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;
251 
252 	pack_fds_tmp = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(int));
253 	if (pack_fds_tmp == NULL)
254 		return got_error_from_errno("calloc");
255 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
256 	if (*pack_fds == NULL) {
257 		free(pack_fds_tmp);
258 		return got_error_from_errno("calloc");
259 	}
260 
261 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
262 		pack_fds_tmp[i] = got_opentempfd();
263 		if (pack_fds_tmp[i] == -1) {
264 			err = got_error_from_errno("got_opentempfd");
265 			got_repo_pack_fds_close(pack_fds_tmp);
266 			return err;
267 		}
268 	}
269 	memcpy(*pack_fds, pack_fds_tmp, GOT_PACK_NUM_TEMPFILES * sizeof(int));
270 	return err;
271 }
272 
273 const struct got_error *
274 got_repo_pack_fds_close(int *pack_fds)
275 {
276 	const struct got_error *err = NULL;
277 	int i;
278 
279 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
280 		if (pack_fds[i] == -1)
281 			continue;
282 		if (close(pack_fds[i]) == -1) {
283 			err = got_error_from_errno("close");
284 			break;
285 		}
286 	}
287 	free(pack_fds);
288 	return err;
289 }
290 
291 const struct got_error *
292 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
293     struct got_object *obj)
294 {
295 #ifndef GOT_NO_OBJ_CACHE
296 	const struct got_error *err = NULL;
297 	err = got_object_cache_add(&repo->objcache, id, obj);
298 	if (err) {
299 		if (err->code == GOT_ERR_OBJ_EXISTS ||
300 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
301 			err = NULL;
302 		return err;
303 	}
304 	obj->refcnt++;
305 #endif
306 	return NULL;
307 }
308 
309 struct got_object *
310 got_repo_get_cached_object(struct got_repository *repo,
311     struct got_object_id *id)
312 {
313 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
314 }
315 
316 const struct got_error *
317 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
318     struct got_tree_object *tree)
319 {
320 #ifndef GOT_NO_OBJ_CACHE
321 	const struct got_error *err = NULL;
322 	err = got_object_cache_add(&repo->treecache, id, tree);
323 	if (err) {
324 		if (err->code == GOT_ERR_OBJ_EXISTS ||
325 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
326 			err = NULL;
327 		return err;
328 	}
329 	tree->refcnt++;
330 #endif
331 	return NULL;
332 }
333 
334 struct got_tree_object *
335 got_repo_get_cached_tree(struct got_repository *repo,
336     struct got_object_id *id)
337 {
338 	return (struct got_tree_object *)got_object_cache_get(
339 	    &repo->treecache, id);
340 }
341 
342 const struct got_error *
343 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
344     struct got_commit_object *commit)
345 {
346 #ifndef GOT_NO_OBJ_CACHE
347 	const struct got_error *err = NULL;
348 	err = got_object_cache_add(&repo->commitcache, id, commit);
349 	if (err) {
350 		if (err->code == GOT_ERR_OBJ_EXISTS ||
351 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
352 			err = NULL;
353 		return err;
354 	}
355 	commit->refcnt++;
356 #endif
357 	return NULL;
358 }
359 
360 struct got_commit_object *
361 got_repo_get_cached_commit(struct got_repository *repo,
362     struct got_object_id *id)
363 {
364 	return (struct got_commit_object *)got_object_cache_get(
365 	    &repo->commitcache, id);
366 }
367 
368 const struct got_error *
369 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
370     struct got_tag_object *tag)
371 {
372 #ifndef GOT_NO_OBJ_CACHE
373 	const struct got_error *err = NULL;
374 	err = got_object_cache_add(&repo->tagcache, id, tag);
375 	if (err) {
376 		if (err->code == GOT_ERR_OBJ_EXISTS ||
377 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
378 			err = NULL;
379 		return err;
380 	}
381 	tag->refcnt++;
382 #endif
383 	return NULL;
384 }
385 
386 struct got_tag_object *
387 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
388 {
389 	return (struct got_tag_object *)got_object_cache_get(
390 	    &repo->tagcache, id);
391 }
392 
393 const struct got_error *
394 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
395     struct got_raw_object *raw)
396 {
397 #ifndef GOT_NO_OBJ_CACHE
398 	const struct got_error *err = NULL;
399 	err = got_object_cache_add(&repo->rawcache, id, raw);
400 	if (err) {
401 		if (err->code == GOT_ERR_OBJ_EXISTS ||
402 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
403 			err = NULL;
404 		return err;
405 	}
406 	raw->refcnt++;
407 #endif
408 	return NULL;
409 }
410 
411 
412 struct got_raw_object *
413 got_repo_get_cached_raw_object(struct got_repository *repo,
414     struct got_object_id *id)
415 {
416 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
417 }
418 
419 
420 static const struct got_error *
421 open_repo(struct got_repository *repo, const char *path)
422 {
423 	const struct got_error *err = NULL;
424 
425 	repo->gitdir_fd = -1;
426 
427 	/* bare git repository? */
428 	repo->path_git_dir = strdup(path);
429 	if (repo->path_git_dir == NULL)
430 		return got_error_from_errno("strdup");
431 	if (is_git_repo(repo)) {
432 		repo->path = strdup(repo->path_git_dir);
433 		if (repo->path == NULL) {
434 			err = got_error_from_errno("strdup");
435 			goto done;
436 		}
437 		repo->gitdir_fd = open(repo->path_git_dir,
438 		    O_DIRECTORY | O_CLOEXEC);
439 		if (repo->gitdir_fd == -1) {
440 			err = got_error_from_errno2("open",
441 			    repo->path_git_dir);
442 			goto done;
443 		}
444 		return NULL;
445 	}
446 
447 	/* git repository with working tree? */
448 	free(repo->path_git_dir);
449 	repo->path_git_dir = NULL;
450 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
451 		err = got_error_from_errno("asprintf");
452 		goto done;
453 	}
454 	if (is_git_repo(repo)) {
455 		repo->path = strdup(path);
456 		if (repo->path == NULL) {
457 			err = got_error_from_errno("strdup");
458 			goto done;
459 		}
460 		repo->gitdir_fd = open(repo->path_git_dir,
461 		    O_DIRECTORY | O_CLOEXEC);
462 		if (repo->gitdir_fd == -1) {
463 			err = got_error_from_errno2("open",
464 			    repo->path_git_dir);
465 			goto done;
466 		}
467 		return NULL;
468 	}
469 
470 	err = got_error(GOT_ERR_NOT_GIT_REPO);
471 done:
472 	if (err) {
473 		free(repo->path);
474 		repo->path = NULL;
475 		free(repo->path_git_dir);
476 		repo->path_git_dir = NULL;
477 		if (repo->gitdir_fd != -1)
478 			close(repo->gitdir_fd);
479 		repo->gitdir_fd = -1;
480 
481 	}
482 	return err;
483 }
484 
485 static const struct got_error *
486 parse_gitconfig_file(int *gitconfig_repository_format_version,
487     char **gitconfig_author_name, char **gitconfig_author_email,
488     struct got_remote_repo **remotes, int *nremotes,
489     char **gitconfig_owner, char ***extensions, int *nextensions,
490     const char *gitconfig_path)
491 {
492 	const struct got_error *err = NULL, *child_err = NULL;
493 	int fd = -1;
494 	int imsg_fds[2] = { -1, -1 };
495 	pid_t pid;
496 	struct imsgbuf *ibuf;
497 
498 	*gitconfig_repository_format_version = 0;
499 	if (extensions)
500 		*extensions = NULL;
501 	if (nextensions)
502 		*nextensions = 0;
503 	*gitconfig_author_name = NULL;
