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 <fcntl.h>
26 #include <fnmatch.h>
27 #include <limits.h>
28 #include <dirent.h>
29 #include <stdlib.h>
30 #include <stdio.h>
31 #include <string.h>
32 #include <time.h>
33 #include <unistd.h>
34 #include <zlib.h>
35 #include <errno.h>
36 #include <libgen.h>
37 #include <stdint.h>
38 #include <uuid.h>
39 
40 #include "got_compat.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 
49 #include "got_lib_delta.h"
50 #include "got_lib_inflate.h"
51 #include "got_lib_object.h"
52 #include "got_lib_object_parse.h"
53 #include "got_lib_object_create.h"
54 #include "got_lib_pack.h"
55 #include "got_lib_privsep.h"
56 #include "got_lib_sha1.h"
57 #include "got_lib_object_cache.h"
58 #include "got_lib_repository.h"
59 #include "got_lib_gotconfig.h"
60 
61 #ifndef nitems
62 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
63 #endif
64 
65 const char *
66 got_repo_get_path(struct got_repository *repo)
67 {
68 	return repo->path;
69 }
70 
71 const char *
72 got_repo_get_path_git_dir(struct got_repository *repo)
73 {
74 	return repo->path_git_dir;
75 }
76 
77 int
78 got_repo_get_fd(struct got_repository *repo)
79 {
80 	return repo->gitdir_fd;
81 }
82 
83 const char *
84 got_repo_get_gitconfig_author_name(struct got_repository *repo)
85 {
86 	return repo->gitconfig_author_name;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_email(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_email;
93 }
94 
95 const char *
96 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
97 {
98 	return repo->global_gitconfig_author_name;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_email;
105 }
106 
107 const char *
108 got_repo_get_gitconfig_owner(struct got_repository *repo)
109 {
110 	return repo->gitconfig_owner;
111 }
112 
113 void
114 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
115     struct got_repository *repo)
116 {
117 	*extensions = repo->extensions;
118 	*nextensions = repo->nextensions;
119 }
120 
121 int
122 got_repo_is_bare(struct got_repository *repo)
123 {
124 	return (strcmp(repo->path, repo->path_git_dir) == 0);
125 }
126 
127 static char *
128 get_path_git_child(struct got_repository *repo, const char *basename)
129 {
130 	char *path_child;
131 
132 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
133 	    basename) == -1)
134 		return NULL;
135 
136 	return path_child;
137 }
138 
139 char *
140 got_repo_get_path_objects(struct got_repository *repo)
141 {
142 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
143 }
144 
145 char *
146 got_repo_get_path_objects_pack(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
149 }
150 
151 char *
152 got_repo_get_path_refs(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_REFS_DIR);
155 }
156 
157 char *
158 got_repo_get_path_packed_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
161 }
162 
163 static char *
164 get_path_head(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_HEAD_FILE);
167 }
168 
169 char *
170 got_repo_get_path_gitconfig(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_GITCONFIG);
173 }
174 
175 char *
176 got_repo_get_path_gotconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
179 }
180 
181 const struct got_gotconfig *
182 got_repo_get_gotconfig(struct got_repository *repo)
183 {
184 	return repo->gotconfig;
185 }
186 
187 void
188 got_repo_get_gitconfig_remotes(int *nremotes,
189     const struct got_remote_repo **remotes, struct got_repository *repo)
190 {
191 	*nremotes = repo->ngitconfig_remotes;
192 	*remotes = repo->gitconfig_remotes;
193 }
194 
195 static int
196 is_git_repo(struct got_repository *repo)
197 {
198 	const char *path_git = got_repo_get_path_git_dir(repo);
199 	char *path_objects = got_repo_get_path_objects(repo);
200 	char *path_refs = got_repo_get_path_refs(repo);
201 	char *path_head = get_path_head(repo);
202 	int ret = 0;
203 	struct stat sb;
204 	struct got_reference *head_ref;
205 
206 	if (lstat(path_git, &sb) == -1)
207 		goto done;
208 	if (!S_ISDIR(sb.st_mode))
209 		goto done;
210 
211 	if (lstat(path_objects, &sb) == -1)
212 		goto done;
213 	if (!S_ISDIR(sb.st_mode))
214 		goto done;
215 
216 	if (lstat(path_refs, &sb) == -1)
217 		goto done;
218 	if (!S_ISDIR(sb.st_mode))
219 		goto done;
220 
221 	if (lstat(path_head, &sb) == -1)
222 		goto done;
223 	if (!S_ISREG(sb.st_mode))
224 		goto done;
225 
226 	/* Check if the HEAD reference can be opened. */
227 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
228 		goto done;
229 	got_ref_close(head_ref);
230 
231 	ret = 1;
232 done:
233 	free(path_objects);
234 	free(path_refs);
235 	free(path_head);
236 	return ret;
237 
238 }
239 
240 const struct got_error *
241 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
242     struct got_object *obj)
243 {
244 #ifndef GOT_NO_OBJ_CACHE
245 	const struct got_error *err = NULL;
246 	err = got_object_cache_add(&repo->objcache, id, obj);
247 	if (err) {
248 		if (err->code == GOT_ERR_OBJ_EXISTS ||
249 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
250 			err = NULL;
251 		return err;
252 	}
253 	obj->refcnt++;
254 #endif
255 	return NULL;
256 }
257 
258 struct got_object *
259 got_repo_get_cached_object(struct got_repository *repo,
260     struct got_object_id *id)
261 {
262 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
263 }
264 
265 const struct got_error *
266 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
267     struct got_tree_object *tree)
268 {
269 #ifndef GOT_NO_OBJ_CACHE
270 	const struct got_error *err = NULL;
271 	err = got_object_cache_add(&repo->treecache, id, tree);
272 	if (err) {
273 		if (err->code == GOT_ERR_OBJ_EXISTS ||
274 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
275 			err = NULL;
276 		return err;
277 	}
278 	tree->refcnt++;
279 #endif
280 	return NULL;
281 }
282 
283 struct got_tree_object *
284 got_repo_get_cached_tree(struct got_repository *repo,
285     struct got_object_id *id)
286 {
287 	return (struct got_tree_object *)got_object_cache_get(
288 	    &repo->treecache, id);
289 }
290 
291 const struct got_error *
292 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
293     struct got_commit_object *commit)
294 {
295 #ifndef GOT_NO_OBJ_CACHE
296 	const struct got_error *err = NULL;
297 	err = got_object_cache_add(&repo->commitcache, id, commit);
298 	if (err) {
299 		if (err->code == GOT_ERR_OBJ_EXISTS ||
300 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
301 			err = NULL;
302 		return err;
303 	}
304 	commit->refcnt++;
305 #endif
306 	return NULL;
307 }
308 
309 struct got_commit_object *
310 got_repo_get_cached_commit(struct got_repository *repo,
311     struct got_object_id *id)
312 {
313 	return (struct got_commit_object *)got_object_cache_get(
314 	    &repo->commitcache, id);
315 }
316 
317 const struct got_error *
318 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
319     struct got_tag_object *tag)
320 {
321 #ifndef GOT_NO_OBJ_CACHE
322 	const struct got_error *err = NULL;
323 	err = got_object_cache_add(&repo->tagcache, id, tag);
324 	if (err) {
325 		if (err->code == GOT_ERR_OBJ_EXISTS ||
326 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
