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/uio.h>
19 #include <sys/socket.h>
20 #include <sys/stat.h>
21 #include <sys/mman.h>
22 #include <sys/resource.h>
23 
24 #include <ctype.h>
25 #include <endian.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 <unistd.h>
36 #include <zlib.h>
37 #include <errno.h>
38 #include <libgen.h>
39 #include <stdint.h>
40 #include <uuid.h>
41 
42 #include "got_compat.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 void
116 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
117     struct got_repository *repo)
118 {
119 	*extensions = repo->extensions;
120 	*nextensions = repo->nextensions;
121 }
122 
123 int
124 got_repo_is_bare(struct got_repository *repo)
125 {
126 	return (strcmp(repo->path, repo->path_git_dir) == 0);
127 }
128 
129 static char *
130 get_path_git_child(struct got_repository *repo, const char *basename)
131 {
132 	char *path_child;
133 
134 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
135 	    basename) == -1)
136 		return NULL;
137 
138 	return path_child;
139 }
140 
141 char *
142 got_repo_get_path_objects(struct got_repository *repo)
143 {
144 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
145 }
146 
147 char *
148 got_repo_get_path_objects_pack(struct got_repository *repo)
149 {
150 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
151 }
152 
153 char *
154 got_repo_get_path_refs(struct got_repository *repo)
155 {
156 	return get_path_git_child(repo, GOT_REFS_DIR);
157 }
158 
159 char *
160 got_repo_get_path_packed_refs(struct got_repository *repo)
161 {
162 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
163 }
164 
165 static char *
166 get_path_head(struct got_repository *repo)
167 {
168 	return get_path_git_child(repo, GOT_HEAD_FILE);
169 }
170 
171 char *
172 got_repo_get_path_gitconfig(struct got_repository *repo)
173 {
174 	return get_path_git_child(repo, GOT_GITCONFIG);
175 }
176 
177 char *
178 got_repo_get_path_gotconfig(struct got_repository *repo)
179 {
180 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
181 }
182 
183 const struct got_gotconfig *
184 got_repo_get_gotconfig(struct got_repository *repo)
185 {
186 	return repo->gotconfig;
187 }
188 
189 void
190 got_repo_get_gitconfig_remotes(int *nremotes,
191     const struct got_remote_repo **remotes, struct got_repository *repo)
192 {
193 	*nremotes = repo->ngitconfig_remotes;
194 	*remotes = repo->gitconfig_remotes;
195 }
196 
197 static int
198 is_git_repo(struct got_repository *repo)
199 {
200 	const char *path_git = got_repo_get_path_git_dir(repo);
201 	char *path_objects = got_repo_get_path_objects(repo);
202 	char *path_refs = got_repo_get_path_refs(repo);
203 	char *path_head = get_path_head(repo);
204 	int ret = 0;
205 	struct stat sb;
206 	struct got_reference *head_ref;
207 
208 	if (lstat(path_git, &sb) == -1)
209 		goto done;
210 	if (!S_ISDIR(sb.st_mode))
211 		goto done;
212 
213 	if (lstat(path_objects, &sb) == -1)
214 		goto done;
215 	if (!S_ISDIR(sb.st_mode))
216 		goto done;
217 
218 	if (lstat(path_refs, &sb) == -1)
219 		goto done;
220 	if (!S_ISDIR(sb.st_mode))
221 		goto done;
222 
223 	if (lstat(path_head, &sb) == -1)
224 		goto done;
225 	if (!S_ISREG(sb.st_mode))
226 		goto done;
227 
228 	/* Check if the HEAD reference can be opened. */
229 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
230 		goto done;
231 	got_ref_close(head_ref);
232 
233 	ret = 1;
234 done:
235 	free(path_objects);
236 	free(path_refs);
237 	free(path_head);
238 	return ret;
239 
240 }
241 
242 const struct got_error *
243 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
244     struct got_object *obj)
245 {
246 #ifndef GOT_NO_OBJ_CACHE
247 	const struct got_error *err = NULL;
248 	err = got_object_cache_add(&repo->objcache, id, obj);
249 	if (err) {
250 		if (err->code == GOT_ERR_OBJ_EXISTS ||
251 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
252 			err = NULL;
253 		return err;
254 	}
255 	obj->refcnt++;
256 #endif
257 	return NULL;
258 }
259 
260 struct got_object *
261 got_repo_get_cached_object(struct got_repository *repo,
262     struct got_object_id *id)
263 {
264 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
265 }
266 
267 const struct got_error *
268 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
269     struct got_tree_object *tree)
270 {
271 #ifndef GOT_NO_OBJ_CACHE
272 	const struct got_error *err = NULL;
273 	err = got_object_cache_add(&repo->treecache, id, tree);
274 	if (err) {
275 		if (err->code == GOT_ERR_OBJ_EXISTS ||
276 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
277 			err = NULL;
278 		return err;
279 	}
280 	tree->refcnt++;
281 #endif
282 	return NULL;
283 }
284 
285 struct got_tree_object *
286 got_repo_get_cached_tree(struct got_repository *repo,
287     struct got_object_id *id)
288 {
289 	return (struct got_tree_object *)got_object_cache_get(
290 	    &repo->treecache, id);
291 }
292 
293 const struct got_error *
294 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
295     struct got_commit_object *commit)
296 {
297 #ifndef GOT_NO_OBJ_CACHE
298 	const struct got_error *err = NULL;
299 	err = got_object_cache_add(&repo->commitcache, id, commit);
300 	if (err) {
301 		if (err->code == GOT_ERR_OBJ_EXISTS ||
302 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
303 			err = NULL;
304 		return err;
305 	}
306 	commit->refcnt++;
307 #endif
308 	return NULL;
309 }
310 
311 struct got_commit_object *
312 got_repo_get_cached_commit(struct got_repository *repo,
313     struct got_object_id *id)
314 {
315 	return (struct got_commit_object *)got_object_cache_get(
316 	    &repo->commitcache, id);
317 }
318 
319 const struct got_error *
320 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
321     struct got_tag_object *tag)
322 {
323 #ifndef GOT_NO_OBJ_CACHE
324 	const struct got_error *err = NULL;
325 	err = got_object_cache_add(&repo->tagcache, id, tag);
326 	if (err) {
327 		if (err->code == GOT_ERR_OBJ_EXISTS ||
328 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
329 			err = NULL;
330 		return err;
331 	}
332 	tag->refcnt++;
333 #endif
334 	return NULL;
335 }
336 
337 struct got_tag_object *
338 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
339 {
340 	return (struct got_tag_object *)got_object_cache_get(
341 	    &repo->tagcache, id);
342 }
343 
344 static const struct got_error *
345 open_repo(struct got_repository *repo, const char *path)
346 {
347 	const struct got_error *err = NULL;
348 
349 	repo->gitdir_fd = -1;
350 
351 	/* bare git repository? */
352 	repo->path_git_dir = strdup(path);
353 	if (repo->path_git_dir == NULL)
354 		return got_error_from_errno("strdup");
355 	if (is_git_repo(repo)) {
356 		repo->path = strdup(repo->path_git_dir);
357 		if (repo->path == NULL) {
358 			err = got_error_from_errno("strdup");
