Gotweb

Blob

Date:: Tue Feb 7 20:14:55 2023 UTC
Message:: introduce got_error_checksum ok stsp@
Actions:: History | Blame | Raw File
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 = GOT_OBJECT_ID_HEX_MAXLEN;
91 
92 	*outbuf = malloc(len);
93 	if (*outbuf == NULL)
94 		return got_error_from_errno("malloc");
95 
96 	if (got_object_id_hex(id, *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 char *
106 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
107 {
108 	return got_sha1_digest_to_str(id->sha1, buf, len);
109 }
110 
111 void
112 got_object_close(struct got_object *obj)
113 {
114 	if (obj->refcnt > 0) {
115 		obj->refcnt--;
116 		if (obj->refcnt > 0)
117 			return;
118 	}
119 
120 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
121 		struct got_delta *delta;
122 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
123 			delta = STAILQ_FIRST(&obj->deltas.entries);
124 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
125 			free(delta);
126 		}
127 	}
128 	free(obj);
129 }
130 
131 const struct got_error *
132 got_object_raw_close(struct got_raw_object *obj)
133 {
134 	const struct got_error *err = NULL;
135 
136 	if (obj->refcnt > 0) {
137 		obj->refcnt--;
138 		if (obj->refcnt > 0)
139 			return NULL;
140 	}
141 
142 	if (obj->close_cb)
143 		obj->close_cb(obj);
144 
145 	if (obj->f == NULL) {
146 		if (obj->fd != -1) {
147 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
148 				err = got_error_from_errno("munmap");
149 			if (close(obj->fd) == -1 && err == NULL)
150 				err = got_error_from_errno("close");
151 		} else
152 			free(obj->data);
153 	} else {
154 		if (fclose(obj->f) == EOF && err == NULL)
155 			err = got_error_from_errno("fclose");
156 	}
157 	free(obj);
158 	return err;
159 }
160 
161 void
162 got_object_qid_free(struct got_object_qid *qid)
163 {
164 	free(qid);
165 }
166 
167 void
168 got_object_id_queue_free(struct got_object_id_queue *ids)
169 {
170 	struct got_object_qid *qid;
171 
172 	while (!STAILQ_EMPTY(ids)) {
173 		qid = STAILQ_FIRST(ids);
174 		STAILQ_REMOVE_HEAD(ids, entry);
175 		got_object_qid_free(qid);
176 	}
177 }
178 
179 const struct got_error *
180 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
181 {
182 	const char *obj_labels[] = {
183 		GOT_OBJ_LABEL_COMMIT,
184 		GOT_OBJ_LABEL_TREE,
185 		GOT_OBJ_LABEL_BLOB,
186 		GOT_OBJ_LABEL_TAG,
187 	};
188 	const int obj_types[] = {
189 		GOT_OBJ_TYPE_COMMIT,
190 		GOT_OBJ_TYPE_TREE,
191 		GOT_OBJ_TYPE_BLOB,
192 		GOT_OBJ_TYPE_TAG,
193 	};
194 	int type = 0;
195 	size_t size = 0;
196 	size_t i;
197 	char *end;
198 
199 	*obj = NULL;
200 
201 	end = memchr(buf, '\0', len);
202 	if (end == NULL)
203 		return got_error(GOT_ERR_BAD_OBJ_HDR);
204 
205 	for (i = 0; i < nitems(obj_labels); i++) {
206 		const char *label = obj_labels[i];
207 		size_t label_len = strlen(label);
208 		const char *errstr;
209 
210 		if (len <= label_len || buf + label_len >= end ||
