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 enum got_hash_algorithm
97 got_repo_get_hash_algorithm(struct got_repository *repo)
98 {
99 	return repo->algo;
100 }
101 
102 const char *
103 got_repo_get_gitconfig_author_name(struct got_repository *repo)
104 {
105 	return repo->gitconfig_author_name;
106 }
107 
108 const char *
109 got_repo_get_gitconfig_author_email(struct got_repository *repo)
110 {
111 	return repo->gitconfig_author_email;
112 }
113 
114 const char *
115 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
116 {
117 	return repo->global_gitconfig_author_name;
118 }
119 
120 const char *
121 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
122 {
123 	return repo->global_gitconfig_author_email;
124 }
125 
126 const char *
127 got_repo_get_gitconfig_owner(struct got_repository *repo)
128 {
129 	return repo->gitconfig_owner;
130 }
131 
132 void
133 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
134     struct got_repository *repo)
135 {
136 	*extensions = repo->extensions;
137 	*nextensions = repo->nextensions;
138 }
139 
140 int
141 got_repo_is_bare(struct got_repository *repo)
142 {
143 	return (strcmp(repo->path, repo->path_git_dir) == 0);
144 }
145 
146 static char *
147 get_path_git_child(struct got_repository *repo, const char *basename)
148 {
149 	char *path_child;
150 
151 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
152 	    basename) == -1)
153 		return NULL;
154 
155 	return path_child;
156 }
157 
158 char *
159 got_repo_get_path_objects(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
162 }
163 
164 char *
165 got_repo_get_path_objects_pack(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
168 }
169 
170 char *
171 got_repo_get_path_refs(struct got_repository *repo)
172 {
173 	return get_path_git_child(repo, GOT_REFS_DIR);
174 }
175 
176 char *
177 got_repo_get_path_packed_refs(struct got_repository *repo)
178 {
179 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
180 }
181 
182 static char *
183 get_path_head(struct got_repository *repo)
184 {
185 	return get_path_git_child(repo, GOT_HEAD_FILE);
186 }
187 
188 char *
189 got_repo_get_path_gitconfig(struct got_repository *repo)
190 {
191 	return get_path_git_child(repo, GOT_GITCONFIG);
192 }
193 
194 char *
195 got_repo_get_path_gotconfig(struct got_repository *repo)
196 {
197 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
198 }
199 
200 const struct got_gotconfig *
201 got_repo_get_gotconfig(struct got_repository *repo)
202 {
203 	return repo->gotconfig;
204 }
205 
206 void
207 got_repo_get_gitconfig_remotes(int *nremotes,
208     const struct got_remote_repo **remotes, struct got_repository *repo)
209 {
210 	*nremotes = repo->ngitconfig_remotes;
211 	*remotes = repo->gitconfig_remotes;
212 }
213 
214 static int
215 is_git_repo(struct got_repository *repo)
216 {
217 	const char *path_git = got_repo_get_path_git_dir(repo);
218 	char *path_objects = got_repo_get_path_objects(repo);
219 	char *path_refs = got_repo_get_path_refs(repo);
220 	char *path_head = get_path_head(repo);
221 	int ret = 0;
222 	struct stat sb;
223 	struct got_reference *head_ref;
224 
225 	if (lstat(path_git, &sb) == -1)
226 		goto done;
227 	if (!S_ISDIR(sb.st_mode))
228 		goto done;
229 
230 	if (lstat(path_objects, &sb) == -1)
231 		goto done;
232 	if (!S_ISDIR(sb.st_mode))
233 		goto done;
234 
235 	if (lstat(path_refs, &sb) == -1)
236 		goto done;
237 	if (!S_ISDIR(sb.st_mode))
238 		goto done;
239 
240 	if (lstat(path_head, &sb) == -1)
241 		goto done;
242 	if (!S_ISREG(sb.st_mode))
243 		goto done;
244 
245 	/* Check if the HEAD reference can be opened. */
246 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
247 		goto done;
248 	got_ref_close(head_ref);
249 
250 	ret = 1;
251 done:
252 	free(path_objects);
253 	free(path_refs);
254 	free(path_head);
255 	return ret;
256 
257 }
258 
259 static const struct got_error *
260 close_tempfiles(int *fds, size_t nfds)
261 {
262 	const struct got_error *err = NULL;
263 	int i;
264 
265 	for (i = 0; i < nfds; i++) {
266 		if (fds[i] == -1)
267 			continue;
268 		if (close(fds[i]) == -1) {
269 			err = got_error_from_errno("close");
270 			break;
271 		}
272 	}
273 	free(fds);
274 	return err;
275 }
276 
277 static const struct got_error *
278 open_tempfiles(int **fds, size_t array_size, size_t nfds)
279 {
280 	const struct got_error *err = NULL;
281 	int i;
282 
283 	*fds = calloc(array_size, sizeof(**fds));
284 	if (*fds == NULL)
285 		return got_error_from_errno("calloc");
286 
287 	for (i = 0; i < array_size; i++)
288 		(*fds)[i] = -1;
289 
290 	for (i = 0; i < nfds; i++) {
291 		(*fds)[i] = got_opentempfd();
292 		if ((*fds)[i] == -1) {
293 			err = got_error_from_errno("got_opentempfd");
294 			close_tempfiles(*fds, nfds);
295 			*fds = NULL;
296 			return err;
297 		}
298 	}
299 
300 	return NULL;
301 }
302 
303 static const struct got_error *
304 get_pack_cache_size(int *pack_cache_size)
305 {
306 	struct rlimit rl;
307 
308 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
309 		return got_error_from_errno("getrlimit");
310 
311 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
312 	if (*pack_cache_size > rl.rlim_cur / 8)
313 		*pack_cache_size = rl.rlim_cur / 8;
314 
315 	return NULL;
316 }
317 
318 const struct got_error *
319 got_repo_pack_fds_open(int **pack_fds)
320 {
321 	const struct got_error *err;
322 	int nfds;
323 
324 	err = get_pack_cache_size(&nfds);
325 	if (err)
326 		return err;
327 
328 	/*
329 	 * We need one basefd and one accumfd per cached pack.
330 	 * Our constants should be set up in a way such that
331 	 * this error never triggers.
