1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 
24 #include <ctype.h>
25 #include <endian.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <sha1.h>
33 #include <string.h>
34 #include <time.h>
35 #include <unistd.h>
36 #include <zlib.h>
37 #include <errno.h>
38 #include <libgen.h>
39 #include <stdint.h>
40 #include <imsg.h>
41 #include <uuid.h>
42 
43 #include "got_error.h"
44 #include "got_reference.h"
45 #include "got_repository.h"
46 #include "got_path.h"
47 #include "got_cancel.h"
48 #include "got_worktree.h"
49 #include "got_object.h"
50 
51 #include "got_lib_delta.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_worktree.h"
59 #include "got_lib_sha1.h"
60 #include "got_lib_object_cache.h"
61 #include "got_lib_repository.h"
62 #include "got_lib_gotconfig.h"
63 
64 #ifndef nitems
65 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
66 #endif
67 
68 const char *
69 got_repo_get_path(struct got_repository *repo)
70 {
71 	return repo->path;
72 }
73 
74 const char *
75 got_repo_get_path_git_dir(struct got_repository *repo)
76 {
77 	return repo->path_git_dir;
78 }
79 
80 const char *
81 got_repo_get_gitconfig_author_name(struct got_repository *repo)
82 {
83 	return repo->gitconfig_author_name;
84 }
85 
86 const char *
87 got_repo_get_gitconfig_author_email(struct got_repository *repo)
88 {
89 	return repo->gitconfig_author_email;
90 }
91 
92 const char *
93 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
94 {
95 	return repo->global_gitconfig_author_name;
96 }
97 
98 const char *
99 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
100 {
101 	return repo->global_gitconfig_author_email;
102 }
103 
104 const char *
105 got_repo_get_gitconfig_owner(struct got_repository *repo)
106 {
107 	return repo->gitconfig_owner;
108 }
109 
110 int
111 got_repo_is_bare(struct got_repository *repo)
112 {
113 	return (strcmp(repo->path, repo->path_git_dir) == 0);
114 }
115 
116 static char *
117 get_path_git_child(struct got_repository *repo, const char *basename)
118 {
119 	char *path_child;
120 
121 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
122 	    basename) == -1)
123 		return NULL;
124 
125 	return path_child;
126 }
127 
128 char *
129 got_repo_get_path_objects(struct got_repository *repo)
130 {
131 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
132 }
133 
134 char *
135 got_repo_get_path_objects_pack(struct got_repository *repo)
136 {
137 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
138 }
139 
140 char *
141 got_repo_get_path_refs(struct got_repository *repo)
142 {
143 	return get_path_git_child(repo, GOT_REFS_DIR);
144 }
145 
146 char *
147 got_repo_get_path_packed_refs(struct got_repository *repo)
148 {
149 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
150 }
151 
152 static char *
153 get_path_head(struct got_repository *repo)
154 {
155 	return get_path_git_child(repo, GOT_HEAD_FILE);
156 }
157 
158 char *
159 got_repo_get_path_gitconfig(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_GITCONFIG);
162 }
163 
164 char *
165 got_repo_get_path_gotconfig(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
168 }
169 
170 const struct got_gotconfig *
171 got_repo_get_gotconfig(struct got_repository *repo)
172 {
173 	return repo->gotconfig;
174 }
175 
176 void
177 got_repo_get_gitconfig_remotes(int *nremotes,
178     const struct got_remote_repo **remotes, struct got_repository *repo)
179 {
180 	*nremotes = repo->ngitconfig_remotes;
181 	*remotes = repo->gitconfig_remotes;
182 }
183 
184 static int
185 is_git_repo(struct got_repository *repo)
186 {
187 	const char *path_git = got_repo_get_path_git_dir(repo);
188 	char *path_objects = got_repo_get_path_objects(repo);
189 	char *path_refs = got_repo_get_path_refs(repo);
190 	char *path_head = get_path_head(repo);
191 	int ret = 0;
192 	struct stat sb;
193 	struct got_reference *head_ref;
194 
195 	if (lstat(path_git, &sb) == -1)
196 		goto done;
197 	if (!S_ISDIR(sb.st_mode))
198 		goto done;
199 
200 	if (lstat(path_objects, &sb) == -1)
201 		goto done;
202 	if (!S_ISDIR(sb.st_mode))
203 		goto done;
204 
205 	if (lstat(path_refs, &sb) == -1)
206 		goto done;
207 	if (!S_ISDIR(sb.st_mode))
208 		goto done;
209 
210 	if (lstat(path_head, &sb) == -1)
211 		goto done;
212 	if (!S_ISREG(sb.st_mode))
213 		goto done;
214 
215 	/* Check if the HEAD reference can be opened. */
216 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
217 		goto done;
218 	got_ref_close(head_ref);
219 
220 	ret = 1;
221 done:
222 	free(path_objects);
223 	free(path_refs);
224 	free(path_head);
225 	return ret;
226 
227 }
228 
229 const struct got_error *
230 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
231     struct got_object *obj)
232 {
233 #ifndef GOT_NO_OBJ_CACHE
234 	const struct got_error *err = NULL;
235 	err = got_object_cache_add(&repo->objcache, id, obj);
236 	if (err) {
237 		if (err->code == GOT_ERR_OBJ_EXISTS ||
238 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
239 			err = NULL;
240 		return err;
241 	}
242 	obj->refcnt++;
243 #endif
244 	return NULL;
245 }
246 
247 struct got_object *
248 got_repo_get_cached_object(struct got_repository *repo,
249     struct got_object_id *id)
250 {
251 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
252 }
253 
254 const struct got_error *
255 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
256     struct got_tree_object *tree)
257 {
258 #ifndef GOT_NO_OBJ_CACHE
259 	const struct got_error *err = NULL;
260 	err = got_object_cache_add(&repo->treecache, id, tree);
261 	if (err) {
262 		if (err->code == GOT_ERR_OBJ_EXISTS ||
263 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
264 			err = NULL;
265 		return err;
266 	}
267 	tree->refcnt++;
268 #endif
269 	return NULL;
270 }
271 
272 struct got_tree_object *
273 got_repo_get_cached_tree(struct got_repository *repo,
274     struct got_object_id *id)
275 {
276 	return (struct got_tree_object *)got_object_cache_get(
277 	    &repo->treecache, id);
278 }
279 
280 const struct got_error *
