n"
1716 "<key>Label</key><string>%s</string>\n"
1717 "<key>ProgramArguments</key>\n"
1718 "<array>\n"
1719 "<string>%s/git</string>\n"
1720 "<string>--exec-path=%s</string>\n"
1721 "<string>for-each-repo</string>\n"
1722 "<string>--config=maintenance.repo</string>\n"
1723 "<string>maintenance</string>\n"
1724 "<string>run</string>\n"
1725 "<string>--schedule=%s</string>\n"
1726 "</array>\n"
1727 "<key>StartCalendarInterval</key>\n"
1728 "<array>\n";
1729 strbuf_addf(&plist, preamble, name, exec_path, exec_path, frequency);
1730
1731 switch (schedule) {
1732 case SCHEDULE_HOURLY:
1733 repeat = "<dict>\n"
1734 "<key>Hour</key><integer>%d</integer>\n"
1735 "<key>Minute</key><integer>0</integer>\n"
1736 "</dict>\n";
1737 for (i = 1; i <= 23; i++)
1738 strbuf_addf(&plist, repeat, i);
1739 break;
1740
1741 case SCHEDULE_DAILY:
1742 repeat = "<dict>\n"
1743 "<key>Day</key><integer>%d</integer>\n"
1744 "<key>Hour</key><integer>0</integer>\n"
1745 "<key>Minute</key><integer>0</integer>\n"
1746 "</dict>\n";
1747 for (i = 1; i <= 6; i++)
1748 strbuf_addf(&plist, repeat, i);
1749 break;
1750
1751 case SCHEDULE_WEEKLY:
1752 strbuf_addstr(&plist,
1753 "<dict>\n"
1754 "<key>Day</key><integer>0</integer>\n"
1755 "<key>Hour</key><integer>0</integer>\n"
1756 "<key>Minute</key><integer>0</integer>\n"
1757 "</dict>\n");
1758 break;
1759
1760 default:
1761 /* unreachable */
1762 break;
1763 }
1764 strbuf_addstr(&plist, "</array>\n</dict>\n</plist>\n");
1765
1766 if (safe_create_leading_directories(filename))
1767 die(_("failed to create directories for '%s'"), filename);
1768
1769 if ((long)lock_file_timeout_ms < 0 &&
1770 git_config_get_ulong("gc.launchctlplistlocktimeoutms",
1771 &lock_file_timeout_ms))
1772 lock_file_timeout_ms = 150;
1773
1774 fd = hold_lock_file_for_update_timeout(&lk, filename, LOCK_DIE_ON_ERROR,
1775 lock_file_timeout_ms);
1776
1777 /*
1778 * Does this file already exist? With the intended contents? Is it
1779 * registered already? Then it does not need to be re-registered.
1780 */
1781 if (!stat(filename, &st) && st.st_size == plist.len &&
1782 strbuf_read_file(&plist2, filename, plist.len) == plist.len &&
1783 !strbuf_cmp(&plist, &plist2) &&
1784 launchctl_list_contains_plist(name, cmd))
1785 rollback_lock_file(&lk);
1786 else {
1787 if (write_in_full(fd, plist.buf, plist.len) < 0 ||
1788 commit_lock_file(&lk))
1789 die_errno(_("could not write '%s'"), filename);
1790
1791 /* bootout might fail if not already running, so ignore */
1792 launchctl_boot_plist(0, filename);
1793 if (launchctl_boot_plist(1, filename))
1794 die(_("failed to bootstrap service %s"), filename);
1795 }
1796
1797 free(filename);
1798 free(name);
1799 strbuf_release(&plist);
1800 strbuf_release(&plist2);
1801 return 0;
1802 }
1803
launchctl_add_plists(void)1804 static int launchctl_add_plists(void)
1805 {
1806 const char *exec_path = git_exec_path();
1807
1808 return launchctl_schedule_plist(exec_path, SCHEDULE_HOURLY) ||
1809 launchctl_schedule_plist(exec_path, SCHEDULE_DAILY) ||
1810 launchctl_schedule_plist(exec_path, SCHEDULE_WEEKLY);
1811 }
1812
launchctl_update_schedule(int run_maintenance,int fd)1813 static int launchctl_update_schedule(int run_maintenance, int fd)
1814 {
1815 if (run_maintenance)
1816 return launchctl_add_plists();
