op public repos

Blob

Date:: Sun Sep 1 11:10:13 2019 UTC
Message:: prevent theoretical double-free with non-OpenBSD asprintf (found by jasper)
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019 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/stat.h>
21 #include <sys/mman.h>
22 #include <sys/syslimits.h>
23 
24 #include <ctype.h>
25 #include <fcntl.h>
26 #include <fnmatch.h>
27 #include <limits.h>
28 #include <dirent.h>
29 #include <stdlib.h>
30 #include <stdio.h>
31 #include <sha1.h>
32 #include <string.h>
33 #include <zlib.h>
34 #include <errno.h>
35 #include <libgen.h>
36 #include <stdint.h>
37 #include <imsg.h>
38 #include <uuid.h>
39 
40 #include "got_error.h"
41 #include "got_reference.h"
42 #include "got_repository.h"
43 #include "got_path.h"
44 #include "got_cancel.h"
45 #include "got_worktree.h"
46 #include "got_object.h"
47 
48 #include "got_lib_delta.h"
49 #include "got_lib_inflate.h"
50 #include "got_lib_object.h"
51 #include "got_lib_object_parse.h"
52 #include "got_lib_object_create.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_privsep.h"
55 #include "got_lib_worktree.h"
56 #include "got_lib_sha1.h"
57 #include "got_lib_object_cache.h"
58 #include "got_lib_repository.h"
59 #include "got_lib_gitconfig.h"
60 
61 #ifndef nitems
62 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
63 #endif
64 
65 #define GOT_GIT_DIR	".git"
66 
67 /* Mandatory files and directories inside the git directory. */
68 #define GOT_OBJECTS_DIR		"objects"
69 #define GOT_REFS_DIR		"refs"
70 #define GOT_HEAD_FILE		"HEAD"
71 #define GOT_GITCONFIG		"config"
72 
73 /* Other files and directories inside the git directory. */
74 #define GOT_FETCH_HEAD_FILE	"FETCH_HEAD"
75 #define GOT_ORIG_HEAD_FILE	"ORIG_HEAD"
76 #define GOT_OBJECTS_PACK_DIR	"objects/pack"
77 #define GOT_PACKED_REFS_FILE	"packed-refs"
78 
79 const char *
80 got_repo_get_path(struct got_repository *repo)
81 {
82 	return repo->path;
83 }
84 
85 const char *
86 got_repo_get_path_git_dir(struct got_repository *repo)
87 {
88 	return repo->path_git_dir;
89 }
90 
91 int
92 got_repo_is_bare(struct got_repository *repo)
93 {
94 	return (strcmp(repo->path, repo->path_git_dir) == 0);
95 }
96 
97 static char *
98 get_path_git_child(struct got_repository *repo, const char *basename)
99 {
100 	char *path_child;
101 	
102 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
103 	    basename) == -1)
104 		return NULL;
105 
106 	return path_child;
107 }
108 
109 char *
110 got_repo_get_path_objects(struct got_repository *repo)
111 {
112 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
113 }
114 
115 char *
116 got_repo_get_path_objects_pack(struct got_repository *repo)
117 {
118 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
119 }
120 
121 char *
122 got_repo_get_path_refs(struct got_repository *repo)
123 {
124 	return get_path_git_child(repo, GOT_REFS_DIR);
125 }
126 
127 char *
128 got_repo_get_path_packed_refs(struct got_repository *repo)
129 {
130 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
131 }
132 
133 static char *
134 get_path_head(struct got_repository *repo)
135 {
136 	return get_path_git_child(repo, GOT_HEAD_FILE);
137 }
138 
139 static const struct got_error *
140 get_path_gitconfig(char **p, struct got_repository *repo)
141 {
142 	*p = get_path_git_child(repo, GOT_GITCONFIG);
143 	if (*p == NULL)
144 		return got_error_from_errno("asprintf");
145 	return NULL;
146 }
147 
148 static int
149 is_git_repo(struct got_repository *repo)
150 {
151 	const char *path_git = got_repo_get_path_git_dir(repo);
152 	char *path_objects = got_repo_get_path_objects(repo);
153 	char *path_refs = got_repo_get_path_refs(repo);
154 	char *path_head = get_path_head(repo);
155 	int ret = 0;
156 	struct stat sb;
157 	struct got_reference *head_ref;
158 
159 	if (lstat(path_git, &sb) == -1)
160 		goto done;
161 	if (!S_ISDIR(sb.st_mode))
162 		goto done;
163 
164 	if (lstat(path_objects, &sb) == -1)
165 		goto done;
166 	if (!S_ISDIR(sb.st_mode))
167 		goto done;
168 
169 	if (lstat(path_refs, &sb) == -1)
170 		goto done;
171 	if (!S_ISDIR(sb.st_mode))
172 		goto done;
173 
174 	if (lstat(path_head, &sb) == -1)
175 		goto done;
176 	if (!S_ISREG(sb.st_mode))
177 		goto done;
178 
179 	/* Check if the HEAD reference can be opened. */
180 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
181 		goto done;
182 	got_ref_close(head_ref);
183 
184 	ret = 1;
185 done:
186 	free(path_objects);
187 	free(path_refs);
188 	free(path_head);
189 	return ret;
190 
