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/stat.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/mman.h>
23 
24 #include <errno.h>
25 #include <fcntl.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <sha1.h>
30 #include <sha2.h>
31 #include <unistd.h>
32 #include <zlib.h>
33 #include <libgen.h>
34 #include <limits.h>
35 #include <imsg.h>
36 
37 #include "got_error.h"
38 #include "got_object.h"
39 #include "got_repository.h"
40 #include "got_opentemp.h"
41 #include "got_path.h"
42 
43 #include "got_lib_hash.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_idcache.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_repository.h"
52 
53 #ifndef MIN
54 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
55 #endif
56 
57 #ifndef nitems
58 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
59 #endif
60 
61 struct got_object_id *
62 got_object_get_id(struct got_object *obj)
63 {
64 	return &obj->id;
65 }
66 
67 const struct got_error *
68 got_object_get_id_str(char **outbuf, struct got_object *obj)
69 {
70 	return got_object_id_str(outbuf, &obj->id);
71 }
72 
73 const struct got_error *
74 got_object_get_type(int *type, struct got_repository *repo,
75     struct got_object_id *id)
76 {
77 	const struct got_error *err = NULL;
78 	struct got_object *obj;
79 
80 	err = got_object_open(&obj, repo, id);
81 	if (err)
82 		return err;
83 
84 	switch (obj->type) {
85 	case GOT_OBJ_TYPE_COMMIT:
86 	case GOT_OBJ_TYPE_TREE:
87 	case GOT_OBJ_TYPE_BLOB:
88 	case GOT_OBJ_TYPE_TAG:
89 		*type = obj->type;
90 		break;
91 	default:
92 		err = got_error(GOT_ERR_OBJ_TYPE);
93 		break;
94 	}
95 
96 	got_object_close(obj);
97 	return err;
98 }
99 
100 const struct got_error *
101 got_object_get_path(char **path, struct got_object_id *id,
102     struct got_repository *repo)
103 {
104 	const struct got_error *err = NULL;
105 	char *hex = NULL;
106 	char *path_objects;
107 
108 	*path = NULL;
109 
110 	path_objects = got_repo_get_path_objects(repo);
111 	if (path_objects == NULL)
112 		return got_error_from_errno("got_repo_get_path_objects");
113 
114 	err = got_object_id_str(&hex, id);
115 	if (err)
116 		goto done;
117 
118 	if (asprintf(path, "%s/%.2x/%s", path_objects,
119 	    id->sha1[0], hex + 2) == -1)
120 		err = got_error_from_errno("asprintf");
121 
122 done:
123 	free(hex);
124 	free(path_objects);
125 	return err;
126 }
127 
128 const struct got_error *
129 got_object_open_loose_fd(int *fd, struct got_object_id *id,
130     struct got_repository *repo)
131 {
132 	const struct got_error *err = NULL;
133 	char *path;
134 
135 	err = got_object_get_path(&path, id, repo);
136 	if (err)
137 		return err;
138 	*fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
139 	if (*fd == -1) {
140 		err = got_error_from_errno2("open", path);
141 		goto done;
142 	}
143 done:
144 	free(path);
145 	return err;
146 }
147 
148 const struct got_error *
149 got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
150     const char *id_str)
151 {
152 	struct got_object_id id;
153 
154 	if (!got_parse_object_id(&id, id_str, GOT_HASH_SHA1))
155 		return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
156 
157 	return got_object_open(obj, repo, &id);
158 }
159 
160 const struct got_error *
161 got_object_resolve_id_str(struct got_object_id **id,
162     struct got_repository *repo, const char *id_str)
163 {
164 	const struct got_error *err = NULL;
165 	struct got_object *obj;
166 
167 	err = got_object_open_by_id_str(&obj, repo, id_str);
168 	if (err)
169 		return err;
170 
171 	*id = got_object_id_dup(got_object_get_id(obj));
172 	got_object_close(obj);
173 	if (*id == NULL)
174 		return got_error_from_errno("got_object_id_dup");
175 
176 	return NULL;
177 }
178 
179 const struct got_error *
180 got_object_qid_alloc(struct got_object_qid **qid, struct got_object_id *id)
181 {
182 	*qid = calloc(1, sizeof(**qid));
183 	if (*qid == NULL)
184 		return got_error_from_errno("calloc");
185 
186 	memcpy(&(*qid)->id, id, sizeof((*qid)->id));
187 	return NULL;
188 }
189 
190 const struct got_error *
191 got_object_id_queue_copy(const struct got_object_id_queue *src,
192     struct got_object_id_queue *dest)
193 {
194 	const struct got_error *err;
195 	struct got_object_qid *qid;
196 
197 	STAILQ_FOREACH(qid, src, entry) {
198 		struct got_object_qid *new;
199 		/*
200 		 * Deep-copy the object ID only. Let the caller deal
201 		 * with setting up the new->data pointer if needed.
