1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
72     got_packidx_bloom_filter_cmp);
73 
74 const char *
75 got_repo_get_path(struct got_repository *repo)
76 {
77 	return repo->path;
78 }
79 
80 const char *
81 got_repo_get_path_git_dir(struct got_repository *repo)
82 {
83 	return repo->path_git_dir;
84 }
85 
86 int
87 got_repo_get_fd(struct got_repository *repo)
88 {
89 	return repo->gitdir_fd;
90 }
91 
92 const char *
93 got_repo_get_gitconfig_author_name(struct got_repository *repo)
94 {
95 	return repo->gitconfig_author_name;
96 }
97 
98 const char *
99 got_repo_get_gitconfig_author_email(struct got_repository *repo)
100 {
101 	return repo->gitconfig_author_email;
102 }
103 
104 const char *
105 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
106 {
107 	return repo->global_gitconfig_author_name;
108 }
109 
110 const char *
111 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
112 {
113 	return repo->global_gitconfig_author_email;
114 }
115 
116 const char *
117 got_repo_get_gitconfig_owner(struct got_repository *repo)
118 {
119 	return repo->gitconfig_owner;
120 }
121 
122 void
123 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
124     struct got_repository *repo)
125 {
126 	*extensions = repo->extensions;
127 	*nextensions = repo->nextensions;
128 }
129 
130 int
131 got_repo_is_bare(struct got_repository *repo)
132 {
133 	return (strcmp(repo->path, repo->path_git_dir) == 0);
134 }
135 
136 static char *
137 get_path_git_child(struct got_repository *repo, const char *basename)
138 {
139 	char *path_child;
140 
141 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
142 	    basename) == -1)
143 		return NULL;
144 
145 	return path_child;
146 }
147 
148 char *
149 got_repo_get_path_objects(struct got_repository *repo)
150 {
151 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
152 }
153 
154 char *
155 got_repo_get_path_objects_pack(struct got_repository *repo)
156 {
157 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
158 }
159 
160 char *
161 got_repo_get_path_refs(struct got_repository *repo)
162 {
163 	return get_path_git_child(repo, GOT_REFS_DIR);
164 }
165 
166 char *
167 got_repo_get_path_packed_refs(struct got_repository *repo)
168 {
169 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
170 }
171 
172 static char *
173 get_path_head(struct got_repository *repo)
174 {
175 	return get_path_git_child(repo, GOT_HEAD_FILE);
176 }
177 
178 char *
179 got_repo_get_path_gitconfig(struct got_repository *repo)
180 {
181 	return get_path_git_child(repo, GOT_GITCONFIG);
182 }
183 
184 char *
185 got_repo_get_path_gotconfig(struct got_repository *repo)
186 {
187 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
188 }
189 
190 const struct got_gotconfig *
191 got_repo_get_gotconfig(struct got_repository *repo)
192 {
193 	return repo->gotconfig;
194 }
195 
196 void
197 got_repo_get_gitconfig_remotes(int *nremotes,
198     const struct got_remote_repo **remotes, struct got_repository *repo)
199 {
200 	*nremotes = repo->ngitconfig_remotes;
201 	*remotes = repo->gitconfig_remotes;
202 }
203 
204 static int
205 is_git_repo(struct got_repository *repo)
206 {
207 	const char *path_git = got_repo_get_path_git_dir(repo);
208 	char *path_objects = got_repo_get_path_objects(repo);
209 	char *path_refs = got_repo_get_path_refs(repo);
210 	char *path_head = get_path_head(repo);
211 	int ret = 0;
212 	struct stat sb;
213 	struct got_reference *head_ref;
214 
215 	if (lstat(path_git, &sb) == -1)
216 		goto done;
217 	if (!S_ISDIR(sb.st_mode))
218 		goto done;
219 
220 	if (lstat(path_objects, &sb) == -1)
221 		goto done;
222 	if (!S_ISDIR(sb.st_mode))
223 		goto done;
224 
225 	if (lstat(path_refs, &sb) == -1)
226 		goto done;
227 	if (!S_ISDIR(sb.st_mode))
228 		goto done;
229 
230 	if (lstat(path_head, &sb) == -1)
231 		goto done;
232 	if (!S_ISREG(sb.st_mode))
233 		goto done;
234 
235 	/* Check if the HEAD reference can be opened. */
236 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
237 		goto done;
238 	got_ref_close(head_ref);
239 
240 	ret = 1;
241 done:
242 	free(path_objects);
243 	free(path_refs);
244 	free(path_head);
245 	return ret;
246 
247 }
248 
249 const struct got_error *
250 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
251     struct got_object *obj)
252 {
253 #ifndef GOT_NO_OBJ_CACHE
254 	const struct got_error *err = NULL;
255 	err = got_object_cache_add(&repo->objcache, id, obj);
256 	if (err) {
257 		if (err->code == GOT_ERR_OBJ_EXISTS ||
258 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
259 			err = NULL;
260 		return err;
261 	}
262 	obj->refcnt++;
263 #endif
264 	return NULL;
265 }
266 
267 struct got_object *
268 got_repo_get_cached_object(struct got_repository *repo,
269     struct got_object_id *id)
270 {
271 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
272 }
273 
274 const struct got_error *
275 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
276     struct got_tree_object *tree)
277 {
278 #ifndef GOT_NO_OBJ_CACHE
279 	const struct got_error *err = NULL;
280 	err = got_object_cache_add(&repo->treecache, id, tree);
281 	if (err) {
282 		if (err->code == GOT_ERR_OBJ_EXISTS ||
283 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
284 			err = NULL;
285 		return err;
286 	}
287 	tree->refcnt++;
288 #endif
289 	return NULL;
290 }
291 
292 struct got_tree_object *
293 got_repo_get_cached_tree(struct got_repository *repo,
294     struct got_object_id *id)
295 {
296 	return (struct got_tree_object *)got_object_cache_get(
297 	    &repo->treecache, id);
298 }
299 
300 const struct got_error *
301 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
302     struct got_commit_object *commit)
303 {
304 #ifndef GOT_NO_OBJ_CACHE
305 	const struct got_error *err = NULL;
306 	err = got_object_cache_add(&repo->commitcache, id, commit);
307 	if (err) {
308 		if (err->code == GOT_ERR_OBJ_EXISTS ||
309 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
310 			err = NULL;
311 		return err;
312 	}
313 	commit->refcnt++;
314 #endif
315 	return NULL;
316 }
317 
318 struct got_commit_object *
319 got_repo_get_cached_commit(struct got_repository *repo,
320     struct got_object_id *id)
321 {
322 	return (struct got_commit_object *)got_object_cache_get(
323 	    &repo->commitcache, id);
324 }
325 
326 const struct got_error *
327 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
328     struct got_tag_object *tag)
329 {
330 #ifndef GOT_NO_OBJ_CACHE
331 	const struct got_error *err = NULL;
332 	err = got_object_cache_add(&repo->tagcache, id, tag);
333 	if (err) {
334 		if (err->code == GOT_ERR_OBJ_EXISTS ||
335 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
336 			err = NULL;
337 		return err;
338 	}
339 	tag->refcnt++;
340 #endif
341 	return NULL;
342 }
343 
344 struct got_tag_object *
345 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
346 {
347 	return (struct got_tag_object *)got_object_cache_get(
348 	    &repo->tagcache, id);
349 }
350 
351 const struct got_error *
352 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
353     struct got_raw_object *raw)
354 {
355 #ifndef GOT_NO_OBJ_CACHE
356 	const struct got_error *err = NULL;
357 	err = got_object_cache_add(&repo->rawcache, id, raw);
358 	if (err) {
359 		if (err->code == GOT_ERR_OBJ_EXISTS ||
360 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
361 			err = NULL;
362 		return err;
363 	}
364 	raw->refcnt++;
365 #endif
366 	return NULL;
367 }
368 
369 
370 struct got_raw_object *
371 got_repo_get_cached_raw_object(struct got_repository *repo,
372     struct got_object_id *id)
373 {
374 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
375 }
376 
377 
378 static const struct got_error *
379 open_repo(struct got_repository *repo, const char *path)
380 {
381 	const struct got_error *err = NULL;