359 			goto done;
360 		}
361 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
362 		if (repo->gitdir_fd == -1) {
363 			err = got_error_from_errno2("open",
364 			    repo->path_git_dir);
365 			goto done;
366 		}
367 		return NULL;
368 	}
369 
370 	/* git repository with working tree? */
371 	free(repo->path_git_dir);
372 	repo->path_git_dir = NULL;
373 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
374 		err = got_error_from_errno("asprintf");
375 		goto done;
376 	}
377 	if (is_git_repo(repo)) {
378 		repo->path = strdup(path);
379 		if (repo->path == NULL) {
380 			err = got_error_from_errno("strdup");
381 			goto done;
382 		}
383 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
384 		if (repo->gitdir_fd == -1) {
385 			err = got_error_from_errno2("open",
386 			    repo->path_git_dir);
387 			goto done;
388 		}
389 		return NULL;
390 	}
391 
392 	err = got_error(GOT_ERR_NOT_GIT_REPO);
393 done:
394 	if (err) {
395 		free(repo->path);
396 		repo->path = NULL;
397 		free(repo->path_git_dir);
398 		repo->path_git_dir = NULL;
399 		if (repo->gitdir_fd != -1)
400 			close(repo->gitdir_fd);
401 		repo->gitdir_fd = -1;
402 
403 	}
404 	return err;
405 }
406 
407 static const struct got_error *
408 parse_gitconfig_file(int *gitconfig_repository_format_version,
409     char **gitconfig_author_name, char **gitconfig_author_email,
410     struct got_remote_repo **remotes, int *nremotes,
411     char **gitconfig_owner, char ***extensions, int *nextensions,
412     const char *gitconfig_path)
413 {
414 	const struct got_error *err = NULL, *child_err = NULL;
415 	int fd = -1;
416 	int imsg_fds[2] = { -1, -1 };
417 	pid_t pid;
418 	struct imsgbuf *ibuf;
419 
420 	*gitconfig_repository_format_version = 0;
421 	if (extensions)
422 		*extensions = NULL;
423 	if (nextensions)
424 		*nextensions = 0;
425 	*gitconfig_author_name = NULL;
426 	*gitconfig_author_email = NULL;
427 	if (remotes)
428 		*remotes = NULL;
429 	if (nremotes)
430 		*nremotes = 0;
431 	if (gitconfig_owner)
432 		*gitconfig_owner = NULL;
433 
434 	fd = open(gitconfig_path, O_RDONLY);
435 	if (fd == -1) {
436 		if (errno == ENOENT)
437 			return NULL;
438 		return got_error_from_errno2("open", gitconfig_path);
439 	}
440 
441 	ibuf = calloc(1, sizeof(*ibuf));
442 	if (ibuf == NULL) {
443 		err = got_error_from_errno("calloc");
444 		goto done;
445 	}
446 
447 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
448 		err = got_error_from_errno("socketpair");
449 		goto done;
450 	}
451 
452 	pid = fork();
453 	if (pid == -1) {
454 		err = got_error_from_errno("fork");
455 		goto done;
456 	} else if (pid == 0) {
457 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
458 		    gitconfig_path);
459 		/* not reached */
460 	}
461 
462 	if (close(imsg_fds[1]) == -1) {
463 		err = got_error_from_errno("close");
464 		goto done;
465 	}
466 	imsg_fds[1] = -1;
467 	imsg_init(ibuf, imsg_fds[0]);
468 
469 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
470 	if (err)
471 		goto done;
472 	fd = -1;
473 
474 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
475 	if (err)
476 		goto done;
477 
478 	err = got_privsep_recv_gitconfig_int(
479 	    gitconfig_repository_format_version, ibuf);
480 	if (err)
481 		goto done;
482 
483 	if (extensions && nextensions) {
484 		err = got_privsep_send_gitconfig_repository_extensions_req(
485 		    ibuf);
486 		if (err)
487 			goto done;
488 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
489 		if (err)
490 			goto done;
491 		if (*nextensions > 0) {
492 			int i;
493 			*extensions = calloc(*nextensions, sizeof(char *));
494 			if (*extensions == NULL) {
495 				err = got_error_from_errno("calloc");
496 				goto done;
497 			}
498 			for (i = 0; i < *nextensions; i++) {
499 				char *ext;
500 				err = got_privsep_recv_gitconfig_str(&ext,
501 				    ibuf);
502 				if (err)
503 					goto done;
504 				(*extensions)[i] = ext;
505 			}
506 		}
507 	}
508 
509 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
510 	if (err)
511 		goto done;
512 
513 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
514 	if (err)
515 		goto done;
516 
517 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
518 	if (err)
519 		goto done;
520 
521 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
522 	if (err)
523 		goto done;
524 
525 	if (remotes && nremotes) {
526 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
527 		if (err)
528 			goto done;
529 
530 		err = got_privsep_recv_gitconfig_remotes(remotes,
531 		    nremotes, ibuf);
532 		if (err)
533 			goto done;
534 	}
535 
536 	if (gitconfig_owner) {
537 		err = got_privsep_send_gitconfig_owner_req(ibuf);
538 		if (err)
539 			goto done;
540 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
541 		if (err)
542 			goto done;
543 	}
544 
545 	imsg_clear(ibuf);
546 	err = got_privsep_send_stop(imsg_fds[0]);
547 	child_err = got_privsep_wait_for_child(pid);
548 	if (child_err && err == NULL)
549 		err = child_err;
550 done:
551 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
552 		err = got_error_from_errno("close");
553 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
554 		err = got_error_from_errno("close");
555 	if (fd != -1 && close(fd) == -1 && err == NULL)
556 		err = got_error_from_errno2("close", gitconfig_path);
557 	free(ibuf);
558 	return err;
559 }
560 
561 static const struct got_error *
562 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
563 {
564 	const struct got_error *err = NULL;
565 	char *repo_gitconfig_path = NULL;
566 
567 	if (global_gitconfig_path) {
568 		/* Read settings from ~/.gitconfig. */
569 		int dummy_repo_version;
570 		err = parse_gitconfig_file(&dummy_repo_version,
571 		    &repo->global_gitconfig_author_name,
572 		    &repo->global_gitconfig_author_email,
573 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
574 		if (err)
575 			return err;
576 	}
577 
578 	/* Read repository's .git/config file. */
579 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
580 	if (repo_gitconfig_path == NULL)
581 		return got_error_from_errno("got_repo_get_path_gitconfig");
582 
583 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
584 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
585 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
586 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
587 	    repo_gitconfig_path);
588 	if (err)
589 		goto done;
590 done:
591 	free(repo_gitconfig_path);
592 	return err;
593 }
594 
595 static const struct got_error *
596 read_gotconfig(struct got_repository *repo)
597 {
598 	const struct got_error *err = NULL;
599 	char *gotconfig_path;
600 
601 	gotconfig_path = got_repo_get_path_gotconfig(repo);
602 	if (gotconfig_path == NULL)
603 		return got_error_from_errno("got_repo_get_path_gotconfig");