1817 else
1818 return launchctl_remove_plists();
1819 }
1820
is_schtasks_available(void)1821 static int is_schtasks_available(void)
1822 {
1823 const char *cmd = "schtasks";
1824 int is_available;
1825 if (get_schedule_cmd(&cmd, &is_available))
1826 return is_available;
1827
1828 #ifdef GIT_WINDOWS_NATIVE
1829 return 1;
1830 #else
1831 return 0;
1832 #endif
1833 }
1834
schtasks_task_name(const char * frequency)1835 static char *schtasks_task_name(const char *frequency)
1836 {
1837 struct strbuf label = STRBUF_INIT;
1838 strbuf_addf(&label, "Git Maintenance (%s)", frequency);
1839 return strbuf_detach(&label, NULL);
1840 }
1841
schtasks_remove_task(enum schedule_priority schedule)1842 static int schtasks_remove_task(enum schedule_priority schedule)
1843 {
1844 const char *cmd = "schtasks";
1845 int result;
1846 struct strvec args = STRVEC_INIT;
1847 const char *frequency = get_frequency(schedule);
1848 char *name = schtasks_task_name(frequency);
1849
1850 get_schedule_cmd(&cmd, NULL);
1851 strvec_split(&args, cmd);
1852 strvec_pushl(&args, "/delete", "/tn", name, "/f", NULL);
1853
1854 result = run_command_v_opt(args.v, 0);
1855
1856 strvec_clear(&args);
1857 free(name);
1858 return result;
1859 }
1860
schtasks_remove_tasks(void)1861 static int schtasks_remove_tasks(void)
1862 {
1863 return schtasks_remove_task(SCHEDULE_HOURLY) ||
1864 schtasks_remove_task(SCHEDULE_DAILY) ||
1865 schtasks_remove_task(SCHEDULE_WEEKLY);
1866 }
1867
schtasks_schedule_task(const char * exec_path,enum schedule_priority schedule)1868 static int schtasks_schedule_task(const char *exec_path, enum schedule_priority schedule)
1869 {
1870 const char *cmd = "schtasks";
1871 int result;
1872 struct child_process child = CHILD_PROCESS_INIT;
1873 const char *xml;
1874 struct tempfile *tfile;
1875 const char *frequency = get_frequency(schedule);
1876 char *name = schtasks_task_name(frequency);
1877 struct strbuf tfilename = STRBUF_INIT;
1878
1879 get_schedule_cmd(&cmd, NULL);
1880
1881 strbuf_addf(&tfilename, "%s/schedule_%s_XXXXXX",
1882 get_git_common_dir(), frequency);
1883 tfile = xmks_tempfile(tfilename.buf);
1884 strbuf_release(&tfilename);
1885
1886 if (!fdopen_tempfile(tfile, "w"))
1887 die(_("failed to create temp xml file"));
1888
1889 xml = "<?xml version=\"1.0\" ?>\n"
1890 "<Task version=\"1.4\" xmlns=\"http://schemas.microsoft.com/windows/2004/02/mit/task\">\n"
1891 "<Triggers>\n"
1892 "<CalendarTrigger>\n";
1893 fputs(xml, tfile->fp);
1894
1895 switch (schedule) {
1896 case SCHEDULE_HOURLY:
1897 fprintf(tfile->fp,
1898 "<StartBoundary>2020-01-01T01:00:00</StartBoundary>\n"
1899 "<Enabled>true</Enabled>\n"
1900 "<ScheduleByDay>\n"
1901 "<DaysInterval>1</DaysInterval>\n"
1902 "</ScheduleByDay>\n"
1903 "<Repetition>\n"
1904 "<Interval>PT1H</Interval>\n"
1905 "<Duration>PT23H</Duration>\n"
1906 "<StopAtDurationEnd>false</StopAtDurationEnd>\n"
1907 "</Repetition>\n");
1908 break;
1909
1910 case SCHEDULE_DAILY:
1911 fprintf(tfile->fp,
1912 "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1913 "<Enabled>true</Enabled>\n"
1914 "<ScheduleByWeek>\n"
1915 "<DaysOfWeek>\n"
1916 "<Monday />\n"
1917 "<Tuesday />\n"
1918 "<Wednesday />\n"
1919 "<Thursday />\n"
1920 "<Friday />\n"
1921 "<Saturday />\n"
1922 "</DaysOfWeek>\n"
1923 "<WeeksInterval>1</WeeksInterval>\n"
