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/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 #include <sys/resource.h>
24 
25 #include <ctype.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <string.h>
33 #include <time.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 
40 #include "bloom.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_delta_cache.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 #include "got_lib_gotconfig.h"
62 
63 #ifndef nitems
64 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
65 #endif
66 
67 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
68 
69 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
70     got_packidx_bloom_filter_cmp);
71 
72 static inline int
73 is_boolean_val(const char *val)
74 {
75 	return (strcasecmp(val, "true") == 0 ||
76 	    strcasecmp(val, "false") == 0 ||
77 	    strcasecmp(val, "on") == 0 ||
78     	    strcasecmp(val, "off") == 0 ||
79 	    strcasecmp(val, "yes") == 0 ||
80     	    strcasecmp(val, "no") == 0 ||
81 	    strcasecmp(val, "1") == 0 ||
82 	    strcasecmp(val, "0") == 0);
83 }
84 
85 static inline int
86 get_boolean_val(const char *val)
87 {
88 	return (strcasecmp(val, "true") == 0 ||
89 	    strcasecmp(val, "on") == 0 ||
90 	    strcasecmp(val, "yes") == 0 ||
91 	    strcasecmp(val, "1") == 0);
92 }
93 
94 const char *
95 got_repo_get_path(struct got_repository *repo)
96 {
97 	return repo->path;
98 }
99 
100 const char *
101 got_repo_get_path_git_dir(struct got_repository *repo)
102 {
103 	return repo->path_git_dir;
104 }
105 
106 int
107 got_repo_get_fd(struct got_repository *repo)
108 {
109 	return repo->gitdir_fd;
110 }
111 
112 const char *
113 got_repo_get_gitconfig_author_name(struct got_repository *repo)
114 {
115 	return repo->gitconfig_author_name;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_author_email(struct got_repository *repo)
120 {
121 	return repo->gitconfig_author_email;
122 }
123 
124 const char *
125 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
126 {
127 	return repo->global_gitconfig_author_name;
128 }
129 
130 const char *
131 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
132 {
133 	return repo->global_gitconfig_author_email;
134 }
135 
136 const char *
137 got_repo_get_gitconfig_owner(struct got_repository *repo)
138 {
139 	return repo->gitconfig_owner;
140 }
141 
142 int
143 got_repo_has_extension(struct got_repository *repo, const char *ext)
144 {
145 	int i;
146 
147 	for (i = 0; i < repo->nextensions; ++i) {
148 		if (!strcasecmp(ext, repo->extnames[i]))
149 			return get_boolean_val(repo->extvals[i]);
150 	}
151 
152 	return 0;
153 }
154 
155 int
156 got_repo_is_bare(struct got_repository *repo)
157 {
158 	return (strcmp(repo->path, repo->path_git_dir) == 0);
159 }
160 
161 static char *
162 get_path_git_child(struct got_repository *repo, const char *basename)
163 {
164 	char *path_child;
165 
166 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
167 	    basename) == -1)
168 		return NULL;
169 
170 	return path_child;
171 }
172 
173 char *
174 got_repo_get_path_objects(struct got_repository *repo)
175 {
176 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
177 }
178 
179 char *
180 got_repo_get_path_objects_pack(struct got_repository *repo)
181 {
182 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
183 }
184 
185 char *
186 got_repo_get_path_refs(struct got_repository *repo)
187 {
188 	return get_path_git_child(repo, GOT_REFS_DIR);
189 }
190 
191 char *
192 got_repo_get_path_packed_refs(struct got_repository *repo)
193 {
194 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
195 }
196 
197 static char *
198 get_path_head(struct got_repository *repo)
199 {
200 	return get_path_git_child(repo, GOT_HEAD_FILE);
201 }
202 
203 char *
204 got_repo_get_path_gitconfig(struct got_repository *repo)
205 {
206 	return get_path_git_child(repo, GOT_GITCONFIG);
207 }
208 
209 char *
210 got_repo_get_path_gotconfig(struct got_repository *repo)
211 {
212 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
213 }
214 
215 const struct got_gotconfig *
216 got_repo_get_gotconfig(struct got_repository *repo)
217 {
218 	return repo->gotconfig;
219 }
220 
221 void
222 got_repo_get_gitconfig_remotes(int *nremotes,
223     const struct got_remote_repo **remotes, struct got_repository *repo)
224 {
225 	*nremotes = repo->ngitconfig_remotes;
226 	*remotes = repo->gitconfig_remotes;
227 }
228 
229 static int
230 is_git_repo(struct got_repository *repo)
231 {
232 	const char *path_git = got_repo_get_path_git_dir(repo);
233 	char *path_objects = got_repo_get_path_objects(repo);
234 	char *path_refs = got_repo_get_path_refs(repo);
235 	char *path_head = get_path_head(repo);
236 	int ret = 0;
237 	struct stat sb;
238 	struct got_reference *head_ref;
239 
240 	if (lstat(path_git, &sb) == -1)
241 		goto done;
242 	if (!S_ISDIR(sb.st_mode))
243 		goto done;
244 
245 	if (lstat(path_objects, &sb) == -1)
246 		goto done;
247 	if (!S_ISDIR(sb.st_mode))
248 		goto done;
249 
250 	if (lstat(path_refs, &sb) == -1)
251 		goto done;
252 	if (!S_ISDIR(sb.st_mode))
253 		goto done;
254 
255 	if (lstat(path_head, &sb) == -1)
256 		goto done;
257 	if (!S_ISREG(sb.st_mode))
258 		goto done;
259 
260 	/* Check if the HEAD reference can be opened. */
261 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
262 		goto done;
263 	got_ref_close(head_ref);
264 
265 	ret = 1;
266 done:
267 	free(path_objects);
268 	free(path_refs);
269 	free(path_head);
270 	return ret;
271 
272 }
273 
274 static const struct got_error *
275 close_tempfiles(int *fds, size_t nfds)
276 {
277 	const struct got_error *err = NULL;
278 	int i;
279 
280 	for (i = 0; i < nfds; i++) {
281 		if (fds[i] == -1)
282 			continue;
283 		if (close(fds[i]) == -1) {
284 			err = got_error_from_errno("close");
285 			break;
286 		}
287 	}
288 	free(fds);
289 	return err;
290 }
291 
292 static const struct got_error *
293 open_tempfiles(int **fds, size_t array_size, size_t nfds)
294 {
295 	const struct got_error *err = NULL;
296 	int i;
297 
298 	*fds = calloc(array_size, sizeof(**fds));
299 	if (*fds == NULL)
300 		return got_error_from_errno("calloc");
301 
302 	for (i = 0; i < array_size; i++)
303 		(*fds)[i] = -1;
304 
305 	for (i = 0; i < nfds; i++) {
306 		(*fds)[i] = got_opentempfd();
307 		if ((*fds)[i] == -1) {
308 			err = got_error_from_errno("got_opentempfd");
309 			close_tempfiles(*fds, nfds);
310 			*fds = NULL;
311 			return err;
312 		}
313 	}
314 
315 	return NULL;
316 }
317 
318 static const struct got_error *
319 get_pack_cache_size(int *pack_cache_size)
320 {
321 	struct rlimit rl;
322 
323 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
324 		return got_error_from_errno("getrlimit");
325 
326 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
327 	if (*pack_cache_size > rl.rlim_cur / 8)
328 		*pack_cache_size = rl.rlim_cur / 8;
329 
330 	return NULL;
331 }
332 
333 const struct got_error *
334 got_repo_pack_fds_open(int **pack_fds)
335 {
336 	const struct got_error *err;
337 	int nfds;
338 
339 	err = get_pack_cache_size(&nfds);
340 	if (err)
341 		return err;
342 
343 	/*
344 	 * We need one basefd and one accumfd per cached pack.
345 	 * Our constants should be set up in a way such that
346 	 * this error never triggers.
