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->fetch_all_branches = repo->fetch_all_branches;
949 
950 	new->nfetch_branches = repo->nfetch_branches;
951 	if (repo->fetch_branches) {
952 		new->fetch_branches = calloc(repo->nfetch_branches,
953 		    sizeof(char *));
954 		if (new->fetch_branches == NULL) {
955 			err = got_error_from_errno("calloc");
956 			goto done;
957 		}
958 		for (i = 0; i < repo->nfetch_branches; i++) {
959 			new->fetch_branches[i] = strdup(
960 			    repo->fetch_branches[i]);
961 			if (new->fetch_branches[i] == NULL) {
962 				err = got_error_from_errno("strdup");
963 				goto done;
964 			}
965 		}
966 	}
967 
968 	new->nsend_branches = repo->nsend_branches;
969 	if (repo->send_branches) {
970 		new->send_branches = calloc(repo->nsend_branches,
971 		    sizeof(char *));
972 		if (new->send_branches == NULL) {
973 			err = got_error_from_errno("calloc");
974 			goto done;
975 		}
976 		for (i = 0; i < repo->nsend_branches; i++) {
977 			new->send_branches[i] = strdup(
978 			    repo->send_branches[i]);
979 			if (new->send_branches[i] == NULL) {
980 				err = got_error_from_errno("strdup");
981 				goto done;
982 			}
983 		}
984 	}
985 
986 	new->nfetch_refs = repo->nfetch_refs;
987 	if (repo->fetch_refs) {
988 		new->fetch_refs = calloc(repo->nfetch_refs,
989 		    sizeof(char *));
990 		if (new->fetch_refs == NULL) {
991 			err = got_error_from_errno("calloc");
992 			goto done;
993 		}
994 		for (i = 0; i < repo->nfetch_refs; i++) {
995 			new->fetch_refs[i] = strdup(
996 			    repo->fetch_refs[i]);
997 			if (new->fetch_refs[i] == NULL) {
998 				err = got_error_from_errno("strdup");
999 				goto done;
1000 			}
1001 		}
1002 	}
1003 done:
1004 	if (err) {
1005 		got_repo_free_remote_repo_data(new);
1006 		free(new);
1007 	} else
1008 		*newp = new;
1009 
1010 	return err;
1011 }
1012 
1013 void
1014 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
1015 {
1016 	int i;
1017 
1018 	if (repo == NULL)
1019 		return;
1020 
1021 	free(repo->name);
1022 	repo->name = NULL;
1023 	free(repo->fetch_url);
1024 	repo->fetch_url = NULL;
1025 	free(repo->send_url);
1026 	repo->send_url = NULL;
1027 	for (i = 0; i < repo->nfetch_branches; i++)
1028 		free(repo->fetch_branches[i]);
1029 	free(repo->fetch_branches);
1030 	repo->fetch_branches = NULL;
1031 	repo->nfetch_branches = 0;
1032 	for (i = 0; i < repo->nsend_branches; i++)
1033 		free(repo->send_branches[i]);
1034 	free(repo->send_branches);
1035 	repo->send_branches = NULL;
1036 	repo->nsend_branches = 0;
1037 	for (i = 0; i < repo->nfetch_refs; i++)
1038 		free(repo->fetch_refs[i]);
1039 	free(repo->fetch_refs);
1040 	repo->fetch_refs = NULL;
1041 	repo->nfetch_refs = 0;
1042 }
1043 
1044 const struct got_error *
1045 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
1046     const char *input_path)
1047 {
1048 	const struct got_error *err = NULL;
1049 	const char *repo_abspath = NULL;
1050 	size_t repolen, len;
1051 	char *canonpath, *path = NULL;
1052 
1053 	*in_repo_path = NULL;
1054 
1055 	canonpath = strdup(input_path);
1056 	if (canonpath == NULL) {
1057 		err = got_error_from_errno("strdup");
1058 		goto done;
1059 	}
1060 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
1061 	if (err)
1062 		goto done;
1063 
1064 	repo_abspath = got_repo_get_path(repo);
1065 
1066 	if (canonpath[0] == '\0') {
1067 		path = strdup(canonpath);
1068 		if (path == NULL) {
1069 			err = got_error_from_errno("strdup");
1070 			goto done;
1071 		}
1072 	} else {
1073 		path = realpath(canonpath, NULL);
1074 		if (path == NULL) {
1075 			if (errno != ENOENT) {
1076 				err = got_error_from_errno2("realpath",
1077 				    canonpath);
1078 				goto done;
1079 			}
1080 			/*
1081 			 * Path is not on disk.
1082 			 * Assume it is already relative to repository root.
1083 			 */
1084 			path = strdup(canonpath);
1085 			if (path == NULL) {
1086 				err = got_error_from_errno("strdup");
1087 				goto done;
1088 			}
1089 		}
1090 
1091 		repolen = strlen(repo_abspath);
1092 		len = strlen(path);
1093 
1094 
1095 		if (strcmp(path, repo_abspath) == 0) {
1096 			free(path);
1097 			path = strdup("");
1098 			if (path == NULL) {
1099 				err = got_error_from_errno("strdup");
1100 				goto done;
1101 			}
1102 		} else if (len > repolen &&
1103 		    got_path_is_child(path, repo_abspath, repolen)) {
1104 			/* Matched an on-disk path inside repository. */
1105 			if (got_repo_is_bare(repo)) {
1106 				/*
1107 				 * Matched an on-disk path inside repository
1108 				 * database. Treat input as repository-relative.
1109 				 */
1110 				free(path);
1111 				path = canonpath;
1112 				canonpath = NULL;
1113 			} else {
1114 				char *child;
1115 				/* Strip common prefix with repository path. */
1116 				err = got_path_skip_common_ancestor(&child,
1117 				    repo_abspath, path);
1118 				if (err)
1119 					goto done;
1120 				free(path);
1121 				path = child;
1122 			}
1123 		} else {
1124 			/*
1125 			 * Matched unrelated on-disk path.
1126 			 * Treat input as repository-relative.
