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 > 1) {
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 (repo->gitconfig_repository_format_version == 1 &&
804 		    strcasecmp(ext, "objectformat") == 0) {
805 			if (strcmp(val, "sha1") == 0)
806 				continue;
807 			if (strcmp(val, "sha256") == 0) {
808 				repo->algo = GOT_HASH_SHA256;
809 				continue;
810 			}
811 			err = got_error_path(val, GOT_ERR_OBJECT_FORMAT);
812 			goto done;
813 		}
814 
815 		if (!is_boolean_val(val)) {
816 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
817 			goto done;
818 		}
819 
820 		if (!get_boolean_val(val))
821 			continue;
822 
823 		for (j = 0; j < nitems(repo_extensions); j++) {
824 			if (strcmp(ext, repo_extensions[j]) == 0) {
825 				supported = 1;
826 				break;
827 			}
828 		}
829 		if (!supported) {
830 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
831 			goto done;
832 		}
833 	}
834 
835 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
836 done:
837 	if (err)
838 		got_repo_close(repo);
839 	else
840 		*repop = repo;
841 	free(repo_path);
842 	return err;
843 }
844 
845 const struct got_error *
846 got_repo_close(struct got_repository *repo)
847 {
848 	const struct got_error *err = NULL, *child_err;
849 	struct got_packidx_bloom_filter *bf;
850 	size_t i;
851 
852 	for (i = 0; i < repo->pack_cache_size; i++) {
853 		if (repo->packidx_cache[i] == NULL)
854 			break;
855 		got_packidx_close(repo->packidx_cache[i]);
856 	}
857 
858 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
859 	    &repo->packidx_bloom_filters))) {
860 		RB_REMOVE(got_packidx_bloom_filter_tree,
861 		    &repo->packidx_bloom_filters, bf);
862 		bloom_free(bf->bloom);
863 		free(bf->bloom);
864 		free(bf);
865 	}
866 
867 	for (i = 0; i < repo->pack_cache_size; i++)
868 		if (repo->packs[i].path_packfile)
869 			if (repo->packs[i].path_packfile)
870 				got_pack_close(&repo->packs[i]);
871 
872 	free(repo->path);
873 	free(repo->path_git_dir);
874 
875 	got_object_cache_close(&repo->objcache);
876 	got_object_cache_close(&repo->treecache);
877 	got_object_cache_close(&repo->commitcache);
878 	got_object_cache_close(&repo->tagcache);
879 	got_object_cache_close(&repo->rawcache);
880 
881 	for (i = 0; i < nitems(repo->privsep_children); i++) {
882 		if (repo->privsep_children[i].imsg_fd == -1)
883 			continue;
884 		imsg_clear(repo->privsep_children[i].ibuf);
885 		free(repo->privsep_children[i].ibuf);
886 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
887 		child_err = got_privsep_wait_for_child(
888 		    repo->privsep_children[i].pid);
889 		if (child_err && err == NULL)
890 			err = child_err;
891 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
892 		    err == NULL)
893 			err = got_error_from_errno("close");
894 	}
895 
896 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
897 	    err == NULL)
898 		err = got_error_from_errno("close");
899 
900 	if (repo->gotconfig)
901 		got_gotconfig_free(repo->gotconfig);
902 	free(repo->gitconfig_author_name);
903 	free(repo->gitconfig_author_email);
904 	for (i = 0; i < repo->ngitconfig_remotes; i++)
905 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
906 	free(repo->gitconfig_remotes);
907 	for (i = 0; i < repo->nextensions; i++) {
908 		free(repo->extnames[i]);
909 		free(repo->extvals[i]);
910 	}
911 	free(repo->extnames);
912 	free(repo->extvals);
913 
914 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
915 	free(repo);
916 
917 	return err;
918 }
919 
920 void
921 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
922 {
923 	int i;
924 
925 	free(repo->name);
926 	repo->name = NULL;
927 	free(repo->fetch_url);
928 	repo->fetch_url = NULL;
929 	free(repo->send_url);
930 	repo->send_url = NULL;
931 	for (i = 0; i < repo->nfetch_branches; i++)
932 		free(repo->fetch_branches[i]);
933 	free(repo->fetch_branches);
934 	repo->fetch_branches = NULL;
935 	repo->nfetch_branches = 0;
936 	for (i = 0; i < repo->nsend_branches; i++)
937 		free(repo->send_branches[i]);
938 	free(repo->send_branches);
939 	repo->send_branches = NULL;
940 	repo->nsend_branches = 0;
941 }
942 
943 const struct got_error *
944 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
945     const char *input_path)
946 {
947 	const struct got_error *err = NULL;
948 	const char *repo_abspath = NULL;
949 	size_t repolen, len;
950 	char *canonpath, *path = NULL;
951 
952 	*in_repo_path = NULL;
953 
954 	canonpath = strdup(input_path);
955 	if (canonpath == NULL) {
956 		err = got_error_from_errno("strdup");
957 		goto done;
958 	}
959 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
960 	if (err)
961 		goto done;
962 
963 	repo_abspath = got_repo_get_path(repo);
964 
965 	if (canonpath[0] == '\0') {
966 		path = strdup(canonpath);
967 		if (path == NULL) {
968 			err = got_error_from_errno("strdup");
969 			goto done;
970 		}
971 	} else {
972 		path = realpath(canonpath, NULL);
973 		if (path == NULL) {
974 			if (errno != ENOENT) {
975 				err = got_error_from_errno2("realpath",
976 				    canonpath);
977 				goto done;
978 			}
979 			/*
980 			 * Path is not on disk.
981 			 * Assume it is already relative to repository root.
982 			 */
983 			path = strdup(canonpath);
984 			if (path == NULL) {
985 				err = got_error_from_errno("strdup");
986 				goto done;
987 			}
988 		}
989 
990 		repolen = strlen(repo_abspath);
991 		len = strlen(path);
992 
993 
994 		if (strcmp(path, repo_abspath) == 0) {
995 			free(path);
996 			path = strdup("");
997 			if (path == NULL) {
998 				err = got_error_from_errno("strdup");
999 				goto done;
1000 			}
1001 		} else if (len > repolen &&
1002 		    got_path_is_child(path, repo_abspath, repolen)) {
1003 			/* Matched an on-disk path inside repository. */
1004 			if (got_repo_is_bare(repo)) {
1005 				/*
1006 				 * Matched an on-disk path inside repository
1007 				 * database. Treat input as repository-relative.