382 
383 	repo->gitdir_fd = -1;
384 
385 	/* bare git repository? */
386 	repo->path_git_dir = strdup(path);
387 	if (repo->path_git_dir == NULL)
388 		return got_error_from_errno("strdup");
389 	if (is_git_repo(repo)) {
390 		repo->path = strdup(repo->path_git_dir);
391 		if (repo->path == NULL) {
392 			err = got_error_from_errno("strdup");
393 			goto done;
394 		}
395 		repo->gitdir_fd = open(repo->path_git_dir,
396 		    O_DIRECTORY | O_CLOEXEC);
397 		if (repo->gitdir_fd == -1) {
398 			err = got_error_from_errno2("open",
399 			    repo->path_git_dir);
400 			goto done;
401 		}
402 		return NULL;
403 	}
404 
405 	/* git repository with working tree? */
406 	free(repo->path_git_dir);
407 	repo->path_git_dir = NULL;
408 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
409 		err = got_error_from_errno("asprintf");
410 		goto done;
411 	}
412 	if (is_git_repo(repo)) {
413 		repo->path = strdup(path);
414 		if (repo->path == NULL) {
415 			err = got_error_from_errno("strdup");
416 			goto done;
417 		}
418 		repo->gitdir_fd = open(repo->path_git_dir,
419 		    O_DIRECTORY | O_CLOEXEC);
420 		if (repo->gitdir_fd == -1) {
421 			err = got_error_from_errno2("open",
422 			    repo->path_git_dir);
423 			goto done;
424 		}
425 		return NULL;
426 	}
427 
428 	err = got_error(GOT_ERR_NOT_GIT_REPO);
429 done:
430 	if (err) {
431 		free(repo->path);
432 		repo->path = NULL;
433 		free(repo->path_git_dir);
434 		repo->path_git_dir = NULL;
435 		if (repo->gitdir_fd != -1)
436 			close(repo->gitdir_fd);
437 		repo->gitdir_fd = -1;
438 
439 	}
440 	return err;
441 }
442 
443 static const struct got_error *
444 parse_gitconfig_file(int *gitconfig_repository_format_version,
445     char **gitconfig_author_name, char **gitconfig_author_email,
446     struct got_remote_repo **remotes, int *nremotes,
447     char **gitconfig_owner, char ***extensions, int *nextensions,
448     const char *gitconfig_path)
449 {
450 	const struct got_error *err = NULL, *child_err = NULL;
451 	int fd = -1;
452 	int imsg_fds[2] = { -1, -1 };
453 	pid_t pid;
454 	struct imsgbuf *ibuf;
455 
456 	*gitconfig_repository_format_version = 0;
457 	if (extensions)
458 		*extensions = NULL;
459 	if (nextensions)
460 		*nextensions = 0;
461 	*gitconfig_author_name = NULL;
462 	*gitconfig_author_email = NULL;
463 	if (remotes)
464 		*remotes = NULL;
465 	if (nremotes)
466 		*nremotes = 0;
467 	if (gitconfig_owner)
468 		*gitconfig_owner = NULL;
469 
470 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
471 	if (fd == -1) {
472 		if (errno == ENOENT)
473 			return NULL;
474 		return got_error_from_errno2("open", gitconfig_path);
475 	}
476 
477 	ibuf = calloc(1, sizeof(*ibuf));
478 	if (ibuf == NULL) {
479 		err = got_error_from_errno("calloc");
480 		goto done;
481 	}
482 
483 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
484 		err = got_error_from_errno("socketpair");
485 		goto done;
486 	}
487 
488 	pid = fork();
489 	if (pid == -1) {
490 		err = got_error_from_errno("fork");
491 		goto done;
492 	} else if (pid == 0) {
493 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
494 		    gitconfig_path);
495 		/* not reached */
496 	}
497 
498 	if (close(imsg_fds[1]) == -1) {
499 		err = got_error_from_errno("close");
500 		goto done;
501 	}
502 	imsg_fds[1] = -1;
503 	imsg_init(ibuf, imsg_fds[0]);
504 
505 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
506 	if (err)
507 		goto done;
508 	fd = -1;
509 
510 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
511 	if (err)
512 		goto done;
513 
514 	err = got_privsep_recv_gitconfig_int(
515 	    gitconfig_repository_format_version, ibuf);
516 	if (err)
517 		goto done;
518 
519 	if (extensions && nextensions) {
520 		err = got_privsep_send_gitconfig_repository_extensions_req(
521 		    ibuf);
522 		if (err)
523 			goto done;
524 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
525 		if (err)
526 			goto done;
527 		if (*nextensions > 0) {
528 			int i;
529 			*extensions = calloc(*nextensions, sizeof(char *));
530 			if (*extensions == NULL) {
531 				err = got_error_from_errno("calloc");
532 				goto done;
533 			}
534 			for (i = 0; i < *nextensions; i++) {
535 				char *ext;
536 				err = got_privsep_recv_gitconfig_str(&ext,
537 				    ibuf);
538 				if (err)
539 					goto done;
540 				(*extensions)[i] = ext;
541 			}
542 		}
543 	}
544 
545 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
546 	if (err)
547 		goto done;
548 
549 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
550 	if (err)
551 		goto done;
552 
553 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
554 	if (err)
555 		goto done;
556 
557 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
558 	if (err)
559 		goto done;
560 
561 	if (remotes && nremotes) {
562 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
563 		if (err)
564 			goto done;
565 
566 		err = got_privsep_recv_gitconfig_remotes(remotes,
567 		    nremotes, ibuf);
568 		if (err)
569 			goto done;
570 	}
571 
572 	if (gitconfig_owner) {
573 		err = got_privsep_send_gitconfig_owner_req(ibuf);
574 		if (err)
575 			goto done;
576 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
577 		if (err)
578 			goto done;
579 	}
580 
581 	err = got_privsep_send_stop(imsg_fds[0]);
582 	child_err = got_privsep_wait_for_child(pid);
583 	if (child_err && err == NULL)
584 		err = child_err;
585 done:
586 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
587 		err = got_error_from_errno("close");
588 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
589 		err = got_error_from_errno("close");
590 	if (fd != -1 && close(fd) == -1 && err == NULL)
591 		err = got_error_from_errno2("close", gitconfig_path);
592 	free(ibuf);
593 	return err;
594 }
595 
596 static const struct got_error *
597 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
598 {
599 	const struct got_error *err = NULL;
600 	char *repo_gitconfig_path = NULL;
601 
602 	if (global_gitconfig_path) {
603 		/* Read settings from ~/.gitconfig. */
604 		int dummy_repo_version;
605 		err = parse_gitconfig_file(&dummy_repo_version,
606 		    &repo->global_gitconfig_author_name,
607 		    &repo->global_gitconfig_author_email,
608 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
609 		if (err)
610 			return err;
611 	}
612 
613 	/* Read repository's .git/config file. */
614 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
615 	if (repo_gitconfig_path == NULL)
616 		return got_error_from_errno("got_repo_get_path_gitconfig");
617 
618 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
619 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
620 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
621 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
622 	    repo_gitconfig_path);
623 	if (err)
624 		goto done;
625 done:
626 	free(repo_gitconfig_path);
627 	return err;
628 }
629 
630 static const struct got_error *
631 read_gotconfig(struct got_repository *repo)
632 {
633 	const struct got_error *err = NULL;
634 	char *gotconfig_path;
635 
636 	gotconfig_path = got_repo_get_path_gotconfig(repo);
637 	if (gotconfig_path == NULL)
638 		return got_error_from_errno("got_repo_get_path_gotconfig");
639 
640 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
641 	free(gotconfig_path);
642 	return err;
643 }
644 
645 /* Supported repository format extensions. */
646 static const char *const repo_extensions[] = {
647 	"noop",			/* Got supports repository format version 1. */
648 	"preciousObjects",	/* Supported by gotadmin cleanup. */
649 	"worktreeConfig",	/* Got does not care about Git work trees. */
650 };
651 
652 const struct got_error *
653 got_repo_open(struct got_repository **repop, const char *path,
654     const char *global_gitconfig_path)
655 {
656 	struct got_repository *repo = NULL;