332 	 */
333 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
334 		return got_error(GOT_ERR_NO_SPACE);
335 
336 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
337 }
338 
339 const struct got_error *
340 got_repo_pack_fds_close(int *pack_fds)
341 {
342 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
343 }
344 
345 const struct got_error *
346 got_repo_temp_fds_open(int **temp_fds)
347 {
348 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
349 	    GOT_REPO_NUM_TEMPFILES);
350 }
351 
352 void
353 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
354 {
355 	int i;
356 
357 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
358 		repo->tempfiles[i] = temp_fds[i];
359 }
360 
361 const struct got_error *
362 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
363 {
364 	int i;
365 
366 	*fd = -1;
367 	*idx = -1;
368 
369 	for (i = 0; i < nitems(repo->tempfiles); i++) {
370 		if (repo->tempfile_use_mask & (1 << i))
371 			continue;
372 		if (repo->tempfiles[i] != -1) {
373 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
374 				return got_error_from_errno("ftruncate");
375 			*fd = repo->tempfiles[i];
376 			*idx = i;
377 			repo->tempfile_use_mask |= (1 << i);
378 			return NULL;
379 		}
380 	}
381 
382 	return got_error(GOT_ERR_REPO_TEMPFILE);
383 }
384 
385 void
386 got_repo_temp_fds_put(int idx, struct got_repository *repo)
387 {
388 	repo->tempfile_use_mask &= ~(1 << idx);
389 }
390 
391 const struct got_error *
392 got_repo_temp_fds_close(int *temp_fds)
393 {
394 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
395 }
396 
397 const struct got_error *
398 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
399     struct got_object *obj)
400 {
401 #ifndef GOT_NO_OBJ_CACHE
402 	const struct got_error *err = NULL;
403 	err = got_object_cache_add(&repo->objcache, id, obj);
404 	if (err) {
405 		if (err->code == GOT_ERR_OBJ_EXISTS ||
406 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
407 			err = NULL;
408 		return err;
409 	}
410 	obj->refcnt++;
411 #endif
412 	return NULL;
413 }
414 
415 struct got_object *
416 got_repo_get_cached_object(struct got_repository *repo,
417     struct got_object_id *id)
418 {
419 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
420 }
421 
422 const struct got_error *
423 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
424     struct got_tree_object *tree)
425 {
426 #ifndef GOT_NO_OBJ_CACHE
427 	const struct got_error *err = NULL;
428 	err = got_object_cache_add(&repo->treecache, id, tree);
429 	if (err) {
430 		if (err->code == GOT_ERR_OBJ_EXISTS ||
431 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
432 			err = NULL;
433 		return err;
434 	}
435 	tree->refcnt++;
436 #endif
437 	return NULL;
438 }
439 
440 struct got_tree_object *
441 got_repo_get_cached_tree(struct got_repository *repo,
442     struct got_object_id *id)
443 {
444 	return (struct got_tree_object *)got_object_cache_get(
445 	    &repo->treecache, id);
446 }
447 
448 const struct got_error *
449 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
450     struct got_commit_object *commit)
451 {
452 #ifndef GOT_NO_OBJ_CACHE
453 	const struct got_error *err = NULL;
454 	err = got_object_cache_add(&repo->commitcache, id, commit);
455 	if (err) {
456 		if (err->code == GOT_ERR_OBJ_EXISTS ||
457 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
458 			err = NULL;
459 		return err;
460 	}
461 	commit->refcnt++;
462 #endif
463 	return NULL;
464 }
465 
466 struct got_commit_object *
467 got_repo_get_cached_commit(struct got_repository *repo,
468     struct got_object_id *id)
469 {
470 	return (struct got_commit_object *)got_object_cache_get(
471 	    &repo->commitcache, id);
472 }
473 
474 const struct got_error *
475 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
476     struct got_tag_object *tag)
477 {
478 #ifndef GOT_NO_OBJ_CACHE
479 	const struct got_error *err = NULL;
480 	err = got_object_cache_add(&repo->tagcache, id, tag);
481 	if (err) {
482 		if (err->code == GOT_ERR_OBJ_EXISTS ||
483 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
484 			err = NULL;
485 		return err;
486 	}
487 	tag->refcnt++;
488 #endif
489 	return NULL;
490 }
491 
492 struct got_tag_object *
493 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
494 {
495 	return (struct got_tag_object *)got_object_cache_get(
496 	    &repo->tagcache, id);
497 }
498 
499 const struct got_error *
500 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
501     struct got_raw_object *raw)
502 {
503 #ifndef GOT_NO_OBJ_CACHE
504 	const struct got_error *err = NULL;
505 	err = got_object_cache_add(&repo->rawcache, id, raw);
506 	if (err) {
507 		if (err->code == GOT_ERR_OBJ_EXISTS ||
508 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
509 			err = NULL;
510 		return err;
511 	}
512 	raw->refcnt++;
513 #endif
514 	return NULL;
515 }
516 
517 
518 struct got_raw_object *
519 got_repo_get_cached_raw_object(struct got_repository *repo,
520     struct got_object_id *id)
521 {
522 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
523 }
524 
525 
526 static const struct got_error *
527 open_repo(struct got_repository *repo, const char *path)
528 {
529 	const struct got_error *err = NULL;
530 
531 	repo->gitdir_fd = -1;
532 
533 	/* bare git repository? */
534 	repo->path_git_dir = strdup(path);
535 	if (repo->path_git_dir == NULL)
536 		return got_error_from_errno("strdup");
537 	if (is_git_repo(repo)) {
538 		repo->path = strdup(repo->path_git_dir);
539 		if (repo->path == NULL) {
540 			err = got_error_from_errno("strdup");
541 			goto done;
542 		}
543 		repo->gitdir_fd = open(repo->path_git_dir,
544 		    O_DIRECTORY | O_CLOEXEC);
545 		if (repo->gitdir_fd == -1) {
546 			err = got_error_from_errno2("open",
547 			    repo->path_git_dir);
548 			goto done;
549 		}
550 		return NULL;
551 	}
552 
553 	/* git repository with working tree? */
554 	free(repo->path_git_dir);
555 	repo->path_git_dir = NULL;
556 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
557 		err = got_error_from_errno("asprintf");
558 		goto done;
559 	}
560 	if (is_git_repo(repo)) {
561 		repo->path = strdup(path);
562 		if (repo->path == NULL) {
563 			err = got_error_from_errno("strdup");
564 			goto done;
565 		}
566 		repo->gitdir_fd = open(repo->path_git_dir,
567 		    O_DIRECTORY | O_CLOEXEC);
568 		if (repo->gitdir_fd == -1) {
569 			err = got_error_from_errno2("open",
570 			    repo->path_git_dir);
571 			goto done;
572 		}
573 		return NULL;
574 	}
575 
576 	err = got_error(GOT_ERR_NOT_GIT_REPO);
577 done:
578 	if (err) {
579 		free(repo->path);
580 		repo->path = NULL;
581 		free(repo->path_git_dir);
582 		repo->path_git_dir = NULL;
583 		if (repo->gitdir_fd != -1)
584 			close(repo->gitdir_fd);
585 		repo->gitdir_fd = -1;
586 
587 	}
588 	return err;
589 }
590 
591 static const struct got_error *
592 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
593 {
594 	const struct got_error *err = NULL;
595 	char *repo_gitconfig_path = NULL;
596 	char *object_format = NULL;
597 
598 	if (global_gitconfig_path) {
599 		/* Read settings from ~/.gitconfig. */
600 		int dummy_repo_version;
601 		err = got_repo_read_gitconfig(&dummy_repo_version,
602 		    &repo->global_gitconfig_author_name,
603 		    &repo->global_gitconfig_author_email,
604 		    NULL, NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
605 		if (err)
606 			return err;
607 	}
608 
609 	/* Read repository's .git/config file. */
610 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
611 	if (repo_gitconfig_path == NULL)
612 		return got_error_from_errno("got_repo_get_path_gitconfig");
613 
614 	err = got_repo_read_gitconfig(
615 	    &repo->gitconfig_repository_format_version,
616 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
617 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
618 	    &repo->gitconfig_owner, &object_format,
619 	    &repo->extensions, &repo->nextensions,
620 	    repo_gitconfig_path);
621 	if (err)
622 		goto done;
623 
624 	if (!object_format)
625 		repo->algo = GOT_HASH_SHA1;
626 	else {
627 		if (!strcmp(object_format, "sha256"))
628 			repo->algo = GOT_HASH_SHA256;
629 		else {
630 			err = got_error_msg(GOT_ERR_OBJECT_FORMAT,
631 			    object_format);
632 			goto done;
633 		}
634 	}
635 
636 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
637 		int i;
638 
639 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
640 			got_repo_free_remote_repo_data(
641 			    &repo->gitconfig_remotes[i]);
642 		}
643 		free(repo->gitconfig_remotes);
644 		repo->gitconfig_remotes = NULL;
645 		repo->ngitconfig_remotes = 0;
646 
647 		free(repo->gitconfig_author_name);
648 		repo->gitconfig_author_name = NULL;
649 		free(repo->gitconfig_author_email);
650 		repo->gitconfig_author_email = NULL;
651 
652 		free(repo->global_gitconfig_author_name);
653 		repo->global_gitconfig_author_name = NULL;
654 		free(repo->global_gitconfig_author_email);
655 		repo->global_gitconfig_author_email = NULL;
656 	}
657 
658 done:
659 	free(repo_gitconfig_path);
660 	return err;
661 }
662 
663 static const struct got_error *
664 read_gotconfig(struct got_repository *repo)
665 {
666 	const struct got_error *err = NULL;
667 	char *gotconfig_path;
668 
669 	gotconfig_path = got_repo_get_path_gotconfig(repo);
