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, cwdlen, len;
661 	char *cwd, *canonpath, *path = NULL;
662 
663 	*in_repo_path = NULL;
664 
665 	cwd = getcwd(NULL, 0);
666 	if (cwd == NULL)
667 		return got_error_from_errno("getcwd");
668 
669 	canonpath = strdup(input_path);
670 	if (canonpath == NULL) {
671 		err = got_error_from_errno("strdup");
672 		goto done;
673 	}
674 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
675 	if (err)
676 		goto done;
677 
678 	repo_abspath = got_repo_get_path(repo);
679 
680 	if (!check_disk || canonpath[0] == '\0') {
681 		path = strdup(canonpath);
682 		if (path == NULL) {
683 			err = got_error_from_errno("strdup");
684 			goto done;
685 		}
686 	} else {
687 		int is_repo_child = 0, is_cwd_child = 0;
688 
689 		path = realpath(canonpath, NULL);
690 		if (path == NULL) {
691 			if (errno != ENOENT) {
692 				err = got_error_from_errno2("realpath",
693 				    canonpath);
694 				goto done;
695 			}
696 			/*
697 			 * Path is not on disk.
698 			 * Assume it is already relative to repository root.
699 			 */
700 			path = strdup(canonpath);
701 			if (path == NULL) {
702 				err = got_error_from_errno("strdup");
703 				goto done;
704 			}
705 		}
706 
707 		repolen = strlen(repo_abspath);
708 		cwdlen = strlen(cwd);
709 		len = strlen(path);
710 
711 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
712 			is_repo_child = 1;
713 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
714 			is_cwd_child = 1;
715 
716 		if (strcmp(path, repo_abspath) == 0) {
717 			free(path);
718 			path = strdup("");
719 			if (path == NULL) {
720 				err = got_error_from_errno("strdup");
721 				goto done;
722 			}
723 		} else if (is_repo_child && is_cwd_child) {
724 			char *child;
725 			/* Strip common prefix with repository path. */
726 			err = got_path_skip_common_ancestor(&child,
727 			    repo_abspath, path);
728 			if (err)
729 				goto done;
730 			free(path);
731 			path = child;
732 		} else if (is_repo_child) {
733 			/* Matched an on-disk path inside repository. */
734 			if (got_repo_is_bare(repo)) {
735 				/*
736 				 * Matched an on-disk path inside repository
737 				 * database. Treat as repository-relative.
738 				 */
739 			} else {
740 				char *child;
741 				/* Strip common prefix with repository path. */
742 				err = got_path_skip_common_ancestor(&child,
743 				    repo_abspath, path);
744 				if (err)
745 					goto done;
746 				free(path);
747 				path = child;
748 			}
749 		} else if (is_cwd_child) {
750 			char *child;
751 			/* Strip common prefix with cwd. */
752 			err = got_path_skip_common_ancestor(&child, cwd,
753 			    path);
754 			if (err)
755 				goto done;
756 			free(path);
757 			path = child;
758 		} else {
759 			/*
760 			 * Matched unrelated on-disk path.
761 			 * Treat it as repository-relative.
762 			 */
763 		}
764 	}
765 
766 	/* Make in-repository path absolute */
767 	if (path[0] != '/') {
768 		char *abspath;
769 		if (asprintf(&abspath, "/%s", path) == -1) {
770 			err = got_error_from_errno("asprintf");
771 			goto done;
772 		}
773 		free(path);
774 		path = abspath;
775 	}
776 
777 done:
778 	free(cwd);
779 	free(canonpath);
780 	if (err)
781 		free(path);
782 	else
783 		*in_repo_path = path;
784 	return err;
785 }
786 
787 static const struct got_error *
788 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
789     const char *path_packidx)
790 {
791 	const struct got_error *err = NULL;
792 	int i;
793 
794 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
795 		if (repo->packidx_cache[i] == NULL)
796 			break;
797 		if (strcmp(repo->packidx_cache[i]->path_packidx,
798 		    path_packidx) == 0) {
799 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
800 		}
801 	}
802 	if (i == nitems(repo->packidx_cache)) {
803 		err = got_packidx_close(repo->packidx_cache[i - 1]);
804 		if (err)
805 			return err;
806 	}
807 
808 	/*
809 	 * Insert the new pack index at the front so it will
810 	 * be searched first in the future.