327 			err = NULL;
328 		return err;
329 	}
330 	tag->refcnt++;
331 #endif
332 	return NULL;
333 }
334 
335 struct got_tag_object *
336 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
337 {
338 	return (struct got_tag_object *)got_object_cache_get(
339 	    &repo->tagcache, id);
340 }
341 
342 static const struct got_error *
343 open_repo(struct got_repository *repo, const char *path)
344 {
345 	const struct got_error *err = NULL;
346 
347 	repo->gitdir_fd = -1;
348 
349 	/* bare git repository? */
350 	repo->path_git_dir = strdup(path);
351 	if (repo->path_git_dir == NULL)
352 		return got_error_from_errno("strdup");
353 	if (is_git_repo(repo)) {
354 		repo->path = strdup(repo->path_git_dir);
355 		if (repo->path == NULL) {
356 			err = got_error_from_errno("strdup");
357 			goto done;
358 		}
359 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
360 		if (repo->gitdir_fd == -1) {
361 			err = got_error_from_errno2("open",
362 			    repo->path_git_dir);
363 			goto done;
364 		}
365 		return NULL;
366 	}
367 
368 	/* git repository with working tree? */
369 	free(repo->path_git_dir);
370 	repo->path_git_dir = NULL;
371 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
372 		err = got_error_from_errno("asprintf");
373 		goto done;
374 	}
375 	if (is_git_repo(repo)) {
376 		repo->path = strdup(path);
377 		if (repo->path == NULL) {
378 			err = got_error_from_errno("strdup");
379 			goto done;
380 		}
381 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
382 		if (repo->gitdir_fd == -1) {
383 			err = got_error_from_errno2("open",
384 			    repo->path_git_dir);
385 			goto done;
386 		}
387 		return NULL;
388 	}
389 
390 	err = got_error(GOT_ERR_NOT_GIT_REPO);
391 done:
392 	if (err) {
393 		free(repo->path);
394 		repo->path = NULL;
395 		free(repo->path_git_dir);
396 		repo->path_git_dir = NULL;
397 		if (repo->gitdir_fd != -1)
398 			close(repo->gitdir_fd);
399 		repo->gitdir_fd = -1;
400 
401 	}
402 	return err;
403 }
404 
405 static const struct got_error *
406 parse_gitconfig_file(int *gitconfig_repository_format_version,
407     char **gitconfig_author_name, char **gitconfig_author_email,
408     struct got_remote_repo **remotes, int *nremotes,
409     char **gitconfig_owner, char ***extensions, int *nextensions,
410     const char *gitconfig_path)
411 {
412 	const struct got_error *err = NULL, *child_err = NULL;
413 	int fd = -1;
414 	int imsg_fds[2] = { -1, -1 };
415 	pid_t pid;
416 	struct imsgbuf *ibuf;
417 
418 	*gitconfig_repository_format_version = 0;
419 	if (extensions)
420 		*extensions = NULL;
421 	if (nextensions)
422 		*nextensions = 0;
423 	*gitconfig_author_name = NULL;
424 	*gitconfig_author_email = NULL;
425 	if (remotes)
426 		*remotes = NULL;
427 	if (nremotes)
428 		*nremotes = 0;
429 	if (gitconfig_owner)
430 		*gitconfig_owner = NULL;
431 
432 	fd = open(gitconfig_path, O_RDONLY);
433 	if (fd == -1) {
434 		if (errno == ENOENT)
435 			return NULL;
436 		return got_error_from_errno2("open", gitconfig_path);
437 	}
438 
439 	ibuf = calloc(1, sizeof(*ibuf));
440 	if (ibuf == NULL) {
441 		err = got_error_from_errno("calloc");
442 		goto done;
443 	}
444 
445 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
446 		err = got_error_from_errno("socketpair");
447 		goto done;
448 	}
449 
450 	pid = fork();
451 	if (pid == -1) {
452 		err = got_error_from_errno("fork");
453 		goto done;
454 	} else if (pid == 0) {
455 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
456 		    gitconfig_path);
457 		/* not reached */
458 	}
459 
460 	if (close(imsg_fds[1]) == -1) {
461 		err = got_error_from_errno("close");
462 		goto done;
463 	}
464 	imsg_fds[1] = -1;
465 	imsg_init(ibuf, imsg_fds[0]);
466 
467 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
468 	if (err)
469 		goto done;
470 	fd = -1;
471 
472 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
473 	if (err)
474 		goto done;
475 
476 	err = got_privsep_recv_gitconfig_int(
477 	    gitconfig_repository_format_version, ibuf);
478 	if (err)
479 		goto done;
480 
481 	if (extensions && nextensions) {
482 		err = got_privsep_send_gitconfig_repository_extensions_req(
483 		    ibuf);
484 		if (err)
485 			goto done;
486 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
487 		if (err)
488 			goto done;
489 		if (*nextensions > 0) {
490 			int i;
491 			*extensions = calloc(*nextensions, sizeof(char *));
492 			if (*extensions == NULL) {
493 				err = got_error_from_errno("calloc");
494 				goto done;
495 			}
496 			for (i = 0; i < *nextensions; i++) {
497 				char *ext;
498 				err = got_privsep_recv_gitconfig_str(&ext,
499 				    ibuf);
500 				if (err)
501 					goto done;
502 				(*extensions)[i] = ext;
503 			}
504 		}
505 	}
506 
507 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
508 	if (err)
509 		goto done;
510 
511 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
512 	if (err)
513 		goto done;
514 
515 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
516 	if (err)
517 		goto done;
518 
519 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
520 	if (err)
521 		goto done;
522 
523 	if (remotes && nremotes) {
524 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
525 		if (err)
526 			goto done;
527 
528 		err = got_privsep_recv_gitconfig_remotes(remotes,
529 		    nremotes, ibuf);
530 		if (err)
531 			goto done;
532 	}
533 
534 	if (gitconfig_owner) {
535 		err = got_privsep_send_gitconfig_owner_req(ibuf);
536 		if (err)
537 			goto done;
538 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
539 		if (err)
540 			goto done;
541 	}
542 
543 	imsg_clear(ibuf);
544 	err = got_privsep_send_stop(imsg_fds[0]);
545 	child_err = got_privsep_wait_for_child(pid);
546 	if (child_err && err == NULL)
547 		err = child_err;
548 done:
549 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
550 		err = got_error_from_errno("close");
551 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
552 		err = got_error_from_errno("close");
553 	if (fd != -1 && close(fd) == -1 && err == NULL)
554 		err = got_error_from_errno2("close", gitconfig_path);
555 	free(ibuf);
556 	return err;
557 }
558 
559 static const struct got_error *
560 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
561 {
562 	const struct got_error *err = NULL;
563 	char *repo_gitconfig_path = NULL;
564 
565 	if (global_gitconfig_path) {
566 		/* Read settings from ~/.gitconfig. */
567 		int dummy_repo_version;
568 		err = parse_gitconfig_file(&dummy_repo_version,
569 		    &repo->global_gitconfig_author_name,
570 		    &repo->global_gitconfig_author_email,
571 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
572 		if (err)
573 			return err;
574 	}
575 
576 	/* Read repository's .git/config file. */
577 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
578 	if (repo_gitconfig_path == NULL)
579 		return got_error_from_errno("got_repo_get_path_gitconfig");
580 
581 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
582 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
583 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
584 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
585 	    repo_gitconfig_path);
586 	if (err)
587 		goto done;
