1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 #include <sys/resource.h>
24 
25 #include <ctype.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <string.h>
33 #include <time.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 
40 #include "bloom.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_sha1.h"
58 #include "got_lib_object_cache.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_gotconfig.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
67 
68 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
69     got_packidx_bloom_filter_cmp);
70 
71 const char *
72 got_repo_get_path(struct got_repository *repo)
73 {
74 	return repo->path;
75 }
76 
77 const char *
78 got_repo_get_path_git_dir(struct got_repository *repo)
79 {
80 	return repo->path_git_dir;
81 }
82 
83 int
84 got_repo_get_fd(struct got_repository *repo)
85 {
86 	return repo->gitdir_fd;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_name(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_name;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_email(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_email;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_name;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_email;
111 }
112 
113 const char *
114 got_repo_get_gitconfig_owner(struct got_repository *repo)
115 {
116 	return repo->gitconfig_owner;
117 }
118 
119 void
120 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
121     struct got_repository *repo)
122 {
123 	*extensions = repo->extensions;
124 	*nextensions = repo->nextensions;
125 }
126 
127 int
128 got_repo_is_bare(struct got_repository *repo)
129 {
130 	return (strcmp(repo->path, repo->path_git_dir) == 0);
131 }
132 
133 static char *
134 get_path_git_child(struct got_repository *repo, const char *basename)
135 {
136 	char *path_child;
137 
138 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
139 	    basename) == -1)
140 		return NULL;
141 
142 	return path_child;
143 }
144 
145 char *
146 got_repo_get_path_objects(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
149 }
150 
151 char *
152 got_repo_get_path_objects_pack(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
155 }
156 
157 char *
158 got_repo_get_path_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_REFS_DIR);
161 }
162 
163 char *
164 got_repo_get_path_packed_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
167 }
168 
169 static char *
170 get_path_head(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_HEAD_FILE);
173 }
174 
175 char *
176 got_repo_get_path_gitconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GITCONFIG);
179 }
180 
181 char *
182 got_repo_get_path_gotconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
185 }
186 
187 const struct got_gotconfig *
188 got_repo_get_gotconfig(struct got_repository *repo)
189 {
190 	return repo->gotconfig;
191 }
192 
193 void
194 got_repo_get_gitconfig_remotes(int *nremotes,
195     const struct got_remote_repo **remotes, struct got_repository *repo)
196 {
197 	*nremotes = repo->ngitconfig_remotes;
198 	*remotes = repo->gitconfig_remotes;
199 }
200 
201 static int
202 is_git_repo(struct got_repository *repo)
203 {
204 	const char *path_git = got_repo_get_path_git_dir(repo);
205 	char *path_objects = got_repo_get_path_objects(repo);
206 	char *path_refs = got_repo_get_path_refs(repo);
207 	char *path_head = get_path_head(repo);
208 	int ret = 0;
209 	struct stat sb;
210 	struct got_reference *head_ref;
211 
212 	if (lstat(path_git, &sb) == -1)
213 		goto done;
214 	if (!S_ISDIR(sb.st_mode))
215 		goto done;
216 
217 	if (lstat(path_objects, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_refs, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_head, &sb) == -1)
228 		goto done;
229 	if (!S_ISREG(sb.st_mode))
230 		goto done;
231 
232 	/* Check if the HEAD reference can be opened. */
233 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
234 		goto done;
235 	got_ref_close(head_ref);
236 
237 	ret = 1;
238 done:
239 	free(path_objects);
240 	free(path_refs);
241 	free(path_head);
242 	return ret;
243 
244 }
245 
246 const struct got_error *
247 got_repo_pack_fds_open(int **pack_fds)
248 {
249 	const struct got_error *err = NULL;
250 	int i, pack_fds_tmp[GOT_PACK_NUM_TEMPFILES];
251 
252 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
253 	if (*pack_fds == NULL)
254 		return got_error_from_errno("calloc");
255 
256 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
257 		pack_fds_tmp[i] = got_opentempfd();
258 		if (pack_fds_tmp[i] == -1) {
259 			err = got_error_from_errno("got_opentempfd");
260 			got_repo_pack_fds_close(pack_fds_tmp);
261 			return err;
262 		}
263 	}
264 	memcpy(*pack_fds, pack_fds_tmp, sizeof(pack_fds_tmp));
265 	return err;
266 }
267 
268 const struct got_error *
269 got_repo_pack_fds_close(int *pack_fds)
270 {
271 	const struct got_error *err = NULL;
272 	int i;
273 
274 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
275 		if (pack_fds[i] == -1)
276 			continue;
277 		if (close(pack_fds[i]) == -1) {
278 			err = got_error_from_errno("close");
279 			break;
280 		}
281 	}
282 	free(pack_fds);
283 	return err;
284 }
285 
286 const struct got_error *
287 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
288     struct got_object *obj)
289 {
290 #ifndef GOT_NO_OBJ_CACHE
291 	const struct got_error *err = NULL;
292 	err = got_object_cache_add(&repo->objcache, id, obj);
293 	if (err) {
294 		if (err->code == GOT_ERR_OBJ_EXISTS ||
295 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
296 			err = NULL;
297 		return err;
298 	}
299 	obj->refcnt++;
300 #endif
301 	return NULL;
302 }
303 
304 struct got_object *
305 got_repo_get_cached_object(struct got_repository *repo,
306     struct got_object_id *id)
307 {
308 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
309 }
310 
311 const struct got_error *
312 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
313     struct got_tree_object *tree)
314 {
315 #ifndef GOT_NO_OBJ_CACHE
316 	const struct got_error *err = NULL;
317 	err = got_object_cache_add(&repo->treecache, id, tree);
318 	if (err) {
319 		if (err->code == GOT_ERR_OBJ_EXISTS ||
320 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
321 			err = NULL;
322 		return err;
323 	}
324 	tree->refcnt++;
325 #endif
326 	return NULL;
327 }
328 
329 struct got_tree_object *
330 got_repo_get_cached_tree(struct got_repository *repo,
331     struct got_object_id *id)
332 {
333 	return (struct got_tree_object *)got_object_cache_get(
334 	    &repo->treecache, id);
335 }
336 
337 const struct got_error *
338 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
339     struct got_commit_object *commit)
340 {
341 #ifndef GOT_NO_OBJ_CACHE
342 	const struct got_error *err = NULL;
343 	err = got_object_cache_add(&repo->commitcache, id, commit);
344 	if (err) {
345 		if (err->code == GOT_ERR_OBJ_EXISTS ||
346 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
347 			err = NULL;
348 		return err;
349 	}
350 	commit->refcnt++;
351 #endif
352 	return NULL;
353 }
354 
355 struct got_commit_object *
356 got_repo_get_cached_commit(struct got_repository *repo,
357     struct got_object_id *id)
358 {
359 	return (struct got_commit_object *)got_object_cache_get(
360 	    &repo->commitcache, id);
361 }
362 
363 const struct got_error *
364 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
365     struct got_tag_object *tag)
366 {
367 #ifndef GOT_NO_OBJ_CACHE
368 	const struct got_error *err = NULL;
369 	err = got_object_cache_add(&repo->tagcache, id, tag);
370 	if (err) {
371 		if (err->code == GOT_ERR_OBJ_EXISTS ||
372 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
373 			err = NULL;
374 		return err;
375 	}
376 	tag->refcnt++;
377 #endif
378 	return NULL;
379 }
380 
381 struct got_tag_object *
382 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
383 {
384 	return (struct got_tag_object *)got_object_cache_get(
385 	    &repo->tagcache, id);
386 }
387 
388 const struct got_error *