811 	 */
812 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
813 	    sizeof(repo->packidx_cache) -
814 	    sizeof(repo->packidx_cache[0]));
815 	repo->packidx_cache[0] = packidx;
816 
817 	return NULL;
818 }
819 
820 static int
821 is_packidx_filename(const char *name, size_t len)
822 {
823 	if (len != GOT_PACKIDX_NAMELEN)
824 		return 0;
825 
826 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
827 		return 0;
828 
829 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
830 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
831 		return 0;
832 
833 	return 1;
834 }
835 
836 const struct got_error *
837 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
838     struct got_repository *repo, struct got_object_id *id)
839 {
840 	const struct got_error *err;
841 	char *path_packdir;
842 	DIR *packdir;
843 	struct dirent *dent;
844 	char *path_packidx;
845 	int i;
846 
847 	/* Search pack index cache. */
848 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
849 		if (repo->packidx_cache[i] == NULL)
850 			break;
851 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
852 		if (*idx != -1) {
853 			*packidx = repo->packidx_cache[i];
854 			/*
855 			 * Move this cache entry to the front. Repeatedly
856 			 * searching a wrong pack index can be expensive.
857 			 */
858 			if (i > 0) {
859 				struct got_packidx *p;
860 				p = repo->packidx_cache[0];
861 				repo->packidx_cache[0] = *packidx;
862 				repo->packidx_cache[i] = p;
863 			}
864 			return NULL;
865 		}
866 	}
867 	/* No luck. Search the filesystem. */
868 
869 	path_packdir = got_repo_get_path_objects_pack(repo);
870 	if (path_packdir == NULL)
871 		return got_error_from_errno("got_repo_get_path_objects_pack");
872 
873 	packdir = opendir(path_packdir);
874 	if (packdir == NULL) {
875 		if (errno == ENOENT)
876 			err = got_error_no_obj(id);
877 		else
878 			err = got_error_from_errno2("opendir", path_packdir);
879 		goto done;
880 	}
881 
882 	while ((dent = readdir(packdir)) != NULL) {
883 		int is_cached = 0;
884 
885 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
886 			continue;
887 
888 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
889 		    dent->d_name) == -1) {
890 			err = got_error_from_errno("asprintf");
891 			goto done;
892 		}
893 
894 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
895 			if (repo->packidx_cache[i] == NULL)
896 				break;
897 			if (strcmp(repo->packidx_cache[i]->path_packidx,
898 			    path_packidx) == 0) {
899 				is_cached = 1;
900 				break;
901 			}
902 		}
903 		if (is_cached) {
904 			free(path_packidx);
905 			continue; /* already searched */
906 		}
907 
908 		err = got_packidx_open(packidx, path_packidx, 0);
909 		if (err) {
910 			free(path_packidx);
911 			goto done;
912 		}
913 
914 		err = cache_packidx(repo, *packidx, path_packidx);
915 		free(path_packidx);
916 		if (err)
917 			goto done;
918 
919 		*idx = got_packidx_get_object_idx(*packidx, id);
920 		if (*idx != -1) {
921 			err = NULL; /* found the object */
922 			goto done;
923 		}
924 	}
925 
926 	err = got_error_no_obj(id);
927 done:
928 	free(path_packdir);
929 	if (packdir && closedir(packdir) != 0 && err == NULL)
930 		err = got_error_from_errno("closedir");
931 	return err;
932 }
933 
934 static const struct got_error *
935 read_packfile_hdr(int fd, struct got_packidx *packidx)
936 {
937 	const struct got_error *err = NULL;
938 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
939 	struct got_packfile_hdr hdr;
940 	ssize_t n;
941 
942 	n = read(fd, &hdr, sizeof(hdr));
943 	if (n < 0)
944 		return got_error_from_errno("read");
945 	if (n != sizeof(hdr))
946 		return got_error(GOT_ERR_BAD_PACKFILE);
947 
948 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
949 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
950 	    betoh32(hdr.nobjects) != totobj)
951 		err = got_error(GOT_ERR_BAD_PACKFILE);
952 
953 	return err;
954 }
955 
956 static const struct got_error *
957 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
958 {
