Merge branch 'nd/the-index' into md/list-objects-filter-by-depth
[git] / submodule.c
1 #define NO_THE_INDEX_COMPATIBILITY_MACROS
2
3 #include "cache.h"
4 #include "repository.h"
5 #include "config.h"
6 #include "submodule-config.h"
7 #include "submodule.h"
8 #include "dir.h"
9 #include "diff.h"
10 #include "commit.h"
11 #include "revision.h"
12 #include "run-command.h"
13 #include "diffcore.h"
14 #include "refs.h"
15 #include "string-list.h"
16 #include "sha1-array.h"
17 #include "argv-array.h"
18 #include "blob.h"
19 #include "thread-utils.h"
20 #include "quote.h"
21 #include "remote.h"
22 #include "worktree.h"
23 #include "parse-options.h"
24 #include "object-store.h"
25 #include "commit-reach.h"
26
27 static int config_update_recurse_submodules = RECURSE_SUBMODULES_OFF;
28 static struct string_list changed_submodule_names = STRING_LIST_INIT_DUP;
29 static int initialized_fetch_ref_tips;
30 static struct oid_array ref_tips_before_fetch;
31 static struct oid_array ref_tips_after_fetch;
32
33 /*
34  * Check if the .gitmodules file is unmerged. Parsing of the .gitmodules file
35  * will be disabled because we can't guess what might be configured in
36  * .gitmodules unless the user resolves the conflict.
37  */
38 int is_gitmodules_unmerged(const struct index_state *istate)
39 {
40         int pos = index_name_pos(istate, GITMODULES_FILE, strlen(GITMODULES_FILE));
41         if (pos < 0) { /* .gitmodules not found or isn't merged */
42                 pos = -1 - pos;
43                 if (istate->cache_nr > pos) {  /* there is a .gitmodules */
44                         const struct cache_entry *ce = istate->cache[pos];
45                         if (ce_namelen(ce) == strlen(GITMODULES_FILE) &&
46                             !strcmp(ce->name, GITMODULES_FILE))
47                                 return 1;
48                 }
49         }
50
51         return 0;
52 }
53
54 /*
55  * Check if the .gitmodules file is safe to write.
56  *
57  * Writing to the .gitmodules file requires that the file exists in the
58  * working tree or, if it doesn't, that a brand new .gitmodules file is going
59  * to be created (i.e. it's neither in the index nor in the current branch).
60  *
61  * It is not safe to write to .gitmodules if it's not in the working tree but
62  * it is in the index or in the current branch, because writing new values
63  * (and staging them) would blindly overwrite ALL the old content.
64  */
65 int is_writing_gitmodules_ok(void)
66 {
67         struct object_id oid;
68         return file_exists(GITMODULES_FILE) ||
69                 (get_oid(GITMODULES_INDEX, &oid) < 0 && get_oid(GITMODULES_HEAD, &oid) < 0);
70 }
71
72 /*
73  * Check if the .gitmodules file has unstaged modifications.  This must be
74  * checked before allowing modifications to the .gitmodules file with the
75  * intention to stage them later, because when continuing we would stage the
76  * modifications the user didn't stage herself too. That might change in a
77  * future version when we learn to stage the changes we do ourselves without
78  * staging any previous modifications.
79  */
80 int is_staging_gitmodules_ok(struct index_state *istate)
81 {
82         int pos = index_name_pos(istate, GITMODULES_FILE, strlen(GITMODULES_FILE));
83
84         if ((pos >= 0) && (pos < istate->cache_nr)) {
85                 struct stat st;
86                 if (lstat(GITMODULES_FILE, &st) == 0 &&
87                     ie_match_stat(istate, istate->cache[pos], &st, 0) & DATA_CHANGED)
88                         return 0;
89         }
90
91         return 1;
92 }
93
94 static int for_each_remote_ref_submodule(const char *submodule,
95                                          each_ref_fn fn, void *cb_data)
96 {
97         return refs_for_each_remote_ref(get_submodule_ref_store(submodule),
98                                         fn, cb_data);
99 }
100
101 /*
102  * Try to update the "path" entry in the "submodule.<name>" section of the
103  * .gitmodules file. Return 0 only if a .gitmodules file was found, a section
104  * with the correct path=<oldpath> setting was found and we could update it.
105  */
106 int update_path_in_gitmodules(const char *oldpath, const char *newpath)
107 {
108         struct strbuf entry = STRBUF_INIT;
109         const struct submodule *submodule;
110         int ret;
111
112         if (!file_exists(GITMODULES_FILE)) /* Do nothing without .gitmodules */
113                 return -1;
114
115         if (is_gitmodules_unmerged(the_repository->index))
116                 die(_("Cannot change unmerged .gitmodules, resolve merge conflicts first"));
117
118         submodule = submodule_from_path(the_repository, &null_oid, oldpath);
119         if (!submodule || !submodule->name) {
120                 warning(_("Could not find section in .gitmodules where path=%s"), oldpath);
121                 return -1;
122         }
123         strbuf_addstr(&entry, "submodule.");
124         strbuf_addstr(&entry, submodule->name);
125         strbuf_addstr(&entry, ".path");
126         ret = config_set_in_gitmodules_file_gently(entry.buf, newpath);
127         strbuf_release(&entry);
128         return ret;
129 }
130
131 /*
132  * Try to remove the "submodule.<name>" section from .gitmodules where the given
133  * path is configured. Return 0 only if a .gitmodules file was found, a section
134  * with the correct path=<path> setting was found and we could remove it.
135  */
136 int remove_path_from_gitmodules(const char *path)
137 {
138         struct strbuf sect = STRBUF_INIT;
139         const struct submodule *submodule;
140
141         if (!file_exists(GITMODULES_FILE)) /* Do nothing without .gitmodules */
142                 return -1;
143
144         if (is_gitmodules_unmerged(the_repository->index))
145                 die(_("Cannot change unmerged .gitmodules, resolve merge conflicts first"));
146
147         submodule = submodule_from_path(the_repository, &null_oid, path);
148         if (!submodule || !submodule->name) {
149                 warning(_("Could not find section in .gitmodules where path=%s"), path);
150                 return -1;
151         }
152         strbuf_addstr(&sect, "submodule.");
153         strbuf_addstr(&sect, submodule->name);
154         if (git_config_rename_section_in_file(GITMODULES_FILE, sect.buf, NULL) < 0) {
155                 /* Maybe the user already did that, don't error out here */
156                 warning(_("Could not remove .gitmodules entry for %s"), path);
157                 strbuf_release(&sect);
158                 return -1;
159         }
160         strbuf_release(&sect);
161         return 0;
162 }
163
164 void stage_updated_gitmodules(struct index_state *istate)
165 {
166         if (add_file_to_index(istate, GITMODULES_FILE, 0))
167                 die(_("staging updated .gitmodules failed"));
168 }
169
170 /* TODO: remove this function, use repo_submodule_init instead. */
171 int add_submodule_odb(const char *path)
172 {
173         struct strbuf objects_directory = STRBUF_INIT;
174         int ret = 0;
175
176         ret = strbuf_git_path_submodule(&objects_directory, path, "objects/");
177         if (ret)
178                 goto done;
179         if (!is_directory(objects_directory.buf)) {
180                 ret = -1;
181                 goto done;
182         }
183         add_to_alternates_memory(objects_directory.buf);
184 done:
185         strbuf_release(&objects_directory);
186         return ret;
187 }
188
189 void set_diffopt_flags_from_submodule_config(struct diff_options *diffopt,
190                                              const char *path)
191 {
192         const struct submodule *submodule = submodule_from_path(the_repository,
193                                                                 &null_oid, path);
194         if (submodule) {
195                 const char *ignore;
196                 char *key;
197
198                 key = xstrfmt("submodule.%s.ignore", submodule->name);
199                 if (repo_config_get_string_const(the_repository, key, &ignore))
200                         ignore = submodule->ignore;
201                 free(key);
202
203                 if (ignore)
204                         handle_ignore_submodules_arg(diffopt, ignore);
205                 else if (is_gitmodules_unmerged(the_repository->index))
206                         diffopt->flags.ignore_submodules = 1;
207         }
208 }
209
210 /* Cheap function that only determines if we're interested in submodules at all */
211 int git_default_submodule_config(const char *var, const char *value, void *cb)
212 {
213         if (!strcmp(var, "submodule.recurse")) {
214                 int v = git_config_bool(var, value) ?
