trace: measure where the time is spent in the index-heavy operations
[git] / submodule-config.c
1 #include "cache.h"
2 #include "repository.h"
3 #include "config.h"
4 #include "submodule-config.h"
5 #include "submodule.h"
6 #include "strbuf.h"
7 #include "parse-options.h"
8
9 /*
10  * submodule cache lookup structure
11  * There is one shared set of 'struct submodule' entries which can be
12  * looked up by their sha1 blob id of the .gitmodule file and either
13  * using path or name as key.
14  * for_path stores submodule entries with path as key
15  * for_name stores submodule entries with name as key
16  */
17 struct submodule_cache {
18         struct hashmap for_path;
19         struct hashmap for_name;
20         unsigned initialized:1;
21         unsigned gitmodules_read:1;
22 };
23
24 /*
25  * thin wrapper struct needed to insert 'struct submodule' entries to
26  * the hashmap
27  */
28 struct submodule_entry {
29         struct hashmap_entry ent;
30         struct submodule *config;
31 };
32
33 enum lookup_type {
34         lookup_name,
35         lookup_path
36 };
37
38 static int config_path_cmp(const void *unused_cmp_data,
39                            const void *entry,
40                            const void *entry_or_key,
41                            const void *unused_keydata)
42 {
43         const struct submodule_entry *a = entry;
44         const struct submodule_entry *b = entry_or_key;
45
46         return strcmp(a->config->path, b->config->path) ||
47                hashcmp(a->config->gitmodules_sha1, b->config->gitmodules_sha1);
48 }
49
50 static int config_name_cmp(const void *unused_cmp_data,
51                            const void *entry,
52                            const void *entry_or_key,
53                            const void *unused_keydata)
54 {
55         const struct submodule_entry *a = entry;
56         const struct submodule_entry *b = entry_or_key;
57
58         return strcmp(a->config->name, b->config->name) ||
59                hashcmp(a->config->gitmodules_sha1, b->config->gitmodules_sha1);
60 }
61
62 static struct submodule_cache *submodule_cache_alloc(void)
63 {
64         return xcalloc(1, sizeof(struct submodule_cache));
65 }
66
67 static void submodule_cache_init(struct submodule_cache *cache)
68 {
69         hashmap_init(&cache->for_path, config_path_cmp, NULL, 0);
70         hashmap_init(&cache->for_name, config_name_cmp, NULL, 0);
71         cache->initialized = 1;
72 }
73
74 static void free_one_config(struct submodule_entry *entry)
75 {
76         free((void *) entry->config->path);
77         free((void *) entry->config->name);
78         free((void *) entry->config->branch);
79         free((void *) entry->config->update_strategy.command);
80         free(entry->config);
81 }
82
83 static void submodule_cache_clear(struct submodule_cache *cache)
84 {
85         struct hashmap_iter iter;
86         struct submodule_entry *entry;
87
88         if (!cache->initialized)
89                 return;
90
91         /*
92          * We iterate over the name hash here to be symmetric with the
93          * allocation of struct submodule entries. Each is allocated by
94          * their .gitmodule blob sha1 and submodule name.
95          */
96         hashmap_iter_init(&cache->for_name, &iter);
97         while ((entry = hashmap_iter_next(&iter)))
98                 free_one_config(entry);
99
100         hashmap_free(&cache->for_path, 1);
101         hashmap_free(&cache->for_name, 1);
102         cache->initialized = 0;
103         cache->gitmodules_read = 0;
104 }
105
106 void submodule_cache_free(struct submodule_cache *cache)
107 {
108         submodule_cache_clear(cache);
109         free(cache);
110 }
111
112 static unsigned int hash_sha1_string(const unsigned char *sha1,
113                                      const char *string)
114 {
115         return memhash(sha1, 20) + strhash(string);
116 }
117
118 static void cache_put_path(struct submodule_cache *cache,
119                            struct submodule *submodule)
120 {
121         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
122                                              submodule->path);
123         struct submodule_entry *e = xmalloc(sizeof(*e));
124         hashmap_entry_init(e, hash);
125         e->config = submodule;
126         hashmap_put(&cache->for_path, e);
127 }
128
129 static void cache_remove_path(struct submodule_cache *cache,
130                               struct submodule *submodule)
131 {
132         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
133                                              submodule->path);
134         struct submodule_entry e;
135         struct submodule_entry *removed;
136         hashmap_entry_init(&e, hash);
137         e.config = submodule;
138         removed = hashmap_remove(&cache->for_path, &e, NULL);
