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 
39 #include "bloom.h"
40 
41 #include "got_error.h"
42 #include "got_reference.h"
43 #include "got_repository.h"
44 #include "got_path.h"
45 #include "got_cancel.h"
46 #include "got_object.h"
47 #include "got_opentemp.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 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
66 
67 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
68     got_packidx_bloom_filter_cmp);
69 
70 const char *
71 got_repo_get_path(struct got_repository *repo)
72 {
73 	return repo->path;
74 }
75 
76 const char *
77 got_repo_get_path_git_dir(struct got_repository *repo)
78 {
79 	return repo->path_git_dir;
80 }
81 
82 int
83 got_repo_get_fd(struct got_repository *repo)
84 {
85 	return repo->gitdir_fd;
86 }
87 
88 const char *
89 got_repo_get_gitconfig_author_name(struct got_repository *repo)
90 {
91 	return repo->gitconfig_author_name;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_email(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_email;
98 }
99 
100 const char *
101 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
102 {
103 	return repo->global_gitconfig_author_name;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_email;
110 }
111 
112 const char *
113 got_repo_get_gitconfig_owner(struct got_repository *repo)
114 {
115 	return repo->gitconfig_owner;
116 }
117 
118 void
119 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
120     struct got_repository *repo)
121 {
122 	*extensions = repo->extensions;
123 	*nextensions = repo->nextensions;
124 }
125 
126 int
127 got_repo_is_bare(struct got_repository *repo)
128 {
129 	return (strcmp(repo->path, repo->path_git_dir) == 0);
130 }
131 
132 static char *
133 get_path_git_child(struct got_repository *repo, const char *basename)
134 {
135 	char *path_child;
136 
137 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
138 	    basename) == -1)
139 		return NULL;
140 
141 	return path_child;
142 }
143 
144 char *
145 got_repo_get_path_objects(struct got_repository *repo)
146 {
147 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
148 }
149 
150 char *
151 got_repo_get_path_objects_pack(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
154 }
155 
156 char *
157 got_repo_get_path_refs(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_REFS_DIR);
160 }
161 
162 char *
163 got_repo_get_path_packed_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
166 }
167 
168 static char *
169 get_path_head(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_HEAD_FILE);
172 }
173 
174 char *
175 got_repo_get_path_gitconfig(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_GITCONFIG);
178 }
179 
180 char *
181 got_repo_get_path_gotconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
184 }
185 
186 const struct got_gotconfig *
187 got_repo_get_gotconfig(struct got_repository *repo)
188 {
189 	return repo->gotconfig;
190 }
191 
192 void
193 got_repo_get_gitconfig_remotes(int *nremotes,
194     const struct got_remote_repo **remotes, struct got_repository *repo)
195 {
196 	*nremotes = repo->ngitconfig_remotes;
197 	*remotes = repo->gitconfig_remotes;
198 }
199 
200 static int
201 is_git_repo(struct got_repository *repo)
202 {
203 	const char *path_git = got_repo_get_path_git_dir(repo);
204 	char *path_objects = got_repo_get_path_objects(repo);
205 	char *path_refs = got_repo_get_path_refs(repo);
206 	char *path_head = get_path_head(repo);
207 	int ret = 0;
208 	struct stat sb;
209 	struct got_reference *head_ref;
210 
211 	if (lstat(path_git, &sb) == -1)
212 		goto done;
213 	if (!S_ISDIR(sb.st_mode))
214 		goto done;
215 
216 	if (lstat(path_objects, &sb) == -1)
217 		goto done;
218 	if (!S_ISDIR(sb.st_mode))
219 		goto done;
220 
221 	if (lstat(path_refs, &sb) == -1)
222 		goto done;
223 	if (!S_ISDIR(sb.st_mode))
224 		goto done;
225 
226 	if (lstat(path_head, &sb) == -1)
227 		goto done;
228 	if (!S_ISREG(sb.st_mode))
229 		goto done;
230 
231 	/* Check if the HEAD reference can be opened. */
232 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
233 		goto done;
234 	got_ref_close(head_ref);
235 
236 	ret = 1;
237 done:
238 	free(path_objects);
239 	free(path_refs);
240 	free(path_head);
241 	return ret;
242 
243 }
244 
245 const struct got_error *
246 got_repo_pack_fds_open(int **pack_fds)
247 {
248 	const struct got_error *err = NULL;
249 	int i, pack_fds_tmp[GOT_PACK_NUM_TEMPFILES];
250 
251 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
252 	if (*pack_fds == NULL)
253 		return got_error_from_errno("calloc");
254 
255 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
256 		pack_fds_tmp[i] = got_opentempfd();
257 		if (pack_fds_tmp[i] == -1) {
258 			err = got_error_from_errno("got_opentempfd");
259 			got_repo_pack_fds_close(pack_fds_tmp);
260 			return err;
261 		}
262 	}
263 	memcpy(*pack_fds, pack_fds_tmp, sizeof(pack_fds_tmp));
264 	return err;
265 }
266 
267 const struct got_error *
268 got_repo_pack_fds_close(int *pack_fds)
269 {
270 	const struct got_error *err = NULL;
271 	int i;
272 
273 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
274 		if (pack_fds[i] == -1)
275 			continue;
276 		if (close(pack_fds[i]) == -1) {
277 			err = got_error_from_errno("close");
278 			break;
279 		}
280 	}
281 	free(pack_fds);
282 	return err;
283 }
284 
285 const struct got_error *
286 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
287     struct got_object *obj)
288 {
289 #ifndef GOT_NO_OBJ_CACHE
290 	const struct got_error *err = NULL;
291 	err = got_object_cache_add(&repo->objcache, id, obj);
292 	if (err) {
293 		if (err->code == GOT_ERR_OBJ_EXISTS ||
294 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
295 			err = NULL;
296 		return err;
297 	}
298 	obj->refcnt++;
299 #endif
300 	return NULL;
301 }
302 
303 struct got_object *
304 got_repo_get_cached_object(struct got_repository *repo,
305     struct got_object_id *id)
306 {
307 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
308 }
309 
310 const struct got_error *
311 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
312     struct got_tree_object *tree)
313 {
314 #ifndef GOT_NO_OBJ_CACHE
315 	const struct got_error *err = NULL;
316 	err = got_object_cache_add(&repo->treecache, id, tree);
317 	if (err) {
318 		if (err->code == GOT_ERR_OBJ_EXISTS ||
319 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
320 			err = NULL;
321 		return err;
322 	}
323 	tree->refcnt++;
324 #endif
325 	return NULL;
326 }
327 
328 struct got_tree_object *
329 got_repo_get_cached_tree(struct got_repository *repo,
330     struct got_object_id *id)
331 {
332 	return (struct got_tree_object *)got_object_cache_get(
333 	    &repo->treecache, id);
334 }
335 
336 const struct got_error *
337 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
338     struct got_commit_object *commit)
339 {
340 #ifndef GOT_NO_OBJ_CACHE
341 	const struct got_error *err = NULL;
342 	err = got_object_cache_add(&repo->commitcache, id, commit);
343 	if (err) {
344 		if (err->code == GOT_ERR_OBJ_EXISTS ||
345 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
346 			err = NULL;
347 		return err;
348 	}
349 	commit->refcnt++;
350 #endif
351 	return NULL;
352 }
353 
354 struct got_commit_object *
355 got_repo_get_cached_commit(struct got_repository *repo,
356     struct got_object_id *id)
357 {
358 	return (struct got_commit_object *)got_object_cache_get(
359 	    &repo->commitcache, id);
360 }
361 
362 const struct got_error *
363 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
364     struct got_tag_object *tag)
365 {
366 #ifndef GOT_NO_OBJ_CACHE
367 	const struct got_error *err = NULL;
368 	err = got_object_cache_add(&repo->tagcache, id, tag);
369 	if (err) {
370 		if (err->code == GOT_ERR_OBJ_EXISTS ||
371 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
372 			err = NULL;
373 		return err;
374 	}
375 	tag->refcnt++;
376 #endif
377 	return NULL;
378 }
379 
380 struct got_tag_object *
381 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
