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 <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <sha1.h>
33 #include <string.h>
34 #include <time.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 #include <imsg.h>
40 #include <uuid.h>
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_worktree.h"
48 #include "got_object.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_worktree.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 const char *
67 got_repo_get_path(struct got_repository *repo)
68 {
69 	return repo->path;
70 }
71 
72 const char *
73 got_repo_get_path_git_dir(struct got_repository *repo)
74 {
75 	return repo->path_git_dir;
76 }
77 
78 const char *
79 got_repo_get_gitconfig_author_name(struct got_repository *repo)
80 {
81 	return repo->gitconfig_author_name;
82 }
83 
84 const char *
85 got_repo_get_gitconfig_author_email(struct got_repository *repo)
86 {
87 	return repo->gitconfig_author_email;
88 }
89 
90 const char *
91 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
92 {
93 	return repo->global_gitconfig_author_name;
94 }
95 
96 const char *
97 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
98 {
99 	return repo->global_gitconfig_author_email;
100 }
101 
102 const char *
103 got_repo_get_gitconfig_owner(struct got_repository *repo)
104 {
105 	return repo->gitconfig_owner;
106 }
107 
108 int
109 got_repo_is_bare(struct got_repository *repo)
110 {
111 	return (strcmp(repo->path, repo->path_git_dir) == 0);
112 }
113 
114 static char *
115 get_path_git_child(struct got_repository *repo, const char *basename)
116 {
117 	char *path_child;
118 	
119 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
120 	    basename) == -1)
121 		return NULL;
122 
123 	return path_child;
124 }
125 
126 char *
127 got_repo_get_path_objects(struct got_repository *repo)
128 {
129 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
130 }
131 
132 char *
133 got_repo_get_path_objects_pack(struct got_repository *repo)
134 {
135 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
136 }
137 
138 char *
139 got_repo_get_path_refs(struct got_repository *repo)
140 {
141 	return get_path_git_child(repo, GOT_REFS_DIR);
142 }
143 
144 char *
145 got_repo_get_path_packed_refs(struct got_repository *repo)
146 {
147 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
148 }
149 
150 static char *
151 get_path_head(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_HEAD_FILE);
154 }
155 
156 static const struct got_error *
157 get_path_gitconfig(char **p, struct got_repository *repo)
158 {
159 	*p = get_path_git_child(repo, GOT_GITCONFIG);
160 	if (*p == NULL)
161 		return got_error_from_errno("asprintf");
162 	return NULL;
163 }
164 
165 void
166 got_repo_get_gitconfig_remotes(int *nremotes, struct got_remote_repo **remotes,
167     struct got_repository *repo)
168 {
169 	*nremotes = repo->ngitconfig_remotes;
170 	*remotes = repo->gitconfig_remotes;
171 }
172 
173 static int
174 is_git_repo(struct got_repository *repo)
175 {
176 	const char *path_git = got_repo_get_path_git_dir(repo);
177 	char *path_objects = got_repo_get_path_objects(repo);
178 	char *path_refs = got_repo_get_path_refs(repo);
179 	char *path_head = get_path_head(repo);
180 	int ret = 0;
181 	struct stat sb;
182 	struct got_reference *head_ref;
183 
184 	if (lstat(path_git, &sb) == -1)
185 		goto done;
186 	if (!S_ISDIR(sb.st_mode))
187 		goto done;
188 
189 	if (lstat(path_objects, &sb) == -1)
190 		goto done;
191 	if (!S_ISDIR(sb.st_mode))
192 		goto done;
193 
194 	if (lstat(path_refs, &sb) == -1)
195 		goto done;
196 	if (!S_ISDIR(sb.st_mode))
197 		goto done;
198 
199 	if (lstat(path_head, &sb) == -1)
200 		goto done;
201 	if (!S_ISREG(sb.st_mode))
202 		goto done;
203 
204 	/* Check if the HEAD reference can be opened. */
205 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
206 		goto done;
207 	got_ref_close(head_ref);
208 
209 	ret = 1;
210 done:
211 	free(path_objects);
212 	free(path_refs);
213 	free(path_head);
214 	return ret;
215 
216 }
217 
218 const struct got_error *
219 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
220     struct got_object *obj)
221 {
222 #ifndef GOT_NO_OBJ_CACHE
223 	const struct got_error *err = NULL;
224 	err = got_object_cache_add(&repo->objcache, id, obj);
225 	if (err) {
226 		if (err->code == GOT_ERR_OBJ_EXISTS ||
227 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
228 			err = NULL;
229 		return err;
230 	}
231 	obj->refcnt++;
232 #endif
233 	return NULL;
234 }
235 
236 struct got_object *
237 got_repo_get_cached_object(struct got_repository *repo,
238     struct got_object_id *id)
239 {
240 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
241 }
242 
243 const struct got_error *
244 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
245     struct got_tree_object *tree)
246 {
247 #ifndef GOT_NO_OBJ_CACHE
248 	const struct got_error *err = NULL;
249 	err = got_object_cache_add(&repo->treecache, id, tree);
250 	if (err) {
251 		if (err->code == GOT_ERR_OBJ_EXISTS ||
252 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
253 			err = NULL;
254 		return err;
255 	}
256 	tree->refcnt++;
257 #endif
258 	return NULL;
259 }
260 
261 struct got_tree_object *
262 got_repo_get_cached_tree(struct got_repository *repo,
263     struct got_object_id *id)
264 {
265 	return (struct got_tree_object *)got_object_cache_get(
266 	    &repo->treecache, id);
267 }
268 
269 const struct got_error *
270 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
271     struct got_commit_object *commit)
272 {
273 #ifndef GOT_NO_OBJ_CACHE
274 	const struct got_error *err = NULL;
275 	err = got_object_cache_add(&repo->commitcache, id, commit);
276 	if (err) {
