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 <unistd.h>
31 #include <zlib.h>
32 #include <libgen.h>
33 #include <limits.h>
34 #include <imsg.h>
35 
36 #include "got_error.h"
37 #include "got_object.h"
38 #include "got_repository.h"
39 #include "got_opentemp.h"
40 #include "got_path.h"
41 
42 #include "got_lib_sha1.h"
43 #include "got_lib_delta.h"
44 #include "got_lib_inflate.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idcache.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_repository.h"
51 
52 #ifndef MIN
53 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
54 #endif
55 
56 #ifndef nitems
57 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
58 #endif
59 
60 struct got_object_id *
61 got_object_get_id(struct got_object *obj)
62 {
63 	return &obj->id;
64 }
65 
66 const struct got_error *
67 got_object_get_id_str(char **outbuf, struct got_object *obj)
68 {
69 	return got_object_id_str(outbuf, &obj->id);
70 }
71 
72 const struct got_error *
73 got_object_get_type(int *type, struct got_repository *repo,
74     struct got_object_id *id)
75 {
76 	const struct got_error *err = NULL;
77 	struct got_object *obj;
78 
79 	err = got_object_open(&obj, repo, id);
80 	if (err)
81 		return err;
82 
83 	switch (obj->type) {
84 	case GOT_OBJ_TYPE_COMMIT:
85 	case GOT_OBJ_TYPE_TREE:
86 	case GOT_OBJ_TYPE_BLOB:
87 	case GOT_OBJ_TYPE_TAG:
88 		*type = obj->type;
89 		break;
90 	default:
91 		err = got_error(GOT_ERR_OBJ_TYPE);
92 		break;
93 	}
94 
95 	got_object_close(obj);
96 	return err;
97 }
98 
99 const struct got_error *
100 got_object_get_path(char **path, struct got_object_id *id,
101     struct got_repository *repo)
102 {
103 	const struct got_error *err = NULL;
104 	char *hex = NULL;
105 	char *path_objects;
106 
107 	*path = NULL;
108 
109 	path_objects = got_repo_get_path_objects(repo);
110 	if (path_objects == NULL)
111 		return got_error_from_errno("got_repo_get_path_objects");
112 
113 	err = got_object_id_str(&hex, id);
114 	if (err)
115 		goto done;
116 
117 	if (asprintf(path, "%s/%.2x/%s", path_objects,
118 	    id->sha1[0], hex + 2) == -1)
119 		err = got_error_from_errno("asprintf");
120 
121 done:
122 	free(hex);
123 	free(path_objects);
124 	return err;
125 }
126 
127 const struct got_error *
128 got_object_open_loose_fd(int *fd, struct got_object_id *id,
129     struct got_repository *repo)
130 {
131 	const struct got_error *err = NULL;
132 	char *path;
133 
134 	err = got_object_get_path(&path, id, repo);
135 	if (err)
136 		return err;
137 	*fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
138 	if (*fd == -1) {
139 		err = got_error_from_errno2("open", path);
140 		goto done;
141 	}
142 done:
143 	free(path);
144 	return err;
145 }
146 
147 const struct got_error *
148 got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
149     const char *id_str)
150 {
151 	struct got_object_id id;
152 
153 	if (!got_parse_sha1_digest(id.sha1, id_str))
154 		return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
155 
156 	return got_object_open(obj, repo, &id);
157 }
158 
159 const struct got_error *
160 got_object_resolve_id_str(struct got_object_id **id,
161     struct got_repository *repo, const char *id_str)
162 {
163 	const struct got_error *err = NULL;
164 	struct got_object *obj;
165 
166 	err = got_object_open_by_id_str(&obj, repo, id_str);
167 	if (err)
168 		return err;
169 
170 	*id = got_object_id_dup(got_object_get_id(obj));
171 	got_object_close(obj);
172 	if (*id == NULL)
173 		return got_error_from_errno("got_object_id_dup");
174 
175 	return NULL;
176 }
177 
178 const struct got_error *
179 got_object_qid_alloc(struct got_object_qid **qid, struct got_object_id *id)
180 {
181 	*qid = calloc(1, sizeof(**qid));
182 	if (*qid == NULL)
183 		return got_error_from_errno("calloc");
184 
185 	memcpy(&(*qid)->id, id, sizeof((*qid)->id));
186 	return NULL;
187 }
188 
189 const struct got_error *
190 got_object_id_queue_copy(const struct got_object_id_queue *src,
191     struct got_object_id_queue *dest)
192 {
193 	const struct got_error *err;
194 	struct got_object_qid *qid;
195 
196 	STAILQ_FOREACH(qid, src, entry) {
197 		struct got_object_qid *new;
198 		/*
199 		 * Deep-copy the object ID only. Let the caller deal
200 		 * with setting up the new->data pointer if needed.
