1 /*
2  * Copyright (c) 2018, 2019, 2020 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/uio.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 #include <sys/mman.h>
24 
25 #include <errno.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <stdint.h>
30 #include <zlib.h>
31 #include <ctype.h>
32 #include <limits.h>
33 #include <time.h>
34 #include <unistd.h>
35 
36 #include "got_compat.h"
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_sha1.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_repository.h"
52 
53 #ifndef nitems
54 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
55 #endif
56 
57 struct got_object_id *
58 got_object_id_dup(struct got_object_id *id1)
59 {
60 	struct got_object_id *id2;
61 
62 	id2 = malloc(sizeof(*id2));
63 	if (id2 == NULL)
64 		return NULL;
65 	memcpy(id2, id1, sizeof(*id2));
66 	return id2;
67 }
68 
69 int
70 got_object_id_cmp(const struct got_object_id *id1,
71     const struct got_object_id *id2)
72 {
73 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
74 }
75 
76 const struct got_error *
77 got_object_qid_alloc_partial(struct got_object_qid **qid)
78 {
79 	*qid = malloc(sizeof(**qid));
80 	if (*qid == NULL)
81 		return got_error_from_errno("malloc");
82 
83 	(*qid)->data = NULL;
84 	return NULL;
85 }
86 
87 const struct got_error *
88 got_object_id_str(char **outbuf, struct got_object_id *id)
89 {
90 	static const size_t len = SHA1_DIGEST_STRING_LENGTH;
91 
92 	*outbuf = malloc(len);
93 	if (*outbuf == NULL)
94 		return got_error_from_errno("malloc");
95 
96 	if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
97 		free(*outbuf);
98 		*outbuf = NULL;
99 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
100 	}
101 
102 	return NULL;
103 }
104 
105 void
106 got_object_close(struct got_object *obj)
107 {
108 	if (obj->refcnt > 0) {
109 		obj->refcnt--;
110 		if (obj->refcnt > 0)
111 			return;
112 	}
113 
114 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
115 		struct got_delta *delta;
116 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
117 			delta = STAILQ_FIRST(&obj->deltas.entries);
118 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
119 			free(delta);
120 		}
121 	}
122 	free(obj);
123 }
124 
125 const struct got_error *
126 got_object_raw_close(struct got_raw_object *obj)
127 {
128 	const struct got_error *err = NULL;
129 
130 	if (obj->refcnt > 0) {
131 		obj->refcnt--;
132 		if (obj->refcnt > 0)
133 			return NULL;
134 	}
135 
136 	if (obj->close_cb)
137 		obj->close_cb(obj);
138 
139 	if (obj->f == NULL) {
140 		if (obj->fd != -1) {
141 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
142 				err = got_error_from_errno("munmap");
143 			if (close(obj->fd) == -1 && err == NULL)
144 				err = got_error_from_errno("close");
145 		} else
146 			free(obj->data);
147 	} else {
148 		if (fclose(obj->f) == EOF && err == NULL)
149 			err = got_error_from_errno("fclose");
150 	}
151 	free(obj);
152 	return err;
153 }
154 
155 void
156 got_object_qid_free(struct got_object_qid *qid)
157 {
158 	free(qid);
159 }
160 
161 void
162 got_object_id_queue_free(struct got_object_id_queue *ids)
163 {
164 	struct got_object_qid *qid;
165 
166 	while (!STAILQ_EMPTY(ids)) {
167 		qid = STAILQ_FIRST(ids);
168 		STAILQ_REMOVE_HEAD(ids, entry);
169 		got_object_qid_free(qid);
170 	}
171 }
172 
173 const struct got_error *
174 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
175 {
176 	const char *obj_labels[] = {
177 		GOT_OBJ_LABEL_COMMIT,
178 		GOT_OBJ_LABEL_TREE,
179 		GOT_OBJ_LABEL_BLOB,
