Git 2.20.4
[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 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                 format_commit_message(commit, format, &sb, &ctx);
472                 strbuf_addch(&sb, '\n');
473                 if (commit->object.flags & SYMMETRIC_LEFT)
474                         diff_emit_submodule_del(o, sb.buf);
475                 else
476                         diff_emit_submodule_add(o, sb.buf);
477         }
478         strbuf_release(&sb);
479 }
480
481 static void prepare_submodule_repo_env_no_git_dir(struct argv_array *out)
482 {
483         const char * const *var;
484
485         for (var = local_repo_env; *var; var++) {
486                 if (strcmp(*var, CONFIG_DATA_ENVIRONMENT))
487                         argv_array_push(out, *var);
488         }
489 }
490
491 void prepare_submodule_repo_env(struct argv_array *out)
492 {
493         prepare_submodule_repo_env_no_git_dir(out);
494         argv_array_pushf(out, "%s=%s", GIT_DIR_ENVIRONMENT,
495                          DEFAULT_GIT_DIR_ENVIRONMENT);
496 }
497
498 /* Helper function to display the submodule header line prior to the full
499  * summary output. If it can locate the submodule objects directory it will
500  * attempt to lookup both the left and right commits and put them into the
501  * left and right pointers.
502  */
503 static void show_submodule_header(struct diff_options *o, const char *path,
504                 struct object_id *one, struct object_id *two,
505                 unsigned dirty_submodule,
506                 struct commit **left, struct commit **right,
507                 struct commit_list **merge_bases)
508 {
509         const char *message = NULL;
510         struct strbuf sb = STRBUF_INIT;
511         int fast_forward = 0, fast_backward = 0;
512
513         if (dirty_submodule & DIRTY_SUBMODULE_UNTRACKED)
514                 diff_emit_submodule_untracked(o, path);
515
516         if (dirty_submodule & DIRTY_SUBMODULE_MODIFIED)
517                 diff_emit_submodule_modified(o, path);
518
519         if (is_null_oid(one))
520                 message = "(new submodule)";
521         else if (is_null_oid(two))
522                 message = "(submodule deleted)";
523
524         if (add_submodule_odb(path)) {
525                 if (!message)
526                         message = "(commits not present)";
527                 goto output_header;
528         }
529
530         /*
531          * Attempt to lookup the commit references, and determine if this is
532          * a fast forward or fast backwards update.
533          */
534         *left = lookup_commit_reference(the_repository, one);
535         *right = lookup_commit_reference(the_repository, two);
536
537         /*
538          * Warn about missing commits in the submodule project, but only if
539          * they aren't null.
540          */
541         if ((!is_null_oid(one) && !*left) ||
542              (!is_null_oid(two) && !*right))
543                 message = "(commits not present)";
544
545         *merge_bases = get_merge_bases(*left, *right);
546         if (*merge_bases) {
547                 if ((*merge_bases)->item == *left)
548                         fast_forward = 1;
549                 else if ((*merge_bases)->item == *right)
550                         fast_backward = 1;
551         }
552
553         if (oideq(one, two)) {
554                 strbuf_release(&sb);
555                 return;
556         }
557
558 output_header:
559         strbuf_addf(&sb, "Submodule %s ", path);
560         strbuf_add_unique_abbrev(&sb, one, DEFAULT_ABBREV);
561         strbuf_addstr(&sb, (fast_backward || fast_forward) ? ".." : "...");
562         strbuf_add_unique_abbrev(&sb, two, DEFAULT_ABBREV);
563         if (message)
564                 strbuf_addf(&sb, " %s\n", message);
565         else
566                 strbuf_addf(&sb, "%s:\n", fast_backward ? " (rewind)" : "");
567         diff_emit_submodule_header(o, sb.buf);
568
569         strbuf_release(&sb);
570 }
571
572 void show_submodule_summary(struct diff_options *o, const char *path,
573                 struct object_id *one, struct object_id *two,
574                 unsigned dirty_submodule)
575 {
576         struct rev_info rev;
577         struct commit *left = NULL, *right = NULL;
578         struct commit_list *merge_bases = NULL;
579
580         show_submodule_header(o, path, one, two, dirty_submodule,
581                               &left, &right, &merge_bases);
582
583         /*
584          * If we don't have both a left and a right pointer, there is no
585          * reason to try and display a summary. The header line should contain
586          * all the information the user needs.
587          */
588         if (!left || !right)
589                 goto out;
590
591         /* Treat revision walker failure the same as missing commits */
592         if (prepare_submodule_summary(&rev, path, left, right, merge_bases)) {
593                 diff_emit_submodule_error(o, "(revision walker failed)\n");
594                 goto out;
595         }
596
597         print_submodule_summary(&rev, o);
598
599 out:
600         if (merge_bases)
601                 free_commit_list(merge_bases);
602         clear_commit_marks(left, ~0);
603         clear_commit_marks(right, ~0);
604 }
605
606 void show_submodule_inline_diff(struct diff_options *o, const char *path,
607                 struct object_id *one, struct object_id *two,
608                 unsigned dirty_submodule)
609 {
610         const struct object_id *old_oid = the_hash_algo->empty_tree, *new_oid = the_hash_algo->empty_tree;
611         struct commit *left = NULL, *right = NULL;
612         struct commit_list *merge_bases = NULL;
613         struct child_process cp = CHILD_PROCESS_INIT;
614         struct strbuf sb = STRBUF_INIT;
615
616         show_submodule_header(o, path, one, two, dirty_submodule,
617                               &left, &right, &merge_bases);
618
619         /* We need a valid left and right commit to display a difference */
620         if (!(left || is_null_oid(one)) ||
621             !(right || is_null_oid(two)))
622                 goto done;
623
624         if (left)
625                 old_oid = one;
626         if (right)
627                 new_oid = two;
628
629         cp.git_cmd = 1;
630         cp.dir = path;
631         cp.out = -1;
632         cp.no_stdin = 1;
633
634         /* TODO: other options may need to be passed here. */
635         argv_array_pushl(&cp.args, "diff", "--submodule=diff", NULL);
636         argv_array_pushf(&cp.args, "--color=%s", want_color(o->use_color) ?
637                          "always" : "never");
638
639         if (o->flags.reverse_diff) {
640                 argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
641                                  o->b_prefix, path);
642                 argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
643                                  o->a_prefix, path);
644         } else {
645                 argv_array_pushf(&cp.args, "--src-prefix=%s%s/",
646                                  o->a_prefix, path);
647                 argv_array_pushf(&cp.args, "--dst-prefix=%s%s/",
648                                  o->b_prefix, path);
649         }
650         argv_array_push(&cp.args, oid_to_hex(old_oid));
651         /*
652          * If the submodule has modified content, we will diff against the
653          * work tree, under the assumption that the user has asked for the
654          * diff format and wishes to actually see all differences even if they
655          * haven't yet been committed to the submodule yet.
