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 		goto done;
1434 	}
1435 
1436 	if (fstat(packdir_fd, &sb) == -1) {
1437 		err = got_error_from_errno("fstat");
1438 		goto done;
1439 	}
1440 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1441 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1442 
1443 	while ((dent = readdir(packdir)) != NULL) {
1444 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1445 			continue;
1446 
1447 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1448 		    dent->d_name) == -1) {
1449 			err = got_error_from_errno("asprintf");
1450 			path_packidx = NULL;
1451 			break;
1452 		}
1453 
1454 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1455 		if (err)
1456 			break;
1457 	}
1458 done:
1459 	if (err)
1460 		free(path_packidx);
1461 	if (packdir && closedir(packdir) != 0 && err == NULL)
1462 		err = got_error_from_errno("closedir");
1463 	return err;
1464 }
1465 
1466 const struct got_error *
1467 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1468     struct got_repository *repo)
1469 {
1470 	const struct got_error *err;
1471 	size_t i;
1472 
1473 	*packidx = NULL;
1474 
1475 	/* Search pack index cache. */
1476 	for (i = 0; i < repo->pack_cache_size; i++) {
1477 		if (repo->packidx_cache[i] == NULL)
1478 			break;
1479 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1480 		    path_packidx) == 0) {
1481 			*packidx = repo->packidx_cache[i];
1482 			return NULL;
1483 		}
1484 	}
1485 	/* No luck. Search the filesystem. */
1486 
1487 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1488 	    path_packidx, 0);
1489 	if (err)
1490 		return err;
1491 
1492 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1493 	if (err)
1494 		goto done;
1495 
1496 	err = cache_packidx(repo, *packidx, path_packidx);
1497 done:
1498 	if (err) {
1499 		got_packidx_close(*packidx);
1500 		*packidx = NULL;
1501 	}
1502 	return err;
1503 }
1504 
1505 static const struct got_error *
1506 read_packfile_hdr(int fd, struct got_packidx *packidx)
1507 {
1508 	const struct got_error *err = NULL;
1509 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1510 	struct got_packfile_hdr hdr;
1511 	ssize_t n;
1512 
1513 	n = read(fd, &hdr, sizeof(hdr));
1514 	if (n < 0)
1515 		return got_error_from_errno("read");
1516 	if (n != sizeof(hdr))
1517 		return got_error(GOT_ERR_BAD_PACKFILE);
1518 
1519 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1520 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1521 	    be32toh(hdr.nobjects) != totobj)
1522 		err = got_error(GOT_ERR_BAD_PACKFILE);
1523 
1524 	return err;
1525 }
1526 
1527 static const struct got_error *
1528 open_packfile(int *fd, struct got_repository *repo,
1529     const char *relpath, struct got_packidx *packidx)
1530 {
1531 	const struct got_error *err = NULL;
1532 
1533 	*fd = openat(got_repo_get_fd(repo), relpath,
1534 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1535 	if (*fd == -1)
1536 		return got_error_from_errno_fmt("openat: %s/%s",
1537 		    got_repo_get_path_git_dir(repo), relpath);
1538 
1539 	if (packidx) {
1540 		err = read_packfile_hdr(*fd, packidx);
1541 		if (err) {
1542 			close(*fd);
1543 			*fd = -1;
1544 		}
1545 	}
1546 
1547 	return err;
1548 }
1549 
1550 const struct got_error *
1551 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1552     const char *path_packfile, struct got_packidx *packidx)
1553 {
1554 	const struct got_error *err = NULL;
1555 	struct got_pack *pack = NULL;
1556 	struct stat sb;
1557 	size_t i;
1558 
1559 	if (packp)
1560 		*packp = NULL;
1561 
1562 	for (i = 0; i < repo->pack_cache_size; i++) {
1563 		pack = &repo->packs[i];
1564 		if (pack->path_packfile == NULL)
1565 			break;
1566 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1567 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1568 	}
1569 
1570 	if (i == repo->pack_cache_size) {
1571 		struct got_pack tmp;
1572 		do {
1573 			i--;
1574 		} while (i > 0 && repo->pinned_pack >= 0 &&
1575 		    i == repo->pinned_pack);
1576 		err = got_pack_close(&repo->packs[i]);
1577 		if (err)
1578 			return err;
1579 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1580 			return got_error_from_errno("ftruncate");