389 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
390     struct got_raw_object *raw)
391 {
392 #ifndef GOT_NO_OBJ_CACHE
393 	const struct got_error *err = NULL;
394 	err = got_object_cache_add(&repo->rawcache, id, raw);
395 	if (err) {
396 		if (err->code == GOT_ERR_OBJ_EXISTS ||
397 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
398 			err = NULL;
399 		return err;
400 	}
401 	raw->refcnt++;
402 #endif
403 	return NULL;
404 }
405 
406 
407 struct got_raw_object *
408 got_repo_get_cached_raw_object(struct got_repository *repo,
409     struct got_object_id *id)
410 {
411 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
412 }
413 
414 
415 static const struct got_error *
416 open_repo(struct got_repository *repo, const char *path)
417 {
418 	const struct got_error *err = NULL;
419 
420 	repo->gitdir_fd = -1;
421 
422 	/* bare git repository? */
423 	repo->path_git_dir = strdup(path);
424 	if (repo->path_git_dir == NULL)
425 		return got_error_from_errno("strdup");
426 	if (is_git_repo(repo)) {
427 		repo->path = strdup(repo->path_git_dir);
428 		if (repo->path == NULL) {
429 			err = got_error_from_errno("strdup");
430 			goto done;
431 		}
432 		repo->gitdir_fd = open(repo->path_git_dir,
433 		    O_DIRECTORY | O_CLOEXEC);
434 		if (repo->gitdir_fd == -1) {
435 			err = got_error_from_errno2("open",
436 			    repo->path_git_dir);
437 			goto done;
438 		}
439 		return NULL;
440 	}
441 
442 	/* git repository with working tree? */
443 	free(repo->path_git_dir);
444 	repo->path_git_dir = NULL;
445 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
446 		err = got_error_from_errno("asprintf");
447 		goto done;
448 	}
449 	if (is_git_repo(repo)) {
450 		repo->path = strdup(path);
451 		if (repo->path == NULL) {
452 			err = got_error_from_errno("strdup");
453 			goto done;
454 		}
455 		repo->gitdir_fd = open(repo->path_git_dir,
456 		    O_DIRECTORY | O_CLOEXEC);
457 		if (repo->gitdir_fd == -1) {
458 			err = got_error_from_errno2("open",
459 			    repo->path_git_dir);
460 			goto done;
461 		}
462 		return NULL;
463 	}
464 
465 	err = got_error(GOT_ERR_NOT_GIT_REPO);
466 done:
467 	if (err) {
468 		free(repo->path);
469 		repo->path = NULL;
470 		free(repo->path_git_dir);
471 		repo->path_git_dir = NULL;
472 		if (repo->gitdir_fd != -1)
473 			close(repo->gitdir_fd);
474 		repo->gitdir_fd = -1;
475 
476 	}
477 	return err;
478 }
479 
480 static const struct got_error *
481 parse_gitconfig_file(int *gitconfig_repository_format_version,
482     char **gitconfig_author_name, char **gitconfig_author_email,
483     struct got_remote_repo **remotes, int *nremotes,
484     char **gitconfig_owner, char ***extensions, int *nextensions,
485     const char *gitconfig_path)
486 {
487 	const struct got_error *err = NULL, *child_err = NULL;
488 	int fd = -1;
489 	int imsg_fds[2] = { -1, -1 };
490 	pid_t pid;
491 	struct imsgbuf *ibuf;
492 
493 	*gitconfig_repository_format_version = 0;
494 	if (extensions)
495 		*extensions = NULL;
496 	if (nextensions)
497 		*nextensions = 0;
498 	*gitconfig_author_name = NULL;
499 	*gitconfig_author_email = NULL;
500 	if (remotes)
501 		*remotes = NULL;
502 	if (nremotes)
503 		*nremotes = 0;
504 	if (gitconfig_owner)
505 		*gitconfig_owner = NULL;
506 
507 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
508 	if (fd == -1) {
509 		if (errno == ENOENT)
510 			return NULL;
511 		return got_error_from_errno2("open", gitconfig_path);
512 	}
513 
514 	ibuf = calloc(1, sizeof(*ibuf));
515 	if (ibuf == NULL) {
516 		err = got_error_from_errno("calloc");
517 		goto done;
518 	}
519 
520 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
521 		err = got_error_from_errno("socketpair");
522 		goto done;
523 	}
524 
525 	pid = fork();
526 	if (pid == -1) {
527 		err = got_error_from_errno("fork");
528 		goto done;
529 	} else if (pid == 0) {
530 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
531 		    gitconfig_path);
532 		/* not reached */
533 	}
534 
535 	if (close(imsg_fds[1]) == -1) {
536 		err = got_error_from_errno("close");
537 		goto done;
538 	}
539 	imsg_fds[1] = -1;
540 	imsg_init(ibuf, imsg_fds[0]);
541 
542 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
543 	if (err)
544 		goto done;
545 	fd = -1;
546 
547 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
548 	if (err)
549 		goto done;
550 
551 	err = got_privsep_recv_gitconfig_int(
552 	    gitconfig_repository_format_version, ibuf);
553 	if (err)
554 		goto done;
555 
556 	if (extensions && nextensions) {
557 		err = got_privsep_send_gitconfig_repository_extensions_req(
558 		    ibuf);
559 		if (err)
560 			goto done;
561 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
562 		if (err)
563 			goto done;
564 		if (*nextensions > 0) {
565 			int i;
566 			*extensions = calloc(*nextensions, sizeof(char *));
567 			if (*extensions == NULL) {
568 				err = got_error_from_errno("calloc");
569 				goto done;
570 			}
571 			for (i = 0; i < *nextensions; i++) {
572 				char *ext;
573 				err = got_privsep_recv_gitconfig_str(&ext,
574 				    ibuf);
575 				if (err)
576 					goto done;
577 				(*extensions)[i] = ext;
578 			}
579 		}
580 	}
581 
582 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
583 	if (err)
584 		goto done;
585 
586 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
587 	if (err)
588 		goto done;
589 
590 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
591 	if (err)
592 		goto done;
593 
594 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
595 	if (err)
596 		goto done;
597 
598 	if (remotes && nremotes) {
599 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
600 		if (err)
601 			goto done;
602 
603 		err = got_privsep_recv_gitconfig_remotes(remotes,
604 		    nremotes, ibuf);
605 		if (err)
606 			goto done;
607 	}
608 
609 	if (gitconfig_owner) {
610 		err = got_privsep_send_gitconfig_owner_req(ibuf);
611 		if (err)
612 			goto done;
613 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
614 		if (err)
615 			goto done;
616 	}
617 
618 	err = got_privsep_send_stop(imsg_fds[0]);
619 	child_err = got_privsep_wait_for_child(pid);
620 	if (child_err && err == NULL)
621 		err = child_err;
622 done:
623 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
624 		err = got_error_from_errno("close");
625 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
626 		err = got_error_from_errno("close");
627 	if (fd != -1 && close(fd) == -1 && err == NULL)
628 		err = got_error_from_errno2("close", gitconfig_path);
629 	free(ibuf);
630 	return err;
631 }
632 
633 static const struct got_error *
634 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
635 {
636 	const struct got_error *err = NULL;
637 	char *repo_gitconfig_path = NULL;
638 
639 	if (global_gitconfig_path) {
640 		/* Read settings from ~/.gitconfig. */
641 		int dummy_repo_version;
642 		err = parse_gitconfig_file(&dummy_repo_version,
643 		    &repo->global_gitconfig_author_name,
644 		    &repo->global_gitconfig_author_email,
645 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
646 		if (err)
647 			return err;
648 	}
649 
650 	/* Read repository's .git/config file. */
651 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
652 	if (repo_gitconfig_path == NULL)
653 		return got_error_from_errno("got_repo_get_path_gitconfig");
654 
655 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
656 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
657 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
658 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
659 	    repo_gitconfig_path);
660 	if (err)
661 		goto done;
662 done:
663 	free(repo_gitconfig_path);
664 	return err;
665 }
666 
667 static const struct got_error *
668 read_gotconfig(struct got_repository *repo)
669 {
670 	const struct got_error *err = NULL;
671 	char *gotconfig_path;
672 
673 	gotconfig_path = got_repo_get_path_gotconfig(repo);
674 	if (gotconfig_path == NULL)
675 		return got_error_from_errno("got_repo_get_path_gotconfig");
676 