382 {
383 	return (struct got_tag_object *)got_object_cache_get(
384 	    &repo->tagcache, id);
385 }
386 
387 const struct got_error *
388 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
389     struct got_raw_object *raw)
390 {
391 #ifndef GOT_NO_OBJ_CACHE
392 	const struct got_error *err = NULL;
393 	err = got_object_cache_add(&repo->rawcache, id, raw);
394 	if (err) {
395 		if (err->code == GOT_ERR_OBJ_EXISTS ||
396 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
397 			err = NULL;
398 		return err;
399 	}
400 	raw->refcnt++;
401 #endif
402 	return NULL;
403 }
404 
405 
406 struct got_raw_object *
407 got_repo_get_cached_raw_object(struct got_repository *repo,
408     struct got_object_id *id)
409 {
410 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
411 }
412 
413 
414 static const struct got_error *
415 open_repo(struct got_repository *repo, const char *path)
416 {
417 	const struct got_error *err = NULL;
418 
419 	repo->gitdir_fd = -1;
420 
421 	/* bare git repository? */
422 	repo->path_git_dir = strdup(path);
423 	if (repo->path_git_dir == NULL)
424 		return got_error_from_errno("strdup");
425 	if (is_git_repo(repo)) {
426 		repo->path = strdup(repo->path_git_dir);
427 		if (repo->path == NULL) {
428 			err = got_error_from_errno("strdup");
429 			goto done;
430 		}
431 		repo->gitdir_fd = open(repo->path_git_dir,
432 		    O_DIRECTORY | O_CLOEXEC);
433 		if (repo->gitdir_fd == -1) {
434 			err = got_error_from_errno2("open",
435 			    repo->path_git_dir);
436 			goto done;
437 		}
438 		return NULL;
439 	}
440 
441 	/* git repository with working tree? */
442 	free(repo->path_git_dir);
443 	repo->path_git_dir = NULL;
444 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
445 		err = got_error_from_errno("asprintf");
446 		goto done;
447 	}
448 	if (is_git_repo(repo)) {
449 		repo->path = strdup(path);
450 		if (repo->path == NULL) {
451 			err = got_error_from_errno("strdup");
452 			goto done;
453 		}
454 		repo->gitdir_fd = open(repo->path_git_dir,
455 		    O_DIRECTORY | O_CLOEXEC);
456 		if (repo->gitdir_fd == -1) {
457 			err = got_error_from_errno2("open",
458 			    repo->path_git_dir);
459 			goto done;
460 		}
461 		return NULL;
462 	}
463 
464 	err = got_error(GOT_ERR_NOT_GIT_REPO);
465 done:
466 	if (err) {
467 		free(repo->path);
468 		repo->path = NULL;
469 		free(repo->path_git_dir);
470 		repo->path_git_dir = NULL;
471 		if (repo->gitdir_fd != -1)
472 			close(repo->gitdir_fd);
473 		repo->gitdir_fd = -1;
474 
475 	}
476 	return err;
477 }
478 
479 static const struct got_error *
480 parse_gitconfig_file(int *gitconfig_repository_format_version,
481     char **gitconfig_author_name, char **gitconfig_author_email,
482     struct got_remote_repo **remotes, int *nremotes,
483     char **gitconfig_owner, char ***extensions, int *nextensions,
484     const char *gitconfig_path)
485 {
486 	const struct got_error *err = NULL, *child_err = NULL;
487 	int fd = -1;
488 	int imsg_fds[2] = { -1, -1 };
489 	pid_t pid;
490 	struct imsgbuf *ibuf;
491 
492 	*gitconfig_repository_format_version = 0;
493 	if (extensions)
494 		*extensions = NULL;
495 	if (nextensions)
496 		*nextensions = 0;
497 	*gitconfig_author_name = NULL;
498 	*gitconfig_author_email = NULL;
499 	if (remotes)
500 		*remotes = NULL;
501 	if (nremotes)
502 		*nremotes = 0;
503 	if (gitconfig_owner)
504 		*gitconfig_owner = NULL;
505 
506 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
507 	if (fd == -1) {
508 		if (errno == ENOENT)
509 			return NULL;
510 		return got_error_from_errno2("open", gitconfig_path);
511 	}
512 
513 	ibuf = calloc(1, sizeof(*ibuf));
514 	if (ibuf == NULL) {
515 		err = got_error_from_errno("calloc");
516 		goto done;
517 	}
518 
519 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
520 		err = got_error_from_errno("socketpair");
521 		goto done;
522 	}
523 
524 	pid = fork();
525 	if (pid == -1) {
526 		err = got_error_from_errno("fork");
527 		goto done;
528 	} else if (pid == 0) {
529 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
530 		    gitconfig_path);
531 		/* not reached */
532 	}
533 
534 	if (close(imsg_fds[1]) == -1) {
535 		err = got_error_from_errno("close");
536 		goto done;
537 	}
538 	imsg_fds[1] = -1;
539 	imsg_init(ibuf, imsg_fds[0]);
540 
541 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
542 	if (err)
543 		goto done;
544 	fd = -1;
545 
546 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
547 	if (err)
548 		goto done;
549 
550 	err = got_privsep_recv_gitconfig_int(
551 	    gitconfig_repository_format_version, ibuf);
552 	if (err)
553 		goto done;
554 
555 	if (extensions && nextensions) {
556 		err = got_privsep_send_gitconfig_repository_extensions_req(
557 		    ibuf);
558 		if (err)
559 			goto done;
560 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
561 		if (err)
562 			goto done;
563 		if (*nextensions > 0) {
564 			int i;
565 			*extensions = calloc(*nextensions, sizeof(char *));
566 			if (*extensions == NULL) {
567 				err = got_error_from_errno("calloc");
568 				goto done;
569 			}
570 			for (i = 0; i < *nextensions; i++) {
571 				char *ext;
572 				err = got_privsep_recv_gitconfig_str(&ext,
573 				    ibuf);
574 				if (err)
575 					goto done;
576 				(*extensions)[i] = ext;
577 			}
578 		}
579 	}
580 
581 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
582 	if (err)
583 		goto done;
584 
585 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
586 	if (err)
587 		goto done;
588 
589 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
590 	if (err)
591 		goto done;
592 
593 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
594 	if (err)
595 		goto done;
596 
597 	if (remotes && nremotes) {
598 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
599 		if (err)
600 			goto done;
601 
602 		err = got_privsep_recv_gitconfig_remotes(remotes,
603 		    nremotes, ibuf);
604 		if (err)
605 			goto done;
606 	}
607 
608 	if (gitconfig_owner) {
609 		err = got_privsep_send_gitconfig_owner_req(ibuf);
610 		if (err)
611 			goto done;
612 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
613 		if (err)
614 			goto done;
615 	}
616 
617 	err = got_privsep_send_stop(imsg_fds[0]);
618 	child_err = got_privsep_wait_for_child(pid);
619 	if (child_err && err == NULL)
620 		err = child_err;
621 done:
622 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
623 		err = got_error_from_errno("close");
624 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
625 		err = got_error_from_errno("close");
626 	if (fd != -1 && close(fd) == -1 && err == NULL)
627 		err = got_error_from_errno2("close", gitconfig_path);
628 	free(ibuf);
629 	return err;
630 }
631 
632 static const struct got_error *
633 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
634 {
635 	const struct got_error *err = NULL;
636 	char *repo_gitconfig_path = NULL;
637 
638 	if (global_gitconfig_path) {
639 		/* Read settings from ~/.gitconfig. */
640 		int dummy_repo_version;
641 		err = parse_gitconfig_file(&dummy_repo_version,
642 		    &repo->global_gitconfig_author_name,
643 		    &repo->global_gitconfig_author_email,
644 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
645 		if (err)
646 			return err;
647 	}
648 
649 	/* Read repository's .git/config file. */
650 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
651 	if (repo_gitconfig_path == NULL)
652 		return got_error_from_errno("got_repo_get_path_gitconfig");
653 
654 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
655 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
656 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
657 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
658 	    repo_gitconfig_path);
659 	if (err)
660 		goto done;
661 done:
662 	free(repo_gitconfig_path);
663 	return err;
664 }
665 
666 static const struct got_error *
667 read_gotconfig(struct got_repository *repo)
668 {
669 	const struct got_error *err = NULL;
670 	char *gotconfig_path;
671 