1581 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1582 			return got_error_from_errno("ftruncate");
1583 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1584 		memcpy(&repo->packs[i], &repo->packs[0],
1585 		    sizeof(repo->packs[i]));
1586 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1587 		if (repo->pinned_pack == 0)
1588 			repo->pinned_pack = i;
1589 		else if (repo->pinned_pack == i)
1590 			repo->pinned_pack = 0;
1591 		i = 0;
1592 	}
1593 
1594 	pack = &repo->packs[i];
1595 
1596 	pack->path_packfile = strdup(path_packfile);
1597 	if (pack->path_packfile == NULL) {
1598 		err = got_error_from_errno("strdup");
1599 		goto done;
1600 	}
1601 
1602 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1603 	if (err)
1604 		goto done;
1605 
1606 	if (fstat(pack->fd, &sb) != 0) {
1607 		err = got_error_from_errno("fstat");
1608 		goto done;
1609 	}
1610 	pack->filesize = sb.st_size;
1611 
1612 	pack->privsep_child = NULL;
1613 
1614 	err = got_delta_cache_alloc(&pack->delta_cache);
1615 	if (err)
1616 		goto done;
1617 
1618 #ifndef GOT_PACK_NO_MMAP
1619 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1620 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1621 		    pack->fd, 0);
1622 		if (pack->map == MAP_FAILED) {
1623 			if (errno != ENOMEM) {
1624 				err = got_error_from_errno("mmap");
1625 				goto done;
1626 			}
1627 			pack->map = NULL; /* fall back to read(2) */
1628 		}
1629 	}
1630 #endif
1631 done:
1632 	if (err) {
1633 		if (pack)
1634 			got_pack_close(pack);
1635 	} else if (packp)
1636 		*packp = pack;
1637 	return err;
1638 }
1639 
1640 struct got_pack *
1641 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1642 {
1643 	struct got_pack *pack = NULL;
1644 	size_t i;
1645 
1646 	for (i = 0; i < repo->pack_cache_size; i++) {
1647 		pack = &repo->packs[i];
1648 		if (pack->path_packfile == NULL)
1649 			break;
1650 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1651 			return pack;
1652 	}
1653 
1654 	return NULL;
1655 }
1656 
1657 const struct got_error *
1658 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1659     struct got_pack *pack)
1660 {
1661 	size_t i;
1662 	int pinned_pack = -1, pinned_packidx = -1;
1663 
1664 	for (i = 0; i < repo->pack_cache_size; i++) {
1665 		if (repo->packidx_cache[i] &&
1666 		    strcmp(repo->packidx_cache[i]->path_packidx,
1667 		    packidx->path_packidx) == 0)
1668 			pinned_packidx = i;
1669 		if (repo->packs[i].path_packfile &&
1670 		    strcmp(repo->packs[i].path_packfile,
1671 		    pack->path_packfile) == 0)
1672 			pinned_pack = i;
1673 	}
1674 
1675 	if (pinned_packidx == -1 || pinned_pack == -1)
1676 		return got_error(GOT_ERR_PIN_PACK);
1677 
1678 	repo->pinned_pack = pinned_pack;
1679 	repo->pinned_packidx = pinned_packidx;
1680 	if (repo->packs[pinned_pack].privsep_child)
1681 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1682 	return NULL;
1683 }
1684 
1685 struct got_pack *
1686 got_repo_get_pinned_pack(struct got_repository *repo)
1687 {
1688 	if (repo->pinned_pack >= 0 &&
1689 	    repo->pinned_pack < repo->pack_cache_size)
1690 		return &repo->packs[repo->pinned_pack];
1691 
1692 	return NULL;
1693 }
1694 
1695 void
1696 got_repo_unpin_pack(struct got_repository *repo)
1697 {
1698 	repo->pinned_packidx = -1;
1699 	repo->pinned_pack = -1;
1700 	repo->pinned_pid = 0;
1701 }
1702 
1703 const struct got_error *
1704 got_repo_init(const char *repo_path, const char *head_name)
1705 {
1706 	const struct got_error *err = NULL;
1707 	const char *dirnames[] = {
1708 		GOT_OBJECTS_DIR,
1709 		GOT_OBJECTS_PACK_DIR,
1710 		GOT_REFS_DIR,
1711 	};
1712 	const char *description_str = "Unnamed repository; "
1713 	    "edit this file 'description' to name the repository.";
1714 	const char *headref = "ref: refs/heads/";
1715 	const char *gitconfig_str = "[core]\n"
1716 	    "\trepositoryformatversion = 0\n"
1717 	    "\tfilemode = true\n"
1718 	    "\tbare = true\n";
1719 	char *headref_str, *path;
1720 	size_t i;
1721 
1722 	if (!got_path_dir_is_empty(repo_path))
1723 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1724 
1725 	for (i = 0; i < nitems(dirnames); i++) {