202 		 */
203 		err = got_object_qid_alloc(&new, &qid->id);
204 		if (err) {
205 			got_object_id_queue_free(dest);
206 			return err;
207 		}
208 		STAILQ_INSERT_TAIL(dest, new, entry);
209 	}
210 
211 	return NULL;
212 }
213 
214 int
215 got_object_tree_get_nentries(struct got_tree_object *tree)
216 {
217 	return tree->nentries;
218 }
219 
220 struct got_tree_entry *
221 got_object_tree_get_first_entry(struct got_tree_object *tree)
222 {
223 	return got_object_tree_get_entry(tree, 0);
224 }
225 
226 struct got_tree_entry *
227 got_object_tree_get_last_entry(struct got_tree_object *tree)
228 {
229 	return got_object_tree_get_entry(tree, tree->nentries - 1);
230 }
231 
232 struct got_tree_entry *
233 got_object_tree_get_entry(struct got_tree_object *tree, int i)
234 {
235 	if (i < 0 || i >= tree->nentries)
236 		return NULL;
237 	return &tree->entries[i];
238 }
239 
240 mode_t
241 got_tree_entry_get_mode(struct got_tree_entry *te)
242 {
243 	return te->mode;
244 }
245 
246 const char *
247 got_tree_entry_get_name(struct got_tree_entry *te)
248 {
249 	return &te->name[0];
250 }
251 
252 struct got_object_id *
253 got_tree_entry_get_id(struct got_tree_entry *te)
254 {
255 	return &te->id;
256 }
257 
258 const struct got_error *
259 got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
260 {
261 	const struct got_error *err = NULL;
262 	size_t len, totlen, hdrlen, offset;
263 
264 	*s = NULL;
265 
266 	hdrlen = got_object_blob_get_hdrlen(blob);
267 	totlen = 0;
268 	offset = 0;
269 	do {
270 		char *p;
271 
272 		err = got_object_blob_read_block(&len, blob);
273 		if (err)
274 			return err;
275 
276 		if (len == 0)
277 			break;
278 
279 		totlen += len - hdrlen;
280 		p = realloc(*s, totlen + 1);
281 		if (p == NULL) {
282 			err = got_error_from_errno("realloc");
283 			free(*s);
284 			*s = NULL;
285 			return err;
286 		}
287 		*s = p;
288 		/* Skip blob object header first time around. */
289 		memcpy(*s + offset,
290 		    got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
291 		hdrlen = 0;
292 		offset = totlen;
293 	} while (len > 0);
294 
295 	(*s)[totlen] = '\0';
296 	return NULL;
297 }
298 
299 const struct got_error *
300 got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
301     struct got_repository *repo)
302 {
303 	const struct got_error *err = NULL;
304 	struct got_blob_object *blob = NULL;
305 	int fd = -1;
306 
307 	*link_target = NULL;
308 
309 	if (!got_object_tree_entry_is_symlink(te))
310 		return got_error(GOT_ERR_TREE_ENTRY_TYPE);
311 
312 	fd = got_opentempfd();
313 	if (fd == -1) {
314 		err = got_error_from_errno("got_opentempfd");
315 		goto done;
316 	}
317 
318 	err = got_object_open_as_blob(&blob, repo,
319 	    got_tree_entry_get_id(te), PATH_MAX, fd);
320 	if (err)
321 		goto done;
322 
323 	err = got_object_blob_read_to_str(link_target, blob);
324 done:
325 	if (fd != -1 && close(fd) == -1 && err == NULL)
326 		err = got_error_from_errno("close");
327 	if (blob)
328 		got_object_blob_close(blob);
329 	if (err) {
330 		free(*link_target);
331 		*link_target = NULL;
332 	}
333 	return err;
334 }
335 
336 int
337 got_tree_entry_get_index(struct got_tree_entry *te)
338 {
339 	return te->idx;
340 }
341 
342 struct got_tree_entry *
343 got_tree_entry_get_next(struct got_tree_object *tree,
344     struct got_tree_entry *te)
345 {
346 	return got_object_tree_get_entry(tree, te->idx + 1);
347 }
348 
349 struct got_tree_entry *
350 got_tree_entry_get_prev(struct got_tree_object *tree,