201 		 */
202 		err = got_object_qid_alloc(&new, &qid->id);
203 		if (err) {
204 			got_object_id_queue_free(dest);
205 			return err;
206 		}
207 		STAILQ_INSERT_TAIL(dest, new, entry);
208 	}
209 
210 	return NULL;
211 }
212 
213 int
214 got_object_tree_get_nentries(struct got_tree_object *tree)
215 {
216 	return tree->nentries;
217 }
218 
219 struct got_tree_entry *
220 got_object_tree_get_first_entry(struct got_tree_object *tree)
221 {
222 	return got_object_tree_get_entry(tree, 0);
223 }
224 
225 struct got_tree_entry *
226 got_object_tree_get_last_entry(struct got_tree_object *tree)
227 {
228 	return got_object_tree_get_entry(tree, tree->nentries - 1);
229 }
230 
231 struct got_tree_entry *
232 got_object_tree_get_entry(struct got_tree_object *tree, int i)
233 {
234 	if (i < 0 || i >= tree->nentries)
235 		return NULL;
236 	return &tree->entries[i];
237 }
238 
239 mode_t
240 got_tree_entry_get_mode(struct got_tree_entry *te)
241 {
242 	return te->mode;
243 }
244 
245 const char *
246 got_tree_entry_get_name(struct got_tree_entry *te)
247 {
248 	return &te->name[0];
249 }
250 
251 struct got_object_id *
252 got_tree_entry_get_id(struct got_tree_entry *te)
253 {
254 	return &te->id;
255 }
256 
257 const struct got_error *
258 got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
259 {
260 	const struct got_error *err = NULL;
261 	size_t len, totlen, hdrlen, offset;
262 
263 	*s = NULL;
264 
265 	hdrlen = got_object_blob_get_hdrlen(blob);
266 	totlen = 0;
267 	offset = 0;
268 	do {
269 		char *p;
270 
271 		err = got_object_blob_read_block(&len, blob);
272 		if (err)
273 			return err;
274 
275 		if (len == 0)
276 			break;
277 
278 		totlen += len - hdrlen;
279 		p = realloc(*s, totlen + 1);
280 		if (p == NULL) {
281 			err = got_error_from_errno("realloc");
282 			free(*s);
283 			*s = NULL;
284 			return err;
285 		}
286 		*s = p;
287 		/* Skip blob object header first time around. */
288 		memcpy(*s + offset,
289 		    got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
290 		hdrlen = 0;
291 		offset = totlen;
292 	} while (len > 0);
293 
294 	(*s)[totlen] = '\0';
295 	return NULL;
296 }
297 
298 const struct got_error *
299 got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
300     struct got_repository *repo)
301 {
302 	const struct got_error *err = NULL;
303 	struct got_blob_object *blob = NULL;
304 	int fd = -1;
305 
306 	*link_target = NULL;
307 
308 	if (!got_object_tree_entry_is_symlink(te))
309 		return got_error(GOT_ERR_TREE_ENTRY_TYPE);
310 
311 	fd = got_opentempfd();
312 	if (fd == -1) {
313 		err = got_error_from_errno("got_opentempfd");
314 		goto done;
315 	}
316 
317 	err = got_object_open_as_blob(&blob, repo,
318 	    got_tree_entry_get_id(te), PATH_MAX, fd);
319 	if (err)
320 		goto done;
321 
322 	err = got_object_blob_read_to_str(link_target, blob);
323 done:
324 	if (fd != -1 && close(fd) == -1 && err == NULL)
325 		err = got_error_from_errno("close");
326 	if (blob)
327 		got_object_blob_close(blob);
328 	if (err) {
329 		free(*link_target);
330 		*link_target = NULL;
331 	}
332 	return err;
333 }
334 
335 int
336 got_tree_entry_get_index(struct got_tree_entry *te)
337 {
338 	return te->idx;
339 }
340 
341 struct got_tree_entry *
342 got_tree_entry_get_next(struct got_tree_object *tree,
343     struct got_tree_entry *te)
344 {
345 	return got_object_tree_get_entry(tree, te->idx + 1);
346 }