281 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
282     struct got_commit_object *commit)
283 {
284 #ifndef GOT_NO_OBJ_CACHE
285 	const struct got_error *err = NULL;
286 	err = got_object_cache_add(&repo->commitcache, id, commit);
287 	if (err) {
288 		if (err->code == GOT_ERR_OBJ_EXISTS ||
289 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
290 			err = NULL;
291 		return err;
292 	}
293 	commit->refcnt++;
294 #endif
295 	return NULL;
296 }
297 
298 struct got_commit_object *
299 got_repo_get_cached_commit(struct got_repository *repo,
300     struct got_object_id *id)
301 {
302 	return (struct got_commit_object *)got_object_cache_get(
303 	    &repo->commitcache, id);
304 }
305 
306 const struct got_error *
307 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
308     struct got_tag_object *tag)
309 {
310 #ifndef GOT_NO_OBJ_CACHE
311 	const struct got_error *err = NULL;
312 	err = got_object_cache_add(&repo->tagcache, id, tag);
313 	if (err) {
314 		if (err->code == GOT_ERR_OBJ_EXISTS ||
315 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
316 			err = NULL;
317 		return err;
318 	}
319 	tag->refcnt++;
320 #endif
321 	return NULL;
322 }
323 
324 struct got_tag_object *
325 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
326 {
327 	return (struct got_tag_object *)got_object_cache_get(
328 	    &repo->tagcache, id);
329 }
330 
331 const struct got_error *
332 open_repo(struct got_repository *repo, const char *path)
333 {
334 	const struct got_error *err = NULL;
335 
336 	/* bare git repository? */
337 	repo->path_git_dir = strdup(path);
338 	if (repo->path_git_dir == NULL)
339 		return got_error_from_errno("strdup");
340 	if (is_git_repo(repo)) {
341 		repo->path = strdup(repo->path_git_dir);
342 		if (repo->path == NULL) {
343 			err = got_error_from_errno("strdup");
344 			goto done;
345 		}
346 		return NULL;
347 	}
348 
349 	/* git repository with working tree? */
350 	free(repo->path_git_dir);
351 	repo->path_git_dir = NULL;
352 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
353 		err = got_error_from_errno("asprintf");
354 		goto done;
355 	}
356 	if (is_git_repo(repo)) {
357 		repo->path = strdup(path);
358 		if (repo->path == NULL) {
359 			err = got_error_from_errno("strdup");
360 			goto done;
361 		}
362 		return NULL;
363 	}
364 
365 	err = got_error(GOT_ERR_NOT_GIT_REPO);
366 done:
367 	if (err) {
368 		free(repo->path);
369 		repo->path = NULL;
370 		free(repo->path_git_dir);
371 		repo->path_git_dir = NULL;
372 	}
373 	return err;
374 }
375 
376 static const struct got_error *
377 parse_gitconfig_file(int *gitconfig_repository_format_version,
378     char **gitconfig_author_name, char **gitconfig_author_email,
379     struct got_remote_repo **remotes, int *nremotes,
380     char **gitconfig_owner, char ***extensions, int *nextensions,
381     const char *gitconfig_path)
382 {
383 	const struct got_error *err = NULL, *child_err = NULL;
384 	int fd = -1;
385 	int imsg_fds[2] = { -1, -1 };
386 	pid_t pid;
387 	struct imsgbuf *ibuf;
388 
389 	*gitconfig_repository_format_version = 0;
390 	if (extensions)
391 		*extensions = NULL;
392 	if (nextensions)
393 		*nextensions = 0;
394 	*gitconfig_author_name = NULL;
395 	*gitconfig_author_email = NULL;
396 	if (remotes)
397 		*remotes = NULL;
398 	if (nremotes)
399 		*nremotes = 0;
400 	if (gitconfig_owner)
401 		*gitconfig_owner = NULL;
402 
403 	fd = open(gitconfig_path, O_RDONLY);
404 	if (fd == -1) {
405 		if (errno == ENOENT)
406 			return NULL;
407 		return got_error_from_errno2("open", gitconfig_path);
408 	}
409 
410 	ibuf = calloc(1, sizeof(*ibuf));
411 	if (ibuf == NULL) {
412 		err = got_error_from_errno("calloc");
413 		goto done;
414 	}
415 
416 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
417 		err = got_error_from_errno("socketpair");
418 		goto done;
419 	}
420 
421 	pid = fork();
422 	if (pid == -1) {
423 		err = got_error_from_errno("fork");
424 		goto done;
425 	} else if (pid == 0) {
426 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
427 		    gitconfig_path);
428 		/* not reached */
429 	}
430 
431 	if (close(imsg_fds[1]) == -1) {
432 		err = got_error_from_errno("close");
433 		goto done;
434 	}
435 	imsg_fds[1] = -1;
436 	imsg_init(ibuf, imsg_fds[0]);
437 
438 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
439 	if (err)
440 		goto done;
441 	fd = -1;
442 
443 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
444 	if (err)
445 		goto done;
446 
447 	err = got_privsep_recv_gitconfig_int(
448 	    gitconfig_repository_format_version, ibuf);
449 	if (err)
450 		goto done;
451 
452 	if (extensions && nextensions) {
453 		err = got_privsep_send_gitconfig_repository_extensions_req(
454 		    ibuf);
455 		if (err)
456 			goto done;
457 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
458 		if (err)
459 			goto done;
460 		if (*nextensions > 0) {
461 			int i;
462 			*extensions = calloc(*nextensions, sizeof(char *));
463 			if (*extensions == NULL) {
464 				err = got_error_from_errno("calloc");
465 				goto done;
466 			}
467 			for (i = 0; i < *nextensions; i++) {
468 				char *ext;
469 				err = got_privsep_recv_gitconfig_str(&ext,
470 				    ibuf);
471 				if (err)
472 					goto done;
473 				(*extensions)[i] = ext;
474 			}
475 		}
476 	}
477 
478 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
479 	if (err)
480 		goto done;
481 
482 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
483 	if (err)
484 		goto done;
485 
486 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
487 	if (err)
488 		goto done;
489 
490 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
491 	if (err)
492 		goto done;
493 
494 	if (remotes && nremotes) {
495 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
496 		if (err)
497 			goto done;
498 
499 		err = got_privsep_recv_gitconfig_remotes(remotes,
500 		    nremotes, ibuf);
501 		if (err)
502 			goto done;
503 	}
504 
505 	if (gitconfig_owner) {
506 		err = got_privsep_send_gitconfig_owner_req(ibuf);