1924 "</ScheduleByWeek>\n");
1925 break;
1926
1927 case SCHEDULE_WEEKLY:
1928 fprintf(tfile->fp,
1929 "<StartBoundary>2020-01-01T00:00:00</StartBoundary>\n"
1930 "<Enabled>true</Enabled>\n"
1931 "<ScheduleByWeek>\n"
1932 "<DaysOfWeek>\n"
1933 "<Sunday />\n"
1934 "</DaysOfWeek>\n"
1935 "<WeeksInterval>1</WeeksInterval>\n"
1936 "</ScheduleByWeek>\n");
1937 break;
1938
1939 default:
1940 break;
1941 }
1942
1943 xml = "</CalendarTrigger>\n"
1944 "</Triggers>\n"
1945 "<Principals>\n"
1946 "<Principal id=\"Author\">\n"
1947 "<LogonType>InteractiveToken</LogonType>\n"
1948 "<RunLevel>LeastPrivilege</RunLevel>\n"
1949 "</Principal>\n"
1950 "</Principals>\n"
1951 "<Settings>\n"
1952 "<MultipleInstancesPolicy>IgnoreNew</MultipleInstancesPolicy>\n"
1953 "<Enabled>true</Enabled>\n"
1954 "<Hidden>true</Hidden>\n"
1955 "<UseUnifiedSchedulingEngine>true</UseUnifiedSchedulingEngine>\n"
1956 "<WakeToRun>false</WakeToRun>\n"
1957 "<ExecutionTimeLimit>PT72H</ExecutionTimeLimit>\n"
1958 "<Priority>7</Priority>\n"
1959 "</Settings>\n"
1960 "<Actions Context=\"Author\">\n"
1961 "<Exec>\n"
1962 "<Command>\"%s\\git.exe\"</Command>\n"
1963 "<Arguments>--exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%s</Arguments>\n"
1964 "</Exec>\n"
1965 "</Actions>\n"
1966 "</Task>\n";
1967 fprintf(tfile->fp, xml, exec_path, exec_path, frequency);
1968 strvec_split(&child.args, cmd);
1969 strvec_pushl(&child.args, "/create", "/tn", name, "/f", "/xml",
1970 get_tempfile_path(tfile), NULL);
1971 close_tempfile_gently(tfile);
1972
1973 child.no_stdout = 1;
1974 child.no_stderr = 1;
1975
1976 if (start_command(&child))
1977 die(_("failed to start schtasks"));
1978 result = finish_command(&child);
1979
1980 delete_tempfile(&tfile);
1981 free(name);
1982 return result;
1983 }
1984
schtasks_schedule_tasks(void)1985 static int schtasks_schedule_tasks(void)
1986 {
1987 const char *exec_path = git_exec_path();
1988
1989 return schtasks_schedule_task(exec_path, SCHEDULE_HOURLY) ||
1990 schtasks_schedule_task(exec_path, SCHEDULE_DAILY) ||
1991 schtasks_schedule_task(exec_path, SCHEDULE_WEEKLY);
1992 }
1993
schtasks_update_schedule(int run_maintenance,int fd)1994 static int schtasks_update_schedule(int run_maintenance, int fd)
1995 {
1996 if (run_maintenance)
1997 return schtasks_schedule_tasks();
1998 else
1999 return schtasks_remove_tasks();
2000 }
2001
2002 MAYBE_UNUSED
check_crontab_process(const char * cmd)2003 static int check_crontab_process(const char *cmd)
2004 {
2005 struct child_process child = CHILD_PROCESS_INIT;
2006
2007 strvec_split(&child.args, cmd);
2008 strvec_push(&child.args, "-l");
2009 child.no_stdin = 1;
2010 child.no_stdout = 1;
2011 child.no_stderr = 1;
2012 child.silent_exec_failure = 1;
2013
2014 if (start_command(&child))
2015 return 0;
2016 /* Ignore exit code, as an empty crontab will return error. */
2017 finish_command(&child);
2018 return 1;
2019 }
2020
is_crontab_available(void)2021 static int is_crontab_available(void)
2022 {
2023 const char *cmd = "crontab";
2024 int is_available;
2025
2026 if (get_schedule_cmd(&cmd, &is_available))
2027 return is_available;
2028
2029 #ifdef __APPLE__
2030 /*
2031 * macOS has cron, but it requires special permissions and will
2032 * create a UI alert when attempting to run this command.