1008 				 */
1009 				free(path);
1010 				path = canonpath;
1011 				canonpath = NULL;
1012 			} else {
1013 				char *child;
1014 				/* Strip common prefix with repository path. */
1015 				err = got_path_skip_common_ancestor(&child,
1016 				    repo_abspath, path);
1017 				if (err)
1018 					goto done;
1019 				free(path);
1020 				path = child;
1021 			}
1022 		} else {
1023 			/*
1024 			 * Matched unrelated on-disk path.
1025 			 * Treat input as repository-relative.
1026 			 */
1027 			free(path);
1028 			path = canonpath;
1029 			canonpath = NULL;
1030 		}
1031 	}
1032 
1033 	/* Make in-repository path absolute */
1034 	if (path[0] != '/') {
1035 		char *abspath;
1036 		if (asprintf(&abspath, "/%s", path) == -1) {
1037 			err = got_error_from_errno("asprintf");
1038 			goto done;
1039 		}
1040 		free(path);
1041 		path = abspath;
1042 	}
1043 
1044 done:
1045 	free(canonpath);
1046 	if (err)
1047 		free(path);
1048 	else
1049 		*in_repo_path = path;
1050 	return err;
1051 }
1052 
1053 static const struct got_error *
1054 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1055     const char *path_packidx)
1056 {
1057 	const struct got_error *err = NULL;
1058 	size_t i;
1059 
1060 	for (i = 0; i < repo->pack_cache_size; i++) {
1061 		if (repo->packidx_cache[i] == NULL)
1062 			break;
1063 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1064 		    path_packidx) == 0) {
1065 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1066 		}
1067 	}
1068 	if (i == repo->pack_cache_size) {
1069 		do {
1070 			i--;
1071 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1072 		    i == repo->pinned_packidx);
1073 		err = got_packidx_close(repo->packidx_cache[i]);
1074 		if (err)
1075 			return err;
1076 	}
1077 
1078 	repo->packidx_cache[i] = packidx;
1079 
1080 	return NULL;
1081 }
1082 
1083 int
1084 got_repo_is_packidx_filename(const char *name, size_t len,
1085     enum got_hash_algorithm algo)
1086 {
1087 	size_t idlen;
1088 
1089 	idlen = got_hash_digest_string_length(algo);
1090 
1091 	if (len != GOT_PACKIDX_NAMELEN(idlen))
1092 		return 0;
1093 
1094 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1095 		return 0;
1096 
1097 	if (strcmp(name + strlen(GOT_PACK_PREFIX) + idlen - 1,
1098 	    GOT_PACKIDX_SUFFIX) != 0)
1099 		return 0;
1100 
1101 	return 1;
1102 }
1103 
1104 static struct got_packidx_bloom_filter *
1105 get_packidx_bloom_filter(struct got_repository *repo,
1106     const char *path, size_t path_len)
1107 {
1108 	struct got_packidx_bloom_filter key;
1109 
1110 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1111 		return NULL; /* XXX */
1112 	key.path_len = path_len;
1113 
1114 	return RB_FIND(got_packidx_bloom_filter_tree,
1115 	    &repo->packidx_bloom_filters, &key);
1116 }
1117 
1118 int
1119 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1120     const char *path_packidx, struct got_object_id *id)
1121 {
1122 	struct got_packidx_bloom_filter *bf;
1123 
1124 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1125 	if (bf)
1126 		return bloom_check(bf->bloom, id->hash,
1127 		    got_hash_digest_length(id->algo));
1128 
1129 	/* No bloom filter means this pack index must be searched. */
1130 	return 1;
1131 }
1132 
1133 static const struct got_error *
1134 add_packidx_bloom_filter(struct got_repository *repo,
1135     struct got_packidx *packidx, const char *path_packidx)
1136 {
1137 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1138 	struct got_packidx_bloom_filter *bf;
1139 	size_t len, idlen;
1140 
1141 	idlen = got_hash_digest_length(repo->algo);
1142 
1143 	/*
1144 	 * Don't use bloom filters for very large pack index files.
1145 	 * Large pack files will contain a relatively large fraction
1146 	 * of our objects so we will likely need to visit them anyway.
1147 	 * The more objects a pack file contains the higher the probability
1148 	 * of a false-positive match from the bloom filter. And reading
1149 	 * all object IDs from a large pack index file can be expensive.
1150 	 */
1151 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1152 		return NULL;
1153 
1154 	/* Do we already have a filter for this pack index? */
1155 	if (get_packidx_bloom_filter(repo, path_packidx,
1156 	    strlen(path_packidx)) != NULL)
1157 		return NULL;
1158 
1159 	bf = calloc(1, sizeof(*bf));
1160 	if (bf == NULL)
1161 		return got_error_from_errno("calloc");
1162 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1163 	if (bf->bloom == NULL) {
1164 		free(bf);
1165 		return got_error_from_errno("calloc");
1166 	}
1167 
1168 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1169 	if (len >= sizeof(bf->path)) {
1170 		free(bf->bloom);
1171 		free(bf);
1172 		return got_error(GOT_ERR_NO_SPACE);
1173 	}
1174 	bf->path_len = len;
1175 
1176 	/* Minimum size supported by our bloom filter is 1000 entries. */
1177 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1178 	for (i = 0; i < nobjects; i++) {
1179 		struct got_packidx_object_id *id;
1180 		id = packidx->hdr.sorted_ids + i * idlen;
1181 		bloom_add(bf->bloom, id->hash, idlen);
1182 	}
1183 
1184 	RB_INSERT(got_packidx_bloom_filter_tree,
1185 	    &repo->packidx_bloom_filters, bf);
1186 	return NULL;
1187 }
1188 
1189 static void
1190 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1191 {
1192 	struct got_pathlist_entry *pe;
1193 
1194 	while (!TAILQ_EMPTY(packidx_paths)) {
1195 		pe = TAILQ_FIRST(packidx_paths);
1196 		TAILQ_REMOVE(packidx_paths, pe, entry);
1197 		free((char *)pe->path);
1198 		free(pe);
1199 	}
1200 }
1201 
1202 static const struct got_error *
1203 refresh_packidx_paths(struct got_repository *repo)
1204 {
1205 	const struct got_error *err = NULL;
1206 	char *objects_pack_dir = NULL;
1207 	struct stat sb;
1208 
1209 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1210 	if (objects_pack_dir == NULL)
1211 		return got_error_from_errno("got_repo_get_path_objects_pack");
1212 
1213 	if (stat(objects_pack_dir, &sb) == -1) {
1214 		if (errno != ENOENT) {
1215 			err = got_error_from_errno2("stat", objects_pack_dir);
1216 			goto done;
1217 		}
1218 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1219 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1220 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1221 		purge_packidx_paths(&repo->packidx_paths);
1222 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1223 		if (err)
1224 			goto done;
1225 	}
1226 done:
1227 	free(objects_pack_dir);
1228 	return err;
1229 }
1230 
1231 const struct got_error *
1232 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1233     struct got_repository *repo, struct got_object_id *id)
1234 {
1235 	const struct got_error *err;
1236 	struct got_pathlist_entry *pe;
1237 	size_t i;
1238 
1239 	/* Search pack index cache. */
1240 	for (i = 0; i < repo->pack_cache_size; i++) {
1241 		if (repo->packidx_cache[i] == NULL)
1242 			break;
1243 		if (!got_repo_check_packidx_bloom_filter(repo,
1244 		    repo->packidx_cache[i]->path_packidx, id))
1245 			continue; /* object will not be found in this index */
1246 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1247 		if (*idx != -1) {
1248 			*packidx = repo->packidx_cache[i];
1249 			/*
1250 			 * Move this cache entry to the front. Repeatedly
1251 			 * searching a wrong pack index can be expensive.
