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/resource.h>
24 
25 #include <ctype.h>
26 #include <endian.h>
27 #include <fcntl.h>
28 #include <fnmatch.h>
29 #include <limits.h>
30 #include <dirent.h>
31 #include <stdlib.h>
32 #include <stdio.h>
33 #include <sha1.h>
34 #include <string.h>
35 #include <time.h>
36 #include <unistd.h>
37 #include <zlib.h>
38 #include <errno.h>
39 #include <libgen.h>
40 #include <stdint.h>
41 #include <imsg.h>
42 #include <uuid.h>
43 
44 #include "got_error.h"
45 #include "got_reference.h"
46 #include "got_repository.h"
47 #include "got_path.h"
48 #include "got_cancel.h"
49 #include "got_object.h"
50 
51 #include "got_lib_delta.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 #include "got_lib_gotconfig.h"
62 
63 #ifndef nitems
64 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
65 #endif
66 
67 const char *
68 got_repo_get_path(struct got_repository *repo)
69 {
70 	return repo->path;
71 }
72 
73 const char *
74 got_repo_get_path_git_dir(struct got_repository *repo)
75 {
76 	return repo->path_git_dir;
77 }
78 
79 int
80 got_repo_get_fd(struct got_repository *repo)
81 {
82 	return repo->gitdir_fd;
83 }
84 
85 const char *
86 got_repo_get_gitconfig_author_name(struct got_repository *repo)
87 {
88 	return repo->gitconfig_author_name;
89 }
90 
91 const char *
92 got_repo_get_gitconfig_author_email(struct got_repository *repo)
93 {
94 	return repo->gitconfig_author_email;
95 }
96 
97 const char *
98 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
99 {
100 	return repo->global_gitconfig_author_name;
101 }
102 
103 const char *
104 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
105 {
106 	return repo->global_gitconfig_author_email;
107 }
108 
109 const char *
110 got_repo_get_gitconfig_owner(struct got_repository *repo)
111 {
112 	return repo->gitconfig_owner;
113 }
114 
115 int
116 got_repo_is_bare(struct got_repository *repo)
117 {
118 	return (strcmp(repo->path, repo->path_git_dir) == 0);
119 }
120 
121 static char *
122 get_path_git_child(struct got_repository *repo, const char *basename)
123 {
124 	char *path_child;
125 
126 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
127 	    basename) == -1)
128 		return NULL;
129 
130 	return path_child;
131 }
132 
133 char *
134 got_repo_get_path_objects(struct got_repository *repo)
135 {
136 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
137 }
138 
139 char *
140 got_repo_get_path_objects_pack(struct got_repository *repo)
141 {
142 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
143 }
144 
145 char *
146 got_repo_get_path_refs(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_REFS_DIR);
149 }
150 
151 char *
152 got_repo_get_path_packed_refs(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
155 }
156 
157 static char *
158 get_path_head(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_HEAD_FILE);
161 }
162 
163 char *
164 got_repo_get_path_gitconfig(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_GITCONFIG);
167 }
168 
169 char *
170 got_repo_get_path_gotconfig(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
173 }
174 
175 const struct got_gotconfig *
176 got_repo_get_gotconfig(struct got_repository *repo)
177 {
178 	return repo->gotconfig;
179 }
180 
181 void
182 got_repo_get_gitconfig_remotes(int *nremotes,
183     const struct got_remote_repo **remotes, struct got_repository *repo)
184 {
185 	*nremotes = repo->ngitconfig_remotes;
186 	*remotes = repo->gitconfig_remotes;
187 }
188 
189 static int
190 is_git_repo(struct got_repository *repo)
191 {
192 	const char *path_git = got_repo_get_path_git_dir(repo);
193 	char *path_objects = got_repo_get_path_objects(repo);
194 	char *path_refs = got_repo_get_path_refs(repo);
195 	char *path_head = get_path_head(repo);
196 	int ret = 0;
197 	struct stat sb;
198 	struct got_reference *head_ref;
199 
200 	if (lstat(path_git, &sb) == -1)
201 		goto done;
202 	if (!S_ISDIR(sb.st_mode))
203 		goto done;
204 
205 	if (lstat(path_objects, &sb) == -1)
206 		goto done;
207 	if (!S_ISDIR(sb.st_mode))
208 		goto done;
209 
210 	if (lstat(path_refs, &sb) == -1)
211 		goto done;
212 	if (!S_ISDIR(sb.st_mode))
213 		goto done;
214 
215 	if (lstat(path_head, &sb) == -1)
216 		goto done;
217 	if (!S_ISREG(sb.st_mode))
218 		goto done;
219 
220 	/* Check if the HEAD reference can be opened. */
221 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
222 		goto done;
223 	got_ref_close(head_ref);
224 
225 	ret = 1;
226 done:
227 	free(path_objects);
228 	free(path_refs);
229 	free(path_head);
230 	return ret;
231 
232 }
233 
234 const struct got_error *
235 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
236     struct got_object *obj)
237 {
238 #ifndef GOT_NO_OBJ_CACHE
239 	const struct got_error *err = NULL;
240 	err = got_object_cache_add(&repo->objcache, id, obj);
241 	if (err) {
242 		if (err->code == GOT_ERR_OBJ_EXISTS ||
243 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
244 			err = NULL;
245 		return err;
246 	}
247 	obj->refcnt++;
248 #endif
249 	return NULL;
250 }
251 
252 struct got_object *
253 got_repo_get_cached_object(struct got_repository *repo,
254     struct got_object_id *id)
255 {
256 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
257 }
258 
259 const struct got_error *
260 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
261     struct got_tree_object *tree)
262 {
263 #ifndef GOT_NO_OBJ_CACHE
264 	const struct got_error *err = NULL;
265 	err = got_object_cache_add(&repo->treecache, id, tree);
266 	if (err) {
267 		if (err->code == GOT_ERR_OBJ_EXISTS ||
268 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
269 			err = NULL;
270 		return err;
271 	}
272 	tree->refcnt++;
273 #endif
274 	return NULL;
275 }
276 
277 struct got_tree_object *
278 got_repo_get_cached_tree(struct got_repository *repo,
279     struct got_object_id *id)
280 {
281 	return (struct got_tree_object *)got_object_cache_get(
282 	    &repo->treecache, id);
283 }
284 
285 const struct got_error *
286 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
287     struct got_commit_object *commit)
288 {
289 #ifndef GOT_NO_OBJ_CACHE
290 	const struct got_error *err = NULL;
291 	err = got_object_cache_add(&repo->commitcache, id, commit);
292 	if (err) {
293 		if (err->code == GOT_ERR_OBJ_EXISTS ||
294 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
295 			err = NULL;
296 		return err;
297 	}
298 	commit->refcnt++;
299 #endif
300 	return NULL;
301 }
302 
303 struct got_commit_object *
304 got_repo_get_cached_commit(struct got_repository *repo,
305     struct got_object_id *id)
306 {
307 	return (struct got_commit_object *)got_object_cache_get(
308 	    &repo->commitcache, id);
309 }
310 
311 const struct got_error *
312 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
313     struct got_tag_object *tag)
314 {
315 #ifndef GOT_NO_OBJ_CACHE
316 	const struct got_error *err = NULL;
317 	err = got_object_cache_add(&repo->tagcache, id, tag);
318 	if (err) {
319 		if (err->code == GOT_ERR_OBJ_EXISTS ||
