Gotweb

Blob

Date:: Thu Feb 23 17:09:15 2023 UTC
Message:: portable: remove sha1.h; found portably Remove sha1.h as this is found portably across systems.
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 <sha2.h>
31 #include <zlib.h>
32 #include <ctype.h>
33 #include <limits.h>
34 #include <time.h>
35 #include <unistd.h>
36 
37 #include "got_compat.h"
38 
39 #include "got_error.h"
40 #include "got_object.h"
41 #include "got_repository.h"
42 #include "got_opentemp.h"
43 #include "got_path.h"
44 
45 #include "got_lib_hash.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_inflate.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_repository.h"
53 
54 #ifndef nitems
55 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
56 #endif
57 
58 struct got_object_id *
59 got_object_id_dup(struct got_object_id *id1)
60 {
61 	struct got_object_id *id2;
62 
63 	id2 = malloc(sizeof(*id2));
64 	if (id2 == NULL)
65 		return NULL;
66 	memcpy(id2, id1, sizeof(*id2));
67 	return id2;
68 }
69 
70 int
71 got_object_id_cmp(const struct got_object_id *id1,
72     const struct got_object_id *id2)
73 {
74 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
75 }
76 
77 const struct got_error *
78 got_object_qid_alloc_partial(struct got_object_qid **qid)
79 {
80 	*qid = malloc(sizeof(**qid));
81 	if (*qid == NULL)
82 		return got_error_from_errno("malloc");
83 
84 	(*qid)->data = NULL;
85 	return NULL;
86 }
87 
88 const struct got_error *
89 got_object_id_str(char **outbuf, struct got_object_id *id)
90 {
91 	static const size_t len = GOT_OBJECT_ID_HEX_MAXLEN;
92 
93 	*outbuf = malloc(len);
94 	if (*outbuf == NULL)
95 		return got_error_from_errno("malloc");
96 
97 	if (got_object_id_hex(id, *outbuf, len) == NULL) {
98 		free(*outbuf);
99 		*outbuf = NULL;
100 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
101 	}
102 
103 	return NULL;
104 }
105 
106 char *
107 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
108 {
109 	return got_sha1_digest_to_str(id->sha1, buf, len);
110 }
111 
112 void
113 got_object_close(struct got_object *obj)
114 {
115 	if (obj->refcnt > 0) {
116 		obj->refcnt--;
117 		if (obj->refcnt > 0)
118 			return;
119 	}
120 
121 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
122 		struct got_delta *delta;
123 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
124 			delta = STAILQ_FIRST(&obj->deltas.entries);
125 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
126 			free(delta);
127 		}
128 	}
129 	free(obj);
130 }
131 
132 const struct got_error *
133 got_object_raw_close(struct got_raw_object *obj)
134 {
135 	const struct got_error *err = NULL;
136 
137 	if (obj->refcnt > 0) {
138 		obj->refcnt--;
139 		if (obj->refcnt > 0)
140 			return NULL;
141 	}
142 
143 	if (obj->close_cb)
144 		obj->close_cb(obj);
145 
146 	if (obj->f == NULL) {
147 		if (obj->fd != -1) {
148 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
149 				err = got_error_from_errno("munmap");
150 			if (close(obj->fd) == -1 && err == NULL)
151 				err = got_error_from_errno("close");
152 		} else
153 			free(obj->data);
154 	} else {
155 		if (fclose(obj->f) == EOF && err == NULL)
156 			err = got_error_from_errno("fclose");
157 	}
158 	free(obj);
159 	return err;
160 }
161 
162 void
163 got_object_qid_free(struct got_object_qid *qid)
164 {
165 	free(qid);
166 }
167 
168 void
169 got_object_id_queue_free(struct got_object_id_queue *ids)
170 {
171 	struct got_object_qid *qid;
172 
173 	while (!STAILQ_EMPTY(ids)) {
174 		qid = STAILQ_FIRST(ids);
175 		STAILQ_REMOVE_HEAD(ids, entry);
176 		got_object_qid_free(qid);
177 	}
178 }
179 
180 const struct got_error *
181 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
