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