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/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
72 
73 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
74     got_packidx_bloom_filter_cmp);
75 
76 const char *
77 got_repo_get_path(struct got_repository *repo)
78 {
79 	return repo->path;
80 }
81 
82 const char *
83 got_repo_get_path_git_dir(struct got_repository *repo)
84 {
85 	return repo->path_git_dir;
86 }
87 
88 int
89 got_repo_get_fd(struct got_repository *repo)
90 {
91 	return repo->gitdir_fd;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_name(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_name;
98 }
99 
100 const char *
101 got_repo_get_gitconfig_author_email(struct got_repository *repo)
102 {
103 	return repo->gitconfig_author_email;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_name;
110 }
111 
112 const char *
113 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
114 {
115 	return repo->global_gitconfig_author_email;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_owner(struct got_repository *repo)
120 {
121 	return repo->gitconfig_owner;
122 }
123 
124 void
125 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
126     struct got_repository *repo)
127 {
128 	*extensions = repo->extensions;
129 	*nextensions = repo->nextensions;
130 }
131 
132 int
133 got_repo_is_bare(struct got_repository *repo)
134 {
135 	return (strcmp(repo->path, repo->path_git_dir) == 0);
136 }
137 
138 static char *
139 get_path_git_child(struct got_repository *repo, const char *basename)
140 {
141 	char *path_child;
142 
143 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
144 	    basename) == -1)
145 		return NULL;
146 
147 	return path_child;
148 }
149 
150 char *
151 got_repo_get_path_objects(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
154 }
155 
156 char *
157 got_repo_get_path_objects_pack(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
160 }
161 
162 char *
163 got_repo_get_path_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_REFS_DIR);
166 }
167 
168 char *
169 got_repo_get_path_packed_refs(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
172 }
173 
174 static char *
175 get_path_head(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_HEAD_FILE);
178 }
179 
180 char *
181 got_repo_get_path_gitconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GITCONFIG);
184 }
185 
186 char *
187 got_repo_get_path_gotconfig(struct got_repository *repo)
188 {
189 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
190 }
191 
192 const struct got_gotconfig *
193 got_repo_get_gotconfig(struct got_repository *repo)
194 {
195 	return repo->gotconfig;
196 }
197 
198 void
199 got_repo_get_gitconfig_remotes(int *nremotes,
200     const struct got_remote_repo **remotes, struct got_repository *repo)
201 {
202 	*nremotes = repo->ngitconfig_remotes;
203 	*remotes = repo->gitconfig_remotes;
204 }
205 
206 static int
207 is_git_repo(struct got_repository *repo)
208 {
209 	const char *path_git = got_repo_get_path_git_dir(repo);
210 	char *path_objects = got_repo_get_path_objects(repo);
211 	char *path_refs = got_repo_get_path_refs(repo);
212 	char *path_head = get_path_head(repo);
213 	int ret = 0;
214 	struct stat sb;
215 	struct got_reference *head_ref;
216 
217 	if (lstat(path_git, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_objects, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_refs, &sb) == -1)
228 		goto done;
229 	if (!S_ISDIR(sb.st_mode))
230 		goto done;
231 
232 	if (lstat(path_head, &sb) == -1)
233 		goto done;
234 	if (!S_ISREG(sb.st_mode))
235 		goto done;
236 
237 	/* Check if the HEAD reference can be opened. */
238 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
239 		goto done;
240 	got_ref_close(head_ref);
241 
242 	ret = 1;
243 done:
244 	free(path_objects);
245 	free(path_refs);
246 	free(path_head);
247 	return ret;
248 
249 }
250 
251 const struct got_error *
252 got_repo_pack_fds_open(int **pack_fds)
253 {
254 	const struct got_error *err = NULL;
255 	int i, pack_fds_tmp[GOT_PACK_NUM_TEMPFILES];
256 
257 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
258 	if (*pack_fds == NULL)
259 		return got_error_from_errno("calloc");
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, sizeof(pack_fds_tmp));
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 	for (i = 0; i < repo->nextensions; i++) {
794 		char *ext = repo->extensions[i];
795 		int j, supported = 0;
796 		for (j = 0; j < nitems(repo_extensions); j++) {
797 			if (strcmp(ext, repo_extensions[j]) == 0) {
798 				supported = 1;
799 				break;
800 			}
801 		}
802 		if (!supported) {
803 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
804 			goto done;
805 		}
806 	}
807 
808 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
809 done:
810 	if (err)
811 		got_repo_close(repo);
812 	else
813 		*repop = repo;
814 	free(repo_path);
815 	return err;
816 }
817 
818 const struct got_error *
819 got_repo_close(struct got_repository *repo)
820 {
821 	const struct got_error *err = NULL, *child_err;
822 	struct got_packidx_bloom_filter *bf;
823 	struct got_pathlist_entry *pe;
824 	size_t i;
825 
826 	for (i = 0; i < repo->pack_cache_size; i++) {
827 		if (repo->packidx_cache[i] == NULL)
828 			break;
829 		got_packidx_close(repo->packidx_cache[i]);
830 	}
831 
832 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
833 	    &repo->packidx_bloom_filters))) {
834 		RB_REMOVE(got_packidx_bloom_filter_tree,
835 		    &repo->packidx_bloom_filters, bf);
836 		free(bf->bloom);
837 		free(bf);
838 	}
839 
840 	for (i = 0; i < repo->pack_cache_size; i++)
841 		if (repo->packs[i].path_packfile)
842 			if (repo->packs[i].path_packfile)
843 				got_pack_close(&repo->packs[i]);
844 
845 	free(repo->path);
846 	free(repo->path_git_dir);
847 
848 	got_object_cache_close(&repo->objcache);
849 	got_object_cache_close(&repo->treecache);
850 	got_object_cache_close(&repo->commitcache);
851 	got_object_cache_close(&repo->tagcache);
852 	got_object_cache_close(&repo->rawcache);
853 
854 	for (i = 0; i < nitems(repo->privsep_children); i++) {
855 		if (repo->privsep_children[i].imsg_fd == -1)
856 			continue;
857 		imsg_clear(repo->privsep_children[i].ibuf);
858 		free(repo->privsep_children[i].ibuf);
859 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
860 		child_err = got_privsep_wait_for_child(
861 		    repo->privsep_children[i].pid);
862 		if (child_err && err == NULL)
863 			err = child_err;
864 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
865 		    err == NULL)
866 			err = got_error_from_errno("close");