1127 			 */
1128 			free(path);
1129 			path = canonpath;
1130 			canonpath = NULL;
1131 		}
1132 	}
1133 
1134 	/* Make in-repository path absolute */
1135 	if (path[0] != '/') {
1136 		char *abspath;
1137 		if (asprintf(&abspath, "/%s", path) == -1) {
1138 			err = got_error_from_errno("asprintf");
1139 			goto done;
1140 		}
1141 		free(path);
1142 		path = abspath;
1143 	}
1144 
1145 done:
1146 	free(canonpath);
1147 	if (err)
1148 		free(path);
1149 	else
1150 		*in_repo_path = path;
1151 	return err;
1152 }
1153 
1154 static const struct got_error *
1155 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1156     const char *path_packidx)
1157 {
1158 	const struct got_error *err = NULL;
1159 	size_t i;
1160 
1161 	for (i = 0; i < repo->pack_cache_size; i++) {
1162 		if (repo->packidx_cache[i] == NULL)
1163 			break;
1164 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1165 		    path_packidx) == 0) {
1166 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1167 		}
1168 	}
1169 	if (i == repo->pack_cache_size) {
1170 		do {
1171 			i--;
1172 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1173 		    i == repo->pinned_packidx);
1174 		err = got_packidx_close(repo->packidx_cache[i]);
1175 		if (err)
1176 			return err;
1177 	}
1178 
1179 	repo->packidx_cache[i] = packidx;
1180 
1181 	return NULL;
1182 }
1183 
1184 int
1185 got_repo_is_packidx_filename(const char *name, size_t len)
1186 {
1187 	if (len != GOT_PACKIDX_NAMELEN)
1188 		return 0;
1189 
1190 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1191 		return 0;
1192 
1193 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1194 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1195 		return 0;
1196 
1197 	return 1;
1198 }
1199 
1200 static struct got_packidx_bloom_filter *
1201 get_packidx_bloom_filter(struct got_repository *repo,
1202     const char *path, size_t path_len)
1203 {
1204 	struct got_packidx_bloom_filter key;
1205 
1206 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1207 		return NULL; /* XXX */
1208 	key.path_len = path_len;
1209 
1210 	return RB_FIND(got_packidx_bloom_filter_tree,
1211 	    &repo->packidx_bloom_filters, &key);
1212 }
1213 
1214 int
1215 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1216     const char *path_packidx, struct got_object_id *id)
1217 {
1218 	struct got_packidx_bloom_filter *bf;
1219 
1220 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1221 	if (bf)
1222 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1223 
1224 	/* No bloom filter means this pack index must be searched. */
1225 	return 1;
1226 }
1227 
1228 static const struct got_error *
1229 add_packidx_bloom_filter(struct got_repository *repo,
1230     struct got_packidx *packidx, const char *path_packidx)
1231 {
1232 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1233 	struct got_packidx_bloom_filter *bf;
1234 	size_t len;
1235 
1236 	/*
1237 	 * Don't use bloom filters for very large pack index files.
1238 	 * Large pack files will contain a relatively large fraction
1239 	 * of our objects so we will likely need to visit them anyway.
1240 	 * The more objects a pack file contains the higher the probability
1241 	 * of a false-positive match from the bloom filter. And reading
1242 	 * all object IDs from a large pack index file can be expensive.
1243 	 */
1244 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1245 		return NULL;
1246 
1247 	/* Do we already have a filter for this pack index? */
1248 	if (get_packidx_bloom_filter(repo, path_packidx,
1249 	    strlen(path_packidx)) != NULL)
1250 		return NULL;
1251 
1252 	bf = calloc(1, sizeof(*bf));
1253 	if (bf == NULL)
1254 		return got_error_from_errno("calloc");
1255 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1256 	if (bf->bloom == NULL) {
1257 		free(bf);
1258 		return got_error_from_errno("calloc");
1259 	}
1260 
1261 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1262 	if (len >= sizeof(bf->path)) {
1263 		free(bf->bloom);
1264 		free(bf);
1265 		return got_error(GOT_ERR_NO_SPACE);
1266 	}
1267 	bf->path_len = len;
1268 
1269 	/* Minimum size supported by our bloom filter is 1000 entries. */
1270 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1271 	for (i = 0; i < nobjects; i++) {
1272 		struct got_packidx_object_id *id;
1273 		id = &packidx->hdr.sorted_ids[i];
1274 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1275 	}
1276 
1277 	RB_INSERT(got_packidx_bloom_filter_tree,
1278 	    &repo->packidx_bloom_filters, bf);
1279 	return NULL;
1280 }
1281 
1282 static void
1283 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1284 {
1285 	struct got_pathlist_entry *pe;
1286 
1287 	while (!TAILQ_EMPTY(packidx_paths)) {
1288 		pe = TAILQ_FIRST(packidx_paths);
1289 		TAILQ_REMOVE(packidx_paths, pe, entry);
1290 		free((char *)pe->path);
1291 		free(pe);
1292 	}
1293 }
1294 
1295 static const struct got_error *
1296 refresh_packidx_paths(struct got_repository *repo)
1297 {
1298 	const struct got_error *err = NULL;
1299 	char *objects_pack_dir = NULL;
1300 	struct stat sb;
1301 
1302 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1303 	if (objects_pack_dir == NULL)
1304 		return got_error_from_errno("got_repo_get_path_objects_pack");
1305 
1306 	if (stat(objects_pack_dir, &sb) == -1) {
1307 		if (errno != ENOENT) {
1308 			err = got_error_from_errno2("stat", objects_pack_dir);
1309 			goto done;
1310 		}
1311 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1312 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1313 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1314 		purge_packidx_paths(&repo->packidx_paths);
1315 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1316 		if (err)
1317 			goto done;
1318 	}
1319 done:
1320 	free(objects_pack_dir);
1321 	return err;
1322 }
1323 
1324 const struct got_error *
1325 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1326     struct got_repository *repo, struct got_object_id *id)
1327 {
1328 	const struct got_error *err;
1329 	struct got_pathlist_entry *pe;
1330 	size_t i;
1331 
1332 	/* Search pack index cache. */
1333 	for (i = 0; i < repo->pack_cache_size; i++) {
1334 		if (repo->packidx_cache[i] == NULL)
1335 			break;
1336 		if (!got_repo_check_packidx_bloom_filter(repo,
1337 		    repo->packidx_cache[i]->path_packidx, id))
1338 			continue; /* object will not be found in this index */
1339 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1340 		if (*idx != -1) {
1341 			*packidx = repo->packidx_cache[i];
1342 			/*
1343 			 * Move this cache entry to the front. Repeatedly
1344 			 * searching a wrong pack index can be expensive.
