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 const char *
79 got_repo_get_path(struct got_repository *repo)
80 {
81 	return repo->path;
82 }
83 
84 const char *
85 got_repo_get_path_git_dir(struct got_repository *repo)
86 {
87 	return repo->path_git_dir;
88 }
89 
90 int
91 got_repo_get_fd(struct got_repository *repo)
92 {
93 	return repo->gitdir_fd;
94 }
95 
96 const char *
97 got_repo_get_gitconfig_author_name(struct got_repository *repo)
98 {
99 	return repo->gitconfig_author_name;
100 }
101 
102 const char *
103 got_repo_get_gitconfig_author_email(struct got_repository *repo)
104 {
105 	return repo->gitconfig_author_email;
106 }
107 
108 const char *
109 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
110 {
111 	return repo->global_gitconfig_author_name;
112 }
113 
114 const char *
115 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
116 {
117 	return repo->global_gitconfig_author_email;
118 }
119 
120 const char *
121 got_repo_get_gitconfig_owner(struct got_repository *repo)
122 {
123 	return repo->gitconfig_owner;
124 }
125 
126 void
127 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
128     struct got_repository *repo)
129 {
130 	*extensions = repo->extensions;
131 	*nextensions = repo->nextensions;
132 }
133 
134 int
135 got_repo_is_bare(struct got_repository *repo)
136 {
137 	return (strcmp(repo->path, repo->path_git_dir) == 0);
138 }
139 
140 static char *
141 get_path_git_child(struct got_repository *repo, const char *basename)
142 {
143 	char *path_child;
144 
145 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
146 	    basename) == -1)
147 		return NULL;
148 
149 	return path_child;
150 }
151 
152 char *
153 got_repo_get_path_objects(struct got_repository *repo)
154 {
155 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
156 }
157 
158 char *
159 got_repo_get_path_objects_pack(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
162 }
163 
164 char *
165 got_repo_get_path_refs(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_REFS_DIR);
168 }
169 
170 char *
171 got_repo_get_path_packed_refs(struct got_repository *repo)
172 {
173 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
174 }
175 
176 static char *
177 get_path_head(struct got_repository *repo)
178 {
179 	return get_path_git_child(repo, GOT_HEAD_FILE);
180 }
181 
182 char *
183 got_repo_get_path_gitconfig(struct got_repository *repo)
184 {
185 	return get_path_git_child(repo, GOT_GITCONFIG);
186 }
187 
188 char *
189 got_repo_get_path_gotconfig(struct got_repository *repo)
190 {
191 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
192 }
193 
194 const struct got_gotconfig *
195 got_repo_get_gotconfig(struct got_repository *repo)
196 {
197 	return repo->gotconfig;
198 }
199 
200 void
201 got_repo_get_gitconfig_remotes(int *nremotes,
202     const struct got_remote_repo **remotes, struct got_repository *repo)
203 {
204 	*nremotes = repo->ngitconfig_remotes;
205 	*remotes = repo->gitconfig_remotes;
206 }
207 
208 static int
209 is_git_repo(struct got_repository *repo)
210 {
211 	const char *path_git = got_repo_get_path_git_dir(repo);
212 	char *path_objects = got_repo_get_path_objects(repo);
213 	char *path_refs = got_repo_get_path_refs(repo);
214 	char *path_head = get_path_head(repo);
215 	int ret = 0;
216 	struct stat sb;
217 	struct got_reference *head_ref;
218 
219 	if (lstat(path_git, &sb) == -1)
220 		goto done;
221 	if (!S_ISDIR(sb.st_mode))
222 		goto done;
223 
224 	if (lstat(path_objects, &sb) == -1)
225 		goto done;
226 	if (!S_ISDIR(sb.st_mode))
227 		goto done;
228 
229 	if (lstat(path_refs, &sb) == -1)
230 		goto done;
231 	if (!S_ISDIR(sb.st_mode))
232 		goto done;
233 
234 	if (lstat(path_head, &sb) == -1)
235 		goto done;
236 	if (!S_ISREG(sb.st_mode))
237 		goto done;
238 
239 	/* Check if the HEAD reference can be opened. */
240 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
241 		goto done;
242 	got_ref_close(head_ref);
243 
244 	ret = 1;
245 done:
246 	free(path_objects);
247 	free(path_refs);
248 	free(path_head);
249 	return ret;
250 
251 }
252 
253 static const struct got_error *
254 close_tempfiles(int *fds, size_t nfds)
255 {
256 	const struct got_error *err = NULL;
257 	int i;
258 
259 	for (i = 0; i < nfds; i++) {
260 		if (fds[i] == -1)
261 			continue;
262 		if (close(fds[i]) == -1) {
263 			err = got_error_from_errno("close");
264 			break;
265 		}
266 	}
267 	free(fds);
268 	return err;
269 }
270 
271 static const struct got_error *
272 open_tempfiles(int **fds, size_t array_size, size_t nfds)
273 {
274 	const struct got_error *err = NULL;
275 	int i;
276 
277 	*fds = calloc(array_size, sizeof(**fds));
278 	if (*fds == NULL)
279 		return got_error_from_errno("calloc");
280 
281 	for (i = 0; i < array_size; i++)
282 		(*fds)[i] = -1;
283 
284 	for (i = 0; i < nfds; i++) {
285 		(*fds)[i] = got_opentempfd();
286 		if ((*fds)[i] == -1) {
287 			err = got_error_from_errno("got_opentempfd");
288 			close_tempfiles(*fds, nfds);
289 			*fds = NULL;
290 			return err;
291 		}
292 	}
293 
294 	return NULL;
295 }
296 
297 static const struct got_error *
298 get_pack_cache_size(int *pack_cache_size)
299 {
300 	struct rlimit rl;
301 
302 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
303 		return got_error_from_errno("getrlimit");
304 
305 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
306 	if (*pack_cache_size > rl.rlim_cur / 8)
307 		*pack_cache_size = rl.rlim_cur / 8;
308 
309 	return NULL;
310 }
311 
312 const struct got_error *
313 got_repo_pack_fds_open(int **pack_fds)
314 {
315 	const struct got_error *err;
316 	int nfds;
317 
318 	err = get_pack_cache_size(&nfds);
319 	if (err)
320 		return err;
321 
322 	/*
323 	 * We need one basefd and one accumfd per cached pack.
324 	 * Our constants should be set up in a way such that
325 	 * this error never triggers.
