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 <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_delta_cache.h"
57 #include "got_lib_inflate.h"
58 #include "got_lib_object.h"
59 #include "got_lib_object_parse.h"
60 #include "got_lib_object_create.h"
61 #include "got_lib_pack.h"
62 #include "got_lib_privsep.h"
63 #include "got_lib_sha1.h"
64 #include "got_lib_object_cache.h"
65 #include "got_lib_repository.h"
66 #include "got_lib_gotconfig.h"
67 
68 #ifndef nitems
69 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
70 #endif
71 
72 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
73 
74 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
75     got_packidx_bloom_filter_cmp);
76 
77 const char *
78 got_repo_get_path(struct got_repository *repo)
79 {
80 	return repo->path;
81 }
82 
83 const char *
84 got_repo_get_path_git_dir(struct got_repository *repo)
85 {
86 	return repo->path_git_dir;
87 }
88 
89 int
90 got_repo_get_fd(struct got_repository *repo)
91 {
92 	return repo->gitdir_fd;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_name(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_name;
99 }
100 
101 const char *
102 got_repo_get_gitconfig_author_email(struct got_repository *repo)
103 {
104 	return repo->gitconfig_author_email;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_name;
111 }
112 
113 const char *
114 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
115 {
116 	return repo->global_gitconfig_author_email;
117 }
118 
119 const char *
120 got_repo_get_gitconfig_owner(struct got_repository *repo)
121 {
122 	return repo->gitconfig_owner;
123 }
124 
125 void
126 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
127     struct got_repository *repo)
128 {
129 	*extensions = repo->extensions;
130 	*nextensions = repo->nextensions;
131 }
132 
133 int
134 got_repo_is_bare(struct got_repository *repo)
135 {
136 	return (strcmp(repo->path, repo->path_git_dir) == 0);
137 }
138 
139 static char *
140 get_path_git_child(struct got_repository *repo, const char *basename)
141 {
142 	char *path_child;
143 
144 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
145 	    basename) == -1)
146 		return NULL;
147 
148 	return path_child;
149 }
150 
151 char *
152 got_repo_get_path_objects(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
155 }
156 
157 char *
158 got_repo_get_path_objects_pack(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
161 }
162 
163 char *
164 got_repo_get_path_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_REFS_DIR);
167 }
168 
169 char *
170 got_repo_get_path_packed_refs(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
173 }
174 
175 static char *
176 get_path_head(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_HEAD_FILE);
179 }
180 
181 char *
182 got_repo_get_path_gitconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GITCONFIG);
185 }
186 
187 char *
188 got_repo_get_path_gotconfig(struct got_repository *repo)
189 {
190 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
191 }
192 
193 const struct got_gotconfig *
194 got_repo_get_gotconfig(struct got_repository *repo)
195 {
196 	return repo->gotconfig;
197 }
198 
199 void
200 got_repo_get_gitconfig_remotes(int *nremotes,
201     const struct got_remote_repo **remotes, struct got_repository *repo)
202 {
203 	*nremotes = repo->ngitconfig_remotes;
204 	*remotes = repo->gitconfig_remotes;
205 }
206 
207 static int
208 is_git_repo(struct got_repository *repo)
209 {
210 	const char *path_git = got_repo_get_path_git_dir(repo);
211 	char *path_objects = got_repo_get_path_objects(repo);
212 	char *path_refs = got_repo_get_path_refs(repo);
213 	char *path_head = get_path_head(repo);
214 	int ret = 0;
215 	struct stat sb;
216 	struct got_reference *head_ref;
217 
218 	if (lstat(path_git, &sb) == -1)
219 		goto done;
220 	if (!S_ISDIR(sb.st_mode))
221 		goto done;
222 
223 	if (lstat(path_objects, &sb) == -1)
224 		goto done;
225 	if (!S_ISDIR(sb.st_mode))
226 		goto done;
227 
228 	if (lstat(path_refs, &sb) == -1)
229 		goto done;
230 	if (!S_ISDIR(sb.st_mode))
231 		goto done;
232 
233 	if (lstat(path_head, &sb) == -1)
234 		goto done;
235 	if (!S_ISREG(sb.st_mode))
236 		goto done;
237 
238 	/* Check if the HEAD reference can be opened. */
239 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
240 		goto done;
241 	got_ref_close(head_ref);
242 
243 	ret = 1;
244 done:
245 	free(path_objects);
246 	free(path_refs);
247 	free(path_head);
248 	return ret;
249 
250 }
251 
252 static const struct got_error *
253 close_tempfiles(int *fds, size_t nfds)
254 {
255 	const struct got_error *err = NULL;
256 	int i;
257 
258 	for (i = 0; i < nfds; i++) {
259 		if (fds[i] == -1)
260 			continue;
261 		if (close(fds[i]) == -1) {
262 			err = got_error_from_errno("close");
263 			break;
264 		}
265 	}
266 	free(fds);
267 	return err;
268 }
269 
270 static const struct got_error *
271 open_tempfiles(int **fds, size_t array_size, size_t nfds)
272 {
273 	const struct got_error *err = NULL;
274 	int i;
275 
276 	*fds = calloc(array_size, sizeof(**fds));
277 	if (*fds == NULL)
278 		return got_error_from_errno("calloc");
279 
280 	for (i = 0; i < array_size; i++)
281 		(*fds)[i] = -1;
282 
283 	for (i = 0; i < nfds; i++) {
284 		(*fds)[i] = got_opentempfd();
285 		if ((*fds)[i] == -1) {
286 			err = got_error_from_errno("got_opentempfd");
287 			close_tempfiles(*fds, nfds);
288 			*fds = NULL;
289 			return err;
290 		}
291 	}
292 
293 	return NULL;
294 }
295 
296 static const struct got_error *
297 get_pack_cache_size(int *pack_cache_size)
298 {
299 	struct rlimit rl;
300 
301 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
302 		return got_error_from_errno("getrlimit");
303 
304 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
305 	if (*pack_cache_size > rl.rlim_cur / 8)
306 		*pack_cache_size = rl.rlim_cur / 8;
307 
308 	return NULL;
309 }
310 
311 const struct got_error *
312 got_repo_pack_fds_open(int **pack_fds)
313 {
314 	const struct got_error *err;
315 	int nfds;
316 
317 	err = get_pack_cache_size(&nfds);
318 	if (err)
319 		return err;
320 
321 	/*
322 	 * We need one basefd and one accumfd per cached pack.
323 	 * Our constants should be set up in a way such that
324 	 * this error never triggers.