670 	if (gotconfig_path == NULL)
671 		return got_error_from_errno("got_repo_get_path_gotconfig");
672 
673 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
674 	free(gotconfig_path);
675 	return err;
676 }
677 
678 /* Supported repository format extensions. */
679 static const char *const repo_extensions[] = {
680 	"noop",			/* Got supports repository format version 1. */
681 	"preciousObjects",	/* Supported by gotadmin cleanup. */
682 	"worktreeConfig",	/* Got does not care about Git work trees. */
683 	"sha256",		/* hack */
684 };
685 
686 const struct got_error *
687 got_repo_open(struct got_repository **repop, const char *path,
688     const char *global_gitconfig_path, int *pack_fds)
689 {
690 	struct got_repository *repo = NULL;
691 	const struct got_error *err = NULL;
692 	char *repo_path = NULL;
693 	size_t i, j = 0;
694 
695 	*repop = NULL;
696 
697 	repo = calloc(1, sizeof(*repo));
698 	if (repo == NULL)
699 		return got_error_from_errno("calloc");
700 
701 	RB_INIT(&repo->packidx_bloom_filters);
702 	TAILQ_INIT(&repo->packidx_paths);
703 
704 	for (i = 0; i < nitems(repo->privsep_children); i++) {
705 		memset(&repo->privsep_children[i], 0,
706 		    sizeof(repo->privsep_children[0]));
707 		repo->privsep_children[i].imsg_fd = -1;
708 	}
709 
710 	err = got_object_cache_init(&repo->objcache,
711 	    GOT_OBJECT_CACHE_TYPE_OBJ);
712 	if (err)
713 		goto done;
714 	err = got_object_cache_init(&repo->treecache,
715 	    GOT_OBJECT_CACHE_TYPE_TREE);
716 	if (err)
717 		goto done;
718 	err = got_object_cache_init(&repo->commitcache,
719 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
720 	if (err)
721 		goto done;
722 	err = got_object_cache_init(&repo->tagcache,
723 	    GOT_OBJECT_CACHE_TYPE_TAG);
724 	if (err)
725 		goto done;
726 	err = got_object_cache_init(&repo->rawcache,
727 	    GOT_OBJECT_CACHE_TYPE_RAW);
728 	if (err)
729 		goto done;
730 
731 	err = get_pack_cache_size(&repo->pack_cache_size);
732 	if (err)
733 		goto done;
734 	for (i = 0; i < nitems(repo->packs); i++) {
735 		if (pack_fds != NULL && i < repo->pack_cache_size) {
736 			repo->packs[i].basefd = pack_fds[j++];
737 			repo->packs[i].accumfd = pack_fds[j++];
738 		} else {
739 			repo->packs[i].basefd = -1;
740 			repo->packs[i].accumfd = -1;
741 		}
742 	}
743 	for (i = 0; i < nitems(repo->tempfiles); i++)
744 		repo->tempfiles[i] = -1;
745 	repo->pinned_pack = -1;
746 	repo->pinned_packidx = -1;
747 	repo->pinned_pid = 0;
748 
749 	repo_path = realpath(path, NULL);
750 	if (repo_path == NULL) {
751 		err = got_error_from_errno2("realpath", path);
752 		goto done;
753 	}
754 
755 	for (;;) {
756 		char *parent_path;
757 
758 		err = open_repo(repo, repo_path);
759 		if (err == NULL)
760 			break;
761 		if (err->code != GOT_ERR_NOT_GIT_REPO)
762 			goto done;
763 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
764 			err = got_error(GOT_ERR_NOT_GIT_REPO);
765 			goto done;
766 		}
767 		err = got_path_dirname(&parent_path, repo_path);
768 		if (err)
769 			goto done;
770 		free(repo_path);
771 		repo_path = parent_path;
772 	}
773 
774 	err = read_gotconfig(repo);
775 	if (err)
776 		goto done;
777 
778 	err = read_gitconfig(repo, global_gitconfig_path);
779 	if (err)
780 		goto done;
781 	if (repo->gitconfig_repository_format_version > 1) {
782 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
783 		goto done;
784 	}
785 	if (repo->gitconfig_repository_format_version == 0 &&
786 	    repo->algo != GOT_HASH_SHA1) {
787 		err = got_error(GOT_ERR_OBJECT_FORMAT);
788 		goto done;
789 	}
790 	for (i = 0; i < repo->nextensions; i++) {
791 		char *ext = repo->extensions[i];
792 		int j, supported = 0;
793 		for (j = 0; j < nitems(repo_extensions); j++) {
794 			if (strcmp(ext, repo_extensions[j]) == 0) {
795 				supported = 1;
796 				break;
797 			}
798 		}
799 		if (!supported) {
800 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
801 			goto done;
802 		}
803 	}
804 
805 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
806 done:
807 	if (err)
808 		got_repo_close(repo);
809 	else
810 		*repop = repo;
811 	free(repo_path);
812 	return err;
813 }
814 
815 const struct got_error *
816 got_repo_close(struct got_repository *repo)
817 {
818 	const struct got_error *err = NULL, *child_err;
819 	struct got_packidx_bloom_filter *bf;
820 	size_t i;
821 
822 	for (i = 0; i < repo->pack_cache_size; i++) {
823 		if (repo->packidx_cache[i] == NULL)
824 			break;
825 		got_packidx_close(repo->packidx_cache[i]);
826 	}
827 
828 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
829 	    &repo->packidx_bloom_filters))) {
830 		RB_REMOVE(got_packidx_bloom_filter_tree,
831 		    &repo->packidx_bloom_filters, bf);
832 		bloom_free(bf->bloom);
833 		free(bf->bloom);
834 		free(bf);
835 	}
836 
837 	for (i = 0; i < repo->pack_cache_size; i++)
838 		if (repo->packs[i].path_packfile)
839 			if (repo->packs[i].path_packfile)
840 				got_pack_close(&repo->packs[i]);
841 
842 	free(repo->path);
843 	free(repo->path_git_dir);
844 
845 	got_object_cache_close(&repo->objcache);
846 	got_object_cache_close(&repo->treecache);
847 	got_object_cache_close(&repo->commitcache);
848 	got_object_cache_close(&repo->tagcache);
849 	got_object_cache_close(&repo->rawcache);
850 
851 	for (i = 0; i < nitems(repo->privsep_children); i++) {
852 		if (repo->privsep_children[i].imsg_fd == -1)
853 			continue;
854 		imsg_clear(repo->privsep_children[i].ibuf);
855 		free(repo->privsep_children[i].ibuf);
856 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
857 		child_err = got_privsep_wait_for_child(
858 		    repo->privsep_children[i].pid);
859 		if (child_err && err == NULL)
860 			err = child_err;
861 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
862 		    err == NULL)
863 			err = got_error_from_errno("close");
864 	}
865 
866 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
867 	    err == NULL)
868 		err = got_error_from_errno("close");
869 
870 	if (repo->gotconfig)
871 		got_gotconfig_free(repo->gotconfig);
872 	free(repo->gitconfig_author_name);
873 	free(repo->gitconfig_author_email);
874 	for (i = 0; i < repo->ngitconfig_remotes; i++)
875 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
876 	free(repo->gitconfig_remotes);
877 	for (i = 0; i < repo->nextensions; i++)
878 		free(repo->extensions[i]);
879 	free(repo->extensions);
880 
881 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
882 	free(repo);
883 
884 	return err;
885 }
886 
887 void
888 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
889 {
890 	int i;
891 
892 	free(repo->name);
893 	repo->name = NULL;
894 	free(repo->fetch_url);
895 	repo->fetch_url = NULL;
896 	free(repo->send_url);
897 	repo->send_url = NULL;
898 	for (i = 0; i < repo->nfetch_branches; i++)
899 		free(repo->fetch_branches[i]);
900 	free(repo->fetch_branches);
901 	repo->fetch_branches = NULL;
902 	repo->nfetch_branches = 0;
903 	for (i = 0; i < repo->nsend_branches; i++)
904 		free(repo->send_branches[i]);
905 	free(repo->send_branches);
906 	repo->send_branches = NULL;
907 	repo->nsend_branches = 0;
908 }
909 
910 const struct got_error *
911 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
912     const char *input_path)
913 {
914 	const struct got_error *err = NULL;
915 	const char *repo_abspath = NULL;
916 	size_t repolen, len;
917 	char *canonpath, *path = NULL;
918 
919 	*in_repo_path = NULL;
920 
921 	canonpath = strdup(input_path);
922 	if (canonpath == NULL) {
923 		err = got_error_from_errno("strdup");
924 		goto done;
925 	}
926 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
927 	if (err)
928 		goto done;
929 
930 	repo_abspath = got_repo_get_path(repo);
931 
932 	if (canonpath[0] == '\0') {
933 		path = strdup(canonpath);
934 		if (path == NULL) {
935 			err = got_error_from_errno("strdup");
936 			goto done;
937 		}
938 	} else {
939 		path = realpath(canonpath, NULL);
940 		if (path == NULL) {
941 			if (errno != ENOENT) {
942 				err = got_error_from_errno2("realpath",
943 				    canonpath);
944 				goto done;
945 			}
946 			/*
947 			 * Path is not on disk.
948 			 * Assume it is already relative to repository root.
949 			 */
950 			path = strdup(canonpath);
951 			if (path == NULL) {
952 				err = got_error_from_errno("strdup");
953 				goto done;
954 			}
955 		}
956 
957 		repolen = strlen(repo_abspath);
958 		len = strlen(path);
959 
960 
961 		if (strcmp(path, repo_abspath) == 0) {
962 			free(path);
963 			path = strdup("");
964 			if (path == NULL) {
965 				err = got_error_from_errno("strdup");
966 				goto done;
967 			}
968 		} else if (len > repolen &&
969 		    got_path_is_child(path, repo_abspath, repolen)) {
970 			/* Matched an on-disk path inside repository. */
971 			if (got_repo_is_bare(repo)) {
972 				/*
973 				 * Matched an on-disk path inside repository
974 				 * database. Treat input as repository-relative.
975 				 */
976 				free(path);
977 				path = canonpath;
978 				canonpath = NULL;
979 			} else {
980 				char *child;
981 				/* Strip common prefix with repository path. */
982 				err = got_path_skip_common_ancestor(&child,
983 				    repo_abspath, path);
984 				if (err)
985 					goto done;
986 				free(path);
987 				path = child;
988 			}
989 		} else {
990 			/*
991 			 * Matched unrelated on-disk path.
