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 "got_compat.h"
18 
19 #include <sys/types.h>
20 #include <sys/queue.h>
21 #include <sys/uio.h>
22 #include <sys/socket.h>
23 #include <sys/stat.h>
24 #include <sys/mman.h>
25 #include <sys/resource.h>
26 
27 #include <ctype.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 <string.h>
35 #include <time.h>
36 #include <unistd.h>
37 #include <zlib.h>
38 #include <errno.h>
39 #include <libgen.h>
40 #include <stdint.h>
41 
42 #include "bloom.h"
43 
44 #include "got_error.h"
45 #include "got_reference.h"
46 #include "got_repository.h"
47 #include "got_path.h"
48 #include "got_cancel.h"
49 #include "got_object.h"
50 #include "got_opentemp.h"
51 
52 #include "got_lib_delta.h"
53 #include "got_lib_delta_cache.h"
54 #include "got_lib_inflate.h"
55 #include "got_lib_object.h"
56 #include "got_lib_object_parse.h"
57 #include "got_lib_object_create.h"
58 #include "got_lib_pack.h"
59 #include "got_lib_privsep.h"
60 #include "got_lib_hash.h"
61 #include "got_lib_object_cache.h"
62 #include "got_lib_repository.h"
63 #include "got_lib_gotconfig.h"
64 
65 #ifndef nitems
66 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
67 #endif
68 
69 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
70 
71 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
72     got_packidx_bloom_filter_cmp);
73 
74 static inline int
75 is_boolean_val(const char *val)
76 {
77 	return (strcasecmp(val, "true") == 0 ||
78 	    strcasecmp(val, "false") == 0 ||
79 	    strcasecmp(val, "on") == 0 ||
80     	    strcasecmp(val, "off") == 0 ||
81 	    strcasecmp(val, "yes") == 0 ||
82     	    strcasecmp(val, "no") == 0 ||
83 	    strcasecmp(val, "1") == 0 ||
84 	    strcasecmp(val, "0") == 0);
85 }
86 
87 static inline int
88 get_boolean_val(const char *val)
89 {
90 	return (strcasecmp(val, "true") == 0 ||
91 	    strcasecmp(val, "on") == 0 ||
92 	    strcasecmp(val, "yes") == 0 ||
93 	    strcasecmp(val, "1") == 0);
94 }
95 
96 const char *
97 got_repo_get_path(struct got_repository *repo)
98 {
99 	return repo->path;
100 }
101 
102 const char *
103 got_repo_get_path_git_dir(struct got_repository *repo)
104 {
105 	return repo->path_git_dir;
106 }
107 
108 int
109 got_repo_get_fd(struct got_repository *repo)
110 {
111 	return repo->gitdir_fd;
112 }
113 
114 enum got_hash_algorithm
115 got_repo_get_object_format(struct got_repository *repo)
116 {
117 	return repo->algo;
118 }
119 
120 const char *
121 got_repo_get_gitconfig_author_name(struct got_repository *repo)
122 {
123 	return repo->gitconfig_author_name;
124 }
125 
126 const char *
127 got_repo_get_gitconfig_author_email(struct got_repository *repo)
128 {
129 	return repo->gitconfig_author_email;
130 }
131 
132 const char *
133 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
134 {
135 	return repo->global_gitconfig_author_name;
136 }
137 
138 const char *
139 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
140 {
141 	return repo->global_gitconfig_author_email;
142 }
143 
144 const char *
145 got_repo_get_gitconfig_owner(struct got_repository *repo)
146 {
147 	return repo->gitconfig_owner;
148 }
149 
150 int
151 got_repo_has_extension(struct got_repository *repo, const char *ext)
152 {
153 	int i;
154 
155 	for (i = 0; i < repo->nextensions; ++i) {
156 		if (!strcasecmp(ext, repo->extnames[i]))
157 			return get_boolean_val(repo->extvals[i]);
158 	}
159 
160 	return 0;
161 }
162 
163 int
164 got_repo_is_bare(struct got_repository *repo)
165 {
166 	return (strcmp(repo->path, repo->path_git_dir) == 0);
167 }
168 
169 static char *
170 get_path_git_child(struct got_repository *repo, const char *basename)
171 {
172 	char *path_child;
173 
174 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
175 	    basename) == -1)
176 		return NULL;
177 
178 	return path_child;
179 }
180 
181 char *
182 got_repo_get_path_objects(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
185 }
186 
187 char *
188 got_repo_get_path_objects_pack(struct got_repository *repo)
189 {
190 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
191 }
192 
193 char *
194 got_repo_get_path_refs(struct got_repository *repo)
195 {
196 	return get_path_git_child(repo, GOT_REFS_DIR);
197 }
198 
199 char *
200 got_repo_get_path_packed_refs(struct got_repository *repo)
201 {
202 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
203 }
204 
205 static char *
206 get_path_head(struct got_repository *repo)
207 {
208 	return get_path_git_child(repo, GOT_HEAD_FILE);
209 }
210 
211 char *
212 got_repo_get_path_gitconfig(struct got_repository *repo)
213 {
214 	return get_path_git_child(repo, GOT_GITCONFIG);
215 }
216 
217 char *
218 got_repo_get_path_gotconfig(struct got_repository *repo)
219 {
220 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
221 }
222 
223 const struct got_gotconfig *
224 got_repo_get_gotconfig(struct got_repository *repo)
225 {
226 	return repo->gotconfig;
227 }
228 
229 void
230 got_repo_get_gitconfig_remotes(int *nremotes,
231     const struct got_remote_repo **remotes, struct got_repository *repo)
232 {
233 	*nremotes = repo->ngitconfig_remotes;
234 	*remotes = repo->gitconfig_remotes;
235 }
236 
237 static int
238 is_git_repo(struct got_repository *repo)
239 {
240 	const char *path_git = got_repo_get_path_git_dir(repo);
241 	char *path_objects = got_repo_get_path_objects(repo);
242 	char *path_refs = got_repo_get_path_refs(repo);
243 	char *path_head = get_path_head(repo);
244 	int ret = 0;
245 	struct stat sb;
246 	struct got_reference *head_ref;
247 
248 	if (lstat(path_git, &sb) == -1)
249 		goto done;
250 	if (!S_ISDIR(sb.st_mode))
251 		goto done;
252 
253 	if (lstat(path_objects, &sb) == -1)
254 		goto done;
255 	if (!S_ISDIR(sb.st_mode))
256 		goto done;
257 
258 	if (lstat(path_refs, &sb) == -1)
259 		goto done;
260 	if (!S_ISDIR(sb.st_mode))
261 		goto done;
262 
263 	if (lstat(path_head, &sb) == -1)
264 		goto done;
265 	if (!S_ISREG(sb.st_mode))
266 		goto done;
267 
268 	/* Check if the HEAD reference can be opened. */
269 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
270 		goto done;
271 	got_ref_close(head_ref);
272 
273 	ret = 1;
274 done:
275 	free(path_objects);
276 	free(path_refs);
277 	free(path_head);
278 	return ret;
279 
280 }
281 
282 static const struct got_error *
283 close_tempfiles(int *fds, size_t nfds)
284 {
285 	const struct got_error *err = NULL;
286 	int i;
287 
288 	for (i = 0; i < nfds; i++) {
289 		if (fds[i] == -1)
290 			continue;
291 		if (close(fds[i]) == -1) {
292 			err = got_error_from_errno("close");
293 			break;
294 		}
295 	}
296 	free(fds);
297 	return err;
298 }
299 
300 static const struct got_error *
301 open_tempfiles(int **fds, size_t array_size, size_t nfds)
302 {
303 	const struct got_error *err = NULL;
304 	int i;
305 
306 	*fds = calloc(array_size, sizeof(**fds));
307 	if (*fds == NULL)
308 		return got_error_from_errno("calloc");
309 
310 	for (i = 0; i < array_size; i++)
311 		(*fds)[i] = -1;
312 
313 	for (i = 0; i < nfds; i++) {
314 		(*fds)[i] = got_opentempfd();
315 		if ((*fds)[i] == -1) {
316 			err = got_error_from_errno("got_opentempfd");
317 			close_tempfiles(*fds, nfds);
318 			*fds = NULL;
319 			return err;
320 		}
321 	}
322 
323 	return NULL;
324 }
325 
326 static const struct got_error *
327 get_pack_cache_size(int *pack_cache_size)
328 {
329 	struct rlimit rl;
330 
331 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
332 		return got_error_from_errno("getrlimit");
333 
334 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
335 	if (*pack_cache_size > rl.rlim_cur / 8)
336 		*pack_cache_size = rl.rlim_cur / 8;
337 
338 	return NULL;
339 }
340 
341 const struct got_error *
342 got_repo_pack_fds_open(int **pack_fds)
343 {
344 	const struct got_error *err;
345 	int nfds;
346 
347 	err = get_pack_cache_size(&nfds);
348 	if (err)
349 		return err;
350 
351 	/*
352 	 * We need one basefd and one accumfd per cached pack.
353 	 * Our constants should be set up in a way such that
354 	 * this error never triggers.