320 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
321 			err = NULL;
322 		return err;
323 	}
324 	tag->refcnt++;
325 #endif
326 	return NULL;
327 }
328 
329 struct got_tag_object *
330 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
331 {
332 	return (struct got_tag_object *)got_object_cache_get(
333 	    &repo->tagcache, id);
334 }
335 
336 const struct got_error *
337 open_repo(struct got_repository *repo, const char *path)
338 {
339 	const struct got_error *err = NULL;
340 
341 	repo->gitdir_fd = -1;
342 
343 	/* bare git repository? */
344 	repo->path_git_dir = strdup(path);
345 	if (repo->path_git_dir == NULL)
346 		return got_error_from_errno("strdup");
347 	if (is_git_repo(repo)) {
348 		repo->path = strdup(repo->path_git_dir);
349 		if (repo->path == NULL) {
350 			err = got_error_from_errno("strdup");
351 			goto done;
352 		}
353 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
354 		if (repo->gitdir_fd == -1) {
355 			err = got_error_from_errno2("open",
356 			    repo->path_git_dir);
357 			goto done;
358 		}
359 		return NULL;
360 	}
361 
362 	/* git repository with working tree? */
363 	free(repo->path_git_dir);
364 	repo->path_git_dir = NULL;
365 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
366 		err = got_error_from_errno("asprintf");
367 		goto done;
368 	}
369 	if (is_git_repo(repo)) {
370 		repo->path = strdup(path);
371 		if (repo->path == NULL) {
372 			err = got_error_from_errno("strdup");
373 			goto done;
374 		}
375 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
376 		if (repo->gitdir_fd == -1) {
377 			err = got_error_from_errno2("open",
378 			    repo->path_git_dir);
379 			goto done;
380 		}
381 		return NULL;
382 	}
383 
384 	err = got_error(GOT_ERR_NOT_GIT_REPO);
385 done:
386 	if (err) {
387 		free(repo->path);
388 		repo->path = NULL;
389 		free(repo->path_git_dir);
390 		repo->path_git_dir = NULL;
391 		if (repo->gitdir_fd != -1)
392 			close(repo->gitdir_fd);
393 		repo->gitdir_fd = -1;
394 
395 	}
396 	return err;
397 }
398 
399 static const struct got_error *
400 parse_gitconfig_file(int *gitconfig_repository_format_version,
401     char **gitconfig_author_name, char **gitconfig_author_email,
402     struct got_remote_repo **remotes, int *nremotes,
403     char **gitconfig_owner, char ***extensions, int *nextensions,
404     const char *gitconfig_path)
405 {
406 	const struct got_error *err = NULL, *child_err = NULL;
407 	int fd = -1;
408 	int imsg_fds[2] = { -1, -1 };
409 	pid_t pid;
410 	struct imsgbuf *ibuf;
411 
412 	*gitconfig_repository_format_version = 0;
413 	if (extensions)
414 		*extensions = NULL;
415 	if (nextensions)
416 		*nextensions = 0;
417 	*gitconfig_author_name = NULL;
418 	*gitconfig_author_email = NULL;
419 	if (remotes)
420 		*remotes = NULL;
421 	if (nremotes)
422 		*nremotes = 0;
423 	if (gitconfig_owner)
424 		*gitconfig_owner = NULL;
425 
426 	fd = open(gitconfig_path, O_RDONLY);
427 	if (fd == -1) {
428 		if (errno == ENOENT)
429 			return NULL;
430 		return got_error_from_errno2("open", gitconfig_path);
431 	}
432 
433 	ibuf = calloc(1, sizeof(*ibuf));
434 	if (ibuf == NULL) {
435 		err = got_error_from_errno("calloc");
436 		goto done;
437 	}
438 
439 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
440 		err = got_error_from_errno("socketpair");
441 		goto done;
442 	}
443 
444 	pid = fork();
445 	if (pid == -1) {
446 		err = got_error_from_errno("fork");
447 		goto done;
448 	} else if (pid == 0) {
449 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
450 		    gitconfig_path);
451 		/* not reached */
452 	}
453 
454 	if (close(imsg_fds[1]) == -1) {
455 		err = got_error_from_errno("close");
456 		goto done;
457 	}
458 	imsg_fds[1] = -1;
459 	imsg_init(ibuf, imsg_fds[0]);
460 
461 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
462 	if (err)
463 		goto done;
464 	fd = -1;
465 
466 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
467 	if (err)
468 		goto done;
469 
470 	err = got_privsep_recv_gitconfig_int(
471 	    gitconfig_repository_format_version, ibuf);
472 	if (err)
473 		goto done;
474 
475 	if (extensions && nextensions) {
476 		err = got_privsep_send_gitconfig_repository_extensions_req(
477 		    ibuf);
478 		if (err)
479 			goto done;
480 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
481 		if (err)
482 			goto done;
483 		if (*nextensions > 0) {
484 			int i;
485 			*extensions = calloc(*nextensions, sizeof(char *));
486 			if (*extensions == NULL) {
487 				err = got_error_from_errno("calloc");
488 				goto done;
489 			}
490 			for (i = 0; i < *nextensions; i++) {
491 				char *ext;
492 				err = got_privsep_recv_gitconfig_str(&ext,
493 				    ibuf);
494 				if (err)
495 					goto done;
496 				(*extensions)[i] = ext;
497 			}
498 		}
499 	}
500 
501 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
502 	if (err)
503 		goto done;
504 
505 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
506 	if (err)
507 		goto done;
508 
509 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
510 	if (err)
511 		goto done;
512 
513 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
514 	if (err)
515 		goto done;
516 
517 	if (remotes && nremotes) {
518 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
519 		if (err)
520 			goto done;
521 
522 		err = got_privsep_recv_gitconfig_remotes(remotes,
523 		    nremotes, ibuf);
524 		if (err)
525 			goto done;
526 	}
527 
528 	if (gitconfig_owner) {
529 		err = got_privsep_send_gitconfig_owner_req(ibuf);
530 		if (err)
531 			goto done;
532 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
533 		if (err)
534 			goto done;
535 	}
536 
537 	imsg_clear(ibuf);
538 	err = got_privsep_send_stop(imsg_fds[0]);
539 	child_err = got_privsep_wait_for_child(pid);
540 	if (child_err && err == NULL)
541 		err = child_err;
542 done:
543 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
544 		err = got_error_from_errno("close");
545 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
546 		err = got_error_from_errno("close");
547 	if (fd != -1 && close(fd) == -1 && err == NULL)
548 		err = got_error_from_errno2("close", gitconfig_path);
549 	free(ibuf);
550 	return err;
551 }
552 
553 static const struct got_error *
554 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
555 {
556 	const struct got_error *err = NULL;
557 	char *repo_gitconfig_path = NULL;
558 
559 	if (global_gitconfig_path) {
560 		/* Read settings from ~/.gitconfig. */
561 		int dummy_repo_version;
562 		err = parse_gitconfig_file(&dummy_repo_version,
563 		    &repo->global_gitconfig_author_name,
564 		    &repo->global_gitconfig_author_email,
565 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
566 		if (err)
567 			return err;
568 	}
569 
570 	/* Read repository's .git/config file. */
571 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
572 	if (repo_gitconfig_path == NULL)
573 		return got_error_from_errno("got_repo_get_path_gitconfig");
574 
575 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
