Merge branch 'pb/set-url-docfix'
[git] / builtin / gc.c
1 /*
2  * git gc builtin command
3  *
4  * Cleanup unreachable files and optimize the repository.
5  *
6  * Copyright (c) 2007 James Bowes
7  *
8  * Based on git-gc.sh, which is
9  *
10  * Copyright (c) 2006 Shawn O. Pearce
11  */
12
13 #include "builtin.h"
14 #include "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
32 #define FAILED_RUN "failed to run %s"
33
34 static const char * const builtin_gc_usage[] = {
35         N_("git gc [<options>]"),
36         NULL
37 };
38
39 static int pack_refs = 1;
40 static int prune_reflogs = 1;
41 static int aggressive_depth = 50;
42 static int aggressive_window = 250;
43 static int gc_auto_threshold = 6700;
44 static int gc_auto_pack_limit = 50;
45 static int detach_auto = 1;
46 static timestamp_t gc_log_expire_time;
47 static const char *gc_log_expire = "1.day.ago";
48 static const char *prune_expire = "2.weeks.ago";
49 static const char *prune_worktrees_expire = "3.months.ago";
50 static unsigned long big_pack_threshold;
51 static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
52
53 static struct strvec pack_refs_cmd = STRVEC_INIT;
54 static struct strvec reflog = STRVEC_INIT;
55 static struct strvec repack = STRVEC_INIT;
56 static struct strvec prune = STRVEC_INIT;
57 static struct strvec prune_worktrees = STRVEC_INIT;
58 static struct strvec rerere = STRVEC_INIT;
59
60 static struct tempfile *pidfile;
61 static struct lock_file log_lock;
62
63 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
64
65 static void clean_pack_garbage(void)
66 {
67         int i;
68         for (i = 0; i < pack_garbage.nr; i++)
69                 unlink_or_warn(pack_garbage.items[i].string);
70         string_list_clear(&pack_garbage, 0);
71 }
72
73 static void report_pack_garbage(unsigned seen_bits, const char *path)
74 {
75         if (seen_bits == PACKDIR_FILE_IDX)
76                 string_list_append(&pack_garbage, path);
77 }
78
79 static void process_log_file(void)
80 {
81         struct stat st;
82         if (fstat(get_lock_file_fd(&log_lock), &st)) {
83                 /*
84                  * Perhaps there was an i/o error or another
85                  * unlikely situation.  Try to make a note of
86                  * this in gc.log along with any existing
87                  * messages.
88                  */
89                 int saved_errno = errno;
90                 fprintf(stderr, _("Failed to fstat %s: %s"),
91                         get_tempfile_path(log_lock.tempfile),
92                         strerror(saved_errno));
93                 fflush(stderr);
94                 commit_lock_file(&log_lock);
95                 errno = saved_errno;
96         } else if (st.st_size) {
97                 /* There was some error recorded in the lock file */
98                 commit_lock_file(&log_lock);
99         } else {
100                 /* No error, clean up any old gc.log */
101                 unlink(git_path("gc.log"));
102                 rollback_lock_file(&log_lock);
103         }
104 }
105
106 static void process_log_file_at_exit(void)
107 {
108         fflush(stderr);
109         process_log_file();
110 }
111
112 static void process_log_file_on_signal(int signo)
113 {
114         process_log_file();
115         sigchain_pop(signo);
116         raise(signo);
117 }
118
119 static int gc_config_is_timestamp_never(const char *var)
120 {
121         const char *value;
122         timestamp_t expire;
123
124         if (!git_config_get_value(var, &value) && value) {
125                 if (parse_expiry_date(value, &expire))
126                         die(_("failed to parse '%s' value '%s'"), var, value);
127                 return expire == 0;
128         }
129         return 0;
130 }
131
132 static void gc_config(void)
133 {
134         const char *value;
135
136         if (!git_config_get_value("gc.packrefs", &value)) {
137                 if (value && !strcmp(value, "notbare"))
138                         pack_refs = -1;
139                 else
140                         pack_refs = git_config_bool("gc.packrefs", value);
141         }
142
143         if (gc_config_is_timestamp_never("gc.reflogexpire") &&
144             gc_config_is_timestamp_never("gc.reflogexpireunreachable"))
145                 prune_reflogs = 0;
146
147         git_config_get_int("gc.aggressivewindow", &aggressive_window);
148         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
149         git_config_get_int("gc.auto", &gc_auto_threshold);
150         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
151         git_config_get_bool("gc.autodetach", &detach_auto);
152         git_config_get_expiry("gc.pruneexpire", &prune_expire);
153         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
154         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
155
156         git_config_get_ulong("gc.bigpackthreshold", &big_pack_threshold);
157         git_config_get_ulong("pack.deltacachesize", &max_delta_cache_size);
158
159         git_config(git_default_config, NULL);
160 }
161
162 static int too_many_loose_objects(void)
163 {
164         /*
165          * Quickly check if a "gc" is needed, by estimating how
166          * many loose objects there are.  Because SHA-1 is evenly
167          * distributed, we can check only one and get a reasonable
168          * estimate.