325 	 */
326 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
327 		return got_error(GOT_ERR_NO_SPACE);
328 
329 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
330 }
331 
332 const struct got_error *
333 got_repo_pack_fds_close(int *pack_fds)
334 {
335 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
336 }
337 
338 const struct got_error *
339 got_repo_temp_fds_open(int **temp_fds)
340 {
341 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
342 	    GOT_REPO_NUM_TEMPFILES);
343 }
344 
345 void
346 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
347 {
348 	int i;
349 
350 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
351 		repo->tempfiles[i] = temp_fds[i];
352 }
353 
354 const struct got_error *
355 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
356 {
357 	int i;
358 
359 	*fd = -1;
360 	*idx = -1;
361 
362 	for (i = 0; i < nitems(repo->tempfiles); i++) {
363 		if (repo->tempfile_use_mask & (1 << i))
364 			continue;
365 		if (repo->tempfiles[i] != -1) {
366 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
367 				return got_error_from_errno("ftruncate");
368 			*fd = repo->tempfiles[i];
369 			*idx = i;
370 			repo->tempfile_use_mask |= (1 << i);
371 			return NULL;
372 		}
373 	}
374 
375 	return got_error(GOT_ERR_REPO_TEMPFILE);
376 }
377 
378 void
379 got_repo_temp_fds_put(int idx, struct got_repository *repo)
380 {
381 	repo->tempfile_use_mask &= ~(1 << idx);
382 }
383 
384 const struct got_error *
385 got_repo_temp_fds_close(int *temp_fds)
386 {
387 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
388 }
389 
390 const struct got_error *
391 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
392     struct got_object *obj)
393 {
394 #ifndef GOT_NO_OBJ_CACHE
395 	const struct got_error *err = NULL;
396 	err = got_object_cache_add(&repo->objcache, id, obj);
397 	if (err) {
398 		if (err->code == GOT_ERR_OBJ_EXISTS ||
399 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
400 			err = NULL;
401 		return err;
402 	}
403 	obj->refcnt++;
404 #endif
405 	return NULL;
406 }
407 
408 struct got_object *
409 got_repo_get_cached_object(struct got_repository *repo,
410     struct got_object_id *id)
411 {
412 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
413 }
414 
415 const struct got_error *
416 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
417     struct got_tree_object *tree)
418 {
419 #ifndef GOT_NO_OBJ_CACHE
420 	const struct got_error *err = NULL;
421 	err = got_object_cache_add(&repo->treecache, id, tree);
422 	if (err) {
423 		if (err->code == GOT_ERR_OBJ_EXISTS ||
424 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
425 			err = NULL;
426 		return err;
427 	}
428 	tree->refcnt++;
429 #endif
430 	return NULL;
431 }
432 
433 struct got_tree_object *
434 got_repo_get_cached_tree(struct got_repository *repo,
435     struct got_object_id *id)
436 {
437 	return (struct got_tree_object *)got_object_cache_get(
438 	    &repo->treecache, id);
439 }
440 
441 const struct got_error *
442 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
443     struct got_commit_object *commit)
444 {
445 #ifndef GOT_NO_OBJ_CACHE
446 	const struct got_error *err = NULL;
447 	err = got_object_cache_add(&repo->commitcache, id, commit);
448 	if (err) {
449 		if (err->code == GOT_ERR_OBJ_EXISTS ||
450 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
451 			err = NULL;
452 		return err;
453 	}
454 	commit->refcnt++;
455 #endif
456 	return NULL;
457 }
458 
459 struct got_commit_object *
460 got_repo_get_cached_commit(struct got_repository *repo,
461     struct got_object_id *id)
462 {
463 	return (struct got_commit_object *)got_object_cache_get(
464 	    &repo->commitcache, id);
465 }
466 
467 const struct got_error *
468 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
469     struct got_tag_object *tag)
470 {
471 #ifndef GOT_NO_OBJ_CACHE
472 	const struct got_error *err = NULL;
473 	err = got_object_cache_add(&repo->tagcache, id, tag);
474 	if (err) {
475 		if (err->code == GOT_ERR_OBJ_EXISTS ||
476 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
477 			err = NULL;
478 		return err;
479 	}
480 	tag->refcnt++;
481 #endif
482 	return NULL;
483 }
484 
485 struct got_tag_object *
486 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
487 {
488 	return (struct got_tag_object *)got_object_cache_get(
489 	    &repo->tagcache, id);
490 }
491 
492 const struct got_error *
493 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
494     struct got_raw_object *raw)
495 {
496 #ifndef GOT_NO_OBJ_CACHE
497 	const struct got_error *err = NULL;
498 	err = got_object_cache_add(&repo->rawcache, id, raw);
499 	if (err) {
500 		if (err->code == GOT_ERR_OBJ_EXISTS ||
501 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
502 			err = NULL;
503 		return err;
504 	}
505 	raw->refcnt++;
506 #endif
507 	return NULL;
508 }
509 
510 
511 struct got_raw_object *
512 got_repo_get_cached_raw_object(struct got_repository *repo,
513     struct got_object_id *id)
514 {
515 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
516 }
517 
518 
519 static const struct got_error *
520 open_repo(struct got_repository *repo, const char *path)
521 {
522 	const struct got_error *err = NULL;
523 
524 	repo->gitdir_fd = -1;
525 
526 	/* bare git repository? */
527 	repo->path_git_dir = strdup(path);
528 	if (repo->path_git_dir == NULL)
529 		return got_error_from_errno("strdup");
530 	if (is_git_repo(repo)) {
531 		repo->path = strdup(repo->path_git_dir);
532 		if (repo->path == NULL) {
533 			err = got_error_from_errno("strdup");
534 			goto done;
535 		}
536 		repo->gitdir_fd = open(repo->path_git_dir,
537 		    O_DIRECTORY | O_CLOEXEC);
538 		if (repo->gitdir_fd == -1) {
539 			err = got_error_from_errno2("open",
540 			    repo->path_git_dir);
541 			goto done;
542 		}
543 		return NULL;
544 	}
545 
546 	/* git repository with working tree? */
547 	free(repo->path_git_dir);
548 	repo->path_git_dir = NULL;
549 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
550 		err = got_error_from_errno("asprintf");
551 		goto done;
552 	}
553 	if (is_git_repo(repo)) {
554 		repo->path = strdup(path);
555 		if (repo->path == NULL) {
556 			err = got_error_from_errno("strdup");
557 			goto done;
558 		}
559 		repo->gitdir_fd = open(repo->path_git_dir,
560 		    O_DIRECTORY | O_CLOEXEC);
561 		if (repo->gitdir_fd == -1) {
562 			err = got_error_from_errno2("open",
563 			    repo->path_git_dir);
564 			goto done;
565 		}
566 		return NULL;
567 	}
568 
569 	err = got_error(GOT_ERR_NOT_GIT_REPO);
570 done:
571 	if (err) {
572 		free(repo->path);
573 		repo->path = NULL;
574 		free(repo->path_git_dir);
575 		repo->path_git_dir = NULL;
576 		if (repo->gitdir_fd != -1)
577 			close(repo->gitdir_fd);
578 		repo->gitdir_fd = -1;
579 
580 	}
581 	return err;
582 }
583 
584 static const struct got_error *
585 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
586 {
587 	const struct got_error *err = NULL;
588 	char *repo_gitconfig_path = NULL;
589 
590 	if (global_gitconfig_path) {
591 		/* Read settings from ~/.gitconfig. */
592 		int dummy_repo_version;
593 		err = got_repo_read_gitconfig(&dummy_repo_version,
594 		    &repo->global_gitconfig_author_name,
595 		    &repo->global_gitconfig_author_email,
596 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
597 		if (err)
598 			return err;
599 	}
600 
601 	/* Read repository's .git/config file. */
602 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
603 	if (repo_gitconfig_path == NULL)
604 		return got_error_from_errno("got_repo_get_path_gitconfig");
605 
606 	err = got_repo_read_gitconfig(
607 	    &repo->gitconfig_repository_format_version,
608 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
609 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
610 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
611 	    repo_gitconfig_path);
612 	if (err)
613 		goto done;
614 
615 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
616 		int i;
617 
618 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
619 			got_repo_free_remote_repo_data(
620 			    &repo->gitconfig_remotes[i]);
621 		}
622 		free(repo->gitconfig_remotes);
623 		repo->gitconfig_remotes = NULL;
624 		repo->ngitconfig_remotes = 0;
625 
626 		free(repo->gitconfig_author_name);
627 		repo->gitconfig_author_name = NULL;
628 		free(repo->gitconfig_author_email);
629 		repo->gitconfig_author_email = NULL;
630 
631 		free(repo->global_gitconfig_author_name);
632 		repo->global_gitconfig_author_name = NULL;
633 		free(repo->global_gitconfig_author_email);
634 		repo->global_gitconfig_author_email = NULL;
635 	}
636 
637 done:
638 	free(repo_gitconfig_path);
639 	return err;
640 }
641 
642 static const struct got_error *
643 read_gotconfig(struct got_repository *repo)
644 {
