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