211 		    strncmp(buf, label, label_len) != 0)
212 			continue;
213 
214 		type = obj_types[i];
215 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
216 		if (errstr != NULL)
217 			return got_error(GOT_ERR_BAD_OBJ_HDR);
218 		break;
219 	}
220 
221 	if (type == 0)
222 		return got_error(GOT_ERR_BAD_OBJ_HDR);
223 
224 	*obj = calloc(1, sizeof(**obj));
225 	if (*obj == NULL)
226 		return got_error_from_errno("calloc");
227 	(*obj)->type = type;
228 	(*obj)->hdrlen = end - buf + 1;
229 	(*obj)->size = size;
230 	return NULL;
231 }
232 
233 const struct got_error *
234 got_object_read_header(struct got_object **obj, int fd)
235 {
236 	const struct got_error *err;
237 	struct got_inflate_buf zb;
238 	uint8_t *buf;
239 	const size_t zbsize = 64;
240 	size_t outlen, totlen;
241 	int nbuf = 1;
242 
243 	*obj = NULL;
244 
245 	buf = malloc(zbsize);
246 	if (buf == NULL)
247 		return got_error_from_errno("malloc");
248 	buf[0] = '\0';
249 
250 	err = got_inflate_init(&zb, buf, zbsize, NULL);
251 	if (err)
252 		return err;
253 
254 	totlen = 0;
255 	do {
256 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
257 		if (err)
258 			goto done;
259 		if (outlen == 0)
260 			break;
261 		totlen += outlen;
262 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
263 			uint8_t *newbuf;
264 			nbuf++;
265 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
266 			if (newbuf == NULL) {
267 				err = got_error_from_errno("recallocarray");
268 				goto done;
269 			}
270 			buf = newbuf;
271 			zb.outbuf = newbuf + totlen;
272 			zb.outlen = (nbuf * zbsize) - totlen;
273 		}
274 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
275 
276 	err = got_object_parse_header(obj, buf, totlen);
277 done:
278 	free(buf);
279 	got_inflate_end(&zb);
280 	return err;
281 }
282 
283 const struct got_error *
284 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
285     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
286     int infd)
287 {
288 	const struct got_error *err = NULL;
289 	struct got_object *obj;
290 	struct got_inflate_checksum csum;
291 	uint8_t sha1[SHA1_DIGEST_LENGTH];
292 	SHA1_CTX sha1_ctx;
293 	size_t len, consumed;
294 	FILE *f = NULL;
295 
296 	*outbuf = NULL;
297 	*size = 0;
298 	*hdrlen = 0;
299 
300 	SHA1Init(&sha1_ctx);
301 	memset(&csum, 0, sizeof(csum));
302 	csum.output_sha1 = &sha1_ctx;
303 
304 	if (lseek(infd, SEEK_SET, 0) == -1)
305 		return got_error_from_errno("lseek");
306 
307 	err = got_object_read_header(&obj, infd);
308 	if (err)
309 		return err;
310 
311 	if (lseek(infd, SEEK_SET, 0) == -1)
312 		return got_error_from_errno("lseek");
313 
314 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
315 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
316 		    obj->size + obj->hdrlen, infd);
317 	} else {
318 		int fd;
319 		/*
320 		 * XXX This uses an extra file descriptor for no good reason.
321 		 * We should have got_inflate_fd_to_fd().