645 	const struct got_error *err = NULL;
646 	char *gotconfig_path;
647 
648 	gotconfig_path = got_repo_get_path_gotconfig(repo);
649 	if (gotconfig_path == NULL)
650 		return got_error_from_errno("got_repo_get_path_gotconfig");
651 
652 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
653 	free(gotconfig_path);
654 	return err;
655 }
656 
657 /* Supported repository format extensions. */
658 static const char *const repo_extensions[] = {
659 	"noop",			/* Got supports repository format version 1. */
660 	"preciousObjects",	/* Supported by gotadmin cleanup. */
661 	"worktreeConfig",	/* Got does not care about Git work trees. */
662 };
663 
664 const struct got_error *
665 got_repo_open(struct got_repository **repop, const char *path,
666     const char *global_gitconfig_path, int *pack_fds)
667 {
668 	struct got_repository *repo = NULL;
669 	const struct got_error *err = NULL;
670 	char *repo_path = NULL;
671 	size_t i, j = 0;
672 
673 	*repop = NULL;
674 
675 	repo = calloc(1, sizeof(*repo));
676 	if (repo == NULL)
677 		return got_error_from_errno("calloc");
678 
679 	RB_INIT(&repo->packidx_bloom_filters);
680 	TAILQ_INIT(&repo->packidx_paths);
681 
682 	for (i = 0; i < nitems(repo->privsep_children); i++) {
683 		memset(&repo->privsep_children[i], 0,
684 		    sizeof(repo->privsep_children[0]));
685 		repo->privsep_children[i].imsg_fd = -1;
686 	}
687 
688 	err = got_object_cache_init(&repo->objcache,
689 	    GOT_OBJECT_CACHE_TYPE_OBJ);
690 	if (err)
691 		goto done;
692 	err = got_object_cache_init(&repo->treecache,
693 	    GOT_OBJECT_CACHE_TYPE_TREE);
694 	if (err)
695 		goto done;
696 	err = got_object_cache_init(&repo->commitcache,
697 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
698 	if (err)
699 		goto done;
700 	err = got_object_cache_init(&repo->tagcache,
701 	    GOT_OBJECT_CACHE_TYPE_TAG);
702 	if (err)
703 		goto done;
704 	err = got_object_cache_init(&repo->rawcache,
705 	    GOT_OBJECT_CACHE_TYPE_RAW);
706 	if (err)
707 		goto done;
708 
709 	err = get_pack_cache_size(&repo->pack_cache_size);
710 	if (err)
711 		goto done;
712 	for (i = 0; i < nitems(repo->packs); i++) {
713 		if (pack_fds != NULL && i < repo->pack_cache_size) {
714 			repo->packs[i].basefd = pack_fds[j++];
715 			repo->packs[i].accumfd = pack_fds[j++];
716 		} else {
717 			repo->packs[i].basefd = -1;
718 			repo->packs[i].accumfd = -1;
719 		}
720 	}
721 	for (i = 0; i < nitems(repo->tempfiles); i++)
722 		repo->tempfiles[i] = -1;
723 	repo->pinned_pack = -1;
724 	repo->pinned_packidx = -1;
725 	repo->pinned_pid = 0;
726 
727 	repo_path = realpath(path, NULL);
728 	if (repo_path == NULL) {
729 		err = got_error_from_errno2("realpath", path);
730 		goto done;
731 	}
732 
733 	for (;;) {
734 		char *parent_path;
735 
736 		err = open_repo(repo, repo_path);
737 		if (err == NULL)
738 			break;
739 		if (err->code != GOT_ERR_NOT_GIT_REPO)
740 			goto done;
741 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
742 			err = got_error(GOT_ERR_NOT_GIT_REPO);
743 			goto done;
744 		}
745 		err = got_path_dirname(&parent_path, repo_path);
746 		if (err)
747 			goto done;
748 		free(repo_path);
749 		repo_path = parent_path;
750 	}
751 
752 	err = read_gotconfig(repo);
753 	if (err)
754 		goto done;
755 
756 	err = read_gitconfig(repo, global_gitconfig_path);
757 	if (err)
758 		goto done;
759 	if (repo->gitconfig_repository_format_version != 0) {
760 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
761 		goto done;
762 	}
763 	for (i = 0; i < repo->nextensions; i++) {
764 		char *ext = repo->extensions[i];
765 		int j, supported = 0;
766 		for (j = 0; j < nitems(repo_extensions); j++) {
767 			if (strcmp(ext, repo_extensions[j]) == 0) {
768 				supported = 1;
769 				break;
770 			}
771 		}
772 		if (!supported) {
773 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
774 			goto done;
775 		}
776 	}
777 
778 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
779 done:
780 	if (err)
781 		got_repo_close(repo);
782 	else
783 		*repop = repo;
784 	free(repo_path);
785 	return err;
786 }
787 
788 const struct got_error *
789 got_repo_close(struct got_repository *repo)
790 {
791 	const struct got_error *err = NULL, *child_err;
792 	struct got_packidx_bloom_filter *bf;
793 	size_t i;
794 
795 	for (i = 0; i < repo->pack_cache_size; i++) {
796 		if (repo->packidx_cache[i] == NULL)
797 			break;
798 		got_packidx_close(repo->packidx_cache[i]);
799 	}
800 
801 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
802 	    &repo->packidx_bloom_filters))) {
803 		RB_REMOVE(got_packidx_bloom_filter_tree,
804 		    &repo->packidx_bloom_filters, bf);
805 		bloom_free(bf->bloom);
806 		free(bf->bloom);
807 		free(bf);
808 	}
809 
810 	for (i = 0; i < repo->pack_cache_size; i++)
811 		if (repo->packs[i].path_packfile)
812 			if (repo->packs[i].path_packfile)
813 				got_pack_close(&repo->packs[i]);
814 
815 	free(repo->path);
816 	free(repo->path_git_dir);
817 
818 	got_object_cache_close(&repo->objcache);
819 	got_object_cache_close(&repo->treecache);
820 	got_object_cache_close(&repo->commitcache);
821 	got_object_cache_close(&repo->tagcache);
822 	got_object_cache_close(&repo->rawcache);
823 
824 	for (i = 0; i < nitems(repo->privsep_children); i++) {
825 		if (repo->privsep_children[i].imsg_fd == -1)
826 			continue;
827 		imsg_clear(repo->privsep_children[i].ibuf);
828 		free(repo->privsep_children[i].ibuf);
829 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
830 		child_err = got_privsep_wait_for_child(
831 		    repo->privsep_children[i].pid);
832 		if (child_err && err == NULL)
833 			err = child_err;
834 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
835 		    err == NULL)
836 			err = got_error_from_errno("close");
837 	}
838 
839 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
840 	    err == NULL)
841 		err = got_error_from_errno("close");
842 
843 	if (repo->gotconfig)
844 		got_gotconfig_free(repo->gotconfig);
845 	free(repo->gitconfig_author_name);
846 	free(repo->gitconfig_author_email);
847 	for (i = 0; i < repo->ngitconfig_remotes; i++)
848 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
849 	free(repo->gitconfig_remotes);
850 	for (i = 0; i < repo->nextensions; i++)
851 		free(repo->extensions[i]);
852 	free(repo->extensions);
853 
854 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
855 	free(repo);
856 
857 	return err;
858 }
859 
860 void
861 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
862 {
863 	int i;
864 
865 	free(repo->name);
866 	repo->name = NULL;
867 	free(repo->fetch_url);
868 	repo->fetch_url = NULL;
869 	free(repo->send_url);
870 	repo->send_url = NULL;
871 	for (i = 0; i < repo->nfetch_branches; i++)
872 		free(repo->fetch_branches[i]);
873 	free(repo->fetch_branches);
874 	repo->fetch_branches = NULL;
875 	repo->nfetch_branches = 0;
876 	for (i = 0; i < repo->nsend_branches; i++)
877 		free(repo->send_branches[i]);
878 	free(repo->send_branches);
879 	repo->send_branches = NULL;
880 	repo->nsend_branches = 0;
881 }
882 
883 const struct got_error *
884 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
885     const char *input_path)
886 {
887 	const struct got_error *err = NULL;
888 	const char *repo_abspath = NULL;
889 	size_t repolen, len;
890 	char *canonpath, *path = NULL;
891 
892 	*in_repo_path = NULL;
893 
894 	canonpath = strdup(input_path);
895 	if (canonpath == NULL) {
896 		err = got_error_from_errno("strdup");
897 		goto done;
898 	}
899 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
900 	if (err)
901 		goto done;
902 
903 	repo_abspath = got_repo_get_path(repo);
904 
905 	if (canonpath[0] == '\0') {
906 		path = strdup(canonpath);
907 		if (path == NULL) {
908 			err = got_error_from_errno("strdup");
909 			goto done;
910 		}
911 	} else {
912 		path = realpath(canonpath, NULL);
913 		if (path == NULL) {
914 			if (errno != ENOENT) {
915 				err = got_error_from_errno2("realpath",
916 				    canonpath);
917 				goto done;
918 			}
919 			/*
920 			 * Path is not on disk.
921 			 * Assume it is already relative to repository root.
922 			 */
923 			path = strdup(canonpath);
924 			if (path == NULL) {
925 				err = got_error_from_errno("strdup");
926 				goto done;
927 			}
928 		}
929 
930 		repolen = strlen(repo_abspath);
931 		len = strlen(path);
932 
933 
934 		if (strcmp(path, repo_abspath) == 0) {
935 			free(path);
936 			path = strdup("");
937 			if (path == NULL) {
938 				err = got_error_from_errno("strdup");
939 				goto done;
940 			}
941 		} else if (len > repolen &&
942 		    got_path_is_child(path, repo_abspath, repolen)) {
943 			/* Matched an on-disk path inside repository. */
944 			if (got_repo_is_bare(repo)) {
945 				/*
946 				 * Matched an on-disk path inside repository
947 				 * database. Treat input as repository-relative.
