op public repos

Blob

Date:: Sat Aug 24 20:14:29 2019 UTC
Message:: don't mask "not a git repo" errors in got_repo_open()
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 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
316 		err = got_error_from_errno("asprintf");
317 		goto done;
318 	}
319 	if (is_git_repo(repo)) {
320 		repo->path = strdup(path);
321 		if (repo->path == NULL) {
322 			err = got_error_from_errno("strdup");
323 			goto done;
324 		}
325 		return NULL;
326 	}
327 
328 	err = got_error(GOT_ERR_NOT_GIT_REPO);
329 done:
330 	if (err) {
331 		free(repo->path);
332 		repo->path = NULL;
333 		free(repo->path_git_dir);
334 		repo->path_git_dir = NULL;
335 	}
336 	return err;
337 }
338 
339 const struct got_error *
340 got_repo_open(struct got_repository **repop, const char *path)
341 {
342 	struct got_repository *repo = NULL;
343 	const struct got_error *err = NULL;
344 	char *abspath, *gitconfig_path = NULL;
345 	int i, tried_root = 0;
346 
347 	*repop = NULL;
348 
349 	if (got_path_is_absolute(path))
350 		abspath = strdup(path);
351 	else
352 		abspath = got_path_get_absolute(path);
353 	if (abspath == NULL)
354 		return got_error(GOT_ERR_BAD_PATH);
355 
356 	repo = calloc(1, sizeof(*repo));
357 	if (repo == NULL) {
358 		err = got_error_from_errno("calloc");
359 		goto done;
360 	}
361 
362 	for (i = 0; i < nitems(repo->privsep_children); i++) {
363 		memset(&repo->privsep_children[i], 0,
364 		    sizeof(repo->privsep_children[0]));
365 		repo->privsep_children[i].imsg_fd = -1;
366 	}
367 
368 	err = got_object_cache_init(&repo->objcache,
369 	    GOT_OBJECT_CACHE_TYPE_OBJ);
370 	if (err)
371 		goto done;
372 	err = got_object_cache_init(&repo->treecache,
373 	    GOT_OBJECT_CACHE_TYPE_TREE);
374 	if (err)
375 		goto done;
376 	err = got_object_cache_init(&repo->commitcache,
377 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
378 	if (err)
379 		goto done;
380 	err = got_object_cache_init(&repo->tagcache,
381 	    GOT_OBJECT_CACHE_TYPE_TAG);
382 	if (err)
383 		goto done;
384 
385 	path = realpath(abspath, NULL);
386 	if (path == NULL) {
387 		err = got_error_from_errno2("realpath", abspath);
388 		goto done;
389 	}
390 
391 	do {
392 		err = open_repo(repo, path);
393 		if (err == NULL)
394 			break;
395 		if (err->code != GOT_ERR_NOT_GIT_REPO)
396 			break;
397 		if (path[0] == '/' && path[1] == '\0') {
398 			if (tried_root) {
399 				err = got_error(GOT_ERR_NOT_GIT_REPO);
400 				goto done;
401 			}
402 			tried_root = 1;
403 		}
404 		path = dirname(path);
405 		if (path == NULL) {
406 			err = got_error_from_errno2("dirname", path);
407 			goto done;
408 		}
409 	} while (path);
410 
411 	err = get_path_gitconfig(&gitconfig_path, repo);
412 	if (err)
413 		goto done;
414 
415 #ifdef notyet
416 	err = got_gitconfig_open(&repo->gitconfig, gitconfig_path);
417 	if (err)
418 		goto done;
419 #else
420 	repo->gitconfig = NULL;
421 #endif
422 done:
423 	if (err)
424 		got_repo_close(repo);
425 	else
426 		*repop = repo;
427 	free(abspath);
428 	free(gitconfig_path);
429 	return err;
430 }
431 
432 const struct got_error *
433 got_repo_close(struct got_repository *repo)
434 {
435 	const struct got_error *err = NULL, *child_err;
436 	int i;
437 
438 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
439 		if (repo->packidx_cache[i] == NULL)
440 			break;
441 		got_packidx_close(repo->packidx_cache[i]);
442 	}
443 
444 	for (i = 0; i < nitems(repo->packs); i++) {
445 		if (repo->packs[i].path_packfile == NULL)
446 			break;
447 		got_pack_close(&repo->packs[i]);
448 	}
449 
450 	free(repo->path);
451 	free(repo->path_git_dir);
452 
453 	got_object_cache_close(&repo->objcache);
454 	got_object_cache_close(&repo->treecache);