326 	 */
327 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
328 		return got_error(GOT_ERR_NO_SPACE);
329 
330 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
331 }
332 
333 const struct got_error *
334 got_repo_pack_fds_close(int *pack_fds)
335 {
336 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
337 }
338 
339 const struct got_error *
340 got_repo_temp_fds_open(int **temp_fds)
341 {
342 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
343 	    GOT_REPO_NUM_TEMPFILES);
344 }
345 
346 void
347 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
348 {
349 	int i;
350 
351 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
352 		repo->tempfiles[i] = temp_fds[i];
353 }
354 
355 const struct got_error *
356 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
357 {
358 	int i;
359 
360 	*fd = -1;
361 	*idx = -1;
362 
363 	for (i = 0; i < nitems(repo->tempfiles); i++) {
364 		if (repo->tempfile_use_mask & (1 << i))
365 			continue;
366 		if (repo->tempfiles[i] != -1) {
367 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
368 				return got_error_from_errno("ftruncate");
369 			*fd = repo->tempfiles[i];
370 			*idx = i;
371 			repo->tempfile_use_mask |= (1 << i);
372 			return NULL;
373 		}
374 	}
375 
376 	return got_error(GOT_ERR_REPO_TEMPFILE);
377 }
378 
379 void
380 got_repo_temp_fds_put(int idx, struct got_repository *repo)
381 {
382 	repo->tempfile_use_mask &= ~(1 << idx);
383 }
384 
385 const struct got_error *
386 got_repo_temp_fds_close(int *temp_fds)
387 {
388 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
389 }
390 
391 const struct got_error *
392 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
393     struct got_object *obj)
394 {
395 #ifndef GOT_NO_OBJ_CACHE
396 	const struct got_error *err = NULL;
397 	err = got_object_cache_add(&repo->objcache, id, obj);
398 	if (err) {
399 		if (err->code == GOT_ERR_OBJ_EXISTS ||
400 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
401 			err = NULL;
402 		return err;
403 	}
404 	obj->refcnt++;
405 #endif
406 	return NULL;
407 }
408 
409 struct got_object *
410 got_repo_get_cached_object(struct got_repository *repo,
411     struct got_object_id *id)
412 {
413 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
414 }
415 
416 const struct got_error *
417 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
418     struct got_tree_object *tree)
419 {
420 #ifndef GOT_NO_OBJ_CACHE
421 	const struct got_error *err = NULL;
422 	err = got_object_cache_add(&repo->treecache, id, tree);
423 	if (err) {
424 		if (err->code == GOT_ERR_OBJ_EXISTS ||
425 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
426 			err = NULL;
427 		return err;
428 	}
429 	tree->refcnt++;
430 #endif
431 	return NULL;
432 }
433 
434 struct got_tree_object *
435 got_repo_get_cached_tree(struct got_repository *repo,
436     struct got_object_id *id)
437 {
438 	return (struct got_tree_object *)got_object_cache_get(
439 	    &repo->treecache, id);
440 }
441 
442 const struct got_error *
443 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
444     struct got_commit_object *commit)
445 {
446 #ifndef GOT_NO_OBJ_CACHE
447 	const struct got_error *err = NULL;
448 	err = got_object_cache_add(&repo->commitcache, id, commit);
449 	if (err) {
450 		if (err->code == GOT_ERR_OBJ_EXISTS ||
451 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
452 			err = NULL;
453 		return err;
454 	}
455 	commit->refcnt++;
456 #endif
457 	return NULL;
458 }
459 
460 struct got_commit_object *
461 got_repo_get_cached_commit(struct got_repository *repo,
462     struct got_object_id *id)
463 {
464 	return (struct got_commit_object *)got_object_cache_get(
465 	    &repo->commitcache, id);
466 }
467 
468 const struct got_error *
469 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
470     struct got_tag_object *tag)
471 {
472 #ifndef GOT_NO_OBJ_CACHE
473 	const struct got_error *err = NULL;
474 	err = got_object_cache_add(&repo->tagcache, id, tag);
475 	if (err) {
476 		if (err->code == GOT_ERR_OBJ_EXISTS ||
477 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
478 			err = NULL;
479 		return err;
480 	}
481 	tag->refcnt++;
482 #endif
483 	return NULL;
484 }
485 
486 struct got_tag_object *
487 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
488 {
489 	return (struct got_tag_object *)got_object_cache_get(
490 	    &repo->tagcache, id);
491 }
492 
493 const struct got_error *
494 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
495     struct got_raw_object *raw)
496 {
497 #ifndef GOT_NO_OBJ_CACHE
498 	const struct got_error *err = NULL;
499 	err = got_object_cache_add(&repo->rawcache, id, raw);
500 	if (err) {
501 		if (err->code == GOT_ERR_OBJ_EXISTS ||
502 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
503 			err = NULL;
504 		return err;
505 	}
506 	raw->refcnt++;
507 #endif
508 	return NULL;
509 }
510 
511 
512 struct got_raw_object *
513 got_repo_get_cached_raw_object(struct got_repository *repo,
514     struct got_object_id *id)
515 {
516 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
517 }
518 
519 
520 static const struct got_error *
521 open_repo(struct got_repository *repo, const char *path)
522 {
523 	const struct got_error *err = NULL;
524 
525 	repo->gitdir_fd = -1;
526 
527 	/* bare git repository? */
528 	repo->path_git_dir = strdup(path);
529 	if (repo->path_git_dir == NULL)
530 		return got_error_from_errno("strdup");
531 	if (is_git_repo(repo)) {
532 		repo->path = strdup(repo->path_git_dir);
533 		if (repo->path == NULL) {
534 			err = got_error_from_errno("strdup");
535 			goto done;
536 		}
537 		repo->gitdir_fd = open(repo->path_git_dir,
538 		    O_DIRECTORY | O_CLOEXEC);
539 		if (repo->gitdir_fd == -1) {
540 			err = got_error_from_errno2("open",
541 			    repo->path_git_dir);
542 			goto done;
543 		}
544 		return NULL;
545 	}
546 
547 	/* git repository with working tree? */
548 	free(repo->path_git_dir);
549 	repo->path_git_dir = NULL;
550 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
551 		err = got_error_from_errno("asprintf");
552 		goto done;
553 	}
554 	if (is_git_repo(repo)) {
555 		repo->path = strdup(path);
556 		if (repo->path == NULL) {
557 			err = got_error_from_errno("strdup");
558 			goto done;
559 		}
560 		repo->gitdir_fd = open(repo->path_git_dir,
561 		    O_DIRECTORY | O_CLOEXEC);
562 		if (repo->gitdir_fd == -1) {
563 			err = got_error_from_errno2("open",
564 			    repo->path_git_dir);
565 			goto done;
566 		}
567 		return NULL;
568 	}
569 
570 	err = got_error(GOT_ERR_NOT_GIT_REPO);
571 done:
572 	if (err) {
573 		free(repo->path);
574 		repo->path = NULL;
575 		free(repo->path_git_dir);
576 		repo->path_git_dir = NULL;
577 		if (repo->gitdir_fd != -1)
578 			close(repo->gitdir_fd);
579 		repo->gitdir_fd = -1;
580 
581 	}
582 	return err;
583 }
584 
585 static const struct got_error *
586 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
587 {
588 	const struct got_error *err = NULL;
589 	char *repo_gitconfig_path = NULL;
590 	char *object_format = NULL;
591 
592 	if (global_gitconfig_path) {
593 		/* Read settings from ~/.gitconfig. */
594 		int dummy_repo_version;
595 		err = got_repo_read_gitconfig(&dummy_repo_version,
596 		    &repo->global_gitconfig_author_name,
597 		    &repo->global_gitconfig_author_email,
598 		    NULL, NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
599 		if (err)
600 			return err;
601 	}
602 
603 	/* Read repository's .git/config file. */
604 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
605 	if (repo_gitconfig_path == NULL)
606 		return got_error_from_errno("got_repo_get_path_gitconfig");
607 
608 	err = got_repo_read_gitconfig(
609 	    &repo->gitconfig_repository_format_version,
610 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
611 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
612 	    &repo->gitconfig_owner, &object_format,
613 	    &repo->extensions, &repo->nextensions,
614 	    repo_gitconfig_path);
615 	if (err)
616 		goto done;
617 
618 	if (!object_format)
619 		repo->algo = GOT_HASH_SHA1;
620 	else {
621 		if (!strcmp(object_format, "sha256"))
622 			repo->algo = GOT_HASH_SHA256;
623 		else {
624 			err = got_error_msg(GOT_ERR_OBJECT_FORMAT,
625 			    object_format);
626 			goto done;
627 		}
628 	}
629 
630 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
631 		int i;
632 
633 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
634 			got_repo_free_remote_repo_data(
635 			    &repo->gitconfig_remotes[i]);
636 		}
637 		free(repo->gitconfig_remotes);
638 		repo->gitconfig_remotes = NULL;
639 		repo->ngitconfig_remotes = 0;
640 
641 		free(repo->gitconfig_author_name);
642 		repo->gitconfig_author_name = NULL;
643 		free(repo->gitconfig_author_email);
644 		repo->gitconfig_author_email = NULL;
645 
646 		free(repo->global_gitconfig_author_name);
647 		repo->global_gitconfig_author_name = NULL;
648 		free(repo->global_gitconfig_author_email);
649 		repo->global_gitconfig_author_email = NULL;
650 	}
651 
652 done:
653 	free(object_format);
654 	free(repo_gitconfig_path);
655 	return err;