604 
605 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
606 	free(gotconfig_path);
607 	return err;
608 }
609 
610 /* Supported repository format extensions. */
611 static const char *repo_extensions[] = {
612 	"noop",			/* Got supports repository format version 1. */
613 	"preciousObjects",	/* Supported by gotadmin cleanup. */
614 	"worktreeConfig",	/* Got does not care about Git work trees. */
615 };
616 
617 const struct got_error *
618 got_repo_open(struct got_repository **repop, const char *path,
619     const char *global_gitconfig_path)
620 {
621 	struct got_repository *repo = NULL;
622 	const struct got_error *err = NULL;
623 	char *repo_path = NULL;
624 	size_t i;
625 	struct rlimit rl;
626 
627 	*repop = NULL;
628 
629 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
630 		return got_error_from_errno("getrlimit");
631 
632 	repo = calloc(1, sizeof(*repo));
633 	if (repo == NULL) {
634 		err = got_error_from_errno("calloc");
635 		goto done;
636 	}
637 
638 	for (i = 0; i < nitems(repo->privsep_children); i++) {
639 		memset(&repo->privsep_children[i], 0,
640 		    sizeof(repo->privsep_children[0]));
641 		repo->privsep_children[i].imsg_fd = -1;
642 	}
643 
644 	err = got_object_cache_init(&repo->objcache,
645 	    GOT_OBJECT_CACHE_TYPE_OBJ);
646 	if (err)
647 		goto done;
648 	err = got_object_cache_init(&repo->treecache,
649 	    GOT_OBJECT_CACHE_TYPE_TREE);
650 	if (err)
651 		goto done;
652 	err = got_object_cache_init(&repo->commitcache,
653 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
654 	if (err)
655 		goto done;
656 	err = got_object_cache_init(&repo->tagcache,
657 	    GOT_OBJECT_CACHE_TYPE_TAG);
658 	if (err)
659 		goto done;
660 
661 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
662 	if (repo->pack_cache_size > rl.rlim_cur / 8)
663 		repo->pack_cache_size = rl.rlim_cur / 8;
664 
665 	repo_path = realpath(path, NULL);
666 	if (repo_path == NULL) {
667 		err = got_error_from_errno2("realpath", path);
668 		goto done;
669 	}
670 
671 	for (;;) {
672 		char *parent_path;
673 
674 		err = open_repo(repo, repo_path);
675 		if (err == NULL)
676 			break;
677 		if (err->code != GOT_ERR_NOT_GIT_REPO)
678 			goto done;
679 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
680 			err = got_error(GOT_ERR_NOT_GIT_REPO);
681 			goto done;
682 		}
683 		err = got_path_dirname(&parent_path, repo_path);
684 		if (err)
685 			goto done;
686 		free(repo_path);
687 		repo_path = parent_path;
688 	}
689 
690 	err = read_gotconfig(repo);
691 	if (err)
692 		goto done;
693 
694 	err = read_gitconfig(repo, global_gitconfig_path);
695 	if (err)
696 		goto done;
697 	if (repo->gitconfig_repository_format_version != 0)
698 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
699 	for (i = 0; i < repo->nextensions; i++) {
700 		char *ext = repo->extensions[i];
701 		int j, supported = 0;
702 		for (j = 0; j < nitems(repo_extensions); j++) {
703 			if (strcmp(ext, repo_extensions[j]) == 0) {
704 				supported = 1;
705 				break;
706 			}
707 		}
708 		if (!supported) {
709 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
710 			goto done;
711 		}
712 	}
713 done:
714 	if (err)
715 		got_repo_close(repo);
716 	else
717 		*repop = repo;
718 	free(repo_path);
719 	return err;
720 }
721 
722 const struct got_error *
723 got_repo_close(struct got_repository *repo)
724 {
725 	const struct got_error *err = NULL, *child_err;
726 	size_t i;
727 
728 	for (i = 0; i < repo->pack_cache_size; i++) {
729 		if (repo->packidx_cache[i] == NULL)
730 			break;
731 		got_packidx_close(repo->packidx_cache[i]);
732 	}
733 
734 	for (i = 0; i < repo->pack_cache_size; i++) {
735 		if (repo->packs[i].path_packfile == NULL)
736 			break;
737 		got_pack_close(&repo->packs[i]);
738 	}
739 
740 	free(repo->path);
741 	free(repo->path_git_dir);
742 
743 	got_object_cache_close(&repo->objcache);
744 	got_object_cache_close(&repo->treecache);
745 	got_object_cache_close(&repo->commitcache);
746 	got_object_cache_close(&repo->tagcache);
747 
748 	for (i = 0; i < nitems(repo->privsep_children); i++) {
749 		if (repo->privsep_children[i].imsg_fd == -1)
750 			continue;
751 		imsg_clear(repo->privsep_children[i].ibuf);
752 		free(repo->privsep_children[i].ibuf);
753 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
754 		child_err = got_privsep_wait_for_child(
755 		    repo->privsep_children[i].pid);
756 		if (child_err && err == NULL)
757 			err = child_err;
758 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
759 		    err == NULL)
760 			err = got_error_from_errno("close");
761 	}
762 
763 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
764 	    err == NULL)
765 		err = got_error_from_errno("close");
766 
767 	if (repo->gotconfig)
768 		got_gotconfig_free(repo->gotconfig);
769 	free(repo->gitconfig_author_name);
770 	free(repo->gitconfig_author_email);
771 	for (i = 0; i < repo->ngitconfig_remotes; i++)
772 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
773 	free(repo->gitconfig_remotes);
774 	for (i = 0; i < repo->nextensions; i++)
775 		free(repo->extensions[i]);
776 	free(repo->extensions);
777 	free(repo);
778 
779 	return err;
780 }
781 
782 void
783 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
784 {
785 	int i;
786 
787 	free(repo->name);
788 	repo->name = NULL;
789 	free(repo->fetch_url);
790 	repo->fetch_url = NULL;
791 	free(repo->send_url);
792 	repo->send_url = NULL;
793 	for (i = 0; i < repo->nfetch_branches; i++)
794 		free(repo->fetch_branches[i]);
795 	free(repo->fetch_branches);
796 	repo->fetch_branches = NULL;
797 	repo->nfetch_branches = 0;
798 	for (i = 0; i < repo->nsend_branches; i++)
799 		free(repo->send_branches[i]);
800 	free(repo->send_branches);
801 	repo->send_branches = NULL;
802 	repo->nsend_branches = 0;
803 }
804 
805 const struct got_error *
806 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
807     const char *input_path)
808 {
809 	const struct got_error *err = NULL;
810 	const char *repo_abspath = NULL;
811 	size_t repolen, len;
812 	char *canonpath, *path = NULL;
813 
814 	*in_repo_path = NULL;
815 
816 	canonpath = strdup(input_path);
817 	if (canonpath == NULL) {
818 		err = got_error_from_errno("strdup");
819 		goto done;
820 	}
821 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
822 	if (err)
823 		goto done;
824 
825 	repo_abspath = got_repo_get_path(repo);
826 
827 	if (canonpath[0] == '\0') {
828 		path = strdup(canonpath);
829 		if (path == NULL) {
830 			err = got_error_from_errno("strdup");
831 			goto done;
832 		}
833 	} else {
834 		path = realpath(canonpath, NULL);
835 		if (path == NULL) {
836 			if (errno != ENOENT) {
837 				err = got_error_from_errno2("realpath",
838 				    canonpath);
839 				goto done;
840 			}
841 			/*
842 			 * Path is not on disk.