677 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
678 	free(gotconfig_path);
679 	return err;
680 }
681 
682 /* Supported repository format extensions. */
683 static const char *const repo_extensions[] = {
684 	"noop",			/* Got supports repository format version 1. */
685 	"preciousObjects",	/* Supported by gotadmin cleanup. */
686 	"worktreeConfig",	/* Got does not care about Git work trees. */
687 };
688 
689 const struct got_error *
690 got_repo_open(struct got_repository **repop, const char *path,
691     const char *global_gitconfig_path, int *pack_fds)
692 {
693 	struct got_repository *repo = NULL;
694 	const struct got_error *err = NULL;
695 	char *repo_path = NULL;
696 	size_t i, j = 0;
697 	struct rlimit rl;
698 
699 	*repop = NULL;
700 
701 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
702 		return got_error_from_errno("getrlimit");
703 
704 	repo = calloc(1, sizeof(*repo));
705 	if (repo == NULL)
706 		return got_error_from_errno("calloc");
707 
708 	RB_INIT(&repo->packidx_bloom_filters);
709 	TAILQ_INIT(&repo->packidx_paths);
710 
711 	for (i = 0; i < nitems(repo->privsep_children); i++) {
712 		memset(&repo->privsep_children[i], 0,
713 		    sizeof(repo->privsep_children[0]));
714 		repo->privsep_children[i].imsg_fd = -1;
715 	}
716 
717 	err = got_object_cache_init(&repo->objcache,
718 	    GOT_OBJECT_CACHE_TYPE_OBJ);
719 	if (err)
720 		goto done;
721 	err = got_object_cache_init(&repo->treecache,
722 	    GOT_OBJECT_CACHE_TYPE_TREE);
723 	if (err)
724 		goto done;
725 	err = got_object_cache_init(&repo->commitcache,
726 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
727 	if (err)
728 		goto done;
729 	err = got_object_cache_init(&repo->tagcache,
730 	    GOT_OBJECT_CACHE_TYPE_TAG);
731 	if (err)
732 		goto done;
733 	err = got_object_cache_init(&repo->rawcache,
734 	    GOT_OBJECT_CACHE_TYPE_RAW);
735 	if (err)
736 		goto done;
737 
738 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
739 	if (repo->pack_cache_size > rl.rlim_cur / 8)
740 		repo->pack_cache_size = rl.rlim_cur / 8;
741 	for (i = 0; i < nitems(repo->packs); i++) {
742 		if (i < repo->pack_cache_size) {
743 			repo->packs[i].basefd = pack_fds[j++];
744 			repo->packs[i].accumfd = pack_fds[j++];
745 		} else {
746 			repo->packs[i].basefd = -1;
747 			repo->packs[i].accumfd = -1;
748 		}
749 	}
750 	repo->pinned_pack = -1;
751 	repo->pinned_packidx = -1;
752 	repo->pinned_pid = 0;
753 
754 	repo_path = realpath(path, NULL);
755 	if (repo_path == NULL) {
756 		err = got_error_from_errno2("realpath", path);
757 		goto done;
758 	}
759 
760 	for (;;) {
761 		char *parent_path;
762 
763 		err = open_repo(repo, repo_path);
764 		if (err == NULL)
765 			break;
766 		if (err->code != GOT_ERR_NOT_GIT_REPO)
767 			goto done;
768 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
769 			err = got_error(GOT_ERR_NOT_GIT_REPO);
770 			goto done;
771 		}
772 		err = got_path_dirname(&parent_path, repo_path);
773 		if (err)
774 			goto done;
775 		free(repo_path);
776 		repo_path = parent_path;
777 	}
778 
779 	err = read_gotconfig(repo);
780 	if (err)
781 		goto done;
782 
783 	err = read_gitconfig(repo, global_gitconfig_path);
784 	if (err)
785 		goto done;
786 	if (repo->gitconfig_repository_format_version != 0) {
787 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
788 		goto done;
789 	}
790 	for (i = 0; i < repo->nextensions; i++) {
791 		char *ext = repo->extensions[i];
792 		int j, supported = 0;
793 		for (j = 0; j < nitems(repo_extensions); j++) {
794 			if (strcmp(ext, repo_extensions[j]) == 0) {
795 				supported = 1;
796 				break;
797 			}
798 		}
799 		if (!supported) {
800 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
801 			goto done;
802 		}
803 	}
804 
805 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
806 done:
807 	if (err)
808 		got_repo_close(repo);
809 	else
810 		*repop = repo;
811 	free(repo_path);
812 	return err;
813 }
814 
815 const struct got_error *
816 got_repo_close(struct got_repository *repo)
817 {
818 	const struct got_error *err = NULL, *child_err;
819 	struct got_packidx_bloom_filter *bf;
820 	struct got_pathlist_entry *pe;
821 	size_t i;
822 
823 	for (i = 0; i < repo->pack_cache_size; i++) {
824 		if (repo->packidx_cache[i] == NULL)
825 			break;
826 		got_packidx_close(repo->packidx_cache[i]);
827 	}
828 
829 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
830 	    &repo->packidx_bloom_filters))) {
831 		RB_REMOVE(got_packidx_bloom_filter_tree,
832 		    &repo->packidx_bloom_filters, bf);
833 		free(bf->bloom);
834 		free(bf);
835 	}
836 
837 	for (i = 0; i < repo->pack_cache_size; i++)
838 		if (repo->packs[i].path_packfile)
839 			if (repo->packs[i].path_packfile)
840 				got_pack_close(&repo->packs[i]);
841 
842 	free(repo->path);
843 	free(repo->path_git_dir);
844 
845 	got_object_cache_close(&repo->objcache);
846 	got_object_cache_close(&repo->treecache);
847 	got_object_cache_close(&repo->commitcache);
848 	got_object_cache_close(&repo->tagcache);
849 	got_object_cache_close(&repo->rawcache);
850 
851 	for (i = 0; i < nitems(repo->privsep_children); i++) {
852 		if (repo->privsep_children[i].imsg_fd == -1)
853 			continue;
854 		imsg_clear(repo->privsep_children[i].ibuf);
855 		free(repo->privsep_children[i].ibuf);
856 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
857 		child_err = got_privsep_wait_for_child(
858 		    repo->privsep_children[i].pid);
859 		if (child_err && err == NULL)
860 			err = child_err;
861 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
862 		    err == NULL)
863 			err = got_error_from_errno("close");
864 	}
865 
866 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
867 	    err == NULL)
868 		err = got_error_from_errno("close");
869 
870 	if (repo->gotconfig)
871 		got_gotconfig_free(repo->gotconfig);
872 	free(repo->gitconfig_author_name);
873 	free(repo->gitconfig_author_email);
874 	for (i = 0; i < repo->ngitconfig_remotes; i++)
875 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
876 	free(repo->gitconfig_remotes);
877 	for (i = 0; i < repo->nextensions; i++)
878 		free(repo->extensions[i]);
879 	free(repo->extensions);
880 
881 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
882 		free((void *)pe->path);
883 	got_pathlist_free(&repo->packidx_paths);
884 	free(repo);
885 
886 	return err;
887 }
888 
889 void
890 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
891 {
892 	int i;
893 
894 	free(repo->name);
895 	repo->name = NULL;
896 	free(repo->fetch_url);
897 	repo->fetch_url = NULL;
898 	free(repo->send_url);
899 	repo->send_url = NULL;
900 	for (i = 0; i < repo->nfetch_branches; i++)
901 		free(repo->fetch_branches[i]);
902 	free(repo->fetch_branches);
903 	repo->fetch_branches = NULL;
904 	repo->nfetch_branches = 0;
905 	for (i = 0; i < repo->nsend_branches; i++)
906 		free(repo->send_branches[i]);
907 	free(repo->send_branches);
908 	repo->send_branches = NULL;
909 	repo->nsend_branches = 0;
910 }
911 
912 const struct got_error *
913 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
914     const char *input_path)
915 {
916 	const struct got_error *err = NULL;
917 	const char *repo_abspath = NULL;
918 	size_t repolen, len;
919 	char *canonpath, *path = NULL;
920 
921 	*in_repo_path = NULL;
922 
923 	canonpath = strdup(input_path);
924 	if (canonpath == NULL) {
925 		err = got_error_from_errno("strdup");
926 		goto done;
927 	}
928 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
929 	if (err)
930 		goto done;
931 
932 	repo_abspath = got_repo_get_path(repo);
933 
934 	if (canonpath[0] == '\0') {
935 		path = strdup(canonpath);
936 		if (path == NULL) {
937 			err = got_error_from_errno("strdup");
938 			goto done;
939 		}
940 	} else {
941 		path = realpath(canonpath, NULL);
942 		if (path == NULL) {
943 			if (errno != ENOENT) {
944 				err = got_error_from_errno2("realpath",
945 				    canonpath);
946 				goto done;
947 			}
948 			/*
949 			 * Path is not on disk.
950 			 * Assume it is already relative to repository root.