504 	*gitconfig_author_email = NULL;
505 	if (remotes)
506 		*remotes = NULL;
507 	if (nremotes)
508 		*nremotes = 0;
509 	if (gitconfig_owner)
510 		*gitconfig_owner = NULL;
511 
512 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
513 	if (fd == -1) {
514 		if (errno == ENOENT)
515 			return NULL;
516 		return got_error_from_errno2("open", gitconfig_path);
517 	}
518 
519 	ibuf = calloc(1, sizeof(*ibuf));
520 	if (ibuf == NULL) {
521 		err = got_error_from_errno("calloc");
522 		goto done;
523 	}
524 
525 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
526 		err = got_error_from_errno("socketpair");
527 		goto done;
528 	}
529 
530 	pid = fork();
531 	if (pid == -1) {
532 		err = got_error_from_errno("fork");
533 		goto done;
534 	} else if (pid == 0) {
535 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
536 		    gitconfig_path);
537 		/* not reached */
538 	}
539 
540 	if (close(imsg_fds[1]) == -1) {
541 		err = got_error_from_errno("close");
542 		goto done;
543 	}
544 	imsg_fds[1] = -1;
545 	imsg_init(ibuf, imsg_fds[0]);
546 
547 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
548 	if (err)
549 		goto done;
550 	fd = -1;
551 
552 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
553 	if (err)
554 		goto done;
555 
556 	err = got_privsep_recv_gitconfig_int(
557 	    gitconfig_repository_format_version, ibuf);
558 	if (err)
559 		goto done;
560 
561 	if (extensions && nextensions) {
562 		err = got_privsep_send_gitconfig_repository_extensions_req(
563 		    ibuf);
564 		if (err)
565 			goto done;
566 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
567 		if (err)
568 			goto done;
569 		if (*nextensions > 0) {
570 			int i;
571 			*extensions = calloc(*nextensions, sizeof(char *));
572 			if (*extensions == NULL) {
573 				err = got_error_from_errno("calloc");
574 				goto done;
575 			}
576 			for (i = 0; i < *nextensions; i++) {
577 				char *ext;
578 				err = got_privsep_recv_gitconfig_str(&ext,
579 				    ibuf);
580 				if (err)
581 					goto done;
582 				(*extensions)[i] = ext;
583 			}
584 		}
585 	}
586 
587 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
588 	if (err)
589 		goto done;
590 
591 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
592 	if (err)
593 		goto done;
594 
595 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
596 	if (err)
597 		goto done;
598 
599 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
600 	if (err)
601 		goto done;
602 
603 	if (remotes && nremotes) {
604 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
605 		if (err)
606 			goto done;
607 
608 		err = got_privsep_recv_gitconfig_remotes(remotes,
609 		    nremotes, ibuf);
610 		if (err)
611 			goto done;
612 	}
613 
614 	if (gitconfig_owner) {
615 		err = got_privsep_send_gitconfig_owner_req(ibuf);
616 		if (err)
617 			goto done;
618 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
619 		if (err)
620 			goto done;
621 	}
622 
623 	err = got_privsep_send_stop(imsg_fds[0]);
624 	child_err = got_privsep_wait_for_child(pid);
625 	if (child_err && err == NULL)
626 		err = child_err;
627 done:
628 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
629 		err = got_error_from_errno("close");
630 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
631 		err = got_error_from_errno("close");
632 	if (fd != -1 && close(fd) == -1 && err == NULL)
633 		err = got_error_from_errno2("close", gitconfig_path);
634 	free(ibuf);
635 	return err;
636 }
637 
638 static const struct got_error *
639 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
640 {
641 	const struct got_error *err = NULL;
642 	char *repo_gitconfig_path = NULL;
643 
644 	if (global_gitconfig_path) {
645 		/* Read settings from ~/.gitconfig. */
646 		int dummy_repo_version;
647 		err = parse_gitconfig_file(&dummy_repo_version,
648 		    &repo->global_gitconfig_author_name,
649 		    &repo->global_gitconfig_author_email,
650 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
651 		if (err)
652 			return err;
653 	}
654 
655 	/* Read repository's .git/config file. */
656 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
657 	if (repo_gitconfig_path == NULL)
658 		return got_error_from_errno("got_repo_get_path_gitconfig");
659 
660 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
661 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
662 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
663 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
664 	    repo_gitconfig_path);
665 	if (err)
666 		goto done;
667 done:
668 	free(repo_gitconfig_path);
669 	return err;
670 }
671 
672 static const struct got_error *
673 read_gotconfig(struct got_repository *repo)
674 {
675 	const struct got_error *err = NULL;
676 	char *gotconfig_path;
677 
678 	gotconfig_path = got_repo_get_path_gotconfig(repo);
679 	if (gotconfig_path == NULL)
680 		return got_error_from_errno("got_repo_get_path_gotconfig");
681 
682 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
683 	free(gotconfig_path);
684 	return err;
685 }
686 
687 /* Supported repository format extensions. */
688 static const char *const repo_extensions[] = {
689 	"noop",			/* Got supports repository format version 1. */
690 	"preciousObjects",	/* Supported by gotadmin cleanup. */
691 	"worktreeConfig",	/* Got does not care about Git work trees. */
692 };
693 
694 const struct got_error *
695 got_repo_open(struct got_repository **repop, const char *path,
696     const char *global_gitconfig_path, int *pack_fds)
697 {
698 	struct got_repository *repo = NULL;
699 	const struct got_error *err = NULL;
700 	char *repo_path = NULL;
701 	size_t i, j = 0;
702 	struct rlimit rl;
703 
704 	*repop = NULL;
705 
706 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
707 		return got_error_from_errno("getrlimit");
708 
709 	repo = calloc(1, sizeof(*repo));
710 	if (repo == NULL)
711 		return got_error_from_errno("calloc");
712 
713 	RB_INIT(&repo->packidx_bloom_filters);
714 	TAILQ_INIT(&repo->packidx_paths);
715 
716 	for (i = 0; i < nitems(repo->privsep_children); i++) {
717 		memset(&repo->privsep_children[i], 0,
718 		    sizeof(repo->privsep_children[0]));
719 		repo->privsep_children[i].imsg_fd = -1;
720 	}
721 
722 	err = got_object_cache_init(&repo->objcache,
723 	    GOT_OBJECT_CACHE_TYPE_OBJ);
724 	if (err)
725 		goto done;
726 	err = got_object_cache_init(&repo->treecache,
727 	    GOT_OBJECT_CACHE_TYPE_TREE);
728 	if (err)
729 		goto done;
730 	err = got_object_cache_init(&repo->commitcache,
731 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
732 	if (err)
733 		goto done;
734 	err = got_object_cache_init(&repo->tagcache,
735 	    GOT_OBJECT_CACHE_TYPE_TAG);
736 	if (err)
737 		goto done;
738 	err = got_object_cache_init(&repo->rawcache,
739 	    GOT_OBJECT_CACHE_TYPE_RAW);
740 	if (err)
741 		goto done;
742 
743 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
744 	if (repo->pack_cache_size > rl.rlim_cur / 8)
745 		repo->pack_cache_size = rl.rlim_cur / 8;
746 	for (i = 0; i < nitems(repo->packs); i++) {
747 		if (i < repo->pack_cache_size) {