948 				 */
949 				free(path);
950 				path = canonpath;
951 				canonpath = NULL;
952 			} else {
953 				char *child;
954 				/* Strip common prefix with repository path. */
955 				err = got_path_skip_common_ancestor(&child,
956 				    repo_abspath, path);
957 				if (err)
958 					goto done;
959 				free(path);
960 				path = child;
961 			}
962 		} else {
963 			/*
964 			 * Matched unrelated on-disk path.
965 			 * Treat input as repository-relative.
966 			 */
967 			free(path);
968 			path = canonpath;
969 			canonpath = NULL;
970 		}
971 	}
972 
973 	/* Make in-repository path absolute */
974 	if (path[0] != '/') {
975 		char *abspath;
976 		if (asprintf(&abspath, "/%s", path) == -1) {
977 			err = got_error_from_errno("asprintf");
978 			goto done;
979 		}
980 		free(path);
981 		path = abspath;
982 	}
983 
984 done:
985 	free(canonpath);
986 	if (err)
987 		free(path);
988 	else
989 		*in_repo_path = path;
990 	return err;
991 }
992 
993 static const struct got_error *
994 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
995     const char *path_packidx)
996 {
997 	const struct got_error *err = NULL;
998 	size_t i;
999 
1000 	for (i = 0; i < repo->pack_cache_size; i++) {
1001 		if (repo->packidx_cache[i] == NULL)
1002 			break;
1003 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1004 		    path_packidx) == 0) {
1005 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1006 		}
1007 	}
1008 	if (i == repo->pack_cache_size) {
1009 		do {
1010 			i--;
1011 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1012 		    i == repo->pinned_packidx);
1013 		err = got_packidx_close(repo->packidx_cache[i]);
1014 		if (err)
1015 			return err;
1016 	}
1017 
1018 	repo->packidx_cache[i] = packidx;
1019 
1020 	return NULL;
1021 }
1022 
1023 int
1024 got_repo_is_packidx_filename(const char *name, size_t len)
1025 {
1026 	if (len != GOT_PACKIDX_NAMELEN)
1027 		return 0;
1028 
1029 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1030 		return 0;
1031 
1032 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1033 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1034 		return 0;
1035 
1036 	return 1;
1037 }
1038 
1039 static struct got_packidx_bloom_filter *
1040 get_packidx_bloom_filter(struct got_repository *repo,
1041     const char *path, size_t path_len)
1042 {
1043 	struct got_packidx_bloom_filter key;
1044 
1045 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1046 		return NULL; /* XXX */
1047 	key.path_len = path_len;
1048 
1049 	return RB_FIND(got_packidx_bloom_filter_tree,
1050 	    &repo->packidx_bloom_filters, &key);
1051 }
1052 
1053 int
1054 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1055     const char *path_packidx, struct got_object_id *id)
1056 {
1057 	struct got_packidx_bloom_filter *bf;
1058 
1059 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1060 	if (bf)
1061 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1062 
1063 	/* No bloom filter means this pack index must be searched. */
1064 	return 1;
1065 }
1066 
1067 static const struct got_error *
1068 add_packidx_bloom_filter(struct got_repository *repo,
1069     struct got_packidx *packidx, const char *path_packidx)
1070 {
1071 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1072 	struct got_packidx_bloom_filter *bf;
1073 	size_t len;
1074 
1075 	/*
1076 	 * Don't use bloom filters for very large pack index files.
1077 	 * Large pack files will contain a relatively large fraction
1078 	 * of our objects so we will likely need to visit them anyway.
1079 	 * The more objects a pack file contains the higher the probability
1080 	 * of a false-positive match from the bloom filter. And reading
1081 	 * all object IDs from a large pack index file can be expensive.
1082 	 */
1083 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1084 		return NULL;
1085 
1086 	/* Do we already have a filter for this pack index? */
1087 	if (get_packidx_bloom_filter(repo, path_packidx,
1088 	    strlen(path_packidx)) != NULL)
1089 		return NULL;
1090 
1091 	bf = calloc(1, sizeof(*bf));
1092 	if (bf == NULL)
1093 		return got_error_from_errno("calloc");
1094 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1095 	if (bf->bloom == NULL) {
1096 		free(bf);
1097 		return got_error_from_errno("calloc");
1098 	}
1099 
1100 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1101 	if (len >= sizeof(bf->path)) {
1102 		free(bf->bloom);
1103 		free(bf);
1104 		return got_error(GOT_ERR_NO_SPACE);
1105 	}
1106 	bf->path_len = len;
1107 
1108 	/* Minimum size supported by our bloom filter is 1000 entries. */
1109 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1110 	for (i = 0; i < nobjects; i++) {
1111 		struct got_packidx_object_id *id;
1112 		id = &packidx->hdr.sorted_ids[i];
1113 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1114 	}
1115 
1116 	RB_INSERT(got_packidx_bloom_filter_tree,
1117 	    &repo->packidx_bloom_filters, bf);
1118 	return NULL;
1119 }
1120 
1121 static void
1122 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1123 {
1124 	struct got_pathlist_entry *pe;
1125 
1126 	while (!TAILQ_EMPTY(packidx_paths)) {
1127 		pe = TAILQ_FIRST(packidx_paths);
1128 		TAILQ_REMOVE(packidx_paths, pe, entry);
1129 		free((char *)pe->path);
1130 		free(pe);
1131 	}
1132 }
1133 
1134 static const struct got_error *
1135 refresh_packidx_paths(struct got_repository *repo)
1136 {
1137 	const struct got_error *err = NULL;
1138 	char *objects_pack_dir = NULL;
1139 	struct stat sb;
1140 
1141 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1142 	if (objects_pack_dir == NULL)
1143 		return got_error_from_errno("got_repo_get_path_objects_pack");
1144 
1145 	if (stat(objects_pack_dir, &sb) == -1) {
1146 		if (errno != ENOENT) {
1147 			err = got_error_from_errno2("stat", objects_pack_dir);
1148 			goto done;
1149 		}
1150 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1151 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1152 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1153 		purge_packidx_paths(&repo->packidx_paths);
1154 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1155 		if (err)
1156 			goto done;
1157 	}
1158 done:
1159 	free(objects_pack_dir);
1160 	return err;
1161 }
1162 
1163 const struct got_error *
1164 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1165     struct got_repository *repo, struct got_object_id *id)
1166 {
1167 	const struct got_error *err;
1168 	struct got_pathlist_entry *pe;
1169 	size_t i;
1170 
1171 	/* Search pack index cache. */
1172 	for (i = 0; i < repo->pack_cache_size; i++) {
1173 		if (repo->packidx_cache[i] == NULL)
1174 			break;
1175 		if (!got_repo_check_packidx_bloom_filter(repo,
1176 		    repo->packidx_cache[i]->path_packidx, id))
1177 			continue; /* object will not be found in this index */
1178 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1179 		if (*idx != -1) {
1180 			*packidx = repo->packidx_cache[i];
1181 			/*
1182 			 * Move this cache entry to the front. Repeatedly
1183 			 * searching a wrong pack index can be expensive.