355 	 */
356 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
357 		return got_error(GOT_ERR_NO_SPACE);
358 
359 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
360 }
361 
362 const struct got_error *
363 got_repo_pack_fds_close(int *pack_fds)
364 {
365 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
366 }
367 
368 const struct got_error *
369 got_repo_temp_fds_open(int **temp_fds)
370 {
371 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
372 	    GOT_REPO_NUM_TEMPFILES);
373 }
374 
375 void
376 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
377 {
378 	int i;
379 
380 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
381 		repo->tempfiles[i] = temp_fds[i];
382 }
383 
384 const struct got_error *
385 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
386 {
387 	int i;
388 
389 	*fd = -1;
390 	*idx = -1;
391 
392 	for (i = 0; i < nitems(repo->tempfiles); i++) {
393 		if (repo->tempfile_use_mask & (1 << i))
394 			continue;
395 		if (repo->tempfiles[i] != -1) {
396 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
397 				return got_error_from_errno("ftruncate");
398 			*fd = repo->tempfiles[i];
399 			*idx = i;
400 			repo->tempfile_use_mask |= (1 << i);
401 			return NULL;
402 		}
403 	}
404 
405 	return got_error(GOT_ERR_REPO_TEMPFILE);
406 }
407 
408 void
409 got_repo_temp_fds_put(int idx, struct got_repository *repo)
410 {
411 	repo->tempfile_use_mask &= ~(1 << idx);
412 }
413 
414 const struct got_error *
415 got_repo_temp_fds_close(int *temp_fds)
416 {
417 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
418 }
419 
420 const struct got_error *
421 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
422     struct got_object *obj)
423 {
424 #ifndef GOT_NO_OBJ_CACHE
425 	const struct got_error *err = NULL;
426 	err = got_object_cache_add(&repo->objcache, id, obj);
427 	if (err) {
428 		if (err->code == GOT_ERR_OBJ_EXISTS ||
429 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
430 			err = NULL;
431 		return err;
432 	}
433 	obj->refcnt++;
434 #endif
435 	return NULL;
436 }
437 
438 struct got_object *
439 got_repo_get_cached_object(struct got_repository *repo,
440     struct got_object_id *id)
441 {
442 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
443 }
444 
445 const struct got_error *
446 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
447     struct got_tree_object *tree)
448 {
449 #ifndef GOT_NO_OBJ_CACHE
450 	const struct got_error *err = NULL;
451 	err = got_object_cache_add(&repo->treecache, id, tree);
452 	if (err) {
453 		if (err->code == GOT_ERR_OBJ_EXISTS ||
454 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
455 			err = NULL;
456 		return err;
457 	}
458 	tree->refcnt++;
459 #endif
460 	return NULL;
461 }
462 
463 struct got_tree_object *
464 got_repo_get_cached_tree(struct got_repository *repo,
465     struct got_object_id *id)
466 {
467 	return (struct got_tree_object *)got_object_cache_get(
468 	    &repo->treecache, id);
469 }
470 
471 const struct got_error *
472 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
473     struct got_commit_object *commit)
474 {
475 #ifndef GOT_NO_OBJ_CACHE
476 	const struct got_error *err = NULL;
477 	err = got_object_cache_add(&repo->commitcache, id, commit);
478 	if (err) {
479 		if (err->code == GOT_ERR_OBJ_EXISTS ||
480 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
481 			err = NULL;
482 		return err;
483 	}
484 	commit->refcnt++;
485 #endif
486 	return NULL;
487 }
488 
489 struct got_commit_object *
490 got_repo_get_cached_commit(struct got_repository *repo,
491     struct got_object_id *id)
492 {
493 	return (struct got_commit_object *)got_object_cache_get(
494 	    &repo->commitcache, id);
495 }
496 
497 const struct got_error *
498 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
499     struct got_tag_object *tag)
500 {
501 #ifndef GOT_NO_OBJ_CACHE
502 	const struct got_error *err = NULL;
503 	err = got_object_cache_add(&repo->tagcache, id, tag);
504 	if (err) {
505 		if (err->code == GOT_ERR_OBJ_EXISTS ||
506 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
507 			err = NULL;
508 		return err;
509 	}
510 	tag->refcnt++;
511 #endif
512 	return NULL;
513 }
514 
515 struct got_tag_object *
516 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
517 {
518 	return (struct got_tag_object *)got_object_cache_get(
519 	    &repo->tagcache, id);
520 }
521 
522 const struct got_error *
523 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
524     struct got_raw_object *raw)
525 {
526 #ifndef GOT_NO_OBJ_CACHE
527 	const struct got_error *err = NULL;
528 	err = got_object_cache_add(&repo->rawcache, id, raw);
529 	if (err) {
530 		if (err->code == GOT_ERR_OBJ_EXISTS ||
531 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
532 			err = NULL;
533 		return err;
534 	}
535 	raw->refcnt++;
536 #endif
537 	return NULL;
538 }
539 
540 
541 struct got_raw_object *
542 got_repo_get_cached_raw_object(struct got_repository *repo,
543     struct got_object_id *id)
544 {
545 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
546 }
547 
548 
549 static const struct got_error *
550 open_repo(struct got_repository *repo, const char *path)
551 {
552 	const struct got_error *err = NULL;
553 
554 	repo->gitdir_fd = -1;
555 
556 	/* bare git repository? */
557 	repo->path_git_dir = strdup(path);
558 	if (repo->path_git_dir == NULL)
559 		return got_error_from_errno("strdup");
560 	if (is_git_repo(repo)) {
561 		repo->path = strdup(repo->path_git_dir);
562 		if (repo->path == NULL) {
563 			err = got_error_from_errno("strdup");
564 			goto done;
565 		}
566 		repo->gitdir_fd = open(repo->path_git_dir,
567 		    O_DIRECTORY | O_CLOEXEC);
568 		if (repo->gitdir_fd == -1) {
569 			err = got_error_from_errno2("open",
570 			    repo->path_git_dir);
571 			goto done;
572 		}
573 		return NULL;
574 	}
575 
576 	/* git repository with working tree? */
577 	free(repo->path_git_dir);
578 	repo->path_git_dir = NULL;
579 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
580 		err = got_error_from_errno("asprintf");
581 		goto done;
582 	}
583 	if (is_git_repo(repo)) {
584 		repo->path = strdup(path);
585 		if (repo->path == NULL) {
586 			err = got_error_from_errno("strdup");
587 			goto done;
588 		}
589 		repo->gitdir_fd = open(repo->path_git_dir,
590 		    O_DIRECTORY | O_CLOEXEC);
591 		if (repo->gitdir_fd == -1) {
592 			err = got_error_from_errno2("open",
593 			    repo->path_git_dir);
594 			goto done;
595 		}
596 		return NULL;
597 	}
598 
599 	err = got_error(GOT_ERR_NOT_GIT_REPO);
600 done:
601 	if (err) {
602 		free(repo->path);
603 		repo->path = NULL;
604 		free(repo->path_git_dir);
605 		repo->path_git_dir = NULL;
606 		if (repo->gitdir_fd != -1)
607 			close(repo->gitdir_fd);
608 		repo->gitdir_fd = -1;
609 
610 	}
611 	return err;
612 }
613 
614 static const struct got_error *
615 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
616 {
617 	const struct got_error *err = NULL;
618 	char *repo_gitconfig_path = NULL;
619 
620 	if (global_gitconfig_path) {
621 		/* Read settings from ~/.gitconfig. */
622 		int dummy_repo_version;
623 		err = got_repo_read_gitconfig(&dummy_repo_version,
624 		    &repo->global_gitconfig_author_name,
625 		    &repo->global_gitconfig_author_email,
626 		    NULL, NULL, NULL, NULL, NULL, NULL,
627 		    global_gitconfig_path);
628 		if (err)
629 			return err;
630 	}
631 
632 	/* Read repository's .git/config file. */
633 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
634 	if (repo_gitconfig_path == NULL)
635 		return got_error_from_errno("got_repo_get_path_gitconfig");
636 
637 	err = got_repo_read_gitconfig(
638 	    &repo->gitconfig_repository_format_version,
639 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
640 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
641 	    &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
642 	    &repo->nextensions, repo_gitconfig_path);
643 	if (err)
644 		goto done;
645 
646 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
647 		int i;
648 
649 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
650 			got_repo_free_remote_repo_data(
651 			    &repo->gitconfig_remotes[i]);
652 		}
653 		free(repo->gitconfig_remotes);
654 		repo->gitconfig_remotes = NULL;
655 		repo->ngitconfig_remotes = 0;
656 
657 		free(repo->gitconfig_author_name);
658 		repo->gitconfig_author_name = NULL;
659 		free(repo->gitconfig_author_email);
660 		repo->gitconfig_author_email = NULL;
661 
662 		free(repo->global_gitconfig_author_name);
663 		repo->global_gitconfig_author_name = NULL;
664 		free(repo->global_gitconfig_author_email);