843 			 * Assume it is already relative to repository root.
844 			 */
845 			path = strdup(canonpath);
846 			if (path == NULL) {
847 				err = got_error_from_errno("strdup");
848 				goto done;
849 			}
850 		}
851 
852 		repolen = strlen(repo_abspath);
853 		len = strlen(path);
854 
855 
856 		if (strcmp(path, repo_abspath) == 0) {
857 			free(path);
858 			path = strdup("");
859 			if (path == NULL) {
860 				err = got_error_from_errno("strdup");
861 				goto done;
862 			}
863 		} else if (len > repolen &&
864 		    got_path_is_child(path, repo_abspath, repolen)) {
865 			/* Matched an on-disk path inside repository. */
866 			if (got_repo_is_bare(repo)) {
867 				/*
868 				 * Matched an on-disk path inside repository
869 				 * database. Treat input as repository-relative.
870 				 */
871 				free(path);
872 				path = canonpath;
873 				canonpath = NULL;
874 			} else {
875 				char *child;
876 				/* Strip common prefix with repository path. */
877 				err = got_path_skip_common_ancestor(&child,
878 				    repo_abspath, path);
879 				if (err)
880 					goto done;
881 				free(path);
882 				path = child;
883 			}
884 		} else {
885 			/*
886 			 * Matched unrelated on-disk path.
887 			 * Treat input as repository-relative.
888 			 */
889 			free(path);
890 			path = canonpath;
891 			canonpath = NULL;
892 		}
893 	}
894 
895 	/* Make in-repository path absolute */
896 	if (path[0] != '/') {
897 		char *abspath;
898 		if (asprintf(&abspath, "/%s", path) == -1) {
899 			err = got_error_from_errno("asprintf");
900 			goto done;
901 		}
902 		free(path);
903 		path = abspath;
904 	}
905 
906 done:
907 	free(canonpath);
908 	if (err)
909 		free(path);
910 	else
911 		*in_repo_path = path;
912 	return err;
913 }
914 
915 static const struct got_error *
916 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
917     const char *path_packidx)
918 {
919 	const struct got_error *err = NULL;
920 	size_t i;
921 
922 	for (i = 0; i < repo->pack_cache_size; i++) {
923 		if (repo->packidx_cache[i] == NULL)
924 			break;
925 		if (strcmp(repo->packidx_cache[i]->path_packidx,
926 		    path_packidx) == 0) {
927 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
928 		}
929 	}
930 	if (i == repo->pack_cache_size) {
931 		err = got_packidx_close(repo->packidx_cache[i - 1]);
932 		if (err)
933 			return err;
934 	}
935 
936 	/*
937 	 * Insert the new pack index at the front so it will
938 	 * be searched first in the future.
939 	 */
940 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
941 	    sizeof(repo->packidx_cache) -
942 	    sizeof(repo->packidx_cache[0]));
943 	repo->packidx_cache[0] = packidx;
944 
945 	return NULL;
946 }
947 
948 int
949 got_repo_is_packidx_filename(const char *name, size_t len)
950 {
951 	if (len != GOT_PACKIDX_NAMELEN)
952 		return 0;
953 
954 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
955 		return 0;
956 
957 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
958 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
959 		return 0;
960 
961 	return 1;
962 }
963 
964 const struct got_error *
965 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
966     struct got_repository *repo, struct got_object_id *id)
967 {
968 	const struct got_error *err;
969 	DIR *packdir = NULL;
970 	struct dirent *dent;
971 	char *path_packidx;
972 	size_t i;
973 	int packdir_fd;
974 
975 	/* Search pack index cache. */
976 	for (i = 0; i < repo->pack_cache_size; i++) {
977 		if (repo->packidx_cache[i] == NULL)
978 			break;
979 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
980 		if (*idx != -1) {
981 			*packidx = repo->packidx_cache[i];
982 			/*
983 			 * Move this cache entry to the front. Repeatedly
984 			 * searching a wrong pack index can be expensive.
985 			 */
986 			if (i > 0) {
987 				struct got_packidx *p;
988 				p = repo->packidx_cache[0];
989 				repo->packidx_cache[0] = *packidx;
990 				repo->packidx_cache[i] = p;
991 			}
992 			return NULL;
993 		}
994 	}
995 	/* No luck. Search the filesystem. */
996 
997 	packdir_fd = openat(got_repo_get_fd(repo),
998 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
999 	if (packdir_fd == -1) {
1000 		if (errno == ENOENT)
1001 			err = got_error_no_obj(id);
1002 		else
1003 			err = got_error_from_errno_fmt("openat: %s/%s",
1004 			    got_repo_get_path_git_dir(repo),
1005 			    GOT_OBJECTS_PACK_DIR);
1006 		goto done;
1007 	}
1008 
1009 	packdir = fdopendir(packdir_fd);
1010 	if (packdir == NULL) {
1011 		err = got_error_from_errno("fdopendir");
1012 		goto done;
1013 	}
1014 
1015 	while ((dent = readdir(packdir)) != NULL) {
1016 		int is_cached = 0;
1017 
1018 		if (!got_repo_is_packidx_filename(dent->d_name,
1019 		    strlen(dent->d_name)))
1020 			continue;
1021 
1022 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1023 		    dent->d_name) == -1) {
1024 			err = got_error_from_errno("asprintf");
1025 			goto done;
1026 		}
1027 
1028 		for (i = 0; i < repo->pack_cache_size; i++) {
1029 			if (repo->packidx_cache[i] == NULL)
1030 				break;
1031 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1032 			    path_packidx) == 0) {
1033 				is_cached = 1;
1034 				break;
1035 			}
1036 		}
1037 		if (is_cached) {
1038 			free(path_packidx);
1039 			continue; /* already searched */
1040 		}
1041 
1042 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1043 		    path_packidx, 0);
1044 		if (err) {
1045 			free(path_packidx);
1046 			goto done;
1047 		}
1048 
1049 		err = cache_packidx(repo, *packidx, path_packidx);
1050 		free(path_packidx);
1051 		if (err)
1052 			goto done;
1053 
1054 		*idx = got_packidx_get_object_idx(*packidx, id);
1055 		if (*idx != -1) {
1056 			err = NULL; /* found the object */
1057 			goto done;
1058 		}
1059 	}
1060 
1061 	err = got_error_no_obj(id);
1062 done:
1063 	if (packdir && closedir(packdir) != 0 && err == NULL)
1064 		err = got_error_from_errno("closedir");
1065 	return err;
1066 }
1067 
1068 static const struct got_error *
1069 read_packfile_hdr(int fd, struct got_packidx *packidx)
1070 {
1071 	const struct got_error *err = NULL;
1072 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1073 	struct got_packfile_hdr hdr;
1074 	ssize_t n;
1075 
1076 	n = read(fd, &hdr, sizeof(hdr));
1077 	if (n < 0)
1078 		return got_error_from_errno("read");
1079 	if (n != sizeof(hdr))
1080 		return got_error(GOT_ERR_BAD_PACKFILE);
1081 
1082 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1083 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1084 	    be32toh(hdr.nobjects) != totobj)