656 }
657 
658 static const struct got_error *
659 read_gotconfig(struct got_repository *repo)
660 {
661 	const struct got_error *err = NULL;
662 	char *gotconfig_path;
663 
664 	gotconfig_path = got_repo_get_path_gotconfig(repo);
665 	if (gotconfig_path == NULL)
666 		return got_error_from_errno("got_repo_get_path_gotconfig");
667 
668 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
669 	free(gotconfig_path);
670 	return err;
671 }
672 
673 /* Supported repository format extensions. */
674 static const char *const repo_extensions[] = {
675 	"noop",			/* Got supports repository format version 1. */
676 	"preciousObjects",	/* Supported by gotadmin cleanup. */
677 	"worktreeConfig",	/* Got does not care about Git work trees. */
678 };
679 
680 const struct got_error *
681 got_repo_open(struct got_repository **repop, const char *path,
682     const char *global_gitconfig_path, int *pack_fds)
683 {
684 	struct got_repository *repo = NULL;
685 	const struct got_error *err = NULL;
686 	char *repo_path = NULL;
687 	size_t i, j = 0;
688 
689 	*repop = NULL;
690 
691 	repo = calloc(1, sizeof(*repo));
692 	if (repo == NULL)
693 		return got_error_from_errno("calloc");
694 
695 	RB_INIT(&repo->packidx_bloom_filters);
696 	TAILQ_INIT(&repo->packidx_paths);
697 
698 	for (i = 0; i < nitems(repo->privsep_children); i++) {
699 		memset(&repo->privsep_children[i], 0,
700 		    sizeof(repo->privsep_children[0]));
701 		repo->privsep_children[i].imsg_fd = -1;
702 	}
703 
704 	err = got_object_cache_init(&repo->objcache,
705 	    GOT_OBJECT_CACHE_TYPE_OBJ);
706 	if (err)
707 		goto done;
708 	err = got_object_cache_init(&repo->treecache,
709 	    GOT_OBJECT_CACHE_TYPE_TREE);
710 	if (err)
711 		goto done;
712 	err = got_object_cache_init(&repo->commitcache,
713 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
714 	if (err)
715 		goto done;
716 	err = got_object_cache_init(&repo->tagcache,
717 	    GOT_OBJECT_CACHE_TYPE_TAG);
718 	if (err)
719 		goto done;
720 	err = got_object_cache_init(&repo->rawcache,
721 	    GOT_OBJECT_CACHE_TYPE_RAW);
722 	if (err)
723 		goto done;
724 
725 	err = get_pack_cache_size(&repo->pack_cache_size);
726 	if (err)
727 		goto done;
728 	for (i = 0; i < nitems(repo->packs); i++) {
729 		if (pack_fds != NULL && i < repo->pack_cache_size) {
730 			repo->packs[i].basefd = pack_fds[j++];
731 			repo->packs[i].accumfd = pack_fds[j++];
732 		} else {
733 			repo->packs[i].basefd = -1;
734 			repo->packs[i].accumfd = -1;
735 		}
736 	}
737 	for (i = 0; i < nitems(repo->tempfiles); i++)
738 		repo->tempfiles[i] = -1;
739 	repo->pinned_pack = -1;
740 	repo->pinned_packidx = -1;
741 	repo->pinned_pid = 0;
742 
743 	repo_path = realpath(path, NULL);
744 	if (repo_path == NULL) {
745 		err = got_error_from_errno2("realpath", path);
746 		goto done;
747 	}
748 
749 	for (;;) {
750 		char *parent_path;
751 
752 		err = open_repo(repo, repo_path);
753 		if (err == NULL)
754 			break;
755 		if (err->code != GOT_ERR_NOT_GIT_REPO)
756 			goto done;
757 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
758 			err = got_error(GOT_ERR_NOT_GIT_REPO);
759 			goto done;
760 		}
761 		err = got_path_dirname(&parent_path, repo_path);
762 		if (err)
763 			goto done;
764 		free(repo_path);
765 		repo_path = parent_path;
766 	}
767 
768 	err = read_gotconfig(repo);
769 	if (err)
770 		goto done;
771 
772 	err = read_gitconfig(repo, global_gitconfig_path);
773 	if (err)
774 		goto done;
775 	if (repo->gitconfig_repository_format_version > 1) {
776 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
777 		goto done;
778 	}
779 	if (repo->gitconfig_repository_format_version == 0 &&
780 	    repo->algo != GOT_HASH_SHA1) {
781 		err = got_error(GOT_ERR_OBJECT_FORMAT);
782 		goto done;
783 	}
784 	for (i = 0; i < repo->nextensions; i++) {
785 		char *ext = repo->extensions[i];
786 		int j, supported = 0;
787 		for (j = 0; j < nitems(repo_extensions); j++) {
788 			if (strcmp(ext, repo_extensions[j]) == 0) {
789 				supported = 1;
790 				break;
791 			}
792 		}
793 		if (!supported) {
794 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
795 			goto done;
796 		}
797 	}
798 
799 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
800 done:
801 	if (err)
802 		got_repo_close(repo);
803 	else
804 		*repop = repo;
805 	free(repo_path);
806 	return err;
807 }
808 
809 const struct got_error *
810 got_repo_close(struct got_repository *repo)
811 {
812 	const struct got_error *err = NULL, *child_err;
813 	struct got_packidx_bloom_filter *bf;
814 	size_t i;
815 
816 	for (i = 0; i < repo->pack_cache_size; i++) {
817 		if (repo->packidx_cache[i] == NULL)
818 			break;
819 		got_packidx_close(repo->packidx_cache[i]);
820 	}
821 
822 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
823 	    &repo->packidx_bloom_filters))) {
824 		RB_REMOVE(got_packidx_bloom_filter_tree,
825 		    &repo->packidx_bloom_filters, bf);
826 		bloom_free(bf->bloom);
827 		free(bf->bloom);
828 		free(bf);
829 	}
830 
831 	for (i = 0; i < repo->pack_cache_size; i++)
832 		if (repo->packs[i].path_packfile)
833 			if (repo->packs[i].path_packfile)
834 				got_pack_close(&repo->packs[i]);
835 
836 	free(repo->path);
837 	free(repo->path_git_dir);
838 
839 	got_object_cache_close(&repo->objcache);
840 	got_object_cache_close(&repo->treecache);
841 	got_object_cache_close(&repo->commitcache);
842 	got_object_cache_close(&repo->tagcache);
843 	got_object_cache_close(&repo->rawcache);
844 
845 	for (i = 0; i < nitems(repo->privsep_children); i++) {
846 		if (repo->privsep_children[i].imsg_fd == -1)
847 			continue;
848 		imsg_clear(repo->privsep_children[i].ibuf);
849 		free(repo->privsep_children[i].ibuf);
850 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
851 		child_err = got_privsep_wait_for_child(
852 		    repo->privsep_children[i].pid);
853 		if (child_err && err == NULL)
854 			err = child_err;
855 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
856 		    err == NULL)
857 			err = got_error_from_errno("close");
858 	}
859 
860 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
861 	    err == NULL)
862 		err = got_error_from_errno("close");
863 
864 	if (repo->gotconfig)
865 		got_gotconfig_free(repo->gotconfig);
866 	free(repo->gitconfig_author_name);
867 	free(repo->gitconfig_author_email);
868 	for (i = 0; i < repo->ngitconfig_remotes; i++)
869 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
870 	free(repo->gitconfig_remotes);
871 	for (i = 0; i < repo->nextensions; i++)
872 		free(repo->extensions[i]);
873 	free(repo->extensions);
874 
875 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
876 	free(repo);
877 
878 	return err;
879 }
880 
881 void
882 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
883 {
884 	int i;
885 
886 	free(repo->name);
887 	repo->name = NULL;
888 	free(repo->fetch_url);
889 	repo->fetch_url = NULL;
890 	free(repo->send_url);
891 	repo->send_url = NULL;
892 	for (i = 0; i < repo->nfetch_branches; i++)
893 		free(repo->fetch_branches[i]);
894 	free(repo->fetch_branches);
895 	repo->fetch_branches = NULL;
896 	repo->nfetch_branches = 0;
897 	for (i = 0; i < repo->nsend_branches; i++)
898 		free(repo->send_branches[i]);
899 	free(repo->send_branches);
900 	repo->send_branches = NULL;
901 	repo->nsend_branches = 0;
902 }
903 
904 const struct got_error *
905 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
906     const char *input_path)
907 {
908 	const struct got_error *err = NULL;
909 	const char *repo_abspath = NULL;
910 	size_t repolen, len;
911 	char *canonpath, *path = NULL;
912 
913 	*in_repo_path = NULL;
914 
915 	canonpath = strdup(input_path);
916 	if (canonpath == NULL) {
917 		err = got_error_from_errno("strdup");
918 		goto done;
919 	}
920 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
921 	if (err)
922 		goto done;
923 
924 	repo_abspath = got_repo_get_path(repo);
925 
926 	if (canonpath[0] == '\0') {
927 		path = strdup(canonpath);
928 		if (path == NULL) {
929 			err = got_error_from_errno("strdup");
930 			goto done;
931 		}
932 	} else {
933 		path = realpath(canonpath, NULL);
934 		if (path == NULL) {
935 			if (errno != ENOENT) {
936 				err = got_error_from_errno2("realpath",
937 				    canonpath);
938 				goto done;
939 			}
940 			/*
941 			 * Path is not on disk.
942 			 * Assume it is already relative to repository root.
943 			 */
944 			path = strdup(canonpath);
945 			if (path == NULL) {
946 				err = got_error_from_errno("strdup");
947 				goto done;
948 			}
949 		}
950 
951 		repolen = strlen(repo_abspath);
952 		len = strlen(path);
953 
954 
955 		if (strcmp(path, repo_abspath) == 0) {
956 			free(path);
957 			path = strdup("");
958 			if (path == NULL) {
959 				err = got_error_from_errno("strdup");
960 				goto done;
961 			}
962 		} else if (len > repolen &&
963 		    got_path_is_child(path, repo_abspath, repolen)) {
964 			/* Matched an on-disk path inside repository. */
965 			if (got_repo_is_bare(repo)) {
966 				/*
967 				 * Matched an on-disk path inside repository
968 				 * database. Treat input as repository-relative.