656          */
657         if (!(dirty_submodule & DIRTY_SUBMODULE_MODIFIED))
658                 argv_array_push(&cp.args, oid_to_hex(new_oid));
659
660         prepare_submodule_repo_env(&cp.env_array);
661         if (start_command(&cp))
662                 diff_emit_submodule_error(o, "(diff failed)\n");
663
664         while (strbuf_getwholeline_fd(&sb, cp.out, '\n') != EOF)
665                 diff_emit_submodule_pipethrough(o, sb.buf, sb.len);
666
667         if (finish_command(&cp))
668                 diff_emit_submodule_error(o, "(diff failed)\n");
669
670 done:
671         strbuf_release(&sb);
672         if (merge_bases)
673                 free_commit_list(merge_bases);
674         if (left)
675                 clear_commit_marks(left, ~0);
676         if (right)
677                 clear_commit_marks(right, ~0);
678 }
679
680 int should_update_submodules(void)
681 {
682         return config_update_recurse_submodules == RECURSE_SUBMODULES_ON;
683 }
684
685 const struct submodule *submodule_from_ce(const struct cache_entry *ce)
686 {
687         if (!S_ISGITLINK(ce->ce_mode))
688                 return NULL;
689
690         if (!should_update_submodules())
691                 return NULL;
692
693         return submodule_from_path(the_repository, &null_oid, ce->name);
694 }
695
696 static struct oid_array *submodule_commits(struct string_list *submodules,
697                                            const char *name)
698 {
699         struct string_list_item *item;
700
701         item = string_list_insert(submodules, name);
702         if (item->util)
703                 return (struct oid_array *) item->util;
704
705         /* NEEDSWORK: should we have oid_array_init()? */
706         item->util = xcalloc(1, sizeof(struct oid_array));
707         return (struct oid_array *) item->util;
708 }
709
710 struct collect_changed_submodules_cb_data {
711         struct repository *repo;
712         struct string_list *changed;
713         const struct object_id *commit_oid;
714 };
715
716 /*
717  * this would normally be two functions: default_name_from_path() and
718  * path_from_default_name(). Since the default name is the same as
719  * the submodule path we can get away with just one function which only
720  * checks whether there is a submodule in the working directory at that
721  * location.
722  */
723 static const char *default_name_or_path(const char *path_or_name)
724 {
725         int error_code;
726
727         if (!is_submodule_populated_gently(path_or_name, &error_code))
728                 return NULL;
729
730         return path_or_name;
731 }
732
733 static void collect_changed_submodules_cb(struct diff_queue_struct *q,
734                                           struct diff_options *options,
735                                           void *data)
736 {
737         struct collect_changed_submodules_cb_data *me = data;
738         struct string_list *changed = me->changed;
739         const struct object_id *commit_oid = me->commit_oid;
740         int i;
741
742         for (i = 0; i < q->nr; i++) {
743                 struct diff_filepair *p = q->queue[i];
744                 struct oid_array *commits;
745                 const struct submodule *submodule;
746                 const char *name;
747
748                 if (!S_ISGITLINK(p->two->mode))
749                         continue;
750
751                 submodule = submodule_from_path(me->repo,
752                                                 commit_oid, p->two->path);
753                 if (submodule)
754                         name = submodule->name;
755                 else {
756                         name = default_name_or_path(p->two->path);
757                         /* make sure name does not collide with existing one */
758                         if (name)
759                                 submodule = submodule_from_name(me->repo,
760                                                                 commit_oid, name);
761                         if (submodule) {
762                                 warning("Submodule in commit %s at path: "
763                                         "'%s' collides with a submodule named "
764                                         "the same. Skipping it.",
765                                         oid_to_hex(commit_oid), p->two->path);
766                                 name = NULL;
767                         }
768                 }
769
770                 if (!name)
771                         continue;
772
773                 commits = submodule_commits(changed, name);
774                 oid_array_append(commits, &p->two->oid);
775         }
776 }
777
778 /*
779  * Collect the paths of submodules in 'changed' which have changed based on
780  * the revisions as specified in 'argv'.  Each entry in 'changed' will also
781  * have a corresponding 'struct oid_array' (in the 'util' field) which lists
782  * what the submodule pointers were updated to during the change.
783  */
784 static void collect_changed_submodules(struct repository *r,
785                                        struct string_list *changed,
786                                        struct argv_array *argv)
787 {
788         struct rev_info rev;
789         const struct commit *commit;
790
791         repo_init_revisions(r, &rev, NULL);
792         setup_revisions(argv->argc, argv->argv, &rev, NULL);
793         if (prepare_revision_walk(&rev))
794                 die("revision walk setup failed");
795
796         while ((commit = get_revision(&rev))) {
797                 struct rev_info diff_rev;
798                 struct collect_changed_submodules_cb_data data;
799                 data.repo = r;
800                 data.changed = changed;
801                 data.commit_oid = &commit->object.oid;
802
803                 repo_init_revisions(r, &diff_rev, NULL);
804                 diff_rev.diffopt.output_format |= DIFF_FORMAT_CALLBACK;
805                 diff_rev.diffopt.format_callback = collect_changed_submodules_cb;
806                 diff_rev.diffopt.format_callback_data = &data;
807                 diff_tree_combined_merge(commit, 1, &diff_rev);
808         }
809
810         reset_revision_walk();
811 }
812
813 static void free_submodules_oids(struct string_list *submodules)
814 {
815         struct string_list_item *item;
816         for_each_string_list_item(item, submodules)
817                 oid_array_clear((struct oid_array *) item->util);
818         string_list_clear(submodules, 1);
819 }
820
821 static int has_remote(const char *refname, const struct object_id *oid,
822                       int flags, void *cb_data)
823 {
824         return 1;
825 }
826
827 static int append_oid_to_argv(const struct object_id *oid, void *data)
828 {
829         struct argv_array *argv = data;
830         argv_array_push(argv, oid_to_hex(oid));
831         return 0;
832 }
833
834 struct has_commit_data {
835         struct repository *repo;
836         int result;
837         const char *path;
838 };
839
840 static int check_has_commit(const struct object_id *oid, void *data)
841 {
842         struct has_commit_data *cb = data;
843
844         enum object_type type = oid_object_info(cb->repo, oid, NULL);
845
846         switch (type) {
847         case OBJ_COMMIT:
848                 return 0;
849         case OBJ_BAD:
850                 /*
851                  * Object is missing or invalid. If invalid, an error message
852                  * has already been printed.
853                  */
854                 cb->result = 0;
855                 return 0;
856         default:
857                 die(_("submodule entry '%s' (%s) is a %s, not a commit"),
858                     cb->path, oid_to_hex(oid), type_name(type));
859         }
860 }
861
862 static int submodule_has_commits(struct repository *r,
863                                  const char *path,
864                                  struct oid_array *commits)
865 {
866         struct has_commit_data has_commit = { r, 1, path };
867
868         /*
869          * Perform a cheap, but incorrect check for the existence of 'commits'.
870          * This is done by adding the submodule's object store to the in-core
871          * object store, and then querying for each commit's existence.  If we
872          * do not have the commit object anywhere, there is no chance we have
873          * it in the object store of the correct submodule and have it
874          * reachable from a ref, so we can fail early without spawning rev-list
875          * which is expensive.
