op public repos

Blob

Date:: Sat Feb 4 14:04:08 2023 UTC
Message:: fix got_object_id_hex for sha256
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/tree.h>
21 #include <sys/uio.h>
22 #include <sys/socket.h>
23 #include <sys/wait.h>
24 #include <sys/mman.h>
25 
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <stdint.h>
31 #include <sha1.h>
32 #include <sha2.h>
33 #include <zlib.h>
34 #include <ctype.h>
35 #include <limits.h>
36 #include <imsg.h>
37 #include <time.h>
38 #include <unistd.h>
39 
40 #include "got_error.h"
41 #include "got_object.h"
42 #include "got_repository.h"
43 #include "got_opentemp.h"
44 #include "got_path.h"
45 
46 #include "got_lib_hash.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_inflate.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_parse.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_repository.h"
54 
55 #ifndef nitems
56 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
57 #endif
58 
59 struct got_object_id *
60 got_object_id_dup(struct got_object_id *id1)
61 {
62 	struct got_object_id *id2;
63 
64 	id2 = malloc(sizeof(*id2));
65 	if (id2 == NULL)
66 		return NULL;
67 	memcpy(id2, id1, sizeof(*id2));
68 	return id2;
69 }
70 
71 int
72 got_object_id_cmp(const struct got_object_id *id1,
73     const struct got_object_id *id2)
74 {
75 	if (id1->algo != id2->algo)
76 		return -1;
77 
78 	if (id1->algo == GOT_HASH_SHA256)
79 		return memcmp(id1->hash, id2->hash, SHA256_DIGEST_LENGTH);
80 	if (id1->algo == GOT_HASH_SHA1)
81 		return memcmp(id1->hash, id2->hash, SHA1_DIGEST_LENGTH);
82 	return -1;
83 }
84 
85 const struct got_error *
86 got_object_qid_alloc_partial(struct got_object_qid **qid)
87 {
88 	*qid = malloc(sizeof(**qid));
89 	if (*qid == NULL)
90 		return got_error_from_errno("malloc");
91 
92 	(*qid)->data = NULL;
93 	return NULL;
94 }
95 
96 const struct got_error *
97 got_object_id_str(char **outbuf, struct got_object_id *id)
98 {
99 	static const size_t len = GOT_OBJECT_ID_HEX_MAXLEN;
100 
101 	*outbuf = malloc(len);
102 	if (*outbuf == NULL)
103 		return got_error_from_errno("malloc");
104 
105 	if (got_object_id_hex(id, *outbuf, len) == NULL) {
106 		free(*outbuf);
107 		*outbuf = NULL;
108 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
109 	}
110 
111 	return NULL;
112 }
113 
114 char *
115 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
116 {
117 	if (id->algo == GOT_HASH_SHA256)
118 		return got_sha256_digest_to_str(id->hash, buf, len);
119 	return got_sha1_digest_to_str(id->hash, buf, len);
120 }
121 
122 void
123 got_object_close(struct got_object *obj)
124 {
125 	if (obj->refcnt > 0) {
126 		obj->refcnt--;
127 		if (obj->refcnt > 0)
128 			return;
129 	}
130 
131 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
132 		struct got_delta *delta;
133 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
134 			delta = STAILQ_FIRST(&obj->deltas.entries);
135 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
136 			free(delta);
137 		}
138 	}
139 	free(obj);
140 }
141 
142 const struct got_error *
143 got_object_raw_close(struct got_raw_object *obj)
144 {
145 	const struct got_error *err = NULL;
146 
147 	if (obj->refcnt > 0) {
148 		obj->refcnt--;
149 		if (obj->refcnt > 0)
150 			return NULL;
151 	}
152 
153 	if (obj->close_cb)
154 		obj->close_cb(obj);
155 
156 	if (obj->f == NULL) {
157 		if (obj->fd != -1) {
158 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
159 				err = got_error_from_errno("munmap");
160 			if (close(obj->fd) == -1 && err == NULL)
161 				err = got_error_from_errno("close");
162 		} else
163 			free(obj->data);
164 	} else {
165 		if (fclose(obj->f) == EOF && err == NULL)
166 			err = got_error_from_errno("fclose");
167 	}
168 	free(obj);
169 	return err;
170 }
171 
172 void
173 got_object_qid_free(struct got_object_qid *qid)
174 {
175 	free(qid);
176 }
177 
178 void
179 got_object_id_queue_free(struct got_object_id_queue *ids)
180 {
181 	struct got_object_qid *qid;
182 
183 	while (!STAILQ_EMPTY(ids)) {