959 	const struct got_error *err = NULL;
960 
961 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
962 	if (*fd == -1)
963 		return got_error_from_errno2("open", path_packfile);
964 
965 	if (packidx) {
966 		err = read_packfile_hdr(*fd, packidx);
967 		if (err) {
968 			close(*fd);
969 			*fd = -1;
970 		}
971 	}
972 
973 	return err;
974 }
975 
976 const struct got_error *
977 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
978     const char *path_packfile, struct got_packidx *packidx)
979 {
980 	const struct got_error *err = NULL;
981 	struct got_pack *pack = NULL;
982 	struct stat sb;
983 	int i;
984 
985 	if (packp)
986 		*packp = NULL;
987 
988 	for (i = 0; i < nitems(repo->packs); i++) {
989 		pack = &repo->packs[i];
990 		if (pack->path_packfile == NULL)
991 			break;
992 		if (strcmp(pack->path_packfile, path_packfile) == 0)
993 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
994 	}
995 
996 	if (i == nitems(repo->packs) - 1) {
997 		err = got_pack_close(&repo->packs[i - 1]);
998 		if (err)
999 			return err;
1000 		memmove(&repo->packs[1], &repo->packs[0],
1001 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1002 		i = 0;
1003 	}
1004 
1005 	pack = &repo->packs[i];
1006 
1007 	pack->path_packfile = strdup(path_packfile);
1008 	if (pack->path_packfile == NULL) {
1009 		err = got_error_from_errno("strdup");
1010 		goto done;
1011 	}
1012 
1013 	err = open_packfile(&pack->fd, path_packfile, packidx);
1014 	if (err)
1015 		goto done;
1016 
1017 	if (fstat(pack->fd, &sb) != 0) {
1018 		err = got_error_from_errno("fstat");
1019 		goto done;
1020 	}
1021 	pack->filesize = sb.st_size;
1022 
1023 	pack->privsep_child = NULL;
1024 
1025 #ifndef GOT_PACK_NO_MMAP
1026 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1027 	    pack->fd, 0);
1028 	if (pack->map == MAP_FAILED) {
1029 		if (errno != ENOMEM) {
1030 			err = got_error_from_errno("mmap");
1031 			goto done;
1032 		}
1033 		pack->map = NULL; /* fall back to read(2) */
1034 	}
1035 #endif
1036 done:
1037 	if (err) {
1038 		if (pack) {
1039 			free(pack->path_packfile);
1040 			memset(pack, 0, sizeof(*pack));
1041 		}
1042 	} else if (packp)
1043 		*packp = pack;
1044 	return err;
1045 }
1046 
1047 struct got_pack *
1048 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1049 {
1050 	struct got_pack *pack = NULL;
1051 	int i;
1052 
1053 	for (i = 0; i < nitems(repo->packs); i++) {
1054 		pack = &repo->packs[i];
1055 		if (pack->path_packfile == NULL)
1056 			break;
1057 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1058 			return pack;
1059 	}
1060 
1061 	return NULL;
1062 }
1063 
1064 const struct got_error *
1065 got_repo_init(const char *repo_path)
1066 {
1067 	const struct got_error *err = NULL;
1068 	const char *dirnames[] = {
1069 		GOT_OBJECTS_DIR,
1070 		GOT_OBJECTS_PACK_DIR,
1071 		GOT_REFS_DIR,
1072 	};
1073 	const char *description_str = "Unnamed repository; "
1074 	    "edit this file 'description' to name the repository.";
1075 	const char *headref_str = "ref: refs/heads/main";
1076 	const char *gitconfig_str = "[core]\n"
1077 	    "\trepositoryformatversion = 0\n"
1078 	    "\tfilemode = true\n"
1079 	    "\tbare = true\n";
1080 	char *path;
1081 	int i;
1082 
1083 	if (!got_path_dir_is_empty(repo_path))
1084 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1085 
1086 	for (i = 0; i < nitems(dirnames); i++) {
1087 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1088 			return got_error_from_errno("asprintf");
1089 		}
1090 		err = got_path_mkdir(path);
1091 		free(path);
1092 		if (err)
1093 			return err;
1094 	}
1095 
1096 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1097 		return got_error_from_errno("asprintf");
1098 	err = got_path_create_file(path, description_str);
1099 	free(path);
1100 	if (err)
1101 		return err;
1102 
1103 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1104 		return got_error_from_errno("asprintf");
1105 	err = got_path_create_file(path, headref_str);
1106 	free(path);
1107 	if (err)
1108 		return err;
1109 