1184 			 */
1185 			if (i > 0) {
1186 				memmove(&repo->packidx_cache[1],
1187 				    &repo->packidx_cache[0],
1188 				    i * sizeof(repo->packidx_cache[0]));
1189 				repo->packidx_cache[0] = *packidx;
1190 				if (repo->pinned_packidx >= 0 &&
1191 				    repo->pinned_packidx < i)
1192 					repo->pinned_packidx++;
1193 				else if (repo->pinned_packidx == i)
1194 					repo->pinned_packidx = 0;
1195 			}
1196 			return NULL;
1197 		}
1198 	}
1199 	/* No luck. Search the filesystem. */
1200 
1201 	err = refresh_packidx_paths(repo);
1202 	if (err)
1203 		return err;
1204 
1205 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1206 		const char *path_packidx = pe->path;
1207 		int is_cached = 0;
1208 
1209 		if (!got_repo_check_packidx_bloom_filter(repo,
1210 		    pe->path, id))
1211 			continue; /* object will not be found in this index */
1212 
1213 		for (i = 0; i < repo->pack_cache_size; i++) {
1214 			if (repo->packidx_cache[i] == NULL)
1215 				break;
1216 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1217 			    path_packidx) == 0) {
1218 				is_cached = 1;
1219 				break;
1220 			}
1221 		}
1222 		if (is_cached)
1223 			continue; /* already searched */
1224 
1225 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1226 		    path_packidx, 0);
1227 		if (err)
1228 			goto done;
1229 
1230 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1231 		if (err)
1232 			goto done;
1233 
1234 		err = cache_packidx(repo, *packidx, path_packidx);
1235 		if (err)
1236 			goto done;
1237 
1238 		*idx = got_packidx_get_object_idx(*packidx, id);
1239 		if (*idx != -1) {
1240 			err = NULL; /* found the object */
1241 			goto done;
1242 		}
1243 	}
1244 
1245 	err = got_error_no_obj(id);
1246 done:
1247 	return err;
1248 }
1249 
1250 const struct got_error *
1251 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1252     struct got_repository *repo)
1253 {
1254 	const struct got_error *err = NULL;
1255 	DIR *packdir = NULL;
1256 	struct dirent *dent;
1257 	char *path_packidx = NULL;
1258 	int packdir_fd;
1259 	struct stat sb;
1260 
1261 	packdir_fd = openat(got_repo_get_fd(repo),
1262 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1263 	if (packdir_fd == -1) {
1264 		return got_error_from_errno_fmt("openat: %s/%s",
1265 		    got_repo_get_path_git_dir(repo),
1266 		    GOT_OBJECTS_PACK_DIR);
1267 	}
1268 
1269 	packdir = fdopendir(packdir_fd);
1270 	if (packdir == NULL) {
1271 		err = got_error_from_errno("fdopendir");
1272 		goto done;
1273 	}
1274 
1275 	if (fstat(packdir_fd, &sb) == -1) {
1276 		err = got_error_from_errno("fstat");
1277 		goto done;
1278 	}
1279 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1280 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1281 
1282 	while ((dent = readdir(packdir)) != NULL) {
1283 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1284 			continue;
1285 
1286 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1287 		    dent->d_name) == -1) {
1288 			err = got_error_from_errno("asprintf");
1289 			path_packidx = NULL;
1290 			break;
1291 		}
1292 
1293 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1294 		if (err)
1295 			break;
1296 	}
1297 done:
1298 	if (err)
1299 		free(path_packidx);
1300 	if (packdir && closedir(packdir) != 0 && err == NULL)
1301 		err = got_error_from_errno("closedir");
1302 	return err;
1303 }
1304 
1305 const struct got_error *
1306 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1307     struct got_repository *repo)
1308 {
1309 	const struct got_error *err;
1310 	size_t i;
1311 
1312 	*packidx = NULL;
1313 
1314 	/* Search pack index cache. */
1315 	for (i = 0; i < repo->pack_cache_size; i++) {
1316 		if (repo->packidx_cache[i] == NULL)
1317 			break;
1318 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1319 		    path_packidx) == 0) {
1320 			*packidx = repo->packidx_cache[i];
1321 			return NULL;
1322 		}
1323 	}
1324 	/* No luck. Search the filesystem. */
1325 
1326 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1327 	    path_packidx, 0);
1328 	if (err)
1329 		return err;
1330 
1331 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1332 	if (err)
1333 		goto done;
1334 
1335 	err = cache_packidx(repo, *packidx, path_packidx);
1336 done:
1337 	if (err) {
1338 		got_packidx_close(*packidx);
1339 		*packidx = NULL;
1340 	}
1341 	return err;
1342 }
1343 
1344 static const struct got_error *
1345 read_packfile_hdr(int fd, struct got_packidx *packidx)
1346 {
1347 	const struct got_error *err = NULL;
1348 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1349 	struct got_packfile_hdr hdr;
1350 	ssize_t n;
1351 
1352 	n = read(fd, &hdr, sizeof(hdr));
1353 	if (n < 0)
1354 		return got_error_from_errno("read");
1355 	if (n != sizeof(hdr))
1356 		return got_error(GOT_ERR_BAD_PACKFILE);
1357 
1358 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1359 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1360 	    be32toh(hdr.nobjects) != totobj)
1361 		err = got_error(GOT_ERR_BAD_PACKFILE);
1362 
1363 	return err;
1364 }
1365 
1366 static const struct got_error *
1367 open_packfile(int *fd, struct got_repository *repo,
1368     const char *relpath, struct got_packidx *packidx)
1369 {
1370 	const struct got_error *err = NULL;
1371 
1372 	*fd = openat(got_repo_get_fd(repo), relpath,
1373 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1374 	if (*fd == -1)
1375 		return got_error_from_errno_fmt("openat: %s/%s",
1376 		    got_repo_get_path_git_dir(repo), relpath);
1377 
1378 	if (packidx) {
1379 		err = read_packfile_hdr(*fd, packidx);
1380 		if (err) {
1381 			close(*fd);
1382 			*fd = -1;
1383 		}
1384 	}
1385 
1386 	return err;
1387 }
1388 
1389 const struct got_error *
1390 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1391     const char *path_packfile, struct got_packidx *packidx)
1392 {
1393 	const struct got_error *err = NULL;
1394 	struct got_pack *pack = NULL;
1395 	struct stat sb;
1396 	size_t i;
1397 
1398 	if (packp)
1399 		*packp = NULL;
1400 
1401 	for (i = 0; i < repo->pack_cache_size; i++) {
1402 		pack = &repo->packs[i];
1403 		if (pack->path_packfile == NULL)
1404 			break;
1405 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1406 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1407 	}
1408 
1409 	if (i == repo->pack_cache_size) {
1410 		struct got_pack tmp;
1411 		do {
1412 			i--;
1413 		} while (i > 0 && repo->pinned_pack >= 0 &&
1414 		    i == repo->pinned_pack);
1415 		err = got_pack_close(&repo->packs[i]);
1416 		if (err)
1417 			return err;
1418 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1419 			return got_error_from_errno("ftruncate");
1420 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1421 			return got_error_from_errno("ftruncate");
1422 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1423 		memcpy(&repo->packs[i], &repo->packs[0],
1424 		    sizeof(repo->packs[i]));
1425 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1426 		if (repo->pinned_pack == 0)
1427 			repo->pinned_pack = i;
1428 		else if (repo->pinned_pack == i)
1429 			repo->pinned_pack = 0;
1430 		i = 0;
1431 	}
1432 
1433 	pack = &repo->packs[i];
1434 
1435 	pack->path_packfile = strdup(path_packfile);
1436 	if (pack->path_packfile == NULL) {
1437 		err = got_error_from_errno("strdup");
1438 		goto done;
1439 	}
1440 
1441 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1442 	if (err)
1443 		goto done;
1444 
1445 	if (fstat(pack->fd, &sb) != 0) {
1446 		err = got_error_from_errno("fstat");
1447 		goto done;
1448 	}
1449 	pack->filesize = sb.st_size;