347 	 */
348 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
349 		return got_error(GOT_ERR_NO_SPACE);
350 
351 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
352 }
353 
354 const struct got_error *
355 got_repo_pack_fds_close(int *pack_fds)
356 {
357 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
358 }
359 
360 const struct got_error *
361 got_repo_temp_fds_open(int **temp_fds)
362 {
363 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
364 	    GOT_REPO_NUM_TEMPFILES);
365 }
366 
367 void
368 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
369 {
370 	int i;
371 
372 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
373 		repo->tempfiles[i] = temp_fds[i];
374 }
375 
376 const struct got_error *
377 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
378 {
379 	int i;
380 
381 	*fd = -1;
382 	*idx = -1;
383 
384 	for (i = 0; i < nitems(repo->tempfiles); i++) {
385 		if (repo->tempfile_use_mask & (1 << i))
386 			continue;
387 		if (repo->tempfiles[i] != -1) {
388 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
389 				return got_error_from_errno("ftruncate");
390 			*fd = repo->tempfiles[i];
391 			*idx = i;
392 			repo->tempfile_use_mask |= (1 << i);
393 			return NULL;
394 		}
395 	}
396 
397 	return got_error(GOT_ERR_REPO_TEMPFILE);
398 }
399 
400 void
401 got_repo_temp_fds_put(int idx, struct got_repository *repo)
402 {
403 	repo->tempfile_use_mask &= ~(1 << idx);
404 }
405 
406 const struct got_error *
407 got_repo_temp_fds_close(int *temp_fds)
408 {
409 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
410 }
411 
412 const struct got_error *
413 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
414     struct got_object *obj)
415 {
416 #ifndef GOT_NO_OBJ_CACHE
417 	const struct got_error *err = NULL;
418 	err = got_object_cache_add(&repo->objcache, id, obj);
419 	if (err) {
420 		if (err->code == GOT_ERR_OBJ_EXISTS ||
421 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
422 			err = NULL;
423 		return err;
424 	}
425 	obj->refcnt++;
426 #endif
427 	return NULL;
428 }
429 
430 struct got_object *
431 got_repo_get_cached_object(struct got_repository *repo,
432     struct got_object_id *id)
433 {
434 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
435 }
436 
437 const struct got_error *
438 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
439     struct got_tree_object *tree)
440 {
441 #ifndef GOT_NO_OBJ_CACHE
442 	const struct got_error *err = NULL;
443 	err = got_object_cache_add(&repo->treecache, id, tree);
444 	if (err) {
445 		if (err->code == GOT_ERR_OBJ_EXISTS ||
446 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
447 			err = NULL;
448 		return err;
449 	}
450 	tree->refcnt++;
451 #endif
452 	return NULL;
453 }
454 
455 struct got_tree_object *
456 got_repo_get_cached_tree(struct got_repository *repo,
457     struct got_object_id *id)
458 {
459 	return (struct got_tree_object *)got_object_cache_get(
460 	    &repo->treecache, id);
461 }
462 
463 const struct got_error *
464 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
465     struct got_commit_object *commit)
466 {
467 #ifndef GOT_NO_OBJ_CACHE
468 	const struct got_error *err = NULL;
469 	err = got_object_cache_add(&repo->commitcache, id, commit);
470 	if (err) {
471 		if (err->code == GOT_ERR_OBJ_EXISTS ||
472 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
473 			err = NULL;
474 		return err;
475 	}
476 	commit->refcnt++;
477 #endif
478 	return NULL;
479 }
480 
481 struct got_commit_object *
482 got_repo_get_cached_commit(struct got_repository *repo,
483     struct got_object_id *id)
484 {
485 	return (struct got_commit_object *)got_object_cache_get(
486 	    &repo->commitcache, id);
487 }
488 
489 const struct got_error *
490 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
491     struct got_tag_object *tag)
492 {
493 #ifndef GOT_NO_OBJ_CACHE
494 	const struct got_error *err = NULL;
495 	err = got_object_cache_add(&repo->tagcache, id, tag);
496 	if (err) {
497 		if (err->code == GOT_ERR_OBJ_EXISTS ||
498 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
499 			err = NULL;
500 		return err;
501 	}
502 	tag->refcnt++;
503 #endif
504 	return NULL;
505 }
506 
507 struct got_tag_object *
508 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
509 {
510 	return (struct got_tag_object *)got_object_cache_get(
511 	    &repo->tagcache, id);
512 }
513 
514 const struct got_error *
515 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
516     struct got_raw_object *raw)
517 {
518 #ifndef GOT_NO_OBJ_CACHE
519 	const struct got_error *err = NULL;
520 	err = got_object_cache_add(&repo->rawcache, id, raw);
521 	if (err) {
522 		if (err->code == GOT_ERR_OBJ_EXISTS ||
523 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
524 			err = NULL;
525 		return err;
526 	}
527 	raw->refcnt++;
528 #endif
529 	return NULL;
530 }
531 
532 
533 struct got_raw_object *
534 got_repo_get_cached_raw_object(struct got_repository *repo,
535     struct got_object_id *id)
536 {
537 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
538 }
539 
540 
541 static const struct got_error *
542 open_repo(struct got_repository *repo, const char *path)
543 {
544 	const struct got_error *err = NULL;
545 
546 	repo->gitdir_fd = -1;
547 
548 	/* bare git repository? */
549 	repo->path_git_dir = strdup(path);
550 	if (repo->path_git_dir == NULL)
551 		return got_error_from_errno("strdup");
552 	if (is_git_repo(repo)) {
553 		repo->path = strdup(repo->path_git_dir);
554 		if (repo->path == NULL) {
555 			err = got_error_from_errno("strdup");
556 			goto done;
557 		}
558 		repo->gitdir_fd = open(repo->path_git_dir,
559 		    O_DIRECTORY | O_CLOEXEC);
560 		if (repo->gitdir_fd == -1) {
561 			err = got_error_from_errno2("open",
562 			    repo->path_git_dir);
563 			goto done;
564 		}
565 		return NULL;
566 	}
567 
568 	/* git repository with working tree? */
569 	free(repo->path_git_dir);
570 	repo->path_git_dir = NULL;
571 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
572 		err = got_error_from_errno("asprintf");
573 		goto done;
574 	}
575 	if (is_git_repo(repo)) {
576 		repo->path = strdup(path);
577 		if (repo->path == NULL) {
578 			err = got_error_from_errno("strdup");
579 			goto done;
580 		}
581 		repo->gitdir_fd = open(repo->path_git_dir,
582 		    O_DIRECTORY | O_CLOEXEC);
583 		if (repo->gitdir_fd == -1) {
584 			err = got_error_from_errno2("open",
585 			    repo->path_git_dir);
586 			goto done;
587 		}
588 		return NULL;
589 	}
590 
591 	err = got_error(GOT_ERR_NOT_GIT_REPO);
592 done:
593 	if (err) {
594 		free(repo->path);
595 		repo->path = NULL;
596 		free(repo->path_git_dir);
597 		repo->path_git_dir = NULL;
598 		if (repo->gitdir_fd != -1)
599 			close(repo->gitdir_fd);
600 		repo->gitdir_fd = -1;
601 
602 	}
603 	return err;
604 }
605 
606 static const struct got_error *
607 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
608 {
609 	const struct got_error *err = NULL;
610 	char *repo_gitconfig_path = NULL;
611 
612 	if (global_gitconfig_path) {
613 		/* Read settings from ~/.gitconfig. */
614 		int dummy_repo_version;
615 		err = got_repo_read_gitconfig(&dummy_repo_version,
616 		    &repo->global_gitconfig_author_name,
617 		    &repo->global_gitconfig_author_email,
618 		    NULL, NULL, NULL, NULL, NULL, NULL,
619 		    global_gitconfig_path);
620 		if (err)
621 			return err;
622 	}
623 
624 	/* Read repository's .git/config file. */
625 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
626 	if (repo_gitconfig_path == NULL)
627 		return got_error_from_errno("got_repo_get_path_gitconfig");
628 
629 	err = got_repo_read_gitconfig(
630 	    &repo->gitconfig_repository_format_version,
631 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
632 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
633 	    &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
634 	    &repo->nextensions, repo_gitconfig_path);
635 	if (err)
636 		goto done;
637 
638 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
639 		int i;
640 
641 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
642 			got_repo_free_remote_repo_data(
643 			    &repo->gitconfig_remotes[i]);
644 		}
645 		free(repo->gitconfig_remotes);
646 		repo->gitconfig_remotes = NULL;
647 		repo->ngitconfig_remotes = 0;
648 
649 		free(repo->gitconfig_author_name);
650 		repo->gitconfig_author_name = NULL;
651 		free(repo->gitconfig_author_email);
652 		repo->gitconfig_author_email = NULL;
653 
654 		free(repo->global_gitconfig_author_name);
655 		repo->global_gitconfig_author_name = NULL;
656 		free(repo->global_gitconfig_author_email);