191 }
192 
193 const struct got_error *
194 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
195     struct got_object *obj)
196 {
197 #ifndef GOT_NO_OBJ_CACHE
198 	const struct got_error *err = NULL;
199 	err = got_object_cache_add(&repo->objcache, id, obj);
200 	if (err) {
201 		if (err->code == GOT_ERR_OBJ_EXISTS ||
202 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
203 			err = NULL;
204 		return err;
205 	}
206 	obj->refcnt++;
207 #endif
208 	return NULL;
209 }
210 
211 struct got_object *
212 got_repo_get_cached_object(struct got_repository *repo,
213     struct got_object_id *id)
214 {
215 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
216 }
217 
218 const struct got_error *
219 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
220     struct got_tree_object *tree)
221 {
222 #ifndef GOT_NO_OBJ_CACHE
223 	const struct got_error *err = NULL;
224 	err = got_object_cache_add(&repo->treecache, id, tree);
225 	if (err) {
226 		if (err->code == GOT_ERR_OBJ_EXISTS ||
227 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
228 			err = NULL;
229 		return err;
230 	}
231 	tree->refcnt++;
232 #endif
233 	return NULL;
234 }
235 
236 struct got_tree_object *
237 got_repo_get_cached_tree(struct got_repository *repo,
238     struct got_object_id *id)
239 {
240 	return (struct got_tree_object *)got_object_cache_get(
241 	    &repo->treecache, id);
242 }
243 
244 const struct got_error *
245 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
246     struct got_commit_object *commit)
247 {
248 #ifndef GOT_NO_OBJ_CACHE
249 	const struct got_error *err = NULL;
250 	err = got_object_cache_add(&repo->commitcache, id, commit);
251 	if (err) {
252 		if (err->code == GOT_ERR_OBJ_EXISTS ||
253 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
254 			err = NULL;
255 		return err;
256 	}
257 	commit->refcnt++;
258 #endif
259 	return NULL;
260 }
261 
262 struct got_commit_object *
263 got_repo_get_cached_commit(struct got_repository *repo,
264     struct got_object_id *id)
265 {
266 	return (struct got_commit_object *)got_object_cache_get(
267 	    &repo->commitcache, id);
268 }
269 
270 const struct got_error *
271 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
272     struct got_tag_object *tag)
273 {
274 #ifndef GOT_NO_OBJ_CACHE
275 	const struct got_error *err = NULL;
276 	err = got_object_cache_add(&repo->tagcache, id, tag);
277 	if (err) {
278 		if (err->code == GOT_ERR_OBJ_EXISTS ||
279 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
280 			err = NULL;
281 		return err;
282 	}
283 	tag->refcnt++;
284 #endif
285 	return NULL;
286 }
287 
288 struct got_tag_object *
289 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
290 {
291 	return (struct got_tag_object *)got_object_cache_get(
292 	    &repo->tagcache, id);
293 }
294 
295 const struct got_error *
296 open_repo(struct got_repository *repo, const char *path)
297 {
298 	const struct got_error *err = NULL;
299 
300 	/* bare git repository? */
301 	repo->path_git_dir = strdup(path);
302 	if (repo->path_git_dir == NULL)
303 		return got_error_from_errno("strdup");
304 	if (is_git_repo(repo)) {
305 		repo->path = strdup(repo->path_git_dir);
306 		if (repo->path == NULL) {
307 			err = got_error_from_errno("strdup");
308 			goto done;
309 		}
310 		return NULL;
311 	}
312 
313 	/* git repository with working tree? */
314 	free(repo->path_git_dir);
315 	repo->path_git_dir = NULL;
316 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
317 		err = got_error_from_errno("asprintf");
318 		goto done;
319 	}
320 	if (is_git_repo(repo)) {
321 		repo->path = strdup(path);
322 		if (repo->path == NULL) {
323 			err = got_error_from_errno("strdup");
324 			goto done;
325 		}
326 		return NULL;
327 	}
328 
329 	err = got_error(GOT_ERR_NOT_GIT_REPO);
330 done:
331 	if (err) {
332 		free(repo->path);
333 		repo->path = NULL;
334 		free(repo->path_git_dir);
335 		repo->path_git_dir = NULL;
336 	}
337 	return err;
338 }
339 
340 const struct got_error *
341 got_repo_open(struct got_repository **repop, const char *path)
342 {
343 	struct got_repository *repo = NULL;
344 	const struct got_error *err = NULL;
345 	char *abspath, *gitconfig_path = NULL;
346 	int i, tried_root = 0;
347 
348 	*repop = NULL;
349 
350 	if (got_path_is_absolute(path))
351 		abspath = strdup(path);
352 	else
353 		abspath = got_path_get_absolute(path);
354 	if (abspath == NULL)
355 		return got_error(GOT_ERR_BAD_PATH);
356 
357 	repo = calloc(1, sizeof(*repo));
358 	if (repo == NULL) {
