1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <sha2.h>
36 #include <string.h>
37 #include <time.h>
38 #include <unistd.h>
39 #include <zlib.h>
40 #include <errno.h>
41 #include <libgen.h>
42 #include <stdint.h>
43 #include <imsg.h>
44 #include <uuid.h>
45 
46 #include "bloom.h"
47 
48 #include "got_error.h"
49 #include "got_reference.h"
50 #include "got_repository.h"
51 #include "got_path.h"
52 #include "got_cancel.h"
53 #include "got_object.h"
54 #include "got_opentemp.h"
55 
56 #include "got_lib_delta.h"
57 #include "got_lib_delta_cache.h"
58 #include "got_lib_inflate.h"
59 #include "got_lib_object.h"
60 #include "got_lib_object_parse.h"
61 #include "got_lib_object_create.h"
62 #include "got_lib_pack.h"
63 #include "got_lib_privsep.h"
64 #include "got_lib_hash.h"
65 #include "got_lib_object_cache.h"
66 #include "got_lib_repository.h"
67 #include "got_lib_gotconfig.h"
68 
69 #ifndef nitems
70 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
71 #endif
72 
73 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
74 
75 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
76     got_packidx_bloom_filter_cmp);
77 
78 static inline int
79 is_boolean_val(const char *val)
80 {
81 	return (strcasecmp(val, "true") == 0 ||
82 	    strcasecmp(val, "false") == 0 ||
83 	    strcasecmp(val, "on") == 0 ||
84     	    strcasecmp(val, "off") == 0 ||
85 	    strcasecmp(val, "yes") == 0 ||
86     	    strcasecmp(val, "no") == 0 ||
87 	    strcasecmp(val, "1") == 0 ||
88 	    strcasecmp(val, "0") == 0);
89 }
90 
91 static inline int
92 get_boolean_val(const char *val)
93 {
94 	return (strcasecmp(val, "true") == 0 ||
95 	    strcasecmp(val, "on") == 0 ||
96 	    strcasecmp(val, "yes") == 0 ||
97 	    strcasecmp(val, "1") == 0);
98 }
99 
100 const char *
101 got_repo_get_path(struct got_repository *repo)
102 {
103 	return repo->path;
104 }
105 
106 const char *
107 got_repo_get_path_git_dir(struct got_repository *repo)
108 {
109 	return repo->path_git_dir;
110 }
111 
112 int
113 got_repo_get_fd(struct got_repository *repo)
114 {
115 	return repo->gitdir_fd;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_author_name(struct got_repository *repo)
120 {
121 	return repo->gitconfig_author_name;
122 }
123 
124 const char *
125 got_repo_get_gitconfig_author_email(struct got_repository *repo)
126 {
127 	return repo->gitconfig_author_email;
128 }
129 
130 const char *
131 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
132 {
133 	return repo->global_gitconfig_author_name;
134 }
135 
136 const char *
137 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
138 {
139 	return repo->global_gitconfig_author_email;
140 }
141 
142 const char *
143 got_repo_get_gitconfig_owner(struct got_repository *repo)
144 {
145 	return repo->gitconfig_owner;
146 }
147 
148 int
149 got_repo_has_extension(struct got_repository *repo, const char *ext)
150 {
151 	int i;
152 
153 	for (i = 0; i < repo->nextensions; ++i) {
154 		if (!strcasecmp(ext, repo->extnames[i]))
155 			return get_boolean_val(repo->extvals[i]);
156 	}
157 
158 	return 0;
159 }
160 
161 int
162 got_repo_is_bare(struct got_repository *repo)
163 {
164 	return (strcmp(repo->path, repo->path_git_dir) == 0);
165 }
166 
167 static char *
168 get_path_git_child(struct got_repository *repo, const char *basename)
169 {
170 	char *path_child;
171 
172 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
173 	    basename) == -1)
174 		return NULL;
175 
176 	return path_child;
177 }
178 
179 char *
180 got_repo_get_path_objects(struct got_repository *repo)
181 {
182 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
183 }
184 
185 char *
186 got_repo_get_path_objects_pack(struct got_repository *repo)
187 {
188 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
189 }
190 
191 char *
192 got_repo_get_path_refs(struct got_repository *repo)
193 {
194 	return get_path_git_child(repo, GOT_REFS_DIR);
195 }
196 
197 char *
198 got_repo_get_path_packed_refs(struct got_repository *repo)
199 {
200 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
201 }
202 
203 static char *
204 get_path_head(struct got_repository *repo)
205 {
206 	return get_path_git_child(repo, GOT_HEAD_FILE);
207 }
208 
209 char *
210 got_repo_get_path_gitconfig(struct got_repository *repo)
211 {
212 	return get_path_git_child(repo, GOT_GITCONFIG);
213 }
214 
215 char *
216 got_repo_get_path_gotconfig(struct got_repository *repo)
217 {
218 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
219 }
220 
221 const struct got_gotconfig *
222 got_repo_get_gotconfig(struct got_repository *repo)
223 {
224 	return repo->gotconfig;
225 }
226 
227 void
228 got_repo_get_gitconfig_remotes(int *nremotes,
229     const struct got_remote_repo **remotes, struct got_repository *repo)
230 {
231 	*nremotes = repo->ngitconfig_remotes;
232 	*remotes = repo->gitconfig_remotes;
233 }
234 
235 static int
236 is_git_repo(struct got_repository *repo)
237 {
238 	const char *path_git = got_repo_get_path_git_dir(repo);
239 	char *path_objects = got_repo_get_path_objects(repo);
240 	char *path_refs = got_repo_get_path_refs(repo);
241 	char *path_head = get_path_head(repo);
242 	int ret = 0;
243 	struct stat sb;
244 	struct got_reference *head_ref;
245 
246 	if (lstat(path_git, &sb) == -1)
247 		goto done;
248 	if (!S_ISDIR(sb.st_mode))
249 		goto done;
250 
251 	if (lstat(path_objects, &sb) == -1)
252 		goto done;
253 	if (!S_ISDIR(sb.st_mode))
254 		goto done;
255 
256 	if (lstat(path_refs, &sb) == -1)
257 		goto done;
258 	if (!S_ISDIR(sb.st_mode))
259 		goto done;
260 
261 	if (lstat(path_head, &sb) == -1)
262 		goto done;
263 	if (!S_ISREG(sb.st_mode))
264 		goto done;
265 
266 	/* Check if the HEAD reference can be opened. */
267 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
268 		goto done;
269 	got_ref_close(head_ref);
270 
271 	ret = 1;
272 done:
273 	free(path_objects);
274 	free(path_refs);
275 	free(path_head);
276 	return ret;
277 
278 }
279 
280 static const struct got_error *
281 close_tempfiles(int *fds, size_t nfds)
282 {
283 	const struct got_error *err = NULL;
284 	int i;
285 
286 	for (i = 0; i < nfds; i++) {
287 		if (fds[i] == -1)
288 			continue;
289 		if (close(fds[i]) == -1) {
290 			err = got_error_from_errno("close");
291 			break;
292 		}
293 	}
294 	free(fds);
295 	return err;
296 }
297 
298 static const struct got_error *
299 open_tempfiles(int **fds, size_t array_size, size_t nfds)
300 {
301 	const struct got_error *err = NULL;
302 	int i;
303 
304 	*fds = calloc(array_size, sizeof(**fds));
305 	if (*fds == NULL)
306 		return got_error_from_errno("calloc");
307 
308 	for (i = 0; i < array_size; i++)
309 		(*fds)[i] = -1;
310 
311 	for (i = 0; i < nfds; i++) {
312 		(*fds)[i] = got_opentempfd();
313 		if ((*fds)[i] == -1) {
314 			err = got_error_from_errno("got_opentempfd");
315 			close_tempfiles(*fds, nfds);
316 			*fds = NULL;
317 			return err;
318 		}
319 	}
320 
321 	return NULL;
322 }
323 
324 static const struct got_error *
325 get_pack_cache_size(int *pack_cache_size)
326 {
327 	struct rlimit rl;
328 
329 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
330 		return got_error_from_errno("getrlimit");
331 
332 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
333 	if (*pack_cache_size > rl.rlim_cur / 8)
334 		*pack_cache_size = rl.rlim_cur / 8;
335 
336 	return NULL;
337 }
338 
339 const struct got_error *
340 got_repo_pack_fds_open(int **pack_fds)
341 {
342 	const struct got_error *err;
343 	int nfds;
344 
345 	err = get_pack_cache_size(&nfds);
346 	if (err)
347 		return err;
348 
349 	/*
350 	 * We need one basefd and one accumfd per cached pack.
351 	 * Our constants should be set up in a way such that
352 	 * this error never triggers.