657 		repo->global_gitconfig_author_email = NULL;
658 	}
659 
660 done:
661 	free(repo_gitconfig_path);
662 	return err;
663 }
664 
665 static const struct got_error *
666 read_gotconfig(struct got_repository *repo)
667 {
668 	const struct got_error *err = NULL;
669 	char *gotconfig_path;
670 
671 	gotconfig_path = got_repo_get_path_gotconfig(repo);
672 	if (gotconfig_path == NULL)
673 		return got_error_from_errno("got_repo_get_path_gotconfig");
674 
675 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
676 	free(gotconfig_path);
677 	return err;
678 }
679 
680 /* Supported repository format extensions. */
681 static const char *const repo_extensions[] = {
682 	"noop",			/* Got supports repository format version 1. */
683 	"preciousObjects",	/* Supported by gotadmin cleanup. */
684 	"worktreeConfig",	/* Got does not care about Git work trees. */
685 };
686 
687 const struct got_error *
688 got_repo_open(struct got_repository **repop, const char *path,
689     const char *global_gitconfig_path, int *pack_fds)
690 {
691 	struct got_repository *repo = NULL;
692 	const struct got_error *err = NULL;
693 	char *repo_path = NULL;
694 	size_t i, j = 0;
695 
696 	*repop = NULL;
697 
698 	repo = calloc(1, sizeof(*repo));
699 	if (repo == NULL)
700 		return got_error_from_errno("calloc");
701 
702 	RB_INIT(&repo->packidx_bloom_filters);
703 	TAILQ_INIT(&repo->packidx_paths);
704 
705 	for (i = 0; i < nitems(repo->privsep_children); i++) {
706 		memset(&repo->privsep_children[i], 0,
707 		    sizeof(repo->privsep_children[0]));
708 		repo->privsep_children[i].imsg_fd = -1;
709 	}
710 
711 	err = got_object_cache_init(&repo->objcache,
712 	    GOT_OBJECT_CACHE_TYPE_OBJ);
713 	if (err)
714 		goto done;
715 	err = got_object_cache_init(&repo->treecache,
716 	    GOT_OBJECT_CACHE_TYPE_TREE);
717 	if (err)
718 		goto done;
719 	err = got_object_cache_init(&repo->commitcache,
720 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
721 	if (err)
722 		goto done;
723 	err = got_object_cache_init(&repo->tagcache,
724 	    GOT_OBJECT_CACHE_TYPE_TAG);
725 	if (err)
726 		goto done;
727 	err = got_object_cache_init(&repo->rawcache,
728 	    GOT_OBJECT_CACHE_TYPE_RAW);
729 	if (err)
730 		goto done;
731 
732 	err = get_pack_cache_size(&repo->pack_cache_size);
733 	if (err)
734 		goto done;
735 	for (i = 0; i < nitems(repo->packs); i++) {
736 		if (pack_fds != NULL && i < repo->pack_cache_size) {
737 			repo->packs[i].basefd = pack_fds[j++];
738 			repo->packs[i].accumfd = pack_fds[j++];
739 		} else {
740 			repo->packs[i].basefd = -1;
741 			repo->packs[i].accumfd = -1;
742 		}
743 	}
744 	for (i = 0; i < nitems(repo->tempfiles); i++)
745 		repo->tempfiles[i] = -1;
746 	repo->pinned_pack = -1;
747 	repo->pinned_packidx = -1;
748 	repo->pinned_pid = 0;
749 
750 	repo_path = realpath(path, NULL);
751 	if (repo_path == NULL) {
752 		err = got_error_from_errno2("realpath", path);
753 		goto done;
754 	}
755 
756 	for (;;) {
757 		char *parent_path;
758 
759 		err = open_repo(repo, repo_path);
760 		if (err == NULL)
761 			break;
762 		if (err->code != GOT_ERR_NOT_GIT_REPO)
763 			goto done;
764 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
765 			err = got_error(GOT_ERR_NOT_GIT_REPO);
766 			goto done;
767 		}
768 		err = got_path_dirname(&parent_path, repo_path);
769 		if (err)
770 			goto done;
771 		free(repo_path);
772 		repo_path = parent_path;
773 	}
774 
775 	err = read_gotconfig(repo);
776 	if (err)
777 		goto done;
778 
779 	err = read_gitconfig(repo, global_gitconfig_path);
780 	if (err)
781 		goto done;
782 	if (repo->gitconfig_repository_format_version != 0) {
783 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
784 		goto done;
785 	}
786 	for (i = 0; i < repo->nextensions; i++) {
787 		char *ext = repo->extnames[i];
788 		char *val = repo->extvals[i];
789 		int j, supported = 0;
790 
791 		if (!is_boolean_val(val)) {
792 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
793 			goto done;
794 		}
795 
796 		if (!get_boolean_val(val))
797 			continue;
798 
799 		for (j = 0; j < nitems(repo_extensions); j++) {
800 			if (strcmp(ext, repo_extensions[j]) == 0) {
801 				supported = 1;
802 				break;
803 			}
804 		}
805 		if (!supported) {
806 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
807 			goto done;
808 		}
809 	}
810 
811 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
812 done:
813 	if (err)
814 		got_repo_close(repo);
815 	else
816 		*repop = repo;
817 	free(repo_path);
818 	return err;
819 }
820 
821 const struct got_error *
822 got_repo_close(struct got_repository *repo)
823 {
824 	const struct got_error *err = NULL, *child_err;
825 	struct got_packidx_bloom_filter *bf;
826 	size_t i;
827 
828 	for (i = 0; i < repo->pack_cache_size; i++) {
829 		if (repo->packidx_cache[i] == NULL)
830 			break;
831 		got_packidx_close(repo->packidx_cache[i]);
832 	}
833 
834 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
835 	    &repo->packidx_bloom_filters))) {
836 		RB_REMOVE(got_packidx_bloom_filter_tree,
837 		    &repo->packidx_bloom_filters, bf);
838 		bloom_free(bf->bloom);
839 		free(bf->bloom);
840 		free(bf);
841 	}
842 
843 	for (i = 0; i < repo->pack_cache_size; i++)
844 		if (repo->packs[i].path_packfile)
845 			if (repo->packs[i].path_packfile)
846 				got_pack_close(&repo->packs[i]);
847 
848 	free(repo->path);
849 	free(repo->path_git_dir);
850 
851 	got_object_cache_close(&repo->objcache);
852 	got_object_cache_close(&repo->treecache);
853 	got_object_cache_close(&repo->commitcache);
854 	got_object_cache_close(&repo->tagcache);
855 	got_object_cache_close(&repo->rawcache);
856 
857 	for (i = 0; i < nitems(repo->privsep_children); i++) {
858 		if (repo->privsep_children[i].imsg_fd == -1)
859 			continue;
860 		imsg_clear(repo->privsep_children[i].ibuf);
861 		free(repo->privsep_children[i].ibuf);
862 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
863 		child_err = got_privsep_wait_for_child(
864 		    repo->privsep_children[i].pid);
865 		if (child_err && err == NULL)
866 			err = child_err;
867 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
868 		    err == NULL)
869 			err = got_error_from_errno("close");
870 	}
871 
872 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
873 	    err == NULL)
874 		err = got_error_from_errno("close");
875 
876 	if (repo->gotconfig)
877 		got_gotconfig_free(repo->gotconfig);
878 	free(repo->gitconfig_author_name);
879 	free(repo->gitconfig_author_email);
880 	for (i = 0; i < repo->ngitconfig_remotes; i++)
881 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
882 	free(repo->gitconfig_remotes);
883 	for (i = 0; i < repo->nextensions; i++) {
884 		free(repo->extnames[i]);
885 		free(repo->extvals[i]);
886 	}
887 	free(repo->extnames);
888 	free(repo->extvals);
889 
890 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
891 	free(repo);
892 
893 	return err;
894 }
895 
896 void
897 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
898 {
899 	int i;
900 
901 	free(repo->name);
902 	repo->name = NULL;
903 	free(repo->fetch_url);
904 	repo->fetch_url = NULL;
905 	free(repo->send_url);
906 	repo->send_url = NULL;
907 	for (i = 0; i < repo->nfetch_branches; i++)
908 		free(repo->fetch_branches[i]);
909 	free(repo->fetch_branches);
910 	repo->fetch_branches = NULL;
911 	repo->nfetch_branches = 0;
912 	for (i = 0; i < repo->nsend_branches; i++)
913 		free(repo->send_branches[i]);
914 	free(repo->send_branches);
915 	repo->send_branches = NULL;
916 	repo->nsend_branches = 0;
917 }
918 
919 const struct got_error *
920 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
921     const char *input_path)
922 {
923 	const struct got_error *err = NULL;
924 	const char *repo_abspath = NULL;
925 	size_t repolen, len;
926 	char *canonpath, *path = NULL;
927 
928 	*in_repo_path = NULL;
929 
930 	canonpath = strdup(input_path);
931 	if (canonpath == NULL) {
932 		err = got_error_from_errno("strdup");
933 		goto done;
934 	}
935 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
936 	if (err)
937 		goto done;
938 
939 	repo_abspath = got_repo_get_path(repo);
940 
941 	if (canonpath[0] == '\0') {
942 		path = strdup(canonpath);
943 		if (path == NULL) {
944 			err = got_error_from_errno("strdup");
945 			goto done;
946 		}
947 	} else {
948 		path = realpath(canonpath, NULL);
949 		if (path == NULL) {
950 			if (errno != ENOENT) {
951 				err = got_error_from_errno2("realpath",
952 				    canonpath);
953 				goto done;
954 			}
955 			/*
956 			 * Path is not on disk.
957 			 * Assume it is already relative to repository root.