969 				 */
970 				free(path);
971 				path = canonpath;
972 				canonpath = NULL;
973 			} else {
974 				char *child;
975 				/* Strip common prefix with repository path. */
976 				err = got_path_skip_common_ancestor(&child,
977 				    repo_abspath, path);
978 				if (err)
979 					goto done;
980 				free(path);
981 				path = child;
982 			}
983 		} else {
984 			/*
985 			 * Matched unrelated on-disk path.
986 			 * Treat input as repository-relative.
987 			 */
988 			free(path);
989 			path = canonpath;
990 			canonpath = NULL;
991 		}
992 	}
993 
994 	/* Make in-repository path absolute */
995 	if (path[0] != '/') {
996 		char *abspath;
997 		if (asprintf(&abspath, "/%s", path) == -1) {
998 			err = got_error_from_errno("asprintf");
999 			goto done;
1000 		}
1001 		free(path);
1002 		path = abspath;
1003 	}
1004 
1005 done:
1006 	free(canonpath);
1007 	if (err)
1008 		free(path);
1009 	else
1010 		*in_repo_path = path;
1011 	return err;
1012 }
1013 
1014 static const struct got_error *
1015 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1016     const char *path_packidx)
1017 {
1018 	const struct got_error *err = NULL;
1019 	size_t i;
1020 
1021 	for (i = 0; i < repo->pack_cache_size; i++) {
1022 		if (repo->packidx_cache[i] == NULL)
1023 			break;
1024 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1025 		    path_packidx) == 0) {
1026 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1027 		}
1028 	}
1029 	if (i == repo->pack_cache_size) {
1030 		do {
1031 			i--;
1032 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1033 		    i == repo->pinned_packidx);
1034 		err = got_packidx_close(repo->packidx_cache[i]);
1035 		if (err)
1036 			return err;
1037 	}
1038 
1039 	repo->packidx_cache[i] = packidx;
1040 
1041 	return NULL;
1042 }
1043 
1044 int
1045 got_repo_is_packidx_filename(const char *name, size_t len)
1046 {
1047 	if (len != GOT_PACKIDX_NAMELEN)
1048 		return 0;
1049 
1050 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1051 		return 0;
1052 
1053 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1054 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1055 		return 0;
1056 
1057 	return 1;
1058 }
1059 
1060 static struct got_packidx_bloom_filter *
1061 get_packidx_bloom_filter(struct got_repository *repo,
1062     const char *path, size_t path_len)
1063 {
1064 	struct got_packidx_bloom_filter key;
1065 
1066 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1067 		return NULL; /* XXX */
1068 	key.path_len = path_len;
1069 
1070 	return RB_FIND(got_packidx_bloom_filter_tree,
1071 	    &repo->packidx_bloom_filters, &key);
1072 }
1073 
1074 int
1075 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1076     const char *path_packidx, struct got_object_id *id)
1077 {
1078 	struct got_packidx_bloom_filter *bf;
1079 
1080 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1081 	if (bf)
1082 		return bloom_check(bf->bloom, id->hash, sizeof(id->hash));
1083 
1084 	/* No bloom filter means this pack index must be searched. */
1085 	return 1;
1086 }
1087 
1088 static const struct got_error *
1089 add_packidx_bloom_filter(struct got_repository *repo,
1090     struct got_packidx *packidx, const char *path_packidx)
1091 {
1092 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1093 	struct got_packidx_bloom_filter *bf;
1094 	size_t len;
1095 
1096 	/*
1097 	 * Don't use bloom filters for very large pack index files.
1098 	 * Large pack files will contain a relatively large fraction
1099 	 * of our objects so we will likely need to visit them anyway.
1100 	 * The more objects a pack file contains the higher the probability
1101 	 * of a false-positive match from the bloom filter. And reading
1102 	 * all object IDs from a large pack index file can be expensive.
1103 	 */
1104 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1105 		return NULL;
1106 
1107 	/* Do we already have a filter for this pack index? */
1108 	if (get_packidx_bloom_filter(repo, path_packidx,
1109 	    strlen(path_packidx)) != NULL)
1110 		return NULL;
1111 
1112 	bf = calloc(1, sizeof(*bf));
1113 	if (bf == NULL)
1114 		return got_error_from_errno("calloc");
1115 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1116 	if (bf->bloom == NULL) {
1117 		free(bf);
1118 		return got_error_from_errno("calloc");
1119 	}
1120 
1121 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1122 	if (len >= sizeof(bf->path)) {
1123 		free(bf->bloom);
1124 		free(bf);
1125 		return got_error(GOT_ERR_NO_SPACE);
1126 	}
1127 	bf->path_len = len;
1128 
1129 	/* Minimum size supported by our bloom filter is 1000 entries. */
1130 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1131 	for (i = 0; i < nobjects; i++) {
1132 		struct got_packidx_object_id *id;
1133 		id = &packidx->hdr.sorted_ids[i];
1134 		bloom_add(bf->bloom, id->hash, sizeof(id->hash));
1135 	}
1136 
1137 	RB_INSERT(got_packidx_bloom_filter_tree,
1138 	    &repo->packidx_bloom_filters, bf);
1139 	return NULL;
1140 }
1141 
1142 static void
1143 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1144 {
1145 	struct got_pathlist_entry *pe;
1146 
1147 	while (!TAILQ_EMPTY(packidx_paths)) {
1148 		pe = TAILQ_FIRST(packidx_paths);
1149 		TAILQ_REMOVE(packidx_paths, pe, entry);
1150 		free((char *)pe->path);
1151 		free(pe);
1152 	}
1153 }
1154 
1155 static const struct got_error *
1156 refresh_packidx_paths(struct got_repository *repo)
1157 {
1158 	const struct got_error *err = NULL;
1159 	char *objects_pack_dir = NULL;
1160 	struct stat sb;
1161 
1162 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1163 	if (objects_pack_dir == NULL)
1164 		return got_error_from_errno("got_repo_get_path_objects_pack");
1165 
1166 	if (stat(objects_pack_dir, &sb) == -1) {
1167 		if (errno != ENOENT) {
1168 			err = got_error_from_errno2("stat", objects_pack_dir);
1169 			goto done;
1170 		}
1171 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1172 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1173 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1174 		purge_packidx_paths(&repo->packidx_paths);
1175 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1176 		if (err)
1177 			goto done;
1178 	}
1179 done:
1180 	free(objects_pack_dir);
1181 	return err;
1182 }
1183 
1184 const struct got_error *
1185 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1186     struct got_repository *repo, struct got_object_id *id)
1187 {
1188 	const struct got_error *err;
1189 	struct got_pathlist_entry *pe;
1190 	size_t i;
1191 
1192 	/* Search pack index cache. */
1193 	for (i = 0; i < repo->pack_cache_size; i++) {
1194 		if (repo->packidx_cache[i] == NULL)
1195 			break;
1196 		if (!got_repo_check_packidx_bloom_filter(repo,
1197 		    repo->packidx_cache[i]->path_packidx, id))
1198 			continue; /* object will not be found in this index */
1199 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1200 		if (*idx != -1) {
1201 			*packidx = repo->packidx_cache[i];
1202 			/*
1203 			 * Move this cache entry to the front. Repeatedly
1204 			 * searching a wrong pack index can be expensive.