353 	 */
354 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
355 		return got_error(GOT_ERR_NO_SPACE);
356 
357 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
358 }
359 
360 const struct got_error *
361 got_repo_pack_fds_close(int *pack_fds)
362 {
363 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
364 }
365 
366 const struct got_error *
367 got_repo_temp_fds_open(int **temp_fds)
368 {
369 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
370 	    GOT_REPO_NUM_TEMPFILES);
371 }
372 
373 void
374 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
375 {
376 	int i;
377 
378 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
379 		repo->tempfiles[i] = temp_fds[i];
380 }
381 
382 const struct got_error *
383 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
384 {
385 	int i;
386 
387 	*fd = -1;
388 	*idx = -1;
389 
390 	for (i = 0; i < nitems(repo->tempfiles); i++) {
391 		if (repo->tempfile_use_mask & (1 << i))
392 			continue;
393 		if (repo->tempfiles[i] != -1) {
394 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
395 				return got_error_from_errno("ftruncate");
396 			*fd = repo->tempfiles[i];
397 			*idx = i;
398 			repo->tempfile_use_mask |= (1 << i);
399 			return NULL;
400 		}
401 	}
402 
403 	return got_error(GOT_ERR_REPO_TEMPFILE);
404 }
405 
406 void
407 got_repo_temp_fds_put(int idx, struct got_repository *repo)
408 {
409 	repo->tempfile_use_mask &= ~(1 << idx);
410 }
411 
412 const struct got_error *
413 got_repo_temp_fds_close(int *temp_fds)
414 {
415 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
416 }
417 
418 const struct got_error *
419 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
420     struct got_object *obj)
421 {
422 #ifndef GOT_NO_OBJ_CACHE
423 	const struct got_error *err = NULL;
424 	err = got_object_cache_add(&repo->objcache, id, obj);
425 	if (err) {
426 		if (err->code == GOT_ERR_OBJ_EXISTS ||
427 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
428 			err = NULL;
429 		return err;
430 	}
431 	obj->refcnt++;
432 #endif
433 	return NULL;
434 }
435 
436 struct got_object *
437 got_repo_get_cached_object(struct got_repository *repo,
438     struct got_object_id *id)
439 {
440 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
441 }
442 
443 const struct got_error *
444 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
445     struct got_tree_object *tree)
446 {
447 #ifndef GOT_NO_OBJ_CACHE
448 	const struct got_error *err = NULL;
449 	err = got_object_cache_add(&repo->treecache, id, tree);
450 	if (err) {
451 		if (err->code == GOT_ERR_OBJ_EXISTS ||
452 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
453 			err = NULL;
454 		return err;
455 	}
456 	tree->refcnt++;
457 #endif
458 	return NULL;
459 }
460 
461 struct got_tree_object *
462 got_repo_get_cached_tree(struct got_repository *repo,
463     struct got_object_id *id)
464 {
465 	return (struct got_tree_object *)got_object_cache_get(
466 	    &repo->treecache, id);
467 }
468 
469 const struct got_error *
470 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
471     struct got_commit_object *commit)
472 {
473 #ifndef GOT_NO_OBJ_CACHE
474 	const struct got_error *err = NULL;
475 	err = got_object_cache_add(&repo->commitcache, id, commit);
476 	if (err) {
477 		if (err->code == GOT_ERR_OBJ_EXISTS ||
478 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
479 			err = NULL;
480 		return err;
481 	}
482 	commit->refcnt++;
483 #endif
484 	return NULL;
485 }
486 
487 struct got_commit_object *
488 got_repo_get_cached_commit(struct got_repository *repo,
489     struct got_object_id *id)
490 {
491 	return (struct got_commit_object *)got_object_cache_get(
492 	    &repo->commitcache, id);
493 }
494 
495 const struct got_error *
496 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
497     struct got_tag_object *tag)
498 {
499 #ifndef GOT_NO_OBJ_CACHE
500 	const struct got_error *err = NULL;
501 	err = got_object_cache_add(&repo->tagcache, id, tag);
502 	if (err) {
503 		if (err->code == GOT_ERR_OBJ_EXISTS ||
504 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
505 			err = NULL;
506 		return err;
507 	}
508 	tag->refcnt++;
509 #endif
510 	return NULL;
511 }
512 
513 struct got_tag_object *
514 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
515 {
516 	return (struct got_tag_object *)got_object_cache_get(
517 	    &repo->tagcache, id);
518 }
519 
520 const struct got_error *
521 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
522     struct got_raw_object *raw)
523 {
524 #ifndef GOT_NO_OBJ_CACHE
525 	const struct got_error *err = NULL;
526 	err = got_object_cache_add(&repo->rawcache, id, raw);
527 	if (err) {
528 		if (err->code == GOT_ERR_OBJ_EXISTS ||
529 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
530 			err = NULL;
531 		return err;
532 	}
533 	raw->refcnt++;
534 #endif
535 	return NULL;
536 }
537 
538 
539 struct got_raw_object *
540 got_repo_get_cached_raw_object(struct got_repository *repo,
541     struct got_object_id *id)
542 {
543 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
544 }
545 
546 
547 static const struct got_error *
548 open_repo(struct got_repository *repo, const char *path)
549 {
550 	const struct got_error *err = NULL;
551 
552 	repo->gitdir_fd = -1;
553 
554 	/* bare git repository? */
555 	repo->path_git_dir = strdup(path);
556 	if (repo->path_git_dir == NULL)
557 		return got_error_from_errno("strdup");
558 	if (is_git_repo(repo)) {
559 		repo->path = strdup(repo->path_git_dir);
560 		if (repo->path == NULL) {
561 			err = got_error_from_errno("strdup");
562 			goto done;
563 		}
564 		repo->gitdir_fd = open(repo->path_git_dir,
565 		    O_DIRECTORY | O_CLOEXEC);
566 		if (repo->gitdir_fd == -1) {
567 			err = got_error_from_errno2("open",
568 			    repo->path_git_dir);
569 			goto done;
570 		}
571 		return NULL;
572 	}
573 
574 	/* git repository with working tree? */
575 	free(repo->path_git_dir);
576 	repo->path_git_dir = NULL;
577 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
578 		err = got_error_from_errno("asprintf");
579 		goto done;
580 	}
581 	if (is_git_repo(repo)) {
582 		repo->path = strdup(path);
583 		if (repo->path == NULL) {
584 			err = got_error_from_errno("strdup");
585 			goto done;
586 		}
587 		repo->gitdir_fd = open(repo->path_git_dir,
588 		    O_DIRECTORY | O_CLOEXEC);
589 		if (repo->gitdir_fd == -1) {
590 			err = got_error_from_errno2("open",
591 			    repo->path_git_dir);
592 			goto done;
593 		}
594 		return NULL;
595 	}
596 
597 	err = got_error(GOT_ERR_NOT_GIT_REPO);
598 done:
599 	if (err) {
600 		free(repo->path);
601 		repo->path = NULL;
602 		free(repo->path_git_dir);
603 		repo->path_git_dir = NULL;
604 		if (repo->gitdir_fd != -1)
605 			close(repo->gitdir_fd);
606 		repo->gitdir_fd = -1;
607 
608 	}
609 	return err;
610 }
611 
612 static const struct got_error *
613 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
614 {
615 	const struct got_error *err = NULL;
616 	char *repo_gitconfig_path = NULL;
617 
618 	if (global_gitconfig_path) {
619 		/* Read settings from ~/.gitconfig. */
620 		int dummy_repo_version;
621 		err = got_repo_read_gitconfig(&dummy_repo_version,
622 		    &repo->global_gitconfig_author_name,
623 		    &repo->global_gitconfig_author_email,
624 		    NULL, NULL, NULL, NULL, NULL, NULL,
625 		    global_gitconfig_path);
626 		if (err)
627 			return err;
628 	}
629 
630 	/* Read repository's .git/config file. */
631 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
632 	if (repo_gitconfig_path == NULL)
633 		return got_error_from_errno("got_repo_get_path_gitconfig");
634 
635 	err = got_repo_read_gitconfig(
636 	    &repo->gitconfig_repository_format_version,
637 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
638 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
639 	    &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
640 	    &repo->nextensions, repo_gitconfig_path);
641 	if (err)
642 		goto done;
643 
644 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
645 		int i;
646 
647 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
648 			got_repo_free_remote_repo_data(
649 			    &repo->gitconfig_remotes[i]);
650 		}
651 		free(repo->gitconfig_remotes);
652 		repo->gitconfig_remotes = NULL;
653 		repo->ngitconfig_remotes = 0;
654 
655 		free(repo->gitconfig_author_name);
656 		repo->gitconfig_author_name = NULL;
657 		free(repo->gitconfig_author_email);
658 		repo->gitconfig_author_email = NULL;
659 
660 		free(repo->global_gitconfig_author_name);
661 		repo->global_gitconfig_author_name = NULL;