1110 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1111 		return got_error_from_errno("asprintf");
1112 	err = got_path_create_file(path, gitconfig_str);
1113 	free(path);
1114 	if (err)
1115 		return err;
1116 
1117 	return NULL;
1118 }
1119 
1120 static const struct got_error *
1121 match_packed_object(struct got_object_id **unique_id,
1122     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1123 {
1124 	const struct got_error *err = NULL;
1125 	char *path_packdir;
1126 	DIR *packdir;
1127 	struct dirent *dent;
1128 	char *path_packidx;
1129 	struct got_object_id_queue matched_ids;
1130 
1131 	SIMPLEQ_INIT(&matched_ids);
1132 
1133 	path_packdir = got_repo_get_path_objects_pack(repo);
1134 	if (path_packdir == NULL)
1135 		return got_error_from_errno("got_repo_get_path_objects_pack");
1136 
1137 	packdir = opendir(path_packdir);
1138 	if (packdir == NULL) {
1139 		if (errno != ENOENT)
1140 			err = got_error_from_errno2("opendir", path_packdir);
1141 		goto done;
1142 	}
1143 
1144 	while ((dent = readdir(packdir)) != NULL) {
1145 		struct got_packidx *packidx;
1146 		struct got_object_qid *qid;
1147 
1148 
1149 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1150 			continue;
1151 
1152 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1153 		    dent->d_name) == -1) {
1154 			err = got_error_from_errno("asprintf");
1155 			break;
1156 		}
1157 
1158 		err = got_packidx_open(&packidx, path_packidx, 0);
1159 		free(path_packidx);
1160 		if (err)
1161 			break;
1162 
1163 		err = got_packidx_match_id_str_prefix(&matched_ids,
1164 		    packidx, id_str_prefix);
1165 		if (err) {
1166 			got_packidx_close(packidx);
1167 			break;
1168 		}
1169 		err = got_packidx_close(packidx);
1170 		if (err)
1171 			break;
1172 
1173 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1174 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1175 				int matched_type;
1176 				err = got_object_get_type(&matched_type, repo,
1177 				    qid->id);
1178 				if (err)
1179 					goto done;
1180 				if (matched_type != obj_type)
1181 					continue;
1182 			}
1183 			if (*unique_id == NULL) {
1184 				*unique_id = got_object_id_dup(qid->id);
1185 				if (*unique_id == NULL) {
1186 					err = got_error_from_errno("malloc");
1187 					goto done;
1188 				}
1189 			} else {
1190 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1191 					continue; /* packed multiple times */
1192 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1193 				goto done;
1194 			}
1195 		}
1196 	}
1197 done:
1198 	got_object_id_queue_free(&matched_ids);
1199 	free(path_packdir);
1200 	if (packdir && closedir(packdir) != 0 && err == NULL)
1201 		err = got_error_from_errno("closedir");
1202 	if (err) {
1203 		free(*unique_id);
1204 		*unique_id = NULL;
1205 	}
1206 	return err;
1207 }
1208 
1209 static const struct got_error *
1210 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1211     const char *object_dir, const char *id_str_prefix, int obj_type,
1212     struct got_repository *repo)
1213 {
1214 	const struct got_error *err = NULL;
1215 	char *path;
1216 	DIR *dir = NULL;
1217 	struct dirent *dent;
1218 	struct got_object_id id;
1219 
1220 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1221 		err = got_error_from_errno("asprintf");
1222 		goto done;
1223 	}
1224 
1225 	dir = opendir(path);
1226 	if (dir == NULL) {
1227 		if (errno == ENOENT) {
1228 			err = NULL;
1229 			goto done;
1230 		}
1231 		err = got_error_from_errno2("opendir", path);
1232 		goto done;
1233 	}
1234 	while ((dent = readdir(dir)) != NULL) {
1235 		char *id_str;
1236 		int cmp;
1237 
1238 		if (strcmp(dent->d_name, ".") == 0 ||
1239 		    strcmp(dent->d_name, "..") == 0)
1240 			continue;
1241 
1242 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1243 			err = got_error_from_errno("asprintf");
1244 			goto done;
1245 		}
1246 
1247 		if (!got_parse_sha1_digest(id.sha1, id_str))
1248 			continue;
1249 
1250 		/*
1251 		 * Directory entries do not necessarily appear in
1252 		 * sorted order, so we must iterate over all of them.