180 		GOT_OBJ_LABEL_TAG,
181 	};
182 	const int obj_types[] = {
183 		GOT_OBJ_TYPE_COMMIT,
184 		GOT_OBJ_TYPE_TREE,
185 		GOT_OBJ_TYPE_BLOB,
186 		GOT_OBJ_TYPE_TAG,
187 	};
188 	int type = 0;
189 	size_t size = 0;
190 	size_t i;
191 	char *end;
192 
193 	*obj = NULL;
194 
195 	end = memchr(buf, '\0', len);
196 	if (end == NULL)
197 		return got_error(GOT_ERR_BAD_OBJ_HDR);
198 
199 	for (i = 0; i < nitems(obj_labels); i++) {
200 		const char *label = obj_labels[i];
201 		size_t label_len = strlen(label);
202 		const char *errstr;
203 
204 		if (len <= label_len || buf + label_len >= end ||
205 		    strncmp(buf, label, label_len) != 0)
206 			continue;
207 
208 		type = obj_types[i];
209 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
210 		if (errstr != NULL)
211 			return got_error(GOT_ERR_BAD_OBJ_HDR);
212 		break;
213 	}
214 
215 	if (type == 0)
216 		return got_error(GOT_ERR_BAD_OBJ_HDR);
217 
218 	*obj = calloc(1, sizeof(**obj));
219 	if (*obj == NULL)
220 		return got_error_from_errno("calloc");
221 	(*obj)->type = type;
222 	(*obj)->hdrlen = end - buf + 1;
223 	(*obj)->size = size;
224 	return NULL;
225 }
226 
227 const struct got_error *
228 got_object_read_header(struct got_object **obj, int fd)
229 {
230 	const struct got_error *err;
231 	struct got_inflate_buf zb;
232 	uint8_t *buf;
233 	const size_t zbsize = 64;
234 	size_t outlen, totlen;
235 	int nbuf = 1;
236 
237 	*obj = NULL;
238 
239 	buf = malloc(zbsize);
240 	if (buf == NULL)
241 		return got_error_from_errno("malloc");
242 	buf[0] = '\0';
243 
244 	err = got_inflate_init(&zb, buf, zbsize, NULL);
245 	if (err)
246 		return err;
247 
248 	totlen = 0;
249 	do {
250 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
251 		if (err)
252 			goto done;
253 		if (outlen == 0)
254 			break;
255 		totlen += outlen;
256 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
257 			uint8_t *newbuf;
258 			nbuf++;
259 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
260 			if (newbuf == NULL) {
261 				err = got_error_from_errno("recallocarray");
262 				goto done;
263 			}
264 			buf = newbuf;
265 			zb.outbuf = newbuf + totlen;
266 			zb.outlen = (nbuf * zbsize) - totlen;
267 		}
268 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
269 
270 	err = got_object_parse_header(obj, buf, totlen);
271 done:
272 	free(buf);
273 	got_inflate_end(&zb);
274 	return err;
275 }
276 
277 const struct got_error *
278 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
279     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
280     int infd)
281 {
282 	const struct got_error *err = NULL;
283 	struct got_object *obj;
284 	struct got_inflate_checksum csum;
285 	uint8_t sha1[SHA1_DIGEST_LENGTH];
286 	SHA1_CTX sha1_ctx;
287 	size_t len, consumed;
288 	FILE *f = NULL;
289 
290 	*outbuf = NULL;
291 	*size = 0;
292 	*hdrlen = 0;
293 
294 	SHA1Init(&sha1_ctx);
295 	memset(&csum, 0, sizeof(csum));
296 	csum.output_sha1 = &sha1_ctx;
297 
298 	if (lseek(infd, SEEK_SET, 0) == -1)
299 		return got_error_from_errno("lseek");
300 
301 	err = got_object_read_header(&obj, infd);
302 	if (err)
303 		return err;
304 
305 	if (lseek(infd, SEEK_SET, 0) == -1)
306 		return got_error_from_errno("lseek");
307 
308 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
309 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
310 		    obj->size + obj->hdrlen, infd);
311 	} else {
312 		int fd;
313 		/*
314 		 * XXX This uses an extra file descriptor for no good reason.
315 		 * We should have got_inflate_fd_to_fd().