662 		free(repo->global_gitconfig_author_email);
663 		repo->global_gitconfig_author_email = NULL;
664 	}
665 
666 done:
667 	free(repo_gitconfig_path);
668 	return err;
669 }
670 
671 static const struct got_error *
672 read_gotconfig(struct got_repository *repo)
673 {
674 	const struct got_error *err = NULL;
675 	char *gotconfig_path;
676 
677 	gotconfig_path = got_repo_get_path_gotconfig(repo);
678 	if (gotconfig_path == NULL)
679 		return got_error_from_errno("got_repo_get_path_gotconfig");
680 
681 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
682 	free(gotconfig_path);
683 	return err;
684 }
685 
686 /* Supported repository format extensions. */
687 static const char *const repo_extensions[] = {
688 	"noop",			/* Got supports repository format version 1. */
689 	"preciousObjects",	/* Supported by gotadmin cleanup. */
690 	"worktreeConfig",	/* Got does not care about Git work trees. */
691 };
692 
693 const struct got_error *
694 got_repo_open(struct got_repository **repop, const char *path,
695     const char *global_gitconfig_path, int *pack_fds)
696 {
697 	struct got_repository *repo = NULL;
698 	const struct got_error *err = NULL;
699 	char *repo_path = NULL;
700 	size_t i, j = 0;
701 
702 	*repop = NULL;
703 
704 	repo = calloc(1, sizeof(*repo));
705 	if (repo == NULL)
706 		return got_error_from_errno("calloc");
707 
708 	RB_INIT(&repo->packidx_bloom_filters);
709 	TAILQ_INIT(&repo->packidx_paths);
710 
711 	for (i = 0; i < nitems(repo->privsep_children); i++) {
712 		memset(&repo->privsep_children[i], 0,
713 		    sizeof(repo->privsep_children[0]));
714 		repo->privsep_children[i].imsg_fd = -1;
715 	}
716 
717 	err = got_object_cache_init(&repo->objcache,
718 	    GOT_OBJECT_CACHE_TYPE_OBJ);
719 	if (err)
720 		goto done;
721 	err = got_object_cache_init(&repo->treecache,
722 	    GOT_OBJECT_CACHE_TYPE_TREE);
723 	if (err)
724 		goto done;
725 	err = got_object_cache_init(&repo->commitcache,
726 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
727 	if (err)
728 		goto done;
729 	err = got_object_cache_init(&repo->tagcache,
730 	    GOT_OBJECT_CACHE_TYPE_TAG);
731 	if (err)
732 		goto done;
733 	err = got_object_cache_init(&repo->rawcache,
734 	    GOT_OBJECT_CACHE_TYPE_RAW);
735 	if (err)
736 		goto done;
737 
738 	err = get_pack_cache_size(&repo->pack_cache_size);
739 	if (err)
740 		goto done;
741 	for (i = 0; i < nitems(repo->packs); i++) {
742 		if (pack_fds != NULL && i < repo->pack_cache_size) {
743 			repo->packs[i].basefd = pack_fds[j++];
744 			repo->packs[i].accumfd = pack_fds[j++];
745 		} else {
746 			repo->packs[i].basefd = -1;
747 			repo->packs[i].accumfd = -1;
748 		}
749 	}
750 	for (i = 0; i < nitems(repo->tempfiles); i++)
751 		repo->tempfiles[i] = -1;
752 	repo->pinned_pack = -1;
753 	repo->pinned_packidx = -1;
754 	repo->pinned_pid = 0;
755 
756 	repo_path = realpath(path, NULL);
757 	if (repo_path == NULL) {
758 		err = got_error_from_errno2("realpath", path);
759 		goto done;
760 	}
761 
762 	for (;;) {
763 		char *parent_path;
764 
765 		err = open_repo(repo, repo_path);
766 		if (err == NULL)
767 			break;
768 		if (err->code != GOT_ERR_NOT_GIT_REPO)
769 			goto done;
770 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
771 			err = got_error(GOT_ERR_NOT_GIT_REPO);
772 			goto done;
773 		}
774 		err = got_path_dirname(&parent_path, repo_path);
775 		if (err)
776 			goto done;
777 		free(repo_path);
778 		repo_path = parent_path;
779 	}
780 
781 	err = read_gotconfig(repo);
782 	if (err)
783 		goto done;
784 
785 	err = read_gitconfig(repo, global_gitconfig_path);
786 	if (err)
787 		goto done;
788 	if (repo->gitconfig_repository_format_version != 0) {
789 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
790 		goto done;
791 	}
792 	for (i = 0; i < repo->nextensions; i++) {
793 		char *ext = repo->extnames[i];
794 		char *val = repo->extvals[i];
795 		int j, supported = 0;
796 
797 		if (!is_boolean_val(val)) {
798 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
799 			goto done;
800 		}
801 
802 		if (!get_boolean_val(val))
803 			continue;
804 
805 		for (j = 0; j < nitems(repo_extensions); j++) {
806 			if (strcmp(ext, repo_extensions[j]) == 0) {
807 				supported = 1;
808 				break;
809 			}
810 		}
811 		if (!supported) {
812 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
813 			goto done;
814 		}
815 	}
816 
817 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
818 done:
819 	if (err)
820 		got_repo_close(repo);
821 	else
822 		*repop = repo;
823 	free(repo_path);
824 	return err;
825 }
826 
827 const struct got_error *
828 got_repo_close(struct got_repository *repo)
829 {
830 	const struct got_error *err = NULL, *child_err;
831 	struct got_packidx_bloom_filter *bf;
832 	size_t i;
833 
834 	for (i = 0; i < repo->pack_cache_size; i++) {
835 		if (repo->packidx_cache[i] == NULL)
836 			break;
837 		got_packidx_close(repo->packidx_cache[i]);
838 	}
839 
840 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
841 	    &repo->packidx_bloom_filters))) {
842 		RB_REMOVE(got_packidx_bloom_filter_tree,
843 		    &repo->packidx_bloom_filters, bf);
844 		bloom_free(bf->bloom);
845 		free(bf->bloom);
846 		free(bf);
847 	}
848 
849 	for (i = 0; i < repo->pack_cache_size; i++)
850 		if (repo->packs[i].path_packfile)
851 			if (repo->packs[i].path_packfile)
852 				got_pack_close(&repo->packs[i]);
853 
854 	free(repo->path);
855 	free(repo->path_git_dir);
856 
857 	got_object_cache_close(&repo->objcache);
858 	got_object_cache_close(&repo->treecache);
859 	got_object_cache_close(&repo->commitcache);
860 	got_object_cache_close(&repo->tagcache);
861 	got_object_cache_close(&repo->rawcache);
862 
863 	for (i = 0; i < nitems(repo->privsep_children); i++) {
864 		if (repo->privsep_children[i].imsg_fd == -1)
865 			continue;
866 		imsg_clear(repo->privsep_children[i].ibuf);
867 		free(repo->privsep_children[i].ibuf);
868 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
869 		child_err = got_privsep_wait_for_child(
870 		    repo->privsep_children[i].pid);
871 		if (child_err && err == NULL)
872 			err = child_err;
873 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
874 		    err == NULL)
875 			err = got_error_from_errno("close");
876 	}
877 
878 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
879 	    err == NULL)
880 		err = got_error_from_errno("close");
881 
882 	if (repo->gotconfig)
883 		got_gotconfig_free(repo->gotconfig);
884 	free(repo->gitconfig_author_name);
885 	free(repo->gitconfig_author_email);
886 	for (i = 0; i < repo->ngitconfig_remotes; i++)
887 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
888 	free(repo->gitconfig_remotes);
889 	for (i = 0; i < repo->nextensions; i++) {
890 		free(repo->extnames[i]);
891 		free(repo->extvals[i]);
892 	}
893 	free(repo->extnames);
894 	free(repo->extvals);
895 
896 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
897 	free(repo);
898 
899 	return err;
900 }
901 
902 void
903 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
904 {
905 	int i;
906 
907 	free(repo->name);
908 	repo->name = NULL;
909 	free(repo->fetch_url);
910 	repo->fetch_url = NULL;
911 	free(repo->send_url);
912 	repo->send_url = NULL;
913 	for (i = 0; i < repo->nfetch_branches; i++)
914 		free(repo->fetch_branches[i]);
915 	free(repo->fetch_branches);
916 	repo->fetch_branches = NULL;
917 	repo->nfetch_branches = 0;
918 	for (i = 0; i < repo->nsend_branches; i++)
919 		free(repo->send_branches[i]);
920 	free(repo->send_branches);
921 	repo->send_branches = NULL;
922 	repo->nsend_branches = 0;
923 }
924 
925 const struct got_error *
926 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
927     const char *input_path)
928 {
929 	const struct got_error *err = NULL;
930 	const char *repo_abspath = NULL;
931 	size_t repolen, len;
932 	char *canonpath, *path = NULL;
933 
934 	*in_repo_path = NULL;
935 
936 	canonpath = strdup(input_path);
937 	if (canonpath == NULL) {
938 		err = got_error_from_errno("strdup");
939 		goto done;
940 	}
941 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
942 	if (err)
943 		goto done;
944 
945 	repo_abspath = got_repo_get_path(repo);
946 
947 	if (canonpath[0] == '\0') {
948 		path = strdup(canonpath);
949 		if (path == NULL) {
950 			err = got_error_from_errno("strdup");
951 			goto done;
952 		}
953 	} else {
954 		path = realpath(canonpath, NULL);
955 		if (path == NULL) {
956 			if (errno != ENOENT) {
957 				err = got_error_from_errno2("realpath",
958 				    canonpath);
959 				goto done;
960 			}
961 			/*
962 			 * Path is not on disk.