1726 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1727 			return got_error_from_errno("asprintf");
1728 		}
1729 		err = got_path_mkdir(path);
1730 		free(path);
1731 		if (err)
1732 			return err;
1733 	}
1734 
1735 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1736 		return got_error_from_errno("asprintf");
1737 	err = got_path_create_file(path, description_str);
1738 	free(path);
1739 	if (err)
1740 		return err;
1741 
1742 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1743 		return got_error_from_errno("asprintf");
1744 	if (asprintf(&headref_str, "%s%s", headref,
1745 	    head_name ? head_name : "main") == -1) {
1746 		free(path);
1747 		return got_error_from_errno("asprintf");
1748 	}
1749 	err = got_path_create_file(path, headref_str);
1750 	free(headref_str);
1751 	free(path);
1752 	if (err)
1753 		return err;
1754 
1755 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1756 		return got_error_from_errno("asprintf");
1757 	err = got_path_create_file(path, gitconfig_str);
1758 	free(path);
1759 	if (err)
1760 		return err;
1761 
1762 	return NULL;
1763 }
1764 
1765 static const struct got_error *
1766 match_packed_object(struct got_object_id **unique_id,
1767     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1768 {
1769 	const struct got_error *err = NULL;
1770 	struct got_object_id_queue matched_ids;
1771 	struct got_pathlist_entry *pe;
1772 	struct timespec tv;
1773 	int retries = 0;
1774 	const int max_retries = 10;
1775 
1776 	STAILQ_INIT(&matched_ids);
1777 
1778 	err = refresh_packidx_paths(repo);
1779 	if (err)
1780 		return err;
1781 
1782 	/*
1783 	 * Opening objects while iterating over the pack-index path
1784 	 * list is racy. If the set of pack files in the repository
1785 	 * changes during loop iteration, refresh_packidx_paths() will
1786 	 * be called again, via got_object_get_type(), invalidating
1787 	 * the packidx_paths list we are iterating over.
1788 	 * To work around this we keep track of the current modification
1789 	 * time and retry the entire loop if it changes.
1790 	 */
1791 retry:
1792 	tv.tv_sec = repo->pack_path_mtime.tv_sec;
1793 	tv.tv_nsec = repo->pack_path_mtime.tv_nsec;
1794 
1795 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1796 		const char *path_packidx;
1797 		struct got_packidx *packidx;
1798 		struct got_object_qid *qid;
1799 
1800 		/*
1801 		 * If the modification time of the 'objects/pack' directory
1802 		 * has changed then 'pe' could now be an invalid pointer.
1803 		 */
1804 		if (tv.tv_sec != repo->pack_path_mtime.tv_sec ||
1805 		    tv.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1806 			if (++retries > max_retries) {
1807 				err = got_error_msg(GOT_ERR_TIMEOUT,
1808 				    "too many concurrent pack file "
1809 				    "modifications");
1810 				goto done;
1811 			}
1812 			goto retry;
1813 		}
1814 
1815 		path_packidx = pe->path;
1816 
1817 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1818 		    path_packidx, 0);
1819 		if (err)
1820 			break;
1821 
1822 		got_object_id_queue_free(&matched_ids);
1823 
1824 		err = got_packidx_match_id_str_prefix(&matched_ids,
1825 		    packidx, id_str_prefix);
1826 		if (err) {
1827 			got_packidx_close(packidx);
1828 			break;
1829 		}
1830 		err = got_packidx_close(packidx);
1831 		if (err)
1832 			break;
1833 
1834 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1835 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1836 				int matched_type;
1837 				err = got_object_get_type(&matched_type, repo,
1838 				    &qid->id);
1839 				if (err)
1840 					goto done;
1841 				if (matched_type != obj_type)
1842 					continue;
1843 			}
1844 			if (*unique_id == NULL) {
1845 				*unique_id = got_object_id_dup(&qid->id);
1846 				if (*unique_id == NULL) {
1847 					err = got_error_from_errno("malloc");
1848 					goto done;
1849 				}
1850 			} else {
1851 				if (got_object_id_cmp(*unique_id,
1852 				    &qid->id) == 0)
1853 					continue; /* packed multiple times */
1854 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1855 				goto done;
1856 			}
1857 		}
1858 	}
1859 done:
1860 	got_object_id_queue_free(&matched_ids);
1861 	if (err) {