316 		 */
317 		fd = dup(infd);
318 		if (fd == -1)
319 			return got_error_from_errno("dup");
320 		f = fdopen(fd, "r");
321 		if (f == NULL) {
322 			err = got_error_from_errno("fdopen");
323 			abort();
324 			close(fd);
325 			goto done;
326 		}
327 		err = got_inflate_to_fd(&len, f, &csum, outfd);
328 	}
329 	if (err)
330 		goto done;
331 
332 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
333 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
334 		goto done;
335 	}
336 
337 	SHA1Final(sha1, &sha1_ctx);
338 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
339 		char buf[SHA1_DIGEST_STRING_LENGTH];
340 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
341 		    "checksum failure for object %s",
342 		    got_sha1_digest_to_str(expected_id->sha1, buf,
343 		    sizeof(buf)));
344 		goto done;
345 	}
346 
347 	*size = obj->size;
348 	*hdrlen = obj->hdrlen;
349 done:
350 	got_object_close(obj);
351 	if (f && fclose(f) == EOF && err == NULL)
352 		err = got_error_from_errno("fclose");
353 	return err;
354 }
355 
356 struct got_commit_object *
357 got_object_commit_alloc_partial(void)
358 {
359 	struct got_commit_object *commit;
360 
361 	commit = calloc(1, sizeof(*commit));
362 	if (commit == NULL)
363 		return NULL;
364 	commit->tree_id = malloc(sizeof(*commit->tree_id));
365 	if (commit->tree_id == NULL) {
366 		free(commit);
367 		return NULL;
368 	}
369 
370 	STAILQ_INIT(&commit->parent_ids);
371 
372 	return commit;
373 }
374 
375 const struct got_error *
376 got_object_commit_add_parent(struct got_commit_object *commit,
377     const char *id_str)
378 {
379 	const struct got_error *err = NULL;
380 	struct got_object_qid *qid;
381 
382 	err = got_object_qid_alloc_partial(&qid);
383 	if (err)
384 		return err;
385 
386 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
387 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
388 		got_object_qid_free(qid);
389 		return err;
390 	}
391 
392 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
393 	commit->nparents++;
394 
395 	return NULL;
396 }
397 
398 static const struct got_error *
399 parse_gmtoff(time_t *gmtoff, const char *tzstr)
400 {
401 	int sign = 1;
402 	const char *p = tzstr;
403 	time_t h, m;
404 
405 	*gmtoff = 0;
406 
407 	if (*p == '-')
408 		sign = -1;
409 	else if (*p != '+')
410 		return got_error(GOT_ERR_BAD_OBJ_DATA);
411 	p++;
412 	if (!isdigit(*p) && !isdigit(*(p + 1)))
413 		return got_error(GOT_ERR_BAD_OBJ_DATA);
414 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
415 
416 	p += 2;
417 	if (!isdigit(*p) && !isdigit(*(p + 1)))
418 		return got_error(GOT_ERR_BAD_OBJ_DATA);
419 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
420 
421 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
422 	return NULL;
423 }
424 
425 static const struct got_error *
426 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
427 {
428 	const struct got_error *err = NULL;
429 	const char *errstr;
430 	char *space, *tzstr;
431 
432 	/* Parse and strip off trailing timezone indicator string. */
433 	space = strrchr(committer, ' ');
434 	if (space == NULL)
435 		return got_error(GOT_ERR_BAD_OBJ_DATA);
436 	tzstr = strdup(space + 1);
437 	if (tzstr == NULL)
438 		return got_error_from_errno("strdup");
439 	err = parse_gmtoff(gmtoff, tzstr);
440 	free(tzstr);
441 	if (err) {
442 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
443 			return err;
444 		/* Old versions of Git omitted the timestamp. */
445 		*time = 0;
446 		*gmtoff = 0;
447 		return NULL;
448 	}
449 	*space = '\0';
450 
451 	/* Timestamp is separated from committer name + email by space. */
452 	space = strrchr(committer, ' ');
453 	if (space == NULL)
454 		return got_error(GOT_ERR_BAD_OBJ_DATA);
455 
456 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
457 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
458 	if (errstr)
459 		return got_error(GOT_ERR_BAD_OBJ_DATA);
460 
461 	/* Strip off parsed time information, leaving just author and email. */
462 	*space = '\0';
463 
464 	return NULL;
465 }
466 
467 void
468 got_object_commit_close(struct got_commit_object *commit)
469 {
470 	if (commit->refcnt > 0) {
471 		commit->refcnt--;
472 		if (commit->refcnt > 0)
