Merge branch 'bp/name-hash-dirname-fix' into jch
[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 "config.h"
15 #include "tempfile.h"
16 #include "lockfile.h"
17 #include "parse-options.h"
18 #include "run-command.h"
19 #include "sigchain.h"
20 #include "argv-array.h"
21 #include "commit.h"
22 #include "packfile.h"
23
24 #define FAILED_RUN "failed to run %s"
25
26 static const char * const builtin_gc_usage[] = {
27         N_("git gc [<options>]"),
28         NULL
29 };
30
31 static int pack_refs = 1;
32 static int prune_reflogs = 1;
33 static int aggressive_depth = 50;
34 static int aggressive_window = 250;
35 static int gc_auto_threshold = 6700;
36 static int gc_auto_pack_limit = 50;
37 static int detach_auto = 1;
38 static timestamp_t gc_log_expire_time;
39 static const char *gc_log_expire = "1.day.ago";
40 static const char *prune_expire = "2.weeks.ago";
41 static const char *prune_worktrees_expire = "3.months.ago";
42
43 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
44 static struct argv_array reflog = ARGV_ARRAY_INIT;
45 static struct argv_array repack = ARGV_ARRAY_INIT;
46 static struct argv_array prune = ARGV_ARRAY_INIT;
47 static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
48 static struct argv_array rerere = ARGV_ARRAY_INIT;
49
50 static struct tempfile *pidfile;
51 static struct lock_file log_lock;
52
53 static struct string_list pack_garbage = STRING_LIST_INIT_DUP;
54
55 static void clean_pack_garbage(void)
56 {
57         int i;
58         for (i = 0; i < pack_garbage.nr; i++)
59                 unlink_or_warn(pack_garbage.items[i].string);
60         string_list_clear(&pack_garbage, 0);
61 }
62
63 static void report_pack_garbage(unsigned seen_bits, const char *path)
64 {
65         if (seen_bits == PACKDIR_FILE_IDX)
66                 string_list_append(&pack_garbage, path);
67 }
68
69 static void process_log_file(void)
70 {
71         struct stat st;
72         if (fstat(get_lock_file_fd(&log_lock), &st)) {
73                 /*
74                  * Perhaps there was an i/o error or another
75                  * unlikely situation.  Try to make a note of
76                  * this in gc.log along with any existing
77                  * messages.
78                  */
79                 int saved_errno = errno;
80                 fprintf(stderr, _("Failed to fstat %s: %s"),
81                         get_tempfile_path(log_lock.tempfile),
82                         strerror(saved_errno));
83                 fflush(stderr);
84                 commit_lock_file(&log_lock);
85                 errno = saved_errno;
86         } else if (st.st_size) {
87                 /* There was some error recorded in the lock file */
88                 commit_lock_file(&log_lock);
89         } else {
90                 /* No error, clean up any old gc.log */
91                 unlink(git_path("gc.log"));
92                 rollback_lock_file(&log_lock);
93         }
94 }
95
96 static void process_log_file_at_exit(void)
97 {
98         fflush(stderr);
99         process_log_file();
100 }
101
102 static void process_log_file_on_signal(int signo)
103 {
104         process_log_file();
105         sigchain_pop(signo);
106         raise(signo);
107 }
108
109 static void gc_config(void)
110 {
111         const char *value;
112
113         if (!git_config_get_value("gc.packrefs", &value)) {
114                 if (value && !strcmp(value, "notbare"))
115                         pack_refs = -1;
116                 else
117                         pack_refs = git_config_bool("gc.packrefs", value);
118         }
119
120         git_config_get_int("gc.aggressivewindow", &aggressive_window);
121         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
122         git_config_get_int("gc.auto", &gc_auto_threshold);
123         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
124         git_config_get_bool("gc.autodetach", &detach_auto);
125         git_config_get_expiry("gc.pruneexpire", &prune_expire);
126         git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
127         git_config_get_expiry("gc.logexpiry", &gc_log_expire);
128
129         git_config(git_default_config, NULL);
130 }
131
132 static int too_many_loose_objects(void)
133 {
134         /*
135          * Quickly check if a "gc" is needed, by estimating how
136          * many loose objects there are.  Because SHA-1 is evenly
137          * distributed, we can check only one and get a reasonable
138          * estimate.
139          */
140         DIR *dir;
141         struct dirent *ent;
142         int auto_threshold;
143         int num_loose = 0;
144         int needed = 0;
145
146         if (gc_auto_threshold <= 0)
147                 return 0;
148
149         dir = opendir(git_path("objects/17"));
150         if (!dir)
151                 return 0;
152
153         auto_threshold = DIV_ROUND_UP(gc_auto_threshold, 256);
154         while ((ent = readdir(dir)) != NULL) {
155                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
156                     ent->d_name[38] != '\0')
157                         continue;
158                 if (++num_loose > auto_threshold) {
159                         needed = 1;
160                         break;
161                 }
162         }
163         closedir(dir);
164         return needed;
165 }
166
167 static int too_many_packs(void)
168 {
169         struct packed_git *p;
170         int cnt;
171
172         if (gc_auto_pack_limit <= 0)
173                 return 0;
174
175         prepare_packed_git();
176         for (cnt = 0, p = packed_git; p; p = p->next) {
177                 if (!p->pack_local)
178                         continue;
179                 if (p->pack_keep)
180                         continue;
181                 /*
182                  * Perhaps check the size of the pack and count only
183                  * very small ones here?