184 		qid = STAILQ_FIRST(ids);
185 		STAILQ_REMOVE_HEAD(ids, entry);
186 		got_object_qid_free(qid);
187 	}
188 }
189 
190 const struct got_error *
191 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
192 {
193 	const char *obj_labels[] = {
194 		GOT_OBJ_LABEL_COMMIT,
195 		GOT_OBJ_LABEL_TREE,
196 		GOT_OBJ_LABEL_BLOB,
197 		GOT_OBJ_LABEL_TAG,
198 	};
199 	const int obj_types[] = {
200 		GOT_OBJ_TYPE_COMMIT,
201 		GOT_OBJ_TYPE_TREE,
202 		GOT_OBJ_TYPE_BLOB,
203 		GOT_OBJ_TYPE_TAG,
204 	};
205 	int type = 0;
206 	size_t size = 0;
207 	size_t i;
208 	char *end;
209 
210 	*obj = NULL;
211 
212 	end = memchr(buf, '\0', len);
213 	if (end == NULL)
214 		return got_error(GOT_ERR_BAD_OBJ_HDR);
215 
216 	for (i = 0; i < nitems(obj_labels); i++) {
217 		const char *label = obj_labels[i];
218 		size_t label_len = strlen(label);
219 		const char *errstr;
220 
221 		if (len <= label_len || buf + label_len >= end ||
222 		    strncmp(buf, label, label_len) != 0)
223 			continue;
224 
225 		type = obj_types[i];
226 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
227 		if (errstr != NULL)
228 			return got_error(GOT_ERR_BAD_OBJ_HDR);
229 		break;
230 	}
231 
232 	if (type == 0)
233 		return got_error(GOT_ERR_BAD_OBJ_HDR);
234 
235 	*obj = calloc(1, sizeof(**obj));
236 	if (*obj == NULL)
237 		return got_error_from_errno("calloc");
238 	(*obj)->type = type;
239 	(*obj)->hdrlen = end - buf + 1;
240 	(*obj)->size = size;
241 	return NULL;
242 }
243 
244 const struct got_error *
245 got_object_read_header(struct got_object **obj, int fd)
246 {
247 	const struct got_error *err;
248 	struct got_inflate_buf zb;
249 	uint8_t *buf;
250 	const size_t zbsize = 64;
251 	size_t outlen, totlen;
252 	int nbuf = 1;
253 
254 	*obj = NULL;
255 
256 	buf = malloc(zbsize);
257 	if (buf == NULL)
258 		return got_error_from_errno("malloc");
259 	buf[0] = '\0';
260 
261 	err = got_inflate_init(&zb, buf, zbsize, NULL);
262 	if (err)
263 		return err;
264 
265 	totlen = 0;
266 	do {
267 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
268 		if (err)
269 			goto done;
270 		if (outlen == 0)
271 			break;
272 		totlen += outlen;
273 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
274 			uint8_t *newbuf;
275 			nbuf++;
276 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
277 			if (newbuf == NULL) {
278 				err = got_error_from_errno("recallocarray");
279 				goto done;
280 			}
281 			buf = newbuf;
282 			zb.outbuf = newbuf + totlen;
283 			zb.outlen = (nbuf * zbsize) - totlen;
284 		}
285 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
286 
287 	err = got_object_parse_header(obj, buf, totlen);
288 done:
289 	free(buf);
290 	got_inflate_end(&zb);
291 	return err;
292 }
293 
294 const struct got_error *
295 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
296     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
297     int infd)
298 {
299 	const struct got_error *err = NULL;
300 	struct got_object *obj;
301 	struct got_inflate_checksum csum;
302 	uint8_t sha256[SHA256_DIGEST_LENGTH];
303 	SHA2_CTX sha256_ctx;
304 	uint8_t sha1[SHA1_DIGEST_LENGTH];
305 	SHA1_CTX sha1_ctx;
306 	size_t len, consumed;
307 	FILE *f = NULL;
308 	int r;
309 
310 	*outbuf = NULL;
311 	*size = 0;
312 	*hdrlen = 0;
313 
314 	memset(&csum, 0, sizeof(csum));
315 	if (expected_id->algo == GOT_HASH_SHA256) {
316 		SHA256Init(&sha256_ctx);
317 		csum.output_sha256 = &sha256_ctx;
318 	} else {
319 		SHA1Init(&sha1_ctx);
320 		csum.output_sha1 = &sha1_ctx;
321 	}
322 
323 	if (lseek(infd, SEEK_SET, 0) == -1)
324 		return got_error_from_errno("lseek");
325 
326 	err = got_object_read_header(&obj, infd);
327 	if (err)
328 		return err;
329 
330 	if (lseek(infd, SEEK_SET, 0) == -1)
331 		return got_error_from_errno("lseek");
332 
333 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
334 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
335 		    obj->size + obj->hdrlen, infd);
336 	} else {
337 		int fd;
338 		/*
339 		 * XXX This uses an extra file descriptor for no good reason.