748 			repo->packs[i].basefd = pack_fds[j++];
749 			repo->packs[i].accumfd = pack_fds[j++];
750 		} else {
751 			repo->packs[i].basefd = -1;
752 			repo->packs[i].accumfd = -1;
753 		}
754 	}
755 	repo->pinned_pack = -1;
756 	repo->pinned_packidx = -1;
757 	repo->pinned_pid = 0;
758 
759 	repo_path = realpath(path, NULL);
760 	if (repo_path == NULL) {
761 		err = got_error_from_errno2("realpath", path);
762 		goto done;
763 	}
764 
765 	for (;;) {
766 		char *parent_path;
767 
768 		err = open_repo(repo, repo_path);
769 		if (err == NULL)
770 			break;
771 		if (err->code != GOT_ERR_NOT_GIT_REPO)
772 			goto done;
773 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
774 			err = got_error(GOT_ERR_NOT_GIT_REPO);
775 			goto done;
776 		}
777 		err = got_path_dirname(&parent_path, repo_path);
778 		if (err)
779 			goto done;
780 		free(repo_path);
781 		repo_path = parent_path;
782 	}
783 
784 	err = read_gotconfig(repo);
785 	if (err)
786 		goto done;
787 
788 	err = read_gitconfig(repo, global_gitconfig_path);
789 	if (err)
790 		goto done;
791 	if (repo->gitconfig_repository_format_version != 0) {
792 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
793 		goto done;
794 	}
795 	for (i = 0; i < repo->nextensions; i++) {
796 		char *ext = repo->extensions[i];
797 		int j, supported = 0;
798 		for (j = 0; j < nitems(repo_extensions); j++) {
799 			if (strcmp(ext, repo_extensions[j]) == 0) {
800 				supported = 1;
801 				break;
802 			}
803 		}
804 		if (!supported) {
805 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
806 			goto done;
807 		}
808 	}
809 
810 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
811 done:
812 	if (err)
813 		got_repo_close(repo);
814 	else
815 		*repop = repo;
816 	free(repo_path);
817 	return err;
818 }
819 
820 const struct got_error *
821 got_repo_close(struct got_repository *repo)
822 {
823 	const struct got_error *err = NULL, *child_err;
824 	struct got_packidx_bloom_filter *bf;
825 	struct got_pathlist_entry *pe;
826 	size_t i;
827 
828 	for (i = 0; i < repo->pack_cache_size; i++) {
829 		if (repo->packidx_cache[i] == NULL)
830 			break;
831 		got_packidx_close(repo->packidx_cache[i]);
832 	}
833 
834 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
835 	    &repo->packidx_bloom_filters))) {
836 		RB_REMOVE(got_packidx_bloom_filter_tree,
837 		    &repo->packidx_bloom_filters, bf);
838 		free(bf->bloom);
839 		free(bf);
840 	}
841 
842 	for (i = 0; i < repo->pack_cache_size; i++)
843 		if (repo->packs[i].path_packfile)
844 			if (repo->packs[i].path_packfile)
845 				got_pack_close(&repo->packs[i]);
846 
847 	free(repo->path);
848 	free(repo->path_git_dir);
849 
850 	got_object_cache_close(&repo->objcache);
851 	got_object_cache_close(&repo->treecache);
852 	got_object_cache_close(&repo->commitcache);
853 	got_object_cache_close(&repo->tagcache);
854 	got_object_cache_close(&repo->rawcache);
855 
856 	for (i = 0; i < nitems(repo->privsep_children); i++) {
857 		if (repo->privsep_children[i].imsg_fd == -1)
858 			continue;
859 		imsg_clear(repo->privsep_children[i].ibuf);
860 		free(repo->privsep_children[i].ibuf);
861 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
862 		child_err = got_privsep_wait_for_child(
863 		    repo->privsep_children[i].pid);
864 		if (child_err && err == NULL)
865 			err = child_err;
866 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
867 		    err == NULL)
868 			err = got_error_from_errno("close");
869 	}
870 
871 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
872 	    err == NULL)
873 		err = got_error_from_errno("close");
874 
875 	if (repo->gotconfig)
876 		got_gotconfig_free(repo->gotconfig);
877 	free(repo->gitconfig_author_name);
878 	free(repo->gitconfig_author_email);
879 	for (i = 0; i < repo->ngitconfig_remotes; i++)
880 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
881 	free(repo->gitconfig_remotes);
882 	for (i = 0; i < repo->nextensions; i++)
883 		free(repo->extensions[i]);
884 	free(repo->extensions);
885 
886 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
887 		free((void *)pe->path);
888 	got_pathlist_free(&repo->packidx_paths);
889 	free(repo);
890 
891 	return err;
892 }
893 
894 void
895 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
896 {
897 	int i;
898 
899 	free(repo->name);
900 	repo->name = NULL;
901 	free(repo->fetch_url);
902 	repo->fetch_url = NULL;
903 	free(repo->send_url);
904 	repo->send_url = NULL;
905 	for (i = 0; i < repo->nfetch_branches; i++)
906 		free(repo->fetch_branches[i]);
907 	free(repo->fetch_branches);
908 	repo->fetch_branches = NULL;
909 	repo->nfetch_branches = 0;
910 	for (i = 0; i < repo->nsend_branches; i++)
911 		free(repo->send_branches[i]);
912 	free(repo->send_branches);
913 	repo->send_branches = NULL;
914 	repo->nsend_branches = 0;
915 }
916 
917 const struct got_error *
918 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
919     const char *input_path)
920 {
921 	const struct got_error *err = NULL;
922 	const char *repo_abspath = NULL;
923 	size_t repolen, len;
924 	char *canonpath, *path = NULL;
925 
926 	*in_repo_path = NULL;
927 
928 	canonpath = strdup(input_path);
929 	if (canonpath == NULL) {
930 		err = got_error_from_errno("strdup");
931 		goto done;
932 	}
933 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
934 	if (err)
935 		goto done;
936 
937 	repo_abspath = got_repo_get_path(repo);
938 
939 	if (canonpath[0] == '\0') {
940 		path = strdup(canonpath);
941 		if (path == NULL) {
942 			err = got_error_from_errno("strdup");
943 			goto done;
944 		}
945 	} else {
946 		path = realpath(canonpath, NULL);
947 		if (path == NULL) {
948 			if (errno != ENOENT) {
949 				err = got_error_from_errno2("realpath",
950 				    canonpath);
951 				goto done;
952 			}
953 			/*
954 			 * Path is not on disk.
955 			 * Assume it is already relative to repository root.
956 			 */
957 			path = strdup(canonpath);
958 			if (path == NULL) {
959 				err = got_error_from_errno("strdup");
960 				goto done;
961 			}
962 		}
963 
964 		repolen = strlen(repo_abspath);
965 		len = strlen(path);
966 
967 
968 		if (strcmp(path, repo_abspath) == 0) {
969 			free(path);
970 			path = strdup("");
971 			if (path == NULL) {
972 				err = got_error_from_errno("strdup");
973 				goto done;
974 			}
975 		} else if (len > repolen &&
976 		    got_path_is_child(path, repo_abspath, repolen)) {
977 			/* Matched an on-disk path inside repository. */
978 			if (got_repo_is_bare(repo)) {
979 				/*
980 				 * Matched an on-disk path inside repository
981 				 * database. Treat input as repository-relative.
982 				 */
983 				free(path);
984 				path = canonpath;
985 				canonpath = NULL;
986 			} else {
987 				char *child;
988 				/* Strip common prefix with repository path. */
989 				err = got_path_skip_common_ancestor(&child,
990 				    repo_abspath, path);
991 				if (err)
992 					goto done;
993 				free(path);
994 				path = child;
995 			}
996 		} else {
997 			/*
998 			 * Matched unrelated on-disk path.
999 			 * Treat input as repository-relative.