1345 			 */
1346 			if (i > 0) {
1347 				memmove(&repo->packidx_cache[1],
1348 				    &repo->packidx_cache[0],
1349 				    i * sizeof(repo->packidx_cache[0]));
1350 				repo->packidx_cache[0] = *packidx;
1351 				if (repo->pinned_packidx >= 0 &&
1352 				    repo->pinned_packidx < i)
1353 					repo->pinned_packidx++;
1354 				else if (repo->pinned_packidx == i)
1355 					repo->pinned_packidx = 0;
1356 			}
1357 			return NULL;
1358 		}
1359 	}
1360 	/* No luck. Search the filesystem. */
1361 
1362 	err = refresh_packidx_paths(repo);
1363 	if (err)
1364 		return err;
1365 
1366 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1367 		const char *path_packidx = pe->path;
1368 		int is_cached = 0;
1369 
1370 		if (!got_repo_check_packidx_bloom_filter(repo,
1371 		    pe->path, id))
1372 			continue; /* object will not be found in this index */
1373 
1374 		for (i = 0; i < repo->pack_cache_size; i++) {
1375 			if (repo->packidx_cache[i] == NULL)
1376 				break;
1377 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1378 			    path_packidx) == 0) {
1379 				is_cached = 1;
1380 				break;
1381 			}
1382 		}
1383 		if (is_cached)
1384 			continue; /* already searched */
1385 
1386 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1387 		    path_packidx, 0);
1388 		if (err)
1389 			goto done;
1390 
1391 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1392 		if (err)
1393 			goto done;
1394 
1395 		err = cache_packidx(repo, *packidx, path_packidx);
1396 		if (err)
1397 			goto done;
1398 
1399 		*idx = got_packidx_get_object_idx(*packidx, id);
1400 		if (*idx != -1) {
1401 			err = NULL; /* found the object */
1402 			goto done;
1403 		}
1404 	}
1405 
1406 	err = got_error_no_obj(id);
1407 done:
1408 	return err;
1409 }
1410 
1411 const struct got_error *
1412 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1413     struct got_repository *repo)
1414 {
1415 	const struct got_error *err = NULL;
1416 	DIR *packdir = NULL;
1417 	struct dirent *dent;
1418 	char *path_packidx = NULL;
1419 	int packdir_fd;
1420 	struct stat sb;
1421 
1422 	packdir_fd = openat(got_repo_get_fd(repo),
1423 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1424 	if (packdir_fd == -1) {
1425 		return got_error_from_errno_fmt("openat: %s/%s",
1426 		    got_repo_get_path_git_dir(repo),
1427 		    GOT_OBJECTS_PACK_DIR);
1428 	}
1429 
1430 	packdir = fdopendir(packdir_fd);
1431 	if (packdir == NULL) {
1432 		err = got_error_from_errno("fdopendir");
1433 		close(packdir_fd);
1434 		goto done;
1435 	}
1436 
1437 	if (fstat(packdir_fd, &sb) == -1) {
1438 		err = got_error_from_errno("fstat");
1439 		goto done;
1440 	}
1441 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1442 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1443 
1444 	while ((dent = readdir(packdir)) != NULL) {
1445 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1446 			continue;
1447 
1448 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1449 		    dent->d_name) == -1) {
1450 			err = got_error_from_errno("asprintf");
1451 			path_packidx = NULL;
1452 			break;
1453 		}
1454 
1455 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1456 		if (err)
1457 			break;
1458 	}
1459 done:
1460 	if (err)
1461 		free(path_packidx);
1462 	if (packdir && closedir(packdir) != 0 && err == NULL)
1463 		err = got_error_from_errno("closedir");
1464 	return err;
1465 }
1466 
1467 const struct got_error *
1468 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1469     struct got_repository *repo)
1470 {
1471 	const struct got_error *err;
1472 	size_t i;
1473 
1474 	*packidx = NULL;
1475 
1476 	/* Search pack index cache. */
1477 	for (i = 0; i < repo->pack_cache_size; i++) {
1478 		if (repo->packidx_cache[i] == NULL)
1479 			break;
1480 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1481 		    path_packidx) == 0) {
1482 			*packidx = repo->packidx_cache[i];
1483 			return NULL;
1484 		}
1485 	}
1486 	/* No luck. Search the filesystem. */
1487 
1488 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1489 	    path_packidx, 0);
1490 	if (err)
1491 		return err;
1492 
1493 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1494 	if (err)
1495 		goto done;
1496 
1497 	err = cache_packidx(repo, *packidx, path_packidx);
1498 done:
1499 	if (err) {
1500 		got_packidx_close(*packidx);
1501 		*packidx = NULL;
1502 	}
1503 	return err;
1504 }
1505 
1506 static const struct got_error *
1507 read_packfile_hdr(int fd, struct got_packidx *packidx)
1508 {
1509 	const struct got_error *err = NULL;
1510 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1511 	struct got_packfile_hdr hdr;
1512 	ssize_t n;
1513 
1514 	n = read(fd, &hdr, sizeof(hdr));
1515 	if (n < 0)
1516 		return got_error_from_errno("read");
1517 	if (n != sizeof(hdr))
1518 		return got_error(GOT_ERR_BAD_PACKFILE);
1519 
1520 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1521 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1522 	    be32toh(hdr.nobjects) != totobj)
1523 		err = got_error(GOT_ERR_BAD_PACKFILE);
1524 
1525 	return err;
1526 }
1527 
1528 static const struct got_error *
1529 open_packfile(int *fd, struct got_repository *repo,
1530     const char *relpath, struct got_packidx *packidx)
1531 {
1532 	const struct got_error *err = NULL;
1533 
1534 	*fd = openat(got_repo_get_fd(repo), relpath,
1535 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1536 	if (*fd == -1)
1537 		return got_error_from_errno_fmt("openat: %s/%s",
1538 		    got_repo_get_path_git_dir(repo), relpath);
1539 
1540 	if (packidx) {
1541 		err = read_packfile_hdr(*fd, packidx);
1542 		if (err) {
1543 			close(*fd);
1544 			*fd = -1;
1545 		}
1546 	}
1547 
1548 	return err;
1549 }
1550 
1551 const struct got_error *
1552 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1553     const char *path_packfile, struct got_packidx *packidx)
1554 {
1555 	const struct got_error *err = NULL;
1556 	struct got_pack *pack = NULL;
1557 	struct stat sb;
1558 	size_t i;
1559 
1560 	if (packp)
1561 		*packp = NULL;
1562 
1563 	for (i = 0; i < repo->pack_cache_size; i++) {
1564 		pack = &repo->packs[i];
1565 		if (pack->path_packfile == NULL)
1566 			break;
1567 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1568 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1569 	}
1570 
1571 	if (i == repo->pack_cache_size) {
1572 		struct got_pack tmp;
1573 		do {