576 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
577 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
578 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
579 	    repo_gitconfig_path);
580 	if (err)
581 		goto done;
582 done:
583 	free(repo_gitconfig_path);
584 	return err;
585 }
586 
587 static const struct got_error *
588 read_gotconfig(struct got_repository *repo)
589 {
590 	const struct got_error *err = NULL;
591 	char *gotconfig_path;
592 
593 	gotconfig_path = got_repo_get_path_gotconfig(repo);
594 	if (gotconfig_path == NULL)
595 		return got_error_from_errno("got_repo_get_path_gotconfig");
596 
597 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
598 	free(gotconfig_path);
599 	return err;
600 }
601 
602 /* Supported repository format extensions. */
603 static const char *repo_extensions[] = {
604 	"noop",			/* Got supports repository format version 1. */
605 	"preciousObjects",	/* Got has no garbage collection yet. */
606 	"worktreeConfig",	/* Got does not care about Git work trees. */
607 };
608 
609 const struct got_error *
610 got_repo_open(struct got_repository **repop, const char *path,
611     const char *global_gitconfig_path)
612 {
613 	struct got_repository *repo = NULL;
614 	const struct got_error *err = NULL;
615 	char *repo_path = NULL;
616 	size_t i;
617 	struct rlimit rl;
618 
619 	*repop = NULL;
620 
621 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
622 		return got_error_from_errno("getrlimit");
623 
624 	repo = calloc(1, sizeof(*repo));
625 	if (repo == NULL) {
626 		err = got_error_from_errno("calloc");
627 		goto done;
628 	}
629 
630 	for (i = 0; i < nitems(repo->privsep_children); i++) {
631 		memset(&repo->privsep_children[i], 0,
632 		    sizeof(repo->privsep_children[0]));
633 		repo->privsep_children[i].imsg_fd = -1;
634 	}
635 
636 	err = got_object_cache_init(&repo->objcache,
637 	    GOT_OBJECT_CACHE_TYPE_OBJ);
638 	if (err)
639 		goto done;
640 	err = got_object_cache_init(&repo->treecache,
641 	    GOT_OBJECT_CACHE_TYPE_TREE);
642 	if (err)
643 		goto done;
644 	err = got_object_cache_init(&repo->commitcache,
645 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
646 	if (err)
647 		goto done;
648 	err = got_object_cache_init(&repo->tagcache,
649 	    GOT_OBJECT_CACHE_TYPE_TAG);
650 	if (err)
651 		goto done;
652 
653 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
654 	if (repo->pack_cache_size > rl.rlim_cur / 8)
655 		repo->pack_cache_size = rl.rlim_cur / 8;
656 
657 	repo_path = realpath(path, NULL);
658 	if (repo_path == NULL) {
659 		err = got_error_from_errno2("realpath", path);
660 		goto done;
661 	}
662 
663 	for (;;) {
664 		char *parent_path;
665 
666 		err = open_repo(repo, repo_path);
667 		if (err == NULL)
668 			break;
669 		if (err->code != GOT_ERR_NOT_GIT_REPO)
670 			goto done;
671 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
672 			err = got_error(GOT_ERR_NOT_GIT_REPO);
673 			goto done;
674 		}
675 		err = got_path_dirname(&parent_path, repo_path);
676 		if (err)
677 			goto done;
678 		free(repo_path);
679 		repo_path = parent_path;
680 	}
681 
682 	err = read_gotconfig(repo);
683 	if (err)
684 		goto done;
685 
686 	err = read_gitconfig(repo, global_gitconfig_path);
687 	if (err)
688 		goto done;
689 	if (repo->gitconfig_repository_format_version != 0)
690 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
691 	for (i = 0; i < repo->nextensions; i++) {
692 		char *ext = repo->extensions[i];
693 		int j, supported = 0;
694 		for (j = 0; j < nitems(repo_extensions); j++) {
695 			if (strcmp(ext, repo_extensions[j]) == 0) {
696 				supported = 1;
697 				break;
698 			}
699 		}
700 		if (!supported) {
701 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
702 			goto done;
703 		}
704 	}
705 done:
706 	if (err)
707 		got_repo_close(repo);
708 	else
709 		*repop = repo;
710 	free(repo_path);
711 	return err;
712 }
713 
714 const struct got_error *
715 got_repo_close(struct got_repository *repo)
716 {
717 	const struct got_error *err = NULL, *child_err;
718 	size_t i;
719 
720 	for (i = 0; i < repo->pack_cache_size; i++) {
721 		if (repo->packidx_cache[i] == NULL)
722 			break;
723 		got_packidx_close(repo->packidx_cache[i]);
724 	}
725 
726 	for (i = 0; i < repo->pack_cache_size; i++) {
727 		if (repo->packs[i].path_packfile == NULL)
728 			break;
729 		got_pack_close(&repo->packs[i]);
730 	}
731 
732 	free(repo->path);
733 	free(repo->path_git_dir);
734 
735 	got_object_cache_close(&repo->objcache);
736 	got_object_cache_close(&repo->treecache);
737 	got_object_cache_close(&repo->commitcache);
738 	got_object_cache_close(&repo->tagcache);
739 
740 	for (i = 0; i < nitems(repo->privsep_children); i++) {
741 		if (repo->privsep_children[i].imsg_fd == -1)
742 			continue;
743 		imsg_clear(repo->privsep_children[i].ibuf);
744 		free(repo->privsep_children[i].ibuf);
745 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
746 		child_err = got_privsep_wait_for_child(
747 		    repo->privsep_children[i].pid);
748 		if (child_err && err == NULL)
749 			err = child_err;
750 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
751 		    err == NULL)
752 			err = got_error_from_errno("close");
753 	}
754 
755 	if (repo->gotconfig)
756 		got_gotconfig_free(repo->gotconfig);
757 	free(repo->gitconfig_author_name);
758 	free(repo->gitconfig_author_email);
759 	for (i = 0; i < repo->ngitconfig_remotes; i++)
760 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
761 	free(repo->gitconfig_remotes);
762 	for (i = 0; i < repo->nextensions; i++)
763 		free(repo->extensions[i]);
764 	free(repo->extensions);
765 	free(repo);
766 
767 	return err;
768 }
769 
770 void
771 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
772 {
773 	int i;
774 
775 	free(repo->name);
776 	repo->name = NULL;
777 	free(repo->url);
778 	repo->url = NULL;
779 	for (i = 0; i < repo->nbranches; i++)
780 		free(repo->branches[i]);
781 	free(repo->branches);
782 	repo->branches = NULL;
783 	repo->nbranches = 0;
784 }
785 
786 const struct got_error *
787 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
788     const char *input_path)
789 {
790 	const struct got_error *err = NULL;
791 	const char *repo_abspath = NULL;
792 	size_t repolen, len;
793 	char *canonpath, *path = NULL;
794 
795 	*in_repo_path = NULL;
796 
797 	canonpath = strdup(input_path);
798 	if (canonpath == NULL) {
799 		err = got_error_from_errno("strdup");
800 		goto done;
801 	}
802 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
803 	if (err)
804 		goto done;
805 
806 	repo_abspath = got_repo_get_path(repo);
807 
808 	if (canonpath[0] == '\0') {
809 		path = strdup(canonpath);
810 		if (path == NULL) {
811 			err = got_error_from_errno("strdup");
812 			goto done;
813 		}
814 	} else {
815 		path = realpath(canonpath, NULL);
816 		if (path == NULL) {
817 			if (errno != ENOENT) {
818 				err = got_error_from_errno2("realpath",
819 				    canonpath);
820 				goto done;
821 			}
822 			/*
823 			 * Path is not on disk.