277 		if (err->code == GOT_ERR_OBJ_EXISTS ||
278 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
279 			err = NULL;
280 		return err;
281 	}
282 	commit->refcnt++;
283 #endif
284 	return NULL;
285 }
286 
287 struct got_commit_object *
288 got_repo_get_cached_commit(struct got_repository *repo,
289     struct got_object_id *id)
290 {
291 	return (struct got_commit_object *)got_object_cache_get(
292 	    &repo->commitcache, id);
293 }
294 
295 const struct got_error *
296 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
297     struct got_tag_object *tag)
298 {
299 #ifndef GOT_NO_OBJ_CACHE
300 	const struct got_error *err = NULL;
301 	err = got_object_cache_add(&repo->tagcache, id, tag);
302 	if (err) {
303 		if (err->code == GOT_ERR_OBJ_EXISTS ||
304 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
305 			err = NULL;
306 		return err;
307 	}
308 	tag->refcnt++;
309 #endif
310 	return NULL;
311 }
312 
313 struct got_tag_object *
314 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
315 {
316 	return (struct got_tag_object *)got_object_cache_get(
317 	    &repo->tagcache, id);
318 }
319 
320 const struct got_error *
321 open_repo(struct got_repository *repo, const char *path)
322 {
323 	const struct got_error *err = NULL;
324 
325 	/* bare git repository? */
326 	repo->path_git_dir = strdup(path);
327 	if (repo->path_git_dir == NULL)
328 		return got_error_from_errno("strdup");
329 	if (is_git_repo(repo)) {
330 		repo->path = strdup(repo->path_git_dir);
331 		if (repo->path == NULL) {
332 			err = got_error_from_errno("strdup");
333 			goto done;
334 		}
335 		return NULL;
336 	}
337 
338 	/* git repository with working tree? */
339 	free(repo->path_git_dir);
340 	repo->path_git_dir = NULL;
341 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
342 		err = got_error_from_errno("asprintf");
343 		goto done;
344 	}
345 	if (is_git_repo(repo)) {
346 		repo->path = strdup(path);
347 		if (repo->path == NULL) {
348 			err = got_error_from_errno("strdup");
349 			goto done;
350 		}
351 		return NULL;
352 	}
353 
354 	err = got_error(GOT_ERR_NOT_GIT_REPO);
355 done:
356 	if (err) {
357 		free(repo->path);
358 		repo->path = NULL;
359 		free(repo->path_git_dir);
360 		repo->path_git_dir = NULL;
361 	}
362 	return err;
363 }
364 
365 static const struct got_error *
366 parse_gitconfig_file(int *gitconfig_repository_format_version,
367     char **gitconfig_author_name, char **gitconfig_author_email,
368     struct got_remote_repo **remotes, int *nremotes,
369     char **gitconfig_owner,
370     const char *gitconfig_path)
371 {
372 	const struct got_error *err = NULL, *child_err = NULL;
373 	int fd = -1;
374 	int imsg_fds[2] = { -1, -1 };
375 	pid_t pid;
376 	struct imsgbuf *ibuf;
377 
378 	*gitconfig_repository_format_version = 0;
379 	*gitconfig_author_name = NULL;
380 	*gitconfig_author_email = NULL;
381 
382 	fd = open(gitconfig_path, O_RDONLY);
383 	if (fd == -1) {
384 		if (errno == ENOENT)
385 			return NULL;
386 		return got_error_from_errno2("open", gitconfig_path);
387 	}
388 
389 	ibuf = calloc(1, sizeof(*ibuf));
390 	if (ibuf == NULL) {
391 		err = got_error_from_errno("calloc");
392 		goto done;
393 	}
394 
395 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
396 		err = got_error_from_errno("socketpair");
397 		goto done;
398 	}
399 
400 	pid = fork();
401 	if (pid == -1) {
402 		err = got_error_from_errno("fork");
403 		goto done;
404 	} else if (pid == 0) {
405 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
406 		    gitconfig_path);
407 		/* not reached */
408 	}
409 
410 	if (close(imsg_fds[1]) == -1) {
411 		err = got_error_from_errno("close");
412 		goto done;
413 	}
414 	imsg_fds[1] = -1;
415 	imsg_init(ibuf, imsg_fds[0]);
416 
417 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
418 	if (err)
419 		goto done;
420 	fd = -1;
421 
422 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
423 	if (err)
424 		goto done;
425 
426 	err = got_privsep_recv_gitconfig_int(
427 	    gitconfig_repository_format_version, ibuf);
428 	if (err)
429 		goto done;
430 
431 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
432 	if (err)
433 		goto done;
434 
435 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
436 	if (err)
437 		goto done;
438 
439 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
440 	if (err)
441 		goto done;
442 
443 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
444 	if (err)
445 		goto done;
446 
447 	if (remotes && nremotes) {
448 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
449 		if (err)
450 			goto done;
451 
452 		err = got_privsep_recv_gitconfig_remotes(remotes,
453 		    nremotes, ibuf);
454 		if (err)
455 			goto done;
456 	}
457 
458 	if (gitconfig_owner) {
459 		err = got_privsep_send_gitconfig_owner_req(ibuf);
460 		if (err)
461 			goto done;
462 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
463 		if (err)
464 			goto done;
465 	}
466 
467 	imsg_clear(ibuf);
468 	err = got_privsep_send_stop(imsg_fds[0]);
469 	child_err = got_privsep_wait_for_child(pid);
470 	if (child_err && err == NULL)
471 		err = child_err;
472 done:
473 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
474 		err = got_error_from_errno("close");
475 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
476 		err = got_error_from_errno("close");
477 	if (fd != -1 && close(fd) == -1 && err == NULL)
478 		err = got_error_from_errno2("close", gitconfig_path);
479 	free(ibuf);
480 	return err;
481 }
482 
483 static const struct got_error *
484 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
485 {
486 	const struct got_error *err = NULL;
487 	char *repo_gitconfig_path = NULL;
488 
489 	if (global_gitconfig_path) {
490 		/* Read settings from ~/.gitconfig. */