2033 */
2034 return 0;
2035 #else
2036 return check_crontab_process(cmd);
2037 #endif
2038 }
2039
2040 #define BEGIN_LINE "# BEGIN GIT MAINTENANCE SCHEDULE"
2041 #define END_LINE "# END GIT MAINTENANCE SCHEDULE"
2042
crontab_update_schedule(int run_maintenance,int fd)2043 static int crontab_update_schedule(int run_maintenance, int fd)
2044 {
2045 const char *cmd = "crontab";
2046 int result = 0;
2047 int in_old_region = 0;
2048 struct child_process crontab_list = CHILD_PROCESS_INIT;
2049 struct child_process crontab_edit = CHILD_PROCESS_INIT;
2050 FILE *cron_list, *cron_in;
2051 struct strbuf line = STRBUF_INIT;
2052
2053 get_schedule_cmd(&cmd, NULL);
2054 strvec_split(&crontab_list.args, cmd);
2055 strvec_push(&crontab_list.args, "-l");
2056 crontab_list.in = -1;
2057 crontab_list.out = dup(fd);
2058 crontab_list.git_cmd = 0;
2059
2060 if (start_command(&crontab_list))
2061 return error(_("failed to run 'crontab -l'; your system might not support 'cron'"));
2062
2063 /* Ignore exit code, as an empty crontab will return error. */
2064 finish_command(&crontab_list);
2065
2066 /*
2067 * Read from the .lock file, filtering out the old
2068 * schedule while appending the new schedule.
2069 */
2070 cron_list = fdopen(fd, "r");
2071 rewind(cron_list);
2072
2073 strvec_split(&crontab_edit.args, cmd);
2074 crontab_edit.in = -1;
2075 crontab_edit.git_cmd = 0;
2076
2077 if (start_command(&crontab_edit))
2078 return error(_("failed to run 'crontab'; your system might not support 'cron'"));
2079
2080 cron_in = fdopen(crontab_edit.in, "w");
2081 if (!cron_in) {
2082 result = error(_("failed to open stdin of 'crontab'"));
2083 goto done_editing;
2084 }
2085
2086 while (!strbuf_getline_lf(&line, cron_list)) {
2087 if (!in_old_region && !strcmp(line.buf, BEGIN_LINE))
2088 in_old_region = 1;
2089 else if (in_old_region && !strcmp(line.buf, END_LINE))
2090 in_old_region = 0;
2091 else if (!in_old_region)
2092 fprintf(cron_in, "%s\n", line.buf);
2093 }
2094 strbuf_release(&line);
2095
2096 if (run_maintenance) {
2097 struct strbuf line_format = STRBUF_INIT;
2098 const char *exec_path = git_exec_path();
2099
2100 fprintf(cron_in, "%s\n", BEGIN_LINE);
2101 fprintf(cron_in,
2102 "# The following schedule was created by Git\n");
2103 fprintf(cron_in, "# Any edits made in this region might be\n");
2104 fprintf(cron_in,
2105 "# replaced in the future by a Git command.\n\n");
2106
2107 strbuf_addf(&line_format,
2108 "%%s %%s * * %%s \"%s/git\" --exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%%s\n",
2109 exec_path, exec_path);
2110 fprintf(cron_in, line_format.buf, "0", "1-23", "*", "hourly");