455 	got_object_cache_close(&repo->commitcache);
456 	got_object_cache_close(&repo->tagcache);
457 
458 	for (i = 0; i < nitems(repo->privsep_children); i++) {
459 		if (repo->privsep_children[i].imsg_fd == -1)
460 			continue;
461 		imsg_clear(repo->privsep_children[i].ibuf);
462 		free(repo->privsep_children[i].ibuf);
463 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
464 		child_err = got_privsep_wait_for_child(
465 		    repo->privsep_children[i].pid);
466 		if (child_err && err == NULL)
467 			err = child_err;
468 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
469 		    err == NULL)
470 			err = got_error_from_errno("close");
471 	}
472 	if (repo->gitconfig)
473 		got_gitconfig_close(repo->gitconfig);
474 	free(repo);
475 
476 	return err;
477 }
478 
479 const struct got_error *
480 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
481     const char *input_path, int check_disk)
482 {
483 	const struct got_error *err = NULL;
484 	const char *repo_abspath = NULL;
485 	size_t repolen, cwdlen, len;
486 	char *cwd, *canonpath, *path = NULL;
487 
488 	*in_repo_path = NULL;
489 
490 	cwd = getcwd(NULL, 0);
491 	if (cwd == NULL)
492 		return got_error_from_errno("getcwd");
493 
494 	canonpath = strdup(input_path);
495 	if (canonpath == NULL) {
496 		err = got_error_from_errno("strdup");
497 		goto done;
498 	}
499 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
500 	if (err)
501 		goto done;
502 
503 	repo_abspath = got_repo_get_path(repo);
504 
505 	if (!check_disk || canonpath[0] == '\0') {
506 		path = strdup(canonpath);
507 		if (path == NULL) {
508 			err = got_error_from_errno("strdup");
509 			goto done;
510 		}
511 	} else {
512 		int is_repo_child = 0, is_cwd_child = 0;
513 
514 		path = realpath(canonpath, NULL);
515 		if (path == NULL) {
516 			if (errno != ENOENT) {
517 				err = got_error_from_errno2("realpath",
518 				    canonpath);
519 				goto done;
520 			}
521 			/*
522 			 * Path is not on disk.
523 			 * Assume it is already relative to repository root.
524 			 */
525 			path = strdup(canonpath);
526 			if (path == NULL) {
527 				err = got_error_from_errno("strdup");
528 				goto done;
529 			}
530 		}
531 
532 		repolen = strlen(repo_abspath);
533 		cwdlen = strlen(cwd);
534 		len = strlen(path);
535 
536 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
537 			is_repo_child = 1;
538 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
539 			is_cwd_child = 1;
540 
541 		if (strcmp(path, repo_abspath) == 0) {
542 			free(path);
543 			path = strdup("");
544 			if (path == NULL) {
545 				err = got_error_from_errno("strdup");
546 				goto done;
547 			}
548 		} else if (is_repo_child && is_cwd_child) {
549 			char *child;
550 			/* Strip common prefix with repository path. */
551 			err = got_path_skip_common_ancestor(&child,
552 			    repo_abspath, path);
553 			if (err)
554 				goto done;
555 			free(path);
556 			path = child;
557 		} else if (is_repo_child) {
558 			/* Matched an on-disk path inside repository. */
559 			if (got_repo_is_bare(repo)) {
560 				/*
561 				 * Matched an on-disk path inside repository
562 				 * database. Treat as repository-relative.
563 				 */
564 			} else {
565 				char *child;
566 				/* Strip common prefix with repository path. */
567 				err = got_path_skip_common_ancestor(&child,
568 				    repo_abspath, path);
569 				if (err)
570 					goto done;
571 				free(path);
572 				path = child;
573 			}
574 		} else if (is_cwd_child) {
575 			char *child;
576 			/* Strip common prefix with cwd. */
577 			err = got_path_skip_common_ancestor(&child, cwd,
578 			    path);
579 			if (err)
580 				goto done;
581 			free(path);
582 			path = child;
583 		} else {
584 			/*
585 			 * Matched unrelated on-disk path.