992 			 * Treat input as repository-relative.
993 			 */
994 			free(path);
995 			path = canonpath;
996 			canonpath = NULL;
997 		}
998 	}
999 
1000 	/* Make in-repository path absolute */
1001 	if (path[0] != '/') {
1002 		char *abspath;
1003 		if (asprintf(&abspath, "/%s", path) == -1) {
1004 			err = got_error_from_errno("asprintf");
1005 			goto done;
1006 		}
1007 		free(path);
1008 		path = abspath;
1009 	}
1010 
1011 done:
1012 	free(canonpath);
1013 	if (err)
1014 		free(path);
1015 	else
1016 		*in_repo_path = path;
1017 	return err;
1018 }
1019 
1020 static const struct got_error *
1021 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1022     const char *path_packidx)
1023 {
1024 	const struct got_error *err = NULL;
1025 	size_t i;
1026 
1027 	for (i = 0; i < repo->pack_cache_size; i++) {
1028 		if (repo->packidx_cache[i] == NULL)
1029 			break;
1030 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1031 		    path_packidx) == 0) {
1032 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1033 		}
1034 	}
1035 	if (i == repo->pack_cache_size) {
1036 		do {
1037 			i--;
1038 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1039 		    i == repo->pinned_packidx);
1040 		err = got_packidx_close(repo->packidx_cache[i]);
1041 		if (err)
1042 			return err;
1043 	}
1044 
1045 	repo->packidx_cache[i] = packidx;
1046 
1047 	return NULL;
1048 }
1049 
1050 int
1051 got_repo_is_packidx_filename(struct got_repository *repo, const char *name,
1052     size_t len)
1053 {
1054 	size_t hashlen = SHA1_DIGEST_STRING_LENGTH;
1055 
1056 	if (repo->algo == GOT_HASH_SHA256)
1057 		hashlen = SHA256_DIGEST_STRING_LENGTH;
1058 
1059 	if (len != GOT_PACKIDX_NAMELEN)
1060 		return 0;
1061 
1062 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1063 		return 0;
1064 
1065 	if (strcmp(name + strlen(GOT_PACK_PREFIX) + hashlen - 1,
1066 	    GOT_PACKIDX_SUFFIX) != 0)
1067 		return 0;
1068 
1069 	return 1;
1070 }
1071 
1072 static struct got_packidx_bloom_filter *
1073 get_packidx_bloom_filter(struct got_repository *repo,
1074     const char *path, size_t path_len)
1075 {
1076 	struct got_packidx_bloom_filter key;
1077 
1078 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1079 		return NULL; /* XXX */
1080 	key.path_len = path_len;
1081 
1082 	return RB_FIND(got_packidx_bloom_filter_tree,
1083 	    &repo->packidx_bloom_filters, &key);
1084 }
1085 
1086 int
1087 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1088     const char *path_packidx, struct got_object_id *id)
1089 {
1090 	struct got_packidx_bloom_filter *bf;
1091 
1092 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1093 	if (bf)
1094 		return bloom_check(bf->bloom, id->hash, sizeof(id->hash));
1095 
1096 	/* No bloom filter means this pack index must be searched. */
1097 	return 1;
1098 }
1099 
1100 static const struct got_error *
1101 add_packidx_bloom_filter(struct got_repository *repo,
1102     struct got_packidx *packidx, const char *path_packidx)
1103 {
1104 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1105 	struct got_packidx_bloom_filter *bf;
1106 	size_t len;
1107 
1108 	/*
1109 	 * Don't use bloom filters for very large pack index files.
1110 	 * Large pack files will contain a relatively large fraction
1111 	 * of our objects so we will likely need to visit them anyway.
1112 	 * The more objects a pack file contains the higher the probability
1113 	 * of a false-positive match from the bloom filter. And reading
1114 	 * all object IDs from a large pack index file can be expensive.
1115 	 */
1116 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1117 		return NULL;
1118 
1119 	/* Do we already have a filter for this pack index? */
1120 	if (get_packidx_bloom_filter(repo, path_packidx,
1121 	    strlen(path_packidx)) != NULL)
1122 		return NULL;
1123 
1124 	bf = calloc(1, sizeof(*bf));
1125 	if (bf == NULL)
1126 		return got_error_from_errno("calloc");
1127 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1128 	if (bf->bloom == NULL) {
1129 		free(bf);
1130 		return got_error_from_errno("calloc");
1131 	}
1132 
1133 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1134 	if (len >= sizeof(bf->path)) {
1135 		free(bf->bloom);
1136 		free(bf);
1137 		return got_error(GOT_ERR_NO_SPACE);
1138 	}
1139 	bf->path_len = len;
1140 
1141 	/* Minimum size supported by our bloom filter is 1000 entries. */
1142 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1143 	for (i = 0; i < nobjects; i++) {
1144 		struct got_packidx_object_id *id;
1145 		id = &packidx->hdr.sorted_ids[i];
1146 		bloom_add(bf->bloom, id->hash, sizeof(id->hash));
1147 	}
1148 
1149 	RB_INSERT(got_packidx_bloom_filter_tree,
1150 	    &repo->packidx_bloom_filters, bf);
1151 	return NULL;
1152 }
1153 
1154 static void
1155 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1156 {
1157 	struct got_pathlist_entry *pe;
1158 
1159 	while (!TAILQ_EMPTY(packidx_paths)) {
1160 		pe = TAILQ_FIRST(packidx_paths);
1161 		TAILQ_REMOVE(packidx_paths, pe, entry);
1162 		free((char *)pe->path);
1163 		free(pe);
1164 	}
1165 }
1166 
1167 static const struct got_error *
1168 refresh_packidx_paths(struct got_repository *repo)
1169 {
1170 	const struct got_error *err = NULL;
1171 	char *objects_pack_dir = NULL;
1172 	struct stat sb;
1173 
1174 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1175 	if (objects_pack_dir == NULL)
1176 		return got_error_from_errno("got_repo_get_path_objects_pack");
1177 
1178 	if (stat(objects_pack_dir, &sb) == -1) {
1179 		if (errno != ENOENT) {
1180 			err = got_error_from_errno2("stat", objects_pack_dir);
1181 			goto done;
1182 		}
1183 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1184 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1185 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1186 		purge_packidx_paths(&repo->packidx_paths);
1187 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1188 		if (err)
1189 			goto done;
1190 	}
1191 done:
1192 	free(objects_pack_dir);
1193 	return err;
1194 }
1195 
1196 const struct got_error *
1197 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1198     struct got_repository *repo, struct got_object_id *id)
1199 {
1200 	const struct got_error *err;
1201 	struct got_pathlist_entry *pe;
1202 	size_t i;
1203 
1204 	/* Search pack index cache. */
1205 	for (i = 0; i < repo->pack_cache_size; i++) {
1206 		if (repo->packidx_cache[i] == NULL)
1207 			break;
1208 		if (!got_repo_check_packidx_bloom_filter(repo,
1209 		    repo->packidx_cache[i]->path_packidx, id))
1210 			continue; /* object will not be found in this index */
1211 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1212 		if (*idx != -1) {
1213 			*packidx = repo->packidx_cache[i];
1214 			/*
1215 			 * Move this cache entry to the front. Repeatedly
1216 			 * searching a wrong pack index can be expensive.
