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 <sha2.h>
36 #include <string.h>
37 #include <time.h>
38 #include <unistd.h>
39 #include <zlib.h>
40 #include <errno.h>
41 #include <libgen.h>
42 #include <stdint.h>
43 #include <imsg.h>
44 #include <uuid.h>
45 
46 #include "bloom.h"
47 
48 #include "got_error.h"
49 #include "got_reference.h"
50 #include "got_repository.h"
51 #include "got_path.h"
52 #include "got_cancel.h"
53 #include "got_object.h"
54 #include "got_opentemp.h"
55 
56 #include "got_lib_delta.h"
57 #include "got_lib_delta_cache.h"
58 #include "got_lib_inflate.h"
59 #include "got_lib_object.h"
60 #include "got_lib_object_parse.h"
61 #include "got_lib_object_create.h"
62 #include "got_lib_pack.h"
63 #include "got_lib_privsep.h"
64 #include "got_lib_hash.h"
65 #include "got_lib_object_cache.h"
66 #include "got_lib_repository.h"
67 #include "got_lib_gotconfig.h"
68 
69 #ifndef nitems
70 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
71 #endif
72 
73 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
74 
75 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
76     got_packidx_bloom_filter_cmp);
77 
78 static inline int
79 is_boolean_val(const char *val)
80 {
81 	return (strcasecmp(val, "true") == 0 ||
82 	    strcasecmp(val, "false") == 0 ||
83 	    strcasecmp(val, "on") == 0 ||
84     	    strcasecmp(val, "off") == 0 ||
85 	    strcasecmp(val, "yes") == 0 ||
86     	    strcasecmp(val, "no") == 0 ||
87 	    strcasecmp(val, "1") == 0 ||
88 	    strcasecmp(val, "0") == 0);
89 }
90 
91 static inline int
92 get_boolean_val(const char *val)
93 {
94 	return (strcasecmp(val, "true") == 0 ||
95 	    strcasecmp(val, "on") == 0 ||
96 	    strcasecmp(val, "yes") == 0 ||
97 	    strcasecmp(val, "1") == 0);
98 }
99 
100 const char *
101 got_repo_get_path(struct got_repository *repo)
102 {
103 	return repo->path;
104 }
105 
106 const char *
107 got_repo_get_path_git_dir(struct got_repository *repo)
108 {
109 	return repo->path_git_dir;
110 }
111 
112 int
113 got_repo_get_fd(struct got_repository *repo)
114 {
115 	return repo->gitdir_fd;
116 }
117 
118 enum got_hash_algorithm
119 got_repo_get_object_format(struct got_repository *repo)
120 {
121 	return repo->algo;
122 }
123 
124 const char *
125 got_repo_get_gitconfig_author_name(struct got_repository *repo)
126 {
127 	return repo->gitconfig_author_name;
128 }
129 
130 const char *
131 got_repo_get_gitconfig_author_email(struct got_repository *repo)
132 {
133 	return repo->gitconfig_author_email;
134 }
135 
136 const char *
137 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
138 {
139 	return repo->global_gitconfig_author_name;
140 }
141 
142 const char *
143 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
144 {
145 	return repo->global_gitconfig_author_email;
146 }
147 
148 const char *
149 got_repo_get_gitconfig_owner(struct got_repository *repo)
150 {
151 	return repo->gitconfig_owner;
152 }
153 
154 int
155 got_repo_has_extension(struct got_repository *repo, const char *ext)
156 {
157 	int i;
158 
159 	for (i = 0; i < repo->nextensions; ++i) {
160 		if (!strcasecmp(ext, repo->extnames[i]))
161 			return get_boolean_val(repo->extvals[i]);
162 	}
163 
164 	return 0;
165 }
166 
167 int
168 got_repo_is_bare(struct got_repository *repo)
169 {
170 	return (strcmp(repo->path, repo->path_git_dir) == 0);
171 }
172 
173 static char *
174 get_path_git_child(struct got_repository *repo, const char *basename)
175 {
176 	char *path_child;
177 
178 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
179 	    basename) == -1)
180 		return NULL;
181 
182 	return path_child;
183 }
184 
185 char *
186 got_repo_get_path_objects(struct got_repository *repo)
187 {
188 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
189 }
190 
191 char *
192 got_repo_get_path_objects_pack(struct got_repository *repo)
193 {
194 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
195 }
196 
197 char *
198 got_repo_get_path_refs(struct got_repository *repo)
199 {
200 	return get_path_git_child(repo, GOT_REFS_DIR);
201 }
202 
203 char *
204 got_repo_get_path_packed_refs(struct got_repository *repo)
205 {
206 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
207 }
208 
209 static char *
210 get_path_head(struct got_repository *repo)
211 {
212 	return get_path_git_child(repo, GOT_HEAD_FILE);
213 }
214 
215 char *
216 got_repo_get_path_gitconfig(struct got_repository *repo)
217 {
218 	return get_path_git_child(repo, GOT_GITCONFIG);
219 }
220 
221 char *
222 got_repo_get_path_gotconfig(struct got_repository *repo)
223 {
224 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
225 }
226 
227 const struct got_gotconfig *
228 got_repo_get_gotconfig(struct got_repository *repo)
229 {
230 	return repo->gotconfig;
231 }
232 
233 void
234 got_repo_get_gitconfig_remotes(int *nremotes,
235     const struct got_remote_repo **remotes, struct got_repository *repo)
236 {
237 	*nremotes = repo->ngitconfig_remotes;
238 	*remotes = repo->gitconfig_remotes;
239 }
240 
241 static int
242 is_git_repo(struct got_repository *repo)
243 {
244 	const char *path_git = got_repo_get_path_git_dir(repo);
245 	char *path_objects = got_repo_get_path_objects(repo);
246 	char *path_refs = got_repo_get_path_refs(repo);
247 	char *path_head = get_path_head(repo);
248 	int ret = 0;
249 	struct stat sb;
250 	struct got_reference *head_ref;
251 
252 	if (lstat(path_git, &sb) == -1)
253 		goto done;
254 	if (!S_ISDIR(sb.st_mode))
255 		goto done;
256 
257 	if (lstat(path_objects, &sb) == -1)
258 		goto done;
259 	if (!S_ISDIR(sb.st_mode))
260 		goto done;
261 
262 	if (lstat(path_refs, &sb) == -1)
263 		goto done;
264 	if (!S_ISDIR(sb.st_mode))
265 		goto done;
266 
267 	if (lstat(path_head, &sb) == -1)
268 		goto done;
269 	if (!S_ISREG(sb.st_mode))
270 		goto done;
271 
272 	/* Check if the HEAD reference can be opened. */
273 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
274 		goto done;
275 	got_ref_close(head_ref);
276 
277 	ret = 1;
278 done:
279 	free(path_objects);
280 	free(path_refs);
281 	free(path_head);
282 	return ret;
283 
284 }
285 
286 static const struct got_error *
287 close_tempfiles(int *fds, size_t nfds)
288 {
289 	const struct got_error *err = NULL;
290 	int i;
291 
292 	for (i = 0; i < nfds; i++) {
293 		if (fds[i] == -1)
294 			continue;
295 		if (close(fds[i]) == -1) {
296 			err = got_error_from_errno("close");
297 			break;
298 		}
299 	}
300 	free(fds);
301 	return err;
302 }
303 
304 static const struct got_error *
305 open_tempfiles(int **fds, size_t array_size, size_t nfds)
306 {
307 	const struct got_error *err = NULL;
308 	int i;
309 
310 	*fds = calloc(array_size, sizeof(**fds));
311 	if (*fds == NULL)
312 		return got_error_from_errno("calloc");
313 
314 	for (i = 0; i < array_size; i++)
315 		(*fds)[i] = -1;
316 
317 	for (i = 0; i < nfds; i++) {
318 		(*fds)[i] = got_opentempfd();
319 		if ((*fds)[i] == -1) {
320 			err = got_error_from_errno("got_opentempfd");
321 			close_tempfiles(*fds, nfds);
322 			*fds = NULL;
323 			return err;
324 		}
325 	}
326 
327 	return NULL;
328 }
329 
330 static const struct got_error *
331 get_pack_cache_size(int *pack_cache_size)
332 {
333 	struct rlimit rl;
334 
335 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
336 		return got_error_from_errno("getrlimit");
337 
338 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
339 	if (*pack_cache_size > rl.rlim_cur / 8)
340 		*pack_cache_size = rl.rlim_cur / 8;
341 
342 	return NULL;
343 }
344 
345 const struct got_error *
346 got_repo_pack_fds_open(int **pack_fds)
347 {
348 	const struct got_error *err;
349 	int nfds;
350 
351 	err = get_pack_cache_size(&nfds);
352 	if (err)
353 		return err;
354 
355 	/*
356 	 * We need one basefd and one accumfd per cached pack.
357 	 * Our constants should be set up in a way such that
358 	 * this error never triggers.