473 			return;
474 	}
475 
476 	got_object_id_queue_free(&commit->parent_ids);
477 	free(commit->tree_id);
478 	free(commit->author);
479 	free(commit->committer);
480 	free(commit->logmsg);
481 	free(commit);
482 }
483 
484 struct got_object_id *
485 got_object_commit_get_tree_id(struct got_commit_object *commit)
486 {
487 	return commit->tree_id;
488 }
489 
490 int
491 got_object_commit_get_nparents(struct got_commit_object *commit)
492 {
493 	return commit->nparents;
494 }
495 
496 const struct got_object_id_queue *
497 got_object_commit_get_parent_ids(struct got_commit_object *commit)
498 {
499 	return &commit->parent_ids;
500 }
501 
502 const char *
503 got_object_commit_get_author(struct got_commit_object *commit)
504 {
505 	return commit->author;
506 }
507 
508 time_t
509 got_object_commit_get_author_time(struct got_commit_object *commit)
510 {
511 	return commit->author_time;
512 }
513 
514 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
515 {
516 	return commit->author_gmtoff;
517 }
518 
519 const char *
520 got_object_commit_get_committer(struct got_commit_object *commit)
521 {
522 	return commit->committer;
523 }
524 
525 time_t
526 got_object_commit_get_committer_time(struct got_commit_object *commit)
527 {
528 	return commit->committer_time;
529 }
530 
531 time_t
532 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
533 {
534 	return commit->committer_gmtoff;
535 }
536 
537 const struct got_error *
538 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
539 {
540 	const struct got_error *err = NULL;
541 	const char *src;
542 	char *dst;
543 	size_t len;
544 
545 	len = strlen(commit->logmsg);
546 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
547 	if (*logmsg == NULL)
548 		return got_error_from_errno("malloc");
549 
550 	/*
551 	 * Strip out unusual headers. Headers are separated from the commit
552 	 * message body by a single empty line.
553 	 */
554 	src = commit->logmsg;
555 	dst = *logmsg;
556 	while (*src != '\0' && *src != '\n') {
557 		int copy_header = 1, eol = 0;
558 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
559 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
560 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
561 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
562 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
563 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
564 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
565 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
566 			copy_header = 0;
567 
568 		while (*src != '\0' && !eol) {
569 			if (copy_header) {
570 				*dst = *src;
571 				dst++;
572 			}
573 			if (*src == '\n')
574 				eol = 1;
575 			src++;
576 		}
577 	}
578 	*dst = '\0';
579 
580 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
581 		err = got_error(GOT_ERR_NO_SPACE);
582 		goto done;
583 	}
584 
585 	/* Trim redundant trailing whitespace. */
586 	len = strlen(*logmsg);
587 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
588 	    isspace((unsigned char)(*logmsg)[len - 1])) {
589 		(*logmsg)[len - 1] = '\0';
590 		len--;
591 	}
592 
593 	/* Append a trailing newline if missing. */
594 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
595 		(*logmsg)[len] = '\n';
596 		(*logmsg)[len + 1] = '\0';
597 	}
598 done:
599 	if (err) {
600 		free(*logmsg);
601 		*logmsg = NULL;
602 	}
603 	return err;
604 }
605 
606 const char *
607 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
608 {
609 	return commit->logmsg;
610 }
611 
612 const struct got_error *
613 got_object_parse_commit(struct got_commit_object **commit, char *buf,
614     size_t len)
615 {
616 	const struct got_error *err = NULL;
617 	char *s = buf;
618 	size_t label_len;
619 	ssize_t remain = (ssize_t)len;
620 
621 	if (remain == 0)
622 		return got_error(GOT_ERR_BAD_OBJ_DATA);
623 
624 	*commit = got_object_commit_alloc_partial();
625 	if (*commit == NULL)
626 		return got_error_from_errno("got_object_commit_alloc_partial");
627 
628 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
629 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
630 		remain -= label_len;
631 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
