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 void
911 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
912 {
913 	int i;
914 
915 	free(repo->name);
916 	repo->name = NULL;
917 	free(repo->fetch_url);
918 	repo->fetch_url = NULL;
919 	free(repo->send_url);
920 	repo->send_url = NULL;
921 	for (i = 0; i < repo->nfetch_branches; i++)
922 		free(repo->fetch_branches[i]);
923 	free(repo->fetch_branches);
924 	repo->fetch_branches = NULL;
925 	repo->nfetch_branches = 0;
926 	for (i = 0; i < repo->nsend_branches; i++)
927 		free(repo->send_branches[i]);
928 	free(repo->send_branches);
929 	repo->send_branches = NULL;
930 	repo->nsend_branches = 0;
931 }
932 
933 const struct got_error *
934 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
935     const char *input_path)
936 {
937 	const struct got_error *err = NULL;
938 	const char *repo_abspath = NULL;
939 	size_t repolen, len;
940 	char *canonpath, *path = NULL;
941 
942 	*in_repo_path = NULL;
943 
944 	canonpath = strdup(input_path);
945 	if (canonpath == NULL) {
946 		err = got_error_from_errno("strdup");
947 		goto done;
948 	}
949 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
950 	if (err)
951 		goto done;
952 
953 	repo_abspath = got_repo_get_path(repo);
954 
955 	if (canonpath[0] == '\0') {
956 		path = strdup(canonpath);
957 		if (path == NULL) {
958 			err = got_error_from_errno("strdup");
959 			goto done;
960 		}
961 	} else {
962 		path = realpath(canonpath, NULL);
963 		if (path == NULL) {
964 			if (errno != ENOENT) {
965 				err = got_error_from_errno2("realpath",
966 				    canonpath);
967 				goto done;
968 			}
969 			/*
970 			 * Path is not on disk.
971 			 * Assume it is already relative to repository root.
972 			 */
973 			path = strdup(canonpath);
974 			if (path == NULL) {
975 				err = got_error_from_errno("strdup");
976 				goto done;
977 			}
978 		}
979 
980 		repolen = strlen(repo_abspath);
981 		len = strlen(path);
982 
983 
984 		if (strcmp(path, repo_abspath) == 0) {
985 			free(path);
986 			path = strdup("");
987 			if (path == NULL) {
988 				err = got_error_from_errno("strdup");
989 				goto done;
990 			}
991 		} else if (len > repolen &&
992 		    got_path_is_child(path, repo_abspath, repolen)) {
993 			/* Matched an on-disk path inside repository. */
994 			if (got_repo_is_bare(repo)) {
995 				/*
996 				 * Matched an on-disk path inside repository
997 				 * database. Treat input as repository-relative.
998 				 */
999 				free(path);
1000 				path = canonpath;
1001 				canonpath = NULL;
1002 			} else {
1003 				char *child;
1004 				/* Strip common prefix with repository path. */
1005 				err = got_path_skip_common_ancestor(&child,
1006 				    repo_abspath, path);
1007 				if (err)
1008 					goto done;
1009 				free(path);
1010 				path = child;
1011 			}
1012 		} else {
1013 			/*
1014 			 * Matched unrelated on-disk path.
1015 			 * Treat input as repository-relative.
1016 			 */
1017 			free(path);
1018 			path = canonpath;
1019 			canonpath = NULL;
1020 		}
1021 	}
1022 
1023 	/* Make in-repository path absolute */
1024 	if (path[0] != '/') {
1025 		char *abspath;
1026 		if (asprintf(&abspath, "/%s", path) == -1) {
1027 			err = got_error_from_errno("asprintf");
1028 			goto done;
1029 		}
1030 		free(path);
1031 		path = abspath;
1032 	}
1033 
1034 done:
1035 	free(canonpath);
1036 	if (err)
1037 		free(path);
1038 	else
1039 		*in_repo_path = path;
1040 	return err;
1041 }
1042 
1043 static const struct got_error *
1044 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1045     const char *path_packidx)
1046 {
1047 	const struct got_error *err = NULL;
1048 	size_t i;
1049 
1050 	for (i = 0; i < repo->pack_cache_size; i++) {
1051 		if (repo->packidx_cache[i] == NULL)
1052 			break;
1053 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1054 		    path_packidx) == 0) {
1055 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1056 		}
1057 	}
1058 	if (i == repo->pack_cache_size) {
1059 		do {
1060 			i--;
1061 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1062 		    i == repo->pinned_packidx);
1063 		err = got_packidx_close(repo->packidx_cache[i]);
1064 		if (err)
1065 			return err;
1066 	}
1067 
1068 	repo->packidx_cache[i] = packidx;
1069 
1070 	return NULL;
1071 }
1072 
1073 int
1074 got_repo_is_packidx_filename(const char *name, size_t len)
1075 {
1076 	if (len != GOT_PACKIDX_NAMELEN)
1077 		return 0;
1078 
1079 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1080 		return 0;
1081 
1082 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1083 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1084 		return 0;
1085 
1086 	return 1;
1087 }
1088 
1089 static struct got_packidx_bloom_filter *
1090 get_packidx_bloom_filter(struct got_repository *repo,
1091     const char *path, size_t path_len)
1092 {
1093 	struct got_packidx_bloom_filter key;
1094 
1095 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1096 		return NULL; /* XXX */
1097 	key.path_len = path_len;
1098 
1099 	return RB_FIND(got_packidx_bloom_filter_tree,
1100 	    &repo->packidx_bloom_filters, &key);
1101 }
1102 
1103 int
1104 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1105     const char *path_packidx, struct got_object_id *id)
1106 {
1107 	struct got_packidx_bloom_filter *bf;
1108 
1109 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1110 	if (bf)
1111 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1112 
1113 	/* No bloom filter means this pack index must be searched. */
1114 	return 1;
1115 }
1116 
1117 static const struct got_error *
1118 add_packidx_bloom_filter(struct got_repository *repo,
1119     struct got_packidx *packidx, const char *path_packidx)
1120 {
1121 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1122 	struct got_packidx_bloom_filter *bf;
1123 	size_t len;
1124 
1125 	/*
1126 	 * Don't use bloom filters for very large pack index files.