1217 			 */
1218 			if (i > 0) {
1219 				memmove(&repo->packidx_cache[1],
1220 				    &repo->packidx_cache[0],
1221 				    i * sizeof(repo->packidx_cache[0]));
1222 				repo->packidx_cache[0] = *packidx;
1223 				if (repo->pinned_packidx >= 0 &&
1224 				    repo->pinned_packidx < i)
1225 					repo->pinned_packidx++;
1226 				else if (repo->pinned_packidx == i)
1227 					repo->pinned_packidx = 0;
1228 			}
1229 			return NULL;
1230 		}
1231 	}
1232 	/* No luck. Search the filesystem. */
1233 
1234 	err = refresh_packidx_paths(repo);
1235 	if (err)
1236 		return err;
1237 
1238 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1239 		const char *path_packidx = pe->path;
1240 		int is_cached = 0;
1241 
1242 		if (!got_repo_check_packidx_bloom_filter(repo,
1243 		    pe->path, id))
1244 			continue; /* object will not be found in this index */
1245 
1246 		for (i = 0; i < repo->pack_cache_size; i++) {
1247 			if (repo->packidx_cache[i] == NULL)
1248 				break;
1249 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1250 			    path_packidx) == 0) {
1251 				is_cached = 1;
1252 				break;
1253 			}
1254 		}
1255 		if (is_cached)
1256 			continue; /* already searched */
1257 
1258 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1259 		    path_packidx, 0, repo->algo);
1260 		if (err)
1261 			goto done;
1262 
1263 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1264 		if (err)
1265 			goto done;
1266 
1267 		err = cache_packidx(repo, *packidx, path_packidx);
1268 		if (err)
1269 			goto done;
1270 
1271 		*idx = got_packidx_get_object_idx(*packidx, id);
1272 		if (*idx != -1) {
1273 			err = NULL; /* found the object */
1274 			goto done;
1275 		}
1276 	}
1277 
1278 	err = got_error_no_obj(id);
1279 done:
1280 	return err;
1281 }
1282 
1283 const struct got_error *
1284 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1285     struct got_repository *repo)
1286 {
1287 	const struct got_error *err = NULL;
1288 	DIR *packdir = NULL;
1289 	struct dirent *dent;
1290 	char *path_packidx = NULL;
1291 	int packdir_fd;
1292 	struct stat sb;
1293 
1294 	packdir_fd = openat(got_repo_get_fd(repo),
1295 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1296 	if (packdir_fd == -1) {
1297 		return got_error_from_errno_fmt("openat: %s/%s",
1298 		    got_repo_get_path_git_dir(repo),
1299 		    GOT_OBJECTS_PACK_DIR);
1300 	}
1301 
1302 	packdir = fdopendir(packdir_fd);
1303 	if (packdir == NULL) {
1304 		err = got_error_from_errno("fdopendir");
1305 		goto done;
1306 	}
1307 
1308 	if (fstat(packdir_fd, &sb) == -1) {
1309 		err = got_error_from_errno("fstat");
1310 		goto done;
1311 	}
1312 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1313 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1314 
1315 	while ((dent = readdir(packdir)) != NULL) {
1316 		if (!got_repo_is_packidx_filename(repo, dent->d_name,
1317 		    dent->d_namlen))
1318 			continue;
1319 
1320 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1321 		    dent->d_name) == -1) {
1322 			err = got_error_from_errno("asprintf");
1323 			path_packidx = NULL;
1324 			break;
1325 		}
1326 
1327 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1328 		if (err)
1329 			break;
1330 	}
1331 done:
1332 	if (err)
1333 		free(path_packidx);
1334 	if (packdir && closedir(packdir) != 0 && err == NULL)
1335 		err = got_error_from_errno("closedir");
1336 	return err;
1337 }
1338 
1339 const struct got_error *
1340 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1341     struct got_repository *repo)
1342 {
1343 	const struct got_error *err;
1344 	size_t i;
1345 
1346 	*packidx = NULL;
1347 
1348 	/* Search pack index cache. */
1349 	for (i = 0; i < repo->pack_cache_size; i++) {
1350 		if (repo->packidx_cache[i] == NULL)
1351 			break;
1352 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1353 		    path_packidx) == 0) {
1354 			*packidx = repo->packidx_cache[i];
1355 			return NULL;
1356 		}
1357 	}
1358 	/* No luck. Search the filesystem. */
1359 
1360 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1361 	    path_packidx, 0, repo->algo);
1362 	if (err)
1363 		return err;
1364 
1365 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1366 	if (err)
1367 		goto done;
1368 
1369 	err = cache_packidx(repo, *packidx, path_packidx);
1370 done:
1371 	if (err) {
1372 		got_packidx_close(*packidx);
1373 		*packidx = NULL;
1374 	}
1375 	return err;
1376 }
1377 
1378 static const struct got_error *
1379 read_packfile_hdr(int fd, struct got_packidx *packidx)
1380 {
1381 	const struct got_error *err = NULL;
1382 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1383 	struct got_packfile_hdr hdr;
1384 	ssize_t n;
1385 
1386 	n = read(fd, &hdr, sizeof(hdr));
1387 	if (n < 0)
1388 		return got_error_from_errno("read");
1389 	if (n != sizeof(hdr))
1390 		return got_error(GOT_ERR_BAD_PACKFILE);
1391 
1392 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1393 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1394 	    be32toh(hdr.nobjects) != totobj)
1395 		err = got_error(GOT_ERR_BAD_PACKFILE);
1396 
1397 	return err;
1398 }
1399 
1400 static const struct got_error *
1401 open_packfile(int *fd, struct got_repository *repo,
1402     const char *relpath, struct got_packidx *packidx)
1403 {
1404 	const struct got_error *err = NULL;
1405 
1406 	*fd = openat(got_repo_get_fd(repo), relpath,
1407 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1408 	if (*fd == -1)
1409 		return got_error_from_errno_fmt("openat: %s/%s",
1410 		    got_repo_get_path_git_dir(repo), relpath);
1411 
1412 	if (packidx) {
1413 		err = read_packfile_hdr(*fd, packidx);
1414 		if (err) {
1415 			close(*fd);
1416 			*fd = -1;
1417 		}
1418 	}
1419 
1420 	return err;
1421 }
1422 
1423 const struct got_error *
1424 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1425     const char *path_packfile, struct got_packidx *packidx)
1426 {
1427 	const struct got_error *err = NULL;
1428 	struct got_pack *pack = NULL;
1429 	struct stat sb;
1430 	size_t i;
1431 
1432 	if (packp)
1433 		*packp = NULL;
1434 
1435 	for (i = 0; i < repo->pack_cache_size; i++) {
1436 		pack = &repo->packs[i];
1437 		if (pack->path_packfile == NULL)
1438 			break;
1439 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1440 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1441 	}
1442 
1443 	if (i == repo->pack_cache_size) {
1444 		struct got_pack tmp;
1445 		do {
1446 			i--;
1447 		} while (i > 0 && repo->pinned_pack >= 0 &&
1448 		    i == repo->pinned_pack);
1449 		err = got_pack_close(&repo->packs[i]);
1450 		if (err)
1451 			return err;
1452 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1453 			return got_error_from_errno("ftruncate");
1454 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1455 			return got_error_from_errno("ftruncate");
1456 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1457 		memcpy(&repo->packs[i], &repo->packs[0],
1458 		    sizeof(repo->packs[i]));
1459 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1460 		if (repo->pinned_pack == 0)
1461 			repo->pinned_pack = i;
1462 		else if (repo->pinned_pack == i)
1463 			repo->pinned_pack = 0;
1464 		i = 0;
1465 	}
1466 
1467 	pack = &repo->packs[i];
1468 
1469 	pack->path_packfile = strdup(path_packfile);
1470 	if (pack->path_packfile == NULL) {
1471 		err = got_error_from_errno("strdup");
1472 		goto done;
1473 	}
1474 
1475 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1476 	if (err)
1477 		goto done;
1478 
1479 	if (fstat(pack->fd, &sb) != 0) {
1480 		err = got_error_from_errno("fstat");
1481 		goto done;
1482 	}
1483 	pack->filesize = sb.st_size;