2111 fprintf(cron_in, line_format.buf, "0", "0", "1-6", "daily");
2112 fprintf(cron_in, line_format.buf, "0", "0", "0", "weekly");
2113 strbuf_release(&line_format);
2114
2115 fprintf(cron_in, "\n%s\n", END_LINE);
2116 }
2117
2118 fflush(cron_in);
2119 fclose(cron_in);
2120 close(crontab_edit.in);
2121
2122 done_editing:
2123 if (finish_command(&crontab_edit))
2124 result = error(_("'crontab' died"));
2125 else
2126 fclose(cron_list);
2127 return result;
2128 }
2129
real_is_systemd_timer_available(void)2130 static int real_is_systemd_timer_available(void)
2131 {
2132 struct child_process child = CHILD_PROCESS_INIT;
2133
2134 strvec_pushl(&child.args, "systemctl", "--user", "list-timers", NULL);
2135 child.no_stdin = 1;
2136 child.no_stdout = 1;
2137 child.no_stderr = 1;
2138 child.silent_exec_failure = 1;
2139
2140 if (start_command(&child))
2141 return 0;
2142 if (finish_command(&child))
2143 return 0;
2144 return 1;
2145 }
2146
is_systemd_timer_available(void)2147 static int is_systemd_timer_available(void)
2148 {
2149 const char *cmd = "systemctl";
2150 int is_available;
2151
2152 if (get_schedule_cmd(&cmd, &is_available))
2153 return is_available;
2154
2155 return real_is_systemd_timer_available();
2156 }
2157
xdg_config_home_systemd(const char * filename)2158 static char *xdg_config_home_systemd(const char *filename)
2159 {
2160 return xdg_config_home_for("systemd/user", filename);
2161 }
2162
systemd_timer_enable_unit(int enable,enum schedule_priority schedule)2163 static int systemd_timer_enable_unit(int enable,
2164 enum schedule_priority schedule)
2165 {
2166 const char *cmd = "systemctl";
2167 struct child_process child = CHILD_PROCESS_INIT;
2168 const char *frequency = get_frequency(schedule);
2169
2170 /*
2171 * Disabling the systemd unit while it is already disabled makes
2172 * systemctl print an error.
2173 * Let's ignore it since it means we already are in the expected state:
2174 * the unit is disabled.
2175 *
2176 * On the other hand, enabling a systemd unit which is already enabled
2177 * produces no error.
2178 */
2179 if (!enable)
2180 child.no_stderr = 1;
2181
2182 get_schedule_cmd(&cmd, NULL);
2183 strvec_split(&child.args, cmd);
2184 strvec_pushl(&child.args, "--user", enable ? "enable" : "disable",
2185 "--now", NULL);
2186 strvec_pushf(&child.args, "git-maintenance@%s.timer", frequency);
2187
2188 if (start_command(&child))
2189 return error(_("failed to start systemctl"));
2190 if (finish_command(&child))
2191 /*
2192 * Disabling an already disabled systemd unit makes
2193 * systemctl fail.
2194 * Let's ignore this failure.
2195 *
2196 * Enabling an enabled systemd unit doesn't fail.