139         free(removed);
140 }
141
142 static void cache_add(struct submodule_cache *cache,
143                       struct submodule *submodule)
144 {
145         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
146                                              submodule->name);
147         struct submodule_entry *e = xmalloc(sizeof(*e));
148         hashmap_entry_init(e, hash);
149         e->config = submodule;
150         hashmap_add(&cache->for_name, e);
151 }
152
153 static const struct submodule *cache_lookup_path(struct submodule_cache *cache,
154                 const unsigned char *gitmodules_sha1, const char *path)
155 {
156         struct submodule_entry *entry;
157         unsigned int hash = hash_sha1_string(gitmodules_sha1, path);
158         struct submodule_entry key;
159         struct submodule key_config;
160
161         hashcpy(key_config.gitmodules_sha1, gitmodules_sha1);
162         key_config.path = path;
163
164         hashmap_entry_init(&key, hash);
165         key.config = &key_config;
166
167         entry = hashmap_get(&cache->for_path, &key, NULL);
168         if (entry)
169                 return entry->config;
170         return NULL;
171 }
172
173 static struct submodule *cache_lookup_name(struct submodule_cache *cache,
174                 const unsigned char *gitmodules_sha1, const char *name)
175 {
176         struct submodule_entry *entry;
177         unsigned int hash = hash_sha1_string(gitmodules_sha1, name);
178         struct submodule_entry key;
179         struct submodule key_config;
180
181         hashcpy(key_config.gitmodules_sha1, gitmodules_sha1);
182         key_config.name = name;
183
184         hashmap_entry_init(&key, hash);
185         key.config = &key_config;
186
187         entry = hashmap_get(&cache->for_name, &key, NULL);
188         if (entry)
189                 return entry->config;
190         return NULL;
191 }
192
193 static int name_and_item_from_var(const char *var, struct strbuf *name,
194                                   struct strbuf *item)
195 {
196         const char *subsection, *key;
197         int subsection_len, parse;
198         parse = parse_config_key(var, "submodule", &subsection,
199                         &subsection_len, &key);
200         if (parse < 0 || !subsection)
201                 return 0;
202
203         strbuf_add(name, subsection, subsection_len);
204         strbuf_addstr(item, key);
205
206         return 1;
207 }
208
209 static struct submodule *lookup_or_create_by_name(struct submodule_cache *cache,
210                 const unsigned char *gitmodules_sha1, const char *name)
211 {
212         struct submodule *submodule;
213         struct strbuf name_buf = STRBUF_INIT;
214
215         submodule = cache_lookup_name(cache, gitmodules_sha1, name);
216         if (submodule)
217                 return submodule;
218
219         submodule = xmalloc(sizeof(*submodule));
220
221         strbuf_addstr(&name_buf, name);
222         submodule->name = strbuf_detach(&name_buf, NULL);
223
224         submodule->path = NULL;
225         submodule->url = NULL;
226         submodule->update_strategy.type = SM_UPDATE_UNSPECIFIED;
227         submodule->update_strategy.command = NULL;
228         submodule->fetch_recurse = RECURSE_SUBMODULES_NONE;
229         submodule->ignore = NULL;
230         submodule->branch = NULL;
231         submodule->recommend_shallow = -1;
232
233         hashcpy(submodule->gitmodules_sha1, gitmodules_sha1);
234
235         cache_add(cache, submodule);
236
237         return submodule;
238 }
239
240 static int parse_fetch_recurse(const char *opt, const char *arg,
241                                int die_on_error)
242 {
243         switch (git_parse_maybe_bool(arg)) {
244         case 1:
245                 return RECURSE_SUBMODULES_ON;
246         case 0:
247                 return RECURSE_SUBMODULES_OFF;
248         default:
249                 if (!strcmp(arg, "on-demand"))
250                         return RECURSE_SUBMODULES_ON_DEMAND;
251
252                 if (die_on_error)
253                         die("bad %s argument: %s", opt, arg);
254                 else
255                         return RECURSE_SUBMODULES_ERROR;
256         }
257 }
258
259 int parse_submodule_fetchjobs(const char *var, const char *value)
260 {
261         int fetchjobs = git_config_int(var, value);
262         if (fetchjobs < 0)
263                 die(_("negative values not allowed for submodule.fetchjobs"));
264         return fetchjobs;
265 }
266
267 int parse_fetch_recurse_submodules_arg(const char *opt, const char *arg)
268 {
269         return parse_fetch_recurse(opt, arg, 1);
270 }
271
272 int option_fetch_parse_recurse_submodules(const struct option *opt,
273                                           const char *arg, int unset)
274 {