951 			 */
952 			path = strdup(canonpath);
953 			if (path == NULL) {
954 				err = got_error_from_errno("strdup");
955 				goto done;
956 			}
957 		}
958 
959 		repolen = strlen(repo_abspath);
960 		len = strlen(path);
961 
962 
963 		if (strcmp(path, repo_abspath) == 0) {
964 			free(path);
965 			path = strdup("");
966 			if (path == NULL) {
967 				err = got_error_from_errno("strdup");
968 				goto done;
969 			}
970 		} else if (len > repolen &&
971 		    got_path_is_child(path, repo_abspath, repolen)) {
972 			/* Matched an on-disk path inside repository. */
973 			if (got_repo_is_bare(repo)) {
974 				/*
975 				 * Matched an on-disk path inside repository
976 				 * database. Treat input as repository-relative.
977 				 */
978 				free(path);
979 				path = canonpath;
980 				canonpath = NULL;
981 			} else {
982 				char *child;
983 				/* Strip common prefix with repository path. */
984 				err = got_path_skip_common_ancestor(&child,
985 				    repo_abspath, path);
986 				if (err)
987 					goto done;
988 				free(path);
989 				path = child;
990 			}
991 		} else {
992 			/*
993 			 * Matched unrelated on-disk path.
994 			 * Treat input as repository-relative.
995 			 */
996 			free(path);
997 			path = canonpath;
998 			canonpath = NULL;
999 		}
1000 	}
1001 
1002 	/* Make in-repository path absolute */
1003 	if (path[0] != '/') {
1004 		char *abspath;
1005 		if (asprintf(&abspath, "/%s", path) == -1) {
1006 			err = got_error_from_errno("asprintf");
1007 			goto done;
1008 		}
1009 		free(path);
1010 		path = abspath;
1011 	}
1012 
1013 done:
1014 	free(canonpath);
1015 	if (err)
1016 		free(path);
1017 	else
1018 		*in_repo_path = path;
1019 	return err;
1020 }
1021 
1022 static const struct got_error *
1023 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1024     const char *path_packidx)
1025 {
1026 	const struct got_error *err = NULL;
1027 	size_t i;
1028 
1029 	for (i = 0; i < repo->pack_cache_size; i++) {
1030 		if (repo->packidx_cache[i] == NULL)
1031 			break;
1032 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1033 		    path_packidx) == 0) {
1034 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1035 		}
1036 	}
1037 	if (i == repo->pack_cache_size) {
1038 		do {
1039 			i--;
1040 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1041 		    i == repo->pinned_packidx);
1042 		err = got_packidx_close(repo->packidx_cache[i]);
1043 		if (err)
1044 			return err;
1045 	}
1046 
1047 	repo->packidx_cache[i] = packidx;
1048 
1049 	return NULL;
1050 }
1051 
1052 int
1053 got_repo_is_packidx_filename(const char *name, size_t len)
1054 {
1055 	if (len != GOT_PACKIDX_NAMELEN)
1056 		return 0;
1057 
1058 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1059 		return 0;
1060 
1061 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1062 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1063 		return 0;
1064 
1065 	return 1;
1066 }
1067 
1068 static struct got_packidx_bloom_filter *
1069 get_packidx_bloom_filter(struct got_repository *repo,
1070     const char *path, size_t path_len)
1071 {
1072 	struct got_packidx_bloom_filter key;
1073 
1074 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1075 		return NULL; /* XXX */
1076 	key.path_len = path_len;
1077 
1078 	return RB_FIND(got_packidx_bloom_filter_tree,
1079 	    &repo->packidx_bloom_filters, &key);
1080 }
1081 
1082 int
1083 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1084     const char *path_packidx, struct got_object_id *id)
1085 {
1086 	struct got_packidx_bloom_filter *bf;
1087 
1088 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1089 	if (bf)
1090 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1091 
1092 	/* No bloom filter means this pack index must be searched. */
1093 	return 1;
1094 }
1095 
1096 static const struct got_error *
1097 add_packidx_bloom_filter(struct got_repository *repo,
1098     struct got_packidx *packidx, const char *path_packidx)
1099 {
1100 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1101 	struct got_packidx_bloom_filter *bf;
1102 	size_t len;
1103 
1104 	/*
1105 	 * Don't use bloom filters for very large pack index files.
1106 	 * Large pack files will contain a relatively large fraction
1107 	 * of our objects so we will likely need to visit them anyway.
1108 	 * The more objects a pack file contains the higher the probability
1109 	 * of a false-positive match from the bloom filter. And reading
1110 	 * all object IDs from a large pack index file can be expensive.
1111 	 */
1112 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1113 		return NULL;
1114 
1115 	/* Do we already have a filter for this pack index? */
1116 	if (get_packidx_bloom_filter(repo, path_packidx,
1117 	    strlen(path_packidx)) != NULL)
1118 		return NULL;
1119 
1120 	bf = calloc(1, sizeof(*bf));
1121 	if (bf == NULL)
1122 		return got_error_from_errno("calloc");
1123 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1124 	if (bf->bloom == NULL) {
1125 		free(bf);
1126 		return got_error_from_errno("calloc");
1127 	}
1128 	
1129 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1130 	if (len >= sizeof(bf->path)) {
1131 		free(bf->bloom);
1132 		free(bf);
1133 		return got_error(GOT_ERR_NO_SPACE);
1134 	}
1135 	bf->path_len = len;
1136 
1137 	/* Minimum size supported by our bloom filter is 1000 entries. */
1138 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1139 	for (i = 0; i < nobjects; i++) {
1140 		struct got_packidx_object_id *id;
1141 		id = &packidx->hdr.sorted_ids[i];
1142 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1143 	}
1144 
1145 	RB_INSERT(got_packidx_bloom_filter_tree,
1146 	    &repo->packidx_bloom_filters, bf);
1147 	return NULL;
1148 }
1149 
1150 const struct got_error *
1151 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1152     struct got_repository *repo, struct got_object_id *id)
1153 {
1154 	const struct got_error *err;
1155 	struct got_pathlist_entry *pe;
1156 	size_t i;
1157 
1158 	/* Search pack index cache. */
1159 	for (i = 0; i < repo->pack_cache_size; i++) {
1160 		if (repo->packidx_cache[i] == NULL)
1161 			break;
1162 		if (!got_repo_check_packidx_bloom_filter(repo,
1163 		    repo->packidx_cache[i]->path_packidx, id))
1164 			continue; /* object will not be found in this index */
1165 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1166 		if (*idx != -1) {
1167 			*packidx = repo->packidx_cache[i];
1168 			/*
1169 			 * Move this cache entry to the front. Repeatedly
1170 			 * searching a wrong pack index can be expensive.