182 {
183 	const char *obj_labels[] = {
184 		GOT_OBJ_LABEL_COMMIT,
185 		GOT_OBJ_LABEL_TREE,
186 		GOT_OBJ_LABEL_BLOB,
187 		GOT_OBJ_LABEL_TAG,
188 	};
189 	const int obj_types[] = {
190 		GOT_OBJ_TYPE_COMMIT,
191 		GOT_OBJ_TYPE_TREE,
192 		GOT_OBJ_TYPE_BLOB,
193 		GOT_OBJ_TYPE_TAG,
194 	};
195 	int type = 0;
196 	size_t size = 0;
197 	size_t i;
198 	char *end;
199 
200 	*obj = NULL;
201 
202 	end = memchr(buf, '\0', len);
203 	if (end == NULL)
204 		return got_error(GOT_ERR_BAD_OBJ_HDR);
205 
206 	for (i = 0; i < nitems(obj_labels); i++) {
207 		const char *label = obj_labels[i];
208 		size_t label_len = strlen(label);
209 		const char *errstr;
210 
211 		if (len <= label_len || buf + label_len >= end ||
212 		    strncmp(buf, label, label_len) != 0)
213 			continue;
214 
215 		type = obj_types[i];
216 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
217 		if (errstr != NULL)
218 			return got_error(GOT_ERR_BAD_OBJ_HDR);
219 		break;
220 	}
221 
222 	if (type == 0)
223 		return got_error(GOT_ERR_BAD_OBJ_HDR);
224 
225 	*obj = calloc(1, sizeof(**obj));
226 	if (*obj == NULL)
227 		return got_error_from_errno("calloc");
228 	(*obj)->type = type;
229 	(*obj)->hdrlen = end - buf + 1;
230 	(*obj)->size = size;
231 	return NULL;
232 }
233 
234 const struct got_error *
235 got_object_read_header(struct got_object **obj, int fd)
236 {
237 	const struct got_error *err;
238 	struct got_inflate_buf zb;
239 	uint8_t *buf;
240 	const size_t zbsize = 64;
241 	size_t outlen, totlen;
242 	int nbuf = 1;
243 
244 	*obj = NULL;
245 
246 	buf = malloc(zbsize);
247 	if (buf == NULL)
248 		return got_error_from_errno("malloc");
249 	buf[0] = '\0';
250 
251 	err = got_inflate_init(&zb, buf, zbsize, NULL);
252 	if (err)
253 		return err;
254 
255 	totlen = 0;
256 	do {
257 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
258 		if (err)
259 			goto done;
260 		if (outlen == 0)
261 			break;
262 		totlen += outlen;
263 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
264 			uint8_t *newbuf;
265 			nbuf++;
266 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
267 			if (newbuf == NULL) {
268 				err = got_error_from_errno("recallocarray");
269 				goto done;
270 			}
271 			buf = newbuf;
272 			zb.outbuf = newbuf + totlen;
273 			zb.outlen = (nbuf * zbsize) - totlen;
274 		}
275 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
276 
277 	err = got_object_parse_header(obj, buf, totlen);
278 done:
279 	free(buf);
280 	got_inflate_end(&zb);
281 	return err;
282 }
283 
284 const struct got_error *
285 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
286     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
287     int infd)
288 {
289 	const struct got_error *err = NULL;
290 	struct got_object *obj;
291 	struct got_inflate_checksum csum;
292 	uint8_t sha1[SHA1_DIGEST_LENGTH];
293 	SHA1_CTX sha1_ctx;
294 	size_t len, consumed;
295 	FILE *f = NULL;
296 
297 	*outbuf = NULL;
298 	*size = 0;
299 	*hdrlen = 0;
300 
301 	SHA1Init(&sha1_ctx);
302 	memset(&csum, 0, sizeof(csum));
303 	csum.output_sha1 = &sha1_ctx;
304 
305 	if (lseek(infd, SEEK_SET, 0) == -1)
306 		return got_error_from_errno("lseek");
307 
308 	err = got_object_read_header(&obj, infd);
309 	if (err)
310 		return err;
311 
312 	if (lseek(infd, SEEK_SET, 0) == -1)
313 		return got_error_from_errno("lseek");
314 
315 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
316 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
317 		    obj->size + obj->hdrlen, infd);
318 	} else {
319 		int fd;
320 		/*
321 		 * XXX This uses an extra file descriptor for no good reason.
322 		 * We should have got_inflate_fd_to_fd().