632 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
633 			goto done;
634 		}
635 		s += label_len;
636 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
637 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
638 			goto done;
639 		}
640 		remain -= SHA1_DIGEST_STRING_LENGTH;
641 		s += SHA1_DIGEST_STRING_LENGTH;
642 	} else {
643 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
644 		goto done;
645 	}
646 
647 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
648 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
649 		remain -= label_len;
650 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
651 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
652 			goto done;
653 		}
654 		s += label_len;
655 		err = got_object_commit_add_parent(*commit, s);
656 		if (err)
657 			goto done;
658 
659 		remain -= SHA1_DIGEST_STRING_LENGTH;
660 		s += SHA1_DIGEST_STRING_LENGTH;
661 	}
662 
663 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
664 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
665 		char *p;
666 		size_t slen;
667 
668 		remain -= label_len;
669 		if (remain <= 0) {
670 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
671 			goto done;
672 		}
673 		s += label_len;
674 		p = memchr(s, '\n', remain);
675 		if (p == NULL) {
676 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
677 			goto done;
678 		}
679 		*p = '\0';
680 		slen = strlen(s);
681 		err = parse_commit_time(&(*commit)->author_time,
682 		    &(*commit)->author_gmtoff, s);
683 		if (err)
684 			goto done;
685 		(*commit)->author = strdup(s);
686 		if ((*commit)->author == NULL) {
687 			err = got_error_from_errno("strdup");
688 			goto done;
689 		}
690 		s += slen + 1;
691 		remain -= slen + 1;
692 	}
693 
694 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
695 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
696 		char *p;
697 		size_t slen;
698 
699 		remain -= label_len;
700 		if (remain <= 0) {
701 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
702 			goto done;
703 		}
704 		s += label_len;
705 		p = memchr(s, '\n', remain);
706 		if (p == NULL) {
707 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
708 			goto done;
709 		}
710 		*p = '\0';
711 		slen = strlen(s);
712 		err = parse_commit_time(&(*commit)->committer_time,
713 		    &(*commit)->committer_gmtoff, s);
714 		if (err)
715 			goto done;
716 		(*commit)->committer = strdup(s);
717 		if ((*commit)->committer == NULL) {
718 			err = got_error_from_errno("strdup");
719 			goto done;
720 		}
721 		s += slen + 1;
722 		remain -= slen + 1;
723 	}
724 
725 	(*commit)->logmsg = strndup(s, remain);
726 	if ((*commit)->logmsg == NULL) {
727 		err = got_error_from_errno("strndup");
728 		goto done;
729 	}
730 done:
731 	if (err) {
732 		got_object_commit_close(*commit);
733 		*commit = NULL;
734 	}
735 	return err;
736 }
737 
738 const struct got_error *
739 got_object_read_commit(struct got_commit_object **commit, int fd,
740     struct got_object_id *expected_id, size_t expected_size)
741 {
742 	struct got_object *obj = NULL;
743 	const struct got_error *err = NULL;
744 	size_t len;
745 	uint8_t *p;
746 	struct got_inflate_checksum csum;
747 	SHA1_CTX sha1_ctx;
748 	struct got_object_id id;
749 
750 	SHA1Init(&sha1_ctx);
751 	memset(&csum, 0, sizeof(csum));
752 	csum.output_sha1 = &sha1_ctx;
753 
754 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
755 	if (err)
756 		return err;
757 
758 	SHA1Final(id.sha1, &sha1_ctx);
759 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
760 		char buf[SHA1_DIGEST_STRING_LENGTH];
761 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
762 		    "checksum failure for object %s",
763 		    got_sha1_digest_to_str(expected_id->sha1, buf,
764 		    sizeof(buf)));
765 		goto done;
766 	}
767 
768 	err = got_object_parse_header(&obj, p, len);
769 	if (err)
770 		goto done;
771 
772 	if (len < obj->hdrlen + obj->size) {
773 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
774 		goto done;
775 	}
776 
777 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
778 		err = got_error(GOT_ERR_OBJ_TYPE);
779 		goto done;
780 	}
781 
782 	/* Skip object header. */
783 	len -= obj->hdrlen;