275         int *v;
276
277         if (!opt->value)
278                 return -1;
279
280         v = opt->value;
281
282         if (unset) {
283                 *v = RECURSE_SUBMODULES_OFF;
284         } else {
285                 if (arg)
286                         *v = parse_fetch_recurse_submodules_arg(opt->long_name, arg);
287                 else
288                         *v = RECURSE_SUBMODULES_ON;
289         }
290         return 0;
291 }
292
293 static int parse_update_recurse(const char *opt, const char *arg,
294                                 int die_on_error)
295 {
296         switch (git_parse_maybe_bool(arg)) {
297         case 1:
298                 return RECURSE_SUBMODULES_ON;
299         case 0:
300                 return RECURSE_SUBMODULES_OFF;
301         default:
302                 if (die_on_error)
303                         die("bad %s argument: %s", opt, arg);
304                 return RECURSE_SUBMODULES_ERROR;
305         }
306 }
307
308 int parse_update_recurse_submodules_arg(const char *opt, const char *arg)
309 {
310         return parse_update_recurse(opt, arg, 1);
311 }
312
313 static int parse_push_recurse(const char *opt, const char *arg,
314                                int die_on_error)
315 {
316         switch (git_parse_maybe_bool(arg)) {
317         case 1:
318                 /* There's no simple "on" value when pushing */
319                 if (die_on_error)
320                         die("bad %s argument: %s", opt, arg);
321                 else
322                         return RECURSE_SUBMODULES_ERROR;
323         case 0:
324                 return RECURSE_SUBMODULES_OFF;
325         default:
326                 if (!strcmp(arg, "on-demand"))
327                         return RECURSE_SUBMODULES_ON_DEMAND;
328                 else if (!strcmp(arg, "check"))
329                         return RECURSE_SUBMODULES_CHECK;
330                 else if (!strcmp(arg, "only"))
331                         return RECURSE_SUBMODULES_ONLY;
332                 else if (die_on_error)
333                         die("bad %s argument: %s", opt, arg);
334                 else
335                         return RECURSE_SUBMODULES_ERROR;
336         }
337 }
338
339 int parse_push_recurse_submodules_arg(const char *opt, const char *arg)
340 {
341         return parse_push_recurse(opt, arg, 1);
342 }
343
344 static void warn_multiple_config(const unsigned char *treeish_name,
345                                  const char *name, const char *option)
346 {
347         const char *commit_string = "WORKTREE";
348         if (treeish_name)
349                 commit_string = sha1_to_hex(treeish_name);
350         warning("%s:.gitmodules, multiple configurations found for "
351                         "'submodule.%s.%s'. Skipping second one!",
352                         commit_string, name, option);
353 }
354
355 struct parse_config_parameter {
356         struct submodule_cache *cache;
357         const unsigned char *treeish_name;
358         const unsigned char *gitmodules_sha1;
359         int overwrite;
360 };
361
362 static int parse_config(const char *var, const char *value, void *data)
363 {
364         struct parse_config_parameter *me = data;
365         struct submodule *submodule;
366         struct strbuf name = STRBUF_INIT, item = STRBUF_INIT;
367         int ret = 0;
368
369         /* this also ensures that we only parse submodule entries */
370         if (!name_and_item_from_var(var, &name, &item))
371                 return 0;
372
373         submodule = lookup_or_create_by_name(me->cache,
374                                              me->gitmodules_sha1,
375                                              name.buf);
376
377         if (!strcmp(item.buf, "path")) {
378                 if (!value)
379                         ret = config_error_nonbool(var);
380                 else if (!me->overwrite && submodule->path)
381                         warn_multiple_config(me->treeish_name, submodule->name,
382                                         "path");
383                 else {
384                         if (submodule->path)
385                                 cache_remove_path(me->cache, submodule);
386                         free((void *) submodule->path);
387                         submodule->path = xstrdup(value);
388                         cache_put_path(me->cache, submodule);
389                 }
390         } else if (!strcmp(item.buf, "fetchrecursesubmodules")) {
391                 /* when parsing worktree configurations we can die early */
392                 int die_on_error = is_null_sha1(me->gitmodules_sha1);
393                 if (!me->overwrite &&
394                     submodule->fetch_recurse != RECURSE_SUBMODULES_NONE)
395                         warn_multiple_config(me->treeish_name, submodule->name,
