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 		child_err = got_privsep_wait_for_child(
876 		    repo->privsep_children[i].pid);
877 		if (child_err && err == NULL)
878 			err = child_err;
879 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
880 		    err == NULL)
881 			err = got_error_from_errno("close");
882 	}
883 
884 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
885 	    err == NULL)
886 		err = got_error_from_errno("close");
887 
888 	if (repo->gotconfig)
889 		got_gotconfig_free(repo->gotconfig);
890 	free(repo->gitconfig_author_name);
891 	free(repo->gitconfig_author_email);
892 	for (i = 0; i < repo->ngitconfig_remotes; i++)
893 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
894 	free(repo->gitconfig_remotes);
895 	for (i = 0; i < repo->nextensions; i++) {
896 		free(repo->extnames[i]);
897 		free(repo->extvals[i]);
898 	}
899 	free(repo->extnames);
900 	free(repo->extvals);
901 
902 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
903 	free(repo);
904 
905 	return err;
906 }
907 
908 void
909 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
910 {
911 	int i;
912 
913 	free(repo->name);
914 	repo->name = NULL;
915 	free(repo->fetch_url);
916 	repo->fetch_url = NULL;
917 	free(repo->send_url);
918 	repo->send_url = NULL;
919 	for (i = 0; i < repo->nfetch_branches; i++)
920 		free(repo->fetch_branches[i]);
921 	free(repo->fetch_branches);
922 	repo->fetch_branches = NULL;
923 	repo->nfetch_branches = 0;
924 	for (i = 0; i < repo->nsend_branches; i++)
925 		free(repo->send_branches[i]);
926 	free(repo->send_branches);
927 	repo->send_branches = NULL;
928 	repo->nsend_branches = 0;
929 }
930 
931 const struct got_error *
932 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
933     const char *input_path)
934 {
935 	const struct got_error *err = NULL;
936 	const char *repo_abspath = NULL;
937 	size_t repolen, len;
938 	char *canonpath, *path = NULL;
939 
940 	*in_repo_path = NULL;
941 
942 	canonpath = strdup(input_path);
943 	if (canonpath == NULL) {
944 		err = got_error_from_errno("strdup");
945 		goto done;
946 	}
947 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
948 	if (err)
949 		goto done;
950 
951 	repo_abspath = got_repo_get_path(repo);
952 
953 	if (canonpath[0] == '\0') {
954 		path = strdup(canonpath);
955 		if (path == NULL) {
956 			err = got_error_from_errno("strdup");
957 			goto done;
958 		}
959 	} else {
960 		path = realpath(canonpath, NULL);
961 		if (path == NULL) {
962 			if (errno != ENOENT) {
963 				err = got_error_from_errno2("realpath",
964 				    canonpath);
965 				goto done;
966 			}
967 			/*
968 			 * Path is not on disk.
969 			 * Assume it is already relative to repository root.
970 			 */
971 			path = strdup(canonpath);
972 			if (path == NULL) {
973 				err = got_error_from_errno("strdup");
974 				goto done;
975 			}
976 		}
977 
978 		repolen = strlen(repo_abspath);
979 		len = strlen(path);
980 
981 
982 		if (strcmp(path, repo_abspath) == 0) {
983 			free(path);
984 			path = strdup("");
985 			if (path == NULL) {
986 				err = got_error_from_errno("strdup");
987 				goto done;
988 			}
989 		} else if (len > repolen &&
990 		    got_path_is_child(path, repo_abspath, repolen)) {
991 			/* Matched an on-disk path inside repository. */
992 			if (got_repo_is_bare(repo)) {
993 				/*
994 				 * Matched an on-disk path inside repository
995 				 * database. Treat input as repository-relative.
996 				 */
997 				free(path);
998 				path = canonpath;
999 				canonpath = NULL;
1000 			} else {
1001 				char *child;
1002 				/* Strip common prefix with repository path. */
1003 				err = got_path_skip_common_ancestor(&child,
1004 				    repo_abspath, path);
1005 				if (err)
1006 					goto done;
1007 				free(path);
1008 				path = child;
1009 			}
1010 		} else {
1011 			/*
1012 			 * Matched unrelated on-disk path.
1013 			 * Treat input as repository-relative.
1014 			 */
1015 			free(path);
1016 			path = canonpath;
1017 			canonpath = NULL;
1018 		}
1019 	}
1020 
1021 	/* Make in-repository path absolute */
1022 	if (path[0] != '/') {
1023 		char *abspath;
1024 		if (asprintf(&abspath, "/%s", path) == -1) {
1025 			err = got_error_from_errno("asprintf");
1026 			goto done;
1027 		}
1028 		free(path);
1029 		path = abspath;
1030 	}
1031 
1032 done:
1033 	free(canonpath);
1034 	if (err)
1035 		free(path);
1036 	else
1037 		*in_repo_path = path;
1038 	return err;
1039 }
1040 
1041 static const struct got_error *
1042 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1043     const char *path_packidx)
1044 {
1045 	const struct got_error *err = NULL;
1046 	size_t i;
1047 
1048 	for (i = 0; i < repo->pack_cache_size; i++) {
1049 		if (repo->packidx_cache[i] == NULL)
1050 			break;
1051 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1052 		    path_packidx) == 0) {
1053 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1054 		}
1055 	}
1056 	if (i == repo->pack_cache_size) {
1057 		do {
1058 			i--;
1059 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1060 		    i == repo->pinned_packidx);
1061 		err = got_packidx_close(repo->packidx_cache[i]);
1062 		if (err)
1063 			return err;
1064 	}
1065 
1066 	repo->packidx_cache[i] = packidx;
1067 
1068 	return NULL;
1069 }
1070 
1071 int
1072 got_repo_is_packidx_filename(const char *name, size_t len)
1073 {
1074 	if (len != GOT_PACKIDX_NAMELEN)
1075 		return 0;
1076 
1077 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1078 		return 0;
1079 
1080 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1081 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1082 		return 0;
1083 
1084 	return 1;
1085 }
1086 
1087 static struct got_packidx_bloom_filter *
1088 get_packidx_bloom_filter(struct got_repository *repo,
1089     const char *path, size_t path_len)
1090 {
1091 	struct got_packidx_bloom_filter key;
1092 
1093 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1094 		return NULL; /* XXX */
1095 	key.path_len = path_len;
1096 
1097 	return RB_FIND(got_packidx_bloom_filter_tree,
1098 	    &repo->packidx_bloom_filters, &key);
1099 }
1100 
1101 int
1102 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1103     const char *path_packidx, struct got_object_id *id)
1104 {
1105 	struct got_packidx_bloom_filter *bf;
1106 
1107 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1108 	if (bf)
1109 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1110 
1111 	/* No bloom filter means this pack index must be searched. */
1112 	return 1;
1113 }
1114 
1115 static const struct got_error *
1116 add_packidx_bloom_filter(struct got_repository *repo,
1117     struct got_packidx *packidx, const char *path_packidx)
1118 {
1119 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1120 	struct got_packidx_bloom_filter *bf;
1121 	size_t len;
1122 
1123 	/*
1124 	 * Don't use bloom filters for very large pack index files.