322 		 */
323 		fd = dup(infd);
324 		if (fd == -1)
325 			return got_error_from_errno("dup");
326 		f = fdopen(fd, "r");
327 		if (f == NULL) {
328 			err = got_error_from_errno("fdopen");
329 			abort();
330 			close(fd);
331 			goto done;
332 		}
333 		err = got_inflate_to_fd(&len, f, &csum, outfd);
334 	}
335 	if (err)
336 		goto done;
337 
338 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
339 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
340 		goto done;
341 	}
342 
343 	SHA1Final(sha1, &sha1_ctx);
344 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
345 		err = got_error_checksum(expected_id);
346 		goto done;
347 	}
348 
349 	*size = obj->size;
350 	*hdrlen = obj->hdrlen;
351 done:
352 	got_object_close(obj);
353 	if (f && fclose(f) == EOF && err == NULL)
354 		err = got_error_from_errno("fclose");
355 	return err;
356 }
357 
358 struct got_commit_object *
359 got_object_commit_alloc_partial(void)
360 {
361 	struct got_commit_object *commit;
362 
363 	commit = calloc(1, sizeof(*commit));
364 	if (commit == NULL)
365 		return NULL;
366 	commit->tree_id = malloc(sizeof(*commit->tree_id));
367 	if (commit->tree_id == NULL) {
368 		free(commit);
369 		return NULL;
370 	}
371 
372 	STAILQ_INIT(&commit->parent_ids);
373 
374 	return commit;
375 }
376 
377 const struct got_error *
378 got_object_commit_add_parent(struct got_commit_object *commit,
379     const char *id_str)
380 {
381 	const struct got_error *err = NULL;
382 	struct got_object_qid *qid;
383 
384 	err = got_object_qid_alloc_partial(&qid);
385 	if (err)
386 		return err;
387 
388 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
389 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
390 		got_object_qid_free(qid);
391 		return err;
392 	}
393 
394 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
395 	commit->nparents++;
396 
397 	return NULL;
398 }
399 
400 static const struct got_error *
401 parse_gmtoff(time_t *gmtoff, const char *tzstr)
402 {
403 	int sign = 1;
404 	const char *p = tzstr;
405 	time_t h, m;
406 
407 	*gmtoff = 0;
408 
409 	if (*p == '-')
410 		sign = -1;
411 	else if (*p != '+')
412 		return got_error(GOT_ERR_BAD_OBJ_DATA);
413 	p++;
414 	if (!isdigit((unsigned char)*p) &&
415 	    !isdigit((unsigned char)*(p + 1)))
416 		return got_error(GOT_ERR_BAD_OBJ_DATA);
417 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
418 
419 	p += 2;
420 	if (!isdigit((unsigned char)*p) &&
421 	    !isdigit((unsigned char)*(p + 1)))
422 		return got_error(GOT_ERR_BAD_OBJ_DATA);
423 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
424 
425 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
426 	return NULL;
427 }
428 
429 static const struct got_error *
430 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
431 {
432 	const struct got_error *err = NULL;
433 	const char *errstr;
434 	char *space, *tzstr;
435 
436 	/* Parse and strip off trailing timezone indicator string. */
437 	space = strrchr(committer, ' ');
438 	if (space == NULL)
439 		return got_error(GOT_ERR_BAD_OBJ_DATA);
440 	tzstr = strdup(space + 1);
441 	if (tzstr == NULL)
442 		return got_error_from_errno("strdup");
443 	err = parse_gmtoff(gmtoff, tzstr);
444 	free(tzstr);
445 	if (err) {
446 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
447 			return err;
448 		/* Old versions of Git omitted the timestamp. */
449 		*time = 0;
450 		*gmtoff = 0;
451 		return NULL;
452 	}
453 	*space = '\0';
454 
455 	/* Timestamp is separated from committer name + email by space. */
456 	space = strrchr(committer, ' ');
457 	if (space == NULL)
458 		return got_error(GOT_ERR_BAD_OBJ_DATA);
459 
460 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
461 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
462 	if (errstr)
463 		return got_error(GOT_ERR_BAD_OBJ_DATA);
464 
465 	/* Strip off parsed time information, leaving just author and email. */
466 	*space = '\0';
467 
468 	return NULL;
469 }
470 
471 void
472 got_object_commit_close(struct got_commit_object *commit)
473 {
474 	if (commit->refcnt > 0) {
475 		commit->refcnt--;
476 		if (commit->refcnt > 0)