867 	}
868 
869 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
870 	    err == NULL)
871 		err = got_error_from_errno("close");
872 
873 	if (repo->gotconfig)
874 		got_gotconfig_free(repo->gotconfig);
875 	free(repo->gitconfig_author_name);
876 	free(repo->gitconfig_author_email);
877 	for (i = 0; i < repo->ngitconfig_remotes; i++)
878 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
879 	free(repo->gitconfig_remotes);
880 	for (i = 0; i < repo->nextensions; i++)
881 		free(repo->extensions[i]);
882 	free(repo->extensions);
883 
884 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
885 		free((void *)pe->path);
886 	got_pathlist_free(&repo->packidx_paths);
887 	free(repo);
888 
889 	return err;
890 }
891 
892 void
893 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
894 {
895 	int i;
896 
897 	free(repo->name);
898 	repo->name = NULL;
899 	free(repo->fetch_url);
900 	repo->fetch_url = NULL;
901 	free(repo->send_url);
902 	repo->send_url = NULL;
903 	for (i = 0; i < repo->nfetch_branches; i++)
904 		free(repo->fetch_branches[i]);
905 	free(repo->fetch_branches);
906 	repo->fetch_branches = NULL;
907 	repo->nfetch_branches = 0;
908 	for (i = 0; i < repo->nsend_branches; i++)
909 		free(repo->send_branches[i]);
910 	free(repo->send_branches);
911 	repo->send_branches = NULL;
912 	repo->nsend_branches = 0;
913 }
914 
915 const struct got_error *
916 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
917     const char *input_path)
918 {
919 	const struct got_error *err = NULL;
920 	const char *repo_abspath = NULL;
921 	size_t repolen, len;
922 	char *canonpath, *path = NULL;
923 
924 	*in_repo_path = NULL;
925 
926 	canonpath = strdup(input_path);
927 	if (canonpath == NULL) {
928 		err = got_error_from_errno("strdup");
929 		goto done;
930 	}
931 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
932 	if (err)
933 		goto done;
934 
935 	repo_abspath = got_repo_get_path(repo);
936 
937 	if (canonpath[0] == '\0') {
938 		path = strdup(canonpath);
939 		if (path == NULL) {
940 			err = got_error_from_errno("strdup");
941 			goto done;
942 		}
943 	} else {
944 		path = realpath(canonpath, NULL);
945 		if (path == NULL) {
946 			if (errno != ENOENT) {
947 				err = got_error_from_errno2("realpath",
948 				    canonpath);
949 				goto done;
950 			}
951 			/*
952 			 * Path is not on disk.
953 			 * Assume it is already relative to repository root.
954 			 */
955 			path = strdup(canonpath);
956 			if (path == NULL) {
957 				err = got_error_from_errno("strdup");
958 				goto done;
959 			}
960 		}
961 
962 		repolen = strlen(repo_abspath);
963 		len = strlen(path);
964 
965 
966 		if (strcmp(path, repo_abspath) == 0) {
967 			free(path);
968 			path = strdup("");
969 			if (path == NULL) {
970 				err = got_error_from_errno("strdup");
971 				goto done;
972 			}
973 		} else if (len > repolen &&
974 		    got_path_is_child(path, repo_abspath, repolen)) {
975 			/* Matched an on-disk path inside repository. */
976 			if (got_repo_is_bare(repo)) {
977 				/*
978 				 * Matched an on-disk path inside repository
979 				 * database. Treat input as repository-relative.
980 				 */
981 				free(path);
982 				path = canonpath;
983 				canonpath = NULL;
984 			} else {
985 				char *child;
986 				/* Strip common prefix with repository path. */
987 				err = got_path_skip_common_ancestor(&child,
988 				    repo_abspath, path);
989 				if (err)
990 					goto done;
991 				free(path);
992 				path = child;
993 			}
994 		} else {
995 			/*
996 			 * Matched unrelated on-disk path.
997 			 * Treat input as repository-relative.
998 			 */
999 			free(path);
1000 			path = canonpath;
1001 			canonpath = NULL;
1002 		}
1003 	}
1004 
1005 	/* Make in-repository path absolute */
1006 	if (path[0] != '/') {
1007 		char *abspath;
1008 		if (asprintf(&abspath, "/%s", path) == -1) {
1009 			err = got_error_from_errno("asprintf");
1010 			goto done;
1011 		}
1012 		free(path);
1013 		path = abspath;
1014 	}
1015 
1016 done:
1017 	free(canonpath);
1018 	if (err)
1019 		free(path);
1020 	else
1021 		*in_repo_path = path;
1022 	return err;
1023 }
1024 
1025 static const struct got_error *
1026 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1027     const char *path_packidx)
1028 {
1029 	const struct got_error *err = NULL;
1030 	size_t i;
1031 
1032 	for (i = 0; i < repo->pack_cache_size; i++) {
1033 		if (repo->packidx_cache[i] == NULL)
1034 			break;
1035 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1036 		    path_packidx) == 0) {
1037 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1038 		}
1039 	}
1040 	if (i == repo->pack_cache_size) {
1041 		do {
1042 			i--;
1043 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1044 		    i == repo->pinned_packidx);
1045 		err = got_packidx_close(repo->packidx_cache[i]);
1046 		if (err)
1047 			return err;
1048 	}
1049 
1050 	repo->packidx_cache[i] = packidx;
1051 
1052 	return NULL;
1053 }
1054 
1055 int
1056 got_repo_is_packidx_filename(const char *name, size_t len)
1057 {
1058 	if (len != GOT_PACKIDX_NAMELEN)
1059 		return 0;
1060 
1061 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1062 		return 0;
1063 
1064 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1065 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1066 		return 0;
1067 
1068 	return 1;
1069 }
1070 
1071 static struct got_packidx_bloom_filter *
1072 get_packidx_bloom_filter(struct got_repository *repo,
1073     const char *path, size_t path_len)
1074 {
1075 	struct got_packidx_bloom_filter key;
1076 
1077 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1078 		return NULL; /* XXX */
1079 	key.path_len = path_len;
1080 
1081 	return RB_FIND(got_packidx_bloom_filter_tree,
1082 	    &repo->packidx_bloom_filters, &key);
1083 }
1084 
1085 int
1086 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1087     const char *path_packidx, struct got_object_id *id)
1088 {
1089 	struct got_packidx_bloom_filter *bf;
1090 
1091 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1092 	if (bf)
1093 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1094 
1095 	/* No bloom filter means this pack index must be searched. */
1096 	return 1;
1097 }
1098 
1099 static const struct got_error *
1100 add_packidx_bloom_filter(struct got_repository *repo,
1101     struct got_packidx *packidx, const char *path_packidx)
1102 {
1103 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1104 	struct got_packidx_bloom_filter *bf;
1105 	size_t len;
1106 
1107 	/*
1108 	 * Don't use bloom filters for very large pack index files.