215                         RECURSE_SUBMODULES_ON : RECURSE_SUBMODULES_OFF;
216                 config_update_recurse_submodules = v;
217         }
218         return 0;
219 }
220
221 int option_parse_recurse_submodules_worktree_updater(const struct option *opt,
222                                                      const char *arg, int unset)
223 {
224         if (unset) {
225                 config_update_recurse_submodules = RECURSE_SUBMODULES_OFF;
226                 return 0;
227         }
228         if (arg)
229                 config_update_recurse_submodules =
230                         parse_update_recurse_submodules_arg(opt->long_name,
231                                                             arg);
232         else
233                 config_update_recurse_submodules = RECURSE_SUBMODULES_ON;
234
235         return 0;
236 }
237
238 /*
239  * Determine if a submodule has been initialized at a given 'path'
240  */
241 int is_submodule_active(struct repository *repo, const char *path)
242 {
243         int ret = 0;
244         char *key = NULL;
245         char *value = NULL;
246         const struct string_list *sl;
247         const struct submodule *module;
248
249         module = submodule_from_path(repo, &null_oid, path);
250
251         /* early return if there isn't a path->module mapping */
252         if (!module)
253                 return 0;
254
255         /* submodule.<name>.active is set */
256         key = xstrfmt("submodule.%s.active", module->name);
257         if (!repo_config_get_bool(repo, key, &ret)) {
258                 free(key);
259                 return ret;
260         }
261         free(key);
262
263         /* submodule.active is set */
264         sl = repo_config_get_value_multi(repo, "submodule.active");
265         if (sl) {
266                 struct pathspec ps;
267                 struct argv_array args = ARGV_ARRAY_INIT;
268                 const struct string_list_item *item;
269
270                 for_each_string_list_item(item, sl) {
271                         argv_array_push(&args, item->string);
272                 }
273
274                 parse_pathspec(&ps, 0, 0, NULL, args.argv);
275                 ret = match_pathspec(repo->index, &ps, path, strlen(path), 0, NULL, 1);
276
277                 argv_array_clear(&args);
278                 clear_pathspec(&ps);
279                 return ret;
280         }
281
282         /* fallback to checking if the URL is set */
283         key = xstrfmt("submodule.%s.url", module->name);
284         ret = !repo_config_get_string(repo, key, &value);
285
286         free(value);
287         free(key);
288         return ret;
289 }
290
291 int is_submodule_populated_gently(const char *path, int *return_error_code)
292 {
293         int ret = 0;
294         char *gitdir = xstrfmt("%s/.git", path);
295
296         if (resolve_gitdir_gently(gitdir, return_error_code))
297                 ret = 1;
298
299         free(gitdir);
300         return ret;
301 }
302
303 /*
304  * Dies if the provided 'prefix' corresponds to an unpopulated submodule
305  */
306 void die_in_unpopulated_submodule(const struct index_state *istate,
307                                   const char *prefix)
308 {
309         int i, prefixlen;
310
311         if (!prefix)
312                 return;
313
314         prefixlen = strlen(prefix);
315
316         for (i = 0; i < istate->cache_nr; i++) {
317                 struct cache_entry *ce = istate->cache[i];
318                 int ce_len = ce_namelen(ce);
319
320                 if (!S_ISGITLINK(ce->ce_mode))
321                         continue;
322                 if (prefixlen <= ce_len)
323                         continue;
324                 if (strncmp(ce->name, prefix, ce_len))
325                         continue;
326                 if (prefix[ce_len] != '/')
327                         continue;
328
329                 die(_("in unpopulated submodule '%s'"), ce->name);
330         }
331 }
332
333 /*
334  * Dies if any paths in the provided pathspec descends into a submodule
335  */
336 void die_path_inside_submodule(const struct index_state *istate,
337                                const struct pathspec *ps)
338 {
339         int i, j;
340
341         for (i = 0; i < istate->cache_nr; i++) {
342                 struct cache_entry *ce = istate->cache[i];
343                 int ce_len = ce_namelen(ce);
344
345                 if (!S_ISGITLINK(ce->ce_mode))
346                         continue;
347
348                 for (j = 0; j < ps->nr ; j++) {
349                         const struct pathspec_item *item = &ps->items[j];
350
351                         if (item->len <= ce_len)
352                                 continue;
353                         if (item->match[ce_len] != '/')
354                                 continue;
355                         if (strncmp(ce->name, item->match, ce_len))
356                                 continue;
357                         if (item->len == ce_len + 1)
358                                 continue;
359
360                         die(_("Pathspec '%s' is in submodule '%.*s'"),
361                             item->original, ce_len, ce->name);
362                 }
363         }
364 }
365
366 enum submodule_update_type parse_submodule_update_type(const char *value)
367 {
368         if (!strcmp(value, "none"))
369                 return SM_UPDATE_NONE;
370         else if (!strcmp(value, "checkout"))
371                 return SM_UPDATE_CHECKOUT;
372         else if (!strcmp(value, "rebase"))
373                 return SM_UPDATE_REBASE;
374         else if (!strcmp(value, "merge"))
375                 return SM_UPDATE_MERGE;
376         else if (*value == '!')
377                 return SM_UPDATE_COMMAND;
378         else
379                 return SM_UPDATE_UNSPECIFIED;
380 }
381
382 int parse_submodule_update_strategy(const char *value,
383                 struct submodule_update_strategy *dst)
384 {
385         enum submodule_update_type type;
386
387         free((void*)dst->command);
388         dst->command = NULL;
389
390         type = parse_submodule_update_type(value);
391         if (type == SM_UPDATE_UNSPECIFIED)
392                 return -1;
393
394         dst->type = type;
395         if (type == SM_UPDATE_COMMAND)
396                 dst->command = xstrdup(value + 1);
397
398         return 0;
399 }
400
401 const char *submodule_strategy_to_string(const struct submodule_update_strategy *s)
402 {
403         struct strbuf sb = STRBUF_INIT;
404         switch (s->type) {
405         case SM_UPDATE_CHECKOUT:
406                 return "checkout";
407         case SM_UPDATE_MERGE:
408                 return "merge";
409         case SM_UPDATE_REBASE:
410                 return "rebase";
411         case SM_UPDATE_NONE:
412                 return "none";
413         case SM_UPDATE_UNSPECIFIED:
414                 return NULL;
415         case SM_UPDATE_COMMAND:
416                 strbuf_addf(&sb, "!%s", s->command);
417                 return strbuf_detach(&sb, NULL);
418         }
419         return NULL;
420 }
421
422 void handle_ignore_submodules_arg(struct diff_options *diffopt,
423                                   const char *arg)
424 {
425         diffopt->flags.ignore_submodules = 0;
426         diffopt->flags.ignore_untracked_in_submodules = 0;
427         diffopt->flags.ignore_dirty_submodules = 0;
428
429         if (!strcmp(arg, "all"))
430                 diffopt->flags.ignore_submodules = 1;
431         else if (!strcmp(arg, "untracked"))
432                 diffopt->flags.ignore_untracked_in_submodules = 1;
433         else if (!strcmp(arg, "dirty"))
434                 diffopt->flags.ignore_dirty_submodules = 1;
435         else if (strcmp(arg, "none"))
436                 die("bad --ignore-submodules argument: %s", arg);
437 }
438
439 static int prepare_submodule_summary(struct rev_info *rev, const char *path,
440                 struct commit *left, struct commit *right,
441                 struct commit_list *merge_bases)
442 {
443         struct commit_list *list;
444
445         repo_init_revisions(the_repository, rev, NULL);
446         setup_revisions(0, NULL, rev, NULL);
447         rev->left_right = 1;
448         rev->first_parent_only = 1;
449         left->object.flags |= SYMMETRIC_LEFT;
450         add_pending_object(rev, &left->object, path);
451         add_pending_object(rev, &right->object, path);
452         for (list = merge_bases; list; list = list->next) {
453                 list->item->object.flags |= UNINTERESTING;
454                 add_pending_object(rev, &list->item->object,
455                         oid_to_hex(&list->item->object.oid));
456         }
457         return prepare_revision_walk(rev);
458 }
459
460 static void print_submodule_summary(struct repository *r, struct rev_info *rev, struct diff_options *o)
461 {
462         static const char format[] = "  %m %s";
463         struct strbuf sb = STRBUF_INIT;
464         struct commit *commit;
465
466         while ((commit = get_revision(rev))) {
467                 struct pretty_print_context ctx = {0};
468                 ctx.date_mode = rev->date_mode;
469                 ctx.output_encoding = get_log_output_encoding();
470                 strbuf_setlen(&sb, 0);
471                 repo_format_commit_message(r, commit, format, &sb,
472                                       &ctx);
473                 strbuf_addch(&sb, '\n');
474                 if (commit->object.flags & SYMMETRIC_LEFT)
475                         diff_emit_submodule_del(o, sb.buf);
476                 else
477                         diff_emit_submodule_add(o, sb.buf);
478         }
479         strbuf_release(&sb);
480 }
481
482 static void prepare_submodule_repo_env_no_git_dir(struct argv_array *out)
483 {
484         const char * const *var;
485
486         for (var = local_repo_env; *var; var++) {
487                 if (strcmp(*var, CONFIG_DATA_ENVIRONMENT))
488                         argv_array_push(out, *var);
489         }
490 }
491
492 void prepare_submodule_repo_env(struct argv_array *out)
493 {
494         prepare_submodule_repo_env_no_git_dir(out);
495         argv_array_pushf(out, "%s=%s", GIT_DIR_ENVIRONMENT,
496                          DEFAULT_GIT_DIR_ENVIRONMENT);
497 }
498
499 /*
500  * Initialize a repository struct for a submodule based on the provided 'path'.