2197 */
2198 if (enable)
2199 return error(_("failed to run systemctl"));
2200 return 0;
2201 }
2202
systemd_timer_delete_unit_templates(void)2203 static int systemd_timer_delete_unit_templates(void)
2204 {
2205 int ret = 0;
2206 char *filename = xdg_config_home_systemd("git-maintenance@.timer");
2207 if (unlink(filename) && !is_missing_file_error(errno))
2208 ret = error_errno(_("failed to delete '%s'"), filename);
2209 FREE_AND_NULL(filename);
2210
2211 filename = xdg_config_home_systemd("git-maintenance@.service");
2212 if (unlink(filename) && !is_missing_file_error(errno))
2213 ret = error_errno(_("failed to delete '%s'"), filename);
2214
2215 free(filename);
2216 return ret;
2217 }
2218
systemd_timer_delete_units(void)2219 static int systemd_timer_delete_units(void)
2220 {
2221 return systemd_timer_enable_unit(0, SCHEDULE_HOURLY) ||
2222 systemd_timer_enable_unit(0, SCHEDULE_DAILY) ||
2223 systemd_timer_enable_unit(0, SCHEDULE_WEEKLY) ||
2224 systemd_timer_delete_unit_templates();
2225 }
2226
systemd_timer_write_unit_templates(const char * exec_path)2227 static int systemd_timer_write_unit_templates(const char *exec_path)
2228 {
2229 char *filename;
2230 FILE *file;
2231 const char *unit;
2232
2233 filename = xdg_config_home_systemd("git-maintenance@.timer");
2234 if (safe_create_leading_directories(filename)) {
2235 error(_("failed to create directories for '%s'"), filename);
2236 goto error;
2237 }
2238 file = fopen_or_warn(filename, "w");
2239 if (file == NULL)
2240 goto error;
2241
2242 unit = "# This file was created and is maintained by Git.\n"
2243 "# Any edits made in this file might be replaced in the future\n"
2244 "# by a Git command.\n"
2245 "\n"
2246 "[Unit]\n"
2247 "Description=Optimize Git repositories data\n"
2248 "\n"
2249 "[Timer]\n"
2250 "OnCalendar=%i\n"
2251 "Persistent=true\n"
2252 "\n"
2253 "[Install]\n"
2254 "WantedBy=timers.target\n";
2255 if (fputs(unit, file) == EOF) {
2256 error(_("failed to write to '%s'"), filename);
2257 fclose(file);
2258 goto error;
2259 }
2260 if (fclose(file) == EOF) {
2261 error_errno(_("failed to flush '%s'"), filename);
2262 goto error;
2263 }
2264 free(filename);
2265
2266 filename = xdg_config_home_systemd("git-maintenance@.service");
2267 file = fopen_or_warn(filename, "w");
2268 if (file == NULL)
2269 goto error;
2270
2271 unit = "# This file was created and is maintained by Git.\n"
2272 "# Any edits made in this file might be replaced in the future\n"
2273 "# by a Git command.\n"
2274 "\n"
2275 "[Unit]\n"
2276 "Description=Optimize Git repositories data\n"
2277 "\n"
2278 "[Service]\n"
2279 "Type=oneshot\n"
2280 "ExecStart=\"%s/git\" --exec-path=\"%s\" for-each-repo --config=maintenance.repo maintenance run --schedule=%%i\n"
2281 "LockPersonality=yes\n"
2282 "MemoryDenyWriteExecute=yes\n"
2283 "NoNewPrivileges=yes\n"
2284 "RestrictAddressFamilies=AF_UNIX AF_INET AF_INET6\n"
2285 "RestrictNamespaces=yes\n"
2286 "RestrictRealtime=yes\n"
2287 "RestrictSUIDSGID=yes\n"
2288 "SystemCallArchitectures=native\n"
2289 "SystemCallFilter=@system-service\n";
2290 if (fprintf(file, unit, exec_path, exec_path) < 0) {
2291 error(_("failed to write to '%s'"), filename);
2292 fclose(file);
2293 goto error;
2294 }
2295 if (fclose(file) == EOF) {
2296 error_errno(_("failed to flush '%s'"), filename);
2297 goto error;
2298 }
2299 free(filename);
2300 return 0;
2301
2302 error:
2303 free(filename);
2304 systemd_timer_delete_unit_templates();
2305 return -1;
