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