876          */
877         if (add_submodule_odb(path))
878                 return 0;
879
880         oid_array_for_each_unique(commits, check_has_commit, &has_commit);
881
882         if (has_commit.result) {
883                 /*
884                  * Even if the submodule is checked out and the commit is
885                  * present, make sure it exists in the submodule's object store
886                  * and that it is reachable from a ref.
887                  */
888                 struct child_process cp = CHILD_PROCESS_INIT;
889                 struct strbuf out = STRBUF_INIT;
890
891                 argv_array_pushl(&cp.args, "rev-list", "-n", "1", NULL);
892                 oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
893                 argv_array_pushl(&cp.args, "--not", "--all", NULL);
894
895                 prepare_submodule_repo_env(&cp.env_array);
896                 cp.git_cmd = 1;
897                 cp.no_stdin = 1;
898                 cp.dir = path;
899
900                 if (capture_command(&cp, &out, GIT_MAX_HEXSZ + 1) || out.len)
901                         has_commit.result = 0;
902
903                 strbuf_release(&out);
904         }
905
906         return has_commit.result;
907 }
908
909 static int submodule_needs_pushing(struct repository *r,
910                                    const char *path,
911                                    struct oid_array *commits)
912 {
913         if (!submodule_has_commits(r, path, commits))
914                 /*
915                  * NOTE: We do consider it safe to return "no" here. The
916                  * correct answer would be "We do not know" instead of
917                  * "No push needed", but it is quite hard to change
918                  * the submodule pointer without having the submodule
919                  * around. If a user did however change the submodules
920                  * without having the submodule around, this indicates
921                  * an expert who knows what they are doing or a
922                  * maintainer integrating work from other people. In
923                  * both cases it should be safe to skip this check.
924                  */
925                 return 0;
926
927         if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
928                 struct child_process cp = CHILD_PROCESS_INIT;
929                 struct strbuf buf = STRBUF_INIT;
930                 int needs_pushing = 0;
931
932                 argv_array_push(&cp.args, "rev-list");
933                 oid_array_for_each_unique(commits, append_oid_to_argv, &cp.args);
934                 argv_array_pushl(&cp.args, "--not", "--remotes", "-n", "1" , NULL);
935
936                 prepare_submodule_repo_env(&cp.env_array);
937                 cp.git_cmd = 1;
938                 cp.no_stdin = 1;
939                 cp.out = -1;
940                 cp.dir = path;
941                 if (start_command(&cp))
942                         die("Could not run 'git rev-list <commits> --not --remotes -n 1' command in submodule %s",
943                                         path);
944                 if (strbuf_read(&buf, cp.out, 41))
945                         needs_pushing = 1;
946                 finish_command(&cp);
947                 close(cp.out);
948                 strbuf_release(&buf);
949                 return needs_pushing;
950         }
951
952         return 0;
953 }
954
955 int find_unpushed_submodules(struct repository *r,
956                              struct oid_array *commits,
957                              const char *remotes_name,
958                              struct string_list *needs_pushing)
959 {
960         struct string_list submodules = STRING_LIST_INIT_DUP;
961         struct string_list_item *name;
962         struct argv_array argv = ARGV_ARRAY_INIT;
963
964         /* argv.argv[0] will be ignored by setup_revisions */
965         argv_array_push(&argv, "find_unpushed_submodules");
966         oid_array_for_each_unique(commits, append_oid_to_argv, &argv);
967         argv_array_push(&argv, "--not");
968         argv_array_pushf(&argv, "--remotes=%s", remotes_name);
969
970         collect_changed_submodules(r, &submodules, &argv);
971
972         for_each_string_list_item(name, &submodules) {
973                 struct oid_array *commits = name->util;
974                 const struct submodule *submodule;
975                 const char *path = NULL;
976
977                 submodule = submodule_from_name(r, &null_oid, name->string);
978                 if (submodule)
979                         path = submodule->path;
980                 else
981                         path = default_name_or_path(name->string);
982
983                 if (!path)
984                         continue;
985
986                 if (submodule_needs_pushing(r, path, commits))
987                         string_list_insert(needs_pushing, path);
988         }
989
990         free_submodules_oids(&submodules);
991         argv_array_clear(&argv);
992
993         return needs_pushing->nr;
994 }
995
996 static int push_submodule(const char *path,
997                           const struct remote *remote,
998                           const struct refspec *rs,
999                           const struct string_list *push_options,
1000                           int dry_run)
1001 {
1002         if (add_submodule_odb(path))
1003                 return 1;
1004
1005         if (for_each_remote_ref_submodule(path, has_remote, NULL) > 0) {
1006                 struct child_process cp = CHILD_PROCESS_INIT;
1007                 argv_array_push(&cp.args, "push");
1008                 if (dry_run)
1009                         argv_array_push(&cp.args, "--dry-run");
1010
1011                 if (push_options && push_options->nr) {
1012                         const struct string_list_item *item;
1013                         for_each_string_list_item(item, push_options)
1014                                 argv_array_pushf(&cp.args, "--push-option=%s",
1015                                                  item->string);
1016                 }
1017
1018                 if (remote->origin != REMOTE_UNCONFIGURED) {
1019                         int i;
1020                         argv_array_push(&cp.args, remote->name);
1021                         for (i = 0; i < rs->raw_nr; i++)
1022                                 argv_array_push(&cp.args, rs->raw[i]);
1023                 }
1024
1025                 prepare_submodule_repo_env(&cp.env_array);
1026                 cp.git_cmd = 1;
1027                 cp.no_stdin = 1;
1028                 cp.dir = path;
1029                 if (run_command(&cp))
1030                         return 0;
1031                 close(cp.out);
1032         }
1033
1034         return 1;
1035 }
1036
1037 /*
1038  * Perform a check in the submodule to see if the remote and refspec work.
1039  * Die if the submodule can't be pushed.
1040  */
1041 static void submodule_push_check(const char *path, const char *head,
1042                                  const struct remote *remote,
1043                                  const struct refspec *rs)
1044 {
1045         struct child_process cp = CHILD_PROCESS_INIT;
1046         int i;
1047
1048         argv_array_push(&cp.args, "submodule--helper");
1049         argv_array_push(&cp.args, "push-check");
1050         argv_array_push(&cp.args, head);
1051         argv_array_push(&cp.args, remote->name);
1052
1053         for (i = 0; i < rs->raw_nr; i++)
1054                 argv_array_push(&cp.args, rs->raw[i]);
1055
1056         prepare_submodule_repo_env(&cp.env_array);
1057         cp.git_cmd = 1;
1058         cp.no_stdin = 1;
1059         cp.no_stdout = 1;
1060         cp.dir = path;
1061
1062         /*
1063          * Simply indicate if 'submodule--helper push-check' failed.
1064          * More detailed error information will be provided by the
1065          * child process.