1125 	 * Large pack files will contain a relatively large fraction
1126 	 * of our objects so we will likely need to visit them anyway.
1127 	 * The more objects a pack file contains the higher the probability
1128 	 * of a false-positive match from the bloom filter. And reading
1129 	 * all object IDs from a large pack index file can be expensive.
1130 	 */
1131 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1132 		return NULL;
1133 
1134 	/* Do we already have a filter for this pack index? */
1135 	if (get_packidx_bloom_filter(repo, path_packidx,
1136 	    strlen(path_packidx)) != NULL)
1137 		return NULL;
1138 
1139 	bf = calloc(1, sizeof(*bf));
1140 	if (bf == NULL)
1141 		return got_error_from_errno("calloc");
1142 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1143 	if (bf->bloom == NULL) {
1144 		free(bf);
1145 		return got_error_from_errno("calloc");
1146 	}
1147 
1148 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1149 	if (len >= sizeof(bf->path)) {
1150 		free(bf->bloom);
1151 		free(bf);
1152 		return got_error(GOT_ERR_NO_SPACE);
1153 	}
1154 	bf->path_len = len;
1155 
1156 	/* Minimum size supported by our bloom filter is 1000 entries. */
1157 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1158 	for (i = 0; i < nobjects; i++) {
1159 		struct got_packidx_object_id *id;
1160 		id = &packidx->hdr.sorted_ids[i];
1161 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1162 	}
1163 
1164 	RB_INSERT(got_packidx_bloom_filter_tree,
1165 	    &repo->packidx_bloom_filters, bf);
1166 	return NULL;
1167 }
1168 
1169 static void
1170 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1171 {
1172 	struct got_pathlist_entry *pe;
1173 
1174 	while (!TAILQ_EMPTY(packidx_paths)) {
1175 		pe = TAILQ_FIRST(packidx_paths);
1176 		TAILQ_REMOVE(packidx_paths, pe, entry);
1177 		free((char *)pe->path);
1178 		free(pe);
1179 	}
1180 }
1181 
1182 static const struct got_error *
1183 refresh_packidx_paths(struct got_repository *repo)
1184 {
1185 	const struct got_error *err = NULL;
1186 	char *objects_pack_dir = NULL;
1187 	struct stat sb;
1188 
1189 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1190 	if (objects_pack_dir == NULL)
1191 		return got_error_from_errno("got_repo_get_path_objects_pack");
1192 
1193 	if (stat(objects_pack_dir, &sb) == -1) {
1194 		if (errno != ENOENT) {
1195 			err = got_error_from_errno2("stat", objects_pack_dir);
1196 			goto done;
1197 		}
1198 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1199 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1200 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1201 		purge_packidx_paths(&repo->packidx_paths);
1202 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1203 		if (err)
1204 			goto done;
1205 	}
1206 done:
1207 	free(objects_pack_dir);
1208 	return err;
1209 }
1210 
1211 const struct got_error *
1212 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1213     struct got_repository *repo, struct got_object_id *id)
1214 {
1215 	const struct got_error *err;
1216 	struct got_pathlist_entry *pe;
1217 	size_t i;
1218 
1219 	/* Search pack index cache. */
1220 	for (i = 0; i < repo->pack_cache_size; i++) {
1221 		if (repo->packidx_cache[i] == NULL)
1222 			break;
1223 		if (!got_repo_check_packidx_bloom_filter(repo,
1224 		    repo->packidx_cache[i]->path_packidx, id))
1225 			continue; /* object will not be found in this index */
1226 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1227 		if (*idx != -1) {
1228 			*packidx = repo->packidx_cache[i];
1229 			/*
1230 			 * Move this cache entry to the front. Repeatedly
1231 			 * searching a wrong pack index can be expensive.