1862 		free(*unique_id);
1863 		*unique_id = NULL;
1864 	}
1865 	return err;
1866 }
1867 
1868 static const struct got_error *
1869 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1870     const char *object_dir, const char *id_str_prefix, int obj_type,
1871     struct got_repository *repo)
1872 {
1873 	const struct got_error *err = NULL;
1874 	char *path, *id_str = NULL;
1875 	DIR *dir = NULL;
1876 	struct dirent *dent;
1877 	struct got_object_id id;
1878 
1879 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1880 		err = got_error_from_errno("asprintf");
1881 		goto done;
1882 	}
1883 
1884 	dir = opendir(path);
1885 	if (dir == NULL) {
1886 		if (errno == ENOENT) {
1887 			err = NULL;
1888 			goto done;
1889 		}
1890 		err = got_error_from_errno2("opendir", path);
1891 		goto done;
1892 	}
1893 	while ((dent = readdir(dir)) != NULL) {
1894 		int cmp;
1895 
1896 		free(id_str);
1897 		id_str = NULL;
1898 
1899 		if (strcmp(dent->d_name, ".") == 0 ||
1900 		    strcmp(dent->d_name, "..") == 0)
1901 			continue;
1902 
1903 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1904 			err = got_error_from_errno("asprintf");
1905 			goto done;
1906 		}
1907 
1908 		if (!got_parse_object_id(&id, id_str, repo->algo))
1909 			continue;
1910 
1911 		/*
1912 		 * Directory entries do not necessarily appear in
1913 		 * sorted order, so we must iterate over all of them.
1914 		 */
1915 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1916 		if (cmp != 0)
1917 			continue;
1918 
1919 		if (*unique_id == NULL) {
1920 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1921 				int matched_type;
1922 				err = got_object_get_type(&matched_type, repo,
1923 				    &id);
1924 				if (err)
1925 					goto done;
1926 				if (matched_type != obj_type)
1927 					continue;
1928 			}
1929 			*unique_id = got_object_id_dup(&id);
1930 			if (*unique_id == NULL) {
1931 				err = got_error_from_errno("got_object_id_dup");
1932 				goto done;
1933 			}
1934 		} else {
1935 			if (got_object_id_cmp(*unique_id, &id) == 0)
1936 				continue; /* both packed and loose */
1937 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1938 			goto done;
1939 		}
1940 	}
1941 done:
1942 	if (dir && closedir(dir) != 0 && err == NULL)
1943 		err = got_error_from_errno("closedir");
1944 	if (err) {
1945 		free(*unique_id);
1946 		*unique_id = NULL;
1947 	}
1948 	free(id_str);
1949 	free(path);
1950 	return err;
1951 }
1952 
1953 const struct got_error *
1954 got_repo_match_object_id_prefix(struct got_object_id **id,
1955     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1956 {
1957 	const struct got_error *err = NULL;
1958 	char *path_objects = NULL, *object_dir = NULL;
1959 	size_t len;
1960 	int i;
1961 
1962 	*id = NULL;
1963 
1964 	path_objects = got_repo_get_path_objects(repo);
1965 
1966 	len = strlen(id_str_prefix);
1967 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1968 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1969 		goto done;
1970 	}
1971 
1972 	for (i = 0; i < len; i++) {
1973 		if (isxdigit((unsigned char)id_str_prefix[i]))
1974 			continue;
1975 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1976 		goto done;
1977 	}
1978 
1979 	if (len >= 2) {
1980 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1981 		if (err)
1982 			goto done;
1983 		object_dir = strndup(id_str_prefix, 2);
1984 		if (object_dir == NULL) {
1985 			err = got_error_from_errno("strdup");
1986 			goto done;
1987 		}
1988 		err = match_loose_object(id, path_objects, object_dir,
1989 		    id_str_prefix, obj_type, repo);
1990 	} else if (len == 1) {
1991 		int i;
1992 		for (i = 0; i < 0xf; i++) {
1993 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1994 			    == -1) {
1995 				err = got_error_from_errno("asprintf");
1996 				goto done;
1997 			}
1998 			err = match_packed_object(id, repo, object_dir,
1999 			    obj_type);
2000 			if (err)
2001 				goto done;
2002 			err = match_loose_object(id, path_objects, object_dir,
2003 			    id_str_prefix, obj_type, repo);
2004 			if (err)
2005 				goto done;
2006 		}
2007 	} else {
2008 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
2009 		goto done;
2010 	}
2011 done:
2012 	free(path_objects);
2013 	free(object_dir);
2014 	if (err) {
2015 		free(*id);
2016 		*id = NULL;
2017 	} else if (*id == NULL) {
2018 		switch (obj_type) {
2019 		case GOT_OBJ_TYPE_BLOB:
2020 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2021 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
2022 			break;
2023 		case GOT_OBJ_TYPE_TREE:
2024 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2025 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
2026 			break;
2027 		case GOT_OBJ_TYPE_COMMIT:
2028 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2029 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
2030 			break;
2031 		case GOT_OBJ_TYPE_TAG:
2032 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2033 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
2034 			break;
2035 		default:
2036 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
2037 			break;
2038 		}
2039 	}
2040 
2041 	return err;
2042 }
2043 
2044 const struct got_error *
2045 got_repo_match_object_id(struct got_object_id **id, char **label,
2046     const char *id_str, int obj_type, struct got_reflist_head *refs,
2047     struct got_repository *repo)
2048 {
2049 	const struct got_error *err;
2050 	struct got_tag_object *tag;
2051 	struct got_reference *ref = NULL;
2052 
2053 	*id = NULL;
2054 	if (label)
2055 		*label = NULL;
2056 
2057 	if (refs) {
2058 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
2059 		    refs, repo);
2060 		if (err == NULL) {
2061 			*id = got_object_id_dup(
2062 			    got_object_tag_get_object_id(tag));
2063 			if (*id == NULL)
2064 				err = got_error_from_errno("got_object_id_dup");
2065 			else if (label && asprintf(label, "refs/tags/%s",
2066 			    got_object_tag_get_name(tag)) == -1) {
2067 				err = got_error_from_errno("asprintf");
2068 				free(*id);
2069 				*id = NULL;
2070 			}
2071 			got_object_tag_close(tag);
2072 			return err;
2073 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
2074 		    err->code != GOT_ERR_NO_OBJ)
2075 			return err;
2076 	}
2077 
2078 	err = got_ref_open(&ref, repo, id_str, 0);
2079 	if (err == NULL) {
2080 		err = got_ref_resolve(id, repo, ref);
2081 		if (err)
2082 			goto done;
2083 		if (label) {
2084 			*label = strdup(got_ref_get_name(ref));
2085 			if (*label == NULL) {
2086 				err = got_error_from_errno("strdup");
2087 				goto done;
2088 			}
2089 		}
2090 	} else {
2091 		if (err->code != GOT_ERR_NOT_REF &&
2092 		    err->code != GOT_ERR_BAD_REF_NAME)
2093 			goto done;
2094 		err = got_repo_match_object_id_prefix(id, id_str,
2095 		    obj_type, repo);
2096 		if (err) {
2097 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
2098 				err = got_error_not_ref(id_str);
2099 			goto done;
2100 		}
2101 		if (label) {
2102 			err = got_object_id_str(label, *id);
2103 			if (*label == NULL) {
2104 				err = got_error_from_errno("strdup");
2105 				goto done;
2106 			}
2107 		}
2108 	}
2109 done:
2110 	if (ref)
2111 		got_ref_close(ref);
2112 	return err;
2113 }
2114 
2115 const struct got_error *
2116 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
2117     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
2118 {
2119 	const struct got_error *err = NULL;
2120 	struct got_reflist_entry *re;
2121 	struct got_object_id *tag_id;
2122 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
2123 
2124 	*tag = NULL;
2125 
2126 	TAILQ_FOREACH(re, refs, entry) {
2127 		const char *refname;
2128 		refname = got_ref_get_name(re->ref);
2129 		if (got_ref_is_symbolic(re->ref))
2130 			continue;
2131 		if (strncmp(refname, "refs/tags/", 10) != 0)
2132 			continue;
2133 		if (!name_is_absolute)
2134 			refname += strlen("refs/tags/");
2135 		if (strcmp(refname, name) != 0)
2136 			continue;
2137 		err = got_ref_resolve(&tag_id, repo, re->ref);
2138 		if (err)
2139 			break;
2140 		err = got_object_open_as_tag(tag, repo, tag_id);
2141 		free(tag_id);
2142 		if (err)
2143 			break;
2144 		if (obj_type == GOT_OBJ_TYPE_ANY ||
2145 		    got_object_tag_get_object_type(*tag) == obj_type)
2146 			break;
2147 		got_object_tag_close(*tag);
2148 		*tag = NULL;
2149 	}
2150 
2151 	if (err == NULL && *tag == NULL)