477 			return;
478 	}
479 
480 	got_object_id_queue_free(&commit->parent_ids);
481 	free(commit->tree_id);
482 	free(commit->author);
483 	free(commit->committer);
484 	free(commit->logmsg);
485 	free(commit);
486 }
487 
488 struct got_object_id *
489 got_object_commit_get_tree_id(struct got_commit_object *commit)
490 {
491 	return commit->tree_id;
492 }
493 
494 int
495 got_object_commit_get_nparents(struct got_commit_object *commit)
496 {
497 	return commit->nparents;
498 }
499 
500 const struct got_object_id_queue *
501 got_object_commit_get_parent_ids(struct got_commit_object *commit)
502 {
503 	return &commit->parent_ids;
504 }
505 
506 const char *
507 got_object_commit_get_author(struct got_commit_object *commit)
508 {
509 	return commit->author;
510 }
511 
512 time_t
513 got_object_commit_get_author_time(struct got_commit_object *commit)
514 {
515 	return commit->author_time;
516 }
517 
518 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
519 {
520 	return commit->author_gmtoff;
521 }
522 
523 const char *
524 got_object_commit_get_committer(struct got_commit_object *commit)
525 {
526 	return commit->committer;
527 }
528 
529 time_t
530 got_object_commit_get_committer_time(struct got_commit_object *commit)
531 {
532 	return commit->committer_time;
533 }
534 
535 time_t
536 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
537 {
538 	return commit->committer_gmtoff;
539 }
540 
541 const struct got_error *
542 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
543 {
544 	const struct got_error *err = NULL;
545 	const char *src;
546 	char *dst;
547 	size_t len;
548 
549 	len = strlen(commit->logmsg);
550 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
551 	if (*logmsg == NULL)
552 		return got_error_from_errno("malloc");
553 
554 	/*
555 	 * Strip out unusual headers. Headers are separated from the commit
556 	 * message body by a single empty line.
557 	 */
558 	src = commit->logmsg;
559 	dst = *logmsg;
560 	while (*src != '\0' && *src != '\n') {
561 		int copy_header = 1, eol = 0;
562 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
563 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
564 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
565 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
566 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
567 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
568 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
569 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
570 			copy_header = 0;
571 
572 		while (*src != '\0' && !eol) {
573 			if (copy_header) {
574 				*dst = *src;
575 				dst++;
576 			}
577 			if (*src == '\n')
578 				eol = 1;
579 			src++;
580 		}
581 	}
582 	*dst = '\0';
583 
584 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
585 		err = got_error(GOT_ERR_NO_SPACE);
586 		goto done;
587 	}
588 
589 	/* Trim redundant trailing whitespace. */
590 	len = strlen(*logmsg);
591 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
592 	    isspace((unsigned char)(*logmsg)[len - 1])) {
593 		(*logmsg)[len - 1] = '\0';
594 		len--;
595 	}
596 
597 	/* Append a trailing newline if missing. */
598 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
599 		(*logmsg)[len] = '\n';
600 		(*logmsg)[len + 1] = '\0';
601 	}
602 done:
603 	if (err) {
604 		free(*logmsg);
605 		*logmsg = NULL;
606 	}
607 	return err;
608 }
609 
610 const char *
611 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
612 {
613 	return commit->logmsg;
614 }
615 
616 const struct got_error *
617 got_object_parse_commit(struct got_commit_object **commit, char *buf,
618     size_t len)
619 {
620 	const struct got_error *err = NULL;
621 	char *s = buf;
622 	size_t label_len;
623 	ssize_t remain = (ssize_t)len;
624 
625 	if (remain == 0)
626 		return got_error(GOT_ERR_BAD_OBJ_DATA);
627 
628 	*commit = got_object_commit_alloc_partial();
629 	if (*commit == NULL)
630 		return got_error_from_errno("got_object_commit_alloc_partial");
631 
632 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
633 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