1066          */
1067         if (run_command(&cp))
1068                 die("process for submodule '%s' failed", path);
1069 }
1070
1071 int push_unpushed_submodules(struct repository *r,
1072                              struct oid_array *commits,
1073                              const struct remote *remote,
1074                              const struct refspec *rs,
1075                              const struct string_list *push_options,
1076                              int dry_run)
1077 {
1078         int i, ret = 1;
1079         struct string_list needs_pushing = STRING_LIST_INIT_DUP;
1080
1081         if (!find_unpushed_submodules(r, commits,
1082                                       remote->name, &needs_pushing))
1083                 return 1;
1084
1085         /*
1086          * Verify that the remote and refspec can be propagated to all
1087          * submodules.  This check can be skipped if the remote and refspec
1088          * won't be propagated due to the remote being unconfigured (e.g. a URL
1089          * instead of a remote name).
1090          */
1091         if (remote->origin != REMOTE_UNCONFIGURED) {
1092                 char *head;
1093                 struct object_id head_oid;
1094
1095                 head = resolve_refdup("HEAD", 0, &head_oid, NULL);
1096                 if (!head)
1097                         die(_("Failed to resolve HEAD as a valid ref."));
1098
1099                 for (i = 0; i < needs_pushing.nr; i++)
1100                         submodule_push_check(needs_pushing.items[i].string,
1101                                              head, remote, rs);
1102                 free(head);
1103         }
1104
1105         /* Actually push the submodules */
1106         for (i = 0; i < needs_pushing.nr; i++) {
1107                 const char *path = needs_pushing.items[i].string;
1108                 fprintf(stderr, "Pushing submodule '%s'\n", path);
1109                 if (!push_submodule(path, remote, rs,
1110                                     push_options, dry_run)) {
1111                         fprintf(stderr, "Unable to push submodule '%s'\n", path);
1112                         ret = 0;
1113                 }
1114         }
1115
1116         string_list_clear(&needs_pushing, 0);
1117
1118         return ret;
1119 }
1120
1121 static int append_oid_to_array(const char *ref, const struct object_id *oid,
1122                                int flags, void *data)
1123 {
1124         struct oid_array *array = data;
1125         oid_array_append(array, oid);
1126         return 0;
1127 }
1128
1129 void check_for_new_submodule_commits(struct object_id *oid)
1130 {
1131         if (!initialized_fetch_ref_tips) {
1132                 for_each_ref(append_oid_to_array, &ref_tips_before_fetch);
1133                 initialized_fetch_ref_tips = 1;
1134         }
1135
1136         oid_array_append(&ref_tips_after_fetch, oid);
1137 }
1138
1139 static void calculate_changed_submodule_paths(struct repository *r)
1140 {
1141         struct argv_array argv = ARGV_ARRAY_INIT;
1142         struct string_list changed_submodules = STRING_LIST_INIT_DUP;
1143         const struct string_list_item *name;
1144
1145         /* No need to check if there are no submodules configured */
1146         if (!submodule_from_path(r, NULL, NULL))
1147                 return;
1148
1149         argv_array_push(&argv, "--"); /* argv[0] program name */
1150         oid_array_for_each_unique(&ref_tips_after_fetch,
1151                                    append_oid_to_argv, &argv);
1152         argv_array_push(&argv, "--not");
1153         oid_array_for_each_unique(&ref_tips_before_fetch,
1154                                    append_oid_to_argv, &argv);
1155
1156         /*
1157          * Collect all submodules (whether checked out or not) for which new
1158          * commits have been recorded upstream in "changed_submodule_names".
1159          */
1160         collect_changed_submodules(r, &changed_submodules, &argv);
1161
1162         for_each_string_list_item(name, &changed_submodules) {
1163                 struct oid_array *commits = name->util;
1164                 const struct submodule *submodule;
1165                 const char *path = NULL;
1166
1167                 submodule = submodule_from_name(r, &null_oid, name->string);
1168                 if (submodule)
1169                         path = submodule->path;
1170                 else
1171                         path = default_name_or_path(name->string);
1172
1173                 if (!path)
1174                         continue;
1175
1176                 if (!submodule_has_commits(r, path, commits))
1177                         string_list_append(&changed_submodule_names, name->string);
1178         }
1179
1180         free_submodules_oids(&changed_submodules);
1181         argv_array_clear(&argv);
1182         oid_array_clear(&ref_tips_before_fetch);
1183         oid_array_clear(&ref_tips_after_fetch);
1184         initialized_fetch_ref_tips = 0;
1185 }
1186
1187 int submodule_touches_in_range(struct repository *r,
1188                                struct object_id *excl_oid,
1189                                struct object_id *incl_oid)
1190 {
1191         struct string_list subs = STRING_LIST_INIT_DUP;
1192         struct argv_array args = ARGV_ARRAY_INIT;
1193         int ret;
1194
1195         /* No need to check if there are no submodules configured */
1196         if (!submodule_from_path(r, NULL, NULL))
1197                 return 0;
1198
1199         argv_array_push(&args, "--"); /* args[0] program name */
1200         argv_array_push(&args, oid_to_hex(incl_oid));
1201         if (!is_null_oid(excl_oid)) {
1202                 argv_array_push(&args, "--not");
1203                 argv_array_push(&args, oid_to_hex(excl_oid));
1204         }
1205
1206         collect_changed_submodules(r, &subs, &args);
1207         ret = subs.nr;
1208
1209         argv_array_clear(&args);
1210
1211         free_submodules_oids(&subs);
1212         return ret;
1213 }
1214
1215 struct submodule_parallel_fetch {
1216         int count;
1217         struct argv_array args;
1218         struct repository *r;
1219         const char *prefix;
1220         int command_line_option;
1221         int default_option;
1222         int quiet;
1223         int result;
1224 };
1225 #define SPF_INIT {0, ARGV_ARRAY_INIT, NULL, NULL, 0, 0, 0, 0}
1226
1227 static int get_fetch_recurse_config(const struct submodule *submodule,
1228                                     struct submodule_parallel_fetch *spf)
1229 {
1230         if (spf->command_line_option != RECURSE_SUBMODULES_DEFAULT)
1231                 return spf->command_line_option;
1232
1233         if (submodule) {
1234                 char *key;
1235                 const char *value;
1236
1237                 int fetch_recurse = submodule->fetch_recurse;
1238                 key = xstrfmt("submodule.%s.fetchRecurseSubmodules", submodule->name);
1239                 if (!repo_config_get_string_const(spf->r, key, &value)) {
1240                         fetch_recurse = parse_fetch_recurse_submodules_arg(key, value);
1241                 }
1242                 free(key);
1243
1244                 if (fetch_recurse != RECURSE_SUBMODULES_NONE)
1245                         /* local config overrules everything except commandline */
1246                         return fetch_recurse;
1247         }
1248
1249         return spf->default_option;
1250 }
1251
1252 static int get_next_submodule(struct child_process *cp,
1253                               struct strbuf *err, void *data, void **task_cb)
1254 {
1255         int ret = 0;
1256         struct submodule_parallel_fetch *spf = data;
1257
1258         for (; spf->count < spf->r->index->cache_nr; spf->count++) {
1259                 struct strbuf submodule_path = STRBUF_INIT;
1260                 struct strbuf submodule_git_dir = STRBUF_INIT;
1261                 struct strbuf submodule_prefix = STRBUF_INIT;
1262                 const struct cache_entry *ce = spf->r->index->cache[spf->count];
1263                 const char *git_dir, *default_argv;
1264                 const struct submodule *submodule;
1265                 struct submodule default_submodule = SUBMODULE_INIT;