2152 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2153 		    GOT_OBJ_LABEL_TAG, name);
2154 	return err;
2155 }
2156 
2157 const struct got_error *
2158 got_repo_find_object_id(struct got_object_id *id, struct got_repository *repo)
2159 {
2160 	const struct got_error *err;
2161 	struct got_object_id *matched_id = NULL;
2162 	char *id_str = NULL;
2163 
2164 	err = got_object_id_str(&id_str, id);
2165 	if (err)
2166 		return err;
2167 
2168 	err = got_repo_match_object_id_prefix(&matched_id, id_str,
2169 	    GOT_OBJ_TYPE_ANY, repo);
2170 	free(id_str);
2171 	return err;
2172 }
2173 
2174 static const struct got_error *
2175 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2176     const char *name, mode_t mode, struct got_object_id *blob_id)
2177 {
2178 	const struct got_error *err = NULL;
2179 
2180 	 *new_te = NULL;
2181 
2182 	*new_te = calloc(1, sizeof(**new_te));
2183 	if (*new_te == NULL)
2184 		return got_error_from_errno("calloc");
2185 
2186 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2187 	    sizeof((*new_te)->name)) {
2188 		err = got_error(GOT_ERR_NO_SPACE);
2189 		goto done;
2190 	}
2191 
2192 	if (S_ISLNK(mode)) {
2193 		(*new_te)->mode = S_IFLNK;
2194 	} else {
2195 		(*new_te)->mode = S_IFREG;
2196 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2197 	}
2198 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2199 done:
2200 	if (err && *new_te) {
2201 		free(*new_te);
2202 		*new_te = NULL;
2203 	}
2204 	return err;
2205 }
2206 
2207 static const struct got_error *
2208 import_file(struct got_tree_entry **new_te, struct dirent *de,
2209     const char *path, struct got_repository *repo)
2210 {
2211 	const struct got_error *err;
2212 	struct got_object_id *blob_id = NULL;
2213 	char *filepath;
2214 	struct stat sb;
2215 
2216 	if (asprintf(&filepath, "%s%s%s", path,
2217 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2218 		return got_error_from_errno("asprintf");
2219 
2220 	if (lstat(filepath, &sb) != 0) {
2221 		err = got_error_from_errno2("lstat", path);
2222 		goto done;
2223 	}
2224 
2225 	err = got_object_blob_create(&blob_id, filepath, repo);
2226 	if (err)
2227 		goto done;
2228 
2229 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2230 	    blob_id);
2231 done:
2232 	free(filepath);
2233 	if (err)
2234 		free(blob_id);
2235 	return err;
2236 }
2237 
2238 static const struct got_error *
2239 insert_tree_entry(struct got_tree_entry *new_te,
2240     struct got_pathlist_head *paths)
2241 {
2242 	const struct got_error *err = NULL;
2243 	struct got_pathlist_entry *new_pe;
2244 
2245 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2246 	if (err)
2247 		return err;
2248 	if (new_pe == NULL)
2249 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2250 	return NULL;
2251 }
2252 
2253 static const struct got_error *write_tree(struct got_object_id **,
2254     const char *, struct got_pathlist_head *, struct got_repository *,
2255     got_repo_import_cb progress_cb, void *progress_arg);
2256 
2257 static const struct got_error *
2258 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2259     const char *path, struct got_pathlist_head *ignores,
2260     struct got_repository *repo,
2261     got_repo_import_cb progress_cb, void *progress_arg)
2262 {
2263 	const struct got_error *err;
2264 	struct got_object_id *id = NULL;
2265 	char *subdirpath;
2266 
2267 	if (asprintf(&subdirpath, "%s%s%s", path,
2268 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2269 		return got_error_from_errno("asprintf");
2270 
2271 	(*new_te) = calloc(1, sizeof(**new_te));
2272 	if (*new_te == NULL)
2273 		return got_error_from_errno("calloc");
2274 	(*new_te)->mode = S_IFDIR;
2275 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2276 	    sizeof((*new_te)->name)) {
2277 		err = got_error(GOT_ERR_NO_SPACE);
2278 		goto done;
2279 	}
2280 	err = write_tree(&id, subdirpath, ignores,  repo,
2281 	    progress_cb, progress_arg);
2282 	if (err)
2283 		goto done;
2284 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2285 
2286 done:
2287 	free(id);
2288 	free(subdirpath);
2289 	if (err) {
2290 		free(*new_te);
2291 		*new_te = NULL;
2292 	}
2293 	return err;
2294 }
2295 