1232 			 */
1233 			if (i > 0) {
1234 				memmove(&repo->packidx_cache[1],
1235 				    &repo->packidx_cache[0],
1236 				    i * sizeof(repo->packidx_cache[0]));
1237 				repo->packidx_cache[0] = *packidx;
1238 				if (repo->pinned_packidx >= 0 &&
1239 				    repo->pinned_packidx < i)
1240 					repo->pinned_packidx++;
1241 				else if (repo->pinned_packidx == i)
1242 					repo->pinned_packidx = 0;
1243 			}
1244 			return NULL;
1245 		}
1246 	}
1247 	/* No luck. Search the filesystem. */
1248 
1249 	err = refresh_packidx_paths(repo);
1250 	if (err)
1251 		return err;
1252 
1253 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1254 		const char *path_packidx = pe->path;
1255 		int is_cached = 0;
1256 
1257 		if (!got_repo_check_packidx_bloom_filter(repo,
1258 		    pe->path, id))
1259 			continue; /* object will not be found in this index */
1260 
1261 		for (i = 0; i < repo->pack_cache_size; i++) {
1262 			if (repo->packidx_cache[i] == NULL)
1263 				break;
1264 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1265 			    path_packidx) == 0) {
1266 				is_cached = 1;
1267 				break;
1268 			}
1269 		}
1270 		if (is_cached)
1271 			continue; /* already searched */
1272 
1273 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1274 		    path_packidx, 0);
1275 		if (err)
1276 			goto done;
1277 
1278 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1279 		if (err)
1280 			goto done;
1281 
1282 		err = cache_packidx(repo, *packidx, path_packidx);
1283 		if (err)
1284 			goto done;
1285 
1286 		*idx = got_packidx_get_object_idx(*packidx, id);
1287 		if (*idx != -1) {
1288 			err = NULL; /* found the object */
1289 			goto done;
1290 		}
1291 	}
1292 
1293 	err = got_error_no_obj(id);
1294 done:
1295 	return err;
1296 }
1297 
1298 const struct got_error *
1299 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1300     struct got_repository *repo)
1301 {
1302 	const struct got_error *err = NULL;
1303 	DIR *packdir = NULL;
1304 	struct dirent *dent;
1305 	char *path_packidx = NULL;
1306 	int packdir_fd;
1307 	struct stat sb;
1308 
1309 	packdir_fd = openat(got_repo_get_fd(repo),
1310 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1311 	if (packdir_fd == -1) {
1312 		return got_error_from_errno_fmt("openat: %s/%s",
1313 		    got_repo_get_path_git_dir(repo),
1314 		    GOT_OBJECTS_PACK_DIR);
1315 	}
1316 
1317 	packdir = fdopendir(packdir_fd);
1318 	if (packdir == NULL) {
1319 		err = got_error_from_errno("fdopendir");
1320 		goto done;
1321 	}
1322 
1323 	if (fstat(packdir_fd, &sb) == -1) {
1324 		err = got_error_from_errno("fstat");
1325 		goto done;
1326 	}
1327 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1328 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1329 
1330 	while ((dent = readdir(packdir)) != NULL) {
1331 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1332 			continue;
1333 
1334 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1335 		    dent->d_name) == -1) {
1336 			err = got_error_from_errno("asprintf");
1337 			path_packidx = NULL;
1338 			break;
1339 		}
1340 
1341 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1342 		if (err)
1343 			break;
1344 	}
1345 done:
1346 	if (err)
1347 		free(path_packidx);
1348 	if (packdir && closedir(packdir) != 0 && err == NULL)
1349 		err = got_error_from_errno("closedir");
1350 	return err;
1351 }
1352 
1353 const struct got_error *
1354 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1355     struct got_repository *repo)
1356 {
1357 	const struct got_error *err;
1358 	size_t i;
1359 
1360 	*packidx = NULL;
1361 
1362 	/* Search pack index cache. */
1363 	for (i = 0; i < repo->pack_cache_size; i++) {
1364 		if (repo->packidx_cache[i] == NULL)
1365 			break;
1366 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1367 		    path_packidx) == 0) {
1368 			*packidx = repo->packidx_cache[i];
1369 			return NULL;
1370 		}
1371 	}
1372 	/* No luck. Search the filesystem. */
1373 
1374 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1375 	    path_packidx, 0);
1376 	if (err)
1377 		return err;
1378 
1379 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1380 	if (err)
1381 		goto done;
1382 
1383 	err = cache_packidx(repo, *packidx, path_packidx);
1384 done:
1385 	if (err) {
1386 		got_packidx_close(*packidx);
1387 		*packidx = NULL;
1388 	}
1389 	return err;
1390 }
1391 
1392 static const struct got_error *
1393 read_packfile_hdr(int fd, struct got_packidx *packidx)
1394 {
1395 	const struct got_error *err = NULL;
1396 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1397 	struct got_packfile_hdr hdr;
1398 	ssize_t n;
1399 
1400 	n = read(fd, &hdr, sizeof(hdr));
1401 	if (n < 0)
1402 		return got_error_from_errno("read");
1403 	if (n != sizeof(hdr))
1404 		return got_error(GOT_ERR_BAD_PACKFILE);
1405 
1406 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1407 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1408 	    be32toh(hdr.nobjects) != totobj)
1409 		err = got_error(GOT_ERR_BAD_PACKFILE);
1410 
1411 	return err;
1412 }