588 done:
589 	free(repo_gitconfig_path);
590 	return err;
591 }
592 
593 static const struct got_error *
594 read_gotconfig(struct got_repository *repo)
595 {
596 	const struct got_error *err = NULL;
597 	char *gotconfig_path;
598 
599 	gotconfig_path = got_repo_get_path_gotconfig(repo);
600 	if (gotconfig_path == NULL)
601 		return got_error_from_errno("got_repo_get_path_gotconfig");
602 
603 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
604 	free(gotconfig_path);
605 	return err;
606 }
607 
608 /* Supported repository format extensions. */
609 static const char *repo_extensions[] = {
610 	"noop",			/* Got supports repository format version 1. */
611 	"preciousObjects",	/* Supported by gotadmin cleanup. */
612 	"worktreeConfig",	/* Got does not care about Git work trees. */
613 };
614 
615 const struct got_error *
616 got_repo_open(struct got_repository **repop, const char *path,
617     const char *global_gitconfig_path)
618 {
619 	struct got_repository *repo = NULL;
620 	const struct got_error *err = NULL;
621 	char *repo_path = NULL;
622 	size_t i;
623 	struct rlimit rl;
624 
625 	*repop = NULL;
626 
627 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
628 		return got_error_from_errno("getrlimit");
629 
630 	repo = calloc(1, sizeof(*repo));
631 	if (repo == NULL) {
632 		err = got_error_from_errno("calloc");
633 		goto done;
634 	}
635 
636 	for (i = 0; i < nitems(repo->privsep_children); i++) {
637 		memset(&repo->privsep_children[i], 0,
638 		    sizeof(repo->privsep_children[0]));
639 		repo->privsep_children[i].imsg_fd = -1;
640 	}
641 
642 	err = got_object_cache_init(&repo->objcache,
643 	    GOT_OBJECT_CACHE_TYPE_OBJ);
644 	if (err)
645 		goto done;
646 	err = got_object_cache_init(&repo->treecache,
647 	    GOT_OBJECT_CACHE_TYPE_TREE);
648 	if (err)
649 		goto done;
650 	err = got_object_cache_init(&repo->commitcache,
651 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
652 	if (err)
653 		goto done;
654 	err = got_object_cache_init(&repo->tagcache,
655 	    GOT_OBJECT_CACHE_TYPE_TAG);
656 	if (err)
657 		goto done;
658 
659 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
660 	if (repo->pack_cache_size > rl.rlim_cur / 8)
661 		repo->pack_cache_size = rl.rlim_cur / 8;
662 
663 	repo_path = realpath(path, NULL);
664 	if (repo_path == NULL) {
665 		err = got_error_from_errno2("realpath", path);
666 		goto done;
667 	}
668 
669 	for (;;) {
670 		char *parent_path;
671 
672 		err = open_repo(repo, repo_path);
673 		if (err == NULL)
674 			break;
675 		if (err->code != GOT_ERR_NOT_GIT_REPO)
676 			goto done;
677 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
678 			err = got_error(GOT_ERR_NOT_GIT_REPO);
679 			goto done;
680 		}
681 		err = got_path_dirname(&parent_path, repo_path);
682 		if (err)
683 			goto done;
684 		free(repo_path);
685 		repo_path = parent_path;
686 	}
687 
688 	err = read_gotconfig(repo);
689 	if (err)
690 		goto done;
691 
692 	err = read_gitconfig(repo, global_gitconfig_path);
693 	if (err)
694 		goto done;
695 	if (repo->gitconfig_repository_format_version != 0)
696 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
697 	for (i = 0; i < repo->nextensions; i++) {
698 		char *ext = repo->extensions[i];
699 		int j, supported = 0;
700 		for (j = 0; j < nitems(repo_extensions); j++) {
701 			if (strcmp(ext, repo_extensions[j]) == 0) {
702 				supported = 1;
703 				break;
704 			}
705 		}
706 		if (!supported) {
707 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
708 			goto done;
709 		}
710 	}
711 done:
712 	if (err)
713 		got_repo_close(repo);
714 	else
715 		*repop = repo;
716 	free(repo_path);
717 	return err;
718 }
719 
720 const struct got_error *
721 got_repo_close(struct got_repository *repo)
722 {
723 	const struct got_error *err = NULL, *child_err;
724 	size_t i;
725 
726 	for (i = 0; i < repo->pack_cache_size; i++) {
727 		if (repo->packidx_cache[i] == NULL)
728 			break;
729 		got_packidx_close(repo->packidx_cache[i]);
730 	}
731 
732 	for (i = 0; i < repo->pack_cache_size; i++) {
733 		if (repo->packs[i].path_packfile == NULL)
734 			break;
735 		got_pack_close(&repo->packs[i]);
736 	}
737 
738 	free(repo->path);
739 	free(repo->path_git_dir);
740 
741 	got_object_cache_close(&repo->objcache);
742 	got_object_cache_close(&repo->treecache);
743 	got_object_cache_close(&repo->commitcache);
744 	got_object_cache_close(&repo->tagcache);
745 
746 	for (i = 0; i < nitems(repo->privsep_children); i++) {
747 		if (repo->privsep_children[i].imsg_fd == -1)
748 			continue;
749 		imsg_clear(repo->privsep_children[i].ibuf);
750 		free(repo->privsep_children[i].ibuf);
751 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
752 		child_err = got_privsep_wait_for_child(
753 		    repo->privsep_children[i].pid);
754 		if (child_err && err == NULL)
755 			err = child_err;
756 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
757 		    err == NULL)
758 			err = got_error_from_errno("close");
759 	}
760 
761 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
762 	    err == NULL)
763 		err = got_error_from_errno("close");
764 
765 	if (repo->gotconfig)
766 		got_gotconfig_free(repo->gotconfig);
767 	free(repo->gitconfig_author_name);
768 	free(repo->gitconfig_author_email);
769 	for (i = 0; i < repo->ngitconfig_remotes; i++)
770 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
771 	free(repo->gitconfig_remotes);
772 	for (i = 0; i < repo->nextensions; i++)
773 		free(repo->extensions[i]);
774 	free(repo->extensions);
775 	free(repo);
776 
777 	return err;
778 }
779 
780 void
781 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
782 {
783 	int i;
784 
785 	free(repo->name);
786 	repo->name = NULL;
787 	free(repo->fetch_url);
788 	repo->fetch_url = NULL;
789 	free(repo->send_url);
790 	repo->send_url = NULL;
791 	for (i = 0; i < repo->nfetch_branches; i++)
792 		free(repo->fetch_branches[i]);
793 	free(repo->fetch_branches);
794 	repo->fetch_branches = NULL;
795 	repo->nfetch_branches = 0;
796 	for (i = 0; i < repo->nsend_branches; i++)
797 		free(repo->send_branches[i]);
798 	free(repo->send_branches);
799 	repo->send_branches = NULL;
800 	repo->nsend_branches = 0;
801 }
802 
803 const struct got_error *
804 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
805     const char *input_path)
806 {
807 	const struct got_error *err = NULL;
808 	const char *repo_abspath = NULL;
809 	size_t repolen, len;
810 	char *canonpath, *path = NULL;
811 
812 	*in_repo_path = NULL;
813 
814 	canonpath = strdup(input_path);
815 	if (canonpath == NULL) {
816 		err = got_error_from_errno("strdup");
817 		goto done;
818 	}
819 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
820 	if (err)
821 		goto done;
822 
823 	repo_abspath = got_repo_get_path(repo);
824 
825 	if (canonpath[0] == '\0') {
826 		path = strdup(canonpath);
827 		if (path == NULL) {
828 			err = got_error_from_errno("strdup");
829 			goto done;
830 		}
831 	} else {
832 		path = realpath(canonpath, NULL);
833 		if (path == NULL) {
834 			if (errno != ENOENT) {
835 				err = got_error_from_errno2("realpath",
836 				    canonpath);
837 				goto done;
838 			}
839 			/*
840 			 * Path is not on disk.