1109 	 * Large pack files will contain a relatively large fraction
1110 	 * of our objects so we will likely need to visit them anyway.
1111 	 * The more objects a pack file contains the higher the probability
1112 	 * of a false-positive match from the bloom filter. And reading
1113 	 * all object IDs from a large pack index file can be expensive.
1114 	 */
1115 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1116 		return NULL;
1117 
1118 	/* Do we already have a filter for this pack index? */
1119 	if (get_packidx_bloom_filter(repo, path_packidx,
1120 	    strlen(path_packidx)) != NULL)
1121 		return NULL;
1122 
1123 	bf = calloc(1, sizeof(*bf));
1124 	if (bf == NULL)
1125 		return got_error_from_errno("calloc");
1126 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1127 	if (bf->bloom == NULL) {
1128 		free(bf);
1129 		return got_error_from_errno("calloc");
1130 	}
1131 	
1132 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1133 	if (len >= sizeof(bf->path)) {
1134 		free(bf->bloom);
1135 		free(bf);
1136 		return got_error(GOT_ERR_NO_SPACE);
1137 	}
1138 	bf->path_len = len;
1139 
1140 	/* Minimum size supported by our bloom filter is 1000 entries. */
1141 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1142 	for (i = 0; i < nobjects; i++) {
1143 		struct got_packidx_object_id *id;
1144 		id = &packidx->hdr.sorted_ids[i];
1145 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1146 	}
1147 
1148 	RB_INSERT(got_packidx_bloom_filter_tree,
1149 	    &repo->packidx_bloom_filters, bf);
1150 	return NULL;
1151 }
1152 
1153 const struct got_error *
1154 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1155     struct got_repository *repo, struct got_object_id *id)
1156 {
1157 	const struct got_error *err;
1158 	struct got_pathlist_entry *pe;
1159 	size_t i;
1160 
1161 	/* Search pack index cache. */
1162 	for (i = 0; i < repo->pack_cache_size; i++) {
1163 		if (repo->packidx_cache[i] == NULL)
1164 			break;
1165 		if (!got_repo_check_packidx_bloom_filter(repo,
1166 		    repo->packidx_cache[i]->path_packidx, id))
1167 			continue; /* object will not be found in this index */
1168 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1169 		if (*idx != -1) {
1170 			*packidx = repo->packidx_cache[i];
1171 			/*
1172 			 * Move this cache entry to the front. Repeatedly
1173 			 * searching a wrong pack index can be expensive.
1174 			 */
1175 			if (i > 0) {
1176 				memmove(&repo->packidx_cache[1],
1177 				    &repo->packidx_cache[0],
1178 				    i * sizeof(repo->packidx_cache[0]));
1179 				repo->packidx_cache[0] = *packidx;
1180 				if (repo->pinned_packidx >= 0 &&
1181 				    repo->pinned_packidx < i)
1182 					repo->pinned_packidx++;
1183 				else if (repo->pinned_packidx == i)
1184 					repo->pinned_packidx = 0;
1185 			}
1186 			return NULL;
1187 		}
1188 	}
1189 	/* No luck. Search the filesystem. */
1190 
1191 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1192 		const char *path_packidx = pe->path;
1193 		int is_cached = 0;
1194 
1195 		if (!got_repo_check_packidx_bloom_filter(repo,
1196 		    pe->path, id))
1197 			continue; /* object will not be found in this index */
1198 
1199 		for (i = 0; i < repo->pack_cache_size; i++) {
1200 			if (repo->packidx_cache[i] == NULL)
1201 				break;
1202 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1203 			    path_packidx) == 0) {
1204 				is_cached = 1;
1205 				break;
1206 			}
1207 		}
1208 		if (is_cached)
1209 			continue; /* already searched */
1210 
1211 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1212 		    path_packidx, 0);
1213 		if (err)
1214 			goto done;
1215 
1216 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1217 		if (err)
1218 			goto done;
1219 
1220 		err = cache_packidx(repo, *packidx, path_packidx);
1221 		if (err)
1222 			goto done;
1223 
1224 		*idx = got_packidx_get_object_idx(*packidx, id);
1225 		if (*idx != -1) {
1226 			err = NULL; /* found the object */
1227 			goto done;
1228 		}
1229 	}
1230 
1231 	err = got_error_no_obj(id);
1232 done:
1233 	return err;
1234 }
1235 
1236 const struct got_error *
1237 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1238     struct got_repository *repo)
1239 {
1240 	const struct got_error *err = NULL;
1241 	DIR *packdir = NULL;
1242 	struct dirent *dent;
1243 	char *path_packidx = NULL;
1244 	int packdir_fd;
1245 
1246 	packdir_fd = openat(got_repo_get_fd(repo),
1247 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1248 	if (packdir_fd == -1) {
1249 		return got_error_from_errno_fmt("openat: %s/%s",
1250 		    got_repo_get_path_git_dir(repo),
1251 		    GOT_OBJECTS_PACK_DIR);
1252 	}
1253 
1254 	packdir = fdopendir(packdir_fd);
1255 	if (packdir == NULL) {
1256 		err = got_error_from_errno("fdopendir");
1257 		goto done;
1258 	}
1259 
1260 	while ((dent = readdir(packdir)) != NULL) {
1261 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1262 			continue;
1263 
1264 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1265 		    dent->d_name) == -1) {
1266 			err = got_error_from_errno("asprintf");
1267 			path_packidx = NULL;
1268 			break;
1269 		}
1270 
1271 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1272 		if (err)
1273 			break;
1274 	}
1275 done:
1276 	if (err)
1277 		free(path_packidx);
1278 	if (packdir && closedir(packdir) != 0 && err == NULL)
1279 		err = got_error_from_errno("closedir");
1280 	return err;
1281 }
1282 
1283 const struct got_error *
1284 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1285     struct got_repository *repo)
1286 {