1266
1267                 if (!S_ISGITLINK(ce->ce_mode))
1268                         continue;
1269
1270                 submodule = submodule_from_path(spf->r, &null_oid, ce->name);
1271                 if (!submodule) {
1272                         const char *name = default_name_or_path(ce->name);
1273                         if (name) {
1274                                 default_submodule.path = default_submodule.name = name;
1275                                 submodule = &default_submodule;
1276                         }
1277                 }
1278
1279                 switch (get_fetch_recurse_config(submodule, spf))
1280                 {
1281                 default:
1282                 case RECURSE_SUBMODULES_DEFAULT:
1283                 case RECURSE_SUBMODULES_ON_DEMAND:
1284                         if (!submodule || !unsorted_string_list_lookup(&changed_submodule_names,
1285                                                          submodule->name))
1286                                 continue;
1287                         default_argv = "on-demand";
1288                         break;
1289                 case RECURSE_SUBMODULES_ON:
1290                         default_argv = "yes";
1291                         break;
1292                 case RECURSE_SUBMODULES_OFF:
1293                         continue;
1294                 }
1295
1296                 strbuf_repo_worktree_path(&submodule_path, spf->r, "%s", ce->name);
1297                 strbuf_addf(&submodule_git_dir, "%s/.git", submodule_path.buf);
1298                 strbuf_addf(&submodule_prefix, "%s%s/", spf->prefix, ce->name);
1299                 git_dir = read_gitfile(submodule_git_dir.buf);
1300                 if (!git_dir)
1301                         git_dir = submodule_git_dir.buf;
1302                 if (is_directory(git_dir)) {
1303                         child_process_init(cp);
1304                         cp->dir = strbuf_detach(&submodule_path, NULL);
1305                         prepare_submodule_repo_env(&cp->env_array);
1306                         cp->git_cmd = 1;
1307                         if (!spf->quiet)
1308                                 strbuf_addf(err, "Fetching submodule %s%s\n",
1309                                             spf->prefix, ce->name);
1310                         argv_array_init(&cp->args);
1311                         argv_array_pushv(&cp->args, spf->args.argv);
1312                         argv_array_push(&cp->args, default_argv);
1313                         argv_array_push(&cp->args, "--submodule-prefix");
1314                         argv_array_push(&cp->args, submodule_prefix.buf);
1315                         ret = 1;
1316                 }
1317                 strbuf_release(&submodule_path);
1318                 strbuf_release(&submodule_git_dir);
1319                 strbuf_release(&submodule_prefix);
1320                 if (ret) {
1321                         spf->count++;
1322                         return 1;
1323                 }
1324         }
1325         return 0;
1326 }
1327
1328 static int fetch_start_failure(struct strbuf *err,
1329                                void *cb, void *task_cb)
1330 {
1331         struct submodule_parallel_fetch *spf = cb;
1332
1333         spf->result = 1;
1334
1335         return 0;
1336 }
1337
1338 static int fetch_finish(int retvalue, struct strbuf *err,
1339                         void *cb, void *task_cb)
1340 {
1341         struct submodule_parallel_fetch *spf = cb;
1342
1343         if (retvalue)
1344                 spf->result = 1;
1345
1346         return 0;
1347 }
1348
1349 int fetch_populated_submodules(struct repository *r,
1350                                const struct argv_array *options,
1351                                const char *prefix, int command_line_option,
1352                                int default_option,
1353                                int quiet, int max_parallel_jobs)
1354 {
1355         int i;
1356         struct submodule_parallel_fetch spf = SPF_INIT;
1357
1358         spf.r = r;
1359         spf.command_line_option = command_line_option;
1360         spf.default_option = default_option;
1361         spf.quiet = quiet;
1362         spf.prefix = prefix;
1363
1364         if (!r->worktree)
1365                 goto out;
1366
1367         if (repo_read_index(r) < 0)
1368                 die("index file corrupt");
1369
1370         argv_array_push(&spf.args, "fetch");
1371         for (i = 0; i < options->argc; i++)
1372                 argv_array_push(&spf.args, options->argv[i]);
1373         argv_array_push(&spf.args, "--recurse-submodules-default");
1374         /* default value, "--submodule-prefix" and its value are added later */
1375
1376         calculate_changed_submodule_paths(r);
1377         run_processes_parallel(max_parallel_jobs,
1378                                get_next_submodule,
1379                                fetch_start_failure,
1380                                fetch_finish,
1381                                &spf);
1382
1383         argv_array_clear(&spf.args);
1384 out:
1385         string_list_clear(&changed_submodule_names, 1);
1386         return spf.result;
1387 }
1388
1389 unsigned is_submodule_modified(const char *path, int ignore_untracked)
1390 {
1391         struct child_process cp = CHILD_PROCESS_INIT;
1392         struct strbuf buf = STRBUF_INIT;
1393         FILE *fp;
1394         unsigned dirty_submodule = 0;
1395         const char *git_dir;
1396         int ignore_cp_exit_code = 0;
1397
1398         strbuf_addf(&buf, "%s/.git", path);
1399         git_dir = read_gitfile(buf.buf);
1400         if (!git_dir)
1401                 git_dir = buf.buf;
1402         if (!is_git_directory(git_dir)) {
1403                 if (is_directory(git_dir))
1404                         die(_("'%s' not recognized as a git repository"), git_dir);
1405                 strbuf_release(&buf);
1406                 /* The submodule is not checked out, so it is not modified */
1407                 return 0;
1408         }
1409         strbuf_reset(&buf);
1410
1411         argv_array_pushl(&cp.args, "status", "--porcelain=2", NULL);
1412         if (ignore_untracked)
1413                 argv_array_push(&cp.args, "-uno");
1414
1415         prepare_submodule_repo_env(&cp.env_array);
1416         cp.git_cmd = 1;
1417         cp.no_stdin = 1;
1418         cp.out = -1;
1419         cp.dir = path;
1420         if (start_command(&cp))
1421                 die("Could not run 'git status --porcelain=2' in submodule %s", path);
1422
1423         fp = xfdopen(cp.out, "r");
1424         while (strbuf_getwholeline(&buf, fp, '\n') != EOF) {
1425                 /* regular untracked files */
1426                 if (buf.buf[0] == '?')
1427                         dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
1428
1429                 if (buf.buf[0] == 'u' ||
1430                     buf.buf[0] == '1' ||
1431                     buf.buf[0] == '2') {
1432                         /* T = line type, XY = status, SSSS = submodule state */
1433                         if (buf.len < strlen("T XY SSSS"))
1434                                 BUG("invalid status --porcelain=2 line %s",
1435                                     buf.buf);
1436
1437                         if (buf.buf[5] == 'S' && buf.buf[8] == 'U')
1438                                 /* nested untracked file */
1439                                 dirty_submodule |= DIRTY_SUBMODULE_UNTRACKED;
1440
1441                         if (buf.buf[0] == 'u' ||
1442                             buf.buf[0] == '2' ||
1443                             memcmp(buf.buf + 5, "S..U", 4))
1444                                 /* other change */
1445                                 dirty_submodule |= DIRTY_SUBMODULE_MODIFIED;
1446                 }
1447
1448                 if ((dirty_submodule & DIRTY_SUBMODULE_MODIFIED) &&
1449                     ((dirty_submodule & DIRTY_SUBMODULE_UNTRACKED) ||
1450                      ignore_untracked)) {
1451                         /*
1452                          * We're not interested in any further information from
1453                          * the child any more, neither output nor its exit code.