586 			 * Treat it as repository-relative.
587 			 */
588 		}
589 	}
590 
591 	/* Make in-repository path absolute */
592 	if (path[0] != '/') {
593 		char *abspath;
594 		if (asprintf(&abspath, "/%s", path) == -1) {
595 			err = got_error_from_errno("asprintf");
596 			goto done;
597 		}
598 		free(path);
599 		path = abspath;
600 	}
601 
602 done:
603 	free(cwd);
604 	free(canonpath);
605 	if (err)
606 		free(path);
607 	else
608 		*in_repo_path = path;
609 	return err;
610 }
611 
612 const struct got_error *
613 got_repo_cache_packidx(struct got_repository *repo, struct got_packidx *packidx)
614 {
615 	const struct got_error *err = NULL;
616 	int i;
617 
618 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
619 		if (repo->packidx_cache[i] == NULL)
620 			break;
621 	}
622 	if (i == nitems(repo->packidx_cache)) {
623 		err = got_packidx_close(repo->packidx_cache[i - 1]);
624 		if (err)
625 			return err;
626 	}
627 
628 	/*
629 	 * Insert the new pack index at the front so it will
630 	 * be searched first in the future.
631 	 */
632 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
633 	    sizeof(repo->packidx_cache) -
634 	    sizeof(repo->packidx_cache[0]));
635 	repo->packidx_cache[0] = packidx;
636 
637 	return NULL;
638 }
639 
640 static int
641 is_packidx_filename(const char *name, size_t len)
642 {
643 	if (len != GOT_PACKIDX_NAMELEN)
644 		return 0;
645 
646 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
647 		return 0;
648 
649 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
650 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
651 		return 0;
652 
653 	return 1;
654 }
655 
656 const struct got_error *
657 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
658     struct got_repository *repo, struct got_object_id *id)
659 {
660 	const struct got_error *err;
661 	char *path_packdir;
662 	DIR *packdir;
663 	struct dirent *dent;
664 	char *path_packidx;
665 	int i;
666 
667 	/* Search pack index cache. */
668 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
669 		if (repo->packidx_cache[i] == NULL)
670 			break;
671 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
672 		if (*idx != -1) {
673 			*packidx = repo->packidx_cache[i];
674 			return NULL;
675 		}
676 	}
677 	/* No luck. Search the filesystem. */
678 
679 	path_packdir = got_repo_get_path_objects_pack(repo);
680 	if (path_packdir == NULL)
681 		return got_error_from_errno("got_repo_get_path_objects_pack");
682 
683 	packdir = opendir(path_packdir);
684 	if (packdir == NULL) {
685 		if (errno == ENOENT)
686 			err = got_error_no_obj(id);
687 		else
688 			err = got_error_from_errno2("opendir", path_packdir);
689 		goto done;
690 	}
691 
692 	while ((dent = readdir(packdir)) != NULL) {
693 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
694 			continue;
695 
696 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
697 		    dent->d_name) == -1) {
698 			err = got_error_from_errno("asprintf");
699 			goto done;
700 		}
701 
702 		err = got_packidx_open(packidx, path_packidx, 0);
703 		free(path_packidx);
704 		if (err)
705 			goto done;
706 
707 		*idx = got_packidx_get_object_idx(*packidx, id);
708 		if (*idx != -1) {
709 			err = NULL; /* found the object */
710 			err = got_repo_cache_packidx(repo, *packidx);
711 			goto done;
712 		}
713 
714 		err = got_packidx_close(*packidx);
715 		*packidx = NULL;
716 		if (err)
717 			goto done;
718 	}
719 
720 	err = got_error_no_obj(id);
721 done:
722 	free(path_packdir);
723 	if (packdir && closedir(packdir) != 0 && err == NULL)
724 		err = got_error_from_errno("closedir");
725 	return err;
726 }
727 
728 static const struct got_error *
729 read_packfile_hdr(int fd, struct got_packidx *packidx)
730 {
731 	const struct got_error *err = NULL;
732 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
733 	struct got_packfile_hdr hdr;
734 	ssize_t n;
735 
736 	n = read(fd, &hdr, sizeof(hdr));
737 	if (n < 0)
738 		return got_error_from_errno("read");
739 	if (n != sizeof(hdr))
740 		return got_error(GOT_ERR_BAD_PACKFILE);
741 