657 	const struct got_error *err = NULL;
658 	char *repo_path = NULL;
659 	size_t i;
660 	struct rlimit rl;
661 
662 	*repop = NULL;
663 
664 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
665 		return got_error_from_errno("getrlimit");
666 
667 	repo = calloc(1, sizeof(*repo));
668 	if (repo == NULL)
669 		return got_error_from_errno("calloc");
670 
671 	RB_INIT(&repo->packidx_bloom_filters);
672 	TAILQ_INIT(&repo->packidx_paths);
673 
674 	for (i = 0; i < nitems(repo->privsep_children); i++) {
675 		memset(&repo->privsep_children[i], 0,
676 		    sizeof(repo->privsep_children[0]));
677 		repo->privsep_children[i].imsg_fd = -1;
678 	}
679 
680 	err = got_object_cache_init(&repo->objcache,
681 	    GOT_OBJECT_CACHE_TYPE_OBJ);
682 	if (err)
683 		goto done;
684 	err = got_object_cache_init(&repo->treecache,
685 	    GOT_OBJECT_CACHE_TYPE_TREE);
686 	if (err)
687 		goto done;
688 	err = got_object_cache_init(&repo->commitcache,
689 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
690 	if (err)
691 		goto done;
692 	err = got_object_cache_init(&repo->tagcache,
693 	    GOT_OBJECT_CACHE_TYPE_TAG);
694 	if (err)
695 		goto done;
696 	err = got_object_cache_init(&repo->rawcache,
697 	    GOT_OBJECT_CACHE_TYPE_RAW);
698 	if (err)
699 		goto done;
700 
701 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
702 	if (repo->pack_cache_size > rl.rlim_cur / 8)
703 		repo->pack_cache_size = rl.rlim_cur / 8;
704 	for (i = 0; i < nitems(repo->packs); i++) {
705 		if (i < repo->pack_cache_size) {
706 			repo->packs[i].basefd = got_opentempfd();
707 			if (repo->packs[i].basefd == -1)
708 				return got_error_from_errno("got_opentempfd");
709 			repo->packs[i].accumfd = got_opentempfd();
710 			if (repo->packs[i].accumfd == -1)
711 				return got_error_from_errno("got_opentempfd");
712 		} else {
713 			repo->packs[i].basefd = -1;
714 			repo->packs[i].accumfd = -1;
715 		}
716 	}
717 
718 	repo_path = realpath(path, NULL);
719 	if (repo_path == NULL) {
720 		err = got_error_from_errno2("realpath", path);
721 		goto done;
722 	}
723 
724 	for (;;) {
725 		char *parent_path;
726 
727 		err = open_repo(repo, repo_path);
728 		if (err == NULL)
729 			break;
730 		if (err->code != GOT_ERR_NOT_GIT_REPO)
731 			goto done;
732 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
733 			err = got_error(GOT_ERR_NOT_GIT_REPO);
734 			goto done;
735 		}
736 		err = got_path_dirname(&parent_path, repo_path);
737 		if (err)
738 			goto done;
739 		free(repo_path);
740 		repo_path = parent_path;
741 	}
742 
743 	err = read_gotconfig(repo);
744 	if (err)
745 		goto done;
746 
747 	err = read_gitconfig(repo, global_gitconfig_path);
748 	if (err)
749 		goto done;
750 	if (repo->gitconfig_repository_format_version != 0)
751 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
752 	for (i = 0; i < repo->nextensions; i++) {
753 		char *ext = repo->extensions[i];
754 		int j, supported = 0;
755 		for (j = 0; j < nitems(repo_extensions); j++) {
756 			if (strcmp(ext, repo_extensions[j]) == 0) {
757 				supported = 1;
758 				break;
759 			}
760 		}
761 		if (!supported) {
762 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
763 			goto done;
764 		}
765 	}
766 
767 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
768 done:
769 	if (err)
770 		got_repo_close(repo);
771 	else
772 		*repop = repo;
773 	free(repo_path);
774 	return err;
775 }
776 
777 const struct got_error *
778 got_repo_close(struct got_repository *repo)
779 {
780 	const struct got_error *err = NULL, *child_err;
781 	struct got_packidx_bloom_filter *bf;
782 	struct got_pathlist_entry *pe;
783 	size_t i;
784 
785 	for (i = 0; i < repo->pack_cache_size; i++) {
786 		if (repo->packidx_cache[i] == NULL)
787 			break;
788 		got_packidx_close(repo->packidx_cache[i]);
789 	}
790 
791 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
792 	    &repo->packidx_bloom_filters))) {
793 		RB_REMOVE(got_packidx_bloom_filter_tree,
794 		    &repo->packidx_bloom_filters, bf);
795 		free(bf->bloom);
796 		free(bf);
797 	}
798 
799 	for (i = 0; i < repo->pack_cache_size; i++) {
800 		if (repo->packs[i].path_packfile == NULL)
801 			break;
802 		got_pack_close(&repo->packs[i]);
803 		if (repo->packs[i].basefd != -1) {
804 			if (close(repo->packs[i].basefd) == -1 && err == NULL)
805 				err = got_error_from_errno("close");
806 			repo->packs[i].basefd = -1;
807 		}
808 		if (repo->packs[i].accumfd != -1) {
809 			if (close(repo->packs[i].accumfd) == -1 && err == NULL)
810 				err = got_error_from_errno("close");
811 			repo->packs[i].accumfd = -1;
812 		}
813 	}
814 
815 	free(repo->path);
816 	free(repo->path_git_dir);
817 
818 	got_object_cache_close(&repo->objcache);
819 	got_object_cache_close(&repo->treecache);
820 	got_object_cache_close(&repo->commitcache);
821 	got_object_cache_close(&repo->tagcache);
822 	got_object_cache_close(&repo->rawcache);
823 
824 	for (i = 0; i < nitems(repo->privsep_children); i++) {
825 		if (repo->privsep_children[i].imsg_fd == -1)
826 			continue;
827 		imsg_clear(repo->privsep_children[i].ibuf);
828 		free(repo->privsep_children[i].ibuf);
829 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
830 		child_err = got_privsep_wait_for_child(
831 		    repo->privsep_children[i].pid);
832 		if (child_err && err == NULL)
833 			err = child_err;
834 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
835 		    err == NULL)
836 			err = got_error_from_errno("close");
837 	}
838 
839 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
840 	    err == NULL)
841 		err = got_error_from_errno("close");
842 
843 	if (repo->gotconfig)
844 		got_gotconfig_free(repo->gotconfig);
845 	free(repo->gitconfig_author_name);
846 	free(repo->gitconfig_author_email);
847 	for (i = 0; i < repo->ngitconfig_remotes; i++)
848 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
849 	free(repo->gitconfig_remotes);
850 	for (i = 0; i < repo->nextensions; i++)
851 		free(repo->extensions[i]);
852 	free(repo->extensions);
853 
854 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
855 		free((void *)pe->path);
856 	got_pathlist_free(&repo->packidx_paths);
857 	free(repo);
858 
859 	return err;
860 }
861 
862 void
863 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
864 {
865 	int i;
866 
867 	free(repo->name);
868 	repo->name = NULL;
869 	free(repo->fetch_url);
870 	repo->fetch_url = NULL;
871 	free(repo->send_url);
872 	repo->send_url = NULL;
873 	for (i = 0; i < repo->nfetch_branches; i++)
874 		free(repo->fetch_branches[i]);
875 	free(repo->fetch_branches);
876 	repo->fetch_branches = NULL;
877 	repo->nfetch_branches = 0;
878 	for (i = 0; i < repo->nsend_branches; i++)
879 		free(repo->send_branches[i]);
880 	free(repo->send_branches);
881 	repo->send_branches = NULL;
882 	repo->nsend_branches = 0;
883 }
884 
885 const struct got_error *
886 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
887     const char *input_path)
888 {
889 	const struct got_error *err = NULL;
890 	const char *repo_abspath = NULL;
891 	size_t repolen, len;
892 	char *canonpath, *path = NULL;
893 
894 	*in_repo_path = NULL;
895 
896 	canonpath = strdup(input_path);
897 	if (canonpath == NULL) {
898 		err = got_error_from_errno("strdup");
899 		goto done;
900 	}
901 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
902 	if (err)
903 		goto done;
904 
905 	repo_abspath = got_repo_get_path(repo);
906 
907 	if (canonpath[0] == '\0') {
908 		path = strdup(canonpath);
909 		if (path == NULL) {
910 			err = got_error_from_errno("strdup");
911 			goto done;
912 		}
913 	} else {
914 		path = realpath(canonpath, NULL);
915 		if (path == NULL) {
916 			if (errno != ENOENT) {
917 				err = got_error_from_errno2("realpath",
918 				    canonpath);
919 				goto done;
920 			}
921 			/*
922 			 * Path is not on disk.