501  *
502  * Unlike repo_submodule_init, this tolerates submodules not present
503  * in .gitmodules. This function exists only to preserve historical behavior,
504  *
505  * Returns the repository struct on success,
506  * NULL when the submodule is not present.
507  */
508 static struct repository *open_submodule(const char *path)
509 {
510         struct strbuf sb = STRBUF_INIT;
511         struct repository *out = xmalloc(sizeof(*out));
512
513         if (submodule_to_gitdir(&sb, path) || repo_init(out, sb.buf, NULL)) {
514                 strbuf_release(&sb);
515                 free(out);
516                 return NULL;
517         }
518
519         /* Mark it as a submodule */
520         out->submodule_prefix = xstrdup(path);
521
522         strbuf_release(&sb);
523         return out;
524 }
525
526 /*
527  * Helper function to display the submodule header line prior to the full
528  * summary output.
529  *
530  * If it can locate the submodule git directory it will create a repository
531  * handle for the submodule and lookup both the left and right commits and
532  * put them into the left and right pointers.
533  */
534 static void show_submodule_header(struct diff_options *o,
535                 const char *path,
536                 struct object_id *one, struct object_id *two,
537                 unsigned dirty_submodule,
538                 struct repository *sub,
539                 struct commit **left, struct commit **right,
540                 struct commit_list **merge_bases)
541 {
542         const char *message = NULL;
543         struct strbuf sb = STRBUF_INIT;
544         int fast_forward = 0, fast_backward = 0;
545
546         if (dirty_submodule & DIRTY_SUBMODULE_UNTRACKED)
547                 diff_emit_submodule_untracked(o, path);
548
549         if (dirty_submodule & DIRTY_SUBMODULE_MODIFIED)
550                 diff_emit_submodule_modified(o, path);
551
552         if (is_null_oid(one))
553                 message = "(new submodule)";
554         else if (is_null_oid(two))
555                 message = "(submodule deleted)";
556
557         if (!sub) {
558                 if (!message)
559                         message = "(commits not present)";
560                 goto output_header;
561         }
562
563         /*
564          * Attempt to lookup the commit references, and determine if this is
565          * a fast forward or fast backwards update.
566          */
567         *left = lookup_commit_reference(sub, one);
568         *right = lookup_commit_reference(sub, two);
569
570         /*
571          * Warn about missing commits in the submodule project, but only if
572          * they aren't null.
573          */
574         if ((!is_null_oid(one) && !*left) ||
575              (!is_null_oid(two) && !*right))
576                 message = "(commits not present)";
577
578         *merge_bases = repo_get_merge_bases(sub, *left, *right);
579         if (*merge_bases) {
580                 if ((*merge_bases)->item == *left)
581                         fast_forward = 1;
582                 else if ((*merge_bases)->item == *right)
583                         fast_backward = 1;
584         }
585
586         if (oideq(one, two)) {
587                 strbuf_release(&sb);
588                 return;
589         }
590
591 output_header:
592         strbuf_addf(&sb, "Submodule %s ", path);
593         strbuf_add_unique_abbrev(&sb, one, DEFAULT_ABBREV);
594         strbuf_addstr(&sb, (fast_backward || fast_forward) ? ".." : "...");
595         strbuf_add_unique_abbrev(&sb, two, DEFAULT_ABBREV);
596         if (message)
597                 strbuf_addf(&sb, " %s\n", message);
598         else
599                 strbuf_addf(&sb, "%s:\n", fast_backward ? " (rewind)" : "");
600         diff_emit_submodule_header(o, sb.buf);
601
602         strbuf_release(&sb);
603 }
604
605 void show_submodule_summary(struct diff_options *o, const char *path,
606                 struct object_id *one, struct object_id *two,
607                 unsigned dirty_submodule)
608 {
609         struct rev_info rev;
610         struct commit *left = NULL, *right = NULL;
611         struct commit_list *merge_bases = NULL;
612         struct repository *sub;
613
614         sub = open_submodule(path);
615         show_submodule_header(o, path, one, two, dirty_submodule,
616                               sub, &left, &right, &merge_bases);
617
618         /*
619          * If we don't have both a left and a right pointer, there is no
620          * reason to try and display a summary. The header line should contain
621          * all the information the user needs.
622          */
623         if (!left || !right || !sub)
624                 goto out;
625
626         /* Treat revision walker failure the same as missing commits */
627         if (prepare_submodule_summary(&rev, path, left, right, merge_bases)) {
628                 diff_emit_submodule_error(o, "(revision walker failed)\n");
629                 goto out;
630         }
631
632         print_submodule_summary(sub, &rev, o);
633
634 out:
635         if (merge_bases)
636                 free_commit_list(merge_bases);
637         clear_commit_marks(left, ~0);
638         clear_commit_marks(right, ~0);
639         if (sub) {
640                 repo_clear(sub);
641                 free(sub);
642         }
643 }
644
645 void show_submodule_inline_diff(struct diff_options *o, const char *path,
646                 struct object_id *one, struct object_id *two,
647                 unsigned dirty_submodule)
648 {
649         const struct object_id *old_oid = the_hash_algo->empty_tree, *new_oid = the_hash_algo->empty_tree;
650         struct commit *left = NULL, *right = NULL;
651         struct commit_list *merge_bases = NULL;
652         struct child_process cp = CHILD_PROCESS_INIT;
653         struct strbuf sb = STRBUF_INIT;
654         struct repository *sub;
655
656         sub = open_submodule(path);
657         show_submodule_header(o, path, one, two, dirty_submodule,
658                               sub, &left, &right, &merge_bases);
659
660         /* We need a valid left and right commit to display a difference */
661         if (!(left || is_null_oid(one)) ||
662             !(right || is_null_oid(two)))
663                 goto done;
664
665         if (left)
666                 old_oid = one;
667         if (right)
668                 new_oid = two;
669
670         cp.git_cmd = 1;
671         cp.dir = path;
672         cp.out = -1;
673         cp.no_stdin = 1;
674
675         /* TODO: other options may need to be passed here. */
676         argv_array_pushl(&cp.args, "diff", "--submodule=diff", NULL);
677         argv_array_pushf(&cp.args, "--color=%s", want_color(o->use_color) ?
678                          "always" : "never");
679
680         if (o->flags.reverse_diff) {
681                 argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
682                                  o->b_prefix, path);
683                 argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
684                                  o->a_prefix, path);
685         } else {
686                 argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
687                                  o->a_prefix, path);
688                 argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
689                                  o->b_prefix, path);
690         }
691         argv_array_push(&cp.args, oid_to_hex(old_oid));
692         /*
693          * If the submodule has modified content, we will diff against the
694          * work tree, under the assumption that the user has asked for the
695          * diff format and wishes to actually see all differences even if they
696          * haven't yet been committed to the submodule yet.
697          */
698         if (!(dirty_submodule & DIRTY_SUBMODULE_MODIFIED))
699                 argv_array_push(&cp.args, oid_to_hex(new_oid));
700
701         prepare_submodule_repo_env(&cp.env_array);
702         if (start_command(&cp))
703                 diff_emit_submodule_error(o, "(diff failed)\n");
704
705         while (strbuf_getwholeline_fd(&sb, cp.out, '\n') != EOF)
706                 diff_emit_submodule_pipethrough(o, sb.buf, sb.len);
707
708         if (finish_command(&cp))
709                 diff_emit_submodule_error(o, "(diff failed)\n");
710
711 done:
712         strbuf_release(&sb);
713         if (merge_bases)
714                 free_commit_list(merge_bases);
715         if (left)
716                 clear_commit_marks(left, ~0);
717         if (right)
718                 clear_commit_marks(right, ~0);
719         if (sub) {
720                 repo_clear(sub);
721                 free(sub);
722         }
723 }
724
725 int should_update_submodules(void)
726 {
727         return config_update_recurse_submodules == RECURSE_SUBMODULES_ON;
728 }
729
730 const struct submodule *submodule_from_ce(const struct cache_entry *ce)
731 {
732         if (!S_ISGITLINK(ce->ce_mode))
733                 return NULL;
734
735         if (!should_update_submodules())
736                 return NULL;
737
738         return submodule_from_path(the_repository, &null_oid, ce->name);
739 }
740
741 static struct oid_array *submodule_commits(struct string_list *submodules,
742                                            const char *name)
743 {
744         struct string_list_item *item;
745
746         item = string_list_insert(submodules, name);
747         if (item->util)
748                 return (struct oid_array *) item->util;
749
750         /* NEEDSWORK: should we have oid_array_init()? */
751         item->util = xcalloc(1, sizeof(struct oid_array));
752         return (struct oid_array *) item->util;
753 }
754
755 struct collect_changed_submodules_cb_data {
756         struct repository *repo;
757         struct string_list *changed;
758         const struct object_id *commit_oid;
759 };
760
761 /*
762  * this would normally be two functions: default_name_from_path() and
763  * path_from_default_name(). Since the default name is the same as
764  * the submodule path we can get away with just one function which only
765  * checks whether there is a submodule in the working directory at that
766  * location.