359 		err = got_error_from_errno("calloc");
360 		goto done;
361 	}
362 
363 	for (i = 0; i < nitems(repo->privsep_children); i++) {
364 		memset(&repo->privsep_children[i], 0,
365 		    sizeof(repo->privsep_children[0]));
366 		repo->privsep_children[i].imsg_fd = -1;
367 	}
368 
369 	err = got_object_cache_init(&repo->objcache,
370 	    GOT_OBJECT_CACHE_TYPE_OBJ);
371 	if (err)
372 		goto done;
373 	err = got_object_cache_init(&repo->treecache,
374 	    GOT_OBJECT_CACHE_TYPE_TREE);
375 	if (err)
376 		goto done;
377 	err = got_object_cache_init(&repo->commitcache,
378 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
379 	if (err)
380 		goto done;
381 	err = got_object_cache_init(&repo->tagcache,
382 	    GOT_OBJECT_CACHE_TYPE_TAG);
383 	if (err)
384 		goto done;
385 
386 	path = realpath(abspath, NULL);
387 	if (path == NULL) {
388 		err = got_error_from_errno2("realpath", abspath);
389 		goto done;
390 	}
391 
392 	do {
393 		err = open_repo(repo, path);
394 		if (err == NULL)
395 			break;
396 		if (err->code != GOT_ERR_NOT_GIT_REPO)
397 			break;
398 		if (path[0] == '/' && path[1] == '\0') {
399 			if (tried_root) {
400 				err = got_error(GOT_ERR_NOT_GIT_REPO);
401 				goto done;
402 			}
403 			tried_root = 1;
404 		}
405 		path = dirname(path);
406 		if (path == NULL) {
407 			err = got_error_from_errno2("dirname", path);
408 			goto done;
409 		}
410 	} while (path);
411 
412 	err = get_path_gitconfig(&gitconfig_path, repo);
413 	if (err)
414 		goto done;
415 
416 #ifdef notyet
417 	err = got_gitconfig_open(&repo->gitconfig, gitconfig_path);
418 	if (err)
419 		goto done;
420 #else
421 	repo->gitconfig = NULL;
422 #endif
423 done:
424 	if (err)
425 		got_repo_close(repo);
426 	else
427 		*repop = repo;
428 	free(abspath);
429 	free(gitconfig_path);
430 	return err;
431 }
432 
433 const struct got_error *
434 got_repo_close(struct got_repository *repo)
435 {
436 	const struct got_error *err = NULL, *child_err;
437 	int i;
438 
439 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
440 		if (repo->packidx_cache[i] == NULL)
441 			break;
442 		got_packidx_close(repo->packidx_cache[i]);
443 	}
444 
445 	for (i = 0; i < nitems(repo->packs); i++) {
446 		if (repo->packs[i].path_packfile == NULL)
447 			break;
448 		got_pack_close(&repo->packs[i]);
449 	}
450 
451 	free(repo->path);
452 	free(repo->path_git_dir);
453 
454 	got_object_cache_close(&repo->objcache);
455 	got_object_cache_close(&repo->treecache);
456 	got_object_cache_close(&repo->commitcache);
457 	got_object_cache_close(&repo->tagcache);
458 
459 	for (i = 0; i < nitems(repo->privsep_children); i++) {
460 		if (repo->privsep_children[i].imsg_fd == -1)
461 			continue;
462 		imsg_clear(repo->privsep_children[i].ibuf);
463 		free(repo->privsep_children[i].ibuf);
464 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
465 		child_err = got_privsep_wait_for_child(
466 		    repo->privsep_children[i].pid);
467 		if (child_err && err == NULL)
468 			err = child_err;
469 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
470 		    err == NULL)
471 			err = got_error_from_errno("close");
472 	}
473 	if (repo->gitconfig)
474 		got_gitconfig_close(repo->gitconfig);
475 	free(repo);
476 
477 	return err;
478 }
479 
480 const struct got_error *
481 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
482     const char *input_path, int check_disk)
483 {
484 	const struct got_error *err = NULL;
485 	const char *repo_abspath = NULL;
486 	size_t repolen, cwdlen, len;
487 	char *cwd, *canonpath, *path = NULL;
488 
489 	*in_repo_path = NULL;
490 
491 	cwd = getcwd(NULL, 0);
492 	if (cwd == NULL)
493 		return got_error_from_errno("getcwd");
494 
495 	canonpath = strdup(input_path);
496 	if (canonpath == NULL) {
497 		err = got_error_from_errno("strdup");
498 		goto done;
499 	}
500 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
501 	if (err)
502 		goto done;
503 
504 	repo_abspath = got_repo_get_path(repo);
505 
506 	if (!check_disk || canonpath[0] == '\0') {
507 		path = strdup(canonpath);
508 		if (path == NULL) {
509 			err = got_error_from_errno("strdup");
510 			goto done;
511 		}
512 	} else {
513 		int is_repo_child = 0, is_cwd_child = 0;
514 
515 		path = realpath(canonpath, NULL);
516 		if (path == NULL) {
517 			if (errno != ENOENT) {
518 				err = got_error_from_errno2("realpath",
519 				    canonpath);
520 				goto done;
521 			}
522 			/*
523 			 * Path is not on disk.
524 			 * Assume it is already relative to repository root.