841 			 * Assume it is already relative to repository root.
842 			 */
843 			path = strdup(canonpath);
844 			if (path == NULL) {
845 				err = got_error_from_errno("strdup");
846 				goto done;
847 			}
848 		}
849 
850 		repolen = strlen(repo_abspath);
851 		len = strlen(path);
852 
853 
854 		if (strcmp(path, repo_abspath) == 0) {
855 			free(path);
856 			path = strdup("");
857 			if (path == NULL) {
858 				err = got_error_from_errno("strdup");
859 				goto done;
860 			}
861 		} else if (len > repolen &&
862 		    got_path_is_child(path, repo_abspath, repolen)) {
863 			/* Matched an on-disk path inside repository. */
864 			if (got_repo_is_bare(repo)) {
865 				/*
866 				 * Matched an on-disk path inside repository
867 				 * database. Treat input as repository-relative.
868 				 */
869 				free(path);
870 				path = canonpath;
871 				canonpath = NULL;
872 			} else {
873 				char *child;
874 				/* Strip common prefix with repository path. */
875 				err = got_path_skip_common_ancestor(&child,
876 				    repo_abspath, path);
877 				if (err)
878 					goto done;
879 				free(path);
880 				path = child;
881 			}
882 		} else {
883 			/*
884 			 * Matched unrelated on-disk path.
885 			 * Treat input as repository-relative.
886 			 */
887 			free(path);
888 			path = canonpath;
889 			canonpath = NULL;
890 		}
891 	}
892 
893 	/* Make in-repository path absolute */
894 	if (path[0] != '/') {
895 		char *abspath;
896 		if (asprintf(&abspath, "/%s", path) == -1) {
897 			err = got_error_from_errno("asprintf");
898 			goto done;
899 		}
900 		free(path);
901 		path = abspath;
902 	}
903 
904 done:
905 	free(canonpath);
906 	if (err)
907 		free(path);
908 	else
909 		*in_repo_path = path;
910 	return err;
911 }
912 
913 static const struct got_error *
914 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
915     const char *path_packidx)
916 {
917 	const struct got_error *err = NULL;
918 	size_t i;
919 
920 	for (i = 0; i < repo->pack_cache_size; i++) {
921 		if (repo->packidx_cache[i] == NULL)
922 			break;
923 		if (strcmp(repo->packidx_cache[i]->path_packidx,
924 		    path_packidx) == 0) {
925 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
926 		}
927 	}
928 	if (i == repo->pack_cache_size) {
929 		err = got_packidx_close(repo->packidx_cache[i - 1]);
930 		if (err)
931 			return err;
932 	}
933 
934 	/*
935 	 * Insert the new pack index at the front so it will
936 	 * be searched first in the future.
937 	 */
938 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
939 	    sizeof(repo->packidx_cache) -
940 	    sizeof(repo->packidx_cache[0]));
941 	repo->packidx_cache[0] = packidx;
942 
943 	return NULL;
944 }
945 
946 int
947 got_repo_is_packidx_filename(const char *name, size_t len)
948 {
949 	if (len != GOT_PACKIDX_NAMELEN)
950 		return 0;
951 
952 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
953 		return 0;
954 
955 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
956 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
957 		return 0;
958 
959 	return 1;
960 }
961 
962 const struct got_error *
963 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
964     struct got_repository *repo, struct got_object_id *id)
965 {
966 	const struct got_error *err;
967 	DIR *packdir = NULL;
968 	struct dirent *dent;
969 	char *path_packidx;
970 	size_t i;
971 	int packdir_fd;
972 
973 	/* Search pack index cache. */
974 	for (i = 0; i < repo->pack_cache_size; i++) {
975 		if (repo->packidx_cache[i] == NULL)
976 			break;
977 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
978 		if (*idx != -1) {
979 			*packidx = repo->packidx_cache[i];
980 			/*
981 			 * Move this cache entry to the front. Repeatedly
982 			 * searching a wrong pack index can be expensive.
983 			 */
984 			if (i > 0) {
985 				struct got_packidx *p;
986 				p = repo->packidx_cache[0];
987 				repo->packidx_cache[0] = *packidx;
988 				repo->packidx_cache[i] = p;
989 			}
990 			return NULL;
991 		}
992 	}
993 	/* No luck. Search the filesystem. */
994 
995 	packdir_fd = openat(got_repo_get_fd(repo),
996 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
997 	if (packdir_fd == -1) {
998 		if (errno == ENOENT)
999 			err = got_error_no_obj(id);
1000 		else
1001 			err = got_error_from_errno_fmt("openat: %s/%s",
1002 			    got_repo_get_path_git_dir(repo),
1003 			    GOT_OBJECTS_PACK_DIR);
1004 		goto done;
1005 	}
1006 
1007 	packdir = fdopendir(packdir_fd);
1008 	if (packdir == NULL) {
1009 		err = got_error_from_errno("fdopendir");
1010 		goto done;
1011 	}
1012 
1013 	while ((dent = readdir(packdir)) != NULL) {
1014 		int is_cached = 0;
1015 
1016 		if (!got_repo_is_packidx_filename(dent->d_name,
1017 		    strlen(dent->d_name)))
1018 			continue;
1019 
1020 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1021 		    dent->d_name) == -1) {
1022 			err = got_error_from_errno("asprintf");
1023 			goto done;
1024 		}
1025 
1026 		for (i = 0; i < repo->pack_cache_size; i++) {
1027 			if (repo->packidx_cache[i] == NULL)
1028 				break;
1029 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1030 			    path_packidx) == 0) {
1031 				is_cached = 1;
1032 				break;
1033 			}
1034 		}
1035 		if (is_cached) {
1036 			free(path_packidx);
1037 			continue; /* already searched */
1038 		}
1039 
1040 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1041 		    path_packidx, 0);
1042 		if (err) {
1043 			free(path_packidx);
1044 			goto done;
1045 		}
1046 
1047 		err = cache_packidx(repo, *packidx, path_packidx);
1048 		free(path_packidx);
1049 		if (err)
1050 			goto done;
1051 
1052 		*idx = got_packidx_get_object_idx(*packidx, id);
1053 		if (*idx != -1) {
1054 			err = NULL; /* found the object */
1055 			goto done;
1056 		}
1057 	}
1058 
1059 	err = got_error_no_obj(id);
1060 done:
1061 	if (packdir && closedir(packdir) != 0 && err == NULL)
1062 		err = got_error_from_errno("closedir");
1063 	return err;
1064 }
1065 
1066 static const struct got_error *
1067 read_packfile_hdr(int fd, struct got_packidx *packidx)
1068 {
1069 	const struct got_error *err = NULL;
1070 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1071 	struct got_packfile_hdr hdr;
1072 	ssize_t n;
1073 
1074 	n = read(fd, &hdr, sizeof(hdr));
1075 	if (n < 0)
1076 		return got_error_from_errno("read");
1077 	if (n != sizeof(hdr))
1078 		return got_error(GOT_ERR_BAD_PACKFILE);
1079 
1080 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1081 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1082 	    be32toh(hdr.nobjects) != totobj)