767  */
768 static const char *default_name_or_path(const char *path_or_name)
769 {
770         int error_code;
771
772         if (!is_submodule_populated_gently(path_or_name, &error_code))
773                 return NULL;
774
775         return path_or_name;
776 }
777
778 static void collect_changed_submodules_cb(struct diff_queue_struct *q,
779                                           struct diff_options *options,
780                                           void *data)
781 {
782         struct collect_changed_submodules_cb_data *me = data;
783         struct string_list *changed = me->changed;
784         const struct object_id *commit_oid = me->commit_oid;
785         int i;
786
787         for (i = 0; i < q->nr; i++) {
788                 struct diff_filepair *p = q->queue[i];
789                 struct oid_array *commits;
790                 const struct submodule *submodule;
791                 const char *name;
792
793                 if (!S_ISGITLINK(p->two->mode))
794                         continue;
795
796                 submodule = submodule_from_path(me->repo,
797                                                 commit_oid, p->two->path);
798                 if (submodule)
799                         name = submodule->name;
800                 else {
801                         name = default_name_or_path(p->two->path);
802                         /* make sure name does not collide with existing one */
803                         if (name)
804                                 submodule = submodule_from_name(me->repo,
805                                                                 commit_oid, name);
806                         if (submodule) {
807                                 warning("Submodule in commit %s at path: "
808                                         "'%s' collides with a submodule named "
809                                         "the same. Skipping it.",
810                                         oid_to_hex(commit_oid), p->two->path);
811                                 name = NULL;
812                         }
813                 }
814
815                 if (!name)
816                         continue;
817
818                 commits = submodule_commits(changed, name);
819                 oid_array_append(commits, &p->two->oid);
820         }
821 }
822
823 /*
824  * Collect the paths of submodules in 'changed' which have changed based on
825  * the revisions as specified in 'argv'.  Each entry in 'changed' will also
826  * have a corresponding 'struct oid_array' (in the 'util' field) which lists
827  * what the submodule pointers were updated to during the change.
828  */
829 static void collect_changed_submodules(struct repository *r,
830                                        struct string_list *changed,
831                                        struct argv_array *argv)
832 {
833         struct rev_info rev;
834         const struct commit *commit;
835
836         repo_init_revisions(r, &rev, NULL);
837         setup_revisions(argv->argc, argv->argv, &rev, NULL);
838         if (prepare_revision_walk(&rev))
839                 die("revision walk setup failed");
840
841         while ((commit = get_revision(&rev))) {
842                 struct rev_info diff_rev;
843                 struct collect_changed_submodules_cb_data data;
844                 data.repo = r;
845                 data.changed = changed;
846                 data.commit_oid = &commit->object.oid;
847
848                 repo_init_revisions(r, &diff_rev, NULL);
849                 diff_rev.diffopt.output_format |= DIFF_FORMAT_CALLBACK;
850                 diff_rev.diffopt.format_callback = collect_changed_submodules_cb;
851                 diff_rev.diffopt.format_callback_data = &data;
852                 diff_tree_combined_merge(commit, 1, &diff_rev);
853         }
854
855         reset_revision_walk();
856 }
857
858 static void free_submodules_oids(struct string_list *submodules)
859 {
860         struct string_list_item *item;
861         for_each_string_list_item(item, submodules)
862                 oid_array_clear((struct oid_array *) item->util);
863         string_list_clear(submodules, 1);
864 }
865
866 static int has_remote(const char *refname, const struct object_id *oid,
867                       int flags, void *cb_data)
868 {
869         return 1;
870 }
871
872 static int append_oid_to_argv(const struct object_id *oid, void *data)
873 {
874         struct argv_array *argv = data;
875         argv_array_push(argv, oid_to_hex(oid));
876         return 0;
877 }
878
879 struct has_commit_data {
880         struct repository *repo;
881         int result;
882         const char *path;
883 };
884
885 static int check_has_commit(const struct object_id *oid, void *data)
886 {
887         struct has_commit_data *cb = data;
888
889         enum object_type type = oid_object_info(cb->repo, oid, NULL);
890
891         switch (type) {
892         case OBJ_COMMIT:
893                 return 0;
894         case OBJ_BAD:
895                 /*
896                  * Object is missing or invalid. If invalid, an error message
897                  * has already been printed.
898                  */
899                 cb->result = 0;
900                 return 0;
901         default:
902                 die(_("submodule entry '%s' (%s) is a %s, not a commit"),
903                     cb->path, oid_to_hex(oid), type_name(type));
904         }
905 }
906
907 static int submodule_has_commits(struct repository *r,
908                                  const char *path,
909                                  struct oid_array *commits)
910 {
911         struct has_commit_data has_commit = { r, 1, path };
912
913         /*
914          * Perform a cheap, but incorrect check for the existence of 'commits'.
915          * This is done by adding the submodule's object store to the in-core
916          * object store, and then querying for each commit's existence.  If we
917          * do not have the commit object anywhere, there is no chance we have
918          * it in the object store of the correct submodule and have it
919          * reachable from a ref, so we can fail early without spawning rev-list
920          * which is expensive.
921          */
922         if (add_submodule_odb(path))
923                 return 0;
924
925         oid_array_for_each_unique(commits, check_has_commit, &has_commit);
926
927         if (has_commit.result) {
928                 /*
929                  * Even if the submodule is checked out and the commit is
930                  * present, make sure it exists in the submodule's object store
931                  * and that it is reachable from a ref.
932                  */
933                 struct child_process cp = CHILD_PROCESS_INIT;
934                 struct strbuf out = STRBUF_INIT;
935
936                 argv_array_pushl(&cp.args, "rev-list", "-n", "1", NULL);
937                 oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
938                 argv_array_pushl(&cp.args, "--not", "--all", NULL);
939
940                 prepare_submodule_repo_env(&cp.env_array);
941                 cp.git_cmd = 1;
942                 cp.no_stdin = 1;
943                 cp.dir = path;
944
945                 if (capture_command(&cp, &out, GIT_MAX_HEXSZ + 1) || out.len)
946                         has_commit.result = 0;
947
948                 strbuf_release(&out);
949         }
950
951         return has_commit.result;
952 }
953
954 static int submodule_needs_pushing(struct repository *r,
955                                    const char *path,
956                                    struct oid_array *commits)
957 {
958         if (!submodule_has_commits(r, path, commits))
959                 /*
960                  * NOTE: We do consider it safe to return "no" here. The
961                  * correct answer would be "We do not know" instead of
962                  * "No push needed", but it is quite hard to change
963                  * the submodule pointer without having the submodule
964                  * around. If a user did however change the submodules
965                  * without having the submodule around, this indicates
966                  * an expert who knows what they are doing or a
967                  * maintainer integrating work from other people. In
968                  * both cases it should be safe to skip this check.
969                  */
970                 return 0;
971
972         if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
973                 struct child_process cp = CHILD_PROCESS_INIT;
974                 struct strbuf buf = STRBUF_INIT;
975                 int needs_pushing = 0;
976
977                 argv_array_push(&cp.args, "rev-list");
978                 oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
979                 argv_array_pushl(&cp.args, "--not", "--remotes", "-n", "1" , NULL);
980
981                 prepare_submodule_repo_env(&cp.env_array);
982                 cp.git_cmd = 1;
983                 cp.no_stdin = 1;
984                 cp.out = -1;
985                 cp.dir = path;
986                 if (start_command(&cp))
987                         die("Could not run 'git rev-list <commits> --not --remotes -n 1' command in submodule %s",
988                                         path);
989                 if (strbuf_read(&buf, cp.out, 41))
990                         needs_pushing = 1;
991                 finish_command(&cp);
992                 close(cp.out);
993                 strbuf_release(&buf);
994                 return needs_pushing;
995         }
996
997         return 0;
998 }
999
1000 int find_unpushed_submodules(struct repository *r,
1001                              struct oid_array *commits,
1002                              const char *remotes_name,
1003                              struct string_list *needs_pushing)
1004 {
1005         struct string_list submodules = STRING_LIST_INIT_DUP;
1006         struct string_list_item *name;
1007         struct argv_array argv = ARGV_ARRAY_INIT;
1008
1009         /* argv.argv[0] will be ignored by setup_revisions */
1010         argv_array_push(&argv, "find_unpushed_submodules");
1011         oid_array_for_each_unique(commits, append_oid_to_argv, &argv);
1012         argv_array_push(&argv, "--not");
1013         argv_array_pushf(&argv, "--remotes=%s", remotes_name);
1014
1015         collect_changed_submodules(r, &submodules, &argv);
1016
1017         for_each_string_list_item(name, &submodules) {
1018                 struct oid_array *commits = name->util;
1019                 const struct submodule *submodule;
1020                 const char *path = NULL;
1021
1022                 submodule = submodule_from_name(r, &null_oid, name->string);
1023                 if (submodule)
1024                         path = submodule->path;
1025                 else
1026                         path = default_name_or_path(name->string);
1027
1028                 if (!path)
1029                         continue;
1030
1031                 if (submodule_needs_pushing(r, path, commits))
1032                         string_list_insert(needs_pushing, path);
1033         }
1034
1035         free_submodules_oids(&submodules);
1036         argv_array_clear(&argv);
1037
1038         return needs_pushing->nr;
1039 }
1040
1041 static int push_submodule(const char *path,
1042                           const struct remote *remote,
1043                           const struct refspec *rs,
1044                           const struct string_list *push_options,
1045                           int dry_run)
1046 {
1047         if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
1048                 struct child_process cp = CHILD_PROCESS_INIT;
1049                 argv_array_push(&cp.args, "push");
1050                 if (dry_run)
1051                         argv_array_push(&cp.args, "--dry-run");
1052
1053                 if (push_options && push_options->nr) {
1054                         const struct string_list_item *item;
1055                         for_each_string_list_item(item, push_options)
1056                                 argv_array_pushf(&cp.args, "--push-option=%s",
1057                                                  item->string);
1058                 }
1059
1060                 if (remote->origin != REMOTE_UNCONFIGURED) {
1061                         int i;
1062                         argv_array_push(&cp.args, remote->name);
1063                         for (i = 0; i < rs->raw_nr; i++)
1064                                 argv_array_push(&cp.args, rs->raw[i]);
1065                 }
1066
1067                 prepare_submodule_repo_env(&cp.env_array);
1068                 cp.git_cmd = 1;
1069                 cp.no_stdin = 1;
1070                 cp.dir = path;
1071                 if (run_command(&cp))
1072                         return 0;
1073                 close(cp.out);
1074         }
1075
1076         return 1;
1077 }
1078
1079 /*
1080  * Perform a check in the submodule to see if the remote and refspec work.