340 		 * We should have got_inflate_fd_to_fd().
341 		 */
342 		fd = dup(infd);
343 		if (fd == -1)
344 			return got_error_from_errno("dup");
345 		f = fdopen(fd, "r");
346 		if (f == NULL) {
347 			err = got_error_from_errno("fdopen");
348 			abort();
349 			close(fd);
350 			goto done;
351 		}
352 		err = got_inflate_to_fd(&len, f, &csum, outfd);
353 	}
354 	if (err)
355 		goto done;
356 
357 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
358 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
359 		goto done;
360 	}
361 
362 	if (expected_id->algo == GOT_HASH_SHA256) {
363 		SHA256Final(sha256, &sha256_ctx);
364 		r = memcmp(expected_id->hash, sha256, sizeof(sha256));
365 	} else {
366 		SHA1Final(sha1, &sha1_ctx);
367 		r = memcmp(expected_id->hash, sha1, sizeof(sha1));
368 	}
369 
370 	if (r != 0) {
371 		err = got_error_checksum(expected_id);
372 		goto done;
373 	}
374 
375 	*size = obj->size;
376 	*hdrlen = obj->hdrlen;
377 done:
378 	got_object_close(obj);
379 	if (f && fclose(f) == EOF && err == NULL)
380 		err = got_error_from_errno("fclose");
381 	return err;
382 }
383 
384 struct got_commit_object *
385 got_object_commit_alloc_partial(void)
386 {
387 	struct got_commit_object *commit;
388 
389 	commit = calloc(1, sizeof(*commit));
390 	if (commit == NULL)
391 		return NULL;
392 	commit->tree_id = malloc(sizeof(*commit->tree_id));
393 	if (commit->tree_id == NULL) {
394 		free(commit);
395 		return NULL;
396 	}
397 
398 	STAILQ_INIT(&commit->parent_ids);
399 
400 	return commit;
401 }
402 
403 const struct got_error *
404 got_object_commit_add_parent(struct got_commit_object *commit,
405     const char *id_str, int algo)
406 {
407 	const struct got_error *err = NULL;
408 	struct got_object_qid *qid;
409 
410 	err = got_object_qid_alloc_partial(&qid);
411 	if (err)
412 		return err;
413 
414 	if (!got_parse_hash_digest(qid->id.hash, id_str, algo, NULL)) {
415 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
416 		got_object_qid_free(qid);
417 		return err;
418 	}
419 	qid->id.algo = algo;
420 
421 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
422 	commit->nparents++;
423 
424 	return NULL;
425 }
426 
427 static const struct got_error *
428 parse_gmtoff(time_t *gmtoff, const char *tzstr)
429 {
430 	int sign = 1;
431 	const char *p = tzstr;
432 	time_t h, m;
433 
434 	*gmtoff = 0;
435 
436 	if (*p == '-')
437 		sign = -1;
438 	else if (*p != '+')
439 		return got_error(GOT_ERR_BAD_OBJ_DATA);
440 	p++;
441 	if (!isdigit((unsigned char)*p) &&
442 	    !isdigit((unsigned char)*(p + 1)))
443 		return got_error(GOT_ERR_BAD_OBJ_DATA);
444 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
445 
446 	p += 2;
447 	if (!isdigit((unsigned char)*p) &&
448 	    !isdigit((unsigned char)*(p + 1)))
449 		return got_error(GOT_ERR_BAD_OBJ_DATA);
450 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
451 
452 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
453 	return NULL;
454 }
455 
456 static const struct got_error *
457 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
458 {
459 	const struct got_error *err = NULL;
460 	const char *errstr;
461 	char *space, *tzstr;
462 
463 	/* Parse and strip off trailing timezone indicator string. */
464 	space = strrchr(committer, ' ');
465 	if (space == NULL)
466 		return got_error(GOT_ERR_BAD_OBJ_DATA);
467 	tzstr = strdup(space + 1);
468 	if (tzstr == NULL)
469 		return got_error_from_errno("strdup");
470 	err = parse_gmtoff(gmtoff, tzstr);
471 	free(tzstr);
472 	if (err) {
473 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
474 			return err;
475 		/* Old versions of Git omitted the timestamp. */
476 		*time = 0;
477 		*gmtoff = 0;
478 		return NULL;
479 	}
480 	*space = '\0';
481 
482 	/* Timestamp is separated from committer name + email by space. */
483 	space = strrchr(committer, ' ');
484 	if (space == NULL)
485 		return got_error(GOT_ERR_BAD_OBJ_DATA);
486 
487 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
488 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
489 	if (errstr)
490 		return got_error(GOT_ERR_BAD_OBJ_DATA);
491 
492 	/* Strip off parsed time information, leaving just author and email. */
493 	*space = '\0';
494 
495 	return NULL;
496 }
497 
498 void
499 got_object_commit_close(struct got_commit_object *commit)
500 {
501 	if (commit->refcnt > 0) {
502 		commit->refcnt--;
503 		if (commit->refcnt > 0)