1083 		err = got_error(GOT_ERR_BAD_PACKFILE);
1084 
1085 	return err;
1086 }
1087 
1088 static const struct got_error *
1089 open_packfile(int *fd, struct got_repository *repo,
1090     const char *relpath, struct got_packidx *packidx)
1091 {
1092 	const struct got_error *err = NULL;
1093 
1094 	*fd = openat(got_repo_get_fd(repo), relpath, O_RDONLY | O_NOFOLLOW);
1095 	if (*fd == -1)
1096 		return got_error_from_errno_fmt("openat: %s/%s",
1097 		    got_repo_get_path_git_dir(repo), relpath);
1098 
1099 	if (packidx) {
1100 		err = read_packfile_hdr(*fd, packidx);
1101 		if (err) {
1102 			close(*fd);
1103 			*fd = -1;
1104 		}
1105 	}
1106 
1107 	return err;
1108 }
1109 
1110 const struct got_error *
1111 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1112     const char *path_packfile, struct got_packidx *packidx)
1113 {
1114 	const struct got_error *err = NULL;
1115 	struct got_pack *pack = NULL;
1116 	struct stat sb;
1117 	size_t i;
1118 
1119 	if (packp)
1120 		*packp = NULL;
1121 
1122 	for (i = 0; i < repo->pack_cache_size; i++) {
1123 		pack = &repo->packs[i];
1124 		if (pack->path_packfile == NULL)
1125 			break;
1126 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1127 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1128 	}
1129 
1130 	if (i == repo->pack_cache_size) {
1131 		err = got_pack_close(&repo->packs[i - 1]);
1132 		if (err)
1133 			return err;
1134 		memmove(&repo->packs[1], &repo->packs[0],
1135 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1136 		i = 0;
1137 	}
1138 
1139 	pack = &repo->packs[i];
1140 
1141 	pack->path_packfile = strdup(path_packfile);
1142 	if (pack->path_packfile == NULL) {
1143 		err = got_error_from_errno("strdup");
1144 		goto done;
1145 	}
1146 
1147 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1148 	if (err)
1149 		goto done;
1150 
1151 	if (fstat(pack->fd, &sb) != 0) {
1152 		err = got_error_from_errno("fstat");
1153 		goto done;
1154 	}
1155 	pack->filesize = sb.st_size;
1156 
1157 	pack->privsep_child = NULL;
1158 
1159 #ifndef GOT_PACK_NO_MMAP
1160 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1161 	    pack->fd, 0);
1162 	if (pack->map == MAP_FAILED) {
1163 		if (errno != ENOMEM) {
1164 			err = got_error_from_errno("mmap");
1165 			goto done;
1166 		}
1167 		pack->map = NULL; /* fall back to read(2) */
1168 	}
1169 #endif
1170 done:
1171 	if (err) {
1172 		if (pack) {
1173 			free(pack->path_packfile);
1174 			memset(pack, 0, sizeof(*pack));
1175 		}
1176 	} else if (packp)
1177 		*packp = pack;
1178 	return err;
1179 }
1180 
1181 struct got_pack *
1182 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1183 {
1184 	struct got_pack *pack = NULL;
1185 	size_t i;
1186 
1187 	for (i = 0; i < repo->pack_cache_size; i++) {
1188 		pack = &repo->packs[i];
1189 		if (pack->path_packfile == NULL)
1190 			break;
1191 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1192 			return pack;
1193 	}
1194 
1195 	return NULL;
1196 }
1197 
1198 const struct got_error *
1199 got_repo_init(const char *repo_path)
1200 {
1201 	const struct got_error *err = NULL;
1202 	const char *dirnames[] = {
1203 		GOT_OBJECTS_DIR,
1204 		GOT_OBJECTS_PACK_DIR,
1205 		GOT_REFS_DIR,
1206 	};
1207 	const char *description_str = "Unnamed repository; "
1208 	    "edit this file 'description' to name the repository.";
1209 	const char *headref_str = "ref: refs/heads/main";
1210 	const char *gitconfig_str = "[core]\n"
1211 	    "\trepositoryformatversion = 0\n"
1212 	    "\tfilemode = true\n"
1213 	    "\tbare = true\n";
1214 	char *path;
1215 	size_t i;
1216 
1217 	if (!got_path_dir_is_empty(repo_path))
1218 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1219 
1220 	for (i = 0; i < nitems(dirnames); i++) {
1221 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1222 			return got_error_from_errno("asprintf");
1223 		}
1224 		err = got_path_mkdir(path);
1225 		free(path);
1226 		if (err)
1227 			return err;
1228 	}
1229 
1230 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1231 		return got_error_from_errno("asprintf");
1232 	err = got_path_create_file(path, description_str);
1233 	free(path);
1234 	if (err)
1235 		return err;
1236 
1237 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1238 		return got_error_from_errno("asprintf");
1239 	err = got_path_create_file(path, headref_str);
1240 	free(path);
1241 	if (err)
1242 		return err;
1243 
1244 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1245 		return got_error_from_errno("asprintf");
1246 	err = got_path_create_file(path, gitconfig_str);
1247 	free(path);
1248 	if (err)
1249 		return err;
1250 
1251 	return NULL;
1252 }
1253 
1254 static const struct got_error *
1255 match_packed_object(struct got_object_id **unique_id,
1256     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1257 {
1258 	const struct got_error *err = NULL;
1259 	DIR *packdir = NULL;
1260 	struct dirent *dent;
1261 	char *path_packidx;
1262 	struct got_object_id_queue matched_ids;
1263 	int packdir_fd;
1264 
1265 	STAILQ_INIT(&matched_ids);
1266 
1267 	packdir_fd = openat(got_repo_get_fd(repo),
1268 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1269 	if (packdir_fd == -1) {
1270 		if (errno != ENOENT)
1271 			err = got_error_from_errno2("openat", GOT_OBJECTS_PACK_DIR);
1272 		goto done;
1273 	}
1274 
1275 	packdir = fdopendir(packdir_fd);
1276 	if (packdir == NULL) {
1277 		err = got_error_from_errno("fdopendir");
1278 		goto done;
1279 	}
1280 
1281 	while ((dent = readdir(packdir)) != NULL) {
1282 		struct got_packidx *packidx;
1283 		struct got_object_qid *qid;
1284 
1285 		if (!got_repo_is_packidx_filename(dent->d_name,
1286 		    strlen(dent->d_name)))
1287 			continue;
1288 
1289 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1290 		    dent->d_name) == -1) {
1291 			err = got_error_from_errno("strdup");
1292 			break;
1293 		}
1294 
1295 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1296 		    path_packidx, 0);
1297 		free(path_packidx);
1298 		if (err)
1299 			break;
1300 
1301 		err = got_packidx_match_id_str_prefix(&matched_ids,
1302 		    packidx, id_str_prefix);
1303 		if (err) {
1304 			got_packidx_close(packidx);
1305 			break;
1306 		}
1307 		err = got_packidx_close(packidx);
1308 		if (err)
1309 			break;
1310 
1311 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1312 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1313 				int matched_type;
1314 				err = got_object_get_type(&matched_type, repo,