672 	gotconfig_path = got_repo_get_path_gotconfig(repo);
673 	if (gotconfig_path == NULL)
674 		return got_error_from_errno("got_repo_get_path_gotconfig");
675 
676 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
677 	free(gotconfig_path);
678 	return err;
679 }
680 
681 /* Supported repository format extensions. */
682 static const char *const repo_extensions[] = {
683 	"noop",			/* Got supports repository format version 1. */
684 	"preciousObjects",	/* Supported by gotadmin cleanup. */
685 	"worktreeConfig",	/* Got does not care about Git work trees. */
686 };
687 
688 const struct got_error *
689 got_repo_open(struct got_repository **repop, const char *path,
690     const char *global_gitconfig_path, int *pack_fds)
691 {
692 	struct got_repository *repo = NULL;
693 	const struct got_error *err = NULL;
694 	char *repo_path = NULL;
695 	size_t i, j = 0;
696 	struct rlimit rl;
697 
698 	*repop = NULL;
699 
700 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
701 		return got_error_from_errno("getrlimit");
702 
703 	repo = calloc(1, sizeof(*repo));
704 	if (repo == NULL)
705 		return got_error_from_errno("calloc");
706 
707 	RB_INIT(&repo->packidx_bloom_filters);
708 	TAILQ_INIT(&repo->packidx_paths);
709 
710 	for (i = 0; i < nitems(repo->privsep_children); i++) {
711 		memset(&repo->privsep_children[i], 0,
712 		    sizeof(repo->privsep_children[0]));
713 		repo->privsep_children[i].imsg_fd = -1;
714 	}
715 
716 	err = got_object_cache_init(&repo->objcache,
717 	    GOT_OBJECT_CACHE_TYPE_OBJ);
718 	if (err)
719 		goto done;
720 	err = got_object_cache_init(&repo->treecache,
721 	    GOT_OBJECT_CACHE_TYPE_TREE);
722 	if (err)
723 		goto done;
724 	err = got_object_cache_init(&repo->commitcache,
725 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
726 	if (err)
727 		goto done;
728 	err = got_object_cache_init(&repo->tagcache,
729 	    GOT_OBJECT_CACHE_TYPE_TAG);
730 	if (err)
731 		goto done;
732 	err = got_object_cache_init(&repo->rawcache,
733 	    GOT_OBJECT_CACHE_TYPE_RAW);
734 	if (err)
735 		goto done;
736 
737 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
738 	if (repo->pack_cache_size > rl.rlim_cur / 8)
739 		repo->pack_cache_size = rl.rlim_cur / 8;
740 	for (i = 0; i < nitems(repo->packs); i++) {
741 		if (i < repo->pack_cache_size) {
742 			repo->packs[i].basefd = pack_fds[j++];
743 			repo->packs[i].accumfd = pack_fds[j++];
744 		} else {
745 			repo->packs[i].basefd = -1;
746 			repo->packs[i].accumfd = -1;
747 		}
748 	}
749 
750 	repo_path = realpath(path, NULL);
751 	if (repo_path == NULL) {
752 		err = got_error_from_errno2("realpath", path);
753 		goto done;
754 	}
755 
756 	for (;;) {
757 		char *parent_path;
758 
759 		err = open_repo(repo, repo_path);
760 		if (err == NULL)
761 			break;
762 		if (err->code != GOT_ERR_NOT_GIT_REPO)
763 			goto done;
764 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
765 			err = got_error(GOT_ERR_NOT_GIT_REPO);
766 			goto done;
767 		}
768 		err = got_path_dirname(&parent_path, repo_path);
769 		if (err)
770 			goto done;
771 		free(repo_path);
772 		repo_path = parent_path;
773 	}
774 
775 	err = read_gotconfig(repo);
776 	if (err)
777 		goto done;
778 
779 	err = read_gitconfig(repo, global_gitconfig_path);
780 	if (err)
781 		goto done;
782 	if (repo->gitconfig_repository_format_version != 0)
783 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
784 	for (i = 0; i < repo->nextensions; i++) {
785 		char *ext = repo->extensions[i];
786 		int j, supported = 0;
787 		for (j = 0; j < nitems(repo_extensions); j++) {
788 			if (strcmp(ext, repo_extensions[j]) == 0) {
789 				supported = 1;
790 				break;
791 			}
792 		}
793 		if (!supported) {
794 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
795 			goto done;
796 		}
797 	}
798 
799 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
800 done:
801 	if (err)
802 		got_repo_close(repo);
803 	else
804 		*repop = repo;
805 	free(repo_path);
806 	return err;
807 }
808 
809 const struct got_error *
810 got_repo_close(struct got_repository *repo)
811 {
812 	const struct got_error *err = NULL, *child_err;
813 	struct got_packidx_bloom_filter *bf;
814 	struct got_pathlist_entry *pe;
815 	size_t i;
816 
817 	for (i = 0; i < repo->pack_cache_size; i++) {
818 		if (repo->packidx_cache[i] == NULL)
819 			break;
820 		got_packidx_close(repo->packidx_cache[i]);
821 	}
822 
823 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
824 	    &repo->packidx_bloom_filters))) {
825 		RB_REMOVE(got_packidx_bloom_filter_tree,
826 		    &repo->packidx_bloom_filters, bf);
827 		free(bf->bloom);
828 		free(bf);
829 	}
830 
831 	for (i = 0; i < repo->pack_cache_size; i++)
832 		if (repo->packs[i].path_packfile)
833 			if (repo->packs[i].path_packfile)
834 				got_pack_close(&repo->packs[i]);
835 
836 	free(repo->path);
837 	free(repo->path_git_dir);
838 
839 	got_object_cache_close(&repo->objcache);
840 	got_object_cache_close(&repo->treecache);
841 	got_object_cache_close(&repo->commitcache);
842 	got_object_cache_close(&repo->tagcache);
843 	got_object_cache_close(&repo->rawcache);
844 
845 	for (i = 0; i < nitems(repo->privsep_children); i++) {
846 		if (repo->privsep_children[i].imsg_fd == -1)
847 			continue;
848 		imsg_clear(repo->privsep_children[i].ibuf);
849 		free(repo->privsep_children[i].ibuf);
850 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
851 		child_err = got_privsep_wait_for_child(
852 		    repo->privsep_children[i].pid);
853 		if (child_err && err == NULL)
854 			err = child_err;
855 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
856 		    err == NULL)
857 			err = got_error_from_errno("close");
858 	}
859 
860 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
861 	    err == NULL)
862 		err = got_error_from_errno("close");
863 
864 	if (repo->gotconfig)
865 		got_gotconfig_free(repo->gotconfig);
866 	free(repo->gitconfig_author_name);
867 	free(repo->gitconfig_author_email);
868 	for (i = 0; i < repo->ngitconfig_remotes; i++)
869 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
870 	free(repo->gitconfig_remotes);
871 	for (i = 0; i < repo->nextensions; i++)
872 		free(repo->extensions[i]);
873 	free(repo->extensions);
874 
875 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
876 		free((void *)pe->path);
877 	got_pathlist_free(&repo->packidx_paths);
878 	free(repo);
879 
880 	return err;
881 }
882 
883 void
884 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
885 {
886 	int i;
887 
888 	free(repo->name);
889 	repo->name = NULL;
890 	free(repo->fetch_url);
891 	repo->fetch_url = NULL;
892 	free(repo->send_url);
893 	repo->send_url = NULL;
894 	for (i = 0; i < repo->nfetch_branches; i++)
895 		free(repo->fetch_branches[i]);
896 	free(repo->fetch_branches);
897 	repo->fetch_branches = NULL;
898 	repo->nfetch_branches = 0;
899 	for (i = 0; i < repo->nsend_branches; i++)
900 		free(repo->send_branches[i]);
901 	free(repo->send_branches);
902 	repo->send_branches = NULL;
903 	repo->nsend_branches = 0;
904 }
905 
906 const struct got_error *
907 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
908     const char *input_path)
909 {
910 	const struct got_error *err = NULL;
911 	const char *repo_abspath = NULL;
912 	size_t repolen, len;
913 	char *canonpath, *path = NULL;
914 
915 	*in_repo_path = NULL;
916 
917 	canonpath = strdup(input_path);
918 	if (canonpath == NULL) {
919 		err = got_error_from_errno("strdup");
920 		goto done;
921 	}
922 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
923 	if (err)
924 		goto done;
925 
926 	repo_abspath = got_repo_get_path(repo);
927 
928 	if (canonpath[0] == '\0') {
929 		path = strdup(canonpath);
930 		if (path == NULL) {
931 			err = got_error_from_errno("strdup");
932 			goto done;
933 		}
934 	} else {
935 		path = realpath(canonpath, NULL);
936 		if (path == NULL) {
937 			if (errno != ENOENT) {
938 				err = got_error_from_errno2("realpath",
939 				    canonpath);
940 				goto done;
941 			}
942 			/*
943 			 * Path is not on disk.