396                                         "fetchrecursesubmodules");
397                 else
398                         submodule->fetch_recurse = parse_fetch_recurse(
399                                                                 var, value,
400                                                                 die_on_error);
401         } else if (!strcmp(item.buf, "ignore")) {
402                 if (!value)
403                         ret = config_error_nonbool(var);
404                 else if (!me->overwrite && submodule->ignore)
405                         warn_multiple_config(me->treeish_name, submodule->name,
406                                         "ignore");
407                 else if (strcmp(value, "untracked") &&
408                          strcmp(value, "dirty") &&
409                          strcmp(value, "all") &&
410                          strcmp(value, "none"))
411                         warning("Invalid parameter '%s' for config option "
412                                         "'submodule.%s.ignore'", value, name.buf);
413                 else {
414                         free((void *) submodule->ignore);
415                         submodule->ignore = xstrdup(value);
416                 }
417         } else if (!strcmp(item.buf, "url")) {
418                 if (!value) {
419                         ret = config_error_nonbool(var);
420                 } else if (!me->overwrite && submodule->url) {
421                         warn_multiple_config(me->treeish_name, submodule->name,
422                                         "url");
423                 } else {
424                         free((void *) submodule->url);
425                         submodule->url = xstrdup(value);
426                 }
427         } else if (!strcmp(item.buf, "update")) {
428                 if (!value)
429                         ret = config_error_nonbool(var);
430                 else if (!me->overwrite &&
431                          submodule->update_strategy.type != SM_UPDATE_UNSPECIFIED)
432                         warn_multiple_config(me->treeish_name, submodule->name,
433                                              "update");
434                 else if (parse_submodule_update_strategy(value,
435                          &submodule->update_strategy) < 0)
436                                 die(_("invalid value for %s"), var);
437         } else if (!strcmp(item.buf, "shallow")) {
438                 if (!me->overwrite && submodule->recommend_shallow != -1)
439                         warn_multiple_config(me->treeish_name, submodule->name,
440                                              "shallow");
441                 else
442                         submodule->recommend_shallow =
443                                 git_config_bool(var, value);
444         } else if (!strcmp(item.buf, "branch")) {
445                 if (!me->overwrite && submodule->branch)
446                         warn_multiple_config(me->treeish_name, submodule->name,
447                                              "branch");
448                 else {
449                         free((void *)submodule->branch);
450                         submodule->branch = xstrdup(value);
451                 }
452         }
453
454         strbuf_release(&name);
455         strbuf_release(&item);
456
457         return ret;
458 }
459
460 static int gitmodule_oid_from_commit(const struct object_id *treeish_name,
461                                      struct object_id *gitmodules_oid,
462                                      struct strbuf *rev)
463 {
464         int ret = 0;
465
466         if (is_null_oid(treeish_name)) {
467                 oidclr(gitmodules_oid);
468                 return 1;
469         }
470
471         strbuf_addf(rev, "%s:.gitmodules", oid_to_hex(treeish_name));
472         if (get_oid(rev->buf, gitmodules_oid) >= 0)
473                 ret = 1;
474
475         return ret;
476 }
477
478 /* This does a lookup of a submodule configuration by name or by path
479  * (key) with on-demand reading of the appropriate .gitmodules from
480  * revisions.
481  */
482 static const struct submodule *config_from(struct submodule_cache *cache,
483                 const struct object_id *treeish_name, const char *key,
484                 enum lookup_type lookup_type)
485 {
486         struct strbuf rev = STRBUF_INIT;
487         unsigned long config_size;
488         char *config = NULL;
489         struct object_id oid;
490         enum object_type type;
491         const struct submodule *submodule = NULL;
492         struct parse_config_parameter parameter;
493
494         /*
495          * If any parameter except the cache is a NULL pointer just
496          * return the first submodule. Can be used to check whether
497          * there are any submodules parsed.