1287 	const struct got_error *err;
1288 	size_t i;
1289 
1290 	*packidx = NULL;
1291 
1292 	/* Search pack index cache. */
1293 	for (i = 0; i < repo->pack_cache_size; i++) {
1294 		if (repo->packidx_cache[i] == NULL)
1295 			break;
1296 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1297 		    path_packidx) == 0) {
1298 			*packidx = repo->packidx_cache[i];
1299 			return NULL;
1300 		}
1301 	}
1302 	/* No luck. Search the filesystem. */
1303 
1304 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1305 	    path_packidx, 0);
1306 	if (err)
1307 		return err;
1308 
1309 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1310 	if (err)
1311 		goto done;
1312 
1313 	err = cache_packidx(repo, *packidx, path_packidx);
1314 done:
1315 	if (err) {
1316 		got_packidx_close(*packidx);
1317 		*packidx = NULL;
1318 	}
1319 	return err;
1320 }
1321 
1322 static const struct got_error *
1323 read_packfile_hdr(int fd, struct got_packidx *packidx)
1324 {
1325 	const struct got_error *err = NULL;
1326 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1327 	struct got_packfile_hdr hdr;
1328 	ssize_t n;
1329 
1330 	n = read(fd, &hdr, sizeof(hdr));
1331 	if (n < 0)
1332 		return got_error_from_errno("read");
1333 	if (n != sizeof(hdr))
1334 		return got_error(GOT_ERR_BAD_PACKFILE);
1335 
1336 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1337 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1338 	    be32toh(hdr.nobjects) != totobj)
1339 		err = got_error(GOT_ERR_BAD_PACKFILE);
1340 
1341 	return err;
1342 }
1343 
1344 static const struct got_error *
1345 open_packfile(int *fd, struct got_repository *repo,
1346     const char *relpath, struct got_packidx *packidx)
1347 {
1348 	const struct got_error *err = NULL;
1349 
1350 	*fd = openat(got_repo_get_fd(repo), relpath,
1351 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1352 	if (*fd == -1)
1353 		return got_error_from_errno_fmt("openat: %s/%s",
1354 		    got_repo_get_path_git_dir(repo), relpath);
1355 
1356 	if (packidx) {
1357 		err = read_packfile_hdr(*fd, packidx);
1358 		if (err) {
1359 			close(*fd);
1360 			*fd = -1;
1361 		}
1362 	}
1363 
1364 	return err;
1365 }
1366 
1367 const struct got_error *
1368 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1369     const char *path_packfile, struct got_packidx *packidx)
1370 {
1371 	const struct got_error *err = NULL;
1372 	struct got_pack *pack = NULL;
1373 	struct stat sb;
1374 	size_t i;
1375 
1376 	if (packp)
1377 		*packp = NULL;
1378 
1379 	for (i = 0; i < repo->pack_cache_size; i++) {
1380 		pack = &repo->packs[i];
1381 		if (pack->path_packfile == NULL)
1382 			break;
1383 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1384 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1385 	}
1386 
1387 	if (i == repo->pack_cache_size) {
1388 		struct got_pack tmp;
1389 		do {
1390 			i--;
1391 		} while (i > 0 && repo->pinned_pack >= 0 &&
1392 		    i == repo->pinned_pack);
1393 		err = got_pack_close(&repo->packs[i]);
1394 		if (err)
1395 			return err;
1396 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1397 			return got_error_from_errno("ftruncate");
1398 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1399 			return got_error_from_errno("ftruncate");
1400 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1401 		memcpy(&repo->packs[i], &repo->packs[0],
1402 		    sizeof(repo->packs[i]));
1403 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1404 		if (repo->pinned_pack == 0)
1405 			repo->pinned_pack = i;
1406 		else if (repo->pinned_pack == i)
1407 			repo->pinned_pack = 0;
1408 		i = 0;
1409 	}
1410 
1411 	pack = &repo->packs[i];
1412 
1413 	pack->path_packfile = strdup(path_packfile);
1414 	if (pack->path_packfile == NULL) {
1415 		err = got_error_from_errno("strdup");
1416 		goto done;
1417 	}
1418 
1419 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1420 	if (err)
1421 		goto done;
1422 
1423 	if (fstat(pack->fd, &sb) != 0) {
1424 		err = got_error_from_errno("fstat");
1425 		goto done;
1426 	}
1427 	pack->filesize = sb.st_size;
1428 
1429 	pack->privsep_child = NULL;
1430 
1431 #ifndef GOT_PACK_NO_MMAP
1432 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1433 	    pack->fd, 0);
1434 	if (pack->map == MAP_FAILED) {
1435 		if (errno != ENOMEM) {
1436 			err = got_error_from_errno("mmap");
1437 			goto done;
1438 		}
1439 		pack->map = NULL; /* fall back to read(2) */
1440 	}
1441 #endif
1442 done:
1443 	if (err) {
1444 		if (pack) {
1445 			free(pack->path_packfile);
1446 			memset(pack, 0, sizeof(*pack));
1447 		}
1448 	} else if (packp)
1449 		*packp = pack;
1450 	return err;
1451 }
1452 
1453 struct got_pack *
1454 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1455 {
1456 	struct got_pack *pack = NULL;
1457 	size_t i;
1458 
1459 	for (i = 0; i < repo->pack_cache_size; i++) {
1460 		pack = &repo->packs[i];
1461 		if (pack->path_packfile == NULL)
1462 			break;
1463 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1464 			return pack;
1465 	}
1466 
1467 	return NULL;
1468 }
1469 
1470 const struct got_error *
1471 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1472     struct got_pack *pack)
1473 {
1474 	size_t i;
1475 	int pinned_pack = -1, pinned_packidx = -1;
1476 
1477 	for (i = 0; i < repo->pack_cache_size; i++) {
1478 		if (repo->packidx_cache[i] &&
1479 		    strcmp(repo->packidx_cache[i]->path_packidx,
1480 		    packidx->path_packidx) == 0)
1481 			pinned_packidx = i;
1482 		if (repo->packs[i].path_packfile &&
1483 		    strcmp(repo->packs[i].path_packfile,