347 
348 struct got_tree_entry *
349 got_tree_entry_get_prev(struct got_tree_object *tree,
350     struct got_tree_entry *te)
351 {
352 	return got_object_tree_get_entry(tree, te->idx - 1);
353 }
354 
355 const struct got_error *
356 got_object_blob_close(struct got_blob_object *blob)
357 {
358 	const struct got_error *err = NULL;
359 	free(blob->read_buf);
360 	if (blob->f && fclose(blob->f) == EOF)
361 		err = got_error_from_errno("fclose");
362 	free(blob->data);
363 	free(blob);
364 	return err;
365 }
366 
367 void
368 got_object_blob_rewind(struct got_blob_object *blob)
369 {
370 	if (blob->f)
371 		rewind(blob->f);
372 }
373 
374 char *
375 got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
376 {
377 	return got_sha1_digest_to_str(blob->id.sha1, buf, size);
378 }
379 
380 size_t
381 got_object_blob_get_hdrlen(struct got_blob_object *blob)
382 {
383 	return blob->hdrlen;
384 }
385 
386 const uint8_t *
387 got_object_blob_get_read_buf(struct got_blob_object *blob)
388 {
389 	return blob->read_buf;
390 }
391 
392 const struct got_error *
393 got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
394 {
395 	size_t n;
396 
397 	n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
398 	if (n == 0 && ferror(blob->f))
399 		return got_ferror(blob->f, GOT_ERR_IO);
400 	*outlenp = n;
401 	return NULL;
402 }
403 
404 const struct got_error *
405 got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
406 {
407 	const struct got_error *err;
408 	size_t hdrlen, len;
409 
410 	*binary = 0;
411 	hdrlen = got_object_blob_get_hdrlen(blob);
412 
413 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
414 		return got_error_from_errno("fseeko");
415 
416 	err = got_object_blob_read_block(&len, blob);
417 	if (err)
418 		return err;
419 
420 	*binary = memchr(blob->read_buf, '\0', len) != NULL;
421 
422 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
423 		return got_error_from_errno("fseeko");
424 	return NULL;
425 }
426 
427 const struct got_error *
428 got_object_blob_dump_to_file(off_t *filesize, int *nlines,
429     off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
430 {
431 	const struct got_error *err = NULL;
432 	size_t n, len, hdrlen;
433 	const uint8_t *buf;
434 	int i;
435 	const int alloc_chunksz = 512;
436 	size_t nalloc = 0;
437 	off_t off = 0, total_len = 0;
438 
439 	if (line_offsets)
440 		*line_offsets = NULL;
441 	if (filesize)
442 		*filesize = 0;
443 	if (nlines)
444 		*nlines = 0;
445 
446 	hdrlen = got_object_blob_get_hdrlen(blob);
447 	do {
448 		err = got_object_blob_read_block(&len, blob);
449 		if (err)
450 			return err;
451 		if (len == 0)
452 			break;
453 		buf = got_object_blob_get_read_buf(blob);
454 		i = hdrlen;
455 		if (nlines) {
456 			if (line_offsets && *line_offsets == NULL) {
457 				/* Have some data but perhaps no '\n'. */
458 				*nlines = 1;
459 				nalloc = alloc_chunksz;
460 				*line_offsets = calloc(nalloc,
461 				    sizeof(**line_offsets));
462 				if (*line_offsets == NULL)
463 					return got_error_from_errno("calloc");
464 
465 				/* Skip forward over end of first line. */
466 				while (i < len) {
467 					if (buf[i] == '\n')
468 						break;
469 					i++;
470 				}
471 			}
472 			/* Scan '\n' offsets in remaining chunk of data. */
473 			while (i < len) {
474 				if (buf[i] != '\n') {
475 					i++;
476 					continue;
477 				}
478 				(*nlines)++;