1252 			 */
1253 			if (i > 0) {
1254 				memmove(&repo->packidx_cache[1],
1255 				    &repo->packidx_cache[0],
1256 				    i * sizeof(repo->packidx_cache[0]));
1257 				repo->packidx_cache[0] = *packidx;
1258 				if (repo->pinned_packidx >= 0 &&
1259 				    repo->pinned_packidx < i)
1260 					repo->pinned_packidx++;
1261 				else if (repo->pinned_packidx == i)
1262 					repo->pinned_packidx = 0;
1263 			}
1264 			return NULL;
1265 		}
1266 	}
1267 	/* No luck. Search the filesystem. */
1268 
1269 	err = refresh_packidx_paths(repo);
1270 	if (err)
1271 		return err;
1272 
1273 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1274 		const char *path_packidx = pe->path;
1275 		int is_cached = 0;
1276 
1277 		if (!got_repo_check_packidx_bloom_filter(repo,
1278 		    pe->path, id))
1279 			continue; /* object will not be found in this index */
1280 
1281 		for (i = 0; i < repo->pack_cache_size; i++) {
1282 			if (repo->packidx_cache[i] == NULL)
1283 				break;
1284 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1285 			    path_packidx) == 0) {
1286 				is_cached = 1;
1287 				break;
1288 			}
1289 		}
1290 		if (is_cached)
1291 			continue; /* already searched */
1292 
1293 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1294 		    path_packidx, 0, repo->algo);
1295 		if (err)
1296 			goto done;
1297 
1298 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1299 		if (err)
1300 			goto done;
1301 
1302 		err = cache_packidx(repo, *packidx, path_packidx);
1303 		if (err)
1304 			goto done;
1305 
1306 		*idx = got_packidx_get_object_idx(*packidx, id);
1307 		if (*idx != -1) {
1308 			err = NULL; /* found the object */
1309 			goto done;
1310 		}
1311 	}
1312 
1313 	err = got_error_no_obj(id);
1314 done:
1315 	return err;
1316 }
1317 
1318 const struct got_error *
1319 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1320     struct got_repository *repo)
1321 {
1322 	const struct got_error *err = NULL;
1323 	DIR *packdir = NULL;
1324 	struct dirent *dent;
1325 	char *path_packidx = NULL;
1326 	int packdir_fd;
1327 	struct stat sb;
1328 
1329 	packdir_fd = openat(got_repo_get_fd(repo),
1330 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1331 	if (packdir_fd == -1) {
1332 		return got_error_from_errno_fmt("openat: %s/%s",
1333 		    got_repo_get_path_git_dir(repo),
1334 		    GOT_OBJECTS_PACK_DIR);
1335 	}
1336 
1337 	packdir = fdopendir(packdir_fd);
1338 	if (packdir == NULL) {
1339 		err = got_error_from_errno("fdopendir");
1340 		goto done;
1341 	}
1342 
1343 	if (fstat(packdir_fd, &sb) == -1) {
1344 		err = got_error_from_errno("fstat");
1345 		goto done;
1346 	}
1347 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1348 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1349 
1350 	while ((dent = readdir(packdir)) != NULL) {
1351 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
1352 		    repo->algo))
1353 			continue;
1354 
1355 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1356 		    dent->d_name) == -1) {
1357 			err = got_error_from_errno("asprintf");
1358 			path_packidx = NULL;
1359 			break;
1360 		}
1361 
1362 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1363 		if (err)
1364 			break;
1365 	}
1366 done:
1367 	if (err)
1368 		free(path_packidx);
1369 	if (packdir && closedir(packdir) != 0 && err == NULL)
1370 		err = got_error_from_errno("closedir");
1371 	return err;
1372 }
1373 
1374 const struct got_error *
1375 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1376     struct got_repository *repo)
1377 {
1378 	const struct got_error *err;
1379 	size_t i;
1380 
1381 	*packidx = NULL;
1382 
1383 	/* Search pack index cache. */
1384 	for (i = 0; i < repo->pack_cache_size; i++) {
1385 		if (repo->packidx_cache[i] == NULL)
1386 			break;
1387 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1388 		    path_packidx) == 0) {
1389 			*packidx = repo->packidx_cache[i];
1390 			return NULL;
1391 		}
1392 	}
1393 	/* No luck. Search the filesystem. */
1394 
1395 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1396 	    path_packidx, 0, repo->algo);
1397 	if (err)
1398 		return err;
1399 
1400 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1401 	if (err)
1402 		goto done;
1403 
1404 	err = cache_packidx(repo, *packidx, path_packidx);
1405 done:
1406 	if (err) {
1407 		got_packidx_close(*packidx);
1408 		*packidx = NULL;
1409 	}
1410 	return err;
1411 }
1412 
1413 static const struct got_error *
1414 read_packfile_hdr(int fd, struct got_packidx *packidx)
1415 {
1416 	const struct got_error *err = NULL;
1417 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1418 	struct got_packfile_hdr hdr;
1419 	ssize_t n;
1420 
1421 	n = read(fd, &hdr, sizeof(hdr));
1422 	if (n < 0)
1423 		return got_error_from_errno("read");
1424 	if (n != sizeof(hdr))
1425 		return got_error(GOT_ERR_BAD_PACKFILE);
1426 
1427 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1428 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1429 	    be32toh(hdr.nobjects) != totobj)
1430 		err = got_error(GOT_ERR_BAD_PACKFILE);
1431 
1432 	return err;
1433 }
1434 