169          */
170         DIR *dir;
171         struct dirent *ent;
172         int auto_threshold;
173         int num_loose = 0;
174         int needed = 0;
175         const unsigned hexsz_loose = the_hash_algo->hexsz - 2;
176
177         dir = opendir(git_path("objects/17"));
178         if (!dir)
179                 return 0;
180
181         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
182         while ((ent = readdir(dir)) != NULL) {
183                 if (strspn(ent->d_name, "0123456789abcdef") != hexsz_loose ||
184                     ent->d_name[hexsz_loose] != '\0')
185                         continue;
186                 if (++num_loose > auto_threshold) {
187                         needed = 1;
188                         break;
189                 }
190         }
191         closedir(dir);
192         return needed;
193 }
194
195 static struct packed_git *find_base_packs(struct string_list *packs,
196                                           unsigned long limit)
197 {
198         struct packed_git *p, *base = NULL;
199
200         for (p = get_all_packs(the_repository); p; p = p->next) {
201                 if (!p->pack_local)
202                         continue;
203                 if (limit) {
204                         if (p->pack_size >= limit)
205                                 string_list_append(packs, p->pack_name);
206                 } else if (!base || base->pack_size < p->pack_size) {
207                         base = p;
208                 }
209         }
210
211         if (base)
212                 string_list_append(packs, base->pack_name);
213
214         return base;
215 }
216
217 static int too_many_packs(void)
218 {
219         struct packed_git *p;
220         int cnt;
221
222         if (gc_auto_pack_limit <= 0)
223                 return 0;
224
225         for (cnt = 0, p = get_all_packs(the_repository); p; p = p->next) {
226                 if (!p->pack_local)
227                         continue;
228                 if (p->pack_keep)
229                         continue;
230                 /*
231                  * Perhaps check the size of the pack and count only
232                  * very small ones here?
233                  */
234                 cnt++;
235         }
236         return gc_auto_pack_limit < cnt;
237 }
238
239 static uint64_t total_ram(void)
240 {
241 #if defined(HAVE_SYSINFO)
242         struct sysinfo si;
243
244         if (!sysinfo(&si))
245                 return si.totalram;
246 #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM))
247         int64_t physical_memory;
248         int mib[2];
249         size_t length;
250
251         mib[0] = CTL_HW;
252 # if defined(HW_MEMSIZE)
253         mib[1] = HW_MEMSIZE;
254 # else
255         mib[1] = HW_PHYSMEM;
256 # endif
257         length = sizeof(int64_t);
258         if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
259                 return physical_memory;
260 #elif defined(GIT_WINDOWS_NATIVE)
261         MEMORYSTATUSEX memInfo;
262
263         memInfo.dwLength = sizeof(MEMORYSTATUSEX);
264         if (GlobalMemoryStatusEx(&memInfo))
265                 return memInfo.ullTotalPhys;
266 #endif
267         return 0;
268 }
269
270 static uint64_t estimate_repack_memory(struct packed_git *pack)
271 {
272         unsigned long nr_objects = approximate_object_count();
273         size_t os_cache, heap;
274
275         if (!pack || !nr_objects)
276                 return 0;
277
278         /*
279          * First we have to scan through at least one pack.
280          * Assume enough room in OS file cache to keep the entire pack
281          * or we may accidentally evict data of other processes from
282          * the cache.