525 			 */
526 			path = strdup(canonpath);
527 			if (path == NULL) {
528 				err = got_error_from_errno("strdup");
529 				goto done;
530 			}
531 		}
532 
533 		repolen = strlen(repo_abspath);
534 		cwdlen = strlen(cwd);
535 		len = strlen(path);
536 
537 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
538 			is_repo_child = 1;
539 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
540 			is_cwd_child = 1;
541 
542 		if (strcmp(path, repo_abspath) == 0) {
543 			free(path);
544 			path = strdup("");
545 			if (path == NULL) {
546 				err = got_error_from_errno("strdup");
547 				goto done;
548 			}
549 		} else if (is_repo_child && is_cwd_child) {
550 			char *child;
551 			/* Strip common prefix with repository path. */
552 			err = got_path_skip_common_ancestor(&child,
553 			    repo_abspath, path);
554 			if (err)
555 				goto done;
556 			free(path);
557 			path = child;
558 		} else if (is_repo_child) {
559 			/* Matched an on-disk path inside repository. */
560 			if (got_repo_is_bare(repo)) {
561 				/*
562 				 * Matched an on-disk path inside repository
563 				 * database. Treat as repository-relative.
564 				 */
565 			} else {
566 				char *child;
567 				/* Strip common prefix with repository path. */
568 				err = got_path_skip_common_ancestor(&child,
569 				    repo_abspath, path);
570 				if (err)
571 					goto done;
572 				free(path);
573 				path = child;
574 			}
575 		} else if (is_cwd_child) {
576 			char *child;
577 			/* Strip common prefix with cwd. */
578 			err = got_path_skip_common_ancestor(&child, cwd,
579 			    path);
580 			if (err)
581 				goto done;
582 			free(path);
583 			path = child;
584 		} else {
585 			/*
586 			 * Matched unrelated on-disk path.
587 			 * Treat it as repository-relative.
588 			 */
589 		}
590 	}
591 
592 	/* Make in-repository path absolute */
593 	if (path[0] != '/') {
594 		char *abspath;
595 		if (asprintf(&abspath, "/%s", path) == -1) {
596 			err = got_error_from_errno("asprintf");
597 			goto done;
598 		}
599 		free(path);
600 		path = abspath;
601 	}
602 
603 done:
604 	free(cwd);
605 	free(canonpath);
606 	if (err)
607 		free(path);
608 	else
609 		*in_repo_path = path;
610 	return err;
611 }
612 
613 const struct got_error *
614 got_repo_cache_packidx(struct got_repository *repo, struct got_packidx *packidx)
615 {
616 	const struct got_error *err = NULL;
617 	int i;
618 
619 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
620 		if (repo->packidx_cache[i] == NULL)
621 			break;
622 	}
623 	if (i == nitems(repo->packidx_cache)) {
624 		err = got_packidx_close(repo->packidx_cache[i - 1]);
625 		if (err)
626 			return err;
627 	}
628 
629 	/*
630 	 * Insert the new pack index at the front so it will
631 	 * be searched first in the future.
632 	 */
633 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
634 	    sizeof(repo->packidx_cache) -
635 	    sizeof(repo->packidx_cache[0]));
636 	repo->packidx_cache[0] = packidx;
637 
638 	return NULL;
639 }
640 
641 static int
642 is_packidx_filename(const char *name, size_t len)
643 {
644 	if (len != GOT_PACKIDX_NAMELEN)
645 		return 0;
646 
647 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
648 		return 0;
649 
650 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
651 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
652 		return 0;
653 
654 	return 1;
655 }
656 
657 const struct got_error *
658 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
659     struct got_repository *repo, struct got_object_id *id)
660 {
661 	const struct got_error *err;
662 	char *path_packdir;
663 	DIR *packdir;
664 	struct dirent *dent;
665 	char *path_packidx;
666 	int i;
667 
668 	/* Search pack index cache. */
669 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
670 		if (repo->packidx_cache[i] == NULL)
671 			break;
672 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
673 		if (*idx != -1) {
674 			*packidx = repo->packidx_cache[i];
675 			return NULL;
676 		}
677 	}
678 	/* No luck. Search the filesystem. */
679 
680 	path_packdir = got_repo_get_path_objects_pack(repo);
681 	if (path_packdir == NULL)
682 		return got_error_from_errno("got_repo_get_path_objects_pack");
683 
684 	packdir = opendir(path_packdir);
685 	if (packdir == NULL) {
686 		if (errno == ENOENT)
687 			err = got_error_no_obj(id);
688 		else
689 			err = got_error_from_errno2("opendir", path_packdir);
690 		goto done;
691 	}
692 
693 	while ((dent = readdir(packdir)) != NULL) {
694 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
695 			continue;
696 
697 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
698 		    dent->d_name) == -1) {