479 				if (line_offsets && nalloc < *nlines) {
480 					size_t n = *nlines + alloc_chunksz;
481 					off_t *o = recallocarray(*line_offsets,
482 					    nalloc, n, sizeof(**line_offsets));
483 					if (o == NULL) {
484 						free(*line_offsets);
485 						*line_offsets = NULL;
486 						return got_error_from_errno(
487 						    "recallocarray");
488 					}
489 					*line_offsets = o;
490 					nalloc = n;
491 				}
492 				if (line_offsets) {
493 					off = total_len + i - hdrlen + 1;
494 					(*line_offsets)[*nlines - 1] = off;
495 				}
496 				i++;
497 			}
498 		}
499 		/* Skip blob object header first time around. */
500 		n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
501 		if (n != len - hdrlen)
502 			return got_ferror(outfile, GOT_ERR_IO);
503 		total_len += len - hdrlen;
504 		hdrlen = 0;
505 	} while (len != 0);
506 
507 	if (fflush(outfile) != 0)
508 		return got_error_from_errno("fflush");
509 	rewind(outfile);
510 
511 	if (filesize)
512 		*filesize = total_len;
513 
514 	return NULL;
515 }
516 
517 const char *
518 got_object_tag_get_name(struct got_tag_object *tag)
519 {
520 	return tag->tag;
521 }
522 
523 int
524 got_object_tag_get_object_type(struct got_tag_object *tag)
525 {
526 	return tag->obj_type;
527 }
528 
529 struct got_object_id *
530 got_object_tag_get_object_id(struct got_tag_object *tag)
531 {
532 	return &tag->id;
533 }
534 
535 time_t
536 got_object_tag_get_tagger_time(struct got_tag_object *tag)
537 {
538 	return tag->tagger_time;
539 }
540 
541 time_t
542 got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
543 {
544 	return tag->tagger_gmtoff;
545 }
546 
547 const char *
548 got_object_tag_get_tagger(struct got_tag_object *tag)
549 {
550 	return tag->tagger;
551 }
552 
553 const char *
554 got_object_tag_get_message(struct got_tag_object *tag)
555 {
556 	return tag->tagmsg;
557 }
558 
559 static struct got_tree_entry *
560 find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
561 {
562 	int i;
563 
564 	/* Note that tree entries are sorted in strncmp() order. */
565 	for (i = 0; i < tree->nentries; i++) {
566 		struct got_tree_entry *te = &tree->entries[i];
567 		int cmp = strncmp(te->name, name, len);
568 		if (cmp < 0)
569 			continue;
570 		if (cmp > 0)
571 			break;
572 		if (te->name[len] == '\0')
573 			return te;
574 	}
575 	return NULL;
576 }
577 
578 struct got_tree_entry *
579 got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
580 {
581 	return find_entry_by_name(tree, name, strlen(name));
582 }
583 
584 const struct got_error *
585 got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
586     struct got_repository *repo, struct got_tree_object *tree,
587     const char *path)
588 {
589 	const struct got_error *err = NULL;
590 	struct got_tree_object *subtree = NULL;
591 	struct got_tree_entry *te = NULL;
592 	const char *seg, *s;
593 	size_t seglen;
594 
595 	*id = NULL;
596 
597 	s = path;
598 	while (s[0] == '/')
599 		s++;
600 	seg = s;
601 	seglen = 0;
602 	subtree = tree;
603 	while (*s) {
604 		struct got_tree_object *next_tree;
605 
606 		if (*s != '/') {
607 			s++;
608 			seglen++;
609 			if (*s)
610 				continue;
611 		}
612 
613 		te = find_entry_by_name(subtree, seg, seglen);
614 		if (te == NULL) {
615 			err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
616 			goto done;
617 		}
618 
619 		if (*s == '\0')
620 			break;
621 
622 		seg = s + 1;
623 		seglen = 0;