1000 			 */
1001 			free(path);
1002 			path = canonpath;
1003 			canonpath = NULL;
1004 		}
1005 	}
1006 
1007 	/* Make in-repository path absolute */
1008 	if (path[0] != '/') {
1009 		char *abspath;
1010 		if (asprintf(&abspath, "/%s", path) == -1) {
1011 			err = got_error_from_errno("asprintf");
1012 			goto done;
1013 		}
1014 		free(path);
1015 		path = abspath;
1016 	}
1017 
1018 done:
1019 	free(canonpath);
1020 	if (err)
1021 		free(path);
1022 	else
1023 		*in_repo_path = path;
1024 	return err;
1025 }
1026 
1027 static const struct got_error *
1028 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1029     const char *path_packidx)
1030 {
1031 	const struct got_error *err = NULL;
1032 	size_t i;
1033 
1034 	for (i = 0; i < repo->pack_cache_size; i++) {
1035 		if (repo->packidx_cache[i] == NULL)
1036 			break;
1037 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1038 		    path_packidx) == 0) {
1039 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1040 		}
1041 	}
1042 	if (i == repo->pack_cache_size) {
1043 		do {
1044 			i--;
1045 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1046 		    i == repo->pinned_packidx);
1047 		err = got_packidx_close(repo->packidx_cache[i]);
1048 		if (err)
1049 			return err;
1050 	}
1051 
1052 	repo->packidx_cache[i] = packidx;
1053 
1054 	return NULL;
1055 }
1056 
1057 int
1058 got_repo_is_packidx_filename(const char *name, size_t len)
1059 {
1060 	if (len != GOT_PACKIDX_NAMELEN)
1061 		return 0;
1062 
1063 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1064 		return 0;
1065 
1066 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1067 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1068 		return 0;
1069 
1070 	return 1;
1071 }
1072 
1073 static struct got_packidx_bloom_filter *
1074 get_packidx_bloom_filter(struct got_repository *repo,
1075     const char *path, size_t path_len)
1076 {
1077 	struct got_packidx_bloom_filter key;
1078 
1079 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1080 		return NULL; /* XXX */
1081 	key.path_len = path_len;
1082 
1083 	return RB_FIND(got_packidx_bloom_filter_tree,
1084 	    &repo->packidx_bloom_filters, &key);
1085 }
1086 
1087 int
1088 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1089     const char *path_packidx, struct got_object_id *id)
1090 {
1091 	struct got_packidx_bloom_filter *bf;
1092 
1093 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1094 	if (bf)
1095 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1096 
1097 	/* No bloom filter means this pack index must be searched. */
1098 	return 1;
1099 }
1100 
1101 static const struct got_error *
1102 add_packidx_bloom_filter(struct got_repository *repo,
1103     struct got_packidx *packidx, const char *path_packidx)
1104 {
1105 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1106 	struct got_packidx_bloom_filter *bf;
1107 	size_t len;
1108 
1109 	/*
1110 	 * Don't use bloom filters for very large pack index files.
1111 	 * Large pack files will contain a relatively large fraction
1112 	 * of our objects so we will likely need to visit them anyway.
1113 	 * The more objects a pack file contains the higher the probability
1114 	 * of a false-positive match from the bloom filter. And reading
1115 	 * all object IDs from a large pack index file can be expensive.
1116 	 */
1117 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1118 		return NULL;
1119 
1120 	/* Do we already have a filter for this pack index? */
1121 	if (get_packidx_bloom_filter(repo, path_packidx,
1122 	    strlen(path_packidx)) != NULL)
1123 		return NULL;
1124 
1125 	bf = calloc(1, sizeof(*bf));
1126 	if (bf == NULL)
1127 		return got_error_from_errno("calloc");
1128 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1129 	if (bf->bloom == NULL) {
1130 		free(bf);
1131 		return got_error_from_errno("calloc");
1132 	}
1133 	
1134 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1135 	if (len >= sizeof(bf->path)) {
1136 		free(bf->bloom);
1137 		free(bf);
1138 		return got_error(GOT_ERR_NO_SPACE);
1139 	}
1140 	bf->path_len = len;
1141 
1142 	/* Minimum size supported by our bloom filter is 1000 entries. */
1143 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1144 	for (i = 0; i < nobjects; i++) {
1145 		struct got_packidx_object_id *id;
1146 		id = &packidx->hdr.sorted_ids[i];
1147 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1148 	}
1149 
1150 	RB_INSERT(got_packidx_bloom_filter_tree,
1151 	    &repo->packidx_bloom_filters, bf);
1152 	return NULL;
1153 }
1154 
1155 const struct got_error *
1156 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1157     struct got_repository *repo, struct got_object_id *id)
1158 {
1159 	const struct got_error *err;
1160 	struct got_pathlist_entry *pe;
1161 	size_t i;
1162 
1163 	/* Search pack index cache. */
1164 	for (i = 0; i < repo->pack_cache_size; i++) {
1165 		if (repo->packidx_cache[i] == NULL)
1166 			break;
1167 		if (!got_repo_check_packidx_bloom_filter(repo,
1168 		    repo->packidx_cache[i]->path_packidx, id))
1169 			continue; /* object will not be found in this index */
1170 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1171 		if (*idx != -1) {
1172 			*packidx = repo->packidx_cache[i];
1173 			/*
1174 			 * Move this cache entry to the front. Repeatedly
1175 			 * searching a wrong pack index can be expensive.
1176 			 */
1177 			if (i > 0) {
1178 				memmove(&repo->packidx_cache[1],
1179 				    &repo->packidx_cache[0],
1180 				    i * sizeof(repo->packidx_cache[0]));
1181 				repo->packidx_cache[0] = *packidx;
1182 				if (repo->pinned_packidx >= 0 &&
1183 				    repo->pinned_packidx < i)
1184 					repo->pinned_packidx++;
1185 				else if (repo->pinned_packidx == i)
1186 					repo->pinned_packidx = 0;
1187 			}
1188 			return NULL;
1189 		}
1190 	}
1191 	/* No luck. Search the filesystem. */
1192 
1193 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1194 		const char *path_packidx = pe->path;
1195 		int is_cached = 0;
1196 
1197 		if (!got_repo_check_packidx_bloom_filter(repo,
1198 		    pe->path, id))
1199 			continue; /* object will not be found in this index */
1200 
1201 		for (i = 0; i < repo->pack_cache_size; i++) {
1202 			if (repo->packidx_cache[i] == NULL)
1203 				break;
1204 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1205 			    path_packidx) == 0) {
1206 				is_cached = 1;
1207 				break;
1208 			}
1209 		}
1210 		if (is_cached)
1211 			continue; /* already searched */
1212 
1213 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1214 		    path_packidx, 0);
1215 		if (err)
1216 			goto done;
1217 
1218 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1219 		if (err)
1220 			goto done;
1221 
1222 		err = cache_packidx(repo, *packidx, path_packidx);
1223 		if (err)
1224 			goto done;
1225 
1226 		*idx = got_packidx_get_object_idx(*packidx, id);
1227 		if (*idx != -1) {
1228 			err = NULL; /* found the object */
1229 			goto done;
1230 		}
1231 	}
1232 
1233 	err = got_error_no_obj(id);
1234 done:
1235 	return err;
1236 }
1237 
1238 const struct got_error *
1239 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1240     struct got_repository *repo)
1241 {
1242 	const struct got_error *err = NULL;
1243 	DIR *packdir = NULL;
1244 	struct dirent *dent;
1245 	char *path_packidx = NULL;
1246 	int packdir_fd;
1247 
1248 	packdir_fd = openat(got_repo_get_fd(repo),
1249 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1250 	if (packdir_fd == -1) {
1251 		return got_error_from_errno_fmt("openat: %s/%s",
1252 		    got_repo_get_path_git_dir(repo),
1253 		    GOT_OBJECTS_PACK_DIR);
1254 	}
1255 
1256 	packdir = fdopendir(packdir_fd);