944 			 * Assume it is already relative to repository root.
945 			 */
946 			path = strdup(canonpath);
947 			if (path == NULL) {
948 				err = got_error_from_errno("strdup");
949 				goto done;
950 			}
951 		}
952 
953 		repolen = strlen(repo_abspath);
954 		len = strlen(path);
955 
956 
957 		if (strcmp(path, repo_abspath) == 0) {
958 			free(path);
959 			path = strdup("");
960 			if (path == NULL) {
961 				err = got_error_from_errno("strdup");
962 				goto done;
963 			}
964 		} else if (len > repolen &&
965 		    got_path_is_child(path, repo_abspath, repolen)) {
966 			/* Matched an on-disk path inside repository. */
967 			if (got_repo_is_bare(repo)) {
968 				/*
969 				 * Matched an on-disk path inside repository
970 				 * database. Treat input as repository-relative.
971 				 */
972 				free(path);
973 				path = canonpath;
974 				canonpath = NULL;
975 			} else {
976 				char *child;
977 				/* Strip common prefix with repository path. */
978 				err = got_path_skip_common_ancestor(&child,
979 				    repo_abspath, path);
980 				if (err)
981 					goto done;
982 				free(path);
983 				path = child;
984 			}
985 		} else {
986 			/*
987 			 * Matched unrelated on-disk path.
988 			 * Treat input as repository-relative.
989 			 */
990 			free(path);
991 			path = canonpath;
992 			canonpath = NULL;
993 		}
994 	}
995 
996 	/* Make in-repository path absolute */
997 	if (path[0] != '/') {
998 		char *abspath;
999 		if (asprintf(&abspath, "/%s", path) == -1) {
1000 			err = got_error_from_errno("asprintf");
1001 			goto done;
1002 		}
1003 		free(path);
1004 		path = abspath;
1005 	}
1006 
1007 done:
1008 	free(canonpath);
1009 	if (err)
1010 		free(path);
1011 	else
1012 		*in_repo_path = path;
1013 	return err;
1014 }
1015 
1016 static const struct got_error *
1017 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1018     const char *path_packidx)
1019 {
1020 	const struct got_error *err = NULL;
1021 	size_t i;
1022 
1023 	for (i = 0; i < repo->pack_cache_size; i++) {
1024 		if (repo->packidx_cache[i] == NULL)
1025 			break;
1026 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1027 		    path_packidx) == 0) {
1028 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1029 		}
1030 	}
1031 	if (i == repo->pack_cache_size) {
1032 		i = repo->pack_cache_size - 1;
1033 		err = got_packidx_close(repo->packidx_cache[i]);
1034 		if (err)
1035 			return err;
1036 	}
1037 
1038 	repo->packidx_cache[i] = packidx;
1039 
1040 	return NULL;
1041 }
1042 
1043 int
1044 got_repo_is_packidx_filename(const char *name, size_t len)
1045 {
1046 	if (len != GOT_PACKIDX_NAMELEN)
1047 		return 0;
1048 
1049 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1050 		return 0;
1051 
1052 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1053 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1054 		return 0;
1055 
1056 	return 1;
1057 }
1058 
1059 static struct got_packidx_bloom_filter *
1060 get_packidx_bloom_filter(struct got_repository *repo,
1061     const char *path, size_t path_len)
1062 {
1063 	struct got_packidx_bloom_filter key;
1064 
1065 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1066 		return NULL; /* XXX */
1067 	key.path_len = path_len;
1068 
1069 	return RB_FIND(got_packidx_bloom_filter_tree,
1070 	    &repo->packidx_bloom_filters, &key);
1071 }
1072 
1073 int
1074 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1075     const char *path_packidx, struct got_object_id *id)
1076 {
1077 	struct got_packidx_bloom_filter *bf;
1078 
1079 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1080 	if (bf)
1081 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1082 
1083 	/* No bloom filter means this pack index must be searched. */
1084 	return 1;
1085 }
1086 
1087 static const struct got_error *
1088 add_packidx_bloom_filter(struct got_repository *repo,
1089     struct got_packidx *packidx, const char *path_packidx)
1090 {
1091 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1092 	struct got_packidx_bloom_filter *bf;
1093 	size_t len;
1094 
1095 	/*
1096 	 * Don't use bloom filters for very large pack index files.
1097 	 * Large pack files will contain a relatively large fraction
1098 	 * of our objects so we will likely need to visit them anyway.
1099 	 * The more objects a pack file contains the higher the probability
1100 	 * of a false-positive match from the bloom filter. And reading
1101 	 * all object IDs from a large pack index file can be expensive.
1102 	 */
1103 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1104 		return NULL;
1105 
1106 	/* Do we already have a filter for this pack index? */
1107 	if (get_packidx_bloom_filter(repo, path_packidx,
1108 	    strlen(path_packidx)) != NULL)
1109 		return NULL;
1110 
1111 	bf = calloc(1, sizeof(*bf));
1112 	if (bf == NULL)
1113 		return got_error_from_errno("calloc");
1114 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1115 	if (bf->bloom == NULL) {
1116 		free(bf);
1117 		return got_error_from_errno("calloc");
1118 	}
1119 	
1120 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1121 	if (len >= sizeof(bf->path)) {
1122 		free(bf->bloom);
1123 		free(bf);
1124 		return got_error(GOT_ERR_NO_SPACE);
1125 	}
1126 	bf->path_len = len;
1127 
1128 	/* Minimum size supported by our bloom filter is 1000 entries. */
1129 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1130 	for (i = 0; i < nobjects; i++) {
1131 		struct got_packidx_object_id *id;
1132 		id = &packidx->hdr.sorted_ids[i];
1133 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1134 	}
1135 
1136 	RB_INSERT(got_packidx_bloom_filter_tree,
1137 	    &repo->packidx_bloom_filters, bf);
1138 	return NULL;
1139 }
1140 
1141 const struct got_error *
1142 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1143     struct got_repository *repo, struct got_object_id *id)
1144 {
1145 	const struct got_error *err;
1146 	struct got_pathlist_entry *pe;
1147 	size_t i;
1148 
1149 	/* Search pack index cache. */
1150 	for (i = 0; i < repo->pack_cache_size; i++) {
1151 		if (repo->packidx_cache[i] == NULL)
1152 			break;
1153 		if (!got_repo_check_packidx_bloom_filter(repo,
1154 		    repo->packidx_cache[i]->path_packidx, id))
1155 			continue; /* object will not be found in this index */
1156 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1157 		if (*idx != -1) {
1158 			*packidx = repo->packidx_cache[i];
1159 			/*
1160 			 * Move this cache entry to the front. Repeatedly
1161 			 * searching a wrong pack index can be expensive.