923 			 * Assume it is already relative to repository root.
924 			 */
925 			path = strdup(canonpath);
926 			if (path == NULL) {
927 				err = got_error_from_errno("strdup");
928 				goto done;
929 			}
930 		}
931 
932 		repolen = strlen(repo_abspath);
933 		len = strlen(path);
934 
935 
936 		if (strcmp(path, repo_abspath) == 0) {
937 			free(path);
938 			path = strdup("");
939 			if (path == NULL) {
940 				err = got_error_from_errno("strdup");
941 				goto done;
942 			}
943 		} else if (len > repolen &&
944 		    got_path_is_child(path, repo_abspath, repolen)) {
945 			/* Matched an on-disk path inside repository. */
946 			if (got_repo_is_bare(repo)) {
947 				/*
948 				 * Matched an on-disk path inside repository
949 				 * database. Treat input as repository-relative.
950 				 */
951 				free(path);
952 				path = canonpath;
953 				canonpath = NULL;
954 			} else {
955 				char *child;
956 				/* Strip common prefix with repository path. */
957 				err = got_path_skip_common_ancestor(&child,
958 				    repo_abspath, path);
959 				if (err)
960 					goto done;
961 				free(path);
962 				path = child;
963 			}
964 		} else {
965 			/*
966 			 * Matched unrelated on-disk path.
967 			 * Treat input as repository-relative.
968 			 */
969 			free(path);
970 			path = canonpath;
971 			canonpath = NULL;
972 		}
973 	}
974 
975 	/* Make in-repository path absolute */
976 	if (path[0] != '/') {
977 		char *abspath;
978 		if (asprintf(&abspath, "/%s", path) == -1) {
979 			err = got_error_from_errno("asprintf");
980 			goto done;
981 		}
982 		free(path);
983 		path = abspath;
984 	}
985 
986 done:
987 	free(canonpath);
988 	if (err)
989 		free(path);
990 	else
991 		*in_repo_path = path;
992 	return err;
993 }
994 
995 static const struct got_error *
996 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
997     const char *path_packidx)
998 {
999 	const struct got_error *err = NULL;
1000 	size_t i;
1001 
1002 	for (i = 0; i < repo->pack_cache_size; i++) {
1003 		if (repo->packidx_cache[i] == NULL)
1004 			break;
1005 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1006 		    path_packidx) == 0) {
1007 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1008 		}
1009 	}
1010 	if (i == repo->pack_cache_size) {
1011 		i = repo->pack_cache_size - 1;
1012 		err = got_packidx_close(repo->packidx_cache[i]);
1013 		if (err)
1014 			return err;
1015 	}
1016 
1017 	repo->packidx_cache[i] = packidx;
1018 
1019 	return NULL;
1020 }
1021 
1022 int
1023 got_repo_is_packidx_filename(const char *name, size_t len)
1024 {
1025 	if (len != GOT_PACKIDX_NAMELEN)
1026 		return 0;
1027 
1028 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1029 		return 0;
1030 
1031 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1032 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1033 		return 0;
1034 
1035 	return 1;
1036 }
1037 
1038 static struct got_packidx_bloom_filter *
1039 get_packidx_bloom_filter(struct got_repository *repo,
1040     const char *path, size_t path_len)
1041 {
1042 	struct got_packidx_bloom_filter key;
1043 
1044 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1045 		return NULL; /* XXX */
1046 	key.path_len = path_len;
1047 
1048 	return RB_FIND(got_packidx_bloom_filter_tree,
1049 	    &repo->packidx_bloom_filters, &key);
1050 }
1051 
1052 int
1053 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1054     const char *path_packidx, struct got_object_id *id)
1055 {
1056 	struct got_packidx_bloom_filter *bf;
1057 
1058 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1059 	if (bf)
1060 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1061 
1062 	/* No bloom filter means this pack index must be searched. */
1063 	return 1;
1064 }
1065 
1066 static const struct got_error *
1067 add_packidx_bloom_filter(struct got_repository *repo,
1068     struct got_packidx *packidx, const char *path_packidx)
1069 {
1070 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1071 	struct got_packidx_bloom_filter *bf;
1072 	size_t len;
1073 
1074 	/*
1075 	 * Don't use bloom filters for very large pack index files.
1076 	 * Large pack files will contain a relatively large fraction
1077 	 * of our objects so we will likely need to visit them anyway.
1078 	 * The more objects a pack file contains the higher the probability
1079 	 * of a false-positive match from the bloom filter. And reading
1080 	 * all object IDs from a large pack index file can be expensive.
1081 	 */
1082 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1083 		return NULL;
1084 
1085 	/* Do we already have a filter for this pack index? */
1086 	if (get_packidx_bloom_filter(repo, path_packidx,
1087 	    strlen(path_packidx)) != NULL)
1088 		return NULL;
1089 
1090 	bf = calloc(1, sizeof(*bf));
1091 	if (bf == NULL)
1092 		return got_error_from_errno("calloc");
1093 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1094 	if (bf->bloom == NULL) {
1095 		free(bf);
1096 		return got_error_from_errno("calloc");
1097 	}
1098 	
1099 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1100 	if (len >= sizeof(bf->path)) {
1101 		free(bf->bloom);
1102 		free(bf);
1103 		return got_error(GOT_ERR_NO_SPACE);
1104 	}
1105 	bf->path_len = len;
1106 
1107 	/* Minimum size supported by our bloom filter is 1000 entries. */
1108 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1109 	for (i = 0; i < nobjects; i++) {
1110 		struct got_packidx_object_id *id;
1111 		id = &packidx->hdr.sorted_ids[i];
1112 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1113 	}
1114 
1115 	RB_INSERT(got_packidx_bloom_filter_tree,
1116 	    &repo->packidx_bloom_filters, bf);
1117 	return NULL;
1118 }
1119 
1120 const struct got_error *
1121 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1122     struct got_repository *repo, struct got_object_id *id)
1123 {
1124 	const struct got_error *err;
1125 	struct got_pathlist_entry *pe;
1126 	size_t i;
1127 
1128 	/* Search pack index cache. */
1129 	for (i = 0; i < repo->pack_cache_size; i++) {
1130 		if (repo->packidx_cache[i] == NULL)
1131 			break;
1132 		if (!got_repo_check_packidx_bloom_filter(repo,
1133 		    repo->packidx_cache[i]->path_packidx, id))
1134 			continue; /* object will not be found in this index */
1135 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1136 		if (*idx != -1) {
1137 			*packidx = repo->packidx_cache[i];
1138 			/*
1139 			 * Move this cache entry to the front. Repeatedly
1140 			 * searching a wrong pack index can be expensive.