1450 
1451 	pack->privsep_child = NULL;
1452 
1453 	err = got_delta_cache_alloc(&pack->delta_cache);
1454 	if (err)
1455 		goto done;
1456 
1457 #ifndef GOT_PACK_NO_MMAP
1458 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1459 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1460 		    pack->fd, 0);
1461 		if (pack->map == MAP_FAILED) {
1462 			if (errno != ENOMEM) {
1463 				err = got_error_from_errno("mmap");
1464 				goto done;
1465 			}
1466 			pack->map = NULL; /* fall back to read(2) */
1467 		}
1468 	}
1469 #endif
1470 done:
1471 	if (err) {
1472 		if (pack)
1473 			got_pack_close(pack);
1474 	} else if (packp)
1475 		*packp = pack;
1476 	return err;
1477 }
1478 
1479 struct got_pack *
1480 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1481 {
1482 	struct got_pack *pack = NULL;
1483 	size_t i;
1484 
1485 	for (i = 0; i < repo->pack_cache_size; i++) {
1486 		pack = &repo->packs[i];
1487 		if (pack->path_packfile == NULL)
1488 			break;
1489 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1490 			return pack;
1491 	}
1492 
1493 	return NULL;
1494 }
1495 
1496 const struct got_error *
1497 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1498     struct got_pack *pack)
1499 {
1500 	size_t i;
1501 	int pinned_pack = -1, pinned_packidx = -1;
1502 
1503 	for (i = 0; i < repo->pack_cache_size; i++) {
1504 		if (repo->packidx_cache[i] &&
1505 		    strcmp(repo->packidx_cache[i]->path_packidx,
1506 		    packidx->path_packidx) == 0)
1507 			pinned_packidx = i;
1508 		if (repo->packs[i].path_packfile &&
1509 		    strcmp(repo->packs[i].path_packfile,
1510 		    pack->path_packfile) == 0)
1511 			pinned_pack = i;
1512 	}
1513 
1514 	if (pinned_packidx == -1 || pinned_pack == -1)
1515 		return got_error(GOT_ERR_PIN_PACK);
1516 
1517 	repo->pinned_pack = pinned_pack;
1518 	repo->pinned_packidx = pinned_packidx;
1519 	if (repo->packs[pinned_pack].privsep_child)
1520 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1521 	return NULL;
1522 }
1523 
1524 struct got_pack *
1525 got_repo_get_pinned_pack(struct got_repository *repo)
1526 {
1527 	if (repo->pinned_pack >= 0 &&
1528 	    repo->pinned_pack < repo->pack_cache_size)
1529 		return &repo->packs[repo->pinned_pack];
1530 
1531 	return NULL;
1532 }
1533 
1534 void
1535 got_repo_unpin_pack(struct got_repository *repo)
1536 {
1537 	repo->pinned_packidx = -1;
1538 	repo->pinned_pack = -1;
1539 	repo->pinned_pid = 0;
1540 }
1541 
1542 const struct got_error *
1543 got_repo_init(const char *repo_path, const char *head_name)
1544 {
1545 	const struct got_error *err = NULL;
1546 	const char *dirnames[] = {
1547 		GOT_OBJECTS_DIR,
1548 		GOT_OBJECTS_PACK_DIR,
1549 		GOT_REFS_DIR,
1550 	};
1551 	const char *description_str = "Unnamed repository; "
1552 	    "edit this file 'description' to name the repository.";
1553 	const char *headref = "ref: refs/heads/";
1554 	const char *gitconfig_str = "[core]\n"
1555 	    "\trepositoryformatversion = 0\n"
1556 	    "\tfilemode = true\n"
1557 	    "\tbare = true\n";
1558 	char *headref_str, *path;
1559 	size_t i;
1560 
1561 	if (!got_path_dir_is_empty(repo_path))
1562 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1563 
1564 	for (i = 0; i < nitems(dirnames); i++) {
1565 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1566 			return got_error_from_errno("asprintf");
1567 		}
1568 		err = got_path_mkdir(path);
1569 		free(path);
1570 		if (err)
1571 			return err;
1572 	}
1573 
1574 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1575 		return got_error_from_errno("asprintf");
1576 	err = got_path_create_file(path, description_str);
1577 	free(path);
1578 	if (err)
1579 		return err;
1580 
1581 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1582 		return got_error_from_errno("asprintf");
1583 	if (asprintf(&headref_str, "%s%s", headref,
1584 	    head_name ? head_name : "main") == -1) {
1585 		free(path);
1586 		return got_error_from_errno("asprintf");
1587 	}
1588 	err = got_path_create_file(path, headref_str);
1589 	free(headref_str);
1590 	free(path);
1591 	if (err)
1592 		return err;
1593 
1594 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1595 		return got_error_from_errno("asprintf");
1596 	err = got_path_create_file(path, gitconfig_str);
1597 	free(path);
1598 	if (err)
1599 		return err;
1600 
1601 	return NULL;
1602 }
1603 
1604 static const struct got_error *
1605 match_packed_object(struct got_object_id **unique_id,
1606     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1607 {
1608 	const struct got_error *err = NULL;
1609 	struct got_object_id_queue matched_ids;
1610 	struct got_pathlist_entry *pe;
1611 
1612 	STAILQ_INIT(&matched_ids);
1613 
1614 	err = refresh_packidx_paths(repo);
1615 	if (err)
1616 		return err;
1617 
1618 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1619 		const char *path_packidx = pe->path;
1620 		struct got_packidx *packidx;
1621 		struct got_object_qid *qid;
1622 
1623 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1624 		    path_packidx, 0);
1625 		if (err)
1626 			break;
1627 
1628 		err = got_packidx_match_id_str_prefix(&matched_ids,
1629 		    packidx, id_str_prefix);
1630 		if (err) {
1631 			got_packidx_close(packidx);
1632 			break;
1633 		}
1634 		err = got_packidx_close(packidx);
1635 		if (err)
1636 			break;
1637 
1638 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1639 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1640 				int matched_type;
1641 				err = got_object_get_type(&matched_type, repo,
1642 				    &qid->id);
1643 				if (err)
1644 					goto done;
1645 				if (matched_type != obj_type)
1646 					continue;
1647 			}
1648 			if (*unique_id == NULL) {
1649 				*unique_id = got_object_id_dup(&qid->id);
1650 				if (*unique_id == NULL) {
1651 					err = got_error_from_errno("malloc");
1652 					goto done;
1653 				}
1654 			} else {
1655 				if (got_object_id_cmp(*unique_id,
1656 				    &qid->id) == 0)
1657 					continue; /* packed multiple times */
1658 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1659 				goto done;
1660 			}
1661 		}
1662 	}
1663 done:
1664 	got_object_id_queue_free(&matched_ids);
1665 	if (err) {
1666 		free(*unique_id);
1667 		*unique_id = NULL;
1668 	}
1669 	return err;
1670 }
1671 
1672 static const struct got_error *
1673 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1674     const char *object_dir, const char *id_str_prefix, int obj_type,
1675     struct got_repository *repo)
1676 {
1677 	const struct got_error *err = NULL;
1678 	char *path, *id_str = NULL;
1679 	DIR *dir = NULL;
1680 	struct dirent *dent;
1681 	struct got_object_id id;
1682 
1683 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1684 		err = got_error_from_errno("asprintf");
1685 		goto done;
1686 	}
1687 
1688 	dir = opendir(path);
1689 	if (dir == NULL) {
1690 		if (errno == ENOENT) {
1691 			err = NULL;
1692 			goto done;
1693 		}
1694 		err = got_error_from_errno2("opendir", path);
1695 		goto done;
1696 	}
1697 	while ((dent = readdir(dir)) != NULL) {
1698 		int cmp;
1699 
1700 		free(id_str);
1701 		id_str = NULL;
1702 
1703 		if (strcmp(dent->d_name, ".") == 0 ||
1704 		    strcmp(dent->d_name, "..") == 0)
1705 			continue;
1706 
1707 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1708 			err = got_error_from_errno("asprintf");
1709 			goto done;
1710 		}
1711 
1712 		if (!got_parse_sha1_digest(id.sha1, id_str))
1713 			continue;
1714 
1715 		/*
1716 		 * Directory entries do not necessarily appear in
1717 		 * sorted order, so we must iterate over all of them.