1162 			 */
1163 			if (i > 0) {
1164 				memmove(&repo->packidx_cache[1],
1165 				    &repo->packidx_cache[0],
1166 				    i * sizeof(repo->packidx_cache[0]));
1167 				repo->packidx_cache[0] = *packidx;
1168 			}
1169 			return NULL;
1170 		}
1171 	}
1172 	/* No luck. Search the filesystem. */
1173 
1174 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1175 		const char *path_packidx = pe->path;
1176 		int is_cached = 0;
1177 
1178 		if (!got_repo_check_packidx_bloom_filter(repo,
1179 		    pe->path, id))
1180 			continue; /* object will not be found in this index */
1181 
1182 		for (i = 0; i < repo->pack_cache_size; i++) {
1183 			if (repo->packidx_cache[i] == NULL)
1184 				break;
1185 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1186 			    path_packidx) == 0) {
1187 				is_cached = 1;
1188 				break;
1189 			}
1190 		}
1191 		if (is_cached)
1192 			continue; /* already searched */
1193 
1194 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1195 		    path_packidx, 0);
1196 		if (err)
1197 			goto done;
1198 
1199 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1200 		if (err)
1201 			goto done;
1202 
1203 		err = cache_packidx(repo, *packidx, path_packidx);
1204 		if (err)
1205 			goto done;
1206 
1207 		*idx = got_packidx_get_object_idx(*packidx, id);
1208 		if (*idx != -1) {
1209 			err = NULL; /* found the object */
1210 			goto done;
1211 		}
1212 	}
1213 
1214 	err = got_error_no_obj(id);
1215 done:
1216 	return err;
1217 }
1218 
1219 const struct got_error *
1220 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1221     struct got_repository *repo)
1222 {
1223 	const struct got_error *err = NULL;
1224 	DIR *packdir = NULL;
1225 	struct dirent *dent;
1226 	char *path_packidx = NULL;
1227 	int packdir_fd;
1228 
1229 	packdir_fd = openat(got_repo_get_fd(repo),
1230 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1231 	if (packdir_fd == -1) {
1232 		return got_error_from_errno_fmt("openat: %s/%s",
1233 		    got_repo_get_path_git_dir(repo),
1234 		    GOT_OBJECTS_PACK_DIR);
1235 	}
1236 
1237 	packdir = fdopendir(packdir_fd);
1238 	if (packdir == NULL) {
1239 		err = got_error_from_errno("fdopendir");
1240 		goto done;
1241 	}
1242 
1243 	while ((dent = readdir(packdir)) != NULL) {
1244 		if (!got_repo_is_packidx_filename(dent->d_name,
1245 		    strlen(dent->d_name)))
1246 			continue;
1247 
1248 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1249 		    dent->d_name) == -1) {
1250 			err = got_error_from_errno("asprintf");
1251 			path_packidx = NULL;
1252 			break;
1253 		}
1254 
1255 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1256 		if (err)
1257 			break;
1258 	}
1259 done:
1260 	if (err)
1261 		free(path_packidx);
1262 	if (packdir && closedir(packdir) != 0 && err == NULL)
1263 		err = got_error_from_errno("closedir");
1264 	return err;
1265 }
1266 
1267 const struct got_error *
1268 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1269     struct got_repository *repo)
1270 {
1271 	const struct got_error *err;
1272 	size_t i;
1273 
1274 	*packidx = NULL;
1275 
1276 	/* Search pack index cache. */
1277 	for (i = 0; i < repo->pack_cache_size; i++) {
1278 		if (repo->packidx_cache[i] == NULL)
1279 			break;
1280 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1281 		    path_packidx) == 0) {
1282 			*packidx = repo->packidx_cache[i];
1283 			return NULL;
1284 		}
1285 	}
1286 	/* No luck. Search the filesystem. */
1287 
1288 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1289 	    path_packidx, 0);
1290 	if (err)
1291 		return err;
1292 
1293 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1294 	if (err)
1295 		goto done;
1296 
1297 	err = cache_packidx(repo, *packidx, path_packidx);
1298 done:
1299 	if (err) {
1300 		got_packidx_close(*packidx);
1301 		*packidx = NULL;
1302 	}
1303 	return err;
1304 }
1305 
1306 static const struct got_error *
1307 read_packfile_hdr(int fd, struct got_packidx *packidx)
1308 {
1309 	const struct got_error *err = NULL;
1310 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1311 	struct got_packfile_hdr hdr;
1312 	ssize_t n;
1313 
1314 	n = read(fd, &hdr, sizeof(hdr));
1315 	if (n < 0)
1316 		return got_error_from_errno("read");
1317 	if (n != sizeof(hdr))
1318 		return got_error(GOT_ERR_BAD_PACKFILE);
1319 
1320 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1321 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1322 	    be32toh(hdr.nobjects) != totobj)
1323 		err = got_error(GOT_ERR_BAD_PACKFILE);
1324 
1325 	return err;
1326 }
1327 
1328 static const struct got_error *
1329 open_packfile(int *fd, struct got_repository *repo,
1330     const char *relpath, struct got_packidx *packidx)
1331 {
1332 	const struct got_error *err = NULL;
1333 
1334 	*fd = openat(got_repo_get_fd(repo), relpath,
1335 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1336 	if (*fd == -1)
1337 		return got_error_from_errno_fmt("openat: %s/%s",
1338 		    got_repo_get_path_git_dir(repo), relpath);
1339 
1340 	if (packidx) {
1341 		err = read_packfile_hdr(*fd, packidx);
1342 		if (err) {
1343 			close(*fd);
1344 			*fd = -1;
1345 		}
1346 	}
1347 
1348 	return err;
1349 }
1350 
1351 const struct got_error *
1352 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1353     const char *path_packfile, struct got_packidx *packidx)
1354 {
1355 	const struct got_error *err = NULL;
1356 	struct got_pack *pack = NULL;
1357 	struct stat sb;
1358 	size_t i;
1359 
1360 	if (packp)
1361 		*packp = NULL;
1362 
1363 	for (i = 0; i < repo->pack_cache_size; i++) {
1364 		pack = &repo->packs[i];
1365 		if (pack->path_packfile == NULL)
1366 			break;
1367 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1368 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1369 	}
1370 
1371 	if (i == repo->pack_cache_size) {
1372 		struct got_pack tmp;
1373 		err = got_pack_close(&repo->packs[i - 1]);
1374 		if (err)
1375 			return err;
1376 		if (ftruncate(repo->packs[i - 1].basefd, 0L) == -1)
1377 			return got_error_from_errno("ftruncate");
1378 		if (ftruncate(repo->packs[i - 1].accumfd, 0L) == -1)
1379 			return got_error_from_errno("ftruncate");
1380 		memcpy(&tmp, &repo->packs[i - 1], sizeof(tmp)); 
1381 		memcpy(&repo->packs[i - 1], &repo->packs[0],
1382 		    sizeof(repo->packs[i - 1]));
1383 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1384 		i = 0;
1385 	}
1386 
1387 	pack = &repo->packs[i];
1388 
1389 	pack->path_packfile = strdup(path_packfile);
1390 	if (pack->path_packfile == NULL) {
1391 		err = got_error_from_errno("strdup");
1392 		goto done;
1393 	}
1394 
1395 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1396 	if (err)
1397 		goto done;
1398 
1399 	if (fstat(pack->fd, &sb) != 0) {
1400 		err = got_error_from_errno("fstat");
1401 		goto done;
1402 	}
1403 	pack->filesize = sb.st_size;
1404 
1405 	pack->privsep_child = NULL;
1406 
1407 #ifndef GOT_PACK_NO_MMAP
1408 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1409 	    pack->fd, 0);
1410 	if (pack->map == MAP_FAILED) {
1411 		if (errno != ENOMEM) {
1412 			err = got_error_from_errno("mmap");
1413 			goto done;
1414 		}
1415 		pack->map = NULL; /* fall back to read(2) */
1416 	}
1417 #endif
1418 done:
1419 	if (err) {
1420 		if (pack) {
1421 			free(pack->path_packfile);
1422 			memset(pack, 0, sizeof(*pack));
1423 		}
1424 	} else if (packp)
1425 		*packp = pack;
1426 	return err;
1427 }
1428 
1429 struct got_pack *
1430 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1431 {
1432 	struct got_pack *pack = NULL;
1433 	size_t i;
1434 
1435 	for (i = 0; i < repo->pack_cache_size; i++) {
1436 		pack = &repo->packs[i];
1437 		if (pack->path_packfile == NULL)
1438 			break;
1439 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1440 			return pack;