824 			 * Assume it is already relative to repository root.
825 			 */
826 			path = strdup(canonpath);
827 			if (path == NULL) {
828 				err = got_error_from_errno("strdup");
829 				goto done;
830 			}
831 		}
832 
833 		repolen = strlen(repo_abspath);
834 		len = strlen(path);
835 
836 
837 		if (strcmp(path, repo_abspath) == 0) {
838 			free(path);
839 			path = strdup("");
840 			if (path == NULL) {
841 				err = got_error_from_errno("strdup");
842 				goto done;
843 			}
844 		} else if (len > repolen &&
845 		    got_path_is_child(path, repo_abspath, repolen)) {
846 			/* Matched an on-disk path inside repository. */
847 			if (got_repo_is_bare(repo)) {
848 				/*
849 				 * Matched an on-disk path inside repository
850 				 * database. Treat input as repository-relative.
851 				 */
852 				 free(path);
853 				 path = canonpath;
854 				 canonpath = NULL;
855 			} else {
856 				char *child;
857 				/* Strip common prefix with repository path. */
858 				err = got_path_skip_common_ancestor(&child,
859 				    repo_abspath, path);
860 				if (err)
861 					goto done;
862 				free(path);
863 				path = child;
864 			}
865 		} else {
866 			/*
867 			 * Matched unrelated on-disk path.
868 			 * Treat input as repository-relative.
869 			 */
870 			 free(path);
871 			 path = canonpath;
872 			 canonpath = NULL;
873 		}
874 	}
875 
876 	/* Make in-repository path absolute */
877 	if (path[0] != '/') {
878 		char *abspath;
879 		if (asprintf(&abspath, "/%s", path) == -1) {
880 			err = got_error_from_errno("asprintf");
881 			goto done;
882 		}
883 		free(path);
884 		path = abspath;
885 	}
886 
887 done:
888 	free(canonpath);
889 	if (err)
890 		free(path);
891 	else
892 		*in_repo_path = path;
893 	return err;
894 }
895 
896 static const struct got_error *
897 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
898     const char *path_packidx)
899 {
900 	const struct got_error *err = NULL;
901 	size_t i;
902 
903 	for (i = 0; i < repo->pack_cache_size; i++) {
904 		if (repo->packidx_cache[i] == NULL)
905 			break;
906 		if (strcmp(repo->packidx_cache[i]->path_packidx,
907 		    path_packidx) == 0) {
908 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
909 		}
910 	}
911 	if (i == repo->pack_cache_size) {
912 		err = got_packidx_close(repo->packidx_cache[i - 1]);
913 		if (err)
914 			return err;
915 	}
916 
917 	/*
918 	 * Insert the new pack index at the front so it will
919 	 * be searched first in the future.
920 	 */
921 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
922 	    sizeof(repo->packidx_cache) -
923 	    sizeof(repo->packidx_cache[0]));
924 	repo->packidx_cache[0] = packidx;
925 
926 	return NULL;
927 }
928 
929 static int
930 is_packidx_filename(const char *name, size_t len)
931 {
932 	if (len != GOT_PACKIDX_NAMELEN)
933 		return 0;
934 
935 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
936 		return 0;
937 
938 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
939 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
940 		return 0;
941 
942 	return 1;
943 }
944 
945 const struct got_error *
946 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
947     struct got_repository *repo, struct got_object_id *id)
948 {
949 	const struct got_error *err;
950 	DIR *packdir = NULL;
951 	struct dirent *dent;
952 	char *path_packidx;
953 	size_t i;
954 	int packdir_fd;
955 
956 	/* Search pack index cache. */
957 	for (i = 0; i < repo->pack_cache_size; i++) {
958 		if (repo->packidx_cache[i] == NULL)
959 			break;
960 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
961 		if (*idx != -1) {
962 			*packidx = repo->packidx_cache[i];
963 			/*
964 			 * Move this cache entry to the front. Repeatedly
965 			 * searching a wrong pack index can be expensive.
966 			 */
967 			if (i > 0) {
968 				struct got_packidx *p;
969 				p = repo->packidx_cache[0];
970 				repo->packidx_cache[0] = *packidx;
971 				repo->packidx_cache[i] = p;
972 			}
973 			return NULL;
974 		}
975 	}
976 	/* No luck. Search the filesystem. */
977 
978 	packdir_fd = openat(got_repo_get_fd(repo),
979 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
980 	if (packdir_fd == -1) {
981 		if (errno == ENOENT)
982 			err = got_error_no_obj(id);
983 		else
984 			err = got_error_from_errno_fmt("openat: %s/%s",
985 			    got_repo_get_path_git_dir(repo),
986 			    GOT_OBJECTS_PACK_DIR);
987 		goto done;
988 	}
989 
990 	packdir = fdopendir(packdir_fd);
991 	if (packdir == NULL) {
992 		err = got_error_from_errno("fdopendir");
993 		goto done;
994 	}
995 
996 	while ((dent = readdir(packdir)) != NULL) {
997 		int is_cached = 0;
998 
999 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1000 			continue;
1001 
1002 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1003 		    dent->d_name) == -1) {
1004 			err = got_error_from_errno("asprintf");
1005 			goto done;
1006 		}
1007 
1008 		for (i = 0; i < repo->pack_cache_size; i++) {
1009 			if (repo->packidx_cache[i] == NULL)
1010 				break;
1011 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1012 			    path_packidx) == 0) {
1013 				is_cached = 1;
1014 				break;
1015 			}
1016 		}
1017 		if (is_cached) {
1018 			free(path_packidx);
1019 			continue; /* already searched */
1020 		}
1021 
1022 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1023 		    path_packidx, 0);
1024 		if (err) {
1025 			free(path_packidx);
1026 			goto done;
1027 		}
1028 
1029 		err = cache_packidx(repo, *packidx, path_packidx);
1030 		free(path_packidx);
1031 		if (err)
1032 			goto done;
1033 
1034 		*idx = got_packidx_get_object_idx(*packidx, id);
1035 		if (*idx != -1) {
1036 			err = NULL; /* found the object */
1037 			goto done;
1038 		}
1039 	}
1040 
1041 	err = got_error_no_obj(id);
1042 done:
1043 	if (packdir && closedir(packdir) != 0 && err == NULL)
1044 		err = got_error_from_errno("closedir");
1045 	return err;
1046 }
1047 
1048 static const struct got_error *
1049 read_packfile_hdr(int fd, struct got_packidx *packidx)
1050 {
1051 	const struct got_error *err = NULL;
1052 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1053 	struct got_packfile_hdr hdr;
1054 	ssize_t n;
1055 
1056 	n = read(fd, &hdr, sizeof(hdr));
1057 	if (n < 0)
1058 		return got_error_from_errno("read");
1059 	if (n != sizeof(hdr))
1060 		return got_error(GOT_ERR_BAD_PACKFILE);
1061 
1062 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1063 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1064 	    be32toh(hdr.nobjects) != totobj)