958 			 */
959 			path = strdup(canonpath);
960 			if (path == NULL) {
961 				err = got_error_from_errno("strdup");
962 				goto done;
963 			}
964 		}
965 
966 		repolen = strlen(repo_abspath);
967 		len = strlen(path);
968 
969 
970 		if (strcmp(path, repo_abspath) == 0) {
971 			free(path);
972 			path = strdup("");
973 			if (path == NULL) {
974 				err = got_error_from_errno("strdup");
975 				goto done;
976 			}
977 		} else if (len > repolen &&
978 		    got_path_is_child(path, repo_abspath, repolen)) {
979 			/* Matched an on-disk path inside repository. */
980 			if (got_repo_is_bare(repo)) {
981 				/*
982 				 * Matched an on-disk path inside repository
983 				 * database. Treat input as repository-relative.
984 				 */
985 				free(path);
986 				path = canonpath;
987 				canonpath = NULL;
988 			} else {
989 				char *child;
990 				/* Strip common prefix with repository path. */
991 				err = got_path_skip_common_ancestor(&child,
992 				    repo_abspath, path);
993 				if (err)
994 					goto done;
995 				free(path);
996 				path = child;
997 			}
998 		} else {
999 			/*
1000 			 * Matched unrelated on-disk path.
1001 			 * Treat input as repository-relative.
1002 			 */
1003 			free(path);
1004 			path = canonpath;
1005 			canonpath = NULL;
1006 		}
1007 	}
1008 
1009 	/* Make in-repository path absolute */
1010 	if (path[0] != '/') {
1011 		char *abspath;
1012 		if (asprintf(&abspath, "/%s", path) == -1) {
1013 			err = got_error_from_errno("asprintf");
1014 			goto done;
1015 		}
1016 		free(path);
1017 		path = abspath;
1018 	}
1019 
1020 done:
1021 	free(canonpath);
1022 	if (err)
1023 		free(path);
1024 	else
1025 		*in_repo_path = path;
1026 	return err;
1027 }
1028 
1029 static const struct got_error *
1030 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1031     const char *path_packidx)
1032 {
1033 	const struct got_error *err = NULL;
1034 	size_t i;
1035 
1036 	for (i = 0; i < repo->pack_cache_size; i++) {
1037 		if (repo->packidx_cache[i] == NULL)
1038 			break;
1039 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1040 		    path_packidx) == 0) {
1041 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1042 		}
1043 	}
1044 	if (i == repo->pack_cache_size) {
1045 		do {
1046 			i--;
1047 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1048 		    i == repo->pinned_packidx);
1049 		err = got_packidx_close(repo->packidx_cache[i]);
1050 		if (err)
1051 			return err;
1052 	}
1053 
1054 	repo->packidx_cache[i] = packidx;
1055 
1056 	return NULL;
1057 }
1058 
1059 int
1060 got_repo_is_packidx_filename(const char *name, size_t len)
1061 {
1062 	if (len != GOT_PACKIDX_NAMELEN)
1063 		return 0;
1064 
1065 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1066 		return 0;
1067 
1068 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1069 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1070 		return 0;
1071 
1072 	return 1;
1073 }
1074 
1075 static struct got_packidx_bloom_filter *
1076 get_packidx_bloom_filter(struct got_repository *repo,
1077     const char *path, size_t path_len)
1078 {
1079 	struct got_packidx_bloom_filter key;
1080 
1081 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1082 		return NULL; /* XXX */
1083 	key.path_len = path_len;
1084 
1085 	return RB_FIND(got_packidx_bloom_filter_tree,
1086 	    &repo->packidx_bloom_filters, &key);
1087 }
1088 
1089 int
1090 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1091     const char *path_packidx, struct got_object_id *id)
1092 {
1093 	struct got_packidx_bloom_filter *bf;
1094 
1095 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1096 	if (bf)
1097 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1098 
1099 	/* No bloom filter means this pack index must be searched. */
1100 	return 1;
1101 }
1102 
1103 static const struct got_error *
1104 add_packidx_bloom_filter(struct got_repository *repo,
1105     struct got_packidx *packidx, const char *path_packidx)
1106 {
1107 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1108 	struct got_packidx_bloom_filter *bf;
1109 	size_t len;
1110 
1111 	/*
1112 	 * Don't use bloom filters for very large pack index files.
1113 	 * Large pack files will contain a relatively large fraction
1114 	 * of our objects so we will likely need to visit them anyway.
1115 	 * The more objects a pack file contains the higher the probability
1116 	 * of a false-positive match from the bloom filter. And reading
1117 	 * all object IDs from a large pack index file can be expensive.
1118 	 */
1119 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1120 		return NULL;
1121 
1122 	/* Do we already have a filter for this pack index? */
1123 	if (get_packidx_bloom_filter(repo, path_packidx,
1124 	    strlen(path_packidx)) != NULL)
1125 		return NULL;
1126 
1127 	bf = calloc(1, sizeof(*bf));
1128 	if (bf == NULL)
1129 		return got_error_from_errno("calloc");
1130 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1131 	if (bf->bloom == NULL) {
1132 		free(bf);
1133 		return got_error_from_errno("calloc");
1134 	}
1135 
1136 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1137 	if (len >= sizeof(bf->path)) {
1138 		free(bf->bloom);
1139 		free(bf);
1140 		return got_error(GOT_ERR_NO_SPACE);
1141 	}
1142 	bf->path_len = len;
1143 
1144 	/* Minimum size supported by our bloom filter is 1000 entries. */
1145 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1146 	for (i = 0; i < nobjects; i++) {
1147 		struct got_packidx_object_id *id;
1148 		id = &packidx->hdr.sorted_ids[i];
1149 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1150 	}
1151 
1152 	RB_INSERT(got_packidx_bloom_filter_tree,
1153 	    &repo->packidx_bloom_filters, bf);
1154 	return NULL;
1155 }
1156 
1157 static void
1158 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1159 {
1160 	struct got_pathlist_entry *pe;
1161 
1162 	while (!TAILQ_EMPTY(packidx_paths)) {
1163 		pe = TAILQ_FIRST(packidx_paths);
1164 		TAILQ_REMOVE(packidx_paths, pe, entry);
1165 		free((char *)pe->path);
1166 		free(pe);
1167 	}
1168 }
1169 
1170 static const struct got_error *
1171 refresh_packidx_paths(struct got_repository *repo)
1172 {
1173 	const struct got_error *err = NULL;
1174 	char *objects_pack_dir = NULL;
1175 	struct stat sb;
1176 
1177 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1178 	if (objects_pack_dir == NULL)
1179 		return got_error_from_errno("got_repo_get_path_objects_pack");
1180 
1181 	if (stat(objects_pack_dir, &sb) == -1) {
1182 		if (errno != ENOENT) {
1183 			err = got_error_from_errno2("stat", objects_pack_dir);
1184 			goto done;
1185 		}
1186 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1187 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1188 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1189 		purge_packidx_paths(&repo->packidx_paths);
1190 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1191 		if (err)
1192 			goto done;
1193 	}
1194 done:
1195 	free(objects_pack_dir);
1196 	return err;
1197 }
1198 
1199 const struct got_error *
1200 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1201     struct got_repository *repo, struct got_object_id *id)
1202 {
1203 	const struct got_error *err;
1204 	struct got_pathlist_entry *pe;
1205 	size_t i;
1206 
1207 	/* Search pack index cache. */
1208 	for (i = 0; i < repo->pack_cache_size; i++) {
1209 		if (repo->packidx_cache[i] == NULL)
1210 			break;
1211 		if (!got_repo_check_packidx_bloom_filter(repo,
1212 		    repo->packidx_cache[i]->path_packidx, id))
1213 			continue; /* object will not be found in this index */
1214 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1215 		if (*idx != -1) {
1216 			*packidx = repo->packidx_cache[i];
1217 			/*
1218 			 * Move this cache entry to the front. Repeatedly
1219 			 * searching a wrong pack index can be expensive.