1205 			 */
1206 			if (i > 0) {
1207 				memmove(&repo->packidx_cache[1],
1208 				    &repo->packidx_cache[0],
1209 				    i * sizeof(repo->packidx_cache[0]));
1210 				repo->packidx_cache[0] = *packidx;
1211 				if (repo->pinned_packidx >= 0 &&
1212 				    repo->pinned_packidx < i)
1213 					repo->pinned_packidx++;
1214 				else if (repo->pinned_packidx == i)
1215 					repo->pinned_packidx = 0;
1216 			}
1217 			return NULL;
1218 		}
1219 	}
1220 	/* No luck. Search the filesystem. */
1221 
1222 	err = refresh_packidx_paths(repo);
1223 	if (err)
1224 		return err;
1225 
1226 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1227 		const char *path_packidx = pe->path;
1228 		int is_cached = 0;
1229 
1230 		if (!got_repo_check_packidx_bloom_filter(repo,
1231 		    pe->path, id))
1232 			continue; /* object will not be found in this index */
1233 
1234 		for (i = 0; i < repo->pack_cache_size; i++) {
1235 			if (repo->packidx_cache[i] == NULL)
1236 				break;
1237 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1238 			    path_packidx) == 0) {
1239 				is_cached = 1;
1240 				break;
1241 			}
1242 		}
1243 		if (is_cached)
1244 			continue; /* already searched */
1245 
1246 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1247 		    path_packidx, 0);
1248 		if (err)
1249 			goto done;
1250 
1251 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1252 		if (err)
1253 			goto done;
1254 
1255 		err = cache_packidx(repo, *packidx, path_packidx);
1256 		if (err)
1257 			goto done;
1258 
1259 		*idx = got_packidx_get_object_idx(*packidx, id);
1260 		if (*idx != -1) {
1261 			err = NULL; /* found the object */
1262 			goto done;
1263 		}
1264 	}
1265 
1266 	err = got_error_no_obj(id);
1267 done:
1268 	return err;
1269 }
1270 
1271 const struct got_error *
1272 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1273     struct got_repository *repo)
1274 {
1275 	const struct got_error *err = NULL;
1276 	DIR *packdir = NULL;
1277 	struct dirent *dent;
1278 	char *path_packidx = NULL;
1279 	int packdir_fd;
1280 	struct stat sb;
1281 
1282 	packdir_fd = openat(got_repo_get_fd(repo),
1283 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1284 	if (packdir_fd == -1) {
1285 		return got_error_from_errno_fmt("openat: %s/%s",
1286 		    got_repo_get_path_git_dir(repo),
1287 		    GOT_OBJECTS_PACK_DIR);
1288 	}
1289 
1290 	packdir = fdopendir(packdir_fd);
1291 	if (packdir == NULL) {
1292 		err = got_error_from_errno("fdopendir");
1293 		goto done;
1294 	}
1295 
1296 	if (fstat(packdir_fd, &sb) == -1) {
1297 		err = got_error_from_errno("fstat");
1298 		goto done;
1299 	}
1300 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1301 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1302 
1303 	while ((dent = readdir(packdir)) != NULL) {
1304 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1305 			continue;
1306 
1307 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1308 		    dent->d_name) == -1) {
1309 			err = got_error_from_errno("asprintf");
1310 			path_packidx = NULL;
1311 			break;
1312 		}
1313 
1314 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1315 		if (err)
1316 			break;
1317 	}
1318 done:
1319 	if (err)
1320 		free(path_packidx);
1321 	if (packdir && closedir(packdir) != 0 && err == NULL)
1322 		err = got_error_from_errno("closedir");
1323 	return err;
1324 }
1325 
1326 const struct got_error *
1327 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1328     struct got_repository *repo)
1329 {
1330 	const struct got_error *err;
1331 	size_t i;
1332 
1333 	*packidx = NULL;
1334 
1335 	/* Search pack index cache. */
1336 	for (i = 0; i < repo->pack_cache_size; i++) {
1337 		if (repo->packidx_cache[i] == NULL)
1338 			break;
1339 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1340 		    path_packidx) == 0) {
1341 			*packidx = repo->packidx_cache[i];
1342 			return NULL;
1343 		}
1344 	}
1345 	/* No luck. Search the filesystem. */
1346 
1347 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1348 	    path_packidx, 0);
1349 	if (err)
1350 		return err;
1351 
1352 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1353 	if (err)
1354 		goto done;
1355 
1356 	err = cache_packidx(repo, *packidx, path_packidx);
1357 done:
1358 	if (err) {
1359 		got_packidx_close(*packidx);
1360 		*packidx = NULL;
1361 	}
1362 	return err;
1363 }
1364 
1365 static const struct got_error *
1366 read_packfile_hdr(int fd, struct got_packidx *packidx)
1367 {
1368 	const struct got_error *err = NULL;
1369 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1370 	struct got_packfile_hdr hdr;
1371 	ssize_t n;
1372 
1373 	n = read(fd, &hdr, sizeof(hdr));
1374 	if (n < 0)
1375 		return got_error_from_errno("read");
1376 	if (n != sizeof(hdr))
1377 		return got_error(GOT_ERR_BAD_PACKFILE);
1378 
1379 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1380 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1381 	    be32toh(hdr.nobjects) != totobj)
1382 		err = got_error(GOT_ERR_BAD_PACKFILE);
1383 
1384 	return err;
1385 }
1386 
1387 static const struct got_error *
1388 open_packfile(int *fd, struct got_repository *repo,
1389     const char *relpath, struct got_packidx *packidx)
1390 {
1391 	const struct got_error *err = NULL;
1392 
1393 	*fd = openat(got_repo_get_fd(repo), relpath,
1394 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1395 	if (*fd == -1)
1396 		return got_error_from_errno_fmt("openat: %s/%s",
1397 		    got_repo_get_path_git_dir(repo), relpath);
1398 
1399 	if (packidx) {
1400 		err = read_packfile_hdr(*fd, packidx);
1401 		if (err) {
1402 			close(*fd);
1403 			*fd = -1;
1404 		}
1405 	}
1406 
1407 	return err;
1408 }
1409 
1410 const struct got_error *
1411 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1412     const char *path_packfile, struct got_packidx *packidx)
1413 {
1414 	const struct got_error *err = NULL;
1415 	struct got_pack *pack = NULL;
1416 	struct stat sb;
1417 	size_t i;
1418 
1419 	if (packp)
1420 		*packp = NULL;
1421 
1422 	for (i = 0; i < repo->pack_cache_size; i++) {
1423 		pack = &repo->packs[i];
1424 		if (pack->path_packfile == NULL)
1425 			break;
1426 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1427 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1428 	}
1429 
1430 	if (i == repo->pack_cache_size) {
1431 		struct got_pack tmp;
1432 		do {
1433 			i--;
1434 		} while (i > 0 && repo->pinned_pack >= 0 &&
1435 		    i == repo->pinned_pack);
1436 		err = got_pack_close(&repo->packs[i]);
1437 		if (err)
1438 			return err;
1439 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1440 			return got_error_from_errno("ftruncate");
1441 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1442 			return got_error_from_errno("ftruncate");
1443 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1444 		memcpy(&repo->packs[i], &repo->packs[0],
1445 		    sizeof(repo->packs[i]));
1446 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1447 		if (repo->pinned_pack == 0)
1448 			repo->pinned_pack = i;
1449 		else if (repo->pinned_pack == i)
1450 			repo->pinned_pack = 0;
1451 		i = 0;
1452 	}
1453 
1454 	pack = &repo->packs[i];
1455 
1456 	pack->path_packfile = strdup(path_packfile);
1457 	if (pack->path_packfile == NULL) {
1458 		err = got_error_from_errno("strdup");
1459 		goto done;
1460 	}
1461 
1462 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1463 	if (err)
1464 		goto done;
1465 
1466 	if (fstat(pack->fd, &sb) != 0) {
1467 		err = got_error_from_errno("fstat");
1468 		goto done;
1469 	}
1470 	pack->filesize = sb.st_size;