1413 
1414 static const struct got_error *
1415 open_packfile(int *fd, struct got_repository *repo,
1416     const char *relpath, struct got_packidx *packidx)
1417 {
1418 	const struct got_error *err = NULL;
1419 
1420 	*fd = openat(got_repo_get_fd(repo), relpath,
1421 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1422 	if (*fd == -1)
1423 		return got_error_from_errno_fmt("openat: %s/%s",
1424 		    got_repo_get_path_git_dir(repo), relpath);
1425 
1426 	if (packidx) {
1427 		err = read_packfile_hdr(*fd, packidx);
1428 		if (err) {
1429 			close(*fd);
1430 			*fd = -1;
1431 		}
1432 	}
1433 
1434 	return err;
1435 }
1436 
1437 const struct got_error *
1438 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1439     const char *path_packfile, struct got_packidx *packidx)
1440 {
1441 	const struct got_error *err = NULL;
1442 	struct got_pack *pack = NULL;
1443 	struct stat sb;
1444 	size_t i;
1445 
1446 	if (packp)
1447 		*packp = NULL;
1448 
1449 	for (i = 0; i < repo->pack_cache_size; i++) {
1450 		pack = &repo->packs[i];
1451 		if (pack->path_packfile == NULL)
1452 			break;
1453 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1454 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1455 	}
1456 
1457 	if (i == repo->pack_cache_size) {
1458 		struct got_pack tmp;
1459 		do {
1460 			i--;
1461 		} while (i > 0 && repo->pinned_pack >= 0 &&
1462 		    i == repo->pinned_pack);
1463 		err = got_pack_close(&repo->packs[i]);
1464 		if (err)
1465 			return err;
1466 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1467 			return got_error_from_errno("ftruncate");
1468 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1469 			return got_error_from_errno("ftruncate");
1470 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1471 		memcpy(&repo->packs[i], &repo->packs[0],
1472 		    sizeof(repo->packs[i]));
1473 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1474 		if (repo->pinned_pack == 0)
1475 			repo->pinned_pack = i;
1476 		else if (repo->pinned_pack == i)
1477 			repo->pinned_pack = 0;
1478 		i = 0;
1479 	}
1480 
1481 	pack = &repo->packs[i];
1482 
1483 	pack->path_packfile = strdup(path_packfile);
1484 	if (pack->path_packfile == NULL) {
1485 		err = got_error_from_errno("strdup");
1486 		goto done;
1487 	}
1488 
1489 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1490 	if (err)
1491 		goto done;
1492 
1493 	if (fstat(pack->fd, &sb) != 0) {
1494 		err = got_error_from_errno("fstat");
1495 		goto done;
1496 	}
1497 	pack->filesize = sb.st_size;
1498 
1499 	pack->privsep_child = NULL;
1500 
1501 	err = got_delta_cache_alloc(&pack->delta_cache);
1502 	if (err)
1503 		goto done;
1504 
1505 #ifndef GOT_PACK_NO_MMAP
1506 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1507 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1508 		    pack->fd, 0);
1509 		if (pack->map == MAP_FAILED) {
1510 			if (errno != ENOMEM) {
1511 				err = got_error_from_errno("mmap");
1512 				goto done;
1513 			}
1514 			pack->map = NULL; /* fall back to read(2) */
1515 		}
1516 	}
1517 #endif
1518 done:
1519 	if (err) {
1520 		if (pack)
1521 			got_pack_close(pack);
1522 	} else if (packp)
1523 		*packp = pack;
1524 	return err;
1525 }
1526 
1527 struct got_pack *
1528 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1529 {
1530 	struct got_pack *pack = NULL;
1531 	size_t i;
1532 
1533 	for (i = 0; i < repo->pack_cache_size; i++) {
1534 		pack = &repo->packs[i];
1535 		if (pack->path_packfile == NULL)
1536 			break;
1537 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1538 			return pack;
1539 	}
1540 
1541 	return NULL;
1542 }
1543 
1544 const struct got_error *
1545 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1546     struct got_pack *pack)
1547 {
1548 	size_t i;
1549 	int pinned_pack = -1, pinned_packidx = -1;
1550 
1551 	for (i = 0; i < repo->pack_cache_size; i++) {
1552 		if (repo->packidx_cache[i] &&
1553 		    strcmp(repo->packidx_cache[i]->path_packidx,
1554 		    packidx->path_packidx) == 0)
1555 			pinned_packidx = i;
1556 		if (repo->packs[i].path_packfile &&
1557 		    strcmp(repo->packs[i].path_packfile,
1558 		    pack->path_packfile) == 0)
1559 			pinned_pack = i;
1560 	}
1561 
1562 	if (pinned_packidx == -1 || pinned_pack == -1)
1563 		return got_error(GOT_ERR_PIN_PACK);
1564 
1565 	repo->pinned_pack = pinned_pack;
1566 	repo->pinned_packidx = pinned_packidx;
1567 	if (repo->packs[pinned_pack].privsep_child)
1568 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1569 	return NULL;
1570 }
1571 
1572 struct got_pack *
1573 got_repo_get_pinned_pack(struct got_repository *repo)
1574 {
1575 	if (repo->pinned_pack >= 0 &&
1576 	    repo->pinned_pack < repo->pack_cache_size)
1577 		return &repo->packs[repo->pinned_pack];
1578 
1579 	return NULL;
1580 }
1581 
1582 void
1583 got_repo_unpin_pack(struct got_repository *repo)
1584 {
1585 	repo->pinned_packidx = -1;
1586 	repo->pinned_pack = -1;
1587 	repo->pinned_pid = 0;
1588 }
1589 
1590 const struct got_error *
1591 got_repo_init(const char *repo_path, const char *head_name)