351     struct got_tree_entry *te)
352 {
353 	return got_object_tree_get_entry(tree, te->idx - 1);
354 }
355 
356 const struct got_error *
357 got_object_blob_close(struct got_blob_object *blob)
358 {
359 	const struct got_error *err = NULL;
360 	free(blob->read_buf);
361 	if (blob->f && fclose(blob->f) == EOF)
362 		err = got_error_from_errno("fclose");
363 	free(blob->data);
364 	free(blob);
365 	return err;
366 }
367 
368 void
369 got_object_blob_rewind(struct got_blob_object *blob)
370 {
371 	if (blob->f)
372 		rewind(blob->f);
373 }
374 
375 char *
376 got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
377 {
378 	return got_object_id_hex(&blob->id, buf, size);
379 }
380 
381 size_t
382 got_object_blob_get_hdrlen(struct got_blob_object *blob)
383 {
384 	return blob->hdrlen;
385 }
386 
387 const uint8_t *
388 got_object_blob_get_read_buf(struct got_blob_object *blob)
389 {
390 	return blob->read_buf;
391 }
392 
393 const struct got_error *
394 got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
395 {
396 	size_t n;
397 
398 	n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
399 	if (n == 0 && ferror(blob->f))
400 		return got_ferror(blob->f, GOT_ERR_IO);
401 	*outlenp = n;
402 	return NULL;
403 }
404 
405 const struct got_error *
406 got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
407 {
408 	const struct got_error *err;
409 	size_t hdrlen, len;
410 
411 	*binary = 0;
412 	hdrlen = got_object_blob_get_hdrlen(blob);
413 
414 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
415 		return got_error_from_errno("fseeko");
416 
417 	err = got_object_blob_read_block(&len, blob);
418 	if (err)
419 		return err;
420 
421 	*binary = memchr(blob->read_buf, '\0', len) != NULL;
422 
423 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
424 		return got_error_from_errno("fseeko");
425 	return NULL;
426 }
427 
428 const struct got_error *
429 got_object_blob_getline(char **line, ssize_t *linelen, size_t *linesize,
430     struct got_blob_object *blob)
431 {
432 	*linelen = getline(line, linesize, blob->f);
433 	if (*linelen == -1 && !feof(blob->f))
434 		return got_error_from_errno("getline");
435 	return NULL;
436 }
437 
438 const struct got_error *
439 got_object_blob_dump_to_file(off_t *filesize, int *nlines,
440     off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
441 {
442 	const struct got_error *err = NULL;
443 	size_t n, len, hdrlen;
444 	const uint8_t *buf;
445 	int i;
446 	const int alloc_chunksz = 512;
447 	size_t nalloc = 0;
448 	off_t off = 0, total_len = 0;
449 
450 	if (line_offsets)
451 		*line_offsets = NULL;
452 	if (filesize)
453 		*filesize = 0;
454 	if (nlines)
455 		*nlines = 0;
456 
457 	hdrlen = got_object_blob_get_hdrlen(blob);
458 	do {
459 		err = got_object_blob_read_block(&len, blob);
460 		if (err)
461 			return err;
462 		if (len == 0)
463 			break;
464 		buf = got_object_blob_get_read_buf(blob);
465 		i = hdrlen;
466 		if (nlines) {
467 			if (line_offsets && *line_offsets == NULL) {
468 				/* Have some data but perhaps no '\n'. */
469 				*nlines = 1;
470 				nalloc = alloc_chunksz;
471 				*line_offsets = calloc(nalloc,
472 				    sizeof(**line_offsets));
473 				if (*line_offsets == NULL)
474 					return got_error_from_errno("calloc");
475 
476 				/* Skip forward over end of first line. */
477 				while (i < len) {
478 					if (buf[i] == '\n')
479 						break;
480 					i++;
481 				}
482 			}