184                  */
185                 cnt++;
186         }
187         return gc_auto_pack_limit < cnt;
188 }
189
190 static void add_repack_all_option(void)
191 {
192         if (prune_expire && !strcmp(prune_expire, "now"))
193                 argv_array_push(&repack, "-a");
194         else {
195                 argv_array_push(&repack, "-A");
196                 if (prune_expire)
197                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
198         }
199 }
200
201 static void add_repack_incremental_option(void)
202 {
203        argv_array_push(&repack, "--no-write-bitmap-index");
204 }
205
206 static int need_to_gc(void)
207 {
208         /*
209          * Setting gc.auto to 0 or negative can disable the
210          * automatic gc.
211          */
212         if (gc_auto_threshold <= 0)
213                 return 0;
214
215         /*
216          * If there are too many loose objects, but not too many
217          * packs, we run "repack -d -l".  If there are too many packs,
218          * we run "repack -A -d -l".  Otherwise we tell the caller
219          * there is no need.
220          */
221         if (too_many_packs())
222                 add_repack_all_option();
223         else if (too_many_loose_objects())
224                 add_repack_incremental_option();
225         else
226                 return 0;
227
228         if (run_hook_le(NULL, "pre-auto-gc", NULL))
229                 return 0;
230         return 1;
231 }
232
233 /* return NULL on success, else hostname running the gc */
234 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
235 {
236         static struct lock_file lock;
237         char my_host[HOST_NAME_MAX + 1];
238         struct strbuf sb = STRBUF_INIT;
239         struct stat st;
240         uintmax_t pid;
241         FILE *fp;
242         int fd;
243         char *pidfile_path;
244
245         if (is_tempfile_active(pidfile))
246                 /* already locked */
247                 return NULL;
248
249         if (xgethostname(my_host, sizeof(my_host)))
250                 xsnprintf(my_host, sizeof(my_host), "unknown");
251
252         pidfile_path = git_pathdup("gc.pid");
253         fd = hold_lock_file_for_update(&lock, pidfile_path,
254                                        LOCK_DIE_ON_ERROR);
255         if (!force) {
256                 static char locking_host[HOST_NAME_MAX + 1];
257                 static char *scan_fmt;
258                 int should_exit;
259
260                 if (!scan_fmt)
261                         scan_fmt = xstrfmt("%s %%%ds", "%"SCNuMAX, HOST_NAME_MAX);
262                 fp = fopen(pidfile_path, "r");
263                 memset(locking_host, 0, sizeof(locking_host));
264                 should_exit =
265                         fp != NULL &&
266                         !fstat(fileno(fp), &st) &&
267                         /*
268                          * 12 hour limit is very generous as gc should
269                          * never take that long. On the other hand we
270                          * don't really need a strict limit here,
271                          * running gc --auto one day late is not a big
272                          * problem. --force can be used in manual gc
273                          * after the user verifies that no gc is
274                          * running.
275                          */
276                         time(NULL) - st.st_mtime <= 12 * 3600 &&
277                         fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
278                         /* be gentle to concurrent "gc" on remote hosts */
279                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
280                 if (fp != NULL)
281                         fclose(fp);
282                 if (should_exit) {
283                         if (fd >= 0)
284                                 rollback_lock_file(&lock);
285                         *ret_pid = pid;
286                         free(pidfile_path);
287                         return locking_host;
288                 }
289         }
290
291         strbuf_addf(&sb, "%"PRIuMAX" %s",
292                     (uintmax_t) getpid(), my_host);
293         write_in_full(fd, sb.buf, sb.len);
294         strbuf_release(&sb);
295         commit_lock_file(&lock);
296         pidfile = register_tempfile(pidfile_path);
297         free(pidfile_path);
298         return NULL;
299 }
300
301 static int report_last_gc_error(void)
302 {
303         struct strbuf sb = STRBUF_INIT;
304         int ret = 0;
305         struct stat st;
306         char *gc_log_path = git_pathdup("gc.log");
307
308         if (stat(gc_log_path, &st)) {
309                 if (errno == ENOENT)
310                         goto done;
311
312                 ret = error_errno(_("Can't stat %s"), gc_log_path);
313                 goto done;
314         }
315
316         if (st.st_mtime < gc_log_expire_time)
317                 goto done;
318
319         ret = strbuf_read_file(&sb, gc_log_path, 0);
320         if (ret > 0)
321                 ret = error(_("The last gc run reported the following. "
322                                "Please correct the root cause\n"
323                                "and remove %s.\n"
324                                "Automatic cleanup will not be performed "
325                                "until the file is removed.\n\n"
326                                "%s"),
327                             gc_log_path, sb.buf);
328         strbuf_release(&sb);
329 done:
330         free(gc_log_path);
331         return ret;
332 }
333
334 static int gc_before_repack(void)
335 {
336         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
337                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
338
339         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
340                 return error(FAILED_RUN, reflog.argv[0]);
341
342         pack_refs = 0;
343         prune_reflogs = 0;
344         return 0;
345 }
346
347 int cmd_gc(int argc, const char **argv, const char *prefix)
348 {
349         int aggressive = 0;
