Merge branch 'jk/color-diff-plain-is-context' into maint
[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 "lockfile.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 prune_reflogs = 1;
30 static int aggressive_depth = 250;
31 static int aggressive_window = 250;
32 static int gc_auto_threshold = 6700;
33 static int gc_auto_pack_limit = 50;
34 static int detach_auto = 1;
35 static const char *prune_expire = "2.weeks.ago";
36
37 static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
38 static struct argv_array reflog = ARGV_ARRAY_INIT;
39 static struct argv_array repack = ARGV_ARRAY_INIT;
40 static struct argv_array prune = ARGV_ARRAY_INIT;
41 static struct argv_array rerere = ARGV_ARRAY_INIT;
42
43 static char *pidfile;
44
45 static void remove_pidfile(void)
46 {
47         if (pidfile)
48                 unlink(pidfile);
49 }
50
51 static void remove_pidfile_on_signal(int signo)
52 {
53         remove_pidfile();
54         sigchain_pop(signo);
55         raise(signo);
56 }
57
58 static void gc_config(void)
59 {
60         const char *value;
61
62         if (!git_config_get_value("gc.packrefs", &value)) {
63                 if (value && !strcmp(value, "notbare"))
64                         pack_refs = -1;
65                 else
66                         pack_refs = git_config_bool("gc.packrefs", value);
67         }
68
69         git_config_get_int("gc.aggressivewindow", &aggressive_window);
70         git_config_get_int("gc.aggressivedepth", &aggressive_depth);
71         git_config_get_int("gc.auto", &gc_auto_threshold);
72         git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
73         git_config_get_bool("gc.autodetach", &detach_auto);
74
75         if (!git_config_get_string_const("gc.pruneexpire", &prune_expire)) {
76                 if (strcmp(prune_expire, "now")) {
77                         unsigned long now = approxidate("now");
78                         if (approxidate(prune_expire) >= now) {
79                                 git_die_config("gc.pruneexpire", _("Invalid gc.pruneexpire: '%s'"),
80                                                 prune_expire);
81                         }
82                 }
83         }
84         git_config(git_default_config, NULL);
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         char my_host[128];
192         struct strbuf sb = STRBUF_INIT;
193         struct stat st;
194         uintmax_t pid;
195         FILE *fp;
196         int fd;
197
198         if (pidfile)
199                 /* already locked */
200                 return NULL;
201
202         if (gethostname(my_host, sizeof(my_host)))
203                 strcpy(my_host, "unknown");
204
205         fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
206                                        LOCK_DIE_ON_ERROR);
207         if (!force) {
208                 static char locking_host[128];
209                 int should_exit;
210                 fp = fopen(git_path("gc.pid"), "r");
211                 memset(locking_host, 0, sizeof(locking_host));
212                 should_exit =
213                         fp != NULL &&
214                         !fstat(fileno(fp), &st) &&
215                         /*
216                          * 12 hour limit is very generous as gc should
217                          * never take that long. On the other hand we
218                          * don't really need a strict limit here,
219                          * running gc --auto one day late is not a big
220                          * problem. --force can be used in manual gc
221                          * after the user verifies that no gc is
222                          * running.
223                          */
224                         time(NULL) - st.st_mtime <= 12 * 3600 &&
225                         fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
226                         /* be gentle to concurrent "gc" on remote hosts */
227                         (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
228                 if (fp != NULL)
229                         fclose(fp);
230                 if (should_exit) {
231                         if (fd >= 0)
232                                 rollback_lock_file(&lock);
233                         *ret_pid = pid;
234                         return locking_host;
235                 }
236         }
237
238         strbuf_addf(&sb, "%"PRIuMAX" %s",
239                     (uintmax_t) getpid(), my_host);
240         write_in_full(fd, sb.buf, sb.len);
241         strbuf_release(&sb);
242         commit_lock_file(&lock);
243
244         pidfile = git_pathdup("gc.pid");
245         sigchain_push_common(remove_pidfile_on_signal);
246         atexit(remove_pidfile);
247
248         return NULL;
249 }
250
251 static int gc_before_repack(void)
252 {
253         if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
254                 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
255
256         if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
257                 return error(FAILED_RUN, reflog.argv[0]);
258
259         pack_refs = 0;
260         prune_reflogs = 0;
261         return 0;
262 }
263
264 int cmd_gc(int argc, const char **argv, const char *prefix)
265 {
266         int aggressive = 0;
267         int auto_gc = 0;
268         int quiet = 0;
269         int force = 0;
270         const char *name;
271         pid_t pid;
272
273         struct option builtin_gc_options[] = {
274                 OPT__QUIET(&quiet, N_("suppress progress reporting")),
275                 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
276                         N_("prune unreferenced objects"),
277                         PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
278                 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
279                 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
280                 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
281                 OPT_END()
282         };
283
284         if (argc == 2 && !strcmp(argv[1], "-h"))
285                 usage_with_options(builtin_gc_usage, builtin_gc_options);
286
287         argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
288         argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
289         argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
290         argv_array_pushl(&prune, "prune", "--expire", NULL );
291         argv_array_pushl(&rerere, "rerere", "gc", NULL);
292
293         gc_config();
294
295         if (pack_refs < 0)
296                 pack_refs = !is_bare_repository();
297
298         argc = parse_options(argc, argv, prefix, builtin_gc_options,
299                              builtin_gc_usage, 0);
300         if (argc > 0)
301                 usage_with_options(builtin_gc_usage, builtin_gc_options);
302
303         if (aggressive) {
304                 argv_array_push(&repack, "-f");
305                 if (aggressive_depth > 0)
306                         argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
307                 if (aggressive_window > 0)
308                         argv_array_pushf(&repack, "--window=%d", aggressive_window);
309         }
310         if (quiet)
311                 argv_array_push(&repack, "-q");
312
313         if (auto_gc) {
314                 /*
315                  * Auto-gc should be least intrusive as possible.
316                  */
317                 if (!need_to_gc())
318                         return 0;
319                 if (!quiet) {
320                         if (detach_auto)
321                                 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
322                         else
323                                 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
324                         fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
325                 }
326                 if (detach_auto) {
327                         if (gc_before_repack())
328                                 return -1;
329                         /*
330                          * failure to daemonize is ok, we'll continue
331                          * in foreground
332                          */
333                         daemonize();
334                 }
335         } else
336                 add_repack_all_option();
337
338         name = lock_repo_for_gc(force, &pid);
339         if (name) {
340                 if (auto_gc)
341                         return 0; /* be quiet on --auto */
342                 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
343                     name, (uintmax_t)pid);
344         }
345
346         if (gc_before_repack())
347                 return -1;
348
349         if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
350                 return error(FAILED_RUN, repack.argv[0]);
351
352         if (prune_expire) {
353                 argv_array_push(&prune, prune_expire);
354                 if (quiet)
355                         argv_array_push(&prune, "--no-progress");
356                 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
357                         return error(FAILED_RUN, prune.argv[0]);
358         }
359
360         if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
361                 return error(FAILED_RUN, rerere.argv[0]);
362
363         if (auto_gc && too_many_loose_objects())
364                 warning(_("There are too many unreachable loose objects; "
365                         "run 'git prune' to remove them."));
366
367         return 0;
368 }