1454                          */
1455                         ignore_cp_exit_code = 1;
1456                         break;
1457                 }
1458         }
1459         fclose(fp);
1460
1461         if (finish_command(&cp) && !ignore_cp_exit_code)
1462                 die("'git status --porcelain=2' failed in submodule %s", path);
1463
1464         strbuf_release(&buf);
1465         return dirty_submodule;
1466 }
1467
1468 int submodule_uses_gitfile(const char *path)
1469 {
1470         struct child_process cp = CHILD_PROCESS_INIT;
1471         const char *argv[] = {
1472                 "submodule",
1473                 "foreach",
1474                 "--quiet",
1475                 "--recursive",
1476                 "test -f .git",
1477                 NULL,
1478         };
1479         struct strbuf buf = STRBUF_INIT;
1480         const char *git_dir;
1481
1482         strbuf_addf(&buf, "%s/.git", path);
1483         git_dir = read_gitfile(buf.buf);
1484         if (!git_dir) {
1485                 strbuf_release(&buf);
1486                 return 0;
1487         }
1488         strbuf_release(&buf);
1489
1490         /* Now test that all nested submodules use a gitfile too */
1491         cp.argv = argv;
1492         prepare_submodule_repo_env(&cp.env_array);
1493         cp.git_cmd = 1;
1494         cp.no_stdin = 1;
1495         cp.no_stderr = 1;
1496         cp.no_stdout = 1;
1497         cp.dir = path;
1498         if (run_command(&cp))
1499                 return 0;
1500
1501         return 1;
1502 }
1503
1504 /*
1505  * Check if it is a bad idea to remove a submodule, i.e. if we'd lose data
1506  * when doing so.
1507  *
1508  * Return 1 if we'd lose data, return 0 if the removal is fine,
1509  * and negative values for errors.
1510  */
1511 int bad_to_remove_submodule(const char *path, unsigned flags)
1512 {
1513         ssize_t len;
1514         struct child_process cp = CHILD_PROCESS_INIT;
1515         struct strbuf buf = STRBUF_INIT;
1516         int ret = 0;
1517
1518         if (!file_exists(path) || is_empty_dir(path))
1519                 return 0;
1520
1521         if (!submodule_uses_gitfile(path))
1522                 return 1;
1523
1524         argv_array_pushl(&cp.args, "status", "--porcelain",
1525                                    "--ignore-submodules=none", NULL);
1526
1527         if (flags & SUBMODULE_REMOVAL_IGNORE_UNTRACKED)
1528                 argv_array_push(&cp.args, "-uno");
1529         else
1530                 argv_array_push(&cp.args, "-uall");
1531
1532         if (!(flags & SUBMODULE_REMOVAL_IGNORE_IGNORED_UNTRACKED))
1533                 argv_array_push(&cp.args, "--ignored");
1534
1535         prepare_submodule_repo_env(&cp.env_array);
1536         cp.git_cmd = 1;
1537         cp.no_stdin = 1;
1538         cp.out = -1;
1539         cp.dir = path;
1540         if (start_command(&cp)) {
1541                 if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
1542                         die(_("could not start 'git status' in submodule '%s'"),
1543                                 path);
1544                 ret = -1;
1545                 goto out;
1546         }
1547
1548         len = strbuf_read(&buf, cp.out, 1024);
1549         if (len > 2)
1550                 ret = 1;
1551         close(cp.out);
1552
1553         if (finish_command(&cp)) {
1554                 if (flags & SUBMODULE_REMOVAL_DIE_ON_ERROR)
1555                         die(_("could not run 'git status' in submodule '%s'"),
1556                                 path);
1557                 ret = -1;
1558         }
1559 out:
1560         strbuf_release(&buf);
1561         return ret;
1562 }
1563
1564 static const char *get_super_prefix_or_empty(void)
1565 {
1566         const char *s = get_super_prefix();
1567         if (!s)
1568                 s = "";
1569         return s;
1570 }
1571
1572 static int submodule_has_dirty_index(const struct submodule *sub)
1573 {
1574         struct child_process cp = CHILD_PROCESS_INIT;
1575
1576         prepare_submodule_repo_env(&cp.env_array);
1577
1578         cp.git_cmd = 1;
1579         argv_array_pushl(&cp.args, "diff-index", "--quiet",
1580                                    "--cached", "HEAD", NULL);
1581         cp.no_stdin = 1;
1582         cp.no_stdout = 1;
1583         cp.dir = sub->path;
1584         if (start_command(&cp))
1585                 die("could not recurse into submodule '%s'", sub->path);
1586
1587         return finish_command(&cp);
1588 }
1589
1590 static void submodule_reset_index(const char *path)
1591 {
1592         struct child_process cp = CHILD_PROCESS_INIT;
1593         prepare_submodule_repo_env(&cp.env_array);
1594
1595         cp.git_cmd = 1;
1596         cp.no_stdin = 1;
1597         cp.dir = path;
1598
1599         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
1600                                    get_super_prefix_or_empty(), path);
1601         argv_array_pushl(&cp.args, "read-tree", "-u", "--reset", NULL);
1602
1603         argv_array_push(&cp.args, empty_tree_oid_hex());
1604
1605         if (run_command(&cp))
1606                 die("could not reset submodule index");
1607 }
1608
1609 /**
1610  * Moves a submodule at a given path from a given head to another new head.
1611  * For edge cases (a submodule coming into existence or removing a submodule)
1612  * pass NULL for old or new respectively.
1613  */
1614 int submodule_move_head(const char *path,
1615                          const char *old_head,
1616                          const char *new_head,
1617                          unsigned flags)
1618 {
1619         int ret = 0;
1620         struct child_process cp = CHILD_PROCESS_INIT;
1621         const struct submodule *sub;
1622         int *error_code_ptr, error_code;
1623
1624         if (!is_submodule_active(the_repository, path))
1625                 return 0;
1626
1627         if (flags & SUBMODULE_MOVE_HEAD_FORCE)
1628                 /*
1629                  * Pass non NULL pointer to is_submodule_populated_gently
1630                  * to prevent die()-ing. We'll use connect_work_tree_and_git_dir
1631                  * to fixup the submodule in the force case later.