1171 			 */
1172 			if (i > 0) {
1173 				memmove(&repo->packidx_cache[1],
1174 				    &repo->packidx_cache[0],
1175 				    i * sizeof(repo->packidx_cache[0]));
1176 				repo->packidx_cache[0] = *packidx;
1177 				if (repo->pinned_packidx >= 0 &&
1178 				    repo->pinned_packidx < i)
1179 					repo->pinned_packidx++;
1180 				else if (repo->pinned_packidx == i)
1181 					repo->pinned_packidx = 0;
1182 			}
1183 			return NULL;
1184 		}
1185 	}
1186 	/* No luck. Search the filesystem. */
1187 
1188 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1189 		const char *path_packidx = pe->path;
1190 		int is_cached = 0;
1191 
1192 		if (!got_repo_check_packidx_bloom_filter(repo,
1193 		    pe->path, id))
1194 			continue; /* object will not be found in this index */
1195 
1196 		for (i = 0; i < repo->pack_cache_size; i++) {
1197 			if (repo->packidx_cache[i] == NULL)
1198 				break;
1199 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1200 			    path_packidx) == 0) {
1201 				is_cached = 1;
1202 				break;
1203 			}
1204 		}
1205 		if (is_cached)
1206 			continue; /* already searched */
1207 
1208 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1209 		    path_packidx, 0);
1210 		if (err)
1211 			goto done;
1212 
1213 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1214 		if (err)
1215 			goto done;
1216 
1217 		err = cache_packidx(repo, *packidx, path_packidx);
1218 		if (err)
1219 			goto done;
1220 
1221 		*idx = got_packidx_get_object_idx(*packidx, id);
1222 		if (*idx != -1) {
1223 			err = NULL; /* found the object */
1224 			goto done;
1225 		}
1226 	}
1227 
1228 	err = got_error_no_obj(id);
1229 done:
1230 	return err;
1231 }
1232 
1233 const struct got_error *
1234 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1235     struct got_repository *repo)
1236 {
1237 	const struct got_error *err = NULL;
1238 	DIR *packdir = NULL;
1239 	struct dirent *dent;
1240 	char *path_packidx = NULL;
1241 	int packdir_fd;
1242 
1243 	packdir_fd = openat(got_repo_get_fd(repo),
1244 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1245 	if (packdir_fd == -1) {
1246 		return got_error_from_errno_fmt("openat: %s/%s",
1247 		    got_repo_get_path_git_dir(repo),
1248 		    GOT_OBJECTS_PACK_DIR);
1249 	}
1250 
1251 	packdir = fdopendir(packdir_fd);
1252 	if (packdir == NULL) {
1253 		err = got_error_from_errno("fdopendir");
1254 		goto done;
1255 	}
1256 
1257 	while ((dent = readdir(packdir)) != NULL) {
1258 		if (!got_repo_is_packidx_filename(dent->d_name,
1259 		    strlen(dent->d_name)))
1260 			continue;
1261 
1262 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1263 		    dent->d_name) == -1) {
1264 			err = got_error_from_errno("asprintf");
1265 			path_packidx = NULL;
1266 			break;
1267 		}
1268 
1269 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1270 		if (err)
1271 			break;
1272 	}
1273 done:
1274 	if (err)
1275 		free(path_packidx);
1276 	if (packdir && closedir(packdir) != 0 && err == NULL)
1277 		err = got_error_from_errno("closedir");
1278 	return err;
1279 }
1280 
1281 const struct got_error *
1282 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1283     struct got_repository *repo)
1284 {
1285 	const struct got_error *err;
1286 	size_t i;
1287 
1288 	*packidx = NULL;
1289 
1290 	/* Search pack index cache. */
1291 	for (i = 0; i < repo->pack_cache_size; i++) {
1292 		if (repo->packidx_cache[i] == NULL)
1293 			break;
1294 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1295 		    path_packidx) == 0) {
1296 			*packidx = repo->packidx_cache[i];
1297 			return NULL;
1298 		}
1299 	}
1300 	/* No luck. Search the filesystem. */
1301 
1302 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1303 	    path_packidx, 0);
1304 	if (err)
1305 		return err;
1306 
1307 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1308 	if (err)
1309 		goto done;
1310 
1311 	err = cache_packidx(repo, *packidx, path_packidx);
1312 done:
1313 	if (err) {
1314 		got_packidx_close(*packidx);
1315 		*packidx = NULL;
1316 	}
1317 	return err;
1318 }
1319 
1320 static const struct got_error *
1321 read_packfile_hdr(int fd, struct got_packidx *packidx)
1322 {
1323 	const struct got_error *err = NULL;
1324 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1325 	struct got_packfile_hdr hdr;
1326 	ssize_t n;
1327 
1328 	n = read(fd, &hdr, sizeof(hdr));
1329 	if (n < 0)
1330 		return got_error_from_errno("read");
1331 	if (n != sizeof(hdr))
1332 		return got_error(GOT_ERR_BAD_PACKFILE);
1333 
1334 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1335 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1336 	    be32toh(hdr.nobjects) != totobj)
1337 		err = got_error(GOT_ERR_BAD_PACKFILE);
1338 
1339 	return err;
1340 }
1341 
1342 static const struct got_error *
1343 open_packfile(int *fd, struct got_repository *repo,
1344     const char *relpath, struct got_packidx *packidx)
1345 {
1346 	const struct got_error *err = NULL;
1347 
1348 	*fd = openat(got_repo_get_fd(repo), relpath,
1349 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1350 	if (*fd == -1)
1351 		return got_error_from_errno_fmt("openat: %s/%s",
1352 		    got_repo_get_path_git_dir(repo), relpath);
1353 
1354 	if (packidx) {
1355 		err = read_packfile_hdr(*fd, packidx);
1356 		if (err) {
1357 			close(*fd);
1358 			*fd = -1;
1359 		}
1360 	}
1361 
1362 	return err;
1363 }
1364 
1365 const struct got_error *
1366 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1367     const char *path_packfile, struct got_packidx *packidx)
1368 {
1369 	const struct got_error *err = NULL;
1370 	struct got_pack *pack = NULL;
1371 	struct stat sb;
1372 	size_t i;
1373 
1374 	if (packp)
1375 		*packp = NULL;
1376 
1377 	for (i = 0; i < repo->pack_cache_size; i++) {
1378 		pack = &repo->packs[i];
1379 		if (pack->path_packfile == NULL)
1380 			break;
1381 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1382 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1383 	}
1384 
1385 	if (i == repo->pack_cache_size) {
1386 		struct got_pack tmp;
1387 		do {
1388 			i--;
1389 		} while (i > 0 && repo->pinned_pack >= 0 &&
1390 		    i == repo->pinned_pack);
1391 		err = got_pack_close(&repo->packs[i]);
1392 		if (err)
1393 			return err;
1394 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1395 			return got_error_from_errno("ftruncate");
1396 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1397 			return got_error_from_errno("ftruncate");
1398 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1399 		memcpy(&repo->packs[i], &repo->packs[0],
1400 		    sizeof(repo->packs[i]));
1401 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1402 		if (repo->pinned_pack == 0)
1403 			repo->pinned_pack = i;
1404 		else if (repo->pinned_pack == i)
1405 			repo->pinned_pack = 0;
1406 		i = 0;
1407 	}
1408 
1409 	pack = &repo->packs[i];
1410 
1411 	pack->path_packfile = strdup(path_packfile);
1412 	if (pack->path_packfile == NULL) {
1413 		err = got_error_from_errno("strdup");
1414 		goto done;
1415 	}
1416 
1417 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1418 	if (err)
1419 		goto done;
1420 
1421 	if (fstat(pack->fd, &sb) != 0) {
1422 		err = got_error_from_errno("fstat");
1423 		goto done;
1424 	}
1425 	pack->filesize = sb.st_size;
1426 
1427 	pack->privsep_child = NULL;
1428 
1429 #ifndef GOT_PACK_NO_MMAP
1430 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1431 	    pack->fd, 0);
1432 	if (pack->map == MAP_FAILED) {
1433 		if (errno != ENOMEM) {
1434 			err = got_error_from_errno("mmap");
1435 			goto done;
1436 		}
1437 		pack->map = NULL; /* fall back to read(2) */
1438 	}
1439 #endif
1440 done:
1441 	if (err) {
1442 		if (pack) {
1443 			free(pack->path_packfile);
1444 			memset(pack, 0, sizeof(*pack));
1445 		}
1446 	} else if (packp)
1447 		*packp = pack;
1448 	return err;
1449 }
1450 
1451 struct got_pack *
1452 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1453 {
1454 	struct got_pack *pack = NULL;
1455 	size_t i;
1456 
1457 	for (i = 0; i < repo->pack_cache_size; i++) {
1458 		pack = &repo->packs[i];
1459 		if (pack->path_packfile == NULL)
1460 			break;
1461 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1462 			return pack;
1463 	}
1464 