963 			 * Assume it is already relative to repository root.
964 			 */
965 			path = strdup(canonpath);
966 			if (path == NULL) {
967 				err = got_error_from_errno("strdup");
968 				goto done;
969 			}
970 		}
971 
972 		repolen = strlen(repo_abspath);
973 		len = strlen(path);
974 
975 
976 		if (strcmp(path, repo_abspath) == 0) {
977 			free(path);
978 			path = strdup("");
979 			if (path == NULL) {
980 				err = got_error_from_errno("strdup");
981 				goto done;
982 			}
983 		} else if (len > repolen &&
984 		    got_path_is_child(path, repo_abspath, repolen)) {
985 			/* Matched an on-disk path inside repository. */
986 			if (got_repo_is_bare(repo)) {
987 				/*
988 				 * Matched an on-disk path inside repository
989 				 * database. Treat input as repository-relative.
990 				 */
991 				free(path);
992 				path = canonpath;
993 				canonpath = NULL;
994 			} else {
995 				char *child;
996 				/* Strip common prefix with repository path. */
997 				err = got_path_skip_common_ancestor(&child,
998 				    repo_abspath, path);
999 				if (err)
1000 					goto done;
1001 				free(path);
1002 				path = child;
1003 			}
1004 		} else {
1005 			/*
1006 			 * Matched unrelated on-disk path.
1007 			 * Treat input as repository-relative.
1008 			 */
1009 			free(path);
1010 			path = canonpath;
1011 			canonpath = NULL;
1012 		}
1013 	}
1014 
1015 	/* Make in-repository path absolute */
1016 	if (path[0] != '/') {
1017 		char *abspath;
1018 		if (asprintf(&abspath, "/%s", path) == -1) {
1019 			err = got_error_from_errno("asprintf");
1020 			goto done;
1021 		}
1022 		free(path);
1023 		path = abspath;
1024 	}
1025 
1026 done:
1027 	free(canonpath);
1028 	if (err)
1029 		free(path);
1030 	else
1031 		*in_repo_path = path;
1032 	return err;
1033 }
1034 
1035 static const struct got_error *
1036 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1037     const char *path_packidx)
1038 {
1039 	const struct got_error *err = NULL;
1040 	size_t i;
1041 
1042 	for (i = 0; i < repo->pack_cache_size; i++) {
1043 		if (repo->packidx_cache[i] == NULL)
1044 			break;
1045 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1046 		    path_packidx) == 0) {
1047 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1048 		}
1049 	}
1050 	if (i == repo->pack_cache_size) {
1051 		do {
1052 			i--;
1053 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1054 		    i == repo->pinned_packidx);
1055 		err = got_packidx_close(repo->packidx_cache[i]);
1056 		if (err)
1057 			return err;
1058 	}
1059 
1060 	repo->packidx_cache[i] = packidx;
1061 
1062 	return NULL;
1063 }
1064 
1065 int
1066 got_repo_is_packidx_filename(const char *name, size_t len)
1067 {
1068 	if (len != GOT_PACKIDX_NAMELEN)
1069 		return 0;
1070 
1071 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1072 		return 0;
1073 
1074 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1075 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1076 		return 0;
1077 
1078 	return 1;
1079 }
1080 
1081 static struct got_packidx_bloom_filter *
1082 get_packidx_bloom_filter(struct got_repository *repo,
1083     const char *path, size_t path_len)
1084 {
1085 	struct got_packidx_bloom_filter key;
1086 
1087 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1088 		return NULL; /* XXX */
1089 	key.path_len = path_len;
1090 
1091 	return RB_FIND(got_packidx_bloom_filter_tree,
1092 	    &repo->packidx_bloom_filters, &key);
1093 }
1094 
1095 int
1096 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1097     const char *path_packidx, struct got_object_id *id)
1098 {
1099 	struct got_packidx_bloom_filter *bf;
1100 
1101 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1102 	if (bf)
1103 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1104 
1105 	/* No bloom filter means this pack index must be searched. */
1106 	return 1;
1107 }
1108 
1109 static const struct got_error *
1110 add_packidx_bloom_filter(struct got_repository *repo,
1111     struct got_packidx *packidx, const char *path_packidx)
1112 {
1113 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1114 	struct got_packidx_bloom_filter *bf;
1115 	size_t len;
1116 
1117 	/*
1118 	 * Don't use bloom filters for very large pack index files.
1119 	 * Large pack files will contain a relatively large fraction
1120 	 * of our objects so we will likely need to visit them anyway.
1121 	 * The more objects a pack file contains the higher the probability
1122 	 * of a false-positive match from the bloom filter. And reading
1123 	 * all object IDs from a large pack index file can be expensive.
1124 	 */
1125 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1126 		return NULL;
1127 
1128 	/* Do we already have a filter for this pack index? */
1129 	if (get_packidx_bloom_filter(repo, path_packidx,
1130 	    strlen(path_packidx)) != NULL)
1131 		return NULL;
1132 
1133 	bf = calloc(1, sizeof(*bf));
1134 	if (bf == NULL)
1135 		return got_error_from_errno("calloc");
1136 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1137 	if (bf->bloom == NULL) {
1138 		free(bf);
1139 		return got_error_from_errno("calloc");
1140 	}
1141 
1142 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1143 	if (len >= sizeof(bf->path)) {
1144 		free(bf->bloom);
1145 		free(bf);
1146 		return got_error(GOT_ERR_NO_SPACE);
1147 	}
1148 	bf->path_len = len;
1149 
1150 	/* Minimum size supported by our bloom filter is 1000 entries. */
1151 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1152 	for (i = 0; i < nobjects; i++) {
1153 		struct got_packidx_object_id *id;
1154 		id = &packidx->hdr.sorted_ids[i];
1155 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1156 	}
1157 
1158 	RB_INSERT(got_packidx_bloom_filter_tree,
1159 	    &repo->packidx_bloom_filters, bf);
1160 	return NULL;
1161 }
1162 
1163 static void
1164 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1165 {
1166 	struct got_pathlist_entry *pe;
1167 
1168 	while (!TAILQ_EMPTY(packidx_paths)) {
1169 		pe = TAILQ_FIRST(packidx_paths);
1170 		TAILQ_REMOVE(packidx_paths, pe, entry);
1171 		free((char *)pe->path);
1172 		free(pe);
1173 	}
1174 }
1175 
1176 static const struct got_error *
1177 refresh_packidx_paths(struct got_repository *repo)
1178 {
1179 	const struct got_error *err = NULL;
1180 	char *objects_pack_dir = NULL;
1181 	struct stat sb;
1182 
1183 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1184 	if (objects_pack_dir == NULL)
1185 		return got_error_from_errno("got_repo_get_path_objects_pack");
1186 
1187 	if (stat(objects_pack_dir, &sb) == -1) {
1188 		if (errno != ENOENT) {
1189 			err = got_error_from_errno2("stat", objects_pack_dir);
1190 			goto done;
1191 		}
1192 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1193 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1194 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1195 		purge_packidx_paths(&repo->packidx_paths);
1196 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1197 		if (err)
1198 			goto done;
1199 	}
1200 done:
1201 	free(objects_pack_dir);
1202 	return err;
1203 }
1204 
1205 const struct got_error *
1206 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1207     struct got_repository *repo, struct got_object_id *id)
1208 {
1209 	const struct got_error *err;
1210 	struct got_pathlist_entry *pe;
1211 	size_t i;
1212 
1213 	/* Search pack index cache. */
1214 	for (i = 0; i < repo->pack_cache_size; i++) {
1215 		if (repo->packidx_cache[i] == NULL)
1216 			break;
1217 		if (!got_repo_check_packidx_bloom_filter(repo,
1218 		    repo->packidx_cache[i]->path_packidx, id))
1219 			continue; /* object will not be found in this index */
1220 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1221 		if (*idx != -1) {
1222 			*packidx = repo->packidx_cache[i];
1223 			/*
1224 			 * Move this cache entry to the front. Repeatedly
1225 			 * searching a wrong pack index can be expensive.