283          */
284         os_cache = pack->pack_size + pack->index_size;
285         /* then pack-objects needs lots more for book keeping */
286         heap = sizeof(struct object_entry) * nr_objects;
287         /*
288          * internal rev-list --all --objects takes up some memory too,
289          * let's say half of it is for blobs
290          */
291         heap += sizeof(struct blob) * nr_objects / 2;
292         /*
293          * and the other half is for trees (commits and tags are
294          * usually insignificant)
295          */
296         heap += sizeof(struct tree) * nr_objects / 2;
297         /* and then obj_hash[], underestimated in fact */
298         heap += sizeof(struct object *) * nr_objects;
299         /* revindex is used also */
300         heap += sizeof(struct revindex_entry) * nr_objects;
301         /*
302          * read_sha1_file() (either at delta calculation phase, or
303          * writing phase) also fills up the delta base cache
304          */
305         heap += delta_base_cache_limit;
306         /* and of course pack-objects has its own delta cache */
307         heap += max_delta_cache_size;
308
309         return os_cache + heap;
310 }
311
312 static int keep_one_pack(struct string_list_item *item, void *data)
313 {
314         strvec_pushf(&repack, "--keep-pack=%s", basename(item->string));
315         return 0;
316 }
317
318 static void add_repack_all_option(struct string_list *keep_pack)
319 {
320         if (prune_expire && !strcmp(prune_expire, "now"))
321                 strvec_push(&repack, "-a");
322         else {
323                 strvec_push(&repack, "-A");
324                 if (prune_expire)
325                         strvec_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
326         }
327
328         if (keep_pack)
329                 for_each_string_list(keep_pack, keep_one_pack, NULL);
330 }
331
332 static void add_repack_incremental_option(void)
333 {
334         strvec_push(&repack, "--no-write-bitmap-index");
335 }
336
337 static int need_to_gc(void)
338 {
339         /*
340          * Setting gc.auto to 0 or negative can disable the
341          * automatic gc.
342          */
343         if (gc_auto_threshold <= 0)
344                 return 0;
345
346         /*
347          * If there are too many loose objects, but not too many
348          * packs, we run "repack -d -l".  If there are too many packs,
349          * we run "repack -A -d -l".  Otherwise we tell the caller
350          * there is no need.
351          */
352         if (too_many_packs()) {
353                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
354
355                 if (big_pack_threshold) {
356                         find_base_packs(&keep_pack, big_pack_threshold);
357                         if (keep_pack.nr >= gc_auto_pack_limit) {
358                                 big_pack_threshold = 0;
359                                 string_list_clear(&keep_pack, 0);
360                                 find_base_packs(&keep_pack, 0);
361                         }
362                 } else {
363                         struct packed_git *p = find_base_packs(&keep_pack, 0);
364                         uint64_t mem_have, mem_want;
365
366                         mem_have = total_ram();
367                         mem_want = estimate_repack_memory(p);
368
369                         /*
370                          * Only allow 1/2 of memory for pack-objects, leave
371                          * the rest for the OS and other processes in the
372                          * system.
373                          */
374                         if (!mem_have || mem_want < mem_have / 2)
375                                 string_list_clear(&keep_pack, 0);
376                 }
377
378                 add_repack_all_option(&keep_pack);
379                 string_list_clear(&keep_pack, 0);
380         } else if (too_many_loose_objects())
381                 add_repack_incremental_option();
382         else
383                 return 0;
384
385         if (run_hook_le(NULL, "pre-auto-gc", NULL))
386                 return 0;
387         return 1;
388 }
389
390 /* return NULL on success, else hostname running the gc */
391 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
392 {
393         struct lock_file lock = LOCK_INIT;
394         char my_host[HOST_NAME_MAX + 1];
395         struct strbuf sb = STRBUF_INIT;
396         struct stat st;
397         uintmax_t pid;
398         FILE *fp;
399         int fd;
400         char *pidfile_path;
401
402         if (is_tempfile_active(pidfile))
403                 /* already locked */
404                 return NULL;
405
406         if (xgethostname(my_host, sizeof(my_host)))
407                 xsnprintf(my_host, sizeof(my_host), "unknown");
408
409         pidfile_path = git_pathdup("gc.pid");
410         fd = hold_lock_file_for_update(&lock, pidfile_path,
411                                        LOCK_DIE_ON_ERROR);
412         if (!force) {
413                 static char locking_host[HOST_NAME_MAX + 1];
414                 static char *scan_fmt;
415                 int should_exit;
416
417                 if (!scan_fmt)
418                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
419                 fp = fopen(pidfile_path, "r");
420                 memset(locking_host, 0, sizeof(locking_host));
421                 should_exit =
422                         fp != NULL &&
423                         !fstat(fileno(fp), &st) &&
424                         /*
425                          * 12 hour limit is very generous as gc should
426                          * never take that long. On the other hand we
427                          * don't really need a strict limit here,
428                          * running gc --auto one day late is not a big
429                          * problem. --force can be used in manual gc
430                          * after the user verifies that no gc is
431                          * running.