2296 static const struct got_error *
2297 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2298     struct got_pathlist_head *ignores, struct got_repository *repo,
2299     got_repo_import_cb progress_cb, void *progress_arg)
2300 {
2301 	const struct got_error *err = NULL;
2302 	DIR *dir;
2303 	struct dirent *de;
2304 	int nentries;
2305 	struct got_tree_entry *new_te = NULL;
2306 	struct got_pathlist_head paths;
2307 	struct got_pathlist_entry *pe;
2308 
2309 	*new_tree_id = NULL;
2310 
2311 	TAILQ_INIT(&paths);
2312 
2313 	dir = opendir(path_dir);
2314 	if (dir == NULL) {
2315 		err = got_error_from_errno2("opendir", path_dir);
2316 		goto done;
2317 	}
2318 
2319 	nentries = 0;
2320 	while ((de = readdir(dir)) != NULL) {
2321 		int ignore = 0;
2322 		int type;
2323 
2324 		if (strcmp(de->d_name, ".") == 0 ||
2325 		    strcmp(de->d_name, "..") == 0)
2326 			continue;
2327 
2328 		err = got_path_dirent_type(&type, path_dir, de);
2329 		if (err)
2330 			goto done;
2331 
2332 		TAILQ_FOREACH(pe, ignores, entry) {
2333 			if (type == DT_DIR && pe->path_len > 0 &&
2334 			    pe->path[pe->path_len - 1] == '/') {
2335 				char stripped[PATH_MAX];
2336 
2337 				if (strlcpy(stripped, pe->path,
2338 				    sizeof(stripped)) >= sizeof(stripped)) {
2339 					err = got_error(GOT_ERR_NO_SPACE);
2340 					goto done;
2341 				}
2342 				got_path_strip_trailing_slashes(stripped);
2343 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2344 					ignore = 1;
2345 					break;
2346 				}
2347 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2348 				ignore = 1;
2349 				break;
2350 			}
2351 		}
2352 		if (ignore)
2353 			continue;
2354 
2355 		if (type == DT_DIR) {
2356 			err = import_subdir(&new_te, de, path_dir,
2357 			    ignores, repo, progress_cb, progress_arg);
2358 			if (err) {
2359 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2360 					goto done;
2361 				err = NULL;
2362 				continue;
2363 			}
2364 		} else if (type == DT_REG || type == DT_LNK) {
2365 			err = import_file(&new_te, de, path_dir, repo);
2366 			if (err)
2367 				goto done;
2368 		} else
2369 			continue;
2370 
2371 		err = insert_tree_entry(new_te, &paths);
2372 		if (err)
2373 			goto done;
2374 		nentries++;
2375 	}
2376 
2377 	if (TAILQ_EMPTY(&paths)) {
2378 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2379 		    "cannot create tree without any entries");
2380 		goto done;
2381 	}
2382 
2383 	TAILQ_FOREACH(pe, &paths, entry) {
2384 		struct got_tree_entry *te = pe->data;
2385 		char *path;
2386 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2387 			continue;
2388 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2389 			err = got_error_from_errno("asprintf");
2390 			goto done;
2391 		}
2392 		err = (*progress_cb)(progress_arg, path);
2393 		free(path);
2394 		if (err)
2395 			goto done;
2396 	}
2397 
2398 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2399 done:
2400 	if (dir)
2401 		closedir(dir);
2402 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2403 	return err;
2404 }
2405 
2406 const struct got_error *
2407 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2408     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2409     struct got_repository *repo, got_repo_import_cb progress_cb,
2410     void *progress_arg)
2411 {
2412 	const struct got_error *err;
2413 	struct got_object_id *new_tree_id;
2414 
2415 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2416 	    progress_cb, progress_arg);
2417 	if (err)
2418 		return err;
2419 
2420 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2421 	    author, time(NULL), author, time(NULL), logmsg, repo);
2422 	free(new_tree_id);
2423 	return err;
2424 }
2425 
2426 const struct got_error *
2427 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2428     struct got_repository *repo)
2429 {
2430 	const struct got_error *err = NULL;
2431 	char *path_objects = NULL, *path = NULL;
2432 	DIR *dir = NULL;
2433 	struct got_object_id id;
2434 	int i;
2435 
2436 	*nobjects = 0;
2437 	*ondisk_size = 0;
2438 
2439 	path_objects = got_repo_get_path_objects(repo);
2440 	if (path_objects == NULL)
2441 		return got_error_from_errno("got_repo_get_path_objects");