742 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
743 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
744 	    betoh32(hdr.nobjects) != totobj)
745 		err = got_error(GOT_ERR_BAD_PACKFILE);
746 
747 	return err;
748 }
749 
750 static const struct got_error *
751 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
752 {
753 	const struct got_error *err = NULL;
754 
755 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
756 	if (*fd == -1)
757 		return got_error_from_errno2("open", path_packfile);
758 
759 	if (packidx) {
760 		err = read_packfile_hdr(*fd, packidx);
761 		if (err) {
762 			close(*fd);
763 			*fd = -1;
764 		}
765 	}
766 
767 	return err;
768 }
769 
770 const struct got_error *
771 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
772     const char *path_packfile, struct got_packidx *packidx)
773 {
774 	const struct got_error *err = NULL;
775 	struct got_pack *pack = NULL;
776 	struct stat sb;
777 	int i;
778 
779 	if (packp)
780 		*packp = NULL;
781 
782 	for (i = 0; i < nitems(repo->packs); i++) {
783 		pack = &repo->packs[i];
784 		if (pack->path_packfile == NULL)
785 			break;
786 		if (strcmp(pack->path_packfile, path_packfile) == 0)
787 			return NULL;
788 	}
789 
790 	if (i == nitems(repo->packs) - 1) {
791 		err = got_pack_close(&repo->packs[i - 1]);
792 		if (err)
793 			return err;
794 		memmove(&repo->packs[1], &repo->packs[0],
795 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
796 		i = 0;
797 	}
798 
799 	pack = &repo->packs[i];
800 
801 	pack->path_packfile = strdup(path_packfile);
802 	if (pack->path_packfile == NULL) {
803 		err = got_error_from_errno("strdup");
804 		goto done;
805 	}
806 
807 	err = open_packfile(&pack->fd, path_packfile, packidx);
808 	if (err)
809 		goto done;
810 
811 	if (fstat(pack->fd, &sb) != 0) {
812 		err = got_error_from_errno("fstat");
813 		goto done;
814 	}
815 	pack->filesize = sb.st_size;
816 
817 	pack->privsep_child = NULL;
818 
819 #ifndef GOT_PACK_NO_MMAP
820 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
821 	    pack->fd, 0);
822 	if (pack->map == MAP_FAILED) {
823 		if (errno != ENOMEM) {
824 			err = got_error_from_errno("mmap");
825 			goto done;
826 		}
827 		pack->map = NULL; /* fall back to read(2) */
828 	}
829 #endif
830 done:
831 	if (err) {
832 		if (pack) {
833 			free(pack->path_packfile);
834 			memset(pack, 0, sizeof(*pack));
835 		}
836 	} else if (packp)
837 		*packp = pack;
838 	return err;
839 }
840 
841 struct got_pack *
842 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
843 {
844 	struct got_pack *pack = NULL;
845 	int i;
846 
847 	for (i = 0; i < nitems(repo->packs); i++) {
848 		pack = &repo->packs[i];
849 		if (pack->path_packfile == NULL)
850 			break;
851 		if (strcmp(pack->path_packfile, path_packfile) == 0)
852 			return pack;
853 	}
854 
855 	return NULL;
856 }
857 
858 const struct got_error *
859 got_repo_init(const char *repo_path)
860 {
861 	const struct got_error *err = NULL;
862 	const char *dirnames[] = {
863 		GOT_OBJECTS_DIR,
864 		GOT_OBJECTS_PACK_DIR,
865 		GOT_REFS_DIR,
866 	};
867 	const char *description_str = "Unnamed repository; "
868 	    "edit this file 'description' to name the repository.";
869 	const char *headref_str = "ref: refs/heads/master";
870 	const char *gitconfig_str = "[core]\n"
871 	    "\trepositoryformatversion = 0\n"
872 	    "\tfilemode = true\n"
873 	    "\tbare = true\n";
874 	char *path;
875 	int i;
876 
877 	if (!got_path_dir_is_empty(repo_path))
878 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
879 
880 	for (i = 0; i < nitems(dirnames); i++) {
881 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
882 			return got_error_from_errno("asprintf");
883 		}
884 		err = got_path_mkdir(path);
885 		free(path);
886 		if (err)
887 			return err;
888 	}
889 
890 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
891 		return got_error_from_errno("asprintf");
892 	err = got_path_create_file(path, description_str);
893 	free(path);
894 	if (err)
895 		return err;
896 
897 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