1220 			 */
1221 			if (i > 0) {
1222 				memmove(&repo->packidx_cache[1],
1223 				    &repo->packidx_cache[0],
1224 				    i * sizeof(repo->packidx_cache[0]));
1225 				repo->packidx_cache[0] = *packidx;
1226 				if (repo->pinned_packidx >= 0 &&
1227 				    repo->pinned_packidx < i)
1228 					repo->pinned_packidx++;
1229 				else if (repo->pinned_packidx == i)
1230 					repo->pinned_packidx = 0;
1231 			}
1232 			return NULL;
1233 		}
1234 	}
1235 	/* No luck. Search the filesystem. */
1236 
1237 	err = refresh_packidx_paths(repo);
1238 	if (err)
1239 		return err;
1240 
1241 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1242 		const char *path_packidx = pe->path;
1243 		int is_cached = 0;
1244 
1245 		if (!got_repo_check_packidx_bloom_filter(repo,
1246 		    pe->path, id))
1247 			continue; /* object will not be found in this index */
1248 
1249 		for (i = 0; i < repo->pack_cache_size; i++) {
1250 			if (repo->packidx_cache[i] == NULL)
1251 				break;
1252 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1253 			    path_packidx) == 0) {
1254 				is_cached = 1;
1255 				break;
1256 			}
1257 		}
1258 		if (is_cached)
1259 			continue; /* already searched */
1260 
1261 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1262 		    path_packidx, 0);
1263 		if (err)
1264 			goto done;
1265 
1266 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1267 		if (err)
1268 			goto done;
1269 
1270 		err = cache_packidx(repo, *packidx, path_packidx);
1271 		if (err)
1272 			goto done;
1273 
1274 		*idx = got_packidx_get_object_idx(*packidx, id);
1275 		if (*idx != -1) {
1276 			err = NULL; /* found the object */
1277 			goto done;
1278 		}
1279 	}
1280 
1281 	err = got_error_no_obj(id);
1282 done:
1283 	return err;
1284 }
1285 
1286 const struct got_error *
1287 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1288     struct got_repository *repo)
1289 {
1290 	const struct got_error *err = NULL;
1291 	DIR *packdir = NULL;
1292 	struct dirent *dent;
1293 	char *path_packidx = NULL;
1294 	int packdir_fd;
1295 	struct stat sb;
1296 
1297 	packdir_fd = openat(got_repo_get_fd(repo),
1298 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1299 	if (packdir_fd == -1) {
1300 		return got_error_from_errno_fmt("openat: %s/%s",
1301 		    got_repo_get_path_git_dir(repo),
1302 		    GOT_OBJECTS_PACK_DIR);
1303 	}
1304 
1305 	packdir = fdopendir(packdir_fd);
1306 	if (packdir == NULL) {
1307 		err = got_error_from_errno("fdopendir");
1308 		goto done;
1309 	}
1310 
1311 	if (fstat(packdir_fd, &sb) == -1) {
1312 		err = got_error_from_errno("fstat");
1313 		goto done;
1314 	}
1315 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1316 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1317 
1318 	while ((dent = readdir(packdir)) != NULL) {
1319 		if (!got_repo_is_packidx_filename(dent->d_name,
1320 		    strlen(dent->d_name)))
1321 			continue;
1322 
1323 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1324 		    dent->d_name) == -1) {
1325 			err = got_error_from_errno("asprintf");
1326 			path_packidx = NULL;
1327 			break;
1328 		}
1329 
1330 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1331 		if (err)
1332 			break;
1333 	}
1334 done:
1335 	if (err)
1336 		free(path_packidx);
1337 	if (packdir && closedir(packdir) != 0 && err == NULL)
1338 		err = got_error_from_errno("closedir");
1339 	return err;
1340 }
1341 
1342 const struct got_error *
1343 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1344     struct got_repository *repo)
1345 {
1346 	const struct got_error *err;
1347 	size_t i;
1348 
1349 	*packidx = NULL;
1350 
1351 	/* Search pack index cache. */
1352 	for (i = 0; i < repo->pack_cache_size; i++) {
1353 		if (repo->packidx_cache[i] == NULL)
1354 			break;
1355 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1356 		    path_packidx) == 0) {
1357 			*packidx = repo->packidx_cache[i];
1358 			return NULL;
1359 		}
1360 	}
1361 	/* No luck. Search the filesystem. */
1362 
1363 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1364 	    path_packidx, 0);
1365 	if (err)
1366 		return err;
1367 
1368 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1369 	if (err)
1370 		goto done;
1371 
1372 	err = cache_packidx(repo, *packidx, path_packidx);
1373 done:
1374 	if (err) {
1375 		got_packidx_close(*packidx);
1376 		*packidx = NULL;
1377 	}
1378 	return err;
1379 }
1380 
1381 static const struct got_error *
1382 read_packfile_hdr(int fd, struct got_packidx *packidx)
1383 {
1384 	const struct got_error *err = NULL;
1385 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1386 	struct got_packfile_hdr hdr;
1387 	ssize_t n;
1388 
1389 	n = read(fd, &hdr, sizeof(hdr));
1390 	if (n < 0)
1391 		return got_error_from_errno("read");
1392 	if (n != sizeof(hdr))
1393 		return got_error(GOT_ERR_BAD_PACKFILE);
1394 
1395 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1396 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1397 	    be32toh(hdr.nobjects) != totobj)
1398 		err = got_error(GOT_ERR_BAD_PACKFILE);
1399 
1400 	return err;
1401 }
1402 
1403 static const struct got_error *
1404 open_packfile(int *fd, struct got_repository *repo,
1405     const char *relpath, struct got_packidx *packidx)
1406 {
1407 	const struct got_error *err = NULL;
1408 
1409 	*fd = openat(got_repo_get_fd(repo), relpath,
1410 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1411 	if (*fd == -1)
1412 		return got_error_from_errno_fmt("openat: %s/%s",
1413 		    got_repo_get_path_git_dir(repo), relpath);
1414 
1415 	if (packidx) {
1416 		err = read_packfile_hdr(*fd, packidx);
1417 		if (err) {
1418 			close(*fd);
1419 			*fd = -1;
1420 		}
1421 	}
1422 
1423 	return err;
1424 }
1425 
1426 const struct got_error *
1427 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1428     const char *path_packfile, struct got_packidx *packidx)
1429 {
1430 	const struct got_error *err = NULL;
1431 	struct got_pack *pack = NULL;
1432 	struct stat sb;
1433 	size_t i;
1434 
1435 	if (packp)
1436 		*packp = NULL;
1437 
1438 	for (i = 0; i < repo->pack_cache_size; i++) {
1439 		pack = &repo->packs[i];
1440 		if (pack->path_packfile == NULL)
1441 			break;
1442 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1443 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1444 	}
1445 
1446 	if (i == repo->pack_cache_size) {
1447 		struct got_pack tmp;
1448 		do {
1449 			i--;
1450 		} while (i > 0 && repo->pinned_pack >= 0 &&
1451 		    i == repo->pinned_pack);
1452 		err = got_pack_close(&repo->packs[i]);
1453 		if (err)
1454 			return err;
1455 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1456 			return got_error_from_errno("ftruncate");
1457 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1458 			return got_error_from_errno("ftruncate");
1459 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1460 		memcpy(&repo->packs[i], &repo->packs[0],
1461 		    sizeof(repo->packs[i]));
1462 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1463 		if (repo->pinned_pack == 0)
1464 			repo->pinned_pack = i;
1465 		else if (repo->pinned_pack == i)
1466 			repo->pinned_pack = 0;
1467 		i = 0;
1468 	}
1469 
1470 	pack = &repo->packs[i];
1471 
1472 	pack->path_packfile = strdup(path_packfile);
1473 	if (pack->path_packfile == NULL) {
1474 		err = got_error_from_errno("strdup");
1475 		goto done;
1476 	}
1477 
1478 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1479 	if (err)
1480 		goto done;
1481 
1482 	if (fstat(pack->fd, &sb) != 0) {
1483 		err = got_error_from_errno("fstat");
1484 		goto done;
1485 	}