359 	 */
360 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
361 		return got_error(GOT_ERR_NO_SPACE);
362 
363 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
364 }
365 
366 const struct got_error *
367 got_repo_pack_fds_close(int *pack_fds)
368 {
369 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
370 }
371 
372 const struct got_error *
373 got_repo_temp_fds_open(int **temp_fds)
374 {
375 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
376 	    GOT_REPO_NUM_TEMPFILES);
377 }
378 
379 void
380 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
381 {
382 	int i;
383 
384 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
385 		repo->tempfiles[i] = temp_fds[i];
386 }
387 
388 const struct got_error *
389 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
390 {
391 	int i;
392 
393 	*fd = -1;
394 	*idx = -1;
395 
396 	for (i = 0; i < nitems(repo->tempfiles); i++) {
397 		if (repo->tempfile_use_mask & (1 << i))
398 			continue;
399 		if (repo->tempfiles[i] != -1) {
400 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
401 				return got_error_from_errno("ftruncate");
402 			*fd = repo->tempfiles[i];
403 			*idx = i;
404 			repo->tempfile_use_mask |= (1 << i);
405 			return NULL;
406 		}
407 	}
408 
409 	return got_error(GOT_ERR_REPO_TEMPFILE);
410 }
411 
412 void
413 got_repo_temp_fds_put(int idx, struct got_repository *repo)
414 {
415 	repo->tempfile_use_mask &= ~(1 << idx);
416 }
417 
418 const struct got_error *
419 got_repo_temp_fds_close(int *temp_fds)
420 {
421 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
422 }
423 
424 const struct got_error *
425 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
426     struct got_object *obj)
427 {
428 #ifndef GOT_NO_OBJ_CACHE
429 	const struct got_error *err = NULL;
430 	err = got_object_cache_add(&repo->objcache, id, obj);
431 	if (err) {
432 		if (err->code == GOT_ERR_OBJ_EXISTS ||
433 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
434 			err = NULL;
435 		return err;
436 	}
437 	obj->refcnt++;
438 #endif
439 	return NULL;
440 }
441 
442 struct got_object *
443 got_repo_get_cached_object(struct got_repository *repo,
444     struct got_object_id *id)
445 {
446 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
447 }
448 
449 const struct got_error *
450 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
451     struct got_tree_object *tree)
452 {
453 #ifndef GOT_NO_OBJ_CACHE
454 	const struct got_error *err = NULL;
455 	err = got_object_cache_add(&repo->treecache, id, tree);
456 	if (err) {
457 		if (err->code == GOT_ERR_OBJ_EXISTS ||
458 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
459 			err = NULL;
460 		return err;
461 	}
462 	tree->refcnt++;
463 #endif
464 	return NULL;
465 }
466 
467 struct got_tree_object *
468 got_repo_get_cached_tree(struct got_repository *repo,
469     struct got_object_id *id)
470 {
471 	return (struct got_tree_object *)got_object_cache_get(
472 	    &repo->treecache, id);
473 }
474 
475 const struct got_error *
476 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
477     struct got_commit_object *commit)
478 {
479 #ifndef GOT_NO_OBJ_CACHE
480 	const struct got_error *err = NULL;
481 	err = got_object_cache_add(&repo->commitcache, id, commit);
482 	if (err) {
483 		if (err->code == GOT_ERR_OBJ_EXISTS ||
484 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
485 			err = NULL;
486 		return err;
487 	}
488 	commit->refcnt++;
489 #endif
490 	return NULL;
491 }
492 
493 struct got_commit_object *
494 got_repo_get_cached_commit(struct got_repository *repo,
495     struct got_object_id *id)
496 {
497 	return (struct got_commit_object *)got_object_cache_get(
498 	    &repo->commitcache, id);
499 }
500 
501 const struct got_error *
502 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
503     struct got_tag_object *tag)
504 {
505 #ifndef GOT_NO_OBJ_CACHE
506 	const struct got_error *err = NULL;
507 	err = got_object_cache_add(&repo->tagcache, id, tag);
508 	if (err) {
509 		if (err->code == GOT_ERR_OBJ_EXISTS ||
510 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
511 			err = NULL;
512 		return err;
513 	}
514 	tag->refcnt++;
515 #endif
516 	return NULL;
517 }
518 
519 struct got_tag_object *
520 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
521 {
522 	return (struct got_tag_object *)got_object_cache_get(
523 	    &repo->tagcache, id);
524 }
525 
526 const struct got_error *
527 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
528     struct got_raw_object *raw)
529 {
530 #ifndef GOT_NO_OBJ_CACHE
531 	const struct got_error *err = NULL;
532 	err = got_object_cache_add(&repo->rawcache, id, raw);
533 	if (err) {
534 		if (err->code == GOT_ERR_OBJ_EXISTS ||
535 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
536 			err = NULL;
537 		return err;
538 	}
539 	raw->refcnt++;
540 #endif
541 	return NULL;
542 }
543 
544 
545 struct got_raw_object *
546 got_repo_get_cached_raw_object(struct got_repository *repo,
547     struct got_object_id *id)
548 {
549 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
550 }
551 
552 
553 static const struct got_error *
554 open_repo(struct got_repository *repo, const char *path)
555 {
556 	const struct got_error *err = NULL;
557 
558 	repo->gitdir_fd = -1;
559 
560 	/* bare git repository? */
561 	repo->path_git_dir = strdup(path);
562 	if (repo->path_git_dir == NULL)
563 		return got_error_from_errno("strdup");
564 	if (is_git_repo(repo)) {
565 		repo->path = strdup(repo->path_git_dir);
566 		if (repo->path == NULL) {
567 			err = got_error_from_errno("strdup");
568 			goto done;
569 		}
570 		repo->gitdir_fd = open(repo->path_git_dir,
571 		    O_DIRECTORY | O_CLOEXEC);
572 		if (repo->gitdir_fd == -1) {
573 			err = got_error_from_errno2("open",
574 			    repo->path_git_dir);
575 			goto done;
576 		}
577 		return NULL;
578 	}
579 
580 	/* git repository with working tree? */
581 	free(repo->path_git_dir);
582 	repo->path_git_dir = NULL;
583 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
584 		err = got_error_from_errno("asprintf");
585 		goto done;
586 	}
587 	if (is_git_repo(repo)) {
588 		repo->path = strdup(path);
589 		if (repo->path == NULL) {
590 			err = got_error_from_errno("strdup");
591 			goto done;
592 		}
593 		repo->gitdir_fd = open(repo->path_git_dir,
594 		    O_DIRECTORY | O_CLOEXEC);
595 		if (repo->gitdir_fd == -1) {
596 			err = got_error_from_errno2("open",
597 			    repo->path_git_dir);
598 			goto done;
599 		}
600 		return NULL;
601 	}
602 
603 	err = got_error(GOT_ERR_NOT_GIT_REPO);
604 done:
605 	if (err) {
606 		free(repo->path);
607 		repo->path = NULL;
608 		free(repo->path_git_dir);
609 		repo->path_git_dir = NULL;
610 		if (repo->gitdir_fd != -1)
611 			close(repo->gitdir_fd);
612 		repo->gitdir_fd = -1;
613 
614 	}
615 	return err;
616 }
617 
618 static const struct got_error *
619 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
620 {
621 	const struct got_error *err = NULL;
622 	char *repo_gitconfig_path = NULL;
623 
624 	if (global_gitconfig_path) {
625 		/* Read settings from ~/.gitconfig. */
626 		int dummy_repo_version;
627 		err = got_repo_read_gitconfig(&dummy_repo_version,
628 		    &repo->global_gitconfig_author_name,
629 		    &repo->global_gitconfig_author_email,
630 		    NULL, NULL, NULL, NULL, NULL, NULL,
631 		    global_gitconfig_path);
632 		if (err)
633 			return err;
634 	}
635 
636 	/* Read repository's .git/config file. */
637 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
638 	if (repo_gitconfig_path == NULL)
639 		return got_error_from_errno("got_repo_get_path_gitconfig");
640 
641 	err = got_repo_read_gitconfig(
642 	    &repo->gitconfig_repository_format_version,
643 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
644 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
645 	    &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
646 	    &repo->nextensions, repo_gitconfig_path);
647 	if (err)
648 		goto done;
649 
650 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
651 		int i;
652 
653 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
654 			got_repo_free_remote_repo_data(
655 			    &repo->gitconfig_remotes[i]);
656 		}
657 		free(repo->gitconfig_remotes);
658 		repo->gitconfig_remotes = NULL;
659 		repo->ngitconfig_remotes = 0;
660 
661 		free(repo->gitconfig_author_name);
662 		repo->gitconfig_author_name = NULL;
663 		free(repo->gitconfig_author_email);
664 		repo->gitconfig_author_email = NULL;
665 
666 		free(repo->global_gitconfig_author_name);
667 		repo->global_gitconfig_author_name = NULL;
668 		free(repo->global_gitconfig_author_email);
669 		repo->global_gitconfig_author_email = NULL;
670 	}
671 
672 done:
673 	free(repo_gitconfig_path);
674 	return err;
675 }
676 
677 static const struct got_error *
678 read_gotconfig(struct got_repository *repo)
679 {
680 	const struct got_error *err = NULL;
681 	char *gotconfig_path;
682 
683 	gotconfig_path = got_repo_get_path_gotconfig(repo);
684 	if (gotconfig_path == NULL)
685 		return got_error_from_errno("got_repo_get_path_gotconfig");
686 
687 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
688 	free(gotconfig_path);
689 	return err;
690 }
691 
692 /* Supported repository format extensions. */
693 static const char *const repo_extensions[] = {
694 	"noop",			/* Got supports repository format version 1. */
695 	"preciousObjects",	/* Supported by gotadmin cleanup. */
696 	"worktreeConfig",	/* Got does not care about Git work trees. */
697 };
698 
699 const struct got_error *
700 got_repo_open(struct got_repository **repop, const char *path,
701     const char *global_gitconfig_path, int *pack_fds)
702 {
703 	struct got_repository *repo = NULL;
704 	const struct got_error *err = NULL;
705 	char *repo_path = NULL;
706 	size_t i, j = 0;
707 
708 	*repop = NULL;
709 
710 	repo = calloc(1, sizeof(*repo));
711 	if (repo == NULL)
712 		return got_error_from_errno("calloc");
713 
714 	RB_INIT(&repo->packidx_bloom_filters);
715 	TAILQ_INIT(&repo->packidx_paths);
716 
717 	for (i = 0; i < nitems(repo->privsep_children); i++) {