350         int auto_gc = 0;
351         int quiet = 0;
352         int force = 0;
353         const char *name;
354         pid_t pid;
355         int daemonized = 0;
356
357         struct option builtin_gc_options[] = {
358                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
359                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
360                         N_("prune unreferenced objects"),
361                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
362                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
363                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
364                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
365                 OPT_END()
366         };
367
368         if (argc == 2 && !strcmp(argv[1], "-h"))
369                 usage_with_options(builtin_gc_usage, builtin_gc_options);
370
371         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
372         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
373         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
374         argv_array_pushl(&prune, "prune", "--expire", NULL);
375         argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
376         argv_array_pushl(&rerere, "rerere", "gc", NULL);
377
378         /* default expiry time, overwritten in gc_config */
379         gc_config();
380         if (parse_expiry_date(gc_log_expire, &gc_log_expire_time))
381                 die(_("Failed to parse gc.logexpiry value %s"), gc_log_expire);
382
383         if (pack_refs < 0)
384                 pack_refs = !is_bare_repository();
385
386         argc = parse_options(argc, argv, prefix, builtin_gc_options,
387                              builtin_gc_usage, 0);
388         if (argc > 0)
389                 usage_with_options(builtin_gc_usage, builtin_gc_options);
390
391         if (aggressive) {
392                 argv_array_push(&repack, "-f");
393                 if (aggressive_depth > 0)
394                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
395                 if (aggressive_window > 0)
396                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
397         }
398         if (quiet)
399                 argv_array_push(&repack, "-q");
400
401         if (auto_gc) {
402                 /*
403                  * Auto-gc should be least intrusive as possible.
404                  */
405                 if (!need_to_gc())
406                         return 0;
407                 if (!quiet) {
408                         if (detach_auto)
409                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
410                         else
411                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
412                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
413                 }
414                 if (detach_auto) {
415                         if (report_last_gc_error())
416                                 return -1;
417
418                         if (lock_repo_for_gc(force, &pid))
419                                 return 0;
420                         if (gc_before_repack())
421                                 return -1;
422                         delete_tempfile(&pidfile);
423
424                         /*
425                          * failure to daemonize is ok, we'll continue
426                          * in foreground
427                          */
428                         daemonized = !daemonize();
429                 }
430         } else
431                 add_repack_all_option();
432
433         name = lock_repo_for_gc(force, &pid);
434         if (name) {
435                 if (auto_gc)
436                         return 0; /* be quiet on --auto */
437                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
438                     name, (uintmax_t)pid);
439         }
440
441         if (daemonized) {
442                 hold_lock_file_for_update(&log_lock,
443                                           git_path("gc.log"),
444                                           LOCK_DIE_ON_ERROR);
445                 dup2(get_lock_file_fd(&log_lock), 2);
446                 sigchain_push_common(process_log_file_on_signal);
447                 atexit(process_log_file_at_exit);
448         }
449
450         if (gc_before_repack())
451                 return -1;
452
453         if (!repository_format_precious_objects) {
454                 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
455                         return error(FAILED_RUN, repack.argv[0]);
456
457                 if (prune_expire) {
458                         argv_array_push(&prune, prune_expire);
459                         if (quiet)
460                                 argv_array_push(&prune, "--no-progress");
461                         if (repository_format_partial_clone)
462                                 argv_array_push(&prune,
463                                                 "--exclude-promisor-objects");
464                         if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
465                                 return error(FAILED_RUN, prune.argv[0]);
466                 }
467         }
468
469         if (prune_worktrees_expire) {
470                 argv_array_push(&prune_worktrees, prune_worktrees_expire);
471                 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
472                         return error(FAILED_RUN, prune_worktrees.argv[0]);
473         }
474
475         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
476                 return error(FAILED_RUN, rerere.argv[0]);
477
478         report_garbage = report_pack_garbage;
479         reprepare_packed_git();
480         if (pack_garbage.nr > 0)
481                 clean_pack_garbage();
482
483         if (auto_gc && too_many_loose_objects())
484                 warning(_("There are too many unreachable loose objects; "
485                         "run 'git prune' to remove them."));
486
487         if (!daemonized)
488                 unlink(git_path("gc.log"));
489
490         return 0;
491 }