504 			return;
505 	}
506 
507 	got_object_id_queue_free(&commit->parent_ids);
508 	free(commit->tree_id);
509 	free(commit->author);
510 	free(commit->committer);
511 	free(commit->logmsg);
512 	free(commit);
513 }
514 
515 struct got_object_id *
516 got_object_commit_get_tree_id(struct got_commit_object *commit)
517 {
518 	return commit->tree_id;
519 }
520 
521 int
522 got_object_commit_get_nparents(struct got_commit_object *commit)
523 {
524 	return commit->nparents;
525 }
526 
527 const struct got_object_id_queue *
528 got_object_commit_get_parent_ids(struct got_commit_object *commit)
529 {
530 	return &commit->parent_ids;
531 }
532 
533 const char *
534 got_object_commit_get_author(struct got_commit_object *commit)
535 {
536 	return commit->author;
537 }
538 
539 time_t
540 got_object_commit_get_author_time(struct got_commit_object *commit)
541 {
542 	return commit->author_time;
543 }
544 
545 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
546 {
547 	return commit->author_gmtoff;
548 }
549 
550 const char *
551 got_object_commit_get_committer(struct got_commit_object *commit)
552 {
553 	return commit->committer;
554 }
555 
556 time_t
557 got_object_commit_get_committer_time(struct got_commit_object *commit)
558 {
559 	return commit->committer_time;
560 }
561 
562 time_t
563 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
564 {
565 	return commit->committer_gmtoff;
566 }
567 
568 const struct got_error *
569 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
570 {
571 	const struct got_error *err = NULL;
572 	const char *src;
573 	char *dst;
574 	size_t len;
575 
576 	len = strlen(commit->logmsg);
577 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
578 	if (*logmsg == NULL)
579 		return got_error_from_errno("malloc");
580 
581 	/*
582 	 * Strip out unusual headers. Headers are separated from the commit
583 	 * message body by a single empty line.
584 	 */
585 	src = commit->logmsg;
586 	dst = *logmsg;
587 	while (*src != '\0' && *src != '\n') {
588 		int copy_header = 1, eol = 0;
589 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
590 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
591 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
592 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
593 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
594 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
595 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
596 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
597 			copy_header = 0;
598 
599 		while (*src != '\0' && !eol) {
600 			if (copy_header) {
601 				*dst = *src;
602 				dst++;
603 			}
604 			if (*src == '\n')
605 				eol = 1;
606 			src++;
607 		}
608 	}
609 	*dst = '\0';
610 
611 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
612 		err = got_error(GOT_ERR_NO_SPACE);
613 		goto done;
614 	}
615 
616 	/* Trim redundant trailing whitespace. */
617 	len = strlen(*logmsg);
618 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
619 	    isspace((unsigned char)(*logmsg)[len - 1])) {
620 		(*logmsg)[len - 1] = '\0';
621 		len--;
622 	}
623 
624 	/* Append a trailing newline if missing. */
625 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
626 		(*logmsg)[len] = '\n';
627 		(*logmsg)[len + 1] = '\0';
628 	}
629 done:
630 	if (err) {
631 		free(*logmsg);
632 		*logmsg = NULL;
633 	}
634 	return err;
635 }
636 
637 const char *
638 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
639 {
640 	return commit->logmsg;
641 }
642 
643 const struct got_error *
644 got_object_parse_commit(struct got_commit_object **commit, char *buf,
645     size_t len, int algo)
646 {
647 	const struct got_error *err = NULL;
648 	char *s = buf;
649 	size_t label_len;
650 	size_t digest_strlen;
651 	ssize_t remain = (ssize_t)len;
652 
653 	if (algo == GOT_HASH_SHA256)
654 		digest_strlen = SHA256_DIGEST_STRING_LENGTH;
655 	else
656 		digest_strlen = SHA1_DIGEST_STRING_LENGTH;
657 
658 	if (remain == 0)
659 		return got_error(GOT_ERR_BAD_OBJ_DATA);
660 
661 	*commit = got_object_commit_alloc_partial();
662 	if (*commit == NULL)
663 		return got_error_from_errno("got_object_commit_alloc_partial");
664 
665 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
666 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
667 		remain -= label_len;
668 		if (remain < digest_strlen) {
669 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