1718 		 */
1719 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1720 		if (cmp != 0)
1721 			continue;
1722 
1723 		if (*unique_id == NULL) {
1724 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1725 				int matched_type;
1726 				err = got_object_get_type(&matched_type, repo,
1727 				    &id);
1728 				if (err)
1729 					goto done;
1730 				if (matched_type != obj_type)
1731 					continue;
1732 			}
1733 			*unique_id = got_object_id_dup(&id);
1734 			if (*unique_id == NULL) {
1735 				err = got_error_from_errno("got_object_id_dup");
1736 				goto done;
1737 			}
1738 		} else {
1739 			if (got_object_id_cmp(*unique_id, &id) == 0)
1740 				continue; /* both packed and loose */
1741 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1742 			goto done;
1743 		}
1744 	}
1745 done:
1746 	if (dir && closedir(dir) != 0 && err == NULL)
1747 		err = got_error_from_errno("closedir");
1748 	if (err) {
1749 		free(*unique_id);
1750 		*unique_id = NULL;
1751 	}
1752 	free(id_str);
1753 	free(path);
1754 	return err;
1755 }
1756 
1757 const struct got_error *
1758 got_repo_match_object_id_prefix(struct got_object_id **id,
1759     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1760 {
1761 	const struct got_error *err = NULL;
1762 	char *path_objects = NULL, *object_dir = NULL;
1763 	size_t len;
1764 	int i;
1765 
1766 	*id = NULL;
1767 
1768 	path_objects = got_repo_get_path_objects(repo);
1769 
1770 	len = strlen(id_str_prefix);
1771 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1772 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1773 		goto done;
1774 	}
1775 
1776 	for (i = 0; i < len; i++) {
1777 		if (isxdigit((unsigned char)id_str_prefix[i]))
1778 			continue;
1779 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1780 		goto done;
1781 	}
1782 
1783 	if (len >= 2) {
1784 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1785 		if (err)
1786 			goto done;
1787 		object_dir = strndup(id_str_prefix, 2);
1788 		if (object_dir == NULL) {
1789 			err = got_error_from_errno("strdup");
1790 			goto done;
1791 		}
1792 		err = match_loose_object(id, path_objects, object_dir,
1793 		    id_str_prefix, obj_type, repo);
1794 	} else if (len == 1) {
1795 		int i;
1796 		for (i = 0; i < 0xf; i++) {
1797 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1798 			    == -1) {
1799 				err = got_error_from_errno("asprintf");
1800 				goto done;
1801 			}
1802 			err = match_packed_object(id, repo, object_dir,
1803 			    obj_type);
1804 			if (err)
1805 				goto done;
1806 			err = match_loose_object(id, path_objects, object_dir,
1807 			    id_str_prefix, obj_type, repo);
1808 			if (err)
1809 				goto done;
1810 		}
1811 	} else {
1812 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1813 		goto done;
1814 	}
1815 done:
1816 	free(path_objects);
1817 	free(object_dir);
1818 	if (err) {
1819 		free(*id);
1820 		*id = NULL;
1821 	} else if (*id == NULL) {
1822 		switch (obj_type) {
1823 		case GOT_OBJ_TYPE_BLOB:
1824 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1825 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1826 			break;
1827 		case GOT_OBJ_TYPE_TREE:
1828 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1829 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1830 			break;
1831 		case GOT_OBJ_TYPE_COMMIT:
1832 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1833 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1834 			break;
1835 		case GOT_OBJ_TYPE_TAG:
1836 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1837 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1838 			break;
1839 		default:
1840 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1841 			break;
1842 		}
1843 	}
1844 
1845 	return err;
1846 }
1847 
1848 const struct got_error *
1849 got_repo_match_object_id(struct got_object_id **id, char **label,
1850     const char *id_str, int obj_type, struct got_reflist_head *refs,
1851     struct got_repository *repo)
1852 {
1853 	const struct got_error *err;
1854 	struct got_tag_object *tag;
1855 	struct got_reference *ref = NULL;
1856 
1857 	*id = NULL;
1858 	if (label)
1859 		*label = NULL;
1860 
1861 	if (refs) {
1862 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1863 		    refs, repo);
1864 		if (err == NULL) {
1865 			*id = got_object_id_dup(
1866 			    got_object_tag_get_object_id(tag));
1867 			if (*id == NULL)
1868 				err = got_error_from_errno("got_object_id_dup");
1869 			else if (label && asprintf(label, "refs/tags/%s",
1870 			    got_object_tag_get_name(tag)) == -1) {
1871 				err = got_error_from_errno("asprintf");
1872 				free(*id);
1873 				*id = NULL;
1874 			}
1875 			got_object_tag_close(tag);
1876 			return err;
1877 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1878 		    err->code != GOT_ERR_NO_OBJ)
1879 			return err;
1880 	}
1881 
1882 	err = got_ref_open(&ref, repo, id_str, 0);
1883 	if (err == NULL) {
1884 		err = got_ref_resolve(id, repo, ref);
1885 		if (err)
1886 			goto done;
1887 		if (label) {
1888 			*label = strdup(got_ref_get_name(ref));
1889 			if (*label == NULL) {
1890 				err = got_error_from_errno("strdup");
1891 				goto done;
1892 			}
1893 		}
1894 	} else {
1895 		if (err->code != GOT_ERR_NOT_REF &&
1896 		    err->code != GOT_ERR_BAD_REF_NAME)
1897 			goto done;
1898 		err = got_repo_match_object_id_prefix(id, id_str,
1899 		    obj_type, repo);
1900 		if (err) {
1901 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1902 				err = got_error_not_ref(id_str);
1903 			goto done;
1904 		}
1905 		if (label) {
1906 			err = got_object_id_str(label, *id);
1907 			if (*label == NULL) {
1908 				err = got_error_from_errno("strdup");
1909 				goto done;
1910 			}
1911 		}
1912 	}
1913 done:
1914 	if (ref)
1915 		got_ref_close(ref);
1916 	return err;
1917 }
1918 
1919 const struct got_error *
1920 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1921     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1922 {
1923 	const struct got_error *err = NULL;
1924 	struct got_reflist_entry *re;
1925 	struct got_object_id *tag_id;
1926 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1927 
1928 	*tag = NULL;
1929 
1930 	TAILQ_FOREACH(re, refs, entry) {
1931 		const char *refname;
1932 		refname = got_ref_get_name(re->ref);
1933 		if (got_ref_is_symbolic(re->ref))
1934 			continue;
1935 		if (strncmp(refname, "refs/tags/", 10) != 0)
1936 			continue;
1937 		if (!name_is_absolute)
1938 			refname += strlen("refs/tags/");
1939 		if (strcmp(refname, name) != 0)
1940 			continue;
1941 		err = got_ref_resolve(&tag_id, repo, re->ref);
1942 		if (err)
1943 			break;
1944 		err = got_object_open_as_tag(tag, repo, tag_id);
1945 		free(tag_id);
1946 		if (err)
1947 			break;
1948 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1949 		    got_object_tag_get_object_type(*tag) == obj_type)
1950 			break;
1951 		got_object_tag_close(*tag);
1952 		*tag = NULL;
1953 	}
1954 
1955 	if (err == NULL && *tag == NULL)
1956 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1957 		    GOT_OBJ_LABEL_TAG, name);
1958 	return err;
1959 }
1960 
1961 static const struct got_error *
1962 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1963     const char *name, mode_t mode, struct got_object_id *blob_id)
1964 {
1965 	const struct got_error *err = NULL;
1966 
1967 	 *new_te = NULL;
1968 
1969 	*new_te = calloc(1, sizeof(**new_te));
1970 	if (*new_te == NULL)
1971 		return got_error_from_errno("calloc");
1972 
1973 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1974 	    sizeof((*new_te)->name)) {
1975 		err = got_error(GOT_ERR_NO_SPACE);
1976 		goto done;
1977 	}
1978 
1979 	if (S_ISLNK(mode)) {
1980 		(*new_te)->mode = S_IFLNK;
1981 	} else {
1982 		(*new_te)->mode = S_IFREG;
1983 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1984 	}
1985 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1986 done:
1987 	if (err && *new_te) {
1988 		free(*new_te);
1989 		*new_te = NULL;
1990 	}
1991 	return err;
1992 }
1993 
1994 static const struct got_error *
1995 import_file(struct got_tree_entry **new_te, struct dirent *de,
1996     const char *path, struct got_repository *repo)
1997 {
1998 	const struct got_error *err;
1999 	struct got_object_id *blob_id = NULL;
2000 	char *filepath;
2001 	struct stat sb;
2002 
2003 	if (asprintf(&filepath, "%s%s%s", path,
2004 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2005 		return got_error_from_errno("asprintf");
2006 
2007 	if (lstat(filepath, &sb) != 0) {
2008 		err = got_error_from_errno2("lstat", path);
2009 		goto done;
2010 	}
2011 
2012 	err = got_object_blob_create(&blob_id, filepath, repo);
2013 	if (err)
2014 		goto done;
2015 
2016 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2017 	    blob_id);
2018 done:
2019 	free(filepath);
2020 	if (err)
2021 		free(blob_id);
2022 	return err;
2023 }
2024 
2025 static const struct got_error *
2026 insert_tree_entry(struct got_tree_entry *new_te,
2027     struct got_pathlist_head *paths)
2028 {
2029 	const struct got_error *err = NULL;
2030 	struct got_pathlist_entry *new_pe;
2031 
2032 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2033 	if (err)
2034 		return err;
2035 	if (new_pe == NULL)
2036 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2037 	return NULL;
2038 }
2039 