2306 }
2307
systemd_timer_setup_units(void)2308 static int systemd_timer_setup_units(void)
2309 {
2310 const char *exec_path = git_exec_path();
2311
2312 int ret = systemd_timer_write_unit_templates(exec_path) ||
2313 systemd_timer_enable_unit(1, SCHEDULE_HOURLY) ||
2314 systemd_timer_enable_unit(1, SCHEDULE_DAILY) ||
2315 systemd_timer_enable_unit(1, SCHEDULE_WEEKLY);
2316 if (ret)
2317 systemd_timer_delete_units();
2318 return ret;
2319 }
2320
systemd_timer_update_schedule(int run_maintenance,int fd)2321 static int systemd_timer_update_schedule(int run_maintenance, int fd)
2322 {
2323 if (run_maintenance)
2324 return systemd_timer_setup_units();
2325 else
2326 return systemd_timer_delete_units();
2327 }
2328
2329 enum scheduler {
2330 SCHEDULER_INVALID = -1,
2331 SCHEDULER_AUTO,
2332 SCHEDULER_CRON,
2333 SCHEDULER_SYSTEMD,
2334 SCHEDULER_LAUNCHCTL,
2335 SCHEDULER_SCHTASKS,
2336 };
2337
2338 static const struct {
2339 const char *name;
2340 int (*is_available)(void);
2341 int (*update_schedule)(int run_maintenance, int fd);
2342 } scheduler_fn[] = {
2343 [SCHEDULER_CRON] = {
2344 .name = "crontab",
2345 .is_available = is_crontab_available,
2346 .update_schedule = crontab_update_schedule,
2347 },
2348 [SCHEDULER_SYSTEMD] = {
2349 .name = "systemctl",
2350 .is_available = is_systemd_timer_available,
2351 .update_schedule = systemd_timer_update_schedule,
2352 },
2353 [SCHEDULER_LAUNCHCTL] = {
2354 .name = "launchctl",
2355 .is_available = is_launchctl_available,
2356 .update_schedule = launchctl_update_schedule,
2357 },
2358 [SCHEDULER_SCHTASKS] = {
2359 .name = "schtasks",
2360 .is_available = is_schtasks_available,
2361 .update_schedule = schtasks_update_schedule,
2362 },
2363 };
2364
parse_scheduler(const char * value)2365 static enum scheduler parse_scheduler(const char *value)
2366 {
2367 if (!value)
2368 return SCHEDULER_INVALID;
2369 else if (!strcasecmp(value, "auto"))
2370 return SCHEDULER_AUTO;
2371 else if (!strcasecmp(value, "cron") || !strcasecmp(value, "crontab"))
2372 return SCHEDULER_CRON;
2373 else if (!strcasecmp(value, "systemd") ||
2374 !strcasecmp(value, "systemd-timer"))
2375 return SCHEDULER_SYSTEMD;
2376 else if (!strcasecmp(value, "launchctl"))
2377 return SCHEDULER_LAUNCHCTL;
2378 else if (!strcasecmp(value, "schtasks"))
2379 return SCHEDULER_SCHTASKS;
2380 else
2381 return SCHEDULER_INVALID;
2382 }
2383
maintenance_opt_scheduler(const struct option * opt,const char * arg,int unset)2384 static int maintenance_opt_scheduler(const struct option *opt, const char *arg,
2385 int unset)
2386 {
2387 enum scheduler *scheduler = opt->value;
2388
2389 BUG_ON_OPT_NEG(unset);
2390
2391 *scheduler = parse_scheduler(arg);
2392 if (*scheduler == SCHEDULER_INVALID)
2393 return error(_("unrecognized --scheduler argument '%s'"), arg);
2394 return 0;
2395 }
2396
2397 struct maintenance_start_opts {
2398 enum scheduler scheduler;
2399 };
2400
resolve_scheduler(enum scheduler scheduler)2401 static enum scheduler resolve_scheduler(enum scheduler scheduler)
2402 {
2403 if (scheduler != SCHEDULER_AUTO)
2404 return scheduler;
2405
2406 #if defined(__APPLE__)
2407 return SCHEDULER_LAUNCHCTL;
2408
2409 #elif defined(GIT_WINDOWS_NATIVE)
2410 return SCHEDULER_SCHTASKS;
2411
2412 #elif defined(__linux__)
2413 if (is_systemd_timer_available())
2414 return SCHEDULER_SYSTEMD;
2415 else if (is_crontab_available())
2416 return SCHEDULER_CRON;
2417 else
2418 die(_("neither systemd timers nor crontab are available"));