784 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
785 done:
786 	free(p);
787 	if (obj)
788 		got_object_close(obj);
789 	return err;
790 }
791 
792 void
793 got_object_tree_close(struct got_tree_object *tree)
794 {
795 	if (tree->refcnt > 0) {
796 		tree->refcnt--;
797 		if (tree->refcnt > 0)
798 			return;
799 	}
800 
801 	free(tree->entries);
802 	free(tree);
803 }
804 
805 static const struct got_error *
806 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
807     size_t maxlen)
808 {
809 	char *p, *space;
810 
811 	*elen = 0;
812 
813 	*elen = strnlen(buf, maxlen) + 1;
814 	if (*elen > maxlen)
815 		return got_error(GOT_ERR_BAD_OBJ_DATA);
816 
817 	space = memchr(buf, ' ', *elen);
818 	if (space == NULL || space <= buf)
819 		return got_error(GOT_ERR_BAD_OBJ_DATA);
820 
821 	pte->mode = 0;
822 	p = buf;
823 	while (p < space) {
824 		if (*p < '0' && *p > '7')
825 			return got_error(GOT_ERR_BAD_OBJ_DATA);
826 		pte->mode <<= 3;
827 		pte->mode |= *p - '0';
828 		p++;
829 	}
830 
831 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
832 		return got_error(GOT_ERR_BAD_OBJ_DATA);
833 
834 	pte->name = space + 1;
835 	pte->namelen = strlen(pte->name);
836 	buf += *elen;
837 	pte->id = buf;
838 	*elen += SHA1_DIGEST_LENGTH;
839 	return NULL;
840 }
841 
842 static int
843 pte_cmp(const void *pa, const void *pb)
844 {
845 	const struct got_parsed_tree_entry *a = pa, *b = pb;
846 
847 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
848 }
849 
850 const struct got_error *
851 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
852     size_t *nentries_alloc, uint8_t *buf, size_t len)
853 {
854 	const struct got_error *err = NULL;
855 	size_t remain = len;
856 	const size_t nalloc = 16;
857 	struct got_parsed_tree_entry *pte;
858 	int i;
859 
860 	*nentries = 0;
861 	if (remain == 0)
862 		return NULL; /* tree is empty */
863 
864 	while (remain > 0) {
865 		size_t elen;
866 
867 		if (*nentries >= *nentries_alloc) {
868 			pte = recallocarray(*entries, *nentries_alloc,
869 			    *nentries_alloc + nalloc, sizeof(**entries));
870 			if (pte == NULL) {
871 				err = got_error_from_errno("recallocarray");
872 				goto done;
873 			}
874 			*entries = pte;
875 			*nentries_alloc += nalloc;
876 		}
877 
878 		pte = &(*entries)[*nentries];
879 		err = parse_tree_entry(pte, &elen, buf, remain);
880 		if (err)
881 			goto done;
882 		buf += elen;
883 		remain -= elen;
884 		(*nentries)++;
885 	}
886 
887 	if (remain != 0) {
888 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
889 		goto done;
890 	}
891 
892 	if (*nentries > 1) {
893 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
894 
895 		for (i = 0; i < *nentries - 1; i++) {
896 			struct got_parsed_tree_entry *prev = &(*entries)[i];
897 			pte = &(*entries)[i + 1];
898 			if (got_path_cmp(prev->name, pte->name,
899 			    prev->namelen, pte->namelen) == 0) {
900 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
901 				break;
902 			}
903 		}
904 	}
905 done:
906 	if (err)
907 		*nentries = 0;
908 	return err;
909 }
910 
911 const struct got_error *
912 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
913     size_t *nentries_alloc, uint8_t **p, int fd,
914     struct got_object_id *expected_id)
915 {
916 	const struct got_error *err = NULL;
917 	struct got_object *obj = NULL;
918 	size_t len;
919 	struct got_inflate_checksum csum;
920 	SHA1_CTX sha1_ctx;
921 	struct got_object_id id;
922 
923 	SHA1Init(&sha1_ctx);
924 	memset(&csum, 0, sizeof(csum));
925 	csum.output_sha1 = &sha1_ctx;
926 
927 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
928 	if (err)
929 		return err;
930 
931 	SHA1Final(id.sha1, &sha1_ctx);
932 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
933 		char buf[SHA1_DIGEST_STRING_LENGTH];
934 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
935 		    "checksum failure for object %s",
936 		    got_sha1_digest_to_str(expected_id->sha1, buf,
937 		    sizeof(buf)));
938 		goto done;
939 	}
940 
941 	err = got_object_parse_header(&obj, *p, len);
942 	if (err)
943 		goto done;
944 
945 	if (len < obj->hdrlen + obj->size) {
946 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
947 		goto done;
948 	}
949 
950 	/* Skip object header. */