507 		if (err)
508 			goto done;
509 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
510 		if (err)
511 			goto done;
512 	}
513 
514 	imsg_clear(ibuf);
515 	err = got_privsep_send_stop(imsg_fds[0]);
516 	child_err = got_privsep_wait_for_child(pid);
517 	if (child_err && err == NULL)
518 		err = child_err;
519 done:
520 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
521 		err = got_error_from_errno("close");
522 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
523 		err = got_error_from_errno("close");
524 	if (fd != -1 && close(fd) == -1 && err == NULL)
525 		err = got_error_from_errno2("close", gitconfig_path);
526 	free(ibuf);
527 	return err;
528 }
529 
530 static const struct got_error *
531 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
532 {
533 	const struct got_error *err = NULL;
534 	char *repo_gitconfig_path = NULL;
535 
536 	if (global_gitconfig_path) {
537 		/* Read settings from ~/.gitconfig. */
538 		int dummy_repo_version;
539 		err = parse_gitconfig_file(&dummy_repo_version,
540 		    &repo->global_gitconfig_author_name,
541 		    &repo->global_gitconfig_author_email,
542 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
543 		if (err)
544 			return err;
545 	}
546 
547 	/* Read repository's .git/config file. */
548 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
549 	if (repo_gitconfig_path == NULL)
550 		return got_error_from_errno("got_repo_get_path_gitconfig");
551 
552 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
553 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
554 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
555 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
556 	    repo_gitconfig_path);
557 	if (err)
558 		goto done;
559 done:
560 	free(repo_gitconfig_path);
561 	return err;
562 }
563 
564 static const struct got_error *
565 read_gotconfig(struct got_repository *repo)
566 {
567 	const struct got_error *err = NULL;
568 	char *gotconfig_path;
569 
570 	gotconfig_path = got_repo_get_path_gotconfig(repo);
571 	if (gotconfig_path == NULL)
572 		return got_error_from_errno("got_repo_get_path_gotconfig");
573 
574 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
575 	free(gotconfig_path);
576 	return err;
577 }
578 
579 /* Supported repository format extensions. */
580 static const char *repo_extensions[] = {
581 	"noop",			/* Got supports repository format version 1. */
582 	"preciousObjects",	/* Got has no garbage collection yet. */
583 	"worktreeConfig",	/* Got does not care about Git work trees. */
584 };
585 
586 const struct got_error *
587 got_repo_open(struct got_repository **repop, const char *path,
588     const char *global_gitconfig_path)
589 {
590 	struct got_repository *repo = NULL;
591 	const struct got_error *err = NULL;
592 	char *repo_path = NULL;
593 	size_t i;
594 
595 	*repop = NULL;
596 
597 	repo = calloc(1, sizeof(*repo));
598 	if (repo == NULL) {
599 		err = got_error_from_errno("calloc");
600 		goto done;
601 	}
602 
603 	for (i = 0; i < nitems(repo->privsep_children); i++) {
604 		memset(&repo->privsep_children[i], 0,
605 		    sizeof(repo->privsep_children[0]));
606 		repo->privsep_children[i].imsg_fd = -1;
607 	}
608 
609 	err = got_object_cache_init(&repo->objcache,
610 	    GOT_OBJECT_CACHE_TYPE_OBJ);
611 	if (err)
612 		goto done;
613 	err = got_object_cache_init(&repo->treecache,
614 	    GOT_OBJECT_CACHE_TYPE_TREE);
615 	if (err)
616 		goto done;
617 	err = got_object_cache_init(&repo->commitcache,
618 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
619 	if (err)
620 		goto done;
621 	err = got_object_cache_init(&repo->tagcache,
622 	    GOT_OBJECT_CACHE_TYPE_TAG);
623 	if (err)
624 		goto done;
625 
626 	repo_path = realpath(path, NULL);
627 	if (repo_path == NULL) {
628 		err = got_error_from_errno2("realpath", path);
629 		goto done;
630 	}
631 
632 	for (;;) {
633 		char *parent_path;
634 
635 		err = open_repo(repo, repo_path);
636 		if (err == NULL)
637 			break;
638 		if (err->code != GOT_ERR_NOT_GIT_REPO)
639 			goto done;
640 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
641 			err = got_error(GOT_ERR_NOT_GIT_REPO);
642 			goto done;
643 		}
644 		err = got_path_dirname(&parent_path, repo_path);
645 		if (err)
646 			goto done;
647 		free(repo_path);
648 		repo_path = parent_path;
649 	}
650 
651 	err = read_gotconfig(repo);
652 	if (err)
653 		goto done;
654 
655 	err = read_gitconfig(repo, global_gitconfig_path);
656 	if (err)
657 		goto done;
658 	if (repo->gitconfig_repository_format_version != 0)
659 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
660 	for (i = 0; i < repo->nextensions; i++) {
661 		char *ext = repo->extensions[i];
662 		int j, supported = 0;
663 		for (j = 0; j < nitems(repo_extensions); j++) {
664 			if (strcmp(ext, repo_extensions[j]) == 0) {
665 				supported = 1;
666 				break;
667 			}
668 		}
669 		if (!supported) {
670 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
671 			goto done;
672 		}
673 	}
674 done:
675 	if (err)
676 		got_repo_close(repo);
677 	else
678 		*repop = repo;
679 	free(repo_path);
680 	return err;
681 }
682 
683 const struct got_error *
684 got_repo_close(struct got_repository *repo)
685 {
686 	const struct got_error *err = NULL, *child_err;
687 	size_t i;
688 
689 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
690 		if (repo->packidx_cache[i] == NULL)
691 			break;
692 		got_packidx_close(repo->packidx_cache[i]);
693 	}
694 
695 	for (i = 0; i < nitems(repo->packs); i++) {
696 		if (repo->packs[i].path_packfile == NULL)
697 			break;
698 		got_pack_close(&repo->packs[i]);
699 	}
700 
701 	free(repo->path);
702 	free(repo->path_git_dir);
703 
704 	got_object_cache_close(&repo->objcache);
705 	got_object_cache_close(&repo->treecache);
706 	got_object_cache_close(&repo->commitcache);
707 	got_object_cache_close(&repo->tagcache);
708 
709 	for (i = 0; i < nitems(repo->privsep_children); i++) {
710 		if (repo->privsep_children[i].imsg_fd == -1)
711 			continue;
