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