624 		s++;
625 		if (*s) {
626 			err = got_object_open_as_tree(&next_tree, repo,
627 			    &te->id);
628 			te = NULL;
629 			if (err)
630 				goto done;
631 			if (subtree != tree)
632 				got_object_tree_close(subtree);
633 			subtree = next_tree;
634 		}
635 	}
636 
637 	if (te) {
638 		*id = got_object_id_dup(&te->id);
639 		if (*id == NULL)
640 			return got_error_from_errno("got_object_id_dup");
641 		if (mode)
642 			*mode = te->mode;
643 	} else
644 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
645 done:
646 	if (subtree && subtree != tree)
647 		got_object_tree_close(subtree);
648 	return err;
649 }
650 
651 const struct got_error *
652 got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
653     struct got_commit_object *commit, const char *path)
654 {
655 	const struct got_error *err = NULL;
656 	struct got_tree_object *tree = NULL;
657 
658 	*id = NULL;
659 
660 	/* Handle opening of root of commit's tree. */
661 	if (got_path_is_root_dir(path)) {
662 		*id = got_object_id_dup(commit->tree_id);
663 		if (*id == NULL)
664 			err = got_error_from_errno("got_object_id_dup");
665 	} else {
666 		err = got_object_open_as_tree(&tree, repo, commit->tree_id);
667 		if (err)
668 			goto done;
669 		err = got_object_tree_find_path(id, NULL, repo, tree, path);
670 	}
671 done:
672 	if (tree)
673 		got_object_tree_close(tree);
674 	return err;
675 }
676 
677 /*
678  * Normalize file mode bits to avoid false positive tree entry differences
679  * in case tree entries have unexpected mode bits set.
680  */
681 static mode_t
682 normalize_mode_for_comparison(mode_t mode)
683 {
684 	/*
685 	 * For directories, the only relevant bit is the IFDIR bit.
686 	 * This allows us to detect paths changing from a directory
687 	 * to a file and vice versa.
688 	 */
689 	if (S_ISDIR(mode))
690 		return mode & S_IFDIR;
691 
692 	/*
693 	 * For symlinks, the only relevant bit is the IFLNK bit.
694 	 * This allows us to detect paths changing from a symlinks
695 	 * to a file or directory and vice versa.
696 	 */
697 	if (S_ISLNK(mode))
698 		return mode & S_IFLNK;
699 
700 	/* For files, the only change we care about is the executable bit. */
701 	return mode & S_IXUSR;
702 }
703 
704 const struct got_error *
705 got_object_tree_path_changed(int *changed,
706     struct got_tree_object *tree01, struct got_tree_object *tree02,
707     const char *path, struct got_repository *repo)
708 {
709 	const struct got_error *err = NULL;
710 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
711 	struct got_tree_entry *te1 = NULL, *te2 = NULL;
712 	const char *seg, *s;
713 	size_t seglen;
714 
715 	*changed = 0;
716 
717 	/* We not do support comparing the root path. */
718 	if (got_path_is_root_dir(path))
719 		return got_error_path(path, GOT_ERR_BAD_PATH);
720 
721 	tree1 = tree01;
722 	tree2 = tree02;
723 	s = path;
724 	while (*s == '/')
725 		s++;
726 	seg = s;
727 	seglen = 0;
728 	while (*s) {
729 		struct got_tree_object *next_tree1, *next_tree2;
730 		mode_t mode1, mode2;
731 
732 		if (*s != '/') {
733 			s++;
734 			seglen++;
735 			if (*s)
736 				continue;
737 		}
738 
739 		te1 = find_entry_by_name(tree1, seg, seglen);
740 		if (te1 == NULL) {
741 			err = got_error(GOT_ERR_NO_OBJ);
742 			goto done;
743 		}
744 
745 		if (tree2)
746 			te2 = find_entry_by_name(tree2, seg, seglen);
747 
748 		if (te2) {