483 			/* Scan '\n' offsets in remaining chunk of data. */
484 			while (i < len) {
485 				if (buf[i] != '\n') {
486 					i++;
487 					continue;
488 				}
489 				(*nlines)++;
490 				if (line_offsets && nalloc < *nlines) {
491 					size_t n = *nlines + alloc_chunksz;
492 					off_t *o = recallocarray(*line_offsets,
493 					    nalloc, n, sizeof(**line_offsets));
494 					if (o == NULL) {
495 						free(*line_offsets);
496 						*line_offsets = NULL;
497 						return got_error_from_errno(
498 						    "recallocarray");
499 					}
500 					*line_offsets = o;
501 					nalloc = n;
502 				}
503 				if (line_offsets) {
504 					off = total_len + i - hdrlen + 1;
505 					(*line_offsets)[*nlines - 1] = off;
506 				}
507 				i++;
508 			}
509 		}
510 		/* Skip blob object header first time around. */
511 		n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
512 		if (n != len - hdrlen)
513 			return got_ferror(outfile, GOT_ERR_IO);
514 		total_len += len - hdrlen;
515 		hdrlen = 0;
516 	} while (len != 0);
517 
518 	if (fflush(outfile) != 0)
519 		return got_error_from_errno("fflush");
520 	rewind(outfile);
521 
522 	if (filesize)
523 		*filesize = total_len;
524 
525 	return NULL;
526 }
527 
528 const char *
529 got_object_tag_get_name(struct got_tag_object *tag)
530 {
531 	return tag->tag;
532 }
533 
534 int
535 got_object_tag_get_object_type(struct got_tag_object *tag)
536 {
537 	return tag->obj_type;
538 }
539 
540 struct got_object_id *
541 got_object_tag_get_object_id(struct got_tag_object *tag)
542 {
543 	return &tag->id;
544 }
545 
546 time_t
547 got_object_tag_get_tagger_time(struct got_tag_object *tag)
548 {
549 	return tag->tagger_time;
550 }
551 
552 time_t
553 got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
554 {
555 	return tag->tagger_gmtoff;
556 }
557 
558 const char *
559 got_object_tag_get_tagger(struct got_tag_object *tag)
560 {
561 	return tag->tagger;
562 }
563 
564 const char *
565 got_object_tag_get_message(struct got_tag_object *tag)
566 {
567 	return tag->tagmsg;
568 }
569 
570 static struct got_tree_entry *
571 find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
572 {
573 	int i;
574 
575 	/* Note that tree entries are sorted in strncmp() order. */
576 	for (i = 0; i < tree->nentries; i++) {
577 		struct got_tree_entry *te = &tree->entries[i];
578 		int cmp = strncmp(te->name, name, len);
579 		if (cmp < 0)
580 			continue;
581 		if (cmp > 0)
582 			break;
583 		if (te->name[len] == '\0')
584 			return te;
585 	}
586 	return NULL;
587 }
588 
589 struct got_tree_entry *
590 got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
591 {
592 	return find_entry_by_name(tree, name, strlen(name));
593 }
594 
595 const struct got_error *
596 got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
597     struct got_repository *repo, struct got_tree_object *tree,
598     const char *path)
599 {
600 	const struct got_error *err = NULL;
601 	struct got_tree_object *subtree = NULL;
602 	struct got_tree_entry *te = NULL;
603 	const char *seg, *s;
604 	size_t seglen;
605 
606 	*id = NULL;
607 
608 	s = path;
609 	while (s[0] == '/')
610 		s++;
611 	seg = s;
612 	seglen = 0;
613 	subtree = tree;
614 	while (*s) {
615 		struct got_tree_object *next_tree;
616 
617 		if (*s != '/') {
618 			s++;
619 			seglen++;
620 			if (*s)
621 				continue;
622 		}
623 
624 		te = find_entry_by_name(subtree, seg, seglen);
625 		if (te == NULL) {
626 			err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