665 		repo->global_gitconfig_author_email = NULL;
666 	}
667 
668 done:
669 	free(repo_gitconfig_path);
670 	return err;
671 }
672 
673 static const struct got_error *
674 read_gotconfig(struct got_repository *repo)
675 {
676 	const struct got_error *err = NULL;
677 	char *gotconfig_path;
678 
679 	gotconfig_path = got_repo_get_path_gotconfig(repo);
680 	if (gotconfig_path == NULL)
681 		return got_error_from_errno("got_repo_get_path_gotconfig");
682 
683 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
684 	free(gotconfig_path);
685 	return err;
686 }
687 
688 /* Supported repository format extensions. */
689 static const char *const repo_extensions[] = {
690 	"noop",			/* Got supports repository format version 1. */
691 	"preciousObjects",	/* Supported by gotadmin cleanup. */
692 	"worktreeConfig",	/* Got does not care about Git work trees. */
693 };
694 
695 const struct got_error *
696 got_repo_open(struct got_repository **repop, const char *path,
697     const char *global_gitconfig_path, int *pack_fds)
698 {
699 	struct got_repository *repo = NULL;
700 	const struct got_error *err = NULL;
701 	char *repo_path = NULL;
702 	size_t i, j = 0;
703 
704 	*repop = NULL;
705 
706 	repo = calloc(1, sizeof(*repo));
707 	if (repo == NULL)
708 		return got_error_from_errno("calloc");
709 
710 	RB_INIT(&repo->packidx_bloom_filters);
711 	TAILQ_INIT(&repo->packidx_paths);
712 
713 	for (i = 0; i < nitems(repo->privsep_children); i++) {
714 		memset(&repo->privsep_children[i], 0,
715 		    sizeof(repo->privsep_children[0]));
716 		repo->privsep_children[i].imsg_fd = -1;
717 	}
718 
719 	err = got_object_cache_init(&repo->objcache,
720 	    GOT_OBJECT_CACHE_TYPE_OBJ);
721 	if (err)
722 		goto done;
723 	err = got_object_cache_init(&repo->treecache,
724 	    GOT_OBJECT_CACHE_TYPE_TREE);
725 	if (err)
726 		goto done;
727 	err = got_object_cache_init(&repo->commitcache,
728 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
729 	if (err)
730 		goto done;
731 	err = got_object_cache_init(&repo->tagcache,
732 	    GOT_OBJECT_CACHE_TYPE_TAG);
733 	if (err)
734 		goto done;
735 	err = got_object_cache_init(&repo->rawcache,
736 	    GOT_OBJECT_CACHE_TYPE_RAW);
737 	if (err)
738 		goto done;
739 
740 	err = get_pack_cache_size(&repo->pack_cache_size);
741 	if (err)
742 		goto done;
743 	for (i = 0; i < nitems(repo->packs); i++) {
744 		if (pack_fds != NULL && i < repo->pack_cache_size) {
745 			repo->packs[i].basefd = pack_fds[j++];
746 			repo->packs[i].accumfd = pack_fds[j++];
747 		} else {
748 			repo->packs[i].basefd = -1;
749 			repo->packs[i].accumfd = -1;
750 		}
751 	}
752 	for (i = 0; i < nitems(repo->tempfiles); i++)
753 		repo->tempfiles[i] = -1;
754 	repo->pinned_pack = -1;
755 	repo->pinned_packidx = -1;
756 	repo->pinned_pid = 0;
757 
758 	repo_path = realpath(path, NULL);
759 	if (repo_path == NULL) {
760 		err = got_error_from_errno2("realpath", path);
761 		goto done;
762 	}
763 
764 	for (;;) {
765 		char *parent_path;
766 
767 		err = open_repo(repo, repo_path);
768 		if (err == NULL)
769 			break;
770 		if (err->code != GOT_ERR_NOT_GIT_REPO)
771 			goto done;
772 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
773 			err = got_error(GOT_ERR_NOT_GIT_REPO);
774 			goto done;
775 		}
776 		err = got_path_dirname(&parent_path, repo_path);
777 		if (err)
778 			goto done;
779 		free(repo_path);
780 		repo_path = parent_path;
781 	}
782 
783 	err = read_gotconfig(repo);
784 	if (err)
785 		goto done;
786 
787 	err = read_gitconfig(repo, global_gitconfig_path);
788 	if (err)
789 		goto done;
790 	if (repo->gitconfig_repository_format_version != 0) {
791 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
792 		goto done;
793 	}
794 	for (i = 0; i < repo->nextensions; i++) {
795 		char *ext = repo->extnames[i];
796 		char *val = repo->extvals[i];
797 		int j, supported = 0;
798 
799 		if (!is_boolean_val(val)) {
800 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
801 			goto done;
802 		}
803 
804 		if (!get_boolean_val(val))
805 			continue;
806 
807 		for (j = 0; j < nitems(repo_extensions); j++) {
808 			if (strcmp(ext, repo_extensions[j]) == 0) {
809 				supported = 1;
810 				break;
811 			}
812 		}
813 		if (!supported) {
814 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
815 			goto done;
816 		}
817 	}
818 
819 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
820 done:
821 	if (err)
822 		got_repo_close(repo);
823 	else
824 		*repop = repo;
825 	free(repo_path);
826 	return err;
827 }
828 
829 const struct got_error *
830 got_repo_close(struct got_repository *repo)
831 {
832 	const struct got_error *err = NULL, *child_err;
833 	struct got_packidx_bloom_filter *bf;
834 	size_t i;
835 
836 	for (i = 0; i < repo->pack_cache_size; i++) {
837 		if (repo->packidx_cache[i] == NULL)
838 			break;
839 		got_packidx_close(repo->packidx_cache[i]);
840 	}
841 
842 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
843 	    &repo->packidx_bloom_filters))) {
844 		RB_REMOVE(got_packidx_bloom_filter_tree,
845 		    &repo->packidx_bloom_filters, bf);
846 		bloom_free(bf->bloom);
847 		free(bf->bloom);
848 		free(bf);
849 	}
850 
851 	for (i = 0; i < repo->pack_cache_size; i++)
852 		if (repo->packs[i].path_packfile)
853 			if (repo->packs[i].path_packfile)
854 				got_pack_close(&repo->packs[i]);
855 
856 	free(repo->path);
857 	free(repo->path_git_dir);
858 
859 	got_object_cache_close(&repo->objcache);
860 	got_object_cache_close(&repo->treecache);
861 	got_object_cache_close(&repo->commitcache);
862 	got_object_cache_close(&repo->tagcache);
863 	got_object_cache_close(&repo->rawcache);
864 
865 	for (i = 0; i < nitems(repo->privsep_children); i++) {
866 		if (repo->privsep_children[i].imsg_fd == -1)
867 			continue;
868 		imsg_clear(repo->privsep_children[i].ibuf);
869 		free(repo->privsep_children[i].ibuf);
870 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
871 		if (err && err->code == GOT_ERR_EOF)
872 			err = NULL;
873 		child_err = got_privsep_wait_for_child(
874 		    repo->privsep_children[i].pid);
875 		if (child_err && err == NULL)
876 			err = child_err;
877 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
878 		    err == NULL)
879 			err = got_error_from_errno("close");
880 	}
881 
882 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
883 	    err == NULL)
884 		err = got_error_from_errno("close");
885 
886 	if (repo->gotconfig)
887 		got_gotconfig_free(repo->gotconfig);
888 	free(repo->gitconfig_author_name);
889 	free(repo->gitconfig_author_email);
890 	for (i = 0; i < repo->ngitconfig_remotes; i++)
891 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
892 	free(repo->gitconfig_remotes);
893 	for (i = 0; i < repo->nextensions; i++) {
894 		free(repo->extnames[i]);
895 		free(repo->extvals[i]);
896 	}
897 	free(repo->extnames);
898 	free(repo->extvals);
899 
900 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
901 	free(repo);
902 
903 	return err;
904 }
905 
906 void
907 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
908 {
909 	int i;
910 
911 	free(repo->name);
912 	repo->name = NULL;
913 	free(repo->fetch_url);
914 	repo->fetch_url = NULL;
915 	free(repo->send_url);
916 	repo->send_url = NULL;
917 	for (i = 0; i < repo->nfetch_branches; i++)
918 		free(repo->fetch_branches[i]);
919 	free(repo->fetch_branches);
920 	repo->fetch_branches = NULL;
921 	repo->nfetch_branches = 0;
922 	for (i = 0; i < repo->nsend_branches; i++)
923 		free(repo->send_branches[i]);
924 	free(repo->send_branches);
925 	repo->send_branches = NULL;
926 	repo->nsend_branches = 0;
927 }
928 
929 const struct got_error *
930 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
931     const char *input_path)
932 {
933 	const struct got_error *err = NULL;
934 	const char *repo_abspath = NULL;
935 	size_t repolen, len;
936 	char *canonpath, *path = NULL;
937 
938 	*in_repo_path = NULL;
939 
940 	canonpath = strdup(input_path);
941 	if (canonpath == NULL) {
942 		err = got_error_from_errno("strdup");
943 		goto done;
944 	}
945 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
946 	if (err)
947 		goto done;
948 
949 	repo_abspath = got_repo_get_path(repo);
950 
951 	if (canonpath[0] == '\0') {
952 		path = strdup(canonpath);
953 		if (path == NULL) {
954 			err = got_error_from_errno("strdup");
955 			goto done;
956 		}
957 	} else {
958 		path = realpath(canonpath, NULL);
959 		if (path == NULL) {
960 			if (errno != ENOENT) {
961 				err = got_error_from_errno2("realpath",
962 				    canonpath);
963 				goto done;
964 			}
965 			/*
966 			 * Path is not on disk.