1484 
1485 	pack->privsep_child = NULL;
1486 
1487 	err = got_delta_cache_alloc(&pack->delta_cache);
1488 	if (err)
1489 		goto done;
1490 
1491 #ifndef GOT_PACK_NO_MMAP
1492 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1493 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1494 		    pack->fd, 0);
1495 		if (pack->map == MAP_FAILED) {
1496 			if (errno != ENOMEM) {
1497 				err = got_error_from_errno("mmap");
1498 				goto done;
1499 			}
1500 			pack->map = NULL; /* fall back to read(2) */
1501 		}
1502 	}
1503 #endif
1504 done:
1505 	if (err) {
1506 		if (pack)
1507 			got_pack_close(pack);
1508 	} else if (packp)
1509 		*packp = pack;
1510 	return err;
1511 }
1512 
1513 struct got_pack *
1514 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1515 {
1516 	struct got_pack *pack = NULL;
1517 	size_t i;
1518 
1519 	for (i = 0; i < repo->pack_cache_size; i++) {
1520 		pack = &repo->packs[i];
1521 		if (pack->path_packfile == NULL)
1522 			break;
1523 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1524 			return pack;
1525 	}
1526 
1527 	return NULL;
1528 }
1529 
1530 const struct got_error *
1531 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1532     struct got_pack *pack)
1533 {
1534 	size_t i;
1535 	int pinned_pack = -1, pinned_packidx = -1;
1536 
1537 	for (i = 0; i < repo->pack_cache_size; i++) {
1538 		if (repo->packidx_cache[i] &&
1539 		    strcmp(repo->packidx_cache[i]->path_packidx,
1540 		    packidx->path_packidx) == 0)
1541 			pinned_packidx = i;
1542 		if (repo->packs[i].path_packfile &&
1543 		    strcmp(repo->packs[i].path_packfile,
1544 		    pack->path_packfile) == 0)
1545 			pinned_pack = i;
1546 	}
1547 
1548 	if (pinned_packidx == -1 || pinned_pack == -1)
1549 		return got_error(GOT_ERR_PIN_PACK);
1550 
1551 	repo->pinned_pack = pinned_pack;
1552 	repo->pinned_packidx = pinned_packidx;
1553 	if (repo->packs[pinned_pack].privsep_child)
1554 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1555 	return NULL;
1556 }
1557 
1558 struct got_pack *
1559 got_repo_get_pinned_pack(struct got_repository *repo)
1560 {
1561 	if (repo->pinned_pack >= 0 &&
1562 	    repo->pinned_pack < repo->pack_cache_size)
1563 		return &repo->packs[repo->pinned_pack];
1564 
1565 	return NULL;
1566 }
1567 
1568 void
1569 got_repo_unpin_pack(struct got_repository *repo)
1570 {
1571 	repo->pinned_packidx = -1;
1572 	repo->pinned_pack = -1;
1573 	repo->pinned_pid = 0;
1574 }
1575 
1576 const struct got_error *
1577 got_repo_init(const char *repo_path, const char *head_name)
1578 {
1579 	const struct got_error *err = NULL;
1580 	const char *dirnames[] = {
1581 		GOT_OBJECTS_DIR,
1582 		GOT_OBJECTS_PACK_DIR,
1583 		GOT_REFS_DIR,
1584 	};
1585 	const char *description_str = "Unnamed repository; "
1586 	    "edit this file 'description' to name the repository.";
1587 	const char *headref = "ref: refs/heads/";
1588 	const char *gitconfig_str = "[core]\n"
1589 	    "\trepositoryformatversion = 0\n"
1590 	    "\tfilemode = true\n"
1591 	    "\tbare = true\n";
1592 	char *headref_str, *path;
1593 	size_t i;
1594 
1595 	if (!got_path_dir_is_empty(repo_path))
1596 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1597 
1598 	for (i = 0; i < nitems(dirnames); i++) {
1599 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1600 			return got_error_from_errno("asprintf");
1601 		}
1602 		err = got_path_mkdir(path);
1603 		free(path);
1604 		if (err)
1605 			return err;
1606 	}
1607 
1608 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1609 		return got_error_from_errno("asprintf");
1610 	err = got_path_create_file(path, description_str);
1611 	free(path);
1612 	if (err)
1613 		return err;
1614 
1615 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1616 		return got_error_from_errno("asprintf");
1617 	if (asprintf(&headref_str, "%s%s", headref,
1618 	    head_name ? head_name : "main") == -1) {
1619 		free(path);
1620 		return got_error_from_errno("asprintf");
1621 	}
1622 	err = got_path_create_file(path, headref_str);
1623 	free(headref_str);
1624 	free(path);
1625 	if (err)
1626 		return err;
1627 
1628 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1629 		return got_error_from_errno("asprintf");
1630 	err = got_path_create_file(path, gitconfig_str);
1631 	free(path);
1632 	if (err)
1633 		return err;
1634 
1635 	return NULL;
1636 }
1637 
1638 static const struct got_error *
1639 match_packed_object(struct got_object_id **unique_id,
1640     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1641 {
1642 	const struct got_error *err = NULL;
1643 	struct got_object_id_queue matched_ids;
1644 	struct got_pathlist_entry *pe;
1645 
1646 	STAILQ_INIT(&matched_ids);
1647 
1648 	err = refresh_packidx_paths(repo);
1649 	if (err)
1650 		return err;
1651 
1652 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1653 		const char *path_packidx = pe->path;
1654 		struct got_packidx *packidx;
1655 		struct got_object_qid *qid;
1656 
1657 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1658 		    path_packidx, 0, repo->algo);
1659 		if (err)
1660 			break;
1661 
1662 		err = got_packidx_match_id_str_prefix(&matched_ids,
1663 		    packidx, id_str_prefix);
1664 		if (err) {
1665 			got_packidx_close(packidx);
1666 			break;
1667 		}
1668 		err = got_packidx_close(packidx);
1669 		if (err)
1670 			break;
1671 
1672 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1673 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1674 				int matched_type;
1675 				err = got_object_get_type(&matched_type, repo,
1676 				    &qid->id);
1677 				if (err)
1678 					goto done;
1679 				if (matched_type != obj_type)
1680 					continue;
1681 			}
1682 			if (*unique_id == NULL) {
1683 				*unique_id = got_object_id_dup(&qid->id);
1684 				if (*unique_id == NULL) {
1685 					err = got_error_from_errno("malloc");
1686 					goto done;
1687 				}
1688 			} else {
1689 				if (got_object_id_cmp(*unique_id,
1690 				    &qid->id) == 0)
1691 					continue; /* packed multiple times */
1692 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1693 				goto done;
1694 			}
1695 		}
1696 	}
1697 done:
1698 	got_object_id_queue_free(&matched_ids);
1699 	if (err) {
1700 		free(*unique_id);
1701 		*unique_id = NULL;
1702 	}
1703 	return err;
1704 }
1705 
1706 static const struct got_error *
1707 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1708     const char *object_dir, const char *id_str_prefix, int obj_type,
1709     struct got_repository *repo)
1710 {
1711 	const struct got_error *err = NULL;
1712 	char *path, *id_str = NULL;
1713 	DIR *dir = NULL;
1714 	struct dirent *dent;
1715 	struct got_object_id id;
1716 
1717 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1718 		err = got_error_from_errno("asprintf");
1719 		goto done;
1720 	}
1721 
1722 	dir = opendir(path);
1723 	if (dir == NULL) {
1724 		if (errno == ENOENT) {
1725 			err = NULL;
1726 			goto done;
1727 		}
1728 		err = got_error_from_errno2("opendir", path);
1729 		goto done;
1730 	}
1731 	while ((dent = readdir(dir)) != NULL) {
1732 		int cmp;
1733 
1734 		free(id_str);
1735 		id_str = NULL;
1736 
1737 		if (strcmp(dent->d_name, ".") == 0 ||
1738 		    strcmp(dent->d_name, "..") == 0)
1739 			continue;
1740 
1741 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1742 			err = got_error_from_errno("asprintf");
1743 			goto done;
1744 		}
1745 
1746 		id.algo = repo->algo;
1747 		if (!got_parse_hash_digest(id.hash, id_str, repo->algo, NULL))
1748 			continue;
1749 
1750 		/*
1751 		 * Directory entries do not necessarily appear in
1752 		 * sorted order, so we must iterate over all of them.