1085 		err = got_error(GOT_ERR_BAD_PACKFILE);
1086 
1087 	return err;
1088 }
1089 
1090 static const struct got_error *
1091 open_packfile(int *fd, struct got_repository *repo,
1092     const char *relpath, struct got_packidx *packidx)
1093 {
1094 	const struct got_error *err = NULL;
1095 
1096 	*fd = openat(got_repo_get_fd(repo), relpath, O_RDONLY | O_NOFOLLOW);
1097 	if (*fd == -1)
1098 		return got_error_from_errno_fmt("openat: %s/%s",
1099 		    got_repo_get_path_git_dir(repo), relpath);
1100 
1101 	if (packidx) {
1102 		err = read_packfile_hdr(*fd, packidx);
1103 		if (err) {
1104 			close(*fd);
1105 			*fd = -1;
1106 		}
1107 	}
1108 
1109 	return err;
1110 }
1111 
1112 const struct got_error *
1113 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1114     const char *path_packfile, struct got_packidx *packidx)
1115 {
1116 	const struct got_error *err = NULL;
1117 	struct got_pack *pack = NULL;
1118 	struct stat sb;
1119 	size_t i;
1120 
1121 	if (packp)
1122 		*packp = NULL;
1123 
1124 	for (i = 0; i < repo->pack_cache_size; i++) {
1125 		pack = &repo->packs[i];
1126 		if (pack->path_packfile == NULL)
1127 			break;
1128 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1129 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1130 	}
1131 
1132 	if (i == repo->pack_cache_size) {
1133 		err = got_pack_close(&repo->packs[i - 1]);
1134 		if (err)
1135 			return err;
1136 		memmove(&repo->packs[1], &repo->packs[0],
1137 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1138 		i = 0;
1139 	}
1140 
1141 	pack = &repo->packs[i];
1142 
1143 	pack->path_packfile = strdup(path_packfile);
1144 	if (pack->path_packfile == NULL) {
1145 		err = got_error_from_errno("strdup");
1146 		goto done;
1147 	}
1148 
1149 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1150 	if (err)
1151 		goto done;
1152 
1153 	if (fstat(pack->fd, &sb) != 0) {
1154 		err = got_error_from_errno("fstat");
1155 		goto done;
1156 	}
1157 	pack->filesize = sb.st_size;
1158 
1159 	pack->privsep_child = NULL;
1160 
1161 #ifndef GOT_PACK_NO_MMAP
1162 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1163 	    pack->fd, 0);
1164 	if (pack->map == MAP_FAILED) {
1165 		if (errno != ENOMEM) {
1166 			err = got_error_from_errno("mmap");
1167 			goto done;
1168 		}
1169 		pack->map = NULL; /* fall back to read(2) */
1170 	}
1171 #endif
1172 done:
1173 	if (err) {
1174 		if (pack) {
1175 			free(pack->path_packfile);
1176 			memset(pack, 0, sizeof(*pack));
1177 		}
1178 	} else if (packp)
1179 		*packp = pack;
1180 	return err;
1181 }
1182 
1183 struct got_pack *
1184 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1185 {
1186 	struct got_pack *pack = NULL;
1187 	size_t i;
1188 
1189 	for (i = 0; i < repo->pack_cache_size; i++) {
1190 		pack = &repo->packs[i];
1191 		if (pack->path_packfile == NULL)
1192 			break;
1193 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1194 			return pack;
1195 	}
1196 
1197 	return NULL;
1198 }
1199 
1200 const struct got_error *
1201 got_repo_init(const char *repo_path)
1202 {
1203 	const struct got_error *err = NULL;
1204 	const char *dirnames[] = {
1205 		GOT_OBJECTS_DIR,
1206 		GOT_OBJECTS_PACK_DIR,
1207 		GOT_REFS_DIR,
1208 	};
1209 	const char *description_str = "Unnamed repository; "
1210 	    "edit this file 'description' to name the repository.";
1211 	const char *headref_str = "ref: refs/heads/main";
1212 	const char *gitconfig_str = "[core]\n"
1213 	    "\trepositoryformatversion = 0\n"
1214 	    "\tfilemode = true\n"
1215 	    "\tbare = true\n";
1216 	char *path;
1217 	size_t i;
1218 
1219 	if (!got_path_dir_is_empty(repo_path))
1220 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1221 
1222 	for (i = 0; i < nitems(dirnames); i++) {
1223 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1224 			return got_error_from_errno("asprintf");
1225 		}
1226 		err = got_path_mkdir(path);
1227 		free(path);
1228 		if (err)
1229 			return err;
1230 	}
1231 
1232 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1233 		return got_error_from_errno("asprintf");
1234 	err = got_path_create_file(path, description_str);
1235 	free(path);
1236 	if (err)
1237 		return err;
1238 
1239 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1240 		return got_error_from_errno("asprintf");
1241 	err = got_path_create_file(path, headref_str);
1242 	free(path);
1243 	if (err)
1244 		return err;
1245 
1246 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1247 		return got_error_from_errno("asprintf");
1248 	err = got_path_create_file(path, gitconfig_str);
1249 	free(path);
1250 	if (err)
1251 		return err;
1252 
1253 	return NULL;
1254 }
1255 
1256 static const struct got_error *
1257 match_packed_object(struct got_object_id **unique_id,
1258     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1259 {
1260 	const struct got_error *err = NULL;
1261 	DIR *packdir = NULL;
1262 	struct dirent *dent;
1263 	char *path_packidx;
1264 	struct got_object_id_queue matched_ids;
1265 	int packdir_fd;
1266 
1267 	STAILQ_INIT(&matched_ids);
1268 
1269 	packdir_fd = openat(got_repo_get_fd(repo),
1270 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1271 	if (packdir_fd == -1) {
1272 		if (errno != ENOENT)
1273 			err = got_error_from_errno2("openat", GOT_OBJECTS_PACK_DIR);
1274 		goto done;
1275 	}
1276 
1277 	packdir = fdopendir(packdir_fd);
1278 	if (packdir == NULL) {
1279 		err = got_error_from_errno("fdopendir");
1280 		goto done;
1281 	}
1282 
1283 	while ((dent = readdir(packdir)) != NULL) {
1284 		struct got_packidx *packidx;
1285 		struct got_object_qid *qid;
1286 
1287 		if (!got_repo_is_packidx_filename(dent->d_name,
1288 		    strlen(dent->d_name)))
1289 			continue;
1290 
1291 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1292 		    dent->d_name) == -1) {
1293 			err = got_error_from_errno("strdup");
1294 			break;
1295 		}
1296 
1297 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1298 		    path_packidx, 0);
1299 		free(path_packidx);
1300 		if (err)
1301 			break;
1302 
1303 		err = got_packidx_match_id_str_prefix(&matched_ids,
1304 		    packidx, id_str_prefix);
1305 		if (err) {
1306 			got_packidx_close(packidx);
1307 			break;
1308 		}
1309 		err = got_packidx_close(packidx);
1310 		if (err)
1311 			break;
1312 
1313 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1314 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1315 				int matched_type;
1316 				err = got_object_get_type(&matched_type, repo,