498          */
499         if (!treeish_name || !key) {
500                 struct hashmap_iter iter;
501                 struct submodule_entry *entry;
502
503                 entry = hashmap_iter_first(&cache->for_name, &iter);
504                 if (!entry)
505                         return NULL;
506                 return entry->config;
507         }
508
509         if (!gitmodule_oid_from_commit(treeish_name, &oid, &rev))
510                 goto out;
511
512         switch (lookup_type) {
513         case lookup_name:
514                 submodule = cache_lookup_name(cache, oid.hash, key);
515                 break;
516         case lookup_path:
517                 submodule = cache_lookup_path(cache, oid.hash, key);
518                 break;
519         }
520         if (submodule)
521                 goto out;
522
523         config = read_sha1_file(oid.hash, &type, &config_size);
524         if (!config || type != OBJ_BLOB)
525                 goto out;
526
527         /* fill the submodule config into the cache */
528         parameter.cache = cache;
529         parameter.treeish_name = treeish_name->hash;
530         parameter.gitmodules_sha1 = oid.hash;
531         parameter.overwrite = 0;
532         git_config_from_mem(parse_config, CONFIG_ORIGIN_SUBMODULE_BLOB, rev.buf,
533                         config, config_size, &parameter);
534         strbuf_release(&rev);
535         free(config);
536
537         switch (lookup_type) {
538         case lookup_name:
539                 return cache_lookup_name(cache, oid.hash, key);
540         case lookup_path:
541                 return cache_lookup_path(cache, oid.hash, key);
542         default:
543                 return NULL;
544         }
545
546 out:
547         strbuf_release(&rev);
548         free(config);
549         return submodule;
550 }
551
552 static void submodule_cache_check_init(struct repository *repo)
553 {
554         if (repo->submodule_cache && repo->submodule_cache->initialized)
555                 return;
556
557         if (!repo->submodule_cache)
558                 repo->submodule_cache = submodule_cache_alloc();
559
560         submodule_cache_init(repo->submodule_cache);
561 }
562
563 static int gitmodules_cb(const char *var, const char *value, void *data)
564 {
565         struct repository *repo = data;
566         struct parse_config_parameter parameter;
567
568         parameter.cache = repo->submodule_cache;
569         parameter.treeish_name = NULL;
570         parameter.gitmodules_sha1 = null_sha1;
571         parameter.overwrite = 1;
572
573         return parse_config(var, value, &parameter);
574 }
575
576 void repo_read_gitmodules(struct repository *repo)
577 {
578         submodule_cache_check_init(repo);
579
580         if (repo->worktree) {
581                 char *gitmodules;
582
583                 if (repo_read_index(repo) < 0)
584                         return;
585
586                 gitmodules = repo_worktree_path(repo, GITMODULES_FILE);
587
588                 if (!is_gitmodules_unmerged(repo->index))
589                         git_config_from_file(gitmodules_cb, gitmodules, repo);
590
591                 free(gitmodules);
592         }
593
594         repo->submodule_cache->gitmodules_read = 1;
595 }
596
597 void gitmodules_config_oid(const struct object_id *commit_oid)
598 {
599         struct strbuf rev = STRBUF_INIT;
600         struct object_id oid;
601
602         submodule_cache_check_init(the_repository);
603
604         if (gitmodule_oid_from_commit(commit_oid, &oid, &rev)) {
605                 git_config_from_blob_oid(gitmodules_cb, rev.buf,
606                                          &oid, the_repository);
607         }
608         strbuf_release(&rev);
609
610         the_repository->submodule_cache->gitmodules_read = 1;
611 }
612
613 static void gitmodules_read_check(struct repository *repo)
614 {
615         submodule_cache_check_init(repo);
616
617         /* read the repo's .gitmodules file if it hasn't been already */
618         if (!repo->submodule_cache->gitmodules_read)
619                 repo_read_gitmodules(repo);
620 }
621
622 const struct submodule *submodule_from_name(const struct object_id *treeish_name,
623                 const char *name)
624 {
625         gitmodules_read_check(the_repository);
626         return config_from(the_repository->submodule_cache, treeish_name, name, lookup_name);
627 }
628
629 const struct submodule *submodule_from_path(const struct object_id *treeish_name,
630                 const char *path)
631 {
632         gitmodules_read_check(the_repository);
633         return config_from(the_repository->submodule_cache, treeish_name, path, lookup_path);
634 }
635
636 const struct submodule *submodule_from_cache(struct repository *repo,
637                                              const struct object_id *treeish_name,
638                                              const char *key)
639 {
640         gitmodules_read_check(repo);
641         return config_from(repo->submodule_cache, treeish_name,
642                            key, lookup_path);
643 }
644
645 void submodule_free(void)
646 {
647         if (the_repository->submodule_cache)
648                 submodule_cache_clear(the_repository->submodule_cache);
649 }