1226 			 */
1227 			if (i > 0) {
1228 				memmove(&repo->packidx_cache[1],
1229 				    &repo->packidx_cache[0],
1230 				    i * sizeof(repo->packidx_cache[0]));
1231 				repo->packidx_cache[0] = *packidx;
1232 				if (repo->pinned_packidx >= 0 &&
1233 				    repo->pinned_packidx < i)
1234 					repo->pinned_packidx++;
1235 				else if (repo->pinned_packidx == i)
1236 					repo->pinned_packidx = 0;
1237 			}
1238 			return NULL;
1239 		}
1240 	}
1241 	/* No luck. Search the filesystem. */
1242 
1243 	err = refresh_packidx_paths(repo);
1244 	if (err)
1245 		return err;
1246 
1247 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1248 		const char *path_packidx = pe->path;
1249 		int is_cached = 0;
1250 
1251 		if (!got_repo_check_packidx_bloom_filter(repo,
1252 		    pe->path, id))
1253 			continue; /* object will not be found in this index */
1254 
1255 		for (i = 0; i < repo->pack_cache_size; i++) {
1256 			if (repo->packidx_cache[i] == NULL)
1257 				break;
1258 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1259 			    path_packidx) == 0) {
1260 				is_cached = 1;
1261 				break;
1262 			}
1263 		}
1264 		if (is_cached)
1265 			continue; /* already searched */
1266 
1267 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1268 		    path_packidx, 0);
1269 		if (err)
1270 			goto done;
1271 
1272 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1273 		if (err)
1274 			goto done;
1275 
1276 		err = cache_packidx(repo, *packidx, path_packidx);
1277 		if (err)
1278 			goto done;
1279 
1280 		*idx = got_packidx_get_object_idx(*packidx, id);
1281 		if (*idx != -1) {
1282 			err = NULL; /* found the object */
1283 			goto done;
1284 		}
1285 	}
1286 
1287 	err = got_error_no_obj(id);
1288 done:
1289 	return err;
1290 }
1291 
1292 const struct got_error *
1293 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1294     struct got_repository *repo)
1295 {
1296 	const struct got_error *err = NULL;
1297 	DIR *packdir = NULL;
1298 	struct dirent *dent;
1299 	char *path_packidx = NULL;
1300 	int packdir_fd;
1301 	struct stat sb;
1302 
1303 	packdir_fd = openat(got_repo_get_fd(repo),
1304 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1305 	if (packdir_fd == -1) {
1306 		return got_error_from_errno_fmt("openat: %s/%s",
1307 		    got_repo_get_path_git_dir(repo),
1308 		    GOT_OBJECTS_PACK_DIR);
1309 	}
1310 
1311 	packdir = fdopendir(packdir_fd);
1312 	if (packdir == NULL) {
1313 		err = got_error_from_errno("fdopendir");
1314 		goto done;
1315 	}
1316 
1317 	if (fstat(packdir_fd, &sb) == -1) {
1318 		err = got_error_from_errno("fstat");
1319 		goto done;
1320 	}
1321 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1322 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1323 
1324 	while ((dent = readdir(packdir)) != NULL) {
1325 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1326 			continue;
1327 
1328 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1329 		    dent->d_name) == -1) {
1330 			err = got_error_from_errno("asprintf");
1331 			path_packidx = NULL;
1332 			break;
1333 		}
1334 
1335 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1336 		if (err)
1337 			break;
1338 	}
1339 done:
1340 	if (err)
1341 		free(path_packidx);
1342 	if (packdir && closedir(packdir) != 0 && err == NULL)
1343 		err = got_error_from_errno("closedir");
1344 	return err;
1345 }
1346 
1347 const struct got_error *
1348 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1349     struct got_repository *repo)
1350 {
1351 	const struct got_error *err;
1352 	size_t i;
1353 
1354 	*packidx = NULL;
1355 
1356 	/* Search pack index cache. */
1357 	for (i = 0; i < repo->pack_cache_size; i++) {
1358 		if (repo->packidx_cache[i] == NULL)
1359 			break;
1360 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1361 		    path_packidx) == 0) {
1362 			*packidx = repo->packidx_cache[i];
1363 			return NULL;
1364 		}
1365 	}
1366 	/* No luck. Search the filesystem. */
1367 
1368 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1369 	    path_packidx, 0);
1370 	if (err)
1371 		return err;
1372 
1373 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1374 	if (err)
1375 		goto done;
1376 
1377 	err = cache_packidx(repo, *packidx, path_packidx);
1378 done:
1379 	if (err) {
1380 		got_packidx_close(*packidx);
1381 		*packidx = NULL;
1382 	}
1383 	return err;
1384 }
1385 
1386 static const struct got_error *
1387 read_packfile_hdr(int fd, struct got_packidx *packidx)
1388 {
1389 	const struct got_error *err = NULL;
1390 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1391 	struct got_packfile_hdr hdr;
1392 	ssize_t n;
1393 
1394 	n = read(fd, &hdr, sizeof(hdr));
1395 	if (n < 0)
1396 		return got_error_from_errno("read");
1397 	if (n != sizeof(hdr))
1398 		return got_error(GOT_ERR_BAD_PACKFILE);
1399 
1400 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1401 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1402 	    be32toh(hdr.nobjects) != totobj)
1403 		err = got_error(GOT_ERR_BAD_PACKFILE);
1404 
1405 	return err;
1406 }
1407 
1408 static const struct got_error *
1409 open_packfile(int *fd, struct got_repository *repo,
1410     const char *relpath, struct got_packidx *packidx)
1411 {
1412 	const struct got_error *err = NULL;
1413 
1414 	*fd = openat(got_repo_get_fd(repo), relpath,
1415 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1416 	if (*fd == -1)
1417 		return got_error_from_errno_fmt("openat: %s/%s",
1418 		    got_repo_get_path_git_dir(repo), relpath);
1419 
1420 	if (packidx) {
1421 		err = read_packfile_hdr(*fd, packidx);
1422 		if (err) {
1423 			close(*fd);
1424 			*fd = -1;
1425 		}
1426 	}
1427 
1428 	return err;
1429 }
1430 
1431 const struct got_error *
1432 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1433     const char *path_packfile, struct got_packidx *packidx)
1434 {
1435 	const struct got_error *err = NULL;
1436 	struct got_pack *pack = NULL;
1437 	struct stat sb;
1438 	size_t i;
1439 
1440 	if (packp)
1441 		*packp = NULL;
1442 
1443 	for (i = 0; i < repo->pack_cache_size; i++) {
1444 		pack = &repo->packs[i];
1445 		if (pack->path_packfile == NULL)
1446 			break;
1447 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1448 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1449 	}
1450 
1451 	if (i == repo->pack_cache_size) {
1452 		struct got_pack tmp;
1453 		do {
1454 			i--;
1455 		} while (i > 0 && repo->pinned_pack >= 0 &&
1456 		    i == repo->pinned_pack);
1457 		err = got_pack_close(&repo->packs[i]);
1458 		if (err)
1459 			return err;
1460 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1461 			return got_error_from_errno("ftruncate");
1462 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1463 			return got_error_from_errno("ftruncate");
1464 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1465 		memcpy(&repo->packs[i], &repo->packs[0],
1466 		    sizeof(repo->packs[i]));
1467 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1468 		if (repo->pinned_pack == 0)
1469 			repo->pinned_pack = i;
1470 		else if (repo->pinned_pack == i)
1471 			repo->pinned_pack = 0;
1472 		i = 0;
1473 	}
1474 
1475 	pack = &repo->packs[i];
1476 
1477 	pack->path_packfile = strdup(path_packfile);
1478 	if (pack->path_packfile == NULL) {
1479 		err = got_error_from_errno("strdup");
1480 		goto done;
1481 	}
1482 
1483 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1484 	if (err)
1485 		goto done;
1486 
1487 	if (fstat(pack->fd, &sb) != 0) {
1488 		err = got_error_from_errno("fstat");
1489 		goto done;
1490 	}
1491 	pack->filesize = sb.st_size;
1492 