1127 	 * Large pack files will contain a relatively large fraction
1128 	 * of our objects so we will likely need to visit them anyway.
1129 	 * The more objects a pack file contains the higher the probability
1130 	 * of a false-positive match from the bloom filter. And reading
1131 	 * all object IDs from a large pack index file can be expensive.
1132 	 */
1133 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1134 		return NULL;
1135 
1136 	/* Do we already have a filter for this pack index? */
1137 	if (get_packidx_bloom_filter(repo, path_packidx,
1138 	    strlen(path_packidx)) != NULL)
1139 		return NULL;
1140 
1141 	bf = calloc(1, sizeof(*bf));
1142 	if (bf == NULL)
1143 		return got_error_from_errno("calloc");
1144 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1145 	if (bf->bloom == NULL) {
1146 		free(bf);
1147 		return got_error_from_errno("calloc");
1148 	}
1149 
1150 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1151 	if (len >= sizeof(bf->path)) {
1152 		free(bf->bloom);
1153 		free(bf);
1154 		return got_error(GOT_ERR_NO_SPACE);
1155 	}
1156 	bf->path_len = len;
1157 
1158 	/* Minimum size supported by our bloom filter is 1000 entries. */
1159 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1160 	for (i = 0; i < nobjects; i++) {
1161 		struct got_packidx_object_id *id;
1162 		id = &packidx->hdr.sorted_ids[i];
1163 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1164 	}
1165 
1166 	RB_INSERT(got_packidx_bloom_filter_tree,
1167 	    &repo->packidx_bloom_filters, bf);
1168 	return NULL;
1169 }
1170 
1171 static void
1172 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1173 {
1174 	struct got_pathlist_entry *pe;
1175 
1176 	while (!TAILQ_EMPTY(packidx_paths)) {
1177 		pe = TAILQ_FIRST(packidx_paths);
1178 		TAILQ_REMOVE(packidx_paths, pe, entry);
1179 		free((char *)pe->path);
1180 		free(pe);
1181 	}
1182 }
1183 
1184 static const struct got_error *
1185 refresh_packidx_paths(struct got_repository *repo)
1186 {
1187 	const struct got_error *err = NULL;
1188 	char *objects_pack_dir = NULL;
1189 	struct stat sb;
1190 
1191 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1192 	if (objects_pack_dir == NULL)
1193 		return got_error_from_errno("got_repo_get_path_objects_pack");
1194 
1195 	if (stat(objects_pack_dir, &sb) == -1) {
1196 		if (errno != ENOENT) {
1197 			err = got_error_from_errno2("stat", objects_pack_dir);
1198 			goto done;
1199 		}
1200 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1201 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1202 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1203 		purge_packidx_paths(&repo->packidx_paths);
1204 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1205 		if (err)
1206 			goto done;
1207 	}
1208 done:
1209 	free(objects_pack_dir);
1210 	return err;
1211 }
1212 
1213 const struct got_error *
1214 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1215     struct got_repository *repo, struct got_object_id *id)
1216 {
1217 	const struct got_error *err;
1218 	struct got_pathlist_entry *pe;
1219 	size_t i;
1220 
1221 	/* Search pack index cache. */
1222 	for (i = 0; i < repo->pack_cache_size; i++) {
1223 		if (repo->packidx_cache[i] == NULL)
1224 			break;
1225 		if (!got_repo_check_packidx_bloom_filter(repo,
1226 		    repo->packidx_cache[i]->path_packidx, id))
1227 			continue; /* object will not be found in this index */
1228 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1229 		if (*idx != -1) {
1230 			*packidx = repo->packidx_cache[i];
1231 			/*
1232 			 * Move this cache entry to the front. Repeatedly
1233 			 * searching a wrong pack index can be expensive.
1234 			 */
1235 			if (i > 0) {
1236 				memmove(&repo->packidx_cache[1],
1237 				    &repo->packidx_cache[0],
1238 				    i * sizeof(repo->packidx_cache[0]));
1239 				repo->packidx_cache[0] = *packidx;
1240 				if (repo->pinned_packidx >= 0 &&
1241 				    repo->pinned_packidx < i)
1242 					repo->pinned_packidx++;
1243 				else if (repo->pinned_packidx == i)
1244 					repo->pinned_packidx = 0;
1245 			}
1246 			return NULL;
1247 		}
1248 	}
1249 	/* No luck. Search the filesystem. */
1250 
1251 	err = refresh_packidx_paths(repo);
1252 	if (err)
1253 		return err;
1254 
1255 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1256 		const char *path_packidx = pe->path;
1257 		int is_cached = 0;
1258 
1259 		if (!got_repo_check_packidx_bloom_filter(repo,
1260 		    pe->path, id))
1261 			continue; /* object will not be found in this index */
1262 
1263 		for (i = 0; i < repo->pack_cache_size; i++) {
1264 			if (repo->packidx_cache[i] == NULL)
1265 				break;
1266 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1267 			    path_packidx) == 0) {
1268 				is_cached = 1;
1269 				break;
1270 			}
1271 		}
1272 		if (is_cached)
1273 			continue; /* already searched */
1274 
1275 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1276 		    path_packidx, 0);
1277 		if (err)
1278 			goto done;
1279 
1280 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1281 		if (err)
1282 			goto done;
1283 
1284 		err = cache_packidx(repo, *packidx, path_packidx);
1285 		if (err)
1286 			goto done;
1287 
1288 		*idx = got_packidx_get_object_idx(*packidx, id);
1289 		if (*idx != -1) {
1290 			err = NULL; /* found the object */
1291 			goto done;
1292 		}
1293 	}
1294 
1295 	err = got_error_no_obj(id);
1296 done:
1297 	return err;