1441 	}
1442 
1443 	return NULL;
1444 }
1445 
1446 const struct got_error *
1447 got_repo_init(const char *repo_path)
1448 {
1449 	const struct got_error *err = NULL;
1450 	const char *dirnames[] = {
1451 		GOT_OBJECTS_DIR,
1452 		GOT_OBJECTS_PACK_DIR,
1453 		GOT_REFS_DIR,
1454 	};
1455 	const char *description_str = "Unnamed repository; "
1456 	    "edit this file 'description' to name the repository.";
1457 	const char *headref_str = "ref: refs/heads/main";
1458 	const char *gitconfig_str = "[core]\n"
1459 	    "\trepositoryformatversion = 0\n"
1460 	    "\tfilemode = true\n"
1461 	    "\tbare = true\n";
1462 	char *path;
1463 	size_t i;
1464 
1465 	if (!got_path_dir_is_empty(repo_path))
1466 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1467 
1468 	for (i = 0; i < nitems(dirnames); i++) {
1469 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1470 			return got_error_from_errno("asprintf");
1471 		}
1472 		err = got_path_mkdir(path);
1473 		free(path);
1474 		if (err)
1475 			return err;
1476 	}
1477 
1478 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1479 		return got_error_from_errno("asprintf");
1480 	err = got_path_create_file(path, description_str);
1481 	free(path);
1482 	if (err)
1483 		return err;
1484 
1485 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1486 		return got_error_from_errno("asprintf");
1487 	err = got_path_create_file(path, headref_str);
1488 	free(path);
1489 	if (err)
1490 		return err;
1491 
1492 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1493 		return got_error_from_errno("asprintf");
1494 	err = got_path_create_file(path, gitconfig_str);
1495 	free(path);
1496 	if (err)
1497 		return err;
1498 
1499 	return NULL;
1500 }
1501 
1502 static const struct got_error *
1503 match_packed_object(struct got_object_id **unique_id,
1504     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1505 {
1506 	const struct got_error *err = NULL;
1507 	struct got_object_id_queue matched_ids;
1508 	struct got_pathlist_entry *pe;
1509 
1510 	STAILQ_INIT(&matched_ids);
1511 
1512 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1513 		const char *path_packidx = pe->path;
1514 		struct got_packidx *packidx;
1515 		struct got_object_qid *qid;
1516 
1517 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1518 		    path_packidx, 0);
1519 		if (err)
1520 			break;
1521 
1522 		err = got_packidx_match_id_str_prefix(&matched_ids,
1523 		    packidx, id_str_prefix);
1524 		if (err) {
1525 			got_packidx_close(packidx);
1526 			break;
1527 		}
1528 		err = got_packidx_close(packidx);
1529 		if (err)
1530 			break;
1531 
1532 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1533 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1534 				int matched_type;
1535 				err = got_object_get_type(&matched_type, repo,
1536 				    &qid->id);
1537 				if (err)
1538 					goto done;
1539 				if (matched_type != obj_type)
1540 					continue;
1541 			}
1542 			if (*unique_id == NULL) {
1543 				*unique_id = got_object_id_dup(&qid->id);
1544 				if (*unique_id == NULL) {
1545 					err = got_error_from_errno("malloc");
1546 					goto done;
1547 				}
1548 			} else {
1549 				if (got_object_id_cmp(*unique_id,
1550 				    &qid->id) == 0)
1551 					continue; /* packed multiple times */
1552 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1553 				goto done;
1554 			}
1555 		}
1556 	}
1557 done:
1558 	got_object_id_queue_free(&matched_ids);
1559 	if (err) {
1560 		free(*unique_id);
1561 		*unique_id = NULL;
1562 	}
1563 	return err;
1564 }
1565 
1566 static const struct got_error *
1567 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1568     const char *object_dir, const char *id_str_prefix, int obj_type,
1569     struct got_repository *repo)
1570 {
1571 	const struct got_error *err = NULL;
1572 	char *path;
1573 	DIR *dir = NULL;
1574 	struct dirent *dent;
1575 	struct got_object_id id;
1576 
1577 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1578 		err = got_error_from_errno("asprintf");
1579 		goto done;
1580 	}
1581 
1582 	dir = opendir(path);
1583 	if (dir == NULL) {
1584 		if (errno == ENOENT) {
1585 			err = NULL;
1586 			goto done;
1587 		}
1588 		err = got_error_from_errno2("opendir", path);
1589 		goto done;
1590 	}
1591 	while ((dent = readdir(dir)) != NULL) {
1592 		char *id_str;
1593 		int cmp;
1594 
1595 		if (strcmp(dent->d_name, ".") == 0 ||
1596 		    strcmp(dent->d_name, "..") == 0)
1597 			continue;
1598 
1599 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1600 			err = got_error_from_errno("asprintf");
1601 			goto done;
1602 		}
1603 
1604 		if (!got_parse_sha1_digest(id.sha1, id_str))
1605 			continue;
1606 
1607 		/*
1608 		 * Directory entries do not necessarily appear in
1609 		 * sorted order, so we must iterate over all of them.
1610 		 */
1611 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1612 		if (cmp != 0) {
1613 			free(id_str);
1614 			continue;
1615 		}
1616 
1617 		if (*unique_id == NULL) {
1618 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1619 				int matched_type;
1620 				err = got_object_get_type(&matched_type, repo,
1621 				    &id);
1622 				if (err)
1623 					goto done;
1624 				if (matched_type != obj_type)
1625 					continue;
1626 			}
1627 			*unique_id = got_object_id_dup(&id);
1628 			if (*unique_id == NULL) {
1629 				err = got_error_from_errno("got_object_id_dup");
1630 				free(id_str);
1631 				goto done;
1632 			}
1633 		} else {
1634 			if (got_object_id_cmp(*unique_id, &id) == 0)
1635 				continue; /* both packed and loose */
1636 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1637 			free(id_str);
1638 			goto done;
1639 		}
1640 	}
1641 done:
1642 	if (dir && closedir(dir) != 0 && err == NULL)
1643 		err = got_error_from_errno("closedir");
1644 	if (err) {
1645 		free(*unique_id);
1646 		*unique_id = NULL;
1647 	}
1648 	free(path);
1649 	return err;
1650 }
1651 
1652 const struct got_error *
1653 got_repo_match_object_id_prefix(struct got_object_id **id,
1654     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1655 {
1656 	const struct got_error *err = NULL;
1657 	char *path_objects = got_repo_get_path_objects(repo);
1658 	char *object_dir = NULL;
1659 	size_t len;
1660 	int i;
1661 
1662 	*id = NULL;
1663 
1664 	len = strlen(id_str_prefix);
1665 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1666 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1667 
1668 	for (i = 0; i < len; i++) {
1669 		if (isxdigit((unsigned char)id_str_prefix[i]))
1670 			continue;
1671 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1672 	}
1673 
1674 	if (len >= 2) {
1675 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1676 		if (err)
1677 			goto done;
1678 		object_dir = strndup(id_str_prefix, 2);
1679 		if (object_dir == NULL) {
1680 			err = got_error_from_errno("strdup");
1681 			goto done;
1682 		}
1683 		err = match_loose_object(id, path_objects, object_dir,
1684 		    id_str_prefix, obj_type, repo);
1685 	} else if (len == 1) {
1686 		int i;
1687 		for (i = 0; i < 0xf; i++) {
1688 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1689 			    == -1) {
1690 				err = got_error_from_errno("asprintf");
1691 				goto done;
1692 			}
1693 			err = match_packed_object(id, repo, object_dir,
1694 			    obj_type);
1695 			if (err)
1696 				goto done;
1697 			err = match_loose_object(id, path_objects, object_dir,
1698 			    id_str_prefix, obj_type, repo);
1699 			if (err)
1700 				goto done;
1701 		}
1702 	} else {
1703 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1704 		goto done;
1705 	}
1706 done:
1707 	free(object_dir);
1708 	if (err) {
1709 		free(*id);
1710 		*id = NULL;
1711 	} else if (*id == NULL) {