749 			mode1 = normalize_mode_for_comparison(te1->mode);
750 			mode2 = normalize_mode_for_comparison(te2->mode);
751 			if (mode1 != mode2) {
752 				*changed = 1;
753 				goto done;
754 			}
755 
756 			if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
757 				*changed = 0;
758 				goto done;
759 			}
760 		}
761 
762 		if (*s == '\0') { /* final path element */
763 			*changed = 1;
764 			goto done;
765 		}
766 
767 		seg = s + 1;
768 		s++;
769 		seglen = 0;
770 		if (*s) {
771 			err = got_object_open_as_tree(&next_tree1, repo,
772 			    &te1->id);
773 			te1 = NULL;
774 			if (err)
775 				goto done;
776 			if (tree1 != tree01)
777 				got_object_tree_close(tree1);
778 			tree1 = next_tree1;
779 
780 			if (te2) {
781 				err = got_object_open_as_tree(&next_tree2, repo,
782 				    &te2->id);
783 				te2 = NULL;
784 				if (err)
785 					goto done;
786 				if (tree2 != tree02)
787 					got_object_tree_close(tree2);
788 				tree2 = next_tree2;
789 			} else if (tree2) {
790 				if (tree2 != tree02)
791 					got_object_tree_close(tree2);
792 				tree2 = NULL;
793 			}
794 		}
795 	}
796 done:
797 	if (tree1 && tree1 != tree01)
798 		got_object_tree_close(tree1);
799 	if (tree2 && tree2 != tree02)
800 		got_object_tree_close(tree2);
801 	return err;
802 }
803 
804 const struct got_error *
805 got_object_tree_entry_dup(struct got_tree_entry **new_te,
806     struct got_tree_entry *te)
807 {
808 	const struct got_error *err = NULL;
809 
810 	*new_te = calloc(1, sizeof(**new_te));
811 	if (*new_te == NULL)
812 		return got_error_from_errno("calloc");
813 
814 	(*new_te)->mode = te->mode;
815 	memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
816 	memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
817 	return err;
818 }
819 
820 int
821 got_object_tree_entry_is_submodule(struct got_tree_entry *te)
822 {
823 	return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
824 }
825 
826 int
827 got_object_tree_entry_is_symlink(struct got_tree_entry *te)
828 {
829 	/* S_IFDIR check avoids confusing symlinks with submodules. */
830 	return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
831 }
832 
833 static const struct got_error *
834 resolve_symlink(char **link_target, const char *path,
835     struct got_commit_object *commit, struct got_repository *repo)
836 {
837 	const struct got_error *err = NULL;
838 	char buf[PATH_MAX];
839 	char *name, *parent_path = NULL;
840 	struct got_object_id *tree_obj_id = NULL;
841 	struct got_tree_object *tree = NULL;
842 	struct got_tree_entry *te = NULL;
843 
844 	*link_target = NULL;
845 
846 	if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
847 		return got_error(GOT_ERR_NO_SPACE);
848 
849 	name = basename(buf);
850 	if (name == NULL)
851 		return got_error_from_errno2("basename", path);
852 
853 	err = got_path_dirname(&parent_path, path);
854 	if (err)
855 		return err;
856 
857 	err = got_object_id_by_path(&tree_obj_id, repo, commit,
858 	    parent_path);
859 	if (err) {
860 		if (err->code == GOT_ERR_NO_TREE_ENTRY) {
861 			/* Display the complete path in error message. */
862 			err = got_error_path(path, err->code);
863 		}
864 		goto done;
865 	}
866 
867 	err = got_object_open_as_tree(&tree, repo, tree_obj_id);
868 	if (err)
869 		goto done;
870 
871 	te = got_object_tree_find_entry(tree, name);
872 	if (te == NULL) {
873 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