1298 }
1299 
1300 const struct got_error *
1301 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1302     struct got_repository *repo)
1303 {
1304 	const struct got_error *err = NULL;
1305 	DIR *packdir = NULL;
1306 	struct dirent *dent;
1307 	char *path_packidx = NULL;
1308 	int packdir_fd;
1309 	struct stat sb;
1310 
1311 	packdir_fd = openat(got_repo_get_fd(repo),
1312 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1313 	if (packdir_fd == -1) {
1314 		return got_error_from_errno_fmt("openat: %s/%s",
1315 		    got_repo_get_path_git_dir(repo),
1316 		    GOT_OBJECTS_PACK_DIR);
1317 	}
1318 
1319 	packdir = fdopendir(packdir_fd);
1320 	if (packdir == NULL) {
1321 		err = got_error_from_errno("fdopendir");
1322 		goto done;
1323 	}
1324 
1325 	if (fstat(packdir_fd, &sb) == -1) {
1326 		err = got_error_from_errno("fstat");
1327 		goto done;
1328 	}
1329 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1330 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1331 
1332 	while ((dent = readdir(packdir)) != NULL) {
1333 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1334 			continue;
1335 
1336 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1337 		    dent->d_name) == -1) {
1338 			err = got_error_from_errno("asprintf");
1339 			path_packidx = NULL;
1340 			break;
1341 		}
1342 
1343 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1344 		if (err)
1345 			break;
1346 	}
1347 done:
1348 	if (err)
1349 		free(path_packidx);
1350 	if (packdir && closedir(packdir) != 0 && err == NULL)
1351 		err = got_error_from_errno("closedir");
1352 	return err;
1353 }
1354 
1355 const struct got_error *
1356 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1357     struct got_repository *repo)
1358 {
1359 	const struct got_error *err;
1360 	size_t i;
1361 
1362 	*packidx = NULL;
1363 
1364 	/* Search pack index cache. */
1365 	for (i = 0; i < repo->pack_cache_size; i++) {
1366 		if (repo->packidx_cache[i] == NULL)
1367 			break;
1368 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1369 		    path_packidx) == 0) {
1370 			*packidx = repo->packidx_cache[i];
1371 			return NULL;
1372 		}
1373 	}
1374 	/* No luck. Search the filesystem. */
1375 
1376 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1377 	    path_packidx, 0);
1378 	if (err)
1379 		return err;
1380 
1381 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1382 	if (err)
1383 		goto done;
1384 
1385 	err = cache_packidx(repo, *packidx, path_packidx);
1386 done:
1387 	if (err) {
1388 		got_packidx_close(*packidx);
1389 		*packidx = NULL;
1390 	}
1391 	return err;
1392 }
1393 
1394 static const struct got_error *
1395 read_packfile_hdr(int fd, struct got_packidx *packidx)
1396 {
1397 	const struct got_error *err = NULL;
1398 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1399 	struct got_packfile_hdr hdr;
1400 	ssize_t n;
1401 
1402 	n = read(fd, &hdr, sizeof(hdr));
1403 	if (n < 0)
1404 		return got_error_from_errno("read");
1405 	if (n != sizeof(hdr))
1406 		return got_error(GOT_ERR_BAD_PACKFILE);
1407 
1408 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1409 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1410 	    be32toh(hdr.nobjects) != totobj)
1411 		err = got_error(GOT_ERR_BAD_PACKFILE);
1412 
1413 	return err;
1414 }
1415 
1416 static const struct got_error *
1417 open_packfile(int *fd, struct got_repository *repo,
1418     const char *relpath, struct got_packidx *packidx)
1419 {
1420 	const struct got_error *err = NULL;
1421 
1422 	*fd = openat(got_repo_get_fd(repo), relpath,
1423 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1424 	if (*fd == -1)
1425 		return got_error_from_errno_fmt("openat: %s/%s",
1426 		    got_repo_get_path_git_dir(repo), relpath);
1427 
1428 	if (packidx) {
1429 		err = read_packfile_hdr(*fd, packidx);
1430 		if (err) {
1431 			close(*fd);
1432 			*fd = -1;
1433 		}
1434 	}
1435 
1436 	return err;
1437 }
1438 
1439 const struct got_error *
1440 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1441     const char *path_packfile, struct got_packidx *packidx)
1442 {
1443 	const struct got_error *err = NULL;
1444 	struct got_pack *pack = NULL;
1445 	struct stat sb;
1446 	size_t i;
1447 
1448 	if (packp)
1449 		*packp = NULL;
1450 
1451 	for (i = 0; i < repo->pack_cache_size; i++) {
1452 		pack = &repo->packs[i];
1453 		if (pack->path_packfile == NULL)
1454 			break;
1455 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1456 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1457 	}
1458 
1459 	if (i == repo->pack_cache_size) {
1460 		struct got_pack tmp;
1461 		do {
1462 			i--;
1463 		} while (i > 0 && repo->pinned_pack >= 0 &&
1464 		    i == repo->pinned_pack);
1465 		err = got_pack_close(&repo->packs[i]);
1466 		if (err)
1467 			return err;
1468 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1469 			return got_error_from_errno("ftruncate");
1470 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1471 			return got_error_from_errno("ftruncate");
1472 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1473 		memcpy(&repo->packs[i], &repo->packs[0],
1474 		    sizeof(repo->packs[i]));
1475 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1476 		if (repo->pinned_pack == 0)
1477 			repo->pinned_pack = i;
1478 		else if (repo->pinned_pack == i)
1479 			repo->pinned_pack = 0;
1480 		i = 0;
1481 	}
1482 
1483 	pack = &repo->packs[i];
1484 