1435 static const struct got_error *
1436 open_packfile(int *fd, struct got_repository *repo,
1437     const char *relpath, struct got_packidx *packidx)
1438 {
1439 	const struct got_error *err = NULL;
1440 
1441 	*fd = openat(got_repo_get_fd(repo), relpath,
1442 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1443 	if (*fd == -1)
1444 		return got_error_from_errno_fmt("openat: %s/%s",
1445 		    got_repo_get_path_git_dir(repo), relpath);
1446 
1447 	if (packidx) {
1448 		err = read_packfile_hdr(*fd, packidx);
1449 		if (err) {
1450 			close(*fd);
1451 			*fd = -1;
1452 		}
1453 	}
1454 
1455 	return err;
1456 }
1457 
1458 const struct got_error *
1459 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1460     const char *path_packfile, struct got_packidx *packidx)
1461 {
1462 	const struct got_error *err = NULL;
1463 	struct got_pack *pack = NULL;
1464 	struct stat sb;
1465 	size_t i;
1466 
1467 	if (packp)
1468 		*packp = NULL;
1469 
1470 	for (i = 0; i < repo->pack_cache_size; i++) {
1471 		pack = &repo->packs[i];
1472 		if (pack->path_packfile == NULL)
1473 			break;
1474 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1475 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1476 	}
1477 
1478 	if (i == repo->pack_cache_size) {
1479 		struct got_pack tmp;
1480 		do {
1481 			i--;
1482 		} while (i > 0 && repo->pinned_pack >= 0 &&
1483 		    i == repo->pinned_pack);
1484 		err = got_pack_close(&repo->packs[i]);
1485 		if (err)
1486 			return err;
1487 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1488 			return got_error_from_errno("ftruncate");
1489 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1490 			return got_error_from_errno("ftruncate");
1491 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1492 		memcpy(&repo->packs[i], &repo->packs[0],
1493 		    sizeof(repo->packs[i]));
1494 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1495 		if (repo->pinned_pack == 0)
1496 			repo->pinned_pack = i;
1497 		else if (repo->pinned_pack == i)
1498 			repo->pinned_pack = 0;
1499 		i = 0;
1500 	}
1501 
1502 	pack = &repo->packs[i];
1503 
1504 	pack->path_packfile = strdup(path_packfile);
1505 	if (pack->path_packfile == NULL) {
1506 		err = got_error_from_errno("strdup");
1507 		goto done;
1508 	}
1509 
1510 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1511 	if (err)
1512 		goto done;
1513 
1514 	if (fstat(pack->fd, &sb) != 0) {
1515 		err = got_error_from_errno("fstat");
1516 		goto done;
1517 	}
1518 	pack->filesize = sb.st_size;
1519 	pack->algo = got_repo_get_object_format(repo);
1520 
1521 	pack->privsep_child = NULL;
1522 
1523 	err = got_delta_cache_alloc(&pack->delta_cache);
1524 	if (err)
1525 		goto done;
1526 
1527 #ifndef GOT_PACK_NO_MMAP
1528 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1529 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1530 		    pack->fd, 0);
1531 		if (pack->map == MAP_FAILED) {
1532 			if (errno != ENOMEM) {
1533 				err = got_error_from_errno("mmap");
1534 				goto done;
1535 			}
1536 			pack->map = NULL; /* fall back to read(2) */
1537 		}
1538 	}
1539 #endif
1540 done:
1541 	if (err) {
1542 		if (pack)
1543 			got_pack_close(pack);
1544 	} else if (packp)
1545 		*packp = pack;
1546 	return err;
1547 }
1548 
1549 struct got_pack *
1550 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1551 {
1552 	struct got_pack *pack = NULL;
1553 	size_t i;
1554 
1555 	for (i = 0; i < repo->pack_cache_size; i++) {
1556 		pack = &repo->packs[i];
1557 		if (pack->path_packfile == NULL)
1558 			break;
1559 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1560 			return pack;
1561 	}
1562 
1563 	return NULL;
1564 }
1565 
1566 const struct got_error *
1567 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1568     struct got_pack *pack)
1569 {
1570 	size_t i;
1571 	int pinned_pack = -1, pinned_packidx = -1;
1572 
1573 	for (i = 0; i < repo->pack_cache_size; i++) {
1574 		if (repo->packidx_cache[i] &&
1575 		    strcmp(repo->packidx_cache[i]->path_packidx,
1576 		    packidx->path_packidx) == 0)
1577 			pinned_packidx = i;
1578 		if (repo->packs[i].path_packfile &&
1579 		    strcmp(repo->packs[i].path_packfile,
1580 		    pack->path_packfile) == 0)
1581 			pinned_pack = i;
1582 	}
1583 
1584 	if (pinned_packidx == -1 || pinned_pack == -1)
1585 		return got_error(GOT_ERR_PIN_PACK);
1586 
1587 	repo->pinned_pack = pinned_pack;
1588 	repo->pinned_packidx = pinned_packidx;
1589 	if (repo->packs[pinned_pack].privsep_child)
1590 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1591 	return NULL;
1592 }
1593 
1594 struct got_pack *
1595 got_repo_get_pinned_pack(struct got_repository *repo)
1596 {
1597 	if (repo->pinned_pack >= 0 &&
1598 	    repo->pinned_pack < repo->pack_cache_size)
1599 		return &repo->packs[repo->pinned_pack];
1600 
1601 	return NULL;
1602 }
1603 
1604 void
1605 got_repo_unpin_pack(struct got_repository *repo)
1606 {
1607 	repo->pinned_packidx = -1;
1608 	repo->pinned_pack = -1;
1609 	repo->pinned_pid = 0;
1610 }
1611 
1612 const struct got_error *
1613 got_repo_init(const char *repo_path, const char *head_name)
1614 {