2040 static const struct got_error *write_tree(struct got_object_id **,
2041     const char *, struct got_pathlist_head *, struct got_repository *,
2042     got_repo_import_cb progress_cb, void *progress_arg);
2043 
2044 static const struct got_error *
2045 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2046     const char *path, struct got_pathlist_head *ignores,
2047     struct got_repository *repo,
2048     got_repo_import_cb progress_cb, void *progress_arg)
2049 {
2050 	const struct got_error *err;
2051 	struct got_object_id *id = NULL;
2052 	char *subdirpath;
2053 
2054 	if (asprintf(&subdirpath, "%s%s%s", path,
2055 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2056 		return got_error_from_errno("asprintf");
2057 
2058 	(*new_te) = calloc(1, sizeof(**new_te));
2059 	if (*new_te == NULL)
2060 		return got_error_from_errno("calloc");
2061 	(*new_te)->mode = S_IFDIR;
2062 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2063 	    sizeof((*new_te)->name)) {
2064 		err = got_error(GOT_ERR_NO_SPACE);
2065 		goto done;
2066 	}
2067 	err = write_tree(&id, subdirpath, ignores,  repo,
2068 	    progress_cb, progress_arg);
2069 	if (err)
2070 		goto done;
2071 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2072 
2073 done:
2074 	free(id);
2075 	free(subdirpath);
2076 	if (err) {
2077 		free(*new_te);
2078 		*new_te = NULL;
2079 	}
2080 	return err;
2081 }
2082 
2083 static const struct got_error *
2084 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2085     struct got_pathlist_head *ignores, struct got_repository *repo,
2086     got_repo_import_cb progress_cb, void *progress_arg)
2087 {
2088 	const struct got_error *err = NULL;
2089 	DIR *dir;
2090 	struct dirent *de;
2091 	int nentries;
2092 	struct got_tree_entry *new_te = NULL;
2093 	struct got_pathlist_head paths;
2094 	struct got_pathlist_entry *pe;
2095 
2096 	*new_tree_id = NULL;
2097 
2098 	TAILQ_INIT(&paths);
2099 
2100 	dir = opendir(path_dir);
2101 	if (dir == NULL) {
2102 		err = got_error_from_errno2("opendir", path_dir);
2103 		goto done;
2104 	}
2105 
2106 	nentries = 0;
2107 	while ((de = readdir(dir)) != NULL) {
2108 		int ignore = 0;
2109 		int type;
2110 
2111 		if (strcmp(de->d_name, ".") == 0 ||
2112 		    strcmp(de->d_name, "..") == 0)
2113 			continue;
2114 
2115 		TAILQ_FOREACH(pe, ignores, entry) {
2116 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2117 				ignore = 1;
2118 				break;
2119 			}
2120 		}
2121 		if (ignore)
2122 			continue;
2123 
2124 		err = got_path_dirent_type(&type, path_dir, de);
2125 		if (err)
2126 			goto done;
2127 
2128 		if (type == DT_DIR) {
2129 			err = import_subdir(&new_te, de, path_dir,
2130 			    ignores, repo, progress_cb, progress_arg);
2131 			if (err) {
2132 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2133 					goto done;
2134 				err = NULL;
2135 				continue;
2136 			}
2137 		} else if (type == DT_REG || type == DT_LNK) {
2138 			err = import_file(&new_te, de, path_dir, repo);
2139 			if (err)
2140 				goto done;
2141 		} else
2142 			continue;
2143 
2144 		err = insert_tree_entry(new_te, &paths);
2145 		if (err)
2146 			goto done;
2147 		nentries++;
2148 	}
2149 
2150 	if (TAILQ_EMPTY(&paths)) {
2151 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2152 		    "cannot create tree without any entries");
2153 		goto done;
2154 	}
2155 
2156 	TAILQ_FOREACH(pe, &paths, entry) {
2157 		struct got_tree_entry *te = pe->data;
2158 		char *path;
2159 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2160 			continue;
2161 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2162 			err = got_error_from_errno("asprintf");
2163 			goto done;
2164 		}
2165 		err = (*progress_cb)(progress_arg, path);
2166 		free(path);
2167 		if (err)
2168 			goto done;
2169 	}
2170 
2171 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2172 done:
2173 	if (dir)
2174 		closedir(dir);
2175 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2176 	return err;
2177 }
2178 
2179 const struct got_error *
2180 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2181     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2182     struct got_repository *repo, got_repo_import_cb progress_cb,
2183     void *progress_arg)
2184 {
2185 	const struct got_error *err;
2186 	struct got_object_id *new_tree_id;
2187 
2188 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2189 	    progress_cb, progress_arg);
2190 	if (err)
2191 		return err;
2192 
2193 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2194 	    author, time(NULL), author, time(NULL), logmsg, repo);
2195 	free(new_tree_id);
2196 	return err;
2197 }
2198 
2199 const struct got_error *
2200 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2201     struct got_repository *repo)
2202 {
2203 	const struct got_error *err = NULL;
2204 	char *path_objects = NULL, *path = NULL;
2205 	DIR *dir = NULL;
2206 	struct got_object_id id;
2207 	int i;
2208 
2209 	*nobjects = 0;
2210 	*ondisk_size = 0;
2211 
2212 	path_objects = got_repo_get_path_objects(repo);
2213 	if (path_objects == NULL)
2214 		return got_error_from_errno("got_repo_get_path_objects");
2215 
2216 	for (i = 0; i <= 0xff; i++) {
2217 		struct dirent *dent;
2218 
2219 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2220 			err = got_error_from_errno("asprintf");
2221 			break;
2222 		}
2223 
2224 		dir = opendir(path);
2225 		if (dir == NULL) {
2226 			if (errno == ENOENT) {
2227 				err = NULL;
2228 				continue;
2229 			}
2230 			err = got_error_from_errno2("opendir", path);
2231 			break;
2232 		}
2233 
2234 		while ((dent = readdir(dir)) != NULL) {
2235 			char *id_str;
2236 			int fd;
2237 			struct stat sb;
2238 
2239 			if (strcmp(dent->d_name, ".") == 0 ||
2240 			    strcmp(dent->d_name, "..") == 0)
2241 				continue;
2242 
2243 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2244 				err = got_error_from_errno("asprintf");
2245 				goto done;
2246 			}
2247 
2248 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2249 				free(id_str);
2250 				continue;
2251 			}
2252 			free(id_str);
2253 
2254 			err = got_object_open_loose_fd(&fd, &id, repo);
2255 			if (err)
2256 				goto done;
2257 
2258 			if (fstat(fd, &sb) == -1) {
2259 				err = got_error_from_errno("fstat");
2260 				close(fd);
2261 				goto done;
2262 			}
2263 			(*nobjects)++;
2264 			(*ondisk_size) += sb.st_size;
2265 
2266 			if (close(fd) == -1) {
2267 				err = got_error_from_errno("close");
2268 				goto done;
2269 			}
2270 		}
2271 
2272 		if (closedir(dir) != 0) {
2273 			err = got_error_from_errno("closedir");
2274 			goto done;
2275 		}
2276 		dir = NULL;
2277 
2278 		free(path);
2279 		path = NULL;
2280 	}
2281 done:
2282 	if (dir && closedir(dir) != 0 && err == NULL)
2283 		err = got_error_from_errno("closedir");
2284 
2285 	if (err) {
2286 		*nobjects = 0;
2287 		*ondisk_size = 0;
2288 	}
2289 	free(path_objects);
2290 	free(path);
2291 	return err;
2292 }
2293 
2294 const struct got_error *
2295 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2296     off_t *total_packsize, struct got_repository *repo)
2297 {
2298 	const struct got_error *err = NULL;
2299 	DIR *packdir = NULL;
2300 	struct dirent *dent;
2301 	struct got_packidx *packidx = NULL;
2302 	char *path_packidx;
2303 	char *path_packfile;
2304 	int packdir_fd;
2305 	struct stat sb;
2306 
2307 	*npackfiles = 0;
2308 	*nobjects = 0;
2309 	*total_packsize = 0;
2310 
2311 	packdir_fd = openat(got_repo_get_fd(repo),
2312 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2313 	if (packdir_fd == -1) {
2314 		return got_error_from_errno_fmt("openat: %s/%s",
2315 		    got_repo_get_path_git_dir(repo),
2316 		    GOT_OBJECTS_PACK_DIR);
2317 	}
2318 
2319 	packdir = fdopendir(packdir_fd);
2320 	if (packdir == NULL) {
2321 		err = got_error_from_errno("fdopendir");
2322 		goto done;
2323 	}
2324 
2325 	while ((dent = readdir(packdir)) != NULL) {
2326 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2327 			continue;
2328 
2329 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2330 		    dent->d_name) == -1) {
2331 			err = got_error_from_errno("asprintf");
2332 			goto done;
2333 		}
2334 
2335 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2336 		    path_packidx, 0);
2337 		free(path_packidx);
2338 		if (err)
2339 			goto done;
2340 
2341 		if (fstat(packidx->fd, &sb) == -1)
2342 			goto done;
2343 		*total_packsize += sb.st_size;
2344 
2345 		err = got_packidx_get_packfile_path(&path_packfile,
2346 		    packidx->path_packidx);
2347 		if (err)
2348 			goto done;
2349 
2350 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2351 		    0) == -1) {
2352 			free(path_packfile);
2353 			goto done;
2354 		}
2355 		free(path_packfile);
2356 		*total_packsize += sb.st_size;
2357 
2358 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2359 
2360 		(*npackfiles)++;
2361 
2362 		got_packidx_close(packidx);
2363 		packidx = NULL;
2364 	}
2365 done:
2366 	if (packidx)
2367 		got_packidx_close(packidx);
2368 	if (packdir && closedir(packdir) != 0 && err == NULL)
2369 		err = got_error_from_errno("closedir");
2370 	if (err) {
2371 		*npackfiles = 0;
2372 		*nobjects = 0;
2373 		*total_packsize = 0;
2374 	}
2375 	return err;
2376 }
2377 
2378 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2379     got_packidx_bloom_filter_cmp);