1485 	pack->path_packfile = strdup(path_packfile);
1486 	if (pack->path_packfile == NULL) {
1487 		err = got_error_from_errno("strdup");
1488 		goto done;
1489 	}
1490 
1491 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1492 	if (err)
1493 		goto done;
1494 
1495 	if (fstat(pack->fd, &sb) != 0) {
1496 		err = got_error_from_errno("fstat");
1497 		goto done;
1498 	}
1499 	pack->filesize = sb.st_size;
1500 
1501 	pack->privsep_child = NULL;
1502 
1503 	err = got_delta_cache_alloc(&pack->delta_cache);
1504 	if (err)
1505 		goto done;
1506 
1507 #ifndef GOT_PACK_NO_MMAP
1508 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1509 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1510 		    pack->fd, 0);
1511 		if (pack->map == MAP_FAILED) {
1512 			if (errno != ENOMEM) {
1513 				err = got_error_from_errno("mmap");
1514 				goto done;
1515 			}
1516 			pack->map = NULL; /* fall back to read(2) */
1517 		}
1518 	}
1519 #endif
1520 done:
1521 	if (err) {
1522 		if (pack)
1523 			got_pack_close(pack);
1524 	} else if (packp)
1525 		*packp = pack;
1526 	return err;
1527 }
1528 
1529 struct got_pack *
1530 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1531 {
1532 	struct got_pack *pack = NULL;
1533 	size_t i;
1534 
1535 	for (i = 0; i < repo->pack_cache_size; i++) {
1536 		pack = &repo->packs[i];
1537 		if (pack->path_packfile == NULL)
1538 			break;
1539 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1540 			return pack;
1541 	}
1542 
1543 	return NULL;
1544 }
1545 
1546 const struct got_error *
1547 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1548     struct got_pack *pack)
1549 {
1550 	size_t i;
1551 	int pinned_pack = -1, pinned_packidx = -1;
1552 
1553 	for (i = 0; i < repo->pack_cache_size; i++) {
1554 		if (repo->packidx_cache[i] &&
1555 		    strcmp(repo->packidx_cache[i]->path_packidx,
1556 		    packidx->path_packidx) == 0)
1557 			pinned_packidx = i;
1558 		if (repo->packs[i].path_packfile &&
1559 		    strcmp(repo->packs[i].path_packfile,
1560 		    pack->path_packfile) == 0)
1561 			pinned_pack = i;
1562 	}
1563 
1564 	if (pinned_packidx == -1 || pinned_pack == -1)
1565 		return got_error(GOT_ERR_PIN_PACK);
1566 
1567 	repo->pinned_pack = pinned_pack;
1568 	repo->pinned_packidx = pinned_packidx;
1569 	if (repo->packs[pinned_pack].privsep_child)
1570 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1571 	return NULL;
1572 }
1573 
1574 struct got_pack *
1575 got_repo_get_pinned_pack(struct got_repository *repo)
1576 {
1577 	if (repo->pinned_pack >= 0 &&
1578 	    repo->pinned_pack < repo->pack_cache_size)
1579 		return &repo->packs[repo->pinned_pack];
1580 
1581 	return NULL;
1582 }
1583 
1584 void
1585 got_repo_unpin_pack(struct got_repository *repo)
1586 {
1587 	repo->pinned_packidx = -1;
1588 	repo->pinned_pack = -1;
1589 	repo->pinned_pid = 0;
1590 }
1591 
1592 const struct got_error *
1593 got_repo_init(const char *repo_path, const char *head_name)
1594 {
1595 	const struct got_error *err = NULL;
1596 	const char *dirnames[] = {
1597 		GOT_OBJECTS_DIR,
1598 		GOT_OBJECTS_PACK_DIR,
1599 		GOT_REFS_DIR,
1600 	};
1601 	const char *description_str = "Unnamed repository; "
1602 	    "edit this file 'description' to name the repository.";
1603 	const char *headref = "ref: refs/heads/";
1604 	const char *gitconfig_str = "[core]\n"
1605 	    "\trepositoryformatversion = 0\n"
1606 	    "\tfilemode = true\n"
1607 	    "\tbare = true\n";
1608 	char *headref_str, *path;
1609 	size_t i;
1610 
1611 	if (!got_path_dir_is_empty(repo_path))
1612 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1613 
1614 	for (i = 0; i < nitems(dirnames); i++) {
1615 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1616 			return got_error_from_errno("asprintf");
1617 		}
1618 		err = got_path_mkdir(path);
1619 		free(path);
1620 		if (err)
1621 			return err;
1622 	}
1623 
1624 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1625 		return got_error_from_errno("asprintf");
1626 	err = got_path_create_file(path, description_str);
1627 	free(path);
1628 	if (err)
1629 		return err;
1630 
1631 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1632 		return got_error_from_errno("asprintf");
1633 	if (asprintf(&headref_str, "%s%s", headref,
1634 	    head_name ? head_name : "main") == -1) {
1635 		free(path);
1636 		return got_error_from_errno("asprintf");
1637 	}
1638 	err = got_path_create_file(path, headref_str);
1639 	free(headref_str);
1640 	free(path);
1641 	if (err)
1642 		return err;
1643 
1644 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1645 		return got_error_from_errno("asprintf");
1646 	err = got_path_create_file(path, gitconfig_str);
1647 	free(path);
1648 	if (err)
1649 		return err;
1650 
1651 	return NULL;
1652 }
1653 
1654 static const struct got_error *
1655 match_packed_object(struct got_object_id **unique_id,
1656     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1657 {
1658 	const struct got_error *err = NULL;
1659 	struct got_object_id_queue matched_ids;
1660 	struct got_pathlist_entry *pe;
1661 
1662 	STAILQ_INIT(&matched_ids);
1663 
1664 	err = refresh_packidx_paths(repo);
1665 	if (err)
1666 		return err;
1667 
1668 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1669 		const char *path_packidx = pe->path;
1670 		struct got_packidx *packidx;
1671 		struct got_object_qid *qid;
1672 
1673 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1674 		    path_packidx, 0);