1317 				    qid->id);
1318 				if (err)
1319 					goto done;
1320 				if (matched_type != obj_type)
1321 					continue;
1322 			}
1323 			if (*unique_id == NULL) {
1324 				*unique_id = got_object_id_dup(qid->id);
1325 				if (*unique_id == NULL) {
1326 					err = got_error_from_errno("malloc");
1327 					goto done;
1328 				}
1329 			} else {
1330 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1331 					continue; /* packed multiple times */
1332 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1333 				goto done;
1334 			}
1335 		}
1336 	}
1337 done:
1338 	got_object_id_queue_free(&matched_ids);
1339 	if (packdir && closedir(packdir) != 0 && err == NULL)
1340 		err = got_error_from_errno("closedir");
1341 	if (err) {
1342 		free(*unique_id);
1343 		*unique_id = NULL;
1344 	}
1345 	return err;
1346 }
1347 
1348 static const struct got_error *
1349 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1350     const char *object_dir, const char *id_str_prefix, int obj_type,
1351     struct got_repository *repo)
1352 {
1353 	const struct got_error *err = NULL;
1354 	char *path;
1355 	DIR *dir = NULL;
1356 	struct dirent *dent;
1357 	struct got_object_id id;
1358 
1359 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1360 		err = got_error_from_errno("asprintf");
1361 		goto done;
1362 	}
1363 
1364 	dir = opendir(path);
1365 	if (dir == NULL) {
1366 		if (errno == ENOENT) {
1367 			err = NULL;
1368 			goto done;
1369 		}
1370 		err = got_error_from_errno2("opendir", path);
1371 		goto done;
1372 	}
1373 	while ((dent = readdir(dir)) != NULL) {
1374 		char *id_str;
1375 		int cmp;
1376 
1377 		if (strcmp(dent->d_name, ".") == 0 ||
1378 		    strcmp(dent->d_name, "..") == 0)
1379 			continue;
1380 
1381 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1382 			err = got_error_from_errno("asprintf");
1383 			goto done;
1384 		}
1385 
1386 		if (!got_parse_sha1_digest(id.sha1, id_str))
1387 			continue;
1388 
1389 		/*
1390 		 * Directory entries do not necessarily appear in
1391 		 * sorted order, so we must iterate over all of them.
1392 		 */
1393 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1394 		if (cmp != 0) {
1395 			free(id_str);
1396 			continue;
1397 		}
1398 
1399 		if (*unique_id == NULL) {
1400 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1401 				int matched_type;
1402 				err = got_object_get_type(&matched_type, repo,
1403 				    &id);
1404 				if (err)
1405 					goto done;
1406 				if (matched_type != obj_type)
1407 					continue;
1408 			}
1409 			*unique_id = got_object_id_dup(&id);
1410 			if (*unique_id == NULL) {
1411 				err = got_error_from_errno("got_object_id_dup");
1412 				free(id_str);
1413 				goto done;
1414 			}
1415 		} else {
1416 			if (got_object_id_cmp(*unique_id, &id) == 0)
1417 				continue; /* both packed and loose */
1418 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1419 			free(id_str);
1420 			goto done;
1421 		}
1422 	}
1423 done:
1424 	if (dir && closedir(dir) != 0 && err == NULL)
1425 		err = got_error_from_errno("closedir");
1426 	if (err) {
1427 		free(*unique_id);
1428 		*unique_id = NULL;
1429 	}
1430 	free(path);
1431 	return err;
1432 }
1433 
1434 const struct got_error *
1435 got_repo_match_object_id_prefix(struct got_object_id **id,
1436     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1437 {
1438 	const struct got_error *err = NULL;
1439 	char *path_objects = got_repo_get_path_objects(repo);
1440 	char *object_dir = NULL;
1441 	size_t len;
1442 	int i;
1443 
1444 	*id = NULL;
1445 
1446 	for (i = 0; i < strlen(id_str_prefix); i++) {
1447 		if (isxdigit((unsigned char)id_str_prefix[i]))
1448 			continue;
1449 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1450 	}
1451 
1452 	len = strlen(id_str_prefix);
1453 	if (len >= 2) {
1454 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1455 		if (err)
1456 			goto done;
1457 		object_dir = strndup(id_str_prefix, 2);
1458 		if (object_dir == NULL) {
1459 			err = got_error_from_errno("strdup");
1460 			goto done;
1461 		}
1462 		err = match_loose_object(id, path_objects, object_dir,
1463 		    id_str_prefix, obj_type, repo);
1464 	} else if (len == 1) {
1465 		int i;
1466 		for (i = 0; i < 0xf; i++) {
1467 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1468 			    == -1) {
1469 				err = got_error_from_errno("asprintf");
1470 				goto done;
1471 			}
1472 			err = match_packed_object(id, repo, object_dir,
1473 			    obj_type);
1474 			if (err)
1475 				goto done;
1476 			err = match_loose_object(id, path_objects, object_dir,
1477 			    id_str_prefix, obj_type, repo);
1478 			if (err)
1479 				goto done;
1480 		}
1481 	} else {
1482 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1483 		goto done;
1484 	}
1485 done:
1486 	free(object_dir);
1487 	if (err) {
1488 		free(*id);
1489 		*id = NULL;
1490 	} else if (*id == NULL)
1491 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1492 
1493 	return err;
1494 }
1495 
1496 const struct got_error *
1497 got_repo_match_object_id(struct got_object_id **id, char **label,
1498     const char *id_str, int obj_type, struct got_reflist_head *refs,
1499     struct got_repository *repo)
1500 {
1501 	const struct got_error *err;
1502 	struct got_tag_object *tag;
1503 	struct got_reference *ref = NULL;
1504 
1505 	*id = NULL;
1506 	if (label)
1507 		*label = NULL;
1508 
1509 	if (refs) {
1510 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1511 		    refs, repo);
1512 		if (err == NULL) {
1513 			*id = got_object_id_dup(
1514 			    got_object_tag_get_object_id(tag));
1515 			if (*id == NULL)
1516 				err = got_error_from_errno("got_object_id_dup");
1517 			else if (label && asprintf(label, "refs/tags/%s",
1518 			    got_object_tag_get_name(tag)) == -1) {
1519 				err = got_error_from_errno("asprintf");
1520 				free(*id);
1521 				*id = NULL;
1522 			}
1523 			got_object_tag_close(tag);
1524 			return err;
1525 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1526 		    err->code != GOT_ERR_NO_OBJ)
1527 			return err;
1528 	}
1529 
1530 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1531 	if (err) {
1532 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1533 			return err;
1534 		err = got_ref_open(&ref, repo, id_str, 0);
1535 		if (err != NULL)
1536 			goto done;
1537 		if (label) {
1538 			*label = strdup(got_ref_get_name(ref));
1539 			if (*label == NULL) {
1540 				err = got_error_from_errno("strdup");
1541 				goto done;
1542 			}