1065 		err = got_error(GOT_ERR_BAD_PACKFILE);
1066 
1067 	return err;
1068 }
1069 
1070 static const struct got_error *
1071 open_packfile(int *fd, struct got_repository *repo,
1072     const char *relpath, struct got_packidx *packidx)
1073 {
1074 	const struct got_error *err = NULL;
1075 
1076 	*fd = openat(got_repo_get_fd(repo), relpath, O_RDONLY | O_NOFOLLOW);
1077 	if (*fd == -1)
1078 		return got_error_from_errno_fmt("openat: %s/%s",
1079 		    got_repo_get_path_git_dir(repo), relpath);
1080 
1081 	if (packidx) {
1082 		err = read_packfile_hdr(*fd, packidx);
1083 		if (err) {
1084 			close(*fd);
1085 			*fd = -1;
1086 		}
1087 	}
1088 
1089 	return err;
1090 }
1091 
1092 const struct got_error *
1093 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1094     const char *path_packfile, struct got_packidx *packidx)
1095 {
1096 	const struct got_error *err = NULL;
1097 	struct got_pack *pack = NULL;
1098 	struct stat sb;
1099 	size_t i;
1100 
1101 	if (packp)
1102 		*packp = NULL;
1103 
1104 	for (i = 0; i < repo->pack_cache_size; i++) {
1105 		pack = &repo->packs[i];
1106 		if (pack->path_packfile == NULL)
1107 			break;
1108 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1109 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1110 	}
1111 
1112 	if (i == repo->pack_cache_size) {
1113 		err = got_pack_close(&repo->packs[i - 1]);
1114 		if (err)
1115 			return err;
1116 		memmove(&repo->packs[1], &repo->packs[0],
1117 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1118 		i = 0;
1119 	}
1120 
1121 	pack = &repo->packs[i];
1122 
1123 	pack->path_packfile = strdup(path_packfile);
1124 	if (pack->path_packfile == NULL) {
1125 		err = got_error_from_errno("strdup");
1126 		goto done;
1127 	}
1128 
1129 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1130 	if (err)
1131 		goto done;
1132 
1133 	if (fstat(pack->fd, &sb) != 0) {
1134 		err = got_error_from_errno("fstat");
1135 		goto done;
1136 	}
1137 	pack->filesize = sb.st_size;
1138 
1139 	pack->privsep_child = NULL;
1140 
1141 #ifndef GOT_PACK_NO_MMAP
1142 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1143 	    pack->fd, 0);
1144 	if (pack->map == MAP_FAILED) {
1145 		if (errno != ENOMEM) {
1146 			err = got_error_from_errno("mmap");
1147 			goto done;
1148 		}
1149 		pack->map = NULL; /* fall back to read(2) */
1150 	}
1151 #endif
1152 done:
1153 	if (err) {
1154 		if (pack) {
1155 			free(pack->path_packfile);
1156 			memset(pack, 0, sizeof(*pack));
1157 		}
1158 	} else if (packp)
1159 		*packp = pack;
1160 	return err;
1161 }
1162 
1163 struct got_pack *
1164 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1165 {
1166 	struct got_pack *pack = NULL;
1167 	size_t i;
1168 
1169 	for (i = 0; i < repo->pack_cache_size; i++) {
1170 		pack = &repo->packs[i];
1171 		if (pack->path_packfile == NULL)
1172 			break;
1173 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1174 			return pack;
1175 	}
1176 
1177 	return NULL;
1178 }
1179 
1180 const struct got_error *
1181 got_repo_init(const char *repo_path)
1182 {
1183 	const struct got_error *err = NULL;
1184 	const char *dirnames[] = {
1185 		GOT_OBJECTS_DIR,
1186 		GOT_OBJECTS_PACK_DIR,
1187 		GOT_REFS_DIR,
1188 	};
1189 	const char *description_str = "Unnamed repository; "
1190 	    "edit this file 'description' to name the repository.";
1191 	const char *headref_str = "ref: refs/heads/main";
1192 	const char *gitconfig_str = "[core]\n"
1193 	    "\trepositoryformatversion = 0\n"
1194 	    "\tfilemode = true\n"
1195 	    "\tbare = true\n";
1196 	char *path;
1197 	size_t i;
1198 
1199 	if (!got_path_dir_is_empty(repo_path))
1200 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1201 
1202 	for (i = 0; i < nitems(dirnames); i++) {
1203 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1204 			return got_error_from_errno("asprintf");
1205 		}
1206 		err = got_path_mkdir(path);
1207 		free(path);
1208 		if (err)
1209 			return err;
1210 	}
1211 
1212 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1213 		return got_error_from_errno("asprintf");
1214 	err = got_path_create_file(path, description_str);
1215 	free(path);
1216 	if (err)
1217 		return err;
1218 
1219 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1220 		return got_error_from_errno("asprintf");
1221 	err = got_path_create_file(path, headref_str);
1222 	free(path);
1223 	if (err)
1224 		return err;
1225 
1226 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1227 		return got_error_from_errno("asprintf");
1228 	err = got_path_create_file(path, gitconfig_str);
1229 	free(path);
1230 	if (err)
1231 		return err;
1232 
1233 	return NULL;
1234 }
1235 
1236 static const struct got_error *
1237 match_packed_object(struct got_object_id **unique_id,
1238     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1239 {
1240 	const struct got_error *err = NULL;
1241 	DIR *packdir = NULL;
1242 	struct dirent *dent;
1243 	char *path_packidx;
1244 	struct got_object_id_queue matched_ids;
1245 	int packdir_fd;
1246 
1247 	SIMPLEQ_INIT(&matched_ids);
1248 
1249 	packdir_fd = openat(got_repo_get_fd(repo),
1250 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1251 	if (packdir_fd == -1) {
1252 		if (errno != ENOENT)
1253 			err = got_error_from_errno2("openat", GOT_OBJECTS_PACK_DIR);
1254 		goto done;
1255 	}
1256 
1257 	packdir = fdopendir(packdir_fd);
1258 	if (packdir == NULL) {
1259 		err = got_error_from_errno("fdopendir");
1260 		goto done;
1261 	}
1262 
1263 	while ((dent = readdir(packdir)) != NULL) {
1264 		struct got_packidx *packidx;
1265 		struct got_object_qid *qid;
1266 
1267 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1268 			continue;
1269 
1270 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1271 		    dent->d_name) == -1) {
1272 			err = got_error_from_errno("strdup");
1273 			break;
1274 		}
1275 
1276 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1277 		    path_packidx, 0);
1278 		free(path_packidx);
1279 		if (err)
1280 			break;
1281 
1282 		err = got_packidx_match_id_str_prefix(&matched_ids,
1283 		    packidx, id_str_prefix);
1284 		if (err) {
1285 			got_packidx_close(packidx);
1286 			break;
1287 		}
1288 		err = got_packidx_close(packidx);
1289 		if (err)
1290 			break;
1291 
1292 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1293 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1294 				int matched_type;
1295 				err = got_object_get_type(&matched_type, repo,