670 			goto done;
671 		}
672 		s += label_len;
673 		if (!got_parse_hash_digest((*commit)->tree_id->hash, s,
674 		    algo, NULL)) {
675 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
676 			goto done;
677 		}
678 		(*commit)->tree_id->algo = algo;
679 		remain -= digest_strlen;
680 		s += digest_strlen;
681 	} else {
682 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
683 		goto done;
684 	}
685 
686 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
687 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
688 		remain -= label_len;
689 		if (remain < digest_strlen) {
690 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
691 			goto done;
692 		}
693 		s += label_len;
694 		err = got_object_commit_add_parent(*commit, s, algo);
695 		if (err)
696 			goto done;
697 
698 		remain -= digest_strlen;
699 		s += digest_strlen;
700 	}
701 
702 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
703 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
704 		char *p;
705 		size_t slen;
706 
707 		remain -= label_len;
708 		if (remain <= 0) {
709 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
710 			goto done;
711 		}
712 		s += label_len;
713 		p = memchr(s, '\n', remain);
714 		if (p == NULL) {
715 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
716 			goto done;
717 		}
718 		*p = '\0';
719 		slen = strlen(s);
720 		err = parse_commit_time(&(*commit)->author_time,
721 		    &(*commit)->author_gmtoff, s);
722 		if (err)
723 			goto done;
724 		(*commit)->author = strdup(s);
725 		if ((*commit)->author == NULL) {
726 			err = got_error_from_errno("strdup");
727 			goto done;
728 		}
729 		s += slen + 1;
730 		remain -= slen + 1;
731 	}
732 
733 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
734 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
735 		char *p;
736 		size_t slen;
737 
738 		remain -= label_len;
739 		if (remain <= 0) {
740 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
741 			goto done;
742 		}
743 		s += label_len;
744 		p = memchr(s, '\n', remain);
745 		if (p == NULL) {
746 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
747 			goto done;
748 		}
749 		*p = '\0';
750 		slen = strlen(s);
751 		err = parse_commit_time(&(*commit)->committer_time,
752 		    &(*commit)->committer_gmtoff, s);
753 		if (err)
754 			goto done;
755 		(*commit)->committer = strdup(s);
756 		if ((*commit)->committer == NULL) {
757 			err = got_error_from_errno("strdup");
758 			goto done;
759 		}
760 		s += slen + 1;
761 		remain -= slen + 1;
762 	}
763 
764 	(*commit)->logmsg = strndup(s, remain);
765 	if ((*commit)->logmsg == NULL) {
766 		err = got_error_from_errno("strndup");
767 		goto done;
768 	}
769 done:
770 	if (err) {
771 		got_object_commit_close(*commit);
772 		*commit = NULL;
773 	}
774 	return err;
775 }
776 
777 const struct got_error *
778 got_object_read_commit(struct got_commit_object **commit, int fd,
779     struct got_object_id *expected_id, size_t expected_size)
780 {
781 	struct got_object *obj = NULL;
782 	const struct got_error *err = NULL;
783 	size_t len;
784 	uint8_t *p;
785 	struct got_inflate_checksum csum;
786 	SHA1_CTX sha1_ctx;
787 	SHA2_CTX sha256_ctx;
788 	struct got_object_id id;
789 
790 	memset(&csum, 0, sizeof(csum));
791 	id.algo = expected_id->algo;
792 	if (expected_id->algo == GOT_HASH_SHA256) {
793 		SHA256Init(&sha256_ctx);
794 		csum.output_sha256 = &sha256_ctx;
795 	} else {
796 		SHA1Init(&sha1_ctx);
797 		csum.output_sha1 = &sha1_ctx;
798 	}
799 
800 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
801 	if (err)
802 		return err;
803 
804 	if (expected_id->algo == GOT_HASH_SHA256)
805 		SHA256Final(id.hash, &sha256_ctx);
806 	else
807 		SHA1Final(id.hash, &sha1_ctx);
808 
809 	if (got_object_id_cmp(expected_id, &id) != 0) {
810 		err = got_error_checksum(expected_id);
811 		goto done;
812 	}
813 
814 	err = got_object_parse_header(&obj, p, len);
815 	if (err)
816 		goto done;
817 
818 	if (len < obj->hdrlen + obj->size) {
819 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
820 		goto done;
821 	}
822 
823 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
824 		err = got_error(GOT_ERR_OBJ_TYPE);
825 		goto done;
826 	}
827 
828 	/* Skip object header. */
829 	len -= obj->hdrlen;
830 	err = got_object_parse_commit(commit, p + obj->hdrlen, len,
831 	    expected_id->algo);
832 done:
833 	free(p);
834 	if (obj)