1632                  */
1633                 error_code_ptr = &error_code;
1634         else
1635                 error_code_ptr = NULL;
1636
1637         if (old_head && !is_submodule_populated_gently(path, error_code_ptr))
1638                 return 0;
1639
1640         sub = submodule_from_path(the_repository, &null_oid, path);
1641
1642         if (!sub)
1643                 BUG("could not get submodule information for '%s'", path);
1644
1645         if (old_head && !(flags & SUBMODULE_MOVE_HEAD_FORCE)) {
1646                 /* Check if the submodule has a dirty index. */
1647                 if (submodule_has_dirty_index(sub))
1648                         return error(_("submodule '%s' has dirty index"), path);
1649         }
1650
1651         if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
1652                 if (old_head) {
1653                         if (!submodule_uses_gitfile(path))
1654                                 absorb_git_dir_into_superproject("", path,
1655                                         ABSORB_GITDIR_RECURSE_SUBMODULES);
1656                 } else {
1657                         char *gitdir = xstrfmt("%s/modules/%s",
1658                                     get_git_common_dir(), sub->name);
1659                         connect_work_tree_and_git_dir(path, gitdir, 0);
1660                         free(gitdir);
1661
1662                         /* make sure the index is clean as well */
1663                         submodule_reset_index(path);
1664                 }
1665
1666                 if (old_head && (flags & SUBMODULE_MOVE_HEAD_FORCE)) {
1667                         char *gitdir = xstrfmt("%s/modules/%s",
1668                                     get_git_common_dir(), sub->name);
1669                         connect_work_tree_and_git_dir(path, gitdir, 1);
1670                         free(gitdir);
1671                 }
1672         }
1673
1674         prepare_submodule_repo_env(&cp.env_array);
1675
1676         cp.git_cmd = 1;
1677         cp.no_stdin = 1;
1678         cp.dir = path;
1679
1680         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
1681                         get_super_prefix_or_empty(), path);
1682         argv_array_pushl(&cp.args, "read-tree", "--recurse-submodules", NULL);
1683
1684         if (flags & SUBMODULE_MOVE_HEAD_DRY_RUN)
1685                 argv_array_push(&cp.args, "-n");
1686         else
1687                 argv_array_push(&cp.args, "-u");
1688
1689         if (flags & SUBMODULE_MOVE_HEAD_FORCE)
1690                 argv_array_push(&cp.args, "--reset");
1691         else
1692                 argv_array_push(&cp.args, "-m");
1693
1694         if (!(flags & SUBMODULE_MOVE_HEAD_FORCE))
1695                 argv_array_push(&cp.args, old_head ? old_head : empty_tree_oid_hex());
1696
1697         argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
1698
1699         if (run_command(&cp)) {
1700                 ret = error(_("Submodule '%s' could not be updated."), path);
1701                 goto out;
1702         }
1703
1704         if (!(flags & SUBMODULE_MOVE_HEAD_DRY_RUN)) {
1705                 if (new_head) {
1706                         child_process_init(&cp);
1707                         /* also set the HEAD accordingly */
1708                         cp.git_cmd = 1;
1709                         cp.no_stdin = 1;
1710                         cp.dir = path;
1711
1712                         prepare_submodule_repo_env(&cp.env_array);
1713                         argv_array_pushl(&cp.args, "update-ref", "HEAD",
1714                                          "--no-deref", new_head, NULL);
1715
1716                         if (run_command(&cp)) {
1717                                 ret = -1;
1718                                 goto out;
1719                         }
1720                 } else {
1721                         struct strbuf sb = STRBUF_INIT;
1722
1723                         strbuf_addf(&sb, "%s/.git", path);
1724                         unlink_or_warn(sb.buf);
1725                         strbuf_release(&sb);
1726
1727                         if (is_empty_dir(path))
1728                                 rmdir_or_warn(path);
1729                 }
1730         }
1731 out:
1732         return ret;
1733 }
1734
1735 int validate_submodule_git_dir(char *git_dir, const char *submodule_name)
1736 {
1737         size_t len = strlen(git_dir), suffix_len = strlen(submodule_name);
1738         char *p;
1739         int ret = 0;
1740
1741         if (len <= suffix_len || (p = git_dir + len - suffix_len)[-1] != '/' ||
1742             strcmp(p, submodule_name))
1743                 BUG("submodule name '%s' not a suffix of git dir '%s'",
1744                     submodule_name, git_dir);
1745
1746         /*
1747          * We prevent the contents of sibling submodules' git directories to
1748          * clash.
1749          *
1750          * Example: having a submodule named `hippo` and another one named
1751          * `hippo/hooks` would result in the git directories
1752          * `.git/modules/hippo/` and `.git/modules/hippo/hooks/`, respectively,
1753          * but the latter directory is already designated to contain the hooks
1754          * of the former.
1755          */
1756         for (; *p; p++) {
1757                 if (is_dir_sep(*p)) {
1758                         char c = *p;
1759
1760                         *p = '\0';
1761                         if (is_git_directory(git_dir))
1762                                 ret = -1;
1763                         *p = c;
1764
1765                         if (ret < 0)
1766                                 return error(_("submodule git dir '%s' is "
1767                                                "inside git dir '%.*s'"),
1768                                              git_dir,
1769                                              (int)(p - git_dir), git_dir);
1770                 }
1771         }
1772
1773         return 0;
1774 }
1775
1776 /*
1777  * Embeds a single submodules git directory into the superprojects git dir,
1778  * non recursively.
1779  */
1780 static void relocate_single_git_dir_into_superproject(const char *prefix,
1781                                                       const char *path)
1782 {
1783         char *old_git_dir = NULL, *real_old_git_dir = NULL, *real_new_git_dir = NULL;
1784         char *new_git_dir;
1785         const struct submodule *sub;
1786
1787         if (submodule_uses_worktrees(path))
1788                 die(_("relocate_gitdir for submodule '%s' with "
1789                       "more than one worktree not supported"), path);
1790
1791         old_git_dir = xstrfmt("%s/.git", path);
1792         if (read_gitfile(old_git_dir))
1793                 /* If it is an actual gitfile, it doesn't need migration. */
1794                 return;
1795
1796         real_old_git_dir = real_pathdup(old_git_dir, 1);
1797
1798         sub = submodule_from_path(the_repository, &null_oid, path);
1799         if (!sub)
1800                 die(_("could not lookup name for submodule '%s'"), path);
1801
1802         new_git_dir = git_pathdup("modules/%s", sub->name);
1803         if (validate_submodule_git_dir(new_git_dir, sub->name) < 0)
1804                 die(_("refusing to move '%s' into an existing git dir"),
1805                     real_old_git_dir);
1806         if (safe_create_leading_directories_const(new_git_dir) < 0)
1807                 die(_("could not create directory '%s'"), new_git_dir);
1808         real_new_git_dir = real_pathdup(new_git_dir, 1);
1809         free(new_git_dir);
1810
1811         fprintf(stderr, _("Migrating git directory of '%s%s' from\n'%s' to\n'%s'\n"),
1812                 get_super_prefix_or_empty(), path,
1813                 real_old_git_dir, real_new_git_dir);
1814
1815         relocate_gitdir(path, real_old_git_dir, real_new_git_dir);
1816
1817         free(old_git_dir);
1818         free(real_old_git_dir);
1819         free(real_new_git_dir);
1820 }
1821
1822 /*
1823  * Migrate the git directory of the submodule given by path from
1824  * having its git directory within the working tree to the git dir nested
1825  * in its superprojects git dir under modules/.