1486 	pack->filesize = sb.st_size;
1487 
1488 	pack->privsep_child = NULL;
1489 
1490 	err = got_delta_cache_alloc(&pack->delta_cache);
1491 	if (err)
1492 		goto done;
1493 
1494 #ifndef GOT_PACK_NO_MMAP
1495 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1496 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1497 		    pack->fd, 0);
1498 		if (pack->map == MAP_FAILED) {
1499 			if (errno != ENOMEM) {
1500 				err = got_error_from_errno("mmap");
1501 				goto done;
1502 			}
1503 			pack->map = NULL; /* fall back to read(2) */
1504 		}
1505 	}
1506 #endif
1507 done:
1508 	if (err) {
1509 		if (pack)
1510 			got_pack_close(pack);
1511 	} else if (packp)
1512 		*packp = pack;
1513 	return err;
1514 }
1515 
1516 struct got_pack *
1517 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1518 {
1519 	struct got_pack *pack = NULL;
1520 	size_t i;
1521 
1522 	for (i = 0; i < repo->pack_cache_size; i++) {
1523 		pack = &repo->packs[i];
1524 		if (pack->path_packfile == NULL)
1525 			break;
1526 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1527 			return pack;
1528 	}
1529 
1530 	return NULL;
1531 }
1532 
1533 const struct got_error *
1534 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1535     struct got_pack *pack)
1536 {
1537 	size_t i;
1538 	int pinned_pack = -1, pinned_packidx = -1;
1539 
1540 	for (i = 0; i < repo->pack_cache_size; i++) {
1541 		if (repo->packidx_cache[i] &&
1542 		    strcmp(repo->packidx_cache[i]->path_packidx,
1543 		    packidx->path_packidx) == 0)
1544 			pinned_packidx = i;
1545 		if (repo->packs[i].path_packfile &&
1546 		    strcmp(repo->packs[i].path_packfile,
1547 		    pack->path_packfile) == 0)
1548 			pinned_pack = i;
1549 	}
1550 
1551 	if (pinned_packidx == -1 || pinned_pack == -1)
1552 		return got_error(GOT_ERR_PIN_PACK);
1553 
1554 	repo->pinned_pack = pinned_pack;
1555 	repo->pinned_packidx = pinned_packidx;
1556 	if (repo->packs[pinned_pack].privsep_child)
1557 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1558 	return NULL;
1559 }
1560 
1561 struct got_pack *
1562 got_repo_get_pinned_pack(struct got_repository *repo)
1563 {
1564 	if (repo->pinned_pack >= 0 &&
1565 	    repo->pinned_pack < repo->pack_cache_size)
1566 		return &repo->packs[repo->pinned_pack];
1567 
1568 	return NULL;
1569 }
1570 
1571 void
1572 got_repo_unpin_pack(struct got_repository *repo)
1573 {
1574 	repo->pinned_packidx = -1;
1575 	repo->pinned_pack = -1;
1576 	repo->pinned_pid = 0;
1577 }
1578 
1579 const struct got_error *
1580 got_repo_init(const char *repo_path, const char *head_name)
1581 {
1582 	const struct got_error *err = NULL;
1583 	const char *dirnames[] = {
1584 		GOT_OBJECTS_DIR,
1585 		GOT_OBJECTS_PACK_DIR,
1586 		GOT_REFS_DIR,
1587 	};
1588 	const char *description_str = "Unnamed repository; "
1589 	    "edit this file 'description' to name the repository.";
1590 	const char *headref = "ref: refs/heads/";
1591 	const char *gitconfig_str = "[core]\n"
1592 	    "\trepositoryformatversion = 0\n"
1593 	    "\tfilemode = true\n"
1594 	    "\tbare = true\n";
1595 	char *headref_str, *path;
1596 	size_t i;
1597 
1598 	if (!got_path_dir_is_empty(repo_path))
1599 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1600 
1601 	for (i = 0; i < nitems(dirnames); i++) {
1602 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1603 			return got_error_from_errno("asprintf");
1604 		}
1605 		err = got_path_mkdir(path);
1606 		free(path);
1607 		if (err)
1608 			return err;
1609 	}
1610 
1611 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1612 		return got_error_from_errno("asprintf");
1613 	err = got_path_create_file(path, description_str);
1614 	free(path);
1615 	if (err)
1616 		return err;
1617 
1618 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1619 		return got_error_from_errno("asprintf");
1620 	if (asprintf(&headref_str, "%s%s", headref,
1621 	    head_name ? head_name : "main") == -1) {
1622 		free(path);
1623 		return got_error_from_errno("asprintf");
1624 	}
1625 	err = got_path_create_file(path, headref_str);
1626 	free(headref_str);
1627 	free(path);
1628 	if (err)
1629 		return err;
1630 
1631 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1632 		return got_error_from_errno("asprintf");
1633 	err = got_path_create_file(path, gitconfig_str);
1634 	free(path);
1635 	if (err)
1636 		return err;
1637 
1638 	return NULL;
1639 }
1640 
1641 static const struct got_error *
1642 match_packed_object(struct got_object_id **unique_id,
1643     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1644 {
1645 	const struct got_error *err = NULL;
1646 	struct got_object_id_queue matched_ids;
1647 	struct got_pathlist_entry *pe;
1648 
1649 	STAILQ_INIT(&matched_ids);
1650 
1651 	err = refresh_packidx_paths(repo);
1652 	if (err)
1653 		return err;
1654 
1655 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1656 		const char *path_packidx = pe->path;
1657 		struct got_packidx *packidx;
1658 		struct got_object_qid *qid;
1659 
1660 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1661 		    path_packidx, 0);
1662 		if (err)
1663 			break;
1664 
1665 		err = got_packidx_match_id_str_prefix(&matched_ids,
1666 		    packidx, id_str_prefix);
1667 		if (err) {
1668 			got_packidx_close(packidx);
1669 			break;
1670 		}
1671 		err = got_packidx_close(packidx);
1672 		if (err)
1673 			break;
1674 
1675 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1676 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1677 				int matched_type;
1678 				err = got_object_get_type(&matched_type, repo,
1679 				    &qid->id);
1680 				if (err)
1681 					goto done;
1682 				if (matched_type != obj_type)
1683 					continue;
1684 			}
1685 			if (*unique_id == NULL) {
1686 				*unique_id = got_object_id_dup(&qid->id);
1687 				if (*unique_id == NULL) {
1688 					err = got_error_from_errno("malloc");
1689 					goto done;
1690 				}
1691 			} else {
1692 				if (got_object_id_cmp(*unique_id,
1693 				    &qid->id) == 0)
1694 					continue; /* packed multiple times */
1695 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1696 				goto done;
1697 			}
1698 		}
1699 	}
1700 done:
1701 	got_object_id_queue_free(&matched_ids);
1702 	if (err) {
1703 		free(*unique_id);
1704 		*unique_id = NULL;
1705 	}
1706 	return err;
1707 }
1708 
1709 static const struct got_error *
1710 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1711     const char *object_dir, const char *id_str_prefix, int obj_type,
1712     struct got_repository *repo)
1713 {
1714 	const struct got_error *err = NULL;
1715 	char *path, *id_str = NULL;
1716 	DIR *dir = NULL;
1717 	struct dirent *dent;
1718 	struct got_object_id id;
1719 
1720 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1721 		err = got_error_from_errno("asprintf");
1722 		goto done;
1723 	}
1724 
1725 	dir = opendir(path);
1726 	if (dir == NULL) {
1727 		if (errno == ENOENT) {
1728 			err = NULL;
1729 			goto done;
1730 		}
1731 		err = got_error_from_errno2("opendir", path);
1732 		goto done;
1733 	}
1734 	while ((dent = readdir(dir)) != NULL) {
1735 		int cmp;
1736 
1737 		free(id_str);
1738 		id_str = NULL;
1739 
1740 		if (strcmp(dent->d_name, ".") == 0 ||
1741 		    strcmp(dent->d_name, "..") == 0)
1742 			continue;
1743 
1744 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1745 			err = got_error_from_errno("asprintf");
1746 			goto done;
1747 		}
1748 
1749 		if (!got_parse_sha1_digest(id.sha1, id_str))
1750 			continue;
1751 
1752 		/*
1753 		 * Directory entries do not necessarily appear in
1754 		 * sorted order, so we must iterate over all of them.