1253 		 */
1254 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1255 		if (cmp != 0) {
1256 			free(id_str);
1257 			continue;
1258 		}
1259 
1260 		if (*unique_id == NULL) {
1261 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1262 				int matched_type;
1263 				err = got_object_get_type(&matched_type, repo,
1264 				    &id);
1265 				if (err)
1266 					goto done;
1267 				if (matched_type != obj_type)
1268 					continue;
1269 			}
1270 			*unique_id = got_object_id_dup(&id);
1271 			if (*unique_id == NULL) {
1272 				err = got_error_from_errno("got_object_id_dup");
1273 				free(id_str);
1274 				goto done;
1275 			}
1276 		} else {
1277 			if (got_object_id_cmp(*unique_id, &id) == 0)
1278 				continue; /* both packed and loose */
1279 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1280 			free(id_str);
1281 			goto done;
1282 		}
1283 	}
1284 done:
1285 	if (dir && closedir(dir) != 0 && err == NULL)
1286 		err = got_error_from_errno("closedir");
1287 	if (err) {
1288 		free(*unique_id);
1289 		*unique_id = NULL;
1290 	}
1291 	free(path);
1292 	return err;
1293 }
1294 
1295 const struct got_error *
1296 got_repo_match_object_id_prefix(struct got_object_id **id,
1297     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1298 {
1299 	const struct got_error *err = NULL;
1300 	char *path_objects = got_repo_get_path_objects(repo);
1301 	char *object_dir = NULL;
1302 	size_t len;
1303 	int i;
1304 
1305 	*id = NULL;
1306 
1307 	for (i = 0; i < strlen(id_str_prefix); i++) {
1308 		if (isxdigit((unsigned char)id_str_prefix[i]))
1309 			continue;
1310 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1311 	}
1312 
1313 	len = strlen(id_str_prefix);
1314 	if (len >= 2) {
1315 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1316 		if (err)
1317 			goto done;
1318 		object_dir = strndup(id_str_prefix, 2);
1319 		if (object_dir == NULL) {
1320 			err = got_error_from_errno("strdup");
1321 			goto done;
1322 		}
1323 		err = match_loose_object(id, path_objects, object_dir,
1324 		    id_str_prefix, obj_type, repo);
1325 	} else if (len == 1) {
1326 		int i;
1327 		for (i = 0; i < 0xf; i++) {
1328 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1329 			    == -1) {
1330 				err = got_error_from_errno("asprintf");
1331 				goto done;
1332 			}
1333 			err = match_packed_object(id, repo, object_dir,
1334 			    obj_type);
1335 			if (err)
1336 				goto done;
1337 			err = match_loose_object(id, path_objects, object_dir,
1338 			    id_str_prefix, obj_type, repo);
1339 			if (err)
1340 				goto done;
1341 		}
1342 	} else {
1343 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1344 		goto done;
1345 	}
1346 done:
1347 	free(object_dir);
1348 	if (err) {
1349 		free(*id);
1350 		*id = NULL;
1351 	} else if (*id == NULL)
1352 		err = got_error(GOT_ERR_NO_OBJ);
1353 
1354 	return err;
1355 }
1356 
1357 const struct got_error *
1358 got_repo_match_object_id(struct got_object_id **id, char **label,
1359     const char *id_str, int obj_type, int resolve_tags,
1360     struct got_repository *repo)
1361 {
1362 	const struct got_error *err;
1363 	struct got_tag_object *tag;
1364 	struct got_reference *ref = NULL;
1365 
1366 	*id = NULL;
1367 	if (label)
1368 		*label = NULL;
1369 
1370 	if (resolve_tags) {
1371 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1372 		    repo);
1373 		if (err == NULL) {
1374 			*id = got_object_id_dup(
1375 			    got_object_tag_get_object_id(tag));
1376 			if (*id == NULL)
1377 				err = got_error_from_errno("got_object_id_dup");
1378 			else if (label && asprintf(label, "refs/tags/%s",
1379 			    got_object_tag_get_name(tag)) == -1) {
1380 				err = got_error_from_errno("asprintf");
1381 				free(*id);
1382 				*id = NULL;
1383 			}
1384 			got_object_tag_close(tag);
1385 			return err;
1386 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1387 		    err->code != GOT_ERR_NO_OBJ)
1388 			return err;
1389 	}
1390 
1391 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1392 	if (err) {
1393 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1394 			return err;
1395 		err = got_ref_open(&ref, repo, id_str, 0);