1592 {
1593 	const struct got_error *err = NULL;
1594 	const char *dirnames[] = {
1595 		GOT_OBJECTS_DIR,
1596 		GOT_OBJECTS_PACK_DIR,
1597 		GOT_REFS_DIR,
1598 	};
1599 	const char *description_str = "Unnamed repository; "
1600 	    "edit this file 'description' to name the repository.";
1601 	const char *headref = "ref: refs/heads/";
1602 	const char *gitconfig_str = "[core]\n"
1603 	    "\trepositoryformatversion = 0\n"
1604 	    "\tfilemode = true\n"
1605 	    "\tbare = true\n";
1606 	char *headref_str, *path;
1607 	size_t i;
1608 
1609 	if (!got_path_dir_is_empty(repo_path))
1610 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1611 
1612 	for (i = 0; i < nitems(dirnames); i++) {
1613 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1614 			return got_error_from_errno("asprintf");
1615 		}
1616 		err = got_path_mkdir(path);
1617 		free(path);
1618 		if (err)
1619 			return err;
1620 	}
1621 
1622 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1623 		return got_error_from_errno("asprintf");
1624 	err = got_path_create_file(path, description_str);
1625 	free(path);
1626 	if (err)
1627 		return err;
1628 
1629 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1630 		return got_error_from_errno("asprintf");
1631 	if (asprintf(&headref_str, "%s%s", headref,
1632 	    head_name ? head_name : "main") == -1) {
1633 		free(path);
1634 		return got_error_from_errno("asprintf");
1635 	}
1636 	err = got_path_create_file(path, headref_str);
1637 	free(headref_str);
1638 	free(path);
1639 	if (err)
1640 		return err;
1641 
1642 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1643 		return got_error_from_errno("asprintf");
1644 	err = got_path_create_file(path, gitconfig_str);
1645 	free(path);
1646 	if (err)
1647 		return err;
1648 
1649 	return NULL;
1650 }
1651 
1652 static const struct got_error *
1653 match_packed_object(struct got_object_id **unique_id,
1654     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1655 {
1656 	const struct got_error *err = NULL;
1657 	struct got_object_id_queue matched_ids;
1658 	struct got_pathlist_entry *pe;
1659 
1660 	STAILQ_INIT(&matched_ids);
1661 
1662 	err = refresh_packidx_paths(repo);
1663 	if (err)
1664 		return err;
1665 
1666 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1667 		const char *path_packidx = pe->path;
1668 		struct got_packidx *packidx;
1669 		struct got_object_qid *qid;
1670 
1671 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1672 		    path_packidx, 0);
1673 		if (err)
1674 			break;
1675 
1676 		err = got_packidx_match_id_str_prefix(&matched_ids,
1677 		    packidx, id_str_prefix);
1678 		if (err) {
1679 			got_packidx_close(packidx);
1680 			break;
1681 		}
1682 		err = got_packidx_close(packidx);
1683 		if (err)
1684 			break;
1685 
1686 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1687 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1688 				int matched_type;
1689 				err = got_object_get_type(&matched_type, repo,
1690 				    &qid->id);
1691 				if (err)
1692 					goto done;
1693 				if (matched_type != obj_type)
1694 					continue;
1695 			}
1696 			if (*unique_id == NULL) {
1697 				*unique_id = got_object_id_dup(&qid->id);
1698 				if (*unique_id == NULL) {
1699 					err = got_error_from_errno("malloc");
1700 					goto done;
1701 				}
1702 			} else {
1703 				if (got_object_id_cmp(*unique_id,
1704 				    &qid->id) == 0)
1705 					continue; /* packed multiple times */
1706 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1707 				goto done;
1708 			}
1709 		}
1710 	}
1711 done:
1712 	got_object_id_queue_free(&matched_ids);
1713 	if (err) {
1714 		free(*unique_id);
1715 		*unique_id = NULL;
1716 	}
1717 	return err;
1718 }
1719 
1720 static const struct got_error *
1721 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1722     const char *object_dir, const char *id_str_prefix, int obj_type,
1723     struct got_repository *repo)
1724 {
1725 	const struct got_error *err = NULL;
1726 	char *path, *id_str = NULL;
1727 	DIR *dir = NULL;
1728 	struct dirent *dent;
1729 	struct got_object_id id;
1730 
1731 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1732 		err = got_error_from_errno("asprintf");
1733 		goto done;
1734 	}
1735 
1736 	dir = opendir(path);
1737 	if (dir == NULL) {
1738 		if (errno == ENOENT) {
1739 			err = NULL;
1740 			goto done;
1741 		}
1742 		err = got_error_from_errno2("opendir", path);
1743 		goto done;
1744 	}
1745 	while ((dent = readdir(dir)) != NULL) {
1746 		int cmp;
1747 
1748 		free(id_str);
1749 		id_str = NULL;
1750 
1751 		if (strcmp(dent->d_name, ".") == 0 ||
1752 		    strcmp(dent->d_name, "..") == 0)
1753 			continue;
1754 
1755 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1756 			err = got_error_from_errno("asprintf");
1757 			goto done;
1758 		}
1759 
1760 		if (!got_parse_object_id(&id, id_str, repo->algo))
1761 			continue;
1762 
1763 		/*
1764 		 * Directory entries do not necessarily appear in
1765 		 * sorted order, so we must iterate over all of them.