712 		imsg_clear(repo->privsep_children[i].ibuf);
713 		free(repo->privsep_children[i].ibuf);
714 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
715 		child_err = got_privsep_wait_for_child(
716 		    repo->privsep_children[i].pid);
717 		if (child_err && err == NULL)
718 			err = child_err;
719 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
720 		    err == NULL)
721 			err = got_error_from_errno("close");
722 	}
723 
724 	if (repo->gotconfig)
725 		got_gotconfig_free(repo->gotconfig);
726 	free(repo->gitconfig_author_name);
727 	free(repo->gitconfig_author_email);
728 	for (i = 0; i < repo->ngitconfig_remotes; i++)
729 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
730 	free(repo->gitconfig_remotes);
731 	for (i = 0; i < repo->nextensions; i++)
732 		free(repo->extensions[i]);
733 	free(repo->extensions);
734 	free(repo);
735 
736 	return err;
737 }
738 
739 void
740 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
741 {
742 	int i;
743 
744 	free(repo->name);
745 	repo->name = NULL;
746 	free(repo->url);
747 	repo->url = NULL;
748 	for (i = 0; i < repo->nbranches; i++)
749 		free(repo->branches[i]);
750 	free(repo->branches);
751 	repo->branches = NULL;
752 	repo->nbranches = 0;
753 }
754 
755 const struct got_error *
756 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
757     const char *input_path)
758 {
759 	const struct got_error *err = NULL;
760 	const char *repo_abspath = NULL;
761 	size_t repolen, len;
762 	char *canonpath, *path = NULL;
763 
764 	*in_repo_path = NULL;
765 
766 	canonpath = strdup(input_path);
767 	if (canonpath == NULL) {
768 		err = got_error_from_errno("strdup");
769 		goto done;
770 	}
771 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
772 	if (err)
773 		goto done;
774 
775 	repo_abspath = got_repo_get_path(repo);
776 
777 	if (canonpath[0] == '\0') {
778 		path = strdup(canonpath);
779 		if (path == NULL) {
780 			err = got_error_from_errno("strdup");
781 			goto done;
782 		}
783 	} else {
784 		path = realpath(canonpath, NULL);
785 		if (path == NULL) {
786 			if (errno != ENOENT) {
787 				err = got_error_from_errno2("realpath",
788 				    canonpath);
789 				goto done;
790 			}
791 			/*
792 			 * Path is not on disk.
793 			 * Assume it is already relative to repository root.
794 			 */
795 			path = strdup(canonpath);
796 			if (path == NULL) {
797 				err = got_error_from_errno("strdup");
798 				goto done;
799 			}
800 		}
801 
802 		repolen = strlen(repo_abspath);
803 		len = strlen(path);
804 
805 
806 		if (strcmp(path, repo_abspath) == 0) {
807 			free(path);
808 			path = strdup("");
809 			if (path == NULL) {
810 				err = got_error_from_errno("strdup");
811 				goto done;
812 			}
813 		} else if (len > repolen &&
814 		    got_path_is_child(path, repo_abspath, repolen)) {
815 			/* Matched an on-disk path inside repository. */
816 			if (got_repo_is_bare(repo)) {
817 				/*
818 				 * Matched an on-disk path inside repository
819 				 * database. Treat input as repository-relative.
820 				 */
821 				 free(path);
822 				 path = canonpath;
823 				 canonpath = NULL;
824 			} else {
825 				char *child;
826 				/* Strip common prefix with repository path. */
827 				err = got_path_skip_common_ancestor(&child,
828 				    repo_abspath, path);
829 				if (err)
830 					goto done;
831 				free(path);
832 				path = child;
833 			}
834 		} else {
835 			/*
836 			 * Matched unrelated on-disk path.
837 			 * Treat input as repository-relative.
838 			 */
839 			 free(path);
840 			 path = canonpath;
841 			 canonpath = NULL;
842 		}
843 	}
844 
845 	/* Make in-repository path absolute */
846 	if (path[0] != '/') {
847 		char *abspath;
848 		if (asprintf(&abspath, "/%s", path) == -1) {
849 			err = got_error_from_errno("asprintf");
850 			goto done;
851 		}
852 		free(path);
853 		path = abspath;
854 	}
855 
856 done:
857 	free(canonpath);
858 	if (err)
859 		free(path);
860 	else
861 		*in_repo_path = path;
862 	return err;
863 }
864 
865 static const struct got_error *
866 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
867     const char *path_packidx)
868 {
869 	const struct got_error *err = NULL;
870 	size_t i;
871 
872 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
873 		if (repo->packidx_cache[i] == NULL)
874 			break;
875 		if (strcmp(repo->packidx_cache[i]->path_packidx,
876 		    path_packidx) == 0) {
877 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
878 		}
879 	}
880 	if (i == nitems(repo->packidx_cache)) {
881 		err = got_packidx_close(repo->packidx_cache[i - 1]);
882 		if (err)
883 			return err;
884 	}
885 
886 	/*
887 	 * Insert the new pack index at the front so it will
888 	 * be searched first in the future.
889 	 */
890 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
891 	    sizeof(repo->packidx_cache) -
892 	    sizeof(repo->packidx_cache[0]));
893 	repo->packidx_cache[0] = packidx;
894 
895 	return NULL;
896 }
897 
898 static int
899 is_packidx_filename(const char *name, size_t len)
900 {
901 	if (len != GOT_PACKIDX_NAMELEN)
902 		return 0;
903 
904 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
905 		return 0;
906 
907 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
908 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
909 		return 0;
910 
911 	return 1;
912 }
913 
914 const struct got_error *
915 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
916     struct got_repository *repo, struct got_object_id *id)
917 {
918 	const struct got_error *err;
919 	char *path_packdir;
920 	DIR *packdir;
921 	struct dirent *dent;
922 	char *path_packidx;
923 	size_t i;
924 
925 	/* Search pack index cache. */
926 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
927 		if (repo->packidx_cache[i] == NULL)
928 			break;
929 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
930 		if (*idx != -1) {
931 			*packidx = repo->packidx_cache[i];
932 			/*
933 			 * Move this cache entry to the front. Repeatedly
934 			 * searching a wrong pack index can be expensive.