1471 
1472 	pack->privsep_child = NULL;
1473 
1474 	err = got_delta_cache_alloc(&pack->delta_cache);
1475 	if (err)
1476 		goto done;
1477 
1478 #ifndef GOT_PACK_NO_MMAP
1479 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1480 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1481 		    pack->fd, 0);
1482 		if (pack->map == MAP_FAILED) {
1483 			if (errno != ENOMEM) {
1484 				err = got_error_from_errno("mmap");
1485 				goto done;
1486 			}
1487 			pack->map = NULL; /* fall back to read(2) */
1488 		}
1489 	}
1490 #endif
1491 done:
1492 	if (err) {
1493 		if (pack)
1494 			got_pack_close(pack);
1495 	} else if (packp)
1496 		*packp = pack;
1497 	return err;
1498 }
1499 
1500 struct got_pack *
1501 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1502 {
1503 	struct got_pack *pack = NULL;
1504 	size_t i;
1505 
1506 	for (i = 0; i < repo->pack_cache_size; i++) {
1507 		pack = &repo->packs[i];
1508 		if (pack->path_packfile == NULL)
1509 			break;
1510 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1511 			return pack;
1512 	}
1513 
1514 	return NULL;
1515 }
1516 
1517 const struct got_error *
1518 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1519     struct got_pack *pack)
1520 {
1521 	size_t i;
1522 	int pinned_pack = -1, pinned_packidx = -1;
1523 
1524 	for (i = 0; i < repo->pack_cache_size; i++) {
1525 		if (repo->packidx_cache[i] &&
1526 		    strcmp(repo->packidx_cache[i]->path_packidx,
1527 		    packidx->path_packidx) == 0)
1528 			pinned_packidx = i;
1529 		if (repo->packs[i].path_packfile &&
1530 		    strcmp(repo->packs[i].path_packfile,
1531 		    pack->path_packfile) == 0)
1532 			pinned_pack = i;
1533 	}
1534 
1535 	if (pinned_packidx == -1 || pinned_pack == -1)
1536 		return got_error(GOT_ERR_PIN_PACK);
1537 
1538 	repo->pinned_pack = pinned_pack;
1539 	repo->pinned_packidx = pinned_packidx;
1540 	if (repo->packs[pinned_pack].privsep_child)
1541 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1542 	return NULL;
1543 }
1544 
1545 struct got_pack *
1546 got_repo_get_pinned_pack(struct got_repository *repo)
1547 {
1548 	if (repo->pinned_pack >= 0 &&
1549 	    repo->pinned_pack < repo->pack_cache_size)
1550 		return &repo->packs[repo->pinned_pack];
1551 
1552 	return NULL;
1553 }
1554 
1555 void
1556 got_repo_unpin_pack(struct got_repository *repo)
1557 {
1558 	repo->pinned_packidx = -1;
1559 	repo->pinned_pack = -1;
1560 	repo->pinned_pid = 0;
1561 }
1562 
1563 const struct got_error *
1564 got_repo_init(const char *repo_path, const char *head_name)
1565 {
1566 	const struct got_error *err = NULL;
1567 	const char *dirnames[] = {
1568 		GOT_OBJECTS_DIR,
1569 		GOT_OBJECTS_PACK_DIR,
1570 		GOT_REFS_DIR,
1571 	};
1572 	const char *description_str = "Unnamed repository; "
1573 	    "edit this file 'description' to name the repository.";
1574 	const char *headref = "ref: refs/heads/";
1575 	const char *gitconfig_str = "[core]\n"
1576 	    "\trepositoryformatversion = 0\n"
1577 	    "\tfilemode = true\n"
1578 	    "\tbare = true\n";
1579 	char *headref_str, *path;
1580 	size_t i;
1581 
1582 	if (!got_path_dir_is_empty(repo_path))
1583 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1584 
1585 	for (i = 0; i < nitems(dirnames); i++) {
1586 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1587 			return got_error_from_errno("asprintf");
1588 		}
1589 		err = got_path_mkdir(path);
1590 		free(path);
1591 		if (err)
1592 			return err;
1593 	}
1594 
1595 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1596 		return got_error_from_errno("asprintf");
1597 	err = got_path_create_file(path, description_str);
1598 	free(path);
1599 	if (err)
1600 		return err;
1601 
1602 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1603 		return got_error_from_errno("asprintf");
1604 	if (asprintf(&headref_str, "%s%s", headref,
1605 	    head_name ? head_name : "main") == -1) {
1606 		free(path);
1607 		return got_error_from_errno("asprintf");
1608 	}
1609 	err = got_path_create_file(path, headref_str);
1610 	free(headref_str);
1611 	free(path);
1612 	if (err)
1613 		return err;
1614 
1615 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1616 		return got_error_from_errno("asprintf");
1617 	err = got_path_create_file(path, gitconfig_str);
1618 	free(path);
1619 	if (err)
1620 		return err;
1621 
1622 	return NULL;
1623 }
1624 
1625 static const struct got_error *
1626 match_packed_object(struct got_object_id **unique_id,
1627     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1628 {
1629 	const struct got_error *err = NULL;
1630 	struct got_object_id_queue matched_ids;
1631 	struct got_pathlist_entry *pe;
1632 
1633 	STAILQ_INIT(&matched_ids);
1634 
1635 	err = refresh_packidx_paths(repo);
1636 	if (err)
1637 		return err;
1638 
1639 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1640 		const char *path_packidx = pe->path;
1641 		struct got_packidx *packidx;
1642 		struct got_object_qid *qid;
1643 
1644 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1645 		    path_packidx, 0);
1646 		if (err)
1647 			break;
1648 
1649 		err = got_packidx_match_id_str_prefix(&matched_ids,
1650 		    packidx, id_str_prefix);
1651 		if (err) {
1652 			got_packidx_close(packidx);
1653 			break;
1654 		}
1655 		err = got_packidx_close(packidx);
1656 		if (err)
1657 			break;
1658 
1659 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1660 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1661 				int matched_type;
1662 				err = got_object_get_type(&matched_type, repo,
1663 				    &qid->id);
1664 				if (err)
1665 					goto done;
1666 				if (matched_type != obj_type)
1667 					continue;
1668 			}
1669 			if (*unique_id == NULL) {
1670 				*unique_id = got_object_id_dup(&qid->id);
1671 				if (*unique_id == NULL) {
1672 					err = got_error_from_errno("malloc");
1673 					goto done;
1674 				}
1675 			} else {
1676 				if (got_object_id_cmp(*unique_id,
1677 				    &qid->id) == 0)
1678 					continue; /* packed multiple times */
1679 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1680 				goto done;
1681 			}
1682 		}
1683 	}
1684 done:
1685 	got_object_id_queue_free(&matched_ids);
1686 	if (err) {
1687 		free(*unique_id);
1688 		*unique_id = NULL;
1689 	}
1690 	return err;
1691 }
1692 
1693 static const struct got_error *
1694 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1695     const char *object_dir, const char *id_str_prefix, int obj_type,
1696     struct got_repository *repo)
1697 {
1698 	const struct got_error *err = NULL;
1699 	char *path, *id_str = NULL;
1700 	DIR *dir = NULL;
1701 	struct dirent *dent;
1702 	struct got_object_id id;
1703 
1704 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1705 		err = got_error_from_errno("asprintf");
1706 		goto done;
1707 	}
1708 
1709 	dir = opendir(path);
1710 	if (dir == NULL) {
1711 		if (errno == ENOENT) {
1712 			err = NULL;
1713 			goto done;
1714 		}
1715 		err = got_error_from_errno2("opendir", path);
1716 		goto done;
1717 	}
1718 	while ((dent = readdir(dir)) != NULL) {
1719 		int cmp;
1720 
1721 		free(id_str);
1722 		id_str = NULL;
1723 
1724 		if (strcmp(dent->d_name, ".") == 0 ||
1725 		    strcmp(dent->d_name, "..") == 0)
1726 			continue;
1727 
1728 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1729 			err = got_error_from_errno("asprintf");
1730 			goto done;
1731 		}
1732 
1733 		if (!got_parse_sha1_digest(id.hash, id_str))
1734 			continue;
1735 
1736 		/*
1737 		 * Directory entries do not necessarily appear in
1738 		 * sorted order, so we must iterate over all of them.