491 		int dummy_repo_version;
492 		err = parse_gitconfig_file(&dummy_repo_version,
493 		    &repo->global_gitconfig_author_name,
494 		    &repo->global_gitconfig_author_email,
495 		    NULL, NULL, NULL, global_gitconfig_path);
496 		if (err)
497 			return err;
498 	}
499 
500 	/* Read repository's .git/config file. */
501 	err = get_path_gitconfig(&repo_gitconfig_path, repo);
502 	if (err)
503 		return err;
504 
505 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
506 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
507 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
508 	    &repo->gitconfig_owner, repo_gitconfig_path);
509 	if (err)
510 		goto done;
511 done:
512 	free(repo_gitconfig_path);
513 	return err;
514 }
515 
516 const struct got_error *
517 got_repo_open(struct got_repository **repop, const char *path,
518     const char *global_gitconfig_path)
519 {
520 	struct got_repository *repo = NULL;
521 	const struct got_error *err = NULL;
522 	char *abspath;
523 	int i, tried_root = 0;
524 
525 	*repop = NULL;
526 
527 	if (got_path_is_absolute(path))
528 		abspath = strdup(path);
529 	else
530 		abspath = got_path_get_absolute(path);
531 	if (abspath == NULL)
532 		return got_error_path(path, GOT_ERR_BAD_PATH);
533 
534 	repo = calloc(1, sizeof(*repo));
535 	if (repo == NULL) {
536 		err = got_error_from_errno("calloc");
537 		goto done;
538 	}
539 
540 	for (i = 0; i < nitems(repo->privsep_children); i++) {
541 		memset(&repo->privsep_children[i], 0,
542 		    sizeof(repo->privsep_children[0]));
543 		repo->privsep_children[i].imsg_fd = -1;
544 	}
545 
546 	err = got_object_cache_init(&repo->objcache,
547 	    GOT_OBJECT_CACHE_TYPE_OBJ);
548 	if (err)
549 		goto done;
550 	err = got_object_cache_init(&repo->treecache,
551 	    GOT_OBJECT_CACHE_TYPE_TREE);
552 	if (err)
553 		goto done;
554 	err = got_object_cache_init(&repo->commitcache,
555 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
556 	if (err)
557 		goto done;
558 	err = got_object_cache_init(&repo->tagcache,
559 	    GOT_OBJECT_CACHE_TYPE_TAG);
560 	if (err)
561 		goto done;
562 
563 	path = realpath(abspath, NULL);
564 	if (path == NULL) {
565 		err = got_error_from_errno2("realpath", abspath);
566 		goto done;
567 	}
568 
569 	do {
570 		err = open_repo(repo, path);
571 		if (err == NULL)
572 			break;
573 		if (err->code != GOT_ERR_NOT_GIT_REPO)
574 			break;
575 		if (path[0] == '/' && path[1] == '\0') {
576 			if (tried_root) {
577 				err = got_error(GOT_ERR_NOT_GIT_REPO);
578 				goto done;
579 			}
580 			tried_root = 1;
581 		}
582 		path = dirname(path);
583 		if (path == NULL) {
584 			err = got_error_from_errno2("dirname", path);
585 			goto done;
586 		}
587 	} while (path);
588 
589 	err = read_gitconfig(repo, global_gitconfig_path);
590 	if (err)
591 		goto done;
592 	if (repo->gitconfig_repository_format_version != 0)
593 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
594 done:
595 	if (err)
596 		got_repo_close(repo);
597 	else
598 		*repop = repo;
599 	free(abspath);
600 	return err;
601 }
602 
603 const struct got_error *
604 got_repo_close(struct got_repository *repo)
605 {
606 	const struct got_error *err = NULL, *child_err;
607 	int i;
608 
609 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
610 		if (repo->packidx_cache[i] == NULL)
611 			break;
612 		got_packidx_close(repo->packidx_cache[i]);
613 	}
614 
615 	for (i = 0; i < nitems(repo->packs); i++) {
616 		if (repo->packs[i].path_packfile == NULL)
617 			break;
618 		got_pack_close(&repo->packs[i]);
619 	}
620 
621 	free(repo->path);
622 	free(repo->path_git_dir);
623 
624 	got_object_cache_close(&repo->objcache);
625 	got_object_cache_close(&repo->treecache);
626 	got_object_cache_close(&repo->commitcache);
627 	got_object_cache_close(&repo->tagcache);
628 
629 	for (i = 0; i < nitems(repo->privsep_children); i++) {
630 		if (repo->privsep_children[i].imsg_fd == -1)
631 			continue;
632 		imsg_clear(repo->privsep_children[i].ibuf);
633 		free(repo->privsep_children[i].ibuf);
634 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
635 		child_err = got_privsep_wait_for_child(
636 		    repo->privsep_children[i].pid);
637 		if (child_err && err == NULL)
638 			err = child_err;
639 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
640 		    err == NULL)
641 			err = got_error_from_errno("close");
642 	}
643 
644 	free(repo->gitconfig_author_name);
645 	free(repo->gitconfig_author_email);
646 	for (i = 0; i < repo->ngitconfig_remotes; i++) {
647 		free(repo->gitconfig_remotes[i].name);
648 		free(repo->gitconfig_remotes[i].url);
649 	}
650 	free(repo->gitconfig_remotes);
651 	free(repo);
652 
653 	return err;
654 }
655 
656 const struct got_error *
657 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
658     const char *input_path, int check_disk)
659 {
660 	const struct got_error *err = NULL;
661 	const char *repo_abspath = NULL;
662 	size_t repolen, len;
663 	char *canonpath, *path = NULL;
664 
665 	*in_repo_path = NULL;
666 
667 	canonpath = strdup(input_path);
668 	if (canonpath == NULL) {
669 		err = got_error_from_errno("strdup");
670 		goto done;
671 	}
672 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
673 	if (err)
674 		goto done;
675 
676 	repo_abspath = got_repo_get_path(repo);
677 
678 	if (!check_disk || canonpath[0] == '\0') {
679 		path = strdup(canonpath);
680 		if (path == NULL) {
681 			err = got_error_from_errno("strdup");
682 			goto done;
683 		}
684 	} else {
685 		path = realpath(canonpath, NULL);
686 		if (path == NULL) {
687 			if (errno != ENOENT) {
688 				err = got_error_from_errno2("realpath",
689 				    canonpath);
690 				goto done;
691 			}
692 			/*
693 			 * Path is not on disk.