1296 				    qid->id);
1297 				if (err)
1298 					goto done;
1299 				if (matched_type != obj_type)
1300 					continue;
1301 			}
1302 			if (*unique_id == NULL) {
1303 				*unique_id = got_object_id_dup(qid->id);
1304 				if (*unique_id == NULL) {
1305 					err = got_error_from_errno("malloc");
1306 					goto done;
1307 				}
1308 			} else {
1309 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1310 					continue; /* packed multiple times */
1311 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1312 				goto done;
1313 			}
1314 		}
1315 	}
1316 done:
1317 	got_object_id_queue_free(&matched_ids);
1318 	if (packdir && closedir(packdir) != 0 && err == NULL)
1319 		err = got_error_from_errno("closedir");
1320 	if (err) {
1321 		free(*unique_id);
1322 		*unique_id = NULL;
1323 	}
1324 	return err;
1325 }
1326 
1327 static const struct got_error *
1328 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1329     const char *object_dir, const char *id_str_prefix, int obj_type,
1330     struct got_repository *repo)
1331 {
1332 	const struct got_error *err = NULL;
1333 	char *path;
1334 	DIR *dir = NULL;
1335 	struct dirent *dent;
1336 	struct got_object_id id;
1337 
1338 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1339 		err = got_error_from_errno("asprintf");
1340 		goto done;
1341 	}
1342 
1343 	dir = opendir(path);
1344 	if (dir == NULL) {
1345 		if (errno == ENOENT) {
1346 			err = NULL;
1347 			goto done;
1348 		}
1349 		err = got_error_from_errno2("opendir", path);
1350 		goto done;
1351 	}
1352 	while ((dent = readdir(dir)) != NULL) {
1353 		char *id_str;
1354 		int cmp;
1355 
1356 		if (strcmp(dent->d_name, ".") == 0 ||
1357 		    strcmp(dent->d_name, "..") == 0)
1358 			continue;
1359 
1360 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1361 			err = got_error_from_errno("asprintf");
1362 			goto done;
1363 		}
1364 
1365 		if (!got_parse_sha1_digest(id.sha1, id_str))
1366 			continue;
1367 
1368 		/*
1369 		 * Directory entries do not necessarily appear in
1370 		 * sorted order, so we must iterate over all of them.
1371 		 */
1372 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1373 		if (cmp != 0) {
1374 			free(id_str);
1375 			continue;
1376 		}
1377 
1378 		if (*unique_id == NULL) {
1379 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1380 				int matched_type;
1381 				err = got_object_get_type(&matched_type, repo,
1382 				    &id);
1383 				if (err)
1384 					goto done;
1385 				if (matched_type != obj_type)
1386 					continue;
1387 			}
1388 			*unique_id = got_object_id_dup(&id);
1389 			if (*unique_id == NULL) {
1390 				err = got_error_from_errno("got_object_id_dup");
1391 				free(id_str);
1392 				goto done;
1393 			}
1394 		} else {
1395 			if (got_object_id_cmp(*unique_id, &id) == 0)
1396 				continue; /* both packed and loose */
1397 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1398 			free(id_str);
1399 			goto done;
1400 		}
1401 	}
1402 done:
1403 	if (dir && closedir(dir) != 0 && err == NULL)
1404 		err = got_error_from_errno("closedir");
1405 	if (err) {
1406 		free(*unique_id);
1407 		*unique_id = NULL;
1408 	}
1409 	free(path);
1410 	return err;
1411 }
1412 
1413 const struct got_error *
1414 got_repo_match_object_id_prefix(struct got_object_id **id,
1415     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1416 {
1417 	const struct got_error *err = NULL;
1418 	char *path_objects = got_repo_get_path_objects(repo);
1419 	char *object_dir = NULL;
1420 	size_t len;
1421 	int i;
1422 
1423 	*id = NULL;
1424 
1425 	for (i = 0; i < strlen(id_str_prefix); i++) {
1426 		if (isxdigit((unsigned char)id_str_prefix[i]))
1427 			continue;
1428 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1429 	}
1430 
1431 	len = strlen(id_str_prefix);
1432 	if (len >= 2) {
1433 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1434 		if (err)
1435 			goto done;
1436 		object_dir = strndup(id_str_prefix, 2);
1437 		if (object_dir == NULL) {
1438 			err = got_error_from_errno("strdup");
1439 			goto done;
1440 		}
1441 		err = match_loose_object(id, path_objects, object_dir,
1442 		    id_str_prefix, obj_type, repo);
1443 	} else if (len == 1) {
1444 		int i;
1445 		for (i = 0; i < 0xf; i++) {
1446 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1447 			    == -1) {
1448 				err = got_error_from_errno("asprintf");
1449 				goto done;
1450 			}
1451 			err = match_packed_object(id, repo, object_dir,
1452 			    obj_type);
1453 			if (err)
1454 				goto done;
1455 			err = match_loose_object(id, path_objects, object_dir,
1456 			    id_str_prefix, obj_type, repo);
1457 			if (err)
1458 				goto done;
1459 		}
1460 	} else {
1461 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1462 		goto done;
1463 	}
1464 done:
1465 	free(object_dir);
1466 	if (err) {
1467 		free(*id);
1468 		*id = NULL;
1469 	} else if (*id == NULL)
1470 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1471 
1472 	return err;
1473 }
1474 
1475 const struct got_error *
1476 got_repo_match_object_id(struct got_object_id **id, char **label,
1477     const char *id_str, int obj_type, struct got_reflist_head *refs,
1478     struct got_repository *repo)
1479 {
1480 	const struct got_error *err;
1481 	struct got_tag_object *tag;
1482 	struct got_reference *ref = NULL;
1483 
1484 	*id = NULL;
1485 	if (label)
1486 		*label = NULL;
1487 
1488 	if (refs) {
1489 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1490 		    refs, repo);
1491 		if (err == NULL) {
1492 			*id = got_object_id_dup(
1493 			    got_object_tag_get_object_id(tag));
1494 			if (*id == NULL)
1495 				err = got_error_from_errno("got_object_id_dup");
1496 			else if (label && asprintf(label, "refs/tags/%s",
1497 			    got_object_tag_get_name(tag)) == -1) {
1498 				err = got_error_from_errno("asprintf");
1499 				free(*id);
1500 				*id = NULL;
1501 			}
1502 			got_object_tag_close(tag);
1503 			return err;
1504 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1505 		    err->code != GOT_ERR_NO_OBJ)
1506 			return err;
1507 	}
1508 
1509 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1510 	if (err) {
1511 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1512 			return err;
1513 		err = got_ref_open(&ref, repo, id_str, 0);
1514 		if (err != NULL)
1515 			goto done;
1516 		if (label) {
1517 			*label = strdup(got_ref_get_name(ref));
1518 			if (*label == NULL) {
1519 				err = got_error_from_errno("strdup");
1520 				goto done;