935 			 */
936 			if (i > 0) {
937 				struct got_packidx *p;
938 				p = repo->packidx_cache[0];
939 				repo->packidx_cache[0] = *packidx;
940 				repo->packidx_cache[i] = p;
941 			}
942 			return NULL;
943 		}
944 	}
945 	/* No luck. Search the filesystem. */
946 
947 	path_packdir = got_repo_get_path_objects_pack(repo);
948 	if (path_packdir == NULL)
949 		return got_error_from_errno("got_repo_get_path_objects_pack");
950 
951 	packdir = opendir(path_packdir);
952 	if (packdir == NULL) {
953 		if (errno == ENOENT)
954 			err = got_error_no_obj(id);
955 		else
956 			err = got_error_from_errno2("opendir", path_packdir);
957 		goto done;
958 	}
959 
960 	while ((dent = readdir(packdir)) != NULL) {
961 		int is_cached = 0;
962 
963 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
964 			continue;
965 
966 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
967 		    dent->d_name) == -1) {
968 			err = got_error_from_errno("asprintf");
969 			goto done;
970 		}
971 
972 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
973 			if (repo->packidx_cache[i] == NULL)
974 				break;
975 			if (strcmp(repo->packidx_cache[i]->path_packidx,
976 			    path_packidx) == 0) {
977 				is_cached = 1;
978 				break;
979 			}
980 		}
981 		if (is_cached) {
982 			free(path_packidx);
983 			continue; /* already searched */
984 		}
985 
986 		err = got_packidx_open(packidx, path_packidx, 0);
987 		if (err) {
988 			free(path_packidx);
989 			goto done;
990 		}
991 
992 		err = cache_packidx(repo, *packidx, path_packidx);
993 		free(path_packidx);
994 		if (err)
995 			goto done;
996 
997 		*idx = got_packidx_get_object_idx(*packidx, id);
998 		if (*idx != -1) {
999 			err = NULL; /* found the object */
1000 			goto done;
1001 		}
1002 	}
1003 
1004 	err = got_error_no_obj(id);
1005 done:
1006 	free(path_packdir);
1007 	if (packdir && closedir(packdir) != 0 && err == NULL)
1008 		err = got_error_from_errno("closedir");
1009 	return err;
1010 }
1011 
1012 static const struct got_error *
1013 read_packfile_hdr(int fd, struct got_packidx *packidx)
1014 {
1015 	const struct got_error *err = NULL;
1016 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1017 	struct got_packfile_hdr hdr;
1018 	ssize_t n;
1019 
1020 	n = read(fd, &hdr, sizeof(hdr));
1021 	if (n < 0)
1022 		return got_error_from_errno("read");
1023 	if (n != sizeof(hdr))
1024 		return got_error(GOT_ERR_BAD_PACKFILE);
1025 
1026 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1027 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1028 	    be32toh(hdr.nobjects) != totobj)
1029 		err = got_error(GOT_ERR_BAD_PACKFILE);
1030 
1031 	return err;
1032 }
1033 
1034 static const struct got_error *
1035 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
1036 {
1037 	const struct got_error *err = NULL;
1038 
1039 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
1040 	if (*fd == -1)
1041 		return got_error_from_errno2("open", path_packfile);
1042 
1043 	if (packidx) {
1044 		err = read_packfile_hdr(*fd, packidx);
1045 		if (err) {
1046 			close(*fd);
1047 			*fd = -1;
1048 		}
1049 	}
1050 
1051 	return err;
1052 }
1053 
1054 const struct got_error *
1055 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1056     const char *path_packfile, struct got_packidx *packidx)
1057 {
1058 	const struct got_error *err = NULL;
1059 	struct got_pack *pack = NULL;
1060 	struct stat sb;
1061 	size_t i;
1062 
1063 	if (packp)
1064 		*packp = NULL;
1065 
1066 	for (i = 0; i < nitems(repo->packs); i++) {
1067 		pack = &repo->packs[i];
1068 		if (pack->path_packfile == NULL)
1069 			break;
1070 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1071 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1072 	}
1073 
1074 	if (i == nitems(repo->packs) - 1) {
1075 		err = got_pack_close(&repo->packs[i - 1]);
1076 		if (err)
1077 			return err;
1078 		memmove(&repo->packs[1], &repo->packs[0],
1079 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1080 		i = 0;
1081 	}
1082 
1083 	pack = &repo->packs[i];
1084 
1085 	pack->path_packfile = strdup(path_packfile);
1086 	if (pack->path_packfile == NULL) {
1087 		err = got_error_from_errno("strdup");
1088 		goto done;
1089 	}
1090 
1091 	err = open_packfile(&pack->fd, path_packfile, packidx);
1092 	if (err)
1093 		goto done;
1094 
1095 	if (fstat(pack->fd, &sb) != 0) {
1096 		err = got_error_from_errno("fstat");
1097 		goto done;
1098 	}
1099 	pack->filesize = sb.st_size;
1100 
1101 	pack->privsep_child = NULL;
1102 
1103 #ifndef GOT_PACK_NO_MMAP
1104 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1105 	    pack->fd, 0);
1106 	if (pack->map == MAP_FAILED) {
1107 		if (errno != ENOMEM) {
1108 			err = got_error_from_errno("mmap");
1109 			goto done;
1110 		}
1111 		pack->map = NULL; /* fall back to read(2) */
1112 	}
1113 #endif
1114 done:
1115 	if (err) {
1116 		if (pack) {
1117 			free(pack->path_packfile);
1118 			memset(pack, 0, sizeof(*pack));
1119 		}
1120 	} else if (packp)
1121 		*packp = pack;
1122 	return err;
1123 }
1124 
1125 struct got_pack *
1126 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1127 {
1128 	struct got_pack *pack = NULL;
1129 	size_t i;
1130 
1131 	for (i = 0; i < nitems(repo->packs); i++) {
1132 		pack = &repo->packs[i];
1133 		if (pack->path_packfile == NULL)
1134 			break;
1135 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1136 			return pack;
1137 	}
1138 
1139 	return NULL;
1140 }
1141 
1142 const struct got_error *
