Merge branch 'maint-1.8.5' into maint-1.9
[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 "cache.h"
15 #include "parse-options.h"
16 #include "run-command.h"
17 #include "sigchain.h"
18 #include "argv-array.h"
19 #include "commit.h"
20
21 #define FAILED_RUN "failed to run %s"
22
23 static const char * const builtin_gc_usage[] = {
24         N_("git gc [options]"),
25         NULL
26 };
27
28 static int pack_refs = 1;
29 static int aggressive_window = 250;
30 static int gc_auto_threshold = 6700;
31 static int gc_auto_pack_limit = 50;
32 static const char *prune_expire = "2.weeks.ago";
33
34 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
35 static struct argv_array reflog = ARGV_ARRAY_INIT;
36 static struct argv_array repack = ARGV_ARRAY_INIT;
37 static struct argv_array prune = ARGV_ARRAY_INIT;
38 static struct argv_array rerere = ARGV_ARRAY_INIT;
39
40 static char *pidfile;
41
42 static void remove_pidfile(void)
43 {
44         if (pidfile)
45                 unlink(pidfile);
46 }
47
48 static void remove_pidfile_on_signal(int signo)
49 {
50         remove_pidfile();
51         sigchain_pop(signo);
52         raise(signo);
53 }
54
55 static int gc_config(const char *var, const char *value, void *cb)
56 {
57         if (!strcmp(var, "gc.packrefs")) {
58                 if (value && !strcmp(value, "notbare"))
59                         pack_refs = -1;
60                 else
61                         pack_refs = git_config_bool(var, value);
62                 return 0;
63         }
64         if (!strcmp(var, "gc.aggressivewindow")) {
65                 aggressive_window = git_config_int(var, value);
66                 return 0;
67         }
68         if (!strcmp(var, "gc.auto")) {
69                 gc_auto_threshold = git_config_int(var, value);
70                 return 0;
71         }
72         if (!strcmp(var, "gc.autopacklimit")) {
73                 gc_auto_pack_limit = git_config_int(var, value);
74                 return 0;
75         }
76         if (!strcmp(var, "gc.pruneexpire")) {
77                 if (value && strcmp(value, "now")) {
78                         unsigned long now = approxidate("now");
79                         if (approxidate(value) >= now)
80                                 return error(_("Invalid %s: '%s'"), var, value);
81                 }
82                 return git_config_string(&prune_expire, var, value);
83         }
84         return git_default_config(var, value, cb);
85 }
86
87 static int too_many_loose_objects(void)
88 {
89         /*
90          * Quickly check if a "gc" is needed, by estimating how
91          * many loose objects there are.  Because SHA-1 is evenly
92          * distributed, we can check only one and get a reasonable
93          * estimate.
94          */
95         char path[PATH_MAX];
96         const char *objdir = get_object_directory();
97         DIR *dir;
98         struct dirent *ent;
99         int auto_threshold;
100         int num_loose = 0;
101         int needed = 0;
102
103         if (gc_auto_threshold <= 0)
104                 return 0;
105
106         if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
107                 warning(_("insanely long object directory %.*s"), 50, objdir);
108                 return 0;
109         }
110         dir = opendir(path);
111         if (!dir)
112                 return 0;
113
114         auto_threshold = (gc_auto_threshold + 255) / 256;
115         while ((ent = readdir(dir)) != NULL) {
116                 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
117                     ent->d_name[38] != '\0')
118                         continue;
119                 if (++num_loose > auto_threshold) {
120                         needed = 1;
121                         break;
122                 }
123         }
124         closedir(dir);
125         return needed;
126 }
127
128 static int too_many_packs(void)
129 {
130         struct packed_git *p;
131         int cnt;
132
133         if (gc_auto_pack_limit <= 0)
134                 return 0;
135
136         prepare_packed_git();
137         for (cnt = 0, p = packed_git; p; p = p->next) {
138                 if (!p->pack_local)
139                         continue;
140                 if (p->pack_keep)
141                         continue;
142                 /*
143                  * Perhaps check the size of the pack and count only
144                  * very small ones here?
145                  */
146                 cnt++;
147         }
148         return gc_auto_pack_limit <= cnt;
149 }
150
151 static void add_repack_all_option(void)
152 {
153         if (prune_expire && !strcmp(prune_expire, "now"))
154                 argv_array_push(&repack, "-a");
155         else {
156                 argv_array_push(&repack, "-A");
157                 if (prune_expire)
158                         argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
159         }
160 }
161
162 static int need_to_gc(void)
163 {
164         /*
165          * Setting gc.auto to 0 or negative can disable the
166          * automatic gc.
167          */
168         if (gc_auto_threshold <= 0)
169                 return 0;
170
171         /*
172          * If there are too many loose objects, but not too many
173          * packs, we run "repack -d -l".  If there are too many packs,
174          * we run "repack -A -d -l".  Otherwise we tell the caller
175          * there is no need.