694 			 * Assume it is already relative to repository root.
695 			 */
696 			path = strdup(canonpath);
697 			if (path == NULL) {
698 				err = got_error_from_errno("strdup");
699 				goto done;
700 			}
701 		}
702 
703 		repolen = strlen(repo_abspath);
704 		len = strlen(path);
705 
706 
707 		if (strcmp(path, repo_abspath) == 0) {
708 			free(path);
709 			path = strdup("");
710 			if (path == NULL) {
711 				err = got_error_from_errno("strdup");
712 				goto done;
713 			}
714 		} else if (len > repolen &&
715 		    got_path_is_child(path, repo_abspath, repolen)) {
716 			/* Matched an on-disk path inside repository. */
717 			if (got_repo_is_bare(repo)) {
718 				/*
719 				 * Matched an on-disk path inside repository
720 				 * database. Treat as repository-relative.
721 				 */
722 			} else {
723 				char *child;
724 				/* Strip common prefix with repository path. */
725 				err = got_path_skip_common_ancestor(&child,
726 				    repo_abspath, path);
727 				if (err)
728 					goto done;
729 				free(path);
730 				path = child;
731 			}
732 		} else {
733 			/*
734 			 * Matched unrelated on-disk path.
735 			 * Treat it as repository-relative.
736 			 */
737 		}
738 	}
739 
740 	/* Make in-repository path absolute */
741 	if (path[0] != '/') {
742 		char *abspath;
743 		if (asprintf(&abspath, "/%s", path) == -1) {
744 			err = got_error_from_errno("asprintf");
745 			goto done;
746 		}
747 		free(path);
748 		path = abspath;
749 	}
750 
751 done:
752 	free(canonpath);
753 	if (err)
754 		free(path);
755 	else
756 		*in_repo_path = path;
757 	return err;
758 }
759 
760 static const struct got_error *
761 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
762     const char *path_packidx)
763 {
764 	const struct got_error *err = NULL;
765 	int i;
766 
767 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
768 		if (repo->packidx_cache[i] == NULL)
769 			break;
770 		if (strcmp(repo->packidx_cache[i]->path_packidx,
771 		    path_packidx) == 0) {
772 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
773 		}
774 	}
775 	if (i == nitems(repo->packidx_cache)) {
776 		err = got_packidx_close(repo->packidx_cache[i - 1]);
777 		if (err)
778 			return err;
779 	}
780 
781 	/*
782 	 * Insert the new pack index at the front so it will
783 	 * be searched first in the future.
784 	 */
785 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
786 	    sizeof(repo->packidx_cache) -
787 	    sizeof(repo->packidx_cache[0]));
788 	repo->packidx_cache[0] = packidx;
789 
790 	return NULL;
791 }
792 
793 static int
794 is_packidx_filename(const char *name, size_t len)
795 {
796 	if (len != GOT_PACKIDX_NAMELEN)
797 		return 0;
798 
799 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
800 		return 0;
801 
802 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
803 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
804 		return 0;
805 
806 	return 1;
807 }
808 
809 const struct got_error *
810 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
811     struct got_repository *repo, struct got_object_id *id)
812 {
813 	const struct got_error *err;
814 	char *path_packdir;
815 	DIR *packdir;
816 	struct dirent *dent;
817 	char *path_packidx;
818 	int i;
819 
820 	/* Search pack index cache. */
821 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
822 		if (repo->packidx_cache[i] == NULL)
823 			break;
824 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
825 		if (*idx != -1) {
826 			*packidx = repo->packidx_cache[i];
827 			/*
828 			 * Move this cache entry to the front. Repeatedly
829 			 * searching a wrong pack index can be expensive.