627 			goto done;
628 		}
629 
630 		if (*s == '\0')
631 			break;
632 
633 		seg = s + 1;
634 		seglen = 0;
635 		s++;
636 		if (*s) {
637 			err = got_object_open_as_tree(&next_tree, repo,
638 			    &te->id);
639 			te = NULL;
640 			if (err)
641 				goto done;
642 			if (subtree != tree)
643 				got_object_tree_close(subtree);
644 			subtree = next_tree;
645 		}
646 	}
647 
648 	if (te) {
649 		*id = got_object_id_dup(&te->id);
650 		if (*id == NULL)
651 			return got_error_from_errno("got_object_id_dup");
652 		if (mode)
653 			*mode = te->mode;
654 	} else
655 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
656 done:
657 	if (subtree && subtree != tree)
658 		got_object_tree_close(subtree);
659 	return err;
660 }
661 
662 const struct got_error *
663 got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
664     struct got_commit_object *commit, const char *path)
665 {
666 	const struct got_error *err = NULL;
667 	struct got_tree_object *tree = NULL;
668 
669 	*id = NULL;
670 
671 	/* Handle opening of root of commit's tree. */
672 	if (got_path_is_root_dir(path)) {
673 		*id = got_object_id_dup(commit->tree_id);
674 		if (*id == NULL)
675 			err = got_error_from_errno("got_object_id_dup");
676 	} else {
677 		err = got_object_open_as_tree(&tree, repo, commit->tree_id);
678 		if (err)
679 			goto done;
680 		err = got_object_tree_find_path(id, NULL, repo, tree, path);
681 	}
682 done:
683 	if (tree)
684 		got_object_tree_close(tree);
685 	return err;
686 }
687 
688 /*
689  * Normalize file mode bits to avoid false positive tree entry differences
690  * in case tree entries have unexpected mode bits set.
691  */
692 static mode_t
693 normalize_mode_for_comparison(mode_t mode)
694 {
695 	/*
696 	 * For directories, the only relevant bit is the IFDIR bit.
697 	 * This allows us to detect paths changing from a directory
698 	 * to a file and vice versa.
699 	 */
700 	if (S_ISDIR(mode))
701 		return mode & S_IFDIR;
702 
703 	/*
704 	 * For symlinks, the only relevant bit is the IFLNK bit.
705 	 * This allows us to detect paths changing from a symlinks
706 	 * to a file or directory and vice versa.
707 	 */
708 	if (S_ISLNK(mode))
709 		return mode & S_IFLNK;
710 
711 	/* For files, the only change we care about is the executable bit. */
712 	return mode & S_IXUSR;
713 }
714 
715 const struct got_error *
716 got_object_tree_path_changed(int *changed,
717     struct got_tree_object *tree01, struct got_tree_object *tree02,
718     const char *path, struct got_repository *repo)
719 {
720 	const struct got_error *err = NULL;
721 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
722 	struct got_tree_entry *te1 = NULL, *te2 = NULL;
723 	const char *seg, *s;
724 	size_t seglen;
725 
726 	*changed = 0;
727 
728 	/* We not do support comparing the root path. */
729 	if (got_path_is_root_dir(path))
730 		return got_error_path(path, GOT_ERR_BAD_PATH);
731 
732 	tree1 = tree01;
733 	tree2 = tree02;
734 	s = path;
735 	while (*s == '/')
736 		s++;
737 	seg = s;
738 	seglen = 0;
739 	while (*s) {
740 		struct got_tree_object *next_tree1, *next_tree2;
741 		mode_t mode1, mode2;
742 
743 		if (*s != '/') {
744 			s++;
745 			seglen++;
746 			if (*s)
747 				continue;
748 		}
749 
750 		te1 = find_entry_by_name(tree1, seg, seglen);
751 		if (te1 == NULL) {
752 			err = got_error(GOT_ERR_NO_OBJ);
753 			goto done;
754 		}
755 
756 		if (tree2)
757 			te2 = find_entry_by_name(tree2, seg, seglen);