432                          */
433                         time(NULL) - st.st_mtime <= 12 * 3600 &&
434                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
435                         /* be gentle to concurrent "gc" on remote hosts */
436                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
437                 if (fp != NULL)
438                         fclose(fp);
439                 if (should_exit) {
440                         if (fd >= 0)
441                                 rollback_lock_file(&lock);
442                         *ret_pid = pid;
443                         free(pidfile_path);
444                         return locking_host;
445                 }
446         }
447
448         strbuf_addf(&sb, "%"PRIuMAX" %s",
449                     (uintmax_t) getpid(), my_host);
450         write_in_full(fd, sb.buf, sb.len);
451         strbuf_release(&sb);
452         commit_lock_file(&lock);
453         pidfile = register_tempfile(pidfile_path);
454         free(pidfile_path);
455         return NULL;
456 }
457
458 /*
459  * Returns 0 if there was no previous error and gc can proceed, 1 if
460  * gc should not proceed due to an error in the last run. Prints a
461  * message and returns -1 if an error occurred while reading gc.log
462  */
463 static int report_last_gc_error(void)
464 {
465         struct strbuf sb = STRBUF_INIT;
466         int ret = 0;
467         ssize_t len;
468         struct stat st;
469         char *gc_log_path = git_pathdup("gc.log");
470
471         if (stat(gc_log_path, &st)) {
472                 if (errno == ENOENT)
473                         goto done;
474
475                 ret = error_errno(_("cannot stat '%s'"), gc_log_path);
476                 goto done;
477         }
478
479         if (st.st_mtime < gc_log_expire_time)
480                 goto done;
481
482         len = strbuf_read_file(&sb, gc_log_path, 0);
483         if (len < 0)
484                 ret = error_errno(_("cannot read '%s'"), gc_log_path);
485         else if (len > 0) {
486                 /*
487                  * A previous gc failed.  Report the error, and don't
488                  * bother with an automatic gc run since it is likely
489                  * to fail in the same way.
490                  */
491                 warning(_("The last gc run reported the following. "
492                                "Please correct the root cause\n"
493                                "and remove %s.\n"
494                                "Automatic cleanup will not be performed "
495                                "until the file is removed.\n\n"
496                                "%s"),
497                             gc_log_path, sb.buf);
498                 ret = 1;
499         }
500         strbuf_release(&sb);
501 done:
502         free(gc_log_path);
503         return ret;
504 }
505
506 static void gc_before_repack(void)
507 {
508         /*
509          * We may be called twice, as both the pre- and
510          * post-daemonized phases will call us, but running these
511          * commands more than once is pointless and wasteful.
512          */
513         static int done = 0;
514         if (done++)
515                 return;
516
517         if (pack_refs && run_command_v_opt(pack_refs_cmd.v, RUN_GIT_CMD))
518                 die(FAILED_RUN, pack_refs_cmd.v[0]);
519
520         if (prune_reflogs && run_command_v_opt(reflog.v, RUN_GIT_CMD))
521                 die(FAILED_RUN, reflog.v[0]);
522 }
523
524 int cmd_gc(int argc, const char **argv, const char *prefix)
525 {
526         int aggressive = 0;
527         int auto_gc = 0;
528         int quiet = 0;
529         int force = 0;
530         const char *name;
531         pid_t pid;
532         int daemonized = 0;
533         int keep_base_pack = -1;
534         timestamp_t dummy;
535
536         struct option builtin_gc_options[] = {
537                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
538                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
539                         N_("prune unreferenced objects"),
540                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
541                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
542                 OPT_BOOL_F(0, "auto", &auto_gc, N_("enable auto-gc mode"),
543                            PARSE_OPT_NOCOMPLETE),
544                 OPT_BOOL_F(0, "force", &force,
545                            N_("force running gc even if there may be another gc running"),
546                            PARSE_OPT_NOCOMPLETE),
547                 OPT_BOOL(0, "keep-largest-pack", &keep_base_pack,
548                          N_("repack all other packs except the largest pack")),
549                 OPT_END()
550         };
551
552         if (argc == 2 && !strcmp(argv[1], "-h"))
553                 usage_with_options(builtin_gc_usage, builtin_gc_options);
554
555         strvec_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
556         strvec_pushl(&reflog, "reflog", "expire", "--all", NULL);
557         strvec_pushl(&repack, "repack", "-d", "-l", NULL);
558         strvec_pushl(&prune, "prune", "--expire", NULL);
559         strvec_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
560         strvec_pushl(&rerere, "rerere", "gc", NULL);
561
562         /* default expiry time, overwritten in gc_config */
563         gc_config();
564         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
565                 die(_("failed to parse gc.logexpiry value %s"), gc_log_expire);
566
567         if (pack_refs < 0)
568                 pack_refs = !is_bare_repository();
569
570         argc = parse_options(argc, argv, prefix, builtin_gc_options,
571                              builtin_gc_usage, 0);
572         if (argc > 0)
573                 usage_with_options(builtin_gc_usage, builtin_gc_options);
574
575         if (prune_expire && parse_expiry_date(prune_expire, &dummy))
576                 die(_("failed to parse prune expiry value %s"), prune_expire);
577
578         if (aggressive) {
579                 strvec_push(&repack, "-f");
580                 if (aggressive_depth > 0)
581                         strvec_pushf(&repack, "--depth=%d", aggressive_depth);
582                 if (aggressive_window > 0)
583                         strvec_pushf(&repack, "--window=%d", aggressive_window);
584         }
585         if (quiet)
586                 strvec_push(&repack, "-q");
587
588         if (auto_gc) {
589                 /*
590                  * Auto-gc should be least intrusive as possible.