1143 got_repo_init(const char *repo_path)
1144 {
1145 	const struct got_error *err = NULL;
1146 	const char *dirnames[] = {
1147 		GOT_OBJECTS_DIR,
1148 		GOT_OBJECTS_PACK_DIR,
1149 		GOT_REFS_DIR,
1150 	};
1151 	const char *description_str = "Unnamed repository; "
1152 	    "edit this file 'description' to name the repository.";
1153 	const char *headref_str = "ref: refs/heads/main";
1154 	const char *gitconfig_str = "[core]\n"
1155 	    "\trepositoryformatversion = 0\n"
1156 	    "\tfilemode = true\n"
1157 	    "\tbare = true\n";
1158 	char *path;
1159 	size_t i;
1160 
1161 	if (!got_path_dir_is_empty(repo_path))
1162 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1163 
1164 	for (i = 0; i < nitems(dirnames); i++) {
1165 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1166 			return got_error_from_errno("asprintf");
1167 		}
1168 		err = got_path_mkdir(path);
1169 		free(path);
1170 		if (err)
1171 			return err;
1172 	}
1173 
1174 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1175 		return got_error_from_errno("asprintf");
1176 	err = got_path_create_file(path, description_str);
1177 	free(path);
1178 	if (err)
1179 		return err;
1180 
1181 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1182 		return got_error_from_errno("asprintf");
1183 	err = got_path_create_file(path, headref_str);
1184 	free(path);
1185 	if (err)
1186 		return err;
1187 
1188 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1189 		return got_error_from_errno("asprintf");
1190 	err = got_path_create_file(path, gitconfig_str);
1191 	free(path);
1192 	if (err)
1193 		return err;
1194 
1195 	return NULL;
1196 }
1197 
1198 static const struct got_error *
1199 match_packed_object(struct got_object_id **unique_id,
1200     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1201 {
1202 	const struct got_error *err = NULL;
1203 	char *path_packdir;
1204 	DIR *packdir;
1205 	struct dirent *dent;
1206 	char *path_packidx;
1207 	struct got_object_id_queue matched_ids;
1208 
1209 	SIMPLEQ_INIT(&matched_ids);
1210 
1211 	path_packdir = got_repo_get_path_objects_pack(repo);
1212 	if (path_packdir == NULL)
1213 		return got_error_from_errno("got_repo_get_path_objects_pack");
1214 
1215 	packdir = opendir(path_packdir);
1216 	if (packdir == NULL) {
1217 		if (errno != ENOENT)
1218 			err = got_error_from_errno2("opendir", path_packdir);
1219 		goto done;
1220 	}
1221 
1222 	while ((dent = readdir(packdir)) != NULL) {
1223 		struct got_packidx *packidx;
1224 		struct got_object_qid *qid;
1225 
1226 
1227 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1228 			continue;
1229 
1230 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1231 		    dent->d_name) == -1) {
1232 			err = got_error_from_errno("asprintf");
1233 			break;
1234 		}
1235 
1236 		err = got_packidx_open(&packidx, path_packidx, 0);
1237 		free(path_packidx);
1238 		if (err)
1239 			break;
1240 
1241 		err = got_packidx_match_id_str_prefix(&matched_ids,
1242 		    packidx, id_str_prefix);
1243 		if (err) {
1244 			got_packidx_close(packidx);
1245 			break;
1246 		}
1247 		err = got_packidx_close(packidx);
1248 		if (err)
1249 			break;
1250 
1251 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1252 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1253 				int matched_type;
1254 				err = got_object_get_type(&matched_type, repo,
1255 				    qid->id);
1256 				if (err)
1257 					goto done;
1258 				if (matched_type != obj_type)
1259 					continue;
1260 			}
1261 			if (*unique_id == NULL) {
1262 				*unique_id = got_object_id_dup(qid->id);
1263 				if (*unique_id == NULL) {
1264 					err = got_error_from_errno("malloc");
1265 					goto done;
1266 				}
1267 			} else {
1268 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1269 					continue; /* packed multiple times */
1270 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1271 				goto done;
1272 			}
1273 		}
1274 	}
1275 done:
1276 	got_object_id_queue_free(&matched_ids);
1277 	free(path_packdir);
1278 	if (packdir && closedir(packdir) != 0 && err == NULL)
1279 		err = got_error_from_errno("closedir");
1280 	if (err) {
1281 		free(*unique_id);
1282 		*unique_id = NULL;
1283 	}
1284 	return err;
1285 }
1286 
1287 static const struct got_error *
1288 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1289     const char *object_dir, const char *id_str_prefix, int obj_type,
1290     struct got_repository *repo)
1291 {
1292 	const struct got_error *err = NULL;
1293 	char *path;
1294 	DIR *dir = NULL;
1295 	struct dirent *dent;
1296 	struct got_object_id id;
1297 
1298 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1299 		err = got_error_from_errno("asprintf");
1300 		goto done;
1301 	}
1302 
1303 	dir = opendir(path);
1304 	if (dir == NULL) {
1305 		if (errno == ENOENT) {
1306 			err = NULL;
1307 			goto done;
1308 		}
1309 		err = got_error_from_errno2("opendir", path);
1310 		goto done;
1311 	}
1312 	while ((dent = readdir(dir)) != NULL) {
1313 		char *id_str;
1314 		int cmp;
1315 
1316 		if (strcmp(dent->d_name, ".") == 0 ||
1317 		    strcmp(dent->d_name, "..") == 0)
1318 			continue;
1319 
1320 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1321 			err = got_error_from_errno("asprintf");
1322 			goto done;
1323 		}
1324 
1325 		if (!got_parse_sha1_digest(id.sha1, id_str))
1326 			continue;
1327 
1328 		/*
1329 		 * Directory entries do not necessarily appear in
1330 		 * sorted order, so we must iterate over all of them.