1465 	return NULL;
1466 }
1467 
1468 const struct got_error *
1469 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1470     struct got_pack *pack)
1471 {
1472 	size_t i;
1473 	int pinned_pack = -1, pinned_packidx = -1;
1474 
1475 	for (i = 0; i < repo->pack_cache_size; i++) {
1476 		if (repo->packidx_cache[i] &&
1477 		    strcmp(repo->packidx_cache[i]->path_packidx,
1478 		    packidx->path_packidx) == 0)
1479 			pinned_packidx = i;
1480 		if (repo->packs[i].path_packfile &&
1481 		    strcmp(repo->packs[i].path_packfile,
1482 		    pack->path_packfile) == 0)
1483 			pinned_pack = i;
1484 	}
1485 
1486 	if (pinned_packidx == -1 || pinned_pack == -1)
1487 		return got_error(GOT_ERR_PIN_PACK);
1488 
1489 	repo->pinned_pack = pinned_pack;
1490 	repo->pinned_packidx = pinned_packidx;
1491 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1492 	return NULL;
1493 }
1494 
1495 struct got_pack *
1496 got_repo_get_pinned_pack(struct got_repository *repo)
1497 {
1498 	if (repo->pinned_pack >= 0 &&
1499 	    repo->pinned_pack < repo->pack_cache_size)
1500 		return &repo->packs[repo->pinned_pack];
1501 
1502 	return NULL;
1503 }
1504 
1505 void
1506 got_repo_unpin_pack(struct got_repository *repo)
1507 {
1508 	repo->pinned_packidx = -1;
1509 	repo->pinned_pack = -1;
1510 	repo->pinned_pid = 0;
1511 }
1512 
1513 const struct got_error *
1514 got_repo_init(const char *repo_path)
1515 {
1516 	const struct got_error *err = NULL;
1517 	const char *dirnames[] = {
1518 		GOT_OBJECTS_DIR,
1519 		GOT_OBJECTS_PACK_DIR,
1520 		GOT_REFS_DIR,
1521 	};
1522 	const char *description_str = "Unnamed repository; "
1523 	    "edit this file 'description' to name the repository.";
1524 	const char *headref_str = "ref: refs/heads/main";
1525 	const char *gitconfig_str = "[core]\n"
1526 	    "\trepositoryformatversion = 0\n"
1527 	    "\tfilemode = true\n"
1528 	    "\tbare = true\n";
1529 	char *path;
1530 	size_t i;
1531 
1532 	if (!got_path_dir_is_empty(repo_path))
1533 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1534 
1535 	for (i = 0; i < nitems(dirnames); i++) {
1536 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1537 			return got_error_from_errno("asprintf");
1538 		}
1539 		err = got_path_mkdir(path);
1540 		free(path);
1541 		if (err)
1542 			return err;
1543 	}
1544 
1545 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1546 		return got_error_from_errno("asprintf");
1547 	err = got_path_create_file(path, description_str);
1548 	free(path);
1549 	if (err)
1550 		return err;
1551 
1552 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1553 		return got_error_from_errno("asprintf");
1554 	err = got_path_create_file(path, headref_str);
1555 	free(path);
1556 	if (err)
1557 		return err;
1558 
1559 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1560 		return got_error_from_errno("asprintf");
1561 	err = got_path_create_file(path, gitconfig_str);
1562 	free(path);
1563 	if (err)
1564 		return err;
1565 
1566 	return NULL;
1567 }
1568 
1569 static const struct got_error *
1570 match_packed_object(struct got_object_id **unique_id,
1571     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1572 {
1573 	const struct got_error *err = NULL;
1574 	struct got_object_id_queue matched_ids;
1575 	struct got_pathlist_entry *pe;
1576 
1577 	STAILQ_INIT(&matched_ids);
1578 
1579 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1580 		const char *path_packidx = pe->path;
1581 		struct got_packidx *packidx;
1582 		struct got_object_qid *qid;
1583 
1584 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1585 		    path_packidx, 0);
1586 		if (err)
1587 			break;
1588 
1589 		err = got_packidx_match_id_str_prefix(&matched_ids,
1590 		    packidx, id_str_prefix);
1591 		if (err) {
1592 			got_packidx_close(packidx);
1593 			break;
1594 		}
1595 		err = got_packidx_close(packidx);
1596 		if (err)
1597 			break;
1598 
1599 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1600 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1601 				int matched_type;
1602 				err = got_object_get_type(&matched_type, repo,
1603 				    &qid->id);
1604 				if (err)
1605 					goto done;
1606 				if (matched_type != obj_type)
1607 					continue;
1608 			}
1609 			if (*unique_id == NULL) {
1610 				*unique_id = got_object_id_dup(&qid->id);
1611 				if (*unique_id == NULL) {
1612 					err = got_error_from_errno("malloc");
1613 					goto done;
1614 				}
1615 			} else {
1616 				if (got_object_id_cmp(*unique_id,
1617 				    &qid->id) == 0)
1618 					continue; /* packed multiple times */
1619 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1620 				goto done;
1621 			}
1622 		}
1623 	}
1624 done:
1625 	got_object_id_queue_free(&matched_ids);
1626 	if (err) {
1627 		free(*unique_id);
1628 		*unique_id = NULL;
1629 	}
1630 	return err;
1631 }
1632 
1633 static const struct got_error *
1634 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1635     const char *object_dir, const char *id_str_prefix, int obj_type,
1636     struct got_repository *repo)
1637 {
1638 	const struct got_error *err = NULL;
1639 	char *path;
1640 	DIR *dir = NULL;
1641 	struct dirent *dent;
1642 	struct got_object_id id;
1643 
1644 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1645 		err = got_error_from_errno("asprintf");
1646 		goto done;
1647 	}
1648 
1649 	dir = opendir(path);
1650 	if (dir == NULL) {
1651 		if (errno == ENOENT) {
1652 			err = NULL;
1653 			goto done;
1654 		}
1655 		err = got_error_from_errno2("opendir", path);
1656 		goto done;
1657 	}
1658 	while ((dent = readdir(dir)) != NULL) {
1659 		char *id_str;
1660 		int cmp;
1661 
1662 		if (strcmp(dent->d_name, ".") == 0 ||
1663 		    strcmp(dent->d_name, "..") == 0)
1664 			continue;
1665 
1666 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1667 			err = got_error_from_errno("asprintf");
1668 			goto done;
1669 		}
1670 
1671 		if (!got_parse_sha1_digest(id.sha1, id_str))
1672 			continue;
1673 
1674 		/*
1675 		 * Directory entries do not necessarily appear in
1676 		 * sorted order, so we must iterate over all of them.
1677 		 */
1678 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1679 		if (cmp != 0) {
1680 			free(id_str);
1681 			continue;
1682 		}
1683 
1684 		if (*unique_id == NULL) {
1685 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1686 				int matched_type;
1687 				err = got_object_get_type(&matched_type, repo,
1688 				    &id);
1689 				if (err)
1690 					goto done;
1691 				if (matched_type != obj_type)
1692 					continue;
1693 			}
1694 			*unique_id = got_object_id_dup(&id);
1695 			if (*unique_id == NULL) {
1696 				err = got_error_from_errno("got_object_id_dup");
1697 				free(id_str);
1698 				goto done;
1699 			}
1700 		} else {
1701 			if (got_object_id_cmp(*unique_id, &id) == 0)
1702 				continue; /* both packed and loose */
1703 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1704 			free(id_str);
1705 			goto done;
1706 		}
1707 	}
1708 done:
1709 	if (dir && closedir(dir) != 0 && err == NULL)
1710 		err = got_error_from_errno("closedir");
1711 	if (err) {
1712 		free(*unique_id);
1713 		*unique_id = NULL;
1714 	}
1715 	free(path);
1716 	return err;
1717 }
1718 
1719 const struct got_error *
1720 got_repo_match_object_id_prefix(struct got_object_id **id,
1721     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1722 {
1723 	const struct got_error *err = NULL;
1724 	char *path_objects = got_repo_get_path_objects(repo);
1725 	char *object_dir = NULL;
1726 	size_t len;
1727 	int i;
1728 
1729 	*id = NULL;
1730 
1731 	len = strlen(id_str_prefix);
1732 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1733 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1734 
1735 	for (i = 0; i < len; i++) {
1736 		if (isxdigit((unsigned char)id_str_prefix[i]))
1737 			continue;
1738 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1739 	}
1740 
1741 	if (len >= 2) {
1742 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1743 		if (err)
1744 			goto done;
1745 		object_dir = strndup(id_str_prefix, 2);
1746 		if (object_dir == NULL) {
1747 			err = got_error_from_errno("strdup");
1748 			goto done;
1749 		}
1750 		err = match_loose_object(id, path_objects, object_dir,