967 			 * Assume it is already relative to repository root.
968 			 */
969 			path = strdup(canonpath);
970 			if (path == NULL) {
971 				err = got_error_from_errno("strdup");
972 				goto done;
973 			}
974 		}
975 
976 		repolen = strlen(repo_abspath);
977 		len = strlen(path);
978 
979 
980 		if (strcmp(path, repo_abspath) == 0) {
981 			free(path);
982 			path = strdup("");
983 			if (path == NULL) {
984 				err = got_error_from_errno("strdup");
985 				goto done;
986 			}
987 		} else if (len > repolen &&
988 		    got_path_is_child(path, repo_abspath, repolen)) {
989 			/* Matched an on-disk path inside repository. */
990 			if (got_repo_is_bare(repo)) {
991 				/*
992 				 * Matched an on-disk path inside repository
993 				 * database. Treat input as repository-relative.
994 				 */
995 				free(path);
996 				path = canonpath;
997 				canonpath = NULL;
998 			} else {
999 				char *child;
1000 				/* Strip common prefix with repository path. */
1001 				err = got_path_skip_common_ancestor(&child,
1002 				    repo_abspath, path);
1003 				if (err)
1004 					goto done;
1005 				free(path);
1006 				path = child;
1007 			}
1008 		} else {
1009 			/*
1010 			 * Matched unrelated on-disk path.
1011 			 * Treat input as repository-relative.
1012 			 */
1013 			free(path);
1014 			path = canonpath;
1015 			canonpath = NULL;
1016 		}
1017 	}
1018 
1019 	/* Make in-repository path absolute */
1020 	if (path[0] != '/') {
1021 		char *abspath;
1022 		if (asprintf(&abspath, "/%s", path) == -1) {
1023 			err = got_error_from_errno("asprintf");
1024 			goto done;
1025 		}
1026 		free(path);
1027 		path = abspath;
1028 	}
1029 
1030 done:
1031 	free(canonpath);
1032 	if (err)
1033 		free(path);
1034 	else
1035 		*in_repo_path = path;
1036 	return err;
1037 }
1038 
1039 static const struct got_error *
1040 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1041     const char *path_packidx)
1042 {
1043 	const struct got_error *err = NULL;
1044 	size_t i;
1045 
1046 	for (i = 0; i < repo->pack_cache_size; i++) {
1047 		if (repo->packidx_cache[i] == NULL)
1048 			break;
1049 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1050 		    path_packidx) == 0) {
1051 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1052 		}
1053 	}
1054 	if (i == repo->pack_cache_size) {
1055 		do {
1056 			i--;
1057 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1058 		    i == repo->pinned_packidx);
1059 		err = got_packidx_close(repo->packidx_cache[i]);
1060 		if (err)
1061 			return err;
1062 	}
1063 
1064 	repo->packidx_cache[i] = packidx;
1065 
1066 	return NULL;
1067 }
1068 
1069 int
1070 got_repo_is_packidx_filename(const char *name, size_t len)
1071 {
1072 	if (len != GOT_PACKIDX_NAMELEN)
1073 		return 0;
1074 
1075 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1076 		return 0;
1077 
1078 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1079 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1080 		return 0;
1081 
1082 	return 1;
1083 }
1084 
1085 static struct got_packidx_bloom_filter *
1086 get_packidx_bloom_filter(struct got_repository *repo,
1087     const char *path, size_t path_len)
1088 {
1089 	struct got_packidx_bloom_filter key;
1090 
1091 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1092 		return NULL; /* XXX */
1093 	key.path_len = path_len;
1094 
1095 	return RB_FIND(got_packidx_bloom_filter_tree,
1096 	    &repo->packidx_bloom_filters, &key);
1097 }
1098 
1099 int
1100 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1101     const char *path_packidx, struct got_object_id *id)
1102 {
1103 	struct got_packidx_bloom_filter *bf;
1104 
1105 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1106 	if (bf)
1107 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1108 
1109 	/* No bloom filter means this pack index must be searched. */
1110 	return 1;
1111 }
1112 
1113 static const struct got_error *
1114 add_packidx_bloom_filter(struct got_repository *repo,
1115     struct got_packidx *packidx, const char *path_packidx)
1116 {
1117 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1118 	struct got_packidx_bloom_filter *bf;
1119 	size_t len;
1120 
1121 	/*
1122 	 * Don't use bloom filters for very large pack index files.
1123 	 * Large pack files will contain a relatively large fraction
1124 	 * of our objects so we will likely need to visit them anyway.
1125 	 * The more objects a pack file contains the higher the probability
1126 	 * of a false-positive match from the bloom filter. And reading
1127 	 * all object IDs from a large pack index file can be expensive.
1128 	 */
1129 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1130 		return NULL;
1131 
1132 	/* Do we already have a filter for this pack index? */
1133 	if (get_packidx_bloom_filter(repo, path_packidx,
1134 	    strlen(path_packidx)) != NULL)
1135 		return NULL;
1136 
1137 	bf = calloc(1, sizeof(*bf));
1138 	if (bf == NULL)
1139 		return got_error_from_errno("calloc");
1140 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1141 	if (bf->bloom == NULL) {
1142 		free(bf);
1143 		return got_error_from_errno("calloc");
1144 	}
1145 
1146 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1147 	if (len >= sizeof(bf->path)) {
1148 		free(bf->bloom);
1149 		free(bf);
1150 		return got_error(GOT_ERR_NO_SPACE);
1151 	}
1152 	bf->path_len = len;
1153 
1154 	/* Minimum size supported by our bloom filter is 1000 entries. */
1155 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1156 	for (i = 0; i < nobjects; i++) {
1157 		struct got_packidx_object_id *id;
1158 		id = &packidx->hdr.sorted_ids[i];
1159 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1160 	}
1161 
1162 	RB_INSERT(got_packidx_bloom_filter_tree,
1163 	    &repo->packidx_bloom_filters, bf);
1164 	return NULL;
1165 }
1166 
1167 static void
1168 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1169 {
1170 	struct got_pathlist_entry *pe;
1171 
1172 	while (!TAILQ_EMPTY(packidx_paths)) {
1173 		pe = TAILQ_FIRST(packidx_paths);
1174 		TAILQ_REMOVE(packidx_paths, pe, entry);
1175 		free((char *)pe->path);
1176 		free(pe);
1177 	}
1178 }
1179 
1180 static const struct got_error *
1181 refresh_packidx_paths(struct got_repository *repo)
1182 {
1183 	const struct got_error *err = NULL;
1184 	char *objects_pack_dir = NULL;
1185 	struct stat sb;
1186 
1187 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1188 	if (objects_pack_dir == NULL)
1189 		return got_error_from_errno("got_repo_get_path_objects_pack");
1190 
1191 	if (stat(objects_pack_dir, &sb) == -1) {
1192 		if (errno != ENOENT) {
1193 			err = got_error_from_errno2("stat", objects_pack_dir);
1194 			goto done;
1195 		}
1196 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1197 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1198 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1199 		purge_packidx_paths(&repo->packidx_paths);
1200 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1201 		if (err)
1202 			goto done;
1203 	}
1204 done:
1205 	free(objects_pack_dir);
1206 	return err;
1207 }
1208 
1209 const struct got_error *
1210 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1211     struct got_repository *repo, struct got_object_id *id)
1212 {
1213 	const struct got_error *err;
1214 	struct got_pathlist_entry *pe;
1215 	size_t i;
1216 
1217 	/* Search pack index cache. */
1218 	for (i = 0; i < repo->pack_cache_size; i++) {
1219 		if (repo->packidx_cache[i] == NULL)
1220 			break;
1221 		if (!got_repo_check_packidx_bloom_filter(repo,
1222 		    repo->packidx_cache[i]->path_packidx, id))
1223 			continue; /* object will not be found in this index */
1224 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1225 		if (*idx != -1) {
1226 			*packidx = repo->packidx_cache[i];
1227 			/*
1228 			 * Move this cache entry to the front. Repeatedly
1229 			 * searching a wrong pack index can be expensive.