1484 		    pack->path_packfile) == 0)
1485 			pinned_pack = i;
1486 	}
1487 
1488 	if (pinned_packidx == -1 || pinned_pack == -1)
1489 		return got_error(GOT_ERR_PIN_PACK);
1490 
1491 	repo->pinned_pack = pinned_pack;
1492 	repo->pinned_packidx = pinned_packidx;
1493 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1494 	return NULL;
1495 }
1496 
1497 struct got_pack *
1498 got_repo_get_pinned_pack(struct got_repository *repo)
1499 {
1500 	if (repo->pinned_pack >= 0 &&
1501 	    repo->pinned_pack < repo->pack_cache_size)
1502 		return &repo->packs[repo->pinned_pack];
1503 
1504 	return NULL;
1505 }
1506 
1507 void
1508 got_repo_unpin_pack(struct got_repository *repo)
1509 {
1510 	repo->pinned_packidx = -1;
1511 	repo->pinned_pack = -1;
1512 	repo->pinned_pid = 0;
1513 }
1514 
1515 const struct got_error *
1516 got_repo_init(const char *repo_path)
1517 {
1518 	const struct got_error *err = NULL;
1519 	const char *dirnames[] = {
1520 		GOT_OBJECTS_DIR,
1521 		GOT_OBJECTS_PACK_DIR,
1522 		GOT_REFS_DIR,
1523 	};
1524 	const char *description_str = "Unnamed repository; "
1525 	    "edit this file 'description' to name the repository.";
1526 	const char *headref_str = "ref: refs/heads/main";
1527 	const char *gitconfig_str = "[core]\n"
1528 	    "\trepositoryformatversion = 0\n"
1529 	    "\tfilemode = true\n"
1530 	    "\tbare = true\n";
1531 	char *path;
1532 	size_t i;
1533 
1534 	if (!got_path_dir_is_empty(repo_path))
1535 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1536 
1537 	for (i = 0; i < nitems(dirnames); i++) {
1538 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1539 			return got_error_from_errno("asprintf");
1540 		}
1541 		err = got_path_mkdir(path);
1542 		free(path);
1543 		if (err)
1544 			return err;
1545 	}
1546 
1547 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1548 		return got_error_from_errno("asprintf");
1549 	err = got_path_create_file(path, description_str);
1550 	free(path);
1551 	if (err)
1552 		return err;
1553 
1554 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1555 		return got_error_from_errno("asprintf");
1556 	err = got_path_create_file(path, headref_str);
1557 	free(path);
1558 	if (err)
1559 		return err;
1560 
1561 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1562 		return got_error_from_errno("asprintf");
1563 	err = got_path_create_file(path, gitconfig_str);
1564 	free(path);
1565 	if (err)
1566 		return err;
1567 
1568 	return NULL;
1569 }
1570 
1571 static const struct got_error *
1572 match_packed_object(struct got_object_id **unique_id,
1573     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1574 {
1575 	const struct got_error *err = NULL;
1576 	struct got_object_id_queue matched_ids;
1577 	struct got_pathlist_entry *pe;
1578 
1579 	STAILQ_INIT(&matched_ids);
1580 
1581 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1582 		const char *path_packidx = pe->path;
1583 		struct got_packidx *packidx;
1584 		struct got_object_qid *qid;
1585 
1586 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1587 		    path_packidx, 0);
1588 		if (err)
1589 			break;
1590 
1591 		err = got_packidx_match_id_str_prefix(&matched_ids,
1592 		    packidx, id_str_prefix);
1593 		if (err) {
1594 			got_packidx_close(packidx);
1595 			break;
1596 		}
1597 		err = got_packidx_close(packidx);
1598 		if (err)
1599 			break;
1600 
1601 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1602 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1603 				int matched_type;
1604 				err = got_object_get_type(&matched_type, repo,
1605 				    &qid->id);
1606 				if (err)
1607 					goto done;
1608 				if (matched_type != obj_type)
1609 					continue;
1610 			}
1611 			if (*unique_id == NULL) {
1612 				*unique_id = got_object_id_dup(&qid->id);
1613 				if (*unique_id == NULL) {
1614 					err = got_error_from_errno("malloc");
1615 					goto done;
1616 				}
1617 			} else {
1618 				if (got_object_id_cmp(*unique_id,
1619 				    &qid->id) == 0)
1620 					continue; /* packed multiple times */
1621 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1622 				goto done;
1623 			}
1624 		}
1625 	}
1626 done:
1627 	got_object_id_queue_free(&matched_ids);
1628 	if (err) {
1629 		free(*unique_id);
1630 		*unique_id = NULL;
1631 	}
1632 	return err;
1633 }
1634 
1635 static const struct got_error *
1636 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1637     const char *object_dir, const char *id_str_prefix, int obj_type,
1638     struct got_repository *repo)
1639 {
1640 	const struct got_error *err = NULL;
1641 	char *path;
1642 	DIR *dir = NULL;
1643 	struct dirent *dent;
1644 	struct got_object_id id;
1645 
1646 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1647 		err = got_error_from_errno("asprintf");
1648 		goto done;
1649 	}
1650 
1651 	dir = opendir(path);
1652 	if (dir == NULL) {
1653 		if (errno == ENOENT) {
1654 			err = NULL;
1655 			goto done;
1656 		}
1657 		err = got_error_from_errno2("opendir", path);
1658 		goto done;
1659 	}
1660 	while ((dent = readdir(dir)) != NULL) {
1661 		char *id_str;
1662 		int cmp;
1663 
1664 		if (strcmp(dent->d_name, ".") == 0 ||
1665 		    strcmp(dent->d_name, "..") == 0)
1666 			continue;
1667 
1668 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1669 			err = got_error_from_errno("asprintf");
1670 			goto done;
1671 		}
1672 
1673 		if (!got_parse_sha1_digest(id.sha1, id_str))
1674 			continue;
1675 
1676 		/*
1677 		 * Directory entries do not necessarily appear in
1678 		 * sorted order, so we must iterate over all of them.