1493 	pack->privsep_child = NULL;
1494 
1495 	err = got_delta_cache_alloc(&pack->delta_cache);
1496 	if (err)
1497 		goto done;
1498 
1499 #ifndef GOT_PACK_NO_MMAP
1500 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1501 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1502 		    pack->fd, 0);
1503 		if (pack->map == MAP_FAILED) {
1504 			if (errno != ENOMEM) {
1505 				err = got_error_from_errno("mmap");
1506 				goto done;
1507 			}
1508 			pack->map = NULL; /* fall back to read(2) */
1509 		}
1510 	}
1511 #endif
1512 done:
1513 	if (err) {
1514 		if (pack)
1515 			got_pack_close(pack);
1516 	} else if (packp)
1517 		*packp = pack;
1518 	return err;
1519 }
1520 
1521 struct got_pack *
1522 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1523 {
1524 	struct got_pack *pack = NULL;
1525 	size_t i;
1526 
1527 	for (i = 0; i < repo->pack_cache_size; i++) {
1528 		pack = &repo->packs[i];
1529 		if (pack->path_packfile == NULL)
1530 			break;
1531 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1532 			return pack;
1533 	}
1534 
1535 	return NULL;
1536 }
1537 
1538 const struct got_error *
1539 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1540     struct got_pack *pack)
1541 {
1542 	size_t i;
1543 	int pinned_pack = -1, pinned_packidx = -1;
1544 
1545 	for (i = 0; i < repo->pack_cache_size; i++) {
1546 		if (repo->packidx_cache[i] &&
1547 		    strcmp(repo->packidx_cache[i]->path_packidx,
1548 		    packidx->path_packidx) == 0)
1549 			pinned_packidx = i;
1550 		if (repo->packs[i].path_packfile &&
1551 		    strcmp(repo->packs[i].path_packfile,
1552 		    pack->path_packfile) == 0)
1553 			pinned_pack = i;
1554 	}
1555 
1556 	if (pinned_packidx == -1 || pinned_pack == -1)
1557 		return got_error(GOT_ERR_PIN_PACK);
1558 
1559 	repo->pinned_pack = pinned_pack;
1560 	repo->pinned_packidx = pinned_packidx;
1561 	if (repo->packs[pinned_pack].privsep_child)
1562 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1563 	return NULL;
1564 }
1565 
1566 struct got_pack *
1567 got_repo_get_pinned_pack(struct got_repository *repo)
1568 {
1569 	if (repo->pinned_pack >= 0 &&
1570 	    repo->pinned_pack < repo->pack_cache_size)
1571 		return &repo->packs[repo->pinned_pack];
1572 
1573 	return NULL;
1574 }
1575 
1576 void
1577 got_repo_unpin_pack(struct got_repository *repo)
1578 {
1579 	repo->pinned_packidx = -1;
1580 	repo->pinned_pack = -1;
1581 	repo->pinned_pid = 0;
1582 }
1583 
1584 const struct got_error *
1585 got_repo_init(const char *repo_path, const char *head_name)
1586 {
1587 	const struct got_error *err = NULL;
1588 	const char *dirnames[] = {
1589 		GOT_OBJECTS_DIR,
1590 		GOT_OBJECTS_PACK_DIR,
1591 		GOT_REFS_DIR,
1592 	};
1593 	const char *description_str = "Unnamed repository; "
1594 	    "edit this file 'description' to name the repository.";
1595 	const char *headref = "ref: refs/heads/";
1596 	const char *gitconfig_str = "[core]\n"
1597 	    "\trepositoryformatversion = 0\n"
1598 	    "\tfilemode = true\n"
1599 	    "\tbare = true\n";
1600 	char *headref_str, *path;
1601 	size_t i;
1602 
1603 	if (!got_path_dir_is_empty(repo_path))
1604 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1605 
1606 	for (i = 0; i < nitems(dirnames); i++) {
1607 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1608 			return got_error_from_errno("asprintf");
1609 		}
1610 		err = got_path_mkdir(path);
1611 		free(path);
1612 		if (err)
1613 			return err;
1614 	}
1615 
1616 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1617 		return got_error_from_errno("asprintf");
1618 	err = got_path_create_file(path, description_str);
1619 	free(path);
1620 	if (err)
1621 		return err;
1622 
1623 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1624 		return got_error_from_errno("asprintf");
1625 	if (asprintf(&headref_str, "%s%s", headref,
1626 	    head_name ? head_name : "main") == -1) {
1627 		free(path);
1628 		return got_error_from_errno("asprintf");
1629 	}
1630 	err = got_path_create_file(path, headref_str);
1631 	free(headref_str);
1632 	free(path);
1633 	if (err)
1634 		return err;
1635 
1636 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1637 		return got_error_from_errno("asprintf");
1638 	err = got_path_create_file(path, gitconfig_str);
1639 	free(path);
1640 	if (err)
1641 		return err;
1642 
1643 	return NULL;
1644 }
1645 
1646 static const struct got_error *
1647 match_packed_object(struct got_object_id **unique_id,
1648     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1649 {
1650 	const struct got_error *err = NULL;
1651 	struct got_object_id_queue matched_ids;
1652 	struct got_pathlist_entry *pe;
1653 
1654 	STAILQ_INIT(&matched_ids);
1655 
1656 	err = refresh_packidx_paths(repo);
1657 	if (err)
1658 		return err;
1659 
1660 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1661 		const char *path_packidx = pe->path;
1662 		struct got_packidx *packidx;
1663 		struct got_object_qid *qid;
1664 
1665 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1666 		    path_packidx, 0);
1667 		if (err)
1668 			break;
1669 
1670 		err = got_packidx_match_id_str_prefix(&matched_ids,
1671 		    packidx, id_str_prefix);
1672 		if (err) {
1673 			got_packidx_close(packidx);
1674 			break;
1675 		}
1676 		err = got_packidx_close(packidx);
1677 		if (err)
1678 			break;
1679 
1680 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1681 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1682 				int matched_type;
1683 				err = got_object_get_type(&matched_type, repo,
1684 				    &qid->id);
1685 				if (err)
1686 					goto done;
1687 				if (matched_type != obj_type)
1688 					continue;
1689 			}
1690 			if (*unique_id == NULL) {
1691 				*unique_id = got_object_id_dup(&qid->id);
1692 				if (*unique_id == NULL) {
1693 					err = got_error_from_errno("malloc");
1694 					goto done;
1695 				}
1696 			} else {
1697 				if (got_object_id_cmp(*unique_id,
1698 				    &qid->id) == 0)
1699 					continue; /* packed multiple times */
1700 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1701 				goto done;
1702 			}
1703 		}
1704 	}
1705 done:
1706 	got_object_id_queue_free(&matched_ids);
1707 	if (err) {
1708 		free(*unique_id);
1709 		*unique_id = NULL;
1710 	}
1711 	return err;
1712 }
1713 
1714 static const struct got_error *
1715 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1716     const char *object_dir, const char *id_str_prefix, int obj_type,
1717     struct got_repository *repo)
1718 {
1719 	const struct got_error *err = NULL;
1720 	char *path, *id_str = NULL;
1721 	DIR *dir = NULL;
1722 	struct dirent *dent;
1723 	struct got_object_id id;
1724 
1725 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1726 		err = got_error_from_errno("asprintf");
1727 		goto done;
1728 	}
1729 
1730 	dir = opendir(path);
1731 	if (dir == NULL) {
1732 		if (errno == ENOENT) {
1733 			err = NULL;
1734 			goto done;
1735 		}
1736 		err = got_error_from_errno2("opendir", path);
1737 		goto done;
1738 	}
1739 	while ((dent = readdir(dir)) != NULL) {
1740 		int cmp;
1741 		enum got_hash_algorithm algo = GOT_HASH_SHA1;
1742 
1743 		free(id_str);
1744 		id_str = NULL;
1745 
1746 		if (strcmp(dent->d_name, ".") == 0 ||
1747 		    strcmp(dent->d_name, "..") == 0)
1748 			continue;
1749 
1750 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1751 			err = got_error_from_errno("asprintf");
1752 			goto done;
1753 		}
1754 
1755 		if (!got_parse_object_id(&id, id_str, algo))
1756 			continue;
1757 
1758 		/*
1759 		 * Directory entries do not necessarily appear in
1760 		 * sorted order, so we must iterate over all of them.