1331 		 */
1332 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1333 		if (cmp != 0) {
1334 			free(id_str);
1335 			continue;
1336 		}
1337 
1338 		if (*unique_id == NULL) {
1339 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1340 				int matched_type;
1341 				err = got_object_get_type(&matched_type, repo,
1342 				    &id);
1343 				if (err)
1344 					goto done;
1345 				if (matched_type != obj_type)
1346 					continue;
1347 			}
1348 			*unique_id = got_object_id_dup(&id);
1349 			if (*unique_id == NULL) {
1350 				err = got_error_from_errno("got_object_id_dup");
1351 				free(id_str);
1352 				goto done;
1353 			}
1354 		} else {
1355 			if (got_object_id_cmp(*unique_id, &id) == 0)
1356 				continue; /* both packed and loose */
1357 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1358 			free(id_str);
1359 			goto done;
1360 		}
1361 	}
1362 done:
1363 	if (dir && closedir(dir) != 0 && err == NULL)
1364 		err = got_error_from_errno("closedir");
1365 	if (err) {
1366 		free(*unique_id);
1367 		*unique_id = NULL;
1368 	}
1369 	free(path);
1370 	return err;
1371 }
1372 
1373 const struct got_error *
1374 got_repo_match_object_id_prefix(struct got_object_id **id,
1375     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1376 {
1377 	const struct got_error *err = NULL;
1378 	char *path_objects = got_repo_get_path_objects(repo);
1379 	char *object_dir = NULL;
1380 	size_t len;
1381 	int i;
1382 
1383 	*id = NULL;
1384 
1385 	for (i = 0; i < strlen(id_str_prefix); i++) {
1386 		if (isxdigit((unsigned char)id_str_prefix[i]))
1387 			continue;
1388 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1389 	}
1390 
1391 	len = strlen(id_str_prefix);
1392 	if (len >= 2) {
1393 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1394 		if (err)
1395 			goto done;
1396 		object_dir = strndup(id_str_prefix, 2);
1397 		if (object_dir == NULL) {
1398 			err = got_error_from_errno("strdup");
1399 			goto done;
1400 		}
1401 		err = match_loose_object(id, path_objects, object_dir,
1402 		    id_str_prefix, obj_type, repo);
1403 	} else if (len == 1) {
1404 		int i;
1405 		for (i = 0; i < 0xf; i++) {
1406 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1407 			    == -1) {
1408 				err = got_error_from_errno("asprintf");
1409 				goto done;
1410 			}
1411 			err = match_packed_object(id, repo, object_dir,
1412 			    obj_type);
1413 			if (err)
1414 				goto done;
1415 			err = match_loose_object(id, path_objects, object_dir,
1416 			    id_str_prefix, obj_type, repo);
1417 			if (err)
1418 				goto done;
1419 		}
1420 	} else {
1421 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1422 		goto done;
1423 	}
1424 done:
1425 	free(object_dir);
1426 	if (err) {
1427 		free(*id);
1428 		*id = NULL;
1429 	} else if (*id == NULL)
1430 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1431 
1432 	return err;
1433 }
1434 
1435 const struct got_error *
1436 got_repo_match_object_id(struct got_object_id **id, char **label,
1437     const char *id_str, int obj_type, int resolve_tags,
1438     struct got_repository *repo)
1439 {
1440 	const struct got_error *err;
1441 	struct got_tag_object *tag;
1442 	struct got_reference *ref = NULL;
1443 
1444 	*id = NULL;
1445 	if (label)
1446 		*label = NULL;
1447 
1448 	if (resolve_tags) {
1449 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1450 		    repo);
1451 		if (err == NULL) {
1452 			*id = got_object_id_dup(
1453 			    got_object_tag_get_object_id(tag));
1454 			if (*id == NULL)
1455 				err = got_error_from_errno("got_object_id_dup");
1456 			else if (label && asprintf(label, "refs/tags/%s",
1457 			    got_object_tag_get_name(tag)) == -1) {
1458 				err = got_error_from_errno("asprintf");
1459 				free(*id);
1460 				*id = NULL;
1461 			}
1462 			got_object_tag_close(tag);
1463 			return err;
1464 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1465 		    err->code != GOT_ERR_NO_OBJ)
1466 			return err;
1467 	}
1468 
1469 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1470 	if (err) {
1471 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1472 			return err;
1473 		err = got_ref_open(&ref, repo, id_str, 0);
1474 		if (err != NULL)
1475 			goto done;
1476 		if (label) {
1477 			*label = strdup(got_ref_get_name(ref));
1478 			if (*label == NULL) {
1479 				err = got_error_from_errno("strdup");
1480 				goto done;
1481 			}
1482 		}
1483 		err = got_ref_resolve(id, repo, ref);
1484 	} else if (label) {
1485 		err = got_object_id_str(label, *id);
1486 		if (*label == NULL) {
1487 			err = got_error_from_errno("strdup");
1488 			goto done;
1489 		}
1490 	}
1491 done:
1492 	if (ref)
1493 		got_ref_close(ref);
1494 	return err;
1495 }
1496 
1497 const struct got_error *
1498 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1499     int obj_type, struct got_repository *repo)
1500 {
1501 	const struct got_error *err;
1502 	struct got_reflist_head refs;
1503 	struct got_reflist_entry *re;
1504 	struct got_object_id *tag_id;
1505 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1506 
1507 	SIMPLEQ_INIT(&refs);
1508 	*tag = NULL;
1509 
1510 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1511 	if (err)
1512 		return err;
1513 
1514 	SIMPLEQ_FOREACH(re, &refs, entry) {
1515 		const char *refname;
1516 		refname = got_ref_get_name(re->ref);
1517 		if (got_ref_is_symbolic(re->ref))
1518 			continue;
1519 		if (!name_is_absolute)
1520 			refname += strlen("refs/tags/");
1521 		if (strcmp(refname, name) != 0)
1522 			continue;
1523 		err = got_ref_resolve(&tag_id, repo, re->ref);
1524 		if (err)
1525 			break;
1526 		err = got_object_open_as_tag(tag, repo, tag_id);
1527 		free(tag_id);
1528 		if (err)
1529 			break;
1530 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1531 		    got_object_tag_get_object_type(*tag) == obj_type)
1532 			break;
1533 		got_object_tag_close(*tag);
1534 		*tag = NULL;
1535 	}
1536 
1537 	got_ref_list_free(&refs);
1538 	if (err == NULL && *tag == NULL)
1539 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1540 	return err;
1541 }
1542 
1543 static const struct got_error *
1544 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1545     const char *name, mode_t mode, struct got_object_id *blob_id)
1546 {
1547 	const struct got_error *err = NULL;
1548 
1549 	 *new_te = NULL;
1550 
1551 	*new_te = calloc(1, sizeof(**new_te));
1552 	if (*new_te == NULL)
1553 		return got_error_from_errno("calloc");