1679 		 */
1680 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1681 		if (cmp != 0) {
1682 			free(id_str);
1683 			continue;
1684 		}
1685 
1686 		if (*unique_id == NULL) {
1687 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1688 				int matched_type;
1689 				err = got_object_get_type(&matched_type, repo,
1690 				    &id);
1691 				if (err)
1692 					goto done;
1693 				if (matched_type != obj_type)
1694 					continue;
1695 			}
1696 			*unique_id = got_object_id_dup(&id);
1697 			if (*unique_id == NULL) {
1698 				err = got_error_from_errno("got_object_id_dup");
1699 				free(id_str);
1700 				goto done;
1701 			}
1702 		} else {
1703 			if (got_object_id_cmp(*unique_id, &id) == 0)
1704 				continue; /* both packed and loose */
1705 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1706 			free(id_str);
1707 			goto done;
1708 		}
1709 	}
1710 done:
1711 	if (dir && closedir(dir) != 0 && err == NULL)
1712 		err = got_error_from_errno("closedir");
1713 	if (err) {
1714 		free(*unique_id);
1715 		*unique_id = NULL;
1716 	}
1717 	free(path);
1718 	return err;
1719 }
1720 
1721 const struct got_error *
1722 got_repo_match_object_id_prefix(struct got_object_id **id,
1723     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1724 {
1725 	const struct got_error *err = NULL;
1726 	char *path_objects = got_repo_get_path_objects(repo);
1727 	char *object_dir = NULL;
1728 	size_t len;
1729 	int i;
1730 
1731 	*id = NULL;
1732 
1733 	len = strlen(id_str_prefix);
1734 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1735 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1736 
1737 	for (i = 0; i < len; i++) {
1738 		if (isxdigit((unsigned char)id_str_prefix[i]))
1739 			continue;
1740 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1741 	}
1742 
1743 	if (len >= 2) {
1744 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1745 		if (err)
1746 			goto done;
1747 		object_dir = strndup(id_str_prefix, 2);
1748 		if (object_dir == NULL) {
1749 			err = got_error_from_errno("strdup");
1750 			goto done;
1751 		}
1752 		err = match_loose_object(id, path_objects, object_dir,
1753 		    id_str_prefix, obj_type, repo);
1754 	} else if (len == 1) {
1755 		int i;
1756 		for (i = 0; i < 0xf; i++) {
1757 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1758 			    == -1) {
1759 				err = got_error_from_errno("asprintf");
1760 				goto done;
1761 			}
1762 			err = match_packed_object(id, repo, object_dir,
1763 			    obj_type);
1764 			if (err)
1765 				goto done;
1766 			err = match_loose_object(id, path_objects, object_dir,
1767 			    id_str_prefix, obj_type, repo);
1768 			if (err)
1769 				goto done;
1770 		}
1771 	} else {
1772 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1773 		goto done;
1774 	}
1775 done:
1776 	free(object_dir);
1777 	if (err) {
1778 		free(*id);
1779 		*id = NULL;
1780 	} else if (*id == NULL) {
1781 		switch (obj_type) {
1782 		case GOT_OBJ_TYPE_BLOB:
1783 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1784 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1785 			break;
1786 		case GOT_OBJ_TYPE_TREE:
1787 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1788 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1789 			break;
1790 		case GOT_OBJ_TYPE_COMMIT:
1791 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1792 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1793 			break;
1794 		case GOT_OBJ_TYPE_TAG:
1795 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1796 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1797 			break;
1798 		default:
1799 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1800 			break;
1801 		}
1802 	}
1803 
1804 	return err;
1805 }
1806 
1807 const struct got_error *
1808 got_repo_match_object_id(struct got_object_id **id, char **label,
1809     const char *id_str, int obj_type, struct got_reflist_head *refs,
1810     struct got_repository *repo)
1811 {
1812 	const struct got_error *err;
1813 	struct got_tag_object *tag;
1814 	struct got_reference *ref = NULL;
1815 
1816 	*id = NULL;
1817 	if (label)
1818 		*label = NULL;
1819 
1820 	if (refs) {
1821 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1822 		    refs, repo);
1823 		if (err == NULL) {
1824 			*id = got_object_id_dup(
1825 			    got_object_tag_get_object_id(tag));
1826 			if (*id == NULL)
1827 				err = got_error_from_errno("got_object_id_dup");
1828 			else if (label && asprintf(label, "refs/tags/%s",
1829 			    got_object_tag_get_name(tag)) == -1) {
1830 				err = got_error_from_errno("asprintf");
1831 				free(*id);
1832 				*id = NULL;
1833 			}
1834 			got_object_tag_close(tag);
1835 			return err;
1836 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1837 		    err->code != GOT_ERR_NO_OBJ)
1838 			return err;
1839 	}
1840 
1841 	err = got_ref_open(&ref, repo, id_str, 0);
1842 	if (err == NULL) {
1843 		err = got_ref_resolve(id, repo, ref);
1844 		if (err)
1845 			goto done;
1846 		if (label) {
1847 			*label = strdup(got_ref_get_name(ref));
1848 			if (*label == NULL) {
1849 				err = got_error_from_errno("strdup");
1850 				goto done;
1851 			}
1852 		}
1853 	} else {
1854 		if (err->code != GOT_ERR_NOT_REF &&
1855 		    err->code != GOT_ERR_BAD_REF_NAME)
1856 			goto done;
1857 		err = got_repo_match_object_id_prefix(id, id_str,
1858 		    obj_type, repo);
1859 		if (err) {
1860 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1861 				err = got_error_not_ref(id_str);
1862 			goto done;
1863 		}
1864 		if (label) {
1865 			err = got_object_id_str(label, *id);
1866 			if (*label == NULL) {
1867 				err = got_error_from_errno("strdup");
1868 				goto done;
1869 			}
1870 		}
1871 	}
1872 done:
1873 	if (ref)
1874 		got_ref_close(ref);
1875 	return err;
1876 }
1877 
1878 const struct got_error *
1879 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1880     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1881 {
1882 	const struct got_error *err = NULL;
1883 	struct got_reflist_entry *re;
1884 	struct got_object_id *tag_id;
1885 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1886 
1887 	*tag = NULL;
1888 
1889 	TAILQ_FOREACH(re, refs, entry) {
1890 		const char *refname;