1543 		}
1544 		err = got_ref_resolve(id, repo, ref);
1545 	} else if (label) {
1546 		err = got_object_id_str(label, *id);
1547 		if (*label == NULL) {
1548 			err = got_error_from_errno("strdup");
1549 			goto done;
1550 		}
1551 	}
1552 done:
1553 	if (ref)
1554 		got_ref_close(ref);
1555 	return err;
1556 }
1557 
1558 const struct got_error *
1559 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1560     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1561 {
1562 	const struct got_error *err = NULL;
1563 	struct got_reflist_entry *re;
1564 	struct got_object_id *tag_id;
1565 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1566 
1567 	*tag = NULL;
1568 
1569 	TAILQ_FOREACH(re, refs, entry) {
1570 		const char *refname;
1571 		refname = got_ref_get_name(re->ref);
1572 		if (got_ref_is_symbolic(re->ref))
1573 			continue;
1574 		if (strncmp(refname, "refs/tags/", 10) != 0)
1575 			continue;
1576 		if (!name_is_absolute)
1577 			refname += strlen("refs/tags/");
1578 		if (strcmp(refname, name) != 0)
1579 			continue;
1580 		err = got_ref_resolve(&tag_id, repo, re->ref);
1581 		if (err)
1582 			break;
1583 		err = got_object_open_as_tag(tag, repo, tag_id);
1584 		free(tag_id);
1585 		if (err)
1586 			break;
1587 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1588 		    got_object_tag_get_object_type(*tag) == obj_type)
1589 			break;
1590 		got_object_tag_close(*tag);
1591 		*tag = NULL;
1592 	}
1593 
1594 	if (err == NULL && *tag == NULL)
1595 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1596 	return err;
1597 }
1598 
1599 static const struct got_error *
1600 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1601     const char *name, mode_t mode, struct got_object_id *blob_id)
1602 {
1603 	const struct got_error *err = NULL;
1604 
1605 	 *new_te = NULL;
1606 
1607 	*new_te = calloc(1, sizeof(**new_te));
1608 	if (*new_te == NULL)
1609 		return got_error_from_errno("calloc");
1610 
1611 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1612 	    sizeof((*new_te)->name)) {
1613 		err = got_error(GOT_ERR_NO_SPACE);
1614 		goto done;
1615 	}
1616 
1617 	if (S_ISLNK(mode)) {
1618 		(*new_te)->mode = S_IFLNK;
1619 	} else {
1620 		(*new_te)->mode = S_IFREG;
1621 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1622 	}
1623 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1624 done:
1625 	if (err && *new_te) {
1626 		free(*new_te);
1627 		*new_te = NULL;
1628 	}
1629 	return err;
1630 }
1631 
1632 static const struct got_error *
1633 import_file(struct got_tree_entry **new_te, struct dirent *de,
1634     const char *path, struct got_repository *repo)
1635 {
1636 	const struct got_error *err;
1637 	struct got_object_id *blob_id = NULL;
1638 	char *filepath;
1639 	struct stat sb;
1640 
1641 	if (asprintf(&filepath, "%s%s%s", path,
1642 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1643 		return got_error_from_errno("asprintf");
1644 
1645 	if (lstat(filepath, &sb) != 0) {
1646 		err = got_error_from_errno2("lstat", path);
1647 		goto done;
1648 	}
1649 
1650 	err = got_object_blob_create(&blob_id, filepath, repo);
1651 	if (err)
1652 		goto done;
1653 
1654 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1655 	    blob_id);
1656 done:
1657 	free(filepath);
1658 	if (err)
1659 		free(blob_id);
1660 	return err;
1661 }
1662 
1663 static const struct got_error *
1664 insert_tree_entry(struct got_tree_entry *new_te,
1665     struct got_pathlist_head *paths)
1666 {
1667 	const struct got_error *err = NULL;
1668 	struct got_pathlist_entry *new_pe;
1669 
1670 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1671 	if (err)
1672 		return err;
1673 	if (new_pe == NULL)
1674 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1675 	return NULL;
1676 }
1677 
1678 static const struct got_error *write_tree(struct got_object_id **,
1679     const char *, struct got_pathlist_head *, struct got_repository *,
1680     got_repo_import_cb progress_cb, void *progress_arg);
1681 
1682 static const struct got_error *
1683 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1684     const char *path, struct got_pathlist_head *ignores,
1685     struct got_repository *repo,
1686     got_repo_import_cb progress_cb, void *progress_arg)
1687 {
1688 	const struct got_error *err;
1689 	struct got_object_id *id = NULL;
1690 	char *subdirpath;
1691 
1692 	if (asprintf(&subdirpath, "%s%s%s", path,
1693 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1694 		return got_error_from_errno("asprintf");
1695 
1696 	(*new_te) = calloc(1, sizeof(**new_te));
1697 	if (*new_te == NULL)
1698 		return got_error_from_errno("calloc");
1699 	(*new_te)->mode = S_IFDIR;
1700 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1701 	    sizeof((*new_te)->name)) {
1702 		err = got_error(GOT_ERR_NO_SPACE);
1703 		goto done;
1704 	}
1705 	err = write_tree(&id, subdirpath, ignores,  repo,
1706 	    progress_cb, progress_arg);
1707 	if (err)
1708 		goto done;
1709 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1710 	
1711 done:
1712 	free(id);
1713 	free(subdirpath);
1714 	if (err) {
1715 		free(*new_te);
1716 		*new_te = NULL;
1717 	}
1718 	return err;
1719 }
1720 
1721 static const struct got_error *
1722 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1723     struct got_pathlist_head *ignores, struct got_repository *repo,
1724     got_repo_import_cb progress_cb, void *progress_arg)
1725 {
1726 	const struct got_error *err = NULL;
1727 	DIR *dir;
1728 	struct dirent *de;
1729 	int nentries;
1730 	struct got_tree_entry *new_te = NULL;
1731 	struct got_pathlist_head paths;
1732 	struct got_pathlist_entry *pe;
1733 
1734 	*new_tree_id = NULL;
1735 
1736 	TAILQ_INIT(&paths);
1737 
1738 	dir = opendir(path_dir);
1739 	if (dir == NULL) {
1740 		err = got_error_from_errno2("opendir", path_dir);
1741 		goto done;
1742 	}
1743 
1744 	nentries = 0;
1745 	while ((de = readdir(dir)) != NULL) {
1746 		int ignore = 0;
1747 		int type;
1748 
1749 		if (strcmp(de->d_name, ".") == 0 ||
1750 		    strcmp(de->d_name, "..") == 0)
1751 			continue;
1752 
1753 		TAILQ_FOREACH(pe, ignores, entry) {
1754 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1755 				ignore = 1;
1756 				break;
1757 			}
1758 		}
1759 		if (ignore)
1760 			continue;
1761 
1762 		err = got_path_dirent_type(&type, path_dir, de);
1763 		if (err)
1764 			goto done;
1765 
1766 		if (type == DT_DIR) {
1767 			err = import_subdir(&new_te, de, path_dir,
1768 			    ignores, repo, progress_cb, progress_arg);