1081  * Die if the submodule can't be pushed.
1082  */
1083 static void submodule_push_check(const char *path, const char *head,
1084                                  const struct remote *remote,
1085                                  const struct refspec *rs)
1086 {
1087         struct child_process cp = CHILD_PROCESS_INIT;
1088         int i;
1089
1090         argv_array_push(&cp.args, "submodule--helper");
1091         argv_array_push(&cp.args, "push-check");
1092         argv_array_push(&cp.args, head);
1093         argv_array_push(&cp.args, remote->name);
1094
1095         for (i = 0; i < rs->raw_nr; i++)
1096                 argv_array_push(&cp.args, rs->raw[i]);
1097
1098         prepare_submodule_repo_env(&cp.env_array);
1099         cp.git_cmd = 1;
1100         cp.no_stdin = 1;
1101         cp.no_stdout = 1;
1102         cp.dir = path;
1103
1104         /*
1105          * Simply indicate if 'submodule--helper push-check' failed.
1106          * More detailed error information will be provided by the
1107          * child process.
1108          */
1109         if (run_command(&cp))
1110                 die("process for submodule '%s' failed", path);
1111 }
1112
1113 int push_unpushed_submodules(struct repository *r,
1114                              struct oid_array *commits,
1115                              const struct remote *remote,
1116                              const struct refspec *rs,
1117                              const struct string_list *push_options,
1118                              int dry_run)
1119 {
1120         int i, ret = 1;
1121         struct string_list needs_pushing = STRING_LIST_INIT_DUP;
1122
1123         if (!find_unpushed_submodules(r, commits,
1124                                       remote->name, &needs_pushing))
1125                 return 1;
1126
1127         /*
1128          * Verify that the remote and refspec can be propagated to all
1129          * submodules.  This check can be skipped if the remote and refspec
1130          * won't be propagated due to the remote being unconfigured (e.g. a URL
1131          * instead of a remote name).
1132          */
1133         if (remote->origin != REMOTE_UNCONFIGURED) {
1134                 char *head;
1135                 struct object_id head_oid;
1136
1137                 head = resolve_refdup("HEAD", 0, &head_oid, NULL);
1138                 if (!head)
1139                         die(_("Failed to resolve HEAD as a valid ref."));
1140
1141                 for (i = 0; i < needs_pushing.nr; i++)
1142                         submodule_push_check(needs_pushing.items[i].string,
1143                                              head, remote, rs);
1144                 free(head);
1145         }
1146
1147         /* Actually push the submodules */
1148         for (i = 0; i < needs_pushing.nr; i++) {
1149                 const char *path = needs_pushing.items[i].string;
1150                 fprintf(stderr, "Pushing submodule '%s'\n", path);
1151                 if (!push_submodule(path, remote, rs,
1152                                     push_options, dry_run)) {
1153                         fprintf(stderr, "Unable to push submodule '%s'\n", path);
1154                         ret = 0;
1155                 }
1156         }
1157
1158         string_list_clear(&needs_pushing, 0);
1159
1160         return ret;
1161 }
1162
1163 static int append_oid_to_array(const char *ref, const struct object_id *oid,
1164                                int flags, void *data)
1165 {
1166         struct oid_array *array = data;
1167         oid_array_append(array, oid);
1168         return 0;
1169 }
1170
1171 void check_for_new_submodule_commits(struct object_id *oid)
1172 {
1173         if (!initialized_fetch_ref_tips) {
1174                 for_each_ref(append_oid_to_array, &ref_tips_before_fetch);
1175                 initialized_fetch_ref_tips = 1;
1176         }
1177
1178         oid_array_append(&ref_tips_after_fetch, oid);
1179 }
1180
1181 static void calculate_changed_submodule_paths(struct repository *r)
1182 {
1183         struct argv_array argv = ARGV_ARRAY_INIT;
1184         struct string_list changed_submodules = STRING_LIST_INIT_DUP;
1185         const struct string_list_item *name;
1186
1187         /* No need to check if there are no submodules configured */
1188         if (!submodule_from_path(r, NULL, NULL))
1189                 return;
1190
1191         argv_array_push(&argv, "--"); /* argv[0] program name */
1192         oid_array_for_each_unique(&ref_tips_after_fetch,
1193                                    append_oid_to_argv, &argv);
1194         argv_array_push(&argv, "--not");
1195         oid_array_for_each_unique(&ref_tips_before_fetch,
1196                                    append_oid_to_argv, &argv);
1197
1198         /*
1199          * Collect all submodules (whether checked out or not) for which new
1200          * commits have been recorded upstream in "changed_submodule_names".
1201          */
1202         collect_changed_submodules(r, &changed_submodules, &argv);
1203
1204         for_each_string_list_item(name, &changed_submodules) {
1205                 struct oid_array *commits = name->util;
1206                 const struct submodule *submodule;
1207                 const char *path = NULL;
1208
1209                 submodule = submodule_from_name(r, &null_oid, name->string);
1210                 if (submodule)
1211                         path = submodule->path;
1212                 else
1213                         path = default_name_or_path(name->string);
1214
1215                 if (!path)
1216                         continue;
1217
1218                 if (!submodule_has_commits(r, path, commits))
1219                         string_list_append(&changed_submodule_names, name->string);
1220         }
1221
1222         free_submodules_oids(&changed_submodules);
1223         argv_array_clear(&argv);
1224         oid_array_clear(&ref_tips_before_fetch);
1225         oid_array_clear(&ref_tips_after_fetch);
1226         initialized_fetch_ref_tips = 0;
1227 }
1228
1229 int submodule_touches_in_range(struct repository *r,
1230                                struct object_id *excl_oid,
1231                                struct object_id *incl_oid)
1232 {
1233         struct string_list subs = STRING_LIST_INIT_DUP;
1234         struct argv_array args = ARGV_ARRAY_INIT;
1235         int ret;
1236
1237         /* No need to check if there are no submodules configured */
1238         if (!submodule_from_path(r, NULL, NULL))
1239                 return 0;
1240
1241         argv_array_push(&args, "--"); /* args[0] program name */
1242         argv_array_push(&args, oid_to_hex(incl_oid));
1243         if (!is_null_oid(excl_oid)) {
1244                 argv_array_push(&args, "--not");
1245                 argv_array_push(&args, oid_to_hex(excl_oid));
1246         }
1247
1248         collect_changed_submodules(r, &subs, &args);
1249         ret = subs.nr;
1250
1251         argv_array_clear(&args);
1252
1253         free_submodules_oids(&subs);
1254         return ret;
1255 }
1256
1257 struct submodule_parallel_fetch {
1258         int count;
1259         struct argv_array args;
1260         struct repository *r;
1261         const char *prefix;
1262         int command_line_option;
1263         int default_option;
1264         int quiet;
1265         int result;
1266 };
1267 #define SPF_INIT {0, ARGV_ARRAY_INIT, NULL, NULL, 0, 0, 0, 0}
1268
1269 static int get_fetch_recurse_config(const struct submodule *submodule,
1270                                     struct submodule_parallel_fetch *spf)
1271 {
1272         if (spf->command_line_option != RECURSE_SUBMODULES_DEFAULT)
1273                 return spf->command_line_option;
1274
1275         if (submodule) {
1276                 char *key;
1277                 const char *value;
1278
1279                 int fetch_recurse = submodule->fetch_recurse;
1280                 key = xstrfmt("submodule.%s.fetchRecurseSubmodules", submodule->name);
1281                 if (!repo_config_get_string_const(spf->r, key, &value)) {
1282                         fetch_recurse = parse_fetch_recurse_submodules_arg(key, value);
1283                 }
1284                 free(key);
1285
1286                 if (fetch_recurse != RECURSE_SUBMODULES_NONE)
1287                         /* local config overrules everything except commandline */
1288                         return fetch_recurse;
1289         }
1290
1291         return spf->default_option;
1292 }
1293
1294 static int get_next_submodule(struct child_process *cp,
1295                               struct strbuf *err, void *data, void **task_cb)
1296 {
1297         int ret = 0;
1298         struct submodule_parallel_fetch *spf = data;
1299