758 
759 		if (te2) {
760 			mode1 = normalize_mode_for_comparison(te1->mode);
761 			mode2 = normalize_mode_for_comparison(te2->mode);
762 			if (mode1 != mode2) {
763 				*changed = 1;
764 				goto done;
765 			}
766 
767 			if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
768 				*changed = 0;
769 				goto done;
770 			}
771 		}
772 
773 		if (*s == '\0') { /* final path element */
774 			*changed = 1;
775 			goto done;
776 		}
777 
778 		seg = s + 1;
779 		s++;
780 		seglen = 0;
781 		if (*s) {
782 			err = got_object_open_as_tree(&next_tree1, repo,
783 			    &te1->id);
784 			te1 = NULL;
785 			if (err)
786 				goto done;
787 			if (tree1 != tree01)
788 				got_object_tree_close(tree1);
789 			tree1 = next_tree1;
790 
791 			if (te2) {
792 				err = got_object_open_as_tree(&next_tree2, repo,
793 				    &te2->id);
794 				te2 = NULL;
795 				if (err)
796 					goto done;
797 				if (tree2 != tree02)
798 					got_object_tree_close(tree2);
799 				tree2 = next_tree2;
800 			} else if (tree2) {
801 				if (tree2 != tree02)
802 					got_object_tree_close(tree2);
803 				tree2 = NULL;
804 			}
805 		}
806 	}
807 done:
808 	if (tree1 && tree1 != tree01)
809 		got_object_tree_close(tree1);
810 	if (tree2 && tree2 != tree02)
811 		got_object_tree_close(tree2);
812 	return err;
813 }
814 
815 const struct got_error *
816 got_object_tree_entry_dup(struct got_tree_entry **new_te,
817     struct got_tree_entry *te)
818 {
819 	const struct got_error *err = NULL;
820 
821 	*new_te = calloc(1, sizeof(**new_te));
822 	if (*new_te == NULL)
823 		return got_error_from_errno("calloc");
824 
825 	(*new_te)->mode = te->mode;
826 	memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
827 	memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
828 	return err;
829 }
830 
831 int
832 got_object_tree_entry_is_submodule(struct got_tree_entry *te)
833 {
834 	return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
835 }
836 
837 int
838 got_object_tree_entry_is_symlink(struct got_tree_entry *te)
839 {
840 	/* S_IFDIR check avoids confusing symlinks with submodules. */
841 	return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
842 }
843 
844 static const struct got_error *
845 resolve_symlink(char **link_target, const char *path,
846     struct got_commit_object *commit, struct got_repository *repo)
847 {
848 	const struct got_error *err = NULL;
849 	char buf[PATH_MAX];
850 	char *name, *parent_path = NULL;
851 	struct got_object_id *tree_obj_id = NULL;
852 	struct got_tree_object *tree = NULL;
853 	struct got_tree_entry *te = NULL;
854 
855 	*link_target = NULL;
856 
857 	if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
858 		return got_error(GOT_ERR_NO_SPACE);
859 
860 	name = basename(buf);
861 	if (name == NULL)
862 		return got_error_from_errno2("basename", path);
863 
864 	err = got_path_dirname(&parent_path, path);
865 	if (err)
866 		return err;
867 
868 	err = got_object_id_by_path(&tree_obj_id, repo, commit,
869 	    parent_path);
870 	if (err) {
871 		if (err->code == GOT_ERR_NO_TREE_ENTRY) {
872 			/* Display the complete path in error message. */
873 			err = got_error_path(path, err->code);
874 		}
875 		goto done;
876 	}
877 
878 	err = got_object_open_as_tree(&tree, repo, tree_obj_id);
879 	if (err)
880 		goto done;
881 
882 	te = got_object_tree_find_entry(tree, name);
883 	if (te == NULL) {
884 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
885 		goto done;
886 	}
887 