1141 			 */
1142 			if (i > 0) {
1143 				memmove(&repo->packidx_cache[1],
1144 				    &repo->packidx_cache[0],
1145 				    i * sizeof(repo->packidx_cache[0]));
1146 				repo->packidx_cache[0] = *packidx;
1147 			}
1148 			return NULL;
1149 		}
1150 	}
1151 	/* No luck. Search the filesystem. */
1152 
1153 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1154 		const char *path_packidx = pe->path;
1155 		int is_cached = 0;
1156 
1157 		if (!got_repo_check_packidx_bloom_filter(repo,
1158 		    pe->path, id))
1159 			continue; /* object will not be found in this index */
1160 
1161 		for (i = 0; i < repo->pack_cache_size; i++) {
1162 			if (repo->packidx_cache[i] == NULL)
1163 				break;
1164 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1165 			    path_packidx) == 0) {
1166 				is_cached = 1;
1167 				break;
1168 			}
1169 		}
1170 		if (is_cached)
1171 			continue; /* already searched */
1172 
1173 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1174 		    path_packidx, 0);
1175 		if (err)
1176 			goto done;
1177 
1178 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1179 		if (err)
1180 			goto done;
1181 
1182 		err = cache_packidx(repo, *packidx, path_packidx);
1183 		if (err)
1184 			goto done;
1185 
1186 		*idx = got_packidx_get_object_idx(*packidx, id);
1187 		if (*idx != -1) {
1188 			err = NULL; /* found the object */
1189 			goto done;
1190 		}
1191 	}
1192 
1193 	err = got_error_no_obj(id);
1194 done:
1195 	return err;
1196 }
1197 
1198 const struct got_error *
1199 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1200     struct got_repository *repo)
1201 {
1202 	const struct got_error *err = NULL;
1203 	DIR *packdir = NULL;
1204 	struct dirent *dent;
1205 	char *path_packidx = NULL;
1206 	int packdir_fd;
1207 
1208 	packdir_fd = openat(got_repo_get_fd(repo),
1209 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1210 	if (packdir_fd == -1) {
1211 		return got_error_from_errno_fmt("openat: %s/%s",
1212 		    got_repo_get_path_git_dir(repo),
1213 		    GOT_OBJECTS_PACK_DIR);
1214 	}
1215 
1216 	packdir = fdopendir(packdir_fd);
1217 	if (packdir == NULL) {
1218 		err = got_error_from_errno("fdopendir");
1219 		goto done;
1220 	}
1221 
1222 	while ((dent = readdir(packdir)) != NULL) {
1223 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1224 			continue;
1225 
1226 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1227 		    dent->d_name) == -1) {
1228 			err = got_error_from_errno("asprintf");
1229 			path_packidx = NULL;
1230 			break;
1231 		}
1232 
1233 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1234 		if (err)
1235 			break;
1236 	}
1237 done:
1238 	if (err)
1239 		free(path_packidx);
1240 	if (packdir && closedir(packdir) != 0 && err == NULL)
1241 		err = got_error_from_errno("closedir");
1242 	return err;
1243 }
1244 
1245 const struct got_error *
1246 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1247     struct got_repository *repo)
1248 {
1249 	const struct got_error *err;
1250 	size_t i;
1251 
1252 	*packidx = NULL;
1253 
1254 	/* Search pack index cache. */
1255 	for (i = 0; i < repo->pack_cache_size; i++) {
1256 		if (repo->packidx_cache[i] == NULL)
1257 			break;
1258 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1259 		    path_packidx) == 0) {
1260 			*packidx = repo->packidx_cache[i];
1261 			return NULL;
1262 		}
1263 	}
1264 	/* No luck. Search the filesystem. */
1265 
1266 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1267 	    path_packidx, 0);
1268 	if (err)
1269 		return err;
1270 
1271 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1272 	if (err)
1273 		goto done;
1274 
1275 	err = cache_packidx(repo, *packidx, path_packidx);
1276 done:
1277 	if (err) {
1278 		got_packidx_close(*packidx);
1279 		*packidx = NULL;
1280 	}
1281 	return err;
1282 }
1283 
1284 static const struct got_error *
1285 read_packfile_hdr(int fd, struct got_packidx *packidx)
1286 {
1287 	const struct got_error *err = NULL;
1288 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1289 	struct got_packfile_hdr hdr;
1290 	ssize_t n;
1291 
1292 	n = read(fd, &hdr, sizeof(hdr));
1293 	if (n < 0)
1294 		return got_error_from_errno("read");
1295 	if (n != sizeof(hdr))
1296 		return got_error(GOT_ERR_BAD_PACKFILE);
1297 
1298 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1299 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1300 	    be32toh(hdr.nobjects) != totobj)
1301 		err = got_error(GOT_ERR_BAD_PACKFILE);
1302 
1303 	return err;
1304 }
1305 
1306 static const struct got_error *
1307 open_packfile(int *fd, struct got_repository *repo,
1308     const char *relpath, struct got_packidx *packidx)
1309 {
1310 	const struct got_error *err = NULL;
1311 
1312 	*fd = openat(got_repo_get_fd(repo), relpath,
1313 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1314 	if (*fd == -1)
1315 		return got_error_from_errno_fmt("openat: %s/%s",
1316 		    got_repo_get_path_git_dir(repo), relpath);
1317 
1318 	if (packidx) {
1319 		err = read_packfile_hdr(*fd, packidx);
1320 		if (err) {
1321 			close(*fd);
1322 			*fd = -1;
1323 		}
1324 	}
1325 
1326 	return err;
1327 }
1328 
1329 const struct got_error *
1330 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1331     const char *path_packfile, struct got_packidx *packidx)
1332 {
1333 	const struct got_error *err = NULL;
1334 	struct got_pack *pack = NULL;
1335 	struct stat sb;
1336 	size_t i;
1337 
1338 	if (packp)
1339 		*packp = NULL;
1340 
1341 	for (i = 0; i < repo->pack_cache_size; i++) {
1342 		pack = &repo->packs[i];
1343 		if (pack->path_packfile == NULL)
1344 			break;
1345 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1346 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1347 	}
1348 
1349 	if (i == repo->pack_cache_size) {
1350 		err = got_pack_close(&repo->packs[i - 1]);
1351 		if (err)
1352 			return err;
1353 		if (ftruncate(repo->packs[i - 1].basefd, 0L) == -1)
1354 			return got_error_from_errno("ftruncate");
1355 		if (ftruncate(repo->packs[i - 1].accumfd, 0L) == -1)
1356 			return got_error_from_errno("ftruncate");
1357 		memmove(&repo->packs[1], &repo->packs[0],
1358 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1359 		i = 0;
1360 	}
1361 
1362 	pack = &repo->packs[i];
1363 
1364 	pack->path_packfile = strdup(path_packfile);
1365 	if (pack->path_packfile == NULL) {
1366 		err = got_error_from_errno("strdup");
1367 		goto done;
1368 	}
1369 
1370 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1371 	if (err)
1372 		goto done;
1373 
1374 	if (fstat(pack->fd, &sb) != 0) {
1375 		err = got_error_from_errno("fstat");
1376 		goto done;
1377 	}
1378 	pack->filesize = sb.st_size;
1379 
1380 	pack->privsep_child = NULL;
1381 
1382 #ifndef GOT_PACK_NO_MMAP
1383 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1384 	    pack->fd, 0);
1385 	if (pack->map == MAP_FAILED) {
1386 		if (errno != ENOMEM) {
1387 			err = got_error_from_errno("mmap");
1388 			goto done;
1389 		}
1390 		pack->map = NULL; /* fall back to read(2) */
1391 	}
1392 #endif
1393 done:
1394 	if (err) {
1395 		if (pack) {
1396 			free(pack->path_packfile);
1397 			memset(pack, 0, sizeof(*pack));
1398 		}
1399 	} else if (packp)
1400 		*packp = pack;
1401 	return err;
1402 }
1403 
1404 struct got_pack *
1405 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1406 {
1407 	struct got_pack *pack = NULL;
1408 	size_t i;
1409 
1410 	for (i = 0; i < repo->pack_cache_size; i++) {
1411 		pack = &repo->packs[i];
1412 		if (pack->path_packfile == NULL)
1413 			break;
1414 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1415 			return pack;
1416 	}
1417 
1418 	return NULL;
1419 }
1420 
1421 const struct got_error *
1422 got_repo_init(const char *repo_path)
1423 {