835 		got_object_close(obj);
836 	return err;
837 }
838 
839 void
840 got_object_tree_close(struct got_tree_object *tree)
841 {
842 	if (tree->refcnt > 0) {
843 		tree->refcnt--;
844 		if (tree->refcnt > 0)
845 			return;
846 	}
847 
848 	free(tree->entries);
849 	free(tree);
850 }
851 
852 static const struct got_error *
853 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
854     size_t maxlen, size_t idlen)
855 {
856 	char *p, *space;
857 
858 	*elen = 0;
859 
860 	*elen = strnlen(buf, maxlen) + 1;
861 	if (*elen > maxlen)
862 		return got_error(GOT_ERR_BAD_OBJ_DATA);
863 
864 	space = memchr(buf, ' ', *elen);
865 	if (space == NULL || space <= buf)
866 		return got_error(GOT_ERR_BAD_OBJ_DATA);
867 
868 	pte->mode = 0;
869 	p = buf;
870 	while (p < space) {
871 		if (*p < '0' || *p > '7')
872 			return got_error(GOT_ERR_BAD_OBJ_DATA);
873 		pte->mode <<= 3;
874 		pte->mode |= *p - '0';
875 		p++;
876 	}
877 
878 	if (*elen > maxlen || maxlen - *elen < idlen)
879 		return got_error(GOT_ERR_BAD_OBJ_DATA);
880 
881 	pte->name = space + 1;
882 	pte->namelen = strlen(pte->name);
883 	buf += *elen;
884 	pte->id = buf;
885 	pte->idlen = idlen;
886 	*elen += idlen;
887 	return NULL;
888 }
889 
890 static int
891 pte_cmp(const void *pa, const void *pb)
892 {
893 	const struct got_parsed_tree_entry *a = pa, *b = pb;
894 
895 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
896 }
897 
898 const struct got_error *
899 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
900     size_t *nentries_alloc, uint8_t *buf, size_t len, int algo)
901 {
902 	const struct got_error *err = NULL;
903 	size_t remain = len;
904 	const size_t nalloc = 16;
905 	struct got_parsed_tree_entry *pte;
906 	size_t idlen = SHA256_DIGEST_LENGTH;
907 	int i;
908 
909 	if (algo != GOT_HASH_SHA256)
910 		idlen = SHA1_DIGEST_LENGTH;
911 
912 	*nentries = 0;
913 	if (remain == 0)
914 		return NULL; /* tree is empty */
915 
916 	while (remain > 0) {
917 		size_t elen;
918 
919 		if (*nentries >= *nentries_alloc) {
920 			pte = recallocarray(*entries, *nentries_alloc,
921 			    *nentries_alloc + nalloc, sizeof(**entries));
922 			if (pte == NULL) {
923 				err = got_error_from_errno("recallocarray");
924 				goto done;
925 			}
926 			*entries = pte;
927 			*nentries_alloc += nalloc;
928 		}
929 
930 		pte = &(*entries)[*nentries];
931 		err = parse_tree_entry(pte, &elen, buf, remain, idlen);
932 		if (err)
933 			goto done;
934 		buf += elen;
935 		remain -= elen;
936 		(*nentries)++;
937 	}
938 
939 	if (remain != 0) {
940 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
941 		goto done;
942 	}
943 
944 	if (*nentries > 1) {
945 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
946 
947 		for (i = 0; i < *nentries - 1; i++) {
948 			struct got_parsed_tree_entry *prev = &(*entries)[i];
949 			pte = &(*entries)[i + 1];
950 			if (got_path_cmp(prev->name, pte->name,
951 			    prev->namelen, pte->namelen) == 0) {
952 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
953 				break;
954 			}
955 		}
956 	}
957 done:
958 	if (err)
959 		*nentries = 0;
960 	return err;
961 }
962 
963 const struct got_error *
964 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
965     size_t *nentries_alloc, uint8_t **p, int fd,
966     struct got_object_id *expected_id)
967 {
968 	const struct got_error *err = NULL;
969 	struct got_object *obj = NULL;
970 	size_t len;
971 	struct got_inflate_checksum csum;
972 	SHA2_CTX sha256_ctx;
973 	SHA1_CTX sha1_ctx;
974 	struct got_object_id id;
975 
976 	memset(&csum, 0, sizeof(csum));
977 	id.algo = expected_id->algo;
978 	if (expected_id->algo == GOT_HASH_SHA256) {
979 		SHA256Init(&sha256_ctx);
980 		csum.output_sha256 = &sha256_ctx;
981 	} else {
982 		SHA1Init(&sha1_ctx);
983 		csum.output_sha1 = &sha1_ctx;
984 	}
985 
986 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
987 	if (err)
988 		return err;
989 
990 	if (expected_id->algo == GOT_HASH_SHA256)
991 		SHA256Final(id.hash, &sha256_ctx);
992 	else
993 		SHA1Final(id.hash, &sha1_ctx);
994 	if (got_object_id_cmp(expected_id, &id) != 0) {
995 		err = got_error_checksum(expected_id);
996 		goto done;
997 	}
998 
999 	err = got_object_parse_header(&obj, *p, len);
1000 	if (err)
1001 		goto done;
1002 
1003 	if (len < obj->hdrlen + obj->size) {
1004 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1005 		goto done;
1006 	}
1007 