1766 		 */
1767 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1768 		if (cmp != 0)
1769 			continue;
1770 
1771 		if (*unique_id == NULL) {
1772 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1773 				int matched_type;
1774 				err = got_object_get_type(&matched_type, repo,
1775 				    &id);
1776 				if (err)
1777 					goto done;
1778 				if (matched_type != obj_type)
1779 					continue;
1780 			}
1781 			*unique_id = got_object_id_dup(&id);
1782 			if (*unique_id == NULL) {
1783 				err = got_error_from_errno("got_object_id_dup");
1784 				goto done;
1785 			}
1786 		} else {
1787 			if (got_object_id_cmp(*unique_id, &id) == 0)
1788 				continue; /* both packed and loose */
1789 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1790 			goto done;
1791 		}
1792 	}
1793 done:
1794 	if (dir && closedir(dir) != 0 && err == NULL)
1795 		err = got_error_from_errno("closedir");
1796 	if (err) {
1797 		free(*unique_id);
1798 		*unique_id = NULL;
1799 	}
1800 	free(id_str);
1801 	free(path);
1802 	return err;
1803 }
1804 
1805 const struct got_error *
1806 got_repo_match_object_id_prefix(struct got_object_id **id,
1807     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1808 {
1809 	const struct got_error *err = NULL;
1810 	char *path_objects = NULL, *object_dir = NULL;
1811 	size_t len;
1812 	int i;
1813 
1814 	*id = NULL;
1815 
1816 	path_objects = got_repo_get_path_objects(repo);
1817 
1818 	len = strlen(id_str_prefix);
1819 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1820 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1821 		goto done;
1822 	}
1823 
1824 	for (i = 0; i < len; i++) {
1825 		if (isxdigit((unsigned char)id_str_prefix[i]))
1826 			continue;
1827 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1828 		goto done;
1829 	}
1830 
1831 	if (len >= 2) {
1832 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1833 		if (err)
1834 			goto done;
1835 		object_dir = strndup(id_str_prefix, 2);
1836 		if (object_dir == NULL) {
1837 			err = got_error_from_errno("strdup");
1838 			goto done;
1839 		}
1840 		err = match_loose_object(id, path_objects, object_dir,
1841 		    id_str_prefix, obj_type, repo);
1842 	} else if (len == 1) {
1843 		int i;
1844 		for (i = 0; i < 0xf; i++) {
1845 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1846 			    == -1) {
1847 				err = got_error_from_errno("asprintf");
1848 				goto done;
1849 			}
1850 			err = match_packed_object(id, repo, object_dir,
1851 			    obj_type);
1852 			if (err)
1853 				goto done;
1854 			err = match_loose_object(id, path_objects, object_dir,
1855 			    id_str_prefix, obj_type, repo);
1856 			if (err)
1857 				goto done;
1858 		}
1859 	} else {
1860 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1861 		goto done;
1862 	}
1863 done:
1864 	free(path_objects);
1865 	free(object_dir);
1866 	if (err) {
1867 		free(*id);
1868 		*id = NULL;
1869 	} else if (*id == NULL) {
1870 		switch (obj_type) {
1871 		case GOT_OBJ_TYPE_BLOB:
1872 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1873 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1874 			break;
1875 		case GOT_OBJ_TYPE_TREE:
1876 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1877 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1878 			break;
1879 		case GOT_OBJ_TYPE_COMMIT:
1880 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1881 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1882 			break;
1883 		case GOT_OBJ_TYPE_TAG:
1884 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1885 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1886 			break;
1887 		default:
1888 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1889 			break;
1890 		}
1891 	}
1892 
1893 	return err;
1894 }
1895 
1896 const struct got_error *
1897 got_repo_match_object_id(struct got_object_id **id, char **label,
1898     const char *id_str, int obj_type, struct got_reflist_head *refs,
1899     struct got_repository *repo)
1900 {
1901 	const struct got_error *err;
1902 	struct got_tag_object *tag;
1903 	struct got_reference *ref = NULL;
1904 
1905 	*id = NULL;
1906 	if (label)
1907 		*label = NULL;
1908 
1909 	if (refs) {
1910 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1911 		    refs, repo);
1912 		if (err == NULL) {
1913 			*id = got_object_id_dup(
1914 			    got_object_tag_get_object_id(tag));
1915 			if (*id == NULL)
1916 				err = got_error_from_errno("got_object_id_dup");
1917 			else if (label && asprintf(label, "refs/tags/%s",
1918 			    got_object_tag_get_name(tag)) == -1) {
1919 				err = got_error_from_errno("asprintf");
1920 				free(*id);
1921 				*id = NULL;
1922 			}
1923 			got_object_tag_close(tag);
1924 			return err;
1925 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1926 		    err->code != GOT_ERR_NO_OBJ)
1927 			return err;
1928 	}
1929 
1930 	err = got_ref_open(&ref, repo, id_str, 0);
1931 	if (err == NULL) {
1932 		err = got_ref_resolve(id, repo, ref);
1933 		if (err)
1934 			goto done;
1935 		if (label) {
1936 			*label = strdup(got_ref_get_name(ref));
1937 			if (*label == NULL) {
1938 				err = got_error_from_errno("strdup");
1939 				goto done;
1940 			}
1941 		}
1942 	} else {
1943 		if (err->code != GOT_ERR_NOT_REF &&
1944 		    err->code != GOT_ERR_BAD_REF_NAME)
1945 			goto done;
1946 		err = got_repo_match_object_id_prefix(id, id_str,
1947 		    obj_type, repo);
1948 		if (err) {
1949 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1950 				err = got_error_not_ref(id_str);
1951 			goto done;
1952 		}
1953 		if (label) {
1954 			err = got_object_id_str(label, *id);
1955 			if (*label == NULL) {
1956 				err = got_error_from_errno("strdup");
1957 				goto done;
1958 			}
1959 		}
1960 	}
1961 done:
1962 	if (ref)
1963 		got_ref_close(ref);
1964 	return err;
1965 }
1966 
1967 const struct got_error *
1968 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1969     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1970 {