830 			 */
831 			if (i > 0) {
832 				struct got_packidx *p;
833 				p = repo->packidx_cache[0];
834 				repo->packidx_cache[0] = *packidx;
835 				repo->packidx_cache[i] = p;
836 			}
837 			return NULL;
838 		}
839 	}
840 	/* No luck. Search the filesystem. */
841 
842 	path_packdir = got_repo_get_path_objects_pack(repo);
843 	if (path_packdir == NULL)
844 		return got_error_from_errno("got_repo_get_path_objects_pack");
845 
846 	packdir = opendir(path_packdir);
847 	if (packdir == NULL) {
848 		if (errno == ENOENT)
849 			err = got_error_no_obj(id);
850 		else
851 			err = got_error_from_errno2("opendir", path_packdir);
852 		goto done;
853 	}
854 
855 	while ((dent = readdir(packdir)) != NULL) {
856 		int is_cached = 0;
857 
858 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
859 			continue;
860 
861 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
862 		    dent->d_name) == -1) {
863 			err = got_error_from_errno("asprintf");
864 			goto done;
865 		}
866 
867 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
868 			if (repo->packidx_cache[i] == NULL)
869 				break;
870 			if (strcmp(repo->packidx_cache[i]->path_packidx,
871 			    path_packidx) == 0) {
872 				is_cached = 1;
873 				break;
874 			}
875 		}
876 		if (is_cached) {
877 			free(path_packidx);
878 			continue; /* already searched */
879 		}
880 
881 		err = got_packidx_open(packidx, path_packidx, 0);
882 		if (err) {
883 			free(path_packidx);
884 			goto done;
885 		}
886 
887 		err = cache_packidx(repo, *packidx, path_packidx);
888 		free(path_packidx);
889 		if (err)
890 			goto done;
891 
892 		*idx = got_packidx_get_object_idx(*packidx, id);
893 		if (*idx != -1) {
894 			err = NULL; /* found the object */
895 			goto done;
896 		}
897 	}
898 
899 	err = got_error_no_obj(id);
900 done:
901 	free(path_packdir);
902 	if (packdir && closedir(packdir) != 0 && err == NULL)
903 		err = got_error_from_errno("closedir");
904 	return err;
905 }
906 
907 static const struct got_error *
908 read_packfile_hdr(int fd, struct got_packidx *packidx)
909 {
910 	const struct got_error *err = NULL;
911 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
912 	struct got_packfile_hdr hdr;
913 	ssize_t n;
914 
915 	n = read(fd, &hdr, sizeof(hdr));
916 	if (n < 0)
917 		return got_error_from_errno("read");
918 	if (n != sizeof(hdr))
919 		return got_error(GOT_ERR_BAD_PACKFILE);
920 
921 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
922 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
923 	    betoh32(hdr.nobjects) != totobj)
924 		err = got_error(GOT_ERR_BAD_PACKFILE);
925 
926 	return err;
927 }
928 
929 static const struct got_error *
930 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
931 {
932 	const struct got_error *err = NULL;
933 
934 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
935 	if (*fd == -1)
936 		return got_error_from_errno2("open", path_packfile);
937 
938 	if (packidx) {
939 		err = read_packfile_hdr(*fd, packidx);
940 		if (err) {
941 			close(*fd);
942 			*fd = -1;
943 		}
944 	}
945 
946 	return err;
947 }
948 
949 const struct got_error *
950 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
951     const char *path_packfile, struct got_packidx *packidx)
952 {
953 	const struct got_error *err = NULL;
954 	struct got_pack *pack = NULL;
955 	struct stat sb;
956 	int i;
957 
958 	if (packp)
959 		*packp = NULL;
960 
961 	for (i = 0; i < nitems(repo->packs); i++) {
962 		pack = &repo->packs[i];
963 		if (pack->path_packfile == NULL)
964 			break;
965 		if (strcmp(pack->path_packfile, path_packfile) == 0)
966 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
967 	}
968 
969 	if (i == nitems(repo->packs) - 1) {
970 		err = got_pack_close(&repo->packs[i - 1]);
971 		if (err)
972 			return err;
973 		memmove(&repo->packs[1], &repo->packs[0],
974 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
975 		i = 0;
976 	}
977 
978 	pack = &repo->packs[i];
979 
980 	pack->path_packfile = strdup(path_packfile);
981 	if (pack->path_packfile == NULL) {
982 		err = got_error_from_errno("strdup");
983 		goto done;
984 	}
985 
986 	err = open_packfile(&pack->fd, path_packfile, packidx);
987 	if (err)
988 		goto done;
989 
990 	if (fstat(pack->fd, &sb) != 0) {
991 		err = got_error_from_errno("fstat");
992 		goto done;
993 	}
994 	pack->filesize = sb.st_size;
995 
996 	pack->privsep_child = NULL;
997 
998 #ifndef GOT_PACK_NO_MMAP
999 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1000 	    pack->fd, 0);
1001 	if (pack->map == MAP_FAILED) {
1002 		if (errno != ENOMEM) {
1003 			err = got_error_from_errno("mmap");
1004 			goto done;
1005 		}
1006 		pack->map = NULL; /* fall back to read(2) */
1007 	}
1008 #endif
1009 done:
1010 	if (err) {
1011 		if (pack) {
1012 			free(pack->path_packfile);
1013 			memset(pack, 0, sizeof(*pack));
1014 		}
1015 	} else if (packp)
1016 		*packp = pack;
1017 	return err;
1018 }
1019 
1020 struct got_pack *
1021 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1022 {
1023 	struct got_pack *pack = NULL;
1024 	int i;
1025 
1026 	for (i = 0; i < nitems(repo->packs); i++) {
1027 		pack = &repo->packs[i];
1028 		if (pack->path_packfile == NULL)
1029 			break;
1030 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1031 			return pack;
1032 	}
1033 
1034 	return NULL;
1035 }
1036 
1037 const struct got_error *
1038 got_repo_init(const char *repo_path)
1039 {