699 			err = got_error_from_errno("asprintf");
700 			goto done;
701 		}
702 
703 		err = got_packidx_open(packidx, path_packidx, 0);
704 		free(path_packidx);
705 		if (err)
706 			goto done;
707 
708 		*idx = got_packidx_get_object_idx(*packidx, id);
709 		if (*idx != -1) {
710 			err = NULL; /* found the object */
711 			err = got_repo_cache_packidx(repo, *packidx);
712 			goto done;
713 		}
714 
715 		err = got_packidx_close(*packidx);
716 		*packidx = NULL;
717 		if (err)
718 			goto done;
719 	}
720 
721 	err = got_error_no_obj(id);
722 done:
723 	free(path_packdir);
724 	if (packdir && closedir(packdir) != 0 && err == NULL)
725 		err = got_error_from_errno("closedir");
726 	return err;
727 }
728 
729 static const struct got_error *
730 read_packfile_hdr(int fd, struct got_packidx *packidx)
731 {
732 	const struct got_error *err = NULL;
733 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
734 	struct got_packfile_hdr hdr;
735 	ssize_t n;
736 
737 	n = read(fd, &hdr, sizeof(hdr));
738 	if (n < 0)
739 		return got_error_from_errno("read");
740 	if (n != sizeof(hdr))
741 		return got_error(GOT_ERR_BAD_PACKFILE);
742 
743 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
744 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
745 	    betoh32(hdr.nobjects) != totobj)
746 		err = got_error(GOT_ERR_BAD_PACKFILE);
747 
748 	return err;
749 }
750 
751 static const struct got_error *
752 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
753 {
754 	const struct got_error *err = NULL;
755 
756 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
757 	if (*fd == -1)
758 		return got_error_from_errno2("open", path_packfile);
759 
760 	if (packidx) {
761 		err = read_packfile_hdr(*fd, packidx);
762 		if (err) {
763 			close(*fd);
764 			*fd = -1;
765 		}
766 	}
767 
768 	return err;
769 }
770 
771 const struct got_error *
772 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
773     const char *path_packfile, struct got_packidx *packidx)
774 {
775 	const struct got_error *err = NULL;
776 	struct got_pack *pack = NULL;
777 	struct stat sb;
778 	int i;
779 
780 	if (packp)
781 		*packp = NULL;
782 
783 	for (i = 0; i < nitems(repo->packs); i++) {
784 		pack = &repo->packs[i];
785 		if (pack->path_packfile == NULL)
786 			break;
787 		if (strcmp(pack->path_packfile, path_packfile) == 0)
788 			return NULL;
789 	}
790 
791 	if (i == nitems(repo->packs) - 1) {
792 		err = got_pack_close(&repo->packs[i - 1]);
793 		if (err)
794 			return err;
795 		memmove(&repo->packs[1], &repo->packs[0],
796 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
797 		i = 0;
798 	}
799 
800 	pack = &repo->packs[i];
801 
802 	pack->path_packfile = strdup(path_packfile);
803 	if (pack->path_packfile == NULL) {
804 		err = got_error_from_errno("strdup");
805 		goto done;
806 	}
807 
808 	err = open_packfile(&pack->fd, path_packfile, packidx);
809 	if (err)
810 		goto done;
811 
812 	if (fstat(pack->fd, &sb) != 0) {
813 		err = got_error_from_errno("fstat");
814 		goto done;
815 	}
816 	pack->filesize = sb.st_size;
817 
818 	pack->privsep_child = NULL;
819 
820 #ifndef GOT_PACK_NO_MMAP
821 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
822 	    pack->fd, 0);
823 	if (pack->map == MAP_FAILED) {
824 		if (errno != ENOMEM) {
825 			err = got_error_from_errno("mmap");
826 			goto done;
827 		}
828 		pack->map = NULL; /* fall back to read(2) */
829 	}
830 #endif
831 done:
832 	if (err) {
833 		if (pack) {
834 			free(pack->path_packfile);
835 			memset(pack, 0, sizeof(*pack));
836 		}
837 	} else if (packp)
838 		*packp = pack;
839 	return err;
840 }
841 
842 struct got_pack *
843 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
844 {
845 	struct got_pack *pack = NULL;
846 	int i;
847 
848 	for (i = 0; i < nitems(repo->packs); i++) {
849 		pack = &repo->packs[i];
850 		if (pack->path_packfile == NULL)
851 			break;
852 		if (strcmp(pack->path_packfile, path_packfile) == 0)
853 			return pack;
854 	}
855 
856 	return NULL;
857 }
858 
859 const struct got_error *
860 got_repo_init(const char *repo_path)
861 {
862 	const struct got_error *err = NULL;
863 	const char *dirnames[] = {
864 		GOT_OBJECTS_DIR,
865 		GOT_OBJECTS_PACK_DIR,
866 		GOT_REFS_DIR,
867 	};
868 	const char *description_str = "Unnamed repository; "
869 	    "edit this file 'description' to name the repository.";
870 	const char *headref_str = "ref: refs/heads/master";
871 	const char *gitconfig_str = "[core]\n"
872 	    "\trepositoryformatversion = 0\n"
873 	    "\tfilemode = true\n"
874 	    "\tbare = true\n";
875 	char *path;
876 	int i;
877 
878 	if (!got_path_dir_is_empty(repo_path))