1315 				    qid->id);
1316 				if (err)
1317 					goto done;
1318 				if (matched_type != obj_type)
1319 					continue;
1320 			}
1321 			if (*unique_id == NULL) {
1322 				*unique_id = got_object_id_dup(qid->id);
1323 				if (*unique_id == NULL) {
1324 					err = got_error_from_errno("malloc");
1325 					goto done;
1326 				}
1327 			} else {
1328 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1329 					continue; /* packed multiple times */
1330 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1331 				goto done;
1332 			}
1333 		}
1334 	}
1335 done:
1336 	got_object_id_queue_free(&matched_ids);
1337 	if (packdir && closedir(packdir) != 0 && err == NULL)
1338 		err = got_error_from_errno("closedir");
1339 	if (err) {
1340 		free(*unique_id);
1341 		*unique_id = NULL;
1342 	}
1343 	return err;
1344 }
1345 
1346 static const struct got_error *
1347 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1348     const char *object_dir, const char *id_str_prefix, int obj_type,
1349     struct got_repository *repo)
1350 {
1351 	const struct got_error *err = NULL;
1352 	char *path;
1353 	DIR *dir = NULL;
1354 	struct dirent *dent;
1355 	struct got_object_id id;
1356 
1357 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1358 		err = got_error_from_errno("asprintf");
1359 		goto done;
1360 	}
1361 
1362 	dir = opendir(path);
1363 	if (dir == NULL) {
1364 		if (errno == ENOENT) {
1365 			err = NULL;
1366 			goto done;
1367 		}
1368 		err = got_error_from_errno2("opendir", path);
1369 		goto done;
1370 	}
1371 	while ((dent = readdir(dir)) != NULL) {
1372 		char *id_str;
1373 		int cmp;
1374 
1375 		if (strcmp(dent->d_name, ".") == 0 ||
1376 		    strcmp(dent->d_name, "..") == 0)
1377 			continue;
1378 
1379 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1380 			err = got_error_from_errno("asprintf");
1381 			goto done;
1382 		}
1383 
1384 		if (!got_parse_sha1_digest(id.sha1, id_str))
1385 			continue;
1386 
1387 		/*
1388 		 * Directory entries do not necessarily appear in
1389 		 * sorted order, so we must iterate over all of them.
1390 		 */
1391 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1392 		if (cmp != 0) {
1393 			free(id_str);
1394 			continue;
1395 		}
1396 
1397 		if (*unique_id == NULL) {
1398 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1399 				int matched_type;
1400 				err = got_object_get_type(&matched_type, repo,
1401 				    &id);
1402 				if (err)
1403 					goto done;
1404 				if (matched_type != obj_type)
1405 					continue;
1406 			}
1407 			*unique_id = got_object_id_dup(&id);
1408 			if (*unique_id == NULL) {
1409 				err = got_error_from_errno("got_object_id_dup");
1410 				free(id_str);
1411 				goto done;
1412 			}
1413 		} else {
1414 			if (got_object_id_cmp(*unique_id, &id) == 0)
1415 				continue; /* both packed and loose */
1416 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1417 			free(id_str);
1418 			goto done;
1419 		}
1420 	}
1421 done:
1422 	if (dir && closedir(dir) != 0 && err == NULL)
1423 		err = got_error_from_errno("closedir");
1424 	if (err) {
1425 		free(*unique_id);
1426 		*unique_id = NULL;
1427 	}
1428 	free(path);
1429 	return err;
1430 }
1431 
1432 const struct got_error *
1433 got_repo_match_object_id_prefix(struct got_object_id **id,
1434     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1435 {
1436 	const struct got_error *err = NULL;
1437 	char *path_objects = got_repo_get_path_objects(repo);
1438 	char *object_dir = NULL;
1439 	size_t len;
1440 	int i;
1441 
1442 	*id = NULL;
1443 
1444 	for (i = 0; i < strlen(id_str_prefix); i++) {
1445 		if (isxdigit((unsigned char)id_str_prefix[i]))
1446 			continue;
1447 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1448 	}
1449 
1450 	len = strlen(id_str_prefix);
1451 	if (len >= 2) {
1452 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1453 		if (err)
1454 			goto done;
1455 		object_dir = strndup(id_str_prefix, 2);
1456 		if (object_dir == NULL) {
1457 			err = got_error_from_errno("strdup");
1458 			goto done;
1459 		}
1460 		err = match_loose_object(id, path_objects, object_dir,
1461 		    id_str_prefix, obj_type, repo);
1462 	} else if (len == 1) {
1463 		int i;
1464 		for (i = 0; i < 0xf; i++) {
1465 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1466 			    == -1) {
1467 				err = got_error_from_errno("asprintf");
1468 				goto done;
1469 			}
1470 			err = match_packed_object(id, repo, object_dir,
1471 			    obj_type);
1472 			if (err)
1473 				goto done;
1474 			err = match_loose_object(id, path_objects, object_dir,
1475 			    id_str_prefix, obj_type, repo);
1476 			if (err)
1477 				goto done;
1478 		}
1479 	} else {
1480 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1481 		goto done;
1482 	}
1483 done:
1484 	free(object_dir);
1485 	if (err) {
1486 		free(*id);
1487 		*id = NULL;
1488 	} else if (*id == NULL)
1489 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1490 
1491 	return err;
1492 }
1493 
1494 const struct got_error *
1495 got_repo_match_object_id(struct got_object_id **id, char **label,
1496     const char *id_str, int obj_type, struct got_reflist_head *refs,
1497     struct got_repository *repo)
1498 {
1499 	const struct got_error *err;
1500 	struct got_tag_object *tag;
1501 	struct got_reference *ref = NULL;
1502 
1503 	*id = NULL;
1504 	if (label)
1505 		*label = NULL;
1506 
1507 	if (refs) {
1508 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1509 		    refs, repo);
1510 		if (err == NULL) {
1511 			*id = got_object_id_dup(
1512 			    got_object_tag_get_object_id(tag));
1513 			if (*id == NULL)
1514 				err = got_error_from_errno("got_object_id_dup");
1515 			else if (label && asprintf(label, "refs/tags/%s",
1516 			    got_object_tag_get_name(tag)) == -1) {
1517 				err = got_error_from_errno("asprintf");
1518 				free(*id);
1519 				*id = NULL;
1520 			}
1521 			got_object_tag_close(tag);
1522 			return err;
1523 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1524 		    err->code != GOT_ERR_NO_OBJ)
1525 			return err;
1526 	}
1527 
1528 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1529 	if (err) {
1530 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1531 			return err;
1532 		err = got_ref_open(&ref, repo, id_str, 0);
1533 		if (err != NULL)
1534 			goto done;
1535 		if (label) {
1536 			*label = strdup(got_ref_get_name(ref));
1537 			if (*label == NULL) {
1538 				err = got_error_from_errno("strdup");
1539 				goto done;
1540 			}