323 		 */
324 		fd = dup(infd);
325 		if (fd == -1)
326 			return got_error_from_errno("dup");
327 		f = fdopen(fd, "r");
328 		if (f == NULL) {
329 			err = got_error_from_errno("fdopen");
330 			abort();
331 			close(fd);
332 			goto done;
333 		}
334 		err = got_inflate_to_fd(&len, f, &csum, outfd);
335 	}
336 	if (err)
337 		goto done;
338 
339 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
340 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
341 		goto done;
342 	}
343 
344 	SHA1Final(sha1, &sha1_ctx);
345 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
346 		err = got_error_checksum(expected_id);
347 		goto done;
348 	}
349 
350 	*size = obj->size;
351 	*hdrlen = obj->hdrlen;
352 done:
353 	got_object_close(obj);
354 	if (f && fclose(f) == EOF && err == NULL)
355 		err = got_error_from_errno("fclose");
356 	return err;
357 }
358 
359 struct got_commit_object *
360 got_object_commit_alloc_partial(void)
361 {
362 	struct got_commit_object *commit;
363 
364 	commit = calloc(1, sizeof(*commit));
365 	if (commit == NULL)
366 		return NULL;
367 	commit->tree_id = malloc(sizeof(*commit->tree_id));
368 	if (commit->tree_id == NULL) {
369 		free(commit);
370 		return NULL;
371 	}
372 
373 	STAILQ_INIT(&commit->parent_ids);
374 
375 	return commit;
376 }
377 
378 const struct got_error *
379 got_object_commit_add_parent(struct got_commit_object *commit,
380     const char *id_str)
381 {
382 	const struct got_error *err = NULL;
383 	struct got_object_qid *qid;
384 
385 	err = got_object_qid_alloc_partial(&qid);
386 	if (err)
387 		return err;
388 
389 	if (!got_parse_object_id(&qid->id, id_str, GOT_HASH_SHA1)) {
390 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
391 		got_object_qid_free(qid);
392 		return err;
393 	}
394 
395 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
396 	commit->nparents++;
397 
398 	return NULL;
399 }
400 
401 static const struct got_error *
402 parse_gmtoff(time_t *gmtoff, const char *tzstr)
403 {
404 	int sign = 1;
405 	const char *p = tzstr;
406 	time_t h, m;
407 
408 	*gmtoff = 0;
409 
410 	if (*p == '-')
411 		sign = -1;
412 	else if (*p != '+')
413 		return got_error(GOT_ERR_BAD_OBJ_DATA);
414 	p++;
415 	if (!isdigit((unsigned char)*p) &&
416 	    !isdigit((unsigned char)*(p + 1)))
417 		return got_error(GOT_ERR_BAD_OBJ_DATA);
418 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
419 
420 	p += 2;
421 	if (!isdigit((unsigned char)*p) &&
422 	    !isdigit((unsigned char)*(p + 1)))
423 		return got_error(GOT_ERR_BAD_OBJ_DATA);
424 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
425 
426 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
427 	return NULL;
428 }
429 
430 static const struct got_error *
431 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
432 {
433 	const struct got_error *err = NULL;
434 	const char *errstr;
435 	char *space, *tzstr;
436 
437 	/* Parse and strip off trailing timezone indicator string. */
438 	space = strrchr(committer, ' ');
439 	if (space == NULL)
440 		return got_error(GOT_ERR_BAD_OBJ_DATA);
441 	tzstr = strdup(space + 1);
442 	if (tzstr == NULL)
443 		return got_error_from_errno("strdup");
444 	err = parse_gmtoff(gmtoff, tzstr);
445 	free(tzstr);
446 	if (err) {
447 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
448 			return err;
449 		/* Old versions of Git omitted the timestamp. */
450 		*time = 0;
451 		*gmtoff = 0;
452 		return NULL;
453 	}
454 	*space = '\0';
455 
456 	/* Timestamp is separated from committer name + email by space. */
457 	space = strrchr(committer, ' ');
458 	if (space == NULL)
459 		return got_error(GOT_ERR_BAD_OBJ_DATA);
460 
461 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
462 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
463 	if (errstr)
464 		return got_error(GOT_ERR_BAD_OBJ_DATA);
465 
466 	/* Strip off parsed time information, leaving just author and email. */
467 	*space = '\0';
468 
469 	return NULL;
470 }
471 
472 void
473 got_object_commit_close(struct got_commit_object *commit)
474 {
475 	if (commit->refcnt > 0) {
476 		commit->refcnt--;
477 		if (commit->refcnt > 0)
478 			return;
479 	}
480 
481 	got_object_id_queue_free(&commit->parent_ids);
482 	free(commit->tree_id);
483 	free(commit->author);
484 	free(commit->committer);
485 	free(commit->logmsg);