898 		return got_error_from_errno("asprintf");
899 	err = got_path_create_file(path, headref_str);
900 	free(path);
901 	if (err)
902 		return err;
903 
904 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
905 		return got_error_from_errno("asprintf");
906 	err = got_path_create_file(path, gitconfig_str);
907 	free(path);
908 	if (err)
909 		return err;
910 
911 	return NULL;
912 }
913 
914 static const struct got_error *
915 match_packed_object(struct got_object_id **unique_id,
916     struct got_repository *repo, const char *id_str_prefix, int obj_type)
917 {
918 	const struct got_error *err = NULL;
919 	char *path_packdir;
920 	DIR *packdir;
921 	struct dirent *dent;
922 	char *path_packidx;
923 	struct got_object_id_queue matched_ids;
924 
925 	SIMPLEQ_INIT(&matched_ids);
926 
927 	path_packdir = got_repo_get_path_objects_pack(repo);
928 	if (path_packdir == NULL)
929 		return got_error_from_errno("got_repo_get_path_objects_pack");
930 
931 	packdir = opendir(path_packdir);
932 	if (packdir == NULL) {
933 		if (errno != ENOENT)
934 			err = got_error_from_errno2("opendir", path_packdir);
935 		goto done;
936 	}
937 
938 	while ((dent = readdir(packdir)) != NULL) {
939 		struct got_packidx *packidx;
940 		struct got_object_qid *qid;
941 
942 
943 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
944 			continue;
945 
946 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
947 		    dent->d_name) == -1) {
948 			err = got_error_from_errno("asprintf");
949 			break;
950 		}
951 
952 		err = got_packidx_open(&packidx, path_packidx, 0);
953 		free(path_packidx);
954 		if (err)
955 			break;
956 
957 		err = got_packidx_match_id_str_prefix(&matched_ids,
958 		    packidx, id_str_prefix);
959 		if (err) {
960 			got_packidx_close(packidx);
961 			break;
962 		}
963 		err = got_packidx_close(packidx);
964 		if (err)
965 			break;
966 
967 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
968 			if (obj_type != GOT_OBJ_TYPE_ANY) {
969 				int matched_type;
970 				err = got_object_get_type(&matched_type, repo,
971 				    qid->id);
972 				if (err)
973 					goto done;
974 				if (matched_type != obj_type)
975 					continue;
976 			}
977 			if (*unique_id == NULL) {
978 				*unique_id = got_object_id_dup(qid->id);
979 				if (*unique_id == NULL) {
980 					err = got_error_from_errno("malloc");
981 					goto done;
982 				}
983 			} else {
984 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
985 				goto done;
986 			}
987 		}
988 	}
989 done:
990 	got_object_id_queue_free(&matched_ids);
991 	free(path_packdir);
992 	if (packdir && closedir(packdir) != 0 && err == NULL)
993 		err = got_error_from_errno("closedir");
994 	if (err) {
995 		free(*unique_id);
996 		*unique_id = NULL;
997 	}
998 	return err;
999 }
1000 
1001 static const struct got_error *
1002 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1003     const char *object_dir, const char *id_str_prefix, int obj_type,
1004     struct got_repository *repo)
1005 {
1006 	const struct got_error *err = NULL;
1007 	char *path;
1008 	DIR *dir = NULL;
1009 	struct dirent *dent;
1010 	struct got_object_id id;
1011 
1012 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1013 		err = got_error_from_errno("asprintf");
1014 		goto done;
1015 	}
1016 
1017 	dir = opendir(path);
1018 	if (dir == NULL) {
1019 		if (errno == ENOENT) {
1020 			err = NULL;
1021 			goto done;
1022 		}
1023 		err = got_error_from_errno2("opendir", path);
1024 		goto done;
1025 	}
1026 	while ((dent = readdir(dir)) != NULL) {
1027 		char *id_str;
1028 		int cmp;
1029 
1030 		if (strcmp(dent->d_name, ".") == 0 ||
1031 		    strcmp(dent->d_name, "..") == 0)
1032 			continue;
1033 
1034 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1035 			err = got_error_from_errno("asprintf");
1036 			goto done;
1037 		}
1038 
1039 		if (!got_parse_sha1_digest(id.sha1, id_str))
1040 			continue;
1041 
1042 		/*
1043 		 * Directory entries do not necessarily appear in
1044 		 * sorted order, so we must iterate over all of them.