879 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
880 
881 	for (i = 0; i < nitems(dirnames); i++) {
882 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
883 			return got_error_from_errno("asprintf");
884 		}
885 		err = got_path_mkdir(path);
886 		free(path);
887 		if (err)
888 			return err;
889 	}
890 
891 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
892 		return got_error_from_errno("asprintf");
893 	err = got_path_create_file(path, description_str);
894 	free(path);
895 	if (err)
896 		return err;
897 
898 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
899 		return got_error_from_errno("asprintf");
900 	err = got_path_create_file(path, headref_str);
901 	free(path);
902 	if (err)
903 		return err;
904 
905 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
906 		return got_error_from_errno("asprintf");
907 	err = got_path_create_file(path, gitconfig_str);
908 	free(path);
909 	if (err)
910 		return err;
911 
912 	return NULL;
913 }
914 
915 static const struct got_error *
916 match_packed_object(struct got_object_id **unique_id,
917     struct got_repository *repo, const char *id_str_prefix, int obj_type)
918 {
919 	const struct got_error *err = NULL;
920 	char *path_packdir;
921 	DIR *packdir;
922 	struct dirent *dent;
923 	char *path_packidx;
924 	struct got_object_id_queue matched_ids;
925 
926 	SIMPLEQ_INIT(&matched_ids);
927 
928 	path_packdir = got_repo_get_path_objects_pack(repo);
929 	if (path_packdir == NULL)
930 		return got_error_from_errno("got_repo_get_path_objects_pack");
931 
932 	packdir = opendir(path_packdir);
933 	if (packdir == NULL) {
934 		if (errno != ENOENT)
935 			err = got_error_from_errno2("opendir", path_packdir);
936 		goto done;
937 	}
938 
939 	while ((dent = readdir(packdir)) != NULL) {
940 		struct got_packidx *packidx;
941 		struct got_object_qid *qid;
942 
943 
944 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
945 			continue;
946 
947 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
948 		    dent->d_name) == -1) {
949 			err = got_error_from_errno("asprintf");
950 			break;
951 		}
952 
953 		err = got_packidx_open(&packidx, path_packidx, 0);
954 		free(path_packidx);
955 		if (err)
956 			break;
957 
958 		err = got_packidx_match_id_str_prefix(&matched_ids,
959 		    packidx, id_str_prefix);
960 		if (err) {
961 			got_packidx_close(packidx);
962 			break;
963 		}
964 		err = got_packidx_close(packidx);
965 		if (err)
966 			break;
967 
968 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
969 			if (obj_type != GOT_OBJ_TYPE_ANY) {
970 				int matched_type;
971 				err = got_object_get_type(&matched_type, repo,
972 				    qid->id);
973 				if (err)
974 					goto done;
975 				if (matched_type != obj_type)
976 					continue;
977 			}
978 			if (*unique_id == NULL) {
979 				*unique_id = got_object_id_dup(qid->id);
980 				if (*unique_id == NULL) {
981 					err = got_error_from_errno("malloc");
982 					goto done;
983 				}
984 			} else {
985 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
986 				goto done;
987 			}
988 		}
989 	}
990 done:
991 	got_object_id_queue_free(&matched_ids);
992 	free(path_packdir);
993 	if (packdir && closedir(packdir) != 0 && err == NULL)
994 		err = got_error_from_errno("closedir");
995 	if (err) {
996 		free(*unique_id);
997 		*unique_id = NULL;
998 	}
999 	return err;
1000 }
1001 
1002 static const struct got_error *
1003 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1004     const char *object_dir, const char *id_str_prefix, int obj_type,
1005     struct got_repository *repo)
1006 {
1007 	const struct got_error *err = NULL;
1008 	char *path;
1009 	DIR *dir = NULL;
1010 	struct dirent *dent;
1011 	struct got_object_id id;
1012 
1013 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1014 		err = got_error_from_errno("asprintf");
1015 		goto done;
1016 	}
1017 
1018 	dir = opendir(path);
1019 	if (dir == NULL) {
1020 		if (errno == ENOENT) {
1021 			err = NULL;
1022 			goto done;
1023 		}
1024 		err = got_error_from_errno2("opendir", path);
1025 		goto done;
1026 	}
1027 	while ((dent = readdir(dir)) != NULL) {
1028 		char *id_str;
1029 		int cmp;
1030 
1031 		if (strcmp(dent->d_name, ".") == 0 ||
1032 		    strcmp(dent->d_name, "..") == 0)
1033 			continue;
1034 
1035 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1036 			err = got_error_from_errno("asprintf");
1037 			goto done;
1038 		}
1039 
1040 		if (!got_parse_sha1_digest(id.sha1, id_str))
1041 			continue;
1042 
1043 		/*
1044 		 * Directory entries do not necessarily appear in
1045 		 * sorted order, so we must iterate over all of them.