1761 		 */
1762 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1763 		if (cmp != 0)
1764 			continue;
1765 
1766 		if (*unique_id == NULL) {
1767 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1768 				int matched_type;
1769 				err = got_object_get_type(&matched_type, repo,
1770 				    &id);
1771 				if (err)
1772 					goto done;
1773 				if (matched_type != obj_type)
1774 					continue;
1775 			}
1776 			*unique_id = got_object_id_dup(&id);
1777 			if (*unique_id == NULL) {
1778 				err = got_error_from_errno("got_object_id_dup");
1779 				goto done;
1780 			}
1781 		} else {
1782 			if (got_object_id_cmp(*unique_id, &id) == 0)
1783 				continue; /* both packed and loose */
1784 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1785 			goto done;
1786 		}
1787 	}
1788 done:
1789 	if (dir && closedir(dir) != 0 && err == NULL)
1790 		err = got_error_from_errno("closedir");
1791 	if (err) {
1792 		free(*unique_id);
1793 		*unique_id = NULL;
1794 	}
1795 	free(id_str);
1796 	free(path);
1797 	return err;
1798 }
1799 
1800 const struct got_error *
1801 got_repo_match_object_id_prefix(struct got_object_id **id,
1802     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1803 {
1804 	const struct got_error *err = NULL;
1805 	char *path_objects = NULL, *object_dir = NULL;
1806 	size_t len;
1807 	int i;
1808 
1809 	*id = NULL;
1810 
1811 	path_objects = got_repo_get_path_objects(repo);
1812 
1813 	len = strlen(id_str_prefix);
1814 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1815 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1816 		goto done;
1817 	}
1818 
1819 	for (i = 0; i < len; i++) {
1820 		if (isxdigit((unsigned char)id_str_prefix[i]))
1821 			continue;
1822 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1823 		goto done;
1824 	}
1825 
1826 	if (len >= 2) {
1827 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1828 		if (err)
1829 			goto done;
1830 		object_dir = strndup(id_str_prefix, 2);
1831 		if (object_dir == NULL) {
1832 			err = got_error_from_errno("strdup");
1833 			goto done;
1834 		}
1835 		err = match_loose_object(id, path_objects, object_dir,
1836 		    id_str_prefix, obj_type, repo);
1837 	} else if (len == 1) {
1838 		int i;
1839 		for (i = 0; i < 0xf; i++) {
1840 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1841 			    == -1) {
1842 				err = got_error_from_errno("asprintf");
1843 				goto done;
1844 			}
1845 			err = match_packed_object(id, repo, object_dir,
1846 			    obj_type);
1847 			if (err)
1848 				goto done;
1849 			err = match_loose_object(id, path_objects, object_dir,
1850 			    id_str_prefix, obj_type, repo);
1851 			if (err)
1852 				goto done;
1853 		}
1854 	} else {
1855 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1856 		goto done;
1857 	}
1858 done:
1859 	free(path_objects);
1860 	free(object_dir);
1861 	if (err) {
1862 		free(*id);
1863 		*id = NULL;
1864 	} else if (*id == NULL) {
1865 		switch (obj_type) {
1866 		case GOT_OBJ_TYPE_BLOB:
1867 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1868 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1869 			break;
1870 		case GOT_OBJ_TYPE_TREE:
1871 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1872 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1873 			break;
1874 		case GOT_OBJ_TYPE_COMMIT:
1875 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1876 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1877 			break;
1878 		case GOT_OBJ_TYPE_TAG:
1879 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1880 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1881 			break;
1882 		default:
1883 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1884 			break;
1885 		}
1886 	}
1887 
1888 	return err;
1889 }
1890 
1891 const struct got_error *
1892 got_repo_match_object_id(struct got_object_id **id, char **label,
1893     const char *id_str, int obj_type, struct got_reflist_head *refs,
1894     struct got_repository *repo)
1895 {
1896 	const struct got_error *err;
1897 	struct got_tag_object *tag;
1898 	struct got_reference *ref = NULL;
1899 
1900 	*id = NULL;
1901 	if (label)
1902 		*label = NULL;
1903 
1904 	if (refs) {
1905 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1906 		    refs, repo);
1907 		if (err == NULL) {
1908 			*id = got_object_id_dup(
1909 			    got_object_tag_get_object_id(tag));
1910 			if (*id == NULL)
1911 				err = got_error_from_errno("got_object_id_dup");
1912 			else if (label && asprintf(label, "refs/tags/%s",
1913 			    got_object_tag_get_name(tag)) == -1) {
1914 				err = got_error_from_errno("asprintf");
1915 				free(*id);
1916 				*id = NULL;
1917 			}
1918 			got_object_tag_close(tag);
1919 			return err;
1920 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1921 		    err->code != GOT_ERR_NO_OBJ)
1922 			return err;
1923 	}
1924 
1925 	err = got_ref_open(&ref, repo, id_str, 0);
1926 	if (err == NULL) {
1927 		err = got_ref_resolve(id, repo, ref);
1928 		if (err)
1929 			goto done;
1930 		if (label) {
1931 			*label = strdup(got_ref_get_name(ref));
1932 			if (*label == NULL) {
1933 				err = got_error_from_errno("strdup");
1934 				goto done;
1935 			}
1936 		}
1937 	} else {
1938 		if (err->code != GOT_ERR_NOT_REF &&
1939 		    err->code != GOT_ERR_BAD_REF_NAME)
1940 			goto done;
1941 		err = got_repo_match_object_id_prefix(id, id_str,
1942 		    obj_type, repo);
1943 		if (err) {
1944 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1945 				err = got_error_not_ref(id_str);
1946 			goto done;
1947 		}
1948 		if (label) {
1949 			err = got_object_id_str(label, *id);
1950 			if (*label == NULL) {
1951 				err = got_error_from_errno("strdup");
1952 				goto done;
1953 			}
1954 		}
1955 	}
1956 done:
1957 	if (ref)
1958 		got_ref_close(ref);
1959 	return err;
1960 }
1961 
1962 const struct got_error *
1963 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1964     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1965 {
1966 	const struct got_error *err = NULL;
1967 	struct got_reflist_entry *re;
1968 	struct got_object_id *tag_id;
1969 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1970 
1971 	*tag = NULL;
1972 
1973 	TAILQ_FOREACH(re, refs, entry) {
1974 		const char *refname;
1975 		refname = got_ref_get_name(re->ref);
1976 		if (got_ref_is_symbolic(re->ref))
1977 			continue;
1978 		if (strncmp(refname, "refs/tags/", 10) != 0)
1979 			continue;
1980 		if (!name_is_absolute)
1981 			refname += strlen("refs/tags/");
1982 		if (strcmp(refname, name) != 0)
1983 			continue;
1984 		err = got_ref_resolve(&tag_id, repo, re->ref);
1985 		if (err)
1986 			break;
1987 		err = got_object_open_as_tag(tag, repo, tag_id);
1988 		free(tag_id);
1989 		if (err)
1990 			break;
1991 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1992 		    got_object_tag_get_object_type(*tag) == obj_type)
1993 			break;
1994 		got_object_tag_close(*tag);
1995 		*tag = NULL;
1996 	}
1997 
1998 	if (err == NULL && *tag == NULL)
1999 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2000 		    GOT_OBJ_LABEL_TAG, name);
2001 	return err;
2002 }
2003 
2004 static const struct got_error *
2005 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2006     const char *name, mode_t mode, struct got_object_id *blob_id)
2007 {
2008 	const struct got_error *err = NULL;
2009 
2010 	 *new_te = NULL;
2011 
2012 	*new_te = calloc(1, sizeof(**new_te));
2013 	if (*new_te == NULL)
2014 		return got_error_from_errno("calloc");
2015 
2016 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2017 	    sizeof((*new_te)->name)) {
2018 		err = got_error(GOT_ERR_NO_SPACE);
2019 		goto done;
2020 	}
2021 
2022 	if (S_ISLNK(mode)) {
2023 		(*new_te)->mode = S_IFLNK;
2024 	} else {
2025 		(*new_te)->mode = S_IFREG;
2026 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2027 	}
2028 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2029 done:
2030 	if (err && *new_te) {
2031 		free(*new_te);
2032 		*new_te = NULL;
2033 	}
2034 	return err;
2035 }
2036 
2037 static const struct got_error *
2038 import_file(struct got_tree_entry **new_te, struct dirent *de,
2039     const char *path, struct got_repository *repo)
2040 {
2041 	const struct got_error *err;
2042 	struct got_object_id *blob_id = NULL;
2043 	char *filepath;
2044 	struct stat sb;
2045 
2046 	if (asprintf(&filepath, "%s%s%s", path,
2047 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2048 		return got_error_from_errno("asprintf");
2049 
2050 	if (lstat(filepath, &sb) != 0) {
2051 		err = got_error_from_errno2("lstat", path);
2052 		goto done;
2053 	}
2054 
2055 	err = got_object_blob_create(&blob_id, filepath, repo);
2056 	if (err)
2057 		goto done;
2058 
2059 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2060 	    blob_id);
2061 done:
2062 	free(filepath);
2063 	if (err)
2064 		free(blob_id);
2065 	return err;
2066 }
2067 
2068 static const struct got_error *
2069 insert_tree_entry(struct got_tree_entry *new_te,
2070     struct got_pathlist_head *paths)
2071 {
2072 	const struct got_error *err = NULL;
2073 	struct got_pathlist_entry *new_pe;
2074 
2075 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2076 	if (err)
2077 		return err;
2078 	if (new_pe == NULL)
2079 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2080 	return NULL;
2081 }
2082 
2083 static const struct got_error *write_tree(struct got_object_id **,
2084     const char *, struct got_pathlist_head *, struct got_repository *,
2085     got_repo_import_cb progress_cb, void *progress_arg);
2086 
2087 static const struct got_error *