1753 		 */
1754 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1755 		if (cmp != 0)
1756 			continue;
1757 
1758 		if (*unique_id == NULL) {
1759 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1760 				int matched_type;
1761 				err = got_object_get_type(&matched_type, repo,
1762 				    &id);
1763 				if (err)
1764 					goto done;
1765 				if (matched_type != obj_type)
1766 					continue;
1767 			}
1768 			*unique_id = got_object_id_dup(&id);
1769 			if (*unique_id == NULL) {
1770 				err = got_error_from_errno("got_object_id_dup");
1771 				goto done;
1772 			}
1773 		} else {
1774 			if (got_object_id_cmp(*unique_id, &id) == 0)
1775 				continue; /* both packed and loose */
1776 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1777 			goto done;
1778 		}
1779 	}
1780 done:
1781 	if (dir && closedir(dir) != 0 && err == NULL)
1782 		err = got_error_from_errno("closedir");
1783 	if (err) {
1784 		free(*unique_id);
1785 		*unique_id = NULL;
1786 	}
1787 	free(id_str);
1788 	free(path);
1789 	return err;
1790 }
1791 
1792 const struct got_error *
1793 got_repo_match_object_id_prefix(struct got_object_id **id,
1794     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1795 {
1796 	const struct got_error *err = NULL;
1797 	char *path_objects = NULL, *object_dir = NULL;
1798 	size_t len;
1799 	int i;
1800 
1801 	*id = NULL;
1802 
1803 	path_objects = got_repo_get_path_objects(repo);
1804 
1805 	len = strlen(id_str_prefix);
1806 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1807 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1808 		goto done;
1809 	}
1810 
1811 	for (i = 0; i < len; i++) {
1812 		if (isxdigit((unsigned char)id_str_prefix[i]))
1813 			continue;
1814 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1815 		goto done;
1816 	}
1817 
1818 	if (len >= 2) {
1819 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1820 		if (err)
1821 			goto done;
1822 		object_dir = strndup(id_str_prefix, 2);
1823 		if (object_dir == NULL) {
1824 			err = got_error_from_errno("strdup");
1825 			goto done;
1826 		}
1827 		err = match_loose_object(id, path_objects, object_dir,
1828 		    id_str_prefix, obj_type, repo);
1829 	} else if (len == 1) {
1830 		int i;
1831 		for (i = 0; i < 0xf; i++) {
1832 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1833 			    == -1) {
1834 				err = got_error_from_errno("asprintf");
1835 				goto done;
1836 			}
1837 			err = match_packed_object(id, repo, object_dir,
1838 			    obj_type);
1839 			if (err)
1840 				goto done;
1841 			err = match_loose_object(id, path_objects, object_dir,
1842 			    id_str_prefix, obj_type, repo);
1843 			if (err)
1844 				goto done;
1845 		}
1846 	} else {
1847 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1848 		goto done;
1849 	}
1850 done:
1851 	free(path_objects);
1852 	free(object_dir);
1853 	if (err) {
1854 		free(*id);
1855 		*id = NULL;
1856 	} else if (*id == NULL) {
1857 		switch (obj_type) {
1858 		case GOT_OBJ_TYPE_BLOB:
1859 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1860 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1861 			break;
1862 		case GOT_OBJ_TYPE_TREE:
1863 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1864 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1865 			break;
1866 		case GOT_OBJ_TYPE_COMMIT:
1867 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1868 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1869 			break;
1870 		case GOT_OBJ_TYPE_TAG:
1871 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1872 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1873 			break;
1874 		default:
1875 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1876 			break;
1877 		}
1878 	}
1879 
1880 	return err;
1881 }
1882 
1883 const struct got_error *
1884 got_repo_match_object_id(struct got_object_id **id, char **label,
1885     const char *id_str, int obj_type, struct got_reflist_head *refs,
1886     struct got_repository *repo)
1887 {
1888 	const struct got_error *err;
1889 	struct got_tag_object *tag;
1890 	struct got_reference *ref = NULL;
1891 
1892 	*id = NULL;
1893 	if (label)
1894 		*label = NULL;
1895 
1896 	if (refs) {
1897 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1898 		    refs, repo);
1899 		if (err == NULL) {
1900 			*id = got_object_id_dup(
1901 			    got_object_tag_get_object_id(tag));
1902 			if (*id == NULL)
1903 				err = got_error_from_errno("got_object_id_dup");
1904 			else if (label && asprintf(label, "refs/tags/%s",
1905 			    got_object_tag_get_name(tag)) == -1) {
1906 				err = got_error_from_errno("asprintf");
1907 				free(*id);
1908 				*id = NULL;
1909 			}
1910 			got_object_tag_close(tag);
1911 			return err;
1912 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1913 		    err->code != GOT_ERR_NO_OBJ)
1914 			return err;
1915 	}
1916 
1917 	err = got_ref_open(&ref, repo, id_str, 0);
1918 	if (err == NULL) {
1919 		err = got_ref_resolve(id, repo, ref);
1920 		if (err)
1921 			goto done;
1922 		if (label) {
1923 			*label = strdup(got_ref_get_name(ref));
1924 			if (*label == NULL) {
1925 				err = got_error_from_errno("strdup");
1926 				goto done;
1927 			}
1928 		}
1929 	} else {
1930 		if (err->code != GOT_ERR_NOT_REF &&
1931 		    err->code != GOT_ERR_BAD_REF_NAME)
1932 			goto done;
1933 		err = got_repo_match_object_id_prefix(id, id_str,
1934 		    obj_type, repo);
1935 		if (err) {
1936 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1937 				err = got_error_not_ref(id_str);
1938 			goto done;
1939 		}
1940 		if (label) {
1941 			err = got_object_id_str(label, *id);
1942 			if (*label == NULL) {
1943 				err = got_error_from_errno("strdup");
1944 				goto done;
1945 			}
1946 		}
1947 	}
1948 done:
1949 	if (ref)
1950 		got_ref_close(ref);
1951 	return err;
1952 }
1953 
1954 const struct got_error *
1955 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1956     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1957 {
1958 	const struct got_error *err = NULL;
1959 	struct got_reflist_entry *re;
1960 	struct got_object_id *tag_id;
1961 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1962 
1963 	*tag = NULL;
1964 
1965 	TAILQ_FOREACH(re, refs, entry) {
1966 		const char *refname;
1967 		refname = got_ref_get_name(re->ref);
1968 		if (got_ref_is_symbolic(re->ref))
1969 			continue;
1970 		if (strncmp(refname, "refs/tags/", 10) != 0)
1971 			continue;
1972 		if (!name_is_absolute)
1973 			refname += strlen("refs/tags/");
1974 		if (strcmp(refname, name) != 0)
1975 			continue;
1976 		err = got_ref_resolve(&tag_id, repo, re->ref);
1977 		if (err)
1978 			break;
1979 		err = got_object_open_as_tag(tag, repo, tag_id);
1980 		free(tag_id);
1981 		if (err)
1982 			break;
1983 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1984 		    got_object_tag_get_object_type(*tag) == obj_type)
1985 			break;
1986 		got_object_tag_close(*tag);
1987 		*tag = NULL;
1988 	}
1989 
1990 	if (err == NULL && *tag == NULL)
1991 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1992 		    GOT_OBJ_LABEL_TAG, name);
1993 	return err;
1994 }
1995 
1996 static const struct got_error *
1997 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1998     const char *name, mode_t mode, struct got_object_id *blob_id)
1999 {
2000 	const struct got_error *err = NULL;
2001 
2002 	 *new_te = NULL;
2003 
2004 	*new_te = calloc(1, sizeof(**new_te));
2005 	if (*new_te == NULL)
2006 		return got_error_from_errno("calloc");
2007 
2008 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2009 	    sizeof((*new_te)->name)) {
2010 		err = got_error(GOT_ERR_NO_SPACE);
2011 		goto done;
2012 	}
2013 
2014 	if (S_ISLNK(mode)) {
2015 		(*new_te)->mode = S_IFLNK;
2016 	} else {
2017 		(*new_te)->mode = S_IFREG;
2018 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2019 	}
2020 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2021 done:
2022 	if (err && *new_te) {
2023 		free(*new_te);
2024 		*new_te = NULL;
2025 	}
2026 	return err;
2027 }
2028 
2029 static const struct got_error *
2030 import_file(struct got_tree_entry **new_te, struct dirent *de,
2031     const char *path, struct got_repository *repo)
2032 {
2033 	const struct got_error *err;
2034 	struct got_object_id *blob_id = NULL;
2035 	char *filepath;
2036 	struct stat sb;
2037 
2038 	if (asprintf(&filepath, "%s%s%s", path,
2039 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2040 		return got_error_from_errno("asprintf");
2041 
2042 	if (lstat(filepath, &sb) != 0) {
2043 		err = got_error_from_errno2("lstat", path);
2044 		goto done;
2045 	}
2046 
2047 	err = got_object_blob_create(&blob_id, filepath, repo);
2048 	if (err)
2049 		goto done;
2050 
2051 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2052 	    blob_id);
2053 done:
2054 	free(filepath);
2055 	if (err)
2056 		free(blob_id);
2057 	return err;
2058 }
2059 
2060 static const struct got_error *
2061 insert_tree_entry(struct got_tree_entry *new_te,
2062     struct got_pathlist_head *paths)
2063 {
2064 	const struct got_error *err = NULL;
2065 	struct got_pathlist_entry *new_pe;
2066 
2067 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2068 	if (err)
2069 		return err;
2070 	if (new_pe == NULL)
2071 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2072 	return NULL;
2073 }
2074 
2075 static const struct got_error *write_tree(struct got_object_id **,
