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