1751 		    id_str_prefix, obj_type, repo);
1752 	} else if (len == 1) {
1753 		int i;
1754 		for (i = 0; i < 0xf; i++) {
1755 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1756 			    == -1) {
1757 				err = got_error_from_errno("asprintf");
1758 				goto done;
1759 			}
1760 			err = match_packed_object(id, repo, object_dir,
1761 			    obj_type);
1762 			if (err)
1763 				goto done;
1764 			err = match_loose_object(id, path_objects, object_dir,
1765 			    id_str_prefix, obj_type, repo);
1766 			if (err)
1767 				goto done;
1768 		}
1769 	} else {
1770 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1771 		goto done;
1772 	}
1773 done:
1774 	free(object_dir);
1775 	if (err) {
1776 		free(*id);
1777 		*id = NULL;
1778 	} else if (*id == NULL) {
1779 		switch (obj_type) {
1780 		case GOT_OBJ_TYPE_BLOB:
1781 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1782 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1783 			break;
1784 		case GOT_OBJ_TYPE_TREE:
1785 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1786 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1787 			break;
1788 		case GOT_OBJ_TYPE_COMMIT:
1789 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1790 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1791 			break;
1792 		case GOT_OBJ_TYPE_TAG:
1793 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1794 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1795 			break;
1796 		default:
1797 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1798 			break;
1799 		}
1800 	}
1801 
1802 	return err;
1803 }
1804 
1805 const struct got_error *
1806 got_repo_match_object_id(struct got_object_id **id, char **label,
1807     const char *id_str, int obj_type, struct got_reflist_head *refs,
1808     struct got_repository *repo)
1809 {
1810 	const struct got_error *err;
1811 	struct got_tag_object *tag;
1812 	struct got_reference *ref = NULL;
1813 
1814 	*id = NULL;
1815 	if (label)
1816 		*label = NULL;
1817 
1818 	if (refs) {
1819 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1820 		    refs, repo);
1821 		if (err == NULL) {
1822 			*id = got_object_id_dup(
1823 			    got_object_tag_get_object_id(tag));
1824 			if (*id == NULL)
1825 				err = got_error_from_errno("got_object_id_dup");
1826 			else if (label && asprintf(label, "refs/tags/%s",
1827 			    got_object_tag_get_name(tag)) == -1) {
1828 				err = got_error_from_errno("asprintf");
1829 				free(*id);
1830 				*id = NULL;
1831 			}
1832 			got_object_tag_close(tag);
1833 			return err;
1834 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1835 		    err->code != GOT_ERR_NO_OBJ)
1836 			return err;
1837 	}
1838 
1839 	err = got_ref_open(&ref, repo, id_str, 0);
1840 	if (err == NULL) {
1841 		err = got_ref_resolve(id, repo, ref);
1842 		if (err)
1843 			goto done;
1844 		if (label) {
1845 			*label = strdup(got_ref_get_name(ref));
1846 			if (*label == NULL) {
1847 				err = got_error_from_errno("strdup");
1848 				goto done;
1849 			}
1850 		}
1851 	} else {
1852 		if (err->code != GOT_ERR_NOT_REF &&
1853 		    err->code != GOT_ERR_BAD_REF_NAME)
1854 			goto done;
1855 		err = got_repo_match_object_id_prefix(id, id_str,
1856 		    obj_type, repo);
1857 		if (err) {
1858 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1859 				err = got_error_not_ref(id_str);
1860 			goto done;
1861 		}
1862 		if (label) {
1863 			err = got_object_id_str(label, *id);
1864 			if (*label == NULL) {
1865 				err = got_error_from_errno("strdup");
1866 				goto done;
1867 			}
1868 		}
1869 	}
1870 done:
1871 	if (ref)
1872 		got_ref_close(ref);
1873 	return err;
1874 }
1875 
1876 const struct got_error *
1877 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1878     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1879 {
1880 	const struct got_error *err = NULL;
1881 	struct got_reflist_entry *re;
1882 	struct got_object_id *tag_id;
1883 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1884 
1885 	*tag = NULL;
1886 
1887 	TAILQ_FOREACH(re, refs, entry) {
1888 		const char *refname;
1889 		refname = got_ref_get_name(re->ref);
1890 		if (got_ref_is_symbolic(re->ref))
1891 			continue;
1892 		if (strncmp(refname, "refs/tags/", 10) != 0)
1893 			continue;
1894 		if (!name_is_absolute)
1895 			refname += strlen("refs/tags/");
1896 		if (strcmp(refname, name) != 0)
1897 			continue;
1898 		err = got_ref_resolve(&tag_id, repo, re->ref);
1899 		if (err)
1900 			break;
1901 		err = got_object_open_as_tag(tag, repo, tag_id);
1902 		free(tag_id);
1903 		if (err)
1904 			break;
1905 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1906 		    got_object_tag_get_object_type(*tag) == obj_type)
1907 			break;
1908 		got_object_tag_close(*tag);
1909 		*tag = NULL;
1910 	}
1911 
1912 	if (err == NULL && *tag == NULL)
1913 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1914 		    GOT_OBJ_LABEL_TAG, name);
1915 	return err;
1916 }
1917 
1918 static const struct got_error *
1919 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1920     const char *name, mode_t mode, struct got_object_id *blob_id)
1921 {
1922 	const struct got_error *err = NULL;
1923 
1924 	 *new_te = NULL;
1925 
1926 	*new_te = calloc(1, sizeof(**new_te));
1927 	if (*new_te == NULL)
1928 		return got_error_from_errno("calloc");
1929 
1930 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1931 	    sizeof((*new_te)->name)) {
1932 		err = got_error(GOT_ERR_NO_SPACE);
1933 		goto done;
1934 	}
1935 
1936 	if (S_ISLNK(mode)) {
1937 		(*new_te)->mode = S_IFLNK;
1938 	} else {
1939 		(*new_te)->mode = S_IFREG;
1940 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1941 	}
1942 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1943 done:
1944 	if (err && *new_te) {
1945 		free(*new_te);
1946 		*new_te = NULL;
1947 	}
1948 	return err;
1949 }
1950 
1951 static const struct got_error *
1952 import_file(struct got_tree_entry **new_te, struct dirent *de,
1953     const char *path, struct got_repository *repo)
1954 {
1955 	const struct got_error *err;
1956 	struct got_object_id *blob_id = NULL;
1957 	char *filepath;
1958 	struct stat sb;
1959 
1960 	if (asprintf(&filepath, "%s%s%s", path,
1961 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1962 		return got_error_from_errno("asprintf");
1963 
1964 	if (lstat(filepath, &sb) != 0) {
1965 		err = got_error_from_errno2("lstat", path);
1966 		goto done;
1967 	}
1968 
1969 	err = got_object_blob_create(&blob_id, filepath, repo);
1970 	if (err)
1971 		goto done;
1972 
1973 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1974 	    blob_id);
1975 done:
1976 	free(filepath);
1977 	if (err)
1978 		free(blob_id);
1979 	return err;
1980 }
1981 
1982 static const struct got_error *
1983 insert_tree_entry(struct got_tree_entry *new_te,
1984     struct got_pathlist_head *paths)
1985 {
1986 	const struct got_error *err = NULL;
1987 	struct got_pathlist_entry *new_pe;
1988 
1989 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1990 	if (err)
1991 		return err;
1992 	if (new_pe == NULL)
1993 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1994 	return NULL;
1995 }
1996 
1997 static const struct got_error *write_tree(struct got_object_id **,
1998     const char *, struct got_pathlist_head *, struct got_repository *,
1999     got_repo_import_cb progress_cb, void *progress_arg);
2000 
2001 static const struct got_error *
2002 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2003     const char *path, struct got_pathlist_head *ignores,
2004     struct got_repository *repo,
2005     got_repo_import_cb progress_cb, void *progress_arg)
2006 {
2007 	const struct got_error *err;
2008 	struct got_object_id *id = NULL;
2009 	char *subdirpath;
2010 
2011 	if (asprintf(&subdirpath, "%s%s%s", path,
2012 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2013 		return got_error_from_errno("asprintf");
2014 
2015 	(*new_te) = calloc(1, sizeof(**new_te));
2016 	if (*new_te == NULL)
2017 		return got_error_from_errno("calloc");
2018 	(*new_te)->mode = S_IFDIR;
2019 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2020 	    sizeof((*new_te)->name)) {
2021 		err = got_error(GOT_ERR_NO_SPACE);
2022 		goto done;
2023 	}
2024 	err = write_tree(&id, subdirpath, ignores,  repo,