1755 		 */
1756 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1757 		if (cmp != 0)
1758 			continue;
1759 
1760 		if (*unique_id == NULL) {
1761 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1762 				int matched_type;
1763 				err = got_object_get_type(&matched_type, repo,
1764 				    &id);
1765 				if (err)
1766 					goto done;
1767 				if (matched_type != obj_type)
1768 					continue;
1769 			}
1770 			*unique_id = got_object_id_dup(&id);
1771 			if (*unique_id == NULL) {
1772 				err = got_error_from_errno("got_object_id_dup");
1773 				goto done;
1774 			}
1775 		} else {
1776 			if (got_object_id_cmp(*unique_id, &id) == 0)
1777 				continue; /* both packed and loose */
1778 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1779 			goto done;
1780 		}
1781 	}
1782 done:
1783 	if (dir && closedir(dir) != 0 && err == NULL)
1784 		err = got_error_from_errno("closedir");
1785 	if (err) {
1786 		free(*unique_id);
1787 		*unique_id = NULL;
1788 	}
1789 	free(id_str);
1790 	free(path);
1791 	return err;
1792 }
1793 
1794 const struct got_error *
1795 got_repo_match_object_id_prefix(struct got_object_id **id,
1796     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1797 {
1798 	const struct got_error *err = NULL;
1799 	char *path_objects = NULL, *object_dir = NULL;
1800 	size_t len;
1801 	int i;
1802 
1803 	*id = NULL;
1804 
1805 	path_objects = got_repo_get_path_objects(repo);
1806 
1807 	len = strlen(id_str_prefix);
1808 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1809 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1810 		goto done;
1811 	}
1812 
1813 	for (i = 0; i < len; i++) {
1814 		if (isxdigit((unsigned char)id_str_prefix[i]))
1815 			continue;
1816 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1817 		goto done;
1818 	}
1819 
1820 	if (len >= 2) {
1821 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1822 		if (err)
1823 			goto done;
1824 		object_dir = strndup(id_str_prefix, 2);
1825 		if (object_dir == NULL) {
1826 			err = got_error_from_errno("strdup");
1827 			goto done;
1828 		}
1829 		err = match_loose_object(id, path_objects, object_dir,
1830 		    id_str_prefix, obj_type, repo);
1831 	} else if (len == 1) {
1832 		int i;
1833 		for (i = 0; i < 0xf; i++) {
1834 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1835 			    == -1) {
1836 				err = got_error_from_errno("asprintf");
1837 				goto done;
1838 			}
1839 			err = match_packed_object(id, repo, object_dir,
1840 			    obj_type);
1841 			if (err)
1842 				goto done;
1843 			err = match_loose_object(id, path_objects, object_dir,
1844 			    id_str_prefix, obj_type, repo);
1845 			if (err)
1846 				goto done;
1847 		}
1848 	} else {
1849 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1850 		goto done;
1851 	}
1852 done:
1853 	free(path_objects);
1854 	free(object_dir);
1855 	if (err) {
1856 		free(*id);
1857 		*id = NULL;
1858 	} else if (*id == NULL) {
1859 		switch (obj_type) {
1860 		case GOT_OBJ_TYPE_BLOB:
1861 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1862 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1863 			break;
1864 		case GOT_OBJ_TYPE_TREE:
1865 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1866 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1867 			break;
1868 		case GOT_OBJ_TYPE_COMMIT:
1869 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1870 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1871 			break;
1872 		case GOT_OBJ_TYPE_TAG:
1873 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1874 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1875 			break;
1876 		default:
1877 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1878 			break;
1879 		}
1880 	}
1881 
1882 	return err;
1883 }
1884 
1885 const struct got_error *
1886 got_repo_match_object_id(struct got_object_id **id, char **label,
1887     const char *id_str, int obj_type, struct got_reflist_head *refs,
1888     struct got_repository *repo)
1889 {
1890 	const struct got_error *err;
1891 	struct got_tag_object *tag;
1892 	struct got_reference *ref = NULL;
1893 
1894 	*id = NULL;
1895 	if (label)
1896 		*label = NULL;
1897 
1898 	if (refs) {
1899 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1900 		    refs, repo);
1901 		if (err == NULL) {
1902 			*id = got_object_id_dup(
1903 			    got_object_tag_get_object_id(tag));
1904 			if (*id == NULL)
1905 				err = got_error_from_errno("got_object_id_dup");
1906 			else if (label && asprintf(label, "refs/tags/%s",
1907 			    got_object_tag_get_name(tag)) == -1) {
1908 				err = got_error_from_errno("asprintf");
1909 				free(*id);
1910 				*id = NULL;
1911 			}
1912 			got_object_tag_close(tag);
1913 			return err;
1914 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1915 		    err->code != GOT_ERR_NO_OBJ)
1916 			return err;
1917 	}
1918 
1919 	err = got_ref_open(&ref, repo, id_str, 0);
1920 	if (err == NULL) {
1921 		err = got_ref_resolve(id, repo, ref);
1922 		if (err)
1923 			goto done;
1924 		if (label) {
1925 			*label = strdup(got_ref_get_name(ref));
1926 			if (*label == NULL) {
1927 				err = got_error_from_errno("strdup");
1928 				goto done;
1929 			}
1930 		}
1931 	} else {
1932 		if (err->code != GOT_ERR_NOT_REF &&
1933 		    err->code != GOT_ERR_BAD_REF_NAME)
1934 			goto done;
1935 		err = got_repo_match_object_id_prefix(id, id_str,
1936 		    obj_type, repo);
1937 		if (err) {
1938 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1939 				err = got_error_not_ref(id_str);
1940 			goto done;
1941 		}
1942 		if (label) {
1943 			err = got_object_id_str(label, *id);
1944 			if (*label == NULL) {
1945 				err = got_error_from_errno("strdup");
1946 				goto done;
1947 			}
1948 		}
1949 	}
1950 done:
1951 	if (ref)
1952 		got_ref_close(ref);
1953 	return err;
1954 }
1955 
1956 const struct got_error *
1957 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1958     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1959 {
1960 	const struct got_error *err = NULL;
1961 	struct got_reflist_entry *re;
1962 	struct got_object_id *tag_id;
1963 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1964 
1965 	*tag = NULL;
1966 
1967 	TAILQ_FOREACH(re, refs, entry) {
1968 		const char *refname;
1969 		refname = got_ref_get_name(re->ref);
1970 		if (got_ref_is_symbolic(re->ref))
1971 			continue;
1972 		if (strncmp(refname, "refs/tags/", 10) != 0)
1973 			continue;
1974 		if (!name_is_absolute)
1975 			refname += strlen("refs/tags/");
1976 		if (strcmp(refname, name) != 0)
1977 			continue;
1978 		err = got_ref_resolve(&tag_id, repo, re->ref);
1979 		if (err)
1980 			break;
1981 		err = got_object_open_as_tag(tag, repo, tag_id);
1982 		free(tag_id);
1983 		if (err)
1984 			break;
1985 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1986 		    got_object_tag_get_object_type(*tag) == obj_type)
1987 			break;
1988 		got_object_tag_close(*tag);
1989 		*tag = NULL;
1990 	}
1991 
1992 	if (err == NULL && *tag == NULL)
1993 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1994 		    GOT_OBJ_LABEL_TAG, name);
1995 	return err;
1996 }
1997 
1998 static const struct got_error *
1999 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2000     const char *name, mode_t mode, struct got_object_id *blob_id)
2001 {
2002 	const struct got_error *err = NULL;
2003 
2004 	 *new_te = NULL;
2005 
2006 	*new_te = calloc(1, sizeof(**new_te));
2007 	if (*new_te == NULL)
2008 		return got_error_from_errno("calloc");
2009 
2010 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2011 	    sizeof((*new_te)->name)) {
2012 		err = got_error(GOT_ERR_NO_SPACE);
2013 		goto done;
2014 	}
2015 
2016 	if (S_ISLNK(mode)) {
2017 		(*new_te)->mode = S_IFLNK;
2018 	} else {
2019 		(*new_te)->mode = S_IFREG;
2020 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2021 	}
2022 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2023 done:
2024 	if (err && *new_te) {
2025 		free(*new_te);
2026 		*new_te = NULL;
2027 	}
2028 	return err;
2029 }
2030 
2031 static const struct got_error *
2032 import_file(struct got_tree_entry **new_te, struct dirent *de,
2033     const char *path, struct got_repository *repo)
2034 {
2035 	const struct got_error *err;
2036 	struct got_object_id *blob_id = NULL;
2037 	char *filepath;
2038 	struct stat sb;
2039 
2040 	if (asprintf(&filepath, "%s%s%s", path,
2041 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2042 		return got_error_from_errno("asprintf");
2043 
2044 	if (lstat(filepath, &sb) != 0) {
2045 		err = got_error_from_errno2("lstat", path);
2046 		goto done;
2047 	}
2048 
2049 	err = got_object_blob_create(&blob_id, filepath, repo);
2050 	if (err)
2051 		goto done;
2052 
2053 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2054 	    blob_id);
2055 done:
2056 	free(filepath);
2057 	if (err)
2058 		free(blob_id);
2059 	return err;
2060 }
2061 
2062 static const struct got_error *
2063 insert_tree_entry(struct got_tree_entry *new_te,
2064     struct got_pathlist_head *paths)
2065 {
2066 	const struct got_error *err = NULL;
2067 	struct got_pathlist_entry *new_pe;
2068 
2069 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2070 	if (err)
2071 		return err;
2072 	if (new_pe == NULL)
2073 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2074 	return NULL;
2075 }
2076 
2077 static const struct got_error *write_tree(struct got_object_id **,
2078     const char *, struct got_pathlist_head *, struct got_repository *,
2079     got_repo_import_cb progress_cb, void *progress_arg);
2080 
2081 static const struct got_error *