951 	len -= obj->hdrlen;
952 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
953 	    *p + obj->hdrlen, len);
954 done:
955 	if (obj)
956 		got_object_close(obj);
957 	return err;
958 }
959 
960 void
961 got_object_tag_close(struct got_tag_object *tag)
962 {
963 	if (tag->refcnt > 0) {
964 		tag->refcnt--;
965 		if (tag->refcnt > 0)
966 			return;
967 	}
968 
969 	free(tag->tag);
970 	free(tag->tagger);
971 	free(tag->tagmsg);
972 	free(tag);
973 }
974 
975 const struct got_error *
976 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
977 {
978 	const struct got_error *err = NULL;
979 	size_t remain = len;
980 	char *s = buf;
981 	size_t label_len;
982 
983 	if (remain == 0)
984 		return got_error(GOT_ERR_BAD_OBJ_DATA);
985 
986 	*tag = calloc(1, sizeof(**tag));
987 	if (*tag == NULL)
988 		return got_error_from_errno("calloc");
989 
990 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
991 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
992 		remain -= label_len;
993 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
994 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
995 			goto done;
996 		}
997 		s += label_len;
998 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
999 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1000 			goto done;
1001 		}
1002 		remain -= SHA1_DIGEST_STRING_LENGTH;
1003 		s += SHA1_DIGEST_STRING_LENGTH;
1004 	} else {
1005 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1006 		goto done;
1007 	}
1008 
1009 	if (remain <= 0) {
1010 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1011 		goto done;
1012 	}
1013 
1014 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1015 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1016 		remain -= label_len;
1017 		if (remain <= 0) {
1018 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1019 			goto done;
1020 		}
1021 		s += label_len;
1022 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1023 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1024 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1025 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1026 			s += label_len;
1027 			remain -= label_len;
1028 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1029 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1030 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1031 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1032 			s += label_len;
1033 			remain -= label_len;
1034 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1035 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1036 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1037 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1038 			s += label_len;
1039 			remain -= label_len;
1040 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1041 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1042 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1043 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1044 			s += label_len;
1045 			remain -= label_len;
1046 		} else {
1047 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1048 			goto done;
1049 		}
1050 
1051 		if (remain <= 0 || *s != '\n') {
1052 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1053 			goto done;
1054 		}
1055 		s++;
1056 		remain--;
1057 		if (remain <= 0) {
1058 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1059 			goto done;
1060 		}
1061 	} else {
1062 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1063 		goto done;
1064 	}
1065 
1066 	label_len = strlen(GOT_TAG_LABEL_TAG);
1067 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1068 		char *p;
1069 		size_t slen;
1070 		remain -= label_len;
1071 		if (remain <= 0) {
1072 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1073 			goto done;
1074 		}
1075 		s += label_len;
1076 		p = memchr(s, '\n', remain);
1077 		if (p == NULL) {
1078 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1079 			goto done;
1080 		}
1081 		*p = '\0';
1082 		slen = strlen(s);
1083 		(*tag)->tag = strndup(s, slen);
1084 		if ((*tag)->tag == NULL) {
1085 			err = got_error_from_errno("strndup");
1086 			goto done;
1087 		}
1088 		s += slen + 1;
1089 		remain -= slen + 1;