634 		remain -= label_len;
635 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
636 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
637 			goto done;
638 		}
639 		s += label_len;
640 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
641 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
642 			goto done;
643 		}
644 		remain -= SHA1_DIGEST_STRING_LENGTH;
645 		s += SHA1_DIGEST_STRING_LENGTH;
646 	} else {
647 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
648 		goto done;
649 	}
650 
651 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
652 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
653 		remain -= label_len;
654 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
655 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
656 			goto done;
657 		}
658 		s += label_len;
659 		err = got_object_commit_add_parent(*commit, s);
660 		if (err)
661 			goto done;
662 
663 		remain -= SHA1_DIGEST_STRING_LENGTH;
664 		s += SHA1_DIGEST_STRING_LENGTH;
665 	}
666 
667 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
668 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
669 		char *p;
670 		size_t slen;
671 
672 		remain -= label_len;
673 		if (remain <= 0) {
674 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
675 			goto done;
676 		}
677 		s += label_len;
678 		p = memchr(s, '\n', remain);
679 		if (p == NULL) {
680 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
681 			goto done;
682 		}
683 		*p = '\0';
684 		slen = strlen(s);
685 		err = parse_commit_time(&(*commit)->author_time,
686 		    &(*commit)->author_gmtoff, s);
687 		if (err)
688 			goto done;
689 		(*commit)->author = strdup(s);
690 		if ((*commit)->author == NULL) {
691 			err = got_error_from_errno("strdup");
692 			goto done;
693 		}
694 		s += slen + 1;
695 		remain -= slen + 1;
696 	}
697 
698 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
699 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
700 		char *p;
701 		size_t slen;
702 
703 		remain -= label_len;
704 		if (remain <= 0) {
705 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
706 			goto done;
707 		}
708 		s += label_len;
709 		p = memchr(s, '\n', remain);
710 		if (p == NULL) {
711 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
712 			goto done;
713 		}
714 		*p = '\0';
715 		slen = strlen(s);
716 		err = parse_commit_time(&(*commit)->committer_time,
717 		    &(*commit)->committer_gmtoff, s);
718 		if (err)
719 			goto done;
720 		(*commit)->committer = strdup(s);
721 		if ((*commit)->committer == NULL) {
722 			err = got_error_from_errno("strdup");
723 			goto done;
724 		}
725 		s += slen + 1;
726 		remain -= slen + 1;
727 	}
728 
729 	(*commit)->logmsg = strndup(s, remain);
730 	if ((*commit)->logmsg == NULL) {
731 		err = got_error_from_errno("strndup");
732 		goto done;
733 	}
734 done:
735 	if (err) {
736 		got_object_commit_close(*commit);
737 		*commit = NULL;
738 	}
739 	return err;
740 }
741 
742 const struct got_error *
743 got_object_read_commit(struct got_commit_object **commit, int fd,
744     struct got_object_id *expected_id, size_t expected_size)
745 {
746 	struct got_object *obj = NULL;
747 	const struct got_error *err = NULL;
748 	size_t len;
749 	uint8_t *p;
750 	struct got_inflate_checksum csum;
751 	SHA1_CTX sha1_ctx;
752 	struct got_object_id id;
753 
754 	SHA1Init(&sha1_ctx);
755 	memset(&csum, 0, sizeof(csum));
756 	csum.output_sha1 = &sha1_ctx;
757 
758 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
759 	if (err)
760 		return err;
761 
762 	SHA1Final(id.sha1, &sha1_ctx);
763 	if (got_object_id_cmp(expected_id, &id) != 0) {
764 		err = got_error_checksum(expected_id);
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 (got_object_id_cmp(expected_id, &id) != 0) {
933 		err = got_error_checksum(expected_id);
934 		goto done;
935 	}
936 
937 	err = got_object_parse_header(&obj, *p, len);
938 	if (err)
939 		goto done;
940 
941 	if (len < obj->hdrlen + obj->size) {
942 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
943 		goto done;
944 	}
945 
946 	/* Skip object header. */