1739 		 */
1740 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1741 		if (cmp != 0)
1742 			continue;
1743 
1744 		if (*unique_id == NULL) {
1745 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1746 				int matched_type;
1747 				err = got_object_get_type(&matched_type, repo,
1748 				    &id);
1749 				if (err)
1750 					goto done;
1751 				if (matched_type != obj_type)
1752 					continue;
1753 			}
1754 			*unique_id = got_object_id_dup(&id);
1755 			if (*unique_id == NULL) {
1756 				err = got_error_from_errno("got_object_id_dup");
1757 				goto done;
1758 			}
1759 		} else {
1760 			if (got_object_id_cmp(*unique_id, &id) == 0)
1761 				continue; /* both packed and loose */
1762 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1763 			goto done;
1764 		}
1765 	}
1766 done:
1767 	if (dir && closedir(dir) != 0 && err == NULL)
1768 		err = got_error_from_errno("closedir");
1769 	if (err) {
1770 		free(*unique_id);
1771 		*unique_id = NULL;
1772 	}
1773 	free(id_str);
1774 	free(path);
1775 	return err;
1776 }
1777 
1778 const struct got_error *
1779 got_repo_match_object_id_prefix(struct got_object_id **id,
1780     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1781 {
1782 	const struct got_error *err = NULL;
1783 	char *path_objects = NULL, *object_dir = NULL;
1784 	size_t len;
1785 	int i;
1786 
1787 	*id = NULL;
1788 
1789 	path_objects = got_repo_get_path_objects(repo);
1790 
1791 	len = strlen(id_str_prefix);
1792 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1793 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1794 		goto done;
1795 	}
1796 
1797 	for (i = 0; i < len; i++) {
1798 		if (isxdigit((unsigned char)id_str_prefix[i]))
1799 			continue;
1800 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1801 		goto done;
1802 	}
1803 
1804 	if (len >= 2) {
1805 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1806 		if (err)
1807 			goto done;
1808 		object_dir = strndup(id_str_prefix, 2);
1809 		if (object_dir == NULL) {
1810 			err = got_error_from_errno("strdup");
1811 			goto done;
1812 		}
1813 		err = match_loose_object(id, path_objects, object_dir,
1814 		    id_str_prefix, obj_type, repo);
1815 	} else if (len == 1) {
1816 		int i;
1817 		for (i = 0; i < 0xf; i++) {
1818 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1819 			    == -1) {
1820 				err = got_error_from_errno("asprintf");
1821 				goto done;
1822 			}
1823 			err = match_packed_object(id, repo, object_dir,
1824 			    obj_type);
1825 			if (err)
1826 				goto done;
1827 			err = match_loose_object(id, path_objects, object_dir,
1828 			    id_str_prefix, obj_type, repo);
1829 			if (err)
1830 				goto done;
1831 		}
1832 	} else {
1833 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1834 		goto done;
1835 	}
1836 done:
1837 	free(path_objects);
1838 	free(object_dir);
1839 	if (err) {
1840 		free(*id);
1841 		*id = NULL;
1842 	} else if (*id == NULL) {
1843 		switch (obj_type) {
1844 		case GOT_OBJ_TYPE_BLOB:
1845 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1846 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1847 			break;
1848 		case GOT_OBJ_TYPE_TREE:
1849 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1850 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1851 			break;
1852 		case GOT_OBJ_TYPE_COMMIT:
1853 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1854 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1855 			break;
1856 		case GOT_OBJ_TYPE_TAG:
1857 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1858 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1859 			break;
1860 		default:
1861 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1862 			break;
1863 		}
1864 	}
1865 
1866 	return err;
1867 }
1868 
1869 const struct got_error *
1870 got_repo_match_object_id(struct got_object_id **id, char **label,
1871     const char *id_str, int obj_type, struct got_reflist_head *refs,
1872     struct got_repository *repo)
1873 {
1874 	const struct got_error *err;
1875 	struct got_tag_object *tag;
1876 	struct got_reference *ref = NULL;
1877 
1878 	*id = NULL;
1879 	if (label)
1880 		*label = NULL;
1881 
1882 	if (refs) {
1883 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1884 		    refs, repo);
1885 		if (err == NULL) {
1886 			*id = got_object_id_dup(
1887 			    got_object_tag_get_object_id(tag));
1888 			if (*id == NULL)
1889 				err = got_error_from_errno("got_object_id_dup");
1890 			else if (label && asprintf(label, "refs/tags/%s",
1891 			    got_object_tag_get_name(tag)) == -1) {
1892 				err = got_error_from_errno("asprintf");
1893 				free(*id);
1894 				*id = NULL;
1895 			}
1896 			got_object_tag_close(tag);
1897 			return err;
1898 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1899 		    err->code != GOT_ERR_NO_OBJ)
1900 			return err;
1901 	}
1902 
1903 	err = got_ref_open(&ref, repo, id_str, 0);
1904 	if (err == NULL) {
1905 		err = got_ref_resolve(id, repo, ref);
1906 		if (err)
1907 			goto done;
1908 		if (label) {
1909 			*label = strdup(got_ref_get_name(ref));
1910 			if (*label == NULL) {
1911 				err = got_error_from_errno("strdup");
1912 				goto done;
1913 			}
1914 		}
1915 	} else {
1916 		if (err->code != GOT_ERR_NOT_REF &&
1917 		    err->code != GOT_ERR_BAD_REF_NAME)
1918 			goto done;
1919 		err = got_repo_match_object_id_prefix(id, id_str,
1920 		    obj_type, repo);
1921 		if (err) {
1922 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1923 				err = got_error_not_ref(id_str);
1924 			goto done;
1925 		}
1926 		if (label) {
1927 			err = got_object_id_str(label, *id);
1928 			if (*label == NULL) {
1929 				err = got_error_from_errno("strdup");
1930 				goto done;
1931 			}
1932 		}
1933 	}
1934 done:
1935 	if (ref)
1936 		got_ref_close(ref);
1937 	return err;
1938 }
1939 
1940 const struct got_error *
1941 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1942     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1943 {
1944 	const struct got_error *err = NULL;
1945 	struct got_reflist_entry *re;
1946 	struct got_object_id *tag_id;
1947 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1948 
1949 	*tag = NULL;
1950 
1951 	TAILQ_FOREACH(re, refs, entry) {
1952 		const char *refname;
1953 		refname = got_ref_get_name(re->ref);
1954 		if (got_ref_is_symbolic(re->ref))
1955 			continue;
1956 		if (strncmp(refname, "refs/tags/", 10) != 0)
1957 			continue;
1958 		if (!name_is_absolute)
1959 			refname += strlen("refs/tags/");
1960 		if (strcmp(refname, name) != 0)
1961 			continue;
1962 		err = got_ref_resolve(&tag_id, repo, re->ref);
1963 		if (err)
1964 			break;
1965 		err = got_object_open_as_tag(tag, repo, tag_id);
1966 		free(tag_id);
1967 		if (err)
1968 			break;
1969 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1970 		    got_object_tag_get_object_type(*tag) == obj_type)
1971 			break;
1972 		got_object_tag_close(*tag);
1973 		*tag = NULL;
1974 	}
1975 
1976 	if (err == NULL && *tag == NULL)
1977 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1978 		    GOT_OBJ_LABEL_TAG, name);
1979 	return err;
1980 }
1981 
1982 static const struct got_error *
1983 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1984     const char *name, mode_t mode, struct got_object_id *blob_id)
1985 {
1986 	const struct got_error *err = NULL;
1987 
1988 	 *new_te = NULL;
1989 
1990 	*new_te = calloc(1, sizeof(**new_te));
1991 	if (*new_te == NULL)
1992 		return got_error_from_errno("calloc");
1993 
1994 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1995 	    sizeof((*new_te)->name)) {
1996 		err = got_error(GOT_ERR_NO_SPACE);
1997 		goto done;
1998 	}
1999 
2000 	if (S_ISLNK(mode)) {
2001 		(*new_te)->mode = S_IFLNK;
2002 	} else {
2003 		(*new_te)->mode = S_IFREG;
2004 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2005 	}
2006 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2007 done:
2008 	if (err && *new_te) {
2009 		free(*new_te);
2010 		*new_te = NULL;
2011 	}
2012 	return err;
2013 }
2014 
2015 static const struct got_error *
2016 import_file(struct got_tree_entry **new_te, struct dirent *de,
2017     const char *path, struct got_repository *repo)
2018 {
2019 	const struct got_error *err;
2020 	struct got_object_id *blob_id = NULL;
2021 	char *filepath;
2022 	struct stat sb;
2023 
2024 	if (asprintf(&filepath, "%s%s%s", path,
2025 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2026 		return got_error_from_errno("asprintf");
2027 
2028 	if (lstat(filepath, &sb) != 0) {
2029 		err = got_error_from_errno2("lstat", path);
2030 		goto done;
2031 	}
2032 
2033 	err = got_object_blob_create(&blob_id, filepath, repo);
2034 	if (err)
2035 		goto done;
2036 
2037 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2038 	    blob_id);
2039 done:
2040 	free(filepath);
2041 	if (err)
2042 		free(blob_id);
2043 	return err;
2044 }
2045 
2046 static const struct got_error *
2047 insert_tree_entry(struct got_tree_entry *new_te,
2048     struct got_pathlist_head *paths)
2049 {
2050 	const struct got_error *err = NULL;
2051 	struct got_pathlist_entry *new_pe;
2052 
2053 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2054 	if (err)
2055 		return err;
2056 	if (new_pe == NULL)
2057 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2058 	return NULL;
2059 }
2060 