1541 		}
1542 		err = got_ref_resolve(id, repo, ref);
1543 	} else if (label) {
1544 		err = got_object_id_str(label, *id);
1545 		if (*label == NULL) {
1546 			err = got_error_from_errno("strdup");
1547 			goto done;
1548 		}
1549 	}
1550 done:
1551 	if (ref)
1552 		got_ref_close(ref);
1553 	return err;
1554 }
1555 
1556 const struct got_error *
1557 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1558     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1559 {
1560 	const struct got_error *err = NULL;
1561 	struct got_reflist_entry *re;
1562 	struct got_object_id *tag_id;
1563 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1564 
1565 	*tag = NULL;
1566 
1567 	TAILQ_FOREACH(re, refs, entry) {
1568 		const char *refname;
1569 		refname = got_ref_get_name(re->ref);
1570 		if (got_ref_is_symbolic(re->ref))
1571 			continue;
1572 		if (strncmp(refname, "refs/tags/", 10) != 0)
1573 			continue;
1574 		if (!name_is_absolute)
1575 			refname += strlen("refs/tags/");
1576 		if (strcmp(refname, name) != 0)
1577 			continue;
1578 		err = got_ref_resolve(&tag_id, repo, re->ref);
1579 		if (err)
1580 			break;
1581 		err = got_object_open_as_tag(tag, repo, tag_id);
1582 		free(tag_id);
1583 		if (err)
1584 			break;
1585 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1586 		    got_object_tag_get_object_type(*tag) == obj_type)
1587 			break;
1588 		got_object_tag_close(*tag);
1589 		*tag = NULL;
1590 	}
1591 
1592 	if (err == NULL && *tag == NULL)
1593 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1594 	return err;
1595 }
1596 
1597 static const struct got_error *
1598 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1599     const char *name, mode_t mode, struct got_object_id *blob_id)
1600 {
1601 	const struct got_error *err = NULL;
1602 
1603 	 *new_te = NULL;
1604 
1605 	*new_te = calloc(1, sizeof(**new_te));
1606 	if (*new_te == NULL)
1607 		return got_error_from_errno("calloc");
1608 
1609 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1610 	    sizeof((*new_te)->name)) {
1611 		err = got_error(GOT_ERR_NO_SPACE);
1612 		goto done;
1613 	}
1614 
1615 	if (S_ISLNK(mode)) {
1616 		(*new_te)->mode = S_IFLNK;
1617 	} else {
1618 		(*new_te)->mode = S_IFREG;
1619 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1620 	}
1621 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1622 done:
1623 	if (err && *new_te) {
1624 		free(*new_te);
1625 		*new_te = NULL;
1626 	}
1627 	return err;
1628 }
1629 
1630 static const struct got_error *
1631 import_file(struct got_tree_entry **new_te, struct dirent *de,
1632     const char *path, struct got_repository *repo)
1633 {
1634 	const struct got_error *err;
1635 	struct got_object_id *blob_id = NULL;
1636 	char *filepath;
1637 	struct stat sb;
1638 
1639 	if (asprintf(&filepath, "%s%s%s", path,
1640 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1641 		return got_error_from_errno("asprintf");
1642 
1643 	if (lstat(filepath, &sb) != 0) {
1644 		err = got_error_from_errno2("lstat", path);
1645 		goto done;
1646 	}
1647 
1648 	err = got_object_blob_create(&blob_id, filepath, repo);
1649 	if (err)
1650 		goto done;
1651 
1652 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1653 	    blob_id);
1654 done:
1655 	free(filepath);
1656 	if (err)
1657 		free(blob_id);
1658 	return err;
1659 }
1660 
1661 static const struct got_error *
1662 insert_tree_entry(struct got_tree_entry *new_te,
1663     struct got_pathlist_head *paths)
1664 {
1665 	const struct got_error *err = NULL;
1666 	struct got_pathlist_entry *new_pe;
1667 
1668 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1669 	if (err)
1670 		return err;
1671 	if (new_pe == NULL)
1672 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1673 	return NULL;
1674 }
1675 
1676 static const struct got_error *write_tree(struct got_object_id **,
1677     const char *, struct got_pathlist_head *, struct got_repository *,
1678     got_repo_import_cb progress_cb, void *progress_arg);
1679 
1680 static const struct got_error *
1681 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1682     const char *path, struct got_pathlist_head *ignores,
1683     struct got_repository *repo,
1684     got_repo_import_cb progress_cb, void *progress_arg)
1685 {
1686 	const struct got_error *err;
1687 	struct got_object_id *id = NULL;
1688 	char *subdirpath;
1689 
1690 	if (asprintf(&subdirpath, "%s%s%s", path,
1691 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1692 		return got_error_from_errno("asprintf");
1693 
1694 	(*new_te) = calloc(1, sizeof(**new_te));
1695 	if (*new_te == NULL)
1696 		return got_error_from_errno("calloc");
1697 	(*new_te)->mode = S_IFDIR;
1698 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1699 	    sizeof((*new_te)->name)) {
1700 		err = got_error(GOT_ERR_NO_SPACE);
1701 		goto done;
1702 	}
1703 	err = write_tree(&id, subdirpath, ignores,  repo,
1704 	    progress_cb, progress_arg);
1705 	if (err)
1706 		goto done;
1707 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1708 	
1709 done:
1710 	free(id);
1711 	free(subdirpath);
1712 	if (err) {
1713 		free(*new_te);
1714 		*new_te = NULL;
1715 	}
1716 	return err;
1717 }
1718 
1719 static const struct got_error *
1720 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1721     struct got_pathlist_head *ignores, struct got_repository *repo,
1722     got_repo_import_cb progress_cb, void *progress_arg)
1723 {
1724 	const struct got_error *err = NULL;
1725 	DIR *dir;
1726 	struct dirent *de;
1727 	int nentries;
1728 	struct got_tree_entry *new_te = NULL;
1729 	struct got_pathlist_head paths;
1730 	struct got_pathlist_entry *pe;
1731 
1732 	*new_tree_id = NULL;
1733 
1734 	TAILQ_INIT(&paths);
1735 
1736 	dir = opendir(path_dir);
1737 	if (dir == NULL) {
1738 		err = got_error_from_errno2("opendir", path_dir);
1739 		goto done;
1740 	}
1741 
1742 	nentries = 0;
1743 	while ((de = readdir(dir)) != NULL) {
1744 		int ignore = 0;
1745 		int type;
1746 
1747 		if (strcmp(de->d_name, ".") == 0 ||
1748 		    strcmp(de->d_name, "..") == 0)
1749 			continue;
1750 
1751 		TAILQ_FOREACH(pe, ignores, entry) {
1752 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1753 				ignore = 1;
1754 				break;
1755 			}
1756 		}
1757 		if (ignore)
1758 			continue;
1759 
1760 		err = got_path_dirent_type(&type, path_dir, de);
1761 		if (err)
1762 			goto done;
1763 
1764 		if (type == DT_DIR) {
1765 			err = import_subdir(&new_te, de, path_dir,
1766 			    ignores, repo, progress_cb, progress_arg);