1615 	const struct got_error *err = NULL;
1616 	const char *dirnames[] = {
1617 		GOT_OBJECTS_DIR,
1618 		GOT_OBJECTS_PACK_DIR,
1619 		GOT_REFS_DIR,
1620 	};
1621 	const char *description_str = "Unnamed repository; "
1622 	    "edit this file 'description' to name the repository.";
1623 	const char *headref = "ref: refs/heads/";
1624 	const char *gitconfig_str = "[core]\n"
1625 	    "\trepositoryformatversion = 0\n"
1626 	    "\tfilemode = true\n"
1627 	    "\tbare = true\n";
1628 	char *headref_str, *path;
1629 	size_t i;
1630 
1631 	if (!got_path_dir_is_empty(repo_path))
1632 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1633 
1634 	for (i = 0; i < nitems(dirnames); i++) {
1635 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1636 			return got_error_from_errno("asprintf");
1637 		}
1638 		err = got_path_mkdir(path);
1639 		free(path);
1640 		if (err)
1641 			return err;
1642 	}
1643 
1644 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1645 		return got_error_from_errno("asprintf");
1646 	err = got_path_create_file(path, description_str);
1647 	free(path);
1648 	if (err)
1649 		return err;
1650 
1651 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1652 		return got_error_from_errno("asprintf");
1653 	if (asprintf(&headref_str, "%s%s", headref,
1654 	    head_name ? head_name : "main") == -1) {
1655 		free(path);
1656 		return got_error_from_errno("asprintf");
1657 	}
1658 	err = got_path_create_file(path, headref_str);
1659 	free(headref_str);
1660 	free(path);
1661 	if (err)
1662 		return err;
1663 
1664 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1665 		return got_error_from_errno("asprintf");
1666 	err = got_path_create_file(path, gitconfig_str);
1667 	free(path);
1668 	if (err)
1669 		return err;
1670 
1671 	return NULL;
1672 }
1673 
1674 static const struct got_error *
1675 match_packed_object(struct got_object_id **unique_id,
1676     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1677 {
1678 	const struct got_error *err = NULL;
1679 	struct got_object_id_queue matched_ids;
1680 	struct got_pathlist_entry *pe;
1681 
1682 	STAILQ_INIT(&matched_ids);
1683 
1684 	err = refresh_packidx_paths(repo);
1685 	if (err)
1686 		return err;
1687 
1688 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1689 		const char *path_packidx = pe->path;
1690 		struct got_packidx *packidx;
1691 		struct got_object_qid *qid;
1692 
1693 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1694 		    path_packidx, 0, repo->algo);
1695 		if (err)
1696 			break;
1697 
1698 		err = got_packidx_match_id_str_prefix(&matched_ids,
1699 		    packidx, id_str_prefix);
1700 		if (err) {
1701 			got_packidx_close(packidx);
1702 			break;
1703 		}
1704 		err = got_packidx_close(packidx);
1705 		if (err)
1706 			break;
1707 
1708 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1709 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1710 				int matched_type;
1711 				err = got_object_get_type(&matched_type, repo,
1712 				    &qid->id);
1713 				if (err)
1714 					goto done;
1715 				if (matched_type != obj_type)
1716 					continue;
1717 			}
1718 			if (*unique_id == NULL) {
1719 				*unique_id = got_object_id_dup(&qid->id);
1720 				if (*unique_id == NULL) {
1721 					err = got_error_from_errno("malloc");
1722 					goto done;
1723 				}
1724 			} else {
1725 				if (got_object_id_cmp(*unique_id,
1726 				    &qid->id) == 0)
1727 					continue; /* packed multiple times */
1728 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1729 				goto done;
1730 			}
1731 		}
1732 	}
1733 done:
1734 	got_object_id_queue_free(&matched_ids);
1735 	if (err) {
1736 		free(*unique_id);
1737 		*unique_id = NULL;
1738 	}
1739 	return err;
1740 }
1741 
1742 static const struct got_error *
1743 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1744     const char *object_dir, const char *id_str_prefix, int obj_type,
1745     struct got_repository *repo)
1746 {
1747 	const struct got_error *err = NULL;
1748 	char *path, *id_str = NULL;
1749 	DIR *dir = NULL;
1750 	struct dirent *dent;
1751 	struct got_object_id id;
1752 
1753 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1754 		err = got_error_from_errno("asprintf");
1755 		goto done;
1756 	}
1757 
1758 	dir = opendir(path);
1759 	if (dir == NULL) {
1760 		if (errno == ENOENT) {
1761 			err = NULL;
1762 			goto done;
1763 		}
1764 		err = got_error_from_errno2("opendir", path);
1765 		goto done;
1766 	}
1767 	while ((dent = readdir(dir)) != NULL) {
1768 		int cmp;
1769 
1770 		free(id_str);
1771 		id_str = NULL;
1772 
1773 		if (strcmp(dent->d_name, ".") == 0 ||
1774 		    strcmp(dent->d_name, "..") == 0)
1775 			continue;
1776 
1777 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1778 			err = got_error_from_errno("asprintf");
1779 			goto done;
1780 		}
1781 
1782 		
1783 		memset(&id, 0, sizeof(id));
1784 		id.algo = repo->algo;
1785 		if (!got_parse_hash_digest(id.hash, id_str, repo->algo))
1786 			continue;
1787 
1788 		/*
1789 		 * Directory entries do not necessarily appear in
1790 		 * sorted order, so we must iterate over all of them.