874 		goto done;
875 	}
876 
877 	if (got_object_tree_entry_is_symlink(te)) {
878 		err = got_tree_entry_get_symlink_target(link_target, te, repo);
879 		if (err)
880 			goto done;
881 		if (!got_path_is_absolute(*link_target)) {
882 			char *abspath;
883 			if (asprintf(&abspath, "%s/%s", parent_path,
884 			    *link_target) == -1) {
885 				err = got_error_from_errno("asprintf");
886 				goto done;
887 			}
888 			free(*link_target);
889 			*link_target = malloc(PATH_MAX);
890 			if (*link_target == NULL) {
891 				err = got_error_from_errno("malloc");
892 				goto done;
893 			}
894 			err = got_canonpath(abspath, *link_target, PATH_MAX);
895 			free(abspath);
896 			if (err)
897 				goto done;
898 		}
899 	}
900 done:
901 	free(parent_path);
902 	free(tree_obj_id);
903 	if (tree)
904 		got_object_tree_close(tree);
905 	if (err) {
906 		free(*link_target);
907 		*link_target = NULL;
908 	}
909 	return err;
910 }
911 
912 const struct got_error *
913 got_object_resolve_symlinks(char **link_target, const char *path,
914     struct got_commit_object *commit, struct got_repository *repo)
915 {
916 	const struct got_error *err = NULL;
917 	char *next_target = NULL;
918 	int max_recursion = 40; /* matches Git */
919 
920 	*link_target = NULL;
921 
922 	do {
923 		err = resolve_symlink(&next_target,
924 		    *link_target ? *link_target : path, commit, repo);
925 		if (err)
926 			break;
927 		if (next_target) {
928 			free(*link_target);
929 			if (--max_recursion == 0) {
930 				err = got_error_path(path, GOT_ERR_RECURSION);
931 				*link_target = NULL;
932 				break;
933 			}
934 			*link_target = next_target;
935 		}
936 	} while (next_target);
937 
938 	return err;
939 }
940 
941 void
942 got_object_commit_retain(struct got_commit_object *commit)
943 {
944 	commit->refcnt++;
945 }
946 
947 const struct got_error *
948 got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
949     size_t hdrlen, off_t size)
950 {
951 	const struct got_error *err = NULL;
952 	off_t tot;
953 
954 	tot = hdrlen + size;
955 
956 	*obj = calloc(1, sizeof(**obj));
957 	if (*obj == NULL) {
958 		err = got_error_from_errno("calloc");
959 		goto done;
960 	}
961 	(*obj)->fd = -1;
962 	(*obj)->tempfile_idx = -1;
963 
964 	if (outbuf) {
965 		(*obj)->data = outbuf;
966 	} else {
967 		struct stat sb;
968 		if (fstat(*outfd, &sb) == -1) {
969 			err = got_error_from_errno("fstat");
970 			goto done;
971 		}
972 
973 		if (sb.st_size != tot) {
974 			err = got_error(GOT_ERR_PRIVSEP_LEN);
975 			goto done;
976 		}
977 #ifndef GOT_PACK_NO_MMAP
978 		if (tot > 0 && tot <= SIZE_MAX) {
979 			(*obj)->data = mmap(NULL, tot, PROT_READ,
980 			    MAP_PRIVATE, *outfd, 0);
981 			if ((*obj)->data == MAP_FAILED) {
982 				if (errno != ENOMEM) {
983 					err = got_error_from_errno("mmap");
984 					goto done;
985 				}
986 				(*obj)->data = NULL;
987 			} else {
988 				(*obj)->fd = *outfd;
989 				*outfd = -1;
990 			}
991 		}
992 #endif
993 		if (*outfd != -1) {
994 			(*obj)->f = fdopen(*outfd, "r");
995 			if ((*obj)->f == NULL) {
996 				err = got_error_from_errno("fdopen");
997 				goto done;
998 			}
999 			*outfd = -1;
1000 		}
1001 	}
1002 	(*obj)->hdrlen = hdrlen;
1003 	(*obj)->size = size;
1004 done:
1005 	if (err) {
1006 		if (*obj) {
1007 			got_object_raw_close(*obj);
1008 			*obj = NULL;
1009 		}
1010 	} else
1011 		(*obj)->refcnt++;
1012 	return err;
1013 }