1675 		if (err)
1676 			break;
1677 
1678 		err = got_packidx_match_id_str_prefix(&matched_ids,
1679 		    packidx, id_str_prefix);
1680 		if (err) {
1681 			got_packidx_close(packidx);
1682 			break;
1683 		}
1684 		err = got_packidx_close(packidx);
1685 		if (err)
1686 			break;
1687 
1688 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1689 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1690 				int matched_type;
1691 				err = got_object_get_type(&matched_type, repo,
1692 				    &qid->id);
1693 				if (err)
1694 					goto done;
1695 				if (matched_type != obj_type)
1696 					continue;
1697 			}
1698 			if (*unique_id == NULL) {
1699 				*unique_id = got_object_id_dup(&qid->id);
1700 				if (*unique_id == NULL) {
1701 					err = got_error_from_errno("malloc");
1702 					goto done;
1703 				}
1704 			} else {
1705 				if (got_object_id_cmp(*unique_id,
1706 				    &qid->id) == 0)
1707 					continue; /* packed multiple times */
1708 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1709 				goto done;
1710 			}
1711 		}
1712 	}
1713 done:
1714 	got_object_id_queue_free(&matched_ids);
1715 	if (err) {
1716 		free(*unique_id);
1717 		*unique_id = NULL;
1718 	}
1719 	return err;
1720 }
1721 
1722 static const struct got_error *
1723 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1724     const char *object_dir, const char *id_str_prefix, int obj_type,
1725     struct got_repository *repo)
1726 {
1727 	const struct got_error *err = NULL;
1728 	char *path, *id_str = NULL;
1729 	DIR *dir = NULL;
1730 	struct dirent *dent;
1731 	struct got_object_id id;
1732 
1733 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1734 		err = got_error_from_errno("asprintf");
1735 		goto done;
1736 	}
1737 
1738 	dir = opendir(path);
1739 	if (dir == NULL) {
1740 		if (errno == ENOENT) {
1741 			err = NULL;
1742 			goto done;
1743 		}
1744 		err = got_error_from_errno2("opendir", path);
1745 		goto done;
1746 	}
1747 	while ((dent = readdir(dir)) != NULL) {
1748 		int cmp;
1749 
1750 		free(id_str);
1751 		id_str = NULL;
1752 
1753 		if (strcmp(dent->d_name, ".") == 0 ||
1754 		    strcmp(dent->d_name, "..") == 0)
1755 			continue;
1756 
1757 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1758 			err = got_error_from_errno("asprintf");
1759 			goto done;
1760 		}
1761 
1762 		if (!got_parse_object_id(&id, id_str, repo->algo))
1763 			continue;
1764 
1765 		/*
1766 		 * Directory entries do not necessarily appear in
1767 		 * sorted order, so we must iterate over all of them.
1768 		 */
1769 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1770 		if (cmp != 0)
1771 			continue;
1772 
1773 		if (*unique_id == NULL) {
1774 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1775 				int matched_type;
1776 				err = got_object_get_type(&matched_type, repo,
1777 				    &id);
1778 				if (err)
1779 					goto done;
1780 				if (matched_type != obj_type)
1781 					continue;
1782 			}
1783 			*unique_id = got_object_id_dup(&id);
1784 			if (*unique_id == NULL) {
1785 				err = got_error_from_errno("got_object_id_dup");
1786 				goto done;
1787 			}
1788 		} else {
1789 			if (got_object_id_cmp(*unique_id, &id) == 0)
1790 				continue; /* both packed and loose */
1791 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1792 			goto done;
1793 		}
1794 	}
1795 done:
1796 	if (dir && closedir(dir) != 0 && err == NULL)
1797 		err = got_error_from_errno("closedir");
1798 	if (err) {
1799 		free(*unique_id);
1800 		*unique_id = NULL;
1801 	}
1802 	free(id_str);
1803 	free(path);
1804 	return err;
1805 }
1806 
1807 const struct got_error *
1808 got_repo_match_object_id_prefix(struct got_object_id **id,
1809     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1810 {
1811 	const struct got_error *err = NULL;
1812 	char *path_objects = NULL, *object_dir = NULL;
1813 	size_t len;
1814 	int i;
1815 
1816 	*id = NULL;
1817 
1818 	path_objects = got_repo_get_path_objects(repo);
1819 
1820 	len = strlen(id_str_prefix);
1821 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1822 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1823 		goto done;
1824 	}
1825 
1826 	for (i = 0; i < len; i++) {
1827 		if (isxdigit((unsigned char)id_str_prefix[i]))
1828 			continue;
1829 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1830 		goto done;
1831 	}
1832 
1833 	if (len >= 2) {
1834 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1835 		if (err)
1836 			goto done;
1837 		object_dir = strndup(id_str_prefix, 2);
1838 		if (object_dir == NULL) {
1839 			err = got_error_from_errno("strdup");
1840 			goto done;
1841 		}
1842 		err = match_loose_object(id, path_objects, object_dir,
1843 		    id_str_prefix, obj_type, repo);
1844 	} else if (len == 1) {
1845 		int i;
1846 		for (i = 0; i < 0xf; i++) {
1847 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1848 			    == -1) {
1849 				err = got_error_from_errno("asprintf");
1850 				goto done;
1851 			}
1852 			err = match_packed_object(id, repo, object_dir,
1853 			    obj_type);
1854 			if (err)
1855 				goto done;
1856 			err = match_loose_object(id, path_objects, object_dir,
1857 			    id_str_prefix, obj_type, repo);
1858 			if (err)
1859 				goto done;
1860 		}
1861 	} else {
1862 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1863 		goto done;
1864 	}
1865 done:
1866 	free(path_objects);
1867 	free(object_dir);