2082 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2083     const char *path, struct got_pathlist_head *ignores,
2084     struct got_repository *repo,
2085     got_repo_import_cb progress_cb, void *progress_arg)
2086 {
2087 	const struct got_error *err;
2088 	struct got_object_id *id = NULL;
2089 	char *subdirpath;
2090 
2091 	if (asprintf(&subdirpath, "%s%s%s", path,
2092 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2093 		return got_error_from_errno("asprintf");
2094 
2095 	(*new_te) = calloc(1, sizeof(**new_te));
2096 	if (*new_te == NULL)
2097 		return got_error_from_errno("calloc");
2098 	(*new_te)->mode = S_IFDIR;
2099 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2100 	    sizeof((*new_te)->name)) {
2101 		err = got_error(GOT_ERR_NO_SPACE);
2102 		goto done;
2103 	}
2104 	err = write_tree(&id, subdirpath, ignores,  repo,
2105 	    progress_cb, progress_arg);
2106 	if (err)
2107 		goto done;
2108 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2109 
2110 done:
2111 	free(id);
2112 	free(subdirpath);
2113 	if (err) {
2114 		free(*new_te);
2115 		*new_te = NULL;
2116 	}
2117 	return err;
2118 }
2119 
2120 static const struct got_error *
2121 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2122     struct got_pathlist_head *ignores, struct got_repository *repo,
2123     got_repo_import_cb progress_cb, void *progress_arg)
2124 {
2125 	const struct got_error *err = NULL;
2126 	DIR *dir;
2127 	struct dirent *de;
2128 	int nentries;
2129 	struct got_tree_entry *new_te = NULL;
2130 	struct got_pathlist_head paths;
2131 	struct got_pathlist_entry *pe;
2132 
2133 	*new_tree_id = NULL;
2134 
2135 	TAILQ_INIT(&paths);
2136 
2137 	dir = opendir(path_dir);
2138 	if (dir == NULL) {
2139 		err = got_error_from_errno2("opendir", path_dir);
2140 		goto done;
2141 	}
2142 
2143 	nentries = 0;
2144 	while ((de = readdir(dir)) != NULL) {
2145 		int ignore = 0;
2146 		int type;
2147 
2148 		if (strcmp(de->d_name, ".") == 0 ||
2149 		    strcmp(de->d_name, "..") == 0)
2150 			continue;
2151 
2152 		err = got_path_dirent_type(&type, path_dir, de);
2153 		if (err)
2154 			goto done;
2155 
2156 		TAILQ_FOREACH(pe, ignores, entry) {
2157 			if (type == DT_DIR && pe->path_len > 0 &&
2158 			    pe->path[pe->path_len - 1] == '/') {
2159 				char stripped[PATH_MAX];
2160 
2161 				if (strlcpy(stripped, pe->path,
2162 				    sizeof(stripped)) >= sizeof(stripped)) {
2163 					err = got_error(GOT_ERR_NO_SPACE);
2164 					goto done;
2165 				}
2166 				got_path_strip_trailing_slashes(stripped);
2167 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2168 					ignore = 1;
2169 					break;
2170 				}
2171 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2172 				ignore = 1;
2173 				break;
2174 			}
2175 		}
2176 		if (ignore)
2177 			continue;
2178 
2179 		if (type == DT_DIR) {
2180 			err = import_subdir(&new_te, de, path_dir,
2181 			    ignores, repo, progress_cb, progress_arg);
2182 			if (err) {
2183 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2184 					goto done;
2185 				err = NULL;
2186 				continue;
2187 			}
2188 		} else if (type == DT_REG || type == DT_LNK) {
2189 			err = import_file(&new_te, de, path_dir, repo);
2190 			if (err)
2191 				goto done;
2192 		} else
2193 			continue;
2194 
2195 		err = insert_tree_entry(new_te, &paths);
2196 		if (err)
2197 			goto done;
2198 		nentries++;
2199 	}
2200 
2201 	if (TAILQ_EMPTY(&paths)) {
2202 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2203 		    "cannot create tree without any entries");
2204 		goto done;
2205 	}
2206 
2207 	TAILQ_FOREACH(pe, &paths, entry) {
2208 		struct got_tree_entry *te = pe->data;
2209 		char *path;
2210 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2211 			continue;
2212 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2213 			err = got_error_from_errno("asprintf");
2214 			goto done;
2215 		}
2216 		err = (*progress_cb)(progress_arg, path);
2217 		free(path);
2218 		if (err)
2219 			goto done;
2220 	}
2221 
2222 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2223 done:
2224 	if (dir)
2225 		closedir(dir);
2226 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2227 	return err;
2228 }
2229 
2230 const struct got_error *
2231 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2232     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2233     struct got_repository *repo, got_repo_import_cb progress_cb,
2234     void *progress_arg)
2235 {
2236 	const struct got_error *err;
2237 	struct got_object_id *new_tree_id;
2238 
2239 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2240 	    progress_cb, progress_arg);
2241 	if (err)
2242 		return err;
2243 
2244 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2245 	    author, time(NULL), author, time(NULL), logmsg, repo);
2246 	free(new_tree_id);
2247 	return err;
2248 }
2249 
2250 const struct got_error *
2251 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2252     struct got_repository *repo)
2253 {
2254 	const struct got_error *err = NULL;
2255 	char *path_objects = NULL, *path = NULL;
2256 	DIR *dir = NULL;
2257 	struct got_object_id id;
2258 	int i;
2259 
2260 	*nobjects = 0;
2261 	*ondisk_size = 0;
2262 
2263 	path_objects = got_repo_get_path_objects(repo);
2264 	if (path_objects == NULL)
2265 		return got_error_from_errno("got_repo_get_path_objects");
2266 
2267 	for (i = 0; i <= 0xff; i++) {
2268 		struct dirent *dent;
2269 
2270 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2271 			err = got_error_from_errno("asprintf");
2272 			break;
2273 		}
2274 
2275 		dir = opendir(path);
2276 		if (dir == NULL) {
2277 			if (errno == ENOENT) {
2278 				err = NULL;
2279 				continue;
2280 			}
2281 			err = got_error_from_errno2("opendir", path);
2282 			break;
2283 		}
2284 
2285 		while ((dent = readdir(dir)) != NULL) {
2286 			char *id_str;
2287 			int fd;
2288 			struct stat sb;
2289 
2290 			if (strcmp(dent->d_name, ".") == 0 ||
2291 			    strcmp(dent->d_name, "..") == 0)
2292 				continue;
2293 
2294 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2295 				err = got_error_from_errno("asprintf");
2296 				goto done;
2297 			}
2298 
2299 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2300 				free(id_str);
2301 				continue;
2302 			}
2303 			free(id_str);
2304 
2305 			err = got_object_open_loose_fd(&fd, &id, repo);
2306 			if (err)
2307 				goto done;
2308 
2309 			if (fstat(fd, &sb) == -1) {
2310 				err = got_error_from_errno("fstat");
2311 				close(fd);
2312 				goto done;
2313 			}
2314 			(*nobjects)++;
2315 			(*ondisk_size) += sb.st_size;
2316 
2317 			if (close(fd) == -1) {
2318 				err = got_error_from_errno("close");
2319 				goto done;
2320 			}
2321 		}
2322 
2323 		if (closedir(dir) != 0) {
2324 			err = got_error_from_errno("closedir");
2325 			goto done;
2326 		}
2327 		dir = NULL;
2328 
2329 		free(path);
2330 		path = NULL;
2331 	}
2332 done:
2333 	if (dir && closedir(dir) != 0 && err == NULL)
2334 		err = got_error_from_errno("closedir");
2335 
2336 	if (err) {
2337 		*nobjects = 0;
2338 		*ondisk_size = 0;
2339 	}
2340 	free(path_objects);
2341 	free(path);
2342 	return err;
2343 }
2344 
2345 const struct got_error *
2346 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2347     off_t *total_packsize, struct got_repository *repo)
2348 {
2349 	const struct got_error *err = NULL;
2350 	DIR *packdir = NULL;
2351 	struct dirent *dent;
2352 	struct got_packidx *packidx = NULL;
2353 	char *path_packidx;
2354 	char *path_packfile;
2355 	int packdir_fd;
2356 	struct stat sb;
2357 
2358 	*npackfiles = 0;
2359 	*nobjects = 0;
2360 	*total_packsize = 0;
2361 
2362 	packdir_fd = openat(got_repo_get_fd(repo),
2363 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2364 	if (packdir_fd == -1) {
2365 		return got_error_from_errno_fmt("openat: %s/%s",
2366 		    got_repo_get_path_git_dir(repo),
2367 		    GOT_OBJECTS_PACK_DIR);
2368 	}
2369 
2370 	packdir = fdopendir(packdir_fd);
2371 	if (packdir == NULL) {
2372 		err = got_error_from_errno("fdopendir");
2373 		goto done;
2374 	}
2375 
2376 	while ((dent = readdir(packdir)) != NULL) {
2377 		if (!got_repo_is_packidx_filename(dent->d_name,
2378 		    strlen(dent->d_name)))
2379 			continue;
2380 
2381 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2382 		    dent->d_name) == -1) {
2383 			err = got_error_from_errno("asprintf");
2384 			goto done;
2385 		}
2386 
2387 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2388 		    path_packidx, 0);
2389 		free(path_packidx);
2390 		if (err)
2391 			goto done;
2392 
2393 		if (fstat(packidx->fd, &sb) == -1)
2394 			goto done;
2395 		*total_packsize += sb.st_size;
2396 
2397 		err = got_packidx_get_packfile_path(&path_packfile,
2398 		    packidx->path_packidx);
2399 		if (err)
2400 			goto done;
2401 
2402 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2403 		    0) == -1) {
2404 			free(path_packfile);
2405 			goto done;
2406 		}
2407 		free(path_packfile);
2408 		*total_packsize += sb.st_size;
2409 
2410 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2411 
2412 		(*npackfiles)++;
2413 
2414 		got_packidx_close(packidx);
2415 		packidx = NULL;
2416 	}
2417 done:
2418 	if (packidx)
2419 		got_packidx_close(packidx);
2420 	if (packdir && closedir(packdir) != 0 && err == NULL)
2421 		err = got_error_from_errno("closedir");
2422 	if (err) {
2423 		*npackfiles = 0;
2424 		*nobjects = 0;
2425 		*total_packsize = 0;
2426 	}
2427 	return err;
2428 }
2429 
2430 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2431     got_packidx_bloom_filter_cmp);