1826  */
1827 void absorb_git_dir_into_superproject(const char *prefix,
1828                                       const char *path,
1829                                       unsigned flags)
1830 {
1831         int err_code;
1832         const char *sub_git_dir;
1833         struct strbuf gitdir = STRBUF_INIT;
1834         strbuf_addf(&gitdir, "%s/.git", path);
1835         sub_git_dir = resolve_gitdir_gently(gitdir.buf, &err_code);
1836
1837         /* Not populated? */
1838         if (!sub_git_dir) {
1839                 const struct submodule *sub;
1840
1841                 if (err_code == READ_GITFILE_ERR_STAT_FAILED) {
1842                         /* unpopulated as expected */
1843                         strbuf_release(&gitdir);
1844                         return;
1845                 }
1846
1847                 if (err_code != READ_GITFILE_ERR_NOT_A_REPO)
1848                         /* We don't know what broke here. */
1849                         read_gitfile_error_die(err_code, path, NULL);
1850
1851                 /*
1852                 * Maybe populated, but no git directory was found?
1853                 * This can happen if the superproject is a submodule
1854                 * itself and was just absorbed. The absorption of the
1855                 * superproject did not rewrite the git file links yet,
1856                 * fix it now.
1857                 */
1858                 sub = submodule_from_path(the_repository, &null_oid, path);
1859                 if (!sub)
1860                         die(_("could not lookup name for submodule '%s'"), path);
1861                 connect_work_tree_and_git_dir(path,
1862                         git_path("modules/%s", sub->name), 0);
1863         } else {
1864                 /* Is it already absorbed into the superprojects git dir? */
1865                 char *real_sub_git_dir = real_pathdup(sub_git_dir, 1);
1866                 char *real_common_git_dir = real_pathdup(get_git_common_dir(), 1);
1867
1868                 if (!starts_with(real_sub_git_dir, real_common_git_dir))
1869                         relocate_single_git_dir_into_superproject(prefix, path);
1870
1871                 free(real_sub_git_dir);
1872                 free(real_common_git_dir);
1873         }
1874         strbuf_release(&gitdir);
1875
1876         if (flags & ABSORB_GITDIR_RECURSE_SUBMODULES) {
1877                 struct child_process cp = CHILD_PROCESS_INIT;
1878                 struct strbuf sb = STRBUF_INIT;
1879
1880                 if (flags & ~ABSORB_GITDIR_RECURSE_SUBMODULES)
1881                         BUG("we don't know how to pass the flags down?");
1882
1883                 strbuf_addstr(&sb, get_super_prefix_or_empty());
1884                 strbuf_addstr(&sb, path);
1885                 strbuf_addch(&sb, '/');
1886
1887                 cp.dir = path;
1888                 cp.git_cmd = 1;
1889                 cp.no_stdin = 1;
1890                 argv_array_pushl(&cp.args, "--super-prefix", sb.buf,
1891                                            "submodule--helper",
1892                                            "absorb-git-dirs", NULL);
1893                 prepare_submodule_repo_env(&cp.env_array);
1894                 if (run_command(&cp))
1895                         die(_("could not recurse into submodule '%s'"), path);
1896
1897                 strbuf_release(&sb);
1898         }
1899 }
1900
1901 const char *get_superproject_working_tree(void)
1902 {
1903         struct child_process cp = CHILD_PROCESS_INIT;
1904         struct strbuf sb = STRBUF_INIT;
1905         const char *one_up = real_path_if_valid("../");
1906         const char *cwd = xgetcwd();
1907         const char *ret = NULL;
1908         const char *subpath;
1909         int code;
1910         ssize_t len;
1911
1912         if (!is_inside_work_tree())
1913                 /*
1914                  * FIXME:
1915                  * We might have a superproject, but it is harder
1916                  * to determine.
1917                  */
1918                 return NULL;
1919
1920         if (!one_up)
1921                 return NULL;
1922
1923         subpath = relative_path(cwd, one_up, &sb);
1924
1925         prepare_submodule_repo_env(&cp.env_array);
1926         argv_array_pop(&cp.env_array);
1927
1928         argv_array_pushl(&cp.args, "--literal-pathspecs", "-C", "..",
1929                         "ls-files", "-z", "--stage", "--full-name", "--",
1930                         subpath, NULL);
1931         strbuf_reset(&sb);
1932
1933         cp.no_stdin = 1;
1934         cp.no_stderr = 1;
1935         cp.out = -1;
1936         cp.git_cmd = 1;
1937
1938         if (start_command(&cp))
1939                 die(_("could not start ls-files in .."));
1940
1941         len = strbuf_read(&sb, cp.out, PATH_MAX);
1942         close(cp.out);
1943
1944         if (starts_with(sb.buf, "160000")) {
1945                 int super_sub_len;
1946                 int cwd_len = strlen(cwd);
1947                 char *super_sub, *super_wt;
1948
1949                 /*
1950                  * There is a superproject having this repo as a submodule.
1951                  * The format is <mode> SP <hash> SP <stage> TAB <full name> \0,
1952                  * We're only interested in the name after the tab.
1953                  */
1954                 super_sub = strchr(sb.buf, '\t') + 1;
1955                 super_sub_len = strlen(super_sub);
1956
1957                 if (super_sub_len > cwd_len ||
1958                     strcmp(&cwd[cwd_len - super_sub_len], super_sub))
1959                         BUG("returned path string doesn't match cwd?");
1960
1961                 super_wt = xstrdup(cwd);
1962                 super_wt[cwd_len - super_sub_len] = '\0';
1963
1964                 ret = real_path(super_wt);
1965                 free(super_wt);
1966         }
1967         strbuf_release(&sb);
1968
1969         code = finish_command(&cp);
1970
1971         if (code == 128)
1972                 /* '../' is not a git repository */
1973                 return NULL;
1974         if (code == 0 && len == 0)
1975                 /* There is an unrelated git repository at '../' */
1976                 return NULL;
1977         if (code)
1978                 die(_("ls-tree returned unexpected return code %d"), code);
1979
1980         return ret;
1981 }
1982
1983 /*
1984  * Put the gitdir for a submodule (given relative to the main
1985  * repository worktree) into `buf`, or return -1 on error.
1986  */
1987 int submodule_to_gitdir(struct strbuf *buf, const char *submodule)
1988 {
1989         const struct submodule *sub;
1990         const char *git_dir;
1991         int ret = 0;
1992
1993         strbuf_reset(buf);
1994         strbuf_addstr(buf, submodule);
1995         strbuf_complete(buf, '/');
1996         strbuf_addstr(buf, ".git");
1997
1998         git_dir = read_gitfile(buf->buf);
1999         if (git_dir) {
2000                 strbuf_reset(buf);
2001                 strbuf_addstr(buf, git_dir);
2002         }
2003         if (!is_git_directory(buf->buf)) {
2004                 sub = submodule_from_path(the_repository, &null_oid, submodule);
2005                 if (!sub) {
2006                         ret = -1;
2007                         goto cleanup;
2008                 }
2009                 strbuf_reset(buf);
2010                 strbuf_git_path(buf, "%s/%s", "modules", sub->name);
2011         }
2012
2013 cleanup:
2014         return ret;
2015 }