1008 	/* Skip object header. */
1009 	len -= obj->hdrlen;
1010 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
1011 	    *p + obj->hdrlen, len, expected_id->algo);
1012 done:
1013 	if (obj)
1014 		got_object_close(obj);
1015 	return err;
1016 }
1017 
1018 void
1019 got_object_tag_close(struct got_tag_object *tag)
1020 {
1021 	if (tag->refcnt > 0) {
1022 		tag->refcnt--;
1023 		if (tag->refcnt > 0)
1024 			return;
1025 	}
1026 
1027 	free(tag->tag);
1028 	free(tag->tagger);
1029 	free(tag->tagmsg);
1030 	free(tag);
1031 }
1032 
1033 const struct got_error *
1034 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len,
1035     int algo)
1036 {
1037 	const struct got_error *err = NULL;
1038 	size_t remain = len;
1039 	char *s = buf;
1040 	size_t label_len, digest_strlen = SHA256_DIGEST_STRING_LENGTH;
1041 
1042 	if (algo != GOT_HASH_SHA256)
1043 		digest_strlen = SHA1_DIGEST_STRING_LENGTH;
1044 
1045 	if (remain == 0)
1046 		return got_error(GOT_ERR_BAD_OBJ_DATA);
1047 
1048 	*tag = calloc(1, sizeof(**tag));
1049 	if (*tag == NULL)
1050 		return got_error_from_errno("calloc");
1051 
1052 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
1053 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
1054 		remain -= label_len;
1055 		if (remain < digest_strlen) {
1056 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1057 			goto done;
1058 		}
1059 		s += label_len;
1060 		if (!got_parse_hash_digest((*tag)->id.hash, s, algo, NULL)) {
1061 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1062 			goto done;
1063 		}
1064 		(*tag)->id.algo = algo;
1065 		remain -= digest_strlen;
1066 		s += digest_strlen;
1067 	} else {
1068 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1069 		goto done;
1070 	}
1071 
1072 	if (remain <= 0) {
1073 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1074 		goto done;
1075 	}
1076 
1077 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1078 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1079 		remain -= label_len;
1080 		if (remain <= 0) {
1081 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1082 			goto done;
1083 		}
1084 		s += label_len;
1085 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1086 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1087 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1088 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1089 			s += label_len;
1090 			remain -= label_len;
1091 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1092 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1093 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1094 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1095 			s += label_len;
1096 			remain -= label_len;
1097 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1098 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1099 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1100 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1101 			s += label_len;
1102 			remain -= label_len;
1103 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1104 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1105 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1106 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1107 			s += label_len;
1108 			remain -= label_len;
1109 		} else {
1110 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1111 			goto done;
1112 		}
1113 
1114 		if (remain <= 0 || *s != '\n') {
1115 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1116 			goto done;
1117 		}
1118 		s++;
1119 		remain--;
1120 		if (remain <= 0) {
1121 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1122 			goto done;
1123 		}
1124 	} else {
1125 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1126 		goto done;
1127 	}
1128 
1129 	label_len = strlen(GOT_TAG_LABEL_TAG);
1130 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1131 		char *p;
1132 		size_t slen;
1133 		remain -= label_len;
1134 		if (remain <= 0) {
1135 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1136 			goto done;
1137 		}
1138 		s += label_len;
1139 		p = memchr(s, '\n', remain);
1140 		if (p == NULL) {
1141 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1142 			goto done;
1143 		}
1144 		*p = '\0';
1145 		slen = strlen(s);
1146 		(*tag)->tag = strndup(s, slen);
1147 		if ((*tag)->tag == NULL) {
1148 			err = got_error_from_errno("strndup");
1149 			goto done;
1150 		}
