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