1040 	const struct got_error *err = NULL;
1041 	const char *dirnames[] = {
1042 		GOT_OBJECTS_DIR,
1043 		GOT_OBJECTS_PACK_DIR,
1044 		GOT_REFS_DIR,
1045 	};
1046 	const char *description_str = "Unnamed repository; "
1047 	    "edit this file 'description' to name the repository.";
1048 	const char *headref_str = "ref: refs/heads/main";
1049 	const char *gitconfig_str = "[core]\n"
1050 	    "\trepositoryformatversion = 0\n"
1051 	    "\tfilemode = true\n"
1052 	    "\tbare = true\n";
1053 	char *path;
1054 	int i;
1055 
1056 	if (!got_path_dir_is_empty(repo_path))
1057 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1058 
1059 	for (i = 0; i < nitems(dirnames); i++) {
1060 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1061 			return got_error_from_errno("asprintf");
1062 		}
1063 		err = got_path_mkdir(path);
1064 		free(path);
1065 		if (err)
1066 			return err;
1067 	}
1068 
1069 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1070 		return got_error_from_errno("asprintf");
1071 	err = got_path_create_file(path, description_str);
1072 	free(path);
1073 	if (err)
1074 		return err;
1075 
1076 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1077 		return got_error_from_errno("asprintf");
1078 	err = got_path_create_file(path, headref_str);
1079 	free(path);
1080 	if (err)
1081 		return err;
1082 
1083 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1084 		return got_error_from_errno("asprintf");
1085 	err = got_path_create_file(path, gitconfig_str);
1086 	free(path);
1087 	if (err)
1088 		return err;
1089 
1090 	return NULL;
1091 }
1092 
1093 static const struct got_error *
1094 match_packed_object(struct got_object_id **unique_id,
1095     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1096 {
1097 	const struct got_error *err = NULL;
1098 	char *path_packdir;
1099 	DIR *packdir;
1100 	struct dirent *dent;
1101 	char *path_packidx;
1102 	struct got_object_id_queue matched_ids;
1103 
1104 	SIMPLEQ_INIT(&matched_ids);
1105 
1106 	path_packdir = got_repo_get_path_objects_pack(repo);
1107 	if (path_packdir == NULL)
1108 		return got_error_from_errno("got_repo_get_path_objects_pack");
1109 
1110 	packdir = opendir(path_packdir);
1111 	if (packdir == NULL) {
1112 		if (errno != ENOENT)
1113 			err = got_error_from_errno2("opendir", path_packdir);
1114 		goto done;
1115 	}
1116 
1117 	while ((dent = readdir(packdir)) != NULL) {
1118 		struct got_packidx *packidx;
1119 		struct got_object_qid *qid;
1120 
1121 
1122 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1123 			continue;
1124 
1125 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1126 		    dent->d_name) == -1) {
1127 			err = got_error_from_errno("asprintf");
1128 			break;
1129 		}
1130 
1131 		err = got_packidx_open(&packidx, path_packidx, 0);
1132 		free(path_packidx);
1133 		if (err)
1134 			break;
1135 
1136 		err = got_packidx_match_id_str_prefix(&matched_ids,
1137 		    packidx, id_str_prefix);
1138 		if (err) {
1139 			got_packidx_close(packidx);
1140 			break;
1141 		}
1142 		err = got_packidx_close(packidx);
1143 		if (err)
1144 			break;
1145 
1146 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1147 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1148 				int matched_type;
1149 				err = got_object_get_type(&matched_type, repo,
1150 				    qid->id);
1151 				if (err)
1152 					goto done;
1153 				if (matched_type != obj_type)
1154 					continue;
1155 			}
1156 			if (*unique_id == NULL) {
1157 				*unique_id = got_object_id_dup(qid->id);
1158 				if (*unique_id == NULL) {
1159 					err = got_error_from_errno("malloc");
1160 					goto done;
1161 				}
1162 			} else {
1163 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1164 					continue; /* packed multiple times */
1165 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1166 				goto done;
1167 			}
1168 		}
1169 	}
1170 done:
1171 	got_object_id_queue_free(&matched_ids);
1172 	free(path_packdir);
1173 	if (packdir && closedir(packdir) != 0 && err == NULL)
1174 		err = got_error_from_errno("closedir");
1175 	if (err) {
1176 		free(*unique_id);
1177 		*unique_id = NULL;
1178 	}
1179 	return err;
1180 }
1181 
1182 static const struct got_error *
1183 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1184     const char *object_dir, const char *id_str_prefix, int obj_type,
1185     struct got_repository *repo)
1186 {
1187 	const struct got_error *err = NULL;
1188 	char *path;
1189 	DIR *dir = NULL;
1190 	struct dirent *dent;
1191 	struct got_object_id id;
1192 
1193 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1194 		err = got_error_from_errno("asprintf");
1195 		goto done;
1196 	}
1197 
1198 	dir = opendir(path);
1199 	if (dir == NULL) {
1200 		if (errno == ENOENT) {
1201 			err = NULL;
1202 			goto done;
1203 		}
1204 		err = got_error_from_errno2("opendir", path);
1205 		goto done;
1206 	}
1207 	while ((dent = readdir(dir)) != NULL) {
1208 		char *id_str;
1209 		int cmp;
1210 
1211 		if (strcmp(dent->d_name, ".") == 0 ||
1212 		    strcmp(dent->d_name, "..") == 0)
1213 			continue;
1214 
1215 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1216 			err = got_error_from_errno("asprintf");
1217 			goto done;
1218 		}
1219 
1220 		if (!got_parse_sha1_digest(id.sha1, id_str))
1221 			continue;
1222 
1223 		/*
1224 		 * Directory entries do not necessarily appear in
1225 		 * sorted order, so we must iterate over all of them.