2061 static const struct got_error *write_tree(struct got_object_id **,
2062     const char *, struct got_pathlist_head *, struct got_repository *,
2063     got_repo_import_cb progress_cb, void *progress_arg);
2064 
2065 static const struct got_error *
2066 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2067     const char *path, struct got_pathlist_head *ignores,
2068     struct got_repository *repo,
2069     got_repo_import_cb progress_cb, void *progress_arg)
2070 {
2071 	const struct got_error *err;
2072 	struct got_object_id *id = NULL;
2073 	char *subdirpath;
2074 
2075 	if (asprintf(&subdirpath, "%s%s%s", path,
2076 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2077 		return got_error_from_errno("asprintf");
2078 
2079 	(*new_te) = calloc(1, sizeof(**new_te));
2080 	if (*new_te == NULL)
2081 		return got_error_from_errno("calloc");
2082 	(*new_te)->mode = S_IFDIR;
2083 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2084 	    sizeof((*new_te)->name)) {
2085 		err = got_error(GOT_ERR_NO_SPACE);
2086 		goto done;
2087 	}
2088 	err = write_tree(&id, subdirpath, ignores,  repo,
2089 	    progress_cb, progress_arg);
2090 	if (err)
2091 		goto done;
2092 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2093 
2094 done:
2095 	free(id);
2096 	free(subdirpath);
2097 	if (err) {
2098 		free(*new_te);
2099 		*new_te = NULL;
2100 	}
2101 	return err;
2102 }
2103 
2104 static const struct got_error *
2105 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2106     struct got_pathlist_head *ignores, struct got_repository *repo,
2107     got_repo_import_cb progress_cb, void *progress_arg)
2108 {
2109 	const struct got_error *err = NULL;
2110 	DIR *dir;
2111 	struct dirent *de;
2112 	int nentries;
2113 	struct got_tree_entry *new_te = NULL;
2114 	struct got_pathlist_head paths;
2115 	struct got_pathlist_entry *pe;
2116 
2117 	*new_tree_id = NULL;
2118 
2119 	TAILQ_INIT(&paths);
2120 
2121 	dir = opendir(path_dir);
2122 	if (dir == NULL) {
2123 		err = got_error_from_errno2("opendir", path_dir);
2124 		goto done;
2125 	}
2126 
2127 	nentries = 0;
2128 	while ((de = readdir(dir)) != NULL) {
2129 		int ignore = 0;
2130 		int type;
2131 
2132 		if (strcmp(de->d_name, ".") == 0 ||
2133 		    strcmp(de->d_name, "..") == 0)
2134 			continue;
2135 
2136 		TAILQ_FOREACH(pe, ignores, entry) {
2137 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2138 				ignore = 1;
2139 				break;
2140 			}
2141 		}
2142 		if (ignore)
2143 			continue;
2144 
2145 		err = got_path_dirent_type(&type, path_dir, de);
2146 		if (err)
2147 			goto done;
2148 
2149 		if (type == DT_DIR) {
2150 			err = import_subdir(&new_te, de, path_dir,
2151 			    ignores, repo, progress_cb, progress_arg);
2152 			if (err) {
2153 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2154 					goto done;
2155 				err = NULL;
2156 				continue;
2157 			}
2158 		} else if (type == DT_REG || type == DT_LNK) {
2159 			err = import_file(&new_te, de, path_dir, repo);
2160 			if (err)
2161 				goto done;
2162 		} else
2163 			continue;
2164 
2165 		err = insert_tree_entry(new_te, &paths);
2166 		if (err)
2167 			goto done;
2168 		nentries++;
2169 	}
2170 
2171 	if (TAILQ_EMPTY(&paths)) {
2172 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2173 		    "cannot create tree without any entries");
2174 		goto done;
2175 	}
2176 
2177 	TAILQ_FOREACH(pe, &paths, entry) {
2178 		struct got_tree_entry *te = pe->data;
2179 		char *path;
2180 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2181 			continue;
2182 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2183 			err = got_error_from_errno("asprintf");
2184 			goto done;
2185 		}
2186 		err = (*progress_cb)(progress_arg, path);
2187 		free(path);
2188 		if (err)
2189 			goto done;
2190 	}
2191 
2192 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2193 done:
2194 	if (dir)
2195 		closedir(dir);
2196 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2197 	return err;
2198 }
2199 
2200 const struct got_error *
2201 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2202     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2203     struct got_repository *repo, got_repo_import_cb progress_cb,
2204     void *progress_arg)
2205 {
2206 	const struct got_error *err;
2207 	struct got_object_id *new_tree_id;
2208 
2209 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2210 	    progress_cb, progress_arg);
2211 	if (err)
2212 		return err;
2213 
2214 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2215 	    author, time(NULL), author, time(NULL), logmsg, repo);
2216 	free(new_tree_id);
2217 	return err;
2218 }
2219 
2220 const struct got_error *
2221 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2222     struct got_repository *repo)
2223 {
2224 	const struct got_error *err = NULL;
2225 	char *path_objects = NULL, *path = NULL;
2226 	DIR *dir = NULL;
2227 	struct got_object_id id;
2228 	int i;
2229 
2230 	*nobjects = 0;
2231 	*ondisk_size = 0;
2232 
2233 	path_objects = got_repo_get_path_objects(repo);
2234 	if (path_objects == NULL)
2235 		return got_error_from_errno("got_repo_get_path_objects");
2236 
2237 	for (i = 0; i <= 0xff; i++) {
2238 		struct dirent *dent;
2239 
2240 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2241 			err = got_error_from_errno("asprintf");
2242 			break;
2243 		}
2244 
2245 		dir = opendir(path);
2246 		if (dir == NULL) {
2247 			if (errno == ENOENT) {
2248 				err = NULL;
2249 				continue;
2250 			}
2251 			err = got_error_from_errno2("opendir", path);
2252 			break;
2253 		}
2254 
2255 		while ((dent = readdir(dir)) != NULL) {
2256 			char *id_str;
2257 			int fd;
2258 			struct stat sb;
2259 
2260 			if (strcmp(dent->d_name, ".") == 0 ||
2261 			    strcmp(dent->d_name, "..") == 0)
2262 				continue;
2263 
2264 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2265 				err = got_error_from_errno("asprintf");
2266 				goto done;
2267 			}
2268 
2269 			if (!got_parse_sha1_digest(id.hash, id_str)) {
2270 				free(id_str);
2271 				continue;
2272 			}
2273 			free(id_str);
2274 
2275 			err = got_object_open_loose_fd(&fd, &id, repo);
2276 			if (err)
2277 				goto done;
2278 
2279 			if (fstat(fd, &sb) == -1) {
2280 				err = got_error_from_errno("fstat");
2281 				close(fd);
2282 				goto done;
2283 			}
2284 			(*nobjects)++;
2285 			(*ondisk_size) += sb.st_size;
2286 
2287 			if (close(fd) == -1) {
2288 				err = got_error_from_errno("close");
2289 				goto done;
2290 			}
2291 		}
2292 
2293 		if (closedir(dir) != 0) {
2294 			err = got_error_from_errno("closedir");
2295 			goto done;
2296 		}
2297 		dir = NULL;
2298 
2299 		free(path);
2300 		path = NULL;
2301 	}
2302 done:
2303 	if (dir && closedir(dir) != 0 && err == NULL)
2304 		err = got_error_from_errno("closedir");
2305 
2306 	if (err) {
2307 		*nobjects = 0;
2308 		*ondisk_size = 0;
2309 	}
2310 	free(path_objects);
2311 	free(path);
2312 	return err;
2313 }
2314 
2315 const struct got_error *
2316 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2317     off_t *total_packsize, struct got_repository *repo)
2318 {
2319 	const struct got_error *err = NULL;
2320 	DIR *packdir = NULL;
2321 	struct dirent *dent;
2322 	struct got_packidx *packidx = NULL;
2323 	char *path_packidx;
2324 	char *path_packfile;
2325 	int packdir_fd;
2326 	struct stat sb;
2327 
2328 	*npackfiles = 0;
2329 	*nobjects = 0;
2330 	*total_packsize = 0;
2331 
2332 	packdir_fd = openat(got_repo_get_fd(repo),
2333 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2334 	if (packdir_fd == -1) {
2335 		return got_error_from_errno_fmt("openat: %s/%s",
2336 		    got_repo_get_path_git_dir(repo),
2337 		    GOT_OBJECTS_PACK_DIR);
2338 	}
2339 
2340 	packdir = fdopendir(packdir_fd);
2341 	if (packdir == NULL) {
2342 		err = got_error_from_errno("fdopendir");
2343 		goto done;
2344 	}
2345 
2346 	while ((dent = readdir(packdir)) != NULL) {
2347 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2348 			continue;
2349 
2350 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2351 		    dent->d_name) == -1) {
2352 			err = got_error_from_errno("asprintf");
2353 			goto done;
2354 		}
2355 
2356 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2357 		    path_packidx, 0);
2358 		free(path_packidx);
2359 		if (err)
2360 			goto done;
2361 
2362 		if (fstat(packidx->fd, &sb) == -1)
2363 			goto done;
2364 		*total_packsize += sb.st_size;
2365 
2366 		err = got_packidx_get_packfile_path(&path_packfile,
2367 		    packidx->path_packidx);
2368 		if (err)
2369 			goto done;
2370 
2371 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2372 		    0) == -1) {
2373 			free(path_packfile);
2374 			goto done;
2375 		}
2376 		free(path_packfile);
2377 		*total_packsize += sb.st_size;
2378 
2379 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2380 
2381 		(*npackfiles)++;
2382 
2383 		got_packidx_close(packidx);
2384 		packidx = NULL;
2385 	}
2386 done:
2387 	if (packidx)
2388 		got_packidx_close(packidx);
2389 	if (packdir && closedir(packdir) != 0 && err == NULL)
2390 		err = got_error_from_errno("closedir");
2391 	if (err) {
2392 		*npackfiles = 0;
2393 		*nobjects = 0;
2394 		*total_packsize = 0;
2395 	}
2396 	return err;
2397 }
2398 
2399 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2400     got_packidx_bloom_filter_cmp);