1574 			i--;
1575 		} while (i > 0 && repo->pinned_pack >= 0 &&
1576 		    i == repo->pinned_pack);
1577 		err = got_pack_close(&repo->packs[i]);
1578 		if (err)
1579 			return err;
1580 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1581 			return got_error_from_errno("ftruncate");
1582 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1583 			return got_error_from_errno("ftruncate");
1584 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1585 		memcpy(&repo->packs[i], &repo->packs[0],
1586 		    sizeof(repo->packs[i]));
1587 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1588 		if (repo->pinned_pack == 0)
1589 			repo->pinned_pack = i;
1590 		else if (repo->pinned_pack == i)
1591 			repo->pinned_pack = 0;
1592 		i = 0;
1593 	}
1594 
1595 	pack = &repo->packs[i];
1596 
1597 	pack->path_packfile = strdup(path_packfile);
1598 	if (pack->path_packfile == NULL) {
1599 		err = got_error_from_errno("strdup");
1600 		goto done;
1601 	}
1602 
1603 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1604 	if (err)
1605 		goto done;
1606 
1607 	if (fstat(pack->fd, &sb) != 0) {
1608 		err = got_error_from_errno("fstat");
1609 		goto done;
1610 	}
1611 	pack->filesize = sb.st_size;
1612 
1613 	pack->privsep_child = NULL;
1614 
1615 	err = got_delta_cache_alloc(&pack->delta_cache);
1616 	if (err)
1617 		goto done;
1618 
1619 #ifndef GOT_PACK_NO_MMAP
1620 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1621 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1622 		    pack->fd, 0);
1623 		if (pack->map == MAP_FAILED) {
1624 			if (errno != ENOMEM) {
1625 				err = got_error_from_errno("mmap");
1626 				goto done;
1627 			}
1628 			pack->map = NULL; /* fall back to read(2) */
1629 		}
1630 	}
1631 #endif
1632 done:
1633 	if (err) {
1634 		if (pack)
1635 			got_pack_close(pack);
1636 	} else if (packp)
1637 		*packp = pack;
1638 	return err;
1639 }
1640 
1641 struct got_pack *
1642 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1643 {
1644 	struct got_pack *pack = NULL;
1645 	size_t i;
1646 
1647 	for (i = 0; i < repo->pack_cache_size; i++) {
1648 		pack = &repo->packs[i];
1649 		if (pack->path_packfile == NULL)
1650 			break;
1651 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1652 			return pack;
1653 	}
1654 
1655 	return NULL;
1656 }
1657 
1658 const struct got_error *
1659 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1660     struct got_pack *pack)
1661 {
1662 	size_t i;
1663 	int pinned_pack = -1, pinned_packidx = -1;
1664 
1665 	for (i = 0; i < repo->pack_cache_size; i++) {
1666 		if (repo->packidx_cache[i] &&
1667 		    strcmp(repo->packidx_cache[i]->path_packidx,
1668 		    packidx->path_packidx) == 0)
1669 			pinned_packidx = i;
1670 		if (repo->packs[i].path_packfile &&
1671 		    strcmp(repo->packs[i].path_packfile,
1672 		    pack->path_packfile) == 0)
1673 			pinned_pack = i;
1674 	}
1675 
1676 	if (pinned_packidx == -1 || pinned_pack == -1)
1677 		return got_error(GOT_ERR_PIN_PACK);
1678 
1679 	repo->pinned_pack = pinned_pack;
1680 	repo->pinned_packidx = pinned_packidx;
1681 	if (repo->packs[pinned_pack].privsep_child)
1682 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1683 	return NULL;
1684 }
1685 
1686 struct got_pack *
1687 got_repo_get_pinned_pack(struct got_repository *repo)
1688 {
1689 	if (repo->pinned_pack >= 0 &&
1690 	    repo->pinned_pack < repo->pack_cache_size)
1691 		return &repo->packs[repo->pinned_pack];
1692 
1693 	return NULL;
1694 }
1695 
1696 void
1697 got_repo_unpin_pack(struct got_repository *repo)
1698 {
1699 	repo->pinned_packidx = -1;
1700 	repo->pinned_pack = -1;
1701 	repo->pinned_pid = 0;
1702 }
1703 
1704 const struct got_error *
1705 got_repo_init(const char *repo_path, const char *head_name)
1706 {
1707 	const struct got_error *err = NULL;
1708 	const char *dirnames[] = {
1709 		GOT_OBJECTS_DIR,
1710 		GOT_OBJECTS_PACK_DIR,
1711 		GOT_REFS_DIR,
1712 	};
1713 	const char *description_str = "Unnamed repository; "
1714 	    "edit this file 'description' to name the repository.";
1715 	const char *headref = "ref: refs/heads/";
1716 	const char *gitconfig_str = "[core]\n"
1717 	    "\trepositoryformatversion = 0\n"
1718 	    "\tfilemode = true\n"
1719 	    "\tbare = true\n";
1720 	char *headref_str, *path;
1721 	size_t i;
1722 
1723 	if (!got_path_dir_is_empty(repo_path))
1724 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1725 
1726 	for (i = 0; i < nitems(dirnames); i++) {
1727 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1728 			return got_error_from_errno("asprintf");
1729 		}
1730 		err = got_path_mkdir(path);
1731 		free(path);
1732 		if (err)
1733 			return err;
1734 	}
1735 
1736 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1737 		return got_error_from_errno("asprintf");
1738 	err = got_path_create_file(path, description_str);
1739 	free(path);
1740 	if (err)
1741 		return err;
1742 
1743 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1744 		return got_error_from_errno("asprintf");
1745 	if (asprintf(&headref_str, "%s%s", headref,
1746 	    head_name ? head_name : "main") == -1) {
1747 		free(path);
1748 		return got_error_from_errno("asprintf");
1749 	}
1750 	err = got_path_create_file(path, headref_str);
1751 	free(headref_str);
1752 	free(path);
1753 	if (err)
1754 		return err;
1755 
1756 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1757 		return got_error_from_errno("asprintf");
1758 	err = got_path_create_file(path, gitconfig_str);
1759 	free(path);
1760 	if (err)
1761 		return err;
1762 
1763 	return NULL;
1764 }
1765 
1766 static const struct got_error *
1767 match_packed_object(struct got_object_id **unique_id,
1768     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1769 {
1770 	const struct got_error *err = NULL;
1771 	struct got_object_id_queue matched_ids;
1772 	struct got_pathlist_entry *pe;
1773 	struct timespec tv;
1774 	int retries = 0;
1775 	const int max_retries = 10;
1776 
1777 	STAILQ_INIT(&matched_ids);
1778 
1779 	err = refresh_packidx_paths(repo);
1780 	if (err)
1781 		return err;
1782 
1783 	/*
1784 	 * Opening objects while iterating over the pack-index path
1785 	 * list is racy. If the set of pack files in the repository
1786 	 * changes during loop iteration, refresh_packidx_paths() will
1787 	 * be called again, via got_object_get_type(), invalidating
1788 	 * the packidx_paths list we are iterating over.