1521 			}
1522 		}
1523 		err = got_ref_resolve(id, repo, ref);
1524 	} else if (label) {
1525 		err = got_object_id_str(label, *id);
1526 		if (*label == NULL) {
1527 			err = got_error_from_errno("strdup");
1528 			goto done;
1529 		}
1530 	}
1531 done:
1532 	if (ref)
1533 		got_ref_close(ref);
1534 	return err;
1535 }
1536 
1537 const struct got_error *
1538 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1539     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1540 {
1541 	const struct got_error *err = NULL;
1542 	struct got_reflist_entry *re;
1543 	struct got_object_id *tag_id;
1544 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1545 
1546 	*tag = NULL;
1547 
1548 	TAILQ_FOREACH(re, refs, entry) {
1549 		const char *refname;
1550 		refname = got_ref_get_name(re->ref);
1551 		if (got_ref_is_symbolic(re->ref))
1552 			continue;
1553 		if (strncmp(refname, "refs/tags/", 10) != 0)
1554 			continue;
1555 		if (!name_is_absolute)
1556 			refname += strlen("refs/tags/");
1557 		if (strcmp(refname, name) != 0)
1558 			continue;
1559 		err = got_ref_resolve(&tag_id, repo, re->ref);
1560 		if (err)
1561 			break;
1562 		err = got_object_open_as_tag(tag, repo, tag_id);
1563 		free(tag_id);
1564 		if (err)
1565 			break;
1566 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1567 		    got_object_tag_get_object_type(*tag) == obj_type)
1568 			break;
1569 		got_object_tag_close(*tag);
1570 		*tag = NULL;
1571 	}
1572 
1573 	if (err == NULL && *tag == NULL)
1574 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1575 	return err;
1576 }
1577 
1578 static const struct got_error *
1579 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1580     const char *name, mode_t mode, struct got_object_id *blob_id)
1581 {
1582 	const struct got_error *err = NULL;
1583 
1584 	 *new_te = NULL;
1585 
1586 	*new_te = calloc(1, sizeof(**new_te));
1587 	if (*new_te == NULL)
1588 		return got_error_from_errno("calloc");
1589 
1590 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1591 	    sizeof((*new_te)->name)) {
1592 		err = got_error(GOT_ERR_NO_SPACE);
1593 		goto done;
1594 	}
1595 
1596 	if (S_ISLNK(mode)) {
1597 		(*new_te)->mode = S_IFLNK;
1598 	} else {
1599 		(*new_te)->mode = S_IFREG;
1600 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1601 	}
1602 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1603 done:
1604 	if (err && *new_te) {
1605 		free(*new_te);
1606 		*new_te = NULL;
1607 	}
1608 	return err;
1609 }
1610 
1611 static const struct got_error *
1612 import_file(struct got_tree_entry **new_te, struct dirent *de,
1613     const char *path, struct got_repository *repo)
1614 {
1615 	const struct got_error *err;
1616 	struct got_object_id *blob_id = NULL;
1617 	char *filepath;
1618 	struct stat sb;
1619 
1620 	if (asprintf(&filepath, "%s%s%s", path,
1621 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1622 		return got_error_from_errno("asprintf");
1623 
1624 	if (lstat(filepath, &sb) != 0) {
1625 		err = got_error_from_errno2("lstat", path);
1626 		goto done;
1627 	}
1628 
1629 	err = got_object_blob_create(&blob_id, filepath, repo);
1630 	if (err)
1631 		goto done;
1632 
1633 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1634 	    blob_id);
1635 done:
1636 	free(filepath);
1637 	if (err)
1638 		free(blob_id);
1639 	return err;
1640 }
1641 
1642 static const struct got_error *
1643 insert_tree_entry(struct got_tree_entry *new_te,
1644     struct got_pathlist_head *paths)
1645 {
1646 	const struct got_error *err = NULL;
1647 	struct got_pathlist_entry *new_pe;
1648 
1649 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1650 	if (err)
1651 		return err;
1652 	if (new_pe == NULL)
1653 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1654 	return NULL;
1655 }
1656 
1657 static const struct got_error *write_tree(struct got_object_id **,
1658     const char *, struct got_pathlist_head *, struct got_repository *,
1659     got_repo_import_cb progress_cb, void *progress_arg);
1660 
1661 static const struct got_error *
1662 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1663     const char *path, struct got_pathlist_head *ignores,
1664     struct got_repository *repo,
1665     got_repo_import_cb progress_cb, void *progress_arg)
1666 {
1667 	const struct got_error *err;
1668 	struct got_object_id *id = NULL;
1669 	char *subdirpath;
1670 
1671 	if (asprintf(&subdirpath, "%s%s%s", path,
1672 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1673 		return got_error_from_errno("asprintf");
1674 
1675 	(*new_te) = calloc(1, sizeof(**new_te));
1676 	if (*new_te == NULL)
1677 		return got_error_from_errno("calloc");
1678 	(*new_te)->mode = S_IFDIR;
1679 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1680 	    sizeof((*new_te)->name)) {
1681 		err = got_error(GOT_ERR_NO_SPACE);
1682 		goto done;
1683 	}
1684 	err = write_tree(&id, subdirpath, ignores,  repo,
1685 	    progress_cb, progress_arg);
1686 	if (err)
1687 		goto done;
1688 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1689 	
1690 done:
1691 	free(id);
1692 	free(subdirpath);
1693 	if (err) {
1694 		free(*new_te);
1695 		*new_te = NULL;
1696 	}
1697 	return err;
1698 }
1699 
1700 static const struct got_error *
1701 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1702     struct got_pathlist_head *ignores, struct got_repository *repo,
1703     got_repo_import_cb progress_cb, void *progress_arg)
1704 {
1705 	const struct got_error *err = NULL;
1706 	DIR *dir;
1707 	struct dirent *de;
1708 	int nentries;
1709 	struct got_tree_entry *new_te = NULL;
1710 	struct got_pathlist_head paths;
1711 	struct got_pathlist_entry *pe;
1712 
1713 	*new_tree_id = NULL;
1714 
1715 	TAILQ_INIT(&paths);
1716 
1717 	dir = opendir(path_dir);
1718 	if (dir == NULL) {
1719 		err = got_error_from_errno2("opendir", path_dir);
1720 		goto done;
1721 	}
1722 
1723 	nentries = 0;
1724 	while ((de = readdir(dir)) != NULL) {
1725 		int ignore = 0;
1726 		int type;
1727 
1728 		if (strcmp(de->d_name, ".") == 0 ||
1729 		    strcmp(de->d_name, "..") == 0)
1730 			continue;
1731 
1732 		TAILQ_FOREACH(pe, ignores, entry) {
1733 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1734 				ignore = 1;
1735 				break;
1736 			}
1737 		}
1738 		if (ignore)
1739 			continue;
1740 
1741 		err = got_path_dirent_type(&type, path_dir, de);
1742 		if (err)
1743 			goto done;
1744 
1745 		if (type == DT_DIR) {
1746 			err = import_subdir(&new_te, de, path_dir,