2419
2420 #else
2421 return SCHEDULER_CRON;
2422 #endif
2423 }
2424
validate_scheduler(enum scheduler scheduler)2425 static void validate_scheduler(enum scheduler scheduler)
2426 {
2427 if (scheduler == SCHEDULER_INVALID)
2428 BUG("invalid scheduler");
2429 if (scheduler == SCHEDULER_AUTO)
2430 BUG("resolve_scheduler should have been called before");
2431
2432 if (!scheduler_fn[scheduler].is_available())
2433 die(_("%s scheduler is not available"),
2434 scheduler_fn[scheduler].name);
2435 }
2436
update_background_schedule(const struct maintenance_start_opts * opts,int enable)2437 static int update_background_schedule(const struct maintenance_start_opts *opts,
2438 int enable)
2439 {
2440 unsigned int i;
2441 int result = 0;
2442 struct lock_file lk;
2443 char *lock_path = xstrfmt("%s/schedule", the_repository->objects->odb->path);
2444
2445 if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
2446 free(lock_path);
2447 return error(_("another process is scheduling background maintenance"));
2448 }
2449
2450 for (i = 1; i < ARRAY_SIZE(scheduler_fn); i++) {
2451 if (enable && opts->scheduler == i)
2452 continue;
2453 if (!scheduler_fn[i].is_available())
2454 continue;
2455 scheduler_fn[i].update_schedule(0, get_lock_file_fd(&lk));
2456 }
2457
2458 if (enable)
2459 result = scheduler_fn[opts->scheduler].update_schedule(
2460 1, get_lock_file_fd(&lk));
2461
2462 rollback_lock_file(&lk);
2463
2464 free(lock_path);
2465 return result;
2466 }
2467
2468 static const char *const builtin_maintenance_start_usage[] = {
2469 N_("git maintenance start [--scheduler=<scheduler>]"),
2470 NULL
2471 };
2472
maintenance_start(int argc,const char ** argv,const char * prefix)2473 static int maintenance_start(int argc, const char **argv, const char *prefix)
2474 {
2475 struct maintenance_start_opts opts = { 0 };
2476 struct option options[] = {
2477 OPT_CALLBACK_F(
2478 0, "scheduler", &opts.scheduler, N_("scheduler"),
2479 N_("scheduler to trigger git maintenance run"),
2480 PARSE_OPT_NONEG, maintenance_opt_scheduler),
2481 OPT_END()
2482 };
2483
2484 argc = parse_options(argc, argv, prefix, options,
2485 builtin_maintenance_start_usage, 0);
2486 if (argc)
2487 usage_with_options(builtin_maintenance_start_usage, options);
2488
2489 opts.scheduler = resolve_scheduler(opts.scheduler);
2490 validate_scheduler(opts.scheduler);
2491
2492 if (maintenance_register())
2493 warning(_("failed to add repo to global config"));
2494 return update_background_schedule(&opts, 1);
2495 }
2496
maintenance_stop(void)2497 static int maintenance_stop(void)
2498 {
2499 return update_background_schedule(NULL, 0);
2500 }
2501
2502 static const char builtin_maintenance_usage[] = N_("git maintenance <subcommand> [<options>]");
2503
cmd_maintenance(int argc,const char ** argv,const char * prefix)2504 int cmd_maintenance(int argc, const char **argv, const char *prefix)
2505 {
2506 if (argc < 2 ||
2507 (argc == 2 && !strcmp(argv[1], "-h")))
2508 usage(builtin_maintenance_usage);
2509
2510 if (!strcmp(argv[1], "run"))
2511 return maintenance_run(argc - 1, argv + 1, prefix);
2512 if (!strcmp(argv[1], "start"))
2513 return maintenance_start(argc - 1, argv + 1, prefix);
2514 if (!strcmp(argv[1], "stop"))
2515 return maintenance_stop();
2516 if (!strcmp(argv[1], "register"))
2517 return maintenance_register();
2518 if (!strcmp(argv[1], "unregister"))
2519 return maintenance_unregister();
2520
2521 die(_("invalid subcommand: %s"), argv[1]);
2522 }
2523