1789 	 * To work around this we keep track of the current modification
1790 	 * time and retry the entire loop if it changes.
1791 	 */
1792 retry:
1793 	tv.tv_sec = repo->pack_path_mtime.tv_sec;
1794 	tv.tv_nsec = repo->pack_path_mtime.tv_nsec;
1795 
1796 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1797 		const char *path_packidx;
1798 		struct got_packidx *packidx;
1799 		struct got_object_qid *qid;
1800 
1801 		/*
1802 		 * If the modification time of the 'objects/pack' directory
1803 		 * has changed then 'pe' could now be an invalid pointer.
1804 		 */
1805 		if (tv.tv_sec != repo->pack_path_mtime.tv_sec ||
1806 		    tv.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1807 			if (++retries > max_retries) {
1808 				err = got_error_msg(GOT_ERR_TIMEOUT,
1809 				    "too many concurrent pack file "
1810 				    "modifications");
1811 				goto done;
1812 			}
1813 			goto retry;
1814 		}
1815 
1816 		path_packidx = pe->path;
1817 
1818 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1819 		    path_packidx, 0);
1820 		if (err)
1821 			break;
1822 
1823 		got_object_id_queue_free(&matched_ids);
1824 
1825 		err = got_packidx_match_id_str_prefix(&matched_ids,
1826 		    packidx, id_str_prefix);
1827 		if (err) {
1828 			got_packidx_close(packidx);
1829 			break;
1830 		}
1831 		err = got_packidx_close(packidx);
1832 		if (err)
1833 			break;
1834 
1835 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1836 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1837 				int matched_type;
1838 				err = got_object_get_type(&matched_type, repo,
1839 				    &qid->id);
1840 				if (err)
1841 					goto done;
1842 				if (matched_type != obj_type)
1843 					continue;
1844 			}
1845 			if (*unique_id == NULL) {
1846 				*unique_id = got_object_id_dup(&qid->id);
1847 				if (*unique_id == NULL) {
1848 					err = got_error_from_errno("malloc");
1849 					goto done;
1850 				}
1851 			} else {
1852 				if (got_object_id_cmp(*unique_id,
1853 				    &qid->id) == 0)
1854 					continue; /* packed multiple times */
1855 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1856 				goto done;
1857 			}
1858 		}
1859 	}
1860 done:
1861 	got_object_id_queue_free(&matched_ids);
1862 	if (err) {
1863 		free(*unique_id);
1864 		*unique_id = NULL;
1865 	}
1866 	return err;
1867 }
1868 
1869 static const struct got_error *
1870 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1871     const char *object_dir, const char *id_str_prefix, int obj_type,
1872     struct got_repository *repo)
1873 {
1874 	const struct got_error *err = NULL;
1875 	char *path, *id_str = NULL;
1876 	DIR *dir = NULL;
1877 	struct dirent *dent;
1878 	struct got_object_id id;
1879 
1880 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1881 		err = got_error_from_errno("asprintf");
1882 		goto done;
1883 	}
1884 
1885 	dir = opendir(path);
1886 	if (dir == NULL) {
1887 		if (errno == ENOENT) {
1888 			err = NULL;
1889 			goto done;
1890 		}
1891 		err = got_error_from_errno2("opendir", path);
1892 		goto done;
1893 	}
1894 	while ((dent = readdir(dir)) != NULL) {
1895 		int cmp;
1896 
1897 		free(id_str);
1898 		id_str = NULL;
1899 
1900 		if (strcmp(dent->d_name, ".") == 0 ||
1901 		    strcmp(dent->d_name, "..") == 0)
1902 			continue;
1903 
1904 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1905 			err = got_error_from_errno("asprintf");
1906 			goto done;
1907 		}
1908 
1909 		if (!got_parse_object_id(&id, id_str, repo->algo))
1910 			continue;
1911 
1912 		/*
1913 		 * Directory entries do not necessarily appear in
1914 		 * sorted order, so we must iterate over all of them.