1712 		switch (obj_type) {
1713 		case GOT_OBJ_TYPE_BLOB:
1714 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1715 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1716 			break;
1717 		case GOT_OBJ_TYPE_TREE:
1718 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1719 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1720 			break;
1721 		case GOT_OBJ_TYPE_COMMIT:
1722 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1723 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1724 			break;
1725 		case GOT_OBJ_TYPE_TAG:
1726 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1727 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1728 			break;
1729 		default:
1730 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1731 			break;
1732 		}
1733 	}
1734 
1735 	return err;
1736 }
1737 
1738 const struct got_error *
1739 got_repo_match_object_id(struct got_object_id **id, char **label,
1740     const char *id_str, int obj_type, struct got_reflist_head *refs,
1741     struct got_repository *repo)
1742 {
1743 	const struct got_error *err;
1744 	struct got_tag_object *tag;
1745 	struct got_reference *ref = NULL;
1746 
1747 	*id = NULL;
1748 	if (label)
1749 		*label = NULL;
1750 
1751 	if (refs) {
1752 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1753 		    refs, repo);
1754 		if (err == NULL) {
1755 			*id = got_object_id_dup(
1756 			    got_object_tag_get_object_id(tag));
1757 			if (*id == NULL)
1758 				err = got_error_from_errno("got_object_id_dup");
1759 			else if (label && asprintf(label, "refs/tags/%s",
1760 			    got_object_tag_get_name(tag)) == -1) {
1761 				err = got_error_from_errno("asprintf");
1762 				free(*id);
1763 				*id = NULL;
1764 			}
1765 			got_object_tag_close(tag);
1766 			return err;
1767 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1768 		    err->code != GOT_ERR_NO_OBJ)
1769 			return err;
1770 	}
1771 
1772 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1773 	if (err) {
1774 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1775 			return err;
1776 		err = got_ref_open(&ref, repo, id_str, 0);
1777 		if (err != NULL)
1778 			goto done;
1779 		if (label) {
1780 			*label = strdup(got_ref_get_name(ref));
1781 			if (*label == NULL) {
1782 				err = got_error_from_errno("strdup");
1783 				goto done;
1784 			}
1785 		}
1786 		err = got_ref_resolve(id, repo, ref);
1787 	} else if (label) {
1788 		err = got_object_id_str(label, *id);
1789 		if (*label == NULL) {
1790 			err = got_error_from_errno("strdup");
1791 			goto done;
1792 		}
1793 	}
1794 done:
1795 	if (ref)
1796 		got_ref_close(ref);
1797 	return err;
1798 }
1799 
1800 const struct got_error *
1801 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1802     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1803 {
1804 	const struct got_error *err = NULL;
1805 	struct got_reflist_entry *re;
1806 	struct got_object_id *tag_id;
1807 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1808 
1809 	*tag = NULL;
1810 
1811 	TAILQ_FOREACH(re, refs, entry) {
1812 		const char *refname;
1813 		refname = got_ref_get_name(re->ref);
1814 		if (got_ref_is_symbolic(re->ref))
1815 			continue;
1816 		if (strncmp(refname, "refs/tags/", 10) != 0)
1817 			continue;
1818 		if (!name_is_absolute)
1819 			refname += strlen("refs/tags/");
1820 		if (strcmp(refname, name) != 0)
1821 			continue;
1822 		err = got_ref_resolve(&tag_id, repo, re->ref);
1823 		if (err)
1824 			break;
1825 		err = got_object_open_as_tag(tag, repo, tag_id);
1826 		free(tag_id);
1827 		if (err)
1828 			break;
1829 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1830 		    got_object_tag_get_object_type(*tag) == obj_type)
1831 			break;
1832 		got_object_tag_close(*tag);
1833 		*tag = NULL;
1834 	}
1835 
1836 	if (err == NULL && *tag == NULL)
1837 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1838 		    GOT_OBJ_LABEL_TAG, name);
1839 	return err;
1840 }
1841 
1842 static const struct got_error *
1843 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1844     const char *name, mode_t mode, struct got_object_id *blob_id)
1845 {
1846 	const struct got_error *err = NULL;
1847 
1848 	 *new_te = NULL;
1849 
1850 	*new_te = calloc(1, sizeof(**new_te));
1851 	if (*new_te == NULL)
1852 		return got_error_from_errno("calloc");
1853 
1854 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1855 	    sizeof((*new_te)->name)) {
1856 		err = got_error(GOT_ERR_NO_SPACE);
1857 		goto done;
1858 	}
1859 
1860 	if (S_ISLNK(mode)) {
1861 		(*new_te)->mode = S_IFLNK;
1862 	} else {
1863 		(*new_te)->mode = S_IFREG;
1864 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1865 	}
1866 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1867 done:
1868 	if (err && *new_te) {
1869 		free(*new_te);
1870 		*new_te = NULL;
1871 	}
1872 	return err;
1873 }
1874 
1875 static const struct got_error *
1876 import_file(struct got_tree_entry **new_te, struct dirent *de,
1877     const char *path, struct got_repository *repo)
1878 {
1879 	const struct got_error *err;
1880 	struct got_object_id *blob_id = NULL;
1881 	char *filepath;
1882 	struct stat sb;
1883 
1884 	if (asprintf(&filepath, "%s%s%s", path,
1885 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1886 		return got_error_from_errno("asprintf");
1887 
1888 	if (lstat(filepath, &sb) != 0) {
1889 		err = got_error_from_errno2("lstat", path);
1890 		goto done;
1891 	}
1892 
1893 	err = got_object_blob_create(&blob_id, filepath, repo);
1894 	if (err)
1895 		goto done;
1896 
1897 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1898 	    blob_id);
1899 done:
1900 	free(filepath);
1901 	if (err)
1902 		free(blob_id);
1903 	return err;
1904 }
1905 
1906 static const struct got_error *
1907 insert_tree_entry(struct got_tree_entry *new_te,
1908     struct got_pathlist_head *paths)
1909 {
1910 	const struct got_error *err = NULL;
1911 	struct got_pathlist_entry *new_pe;
1912 
1913 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1914 	if (err)
1915 		return err;
1916 	if (new_pe == NULL)
1917 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1918 	return NULL;
1919 }
1920 
1921 static const struct got_error *write_tree(struct got_object_id **,
1922     const char *, struct got_pathlist_head *, struct got_repository *,
1923     got_repo_import_cb progress_cb, void *progress_arg);
1924 
1925 static const struct got_error *
1926 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1927     const char *path, struct got_pathlist_head *ignores,
1928     struct got_repository *repo,
1929     got_repo_import_cb progress_cb, void *progress_arg)
1930 {
1931 	const struct got_error *err;
1932 	struct got_object_id *id = NULL;
1933 	char *subdirpath;
1934 
1935 	if (asprintf(&subdirpath, "%s%s%s", path,
1936 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1937 		return got_error_from_errno("asprintf");
1938 
1939 	(*new_te) = calloc(1, sizeof(**new_te));
1940 	if (*new_te == NULL)
1941 		return got_error_from_errno("calloc");
1942 	(*new_te)->mode = S_IFDIR;
1943 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1944 	    sizeof((*new_te)->name)) {
1945 		err = got_error(GOT_ERR_NO_SPACE);
1946 		goto done;
1947 	}
1948 	err = write_tree(&id, subdirpath, ignores,  repo,
1949 	    progress_cb, progress_arg);
1950 	if (err)
1951 		goto done;
1952 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1953 	
1954 done:
1955 	free(id);
1956 	free(subdirpath);
1957 	if (err) {
1958 		free(*new_te);
1959 		*new_te = NULL;
1960 	}
1961 	return err;
1962 }
1963 
1964 static const struct got_error *
1965 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1966     struct got_pathlist_head *ignores, struct got_repository *repo,