1891 		refname = got_ref_get_name(re->ref);
1892 		if (got_ref_is_symbolic(re->ref))
1893 			continue;
1894 		if (strncmp(refname, "refs/tags/", 10) != 0)
1895 			continue;
1896 		if (!name_is_absolute)
1897 			refname += strlen("refs/tags/");
1898 		if (strcmp(refname, name) != 0)
1899 			continue;
1900 		err = got_ref_resolve(&tag_id, repo, re->ref);
1901 		if (err)
1902 			break;
1903 		err = got_object_open_as_tag(tag, repo, tag_id);
1904 		free(tag_id);
1905 		if (err)
1906 			break;
1907 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1908 		    got_object_tag_get_object_type(*tag) == obj_type)
1909 			break;
1910 		got_object_tag_close(*tag);
1911 		*tag = NULL;
1912 	}
1913 
1914 	if (err == NULL && *tag == NULL)
1915 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1916 		    GOT_OBJ_LABEL_TAG, name);
1917 	return err;
1918 }
1919 
1920 static const struct got_error *
1921 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1922     const char *name, mode_t mode, struct got_object_id *blob_id)
1923 {
1924 	const struct got_error *err = NULL;
1925 
1926 	 *new_te = NULL;
1927 
1928 	*new_te = calloc(1, sizeof(**new_te));
1929 	if (*new_te == NULL)
1930 		return got_error_from_errno("calloc");
1931 
1932 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1933 	    sizeof((*new_te)->name)) {
1934 		err = got_error(GOT_ERR_NO_SPACE);
1935 		goto done;
1936 	}
1937 
1938 	if (S_ISLNK(mode)) {
1939 		(*new_te)->mode = S_IFLNK;
1940 	} else {
1941 		(*new_te)->mode = S_IFREG;
1942 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1943 	}
1944 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1945 done:
1946 	if (err && *new_te) {
1947 		free(*new_te);
1948 		*new_te = NULL;
1949 	}
1950 	return err;
1951 }
1952 
1953 static const struct got_error *
1954 import_file(struct got_tree_entry **new_te, struct dirent *de,
1955     const char *path, struct got_repository *repo)
1956 {
1957 	const struct got_error *err;
1958 	struct got_object_id *blob_id = NULL;
1959 	char *filepath;
1960 	struct stat sb;
1961 
1962 	if (asprintf(&filepath, "%s%s%s", path,
1963 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1964 		return got_error_from_errno("asprintf");
1965 
1966 	if (lstat(filepath, &sb) != 0) {
1967 		err = got_error_from_errno2("lstat", path);
1968 		goto done;
1969 	}
1970 
1971 	err = got_object_blob_create(&blob_id, filepath, repo);
1972 	if (err)
1973 		goto done;
1974 
1975 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1976 	    blob_id);
1977 done:
1978 	free(filepath);
1979 	if (err)
1980 		free(blob_id);
1981 	return err;
1982 }
1983 
1984 static const struct got_error *
1985 insert_tree_entry(struct got_tree_entry *new_te,
1986     struct got_pathlist_head *paths)
1987 {
1988 	const struct got_error *err = NULL;
1989 	struct got_pathlist_entry *new_pe;
1990 
1991 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1992 	if (err)
1993 		return err;
1994 	if (new_pe == NULL)
1995 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1996 	return NULL;
1997 }
1998 
1999 static const struct got_error *write_tree(struct got_object_id **,
2000     const char *, struct got_pathlist_head *, struct got_repository *,
2001     got_repo_import_cb progress_cb, void *progress_arg);
2002 
2003 static const struct got_error *
2004 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2005     const char *path, struct got_pathlist_head *ignores,
2006     struct got_repository *repo,
2007     got_repo_import_cb progress_cb, void *progress_arg)
2008 {
2009 	const struct got_error *err;
2010 	struct got_object_id *id = NULL;
2011 	char *subdirpath;
2012 
2013 	if (asprintf(&subdirpath, "%s%s%s", path,
2014 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2015 		return got_error_from_errno("asprintf");
2016 
2017 	(*new_te) = calloc(1, sizeof(**new_te));
2018 	if (*new_te == NULL)
2019 		return got_error_from_errno("calloc");
2020 	(*new_te)->mode = S_IFDIR;
2021 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2022 	    sizeof((*new_te)->name)) {
2023 		err = got_error(GOT_ERR_NO_SPACE);
2024 		goto done;
2025 	}
2026 	err = write_tree(&id, subdirpath, ignores,  repo,
2027 	    progress_cb, progress_arg);
2028 	if (err)
2029 		goto done;
2030 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2031 	
2032 done:
2033 	free(id);
2034 	free(subdirpath);
2035 	if (err) {
2036 		free(*new_te);
2037 		*new_te = NULL;
2038 	}
2039 	return err;
2040 }
2041 
2042 static const struct got_error *
2043 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2044     struct got_pathlist_head *ignores, struct got_repository *repo,
2045     got_repo_import_cb progress_cb, void *progress_arg)
2046 {
2047 	const struct got_error *err = NULL;
2048 	DIR *dir;
2049 	struct dirent *de;
2050 	int nentries;
2051 	struct got_tree_entry *new_te = NULL;
2052 	struct got_pathlist_head paths;
2053 	struct got_pathlist_entry *pe;
2054 
2055 	*new_tree_id = NULL;
2056 
2057 	TAILQ_INIT(&paths);
2058 
2059 	dir = opendir(path_dir);
2060 	if (dir == NULL) {
2061 		err = got_error_from_errno2("opendir", path_dir);
2062 		goto done;
2063 	}
2064 
2065 	nentries = 0;
2066 	while ((de = readdir(dir)) != NULL) {
2067 		int ignore = 0;
2068 		int type;
2069 
2070 		if (strcmp(de->d_name, ".") == 0 ||
2071 		    strcmp(de->d_name, "..") == 0)
2072 			continue;
2073 
2074 		TAILQ_FOREACH(pe, ignores, entry) {
2075 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2076 				ignore = 1;
2077 				break;
2078 			}
2079 		}
2080 		if (ignore)
2081 			continue;
2082 
2083 		err = got_path_dirent_type(&type, path_dir, de);
2084 		if (err)
2085 			goto done;
2086 
2087 		if (type == DT_DIR) {
2088 			err = import_subdir(&new_te, de, path_dir,
2089 			    ignores, repo, progress_cb, progress_arg);
2090 			if (err) {
2091 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2092 					goto done;
2093 				err = NULL;
2094 				continue;
2095 			}
2096 		} else if (type == DT_REG || type == DT_LNK) {
2097 			err = import_file(&new_te, de, path_dir, repo);
2098 			if (err)
2099 				goto done;
2100 		} else
2101 			continue;
2102 
2103 		err = insert_tree_entry(new_te, &paths);