1300         for (; spf->count < spf->r->index->cache_nr; spf->count++) {
1301                 struct strbuf submodule_path = STRBUF_INIT;
1302                 struct strbuf submodule_git_dir = STRBUF_INIT;
1303                 struct strbuf submodule_prefix = STRBUF_INIT;
1304                 const struct cache_entry *ce = spf->r->index->cache[spf->count];
1305                 const char *git_dir, *default_argv;
1306                 const struct submodule *submodule;
1307                 struct submodule default_submodule = SUBMODULE_INIT;
1308
1309                 if (!S_ISGITLINK(ce->ce_mode))
1310                         continue;
1311
1312                 submodule = submodule_from_path(spf->r, &null_oid, ce->name);
1313                 if (!submodule) {
1314                         const char *name = default_name_or_path(ce->name);
1315                         if (name) {
1316                                 default_submodule.path = default_submodule.name = name;
1317                                 submodule = &default_submodule;
1318                         }
1319                 }
1320
1321                 switch (get_fetch_recurse_config(submodule, spf))
1322                 {
1323                 default:
1324                 case RECURSE_SUBMODULES_DEFAULT:
1325                 case RECURSE_SUBMODULES_ON_DEMAND:
1326                         if (!submodule || !unsorted_string_list_lookup(&changed_submodule_names,
1327                                                          submodule->name))
1328                                 continue;
1329                         default_argv = "on-demand";
1330                         break;
1331                 case RECURSE_SUBMODULES_ON:
1332                         default_argv = "yes";
1333                         break;
1334                 case RECURSE_SUBMODULES_OFF:
1335                         continue;
1336                 }
1337
1338                 strbuf_repo_worktree_path(&submodule_path, spf->r, "%s", ce->name);
1339                 strbuf_addf(&submodule_git_dir, "%s/.git", submodule_path.buf);
1340                 strbuf_addf(&submodule_prefix, "%s%s/", spf->prefix, ce->name);
1341                 git_dir = read_gitfile(submodule_git_dir.buf);
1342                 if (!git_dir)
1343                         git_dir = submodule_git_dir.buf;
1344                 if (is_directory(git_dir)) {
1345                         child_process_init(cp);
1346                         cp->dir = strbuf_detach(&submodule_path, NULL);
1347                         prepare_submodule_repo_env(&cp->env_array);
1348                         cp->git_cmd = 1;
1349                         if (!spf->quiet)
1350                                 strbuf_addf(err, "Fetching submodule %s%s\n",
1351                                             spf->prefix, ce->name);
1352                         argv_array_init(&cp->args);
1353                         argv_array_pushv(&cp->args, spf->args.argv);
1354                         argv_array_push(&cp->args, default_argv);
1355                         argv_array_push(&cp->args, "--submodule-prefix");
1356                         argv_array_push(&cp->args, submodule_prefix.buf);
1357                         ret = 1;
1358                 }
1359                 strbuf_release(&submodule_path);
1360                 strbuf_release(&submodule_git_dir);
1361                 strbuf_release(&submodule_prefix);
1362                 if (ret) {
1363                         spf->count++;
1364                         return 1;
1365                 }
1366         }
1367         return 0;
1368 }
1369
1370 static int fetch_start_failure(struct strbuf *err,
1371                                void *cb, void *task_cb)
1372 {
1373         struct submodule_parallel_fetch *spf = cb;
1374
1375         spf->result = 1;
1376
1377         return 0;
1378 }
1379
1380 static int fetch_finish(int retvalue, struct strbuf *err,
1381                         void *cb, void *task_cb)
1382 {
1383         struct submodule_parallel_fetch *spf = cb;
1384
1385         if (retvalue)
1386                 spf->result = 1;
1387
1388         return 0;
1389 }
1390
1391 int fetch_populated_submodules(struct repository *r,
1392                                const struct argv_array *options,
1393                                const char *prefix, int command_line_option,
1394                                int default_option,
1395                                int quiet, int max_parallel_jobs)
1396 {
1397         int i;
1398         struct submodule_parallel_fetch spf = SPF_INIT;
1399
1400         spf.r = r;
1401         spf.command_line_option = command_line_option;
1402         spf.default_option = default_option;
1403         spf.quiet = quiet;
1404         spf.prefix = prefix;
1405
1406         if (!r->worktree)
1407                 goto out;
1408
1409         if (repo_read_index(r) < 0)
1410                 die("index file corrupt");
1411
1412         argv_array_push(&spf.args, "fetch");
1413         for (i = 0; i < options->argc; i++)
1414                 argv_array_push(&spf.args, options->argv[i]);
1415         argv_array_push(&spf.args, "--recurse-submodules-default");
1416         /* default value, "--submodule-prefix" and its value are added later */
1417
1418         calculate_changed_submodule_paths(r);
1419         run_processes_parallel(max_parallel_jobs,
1420                                get_next_submodule,
1421                                fetch_start_failure,
1422                                fetch_finish,
1423                                &spf);
1424
1425         argv_array_clear(&spf.args);
1426 out:
1427         string_list_clear(&changed_submodule_names, 1);
1428         return spf.result;
1429 }
1430
1431 unsigned is_submodule_modified(const char *path, int ignore_untracked)
1432 {
1433         struct child_process cp = CHILD_PROCESS_INIT;
1434         struct strbuf buf = STRBUF_INIT;
1435         FILE *fp;
1436         unsigned dirty_submodule = 0;
1437         const char *git_dir;
1438         int ignore_cp_exit_code = 0;
1439
1440         strbuf_addf(&buf, "%s/.git", path);
1441         git_dir = read_gitfile(buf.buf);
1442         if (!git_dir)
1443                 git_dir = buf.buf;
1444         if (!is_git_directory(git_dir)) {
1445                 if (is_directory(git_dir))
1446                         die(_("'%s' not recognized as a git repository"), git_dir);
1447                 strbuf_release(&buf);
1448                 /* The submodule is not checked out, so it is not modified */
1449                 return 0;
1450         }
1451         strbuf_reset(&buf);
1452
1453         argv_array_pushl(&cp.args, "status", "--porcelain=2", NULL);
1454         if (ignore_untracked)
1455                 argv_array_push(&cp.args, "-uno");
1456
1457         prepare_submodule_repo_env(&cp.env_array);
1458         cp.git_cmd = 1;
1459         cp.no_stdin = 1;
1460         cp.out = -1;
1461         cp.dir = path;
1462         if (start_command(&cp))
1463                 die("Could not run 'git status --porcelain=2' in submodule %s", path);
1464
1465         fp = xfdopen(cp.out, "r");
1466         while (strbuf_getwholeline(&buf, fp, '\n') != EOF) {
1467                 /* regular untracked files */
1468                 if (buf.buf[0] == '?')
1469                         dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
1470
1471                 if (buf.buf[0] == 'u' ||
1472                     buf.buf[0] == '1' ||
1473                     buf.buf[0] == '2') {
1474                         /* T = line type, XY = status, SSSS = submodule state */
1475                         if (buf.len < strlen("T XY SSSS"))
1476                                 BUG("invalid status --porcelain=2 line %s",
1477                                     buf.buf);
1478
1479                         if (buf.buf[5] == 'S' && buf.buf[8] == 'U')
1480                                 /* nested untracked file */
1481                                 dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
1482
1483                         if (buf.buf[0] == 'u' ||
1484                             buf.buf[0] == '2' ||
1485                             memcmp(buf.buf + 5, "S..U", 4))
1486                                 /* other change */
1487                                 dirty_submodule |= DIRTY_SUBMODULE_MODIFIED;
1488                 }
1489
1490                 if ((dirty_submodule & DIRTY_SUBMODULE_MODIFIED) &&
1491                     ((dirty_submodule & DIRTY_SUBMODULE_UNTRACKED) ||
1492                      ignore_untracked)) {
1493                         /*
1494                          * We're not interested in any further information from
1495                          * the child any more, neither output nor its exit code.