1230 			 */
1231 			if (i > 0) {
1232 				memmove(&repo->packidx_cache[1],
1233 				    &repo->packidx_cache[0],
1234 				    i * sizeof(repo->packidx_cache[0]));
1235 				repo->packidx_cache[0] = *packidx;
1236 				if (repo->pinned_packidx >= 0 &&
1237 				    repo->pinned_packidx < i)
1238 					repo->pinned_packidx++;
1239 				else if (repo->pinned_packidx == i)
1240 					repo->pinned_packidx = 0;
1241 			}
1242 			return NULL;
1243 		}
1244 	}
1245 	/* No luck. Search the filesystem. */
1246 
1247 	err = refresh_packidx_paths(repo);
1248 	if (err)
1249 		return err;
1250 
1251 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1252 		const char *path_packidx = pe->path;
1253 		int is_cached = 0;
1254 
1255 		if (!got_repo_check_packidx_bloom_filter(repo,
1256 		    pe->path, id))
1257 			continue; /* object will not be found in this index */
1258 
1259 		for (i = 0; i < repo->pack_cache_size; i++) {
1260 			if (repo->packidx_cache[i] == NULL)
1261 				break;
1262 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1263 			    path_packidx) == 0) {
1264 				is_cached = 1;
1265 				break;
1266 			}
1267 		}
1268 		if (is_cached)
1269 			continue; /* already searched */
1270 
1271 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1272 		    path_packidx, 0);
1273 		if (err)
1274 			goto done;
1275 
1276 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1277 		if (err)
1278 			goto done;
1279 
1280 		err = cache_packidx(repo, *packidx, path_packidx);
1281 		if (err)
1282 			goto done;
1283 
1284 		*idx = got_packidx_get_object_idx(*packidx, id);
1285 		if (*idx != -1) {
1286 			err = NULL; /* found the object */
1287 			goto done;
1288 		}
1289 	}
1290 
1291 	err = got_error_no_obj(id);
1292 done:
1293 	return err;
1294 }
1295 
1296 const struct got_error *
1297 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1298     struct got_repository *repo)
1299 {
1300 	const struct got_error *err = NULL;
1301 	DIR *packdir = NULL;
1302 	struct dirent *dent;
1303 	char *path_packidx = NULL;
1304 	int packdir_fd;
1305 	struct stat sb;
1306 
1307 	packdir_fd = openat(got_repo_get_fd(repo),
1308 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1309 	if (packdir_fd == -1) {
1310 		return got_error_from_errno_fmt("openat: %s/%s",
1311 		    got_repo_get_path_git_dir(repo),
1312 		    GOT_OBJECTS_PACK_DIR);
1313 	}
1314 
1315 	packdir = fdopendir(packdir_fd);
1316 	if (packdir == NULL) {
1317 		err = got_error_from_errno("fdopendir");
1318 		goto done;
1319 	}
1320 
1321 	if (fstat(packdir_fd, &sb) == -1) {
1322 		err = got_error_from_errno("fstat");
1323 		goto done;
1324 	}
1325 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1326 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1327 
1328 	while ((dent = readdir(packdir)) != NULL) {
1329 		if (!got_repo_is_packidx_filename(dent->d_name,
1330 		    strlen(dent->d_name)))
1331 			continue;
1332 
1333 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1334 		    dent->d_name) == -1) {
1335 			err = got_error_from_errno("asprintf");
1336 			path_packidx = NULL;
1337 			break;
1338 		}
1339 
1340 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1341 		if (err)
1342 			break;
1343 	}
1344 done:
1345 	if (err)
1346 		free(path_packidx);
1347 	if (packdir && closedir(packdir) != 0 && err == NULL)
1348 		err = got_error_from_errno("closedir");
1349 	return err;
1350 }
1351 
1352 const struct got_error *
1353 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1354     struct got_repository *repo)
1355 {
1356 	const struct got_error *err;
1357 	size_t i;
1358 
1359 	*packidx = NULL;
1360 
1361 	/* Search pack index cache. */
1362 	for (i = 0; i < repo->pack_cache_size; i++) {
1363 		if (repo->packidx_cache[i] == NULL)
1364 			break;
1365 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1366 		    path_packidx) == 0) {
1367 			*packidx = repo->packidx_cache[i];
1368 			return NULL;
1369 		}
1370 	}
1371 	/* No luck. Search the filesystem. */
1372 
1373 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1374 	    path_packidx, 0);
1375 	if (err)
1376 		return err;
1377 
1378 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1379 	if (err)
1380 		goto done;
1381 
1382 	err = cache_packidx(repo, *packidx, path_packidx);
1383 done:
1384 	if (err) {
1385 		got_packidx_close(*packidx);
1386 		*packidx = NULL;
1387 	}
1388 	return err;
1389 }
1390 
1391 static const struct got_error *
1392 read_packfile_hdr(int fd, struct got_packidx *packidx)
1393 {
1394 	const struct got_error *err = NULL;
1395 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1396 	struct got_packfile_hdr hdr;
1397 	ssize_t n;
1398 
1399 	n = read(fd, &hdr, sizeof(hdr));
1400 	if (n < 0)
1401 		return got_error_from_errno("read");
1402 	if (n != sizeof(hdr))
1403 		return got_error(GOT_ERR_BAD_PACKFILE);
1404 
1405 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1406 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1407 	    be32toh(hdr.nobjects) != totobj)
1408 		err = got_error(GOT_ERR_BAD_PACKFILE);
1409 
1410 	return err;
1411 }
1412 
1413 static const struct got_error *
1414 open_packfile(int *fd, struct got_repository *repo,
1415     const char *relpath, struct got_packidx *packidx)
1416 {
1417 	const struct got_error *err = NULL;
1418 
1419 	*fd = openat(got_repo_get_fd(repo), relpath,
1420 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1421 	if (*fd == -1)
1422 		return got_error_from_errno_fmt("openat: %s/%s",
1423 		    got_repo_get_path_git_dir(repo), relpath);
1424 
1425 	if (packidx) {
1426 		err = read_packfile_hdr(*fd, packidx);
1427 		if (err) {
1428 			close(*fd);
1429 			*fd = -1;
1430 		}
1431 	}
1432 
1433 	return err;
1434 }
1435 
1436 const struct got_error *
1437 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1438     const char *path_packfile, struct got_packidx *packidx)
1439 {
1440 	const struct got_error *err = NULL;
1441 	struct got_pack *pack = NULL;
1442 	struct stat sb;
1443 	size_t i;
1444 
1445 	if (packp)
1446 		*packp = NULL;
1447 
1448 	for (i = 0; i < repo->pack_cache_size; i++) {
1449 		pack = &repo->packs[i];
1450 		if (pack->path_packfile == NULL)
1451 			break;
1452 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1453 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1454 	}
1455 
1456 	if (i == repo->pack_cache_size) {
1457 		struct got_pack tmp;
1458 		do {
1459 			i--;
1460 		} while (i > 0 && repo->pinned_pack >= 0 &&
1461 		    i == repo->pinned_pack);
1462 		err = got_pack_close(&repo->packs[i]);
1463 		if (err)
1464 			return err;
1465 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1466 			return got_error_from_errno("ftruncate");
1467 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1468 			return got_error_from_errno("ftruncate");
1469 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1470 		memcpy(&repo->packs[i], &repo->packs[0],
1471 		    sizeof(repo->packs[i]));
1472 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1473 		if (repo->pinned_pack == 0)
1474 			repo->pinned_pack = i;
1475 		else if (repo->pinned_pack == i)
1476 			repo->pinned_pack = 0;
1477 		i = 0;
1478 	}
1479 
1480 	pack = &repo->packs[i];
1481 
1482 	pack->path_packfile = strdup(path_packfile);
1483 	if (pack->path_packfile == NULL) {
1484 		err = got_error_from_errno("strdup");
1485 		goto done;
1486 	}
1487 
1488 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1489 	if (err)
1490 		goto done;
1491 
1492 	if (fstat(pack->fd, &sb) != 0) {
1493 		err = got_error_from_errno("fstat");
1494 		goto done;
1495 	}