1791 		 */
1792 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1793 		if (cmp != 0)
1794 			continue;
1795 
1796 		if (*unique_id == NULL) {
1797 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1798 				int matched_type;
1799 				err = got_object_get_type(&matched_type, repo,
1800 				    &id);
1801 				if (err)
1802 					goto done;
1803 				if (matched_type != obj_type)
1804 					continue;
1805 			}
1806 			*unique_id = got_object_id_dup(&id);
1807 			if (*unique_id == NULL) {
1808 				err = got_error_from_errno("got_object_id_dup");
1809 				goto done;
1810 			}
1811 		} else {
1812 			if (got_object_id_cmp(*unique_id, &id) == 0)
1813 				continue; /* both packed and loose */
1814 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1815 			goto done;
1816 		}
1817 	}
1818 done:
1819 	if (dir && closedir(dir) != 0 && err == NULL)
1820 		err = got_error_from_errno("closedir");
1821 	if (err) {
1822 		free(*unique_id);
1823 		*unique_id = NULL;
1824 	}
1825 	free(id_str);
1826 	free(path);
1827 	return err;
1828 }
1829 
1830 const struct got_error *
1831 got_repo_match_object_id_prefix(struct got_object_id **id,
1832     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1833 {
1834 	const struct got_error *err = NULL;
1835 	char *path_objects = NULL, *object_dir = NULL;
1836 	size_t len;
1837 	int i;
1838 
1839 	*id = NULL;
1840 
1841 	path_objects = got_repo_get_path_objects(repo);
1842 
1843 	len = strlen(id_str_prefix);
1844 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1845 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1846 		goto done;
1847 	}
1848 
1849 	for (i = 0; i < len; i++) {
1850 		if (isxdigit((unsigned char)id_str_prefix[i]))
1851 			continue;
1852 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1853 		goto done;
1854 	}
1855 
1856 	if (len >= 2) {
1857 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1858 		if (err)
1859 			goto done;
1860 		object_dir = strndup(id_str_prefix, 2);
1861 		if (object_dir == NULL) {
1862 			err = got_error_from_errno("strdup");
1863 			goto done;
1864 		}
1865 		err = match_loose_object(id, path_objects, object_dir,
1866 		    id_str_prefix, obj_type, repo);
1867 	} else if (len == 1) {
1868 		int i;
1869 		for (i = 0; i < 0xf; i++) {
1870 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1871 			    == -1) {
1872 				err = got_error_from_errno("asprintf");
1873 				goto done;
1874 			}
1875 			err = match_packed_object(id, repo, object_dir,
1876 			    obj_type);
1877 			if (err)
1878 				goto done;
1879 			err = match_loose_object(id, path_objects, object_dir,
1880 			    id_str_prefix, obj_type, repo);
1881 			if (err)
1882 				goto done;
1883 		}
1884 	} else {
1885 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1886 		goto done;
1887 	}
1888 done:
1889 	free(path_objects);
1890 	free(object_dir);
1891 	if (err) {
1892 		free(*id);
1893 		*id = NULL;
1894 	} else if (*id == NULL) {
1895 		switch (obj_type) {
1896 		case GOT_OBJ_TYPE_BLOB:
1897 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1898 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1899 			break;
1900 		case GOT_OBJ_TYPE_TREE:
1901 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1902 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1903 			break;
1904 		case GOT_OBJ_TYPE_COMMIT:
1905 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1906 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1907 			break;
1908 		case GOT_OBJ_TYPE_TAG:
1909 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1910 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1911 			break;
1912 		default:
1913 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1914 			break;
1915 		}
1916 	}
1917 
1918 	return err;
1919 }
1920 
1921 const struct got_error *
1922 got_repo_match_object_id(struct got_object_id **id, char **label,
1923     const char *id_str, int obj_type, struct got_reflist_head *refs,
1924     struct got_repository *repo)
1925 {
1926 	const struct got_error *err;
1927 	struct got_tag_object *tag;
1928 	struct got_reference *ref = NULL;
1929 
1930 	*id = NULL;
1931 	if (label)
1932 		*label = NULL;
1933 
1934 	if (refs) {
1935 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1936 		    refs, repo);
1937 		if (err == NULL) {
1938 			*id = got_object_id_dup(
1939 			    got_object_tag_get_object_id(tag));
1940 			if (*id == NULL)
1941 				err = got_error_from_errno("got_object_id_dup");
1942 			else if (label && asprintf(label, "refs/tags/%s",
1943 			    got_object_tag_get_name(tag)) == -1) {
1944 				err = got_error_from_errno("asprintf");
1945 				free(*id);
1946 				*id = NULL;
1947 			}
1948 			got_object_tag_close(tag);
1949 			return err;
1950 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1951 		    err->code != GOT_ERR_NO_OBJ)
1952 			return err;
1953 	}
1954 
1955 	err = got_ref_open(&ref, repo, id_str, 0);
1956 	if (err == NULL) {
1957 		err = got_ref_resolve(id, repo, ref);
1958 		if (err)
1959 			goto done;
1960 		if (label) {
1961 			*label = strdup(got_ref_get_name(ref));
1962 			if (*label == NULL) {
1963 				err = got_error_from_errno("strdup");
1964 				goto done;
1965 			}
1966 		}
1967 	} else {
1968 		if (err->code != GOT_ERR_NOT_REF &&
1969 		    err->code != GOT_ERR_BAD_REF_NAME)
1970 			goto done;
1971 		err = got_repo_match_object_id_prefix(id, id_str,
1972 		    obj_type, repo);
1973 		if (err) {
1974 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1975 				err = got_error_not_ref(id_str);
1976 			goto done;
1977 		}
1978 		if (label) {
1979 			err = got_object_id_str(label, *id);
1980 			if (*label == NULL) {
1981 				err = got_error_from_errno("strdup");
1982 				goto done;
1983 			}
1984 		}
1985 	}
1986 done:
1987 	if (ref)
1988 		got_ref_close(ref);
1989 	return err;
1990 }
1991 
1992 const struct got_error *
1993 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1994     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1995 {
1996 	const struct got_error *err = NULL;
1997 	struct got_reflist_entry *re;
1998 	struct got_object_id *tag_id;
1999 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
2000 
2001 	*tag = NULL;
2002 
2003 	TAILQ_FOREACH(re, refs, entry) {
2004 		const char *refname;
2005 		refname = got_ref_get_name(re->ref);
2006 		if (got_ref_is_symbolic(re->ref))
2007 			continue;
2008 		if (strncmp(refname, "refs/tags/", 10) != 0)
2009 			continue;