2104 		if (err)
2105 			goto done;
2106 		nentries++;
2107 	}
2108 
2109 	if (TAILQ_EMPTY(&paths)) {
2110 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2111 		    "cannot create tree without any entries");
2112 		goto done;
2113 	}
2114 
2115 	TAILQ_FOREACH(pe, &paths, entry) {
2116 		struct got_tree_entry *te = pe->data;
2117 		char *path;
2118 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2119 			continue;
2120 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2121 			err = got_error_from_errno("asprintf");
2122 			goto done;
2123 		}
2124 		err = (*progress_cb)(progress_arg, path);
2125 		free(path);
2126 		if (err)
2127 			goto done;
2128 	}
2129 
2130 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2131 done:
2132 	if (dir)
2133 		closedir(dir);
2134 	got_pathlist_free(&paths);
2135 	return err;
2136 }
2137 
2138 const struct got_error *
2139 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2140     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2141     struct got_repository *repo, got_repo_import_cb progress_cb,
2142     void *progress_arg)
2143 {
2144 	const struct got_error *err;
2145 	struct got_object_id *new_tree_id;
2146 
2147 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2148 	    progress_cb, progress_arg);
2149 	if (err)
2150 		return err;
2151 
2152 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2153 	    author, time(NULL), author, time(NULL), logmsg, repo);
2154 	free(new_tree_id);
2155 	return err;
2156 }
2157 
2158 const struct got_error *
2159 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2160     struct got_repository *repo)
2161 {
2162 	const struct got_error *err = NULL;
2163 	char *path_objects = NULL, *path = NULL;
2164 	DIR *dir = NULL;
2165 	struct got_object_id id;
2166 	int i;
2167 
2168 	*nobjects = 0;
2169 	*ondisk_size = 0;
2170 
2171 	path_objects = got_repo_get_path_objects(repo);
2172 	if (path_objects == NULL)
2173 		return got_error_from_errno("got_repo_get_path_objects");
2174 
2175 	for (i = 0; i <= 0xff; i++) {
2176 		struct dirent *dent;
2177 
2178 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2179 			err = got_error_from_errno("asprintf");
2180 			break;
2181 		}
2182 
2183 		dir = opendir(path);
2184 		if (dir == NULL) {
2185 			if (errno == ENOENT) {
2186 				err = NULL;
2187 				continue;
2188 			}
2189 			err = got_error_from_errno2("opendir", path);
2190 			break;
2191 		}
2192 
2193 		while ((dent = readdir(dir)) != NULL) {
2194 			char *id_str;
2195 			int fd;
2196 			struct stat sb;
2197 
2198 			if (strcmp(dent->d_name, ".") == 0 ||
2199 			    strcmp(dent->d_name, "..") == 0)
2200 				continue;
2201 
2202 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2203 				err = got_error_from_errno("asprintf");
2204 				goto done;
2205 			}
2206 
2207 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2208 				free(id_str);
2209 				continue;
2210 			}
2211 			free(id_str);
2212 
2213 			err = got_object_open_loose_fd(&fd, &id, repo);
2214 			if (err)
2215 				goto done;
2216 
2217 			if (fstat(fd, &sb) == -1) {
2218 				err = got_error_from_errno("fstat");
2219 				close(fd);
2220 				goto done;
2221 			}
2222 			(*nobjects)++;
2223 			(*ondisk_size) += sb.st_size;
2224 
2225 			if (close(fd) == -1) {
2226 				err = got_error_from_errno("close");
2227 				goto done;
2228 			}
2229 		}
2230 
2231 		if (closedir(dir) != 0) {
2232 			err = got_error_from_errno("closedir");
2233 			goto done;
2234 		}
2235 		dir = NULL;
2236 
2237 		free(path);
2238 		path = NULL;
2239 	}
2240 done:
2241 	if (dir && closedir(dir) != 0 && err == NULL)
2242 		err = got_error_from_errno("closedir");
2243 
2244 	if (err) {
2245 		*nobjects = 0;
2246 		*ondisk_size = 0;
2247 	}
2248 	free(path_objects);
2249 	free(path);
2250 	return err;
2251 }
2252 
2253 const struct got_error *
2254 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2255     off_t *total_packsize, struct got_repository *repo)
2256 {
2257 	const struct got_error *err = NULL;
2258 	DIR *packdir = NULL;
2259 	struct dirent *dent;
2260 	struct got_packidx *packidx = NULL;
2261 	char *path_packidx;
2262 	char *path_packfile;
2263 	int packdir_fd;
2264 	struct stat sb;
2265 
2266 	*npackfiles = 0;
2267 	*nobjects = 0;
2268 	*total_packsize = 0;
2269 
2270 	packdir_fd = openat(got_repo_get_fd(repo),
2271 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2272 	if (packdir_fd == -1) {
2273 		return got_error_from_errno_fmt("openat: %s/%s",
2274 		    got_repo_get_path_git_dir(repo),
2275 		    GOT_OBJECTS_PACK_DIR);
2276 	}
2277 
2278 	packdir = fdopendir(packdir_fd);
2279 	if (packdir == NULL) {
2280 		err = got_error_from_errno("fdopendir");
2281 		goto done;
2282 	}
2283 
2284 	while ((dent = readdir(packdir)) != NULL) {
2285 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2286 			continue;
2287 
2288 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2289 		    dent->d_name) == -1) {
2290 			err = got_error_from_errno("asprintf");
2291 			goto done;
2292 		}
2293 
2294 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2295 		    path_packidx, 0);
2296 		free(path_packidx);
2297 		if (err)
2298 			goto done;
2299 
2300 		if (fstat(packidx->fd, &sb) == -1)
2301 			goto done;
2302 		*total_packsize += sb.st_size;
2303 
2304 		err = got_packidx_get_packfile_path(&path_packfile,
2305 		    packidx->path_packidx);
2306 		if (err)
2307 			goto done;
2308 
2309 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2310 		    0) == -1) {
2311 			free(path_packfile);
2312 			goto done;
2313 		}
2314 		free(path_packfile);
2315 		*total_packsize += sb.st_size;
2316 
2317 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2318 
2319 		(*npackfiles)++;
2320 
2321 		got_packidx_close(packidx);
2322 		packidx = NULL;
2323 	}
2324 done:
2325 	if (packidx)
2326 		got_packidx_close(packidx);
2327 	if (packdir && closedir(packdir) != 0 && err == NULL)
2328 		err = got_error_from_errno("closedir");
2329 	if (err) {
2330 		*npackfiles = 0;
2331 		*nobjects = 0;
2332 		*total_packsize = 0;
2333 	}
2334 	return err;
2335 }
2336 
2337 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2338     got_packidx_bloom_filter_cmp);