718 		memset(&repo->privsep_children[i], 0,
719 		    sizeof(repo->privsep_children[0]));
720 		repo->privsep_children[i].imsg_fd = -1;
721 	}
722 
723 	err = got_object_cache_init(&repo->objcache,
724 	    GOT_OBJECT_CACHE_TYPE_OBJ);
725 	if (err)
726 		goto done;
727 	err = got_object_cache_init(&repo->treecache,
728 	    GOT_OBJECT_CACHE_TYPE_TREE);
729 	if (err)
730 		goto done;
731 	err = got_object_cache_init(&repo->commitcache,
732 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
733 	if (err)
734 		goto done;
735 	err = got_object_cache_init(&repo->tagcache,
736 	    GOT_OBJECT_CACHE_TYPE_TAG);
737 	if (err)
738 		goto done;
739 	err = got_object_cache_init(&repo->rawcache,
740 	    GOT_OBJECT_CACHE_TYPE_RAW);
741 	if (err)
742 		goto done;
743 
744 	err = get_pack_cache_size(&repo->pack_cache_size);
745 	if (err)
746 		goto done;
747 	for (i = 0; i < nitems(repo->packs); i++) {
748 		if (pack_fds != NULL && i < repo->pack_cache_size) {
749 			repo->packs[i].basefd = pack_fds[j++];
750 			repo->packs[i].accumfd = pack_fds[j++];
751 		} else {
752 			repo->packs[i].basefd = -1;
753 			repo->packs[i].accumfd = -1;
754 		}
755 	}
756 	for (i = 0; i < nitems(repo->tempfiles); i++)
757 		repo->tempfiles[i] = -1;
758 	repo->pinned_pack = -1;
759 	repo->pinned_packidx = -1;
760 	repo->pinned_pid = 0;
761 
762 	repo_path = realpath(path, NULL);
763 	if (repo_path == NULL) {
764 		err = got_error_from_errno2("realpath", path);
765 		goto done;
766 	}
767 
768 	for (;;) {
769 		char *parent_path;
770 
771 		err = open_repo(repo, repo_path);
772 		if (err == NULL)
773 			break;
774 		if (err->code != GOT_ERR_NOT_GIT_REPO)
775 			goto done;
776 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
777 			err = got_error(GOT_ERR_NOT_GIT_REPO);
778 			goto done;
779 		}
780 		err = got_path_dirname(&parent_path, repo_path);
781 		if (err)
782 			goto done;
783 		free(repo_path);
784 		repo_path = parent_path;
785 	}
786 
787 	err = read_gotconfig(repo);
788 	if (err)
789 		goto done;
790 
791 	err = read_gitconfig(repo, global_gitconfig_path);
792 	if (err)
793 		goto done;
794 	if (repo->gitconfig_repository_format_version != 0) {
795 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
796 		goto done;
797 	}
798 	for (i = 0; i < repo->nextensions; i++) {
799 		char *ext = repo->extnames[i];
800 		char *val = repo->extvals[i];
801 		int j, supported = 0;
802 
803 		if (!is_boolean_val(val)) {
804 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
805 			goto done;
806 		}
807 
808 		if (!get_boolean_val(val))
809 			continue;
810 
811 		for (j = 0; j < nitems(repo_extensions); j++) {
812 			if (strcmp(ext, repo_extensions[j]) == 0) {
813 				supported = 1;
814 				break;
815 			}
816 		}
817 		if (!supported) {
818 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
819 			goto done;
820 		}
821 	}
822 
823 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
824 done:
825 	if (err)
826 		got_repo_close(repo);
827 	else
828 		*repop = repo;
829 	free(repo_path);
830 	return err;
831 }
832 
833 const struct got_error *
834 got_repo_close(struct got_repository *repo)
835 {
836 	const struct got_error *err = NULL, *child_err;
837 	struct got_packidx_bloom_filter *bf;
838 	size_t i;
839 
840 	for (i = 0; i < repo->pack_cache_size; i++) {
841 		if (repo->packidx_cache[i] == NULL)
842 			break;
843 		got_packidx_close(repo->packidx_cache[i]);
844 	}
845 
846 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
847 	    &repo->packidx_bloom_filters))) {
848 		RB_REMOVE(got_packidx_bloom_filter_tree,
849 		    &repo->packidx_bloom_filters, bf);
850 		bloom_free(bf->bloom);
851 		free(bf->bloom);
852 		free(bf);
853 	}
854 
855 	for (i = 0; i < repo->pack_cache_size; i++)
856 		if (repo->packs[i].path_packfile)
857 			if (repo->packs[i].path_packfile)
858 				got_pack_close(&repo->packs[i]);
859 
860 	free(repo->path);
861 	free(repo->path_git_dir);
862 
863 	got_object_cache_close(&repo->objcache);
864 	got_object_cache_close(&repo->treecache);
865 	got_object_cache_close(&repo->commitcache);
866 	got_object_cache_close(&repo->tagcache);
867 	got_object_cache_close(&repo->rawcache);
868 
869 	for (i = 0; i < nitems(repo->privsep_children); i++) {
870 		if (repo->privsep_children[i].imsg_fd == -1)
871 			continue;
872 		imsg_clear(repo->privsep_children[i].ibuf);
873 		free(repo->privsep_children[i].ibuf);
874 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
875 		if (err && err->code == GOT_ERR_EOF)
876 			err = NULL;
877 		child_err = got_privsep_wait_for_child(
878 		    repo->privsep_children[i].pid);
879 		if (child_err && err == NULL)
880 			err = child_err;
881 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
882 		    err == NULL)
883 			err = got_error_from_errno("close");
884 	}
885 
886 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
887 	    err == NULL)
888 		err = got_error_from_errno("close");
889 
890 	if (repo->gotconfig)
891 		got_gotconfig_free(repo->gotconfig);
892 	free(repo->gitconfig_author_name);
893 	free(repo->gitconfig_author_email);
894 	for (i = 0; i < repo->ngitconfig_remotes; i++)
895 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
896 	free(repo->gitconfig_remotes);
897 	for (i = 0; i < repo->nextensions; i++) {
898 		free(repo->extnames[i]);
899 		free(repo->extvals[i]);
900 	}
901 	free(repo->extnames);
902 	free(repo->extvals);
903 
904 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
905 	free(repo);
906 
907 	return err;
908 }
909 
910 const struct got_error *
911 got_repo_remote_repo_dup(struct got_remote_repo **newp,
912     const struct got_remote_repo *repo)
913 {
914 	const struct got_error *err = NULL;
915 	struct got_remote_repo *new;
916 	int i;
917 
918 	new = calloc(1, sizeof(*new));
919 	if (new == NULL)
920 		return got_error_from_errno("calloc");
921 	
922 	if (repo->name) {
923 		new->name = strdup(repo->name);
924 		if (new->name == NULL) {
925 			err = got_error_from_errno("strdup");
926 			goto done;
927 		}
928 	}
929 
930 	if (repo->fetch_url) {
931 		new->fetch_url = strdup(repo->fetch_url);
932 		if (new->fetch_url == NULL) {
933 			err = got_error_from_errno("strdup");
934 			goto done;
935 		}
936 	}
937 
938 	if (repo->send_url) {
939 		new->send_url = strdup(repo->send_url);
940 		if (new->send_url == NULL) {
941 			err = got_error_from_errno("strdup");
942 			goto done;
943 		}
944 	}
945 
946 	new->mirror_references = repo->mirror_references;
947 
948 	new->nfetch_branches = repo->nfetch_branches;
949 	if (repo->fetch_branches) {
950 		new->fetch_branches = calloc(repo->nfetch_branches,
951 		    sizeof(char *));
952 		if (new->fetch_branches == NULL) {
953 			err = got_error_from_errno("calloc");
954 			goto done;
955 		}
956 		for (i = 0; i < repo->nfetch_branches; i++) {
957 			new->fetch_branches[i] = strdup(
958 			    repo->fetch_branches[i]);
959 			if (new->fetch_branches[i] == NULL) {
960 				err = got_error_from_errno("strdup");
961 				goto done;
962 			}
963 		}
964 	}
965 
966 	new->nsend_branches = repo->nsend_branches;
967 	if (repo->send_branches) {
968 		new->send_branches = calloc(repo->nsend_branches,
969 		    sizeof(char *));
970 		if (new->send_branches == NULL) {
971 			err = got_error_from_errno("calloc");
972 			goto done;
973 		}
974 		for (i = 0; i < repo->nsend_branches; i++) {
975 			new->send_branches[i] = strdup(
976 			    repo->send_branches[i]);
977 			if (new->send_branches[i] == NULL) {
978 				err = got_error_from_errno("strdup");
979 				goto done;
980 			}
981 		}
982 	}
983 
984 	new->nfetch_refs = repo->nfetch_refs;
985 	if (repo->fetch_refs) {
986 		new->fetch_refs = calloc(repo->nfetch_refs,
987 		    sizeof(char *));
988 		if (new->fetch_refs == NULL) {
989 			err = got_error_from_errno("calloc");
990 			goto done;
991 		}
992 		for (i = 0; i < repo->nfetch_refs; i++) {
993 			new->fetch_refs[i] = strdup(
994 			    repo->fetch_refs[i]);
995 			if (new->fetch_refs[i] == NULL) {
996 				err = got_error_from_errno("strdup");
997 				goto done;
998 			}
999 		}
1000 	}
1001 done:
1002 	if (err) {
1003 		got_repo_free_remote_repo_data(new);
1004 		free(new);
1005 	} else
1006 		*newp = new;
1007 
1008 	return err;
1009 }
1010 
1011 void
1012 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
1013 {
1014 	int i;
1015 
1016 	if (repo == NULL)
1017 		return;
1018 
1019 	free(repo->name);
1020 	repo->name = NULL;
1021 	free(repo->fetch_url);
1022 	repo->fetch_url = NULL;
1023 	free(repo->send_url);
1024 	repo->send_url = NULL;
1025 	for (i = 0; i < repo->nfetch_branches; i++)
1026 		free(repo->fetch_branches[i]);
1027 	free(repo->fetch_branches);
1028 	repo->fetch_branches = NULL;
1029 	repo->nfetch_branches = 0;
1030 	for (i = 0; i < repo->nsend_branches; i++)
1031 		free(repo->send_branches[i]);
1032 	free(repo->send_branches);
1033 	repo->send_branches = NULL;
1034 	repo->nsend_branches = 0;
1035 	for (i = 0; i < repo->nfetch_refs; i++)
1036 		free(repo->fetch_refs[i]);
1037 	free(repo->fetch_refs);
1038 	repo->fetch_refs = NULL;
1039 	repo->nfetch_refs = 0;
1040 }
1041 
1042 const struct got_error *
1043 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
1044     const char *input_path)
1045 {
1046 	const struct got_error *err = NULL;
1047 	const char *repo_abspath = NULL;
1048 	size_t repolen, len;
1049 	char *canonpath, *path = NULL;
1050 
1051 	*in_repo_path = NULL;
1052 
1053 	canonpath = strdup(input_path);
1054 	if (canonpath == NULL) {
1055 		err = got_error_from_errno("strdup");
1056 		goto done;
1057 	}
1058 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
1059 	if (err)
1060 		goto done;
1061 
1062 	repo_abspath = got_repo_get_path(repo);
1063 
1064 	if (canonpath[0] == '\0') {
1065 		path = strdup(canonpath);
1066 		if (path == NULL) {
1067 			err = got_error_from_errno("strdup");
1068 			goto done;
1069 		}
1070 	} else {
1071 		path = realpath(canonpath, NULL);
1072 		if (path == NULL) {
1073 			if (errno != ENOENT) {
1074 				err = got_error_from_errno2("realpath",
1075 				    canonpath);
1076 				goto done;
1077 			}
1078 			/*
1079 			 * Path is not on disk.