2010 		if (!name_is_absolute)
2011 			refname += strlen("refs/tags/");
2012 		if (strcmp(refname, name) != 0)
2013 			continue;
2014 		err = got_ref_resolve(&tag_id, repo, re->ref);
2015 		if (err)
2016 			break;
2017 		err = got_object_open_as_tag(tag, repo, tag_id);
2018 		free(tag_id);
2019 		if (err)
2020 			break;
2021 		if (obj_type == GOT_OBJ_TYPE_ANY ||
2022 		    got_object_tag_get_object_type(*tag) == obj_type)
2023 			break;
2024 		got_object_tag_close(*tag);
2025 		*tag = NULL;
2026 	}
2027 
2028 	if (err == NULL && *tag == NULL)
2029 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2030 		    GOT_OBJ_LABEL_TAG, name);
2031 	return err;
2032 }
2033 
2034 static const struct got_error *
2035 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2036     const char *name, mode_t mode, struct got_object_id *blob_id)
2037 {
2038 	const struct got_error *err = NULL;
2039 
2040 	 *new_te = NULL;
2041 
2042 	*new_te = calloc(1, sizeof(**new_te));
2043 	if (*new_te == NULL)
2044 		return got_error_from_errno("calloc");
2045 
2046 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2047 	    sizeof((*new_te)->name)) {
2048 		err = got_error(GOT_ERR_NO_SPACE);
2049 		goto done;
2050 	}
2051 
2052 	if (S_ISLNK(mode)) {
2053 		(*new_te)->mode = S_IFLNK;
2054 	} else {
2055 		(*new_te)->mode = S_IFREG;
2056 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2057 	}
2058 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2059 done:
2060 	if (err && *new_te) {
2061 		free(*new_te);
2062 		*new_te = NULL;
2063 	}
2064 	return err;
2065 }
2066 
2067 static const struct got_error *
2068 import_file(struct got_tree_entry **new_te, struct dirent *de,
2069     const char *path, struct got_repository *repo)
2070 {
2071 	const struct got_error *err;
2072 	struct got_object_id *blob_id = NULL;
2073 	char *filepath;
2074 	struct stat sb;
2075 
2076 	if (asprintf(&filepath, "%s%s%s", path,
2077 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2078 		return got_error_from_errno("asprintf");
2079 
2080 	if (lstat(filepath, &sb) != 0) {
2081 		err = got_error_from_errno2("lstat", path);
2082 		goto done;
2083 	}
2084 
2085 	err = got_object_blob_create(&blob_id, filepath, repo);
2086 	if (err)
2087 		goto done;
2088 
2089 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2090 	    blob_id);
2091 done:
2092 	free(filepath);
2093 	if (err)
2094 		free(blob_id);
2095 	return err;
2096 }
2097 
2098 static const struct got_error *
2099 insert_tree_entry(struct got_tree_entry *new_te,
2100     struct got_pathlist_head *paths)
2101 {
2102 	const struct got_error *err = NULL;
2103 	struct got_pathlist_entry *new_pe;
2104 
2105 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2106 	if (err)
2107 		return err;
2108 	if (new_pe == NULL)
2109 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2110 	return NULL;
2111 }
2112 
2113 static const struct got_error *write_tree(struct got_object_id **,
2114     const char *, struct got_pathlist_head *, struct got_repository *,
2115     got_repo_import_cb progress_cb, void *progress_arg);
2116 
2117 static const struct got_error *
2118 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2119     const char *path, struct got_pathlist_head *ignores,
2120     struct got_repository *repo,
2121     got_repo_import_cb progress_cb, void *progress_arg)
2122 {
2123 	const struct got_error *err;
2124 	struct got_object_id *id = NULL;
2125 	char *subdirpath;
2126 
2127 	if (asprintf(&subdirpath, "%s%s%s", path,
2128 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2129 		return got_error_from_errno("asprintf");
2130 
2131 	(*new_te) = calloc(1, sizeof(**new_te));
2132 	if (*new_te == NULL)
2133 		return got_error_from_errno("calloc");
2134 	(*new_te)->mode = S_IFDIR;
2135 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2136 	    sizeof((*new_te)->name)) {
2137 		err = got_error(GOT_ERR_NO_SPACE);
2138 		goto done;
2139 	}
2140 	err = write_tree(&id, subdirpath, ignores,  repo,
2141 	    progress_cb, progress_arg);
2142 	if (err)
2143 		goto done;
2144 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2145 
2146 done:
2147 	free(id);
2148 	free(subdirpath);
2149 	if (err) {
2150 		free(*new_te);
2151 		*new_te = NULL;
2152 	}
2153 	return err;
2154 }
2155 
2156 static const struct got_error *
2157 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2158     struct got_pathlist_head *ignores, struct got_repository *repo,
2159     got_repo_import_cb progress_cb, void *progress_arg)
2160 {
2161 	const struct got_error *err = NULL;
2162 	DIR *dir;
2163 	struct dirent *de;
2164 	int nentries;
2165 	struct got_tree_entry *new_te = NULL;
2166 	struct got_pathlist_head paths;
2167 	struct got_pathlist_entry *pe;
2168 
2169 	*new_tree_id = NULL;
2170 
2171 	TAILQ_INIT(&paths);
2172 
2173 	dir = opendir(path_dir);
2174 	if (dir == NULL) {
2175 		err = got_error_from_errno2("opendir", path_dir);
2176 		goto done;
2177 	}
2178 
2179 	nentries = 0;
2180 	while ((de = readdir(dir)) != NULL) {
2181 		int ignore = 0;
2182 		int type;
2183 
2184 		if (strcmp(de->d_name, ".") == 0 ||
2185 		    strcmp(de->d_name, "..") == 0)
2186 			continue;
2187 
2188 		err = got_path_dirent_type(&type, path_dir, de);
2189 		if (err)
2190 			goto done;
2191 
2192 		TAILQ_FOREACH(pe, ignores, entry) {
2193 			if (type == DT_DIR && pe->path_len > 0 &&
2194 			    pe->path[pe->path_len - 1] == '/') {
2195 				char stripped[PATH_MAX];
2196 
2197 				if (strlcpy(stripped, pe->path,
2198 				    sizeof(stripped)) >= sizeof(stripped)) {
2199 					err = got_error(GOT_ERR_NO_SPACE);
2200 					goto done;
2201 				}
2202 				got_path_strip_trailing_slashes(stripped);
2203 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2204 					ignore = 1;
2205 					break;
2206 				}
2207 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2208 				ignore = 1;
2209 				break;
2210 			}
2211 		}
2212 		if (ignore)
2213 			continue;
2214 
2215 		if (type == DT_DIR) {
2216 			err = import_subdir(&new_te, de, path_dir,
2217 			    ignores, repo, progress_cb, progress_arg);
2218 			if (err) {
2219 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2220 					goto done;
2221 				err = NULL;
2222 				continue;
2223 			}
2224 		} else if (type == DT_REG || type == DT_LNK) {
2225 			err = import_file(&new_te, de, path_dir, repo);
2226 			if (err)
2227 				goto done;
2228 		} else
2229 			continue;