2076     const char *, struct got_pathlist_head *, struct got_repository *,
2077     got_repo_import_cb progress_cb, void *progress_arg);
2078 
2079 static const struct got_error *
2080 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2081     const char *path, struct got_pathlist_head *ignores,
2082     struct got_repository *repo,
2083     got_repo_import_cb progress_cb, void *progress_arg)
2084 {
2085 	const struct got_error *err;
2086 	struct got_object_id *id = NULL;
2087 	char *subdirpath;
2088 
2089 	if (asprintf(&subdirpath, "%s%s%s", path,
2090 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2091 		return got_error_from_errno("asprintf");
2092 
2093 	(*new_te) = calloc(1, sizeof(**new_te));
2094 	if (*new_te == NULL)
2095 		return got_error_from_errno("calloc");
2096 	(*new_te)->mode = S_IFDIR;
2097 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2098 	    sizeof((*new_te)->name)) {
2099 		err = got_error(GOT_ERR_NO_SPACE);
2100 		goto done;
2101 	}
2102 	err = write_tree(&id, subdirpath, ignores,  repo,
2103 	    progress_cb, progress_arg);
2104 	if (err)
2105 		goto done;
2106 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2107 
2108 done:
2109 	free(id);
2110 	free(subdirpath);
2111 	if (err) {
2112 		free(*new_te);
2113 		*new_te = NULL;
2114 	}
2115 	return err;
2116 }
2117 
2118 static const struct got_error *
2119 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2120     struct got_pathlist_head *ignores, struct got_repository *repo,
2121     got_repo_import_cb progress_cb, void *progress_arg)
2122 {
2123 	const struct got_error *err = NULL;
2124 	DIR *dir;
2125 	struct dirent *de;
2126 	int nentries;
2127 	struct got_tree_entry *new_te = NULL;
2128 	struct got_pathlist_head paths;
2129 	struct got_pathlist_entry *pe;
2130 
2131 	*new_tree_id = NULL;
2132 
2133 	TAILQ_INIT(&paths);
2134 
2135 	dir = opendir(path_dir);
2136 	if (dir == NULL) {
2137 		err = got_error_from_errno2("opendir", path_dir);
2138 		goto done;
2139 	}
2140 
2141 	nentries = 0;
2142 	while ((de = readdir(dir)) != NULL) {
2143 		int ignore = 0;
2144 		int type;
2145 
2146 		if (strcmp(de->d_name, ".") == 0 ||
2147 		    strcmp(de->d_name, "..") == 0)
2148 			continue;
2149 
2150 		TAILQ_FOREACH(pe, ignores, entry) {
2151 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2152 				ignore = 1;
2153 				break;
2154 			}
2155 		}
2156 		if (ignore)
2157 			continue;
2158 
2159 		err = got_path_dirent_type(&type, path_dir, de);
2160 		if (err)
2161 			goto done;
2162 
2163 		if (type == DT_DIR) {
2164 			err = import_subdir(&new_te, de, path_dir,
2165 			    ignores, repo, progress_cb, progress_arg);
2166 			if (err) {
2167 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2168 					goto done;
2169 				err = NULL;
2170 				continue;
2171 			}
2172 		} else if (type == DT_REG || type == DT_LNK) {
2173 			err = import_file(&new_te, de, path_dir, repo);
2174 			if (err)
2175 				goto done;
2176 		} else
2177 			continue;
2178 
2179 		err = insert_tree_entry(new_te, &paths);
2180 		if (err)
2181 			goto done;
2182 		nentries++;
2183 	}
2184 
2185 	if (TAILQ_EMPTY(&paths)) {
2186 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2187 		    "cannot create tree without any entries");
2188 		goto done;
2189 	}
2190 
2191 	TAILQ_FOREACH(pe, &paths, entry) {
2192 		struct got_tree_entry *te = pe->data;
2193 		char *path;
2194 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2195 			continue;
2196 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2197 			err = got_error_from_errno("asprintf");
2198 			goto done;
2199 		}
2200 		err = (*progress_cb)(progress_arg, path);
2201 		free(path);
2202 		if (err)
2203 			goto done;
2204 	}
2205 
2206 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2207 done:
2208 	if (dir)
2209 		closedir(dir);
2210 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2211 	return err;
2212 }
2213 
2214 const struct got_error *
2215 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2216     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2217     struct got_repository *repo, got_repo_import_cb progress_cb,
2218     void *progress_arg)
2219 {
2220 	const struct got_error *err;
2221 	struct got_object_id *new_tree_id;
2222 
2223 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2224 	    progress_cb, progress_arg);
2225 	if (err)
2226 		return err;
2227 
2228 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2229 	    author, time(NULL), author, time(NULL), logmsg, repo);
2230 	free(new_tree_id);
2231 	return err;
2232 }
2233 
2234 const struct got_error *
2235 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2236     struct got_repository *repo)
2237 {
2238 	const struct got_error *err = NULL;
2239 	char *path_objects = NULL, *path = NULL;
2240 	DIR *dir = NULL;
2241 	struct got_object_id id;
2242 	int i;
2243 
2244 	*nobjects = 0;
2245 	*ondisk_size = 0;
2246 
2247 	path_objects = got_repo_get_path_objects(repo);
2248 	if (path_objects == NULL)
2249 		return got_error_from_errno("got_repo_get_path_objects");
2250 
2251 	for (i = 0; i <= 0xff; i++) {
2252 		struct dirent *dent;
2253 
2254 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2255 			err = got_error_from_errno("asprintf");
2256 			break;
2257 		}
2258 
2259 		dir = opendir(path);
2260 		if (dir == NULL) {
2261 			if (errno == ENOENT) {
2262 				err = NULL;
2263 				continue;
2264 			}
2265 			err = got_error_from_errno2("opendir", path);
2266 			break;
2267 		}
2268 
2269 		while ((dent = readdir(dir)) != NULL) {
2270 			char *id_str;
2271 			int fd;
2272 			struct stat sb;
2273 
2274 			if (strcmp(dent->d_name, ".") == 0 ||
2275 			    strcmp(dent->d_name, "..") == 0)
2276 				continue;
2277 
2278 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2279 				err = got_error_from_errno("asprintf");
2280 				goto done;
2281 			}
2282 
2283 			/* XXX: sha256 */
2284 			if (!got_parse_sha1_digest(id.hash, id_str)) {
2285 				free(id_str);
2286 				continue;
2287 			}
2288 			free(id_str);
2289 
2290 			err = got_object_open_loose_fd(&fd, &id, repo);
2291 			if (err)
2292 				goto done;
2293 
2294 			if (fstat(fd, &sb) == -1) {
2295 				err = got_error_from_errno("fstat");
2296 				close(fd);
2297 				goto done;
2298 			}
2299 			(*nobjects)++;
2300 			(*ondisk_size) += sb.st_size;
2301 
2302 			if (close(fd) == -1) {
2303 				err = got_error_from_errno("close");
2304 				goto done;
2305 			}
2306 		}
2307 
2308 		if (closedir(dir) != 0) {
2309 			err = got_error_from_errno("closedir");
2310 			goto done;
2311 		}
2312 		dir = NULL;
2313 
2314 		free(path);
2315 		path = NULL;
2316 	}
2317 done:
2318 	if (dir && closedir(dir) != 0 && err == NULL)
2319 		err = got_error_from_errno("closedir");
2320 
2321 	if (err) {
2322 		*nobjects = 0;
2323 		*ondisk_size = 0;
2324 	}
2325 	free(path_objects);
2326 	free(path);
2327 	return err;
2328 }
2329 
2330 const struct got_error *
2331 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2332     off_t *total_packsize, struct got_repository *repo)
2333 {
2334 	const struct got_error *err = NULL;
2335 	DIR *packdir = NULL;
2336 	struct dirent *dent;
2337 	struct got_packidx *packidx = NULL;
2338 	char *path_packidx;
2339 	char *path_packfile;
2340 	int packdir_fd;
2341 	struct stat sb;
2342 
2343 	*npackfiles = 0;
2344 	*nobjects = 0;
2345 	*total_packsize = 0;
2346 
2347 	packdir_fd = openat(got_repo_get_fd(repo),
2348 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2349 	if (packdir_fd == -1) {
2350 		return got_error_from_errno_fmt("openat: %s/%s",
2351 		    got_repo_get_path_git_dir(repo),
2352 		    GOT_OBJECTS_PACK_DIR);
2353 	}
2354 
2355 	packdir = fdopendir(packdir_fd);
2356 	if (packdir == NULL) {
2357 		err = got_error_from_errno("fdopendir");
2358 		goto done;
2359 	}
2360 
2361 	while ((dent = readdir(packdir)) != NULL) {
2362 		if (!got_repo_is_packidx_filename(repo, dent->d_name,
2363 		    dent->d_namlen))
2364 			continue;
2365 
2366 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2367 		    dent->d_name) == -1) {
2368 			err = got_error_from_errno("asprintf");
2369 			goto done;
2370 		}
2371 
2372 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2373 		    path_packidx, 0, repo->algo);
2374 		free(path_packidx);
2375 		if (err)
2376 			goto done;
2377 
2378 		if (fstat(packidx->fd, &sb) == -1)
2379 			goto done;
2380 		*total_packsize += sb.st_size;
2381 
2382 		err = got_packidx_get_packfile_path(&path_packfile,
2383 		    packidx->path_packidx);
2384 		if (err)
2385 			goto done;
2386 
2387 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2388 		    0) == -1) {
2389 			free(path_packfile);
2390 			goto done;
2391 		}
2392 		free(path_packfile);
2393 		*total_packsize += sb.st_size;
2394 
2395 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2396 
2397 		(*npackfiles)++;
2398 
2399 		got_packidx_close(packidx);
2400 		packidx = NULL;
2401 	}
2402 done:
2403 	if (packidx)
2404 		got_packidx_close(packidx);
2405 	if (packdir && closedir(packdir) != 0 && err == NULL)
2406 		err = got_error_from_errno("closedir");
2407 	if (err) {
2408 		*npackfiles = 0;
2409 		*nobjects = 0;
2410 		*total_packsize = 0;
2411 	}
2412 	return err;
2413 }
2414 
2415 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2416     got_packidx_bloom_filter_cmp);