947 	len -= obj->hdrlen;
948 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
949 	    *p + obj->hdrlen, len);
950 done:
951 	if (obj)
952 		got_object_close(obj);
953 	return err;
954 }
955 
956 void
957 got_object_tag_close(struct got_tag_object *tag)
958 {
959 	if (tag->refcnt > 0) {
960 		tag->refcnt--;
961 		if (tag->refcnt > 0)
962 			return;
963 	}
964 
965 	free(tag->tag);
966 	free(tag->tagger);
967 	free(tag->tagmsg);
968 	free(tag);
969 }
970 
971 const struct got_error *
972 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
973 {
974 	const struct got_error *err = NULL;
975 	size_t remain = len;
976 	char *s = buf;
977 	size_t label_len;
978 
979 	if (remain == 0)
980 		return got_error(GOT_ERR_BAD_OBJ_DATA);
981 
982 	*tag = calloc(1, sizeof(**tag));
983 	if (*tag == NULL)
984 		return got_error_from_errno("calloc");
985 
986 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
987 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
988 		remain -= label_len;
989 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
990 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
991 			goto done;
992 		}
993 		s += label_len;
994 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
995 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
996 			goto done;
997 		}
998 		remain -= SHA1_DIGEST_STRING_LENGTH;
999 		s += SHA1_DIGEST_STRING_LENGTH;
1000 	} else {
1001 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1002 		goto done;
1003 	}
1004 
1005 	if (remain <= 0) {
1006 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1007 		goto done;
1008 	}
1009 
1010 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1011 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1012 		remain -= label_len;
1013 		if (remain <= 0) {
1014 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1015 			goto done;
1016 		}
1017 		s += label_len;
1018 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1019 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1020 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1021 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1022 			s += label_len;
1023 			remain -= label_len;
1024 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1025 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1026 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1027 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1028 			s += label_len;
1029 			remain -= label_len;
1030 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1031 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1032 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1033 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1034 			s += label_len;
1035 			remain -= label_len;
1036 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1037 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1038 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1039 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1040 			s += label_len;
1041 			remain -= label_len;
1042 		} else {
1043 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1044 			goto done;
1045 		}
1046 
1047 		if (remain <= 0 || *s != '\n') {
1048 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1049 			goto done;
1050 		}
1051 		s++;
1052 		remain--;
1053 		if (remain <= 0) {
1054 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1055 			goto done;
1056 		}
1057 	} else {
1058 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1059 		goto done;
1060 	}
1061 
1062 	label_len = strlen(GOT_TAG_LABEL_TAG);
1063 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1064 		char *p;
1065 		size_t slen;
1066 		remain -= label_len;
1067 		if (remain <= 0) {
1068 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1069 			goto done;
1070 		}
1071 		s += label_len;
1072 		p = memchr(s, '\n', remain);
1073 		if (p == NULL) {
1074 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1075 			goto done;
1076 		}
1077 		*p = '\0';
1078 		slen = strlen(s);
1079 		(*tag)->tag = strndup(s, slen);
1080 		if ((*tag)->tag == NULL) {
1081 			err = got_error_from_errno("strndup");