1915 		 */
1916 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1917 		if (cmp != 0)
1918 			continue;
1919 
1920 		if (*unique_id == NULL) {
1921 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1922 				int matched_type;
1923 				err = got_object_get_type(&matched_type, repo,
1924 				    &id);
1925 				if (err)
1926 					goto done;
1927 				if (matched_type != obj_type)
1928 					continue;
1929 			}
1930 			*unique_id = got_object_id_dup(&id);
1931 			if (*unique_id == NULL) {
1932 				err = got_error_from_errno("got_object_id_dup");
1933 				goto done;
1934 			}
1935 		} else {
1936 			if (got_object_id_cmp(*unique_id, &id) == 0)
1937 				continue; /* both packed and loose */
1938 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1939 			goto done;
1940 		}
1941 	}
1942 done:
1943 	if (dir && closedir(dir) != 0 && err == NULL)
1944 		err = got_error_from_errno("closedir");
1945 	if (err) {
1946 		free(*unique_id);
1947 		*unique_id = NULL;
1948 	}
1949 	free(id_str);
1950 	free(path);
1951 	return err;
1952 }
1953 
1954 const struct got_error *
1955 got_repo_match_object_id_prefix(struct got_object_id **id,
1956     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1957 {
1958 	const struct got_error *err = NULL;
1959 	char *path_objects = NULL, *object_dir = NULL;
1960 	size_t len;
1961 	int i;
1962 
1963 	*id = NULL;
1964 
1965 	path_objects = got_repo_get_path_objects(repo);
1966 
1967 	len = strlen(id_str_prefix);
1968 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1969 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1970 		goto done;
1971 	}
1972 
1973 	for (i = 0; i < len; i++) {
1974 		if (isxdigit((unsigned char)id_str_prefix[i]))
1975 			continue;
1976 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1977 		goto done;
1978 	}
1979 
1980 	if (len >= 2) {
1981 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1982 		if (err)
1983 			goto done;
1984 		object_dir = strndup(id_str_prefix, 2);
1985 		if (object_dir == NULL) {
1986 			err = got_error_from_errno("strdup");
1987 			goto done;
1988 		}
1989 		err = match_loose_object(id, path_objects, object_dir,
1990 		    id_str_prefix, obj_type, repo);
1991 	} else if (len == 1) {
1992 		int i;
1993 		for (i = 0; i < 0xf; i++) {
1994 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1995 			    == -1) {
1996 				err = got_error_from_errno("asprintf");
1997 				goto done;
1998 			}
1999 			err = match_packed_object(id, repo, object_dir,
2000 			    obj_type);
2001 			if (err)
2002 				goto done;
2003 			err = match_loose_object(id, path_objects, object_dir,
2004 			    id_str_prefix, obj_type, repo);
2005 			if (err)
2006 				goto done;
2007 		}
2008 	} else {
2009 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
2010 		goto done;
2011 	}
2012 done:
2013 	free(path_objects);
2014 	free(object_dir);
2015 	if (err) {
2016 		free(*id);
2017 		*id = NULL;
2018 	} else if (*id == NULL) {
2019 		switch (obj_type) {
2020 		case GOT_OBJ_TYPE_BLOB:
2021 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2022 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
2023 			break;
2024 		case GOT_OBJ_TYPE_TREE:
2025 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2026 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
2027 			break;
2028 		case GOT_OBJ_TYPE_COMMIT:
2029 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2030 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
2031 			break;
2032 		case GOT_OBJ_TYPE_TAG:
2033 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2034 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
2035 			break;
2036 		default:
2037 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
2038 			break;
2039 		}
2040 	}
2041 
2042 	return err;
2043 }
2044 
2045 const struct got_error *
2046 got_repo_match_object_id(struct got_object_id **id, char **label,
2047     const char *id_str, int obj_type, struct got_reflist_head *refs,
2048     struct got_repository *repo)
2049 {
2050 	const struct got_error *err;
2051 	struct got_tag_object *tag;
2052 	struct got_reference *ref = NULL;
2053 
2054 	*id = NULL;
2055 	if (label)
2056 		*label = NULL;
2057 
2058 	if (refs) {
2059 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
2060 		    refs, repo);
2061 		if (err == NULL) {
2062 			*id = got_object_id_dup(
2063 			    got_object_tag_get_object_id(tag));
2064 			if (*id == NULL)
2065 				err = got_error_from_errno("got_object_id_dup");
2066 			else if (label && asprintf(label, "refs/tags/%s",
2067 			    got_object_tag_get_name(tag)) == -1) {
2068 				err = got_error_from_errno("asprintf");
2069 				free(*id);
2070 				*id = NULL;
2071 			}
2072 			got_object_tag_close(tag);
2073 			return err;
2074 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
2075 		    err->code != GOT_ERR_NO_OBJ)
2076 			return err;
2077 	}
2078 
2079 	err = got_ref_open(&ref, repo, id_str, 0);
2080 	if (err == NULL) {
2081 		err = got_ref_resolve(id, repo, ref);
2082 		if (err)
2083 			goto done;
2084 		if (label) {
2085 			*label = strdup(got_ref_get_name(ref));
2086 			if (*label == NULL) {
2087 				err = got_error_from_errno("strdup");
2088 				goto done;
2089 			}
2090 		}
2091 	} else {
2092 		if (err->code != GOT_ERR_NOT_REF &&
2093 		    err->code != GOT_ERR_BAD_REF_NAME)
2094 			goto done;
2095 		err = got_repo_match_object_id_prefix(id, id_str,
2096 		    obj_type, repo);
2097 		if (err) {
2098 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
2099 				err = got_error_not_ref(id_str);
2100 			goto done;
2101 		}
2102 		if (label) {
2103 			err = got_object_id_str(label, *id);
2104 			if (*label == NULL) {
2105 				err = got_error_from_errno("strdup");
2106 				goto done;
2107 			}
2108 		}
2109 	}
2110 done:
2111 	if (ref)
2112 		got_ref_close(ref);
2113 	return err;
2114 }
2115 
2116 const struct got_error *
2117 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
2118     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
2119 {
2120 	const struct got_error *err = NULL;
2121 	struct got_reflist_entry *re;
2122 	struct got_object_id *tag_id;
2123 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
2124 
2125 	*tag = NULL;
2126 
2127 	TAILQ_FOREACH(re, refs, entry) {
2128 		const char *refname;
2129 		refname = got_ref_get_name(re->ref);
2130 		if (got_ref_is_symbolic(re->ref))
2131 			continue;
2132 		if (strncmp(refname, "refs/tags/", 10) != 0)
2133 			continue;
2134 		if (!name_is_absolute)
2135 			refname += strlen("refs/tags/");
2136 		if (strcmp(refname, name) != 0)