1496 	pack->filesize = sb.st_size;
1497 
1498 	pack->privsep_child = NULL;
1499 
1500 	err = got_delta_cache_alloc(&pack->delta_cache);
1501 	if (err)
1502 		goto done;
1503 
1504 #ifndef GOT_PACK_NO_MMAP
1505 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1506 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1507 		    pack->fd, 0);
1508 		if (pack->map == MAP_FAILED) {
1509 			if (errno != ENOMEM) {
1510 				err = got_error_from_errno("mmap");
1511 				goto done;
1512 			}
1513 			pack->map = NULL; /* fall back to read(2) */
1514 		}
1515 	}
1516 #endif
1517 done:
1518 	if (err) {
1519 		if (pack)
1520 			got_pack_close(pack);
1521 	} else if (packp)
1522 		*packp = pack;
1523 	return err;
1524 }
1525 
1526 struct got_pack *
1527 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1528 {
1529 	struct got_pack *pack = NULL;
1530 	size_t i;
1531 
1532 	for (i = 0; i < repo->pack_cache_size; i++) {
1533 		pack = &repo->packs[i];
1534 		if (pack->path_packfile == NULL)
1535 			break;
1536 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1537 			return pack;
1538 	}
1539 
1540 	return NULL;
1541 }
1542 
1543 const struct got_error *
1544 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1545     struct got_pack *pack)
1546 {
1547 	size_t i;
1548 	int pinned_pack = -1, pinned_packidx = -1;
1549 
1550 	for (i = 0; i < repo->pack_cache_size; i++) {
1551 		if (repo->packidx_cache[i] &&
1552 		    strcmp(repo->packidx_cache[i]->path_packidx,
1553 		    packidx->path_packidx) == 0)
1554 			pinned_packidx = i;
1555 		if (repo->packs[i].path_packfile &&
1556 		    strcmp(repo->packs[i].path_packfile,
1557 		    pack->path_packfile) == 0)
1558 			pinned_pack = i;
1559 	}
1560 
1561 	if (pinned_packidx == -1 || pinned_pack == -1)
1562 		return got_error(GOT_ERR_PIN_PACK);
1563 
1564 	repo->pinned_pack = pinned_pack;
1565 	repo->pinned_packidx = pinned_packidx;
1566 	if (repo->packs[pinned_pack].privsep_child)
1567 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1568 	return NULL;
1569 }
1570 
1571 struct got_pack *
1572 got_repo_get_pinned_pack(struct got_repository *repo)
1573 {
1574 	if (repo->pinned_pack >= 0 &&
1575 	    repo->pinned_pack < repo->pack_cache_size)
1576 		return &repo->packs[repo->pinned_pack];
1577 
1578 	return NULL;
1579 }
1580 
1581 void
1582 got_repo_unpin_pack(struct got_repository *repo)
1583 {
1584 	repo->pinned_packidx = -1;
1585 	repo->pinned_pack = -1;
1586 	repo->pinned_pid = 0;
1587 }
1588 
1589 const struct got_error *
1590 got_repo_init(const char *repo_path, const char *head_name)
1591 {
1592 	const struct got_error *err = NULL;
1593 	const char *dirnames[] = {
1594 		GOT_OBJECTS_DIR,
1595 		GOT_OBJECTS_PACK_DIR,
1596 		GOT_REFS_DIR,
1597 	};
1598 	const char *description_str = "Unnamed repository; "
1599 	    "edit this file 'description' to name the repository.";
1600 	const char *headref = "ref: refs/heads/";
1601 	const char *gitconfig_str = "[core]\n"
1602 	    "\trepositoryformatversion = 0\n"
1603 	    "\tfilemode = true\n"
1604 	    "\tbare = true\n";
1605 	char *headref_str, *path;
1606 	size_t i;
1607 
1608 	if (!got_path_dir_is_empty(repo_path))
1609 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1610 
1611 	for (i = 0; i < nitems(dirnames); i++) {
1612 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1613 			return got_error_from_errno("asprintf");
1614 		}
1615 		err = got_path_mkdir(path);
1616 		free(path);
1617 		if (err)
1618 			return err;
1619 	}
1620 
1621 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1622 		return got_error_from_errno("asprintf");
1623 	err = got_path_create_file(path, description_str);
1624 	free(path);
1625 	if (err)
1626 		return err;
1627 
1628 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1629 		return got_error_from_errno("asprintf");
1630 	if (asprintf(&headref_str, "%s%s", headref,
1631 	    head_name ? head_name : "main") == -1) {
1632 		free(path);
1633 		return got_error_from_errno("asprintf");
1634 	}
1635 	err = got_path_create_file(path, headref_str);
1636 	free(headref_str);
1637 	free(path);
1638 	if (err)
1639 		return err;
1640 
1641 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1642 		return got_error_from_errno("asprintf");
1643 	err = got_path_create_file(path, gitconfig_str);
1644 	free(path);
1645 	if (err)
1646 		return err;
1647 
1648 	return NULL;
1649 }
1650 
1651 static const struct got_error *
1652 match_packed_object(struct got_object_id **unique_id,
1653     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1654 {
1655 	const struct got_error *err = NULL;
1656 	struct got_object_id_queue matched_ids;
1657 	struct got_pathlist_entry *pe;
1658 
1659 	STAILQ_INIT(&matched_ids);
1660 
1661 	err = refresh_packidx_paths(repo);
1662 	if (err)
1663 		return err;
1664 
1665 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1666 		const char *path_packidx = pe->path;
1667 		struct got_packidx *packidx;
1668 		struct got_object_qid *qid;
1669 
1670 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1671 		    path_packidx, 0);
1672 		if (err)
1673 			break;
1674 
1675 		err = got_packidx_match_id_str_prefix(&matched_ids,
1676 		    packidx, id_str_prefix);
1677 		if (err) {
1678 			got_packidx_close(packidx);
1679 			break;
1680 		}
1681 		err = got_packidx_close(packidx);
1682 		if (err)
1683 			break;
1684 
1685 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1686 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1687 				int matched_type;
1688 				err = got_object_get_type(&matched_type, repo,
1689 				    &qid->id);
1690 				if (err)
1691 					goto done;
1692 				if (matched_type != obj_type)
1693 					continue;
1694 			}
1695 			if (*unique_id == NULL) {
1696 				*unique_id = got_object_id_dup(&qid->id);
1697 				if (*unique_id == NULL) {
1698 					err = got_error_from_errno("malloc");
1699 					goto done;
1700 				}
1701 			} else {
1702 				if (got_object_id_cmp(*unique_id,
1703 				    &qid->id) == 0)
1704 					continue; /* packed multiple times */
1705 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1706 				goto done;
1707 			}
1708 		}
1709 	}
1710 done:
1711 	got_object_id_queue_free(&matched_ids);
1712 	if (err) {
1713 		free(*unique_id);
1714 		*unique_id = NULL;
1715 	}
1716 	return err;
1717 }
1718 
1719 static const struct got_error *
1720 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1721     const char *object_dir, const char *id_str_prefix, int obj_type,
1722     struct got_repository *repo)
1723 {
1724 	const struct got_error *err = NULL;
1725 	char *path, *id_str = NULL;
1726 	DIR *dir = NULL;
1727 	struct dirent *dent;
1728 	struct got_object_id id;
1729 
1730 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1731 		err = got_error_from_errno("asprintf");
1732 		goto done;
1733 	}
1734 
1735 	dir = opendir(path);
1736 	if (dir == NULL) {
1737 		if (errno == ENOENT) {
1738 			err = NULL;
1739 			goto done;
1740 		}
1741 		err = got_error_from_errno2("opendir", path);
1742 		goto done;
1743 	}
1744 	while ((dent = readdir(dir)) != NULL) {
1745 		int cmp;
1746 
1747 		free(id_str);
1748 		id_str = NULL;
1749 
1750 		if (strcmp(dent->d_name, ".") == 0 ||
1751 		    strcmp(dent->d_name, "..") == 0)
1752 			continue;
1753 
1754 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1755 			err = got_error_from_errno("asprintf");
1756 			goto done;
1757 		}
1758 
1759 		if (!got_parse_object_id(&id, id_str, repo->algo))
1760 			continue;
1761 
1762 		/*
1763 		 * Directory entries do not necessarily appear in
1764 		 * sorted order, so we must iterate over all of them.