2025 	    progress_cb, progress_arg);
2026 	if (err)
2027 		goto done;
2028 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2029 	
2030 done:
2031 	free(id);
2032 	free(subdirpath);
2033 	if (err) {
2034 		free(*new_te);
2035 		*new_te = NULL;
2036 	}
2037 	return err;
2038 }
2039 
2040 static const struct got_error *
2041 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2042     struct got_pathlist_head *ignores, struct got_repository *repo,
2043     got_repo_import_cb progress_cb, void *progress_arg)
2044 {
2045 	const struct got_error *err = NULL;
2046 	DIR *dir;
2047 	struct dirent *de;
2048 	int nentries;
2049 	struct got_tree_entry *new_te = NULL;
2050 	struct got_pathlist_head paths;
2051 	struct got_pathlist_entry *pe;
2052 
2053 	*new_tree_id = NULL;
2054 
2055 	TAILQ_INIT(&paths);
2056 
2057 	dir = opendir(path_dir);
2058 	if (dir == NULL) {
2059 		err = got_error_from_errno2("opendir", path_dir);
2060 		goto done;
2061 	}
2062 
2063 	nentries = 0;
2064 	while ((de = readdir(dir)) != NULL) {
2065 		int ignore = 0;
2066 		int type;
2067 
2068 		if (strcmp(de->d_name, ".") == 0 ||
2069 		    strcmp(de->d_name, "..") == 0)
2070 			continue;
2071 
2072 		TAILQ_FOREACH(pe, ignores, entry) {
2073 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2074 				ignore = 1;
2075 				break;
2076 			}
2077 		}
2078 		if (ignore)
2079 			continue;
2080 
2081 		err = got_path_dirent_type(&type, path_dir, de);
2082 		if (err)
2083 			goto done;
2084 
2085 		if (type == DT_DIR) {
2086 			err = import_subdir(&new_te, de, path_dir,
2087 			    ignores, repo, progress_cb, progress_arg);
2088 			if (err) {
2089 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2090 					goto done;
2091 				err = NULL;
2092 				continue;
2093 			}
2094 		} else if (type == DT_REG || type == DT_LNK) {
2095 			err = import_file(&new_te, de, path_dir, repo);
2096 			if (err)
2097 				goto done;
2098 		} else
2099 			continue;
2100 
2101 		err = insert_tree_entry(new_te, &paths);
2102 		if (err)
2103 			goto done;
2104 		nentries++;
2105 	}
2106 
2107 	if (TAILQ_EMPTY(&paths)) {
2108 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2109 		    "cannot create tree without any entries");
2110 		goto done;
2111 	}
2112 
2113 	TAILQ_FOREACH(pe, &paths, entry) {
2114 		struct got_tree_entry *te = pe->data;
2115 		char *path;
2116 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2117 			continue;
2118 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2119 			err = got_error_from_errno("asprintf");
2120 			goto done;
2121 		}
2122 		err = (*progress_cb)(progress_arg, path);
2123 		free(path);
2124 		if (err)
2125 			goto done;
2126 	}
2127 
2128 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2129 done:
2130 	if (dir)
2131 		closedir(dir);
2132 	got_pathlist_free(&paths);
2133 	return err;
2134 }
2135 
2136 const struct got_error *
2137 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2138     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2139     struct got_repository *repo, got_repo_import_cb progress_cb,
2140     void *progress_arg)
2141 {
2142 	const struct got_error *err;
2143 	struct got_object_id *new_tree_id;
2144 
2145 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2146 	    progress_cb, progress_arg);
2147 	if (err)
2148 		return err;
2149 
2150 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2151 	    author, time(NULL), author, time(NULL), logmsg, repo);
2152 	free(new_tree_id);
2153 	return err;
2154 }
2155 
2156 const struct got_error *
2157 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2158     struct got_repository *repo)
2159 {
2160 	const struct got_error *err = NULL;
2161 	char *path_objects = NULL, *path = NULL;
2162 	DIR *dir = NULL;
2163 	struct got_object_id id;
2164 	int i;
2165 
2166 	*nobjects = 0;
2167 	*ondisk_size = 0;
2168 
2169 	path_objects = got_repo_get_path_objects(repo);
2170 	if (path_objects == NULL)
2171 		return got_error_from_errno("got_repo_get_path_objects");
2172 
2173 	for (i = 0; i <= 0xff; i++) {
2174 		struct dirent *dent;
2175 
2176 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2177 			err = got_error_from_errno("asprintf");
2178 			break;
2179 		}
2180 
2181 		dir = opendir(path);
2182 		if (dir == NULL) {
2183 			if (errno == ENOENT) {
2184 				err = NULL;
2185 				continue;
2186 			}
2187 			err = got_error_from_errno2("opendir", path);
2188 			break;
2189 		}
2190 
2191 		while ((dent = readdir(dir)) != NULL) {
2192 			char *id_str;
2193 			int fd;
2194 			struct stat sb;
2195 
2196 			if (strcmp(dent->d_name, ".") == 0 ||
2197 			    strcmp(dent->d_name, "..") == 0)
2198 				continue;
2199 
2200 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2201 				err = got_error_from_errno("asprintf");
2202 				goto done;
2203 			}
2204 
2205 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2206 				free(id_str);
2207 				continue;
2208 			}
2209 			free(id_str);
2210 
2211 			err = got_object_open_loose_fd(&fd, &id, repo);
2212 			if (err)
2213 				goto done;
2214 
2215 			if (fstat(fd, &sb) == -1) {
2216 				err = got_error_from_errno("fstat");
2217 				close(fd);
2218 				goto done;
2219 			}
2220 			(*nobjects)++;
2221 			(*ondisk_size) += sb.st_size;
2222 
2223 			if (close(fd) == -1) {
2224 				err = got_error_from_errno("close");
2225 				goto done;
2226 			}
2227 		}
2228 
2229 		if (closedir(dir) != 0) {
2230 			err = got_error_from_errno("closedir");
2231 			goto done;
2232 		}
2233 		dir = NULL;
2234 
2235 		free(path);
2236 		path = NULL;
2237 	}
2238 done:
2239 	if (dir && closedir(dir) != 0 && err == NULL)
2240 		err = got_error_from_errno("closedir");
2241 
2242 	if (err) {
2243 		*nobjects = 0;
2244 		*ondisk_size = 0;
2245 	}
2246 	free(path_objects);
2247 	free(path);
2248 	return err;
2249 }
2250 
2251 const struct got_error *
2252 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2253     off_t *total_packsize, struct got_repository *repo)
2254 {
2255 	const struct got_error *err = NULL;
2256 	DIR *packdir = NULL;
2257 	struct dirent *dent;
2258 	struct got_packidx *packidx = NULL;
2259 	char *path_packidx;
2260 	char *path_packfile;
2261 	int packdir_fd;
2262 	struct stat sb;
2263 
2264 	*npackfiles = 0;
2265 	*nobjects = 0;
2266 	*total_packsize = 0;
2267 
2268 	packdir_fd = openat(got_repo_get_fd(repo),
2269 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2270 	if (packdir_fd == -1) {
2271 		return got_error_from_errno_fmt("openat: %s/%s",
2272 		    got_repo_get_path_git_dir(repo),
2273 		    GOT_OBJECTS_PACK_DIR);
2274 	}
2275 
2276 	packdir = fdopendir(packdir_fd);
2277 	if (packdir == NULL) {
2278 		err = got_error_from_errno("fdopendir");
2279 		goto done;
2280 	}
2281 
2282 	while ((dent = readdir(packdir)) != NULL) {
2283 		if (!got_repo_is_packidx_filename(dent->d_name,
2284 		    strlen(dent->d_name)))
2285 			continue;
2286 
2287 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2288 		    dent->d_name) == -1) {
2289 			err = got_error_from_errno("asprintf");
2290 			goto done;
2291 		}
2292 
2293 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2294 		    path_packidx, 0);
2295 		free(path_packidx);
2296 		if (err)
2297 			goto done;
2298 
2299 		if (fstat(packidx->fd, &sb) == -1)
2300 			goto done;
2301 		*total_packsize += sb.st_size;
2302 
2303 		err = got_packidx_get_packfile_path(&path_packfile,
2304 		    packidx->path_packidx);
2305 		if (err)
2306 			goto done;
2307 
2308 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2309 		    0) == -1) {
2310 			free(path_packfile);
2311 			goto done;
2312 		}
2313 		free(path_packfile);
2314 		*total_packsize += sb.st_size;
2315 
2316 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2317 
2318 		(*npackfiles)++;
2319 
2320 		got_packidx_close(packidx);
2321 		packidx = NULL;
2322 	}
2323 done:
2324 	if (packidx)
2325 		got_packidx_close(packidx);
2326 	if (packdir && closedir(packdir) != 0 && err == NULL)
2327 		err = got_error_from_errno("closedir");
2328 	if (err) {
2329 		*npackfiles = 0;
2330 		*nobjects = 0;
2331 		*total_packsize = 0;
2332 	}
2333 	return err;
2334 }
2335 
2336 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2337     got_packidx_bloom_filter_cmp);