591                  */
592                 if (!need_to_gc())
593                         return 0;
594                 if (!quiet) {
595                         if (detach_auto)
596                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
597                         else
598                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
599                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
600                 }
601                 if (detach_auto) {
602                         int ret = report_last_gc_error();
603                         if (ret < 0)
604                                 /* an I/O error occurred, already reported */
605                                 exit(128);
606                         if (ret == 1)
607                                 /* Last gc --auto failed. Skip this one. */
608                                 return 0;
609
610                         if (lock_repo_for_gc(force, &pid))
611                                 return 0;
612                         gc_before_repack(); /* dies on failure */
613                         delete_tempfile(&pidfile);
614
615                         /*
616                          * failure to daemonize is ok, we'll continue
617                          * in foreground
618                          */
619                         daemonized = !daemonize();
620                 }
621         } else {
622                 struct string_list keep_pack = STRING_LIST_INIT_NODUP;
623
624                 if (keep_base_pack != -1) {
625                         if (keep_base_pack)
626                                 find_base_packs(&keep_pack, 0);
627                 } else if (big_pack_threshold) {
628                         find_base_packs(&keep_pack, big_pack_threshold);
629                 }
630
631                 add_repack_all_option(&keep_pack);
632                 string_list_clear(&keep_pack, 0);
633         }
634
635         name = lock_repo_for_gc(force, &pid);
636         if (name) {
637                 if (auto_gc)
638                         return 0; /* be quiet on --auto */
639                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
640                     name, (uintmax_t)pid);
641         }
642
643         if (daemonized) {
644                 hold_lock_file_for_update(&log_lock,
645                                           git_path("gc.log"),
646                                           LOCK_DIE_ON_ERROR);
647                 dup2(get_lock_file_fd(&log_lock), 2);
648                 sigchain_push_common(process_log_file_on_signal);
649                 atexit(process_log_file_at_exit);
650         }
651
652         gc_before_repack();
653
654         if (!repository_format_precious_objects) {
655                 close_object_store(the_repository->objects);
656                 if (run_command_v_opt(repack.v, RUN_GIT_CMD))
657                         die(FAILED_RUN, repack.v[0]);
658
659                 if (prune_expire) {
660                         strvec_push(&prune, prune_expire);
661                         if (quiet)
662                                 strvec_push(&prune, "--no-progress");
663                         if (has_promisor_remote())
664                                 strvec_push(&prune,
665                                             "--exclude-promisor-objects");
666                         if (run_command_v_opt(prune.v, RUN_GIT_CMD))
667                                 die(FAILED_RUN, prune.v[0]);
668                 }
669         }
670
671         if (prune_worktrees_expire) {
672                 strvec_push(&prune_worktrees, prune_worktrees_expire);
673                 if (run_command_v_opt(prune_worktrees.v, RUN_GIT_CMD))
674                         die(FAILED_RUN, prune_worktrees.v[0]);
675         }
676
677         if (run_command_v_opt(rerere.v, RUN_GIT_CMD))
678                 die(FAILED_RUN, rerere.v[0]);
679
680         report_garbage = report_pack_garbage;
681         reprepare_packed_git(the_repository);
682         if (pack_garbage.nr > 0) {
683                 close_object_store(the_repository->objects);
684                 clean_pack_garbage();
685         }
686
687         prepare_repo_settings(the_repository);
688         if (the_repository->settings.gc_write_commit_graph == 1)
689                 write_commit_graph_reachable(the_repository->objects->odb,
690                                              !quiet && !daemonized ? COMMIT_GRAPH_WRITE_PROGRESS : 0,
691                                              NULL);
692
693         if (auto_gc && too_many_loose_objects())
694                 warning(_("There are too many unreachable loose objects; "
695                         "run 'git prune' to remove them."));
696
697         if (!daemonized)
698                 unlink(git_path("gc.log"));
699
700         return 0;
701 }