1967     got_repo_import_cb progress_cb, void *progress_arg)
1968 {
1969 	const struct got_error *err = NULL;
1970 	DIR *dir;
1971 	struct dirent *de;
1972 	int nentries;
1973 	struct got_tree_entry *new_te = NULL;
1974 	struct got_pathlist_head paths;
1975 	struct got_pathlist_entry *pe;
1976 
1977 	*new_tree_id = NULL;
1978 
1979 	TAILQ_INIT(&paths);
1980 
1981 	dir = opendir(path_dir);
1982 	if (dir == NULL) {
1983 		err = got_error_from_errno2("opendir", path_dir);
1984 		goto done;
1985 	}
1986 
1987 	nentries = 0;
1988 	while ((de = readdir(dir)) != NULL) {
1989 		int ignore = 0;
1990 		int type;
1991 
1992 		if (strcmp(de->d_name, ".") == 0 ||
1993 		    strcmp(de->d_name, "..") == 0)
1994 			continue;
1995 
1996 		TAILQ_FOREACH(pe, ignores, entry) {
1997 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1998 				ignore = 1;
1999 				break;
2000 			}
2001 		}
2002 		if (ignore)
2003 			continue;
2004 
2005 		err = got_path_dirent_type(&type, path_dir, de);
2006 		if (err)
2007 			goto done;
2008 
2009 		if (type == DT_DIR) {
2010 			err = import_subdir(&new_te, de, path_dir,
2011 			    ignores, repo, progress_cb, progress_arg);
2012 			if (err) {
2013 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2014 					goto done;
2015 				err = NULL;
2016 				continue;
2017 			}
2018 		} else if (type == DT_REG || type == DT_LNK) {
2019 			err = import_file(&new_te, de, path_dir, repo);
2020 			if (err)
2021 				goto done;
2022 		} else
2023 			continue;
2024 
2025 		err = insert_tree_entry(new_te, &paths);
2026 		if (err)
2027 			goto done;
2028 		nentries++;
2029 	}
2030 
2031 	if (TAILQ_EMPTY(&paths)) {
2032 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2033 		    "cannot create tree without any entries");
2034 		goto done;
2035 	}
2036 
2037 	TAILQ_FOREACH(pe, &paths, entry) {
2038 		struct got_tree_entry *te = pe->data;
2039 		char *path;
2040 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2041 			continue;
2042 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2043 			err = got_error_from_errno("asprintf");
2044 			goto done;
2045 		}
2046 		err = (*progress_cb)(progress_arg, path);
2047 		free(path);
2048 		if (err)
2049 			goto done;
2050 	}
2051 
2052 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2053 done:
2054 	if (dir)
2055 		closedir(dir);
2056 	got_pathlist_free(&paths);
2057 	return err;
2058 }
2059 
2060 const struct got_error *
2061 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2062     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2063     struct got_repository *repo, got_repo_import_cb progress_cb,
2064     void *progress_arg)
2065 {
2066 	const struct got_error *err;
2067 	struct got_object_id *new_tree_id;
2068 
2069 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2070 	    progress_cb, progress_arg);
2071 	if (err)
2072 		return err;
2073 
2074 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2075 	    author, time(NULL), author, time(NULL), logmsg, repo);
2076 	free(new_tree_id);
2077 	return err;
2078 }
2079 
2080 const struct got_error *
2081 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2082     struct got_repository *repo)
2083 {
2084 	const struct got_error *err = NULL;
2085 	char *path_objects = NULL, *path = NULL;
2086 	DIR *dir = NULL;
2087 	struct got_object_id id;
2088 	int i;
2089 
2090 	*nobjects = 0;
2091 	*ondisk_size = 0;
2092 
2093 	path_objects = got_repo_get_path_objects(repo);
2094 	if (path_objects == NULL)
2095 		return got_error_from_errno("got_repo_get_path_objects");
2096 
2097 	for (i = 0; i <= 0xff; i++) {
2098 		struct dirent *dent;
2099 
2100 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2101 			err = got_error_from_errno("asprintf");
2102 			break;
2103 		}
2104 
2105 		dir = opendir(path);
2106 		if (dir == NULL) {
2107 			if (errno == ENOENT) {
2108 				err = NULL;
2109 				continue;
2110 			}
2111 			err = got_error_from_errno2("opendir", path);
2112 			break;
2113 		}
2114 
2115 		while ((dent = readdir(dir)) != NULL) {
2116 			char *id_str;
2117 			int fd;
2118 			struct stat sb;
2119 
2120 			if (strcmp(dent->d_name, ".") == 0 ||
2121 			    strcmp(dent->d_name, "..") == 0)
2122 				continue;
2123 
2124 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2125 				err = got_error_from_errno("asprintf");
2126 				goto done;
2127 			}
2128 
2129 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2130 				free(id_str);
2131 				continue;
2132 			}
2133 			free(id_str);
2134 
2135 			err = got_object_open_loose_fd(&fd, &id, repo);
2136 			if (err)
2137 				goto done;
2138 
2139 			if (fstat(fd, &sb) == -1) {
2140 				err = got_error_from_errno("fstat");
2141 				close(fd);
2142 				goto done;
2143 			}
2144 			(*nobjects)++;
2145 			(*ondisk_size) += sb.st_size;
2146 
2147 			if (close(fd) == -1) {
2148 				err = got_error_from_errno("close");
2149 				goto done;
2150 			}
2151 		}
2152 
2153 		if (closedir(dir) != 0) {
2154 			err = got_error_from_errno("closedir");
2155 			goto done;
2156 		}
2157 		dir = NULL;
2158 
2159 		free(path);
2160 		path = NULL;
2161 	}
2162 done:
2163 	if (dir && closedir(dir) != 0 && err == NULL)
2164 		err = got_error_from_errno("closedir");
2165 
2166 	if (err) {
2167 		*nobjects = 0;
2168 		*ondisk_size = 0;
2169 	}
2170 	free(path_objects);
2171 	free(path);
2172 	return err;
2173 }
2174 
2175 const struct got_error *
2176 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2177     off_t *total_packsize, struct got_repository *repo)
2178 {
2179 	const struct got_error *err = NULL;
2180 	DIR *packdir = NULL;
2181 	struct dirent *dent;
2182 	struct got_packidx *packidx = NULL;
2183 	char *path_packidx;
2184 	char *path_packfile;
2185 	int packdir_fd;
2186 	struct stat sb;
2187 
2188 	*npackfiles = 0;
2189 	*nobjects = 0;
2190 	*total_packsize = 0;
2191 
2192 	packdir_fd = openat(got_repo_get_fd(repo),
2193 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2194 	if (packdir_fd == -1) {
2195 		return got_error_from_errno_fmt("openat: %s/%s",
2196 		    got_repo_get_path_git_dir(repo),
2197 		    GOT_OBJECTS_PACK_DIR);
2198 	}
2199 
2200 	packdir = fdopendir(packdir_fd);
2201 	if (packdir == NULL) {
2202 		err = got_error_from_errno("fdopendir");
2203 		goto done;
2204 	}
2205 
2206 	while ((dent = readdir(packdir)) != NULL) {
2207 		if (!got_repo_is_packidx_filename(dent->d_name,
2208 		    strlen(dent->d_name)))
2209 			continue;
2210 
2211 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2212 		    dent->d_name) == -1) {
2213 			err = got_error_from_errno("asprintf");
2214 			goto done;
2215 		}
2216 
2217 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2218 		    path_packidx, 0);
2219 		free(path_packidx);
2220 		if (err)
2221 			goto done;
2222 
2223 		if (fstat(packidx->fd, &sb) == -1)
2224 			goto done;
2225 		*total_packsize += sb.st_size;
2226 
2227 		err = got_packidx_get_packfile_path(&path_packfile,
2228 		    packidx->path_packidx);
2229 		if (err)
2230 			goto done;
2231 
2232 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2233 		    0) == -1) {
2234 			free(path_packfile);
2235 			goto done;
2236 		}
2237 		free(path_packfile);
2238 		*total_packsize += sb.st_size;
2239 
2240 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2241 
2242 		(*npackfiles)++;
2243 
2244 		got_packidx_close(packidx);
2245 		packidx = NULL;
2246 	}
2247 done:
2248 	if (packidx)
2249 		got_packidx_close(packidx);
2250 	if (packdir && closedir(packdir) != 0 && err == NULL)
2251 		err = got_error_from_errno("closedir");
2252 	if (err) {
2253 		*npackfiles = 0;
2254 		*nobjects = 0;
2255 		*total_packsize = 0;
2256 	}
2257 	return err;
2258 }
2259 
2260 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2261     got_packidx_bloom_filter_cmp);