1082 			goto done;
1083 		}
1084 		s += slen + 1;
1085 		remain -= slen + 1;
1086 		if (remain <= 0) {
1087 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1088 			goto done;
1089 		}
1090 	} else {
1091 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1092 		goto done;
1093 	}
1094 
1095 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1096 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1097 		char *p;
1098 		size_t slen;
1099 
1100 		remain -= label_len;
1101 		if (remain <= 0) {
1102 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1103 			goto done;
1104 		}
1105 		s += label_len;
1106 		p = memchr(s, '\n', remain);
1107 		if (p == NULL) {
1108 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1109 			goto done;
1110 		}
1111 		*p = '\0';
1112 		slen = strlen(s);
1113 		err = parse_commit_time(&(*tag)->tagger_time,
1114 		    &(*tag)->tagger_gmtoff, s);
1115 		if (err)
1116 			goto done;
1117 		(*tag)->tagger = strdup(s);
1118 		if ((*tag)->tagger == NULL) {
1119 			err = got_error_from_errno("strdup");
1120 			goto done;
1121 		}
1122 		s += slen + 1;
1123 		remain -= slen + 1;
1124 		if (remain < 0) {
1125 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1126 			goto done;
1127 		}
1128 	} else {
1129 		/* Some old tags in the Linux git repo have no tagger. */
1130 		(*tag)->tagger = strdup("");
1131 		if ((*tag)->tagger == NULL) {
1132 			err = got_error_from_errno("strdup");
1133 			goto done;
1134 		}
1135 	}
1136 
1137 	(*tag)->tagmsg = strndup(s, remain);
1138 	if ((*tag)->tagmsg == NULL) {
1139 		err = got_error_from_errno("strndup");
1140 		goto done;
1141 	}
1142 done:
1143 	if (err) {
1144 		got_object_tag_close(*tag);
1145 		*tag = NULL;
1146 	}
1147 	return err;
1148 }
1149 
1150 const struct got_error *
1151 got_object_read_tag(struct got_tag_object **tag, int fd,
1152     struct got_object_id *expected_id, size_t expected_size)
1153 {
1154 	const struct got_error *err = NULL;
1155 	struct got_object *obj = NULL;
1156 	size_t len;
1157 	uint8_t *p;
1158 	struct got_inflate_checksum csum;
1159 	SHA1_CTX sha1_ctx;
1160 	struct got_object_id id;
1161 
1162 	SHA1Init(&sha1_ctx);
1163 	memset(&csum, 0, sizeof(csum));
1164 	csum.output_sha1 = &sha1_ctx;
1165 
1166 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1167 	    expected_size, fd);
1168 	if (err)
1169 		return err;
1170 
1171 	SHA1Final(id.sha1, &sha1_ctx);
1172 	if (got_object_id_cmp(expected_id, &id) != 0) {
1173 		err = got_error_checksum(expected_id);
1174 		goto done;
1175 	}
1176 
1177 	err = got_object_parse_header(&obj, p, len);
1178 	if (err)
1179 		goto done;
1180 
1181 	if (len < obj->hdrlen + obj->size) {
1182 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1183 		goto done;
1184 	}
1185 
1186 	/* Skip object header. */
1187 	len -= obj->hdrlen;
1188 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1189 done:
1190 	free(p);
1191 	if (obj)
1192 		got_object_close(obj);
1193 	return err;
1194 }
1195 
1196 const struct got_error *
1197 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1198 {
1199 	const struct got_error *err = NULL;
1200 	static const size_t blocksize = 512;
1201 	size_t n, total, remain;
1202 	uint8_t *buf;
1203 
1204 	*outbuf = NULL;
1205 	*outlen = 0;
1206 
1207 	buf = malloc(blocksize);
1208 	if (buf == NULL)
1209 		return got_error_from_errno("malloc");
1210 
1211 	remain = blocksize;
1212 	total = 0;
1213 	for (;;) {
1214 		if (remain == 0) {
1215 			uint8_t *newbuf;
1216 			newbuf = reallocarray(buf, 1, total + blocksize);
1217 			if (newbuf == NULL) {
1218 				err = got_error_from_errno("reallocarray");
1219 				goto done;
1220 			}
1221 			buf = newbuf;
1222 			remain += blocksize;
1223 		}
1224 		n = fread(buf + total, 1, remain, f);
1225 		if (n == 0) {
1226 			if (ferror(f)) {
1227 				err = got_ferror(f, GOT_ERR_IO);
1228 				goto done;
1229 			}
1230 			break; /* EOF */
1231 		}
1232 		remain -= n;
1233 		total += n;
1234 	};
1235 
1236 done:
1237 	if (err == NULL) {
1238 		*outbuf = buf;
1239 		*outlen = total;
1240 	} else
1241 		free(buf);
1242 	return err;
1243 }