1424 	const struct got_error *err = NULL;
1425 	const char *dirnames[] = {
1426 		GOT_OBJECTS_DIR,
1427 		GOT_OBJECTS_PACK_DIR,
1428 		GOT_REFS_DIR,
1429 	};
1430 	const char *description_str = "Unnamed repository; "
1431 	    "edit this file 'description' to name the repository.";
1432 	const char *headref_str = "ref: refs/heads/main";
1433 	const char *gitconfig_str = "[core]\n"
1434 	    "\trepositoryformatversion = 0\n"
1435 	    "\tfilemode = true\n"
1436 	    "\tbare = true\n";
1437 	char *path;
1438 	size_t i;
1439 
1440 	if (!got_path_dir_is_empty(repo_path))
1441 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1442 
1443 	for (i = 0; i < nitems(dirnames); i++) {
1444 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1445 			return got_error_from_errno("asprintf");
1446 		}
1447 		err = got_path_mkdir(path);
1448 		free(path);
1449 		if (err)
1450 			return err;
1451 	}
1452 
1453 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1454 		return got_error_from_errno("asprintf");
1455 	err = got_path_create_file(path, description_str);
1456 	free(path);
1457 	if (err)
1458 		return err;
1459 
1460 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1461 		return got_error_from_errno("asprintf");
1462 	err = got_path_create_file(path, headref_str);
1463 	free(path);
1464 	if (err)
1465 		return err;
1466 
1467 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1468 		return got_error_from_errno("asprintf");
1469 	err = got_path_create_file(path, gitconfig_str);
1470 	free(path);
1471 	if (err)
1472 		return err;
1473 
1474 	return NULL;
1475 }
1476 
1477 static const struct got_error *
1478 match_packed_object(struct got_object_id **unique_id,
1479     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1480 {
1481 	const struct got_error *err = NULL;
1482 	struct got_object_id_queue matched_ids;
1483 	struct got_pathlist_entry *pe;
1484 
1485 	STAILQ_INIT(&matched_ids);
1486 
1487 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1488 		const char *path_packidx = pe->path;
1489 		struct got_packidx *packidx;
1490 		struct got_object_qid *qid;
1491 
1492 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1493 		    path_packidx, 0);
1494 		if (err)
1495 			break;
1496 
1497 		err = got_packidx_match_id_str_prefix(&matched_ids,
1498 		    packidx, id_str_prefix);
1499 		if (err) {
1500 			got_packidx_close(packidx);
1501 			break;
1502 		}
1503 		err = got_packidx_close(packidx);
1504 		if (err)
1505 			break;
1506 
1507 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1508 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1509 				int matched_type;
1510 				err = got_object_get_type(&matched_type, repo,
1511 				    &qid->id);
1512 				if (err)
1513 					goto done;
1514 				if (matched_type != obj_type)
1515 					continue;
1516 			}
1517 			if (*unique_id == NULL) {
1518 				*unique_id = got_object_id_dup(&qid->id);
1519 				if (*unique_id == NULL) {
1520 					err = got_error_from_errno("malloc");
1521 					goto done;
1522 				}
1523 			} else {
1524 				if (got_object_id_cmp(*unique_id,
1525 				    &qid->id) == 0)
1526 					continue; /* packed multiple times */
1527 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1528 				goto done;
1529 			}
1530 		}
1531 	}
1532 done:
1533 	got_object_id_queue_free(&matched_ids);
1534 	if (err) {
1535 		free(*unique_id);
1536 		*unique_id = NULL;
1537 	}
1538 	return err;
1539 }
1540 
1541 static const struct got_error *
1542 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1543     const char *object_dir, const char *id_str_prefix, int obj_type,
1544     struct got_repository *repo)
1545 {
1546 	const struct got_error *err = NULL;
1547 	char *path;
1548 	DIR *dir = NULL;
1549 	struct dirent *dent;
1550 	struct got_object_id id;
1551 
1552 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1553 		err = got_error_from_errno("asprintf");
1554 		goto done;
1555 	}
1556 
1557 	dir = opendir(path);
1558 	if (dir == NULL) {
1559 		if (errno == ENOENT) {
1560 			err = NULL;
1561 			goto done;
1562 		}
1563 		err = got_error_from_errno2("opendir", path);
1564 		goto done;
1565 	}
1566 	while ((dent = readdir(dir)) != NULL) {
1567 		char *id_str;
1568 		int cmp;
1569 
1570 		if (strcmp(dent->d_name, ".") == 0 ||
1571 		    strcmp(dent->d_name, "..") == 0)
1572 			continue;
1573 
1574 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1575 			err = got_error_from_errno("asprintf");
1576 			goto done;
1577 		}
1578 
1579 		if (!got_parse_sha1_digest(id.sha1, id_str))
1580 			continue;
1581 
1582 		/*
1583 		 * Directory entries do not necessarily appear in
1584 		 * sorted order, so we must iterate over all of them.
1585 		 */
1586 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1587 		if (cmp != 0) {
1588 			free(id_str);
1589 			continue;
1590 		}
1591 
1592 		if (*unique_id == NULL) {
1593 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1594 				int matched_type;
1595 				err = got_object_get_type(&matched_type, repo,
1596 				    &id);
1597 				if (err)
1598 					goto done;
1599 				if (matched_type != obj_type)
1600 					continue;
1601 			}
1602 			*unique_id = got_object_id_dup(&id);
1603 			if (*unique_id == NULL) {
1604 				err = got_error_from_errno("got_object_id_dup");
1605 				free(id_str);
1606 				goto done;
1607 			}
1608 		} else {
1609 			if (got_object_id_cmp(*unique_id, &id) == 0)
1610 				continue; /* both packed and loose */
1611 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1612 			free(id_str);
1613 			goto done;
1614 		}
1615 	}
1616 done:
1617 	if (dir && closedir(dir) != 0 && err == NULL)
1618 		err = got_error_from_errno("closedir");
1619 	if (err) {
1620 		free(*unique_id);
1621 		*unique_id = NULL;
1622 	}
1623 	free(path);
1624 	return err;
1625 }
1626 
1627 const struct got_error *
1628 got_repo_match_object_id_prefix(struct got_object_id **id,
1629     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1630 {
1631 	const struct got_error *err = NULL;
1632 	char *path_objects = got_repo_get_path_objects(repo);
1633 	char *object_dir = NULL;
1634 	size_t len;
1635 	int i;
1636 
1637 	*id = NULL;
1638 
1639 	len = strlen(id_str_prefix);
1640 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1641 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1642 
1643 	for (i = 0; i < len; i++) {
1644 		if (isxdigit((unsigned char)id_str_prefix[i]))
1645 			continue;
1646 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1647 	}
1648 
1649 	if (len >= 2) {
1650 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1651 		if (err)
1652 			goto done;
1653 		object_dir = strndup(id_str_prefix, 2);
1654 		if (object_dir == NULL) {
1655 			err = got_error_from_errno("strdup");
1656 			goto done;
1657 		}
1658 		err = match_loose_object(id, path_objects, object_dir,
1659 		    id_str_prefix, obj_type, repo);
1660 	} else if (len == 1) {
1661 		int i;
1662 		for (i = 0; i < 0xf; i++) {
1663 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1664 			    == -1) {
1665 				err = got_error_from_errno("asprintf");
1666 				goto done;
1667 			}
1668 			err = match_packed_object(id, repo, object_dir,
1669 			    obj_type);
1670 			if (err)
1671 				goto done;
1672 			err = match_loose_object(id, path_objects, object_dir,
1673 			    id_str_prefix, obj_type, repo);
1674 			if (err)
1675 				goto done;
1676 		}
1677 	} else {
1678 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1679 		goto done;
1680 	}
1681 done:
1682 	free(object_dir);
1683 	if (err) {
1684 		free(*id);
1685 		*id = NULL;
1686 	} else if (*id == NULL) {
1687 		switch (obj_type) {
1688 		case GOT_OBJ_TYPE_BLOB:
1689 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1690 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1691 			break;