1046 		 */
1047 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1048 		if (cmp != 0) {
1049 			free(id_str);
1050 			continue;
1051 		}
1052 
1053 		if (*unique_id == NULL) {
1054 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1055 				int matched_type;
1056 				err = got_object_get_type(&matched_type, repo,
1057 				    &id);
1058 				if (err)
1059 					goto done;
1060 				if (matched_type != obj_type)
1061 					continue;
1062 			}
1063 			*unique_id = got_object_id_dup(&id);
1064 			if (*unique_id == NULL) {
1065 				err = got_error_from_errno("got_object_id_dup");
1066 				free(id_str);
1067 				goto done;
1068 			}
1069 		} else {
1070 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1071 			free(id_str);
1072 			goto done;
1073 		}
1074 	}
1075 done:
1076 	if (dir && closedir(dir) != 0 && err == NULL)
1077 		err = got_error_from_errno("closedir");
1078 	if (err) {
1079 		free(*unique_id);
1080 		*unique_id = NULL;
1081 	}
1082 	free(path);
1083 	return err;
1084 }
1085 
1086 const struct got_error *
1087 got_repo_match_object_id_prefix(struct got_object_id **id,
1088     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1089 {
1090 	const struct got_error *err = NULL;
1091 	char *path_objects = got_repo_get_path_objects(repo);
1092 	char *object_dir = NULL;
1093 	size_t len;
1094 	int i;
1095 
1096 	*id = NULL;
1097 
1098 	for (i = 0; i < strlen(id_str_prefix); i++) {
1099 		if (isxdigit((unsigned char)id_str_prefix[i]))
1100 			continue;
1101 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
1102 	}
1103 
1104 	len = strlen(id_str_prefix);
1105 	if (len >= 2) {
1106 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1107 		if (err)
1108 			goto done;
1109 		object_dir = strndup(id_str_prefix, 2);
1110 		if (object_dir == NULL) {
1111 			err = got_error_from_errno("strdup");
1112 			goto done;
1113 		}
1114 		err = match_loose_object(id, path_objects, object_dir,
1115 		    id_str_prefix, obj_type, repo);
1116 	} else if (len == 1) {
1117 		int i;
1118 		for (i = 0; i < 0xf; i++) {
1119 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1120 			    == -1) {
1121 				err = got_error_from_errno("asprintf");
1122 				goto done;
1123 			}
1124 			err = match_packed_object(id, repo, object_dir,
1125 			    obj_type);
1126 			if (err)
1127 				goto done;
1128 			err = match_loose_object(id, path_objects, object_dir,
1129 			    id_str_prefix, obj_type, repo);
1130 			if (err)
1131 				goto done;
1132 		}
1133 	} else {
1134 		err = got_error(GOT_ERR_BAD_OBJ_ID_STR);
1135 		goto done;
1136 	}
1137 done:
1138 	free(object_dir);
1139 	if (err) {
1140 		free(*id);
1141 		*id = NULL;
1142 	} else if (*id == NULL)
1143 		err = got_error(GOT_ERR_NO_OBJ);
1144 
1145 	return err;
1146 }
1147 
1148 const struct got_error *
1149 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1150     int obj_type, struct got_repository *repo)
1151 {
1152 	const struct got_error *err;
1153 	struct got_reflist_head refs;
1154 	struct got_reflist_entry *re;
1155 	struct got_object_id *tag_id;
1156 
1157 	SIMPLEQ_INIT(&refs);
1158 	*tag = NULL;
1159 
1160 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1161 	if (err)
1162 		return err;
1163 
1164 	SIMPLEQ_FOREACH(re, &refs, entry) {
1165 		const char *refname;
1166 		refname = got_ref_get_name(re->ref);
1167 		if (got_ref_is_symbolic(re->ref))
1168 			continue;
1169 		refname += strlen("refs/tags/");
1170 		if (strcmp(refname, name) != 0)
1171 			continue;
1172 		err = got_ref_resolve(&tag_id, repo, re->ref);
1173 		if (err)
1174 			break;
1175 		err = got_object_open_as_tag(tag, repo, tag_id);
1176 		free(tag_id);
1177 		if (err)
1178 			break;
1179 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1180 		    got_object_tag_get_object_type(*tag) == obj_type)
1181 			break;
1182 		got_object_tag_close(*tag);
1183 		*tag = NULL;
1184 	}
1185 
1186 	got_ref_list_free(&refs);
1187 	if (err == NULL && *tag == NULL)
1188 		err = got_error(GOT_ERR_NO_OBJ);
1189 	return err;
1190 }
1191 
1192 static const struct got_error *
1193 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1194     const char *name, mode_t mode, struct got_object_id *blob_id)
1195 {
1196 	const struct got_error *err = NULL;
1197 
1198 	 *new_te = NULL;
1199 
1200 	*new_te = calloc(1, sizeof(**new_te));
1201 	if (*new_te == NULL)
1202 		return got_error_from_errno("calloc");
1203 
1204 	(*new_te)->name = strdup(name);
1205 	if ((*new_te)->name == NULL) {
1206 		err = got_error_from_errno("strdup");
1207 		goto done;
1208 	}
1209 
1210 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1211 	(*new_te)->id = blob_id;
1212 done:
1213 	if (err && *new_te) {
1214 		got_object_tree_entry_close(*new_te);
1215 		*new_te = NULL;
1216 	}
1217 	return err;
1218 }
1219 
1220 static const struct got_error *
1221 import_file(struct got_tree_entry **new_te, struct dirent *de,
1222     const char *path, struct got_repository *repo)
1223 {
1224 	const struct got_error *err;
1225 	struct got_object_id *blob_id = NULL;
1226 	char *filepath;
1227 	struct stat sb;
1228 
1229 	if (asprintf(&filepath, "%s%s%s", path,