2137 			continue;
2138 		err = got_ref_resolve(&tag_id, repo, re->ref);
2139 		if (err)
2140 			break;
2141 		err = got_object_open_as_tag(tag, repo, tag_id);
2142 		free(tag_id);
2143 		if (err)
2144 			break;
2145 		if (obj_type == GOT_OBJ_TYPE_ANY ||
2146 		    got_object_tag_get_object_type(*tag) == obj_type)
2147 			break;
2148 		got_object_tag_close(*tag);
2149 		*tag = NULL;
2150 	}
2151 
2152 	if (err == NULL && *tag == NULL)
2153 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2154 		    GOT_OBJ_LABEL_TAG, name);
2155 	return err;
2156 }
2157 
2158 const struct got_error *
2159 got_repo_find_object_id(struct got_object_id *id, struct got_repository *repo)
2160 {
2161 	const struct got_error *err;
2162 	struct got_object_id *matched_id = NULL;
2163 	char *id_str = NULL;
2164 
2165 	err = got_object_id_str(&id_str, id);
2166 	if (err)
2167 		return err;
2168 
2169 	err = got_repo_match_object_id_prefix(&matched_id, id_str,
2170 	    GOT_OBJ_TYPE_ANY, repo);
2171 	free(id_str);
2172 	return err;
2173 }
2174 
2175 static const struct got_error *
2176 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2177     const char *name, mode_t mode, struct got_object_id *blob_id)
2178 {
2179 	const struct got_error *err = NULL;
2180 
2181 	*new_te = NULL;
2182 
2183 	*new_te = calloc(1, sizeof(**new_te));
2184 	if (*new_te == NULL)
2185 		return got_error_from_errno("calloc");
2186 
2187 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2188 	    sizeof((*new_te)->name)) {
2189 		err = got_error(GOT_ERR_NO_SPACE);
2190 		goto done;
2191 	}
2192 
2193 	if (S_ISLNK(mode)) {
2194 		(*new_te)->mode = S_IFLNK;
2195 	} else {
2196 		(*new_te)->mode = S_IFREG;
2197 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2198 	}
2199 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2200 done:
2201 	if (err && *new_te) {
2202 		free(*new_te);
2203 		*new_te = NULL;
2204 	}
2205 	return err;
2206 }
2207 
2208 static const struct got_error *
2209 import_file(struct got_tree_entry **new_te, struct dirent *de,
2210     const char *path, struct got_repository *repo)
2211 {
2212 	const struct got_error *err;
2213 	struct got_object_id *blob_id = NULL;
2214 	char *filepath;
2215 	struct stat sb;
2216 
2217 	if (asprintf(&filepath, "%s%s%s", path,
2218 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2219 		return got_error_from_errno("asprintf");
2220 
2221 	if (lstat(filepath, &sb) != 0) {
2222 		err = got_error_from_errno2("lstat", path);
2223 		goto done;
2224 	}
2225 
2226 	err = got_object_blob_create(&blob_id, filepath, repo);
2227 	if (err)
2228 		goto done;
2229 
2230 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2231 	    blob_id);
2232 done:
2233 	free(filepath);
2234 	if (err)
2235 		free(blob_id);
2236 	return err;
2237 }
2238 
2239 static const struct got_error *
2240 insert_tree_entry(struct got_tree_entry *new_te,
2241     struct got_pathlist_head *paths)
2242 {
2243 	const struct got_error *err = NULL;
2244 	struct got_pathlist_entry *new_pe;
2245 
2246 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2247 	if (err)
2248 		return err;
2249 	if (new_pe == NULL)
2250 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2251 	return NULL;
2252 }
2253 
2254 static const struct got_error *write_tree(struct got_object_id **,
2255     const char *, struct got_pathlist_head *, struct got_repository *,
2256     got_repo_import_cb progress_cb, void *progress_arg);
2257 
2258 static const struct got_error *
2259 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2260     const char *path, struct got_pathlist_head *ignores,
2261     struct got_repository *repo,
2262     got_repo_import_cb progress_cb, void *progress_arg)
2263 {
2264 	const struct got_error *err;
2265 	struct got_object_id *id = NULL;
2266 	char *subdirpath;
2267 
2268 	if (asprintf(&subdirpath, "%s%s%s", path,
2269 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2270 		return got_error_from_errno("asprintf");
2271 
2272 	(*new_te) = calloc(1, sizeof(**new_te));
2273 	if (*new_te == NULL)
2274 		return got_error_from_errno("calloc");
2275 	(*new_te)->mode = S_IFDIR;
2276 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2277 	    sizeof((*new_te)->name)) {
2278 		err = got_error(GOT_ERR_NO_SPACE);
2279 		goto done;
2280 	}
2281 	err = write_tree(&id, subdirpath, ignores,  repo,
2282 	    progress_cb, progress_arg);
2283 	if (err)
2284 		goto done;
2285 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2286 
2287 done:
2288 	free(id);
2289 	free(subdirpath);
2290 	if (err) {
2291 		free(*new_te);
2292 		*new_te = NULL;
2293 	}
2294 	return err;
2295 }
2296 
2297 static const struct got_error *
2298 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2299     struct got_pathlist_head *ignores, struct got_repository *repo,
2300     got_repo_import_cb progress_cb, void *progress_arg)
2301 {
2302 	const struct got_error *err = NULL;
2303 	DIR *dir;
2304 	struct dirent *de;
2305 	int nentries;
2306 	struct got_tree_entry *new_te = NULL;
2307 	struct got_pathlist_head paths;
2308 	struct got_pathlist_entry *pe;
2309 
2310 	*new_tree_id = NULL;
2311 
2312 	TAILQ_INIT(&paths);
2313 
2314 	dir = opendir(path_dir);
2315 	if (dir == NULL) {
2316 		err = got_error_from_errno2("opendir", path_dir);
2317 		goto done;
2318 	}
2319 
2320 	nentries = 0;
2321 	while ((de = readdir(dir)) != NULL) {
2322 		int ignore = 0;
2323 		int type;
2324 
2325 		if (strcmp(de->d_name, ".") == 0 ||
2326 		    strcmp(de->d_name, "..") == 0)
2327 			continue;
2328 
2329 		err = got_path_dirent_type(&type, path_dir, de);
2330 		if (err)
2331 			goto done;
2332 
2333 		TAILQ_FOREACH(pe, ignores, entry) {
2334 			if (type == DT_DIR && pe->path_len > 0 &&
2335 			    pe->path[pe->path_len - 1] == '/') {
2336 				char stripped[PATH_MAX];
2337 
2338 				if (strlcpy(stripped, pe->path,
2339 				    sizeof(stripped)) >= sizeof(stripped)) {
2340 					err = got_error(GOT_ERR_NO_SPACE);
2341 					goto done;
2342 				}
2343 				got_path_strip_trailing_slashes(stripped);
2344 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2345 					ignore = 1;
2346 					break;
2347 				}
2348 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2349 				ignore = 1;
2350 				break;
2351 			}
2352 		}
2353 		if (ignore)
2354 			continue;
2355 
2356 		if (type == DT_DIR) {
2357 			err = import_subdir(&new_te, de, path_dir,
2358 			    ignores, repo, progress_cb, progress_arg);
2359 			if (err) {
2360 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2361 					goto done;