1769 			if (err) {
1770 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1771 					goto done;
1772 				err = NULL;
1773 				continue;
1774 			}
1775 		} else if (type == DT_REG || type == DT_LNK) {
1776 			err = import_file(&new_te, de, path_dir, repo);
1777 			if (err)
1778 				goto done;
1779 		} else
1780 			continue;
1781 
1782 		err = insert_tree_entry(new_te, &paths);
1783 		if (err)
1784 			goto done;
1785 		nentries++;
1786 	}
1787 
1788 	if (TAILQ_EMPTY(&paths)) {
1789 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1790 		    "cannot create tree without any entries");
1791 		goto done;
1792 	}
1793 
1794 	TAILQ_FOREACH(pe, &paths, entry) {
1795 		struct got_tree_entry *te = pe->data;
1796 		char *path;
1797 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1798 			continue;
1799 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1800 			err = got_error_from_errno("asprintf");
1801 			goto done;
1802 		}
1803 		err = (*progress_cb)(progress_arg, path);
1804 		free(path);
1805 		if (err)
1806 			goto done;
1807 	}
1808 
1809 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1810 done:
1811 	if (dir)
1812 		closedir(dir);
1813 	got_pathlist_free(&paths);
1814 	return err;
1815 }
1816 
1817 const struct got_error *
1818 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1819     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1820     struct got_repository *repo, got_repo_import_cb progress_cb,
1821     void *progress_arg)
1822 {
1823 	const struct got_error *err;
1824 	struct got_object_id *new_tree_id;
1825 
1826 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1827 	    progress_cb, progress_arg);
1828 	if (err)
1829 		return err;
1830 
1831 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1832 	    author, time(NULL), author, time(NULL), logmsg, repo);
1833 	free(new_tree_id);
1834 	return err;
1835 }
1836 
1837 const struct got_error *
1838 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
1839     struct got_repository *repo)
1840 {
1841 	const struct got_error *err = NULL;
1842 	char *path_objects = NULL, *path = NULL;
1843 	DIR *dir = NULL;
1844 	struct got_object_id id;
1845 	int i;
1846 
1847 	*nobjects = 0;
1848 	*ondisk_size = 0;
1849 
1850 	path_objects = got_repo_get_path_objects(repo);
1851 	if (path_objects == NULL)
1852 		return got_error_from_errno("got_repo_get_path_objects");
1853 
1854 	for (i = 0; i <= 0xff; i++) {
1855 		struct dirent *dent;
1856 
1857 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
1858 			err = got_error_from_errno("asprintf");
1859 			break;
1860 		}
1861 
1862 		dir = opendir(path);
1863 		if (dir == NULL) {
1864 			if (errno == ENOENT) {
1865 				err = NULL;
1866 				continue;
1867 			}
1868 			err = got_error_from_errno2("opendir", path);
1869 			break;
1870 		}
1871 
1872 		while ((dent = readdir(dir)) != NULL) {
1873 			char *id_str;
1874 			int fd;
1875 			struct stat sb;
1876 
1877 			if (strcmp(dent->d_name, ".") == 0 ||
1878 			    strcmp(dent->d_name, "..") == 0)
1879 				continue;
1880 
1881 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
1882 				err = got_error_from_errno("asprintf");
1883 				goto done;
1884 			}
1885 
1886 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
1887 				free(id_str);
1888 				continue;
1889 			}
1890 			free(id_str);
1891 
1892 			err = got_object_open_loose_fd(&fd, &id, repo);
1893 			if (err)
1894 				goto done;
1895 
1896 			if (fstat(fd, &sb) == -1) {
1897 				err = got_error_from_errno("fstat");
1898 				close(fd);
1899 				goto done;
1900 			}
1901 			(*nobjects)++;
1902 			(*ondisk_size) += sb.st_size;
1903 
1904 			if (close(fd) == -1) {
1905 				err = got_error_from_errno("close");
1906 				goto done;
1907 			}
1908 		}
1909 
1910 		if (closedir(dir) != 0) {
1911 			err = got_error_from_errno("closedir");
1912 			goto done;
1913 		}
1914 		dir = NULL;
1915 
1916 		free(path);
1917 		path = NULL;
1918 	}
1919 done:
1920 	if (dir && closedir(dir) != 0 && err == NULL)
1921 		err = got_error_from_errno("closedir");
1922 
1923 	if (err) {
1924 		*nobjects = 0;
1925 		*ondisk_size = 0;
1926 	}
1927 	free(path_objects);
1928 	free(path);
1929 	return err;
1930 }
1931 
1932 const struct got_error *
1933 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
1934     off_t *total_packsize, struct got_repository *repo)
1935 {
1936 	const struct got_error *err = NULL;
1937 	DIR *packdir = NULL;
1938 	struct dirent *dent;
1939 	struct got_packidx *packidx = NULL;
1940 	char *path_packidx;
1941 	char *path_packfile;
1942 	int packdir_fd;
1943 	struct stat sb;
1944 
1945 	*npackfiles = 0;
1946 	*nobjects = 0;
1947 	*total_packsize = 0;
1948 
1949 	packdir_fd = openat(got_repo_get_fd(repo),
1950 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1951 	if (packdir_fd == -1) {
1952 		return got_error_from_errno_fmt("openat: %s/%s",
1953 		    got_repo_get_path_git_dir(repo),
1954 		    GOT_OBJECTS_PACK_DIR);
1955 	}
1956 
1957 	packdir = fdopendir(packdir_fd);
1958 	if (packdir == NULL) {
1959 		err = got_error_from_errno("fdopendir");
1960 		goto done;
1961 	}
1962 
1963 	while ((dent = readdir(packdir)) != NULL) {
1964 		if (!got_repo_is_packidx_filename(dent->d_name,
1965 		    strlen(dent->d_name)))
1966 			continue;
1967 
1968 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1969 		    dent->d_name) == -1) {
1970 			err = got_error_from_errno("asprintf");
1971 			goto done;
1972 		}
1973 
1974 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1975 		    path_packidx, 0);
1976 		free(path_packidx);
1977 		if (err)
1978 			goto done;
1979 
1980 		if (fstat(packidx->fd, &sb) == -1)
1981 			goto done;
1982 		*total_packsize += sb.st_size;
1983 
1984 		err = got_packidx_get_packfile_path(&path_packfile,
1985 		    packidx->path_packidx);
1986 		if (err)
1987 			goto done;
1988 
1989 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
1990 		    0) == -1) {
1991 			free(path_packfile);
1992 			goto done;
1993 		}
1994 		free(path_packfile);
1995 		*total_packsize += sb.st_size;
1996 
1997 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
1998 
1999 		(*npackfiles)++;
2000 
2001 		got_packidx_close(packidx);
2002 		packidx = NULL;
2003 	}
2004 done:
2005 	if (packidx)
2006 		got_packidx_close(packidx);
2007 	if (packdir && closedir(packdir) != 0 && err == NULL)
2008 		err = got_error_from_errno("closedir");
2009 	if (err) {
2010 		*npackfiles = 0;
2011 		*nobjects = 0;
2012 		*total_packsize = 0;
2013 	}
2014 	return err;
2015 }