1554 
1555 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1556 	    sizeof((*new_te)->name)) {
1557 		err = got_error(GOT_ERR_NO_SPACE);
1558 		goto done;
1559 	}
1560 
1561 	if (S_ISLNK(mode)) {
1562 		(*new_te)->mode = S_IFLNK;
1563 	} else {
1564 		(*new_te)->mode = S_IFREG;
1565 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1566 	}
1567 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1568 done:
1569 	if (err && *new_te) {
1570 		free(*new_te);
1571 		*new_te = NULL;
1572 	}
1573 	return err;
1574 }
1575 
1576 static const struct got_error *
1577 import_file(struct got_tree_entry **new_te, struct dirent *de,
1578     const char *path, struct got_repository *repo)
1579 {
1580 	const struct got_error *err;
1581 	struct got_object_id *blob_id = NULL;
1582 	char *filepath;
1583 	struct stat sb;
1584 
1585 	if (asprintf(&filepath, "%s%s%s", path,
1586 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1587 		return got_error_from_errno("asprintf");
1588 
1589 	if (lstat(filepath, &sb) != 0) {
1590 		err = got_error_from_errno2("lstat", path);
1591 		goto done;
1592 	}
1593 
1594 	err = got_object_blob_create(&blob_id, filepath, repo);
1595 	if (err)
1596 		goto done;
1597 
1598 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1599 	    blob_id);
1600 done:
1601 	free(filepath);
1602 	if (err)
1603 		free(blob_id);
1604 	return err;
1605 }
1606 
1607 static const struct got_error *
1608 insert_tree_entry(struct got_tree_entry *new_te,
1609     struct got_pathlist_head *paths)
1610 {
1611 	const struct got_error *err = NULL;
1612 	struct got_pathlist_entry *new_pe;
1613 
1614 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1615 	if (err)
1616 		return err;
1617 	if (new_pe == NULL)
1618 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1619 	return NULL;
1620 }
1621 
1622 static const struct got_error *write_tree(struct got_object_id **,
1623     const char *, struct got_pathlist_head *, struct got_repository *,
1624     got_repo_import_cb progress_cb, void *progress_arg);
1625 
1626 static const struct got_error *
1627 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1628     const char *path, struct got_pathlist_head *ignores,
1629     struct got_repository *repo,
1630     got_repo_import_cb progress_cb, void *progress_arg)
1631 {
1632 	const struct got_error *err;
1633 	struct got_object_id *id = NULL;
1634 	char *subdirpath;
1635 
1636 	if (asprintf(&subdirpath, "%s%s%s", path,
1637 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1638 		return got_error_from_errno("asprintf");
1639 
1640 	(*new_te) = calloc(1, sizeof(**new_te));
1641 	if (*new_te == NULL)
1642 		return got_error_from_errno("calloc");
1643 	(*new_te)->mode = S_IFDIR;
1644 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1645 	    sizeof((*new_te)->name)) {
1646 		err = got_error(GOT_ERR_NO_SPACE);
1647 		goto done;
1648 	}
1649 	err = write_tree(&id, subdirpath, ignores,  repo,
1650 	    progress_cb, progress_arg);
1651 	if (err)
1652 		goto done;
1653 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1654 	
1655 done:
1656 	free(id);
1657 	free(subdirpath);
1658 	if (err) {
1659 		free(*new_te);
1660 		*new_te = NULL;
1661 	}
1662 	return err;
1663 }
1664 
1665 static const struct got_error *
1666 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1667     struct got_pathlist_head *ignores, struct got_repository *repo,
1668     got_repo_import_cb progress_cb, void *progress_arg)
1669 {
1670 	const struct got_error *err = NULL;
1671 	DIR *dir;
1672 	struct dirent *de;
1673 	int nentries;
1674 	struct got_tree_entry *new_te = NULL;
1675 	struct got_pathlist_head paths;
1676 	struct got_pathlist_entry *pe;
1677 
1678 	*new_tree_id = NULL;
1679 
1680 	TAILQ_INIT(&paths);
1681 
1682 	dir = opendir(path_dir);
1683 	if (dir == NULL) {
1684 		err = got_error_from_errno2("opendir", path_dir);
1685 		goto done;
1686 	}
1687 
1688 	nentries = 0;
1689 	while ((de = readdir(dir)) != NULL) {
1690 		int ignore = 0;
1691 		int type;
1692 
1693 		if (strcmp(de->d_name, ".") == 0 ||
1694 		    strcmp(de->d_name, "..") == 0)
1695 			continue;
1696 
1697 		TAILQ_FOREACH(pe, ignores, entry) {
1698 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1699 				ignore = 1;
1700 				break;
1701 			}
1702 		}
1703 		if (ignore)
1704 			continue;
1705 
1706 		err = got_path_dirent_type(&type, path_dir, de);
1707 		if (err)
1708 			goto done;
1709 
1710 		if (type == DT_DIR) {
1711 			err = import_subdir(&new_te, de, path_dir,
1712 			    ignores, repo, progress_cb, progress_arg);
1713 			if (err) {
1714 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1715 					goto done;
1716 				err = NULL;
1717 				continue;
1718 			}
1719 		} else if (type == DT_REG || type == DT_LNK) {
1720 			err = import_file(&new_te, de, path_dir, repo);
1721 			if (err)
1722 				goto done;
1723 		} else
1724 			continue;
1725 
1726 		err = insert_tree_entry(new_te, &paths);
1727 		if (err)
1728 			goto done;
1729 		nentries++;
1730 	}
1731 
1732 	if (TAILQ_EMPTY(&paths)) {
1733 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1734 		    "cannot create tree without any entries");
1735 		goto done;
1736 	}
1737 
1738 	TAILQ_FOREACH(pe, &paths, entry) {
1739 		struct got_tree_entry *te = pe->data;
1740 		char *path;
1741 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1742 			continue;
1743 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1744 			err = got_error_from_errno("asprintf");
1745 			goto done;
1746 		}
1747 		err = (*progress_cb)(progress_arg, path);
1748 		free(path);
1749 		if (err)
1750 			goto done;
1751 	}
1752 
1753 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1754 done:
1755 	if (dir)
1756 		closedir(dir);
1757 	got_pathlist_free(&paths);
1758 	return err;
1759 }
1760 
1761 const struct got_error *
1762 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1763     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1764     struct got_repository *repo, got_repo_import_cb progress_cb,
1765     void *progress_arg)
1766 {
1767 	const struct got_error *err;
1768 	struct got_object_id *new_tree_id;
1769 
1770 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1771 	    progress_cb, progress_arg);
1772 	if (err)
1773 		return err;
1774 
1775 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1776 	    author, time(NULL), author, time(NULL), logmsg, repo);
1777 	free(new_tree_id);
1778 	return err;
1779 }