176          */
177         if (too_many_packs())
178                 add_repack_all_option();
179         else if (!too_many_loose_objects())
180                 return 0;
181
182         if (run_hook_le(NULL, "pre-auto-gc", NULL))
183                 return 0;
184         return 1;
185 }
186
187 /* return NULL on success, else hostname running the gc */
188 static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
189 {
190         static struct lock_file lock;
191         static char locking_host[128];
192         char my_host[128];
193         struct strbuf sb = STRBUF_INIT;
194         struct stat st;
195         uintmax_t pid;
196         FILE *fp;
197         int fd, should_exit;
198
199         if (pidfile)
200                 /* already locked */
201                 return NULL;
202
203         if (gethostname(my_host, sizeof(my_host)))
204                 strcpy(my_host, "unknown");
205
206         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
207                                        LOCK_DIE_ON_ERROR);
208         if (!force) {
209                 fp = fopen(git_path("gc.pid"), "r");
210                 memset(locking_host, 0, sizeof(locking_host));
211                 should_exit =
212                         fp != NULL &&
213                         !fstat(fileno(fp), &st) &&
214                         /*
215                          * 12 hour limit is very generous as gc should
216                          * never take that long. On the other hand we
217                          * don't really need a strict limit here,
218                          * running gc --auto one day late is not a big
219                          * problem. --force can be used in manual gc
220                          * after the user verifies that no gc is
221                          * running.
222                          */
223                         time(NULL) - st.st_mtime <= 12 * 3600 &&
224                         fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
225                         /* be gentle to concurrent "gc" on remote hosts */
226                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
227                 if (fp != NULL)
228                         fclose(fp);
229                 if (should_exit) {
230                         if (fd >= 0)
231                                 rollback_lock_file(&lock);
232                         *ret_pid = pid;
233                         return locking_host;
234                 }
235         }
236
237         strbuf_addf(&sb, "%"PRIuMAX" %s",
238                     (uintmax_t) getpid(), my_host);
239         write_in_full(fd, sb.buf, sb.len);
240         strbuf_release(&sb);
241         commit_lock_file(&lock);
242
243         pidfile = git_pathdup("gc.pid");
244         sigchain_push_common(remove_pidfile_on_signal);
245         atexit(remove_pidfile);
246
247         return NULL;
248 }
249
250 int cmd_gc(int argc, const char **argv, const char *prefix)
251 {
252         int aggressive = 0;
253         int auto_gc = 0;
254         int quiet = 0;
255         int force = 0;
256         const char *name;
257         pid_t pid;
258
259         struct option builtin_gc_options[] = {
260                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
261                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
262                         N_("prune unreferenced objects"),
263                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
264                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
265                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
266                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
267                 OPT_END()
268         };
269
270         if (argc == 2 && !strcmp(argv[1], "-h"))
271                 usage_with_options(builtin_gc_usage, builtin_gc_options);
272
273         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
274         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
275         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
276         argv_array_pushl(&prune, "prune", "--expire", NULL );
277         argv_array_pushl(&rerere, "rerere", "gc", NULL);
278
279         git_config(gc_config, NULL);
280
281         if (pack_refs < 0)
282                 pack_refs = !is_bare_repository();
283
284         argc = parse_options(argc, argv, prefix, builtin_gc_options,
285                              builtin_gc_usage, 0);
286         if (argc > 0)
287                 usage_with_options(builtin_gc_usage, builtin_gc_options);
288
289         if (aggressive) {
290                 argv_array_push(&repack, "-f");
291                 argv_array_push(&repack, "--depth=250");
292                 if (aggressive_window > 0)
293                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
294         }
295         if (quiet)
296                 argv_array_push(&repack, "-q");
297
298         if (auto_gc) {
299                 /*
300                  * Auto-gc should be least intrusive as possible.
301                  */
302                 if (!need_to_gc())
303                         return 0;
304                 if (!quiet)
305                         fprintf(stderr,
306                                         _("Auto packing the repository for optimum performance. You may also\n"
307                                         "run \"git gc\" manually. See "
308                                         "\"git help gc\" for more information.\n"));
309         } else
310                 add_repack_all_option();
311
312         name = lock_repo_for_gc(force, &pid);
313         if (name) {
314                 if (auto_gc)
315                         return 0; /* be quiet on --auto */
316                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
317                     name, (uintmax_t)pid);
318         }
319
320         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
321                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
322
323         if (run_command_v_opt(reflog.argv, RUN_GIT_CMD))
324                 return error(FAILED_RUN, reflog.argv[0]);
325
326         if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
327                 return error(FAILED_RUN, repack.argv[0]);
328
329         if (prune_expire) {
330                 argv_array_push(&prune, prune_expire);
331                 if (quiet)
332                         argv_array_push(&prune, "--no-progress");
333                 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
334                         return error(FAILED_RUN, prune.argv[0]);
335         }
336
337         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
338                 return error(FAILED_RUN, rerere.argv[0]);
339
340         if (auto_gc && too_many_loose_objects())
341                 warning(_("There are too many unreachable loose objects; "
342                         "run 'git prune' to remove them."));
343
344         return 0;
345 }