1692 		case GOT_OBJ_TYPE_TREE:
1693 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1694 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1695 			break;
1696 		case GOT_OBJ_TYPE_COMMIT:
1697 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1698 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1699 			break;
1700 		case GOT_OBJ_TYPE_TAG:
1701 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1702 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1703 			break;
1704 		default:
1705 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1706 			break;
1707 		}
1708 	}
1709 
1710 	return err;
1711 }
1712 
1713 const struct got_error *
1714 got_repo_match_object_id(struct got_object_id **id, char **label,
1715     const char *id_str, int obj_type, struct got_reflist_head *refs,
1716     struct got_repository *repo)
1717 {
1718 	const struct got_error *err;
1719 	struct got_tag_object *tag;
1720 	struct got_reference *ref = NULL;
1721 
1722 	*id = NULL;
1723 	if (label)
1724 		*label = NULL;
1725 
1726 	if (refs) {
1727 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1728 		    refs, repo);
1729 		if (err == NULL) {
1730 			*id = got_object_id_dup(
1731 			    got_object_tag_get_object_id(tag));
1732 			if (*id == NULL)
1733 				err = got_error_from_errno("got_object_id_dup");
1734 			else if (label && asprintf(label, "refs/tags/%s",
1735 			    got_object_tag_get_name(tag)) == -1) {
1736 				err = got_error_from_errno("asprintf");
1737 				free(*id);
1738 				*id = NULL;
1739 			}
1740 			got_object_tag_close(tag);
1741 			return err;
1742 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1743 		    err->code != GOT_ERR_NO_OBJ)
1744 			return err;
1745 	}
1746 
1747 	err = got_ref_open(&ref, repo, id_str, 0);
1748 	if (err == NULL) {
1749 		err = got_ref_resolve(id, repo, ref);
1750 		if (err)
1751 			goto done;
1752 		if (label) {
1753 			*label = strdup(got_ref_get_name(ref));
1754 			if (*label == NULL) {
1755 				err = got_error_from_errno("strdup");
1756 				goto done;
1757 			}
1758 		}
1759 	} else {
1760 		if (err->code != GOT_ERR_NOT_REF &&
1761 		    err->code != GOT_ERR_BAD_REF_NAME)
1762 			goto done;
1763 		err = got_repo_match_object_id_prefix(id, id_str,
1764 		    obj_type, repo);
1765 		if (err) {
1766 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1767 				err = got_error_not_ref(id_str);
1768 			goto done;
1769 		}
1770 		if (label) {
1771 			err = got_object_id_str(label, *id);
1772 			if (*label == NULL) {
1773 				err = got_error_from_errno("strdup");
1774 				goto done;
1775 			}
1776 		}
1777 	}
1778 done:
1779 	if (ref)
1780 		got_ref_close(ref);
1781 	return err;
1782 }
1783 
1784 const struct got_error *
1785 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1786     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1787 {
1788 	const struct got_error *err = NULL;
1789 	struct got_reflist_entry *re;
1790 	struct got_object_id *tag_id;
1791 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1792 
1793 	*tag = NULL;
1794 
1795 	TAILQ_FOREACH(re, refs, entry) {
1796 		const char *refname;
1797 		refname = got_ref_get_name(re->ref);
1798 		if (got_ref_is_symbolic(re->ref))
1799 			continue;
1800 		if (strncmp(refname, "refs/tags/", 10) != 0)
1801 			continue;
1802 		if (!name_is_absolute)
1803 			refname += strlen("refs/tags/");
1804 		if (strcmp(refname, name) != 0)
1805 			continue;
1806 		err = got_ref_resolve(&tag_id, repo, re->ref);
1807 		if (err)
1808 			break;
1809 		err = got_object_open_as_tag(tag, repo, tag_id);
1810 		free(tag_id);
1811 		if (err)
1812 			break;
1813 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1814 		    got_object_tag_get_object_type(*tag) == obj_type)
1815 			break;
1816 		got_object_tag_close(*tag);
1817 		*tag = NULL;
1818 	}
1819 
1820 	if (err == NULL && *tag == NULL)
1821 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1822 		    GOT_OBJ_LABEL_TAG, name);
1823 	return err;
1824 }
1825 
1826 static const struct got_error *
1827 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1828     const char *name, mode_t mode, struct got_object_id *blob_id)
1829 {
1830 	const struct got_error *err = NULL;
1831 
1832 	 *new_te = NULL;
1833 
1834 	*new_te = calloc(1, sizeof(**new_te));
1835 	if (*new_te == NULL)
1836 		return got_error_from_errno("calloc");
1837 
1838 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1839 	    sizeof((*new_te)->name)) {
1840 		err = got_error(GOT_ERR_NO_SPACE);
1841 		goto done;
1842 	}
1843 
1844 	if (S_ISLNK(mode)) {
1845 		(*new_te)->mode = S_IFLNK;
1846 	} else {
1847 		(*new_te)->mode = S_IFREG;
1848 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1849 	}
1850 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1851 done:
1852 	if (err && *new_te) {
1853 		free(*new_te);
1854 		*new_te = NULL;
1855 	}
1856 	return err;
1857 }
1858 
1859 static const struct got_error *
1860 import_file(struct got_tree_entry **new_te, struct dirent *de,
1861     const char *path, struct got_repository *repo)
1862 {
1863 	const struct got_error *err;
1864 	struct got_object_id *blob_id = NULL;
1865 	char *filepath;
1866 	struct stat sb;
1867 
1868 	if (asprintf(&filepath, "%s%s%s", path,
1869 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1870 		return got_error_from_errno("asprintf");
1871 
1872 	if (lstat(filepath, &sb) != 0) {
1873 		err = got_error_from_errno2("lstat", path);
1874 		goto done;
1875 	}
1876 
1877 	err = got_object_blob_create(&blob_id, filepath, repo);
1878 	if (err)
1879 		goto done;
1880 
1881 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1882 	    blob_id);
1883 done:
1884 	free(filepath);
1885 	if (err)
1886 		free(blob_id);
1887 	return err;
1888 }
1889 
1890 static const struct got_error *
1891 insert_tree_entry(struct got_tree_entry *new_te,
1892     struct got_pathlist_head *paths)
1893 {
1894 	const struct got_error *err = NULL;
1895 	struct got_pathlist_entry *new_pe;
1896 
1897 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1898 	if (err)
1899 		return err;
1900 	if (new_pe == NULL)
1901 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1902 	return NULL;
1903 }
1904 
1905 static const struct got_error *write_tree(struct got_object_id **,
1906     const char *, struct got_pathlist_head *, struct got_repository *,
1907     got_repo_import_cb progress_cb, void *progress_arg);
1908 
1909 static const struct got_error *
1910 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1911     const char *path, struct got_pathlist_head *ignores,
1912     struct got_repository *repo,
1913     got_repo_import_cb progress_cb, void *progress_arg)
1914 {
1915 	const struct got_error *err;
1916 	struct got_object_id *id = NULL;
1917 	char *subdirpath;
1918 
1919 	if (asprintf(&subdirpath, "%s%s%s", path,
1920 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1921 		return got_error_from_errno("asprintf");
1922 
1923 	(*new_te) = calloc(1, sizeof(**new_te));
1924 	if (*new_te == NULL)
1925 		return got_error_from_errno("calloc");
1926 	(*new_te)->mode = S_IFDIR;
1927 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1928 	    sizeof((*new_te)->name)) {
1929 		err = got_error(GOT_ERR_NO_SPACE);
1930 		goto done;
1931 	}
1932 	err = write_tree(&id, subdirpath, ignores,  repo,
1933 	    progress_cb, progress_arg);
1934 	if (err)
1935 		goto done;
1936 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1937 	
1938 done:
1939 	free(id);
1940 	free(subdirpath);
1941 	if (err) {
1942 		free(*new_te);
1943 		*new_te = NULL;
1944 	}
1945 	return err;
1946 }
1947 
1948 static const struct got_error *
1949 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1950     struct got_pathlist_head *ignores, struct got_repository *repo,