888 	if (got_object_tree_entry_is_symlink(te)) {
889 		err = got_tree_entry_get_symlink_target(link_target, te, repo);
890 		if (err)
891 			goto done;
892 		if (!got_path_is_absolute(*link_target)) {
893 			char *abspath;
894 			if (asprintf(&abspath, "%s/%s", parent_path,
895 			    *link_target) == -1) {
896 				err = got_error_from_errno("asprintf");
897 				goto done;
898 			}
899 			free(*link_target);
900 			*link_target = malloc(PATH_MAX);
901 			if (*link_target == NULL) {
902 				err = got_error_from_errno("malloc");
903 				goto done;
904 			}
905 			err = got_canonpath(abspath, *link_target, PATH_MAX);
906 			free(abspath);
907 			if (err)
908 				goto done;
909 		}
910 	}
911 done:
912 	free(parent_path);
913 	free(tree_obj_id);
914 	if (tree)
915 		got_object_tree_close(tree);
916 	if (err) {
917 		free(*link_target);
918 		*link_target = NULL;
919 	}
920 	return err;
921 }
922 
923 const struct got_error *
924 got_object_resolve_symlinks(char **link_target, const char *path,
925     struct got_commit_object *commit, struct got_repository *repo)
926 {
927 	const struct got_error *err = NULL;
928 	char *next_target = NULL;
929 	int max_recursion = 40; /* matches Git */
930 
931 	*link_target = NULL;
932 
933 	do {
934 		err = resolve_symlink(&next_target,
935 		    *link_target ? *link_target : path, commit, repo);
936 		if (err)
937 			break;
938 		if (next_target) {
939 			free(*link_target);
940 			if (--max_recursion == 0) {
941 				err = got_error_path(path, GOT_ERR_RECURSION);
942 				*link_target = NULL;
943 				break;
944 			}
945 			*link_target = next_target;
946 		}
947 	} while (next_target);
948 
949 	return err;
950 }
951 
952 void
953 got_object_commit_retain(struct got_commit_object *commit)
954 {
955 	commit->refcnt++;
956 }
957 
958 const struct got_error *
959 got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
960     size_t max_in_mem_size, size_t hdrlen, off_t size)
961 {
962 	const struct got_error *err = NULL;
963 	off_t tot;
964 
965 	tot = hdrlen + size;
966 
967 	*obj = calloc(1, sizeof(**obj));
968 	if (*obj == NULL) {
969 		err = got_error_from_errno("calloc");
970 		goto done;
971 	}
972 	(*obj)->fd = -1;
973 	(*obj)->tempfile_idx = -1;
974 
975 	if (outbuf) {
976 		(*obj)->data = outbuf;
977 	} else {
978 		struct stat sb;
979 		if (fstat(*outfd, &sb) == -1) {
980 			err = got_error_from_errno("fstat");
981 			goto done;
982 		}
983 
984 		if (sb.st_size != tot) {
985 			err = got_error_msg(GOT_ERR_BAD_OBJ_HDR,
986 			    "raw object has unexpected size");
987 			goto done;
988 		}
989 #ifndef GOT_PACK_NO_MMAP
990 		if (tot > 0 && tot <= max_in_mem_size) {
991 			(*obj)->data = mmap(NULL, tot, PROT_READ,
992 			    MAP_PRIVATE, *outfd, 0);
993 			if ((*obj)->data == MAP_FAILED) {
994 				if (errno != ENOMEM) {
995 					err = got_error_from_errno("mmap");
996 					goto done;
997 				}
998 				(*obj)->data = NULL;
999 			} else {
1000 				(*obj)->fd = *outfd;
1001 				*outfd = -1;
1002 			}
1003 		}
1004 #endif
1005 		if (*outfd != -1) {
1006 			(*obj)->f = fdopen(*outfd, "r");
1007 			if ((*obj)->f == NULL) {
1008 				err = got_error_from_errno("fdopen");
1009 				goto done;
1010 			}
1011 			*outfd = -1;
1012 		}
1013 	}
1014 	(*obj)->hdrlen = hdrlen;
1015 	(*obj)->size = size;
1016 done:
1017 	if (err) {
1018 		if (*obj) {
1019 			got_object_raw_close(*obj);
1020 			*obj = NULL;
1021 		}
1022 	} else
1023 		(*obj)->refcnt++;
1024 	return err;
1025 }