2442 
2443 	for (i = 0; i <= 0xff; i++) {
2444 		struct dirent *dent;
2445 
2446 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2447 			err = got_error_from_errno("asprintf");
2448 			break;
2449 		}
2450 
2451 		dir = opendir(path);
2452 		if (dir == NULL) {
2453 			if (errno == ENOENT) {
2454 				err = NULL;
2455 				continue;
2456 			}
2457 			err = got_error_from_errno2("opendir", path);
2458 			break;
2459 		}
2460 
2461 		while ((dent = readdir(dir)) != NULL) {
2462 			char *id_str;
2463 			int fd;
2464 			struct stat sb;
2465 
2466 			if (strcmp(dent->d_name, ".") == 0 ||
2467 			    strcmp(dent->d_name, "..") == 0)
2468 				continue;
2469 
2470 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2471 				err = got_error_from_errno("asprintf");
2472 				goto done;
2473 			}
2474 
2475 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2476 				free(id_str);
2477 				continue;
2478 			}
2479 			free(id_str);
2480 
2481 			err = got_object_open_loose_fd(&fd, &id, repo);
2482 			if (err)
2483 				goto done;
2484 
2485 			if (fstat(fd, &sb) == -1) {
2486 				err = got_error_from_errno("fstat");
2487 				close(fd);
2488 				goto done;
2489 			}
2490 			(*nobjects)++;
2491 			(*ondisk_size) += sb.st_size;
2492 
2493 			if (close(fd) == -1) {
2494 				err = got_error_from_errno("close");
2495 				goto done;
2496 			}
2497 		}
2498 
2499 		if (closedir(dir) != 0) {
2500 			err = got_error_from_errno("closedir");
2501 			goto done;
2502 		}
2503 		dir = NULL;
2504 
2505 		free(path);
2506 		path = NULL;
2507 	}
2508 done:
2509 	if (dir && closedir(dir) != 0 && err == NULL)
2510 		err = got_error_from_errno("closedir");
2511 
2512 	if (err) {
2513 		*nobjects = 0;
2514 		*ondisk_size = 0;
2515 	}
2516 	free(path_objects);
2517 	free(path);
2518 	return err;
2519 }
2520 
2521 const struct got_error *
2522 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2523     off_t *total_packsize, struct got_repository *repo)
2524 {
2525 	const struct got_error *err = NULL;
2526 	DIR *packdir = NULL;
2527 	struct dirent *dent;
2528 	struct got_packidx *packidx = NULL;
2529 	char *path_packidx;
2530 	char *path_packfile;
2531 	int packdir_fd;
2532 	struct stat sb;
2533 
2534 	*npackfiles = 0;
2535 	*nobjects = 0;
2536 	*total_packsize = 0;
2537 
2538 	packdir_fd = openat(got_repo_get_fd(repo),
2539 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2540 	if (packdir_fd == -1) {
2541 		return got_error_from_errno_fmt("openat: %s/%s",
2542 		    got_repo_get_path_git_dir(repo),
2543 		    GOT_OBJECTS_PACK_DIR);
2544 	}
2545 
2546 	packdir = fdopendir(packdir_fd);
2547 	if (packdir == NULL) {
2548 		err = got_error_from_errno("fdopendir");
2549 		goto done;
2550 	}
2551 
2552 	while ((dent = readdir(packdir)) != NULL) {
2553 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2554 			continue;
2555 
2556 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2557 		    dent->d_name) == -1) {
2558 			err = got_error_from_errno("asprintf");
2559 			goto done;
2560 		}
2561 
2562 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2563 		    path_packidx, 0);
2564 		free(path_packidx);
2565 		if (err)
2566 			goto done;
2567 
2568 		if (fstat(packidx->fd, &sb) == -1)
2569 			goto done;
2570 		*total_packsize += sb.st_size;
2571 
2572 		err = got_packidx_get_packfile_path(&path_packfile,
2573 		    packidx->path_packidx);
2574 		if (err)
2575 			goto done;
2576 
2577 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2578 		    0) == -1) {
2579 			free(path_packfile);
2580 			goto done;
2581 		}
2582 		free(path_packfile);
2583 		*total_packsize += sb.st_size;
2584 
2585 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2586 
2587 		(*npackfiles)++;
2588 
2589 		got_packidx_close(packidx);
2590 		packidx = NULL;
2591 	}
2592 done:
2593 	if (packidx)
2594 		got_packidx_close(packidx);
2595 	if (packdir && closedir(packdir) != 0 && err == NULL)
2596 		err = got_error_from_errno("closedir");
2597 	if (err) {
2598 		*npackfiles = 0;
2599 		*nobjects = 0;
2600 		*total_packsize = 0;
2601 	}
2602 	return err;
2603 }
2604 
2605 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2606     got_packidx_bloom_filter_cmp);