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