1496                          */
1497                         ignore_cp_exit_code = 1;
1498                         break;
1499                 }
1500         }
1501         fclose(fp);
1502
1503         if (finish_command(&cp) && !ignore_cp_exit_code)
1504                 die("'git status --porcelain=2' failed in submodule %s", path);
1505
1506         strbuf_release(&buf);
1507         return dirty_submodule;
1508 }
1509
1510 int submodule_uses_gitfile(const char *path)
1511 {
1512         struct child_process cp = CHILD_PROCESS_INIT;
1513         const char *argv[] = {
1514                 "submodule",
1515                 "foreach",
1516                 "--quiet",
1517                 "--recursive",
1518                 "test -f .git",
1519                 NULL,
1520         };
1521         struct strbuf buf = STRBUF_INIT;
1522         const char *git_dir;
1523
1524         strbuf_addf(&buf, "%s/.git", path);
1525         git_dir = read_gitfile(buf.buf);
1526         if (!git_dir) {
1527                 strbuf_release(&buf);
1528                 return 0;
1529         }
1530         strbuf_release(&buf);
1531
1532         /* Now test that all nested submodules use a gitfile too */
1533         cp.argv = argv;
1534         prepare_submodule_repo_env(&cp.env_array);
1535         cp.git_cmd = 1;
1536         cp.no_stdin = 1;
1537         cp.no_stderr = 1;
1538         cp.no_stdout = 1;
1539         cp.dir = path;
1540         if (run_command(&cp))
1541                 return 0;
1542
1543         return 1;
1544 }
1545
1546 /*
1547  * Check if it is a bad idea to remove a submodule, i.e. if we'd lose data
1548  * when doing so.
1549  *
1550  * Return 1 if we'd lose data, return 0 if the removal is fine,
1551  * and negative values for errors.
1552  */
1553 int bad_to_remove_submodule(const char *path, unsigned flags)
1554 {
1555         ssize_t len;
1556         struct child_process cp = CHILD_PROCESS_INIT;
1557         struct strbuf buf = STRBUF_INIT;
1558         int ret = 0;
1559
1560         if (!file_exists(path) || is_empty_dir(path))
1561                 return 0;
1562
1563         if (!submodule_uses_gitfile(path))
1564                 return 1;
1565
1566         argv_array_pushl(&cp.args, "status", "--porcelain",
1567                                    "--ignore-submodules=none", NULL);
1568
1569         if (flags & SUBMODULE_REMOVAL_IGNORE_UNTRACKED)
1570                 argv_array_push(&cp.args, "-uno");
1571         else
1572                 argv_array_push(&cp.args, "-uall");
1573
1574         if (!(flags & SUBMODULE_REMOVAL_IGNORE_IGNORED_UNTRACKED))
1575                 argv_array_push(&cp.args, "--ignored");
1576
1577         prepare_submodule_repo_env(&cp.env_array);
1578         cp.git_cmd = 1;
1579         cp.no_stdin = 1;
1580         cp.out = -1;
1581         cp.dir = path;
1582         if (start_command(&cp)) {
1583                 if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
1584                         die(_("could not start 'git status' in submodule '%s'"),
1585                                 path);
1586                 ret = -1;
1587                 goto out;
1588         }
1589
1590         len = strbuf_read(&buf, cp.out, 1024);
1591         if (len > 2)
1592                 ret = 1;
1593         close(cp.out);
1594
1595         if (finish_command(&cp)) {
1596                 if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
1597                         die(_("could not run 'git status' in submodule '%s'"),
1598                                 path);
1599                 ret = -1;
1600         }
1601 out:
1602         strbuf_release(&buf);
1603         return ret;
1604 }
1605
1606 static const char *get_super_prefix_or_empty(void)
1607 {
1608         const char *s = get_super_prefix();
1609         if (!s)
1610                 s = "";
1611         return s;
1612 }
1613
1614 static int submodule_has_dirty_index(const struct submodule *sub)
1615 {
1616         struct child_process cp = CHILD_PROCESS_INIT;
1617
1618         prepare_submodule_repo_env(&cp.env_array);
1619
1620         cp.git_cmd = 1;
1621         argv_array_pushl(&cp.args, "diff-index", "--quiet",
1622                                    "--cached", "HEAD", NULL);
1623         cp.no_stdin = 1;
1624         cp.no_stdout = 1;
1625         cp.dir = sub->path;
1626         if (start_command(&cp))
1627                 die("could not recurse into submodule '%s'", sub->path);
1628
1629         return finish_command(&cp);
1630 }
1631
1632 static void submodule_reset_index(const char *path)
1633 {
1634         struct child_process cp = CHILD_PROCESS_INIT;
1635         prepare_submodule_repo_env(&cp.env_array);
1636
1637         cp.git_cmd = 1;
1638         cp.no_stdin = 1;
1639         cp.dir = path;
1640
1641         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
1642                                    get_super_prefix_or_empty(), path);
1643         argv_array_pushl(&cp.args, "read-tree", "-u", "--reset", NULL);
1644
1645         argv_array_push(&cp.args, empty_tree_oid_hex());
1646
1647         if (run_command(&cp))
1648                 die("could not reset submodule index");
1649 }
1650
1651 /**
1652  * Moves a submodule at a given path from a given head to another new head.
1653  * For edge cases (a submodule coming into existence or removing a submodule)
1654  * pass NULL for old or new respectively.
1655  */
1656 int submodule_move_head(const char *path,
1657                          const char *old_head,
1658                          const char *new_head,
1659                          unsigned flags)
1660 {
1661         int ret = 0;
1662         struct child_process cp = CHILD_PROCESS_INIT;
1663         const struct submodule *sub;
1664         int *error_code_ptr, error_code;
1665
1666         if (!is_submodule_active(the_repository, path))
1667                 return 0;
1668
1669         if (flags & SUBMODULE_MOVE_HEAD_FORCE)
1670                 /*
1671                  * Pass non NULL pointer to is_submodule_populated_gently
1672                  * to prevent die()-ing. We'll use connect_work_tree_and_git_dir
1673                  * to fixup the submodule in the force case later.
1674                  */
1675                 error_code_ptr = &error_code;
1676         else
1677                 error_code_ptr = NULL;
1678
1679         if (old_head && !is_submodule_populated_gently(path, error_code_ptr))
1680                 return 0;
1681
1682         sub = submodule_from_path(the_repository, &null_oid, path);
1683
1684         if (!sub)
1685                 BUG("could not get submodule information for '%s'", path);
1686
1687         if (old_head && !(flags & SUBMODULE_MOVE_HEAD_FORCE)) {
1688                 /* Check if the submodule has a dirty index. */
1689                 if (submodule_has_dirty_index(sub))
1690                         return error(_("submodule '%s' has dirty index"), path);
1691         }
1692
1693         if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
1694                 if (old_head) {
1695                         if (!submodule_uses_gitfile(path))
1696                                 absorb_git_dir_into_superproject("", path,
1697                                         ABSORB_GITDIR_RECURSE_SUBMODULES);
1698                 } else {
1699                         char *gitdir = xstrfmt("%s/modules/%s",
1700                                     get_git_common_dir(), sub->name);
1701                         connect_work_tree_and_git_dir(path, gitdir, 0);
1702                         free(gitdir);
1703
1704                         /* make sure the index is clean as well */
1705                         submodule_reset_index(path);
1706                 }
1707
1708                 if (old_head && (flags & SUBMODULE_MOVE_HEAD_FORCE)) {
1709                         char *gitdir = xstrfmt("%s/modules/%s",
1710                                     get_git_common_dir(), sub->name);
1711                         connect_work_tree_and_git_dir(path, gitdir, 1);
1712                         free(gitdir);
1713                 }
1714         }
1715
1716         prepare_submodule_repo_env(&cp.env_array);
1717
1718         cp.git_cmd = 1;
1719         cp.no_stdin = 1;
1720         cp.dir = path;
1721
1722         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
1723                         get_super_prefix_or_empty(), path);
1724         argv_array_pushl(&cp.args, "read-tree", "--recurse-submodules", NULL);
1725
1726         if (flags & SUBMODULE_MOVE_HEAD_DRY_RUN)
1727                 argv_array_push(&cp.args, "-n");
1728         else
1729                 argv_array_push(&cp.args, "-u");
1730
1731         if (flags & SUBMODULE_MOVE_HEAD_FORCE)
1732                 argv_array_push(&cp.args, "--reset");
1733         else
1734                 argv_array_push(&cp.args, "-m");
1735
1736         if (!(flags & SUBMODULE_MOVE_HEAD_FORCE))
1737                 argv_array_push(&cp.args, old_head ? old_head : empty_tree_oid_hex());
1738
1739         argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
1740
1741         if (run_command(&cp)) {
1742                 ret = error(_("Submodule '%s' could not be updated."), path);
1743                 goto out;
1744         }
1745
1746         if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
1747                 if (new_head) {
1748                         child_process_init(&cp);
1749                         /* also set the HEAD accordingly */
1750                         cp.git_cmd = 1;
1751                         cp.no_stdin = 1;
1752                         cp.dir = path;
1753
1754                         prepare_submodule_repo_env(&cp.env_array);
1755                         argv_array_pushl(&cp.args, "update-ref", "HEAD",
1756                                          "--no-deref", new_head, NULL);
1757
1758                         if (run_command(&cp)) {
1759                                 ret = -1;
1760                                 goto out;
1761                         }
1762                 } else {
1763                         struct strbuf sb = STRBUF_INIT;
1764
1765                         strbuf_addf(&sb, "%s/.git", path);
1766                         unlink_or_warn(sb.buf);
1767                         strbuf_release(&sb);
1768
1769                         if (is_empty_dir(path))
1770                                 rmdir_or_warn(path);
1771                 }
1772         }
1773 out:
1774         return ret;
1775 }
1776
1777 /*
1778  * Embeds a single submodules git directory into the superprojects git dir,
1779  * non recursively.