1257 	if (packdir == NULL) {
1258 		err = got_error_from_errno("fdopendir");
1259 		goto done;
1260 	}
1261 
1262 	while ((dent = readdir(packdir)) != NULL) {
1263 		if (!got_repo_is_packidx_filename(dent->d_name,
1264 		    strlen(dent->d_name)))
1265 			continue;
1266 
1267 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1268 		    dent->d_name) == -1) {
1269 			err = got_error_from_errno("asprintf");
1270 			path_packidx = NULL;
1271 			break;
1272 		}
1273 
1274 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1275 		if (err)
1276 			break;
1277 	}
1278 done:
1279 	if (err)
1280 		free(path_packidx);
1281 	if (packdir && closedir(packdir) != 0 && err == NULL)
1282 		err = got_error_from_errno("closedir");
1283 	return err;
1284 }
1285 
1286 const struct got_error *
1287 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1288     struct got_repository *repo)
1289 {
1290 	const struct got_error *err;
1291 	size_t i;
1292 
1293 	*packidx = NULL;
1294 
1295 	/* Search pack index cache. */
1296 	for (i = 0; i < repo->pack_cache_size; i++) {
1297 		if (repo->packidx_cache[i] == NULL)
1298 			break;
1299 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1300 		    path_packidx) == 0) {
1301 			*packidx = repo->packidx_cache[i];
1302 			return NULL;
1303 		}
1304 	}
1305 	/* No luck. Search the filesystem. */
1306 
1307 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1308 	    path_packidx, 0);
1309 	if (err)
1310 		return err;
1311 
1312 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1313 	if (err)
1314 		goto done;
1315 
1316 	err = cache_packidx(repo, *packidx, path_packidx);
1317 done:
1318 	if (err) {
1319 		got_packidx_close(*packidx);
1320 		*packidx = NULL;
1321 	}
1322 	return err;
1323 }
1324 
1325 static const struct got_error *
1326 read_packfile_hdr(int fd, struct got_packidx *packidx)
1327 {
1328 	const struct got_error *err = NULL;
1329 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1330 	struct got_packfile_hdr hdr;
1331 	ssize_t n;
1332 
1333 	n = read(fd, &hdr, sizeof(hdr));
1334 	if (n < 0)
1335 		return got_error_from_errno("read");
1336 	if (n != sizeof(hdr))
1337 		return got_error(GOT_ERR_BAD_PACKFILE);
1338 
1339 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1340 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1341 	    be32toh(hdr.nobjects) != totobj)
1342 		err = got_error(GOT_ERR_BAD_PACKFILE);
1343 
1344 	return err;
1345 }
1346 
1347 static const struct got_error *
1348 open_packfile(int *fd, struct got_repository *repo,
1349     const char *relpath, struct got_packidx *packidx)
1350 {
1351 	const struct got_error *err = NULL;
1352 
1353 	*fd = openat(got_repo_get_fd(repo), relpath,
1354 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1355 	if (*fd == -1)
1356 		return got_error_from_errno_fmt("openat: %s/%s",
1357 		    got_repo_get_path_git_dir(repo), relpath);
1358 
1359 	if (packidx) {
1360 		err = read_packfile_hdr(*fd, packidx);
1361 		if (err) {
1362 			close(*fd);
1363 			*fd = -1;
1364 		}
1365 	}
1366 
1367 	return err;
1368 }
1369 
1370 const struct got_error *
1371 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1372     const char *path_packfile, struct got_packidx *packidx)
1373 {
1374 	const struct got_error *err = NULL;
1375 	struct got_pack *pack = NULL;
1376 	struct stat sb;
1377 	size_t i;
1378 
1379 	if (packp)
1380 		*packp = NULL;
1381 
1382 	for (i = 0; i < repo->pack_cache_size; i++) {
1383 		pack = &repo->packs[i];
1384 		if (pack->path_packfile == NULL)
1385 			break;
1386 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1387 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1388 	}
1389 
1390 	if (i == repo->pack_cache_size) {
1391 		struct got_pack tmp;
1392 		do {
1393 			i--;
1394 		} while (i > 0 && repo->pinned_pack >= 0 &&
1395 		    i == repo->pinned_pack);
1396 		err = got_pack_close(&repo->packs[i]);
1397 		if (err)
1398 			return err;
1399 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1400 			return got_error_from_errno("ftruncate");
1401 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1402 			return got_error_from_errno("ftruncate");
1403 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1404 		memcpy(&repo->packs[i], &repo->packs[0],
1405 		    sizeof(repo->packs[i]));
1406 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1407 		if (repo->pinned_pack == 0)
1408 			repo->pinned_pack = i;
1409 		else if (repo->pinned_pack == i)
1410 			repo->pinned_pack = 0;
1411 		i = 0;
1412 	}
1413 
1414 	pack = &repo->packs[i];
1415 
1416 	pack->path_packfile = strdup(path_packfile);
1417 	if (pack->path_packfile == NULL) {
1418 		err = got_error_from_errno("strdup");
1419 		goto done;
1420 	}
1421 
1422 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1423 	if (err)
1424 		goto done;
1425 
1426 	if (fstat(pack->fd, &sb) != 0) {
1427 		err = got_error_from_errno("fstat");
1428 		goto done;
1429 	}
1430 	pack->filesize = sb.st_size;
1431 
1432 	pack->privsep_child = NULL;
1433 
1434 #ifndef GOT_PACK_NO_MMAP
1435 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1436 	    pack->fd, 0);
1437 	if (pack->map == MAP_FAILED) {
1438 		if (errno != ENOMEM) {
1439 			err = got_error_from_errno("mmap");
1440 			goto done;
1441 		}
1442 		pack->map = NULL; /* fall back to read(2) */
1443 	}
1444 #endif
1445 done:
1446 	if (err) {
1447 		if (pack) {
1448 			free(pack->path_packfile);
1449 			memset(pack, 0, sizeof(*pack));
1450 		}
1451 	} else if (packp)
1452 		*packp = pack;
1453 	return err;
1454 }
1455 
1456 struct got_pack *
1457 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1458 {
1459 	struct got_pack *pack = NULL;
1460 	size_t i;
1461 
1462 	for (i = 0; i < repo->pack_cache_size; i++) {
1463 		pack = &repo->packs[i];
1464 		if (pack->path_packfile == NULL)
1465 			break;
1466 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1467 			return pack;
1468 	}
1469 
1470 	return NULL;
1471 }
1472 
1473 const struct got_error *
1474 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1475     struct got_pack *pack)
1476 {
1477 	size_t i;
1478 	int pinned_pack = -1, pinned_packidx = -1;
1479 
1480 	for (i = 0; i < repo->pack_cache_size; i++) {
1481 		if (repo->packidx_cache[i] &&
1482 		    strcmp(repo->packidx_cache[i]->path_packidx,
1483 		    packidx->path_packidx) == 0)
1484 			pinned_packidx = i;
1485 		if (repo->packs[i].path_packfile &&
1486 		    strcmp(repo->packs[i].path_packfile,
1487 		    pack->path_packfile) == 0)
1488 			pinned_pack = i;
1489 	}
1490 
1491 	if (pinned_packidx == -1 || pinned_pack == -1)
1492 		return got_error(GOT_ERR_PIN_PACK);
1493 
1494 	repo->pinned_pack = pinned_pack;
1495 	repo->pinned_packidx = pinned_packidx;
1496 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1497 	return NULL;
1498 }
1499 
1500 struct got_pack *
1501 got_repo_get_pinned_pack(struct got_repository *repo)
1502 {
1503 	if (repo->pinned_pack >= 0 &&
1504 	    repo->pinned_pack < repo->pack_cache_size)
1505 		return &repo->packs[repo->pinned_pack];
1506 
1507 	return NULL;
1508 }
1509 
1510 void
1511 got_repo_unpin_pack(struct got_repository *repo)
1512 {
1513 	repo->pinned_packidx = -1;
1514 	repo->pinned_pack = -1;
1515 	repo->pinned_pid = 0;
1516 }
1517 
1518 const struct got_error *
1519 got_repo_init(const char *repo_path)
1520 {
1521 	const struct got_error *err = NULL;
1522 	const char *dirnames[] = {
1523 		GOT_OBJECTS_DIR,
1524 		GOT_OBJECTS_PACK_DIR,
1525 		GOT_REFS_DIR,
1526 	};
1527 	const char *description_str = "Unnamed repository; "
1528 	    "edit this file 'description' to name the repository.";
1529 	const char *headref_str = "ref: refs/heads/main";