1747 			    ignores, repo, progress_cb, progress_arg);
1748 			if (err) {
1749 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1750 					goto done;
1751 				err = NULL;
1752 				continue;
1753 			}
1754 		} else if (type == DT_REG || type == DT_LNK) {
1755 			err = import_file(&new_te, de, path_dir, repo);
1756 			if (err)
1757 				goto done;
1758 		} else
1759 			continue;
1760 
1761 		err = insert_tree_entry(new_te, &paths);
1762 		if (err)
1763 			goto done;
1764 		nentries++;
1765 	}
1766 
1767 	if (TAILQ_EMPTY(&paths)) {
1768 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1769 		    "cannot create tree without any entries");
1770 		goto done;
1771 	}
1772 
1773 	TAILQ_FOREACH(pe, &paths, entry) {
1774 		struct got_tree_entry *te = pe->data;
1775 		char *path;
1776 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1777 			continue;
1778 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1779 			err = got_error_from_errno("asprintf");
1780 			goto done;
1781 		}
1782 		err = (*progress_cb)(progress_arg, path);
1783 		free(path);
1784 		if (err)
1785 			goto done;
1786 	}
1787 
1788 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1789 done:
1790 	if (dir)
1791 		closedir(dir);
1792 	got_pathlist_free(&paths);
1793 	return err;
1794 }
1795 
1796 const struct got_error *
1797 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1798     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1799     struct got_repository *repo, got_repo_import_cb progress_cb,
1800     void *progress_arg)
1801 {
1802 	const struct got_error *err;
1803 	struct got_object_id *new_tree_id;
1804 
1805 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1806 	    progress_cb, progress_arg);
1807 	if (err)
1808 		return err;
1809 
1810 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1811 	    author, time(NULL), author, time(NULL), logmsg, repo);
1812 	free(new_tree_id);
1813 	return err;
1814 }
1815 
1816 const struct got_error *
1817 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
1818     struct got_repository *repo)
1819 {
1820 	const struct got_error *err = NULL;
1821 	char *path_objects = NULL, *path = NULL;
1822 	DIR *dir = NULL;
1823 	struct got_object_id id;
1824 	int i;
1825 
1826 	*nobjects = 0;
1827 	*ondisk_size = 0;
1828 
1829 	path_objects = got_repo_get_path_objects(repo);
1830 	if (path_objects == NULL)
1831 		return got_error_from_errno("got_repo_get_path_objects");
1832 
1833 	for (i = 0; i <= 0xff; i++) {
1834 		struct dirent *dent;
1835 
1836 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
1837 			err = got_error_from_errno("asprintf");
1838 			break;
1839 		}
1840 
1841 		dir = opendir(path);
1842 		if (dir == NULL) {
1843 			if (errno == ENOENT) {
1844 				err = NULL;
1845 				continue;
1846 			}
1847 			err = got_error_from_errno2("opendir", path);
1848 			break;
1849 		}
1850 
1851 		while ((dent = readdir(dir)) != NULL) {
1852 			char *id_str;
1853 			int fd;
1854 			struct stat sb;
1855 
1856 			if (strcmp(dent->d_name, ".") == 0 ||
1857 			    strcmp(dent->d_name, "..") == 0)
1858 				continue;
1859 
1860 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
1861 				err = got_error_from_errno("asprintf");
1862 				goto done;
1863 			}
1864 
1865 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
1866 				free(id_str);
1867 				continue;
1868 			}
1869 			free(id_str);
1870 
1871 			err = got_object_open_loose_fd(&fd, &id, repo);
1872 			if (err)
1873 				goto done;
1874 
1875 			if (fstat(fd, &sb) == -1) {
1876 				err = got_error_from_errno("fstat");
1877 				close(fd);
1878 				goto done;
1879 			}
1880 			(*nobjects)++;
1881 			(*ondisk_size) += sb.st_size;
1882 
1883 			if (close(fd) == -1) {
1884 				err = got_error_from_errno("close");
1885 				goto done;
1886 			}
1887 		}
1888 
1889 		if (closedir(dir) != 0) {
1890 			err = got_error_from_errno("closedir");
1891 			goto done;
1892 		}
1893 		dir = NULL;
1894 
1895 		free(path);
1896 		path = NULL;
1897 	}
1898 done:
1899 	if (dir && closedir(dir) != 0 && err == NULL)
1900 		err = got_error_from_errno("closedir");
1901 
1902 	if (err) {
1903 		*nobjects = 0;
1904 		*ondisk_size = 0;
1905 	}
1906 	free(path_objects);
1907 	free(path);
1908 	return err;
1909 }
1910 
1911 const struct got_error *
1912 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
1913     off_t *total_packsize, struct got_repository *repo)
1914 {
1915 	const struct got_error *err;
1916 	DIR *packdir = NULL;
1917 	struct dirent *dent;
1918 	struct got_packidx *packidx = NULL;
1919 	char *path_packidx;
1920 	char *path_packfile;
1921 	int packdir_fd;
1922 	struct stat sb;
1923 
1924 	*npackfiles = 0;
1925 	*nobjects = 0;
1926 	*total_packsize = 0;
1927 
1928 	packdir_fd = openat(got_repo_get_fd(repo),
1929 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1930 	if (packdir_fd == -1) {
1931 		return got_error_from_errno_fmt("openat: %s/%s",
1932 		    got_repo_get_path_git_dir(repo),
1933 		    GOT_OBJECTS_PACK_DIR);
1934 	}
1935 
1936 	packdir = fdopendir(packdir_fd);
1937 	if (packdir == NULL) {
1938 		err = got_error_from_errno("fdopendir");
1939 		goto done;
1940 	}
1941 
1942 	while ((dent = readdir(packdir)) != NULL) {
1943 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1944 			continue;
1945 
1946 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1947 		    dent->d_name) == -1) {
1948 			err = got_error_from_errno("asprintf");
1949 			goto done;
1950 		}
1951 
1952 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1953 		    path_packidx, 0);
1954 		free(path_packidx);
1955 		if (err)
1956 			goto done;
1957 
1958 		if (fstat(packidx->fd, &sb) == -1)
1959 			goto done;
1960 		*total_packsize += sb.st_size;
1961 
1962 		err = got_packidx_get_packfile_path(&path_packfile, packidx);
1963 		if (err)
1964 			goto done;
1965 
1966 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
1967 		    0) == -1) {
1968 			free(path_packfile);
1969 			goto done;
1970 		}
1971 		free(path_packfile);
1972 		*total_packsize += sb.st_size;
1973 
1974 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
1975 
1976 		(*npackfiles)++;
1977 
1978 		got_packidx_close(packidx);
1979 		packidx = NULL;
1980 	}
1981 done:
1982 	if (packidx)
1983 		got_packidx_close(packidx);
1984 	if (packdir && closedir(packdir) != 0 && err == NULL)
1985 		err = got_error_from_errno("closedir");
1986 	if (err) {
1987 		*npackfiles = 0;
1988 		*nobjects = 0;
1989 		*total_packsize = 0;
1990 	}
1991 	return err;
1992 }