1868 	if (err) {
1869 		free(*id);
1870 		*id = NULL;
1871 	} else if (*id == NULL) {
1872 		switch (obj_type) {
1873 		case GOT_OBJ_TYPE_BLOB:
1874 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1875 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1876 			break;
1877 		case GOT_OBJ_TYPE_TREE:
1878 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1879 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1880 			break;
1881 		case GOT_OBJ_TYPE_COMMIT:
1882 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1883 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1884 			break;
1885 		case GOT_OBJ_TYPE_TAG:
1886 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1887 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1888 			break;
1889 		default:
1890 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1891 			break;
1892 		}
1893 	}
1894 
1895 	return err;
1896 }
1897 
1898 const struct got_error *
1899 got_repo_match_object_id(struct got_object_id **id, char **label,
1900     const char *id_str, int obj_type, struct got_reflist_head *refs,
1901     struct got_repository *repo)
1902 {
1903 	const struct got_error *err;
1904 	struct got_tag_object *tag;
1905 	struct got_reference *ref = NULL;
1906 
1907 	*id = NULL;
1908 	if (label)
1909 		*label = NULL;
1910 
1911 	if (refs) {
1912 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1913 		    refs, repo);
1914 		if (err == NULL) {
1915 			*id = got_object_id_dup(
1916 			    got_object_tag_get_object_id(tag));
1917 			if (*id == NULL)
1918 				err = got_error_from_errno("got_object_id_dup");
1919 			else if (label && asprintf(label, "refs/tags/%s",
1920 			    got_object_tag_get_name(tag)) == -1) {
1921 				err = got_error_from_errno("asprintf");
1922 				free(*id);
1923 				*id = NULL;
1924 			}
1925 			got_object_tag_close(tag);
1926 			return err;
1927 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1928 		    err->code != GOT_ERR_NO_OBJ)
1929 			return err;
1930 	}
1931 
1932 	err = got_ref_open(&ref, repo, id_str, 0);
1933 	if (err == NULL) {
1934 		err = got_ref_resolve(id, repo, ref);
1935 		if (err)
1936 			goto done;
1937 		if (label) {
1938 			*label = strdup(got_ref_get_name(ref));
1939 			if (*label == NULL) {
1940 				err = got_error_from_errno("strdup");
1941 				goto done;
1942 			}
1943 		}
1944 	} else {
1945 		if (err->code != GOT_ERR_NOT_REF &&
1946 		    err->code != GOT_ERR_BAD_REF_NAME)
1947 			goto done;
1948 		err = got_repo_match_object_id_prefix(id, id_str,
1949 		    obj_type, repo);
1950 		if (err) {
1951 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1952 				err = got_error_not_ref(id_str);
1953 			goto done;
1954 		}
1955 		if (label) {
1956 			err = got_object_id_str(label, *id);
1957 			if (*label == NULL) {
1958 				err = got_error_from_errno("strdup");
1959 				goto done;
1960 			}
1961 		}
1962 	}
1963 done:
1964 	if (ref)
1965 		got_ref_close(ref);
1966 	return err;
1967 }
1968 
1969 const struct got_error *
1970 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1971     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1972 {
1973 	const struct got_error *err = NULL;
1974 	struct got_reflist_entry *re;
1975 	struct got_object_id *tag_id;
1976 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1977 
1978 	*tag = NULL;
1979 
1980 	TAILQ_FOREACH(re, refs, entry) {
1981 		const char *refname;
1982 		refname = got_ref_get_name(re->ref);
1983 		if (got_ref_is_symbolic(re->ref))
1984 			continue;
1985 		if (strncmp(refname, "refs/tags/", 10) != 0)
1986 			continue;
1987 		if (!name_is_absolute)
1988 			refname += strlen("refs/tags/");
1989 		if (strcmp(refname, name) != 0)
1990 			continue;
1991 		err = got_ref_resolve(&tag_id, repo, re->ref);
1992 		if (err)
1993 			break;
1994 		err = got_object_open_as_tag(tag, repo, tag_id);
1995 		free(tag_id);
1996 		if (err)
1997 			break;
1998 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1999 		    got_object_tag_get_object_type(*tag) == obj_type)
2000 			break;
2001 		got_object_tag_close(*tag);
2002 		*tag = NULL;
2003 	}
2004 
2005 	if (err == NULL && *tag == NULL)
2006 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2007 		    GOT_OBJ_LABEL_TAG, name);
2008 	return err;
2009 }
2010 
2011 static const struct got_error *
2012 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2013     const char *name, mode_t mode, struct got_object_id *blob_id)
2014 {
2015 	const struct got_error *err = NULL;
2016 
2017 	 *new_te = NULL;
2018 
2019 	*new_te = calloc(1, sizeof(**new_te));
2020 	if (*new_te == NULL)
2021 		return got_error_from_errno("calloc");
2022 
2023 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2024 	    sizeof((*new_te)->name)) {
2025 		err = got_error(GOT_ERR_NO_SPACE);
2026 		goto done;
2027 	}
2028 
2029 	if (S_ISLNK(mode)) {
2030 		(*new_te)->mode = S_IFLNK;
2031 	} else {
2032 		(*new_te)->mode = S_IFREG;
2033 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2034 	}
2035 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2036 done:
2037 	if (err && *new_te) {
2038 		free(*new_te);
2039 		*new_te = NULL;
2040 	}
2041 	return err;
2042 }
2043 
2044 static const struct got_error *
2045 import_file(struct got_tree_entry **new_te, struct dirent *de,
2046     const char *path, struct got_repository *repo)
2047 {
2048 	const struct got_error *err;
2049 	struct got_object_id *blob_id = NULL;
2050 	char *filepath;