1530 	const char *gitconfig_str = "[core]\n"
1531 	    "\trepositoryformatversion = 0\n"
1532 	    "\tfilemode = true\n"
1533 	    "\tbare = true\n";
1534 	char *path;
1535 	size_t i;
1536 
1537 	if (!got_path_dir_is_empty(repo_path))
1538 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1539 
1540 	for (i = 0; i < nitems(dirnames); i++) {
1541 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1542 			return got_error_from_errno("asprintf");
1543 		}
1544 		err = got_path_mkdir(path);
1545 		free(path);
1546 		if (err)
1547 			return err;
1548 	}
1549 
1550 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1551 		return got_error_from_errno("asprintf");
1552 	err = got_path_create_file(path, description_str);
1553 	free(path);
1554 	if (err)
1555 		return err;
1556 
1557 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1558 		return got_error_from_errno("asprintf");
1559 	err = got_path_create_file(path, headref_str);
1560 	free(path);
1561 	if (err)
1562 		return err;
1563 
1564 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1565 		return got_error_from_errno("asprintf");
1566 	err = got_path_create_file(path, gitconfig_str);
1567 	free(path);
1568 	if (err)
1569 		return err;
1570 
1571 	return NULL;
1572 }
1573 
1574 static const struct got_error *
1575 match_packed_object(struct got_object_id **unique_id,
1576     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1577 {
1578 	const struct got_error *err = NULL;
1579 	struct got_object_id_queue matched_ids;
1580 	struct got_pathlist_entry *pe;
1581 
1582 	STAILQ_INIT(&matched_ids);
1583 
1584 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1585 		const char *path_packidx = pe->path;
1586 		struct got_packidx *packidx;
1587 		struct got_object_qid *qid;
1588 
1589 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1590 		    path_packidx, 0);
1591 		if (err)
1592 			break;
1593 
1594 		err = got_packidx_match_id_str_prefix(&matched_ids,
1595 		    packidx, id_str_prefix);
1596 		if (err) {
1597 			got_packidx_close(packidx);
1598 			break;
1599 		}
1600 		err = got_packidx_close(packidx);
1601 		if (err)
1602 			break;
1603 
1604 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1605 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1606 				int matched_type;
1607 				err = got_object_get_type(&matched_type, repo,
1608 				    &qid->id);
1609 				if (err)
1610 					goto done;
1611 				if (matched_type != obj_type)
1612 					continue;
1613 			}
1614 			if (*unique_id == NULL) {
1615 				*unique_id = got_object_id_dup(&qid->id);
1616 				if (*unique_id == NULL) {
1617 					err = got_error_from_errno("malloc");
1618 					goto done;
1619 				}
1620 			} else {
1621 				if (got_object_id_cmp(*unique_id,
1622 				    &qid->id) == 0)
1623 					continue; /* packed multiple times */
1624 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1625 				goto done;
1626 			}
1627 		}
1628 	}
1629 done:
1630 	got_object_id_queue_free(&matched_ids);
1631 	if (err) {
1632 		free(*unique_id);
1633 		*unique_id = NULL;
1634 	}
1635 	return err;
1636 }
1637 
1638 static const struct got_error *
1639 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1640     const char *object_dir, const char *id_str_prefix, int obj_type,
1641     struct got_repository *repo)
1642 {
1643 	const struct got_error *err = NULL;
1644 	char *path;
1645 	DIR *dir = NULL;
1646 	struct dirent *dent;
1647 	struct got_object_id id;
1648 
1649 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1650 		err = got_error_from_errno("asprintf");
1651 		goto done;
1652 	}
1653 
1654 	dir = opendir(path);
1655 	if (dir == NULL) {
1656 		if (errno == ENOENT) {
1657 			err = NULL;
1658 			goto done;
1659 		}
1660 		err = got_error_from_errno2("opendir", path);
1661 		goto done;
1662 	}
1663 	while ((dent = readdir(dir)) != NULL) {
1664 		char *id_str;
1665 		int cmp;
1666 
1667 		if (strcmp(dent->d_name, ".") == 0 ||
1668 		    strcmp(dent->d_name, "..") == 0)
1669 			continue;
1670 
1671 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1672 			err = got_error_from_errno("asprintf");
1673 			goto done;
1674 		}
1675 
1676 		if (!got_parse_sha1_digest(id.sha1, id_str))
1677 			continue;
1678 
1679 		/*
1680 		 * Directory entries do not necessarily appear in
1681 		 * sorted order, so we must iterate over all of them.
1682 		 */
1683 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1684 		if (cmp != 0) {
1685 			free(id_str);
1686 			continue;
1687 		}
1688 
1689 		if (*unique_id == NULL) {
1690 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1691 				int matched_type;
1692 				err = got_object_get_type(&matched_type, repo,
1693 				    &id);
1694 				if (err)
1695 					goto done;
1696 				if (matched_type != obj_type)
1697 					continue;
1698 			}
1699 			*unique_id = got_object_id_dup(&id);
1700 			if (*unique_id == NULL) {
1701 				err = got_error_from_errno("got_object_id_dup");
1702 				free(id_str);
1703 				goto done;
1704 			}
1705 		} else {
1706 			if (got_object_id_cmp(*unique_id, &id) == 0)
1707 				continue; /* both packed and loose */
1708 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1709 			free(id_str);
1710 			goto done;
1711 		}
1712 	}
1713 done:
1714 	if (dir && closedir(dir) != 0 && err == NULL)
1715 		err = got_error_from_errno("closedir");
1716 	if (err) {
1717 		free(*unique_id);
1718 		*unique_id = NULL;
1719 	}
1720 	free(path);
1721 	return err;
1722 }
1723 
1724 const struct got_error *
1725 got_repo_match_object_id_prefix(struct got_object_id **id,
1726     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1727 {
1728 	const struct got_error *err = NULL;
1729 	char *path_objects = got_repo_get_path_objects(repo);
1730 	char *object_dir = NULL;
1731 	size_t len;
1732 	int i;
1733 
1734 	*id = NULL;
1735 
1736 	len = strlen(id_str_prefix);
1737 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1738 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1739 
1740 	for (i = 0; i < len; i++) {
1741 		if (isxdigit((unsigned char)id_str_prefix[i]))
1742 			continue;
1743 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1744 	}
1745 
1746 	if (len >= 2) {
1747 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1748 		if (err)
1749 			goto done;
1750 		object_dir = strndup(id_str_prefix, 2);
1751 		if (object_dir == NULL) {
1752 			err = got_error_from_errno("strdup");
1753 			goto done;
1754 		}
1755 		err = match_loose_object(id, path_objects, object_dir,
1756 		    id_str_prefix, obj_type, repo);
1757 	} else if (len == 1) {
1758 		int i;
1759 		for (i = 0; i < 0xf; i++) {
1760 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1761 			    == -1) {
1762 				err = got_error_from_errno("asprintf");
1763 				goto done;
1764 			}
1765 			err = match_packed_object(id, repo, object_dir,
1766 			    obj_type);
1767 			if (err)
1768 				goto done;
1769 			err = match_loose_object(id, path_objects, object_dir,
1770 			    id_str_prefix, obj_type, repo);
1771 			if (err)
1772 				goto done;
1773 		}
1774 	} else {
1775 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1776 		goto done;
1777 	}
1778 done:
1779 	free(object_dir);
1780 	if (err) {
1781 		free(*id);
1782 		*id = NULL;
1783 	} else if (*id == NULL) {
1784 		switch (obj_type) {
1785 		case GOT_OBJ_TYPE_BLOB:
1786 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1787 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1788 			break;
1789 		case GOT_OBJ_TYPE_TREE:
1790 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1791 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1792 			break;
1793 		case GOT_OBJ_TYPE_COMMIT:
1794 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1795 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1796 			break;
1797 		case GOT_OBJ_TYPE_TAG:
1798 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1799 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1800 			break;
1801 		default:
1802 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1803 			break;
1804 		}
1805 	}
1806 
1807 	return err;
1808 }
1809 
1810 const struct got_error *
1811 got_repo_match_object_id(struct got_object_id **id, char **label,
1812     const char *id_str, int obj_type, struct got_reflist_head *refs,
1813     struct got_repository *repo)
1814 {
1815 	const struct got_error *err;
1816 	struct got_tag_object *tag;
1817 	struct got_reference *ref = NULL;
1818 
1819 	*id = NULL;
1820 	if (label)
1821 		*label = NULL;
1822 
1823 	if (refs) {
1824 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1825 		    refs, repo);
1826 		if (err == NULL) {
1827 			*id = got_object_id_dup(
1828 			    got_object_tag_get_object_id(tag));
1829 			if (*id == NULL)
1830 				err = got_error_from_errno("got_object_id_dup");
1831 			else if (label && asprintf(label, "refs/tags/%s",
1832 			    got_object_tag_get_name(tag)) == -1) {
1833 				err = got_error_from_errno("asprintf");
1834 				free(*id);
1835 				*id = NULL;
1836 			}
1837 			got_object_tag_close(tag);
1838 			return err;
1839 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1840 		    err->code != GOT_ERR_NO_OBJ)
1841 			return err;
1842 	}
1843 
1844 	err = got_ref_open(&ref, repo, id_str, 0);
1845 	if (err == NULL) {
1846 		err = got_ref_resolve(id, repo, ref);
1847 		if (err)
1848 			goto done;
1849 		if (label) {
1850 			*label = strdup(got_ref_get_name(ref));
1851 			if (*label == NULL) {
1852 				err = got_error_from_errno("strdup");
1853 				goto done;
1854 			}
1855 		}
1856 	} else {
1857 		if (err->code != GOT_ERR_NOT_REF &&
1858 		    err->code != GOT_ERR_BAD_REF_NAME)
1859 			goto done;
1860 		err = got_repo_match_object_id_prefix(id, id_str,
1861 		    obj_type, repo);
1862 		if (err) {
1863 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1864 				err = got_error_not_ref(id_str);
1865 			goto done;
1866 		}
1867 		if (label) {
1868 			err = got_object_id_str(label, *id);
1869 			if (*label == NULL) {
1870 				err = got_error_from_errno("strdup");
1871 				goto done;
1872 			}
1873 		}
1874 	}
1875 done:
1876 	if (ref)
1877 		got_ref_close(ref);
1878 	return err;
1879 }
1880 
1881 const struct got_error *
1882 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1883     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1884 {
1885 	const struct got_error *err = NULL;
1886 	struct got_reflist_entry *re;
1887 	struct got_object_id *tag_id;
1888 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1889 
1890 	*tag = NULL;
1891 
1892 	TAILQ_FOREACH(re, refs, entry) {
1893 		const char *refname;
1894 		refname = got_ref_get_name(re->ref);
1895 		if (got_ref_is_symbolic(re->ref))
1896 			continue;
1897 		if (strncmp(refname, "refs/tags/", 10) != 0)
1898 			continue;
1899 		if (!name_is_absolute)
1900 			refname += strlen("refs/tags/");
1901 		if (strcmp(refname, name) != 0)
1902 			continue;
1903 		err = got_ref_resolve(&tag_id, repo, re->ref);
1904 		if (err)
1905 			break;
1906 		err = got_object_open_as_tag(tag, repo, tag_id);
1907 		free(tag_id);
1908 		if (err)
1909 			break;
1910 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1911 		    got_object_tag_get_object_type(*tag) == obj_type)
1912 			break;
1913 		got_object_tag_close(*tag);
1914 		*tag = NULL;
1915 	}
1916 
1917 	if (err == NULL && *tag == NULL)
1918 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1919 		    GOT_OBJ_LABEL_TAG, name);
1920 	return err;
1921 }
1922 
1923 static const struct got_error *
1924 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1925     const char *name, mode_t mode, struct got_object_id *blob_id)
1926 {
1927 	const struct got_error *err = NULL;
1928 
1929 	 *new_te = NULL;
1930 
1931 	*new_te = calloc(1, sizeof(**new_te));
1932 	if (*new_te == NULL)
1933 		return got_error_from_errno("calloc");
1934 
1935 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1936 	    sizeof((*new_te)->name)) {
1937 		err = got_error(GOT_ERR_NO_SPACE);
1938 		goto done;
1939 	}
1940 
1941 	if (S_ISLNK(mode)) {
1942 		(*new_te)->mode = S_IFLNK;
1943 	} else {
1944 		(*new_te)->mode = S_IFREG;
1945 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1946 	}
1947 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1948 done:
1949 	if (err && *new_te) {
1950 		free(*new_te);
1951 		*new_te = NULL;
1952 	}
1953 	return err;
1954 }
1955 
1956 static const struct got_error *
1957 import_file(struct got_tree_entry **new_te, struct dirent *de,
1958     const char *path, struct got_repository *repo)
1959 {
1960 	const struct got_error *err;
1961 	struct got_object_id *blob_id = NULL;
1962 	char *filepath;
1963 	struct stat sb;
1964 
1965 	if (asprintf(&filepath, "%s%s%s", path,
1966 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1967 		return got_error_from_errno("asprintf");
1968 
1969 	if (lstat(filepath, &sb) != 0) {
1970 		err = got_error_from_errno2("lstat", path);
1971 		goto done;
1972 	}
1973 
1974 	err = got_object_blob_create(&blob_id, filepath, repo);
1975 	if (err)
1976 		goto done;
1977 
1978 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1979 	    blob_id);
1980 done:
1981 	free(filepath);
1982 	if (err)
1983 		free(blob_id);
1984 	return err;
1985 }
1986 
1987 static const struct got_error *
1988 insert_tree_entry(struct got_tree_entry *new_te,
1989     struct got_pathlist_head *paths)
1990 {
1991 	const struct got_error *err = NULL;
1992 	struct got_pathlist_entry *new_pe;
1993 
1994 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1995 	if (err)
1996 		return err;
1997 	if (new_pe == NULL)
1998 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1999 	return NULL;
2000 }
2001 
2002 static const struct got_error *write_tree(struct got_object_id **,
2003     const char *, struct got_pathlist_head *, struct got_repository *,
2004     got_repo_import_cb progress_cb, void *progress_arg);
2005 
2006 static const struct got_error *
2007 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2008     const char *path, struct got_pathlist_head *ignores,
2009     struct got_repository *repo,
2010     got_repo_import_cb progress_cb, void *progress_arg)
2011 {
2012 	const struct got_error *err;
2013 	struct got_object_id *id = NULL;
2014 	char *subdirpath;
2015 
2016 	if (asprintf(&subdirpath, "%s%s%s", path,
2017 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2018 		return got_error_from_errno("asprintf");
2019 
2020 	(*new_te) = calloc(1, sizeof(**new_te));
2021 	if (*new_te == NULL)
2022 		return got_error_from_errno("calloc");
2023 	(*new_te)->mode = S_IFDIR;
2024 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2025 	    sizeof((*new_te)->name)) {
2026 		err = got_error(GOT_ERR_NO_SPACE);
2027 		goto done;
2028 	}
2029 	err = write_tree(&id, subdirpath, ignores,  repo,
2030 	    progress_cb, progress_arg);
2031 	if (err)
2032 		goto done;
2033 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2034 	
2035 done:
2036 	free(id);
2037 	free(subdirpath);
2038 	if (err) {
2039 		free(*new_te);
2040 		*new_te = NULL;
2041 	}
2042 	return err;
2043 }
2044 
2045 static const struct got_error *
2046 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2047     struct got_pathlist_head *ignores, struct got_repository *repo,
2048     got_repo_import_cb progress_cb, void *progress_arg)
2049 {
2050 	const struct got_error *err = NULL;