1971 	const struct got_error *err = NULL;
1972 	struct got_reflist_entry *re;
1973 	struct got_object_id *tag_id;
1974 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1975 
1976 	*tag = NULL;
1977 
1978 	TAILQ_FOREACH(re, refs, entry) {
1979 		const char *refname;
1980 		refname = got_ref_get_name(re->ref);
1981 		if (got_ref_is_symbolic(re->ref))
1982 			continue;
1983 		if (strncmp(refname, "refs/tags/", 10) != 0)
1984 			continue;
1985 		if (!name_is_absolute)
1986 			refname += strlen("refs/tags/");
1987 		if (strcmp(refname, name) != 0)
1988 			continue;
1989 		err = got_ref_resolve(&tag_id, repo, re->ref);
1990 		if (err)
1991 			break;
1992 		err = got_object_open_as_tag(tag, repo, tag_id);
1993 		free(tag_id);
1994 		if (err)
1995 			break;
1996 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1997 		    got_object_tag_get_object_type(*tag) == obj_type)
1998 			break;
1999 		got_object_tag_close(*tag);
2000 		*tag = NULL;
2001 	}
2002 
2003 	if (err == NULL && *tag == NULL)
2004 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2005 		    GOT_OBJ_LABEL_TAG, name);
2006 	return err;
2007 }
2008 
2009 static const struct got_error *
2010 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2011     const char *name, mode_t mode, struct got_object_id *blob_id)
2012 {
2013 	const struct got_error *err = NULL;
2014 
2015 	 *new_te = NULL;
2016 
2017 	*new_te = calloc(1, sizeof(**new_te));
2018 	if (*new_te == NULL)
2019 		return got_error_from_errno("calloc");
2020 
2021 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2022 	    sizeof((*new_te)->name)) {
2023 		err = got_error(GOT_ERR_NO_SPACE);
2024 		goto done;
2025 	}
2026 
2027 	if (S_ISLNK(mode)) {
2028 		(*new_te)->mode = S_IFLNK;
2029 	} else {
2030 		(*new_te)->mode = S_IFREG;
2031 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2032 	}
2033 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2034 done:
2035 	if (err && *new_te) {
2036 		free(*new_te);
2037 		*new_te = NULL;
2038 	}
2039 	return err;
2040 }
2041 
2042 static const struct got_error *
2043 import_file(struct got_tree_entry **new_te, struct dirent *de,
2044     const char *path, struct got_repository *repo)
2045 {
2046 	const struct got_error *err;
2047 	struct got_object_id *blob_id = NULL;
2048 	char *filepath;
2049 	struct stat sb;
2050 
2051 	if (asprintf(&filepath, "%s%s%s", path,
2052 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2053 		return got_error_from_errno("asprintf");
2054 
2055 	if (lstat(filepath, &sb) != 0) {
2056 		err = got_error_from_errno2("lstat", path);
2057 		goto done;
2058 	}
2059 
2060 	err = got_object_blob_create(&blob_id, filepath, repo);
2061 	if (err)
2062 		goto done;
2063 
2064 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2065 	    blob_id);
2066 done:
2067 	free(filepath);
2068 	if (err)
2069 		free(blob_id);
2070 	return err;
2071 }
2072 
2073 static const struct got_error *
2074 insert_tree_entry(struct got_tree_entry *new_te,
2075     struct got_pathlist_head *paths)
2076 {
2077 	const struct got_error *err = NULL;
2078 	struct got_pathlist_entry *new_pe;
2079 
2080 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2081 	if (err)
2082 		return err;
2083 	if (new_pe == NULL)
2084 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2085 	return NULL;
2086 }
2087 
2088 static const struct got_error *write_tree(struct got_object_id **,
2089     const char *, struct got_pathlist_head *, struct got_repository *,
2090     got_repo_import_cb progress_cb, void *progress_arg);
2091 
2092 static const struct got_error *
2093 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2094     const char *path, struct got_pathlist_head *ignores,
2095     struct got_repository *repo,
2096     got_repo_import_cb progress_cb, void *progress_arg)
2097 {
2098 	const struct got_error *err;
2099 	struct got_object_id *id = NULL;
2100 	char *subdirpath;
2101 
2102 	if (asprintf(&subdirpath, "%s%s%s", path,
2103 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2104 		return got_error_from_errno("asprintf");
2105 
2106 	(*new_te) = calloc(1, sizeof(**new_te));
2107 	if (*new_te == NULL)
2108 		return got_error_from_errno("calloc");
2109 	(*new_te)->mode = S_IFDIR;
2110 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2111 	    sizeof((*new_te)->name)) {
2112 		err = got_error(GOT_ERR_NO_SPACE);
2113 		goto done;
2114 	}
2115 	err = write_tree(&id, subdirpath, ignores,  repo,
2116 	    progress_cb, progress_arg);
2117 	if (err)
2118 		goto done;
2119 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2120 
2121 done:
2122 	free(id);
2123 	free(subdirpath);
2124 	if (err) {
2125 		free(*new_te);
2126 		*new_te = NULL;
2127 	}
2128 	return err;
2129 }
2130 
2131 static const struct got_error *
2132 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2133     struct got_pathlist_head *ignores, struct got_repository *repo,
2134     got_repo_import_cb progress_cb, void *progress_arg)
2135 {
2136 	const struct got_error *err = NULL;
2137 	DIR *dir;
2138 	struct dirent *de;
2139 	int nentries;
2140 	struct got_tree_entry *new_te = NULL;
2141 	struct got_pathlist_head paths;
2142 	struct got_pathlist_entry *pe;
2143 
2144 	*new_tree_id = NULL;
2145 
2146 	TAILQ_INIT(&paths);
2147 
2148 	dir = opendir(path_dir);
2149 	if (dir == NULL) {
2150 		err = got_error_from_errno2("opendir", path_dir);
2151 		goto done;
2152 	}
2153 
2154 	nentries = 0;
2155 	while ((de = readdir(dir)) != NULL) {
2156 		int ignore = 0;
2157 		int type;
2158 
2159 		if (strcmp(de->d_name, ".") == 0 ||
2160 		    strcmp(de->d_name, "..") == 0)
2161 			continue;
2162 
2163 		err = got_path_dirent_type(&type, path_dir, de);
2164 		if (err)
2165 			goto done;
2166 
2167 		TAILQ_FOREACH(pe, ignores, entry) {
2168 			if (type == DT_DIR && pe->path_len > 0 &&
2169 			    pe->path[pe->path_len - 1] == '/') {
2170 				char stripped[PATH_MAX];