2051 	struct stat sb;
2052 
2053 	if (asprintf(&filepath, "%s%s%s", path,
2054 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2055 		return got_error_from_errno("asprintf");
2056 
2057 	if (lstat(filepath, &sb) != 0) {
2058 		err = got_error_from_errno2("lstat", path);
2059 		goto done;
2060 	}
2061 
2062 	err = got_object_blob_create(&blob_id, filepath, repo);
2063 	if (err)
2064 		goto done;
2065 
2066 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2067 	    blob_id);
2068 done:
2069 	free(filepath);
2070 	if (err)
2071 		free(blob_id);
2072 	return err;
2073 }
2074 
2075 static const struct got_error *
2076 insert_tree_entry(struct got_tree_entry *new_te,
2077     struct got_pathlist_head *paths)
2078 {
2079 	const struct got_error *err = NULL;
2080 	struct got_pathlist_entry *new_pe;
2081 
2082 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2083 	if (err)
2084 		return err;
2085 	if (new_pe == NULL)
2086 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2087 	return NULL;
2088 }
2089 
2090 static const struct got_error *write_tree(struct got_object_id **,
2091     const char *, struct got_pathlist_head *, struct got_repository *,
2092     got_repo_import_cb progress_cb, void *progress_arg);
2093 
2094 static const struct got_error *
2095 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2096     const char *path, struct got_pathlist_head *ignores,
2097     struct got_repository *repo,
2098     got_repo_import_cb progress_cb, void *progress_arg)
2099 {
2100 	const struct got_error *err;
2101 	struct got_object_id *id = NULL;
2102 	char *subdirpath;
2103 
2104 	if (asprintf(&subdirpath, "%s%s%s", path,
2105 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2106 		return got_error_from_errno("asprintf");
2107 
2108 	(*new_te) = calloc(1, sizeof(**new_te));
2109 	if (*new_te == NULL)
2110 		return got_error_from_errno("calloc");
2111 	(*new_te)->mode = S_IFDIR;
2112 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2113 	    sizeof((*new_te)->name)) {
2114 		err = got_error(GOT_ERR_NO_SPACE);
2115 		goto done;
2116 	}
2117 	err = write_tree(&id, subdirpath, ignores,  repo,
2118 	    progress_cb, progress_arg);
2119 	if (err)
2120 		goto done;
2121 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2122 
2123 done:
2124 	free(id);
2125 	free(subdirpath);
2126 	if (err) {
2127 		free(*new_te);
2128 		*new_te = NULL;
2129 	}
2130 	return err;
2131 }
2132 
2133 static const struct got_error *
2134 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2135     struct got_pathlist_head *ignores, struct got_repository *repo,
2136     got_repo_import_cb progress_cb, void *progress_arg)
2137 {
2138 	const struct got_error *err = NULL;
2139 	DIR *dir;
2140 	struct dirent *de;
2141 	int nentries;
2142 	struct got_tree_entry *new_te = NULL;
2143 	struct got_pathlist_head paths;
2144 	struct got_pathlist_entry *pe;
2145 
2146 	*new_tree_id = NULL;
2147 
2148 	TAILQ_INIT(&paths);
2149 
2150 	dir = opendir(path_dir);
2151 	if (dir == NULL) {
2152 		err = got_error_from_errno2("opendir", path_dir);
2153 		goto done;
2154 	}
2155 
2156 	nentries = 0;
2157 	while ((de = readdir(dir)) != NULL) {
2158 		int ignore = 0;
2159 		int type;
2160 
2161 		if (strcmp(de->d_name, ".") == 0 ||
2162 		    strcmp(de->d_name, "..") == 0)
2163 			continue;
2164 
2165 		err = got_path_dirent_type(&type, path_dir, de);
2166 		if (err)
2167 			goto done;
2168 
2169 		TAILQ_FOREACH(pe, ignores, entry) {
2170 			if (type == DT_DIR && pe->path_len > 0 &&
2171 			    pe->path[pe->path_len - 1] == '/') {
2172 				char stripped[PATH_MAX];
2173 
2174 				if (strlcpy(stripped, pe->path,
2175 				    sizeof(stripped)) >= sizeof(stripped)) {
2176 					err = got_error(GOT_ERR_NO_SPACE);
2177 					goto done;
2178 				}
2179 				got_path_strip_trailing_slashes(stripped);
2180 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2181 					ignore = 1;
2182 					break;
2183 				}
2184 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2185 				ignore = 1;
2186 				break;
2187 			}
2188 		}
2189 		if (ignore)
2190 			continue;
2191 
2192 		if (type == DT_DIR) {
2193 			err = import_subdir(&new_te, de, path_dir,
2194 			    ignores, repo, progress_cb, progress_arg);
2195 			if (err) {
2196 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2197 					goto done;
2198 				err = NULL;
2199 				continue;
2200 			}
2201 		} else if (type == DT_REG || type == DT_LNK) {
2202 			err = import_file(&new_te, de, path_dir, repo);
2203 			if (err)
2204 				goto done;
2205 		} else
2206 			continue;
2207 
2208 		err = insert_tree_entry(new_te, &paths);
2209 		if (err)
2210 			goto done;
2211 		nentries++;
2212 	}
2213 
2214 	if (TAILQ_EMPTY(&paths)) {
2215 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2216 		    "cannot create tree without any entries");
2217 		goto done;
2218 	}
2219 
2220 	TAILQ_FOREACH(pe, &paths, entry) {
2221 		struct got_tree_entry *te = pe->data;
2222 		char *path;
2223 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2224 			continue;
2225 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2226 			err = got_error_from_errno("asprintf");
2227 			goto done;
2228 		}
2229 		err = (*progress_cb)(progress_arg, path);
2230 		free(path);
2231 		if (err)
2232 			goto done;
2233 	}
2234 
2235 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2236 done:
2237 	if (dir)
2238 		closedir(dir);