1765 		 */
1766 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1767 		if (cmp != 0)
1768 			continue;
1769 
1770 		if (*unique_id == NULL) {
1771 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1772 				int matched_type;
1773 				err = got_object_get_type(&matched_type, repo,
1774 				    &id);
1775 				if (err)
1776 					goto done;
1777 				if (matched_type != obj_type)
1778 					continue;
1779 			}
1780 			*unique_id = got_object_id_dup(&id);
1781 			if (*unique_id == NULL) {
1782 				err = got_error_from_errno("got_object_id_dup");
1783 				goto done;
1784 			}
1785 		} else {
1786 			if (got_object_id_cmp(*unique_id, &id) == 0)
1787 				continue; /* both packed and loose */
1788 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1789 			goto done;
1790 		}
1791 	}
1792 done:
1793 	if (dir && closedir(dir) != 0 && err == NULL)
1794 		err = got_error_from_errno("closedir");
1795 	if (err) {
1796 		free(*unique_id);
1797 		*unique_id = NULL;
1798 	}
1799 	free(id_str);
1800 	free(path);
1801 	return err;
1802 }
1803 
1804 const struct got_error *
1805 got_repo_match_object_id_prefix(struct got_object_id **id,
1806     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1807 {
1808 	const struct got_error *err = NULL;
1809 	char *path_objects = NULL, *object_dir = NULL;
1810 	size_t len;
1811 	int i;
1812 
1813 	*id = NULL;
1814 
1815 	path_objects = got_repo_get_path_objects(repo);
1816 
1817 	len = strlen(id_str_prefix);
1818 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1819 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1820 		goto done;
1821 	}
1822 
1823 	for (i = 0; i < len; i++) {
1824 		if (isxdigit((unsigned char)id_str_prefix[i]))
1825 			continue;
1826 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1827 		goto done;
1828 	}
1829 
1830 	if (len >= 2) {
1831 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1832 		if (err)
1833 			goto done;
1834 		object_dir = strndup(id_str_prefix, 2);
1835 		if (object_dir == NULL) {
1836 			err = got_error_from_errno("strdup");
1837 			goto done;
1838 		}
1839 		err = match_loose_object(id, path_objects, object_dir,
1840 		    id_str_prefix, obj_type, repo);
1841 	} else if (len == 1) {
1842 		int i;
1843 		for (i = 0; i < 0xf; i++) {
1844 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1845 			    == -1) {
1846 				err = got_error_from_errno("asprintf");
1847 				goto done;
1848 			}
1849 			err = match_packed_object(id, repo, object_dir,
1850 			    obj_type);
1851 			if (err)
1852 				goto done;
1853 			err = match_loose_object(id, path_objects, object_dir,
1854 			    id_str_prefix, obj_type, repo);
1855 			if (err)
1856 				goto done;
1857 		}
1858 	} else {
1859 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1860 		goto done;
1861 	}
1862 done:
1863 	free(path_objects);
1864 	free(object_dir);
1865 	if (err) {
1866 		free(*id);
1867 		*id = NULL;
1868 	} else if (*id == NULL) {
1869 		switch (obj_type) {
1870 		case GOT_OBJ_TYPE_BLOB:
1871 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1872 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1873 			break;
1874 		case GOT_OBJ_TYPE_TREE:
1875 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1876 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1877 			break;
1878 		case GOT_OBJ_TYPE_COMMIT:
1879 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1880 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1881 			break;
1882 		case GOT_OBJ_TYPE_TAG:
1883 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1884 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1885 			break;
1886 		default:
1887 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1888 			break;
1889 		}
1890 	}
1891 
1892 	return err;
1893 }
1894 
1895 const struct got_error *
1896 got_repo_match_object_id(struct got_object_id **id, char **label,
1897     const char *id_str, int obj_type, struct got_reflist_head *refs,
1898     struct got_repository *repo)
1899 {
1900 	const struct got_error *err;
1901 	struct got_tag_object *tag;
1902 	struct got_reference *ref = NULL;
1903 
1904 	*id = NULL;
1905 	if (label)
1906 		*label = NULL;
1907 
1908 	if (refs) {
1909 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1910 		    refs, repo);
1911 		if (err == NULL) {
1912 			*id = got_object_id_dup(
1913 			    got_object_tag_get_object_id(tag));
1914 			if (*id == NULL)
1915 				err = got_error_from_errno("got_object_id_dup");
1916 			else if (label && asprintf(label, "refs/tags/%s",
1917 			    got_object_tag_get_name(tag)) == -1) {
1918 				err = got_error_from_errno("asprintf");
1919 				free(*id);
1920 				*id = NULL;
1921 			}
1922 			got_object_tag_close(tag);
1923 			return err;
1924 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1925 		    err->code != GOT_ERR_NO_OBJ)
1926 			return err;
1927 	}
1928 
1929 	err = got_ref_open(&ref, repo, id_str, 0);
1930 	if (err == NULL) {
1931 		err = got_ref_resolve(id, repo, ref);
1932 		if (err)
1933 			goto done;
1934 		if (label) {
1935 			*label = strdup(got_ref_get_name(ref));
1936 			if (*label == NULL) {
1937 				err = got_error_from_errno("strdup");
1938 				goto done;
1939 			}
1940 		}
1941 	} else {
1942 		if (err->code != GOT_ERR_NOT_REF &&
1943 		    err->code != GOT_ERR_BAD_REF_NAME)
1944 			goto done;
1945 		err = got_repo_match_object_id_prefix(id, id_str,
1946 		    obj_type, repo);
1947 		if (err) {
1948 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1949 				err = got_error_not_ref(id_str);
1950 			goto done;
1951 		}
1952 		if (label) {
1953 			err = got_object_id_str(label, *id);
1954 			if (*label == NULL) {
1955 				err = got_error_from_errno("strdup");
1956 				goto done;
1957 			}
1958 		}
1959 	}
1960 done:
1961 	if (ref)
1962 		got_ref_close(ref);
1963 	return err;
1964 }
1965 
1966 const struct got_error *
1967 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1968     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1969 {
1970 	const struct got_error *err = NULL;
1971 	struct got_reflist_entry *re;
1972 	struct got_object_id *tag_id;
1973 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1974 
1975 	*tag = NULL;
1976 
1977 	TAILQ_FOREACH(re, refs, entry) {
1978 		const char *refname;
1979 		refname = got_ref_get_name(re->ref);
1980 		if (got_ref_is_symbolic(re->ref))
1981 			continue;
1982 		if (strncmp(refname, "refs/tags/", 10) != 0)
1983 			continue;
1984 		if (!name_is_absolute)
1985 			refname += strlen("refs/tags/");
1986 		if (strcmp(refname, name) != 0)
1987 			continue;
1988 		err = got_ref_resolve(&tag_id, repo, re->ref);
1989 		if (err)
1990 			break;
1991 		err = got_object_open_as_tag(tag, repo, tag_id);
1992 		free(tag_id);
1993 		if (err)
1994 			break;
1995 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1996 		    got_object_tag_get_object_type(*tag) == obj_type)
1997 			break;
1998 		got_object_tag_close(*tag);
1999 		*tag = NULL;
2000 	}
2001 
2002 	if (err == NULL && *tag == NULL)
2003 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2004 		    GOT_OBJ_LABEL_TAG, name);
2005 	return err;
2006 }
2007 
2008 static const struct got_error *
2009 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2010     const char *name, mode_t mode, struct got_object_id *blob_id)
2011 {
2012 	const struct got_error *err = NULL;
2013 
2014 	 *new_te = NULL;
2015 
2016 	*new_te = calloc(1, sizeof(**new_te));
2017 	if (*new_te == NULL)
2018 		return got_error_from_errno("calloc");
2019 
2020 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2021 	    sizeof((*new_te)->name)) {
2022 		err = got_error(GOT_ERR_NO_SPACE);
2023 		goto done;
2024 	}
2025 
2026 	if (S_ISLNK(mode)) {
2027 		(*new_te)->mode = S_IFLNK;
2028 	} else {
2029 		(*new_te)->mode = S_IFREG;
2030 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2031 	}
2032 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2033 done:
2034 	if (err && *new_te) {
2035 		free(*new_te);
2036 		*new_te = NULL;
2037 	}
2038 	return err;
2039 }
2040 
2041 static const struct got_error *
2042 import_file(struct got_tree_entry **new_te, struct dirent *de,
2043     const char *path, struct got_repository *repo)
2044 {
2045 	const struct got_error *err;
2046 	struct got_object_id *blob_id = NULL;
2047 	char *filepath;
2048 	struct stat sb;
2049 
2050 	if (asprintf(&filepath, "%s%s%s", path,
2051 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2052 		return got_error_from_errno("asprintf");
2053 
2054 	if (lstat(filepath, &sb) != 0) {
2055 		err = got_error_from_errno2("lstat", path);
2056 		goto done;
2057 	}
2058 
2059 	err = got_object_blob_create(&blob_id, filepath, repo);
2060 	if (err)
2061 		goto done;
2062 
2063 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2064 	    blob_id);
2065 done:
2066 	free(filepath);
2067 	if (err)
2068 		free(blob_id);
2069 	return err;
2070 }
2071 
2072 static const struct got_error *
2073 insert_tree_entry(struct got_tree_entry *new_te,
2074     struct got_pathlist_head *paths)
2075 {
2076 	const struct got_error *err = NULL;
2077 	struct got_pathlist_entry *new_pe;
2078 
2079 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2080 	if (err)
2081 		return err;
2082 	if (new_pe == NULL)
2083 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2084 	return NULL;
2085 }
2086 
2087 static const struct got_error *write_tree(struct got_object_id **,
2088     const char *, struct got_pathlist_head *, struct got_repository *,
2089     got_repo_import_cb progress_cb, void *progress_arg);
2090 
2091 static const struct got_error *