1226 		 */
1227 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1228 		if (cmp != 0) {
1229 			free(id_str);
1230 			continue;
1231 		}
1232 
1233 		if (*unique_id == NULL) {
1234 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1235 				int matched_type;
1236 				err = got_object_get_type(&matched_type, repo,
1237 				    &id);
1238 				if (err)
1239 					goto done;
1240 				if (matched_type != obj_type)
1241 					continue;
1242 			}
1243 			*unique_id = got_object_id_dup(&id);
1244 			if (*unique_id == NULL) {
1245 				err = got_error_from_errno("got_object_id_dup");
1246 				free(id_str);
1247 				goto done;
1248 			}
1249 		} else {
1250 			if (got_object_id_cmp(*unique_id, &id) == 0)
1251 				continue; /* both packed and loose */
1252 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1253 			free(id_str);
1254 			goto done;
1255 		}
1256 	}
1257 done:
1258 	if (dir && closedir(dir) != 0 && err == NULL)
1259 		err = got_error_from_errno("closedir");
1260 	if (err) {
1261 		free(*unique_id);
1262 		*unique_id = NULL;
1263 	}
1264 	free(path);
1265 	return err;
1266 }
1267 
1268 const struct got_error *
1269 got_repo_match_object_id_prefix(struct got_object_id **id,
1270     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1271 {
1272 	const struct got_error *err = NULL;
1273 	char *path_objects = got_repo_get_path_objects(repo);
1274 	char *object_dir = NULL;
1275 	size_t len;
1276 	int i;
1277 
1278 	*id = NULL;
1279 
1280 	for (i = 0; i < strlen(id_str_prefix); i++) {
1281 		if (isxdigit((unsigned char)id_str_prefix[i]))
1282 			continue;
1283 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1284 	}
1285 
1286 	len = strlen(id_str_prefix);
1287 	if (len >= 2) {
1288 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1289 		if (err)
1290 			goto done;
1291 		object_dir = strndup(id_str_prefix, 2);
1292 		if (object_dir == NULL) {
1293 			err = got_error_from_errno("strdup");
1294 			goto done;
1295 		}
1296 		err = match_loose_object(id, path_objects, object_dir,
1297 		    id_str_prefix, obj_type, repo);
1298 	} else if (len == 1) {
1299 		int i;
1300 		for (i = 0; i < 0xf; i++) {
1301 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1302 			    == -1) {
1303 				err = got_error_from_errno("asprintf");
1304 				goto done;
1305 			}
1306 			err = match_packed_object(id, repo, object_dir,
1307 			    obj_type);
1308 			if (err)
1309 				goto done;
1310 			err = match_loose_object(id, path_objects, object_dir,
1311 			    id_str_prefix, obj_type, repo);
1312 			if (err)
1313 				goto done;
1314 		}
1315 	} else {
1316 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1317 		goto done;
1318 	}
1319 done:
1320 	free(object_dir);
1321 	if (err) {
1322 		free(*id);
1323 		*id = NULL;
1324 	} else if (*id == NULL)
1325 		err = got_error(GOT_ERR_NO_OBJ);
1326 
1327 	return err;
1328 }
1329 
1330 const struct got_error *
1331 got_repo_match_object_id(struct got_object_id **id, char **label,
1332     const char *id_str, int obj_type, int resolve_tags,
1333     struct got_repository *repo)
1334 {
1335 	const struct got_error *err;
1336 	struct got_tag_object *tag;
1337 	struct got_reference *ref = NULL;
1338 
1339 	*id = NULL;
1340 	if (label)
1341 		*label = NULL;
1342 
1343 	if (resolve_tags) {
1344 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1345 		    repo);
1346 		if (err == NULL) {
1347 			*id = got_object_id_dup(
1348 			    got_object_tag_get_object_id(tag));
1349 			if (*id == NULL)
1350 				err = got_error_from_errno("got_object_id_dup");
1351 			else if (label && asprintf(label, "refs/tags/%s",
1352 			    got_object_tag_get_name(tag)) == -1) {
1353 				err = got_error_from_errno("asprintf");
1354 				free(*id);
1355 				*id = NULL;
1356 			}
1357 			got_object_tag_close(tag);
1358 			return err;
1359 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1360 		    err->code != GOT_ERR_NO_OBJ)
1361 			return err;
1362 	}
1363 
1364 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1365 	if (err) {
1366 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1367 			return err;
1368 		err = got_ref_open(&ref, repo, id_str, 0);
1369 		if (err != NULL)
1370 			goto done;
1371 		if (label) {
1372 			*label = strdup(got_ref_get_name(ref));
1373 			if (*label == NULL) {
1374 				err = got_error_from_errno("strdup");
1375 				goto done;
1376 			}
1377 		}
1378 		err = got_ref_resolve(id, repo, ref);
1379 	} else if (label) {
1380 		err = got_object_id_str(label, *id);
1381 		if (*label == NULL) {
1382 			err = got_error_from_errno("strdup");
1383 			goto done;
1384 		}
1385 	}
1386 done:
1387 	if (ref)
1388 		got_ref_close(ref);
1389 	return err;
1390 }
1391 
1392 const struct got_error *
1393 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1394     int obj_type, struct got_repository *repo)
1395 {
1396 	const struct got_error *err;
1397 	struct got_reflist_head refs;
1398 	struct got_reflist_entry *re;
1399 	struct got_object_id *tag_id;
1400 
1401 	SIMPLEQ_INIT(&refs);
1402 	*tag = NULL;
1403 
1404 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1405 	if (err)
1406 		return err;
1407 
1408 	SIMPLEQ_FOREACH(re, &refs, entry) {
1409 		const char *refname;
1410 		refname = got_ref_get_name(re->ref);
1411 		if (got_ref_is_symbolic(re->ref))
1412 			continue;
1413 		refname += strlen("refs/tags/");
1414 		if (strcmp(refname, name) != 0)
1415 			continue;
1416 		err = got_ref_resolve(&tag_id, repo, re->ref);
1417 		if (err)
1418 			break;
1419 		err = got_object_open_as_tag(tag, repo, tag_id);
1420 		free(tag_id);
1421 		if (err)
1422 			break;
1423 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1424 		    got_object_tag_get_object_type(*tag) == obj_type)
1425 			break;
1426 		got_object_tag_close(*tag);
1427 		*tag = NULL;
1428 	}
1429 
1430 	got_ref_list_free(&refs);
1431 	if (err == NULL && *tag == NULL)
1432 		err = got_error(GOT_ERR_NO_OBJ);
1433 	return err;
1434 }
1435 
1436 static const struct got_error *
1437 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1438     const char *name, mode_t mode, struct got_object_id *blob_id)
1439 {
1440 	const struct got_error *err = NULL;
1441 
1442 	 *new_te = NULL;
1443 