2239 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2240 	return err;
2241 }
2242 
2243 const struct got_error *
2244 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2245     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2246     struct got_repository *repo, got_repo_import_cb progress_cb,
2247     void *progress_arg)
2248 {
2249 	const struct got_error *err;
2250 	struct got_object_id *new_tree_id;
2251 
2252 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2253 	    progress_cb, progress_arg);
2254 	if (err)
2255 		return err;
2256 
2257 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2258 	    author, time(NULL), author, time(NULL), logmsg, repo);
2259 	free(new_tree_id);
2260 	return err;
2261 }
2262 
2263 const struct got_error *
2264 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2265     struct got_repository *repo)
2266 {
2267 	const struct got_error *err = NULL;
2268 	char *path_objects = NULL, *path = NULL;
2269 	DIR *dir = NULL;
2270 	struct got_object_id id;
2271 	int i;
2272 
2273 	*nobjects = 0;
2274 	*ondisk_size = 0;
2275 
2276 	path_objects = got_repo_get_path_objects(repo);
2277 	if (path_objects == NULL)
2278 		return got_error_from_errno("got_repo_get_path_objects");
2279 
2280 	for (i = 0; i <= 0xff; i++) {
2281 		struct dirent *dent;
2282 
2283 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2284 			err = got_error_from_errno("asprintf");
2285 			break;
2286 		}
2287 
2288 		dir = opendir(path);
2289 		if (dir == NULL) {
2290 			if (errno == ENOENT) {
2291 				err = NULL;
2292 				continue;
2293 			}
2294 			err = got_error_from_errno2("opendir", path);
2295 			break;
2296 		}
2297 
2298 		while ((dent = readdir(dir)) != NULL) {
2299 			char *id_str;
2300 			int fd;
2301 			struct stat sb;
2302 
2303 			if (strcmp(dent->d_name, ".") == 0 ||
2304 			    strcmp(dent->d_name, "..") == 0)
2305 				continue;
2306 
2307 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2308 				err = got_error_from_errno("asprintf");
2309 				goto done;
2310 			}
2311 
2312 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2313 				free(id_str);
2314 				continue;
2315 			}
2316 			free(id_str);
2317 
2318 			err = got_object_open_loose_fd(&fd, &id, repo);
2319 			if (err)
2320 				goto done;
2321 
2322 			if (fstat(fd, &sb) == -1) {
2323 				err = got_error_from_errno("fstat");
2324 				close(fd);
2325 				goto done;
2326 			}
2327 			(*nobjects)++;
2328 			(*ondisk_size) += sb.st_size;
2329 
2330 			if (close(fd) == -1) {
2331 				err = got_error_from_errno("close");
2332 				goto done;
2333 			}
2334 		}
2335 
2336 		if (closedir(dir) != 0) {
2337 			err = got_error_from_errno("closedir");
2338 			goto done;
2339 		}
2340 		dir = NULL;
2341 
2342 		free(path);
2343 		path = NULL;
2344 	}
2345 done:
2346 	if (dir && closedir(dir) != 0 && err == NULL)
2347 		err = got_error_from_errno("closedir");
2348 
2349 	if (err) {
2350 		*nobjects = 0;
2351 		*ondisk_size = 0;
2352 	}
2353 	free(path_objects);
2354 	free(path);
2355 	return err;
2356 }
2357 
2358 const struct got_error *
2359 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2360     off_t *total_packsize, struct got_repository *repo)
2361 {
2362 	const struct got_error *err = NULL;
2363 	DIR *packdir = NULL;
2364 	struct dirent *dent;
2365 	struct got_packidx *packidx = NULL;
2366 	char *path_packidx;
2367 	char *path_packfile;
2368 	int packdir_fd;
2369 	struct stat sb;
2370 
2371 	*npackfiles = 0;
2372 	*nobjects = 0;
2373 	*total_packsize = 0;
2374 
2375 	packdir_fd = openat(got_repo_get_fd(repo),
2376 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2377 	if (packdir_fd == -1) {
2378 		return got_error_from_errno_fmt("openat: %s/%s",
2379 		    got_repo_get_path_git_dir(repo),
2380 		    GOT_OBJECTS_PACK_DIR);
2381 	}
2382 
2383 	packdir = fdopendir(packdir_fd);
2384 	if (packdir == NULL) {
2385 		err = got_error_from_errno("fdopendir");
2386 		goto done;
2387 	}
2388 
2389 	while ((dent = readdir(packdir)) != NULL) {
2390 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2391 			continue;
2392 
2393 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2394 		    dent->d_name) == -1) {
2395 			err = got_error_from_errno("asprintf");
2396 			goto done;
2397 		}
2398 
2399 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2400 		    path_packidx, 0);
2401 		free(path_packidx);
2402 		if (err)
2403 			goto done;
2404 
2405 		if (fstat(packidx->fd, &sb) == -1)
2406 			goto done;
2407 		*total_packsize += sb.st_size;
2408 
2409 		err = got_packidx_get_packfile_path(&path_packfile,
2410 		    packidx->path_packidx);
2411 		if (err)
2412 			goto done;
2413 
2414 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2415 		    0) == -1) {
2416 			free(path_packfile);
2417 			goto done;
2418 		}
2419 		free(path_packfile);
2420 		*total_packsize += sb.st_size;
2421 
2422 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2423 
2424 		(*npackfiles)++;
2425 
2426 		got_packidx_close(packidx);
2427 		packidx = NULL;
2428 	}
2429 done:
2430 	if (packidx)
2431 		got_packidx_close(packidx);
2432 	if (packdir && closedir(packdir) != 0 && err == NULL)
2433 		err = got_error_from_errno("closedir");
2434 	if (err) {
2435 		*npackfiles = 0;
2436 		*nobjects = 0;
2437 		*total_packsize = 0;
2438 	}
2439 	return err;
2440 }
2441 
2442 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2443     got_packidx_bloom_filter_cmp);