1080 			 * Assume it is already relative to repository root.
1081 			 */
1082 			path = strdup(canonpath);
1083 			if (path == NULL) {
1084 				err = got_error_from_errno("strdup");
1085 				goto done;
1086 			}
1087 		}
1088 
1089 		repolen = strlen(repo_abspath);
1090 		len = strlen(path);
1091 
1092 
1093 		if (strcmp(path, repo_abspath) == 0) {
1094 			free(path);
1095 			path = strdup("");
1096 			if (path == NULL) {
1097 				err = got_error_from_errno("strdup");
1098 				goto done;
1099 			}
1100 		} else if (len > repolen &&
1101 		    got_path_is_child(path, repo_abspath, repolen)) {
1102 			/* Matched an on-disk path inside repository. */
1103 			if (got_repo_is_bare(repo)) {
1104 				/*
1105 				 * Matched an on-disk path inside repository
1106 				 * database. Treat input as repository-relative.
1107 				 */
1108 				free(path);
1109 				path = canonpath;
1110 				canonpath = NULL;
1111 			} else {
1112 				char *child;
1113 				/* Strip common prefix with repository path. */
1114 				err = got_path_skip_common_ancestor(&child,
1115 				    repo_abspath, path);
1116 				if (err)
1117 					goto done;
1118 				free(path);
1119 				path = child;
1120 			}
1121 		} else {
1122 			/*
1123 			 * Matched unrelated on-disk path.
1124 			 * Treat input as repository-relative.
1125 			 */
1126 			free(path);
1127 			path = canonpath;
1128 			canonpath = NULL;
1129 		}
1130 	}
1131 
1132 	/* Make in-repository path absolute */
1133 	if (path[0] != '/') {
1134 		char *abspath;
1135 		if (asprintf(&abspath, "/%s", path) == -1) {
1136 			err = got_error_from_errno("asprintf");
1137 			goto done;
1138 		}
1139 		free(path);
1140 		path = abspath;
1141 	}
1142 
1143 done:
1144 	free(canonpath);
1145 	if (err)
1146 		free(path);
1147 	else
1148 		*in_repo_path = path;
1149 	return err;
1150 }
1151 
1152 static const struct got_error *
1153 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1154     const char *path_packidx)
1155 {
1156 	const struct got_error *err = NULL;
1157 	size_t i;
1158 
1159 	for (i = 0; i < repo->pack_cache_size; i++) {
1160 		if (repo->packidx_cache[i] == NULL)
1161 			break;
1162 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1163 		    path_packidx) == 0) {
1164 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1165 		}
1166 	}
1167 	if (i == repo->pack_cache_size) {
1168 		do {
1169 			i--;
1170 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1171 		    i == repo->pinned_packidx);
1172 		err = got_packidx_close(repo->packidx_cache[i]);
1173 		if (err)
1174 			return err;
1175 	}
1176 
1177 	repo->packidx_cache[i] = packidx;
1178 
1179 	return NULL;
1180 }
1181 
1182 int
1183 got_repo_is_packidx_filename(const char *name, size_t len)
1184 {
1185 	if (len != GOT_PACKIDX_NAMELEN)
1186 		return 0;
1187 
1188 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1189 		return 0;
1190 
1191 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1192 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1193 		return 0;
1194 
1195 	return 1;
1196 }
1197 
1198 static struct got_packidx_bloom_filter *
1199 get_packidx_bloom_filter(struct got_repository *repo,
1200     const char *path, size_t path_len)
1201 {
1202 	struct got_packidx_bloom_filter key;
1203 
1204 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1205 		return NULL; /* XXX */
1206 	key.path_len = path_len;
1207 
1208 	return RB_FIND(got_packidx_bloom_filter_tree,
1209 	    &repo->packidx_bloom_filters, &key);
1210 }
1211 
1212 int
1213 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1214     const char *path_packidx, struct got_object_id *id)
1215 {
1216 	struct got_packidx_bloom_filter *bf;
1217 
1218 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1219 	if (bf)
1220 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1221 
1222 	/* No bloom filter means this pack index must be searched. */
1223 	return 1;
1224 }
1225 
1226 static const struct got_error *
1227 add_packidx_bloom_filter(struct got_repository *repo,
1228     struct got_packidx *packidx, const char *path_packidx)
1229 {
1230 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1231 	struct got_packidx_bloom_filter *bf;
1232 	size_t len;
1233 
1234 	/*
1235 	 * Don't use bloom filters for very large pack index files.
1236 	 * Large pack files will contain a relatively large fraction
1237 	 * of our objects so we will likely need to visit them anyway.
1238 	 * The more objects a pack file contains the higher the probability
1239 	 * of a false-positive match from the bloom filter. And reading
1240 	 * all object IDs from a large pack index file can be expensive.
1241 	 */
1242 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1243 		return NULL;
1244 
1245 	/* Do we already have a filter for this pack index? */
1246 	if (get_packidx_bloom_filter(repo, path_packidx,
1247 	    strlen(path_packidx)) != NULL)
1248 		return NULL;
1249 
1250 	bf = calloc(1, sizeof(*bf));
1251 	if (bf == NULL)
1252 		return got_error_from_errno("calloc");
1253 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1254 	if (bf->bloom == NULL) {
1255 		free(bf);
1256 		return got_error_from_errno("calloc");
1257 	}
1258 
1259 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1260 	if (len >= sizeof(bf->path)) {
1261 		free(bf->bloom);
1262 		free(bf);
1263 		return got_error(GOT_ERR_NO_SPACE);
1264 	}
1265 	bf->path_len = len;
1266 
1267 	/* Minimum size supported by our bloom filter is 1000 entries. */
1268 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1269 	for (i = 0; i < nobjects; i++) {
1270 		struct got_packidx_object_id *id;
1271 		id = &packidx->hdr.sorted_ids[i];
1272 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1273 	}
1274 
1275 	RB_INSERT(got_packidx_bloom_filter_tree,
1276 	    &repo->packidx_bloom_filters, bf);
1277 	return NULL;
1278 }
1279 
1280 static void
1281 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1282 {
1283 	struct got_pathlist_entry *pe;
1284 
1285 	while (!TAILQ_EMPTY(packidx_paths)) {
1286 		pe = TAILQ_FIRST(packidx_paths);
1287 		TAILQ_REMOVE(packidx_paths, pe, entry);
1288 		free((char *)pe->path);
1289 		free(pe);
1290 	}
1291 }
1292 
1293 static const struct got_error *
1294 refresh_packidx_paths(struct got_repository *repo)
1295 {
1296 	const struct got_error *err = NULL;
1297 	char *objects_pack_dir = NULL;
1298 	struct stat sb;
1299 
1300 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1301 	if (objects_pack_dir == NULL)
1302 		return got_error_from_errno("got_repo_get_path_objects_pack");
1303 
1304 	if (stat(objects_pack_dir, &sb) == -1) {
1305 		if (errno != ENOENT) {
1306 			err = got_error_from_errno2("stat", objects_pack_dir);
1307 			goto done;
1308 		}
1309 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1310 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1311 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1312 		purge_packidx_paths(&repo->packidx_paths);
1313 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1314 		if (err)
1315 			goto done;
1316 	}
1317 done:
1318 	free(objects_pack_dir);
1319 	return err;
1320 }
1321 
1322 const struct got_error *
1323 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1324     struct got_repository *repo, struct got_object_id *id)
1325 {
1326 	const struct got_error *err;
1327 	struct got_pathlist_entry *pe;
1328 	size_t i;
1329 
1330 	/* Search pack index cache. */
1331 	for (i = 0; i < repo->pack_cache_size; i++) {
1332 		if (repo->packidx_cache[i] == NULL)
1333 			break;
1334 		if (!got_repo_check_packidx_bloom_filter(repo,
1335 		    repo->packidx_cache[i]->path_packidx, id))
1336 			continue; /* object will not be found in this index */
1337 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1338 		if (*idx != -1) {
1339 			*packidx = repo->packidx_cache[i];
1340 			/*
1341 			 * Move this cache entry to the front. Repeatedly
1342 			 * searching a wrong pack index can be expensive.