1951     got_repo_import_cb progress_cb, void *progress_arg)
1952 {
1953 	const struct got_error *err = NULL;
1954 	DIR *dir;
1955 	struct dirent *de;
1956 	int nentries;
1957 	struct got_tree_entry *new_te = NULL;
1958 	struct got_pathlist_head paths;
1959 	struct got_pathlist_entry *pe;
1960 
1961 	*new_tree_id = NULL;
1962 
1963 	TAILQ_INIT(&paths);
1964 
1965 	dir = opendir(path_dir);
1966 	if (dir == NULL) {
1967 		err = got_error_from_errno2("opendir", path_dir);
1968 		goto done;
1969 	}
1970 
1971 	nentries = 0;
1972 	while ((de = readdir(dir)) != NULL) {
1973 		int ignore = 0;
1974 		int type;
1975 
1976 		if (strcmp(de->d_name, ".") == 0 ||
1977 		    strcmp(de->d_name, "..") == 0)
1978 			continue;
1979 
1980 		TAILQ_FOREACH(pe, ignores, entry) {
1981 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1982 				ignore = 1;
1983 				break;
1984 			}
1985 		}
1986 		if (ignore)
1987 			continue;
1988 
1989 		err = got_path_dirent_type(&type, path_dir, de);
1990 		if (err)
1991 			goto done;
1992 
1993 		if (type == DT_DIR) {
1994 			err = import_subdir(&new_te, de, path_dir,
1995 			    ignores, repo, progress_cb, progress_arg);
1996 			if (err) {
1997 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1998 					goto done;
1999 				err = NULL;
2000 				continue;
2001 			}
2002 		} else if (type == DT_REG || type == DT_LNK) {
2003 			err = import_file(&new_te, de, path_dir, repo);
2004 			if (err)
2005 				goto done;
2006 		} else
2007 			continue;
2008 
2009 		err = insert_tree_entry(new_te, &paths);
2010 		if (err)
2011 			goto done;
2012 		nentries++;
2013 	}
2014 
2015 	if (TAILQ_EMPTY(&paths)) {
2016 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2017 		    "cannot create tree without any entries");
2018 		goto done;
2019 	}
2020 
2021 	TAILQ_FOREACH(pe, &paths, entry) {
2022 		struct got_tree_entry *te = pe->data;
2023 		char *path;
2024 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2025 			continue;
2026 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2027 			err = got_error_from_errno("asprintf");
2028 			goto done;
2029 		}
2030 		err = (*progress_cb)(progress_arg, path);
2031 		free(path);
2032 		if (err)
2033 			goto done;
2034 	}
2035 
2036 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2037 done:
2038 	if (dir)
2039 		closedir(dir);
2040 	got_pathlist_free(&paths);
2041 	return err;
2042 }
2043 
2044 const struct got_error *
2045 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2046     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2047     struct got_repository *repo, got_repo_import_cb progress_cb,
2048     void *progress_arg)
2049 {
2050 	const struct got_error *err;
2051 	struct got_object_id *new_tree_id;
2052 
2053 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2054 	    progress_cb, progress_arg);
2055 	if (err)
2056 		return err;
2057 
2058 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2059 	    author, time(NULL), author, time(NULL), logmsg, repo);
2060 	free(new_tree_id);
2061 	return err;
2062 }
2063 
2064 const struct got_error *
2065 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2066     struct got_repository *repo)
2067 {
2068 	const struct got_error *err = NULL;
2069 	char *path_objects = NULL, *path = NULL;
2070 	DIR *dir = NULL;
2071 	struct got_object_id id;
2072 	int i;
2073 
2074 	*nobjects = 0;
2075 	*ondisk_size = 0;
2076 
2077 	path_objects = got_repo_get_path_objects(repo);
2078 	if (path_objects == NULL)
2079 		return got_error_from_errno("got_repo_get_path_objects");
2080 
2081 	for (i = 0; i <= 0xff; i++) {
2082 		struct dirent *dent;
2083 
2084 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2085 			err = got_error_from_errno("asprintf");
2086 			break;
2087 		}
2088 
2089 		dir = opendir(path);
2090 		if (dir == NULL) {
2091 			if (errno == ENOENT) {
2092 				err = NULL;
2093 				continue;
2094 			}
2095 			err = got_error_from_errno2("opendir", path);
2096 			break;
2097 		}
2098 
2099 		while ((dent = readdir(dir)) != NULL) {
2100 			char *id_str;
2101 			int fd;
2102 			struct stat sb;
2103 
2104 			if (strcmp(dent->d_name, ".") == 0 ||
2105 			    strcmp(dent->d_name, "..") == 0)
2106 				continue;
2107 
2108 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2109 				err = got_error_from_errno("asprintf");
2110 				goto done;
2111 			}
2112 
2113 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2114 				free(id_str);
2115 				continue;
2116 			}
2117 			free(id_str);
2118 
2119 			err = got_object_open_loose_fd(&fd, &id, repo);
2120 			if (err)
2121 				goto done;
2122 
2123 			if (fstat(fd, &sb) == -1) {
2124 				err = got_error_from_errno("fstat");
2125 				close(fd);
2126 				goto done;
2127 			}
2128 			(*nobjects)++;
2129 			(*ondisk_size) += sb.st_size;
2130 
2131 			if (close(fd) == -1) {
2132 				err = got_error_from_errno("close");
2133 				goto done;
2134 			}
2135 		}
2136 
2137 		if (closedir(dir) != 0) {
2138 			err = got_error_from_errno("closedir");
2139 			goto done;
2140 		}
2141 		dir = NULL;
2142 
2143 		free(path);
2144 		path = NULL;
2145 	}
2146 done:
2147 	if (dir && closedir(dir) != 0 && err == NULL)
2148 		err = got_error_from_errno("closedir");
2149 
2150 	if (err) {
2151 		*nobjects = 0;
2152 		*ondisk_size = 0;
2153 	}
2154 	free(path_objects);
2155 	free(path);
2156 	return err;
2157 }
2158 
2159 const struct got_error *
2160 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2161     off_t *total_packsize, struct got_repository *repo)
2162 {
2163 	const struct got_error *err = NULL;
2164 	DIR *packdir = NULL;
2165 	struct dirent *dent;
2166 	struct got_packidx *packidx = NULL;
2167 	char *path_packidx;
2168 	char *path_packfile;
2169 	int packdir_fd;
2170 	struct stat sb;
2171 
2172 	*npackfiles = 0;
2173 	*nobjects = 0;
2174 	*total_packsize = 0;
2175 
2176 	packdir_fd = openat(got_repo_get_fd(repo),
2177 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2178 	if (packdir_fd == -1) {
2179 		return got_error_from_errno_fmt("openat: %s/%s",
2180 		    got_repo_get_path_git_dir(repo),
2181 		    GOT_OBJECTS_PACK_DIR);
2182 	}
2183 
2184 	packdir = fdopendir(packdir_fd);
2185 	if (packdir == NULL) {
2186 		err = got_error_from_errno("fdopendir");
2187 		goto done;
2188 	}
2189 
2190 	while ((dent = readdir(packdir)) != NULL) {
2191 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2192 			continue;
2193 
2194 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2195 		    dent->d_name) == -1) {
2196 			err = got_error_from_errno("asprintf");
2197 			goto done;
2198 		}
2199 
2200 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2201 		    path_packidx, 0);
2202 		free(path_packidx);
2203 		if (err)
2204 			goto done;
2205 
2206 		if (fstat(packidx->fd, &sb) == -1)
2207 			goto done;
2208 		*total_packsize += sb.st_size;
2209 
2210 		err = got_packidx_get_packfile_path(&path_packfile,
2211 		    packidx->path_packidx);
2212 		if (err)
2213 			goto done;
2214 
2215 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2216 		    0) == -1) {
2217 			free(path_packfile);
2218 			goto done;
2219 		}
2220 		free(path_packfile);
2221 		*total_packsize += sb.st_size;
2222 
2223 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2224 
2225 		(*npackfiles)++;
2226 
2227 		got_packidx_close(packidx);
2228 		packidx = NULL;
2229 	}
2230 done:
2231 	if (packidx)
2232 		got_packidx_close(packidx);
2233 	if (packdir && closedir(packdir) != 0 && err == NULL)
2234 		err = got_error_from_errno("closedir");
2235 	if (err) {
2236 		*npackfiles = 0;
2237 		*nobjects = 0;
2238 		*total_packsize = 0;
2239 	}
2240 	return err;
2241 }
2242 
2243 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2244     got_packidx_bloom_filter_cmp);