2051 	DIR *dir;
2052 	struct dirent *de;
2053 	int nentries;
2054 	struct got_tree_entry *new_te = NULL;
2055 	struct got_pathlist_head paths;
2056 	struct got_pathlist_entry *pe;
2057 
2058 	*new_tree_id = NULL;
2059 
2060 	TAILQ_INIT(&paths);
2061 
2062 	dir = opendir(path_dir);
2063 	if (dir == NULL) {
2064 		err = got_error_from_errno2("opendir", path_dir);
2065 		goto done;
2066 	}
2067 
2068 	nentries = 0;
2069 	while ((de = readdir(dir)) != NULL) {
2070 		int ignore = 0;
2071 		int type;
2072 
2073 		if (strcmp(de->d_name, ".") == 0 ||
2074 		    strcmp(de->d_name, "..") == 0)
2075 			continue;
2076 
2077 		TAILQ_FOREACH(pe, ignores, entry) {
2078 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2079 				ignore = 1;
2080 				break;
2081 			}
2082 		}
2083 		if (ignore)
2084 			continue;
2085 
2086 		err = got_path_dirent_type(&type, path_dir, de);
2087 		if (err)
2088 			goto done;
2089 
2090 		if (type == DT_DIR) {
2091 			err = import_subdir(&new_te, de, path_dir,
2092 			    ignores, repo, progress_cb, progress_arg);
2093 			if (err) {
2094 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2095 					goto done;
2096 				err = NULL;
2097 				continue;
2098 			}
2099 		} else if (type == DT_REG || type == DT_LNK) {
2100 			err = import_file(&new_te, de, path_dir, repo);
2101 			if (err)
2102 				goto done;
2103 		} else
2104 			continue;
2105 
2106 		err = insert_tree_entry(new_te, &paths);
2107 		if (err)
2108 			goto done;
2109 		nentries++;
2110 	}
2111 
2112 	if (TAILQ_EMPTY(&paths)) {
2113 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2114 		    "cannot create tree without any entries");
2115 		goto done;
2116 	}
2117 
2118 	TAILQ_FOREACH(pe, &paths, entry) {
2119 		struct got_tree_entry *te = pe->data;
2120 		char *path;
2121 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2122 			continue;
2123 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2124 			err = got_error_from_errno("asprintf");
2125 			goto done;
2126 		}
2127 		err = (*progress_cb)(progress_arg, path);
2128 		free(path);
2129 		if (err)
2130 			goto done;
2131 	}
2132 
2133 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2134 done:
2135 	if (dir)
2136 		closedir(dir);
2137 	got_pathlist_free(&paths);
2138 	return err;
2139 }
2140 
2141 const struct got_error *
2142 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2143     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2144     struct got_repository *repo, got_repo_import_cb progress_cb,
2145     void *progress_arg)
2146 {
2147 	const struct got_error *err;
2148 	struct got_object_id *new_tree_id;
2149 
2150 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2151 	    progress_cb, progress_arg);
2152 	if (err)
2153 		return err;
2154 
2155 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2156 	    author, time(NULL), author, time(NULL), logmsg, repo);
2157 	free(new_tree_id);
2158 	return err;
2159 }
2160 
2161 const struct got_error *
2162 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2163     struct got_repository *repo)
2164 {
2165 	const struct got_error *err = NULL;
2166 	char *path_objects = NULL, *path = NULL;
2167 	DIR *dir = NULL;
2168 	struct got_object_id id;
2169 	int i;
2170 
2171 	*nobjects = 0;
2172 	*ondisk_size = 0;
2173 
2174 	path_objects = got_repo_get_path_objects(repo);
2175 	if (path_objects == NULL)
2176 		return got_error_from_errno("got_repo_get_path_objects");
2177 
2178 	for (i = 0; i <= 0xff; i++) {
2179 		struct dirent *dent;
2180 
2181 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2182 			err = got_error_from_errno("asprintf");
2183 			break;
2184 		}
2185 
2186 		dir = opendir(path);
2187 		if (dir == NULL) {
2188 			if (errno == ENOENT) {
2189 				err = NULL;
2190 				continue;
2191 			}
2192 			err = got_error_from_errno2("opendir", path);
2193 			break;
2194 		}
2195 
2196 		while ((dent = readdir(dir)) != NULL) {
2197 			char *id_str;
2198 			int fd;
2199 			struct stat sb;
2200 
2201 			if (strcmp(dent->d_name, ".") == 0 ||
2202 			    strcmp(dent->d_name, "..") == 0)
2203 				continue;
2204 
2205 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2206 				err = got_error_from_errno("asprintf");
2207 				goto done;
2208 			}
2209 
2210 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2211 				free(id_str);
2212 				continue;
2213 			}
2214 			free(id_str);
2215 
2216 			err = got_object_open_loose_fd(&fd, &id, repo);
2217 			if (err)
2218 				goto done;
2219 
2220 			if (fstat(fd, &sb) == -1) {
2221 				err = got_error_from_errno("fstat");
2222 				close(fd);
2223 				goto done;
2224 			}
2225 			(*nobjects)++;
2226 			(*ondisk_size) += sb.st_size;
2227 
2228 			if (close(fd) == -1) {
2229 				err = got_error_from_errno("close");
2230 				goto done;
2231 			}
2232 		}
2233 
2234 		if (closedir(dir) != 0) {
2235 			err = got_error_from_errno("closedir");
2236 			goto done;
2237 		}
2238 		dir = NULL;
2239 
2240 		free(path);
2241 		path = NULL;
2242 	}
2243 done:
2244 	if (dir && closedir(dir) != 0 && err == NULL)
2245 		err = got_error_from_errno("closedir");
2246 
2247 	if (err) {
2248 		*nobjects = 0;
2249 		*ondisk_size = 0;
2250 	}
2251 	free(path_objects);
2252 	free(path);
2253 	return err;
2254 }
2255 
2256 const struct got_error *
2257 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2258     off_t *total_packsize, struct got_repository *repo)
2259 {
2260 	const struct got_error *err = NULL;
2261 	DIR *packdir = NULL;
2262 	struct dirent *dent;
2263 	struct got_packidx *packidx = NULL;
2264 	char *path_packidx;
2265 	char *path_packfile;
2266 	int packdir_fd;
2267 	struct stat sb;
2268 
2269 	*npackfiles = 0;
2270 	*nobjects = 0;
2271 	*total_packsize = 0;
2272 
2273 	packdir_fd = openat(got_repo_get_fd(repo),
2274 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2275 	if (packdir_fd == -1) {
2276 		return got_error_from_errno_fmt("openat: %s/%s",
2277 		    got_repo_get_path_git_dir(repo),
2278 		    GOT_OBJECTS_PACK_DIR);
2279 	}
2280 
2281 	packdir = fdopendir(packdir_fd);
2282 	if (packdir == NULL) {
2283 		err = got_error_from_errno("fdopendir");
2284 		goto done;
2285 	}
2286 
2287 	while ((dent = readdir(packdir)) != NULL) {
2288 		if (!got_repo_is_packidx_filename(dent->d_name,
2289 		    strlen(dent->d_name)))
2290 			continue;
2291 
2292 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2293 		    dent->d_name) == -1) {
2294 			err = got_error_from_errno("asprintf");
2295 			goto done;
2296 		}
2297 
2298 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2299 		    path_packidx, 0);
2300 		free(path_packidx);
2301 		if (err)
2302 			goto done;
2303 
2304 		if (fstat(packidx->fd, &sb) == -1)
2305 			goto done;
2306 		*total_packsize += sb.st_size;
2307 
2308 		err = got_packidx_get_packfile_path(&path_packfile,
2309 		    packidx->path_packidx);
2310 		if (err)
2311 			goto done;
2312 
2313 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2314 		    0) == -1) {
2315 			free(path_packfile);
2316 			goto done;
2317 		}
2318 		free(path_packfile);
2319 		*total_packsize += sb.st_size;
2320 
2321 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2322 
2323 		(*npackfiles)++;
2324 
2325 		got_packidx_close(packidx);
2326 		packidx = NULL;
2327 	}
2328 done:
2329 	if (packidx)
2330 		got_packidx_close(packidx);
2331 	if (packdir && closedir(packdir) != 0 && err == NULL)
2332 		err = got_error_from_errno("closedir");
2333 	if (err) {
2334 		*npackfiles = 0;
2335 		*nobjects = 0;
2336 		*total_packsize = 0;
2337 	}
2338 	return err;
2339 }
2340 
2341 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2342     got_packidx_bloom_filter_cmp);