1045 		 */
1046 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1047 		if (cmp != 0) {
1048 			free(id_str);
1049 			continue;
1050 		}
1051 
1052 		if (*unique_id == NULL) {
1053 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1054 				int matched_type;
1055 				err = got_object_get_type(&matched_type, repo,
1056 				    &id);
1057 				if (err)
1058 					goto done;
1059 				if (matched_type != obj_type)
1060 					continue;
1061 			}
1062 			*unique_id = got_object_id_dup(&id);
1063 			if (*unique_id == NULL) {
1064 				err = got_error_from_errno("got_object_id_dup");
1065 				free(id_str);
1066 				goto done;
1067 			}
1068 		} else {
1069 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1070 			free(id_str);
1071 			goto done;
1072 		}
1073 	}
1074 done:
1075 	if (dir && closedir(dir) != 0 && err == NULL)
1076 		err = got_error_from_errno("closedir");
1077 	if (err) {
1078 		free(*unique_id);
1079 		*unique_id = NULL;
1080 	}
1081 	free(path);
1082 	return err;
1083 }
1084 
1085 const struct got_error *
1086 got_repo_match_object_id_prefix(struct got_object_id **id,
1087     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1088 {
1089 	const struct got_error *err = NULL;
1090 	char *path_objects = got_repo_get_path_objects(repo);
1091 	char *object_dir = NULL;
1092 	size_t len;
1093 	int i;
1094 
1095 	*id = NULL;
1096 
1097 	for (i = 0; i < strlen(id_str_prefix); i++) {
1098 		if (isxdigit((unsigned char)id_str_prefix[i]))
1099 			continue;
1100 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
1101 	}
1102 
1103 	len = strlen(id_str_prefix);
1104 	if (len >= 2) {
1105 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1106 		if (err)
1107 			goto done;
1108 		object_dir = strndup(id_str_prefix, 2);
1109 		if (object_dir == NULL) {
1110 			err = got_error_from_errno("strdup");
1111 			goto done;
1112 		}
1113 		err = match_loose_object(id, path_objects, object_dir,
1114 		    id_str_prefix, obj_type, repo);
1115 	} else if (len == 1) {
1116 		int i;
1117 		for (i = 0; i < 0xf; i++) {
1118 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1119 			    == -1) {
1120 				err = got_error_from_errno("asprintf");
1121 				goto done;
1122 			}
1123 			err = match_packed_object(id, repo, object_dir,
1124 			    obj_type);
1125 			if (err)
1126 				goto done;
1127 			err = match_loose_object(id, path_objects, object_dir,
1128 			    id_str_prefix, obj_type, repo);
1129 			if (err)
1130 				goto done;
1131 		}
1132 	} else {
1133 		err = got_error(GOT_ERR_BAD_OBJ_ID_STR);
1134 		goto done;
1135 	}
1136 done:
1137 	free(object_dir);
1138 	if (err) {
1139 		free(*id);
1140 		*id = NULL;
1141 	} else if (*id == NULL)
1142 		err = got_error(GOT_ERR_NO_OBJ);
1143 
1144 	return err;
1145 }
1146 
1147 const struct got_error *
1148 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1149     int obj_type, struct got_repository *repo)
1150 {
1151 	const struct got_error *err;
1152 	struct got_reflist_head refs;
1153 	struct got_reflist_entry *re;
1154 	struct got_object_id *tag_id;
1155 
1156 	SIMPLEQ_INIT(&refs);
1157 	*tag = NULL;
1158 
1159 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1160 	if (err)
1161 		return err;
1162 
1163 	SIMPLEQ_FOREACH(re, &refs, entry) {
1164 		const char *refname;
1165 		refname = got_ref_get_name(re->ref);
1166 		if (got_ref_is_symbolic(re->ref))
1167 			continue;
1168 		refname += strlen("refs/tags/");