1396 		if (err != NULL)
1397 			goto done;
1398 		if (label) {
1399 			*label = strdup(got_ref_get_name(ref));
1400 			if (*label == NULL) {
1401 				err = got_error_from_errno("strdup");
1402 				goto done;
1403 			}
1404 		}
1405 		err = got_ref_resolve(id, repo, ref);
1406 	} else if (label) {
1407 		err = got_object_id_str(label, *id);
1408 		if (*label == NULL) {
1409 			err = got_error_from_errno("strdup");
1410 			goto done;
1411 		}
1412 	}
1413 done:
1414 	if (ref)
1415 		got_ref_close(ref);
1416 	return err;
1417 }
1418 
1419 const struct got_error *
1420 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1421     int obj_type, struct got_repository *repo)
1422 {
1423 	const struct got_error *err;
1424 	struct got_reflist_head refs;
1425 	struct got_reflist_entry *re;
1426 	struct got_object_id *tag_id;
1427 
1428 	SIMPLEQ_INIT(&refs);
1429 	*tag = NULL;
1430 
1431 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1432 	if (err)
1433 		return err;
1434 
1435 	SIMPLEQ_FOREACH(re, &refs, entry) {
1436 		const char *refname;
1437 		refname = got_ref_get_name(re->ref);
1438 		if (got_ref_is_symbolic(re->ref))
1439 			continue;
1440 		refname += strlen("refs/tags/");
1441 		if (strcmp(refname, name) != 0)
1442 			continue;
1443 		err = got_ref_resolve(&tag_id, repo, re->ref);
1444 		if (err)
1445 			break;
1446 		err = got_object_open_as_tag(tag, repo, tag_id);
1447 		free(tag_id);
1448 		if (err)
1449 			break;
1450 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1451 		    got_object_tag_get_object_type(*tag) == obj_type)
1452 			break;
1453 		got_object_tag_close(*tag);
1454 		*tag = NULL;
1455 	}
1456 
1457 	got_ref_list_free(&refs);
1458 	if (err == NULL && *tag == NULL)
1459 		err = got_error(GOT_ERR_NO_OBJ);
1460 	return err;
1461 }
1462 
1463 static const struct got_error *
1464 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1465     const char *name, mode_t mode, struct got_object_id *blob_id)
1466 {
1467 	const struct got_error *err = NULL;
1468 
1469 	 *new_te = NULL;
1470 
1471 	*new_te = calloc(1, sizeof(**new_te));
1472 	if (*new_te == NULL)
1473 		return got_error_from_errno("calloc");
1474 
1475 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1476 	    sizeof((*new_te)->name)) {
1477 		err = got_error(GOT_ERR_NO_SPACE);
1478 		goto done;
1479 	}
1480 
1481 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1482 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1483 done:
1484 	if (err && *new_te) {
1485 		free(*new_te);
1486 		*new_te = NULL;
1487 	}
1488 	return err;
1489 }
1490 
1491 static const struct got_error *
1492 import_file(struct got_tree_entry **new_te, struct dirent *de,
1493     const char *path, struct got_repository *repo)
1494 {
1495 	const struct got_error *err;
1496 	struct got_object_id *blob_id = NULL;
1497 	char *filepath;
1498 	struct stat sb;
1499 
1500 	if (asprintf(&filepath, "%s%s%s", path,
1501 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1502 		return got_error_from_errno("asprintf");
1503 
1504 	if (lstat(filepath, &sb) != 0) {
1505 		err = got_error_from_errno2("lstat", path);
1506 		goto done;
1507 	}
1508 
1509 	err = got_object_blob_create(&blob_id, filepath, repo);
1510 	if (err)
1511 		goto done;
1512 
1513 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1514 	    blob_id);
1515 done:
1516 	free(filepath);
1517 	if (err)
1518 		free(blob_id);
1519 	return err;
1520 }
1521 
1522 static const struct got_error *
1523 insert_tree_entry(struct got_tree_entry *new_te,
1524     struct got_pathlist_head *paths)
1525 {
1526 	const struct got_error *err = NULL;
1527 	struct got_pathlist_entry *new_pe;
1528 
1529 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1530 	if (err)
1531 		return err;
1532 	if (new_pe == NULL)
1533 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1534 	return NULL;
1535 }
1536 
1537 static const struct got_error *write_tree(struct got_object_id **,
1538     const char *, struct got_pathlist_head *, struct got_repository *,
1539     got_repo_import_cb progress_cb, void *progress_arg);
1540 
1541 static const struct got_error *