1780  */
1781 static void relocate_single_git_dir_into_superproject(const char *prefix,
1782                                                       const char *path)
1783 {
1784         char *old_git_dir = NULL, *real_old_git_dir = NULL, *real_new_git_dir = NULL;
1785         const char *new_git_dir;
1786         const struct submodule *sub;
1787
1788         if (submodule_uses_worktrees(path))
1789                 die(_("relocate_gitdir for submodule '%s' with "
1790                       "more than one worktree not supported"), path);
1791
1792         old_git_dir = xstrfmt("%s/.git", path);
1793         if (read_gitfile(old_git_dir))
1794                 /* If it is an actual gitfile, it doesn't need migration. */
1795                 return;
1796
1797         real_old_git_dir = real_pathdup(old_git_dir, 1);
1798
1799         sub = submodule_from_path(the_repository, &null_oid, path);
1800         if (!sub)
1801                 die(_("could not lookup name for submodule '%s'"), path);
1802
1803         new_git_dir = git_path("modules/%s", sub->name);
1804         if (safe_create_leading_directories_const(new_git_dir) < 0)
1805                 die(_("could not create directory '%s'"), new_git_dir);
1806         real_new_git_dir = real_pathdup(new_git_dir, 1);
1807
1808         fprintf(stderr, _("Migrating git directory of '%s%s' from\n'%s' to\n'%s'\n"),
1809                 get_super_prefix_or_empty(), path,
1810                 real_old_git_dir, real_new_git_dir);
1811
1812         relocate_gitdir(path, real_old_git_dir, real_new_git_dir);
1813
1814         free(old_git_dir);
1815         free(real_old_git_dir);
1816         free(real_new_git_dir);
1817 }
1818
1819 /*
1820  * Migrate the git directory of the submodule given by path from
1821  * having its git directory within the working tree to the git dir nested
1822  * in its superprojects git dir under modules/.
1823  */
1824 void absorb_git_dir_into_superproject(const char *prefix,
1825                                       const char *path,
1826                                       unsigned flags)
1827 {
1828         int err_code;
1829         const char *sub_git_dir;
1830         struct strbuf gitdir = STRBUF_INIT;
1831         strbuf_addf(&gitdir, "%s/.git", path);
1832         sub_git_dir = resolve_gitdir_gently(gitdir.buf, &err_code);
1833
1834         /* Not populated? */
1835         if (!sub_git_dir) {
1836                 const struct submodule *sub;
1837
1838                 if (err_code == READ_GITFILE_ERR_STAT_FAILED) {
1839                         /* unpopulated as expected */
1840                         strbuf_release(&gitdir);
1841                         return;
1842                 }
1843
1844                 if (err_code != READ_GITFILE_ERR_NOT_A_REPO)
1845                         /* We don't know what broke here. */
1846                         read_gitfile_error_die(err_code, path, NULL);
1847
1848                 /*
1849                 * Maybe populated, but no git directory was found?
1850                 * This can happen if the superproject is a submodule
1851                 * itself and was just absorbed. The absorption of the
1852                 * superproject did not rewrite the git file links yet,
1853                 * fix it now.
1854                 */
1855                 sub = submodule_from_path(the_repository, &null_oid, path);
1856                 if (!sub)
1857                         die(_("could not lookup name for submodule '%s'"), path);
1858                 connect_work_tree_and_git_dir(path,
1859                         git_path("modules/%s", sub->name), 0);
1860         } else {
1861                 /* Is it already absorbed into the superprojects git dir? */
1862                 char *real_sub_git_dir = real_pathdup(sub_git_dir, 1);
1863                 char *real_common_git_dir = real_pathdup(get_git_common_dir(), 1);
1864
1865                 if (!starts_with(real_sub_git_dir, real_common_git_dir))
1866                         relocate_single_git_dir_into_superproject(prefix, path);
1867
1868                 free(real_sub_git_dir);
1869                 free(real_common_git_dir);
1870         }
1871         strbuf_release(&gitdir);
1872
1873         if (flags & ABSORB_GITDIR_RECURSE_SUBMODULES) {
1874                 struct child_process cp = CHILD_PROCESS_INIT;
1875                 struct strbuf sb = STRBUF_INIT;
1876
1877                 if (flags & ~ABSORB_GITDIR_RECURSE_SUBMODULES)
1878                         BUG("we don't know how to pass the flags down?");
1879
1880                 strbuf_addstr(&sb, get_super_prefix_or_empty());
1881                 strbuf_addstr(&sb, path);
1882                 strbuf_addch(&sb, '/');
1883
1884                 cp.dir = path;
1885                 cp.git_cmd = 1;
1886                 cp.no_stdin = 1;
1887                 argv_array_pushl(&cp.args, "--super-prefix", sb.buf,
1888                                            "submodule--helper",
1889                                            "absorb-git-dirs", NULL);
1890                 prepare_submodule_repo_env(&cp.env_array);
1891                 if (run_command(&cp))
1892                         die(_("could not recurse into submodule '%s'"), path);
1893
1894                 strbuf_release(&sb);
1895         }
1896 }
1897
1898 const char *get_superproject_working_tree(void)
1899 {
1900         struct child_process cp = CHILD_PROCESS_INIT;
1901         struct strbuf sb = STRBUF_INIT;
1902         const char *one_up = real_path_if_valid("../");
1903         const char *cwd = xgetcwd();
1904         const char *ret = NULL;
1905         const char *subpath;
1906         int code;
1907         ssize_t len;
1908
1909         if (!is_inside_work_tree())
1910                 /*
1911                  * FIXME:
1912                  * We might have a superproject, but it is harder
1913                  * to determine.
1914                  */
1915                 return NULL;
1916
1917         if (!one_up)
1918                 return NULL;
1919
1920         subpath = relative_path(cwd, one_up, &sb);
1921
1922         prepare_submodule_repo_env(&cp.env_array);
1923         argv_array_pop(&cp.env_array);
1924
1925         argv_array_pushl(&cp.args, "--literal-pathspecs", "-C", "..",
1926                         "ls-files", "-z", "--stage", "--full-name", "--",
1927                         subpath, NULL);
1928         strbuf_reset(&sb);
1929
1930         cp.no_stdin = 1;
1931         cp.no_stderr = 1;
1932         cp.out = -1;
1933         cp.git_cmd = 1;
1934
1935         if (start_command(&cp))
1936                 die(_("could not start ls-files in .."));
1937
1938         len = strbuf_read(&sb, cp.out, PATH_MAX);
1939         close(cp.out);
1940
1941         if (starts_with(sb.buf, "160000")) {
1942                 int super_sub_len;
1943                 int cwd_len = strlen(cwd);
1944                 char *super_sub, *super_wt;
1945
1946                 /*
1947                  * There is a superproject having this repo as a submodule.
1948                  * The format is <mode> SP <hash> SP <stage> TAB <full name> \0,
1949                  * We're only interested in the name after the tab.
1950                  */
1951                 super_sub = strchr(sb.buf, '\t') + 1;
1952                 super_sub_len = strlen(super_sub);
1953
1954                 if (super_sub_len > cwd_len ||
1955                     strcmp(&cwd[cwd_len - super_sub_len], super_sub))
1956                         BUG("returned path string doesn't match cwd?");
1957
1958                 super_wt = xstrdup(cwd);
1959                 super_wt[cwd_len - super_sub_len] = '\0';
1960
1961                 ret = real_path(super_wt);
1962                 free(super_wt);
1963         }
1964         strbuf_release(&sb);
1965
1966         code = finish_command(&cp);
1967
1968         if (code == 128)
1969                 /* '../' is not a git repository */
1970                 return NULL;
1971         if (code == 0 && len == 0)
1972                 /* There is an unrelated git repository at '../' */
1973                 return NULL;
1974         if (code)
1975                 die(_("ls-tree returned unexpected return code %d"), code);
1976
1977         return ret;
1978 }
1979
1980 /*
1981  * Put the gitdir for a submodule (given relative to the main
1982  * repository worktree) into `buf`, or return -1 on error.
1983  */
1984 int submodule_to_gitdir(struct strbuf *buf, const char *submodule)
1985 {
1986         const struct submodule *sub;
1987         const char *git_dir;
1988         int ret = 0;
1989
1990         strbuf_reset(buf);
1991         strbuf_addstr(buf, submodule);
1992         strbuf_complete(buf, '/');
1993         strbuf_addstr(buf, ".git");
1994
1995         git_dir = read_gitfile(buf->buf);
1996         if (git_dir) {
1997                 strbuf_reset(buf);
1998                 strbuf_addstr(buf, git_dir);
1999         }
2000         if (!is_git_directory(buf->buf)) {
2001                 sub = submodule_from_path(the_repository, &null_oid, submodule);
2002                 if (!sub) {
2003                         ret = -1;
2004                         goto cleanup;
2005                 }
2006                 strbuf_reset(buf);
2007                 strbuf_git_path(buf, "%s/%s", "modules", sub->name);
2008         }
2009
2010 cleanup:
2011         return ret;
2012 }