1090 		if (remain <= 0) {
1091 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1092 			goto done;
1093 		}
1094 	} else {
1095 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1096 		goto done;
1097 	}
1098 
1099 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1100 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1101 		char *p;
1102 		size_t slen;
1103 
1104 		remain -= label_len;
1105 		if (remain <= 0) {
1106 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1107 			goto done;
1108 		}
1109 		s += label_len;
1110 		p = memchr(s, '\n', remain);
1111 		if (p == NULL) {
1112 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1113 			goto done;
1114 		}
1115 		*p = '\0';
1116 		slen = strlen(s);
1117 		err = parse_commit_time(&(*tag)->tagger_time,
1118 		    &(*tag)->tagger_gmtoff, s);
1119 		if (err)
1120 			goto done;
1121 		(*tag)->tagger = strdup(s);
1122 		if ((*tag)->tagger == NULL) {
1123 			err = got_error_from_errno("strdup");
1124 			goto done;
1125 		}
1126 		s += slen + 1;
1127 		remain -= slen + 1;
1128 		if (remain < 0) {
1129 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1130 			goto done;
1131 		}
1132 	} else {
1133 		/* Some old tags in the Linux git repo have no tagger. */
1134 		(*tag)->tagger = strdup("");
1135 		if ((*tag)->tagger == NULL) {
1136 			err = got_error_from_errno("strdup");
1137 			goto done;
1138 		}
1139 	}
1140 
1141 	(*tag)->tagmsg = strndup(s, remain);
1142 	if ((*tag)->tagmsg == NULL) {
1143 		err = got_error_from_errno("strndup");
1144 		goto done;
1145 	}
1146 done:
1147 	if (err) {
1148 		got_object_tag_close(*tag);
1149 		*tag = NULL;
1150 	}
1151 	return err;
1152 }
1153 
1154 const struct got_error *
1155 got_object_read_tag(struct got_tag_object **tag, int fd, 
1156     struct got_object_id *expected_id, size_t expected_size)
1157 {
1158 	const struct got_error *err = NULL;
1159 	struct got_object *obj = NULL;
1160 	size_t len;
1161 	uint8_t *p;
1162 	struct got_inflate_checksum csum;
1163 	SHA1_CTX sha1_ctx;
1164 	struct got_object_id id;
1165 
1166 	SHA1Init(&sha1_ctx);
1167 	memset(&csum, 0, sizeof(csum));
1168 	csum.output_sha1 = &sha1_ctx;
1169 
1170 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1171 	    expected_size, fd);
1172 	if (err)
1173 		return err;
1174 
1175 	SHA1Final(id.sha1, &sha1_ctx);
1176 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
1177 		char buf[SHA1_DIGEST_STRING_LENGTH];
1178 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
1179 		    "checksum failure for object %s",
1180 		    got_sha1_digest_to_str(expected_id->sha1, buf,
1181 		    sizeof(buf)));
1182 		goto done;
1183 	}
1184 
1185 	err = got_object_parse_header(&obj, p, len);
1186 	if (err)
1187 		goto done;
1188 
1189 	if (len < obj->hdrlen + obj->size) {
1190 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1191 		goto done;
1192 	}
1193 
1194 	/* Skip object header. */
1195 	len -= obj->hdrlen;
1196 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1197 done:
1198 	free(p);
1199 	if (obj)
1200 		got_object_close(obj);
1201 	return err;
1202 }
1203 
1204 const struct got_error *
1205 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1206 {
1207 	const struct got_error *err = NULL;
1208 	static const size_t blocksize = 512;
1209 	size_t n, total, remain;
1210 	uint8_t *buf;
1211 
1212 	*outbuf = NULL;
1213 	*outlen = 0;
1214 
1215 	buf = malloc(blocksize);
1216 	if (buf == NULL)
1217 		return got_error_from_errno("malloc");
1218 
1219 	remain = blocksize;
1220 	total = 0;
1221 	for (;;) {
1222 		if (remain == 0) {
1223 			uint8_t *newbuf;
1224 			newbuf = reallocarray(buf, 1, total + blocksize);
1225 			if (newbuf == NULL) {
1226 				err = got_error_from_errno("reallocarray");
1227 				goto done;
1228 			}
1229 			buf = newbuf;
1230 			remain += blocksize;
1231 		}
1232 		n = fread(buf + total, 1, remain, f);
1233 		if (n == 0) {
1234 			if (ferror(f)) {
1235 				err = got_ferror(f, GOT_ERR_IO);
1236 				goto done;
1237 			}
1238 			break; /* EOF */
1239 		}
1240 		remain -= n;
1241 		total += n;
1242 	};
1243 
1244 done:
1245 	if (err == NULL) {
1246 		*outbuf = buf;
1247 		*outlen = total;
1248 	} else
1249 		free(buf);
1250 	return err;
1251 }