1230 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1231 		return got_error_from_errno("asprintf");
1232 
1233 	if (lstat(filepath, &sb) != 0) {
1234 		err = got_error_from_errno2("lstat", path);
1235 		goto done;
1236 	}
1237 
1238 	err = got_object_blob_create(&blob_id, filepath, repo);
1239 	if (err)
1240 		goto done;
1241 
1242 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1243 	    blob_id);
1244 done:
1245 	free(filepath);
1246 	if (err)
1247 		free(blob_id);
1248 	return err;
1249 }
1250 
1251 static const struct got_error *
1252 insert_tree_entry(struct got_tree_entry *new_te,
1253     struct got_pathlist_head *paths)
1254 {
1255 	const struct got_error *err = NULL;
1256 	struct got_pathlist_entry *new_pe;
1257 
1258 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1259 	if (err)
1260 		return err;
1261 	if (new_pe == NULL)
1262 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1263 	return NULL;
1264 }
1265 
1266 static const struct got_error *write_tree(struct got_object_id **,
1267     const char *, struct got_pathlist_head *, struct got_repository *,
1268     got_repo_import_cb progress_cb, void *progress_arg);
1269 
1270 static const struct got_error *
1271 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1272     const char *path, struct got_pathlist_head *ignores,
1273     struct got_repository *repo,
1274     got_repo_import_cb progress_cb, void *progress_arg)
1275 {
1276 	const struct got_error *err;
1277 	char *subdirpath;
1278 
1279 	if (asprintf(&subdirpath, "%s%s%s", path,
1280 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1281 		return got_error_from_errno("asprintf");
1282 
1283 	(*new_te) = calloc(1, sizeof(**new_te));
1284 	(*new_te)->mode = S_IFDIR;
1285 	(*new_te)->name = strdup(de->d_name);
1286 	if ((*new_te)->name == NULL) {
1287 		err = got_error_from_errno("strdup");
1288 		goto done;
1289 	}
1290 
1291 	err = write_tree(&(*new_te)->id, subdirpath, ignores,  repo,
1292 	    progress_cb, progress_arg);
1293 done:
1294 	free(subdirpath);
1295 	if (err) {
1296 		got_object_tree_entry_close(*new_te);
1297 		*new_te = NULL;
1298 	}
1299 	return err;
1300 }
1301 
1302 static const struct got_error *
1303 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1304     struct got_pathlist_head *ignores, struct got_repository *repo,
1305     got_repo_import_cb progress_cb, void *progress_arg)
1306 {
1307 	const struct got_error *err = NULL;
1308 	DIR *dir;
1309 	struct dirent *de;
1310 	struct got_tree_entries new_tree_entries;
1311 	struct got_tree_entry *new_te = NULL;
1312 	struct got_pathlist_head paths;
1313 	struct got_pathlist_entry *pe;
1314 
1315 	*new_tree_id = NULL;
1316 
1317 	TAILQ_INIT(&paths);
1318 	new_tree_entries.nentries = 0;
1319 	SIMPLEQ_INIT(&new_tree_entries.head);
1320 
1321 	dir = opendir(path_dir);
1322 	if (dir == NULL) {
1323 		err = got_error_from_errno2("opendir", path_dir);
1324 		goto done;
1325 	}
1326 
1327 	while ((de = readdir(dir)) != NULL) {
1328 		int ignore = 0;
1329 
1330 		if (strcmp(de->d_name, ".") == 0 ||
1331 		    strcmp(de->d_name, "..") == 0)
1332 			continue;
1333 
1334 		TAILQ_FOREACH(pe, ignores, entry) {
1335 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1336 				ignore = 1;
1337 				break;
1338 			}
1339 		}
1340 		if (ignore)
1341 			continue;
1342 		if (de->d_type == DT_DIR) {
1343 			err = import_subdir(&new_te, de, path_dir,
1344 			    ignores, repo, progress_cb, progress_arg);
1345 			if (err)
1346 				goto done;
1347 		} else if (de->d_type == DT_REG) {
1348 			err = import_file(&new_te, de, path_dir, repo);
1349 			if (err)
1350 				goto done;
1351 		} else
1352 			continue;
1353 
1354 		err = insert_tree_entry(new_te, &paths);
1355 		if (err)
1356 			goto done;
1357 	}
1358 
1359 	TAILQ_FOREACH(pe, &paths, entry) {
1360 		struct got_tree_entry *te = pe->data;
1361 		char *path;
1362 		new_tree_entries.nentries++;
1363 		SIMPLEQ_INSERT_TAIL(&new_tree_entries.head, te, entry);
1364 		if (!S_ISREG(te->mode))
1365 			continue;
1366 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1367 			err = got_error_from_errno("asprintf");
1368 			goto done;
1369 		}
1370 		err = (*progress_cb)(progress_arg, path);
1371 		free(path);
1372 		if (err)
1373 			goto done;
1374 	}
1375 
1376 	err = got_object_tree_create(new_tree_id, &new_tree_entries, repo);
1377 done:
1378 	if (dir)
1379 		closedir(dir);
1380 	got_object_tree_entries_close(&new_tree_entries);
1381 	got_pathlist_free(&paths);
1382 	return err;
1383 }
1384 
1385 const struct got_error *
1386 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1387     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1388     struct got_repository *repo, got_repo_import_cb progress_cb,
1389     void *progress_arg)
1390 {
1391 	const struct got_error *err;
1392 	struct got_object_id *new_tree_id;
1393 
1394 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1395 	    progress_cb, progress_arg);
1396 	if (err)
1397 		return err;
1398 
1399 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1400 	    author, time(NULL), author, time(NULL), logmsg, repo);
1401 	free(new_tree_id);
1402 	return err;
1403 }