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