2092 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2093     const char *path, struct got_pathlist_head *ignores,
2094     struct got_repository *repo,
2095     got_repo_import_cb progress_cb, void *progress_arg)
2096 {
2097 	const struct got_error *err;
2098 	struct got_object_id *id = NULL;
2099 	char *subdirpath;
2100 
2101 	if (asprintf(&subdirpath, "%s%s%s", path,
2102 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2103 		return got_error_from_errno("asprintf");
2104 
2105 	(*new_te) = calloc(1, sizeof(**new_te));
2106 	if (*new_te == NULL)
2107 		return got_error_from_errno("calloc");
2108 	(*new_te)->mode = S_IFDIR;
2109 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2110 	    sizeof((*new_te)->name)) {
2111 		err = got_error(GOT_ERR_NO_SPACE);
2112 		goto done;
2113 	}
2114 	err = write_tree(&id, subdirpath, ignores,  repo,
2115 	    progress_cb, progress_arg);
2116 	if (err)
2117 		goto done;
2118 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2119 
2120 done:
2121 	free(id);
2122 	free(subdirpath);
2123 	if (err) {
2124 		free(*new_te);
2125 		*new_te = NULL;
2126 	}
2127 	return err;
2128 }
2129 
2130 static const struct got_error *
2131 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2132     struct got_pathlist_head *ignores, struct got_repository *repo,
2133     got_repo_import_cb progress_cb, void *progress_arg)
2134 {
2135 	const struct got_error *err = NULL;
2136 	DIR *dir;
2137 	struct dirent *de;
2138 	int nentries;
2139 	struct got_tree_entry *new_te = NULL;
2140 	struct got_pathlist_head paths;
2141 	struct got_pathlist_entry *pe;
2142 
2143 	*new_tree_id = NULL;
2144 
2145 	TAILQ_INIT(&paths);
2146 
2147 	dir = opendir(path_dir);
2148 	if (dir == NULL) {
2149 		err = got_error_from_errno2("opendir", path_dir);
2150 		goto done;
2151 	}
2152 
2153 	nentries = 0;
2154 	while ((de = readdir(dir)) != NULL) {
2155 		int ignore = 0;
2156 		int type;
2157 
2158 		if (strcmp(de->d_name, ".") == 0 ||
2159 		    strcmp(de->d_name, "..") == 0)
2160 			continue;
2161 
2162 		err = got_path_dirent_type(&type, path_dir, de);
2163 		if (err)
2164 			goto done;
2165 
2166 		TAILQ_FOREACH(pe, ignores, entry) {
2167 			if (type == DT_DIR && pe->path_len > 0 &&
2168 			    pe->path[pe->path_len - 1] == '/') {
2169 				char stripped[PATH_MAX];
2170 
2171 				if (strlcpy(stripped, pe->path,
2172 				    sizeof(stripped)) >= sizeof(stripped)) {
2173 					err = got_error(GOT_ERR_NO_SPACE);
2174 					goto done;
2175 				}
2176 				got_path_strip_trailing_slashes(stripped);
2177 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2178 					ignore = 1;
2179 					break;
2180 				}
2181 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2182 				ignore = 1;
2183 				break;
2184 			}
2185 		}
2186 		if (ignore)
2187 			continue;
2188 
2189 		if (type == DT_DIR) {
2190 			err = import_subdir(&new_te, de, path_dir,
2191 			    ignores, repo, progress_cb, progress_arg);
2192 			if (err) {
2193 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2194 					goto done;
2195 				err = NULL;
2196 				continue;
2197 			}
2198 		} else if (type == DT_REG || type == DT_LNK) {
2199 			err = import_file(&new_te, de, path_dir, repo);
2200 			if (err)
2201 				goto done;
2202 		} else
2203 			continue;
2204 
2205 		err = insert_tree_entry(new_te, &paths);
2206 		if (err)
2207 			goto done;
2208 		nentries++;
2209 	}
2210 
2211 	if (TAILQ_EMPTY(&paths)) {
2212 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2213 		    "cannot create tree without any entries");
2214 		goto done;
2215 	}
2216 
2217 	TAILQ_FOREACH(pe, &paths, entry) {
2218 		struct got_tree_entry *te = pe->data;
2219 		char *path;
2220 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2221 			continue;
2222 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2223 			err = got_error_from_errno("asprintf");
2224 			goto done;
2225 		}
2226 		err = (*progress_cb)(progress_arg, path);
2227 		free(path);
2228 		if (err)
2229 			goto done;
2230 	}
2231 
2232 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2233 done:
2234 	if (dir)
2235 		closedir(dir);
2236 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2237 	return err;
2238 }
2239 
2240 const struct got_error *
2241 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2242     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2243     struct got_repository *repo, got_repo_import_cb progress_cb,
2244     void *progress_arg)
2245 {
2246 	const struct got_error *err;
2247 	struct got_object_id *new_tree_id;
2248 
2249 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2250 	    progress_cb, progress_arg);
2251 	if (err)
2252 		return err;
2253 
2254 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2255 	    author, time(NULL), author, time(NULL), logmsg, repo);
2256 	free(new_tree_id);
2257 	return err;
2258 }
2259 
2260 const struct got_error *
2261 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2262     struct got_repository *repo)
2263 {
2264 	const struct got_error *err = NULL;
2265 	char *path_objects = NULL, *path = NULL;
2266 	DIR *dir = NULL;
2267 	struct got_object_id id;
2268 	int i;
2269 
2270 	*nobjects = 0;
2271 	*ondisk_size = 0;
2272 
2273 	path_objects = got_repo_get_path_objects(repo);
2274 	if (path_objects == NULL)
2275 		return got_error_from_errno("got_repo_get_path_objects");
2276 
2277 	for (i = 0; i <= 0xff; i++) {
2278 		struct dirent *dent;
2279 
2280 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2281 			err = got_error_from_errno("asprintf");
2282 			break;
2283 		}
2284 
2285 		dir = opendir(path);
2286 		if (dir == NULL) {
2287 			if (errno == ENOENT) {
2288 				err = NULL;
2289 				continue;
2290 			}
2291 			err = got_error_from_errno2("opendir", path);
2292 			break;
2293 		}
2294 
2295 		while ((dent = readdir(dir)) != NULL) {
2296 			char *id_str;
2297 			int fd;
2298 			struct stat sb;
2299 
2300 			if (strcmp(dent->d_name, ".") == 0 ||
2301 			    strcmp(dent->d_name, "..") == 0)
2302 				continue;
2303 
2304 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2305 				err = got_error_from_errno("asprintf");
2306 				goto done;
2307 			}
2308 
2309 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2310 				free(id_str);
2311 				continue;
2312 			}
2313 			free(id_str);
2314 
2315 			err = got_object_open_loose_fd(&fd, &id, repo);
2316 			if (err)
2317 				goto done;
2318 
2319 			if (fstat(fd, &sb) == -1) {
2320 				err = got_error_from_errno("fstat");
2321 				close(fd);
2322 				goto done;
2323 			}
2324 			(*nobjects)++;
2325 			(*ondisk_size) += sb.st_size;
2326 
2327 			if (close(fd) == -1) {
2328 				err = got_error_from_errno("close");
2329 				goto done;
2330 			}
2331 		}
2332 
2333 		if (closedir(dir) != 0) {
2334 			err = got_error_from_errno("closedir");
2335 			goto done;
2336 		}
2337 		dir = NULL;
2338 
2339 		free(path);
2340 		path = NULL;
2341 	}
2342 done:
2343 	if (dir && closedir(dir) != 0 && err == NULL)
2344 		err = got_error_from_errno("closedir");
2345 
2346 	if (err) {
2347 		*nobjects = 0;
2348 		*ondisk_size = 0;
2349 	}
2350 	free(path_objects);
2351 	free(path);
2352 	return err;
2353 }
2354 
2355 const struct got_error *
2356 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2357     off_t *total_packsize, struct got_repository *repo)
2358 {
2359 	const struct got_error *err = NULL;
2360 	DIR *packdir = NULL;
2361 	struct dirent *dent;
2362 	struct got_packidx *packidx = NULL;
2363 	char *path_packidx;
2364 	char *path_packfile;
2365 	int packdir_fd;
2366 	struct stat sb;
2367 
2368 	*npackfiles = 0;
2369 	*nobjects = 0;
2370 	*total_packsize = 0;
2371 
2372 	packdir_fd = openat(got_repo_get_fd(repo),
2373 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2374 	if (packdir_fd == -1) {
2375 		return got_error_from_errno_fmt("openat: %s/%s",
2376 		    got_repo_get_path_git_dir(repo),
2377 		    GOT_OBJECTS_PACK_DIR);
2378 	}
2379 
2380 	packdir = fdopendir(packdir_fd);
2381 	if (packdir == NULL) {
2382 		err = got_error_from_errno("fdopendir");
2383 		goto done;
2384 	}
2385 
2386 	while ((dent = readdir(packdir)) != NULL) {
2387 		if (!got_repo_is_packidx_filename(dent->d_name,
2388 		    strlen(dent->d_name)))
2389 			continue;
2390 
2391 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2392 		    dent->d_name) == -1) {
2393 			err = got_error_from_errno("asprintf");
2394 			goto done;
2395 		}
2396 
2397 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2398 		    path_packidx, 0);
2399 		free(path_packidx);
2400 		if (err)
2401 			goto done;
2402 
2403 		if (fstat(packidx->fd, &sb) == -1)
2404 			goto done;
2405 		*total_packsize += sb.st_size;
2406 
2407 		err = got_packidx_get_packfile_path(&path_packfile,
2408 		    packidx->path_packidx);
2409 		if (err)
2410 			goto done;
2411 
2412 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2413 		    0) == -1) {
2414 			free(path_packfile);
2415 			goto done;
2416 		}
2417 		free(path_packfile);
2418 		*total_packsize += sb.st_size;
2419 
2420 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2421 
2422 		(*npackfiles)++;
2423 
2424 		got_packidx_close(packidx);
2425 		packidx = NULL;
2426 	}
2427 done:
2428 	if (packidx)
2429 		got_packidx_close(packidx);
2430 	if (packdir && closedir(packdir) != 0 && err == NULL)
2431 		err = got_error_from_errno("closedir");
2432 	if (err) {
2433 		*npackfiles = 0;
2434 		*nobjects = 0;
2435 		*total_packsize = 0;
2436 	}
2437 	return err;
2438 }
2439 
2440 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2441     got_packidx_bloom_filter_cmp);