1343 			 */
1344 			if (i > 0) {
1345 				memmove(&repo->packidx_cache[1],
1346 				    &repo->packidx_cache[0],
1347 				    i * sizeof(repo->packidx_cache[0]));
1348 				repo->packidx_cache[0] = *packidx;
1349 				if (repo->pinned_packidx >= 0 &&
1350 				    repo->pinned_packidx < i)
1351 					repo->pinned_packidx++;
1352 				else if (repo->pinned_packidx == i)
1353 					repo->pinned_packidx = 0;
1354 			}
1355 			return NULL;
1356 		}
1357 	}
1358 	/* No luck. Search the filesystem. */
1359 
1360 	err = refresh_packidx_paths(repo);
1361 	if (err)
1362 		return err;
1363 
1364 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1365 		const char *path_packidx = pe->path;
1366 		int is_cached = 0;
1367 
1368 		if (!got_repo_check_packidx_bloom_filter(repo,
1369 		    pe->path, id))
1370 			continue; /* object will not be found in this index */
1371 
1372 		for (i = 0; i < repo->pack_cache_size; i++) {
1373 			if (repo->packidx_cache[i] == NULL)
1374 				break;
1375 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1376 			    path_packidx) == 0) {
1377 				is_cached = 1;
1378 				break;
1379 			}
1380 		}
1381 		if (is_cached)
1382 			continue; /* already searched */
1383 
1384 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1385 		    path_packidx, 0);
1386 		if (err)
1387 			goto done;
1388 
1389 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1390 		if (err)
1391 			goto done;
1392 
1393 		err = cache_packidx(repo, *packidx, path_packidx);
1394 		if (err)
1395 			goto done;
1396 
1397 		*idx = got_packidx_get_object_idx(*packidx, id);
1398 		if (*idx != -1) {
1399 			err = NULL; /* found the object */
1400 			goto done;
1401 		}
1402 	}
1403 
1404 	err = got_error_no_obj(id);
1405 done:
1406 	return err;
1407 }
1408 
1409 const struct got_error *
1410 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1411     struct got_repository *repo)
1412 {
1413 	const struct got_error *err = NULL;
1414 	DIR *packdir = NULL;
1415 	struct dirent *dent;
1416 	char *path_packidx = NULL;
1417 	int packdir_fd;
1418 	struct stat sb;
1419 
1420 	packdir_fd = openat(got_repo_get_fd(repo),
1421 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1422 	if (packdir_fd == -1) {
1423 		return got_error_from_errno_fmt("openat: %s/%s",
1424 		    got_repo_get_path_git_dir(repo),
1425 		    GOT_OBJECTS_PACK_DIR);
1426 	}
1427 
1428 	packdir = fdopendir(packdir_fd);
1429 	if (packdir == NULL) {
1430 		err = got_error_from_errno("fdopendir");
1431 		goto done;
1432 	}
1433 
1434 	if (fstat(packdir_fd, &sb) == -1) {
1435 		err = got_error_from_errno("fstat");
1436 		goto done;
1437 	}
1438 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1439 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1440 
1441 	while ((dent = readdir(packdir)) != NULL) {
1442 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1443 			continue;
1444 
1445 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1446 		    dent->d_name) == -1) {
1447 			err = got_error_from_errno("asprintf");
1448 			path_packidx = NULL;
1449 			break;
1450 		}
1451 
1452 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1453 		if (err)
1454 			break;
1455 	}
1456 done:
1457 	if (err)
1458 		free(path_packidx);
1459 	if (packdir && closedir(packdir) != 0 && err == NULL)
1460 		err = got_error_from_errno("closedir");
1461 	return err;
1462 }
1463 
1464 const struct got_error *
1465 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1466     struct got_repository *repo)
1467 {
1468 	const struct got_error *err;
1469 	size_t i;
1470 
1471 	*packidx = NULL;
1472 
1473 	/* Search pack index cache. */
1474 	for (i = 0; i < repo->pack_cache_size; i++) {
1475 		if (repo->packidx_cache[i] == NULL)
1476 			break;
1477 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1478 		    path_packidx) == 0) {
1479 			*packidx = repo->packidx_cache[i];
1480 			return NULL;
1481 		}
1482 	}
1483 	/* No luck. Search the filesystem. */
1484 
1485 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1486 	    path_packidx, 0);
1487 	if (err)
1488 		return err;
1489 
1490 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1491 	if (err)
1492 		goto done;
1493 
1494 	err = cache_packidx(repo, *packidx, path_packidx);
1495 done:
1496 	if (err) {
1497 		got_packidx_close(*packidx);
1498 		*packidx = NULL;
1499 	}
1500 	return err;
1501 }
1502 
1503 static const struct got_error *
1504 read_packfile_hdr(int fd, struct got_packidx *packidx)
1505 {
1506 	const struct got_error *err = NULL;
1507 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1508 	struct got_packfile_hdr hdr;
1509 	ssize_t n;
1510 
1511 	n = read(fd, &hdr, sizeof(hdr));
1512 	if (n < 0)
1513 		return got_error_from_errno("read");
1514 	if (n != sizeof(hdr))
1515 		return got_error(GOT_ERR_BAD_PACKFILE);
1516 
1517 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1518 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1519 	    be32toh(hdr.nobjects) != totobj)
1520 		err = got_error(GOT_ERR_BAD_PACKFILE);
1521 
1522 	return err;
1523 }
1524 
1525 static const struct got_error *
1526 open_packfile(int *fd, struct got_repository *repo,
1527     const char *relpath, struct got_packidx *packidx)
1528 {
1529 	const struct got_error *err = NULL;
1530 
1531 	*fd = openat(got_repo_get_fd(repo), relpath,
1532 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1533 	if (*fd == -1)
1534 		return got_error_from_errno_fmt("openat: %s/%s",
1535 		    got_repo_get_path_git_dir(repo), relpath);
1536 
1537 	if (packidx) {
1538 		err = read_packfile_hdr(*fd, packidx);
1539 		if (err) {
1540 			close(*fd);
1541 			*fd = -1;
1542 		}
1543 	}
1544 
1545 	return err;
1546 }
1547 
1548 const struct got_error *
1549 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1550     const char *path_packfile, struct got_packidx *packidx)
1551 {
1552 	const struct got_error *err = NULL;
1553 	struct got_pack *pack = NULL;
1554 	struct stat sb;
1555 	size_t i;
1556 
1557 	if (packp)
1558 		*packp = NULL;
1559 
1560 	for (i = 0; i < repo->pack_cache_size; i++) {
1561 		pack = &repo->packs[i];
1562 		if (pack->path_packfile == NULL)
1563 			break;
1564 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1565 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1566 	}
1567 
1568 	if (i == repo->pack_cache_size) {
1569 		struct got_pack tmp;
1570 		do {
1571 			i--;
1572 		} while (i > 0 && repo->pinned_pack >= 0 &&
1573 		    i == repo->pinned_pack);
1574 		err = got_pack_close(&repo->packs[i]);
1575 		if (err)
1576 			return err;
1577 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1578 			return got_error_from_errno("ftruncate");
1579 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1580 			return got_error_from_errno("ftruncate");
1581 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1582 		memcpy(&repo->packs[i], &repo->packs[0],
1583 		    sizeof(repo->packs[i]));
1584 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1585 		if (repo->pinned_pack == 0)
1586 			repo->pinned_pack = i;
1587 		else if (repo->pinned_pack == i)
1588 			repo->pinned_pack = 0;
1589 		i = 0;
1590 	}
1591 
1592 	pack = &repo->packs[i];
1593 
1594 	pack->path_packfile = strdup(path_packfile);
1595 	if (pack->path_packfile == NULL) {
1596 		err = got_error_from_errno("strdup");
1597 		goto done;
1598 	}
1599 
1600 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1601 	if (err)
1602 		goto done;
1603 
1604 	if (fstat(pack->fd, &sb) != 0) {
1605 		err = got_error_from_errno("fstat");
1606 		goto done;
1607 	}
1608 	pack->filesize = sb.st_size;