1169 		if (strcmp(refname, name) != 0)
1170 			continue;
1171 		err = got_ref_resolve(&tag_id, repo, re->ref);
1172 		if (err)
1173 			break;
1174 		err = got_object_open_as_tag(tag, repo, tag_id);
1175 		free(tag_id);
1176 		if (err)
1177 			break;
1178 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1179 		    got_object_tag_get_object_type(*tag) == obj_type)
1180 			break;
1181 		got_object_tag_close(*tag);
1182 		*tag = NULL;
1183 	}
1184 
1185 	got_ref_list_free(&refs);
1186 	if (err == NULL && *tag == NULL)
1187 		err = got_error(GOT_ERR_NO_OBJ);
1188 	return err;
1189 }
1190 
1191 static const struct got_error *
1192 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1193     const char *name, mode_t mode, struct got_object_id *blob_id)
1194 {
1195 	const struct got_error *err = NULL;
1196 
1197 	 *new_te = NULL;
1198 
1199 	*new_te = calloc(1, sizeof(**new_te));
1200 	if (*new_te == NULL)
1201 		return got_error_from_errno("calloc");
1202 
1203 	(*new_te)->name = strdup(name);
1204 	if ((*new_te)->name == NULL) {
1205 		err = got_error_from_errno("strdup");
1206 		goto done;
1207 	}
1208 
1209 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1210 	(*new_te)->id = blob_id;
1211 done:
1212 	if (err && *new_te) {
1213 		got_object_tree_entry_close(*new_te);
1214 		*new_te = NULL;
1215 	}
1216 	return err;
1217 }
1218 
1219 static const struct got_error *
1220 import_file(struct got_tree_entry **new_te, struct dirent *de,
1221     const char *path, struct got_repository *repo)
1222 {
1223 	const struct got_error *err;
1224 	struct got_object_id *blob_id = NULL;
1225 	char *filepath;
1226 	struct stat sb;
1227 
1228 	if (asprintf(&filepath, "%s%s%s", path,
1229 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1230 		return got_error_from_errno("asprintf");
1231 
1232 	if (lstat(filepath, &sb) != 0) {
1233 		err = got_error_from_errno2("lstat", path);
1234 		goto done;
1235 	}
1236 
1237 	err = got_object_blob_create(&blob_id, filepath, repo);
1238 	if (err)
1239 		goto done;
1240 
1241 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1242 	    blob_id);
1243 done:
1244 	free(filepath);
1245 	if (err)
1246 		free(blob_id);
1247 	return err;
1248 }
1249 
1250 static const struct got_error *
1251 insert_tree_entry(struct got_tree_entry *new_te,
1252     struct got_pathlist_head *paths)
1253 {
1254 	const struct got_error *err = NULL;
1255 	struct got_pathlist_entry *new_pe;
1256 
1257 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1258 	if (err)
1259 		return err;
1260 	if (new_pe == NULL)
1261 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1262 	return NULL;
1263 }
1264 
1265 static const struct got_error *write_tree(struct got_object_id **,
1266     const char *, struct got_pathlist_head *, struct got_repository *,
1267     got_repo_import_cb progress_cb, void *progress_arg);
1268 
1269 static const struct got_error *
1270 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1271     const char *path, struct got_pathlist_head *ignores,
1272     struct got_repository *repo,
1273     got_repo_import_cb progress_cb, void *progress_arg)
1274 {
1275 	const struct got_error *err;
1276 	char *subdirpath;
1277 
1278 	if (asprintf(&subdirpath, "%s%s%s", path,
1279 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1280 		return got_error_from_errno("asprintf");
1281 
1282 	(*new_te) = calloc(1, sizeof(**new_te));
1283 	(*new_te)->mode = S_IFDIR;