1767 			if (err) {
1768 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1769 					goto done;
1770 				err = NULL;
1771 				continue;
1772 			}
1773 		} else if (type == DT_REG || type == DT_LNK) {
1774 			err = import_file(&new_te, de, path_dir, repo);
1775 			if (err)
1776 				goto done;
1777 		} else
1778 			continue;
1779 
1780 		err = insert_tree_entry(new_te, &paths);
1781 		if (err)
1782 			goto done;
1783 		nentries++;
1784 	}
1785 
1786 	if (TAILQ_EMPTY(&paths)) {
1787 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1788 		    "cannot create tree without any entries");
1789 		goto done;
1790 	}
1791 
1792 	TAILQ_FOREACH(pe, &paths, entry) {
1793 		struct got_tree_entry *te = pe->data;
1794 		char *path;
1795 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1796 			continue;
1797 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1798 			err = got_error_from_errno("asprintf");
1799 			goto done;
1800 		}
1801 		err = (*progress_cb)(progress_arg, path);
1802 		free(path);
1803 		if (err)
1804 			goto done;
1805 	}
1806 
1807 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1808 done:
1809 	if (dir)
1810 		closedir(dir);
1811 	got_pathlist_free(&paths);
1812 	return err;
1813 }
1814 
1815 const struct got_error *
1816 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1817     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1818     struct got_repository *repo, got_repo_import_cb progress_cb,
1819     void *progress_arg)
1820 {
1821 	const struct got_error *err;
1822 	struct got_object_id *new_tree_id;
1823 
1824 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1825 	    progress_cb, progress_arg);
1826 	if (err)
1827 		return err;
1828 
1829 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1830 	    author, time(NULL), author, time(NULL), logmsg, repo);
1831 	free(new_tree_id);
1832 	return err;
1833 }
1834 
1835 const struct got_error *
1836 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
1837     struct got_repository *repo)
1838 {
1839 	const struct got_error *err = NULL;
1840 	char *path_objects = NULL, *path = NULL;
1841 	DIR *dir = NULL;
1842 	struct got_object_id id;
1843 	int i;
1844 
1845 	*nobjects = 0;
1846 	*ondisk_size = 0;
1847 
1848 	path_objects = got_repo_get_path_objects(repo);
1849 	if (path_objects == NULL)
1850 		return got_error_from_errno("got_repo_get_path_objects");
1851 
1852 	for (i = 0; i <= 0xff; i++) {
1853 		struct dirent *dent;
1854 
1855 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
1856 			err = got_error_from_errno("asprintf");
1857 			break;
1858 		}
1859 
1860 		dir = opendir(path);
1861 		if (dir == NULL) {
1862 			if (errno == ENOENT) {
1863 				err = NULL;
1864 				continue;
1865 			}
1866 			err = got_error_from_errno2("opendir", path);
1867 			break;
1868 		}
1869 
1870 		while ((dent = readdir(dir)) != NULL) {
1871 			char *id_str;
1872 			int fd;
1873 			struct stat sb;
1874 
1875 			if (strcmp(dent->d_name, ".") == 0 ||
1876 			    strcmp(dent->d_name, "..") == 0)
1877 				continue;
1878 
1879 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
1880 				err = got_error_from_errno("asprintf");
1881 				goto done;
1882 			}
1883 
1884 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
1885 				free(id_str);
1886 				continue;
1887 			}
1888 			free(id_str);
1889 
1890 			err = got_object_open_loose_fd(&fd, &id, repo);
1891 			if (err)
1892 				goto done;
1893 
1894 			if (fstat(fd, &sb) == -1) {
1895 				err = got_error_from_errno("fstat");
1896 				close(fd);
1897 				goto done;
1898 			}
1899 			(*nobjects)++;
1900 			(*ondisk_size) += sb.st_size;
1901 
1902 			if (close(fd) == -1) {
1903 				err = got_error_from_errno("close");
1904 				goto done;
1905 			}
1906 		}
1907 
1908 		if (closedir(dir) != 0) {
1909 			err = got_error_from_errno("closedir");
1910 			goto done;
1911 		}
1912 		dir = NULL;
1913 
1914 		free(path);
1915 		path = NULL;
1916 	}
1917 done:
1918 	if (dir && closedir(dir) != 0 && err == NULL)
1919 		err = got_error_from_errno("closedir");
1920 
1921 	if (err) {
1922 		*nobjects = 0;
1923 		*ondisk_size = 0;
1924 	}
1925 	free(path_objects);
1926 	free(path);
1927 	return err;
1928 }
1929 
1930 const struct got_error *
1931 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
1932     off_t *total_packsize, struct got_repository *repo)
1933 {
1934 	const struct got_error *err = NULL;
1935 	DIR *packdir = NULL;
1936 	struct dirent *dent;
1937 	struct got_packidx *packidx = NULL;
1938 	char *path_packidx;
1939 	char *path_packfile;
1940 	int packdir_fd;
1941 	struct stat sb;
1942 
1943 	*npackfiles = 0;
1944 	*nobjects = 0;
1945 	*total_packsize = 0;
1946 
1947 	packdir_fd = openat(got_repo_get_fd(repo),
1948 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1949 	if (packdir_fd == -1) {
1950 		return got_error_from_errno_fmt("openat: %s/%s",
1951 		    got_repo_get_path_git_dir(repo),
1952 		    GOT_OBJECTS_PACK_DIR);
1953 	}
1954 
1955 	packdir = fdopendir(packdir_fd);
1956 	if (packdir == NULL) {
1957 		err = got_error_from_errno("fdopendir");
1958 		goto done;
1959 	}
1960 
1961 	while ((dent = readdir(packdir)) != NULL) {
1962 		if (!got_repo_is_packidx_filename(dent->d_name,
1963 		    strlen(dent->d_name)))
1964 			continue;
1965 
1966 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1967 		    dent->d_name) == -1) {
1968 			err = got_error_from_errno("asprintf");
1969 			goto done;
1970 		}
1971 
1972 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1973 		    path_packidx, 0);
1974 		free(path_packidx);
1975 		if (err)
1976 			goto done;
1977 
1978 		if (fstat(packidx->fd, &sb) == -1)
1979 			goto done;
1980 		*total_packsize += sb.st_size;
1981 
1982 		err = got_packidx_get_packfile_path(&path_packfile,
1983 		    packidx->path_packidx);
1984 		if (err)
1985 			goto done;
1986 
1987 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
1988 		    0) == -1) {
1989 			free(path_packfile);
1990 			goto done;
1991 		}
1992 		free(path_packfile);
1993 		*total_packsize += sb.st_size;
1994 
1995 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
1996 
1997 		(*npackfiles)++;
1998 
1999 		got_packidx_close(packidx);
2000 		packidx = NULL;
2001 	}
2002 done:
2003 	if (packidx)
2004 		got_packidx_close(packidx);
2005 	if (packdir && closedir(packdir) != 0 && err == NULL)
2006 		err = got_error_from_errno("closedir");
2007 	if (err) {
2008 		*npackfiles = 0;
2009 		*nobjects = 0;
2010 		*total_packsize = 0;
2011 	}
2012 	return err;
2013 }