1609 
1610 	pack->privsep_child = NULL;
1611 
1612 	err = got_delta_cache_alloc(&pack->delta_cache);
1613 	if (err)
1614 		goto done;
1615 
1616 #ifndef GOT_PACK_NO_MMAP
1617 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1618 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1619 		    pack->fd, 0);
1620 		if (pack->map == MAP_FAILED) {
1621 			if (errno != ENOMEM) {
1622 				err = got_error_from_errno("mmap");
1623 				goto done;
1624 			}
1625 			pack->map = NULL; /* fall back to read(2) */
1626 		}
1627 	}
1628 #endif
1629 done:
1630 	if (err) {
1631 		if (pack)
1632 			got_pack_close(pack);
1633 	} else if (packp)
1634 		*packp = pack;
1635 	return err;
1636 }
1637 
1638 struct got_pack *
1639 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1640 {
1641 	struct got_pack *pack = NULL;
1642 	size_t i;
1643 
1644 	for (i = 0; i < repo->pack_cache_size; i++) {
1645 		pack = &repo->packs[i];
1646 		if (pack->path_packfile == NULL)
1647 			break;
1648 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1649 			return pack;
1650 	}
1651 
1652 	return NULL;
1653 }
1654 
1655 const struct got_error *
1656 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1657     struct got_pack *pack)
1658 {
1659 	size_t i;
1660 	int pinned_pack = -1, pinned_packidx = -1;
1661 
1662 	for (i = 0; i < repo->pack_cache_size; i++) {
1663 		if (repo->packidx_cache[i] &&
1664 		    strcmp(repo->packidx_cache[i]->path_packidx,
1665 		    packidx->path_packidx) == 0)
1666 			pinned_packidx = i;
1667 		if (repo->packs[i].path_packfile &&
1668 		    strcmp(repo->packs[i].path_packfile,
1669 		    pack->path_packfile) == 0)
1670 			pinned_pack = i;
1671 	}
1672 
1673 	if (pinned_packidx == -1 || pinned_pack == -1)
1674 		return got_error(GOT_ERR_PIN_PACK);
1675 
1676 	repo->pinned_pack = pinned_pack;
1677 	repo->pinned_packidx = pinned_packidx;
1678 	if (repo->packs[pinned_pack].privsep_child)
1679 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1680 	return NULL;
1681 }
1682 
1683 struct got_pack *
1684 got_repo_get_pinned_pack(struct got_repository *repo)
1685 {
1686 	if (repo->pinned_pack >= 0 &&
1687 	    repo->pinned_pack < repo->pack_cache_size)
1688 		return &repo->packs[repo->pinned_pack];
1689 
1690 	return NULL;
1691 }
1692 
1693 void
1694 got_repo_unpin_pack(struct got_repository *repo)
1695 {
1696 	repo->pinned_packidx = -1;
1697 	repo->pinned_pack = -1;
1698 	repo->pinned_pid = 0;
1699 }
1700 
1701 const struct got_error *
1702 got_repo_init(const char *repo_path, const char *head_name)
1703 {
1704 	const struct got_error *err = NULL;
1705 	const char *dirnames[] = {
1706 		GOT_OBJECTS_DIR,
1707 		GOT_OBJECTS_PACK_DIR,
1708 		GOT_REFS_DIR,
1709 	};
1710 	const char *description_str = "Unnamed repository; "
1711 	    "edit this file 'description' to name the repository.";
1712 	const char *headref = "ref: refs/heads/";
1713 	const char *gitconfig_str = "[core]\n"
1714 	    "\trepositoryformatversion = 0\n"
1715 	    "\tfilemode = true\n"
1716 	    "\tbare = true\n";
1717 	char *headref_str, *path;
1718 	size_t i;
1719 
1720 	if (!got_path_dir_is_empty(repo_path))
1721 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1722 
1723 	for (i = 0; i < nitems(dirnames); i++) {
1724 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1725 			return got_error_from_errno("asprintf");
1726 		}
1727 		err = got_path_mkdir(path);
1728 		free(path);
1729 		if (err)
1730 			return err;
1731 	}
1732 
1733 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1734 		return got_error_from_errno("asprintf");
1735 	err = got_path_create_file(path, description_str);
1736 	free(path);
1737 	if (err)
1738 		return err;
1739 
1740 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1741 		return got_error_from_errno("asprintf");
1742 	if (asprintf(&headref_str, "%s%s", headref,
1743 	    head_name ? head_name : "main") == -1) {
1744 		free(path);
1745 		return got_error_from_errno("asprintf");
1746 	}
1747 	err = got_path_create_file(path, headref_str);
1748 	free(headref_str);
1749 	free(path);
1750 	if (err)
1751 		return err;
1752 
1753 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1754 		return got_error_from_errno("asprintf");
1755 	err = got_path_create_file(path, gitconfig_str);
1756 	free(path);
1757 	if (err)
1758 		return err;
1759 
1760 	return NULL;
1761 }
1762 
1763 static const struct got_error *
1764 match_packed_object(struct got_object_id **unique_id,
1765     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1766 {
1767 	const struct got_error *err = NULL;
1768 	struct got_object_id_queue matched_ids;
1769 	struct got_pathlist_entry *pe;
1770 
1771 	STAILQ_INIT(&matched_ids);
1772 
1773 	err = refresh_packidx_paths(repo);
1774 	if (err)
1775 		return err;
1776 
1777 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1778 		const char *path_packidx = pe->path;
1779 		struct got_packidx *packidx;
1780 		struct got_object_qid *qid;
1781 
1782 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1783 		    path_packidx, 0);
1784 		if (err)
1785 			break;
1786 
1787 		err = got_packidx_match_id_str_prefix(&matched_ids,
1788 		    packidx, id_str_prefix);
1789 		if (err) {
1790 			got_packidx_close(packidx);
1791 			break;
1792 		}
1793 		err = got_packidx_close(packidx);
1794 		if (err)
1795 			break;
1796 
1797 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1798 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1799 				int matched_type;
1800 				err = got_object_get_type(&matched_type, repo,
1801 				    &qid->id);
1802 				if (err)
1803 					goto done;
1804 				if (matched_type != obj_type)
1805 					continue;
1806 			}
1807 			if (*unique_id == NULL) {
1808 				*unique_id = got_object_id_dup(&qid->id);
1809 				if (*unique_id == NULL) {
1810 					err = got_error_from_errno("malloc");
1811 					goto done;
1812 				}
1813 			} else {
1814 				if (got_object_id_cmp(*unique_id,
1815 				    &qid->id) == 0)
1816 					continue; /* packed multiple times */
1817 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1818 				goto done;
1819 			}
1820 		}
1821 	}
1822 done:
1823 	got_object_id_queue_free(&matched_ids);
1824 	if (err) {
1825 		free(*unique_id);
1826 		*unique_id = NULL;
1827 	}
1828 	return err;
1829 }
1830 
1831 static const struct got_error *
1832 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1833     const char *object_dir, const char *id_str_prefix, int obj_type,
1834     struct got_repository *repo)
1835 {
1836 	const struct got_error *err = NULL;
1837 	char *path, *id_str = NULL;
1838 	DIR *dir = NULL;
1839 	struct dirent *dent;
1840 	struct got_object_id id;
1841 
1842 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1843 		err = got_error_from_errno("asprintf");
1844 		goto done;
1845 	}
1846 
1847 	dir = opendir(path);
1848 	if (dir == NULL) {
1849 		if (errno == ENOENT) {
1850 			err = NULL;
1851 			goto done;
1852 		}
1853 		err = got_error_from_errno2("opendir", path);
1854 		goto done;
1855 	}
1856 	while ((dent = readdir(dir)) != NULL) {
1857 		int cmp;
1858 
1859 		free(id_str);
1860 		id_str = NULL;
1861 
1862 		if (strcmp(dent->d_name, ".") == 0 ||
1863 		    strcmp(dent->d_name, "..") == 0)
1864 			continue;
1865 
1866 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1867 			err = got_error_from_errno("asprintf");
1868 			goto done;
1869 		}
1870 
1871 		if (!got_parse_object_id(&id, id_str, repo->algo))
1872 			continue;
1873 
1874 		/*
1875 		 * Directory entries do not necessarily appear in
1876 		 * sorted order, so we must iterate over all of them.