1444 	*new_te = calloc(1, sizeof(**new_te));
1445 	if (*new_te == NULL)
1446 		return got_error_from_errno("calloc");
1447 
1448 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1449 	    sizeof((*new_te)->name)) {
1450 		err = got_error(GOT_ERR_NO_SPACE);
1451 		goto done;
1452 	}
1453 
1454 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1455 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1456 done:
1457 	if (err && *new_te) {
1458 		free(*new_te);
1459 		*new_te = NULL;
1460 	}
1461 	return err;
1462 }
1463 
1464 static const struct got_error *
1465 import_file(struct got_tree_entry **new_te, struct dirent *de,
1466     const char *path, struct got_repository *repo)
1467 {
1468 	const struct got_error *err;
1469 	struct got_object_id *blob_id = NULL;
1470 	char *filepath;
1471 	struct stat sb;
1472 
1473 	if (asprintf(&filepath, "%s%s%s", path,
1474 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1475 		return got_error_from_errno("asprintf");
1476 
1477 	if (lstat(filepath, &sb) != 0) {
1478 		err = got_error_from_errno2("lstat", path);
1479 		goto done;
1480 	}
1481 
1482 	err = got_object_blob_create(&blob_id, filepath, repo);
1483 	if (err)
1484 		goto done;
1485 
1486 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1487 	    blob_id);
1488 done:
1489 	free(filepath);
1490 	if (err)
1491 		free(blob_id);
1492 	return err;
1493 }
1494 
1495 static const struct got_error *
1496 insert_tree_entry(struct got_tree_entry *new_te,
1497     struct got_pathlist_head *paths)
1498 {
1499 	const struct got_error *err = NULL;
1500 	struct got_pathlist_entry *new_pe;
1501 
1502 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1503 	if (err)
1504 		return err;
1505 	if (new_pe == NULL)
1506 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1507 	return NULL;
1508 }
1509 
1510 static const struct got_error *write_tree(struct got_object_id **,
1511     const char *, struct got_pathlist_head *, struct got_repository *,
1512     got_repo_import_cb progress_cb, void *progress_arg);
1513 
1514 static const struct got_error *
1515 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1516     const char *path, struct got_pathlist_head *ignores,
1517     struct got_repository *repo,
1518     got_repo_import_cb progress_cb, void *progress_arg)
1519 {
1520 	const struct got_error *err;
1521 	struct got_object_id *id = NULL;
1522 	char *subdirpath;
1523 
1524 	if (asprintf(&subdirpath, "%s%s%s", path,
1525 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1526 		return got_error_from_errno("asprintf");
1527 
1528 	(*new_te) = calloc(1, sizeof(**new_te));
1529 	if (*new_te == NULL)
1530 		return got_error_from_errno("calloc");
1531 	(*new_te)->mode = S_IFDIR;
1532 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1533 	    sizeof((*new_te)->name)) {
1534 		err = got_error(GOT_ERR_NO_SPACE);
1535 		goto done;
1536 	}
1537 	err = write_tree(&id, subdirpath, ignores,  repo,
1538 	    progress_cb, progress_arg);
1539 	if (err)
1540 		goto done;
1541 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1542 	
1543 done:
1544 	free(id);
1545 	free(subdirpath);
1546 	if (err) {
1547 		free(*new_te);
1548 		*new_te = NULL;
1549 	}
1550 	return err;
1551 }
1552 
1553 static const struct got_error *
1554 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1555     struct got_pathlist_head *ignores, struct got_repository *repo,
1556     got_repo_import_cb progress_cb, void *progress_arg)
1557 {
1558 	const struct got_error *err = NULL;
1559 	DIR *dir;
1560 	struct dirent *de;
1561 	int nentries;
1562 	struct got_tree_entry *new_te = NULL;
1563 	struct got_pathlist_head paths;
1564 	struct got_pathlist_entry *pe;
1565 
1566 	*new_tree_id = NULL;
1567 
1568 	TAILQ_INIT(&paths);
1569 
1570 	dir = opendir(path_dir);
1571 	if (dir == NULL) {
1572 		err = got_error_from_errno2("opendir", path_dir);
1573 		goto done;
1574 	}
1575 
1576 	nentries = 0;
1577 	while ((de = readdir(dir)) != NULL) {
1578 		int ignore = 0;
1579 
1580 		if (strcmp(de->d_name, ".") == 0 ||
1581 		    strcmp(de->d_name, "..") == 0)
1582 			continue;
1583 
1584 		TAILQ_FOREACH(pe, ignores, entry) {
1585 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1586 				ignore = 1;
1587 				break;
1588 			}
1589 		}
1590 		if (ignore)
1591 			continue;
1592 		if (de->d_type == DT_DIR) {
1593 			err = import_subdir(&new_te, de, path_dir,
1594 			    ignores, repo, progress_cb, progress_arg);
1595 			if (err) {
1596 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1597 					goto done;
1598 				err = NULL;
1599 				continue;
1600 			}
1601 		} else if (de->d_type == DT_REG) {
1602 			err = import_file(&new_te, de, path_dir, repo);
1603 			if (err)
1604 				goto done;
1605 		} else
1606 			continue;
1607 
1608 		err = insert_tree_entry(new_te, &paths);
1609 		if (err)
1610 			goto done;
1611 		nentries++;
1612 	}
1613 
1614 	if (TAILQ_EMPTY(&paths)) {
1615 		err = got_error(GOT_ERR_NO_TREE_ENTRY);
1616 		goto done;
1617 	}
1618 
1619 	TAILQ_FOREACH(pe, &paths, entry) {
1620 		struct got_tree_entry *te = pe->data;
1621 		char *path;
1622 		if (!S_ISREG(te->mode))
1623 			continue;
1624 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1625 			err = got_error_from_errno("asprintf");
1626 			goto done;
1627 		}
1628 		err = (*progress_cb)(progress_arg, path);
1629 		free(path);
1630 		if (err)
1631 			goto done;
1632 	}
1633 
1634 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1635 done:
1636 	if (dir)
1637 		closedir(dir);
1638 	got_pathlist_free(&paths);
1639 	return err;
1640 }
1641 
1642 const struct got_error *
1643 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1644     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1645     struct got_repository *repo, got_repo_import_cb progress_cb,
1646     void *progress_arg)
1647 {
1648 	const struct got_error *err;
1649 	struct got_object_id *new_tree_id;
1650 
1651 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1652 	    progress_cb, progress_arg);
1653 	if (err)
1654 		return err;
1655 
1656 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1657 	    author, time(NULL), author, time(NULL), logmsg, repo);
1658 	free(new_tree_id);
1659 	return err;
1660 }