1151 		s += slen + 1;
1152 		remain -= slen + 1;
1153 		if (remain <= 0) {
1154 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1155 			goto done;
1156 		}
1157 	} else {
1158 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1159 		goto done;
1160 	}
1161 
1162 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1163 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1164 		char *p;
1165 		size_t slen;
1166 
1167 		remain -= label_len;
1168 		if (remain <= 0) {
1169 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1170 			goto done;
1171 		}
1172 		s += label_len;
1173 		p = memchr(s, '\n', remain);
1174 		if (p == NULL) {
1175 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1176 			goto done;
1177 		}
1178 		*p = '\0';
1179 		slen = strlen(s);
1180 		err = parse_commit_time(&(*tag)->tagger_time,
1181 		    &(*tag)->tagger_gmtoff, s);
1182 		if (err)
1183 			goto done;
1184 		(*tag)->tagger = strdup(s);
1185 		if ((*tag)->tagger == NULL) {
1186 			err = got_error_from_errno("strdup");
1187 			goto done;
1188 		}
1189 		s += slen + 1;
1190 		remain -= slen + 1;
1191 		if (remain < 0) {
1192 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1193 			goto done;
1194 		}
1195 	} else {
1196 		/* Some old tags in the Linux git repo have no tagger. */
1197 		(*tag)->tagger = strdup("");
1198 		if ((*tag)->tagger == NULL) {
1199 			err = got_error_from_errno("strdup");
1200 			goto done;
1201 		}
1202 	}
1203 
1204 	(*tag)->tagmsg = strndup(s, remain);
1205 	if ((*tag)->tagmsg == NULL) {
1206 		err = got_error_from_errno("strndup");
1207 		goto done;
1208 	}
1209 done:
1210 	if (err) {
1211 		got_object_tag_close(*tag);
1212 		*tag = NULL;
1213 	}
1214 	return err;
1215 }
1216 
1217 const struct got_error *
1218 got_object_read_tag(struct got_tag_object **tag, int fd,
1219     struct got_object_id *expected_id, size_t expected_size)
1220 {
1221 	const struct got_error *err = NULL;
1222 	struct got_object *obj = NULL;
1223 	size_t len;
1224 	uint8_t *p;
1225 	struct got_inflate_checksum csum;
1226 	SHA2_CTX sha256_ctx;
1227 	SHA1_CTX sha1_ctx;
1228 	struct got_object_id id;
1229 
1230 	memset(&csum, 0, sizeof(csum));
1231 	id.algo = expected_id->algo;
1232 	if (expected_id->algo == GOT_HASH_SHA256) {
1233 		SHA256Init(&sha256_ctx);
1234 		csum.output_sha256 = &sha256_ctx;
1235 	} else {
1236 		SHA1Init(&sha1_ctx);
1237 		csum.output_sha1 = &sha1_ctx;
1238 	}
1239 
1240 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1241 	    expected_size, fd);
1242 	if (err)
1243 		return err;
1244 
1245 	if (expected_id->algo == GOT_HASH_SHA256)
1246 		SHA256Final(id.hash, &sha256_ctx);
1247 	else
1248 		SHA1Final(id.hash, &sha1_ctx);
1249 	if (got_object_id_cmp(expected_id, &id) != 0) {
1250 		err = got_error_checksum(expected_id);
1251 		goto done;
1252 	}
1253 
1254 	err = got_object_parse_header(&obj, p, len);
1255 	if (err)
1256 		goto done;
1257 
1258 	if (len < obj->hdrlen + obj->size) {
1259 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1260 		goto done;
1261 	}
1262 
1263 	/* Skip object header. */
1264 	len -= obj->hdrlen;
1265 	err = got_object_parse_tag(tag, p + obj->hdrlen, len, expected_id->algo);
1266 done:
1267 	free(p);
1268 	if (obj)
1269 		got_object_close(obj);
1270 	return err;
1271 }
1272 
1273 const struct got_error *
1274 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1275 {
1276 	const struct got_error *err = NULL;
1277 	static const size_t blocksize = 512;
1278 	size_t n, total, remain;
1279 	uint8_t *buf;
1280 
1281 	*outbuf = NULL;
1282 	*outlen = 0;
1283 
1284 	buf = malloc(blocksize);
1285 	if (buf == NULL)
1286 		return got_error_from_errno("malloc");
1287 
1288 	remain = blocksize;
1289 	total = 0;
1290 	for (;;) {
1291 		if (remain == 0) {
1292 			uint8_t *newbuf;
1293 			newbuf = reallocarray(buf, 1, total + blocksize);
1294 			if (newbuf == NULL) {
1295 				err = got_error_from_errno("reallocarray");
1296 				goto done;
1297 			}
1298 			buf = newbuf;
1299 			remain += blocksize;
1300 		}
1301 		n = fread(buf + total, 1, remain, f);
1302 		if (n == 0) {
1303 			if (ferror(f)) {
1304 				err = got_ferror(f, GOT_ERR_IO);
1305 				goto done;
1306 			}
1307 			break; /* EOF */
1308 		}
1309 		remain -= n;
1310 		total += n;
1311 	};
1312 
1313 done:
1314 	if (err == NULL) {
1315 		*outbuf = buf;
1316 		*outlen = total;
1317 	} else
1318 		free(buf);
1319 	return err;
1320 }