1877 		 */
1878 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1879 		if (cmp != 0)
1880 			continue;
1881 
1882 		if (*unique_id == NULL) {
1883 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1884 				int matched_type;
1885 				err = got_object_get_type(&matched_type, repo,
1886 				    &id);
1887 				if (err)
1888 					goto done;
1889 				if (matched_type != obj_type)
1890 					continue;
1891 			}
1892 			*unique_id = got_object_id_dup(&id);
1893 			if (*unique_id == NULL) {
1894 				err = got_error_from_errno("got_object_id_dup");
1895 				goto done;
1896 			}
1897 		} else {
1898 			if (got_object_id_cmp(*unique_id, &id) == 0)
1899 				continue; /* both packed and loose */
1900 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1901 			goto done;
1902 		}
1903 	}
1904 done:
1905 	if (dir && closedir(dir) != 0 && err == NULL)
1906 		err = got_error_from_errno("closedir");
1907 	if (err) {
1908 		free(*unique_id);
1909 		*unique_id = NULL;
1910 	}
1911 	free(id_str);
1912 	free(path);
1913 	return err;
1914 }
1915 
1916 const struct got_error *
1917 got_repo_match_object_id_prefix(struct got_object_id **id,
1918     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1919 {
1920 	const struct got_error *err = NULL;
1921 	char *path_objects = NULL, *object_dir = NULL;
1922 	size_t len;
1923 	int i;
1924 
1925 	*id = NULL;
1926 
1927 	path_objects = got_repo_get_path_objects(repo);
1928 
1929 	len = strlen(id_str_prefix);
1930 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1931 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1932 		goto done;
1933 	}
1934 
1935 	for (i = 0; i < len; i++) {
1936 		if (isxdigit((unsigned char)id_str_prefix[i]))
1937 			continue;
1938 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1939 		goto done;
1940 	}
1941 
1942 	if (len >= 2) {
1943 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1944 		if (err)
1945 			goto done;
1946 		object_dir = strndup(id_str_prefix, 2);
1947 		if (object_dir == NULL) {
1948 			err = got_error_from_errno("strdup");
1949 			goto done;
1950 		}
1951 		err = match_loose_object(id, path_objects, object_dir,
1952 		    id_str_prefix, obj_type, repo);
1953 	} else if (len == 1) {
1954 		int i;
1955 		for (i = 0; i < 0xf; i++) {
1956 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1957 			    == -1) {
1958 				err = got_error_from_errno("asprintf");
1959 				goto done;
1960 			}
1961 			err = match_packed_object(id, repo, object_dir,
1962 			    obj_type);
1963 			if (err)
1964 				goto done;
1965 			err = match_loose_object(id, path_objects, object_dir,
1966 			    id_str_prefix, obj_type, repo);
1967 			if (err)
1968 				goto done;
1969 		}
1970 	} else {
1971 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1972 		goto done;
1973 	}
1974 done:
1975 	free(path_objects);
1976 	free(object_dir);
1977 	if (err) {
1978 		free(*id);
1979 		*id = NULL;
1980 	} else if (*id == NULL) {
1981 		switch (obj_type) {
1982 		case GOT_OBJ_TYPE_BLOB:
1983 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1984 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1985 			break;
1986 		case GOT_OBJ_TYPE_TREE:
1987 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1988 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1989 			break;
1990 		case GOT_OBJ_TYPE_COMMIT:
1991 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1992 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1993 			break;
1994 		case GOT_OBJ_TYPE_TAG:
1995 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1996 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1997 			break;
1998 		default:
1999 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
2000 			break;
2001 		}
2002 	}
2003 
2004 	return err;
2005 }
2006 
2007 const struct got_error *
2008 got_repo_match_object_id(struct got_object_id **id, char **label,
2009     const char *id_str, int obj_type, struct got_reflist_head *refs,
2010     struct got_repository *repo)
2011 {
2012 	const struct got_error *err;
2013 	struct got_tag_object *tag;
2014 	struct got_reference *ref = NULL;
2015 
2016 	*id = NULL;
2017 	if (label)
2018 		*label = NULL;
2019 
2020 	if (refs) {
2021 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
2022 		    refs, repo);
2023 		if (err == NULL) {
2024 			*id = got_object_id_dup(
2025 			    got_object_tag_get_object_id(tag));
2026 			if (*id == NULL)
2027 				err = got_error_from_errno("got_object_id_dup");
2028 			else if (label && asprintf(label, "refs/tags/%s",
2029 			    got_object_tag_get_name(tag)) == -1) {
2030 				err = got_error_from_errno("asprintf");
2031 				free(*id);
2032 				*id = NULL;
2033 			}
2034 			got_object_tag_close(tag);
2035 			return err;
2036 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
2037 		    err->code != GOT_ERR_NO_OBJ)
2038 			return err;
2039 	}
2040 
2041 	err = got_ref_open(&ref, repo, id_str, 0);
2042 	if (err == NULL) {
2043 		err = got_ref_resolve(id, repo, ref);
2044 		if (err)
2045 			goto done;
2046 		if (label) {
2047 			*label = strdup(got_ref_get_name(ref));
2048 			if (*label == NULL) {
2049 				err = got_error_from_errno("strdup");
2050 				goto done;
2051 			}
2052 		}
2053 	} else {
2054 		if (err->code != GOT_ERR_NOT_REF &&
2055 		    err->code != GOT_ERR_BAD_REF_NAME)
2056 			goto done;
2057 		err = got_repo_match_object_id_prefix(id, id_str,
2058 		    obj_type, repo);
2059 		if (err) {
2060 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
2061 				err = got_error_not_ref(id_str);
2062 			goto done;
2063 		}
2064 		if (label) {
2065 			err = got_object_id_str(label, *id);
2066 			if (*label == NULL) {
2067 				err = got_error_from_errno("strdup");
2068 				goto done;
2069 			}
2070 		}
2071 	}
2072 done:
2073 	if (ref)
2074 		got_ref_close(ref);
2075 	return err;
2076 }
2077 
2078 const struct got_error *
2079 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
2080     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
2081 {
2082 	const struct got_error *err = NULL;
2083 	struct got_reflist_entry *re;
2084 	struct got_object_id *tag_id;
2085 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
2086 
2087 	*tag = NULL;
2088 
2089 	TAILQ_FOREACH(re, refs, entry) {
2090 		const char *refname;
2091 		refname = got_ref_get_name(re->ref);
2092 		if (got_ref_is_symbolic(re->ref))
2093 			continue;
2094 		if (strncmp(refname, "refs/tags/", 10) != 0)
2095 			continue;
2096 		if (!name_is_absolute)
2097 			refname += strlen("refs/tags/");
2098 		if (strcmp(refname, name) != 0)
2099 			continue;
2100 		err = got_ref_resolve(&tag_id, repo, re->ref);
2101 		if (err)
2102 			break;
2103 		err = got_object_open_as_tag(tag, repo, tag_id);
2104 		free(tag_id);
2105 		if (err)
2106 			break;
2107 		if (obj_type == GOT_OBJ_TYPE_ANY ||
2108 		    got_object_tag_get_object_type(*tag) == obj_type)
2109 			break;
2110 		got_object_tag_close(*tag);
2111 		*tag = NULL;
2112 	}
2113 
2114 	if (err == NULL && *tag == NULL)
2115 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2116 		    GOT_OBJ_LABEL_TAG, name);
2117 	return err;
2118 }
2119 
2120 static const struct got_error *
2121 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2122     const char *name, mode_t mode, struct got_object_id *blob_id)
2123 {
2124 	const struct got_error *err = NULL;
2125 
2126 	 *new_te = NULL;
2127 
2128 	*new_te = calloc(1, sizeof(**new_te));
2129 	if (*new_te == NULL)
2130 		return got_error_from_errno("calloc");
2131 
2132 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2133 	    sizeof((*new_te)->name)) {
2134 		err = got_error(GOT_ERR_NO_SPACE);
2135 		goto done;
2136 	}
2137 
2138 	if (S_ISLNK(mode)) {
2139 		(*new_te)->mode = S_IFLNK;
2140 	} else {
2141 		(*new_te)->mode = S_IFREG;
2142 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2143 	}
2144 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2145 done:
2146 	if (err && *new_te) {
2147 		free(*new_te);
2148 		*new_te = NULL;
2149 	}
2150 	return err;
2151 }
2152 
2153 static const struct got_error *
2154 import_file(struct got_tree_entry **new_te, struct dirent *de,
2155     const char *path, struct got_repository *repo)
2156 {
2157 	const struct got_error *err;
2158 	struct got_object_id *blob_id = NULL;
2159 	char *filepath;
2160 	struct stat sb;
2161 
2162 	if (asprintf(&filepath, "%s%s%s", path,
2163 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2164 		return got_error_from_errno("asprintf");
2165 
2166 	if (lstat(filepath, &sb) != 0) {
2167 		err = got_error_from_errno2("lstat", path);
2168 		goto done;
2169 	}
2170 
2171 	err = got_object_blob_create(&blob_id, filepath, repo);
2172 	if (err)
2173 		goto done;
2174 
2175 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2176 	    blob_id);
2177 done:
2178 	free(filepath);
2179 	if (err)
2180 		free(blob_id);
2181 	return err;
2182 }
2183 
2184 static const struct got_error *
2185 insert_tree_entry(struct got_tree_entry *new_te,
2186     struct got_pathlist_head *paths)
2187 {
2188 	const struct got_error *err = NULL;
2189 	struct got_pathlist_entry *new_pe;
2190 
2191 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2192 	if (err)
2193 		return err;
2194 	if (new_pe == NULL)
2195 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2196 	return NULL;
2197 }
2198 
2199 static const struct got_error *write_tree(struct got_object_id **,
2200     const char *, struct got_pathlist_head *, struct got_repository *,
2201     got_repo_import_cb progress_cb, void *progress_arg);
2202 
2203 static const struct got_error *