1542 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1543     const char *path, struct got_pathlist_head *ignores,
1544     struct got_repository *repo,
1545     got_repo_import_cb progress_cb, void *progress_arg)
1546 {
1547 	const struct got_error *err;
1548 	struct got_object_id *id = NULL;
1549 	char *subdirpath;
1550 
1551 	if (asprintf(&subdirpath, "%s%s%s", path,
1552 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1553 		return got_error_from_errno("asprintf");
1554 
1555 	(*new_te) = calloc(1, sizeof(**new_te));
1556 	if (*new_te == NULL)
1557 		return got_error_from_errno("calloc");
1558 	(*new_te)->mode = S_IFDIR;
1559 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1560 	    sizeof((*new_te)->name)) {
1561 		err = got_error(GOT_ERR_NO_SPACE);
1562 		goto done;
1563 	}
1564 	err = write_tree(&id, subdirpath, ignores,  repo,
1565 	    progress_cb, progress_arg);
1566 	if (err)
1567 		goto done;
1568 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1569 	
1570 done:
1571 	free(id);
1572 	free(subdirpath);
1573 	if (err) {
1574 		free(*new_te);
1575 		*new_te = NULL;
1576 	}
1577 	return err;
1578 }
1579 
1580 static const struct got_error *
1581 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1582     struct got_pathlist_head *ignores, struct got_repository *repo,
1583     got_repo_import_cb progress_cb, void *progress_arg)
1584 {
1585 	const struct got_error *err = NULL;
1586 	DIR *dir;
1587 	struct dirent *de;
1588 	int nentries;
1589 	struct got_tree_entry *new_te = NULL;
1590 	struct got_pathlist_head paths;
1591 	struct got_pathlist_entry *pe;
1592 
1593 	*new_tree_id = NULL;
1594 
1595 	TAILQ_INIT(&paths);
1596 
1597 	dir = opendir(path_dir);
1598 	if (dir == NULL) {
1599 		err = got_error_from_errno2("opendir", path_dir);
1600 		goto done;
1601 	}
1602 
1603 	nentries = 0;
1604 	while ((de = readdir(dir)) != NULL) {
1605 		int ignore = 0;
1606 
1607 		if (strcmp(de->d_name, ".") == 0 ||
1608 		    strcmp(de->d_name, "..") == 0)
1609 			continue;
1610 
1611 		TAILQ_FOREACH(pe, ignores, entry) {
1612 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1613 				ignore = 1;
1614 				break;
1615 			}
1616 		}
1617 		if (ignore)
1618 			continue;
1619 		if (de->d_type == DT_DIR) {
1620 			err = import_subdir(&new_te, de, path_dir,
1621 			    ignores, repo, progress_cb, progress_arg);
1622 			if (err) {
1623 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1624 					goto done;
1625 				err = NULL;
1626 				continue;
1627 			}
1628 		} else if (de->d_type == DT_REG) {
1629 			err = import_file(&new_te, de, path_dir, repo);
1630 			if (err)
1631 				goto done;
1632 		} else
1633 			continue;
1634 
1635 		err = insert_tree_entry(new_te, &paths);
1636 		if (err)
1637 			goto done;
1638 		nentries++;
1639 	}
1640 
1641 	if (TAILQ_EMPTY(&paths)) {
1642 		err = got_error(GOT_ERR_NO_TREE_ENTRY);
1643 		goto done;
1644 	}
1645 
1646 	TAILQ_FOREACH(pe, &paths, entry) {
1647 		struct got_tree_entry *te = pe->data;
1648 		char *path;
1649 		if (!S_ISREG(te->mode))
1650 			continue;
1651 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1652 			err = got_error_from_errno("asprintf");
1653 			goto done;
1654 		}
1655 		err = (*progress_cb)(progress_arg, path);
1656 		free(path);
1657 		if (err)
1658 			goto done;
1659 	}
1660 
1661 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1662 done:
1663 	if (dir)
1664 		closedir(dir);
1665 	got_pathlist_free(&paths);
1666 	return err;
1667 }
1668 
1669 const struct got_error *
1670 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1671     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1672     struct got_repository *repo, got_repo_import_cb progress_cb,
1673     void *progress_arg)
1674 {
1675 	const struct got_error *err;
1676 	struct got_object_id *new_tree_id;
1677 
1678 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1679 	    progress_cb, progress_arg);
1680 	if (err)
1681 		return err;
1682 
1683 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1684 	    author, time(NULL), author, time(NULL), logmsg, repo);
1685 	free(new_tree_id);
1686 	return err;
1687 }