1284 	(*new_te)->name = strdup(de->d_name);
1285 	if ((*new_te)->name == NULL) {
1286 		err = got_error_from_errno("strdup");
1287 		goto done;
1288 	}
1289 
1290 	err = write_tree(&(*new_te)->id, subdirpath, ignores,  repo,
1291 	    progress_cb, progress_arg);
1292 done:
1293 	free(subdirpath);
1294 	if (err) {
1295 		got_object_tree_entry_close(*new_te);
1296 		*new_te = NULL;
1297 	}
1298 	return err;
1299 }
1300 
1301 static const struct got_error *
1302 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1303     struct got_pathlist_head *ignores, struct got_repository *repo,
1304     got_repo_import_cb progress_cb, void *progress_arg)
1305 {
1306 	const struct got_error *err = NULL;
1307 	DIR *dir;
1308 	struct dirent *de;
1309 	struct got_tree_entries new_tree_entries;
1310 	struct got_tree_entry *new_te = NULL;
1311 	struct got_pathlist_head paths;
1312 	struct got_pathlist_entry *pe;
1313 
1314 	*new_tree_id = NULL;
1315 
1316 	TAILQ_INIT(&paths);
1317 	new_tree_entries.nentries = 0;
1318 	SIMPLEQ_INIT(&new_tree_entries.head);
1319 
1320 	dir = opendir(path_dir);
1321 	if (dir == NULL) {
1322 		err = got_error_from_errno2("opendir", path_dir);
1323 		goto done;
1324 	}
1325 
1326 	while ((de = readdir(dir)) != NULL) {
1327 		int ignore = 0;
1328 
1329 		if (strcmp(de->d_name, ".") == 0 ||
1330 		    strcmp(de->d_name, "..") == 0)
1331 			continue;
1332 
1333 		TAILQ_FOREACH(pe, ignores, entry) {
1334 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1335 				ignore = 1;
1336 				break;
1337 			}
1338 		}
1339 		if (ignore)
1340 			continue;
1341 		if (de->d_type == DT_DIR) {
1342 			err = import_subdir(&new_te, de, path_dir,
1343 			    ignores, repo, progress_cb, progress_arg);
1344 			if (err)
1345 				goto done;
1346 		} else if (de->d_type == DT_REG) {
1347 			err = import_file(&new_te, de, path_dir, repo);
1348 			if (err)
1349 				goto done;
1350 		} else
1351 			continue;
1352 
1353 		err = insert_tree_entry(new_te, &paths);
1354 		if (err)
1355 			goto done;
1356 	}
1357 
1358 	TAILQ_FOREACH(pe, &paths, entry) {
1359 		struct got_tree_entry *te = pe->data;
1360 		char *path;
1361 		new_tree_entries.nentries++;
1362 		SIMPLEQ_INSERT_TAIL(&new_tree_entries.head, te, entry);
1363 		if (!S_ISREG(te->mode))
1364 			continue;
1365 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1366 			err = got_error_from_errno("asprintf");
1367 			goto done;
1368 		}
1369 		err = (*progress_cb)(progress_arg, path);
1370 		free(path);
1371 		if (err)
1372 			goto done;
1373 	}
1374 
1375 	err = got_object_tree_create(new_tree_id, &new_tree_entries, repo);
1376 done:
1377 	if (dir)
1378 		closedir(dir);
1379 	got_object_tree_entries_close(&new_tree_entries);
1380 	got_pathlist_free(&paths);
1381 	return err;
1382 }
1383 
1384 const struct got_error *
1385 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1386     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1387     struct got_repository *repo, got_repo_import_cb progress_cb,
1388     void *progress_arg)
1389 {
1390 	const struct got_error *err;
1391 	struct got_object_id *new_tree_id;
1392 
1393 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1394 	    progress_cb, progress_arg);
1395 	if (err)
1396 		return err;
1397 
1398 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1399 	    author, time(NULL), author, time(NULL), logmsg, repo);
1400 	free(new_tree_id);
1401 	return err;
1402 }