2088 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2089     const char *path, struct got_pathlist_head *ignores,
2090     struct got_repository *repo,
2091     got_repo_import_cb progress_cb, void *progress_arg)
2092 {
2093 	const struct got_error *err;
2094 	struct got_object_id *id = NULL;
2095 	char *subdirpath;
2096 
2097 	if (asprintf(&subdirpath, "%s%s%s", path,
2098 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2099 		return got_error_from_errno("asprintf");
2100 
2101 	(*new_te) = calloc(1, sizeof(**new_te));
2102 	if (*new_te == NULL)
2103 		return got_error_from_errno("calloc");
2104 	(*new_te)->mode = S_IFDIR;
2105 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2106 	    sizeof((*new_te)->name)) {
2107 		err = got_error(GOT_ERR_NO_SPACE);
2108 		goto done;
2109 	}
2110 	err = write_tree(&id, subdirpath, ignores,  repo,
2111 	    progress_cb, progress_arg);
2112 	if (err)
2113 		goto done;
2114 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2115 
2116 done:
2117 	free(id);
2118 	free(subdirpath);
2119 	if (err) {
2120 		free(*new_te);
2121 		*new_te = NULL;
2122 	}
2123 	return err;
2124 }
2125 
2126 static const struct got_error *
2127 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2128     struct got_pathlist_head *ignores, struct got_repository *repo,
2129     got_repo_import_cb progress_cb, void *progress_arg)
2130 {
2131 	const struct got_error *err = NULL;
2132 	DIR *dir;
2133 	struct dirent *de;
2134 	int nentries;
2135 	struct got_tree_entry *new_te = NULL;
2136 	struct got_pathlist_head paths;
2137 	struct got_pathlist_entry *pe;
2138 
2139 	*new_tree_id = NULL;
2140 
2141 	TAILQ_INIT(&paths);
2142 
2143 	dir = opendir(path_dir);
2144 	if (dir == NULL) {
2145 		err = got_error_from_errno2("opendir", path_dir);
2146 		goto done;
2147 	}
2148 
2149 	nentries = 0;
2150 	while ((de = readdir(dir)) != NULL) {
2151 		int ignore = 0;
2152 		int type;
2153 
2154 		if (strcmp(de->d_name, ".") == 0 ||
2155 		    strcmp(de->d_name, "..") == 0)
2156 			continue;
2157 
2158 		err = got_path_dirent_type(&type, path_dir, de);
2159 		if (err)
2160 			goto done;
2161 
2162 		TAILQ_FOREACH(pe, ignores, entry) {
2163 			if (type == DT_DIR && pe->path_len > 0 &&
2164 			    pe->path[pe->path_len - 1] == '/') {
2165 				char stripped[PATH_MAX];
2166 
2167 				if (strlcpy(stripped, pe->path,
2168 				    sizeof(stripped)) >= sizeof(stripped)) {
2169 					err = got_error(GOT_ERR_NO_SPACE);
2170 					goto done;
2171 				}
2172 				got_path_strip_trailing_slashes(stripped);
2173 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2174 					ignore = 1;
2175 					break;
2176 				}
2177 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2178 				ignore = 1;
2179 				break;
2180 			}
2181 		}
2182 		if (ignore)
2183 			continue;
2184 
2185 		if (type == DT_DIR) {
2186 			err = import_subdir(&new_te, de, path_dir,
2187 			    ignores, repo, progress_cb, progress_arg);
2188 			if (err) {
2189 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2190 					goto done;
2191 				err = NULL;
2192 				continue;
2193 			}
2194 		} else if (type == DT_REG || type == DT_LNK) {
2195 			err = import_file(&new_te, de, path_dir, repo);
2196 			if (err)
2197 				goto done;
2198 		} else
2199 			continue;
2200 
2201 		err = insert_tree_entry(new_te, &paths);
2202 		if (err)
2203 			goto done;
2204 		nentries++;
2205 	}
2206 
2207 	if (TAILQ_EMPTY(&paths)) {
2208 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2209 		    "cannot create tree without any entries");
2210 		goto done;
2211 	}
2212 
2213 	TAILQ_FOREACH(pe, &paths, entry) {
2214 		struct got_tree_entry *te = pe->data;
2215 		char *path;
2216 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2217 			continue;
2218 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2219 			err = got_error_from_errno("asprintf");
2220 			goto done;
2221 		}
2222 		err = (*progress_cb)(progress_arg, path);
2223 		free(path);
2224 		if (err)
2225 			goto done;
2226 	}
2227 
2228 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2229 done:
2230 	if (dir)
2231 		closedir(dir);
2232 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2233 	return err;
2234 }
2235 
2236 const struct got_error *
2237 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2238     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2239     struct got_repository *repo, got_repo_import_cb progress_cb,
2240     void *progress_arg)
2241 {
2242 	const struct got_error *err;
2243 	struct got_object_id *new_tree_id;
2244 
2245 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2246 	    progress_cb, progress_arg);
2247 	if (err)
2248 		return err;
2249 
2250 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2251 	    author, time(NULL), author, time(NULL), logmsg, repo);
2252 	free(new_tree_id);
2253 	return err;
2254 }
2255 
2256 const struct got_error *
2257 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2258     struct got_repository *repo)
2259 {
2260 	const struct got_error *err = NULL;
2261 	char *path_objects = NULL, *path = NULL;
2262 	DIR *dir = NULL;
2263 	struct got_object_id id;
2264 	int i;
2265 
2266 	*nobjects = 0;
2267 	*ondisk_size = 0;
2268 
2269 	path_objects = got_repo_get_path_objects(repo);
2270 	if (path_objects == NULL)
2271 		return got_error_from_errno("got_repo_get_path_objects");
2272 
2273 	for (i = 0; i <= 0xff; i++) {
2274 		struct dirent *dent;
2275 
2276 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2277 			err = got_error_from_errno("asprintf");
2278 			break;
2279 		}
2280 
2281 		dir = opendir(path);
2282 		if (dir == NULL) {
2283 			if (errno == ENOENT) {
2284 				err = NULL;
2285 				continue;
2286 			}
2287 			err = got_error_from_errno2("opendir", path);
2288 			break;
2289 		}
2290 
2291 		while ((dent = readdir(dir)) != NULL) {
2292 			char *id_str;
2293 			int fd;
2294 			struct stat sb;
2295 			enum got_hash_algorithm algo = GOT_HASH_SHA1;
2296 
2297 			if (strcmp(dent->d_name, ".") == 0 ||
2298 			    strcmp(dent->d_name, "..") == 0)
2299 				continue;
2300 
2301 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2302 				err = got_error_from_errno("asprintf");
2303 				goto done;
2304 			}
2305 
2306 			if (!got_parse_object_id(&id, id_str, algo)) {
2307 				free(id_str);
2308 				continue;
2309 			}
2310 			free(id_str);
2311 
2312 			err = got_object_open_loose_fd(&fd, &id, repo);
2313 			if (err)
2314 				goto done;
2315 
2316 			if (fstat(fd, &sb) == -1) {
2317 				err = got_error_from_errno("fstat");
2318 				close(fd);
2319 				goto done;
2320 			}
2321 			(*nobjects)++;
2322 			(*ondisk_size) += sb.st_size;
2323 
2324 			if (close(fd) == -1) {
2325 				err = got_error_from_errno("close");
2326 				goto done;
2327 			}
2328 		}
2329 
2330 		if (closedir(dir) != 0) {
2331 			err = got_error_from_errno("closedir");
2332 			goto done;
2333 		}
2334 		dir = NULL;
2335 
2336 		free(path);
2337 		path = NULL;
2338 	}
2339 done:
2340 	if (dir && closedir(dir) != 0 && err == NULL)
2341 		err = got_error_from_errno("closedir");
2342 
2343 	if (err) {
2344 		*nobjects = 0;
2345 		*ondisk_size = 0;
2346 	}
2347 	free(path_objects);
2348 	free(path);
2349 	return err;
2350 }
2351 
2352 const struct got_error *
2353 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2354     off_t *total_packsize, struct got_repository *repo)
2355 {
2356 	const struct got_error *err = NULL;
2357 	DIR *packdir = NULL;
2358 	struct dirent *dent;
2359 	struct got_packidx *packidx = NULL;
2360 	char *path_packidx;
2361 	char *path_packfile;
2362 	int packdir_fd;
2363 	struct stat sb;
2364 
2365 	*npackfiles = 0;
2366 	*nobjects = 0;
2367 	*total_packsize = 0;
2368 
2369 	packdir_fd = openat(got_repo_get_fd(repo),
2370 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2371 	if (packdir_fd == -1) {
2372 		return got_error_from_errno_fmt("openat: %s/%s",
2373 		    got_repo_get_path_git_dir(repo),
2374 		    GOT_OBJECTS_PACK_DIR);
2375 	}
2376 
2377 	packdir = fdopendir(packdir_fd);
2378 	if (packdir == NULL) {
2379 		err = got_error_from_errno("fdopendir");
2380 		goto done;
2381 	}
2382 
2383 	while ((dent = readdir(packdir)) != NULL) {
2384 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2385 			continue;
2386 
2387 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2388 		    dent->d_name) == -1) {
2389 			err = got_error_from_errno("asprintf");
2390 			goto done;
2391 		}
2392 
2393 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2394 		    path_packidx, 0);
2395 		free(path_packidx);
2396 		if (err)
2397 			goto done;
2398 
2399 		if (fstat(packidx->fd, &sb) == -1)
2400 			goto done;
2401 		*total_packsize += sb.st_size;
2402 
2403 		err = got_packidx_get_packfile_path(&path_packfile,
2404 		    packidx->path_packidx);
2405 		if (err)
2406 			goto done;
2407 
2408 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2409 		    0) == -1) {
2410 			free(path_packfile);
2411 			goto done;
2412 		}
2413 		free(path_packfile);
2414 		*total_packsize += sb.st_size;
2415 
2416 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2417 
2418 		(*npackfiles)++;
2419 
2420 		got_packidx_close(packidx);
2421 		packidx = NULL;
2422 	}
2423 done:
2424 	if (packidx)
2425 		got_packidx_close(packidx);
2426 	if (packdir && closedir(packdir) != 0 && err == NULL)
2427 		err = got_error_from_errno("closedir");
2428 	if (err) {
2429 		*npackfiles = 0;
2430 		*nobjects = 0;
2431 		*total_packsize = 0;
2432 	}
2433 	return err;
2434 }
2435 
2436 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2437     got_packidx_bloom_filter_cmp);