2362 				err = NULL;
2363 				continue;
2364 			}
2365 		} else if (type == DT_REG || type == DT_LNK) {
2366 			err = import_file(&new_te, de, path_dir, repo);
2367 			if (err)
2368 				goto done;
2369 		} else
2370 			continue;
2371 
2372 		err = insert_tree_entry(new_te, &paths);
2373 		if (err)
2374 			goto done;
2375 		nentries++;
2376 	}
2377 
2378 	if (TAILQ_EMPTY(&paths)) {
2379 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2380 		    "cannot create tree without any entries");
2381 		goto done;
2382 	}
2383 
2384 	TAILQ_FOREACH(pe, &paths, entry) {
2385 		struct got_tree_entry *te = pe->data;
2386 		char *path;
2387 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2388 			continue;
2389 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2390 			err = got_error_from_errno("asprintf");
2391 			goto done;
2392 		}
2393 		err = (*progress_cb)(progress_arg, path);
2394 		free(path);
2395 		if (err)
2396 			goto done;
2397 	}
2398 
2399 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2400 done:
2401 	if (dir)
2402 		closedir(dir);
2403 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2404 	return err;
2405 }
2406 
2407 const struct got_error *
2408 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2409     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2410     struct got_repository *repo, got_repo_import_cb progress_cb,
2411     void *progress_arg)
2412 {
2413 	const struct got_error *err;
2414 	struct got_object_id *new_tree_id;
2415 
2416 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2417 	    progress_cb, progress_arg);
2418 	if (err)
2419 		return err;
2420 
2421 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2422 	    author, time(NULL), author, time(NULL), logmsg, repo);
2423 	free(new_tree_id);
2424 	return err;
2425 }
2426 
2427 const struct got_error *
2428 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2429     struct got_repository *repo)
2430 {
2431 	const struct got_error *err = NULL;
2432 	char *path_objects = NULL, *path = NULL;
2433 	DIR *dir = NULL;
2434 	struct got_object_id id;
2435 	int i;
2436 
2437 	*nobjects = 0;
2438 	*ondisk_size = 0;
2439 
2440 	path_objects = got_repo_get_path_objects(repo);
2441 	if (path_objects == NULL)
2442 		return got_error_from_errno("got_repo_get_path_objects");
2443 
2444 	for (i = 0; i <= 0xff; i++) {
2445 		struct dirent *dent;
2446 
2447 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2448 			err = got_error_from_errno("asprintf");
2449 			break;
2450 		}
2451 
2452 		dir = opendir(path);
2453 		if (dir == NULL) {
2454 			if (errno == ENOENT) {
2455 				err = NULL;
2456 				continue;
2457 			}
2458 			err = got_error_from_errno2("opendir", path);
2459 			break;
2460 		}
2461 
2462 		while ((dent = readdir(dir)) != NULL) {
2463 			char *id_str;
2464 			int fd;
2465 			struct stat sb;
2466 
2467 			if (strcmp(dent->d_name, ".") == 0 ||
2468 			    strcmp(dent->d_name, "..") == 0)
2469 				continue;
2470 
2471 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2472 				err = got_error_from_errno("asprintf");
2473 				goto done;
2474 			}
2475 
2476 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2477 				free(id_str);
2478 				continue;
2479 			}
2480 			free(id_str);
2481 
2482 			err = got_object_open_loose_fd(&fd, &id, repo);
2483 			if (err)
2484 				goto done;
2485 
2486 			if (fstat(fd, &sb) == -1) {
2487 				err = got_error_from_errno("fstat");
2488 				close(fd);
2489 				goto done;
2490 			}
2491 			(*nobjects)++;
2492 			(*ondisk_size) += sb.st_size;
2493 
2494 			if (close(fd) == -1) {
2495 				err = got_error_from_errno("close");
2496 				goto done;
2497 			}
2498 		}
2499 
2500 		if (closedir(dir) != 0) {
2501 			err = got_error_from_errno("closedir");
2502 			goto done;
2503 		}
2504 		dir = NULL;
2505 
2506 		free(path);
2507 		path = NULL;
2508 	}
2509 done:
2510 	if (dir && closedir(dir) != 0 && err == NULL)
2511 		err = got_error_from_errno("closedir");
2512 
2513 	if (err) {
2514 		*nobjects = 0;
2515 		*ondisk_size = 0;
2516 	}
2517 	free(path_objects);
2518 	free(path);
2519 	return err;
2520 }
2521 
2522 const struct got_error *
2523 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2524     off_t *total_packsize, struct got_repository *repo)
2525 {
2526 	const struct got_error *err = NULL;
2527 	DIR *packdir = NULL;
2528 	struct dirent *dent;
2529 	struct got_packidx *packidx = NULL;
2530 	char *path_packidx;
2531 	char *path_packfile;
2532 	int packdir_fd;
2533 	struct stat sb;
2534 
2535 	*npackfiles = 0;
2536 	*nobjects = 0;
2537 	*total_packsize = 0;
2538 
2539 	packdir_fd = openat(got_repo_get_fd(repo),
2540 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2541 	if (packdir_fd == -1) {
2542 		return got_error_from_errno_fmt("openat: %s/%s",
2543 		    got_repo_get_path_git_dir(repo),
2544 		    GOT_OBJECTS_PACK_DIR);
2545 	}
2546 
2547 	packdir = fdopendir(packdir_fd);
2548 	if (packdir == NULL) {
2549 		err = got_error_from_errno("fdopendir");
2550 		close(packdir_fd);
2551 		goto done;
2552 	}
2553 
2554 	while ((dent = readdir(packdir)) != NULL) {
2555 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2556 			continue;
2557 
2558 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2559 		    dent->d_name) == -1) {
2560 			err = got_error_from_errno("asprintf");
2561 			goto done;
2562 		}
2563 
2564 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2565 		    path_packidx, 0);
2566 		free(path_packidx);
2567 		if (err)
2568 			goto done;
2569 
2570 		if (fstat(packidx->fd, &sb) == -1)
2571 			goto done;
2572 		*total_packsize += sb.st_size;
2573 
2574 		err = got_packidx_get_packfile_path(&path_packfile,
2575 		    packidx->path_packidx);
2576 		if (err)
2577 			goto done;
2578 
2579 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2580 		    0) == -1) {
2581 			free(path_packfile);
2582 			goto done;
2583 		}
2584 		free(path_packfile);
2585 		*total_packsize += sb.st_size;
2586 
2587 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2588 
2589 		(*npackfiles)++;
2590 
2591 		got_packidx_close(packidx);
2592 		packidx = NULL;
2593 	}
2594 done:
2595 	if (packidx)
2596 		got_packidx_close(packidx);
2597 	if (packdir && closedir(packdir) != 0 && err == NULL)
2598 		err = got_error_from_errno("closedir");
2599 	if (err) {
2600 		*npackfiles = 0;
2601 		*nobjects = 0;
2602 		*total_packsize = 0;
2603 	}
2604 	return err;
2605 }
2606 
2607 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2608     got_packidx_bloom_filter_cmp);