2171 
2172 				if (strlcpy(stripped, pe->path,
2173 				    sizeof(stripped)) >= sizeof(stripped)) {
2174 					err = got_error(GOT_ERR_NO_SPACE);
2175 					goto done;
2176 				}
2177 				got_path_strip_trailing_slashes(stripped);
2178 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2179 					ignore = 1;
2180 					break;
2181 				}
2182 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2183 				ignore = 1;
2184 				break;
2185 			}
2186 		}
2187 		if (ignore)
2188 			continue;
2189 
2190 		if (type == DT_DIR) {
2191 			err = import_subdir(&new_te, de, path_dir,
2192 			    ignores, repo, progress_cb, progress_arg);
2193 			if (err) {
2194 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2195 					goto done;
2196 				err = NULL;
2197 				continue;
2198 			}
2199 		} else if (type == DT_REG || type == DT_LNK) {
2200 			err = import_file(&new_te, de, path_dir, repo);
2201 			if (err)
2202 				goto done;
2203 		} else
2204 			continue;
2205 
2206 		err = insert_tree_entry(new_te, &paths);
2207 		if (err)
2208 			goto done;
2209 		nentries++;
2210 	}
2211 
2212 	if (TAILQ_EMPTY(&paths)) {
2213 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2214 		    "cannot create tree without any entries");
2215 		goto done;
2216 	}
2217 
2218 	TAILQ_FOREACH(pe, &paths, entry) {
2219 		struct got_tree_entry *te = pe->data;
2220 		char *path;
2221 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2222 			continue;
2223 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2224 			err = got_error_from_errno("asprintf");
2225 			goto done;
2226 		}
2227 		err = (*progress_cb)(progress_arg, path);
2228 		free(path);
2229 		if (err)
2230 			goto done;
2231 	}
2232 
2233 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2234 done:
2235 	if (dir)
2236 		closedir(dir);
2237 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2238 	return err;
2239 }
2240 
2241 const struct got_error *
2242 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2243     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2244     struct got_repository *repo, got_repo_import_cb progress_cb,
2245     void *progress_arg)
2246 {
2247 	const struct got_error *err;
2248 	struct got_object_id *new_tree_id;
2249 
2250 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2251 	    progress_cb, progress_arg);
2252 	if (err)
2253 		return err;
2254 
2255 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2256 	    author, time(NULL), author, time(NULL), logmsg, repo);
2257 	free(new_tree_id);
2258 	return err;
2259 }
2260 
2261 const struct got_error *
2262 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2263     struct got_repository *repo)
2264 {
2265 	const struct got_error *err = NULL;
2266 	char *path_objects = NULL, *path = NULL;
2267 	DIR *dir = NULL;
2268 	struct got_object_id id;
2269 	int i;
2270 
2271 	*nobjects = 0;
2272 	*ondisk_size = 0;
2273 
2274 	path_objects = got_repo_get_path_objects(repo);
2275 	if (path_objects == NULL)
2276 		return got_error_from_errno("got_repo_get_path_objects");
2277 
2278 	for (i = 0; i <= 0xff; i++) {
2279 		struct dirent *dent;
2280 
2281 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2282 			err = got_error_from_errno("asprintf");
2283 			break;
2284 		}
2285 
2286 		dir = opendir(path);
2287 		if (dir == NULL) {
2288 			if (errno == ENOENT) {
2289 				err = NULL;
2290 				continue;
2291 			}
2292 			err = got_error_from_errno2("opendir", path);
2293 			break;
2294 		}
2295 
2296 		while ((dent = readdir(dir)) != NULL) {
2297 			char *id_str;
2298 			int fd;
2299 			struct stat sb;
2300 
2301 			if (strcmp(dent->d_name, ".") == 0 ||
2302 			    strcmp(dent->d_name, "..") == 0)
2303 				continue;
2304 
2305 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2306 				err = got_error_from_errno("asprintf");
2307 				goto done;
2308 			}
2309 
2310 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
2311 				free(id_str);
2312 				continue;
2313 			}
2314 			free(id_str);
2315 
2316 			err = got_object_open_loose_fd(&fd, &id, repo);
2317 			if (err)
2318 				goto done;
2319 
2320 			if (fstat(fd, &sb) == -1) {
2321 				err = got_error_from_errno("fstat");
2322 				close(fd);
2323 				goto done;
2324 			}
2325 			(*nobjects)++;
2326 			(*ondisk_size) += sb.st_size;
2327 
2328 			if (close(fd) == -1) {
2329 				err = got_error_from_errno("close");
2330 				goto done;
2331 			}
2332 		}
2333 
2334 		if (closedir(dir) != 0) {
2335 			err = got_error_from_errno("closedir");
2336 			goto done;
2337 		}
2338 		dir = NULL;
2339 
2340 		free(path);
2341 		path = NULL;
2342 	}
2343 done:
2344 	if (dir && closedir(dir) != 0 && err == NULL)
2345 		err = got_error_from_errno("closedir");
2346 
2347 	if (err) {
2348 		*nobjects = 0;
2349 		*ondisk_size = 0;
2350 	}
2351 	free(path_objects);
2352 	free(path);
2353 	return err;
2354 }
2355 
2356 const struct got_error *
2357 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2358     off_t *total_packsize, struct got_repository *repo)
2359 {
2360 	const struct got_error *err = NULL;
2361 	DIR *packdir = NULL;
2362 	struct dirent *dent;
2363 	struct got_packidx *packidx = NULL;
2364 	char *path_packidx;
2365 	char *path_packfile;
2366 	int packdir_fd;
2367 	struct stat sb;
2368 
2369 	*npackfiles = 0;
2370 	*nobjects = 0;
2371 	*total_packsize = 0;
2372 
2373 	packdir_fd = openat(got_repo_get_fd(repo),
2374 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2375 	if (packdir_fd == -1) {
2376 		return got_error_from_errno_fmt("openat: %s/%s",
2377 		    got_repo_get_path_git_dir(repo),
2378 		    GOT_OBJECTS_PACK_DIR);
2379 	}
2380 
2381 	packdir = fdopendir(packdir_fd);
2382 	if (packdir == NULL) {
2383 		err = got_error_from_errno("fdopendir");
2384 		goto done;
2385 	}
2386 
2387 	while ((dent = readdir(packdir)) != NULL) {
2388 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
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);