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