2204 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2205     const char *path, struct got_pathlist_head *ignores,
2206     struct got_repository *repo,
2207     got_repo_import_cb progress_cb, void *progress_arg)
2208 {
2209 	const struct got_error *err;
2210 	struct got_object_id *id = NULL;
2211 	char *subdirpath;
2212 
2213 	if (asprintf(&subdirpath, "%s%s%s", path,
2214 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2215 		return got_error_from_errno("asprintf");
2216 
2217 	(*new_te) = calloc(1, sizeof(**new_te));
2218 	if (*new_te == NULL)
2219 		return got_error_from_errno("calloc");
2220 	(*new_te)->mode = S_IFDIR;
2221 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2222 	    sizeof((*new_te)->name)) {
2223 		err = got_error(GOT_ERR_NO_SPACE);
2224 		goto done;
2225 	}
2226 	err = write_tree(&id, subdirpath, ignores,  repo,
2227 	    progress_cb, progress_arg);
2228 	if (err)
2229 		goto done;
2230 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2231 
2232 done:
2233 	free(id);
2234 	free(subdirpath);
2235 	if (err) {
2236 		free(*new_te);
2237 		*new_te = NULL;
2238 	}
2239 	return err;
2240 }
2241 
2242 static const struct got_error *
2243 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2244     struct got_pathlist_head *ignores, struct got_repository *repo,
2245     got_repo_import_cb progress_cb, void *progress_arg)
2246 {
2247 	const struct got_error *err = NULL;
2248 	DIR *dir;
2249 	struct dirent *de;
2250 	int nentries;
2251 	struct got_tree_entry *new_te = NULL;
2252 	struct got_pathlist_head paths;
2253 	struct got_pathlist_entry *pe;
2254 
2255 	*new_tree_id = NULL;
2256 
2257 	TAILQ_INIT(&paths);
2258 
2259 	dir = opendir(path_dir);
2260 	if (dir == NULL) {
2261 		err = got_error_from_errno2("opendir", path_dir);
2262 		goto done;
2263 	}
2264 
2265 	nentries = 0;
2266 	while ((de = readdir(dir)) != NULL) {
2267 		int ignore = 0;
2268 		int type;
2269 
2270 		if (strcmp(de->d_name, ".") == 0 ||
2271 		    strcmp(de->d_name, "..") == 0)
2272 			continue;
2273 
2274 		err = got_path_dirent_type(&type, path_dir, de);
2275 		if (err)
2276 			goto done;
2277 
2278 		TAILQ_FOREACH(pe, ignores, entry) {
2279 			if (type == DT_DIR && pe->path_len > 0 &&
2280 			    pe->path[pe->path_len - 1] == '/') {
2281 				char stripped[PATH_MAX];
2282 
2283 				if (strlcpy(stripped, pe->path,
2284 				    sizeof(stripped)) >= sizeof(stripped)) {
2285 					err = got_error(GOT_ERR_NO_SPACE);
2286 					goto done;
2287 				}
2288 				got_path_strip_trailing_slashes(stripped);
2289 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2290 					ignore = 1;
2291 					break;
2292 				}
2293 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2294 				ignore = 1;
2295 				break;
2296 			}
2297 		}
2298 		if (ignore)
2299 			continue;
2300 
2301 		if (type == DT_DIR) {
2302 			err = import_subdir(&new_te, de, path_dir,
2303 			    ignores, repo, progress_cb, progress_arg);
2304 			if (err) {
2305 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2306 					goto done;
2307 				err = NULL;
2308 				continue;
2309 			}
2310 		} else if (type == DT_REG || type == DT_LNK) {
2311 			err = import_file(&new_te, de, path_dir, repo);
2312 			if (err)
2313 				goto done;
2314 		} else
2315 			continue;
2316 
2317 		err = insert_tree_entry(new_te, &paths);
2318 		if (err)
2319 			goto done;
2320 		nentries++;
2321 	}
2322 
2323 	if (TAILQ_EMPTY(&paths)) {
2324 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2325 		    "cannot create tree without any entries");
2326 		goto done;
2327 	}
2328 
2329 	TAILQ_FOREACH(pe, &paths, entry) {
2330 		struct got_tree_entry *te = pe->data;
2331 		char *path;
2332 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2333 			continue;
2334 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2335 			err = got_error_from_errno("asprintf");
2336 			goto done;
2337 		}
2338 		err = (*progress_cb)(progress_arg, path);
2339 		free(path);
2340 		if (err)
2341 			goto done;
2342 	}
2343 
2344 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2345 done:
2346 	if (dir)
2347 		closedir(dir);
2348 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2349 	return err;
2350 }
2351 
2352 const struct got_error *
2353 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2354     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2355     struct got_repository *repo, got_repo_import_cb progress_cb,
2356     void *progress_arg)
2357 {
2358 	const struct got_error *err;
2359 	struct got_object_id *new_tree_id;
2360 
2361 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2362 	    progress_cb, progress_arg);
2363 	if (err)
2364 		return err;
2365 
2366 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2367 	    author, time(NULL), author, time(NULL), logmsg, repo);
2368 	free(new_tree_id);
2369 	return err;
2370 }
2371 
2372 const struct got_error *
2373 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2374     struct got_repository *repo)
2375 {
2376 	const struct got_error *err = NULL;
2377 	char *path_objects = NULL, *path = NULL;
2378 	DIR *dir = NULL;
2379 	struct got_object_id id;
2380 	int i;
2381 
2382 	*nobjects = 0;
2383 	*ondisk_size = 0;
2384 
2385 	path_objects = got_repo_get_path_objects(repo);
2386 	if (path_objects == NULL)
2387 		return got_error_from_errno("got_repo_get_path_objects");
2388 
2389 	for (i = 0; i <= 0xff; i++) {
2390 		struct dirent *dent;
2391 
2392 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2393 			err = got_error_from_errno("asprintf");
2394 			break;
2395 		}
2396 
2397 		dir = opendir(path);
2398 		if (dir == NULL) {
2399 			if (errno == ENOENT) {
2400 				err = NULL;
2401 				continue;
2402 			}
2403 			err = got_error_from_errno2("opendir", path);
2404 			break;
2405 		}
2406 
2407 		while ((dent = readdir(dir)) != NULL) {
2408 			char *id_str;
2409 			int fd;
2410 			struct stat sb;
2411 
2412 			if (strcmp(dent->d_name, ".") == 0 ||
2413 			    strcmp(dent->d_name, "..") == 0)
2414 				continue;
2415 
2416 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2417 				err = got_error_from_errno("asprintf");
2418 				goto done;
2419 			}
2420 
2421 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2422 				free(id_str);
2423 				continue;
2424 			}
2425 			free(id_str);
2426 
2427 			err = got_object_open_loose_fd(&fd, &id, repo);
2428 			if (err)
2429 				goto done;
2430 
2431 			if (fstat(fd, &sb) == -1) {
2432 				err = got_error_from_errno("fstat");
2433 				close(fd);
2434 				goto done;
2435 			}
2436 			(*nobjects)++;
2437 			(*ondisk_size) += sb.st_size;
2438 
2439 			if (close(fd) == -1) {
2440 				err = got_error_from_errno("close");
2441 				goto done;
2442 			}
2443 		}
2444 
2445 		if (closedir(dir) != 0) {
2446 			err = got_error_from_errno("closedir");
2447 			goto done;
2448 		}
2449 		dir = NULL;
2450 
2451 		free(path);
2452 		path = NULL;
2453 	}
2454 done:
2455 	if (dir && closedir(dir) != 0 && err == NULL)
2456 		err = got_error_from_errno("closedir");
2457 
2458 	if (err) {
2459 		*nobjects = 0;
2460 		*ondisk_size = 0;
2461 	}
2462 	free(path_objects);
2463 	free(path);
2464 	return err;
2465 }
2466 
2467 const struct got_error *
2468 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2469     off_t *total_packsize, struct got_repository *repo)
2470 {
2471 	const struct got_error *err = NULL;
2472 	DIR *packdir = NULL;
2473 	struct dirent *dent;
2474 	struct got_packidx *packidx = NULL;
2475 	char *path_packidx;
2476 	char *path_packfile;
2477 	int packdir_fd;
2478 	struct stat sb;
2479 
2480 	*npackfiles = 0;
2481 	*nobjects = 0;
2482 	*total_packsize = 0;
2483 
2484 	packdir_fd = openat(got_repo_get_fd(repo),
2485 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2486 	if (packdir_fd == -1) {
2487 		return got_error_from_errno_fmt("openat: %s/%s",
2488 		    got_repo_get_path_git_dir(repo),
2489 		    GOT_OBJECTS_PACK_DIR);
2490 	}
2491 
2492 	packdir = fdopendir(packdir_fd);
2493 	if (packdir == NULL) {
2494 		err = got_error_from_errno("fdopendir");
2495 		goto done;
2496 	}
2497 
2498 	while ((dent = readdir(packdir)) != NULL) {
2499 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2500 			continue;
2501 
2502 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2503 		    dent->d_name) == -1) {
2504 			err = got_error_from_errno("asprintf");
2505 			goto done;
2506 		}
2507 
2508 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2509 		    path_packidx, 0);
2510 		free(path_packidx);
2511 		if (err)
2512 			goto done;
2513 
2514 		if (fstat(packidx->fd, &sb) == -1)
2515 			goto done;
2516 		*total_packsize += sb.st_size;
2517 
2518 		err = got_packidx_get_packfile_path(&path_packfile,
2519 		    packidx->path_packidx);
2520 		if (err)
2521 			goto done;
2522 
2523 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2524 		    0) == -1) {
2525 			free(path_packfile);
2526 			goto done;
2527 		}
2528 		free(path_packfile);
2529 		*total_packsize += sb.st_size;
2530 
2531 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2532 
2533 		(*npackfiles)++;
2534 
2535 		got_packidx_close(packidx);
2536 		packidx = NULL;
2537 	}
2538 done:
2539 	if (packidx)
2540 		got_packidx_close(packidx);
2541 	if (packdir && closedir(packdir) != 0 && err == NULL)
2542 		err = got_error_from_errno("closedir");
2543 	if (err) {
2544 		*npackfiles = 0;
2545 		*nobjects = 0;
2546 		*total_packsize = 0;
2547 	}
2548 	return err;
2549 }
2550 
2551 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2552     got_packidx_bloom_filter_cmp);