2230 
2231 		err = insert_tree_entry(new_te, &paths);
2232 		if (err)
2233 			goto done;
2234 		nentries++;
2235 	}
2236 
2237 	if (TAILQ_EMPTY(&paths)) {
2238 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2239 		    "cannot create tree without any entries");
2240 		goto done;
2241 	}
2242 
2243 	TAILQ_FOREACH(pe, &paths, entry) {
2244 		struct got_tree_entry *te = pe->data;
2245 		char *path;
2246 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2247 			continue;
2248 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2249 			err = got_error_from_errno("asprintf");
2250 			goto done;
2251 		}
2252 		err = (*progress_cb)(progress_arg, path);
2253 		free(path);
2254 		if (err)
2255 			goto done;
2256 	}
2257 
2258 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2259 done:
2260 	if (dir)
2261 		closedir(dir);
2262 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2263 	return err;
2264 }
2265 
2266 const struct got_error *
2267 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2268     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2269     struct got_repository *repo, got_repo_import_cb progress_cb,
2270     void *progress_arg)
2271 {
2272 	const struct got_error *err;
2273 	struct got_object_id *new_tree_id;
2274 
2275 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2276 	    progress_cb, progress_arg);
2277 	if (err)
2278 		return err;
2279 
2280 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2281 	    author, time(NULL), author, time(NULL), logmsg, repo);
2282 	free(new_tree_id);
2283 	return err;
2284 }
2285 
2286 const struct got_error *
2287 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2288     struct got_repository *repo)
2289 {
2290 	const struct got_error *err = NULL;
2291 	char *path_objects = NULL, *path = NULL;
2292 	DIR *dir = NULL;
2293 	struct got_object_id id;
2294 	int i;
2295 
2296 	*nobjects = 0;
2297 	*ondisk_size = 0;
2298 
2299 	path_objects = got_repo_get_path_objects(repo);
2300 	if (path_objects == NULL)
2301 		return got_error_from_errno("got_repo_get_path_objects");
2302 
2303 	for (i = 0; i <= 0xff; i++) {
2304 		struct dirent *dent;
2305 
2306 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2307 			err = got_error_from_errno("asprintf");
2308 			break;
2309 		}
2310 
2311 		dir = opendir(path);
2312 		if (dir == NULL) {
2313 			if (errno == ENOENT) {
2314 				err = NULL;
2315 				continue;
2316 			}
2317 			err = got_error_from_errno2("opendir", path);
2318 			break;
2319 		}
2320 
2321 		while ((dent = readdir(dir)) != NULL) {
2322 			char *id_str;
2323 			int fd;
2324 			struct stat sb;
2325 
2326 			if (strcmp(dent->d_name, ".") == 0 ||
2327 			    strcmp(dent->d_name, "..") == 0)
2328 				continue;
2329 
2330 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2331 				err = got_error_from_errno("asprintf");
2332 				goto done;
2333 			}
2334 
2335 			memset(&id, 0, sizeof(id));
2336 			id.algo = repo->algo;
2337 			if (!got_parse_hash_digest(id.hash, id_str,
2338 			    repo->algo)) {
2339 				free(id_str);
2340 				continue;
2341 			}
2342 			free(id_str);
2343 
2344 			err = got_object_open_loose_fd(&fd, &id, repo);
2345 			if (err)
2346 				goto done;
2347 
2348 			if (fstat(fd, &sb) == -1) {
2349 				err = got_error_from_errno("fstat");
2350 				close(fd);
2351 				goto done;
2352 			}
2353 			(*nobjects)++;
2354 			(*ondisk_size) += sb.st_size;
2355 
2356 			if (close(fd) == -1) {
2357 				err = got_error_from_errno("close");
2358 				goto done;
2359 			}
2360 		}
2361 
2362 		if (closedir(dir) != 0) {
2363 			err = got_error_from_errno("closedir");
2364 			goto done;
2365 		}
2366 		dir = NULL;
2367 
2368 		free(path);
2369 		path = NULL;
2370 	}
2371 done:
2372 	if (dir && closedir(dir) != 0 && err == NULL)
2373 		err = got_error_from_errno("closedir");
2374 
2375 	if (err) {
2376 		*nobjects = 0;
2377 		*ondisk_size = 0;
2378 	}
2379 	free(path_objects);
2380 	free(path);
2381 	return err;
2382 }
2383 
2384 const struct got_error *
2385 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2386     off_t *total_packsize, struct got_repository *repo)
2387 {
2388 	const struct got_error *err = NULL;
2389 	DIR *packdir = NULL;
2390 	struct dirent *dent;
2391 	struct got_packidx *packidx = NULL;
2392 	char *path_packidx;
2393 	char *path_packfile;
2394 	int packdir_fd;
2395 	struct stat sb;
2396 
2397 	*npackfiles = 0;
2398 	*nobjects = 0;
2399 	*total_packsize = 0;
2400 
2401 	packdir_fd = openat(got_repo_get_fd(repo),
2402 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2403 	if (packdir_fd == -1) {
2404 		return got_error_from_errno_fmt("openat: %s/%s",
2405 		    got_repo_get_path_git_dir(repo),
2406 		    GOT_OBJECTS_PACK_DIR);
2407 	}
2408 
2409 	packdir = fdopendir(packdir_fd);
2410 	if (packdir == NULL) {
2411 		err = got_error_from_errno("fdopendir");
2412 		goto done;
2413 	}
2414 
2415 	while ((dent = readdir(packdir)) != NULL) {
2416 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
2417 		    repo->algo))
2418 			continue;
2419 
2420 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2421 		    dent->d_name) == -1) {
2422 			err = got_error_from_errno("asprintf");
2423 			goto done;
2424 		}
2425 
2426 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2427 		    path_packidx, 0, repo->algo);
2428 		free(path_packidx);
2429 		if (err)
2430 			goto done;
2431 
2432 		if (fstat(packidx->fd, &sb) == -1)
2433 			goto done;
2434 		*total_packsize += sb.st_size;
2435 
2436 		err = got_packidx_get_packfile_path(&path_packfile,
2437 		    packidx->path_packidx);
2438 		if (err)
2439 			goto done;
2440 
2441 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2442 		    0) == -1) {
2443 			free(path_packfile);
2444 			goto done;
2445 		}
2446 		free(path_packfile);
2447 		*total_packsize += sb.st_size;
2448 
2449 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2450 
2451 		(*npackfiles)++;
2452 
2453 		got_packidx_close(packidx);
2454 		packidx = NULL;
2455 	}
2456 done:
2457 	if (packidx)
2458 		got_packidx_close(packidx);
2459 	if (packdir && closedir(packdir) != 0 && err == NULL)
2460 		err = got_error_from_errno("closedir");
2461 	if (err) {
2462 		*npackfiles = 0;
2463 		*nobjects = 0;
2464 		*total_packsize = 0;
2465 	}
2466 	return err;
2467 }
2468 
2469 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2470     got_packidx_bloom_filter_cmp);