486 	free(commit);
487 }
488 
489 struct got_object_id *
490 got_object_commit_get_tree_id(struct got_commit_object *commit)
491 {
492 	return commit->tree_id;
493 }
494 
495 int
496 got_object_commit_get_nparents(struct got_commit_object *commit)
497 {
498 	return commit->nparents;
499 }
500 
501 const struct got_object_id_queue *
502 got_object_commit_get_parent_ids(struct got_commit_object *commit)
503 {
504 	return &commit->parent_ids;
505 }
506 
507 const char *
508 got_object_commit_get_author(struct got_commit_object *commit)
509 {
510 	return commit->author;
511 }
512 
513 time_t
514 got_object_commit_get_author_time(struct got_commit_object *commit)
515 {
516 	return commit->author_time;
517 }
518 
519 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
520 {
521 	return commit->author_gmtoff;
522 }
523 
524 const char *
525 got_object_commit_get_committer(struct got_commit_object *commit)
526 {
527 	return commit->committer;
528 }
529 
530 time_t
531 got_object_commit_get_committer_time(struct got_commit_object *commit)
532 {
533 	return commit->committer_time;
534 }
535 
536 time_t
537 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
538 {
539 	return commit->committer_gmtoff;
540 }
541 
542 const struct got_error *
543 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
544 {
545 	const struct got_error *err = NULL;
546 	const char *src;
547 	char *dst;
548 	size_t len;
549 
550 	len = strlen(commit->logmsg);
551 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
552 	if (*logmsg == NULL)
553 		return got_error_from_errno("malloc");
554 
555 	/*
556 	 * Strip out unusual headers. Headers are separated from the commit
557 	 * message body by a single empty line.
558 	 */
559 	src = commit->logmsg;
560 	dst = *logmsg;
561 	while (*src != '\0' && *src != '\n') {
562 		int copy_header = 1, eol = 0;
563 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
564 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
565 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
566 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
567 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
568 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
569 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
570 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
571 			copy_header = 0;
572 
573 		while (*src != '\0' && !eol) {
574 			if (copy_header) {
575 				*dst = *src;
576 				dst++;
577 			}
578 			if (*src == '\n')
579 				eol = 1;
580 			src++;
581 		}
582 	}
583 	*dst = '\0';
584 
585 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
586 		err = got_error(GOT_ERR_NO_SPACE);
587 		goto done;
588 	}
589 
590 	/* Trim redundant trailing whitespace. */
591 	len = strlen(*logmsg);
592 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
593 	    isspace((unsigned char)(*logmsg)[len - 1])) {
594 		(*logmsg)[len - 1] = '\0';
595 		len--;
596 	}
597 
598 	/* Append a trailing newline if missing. */
599 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
600 		(*logmsg)[len] = '\n';
601 		(*logmsg)[len + 1] = '\0';
602 	}
603 done:
604 	if (err) {
605 		free(*logmsg);
606 		*logmsg = NULL;
607 	}
608 	return err;
609 }
610 
611 const char *
612 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
613 {
614 	return commit->logmsg;
615 }
616 
617 const struct got_error *
618 got_object_parse_commit(struct got_commit_object **commit, char *buf,
619     size_t len)
620 {
621 	const struct got_error *err = NULL;
622 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
623 	char *s = buf;
624 	size_t label_len;
625 	ssize_t remain = (ssize_t)len;
626 
627 	if (remain == 0)
628 		return got_error(GOT_ERR_BAD_OBJ_DATA);
629 
630 	*commit = got_object_commit_alloc_partial();
631 	if (*commit == NULL)
632 		return got_error_from_errno("got_object_commit_alloc_partial");
633 
634 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
635 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
636 		remain -= label_len;
637 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
638 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
639 			goto done;
640 		}
641 		s += label_len;
642 		if (!got_parse_object_id((*commit)->tree_id, s, algo)) {
643 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
644 			goto done;
645 		}
646 		remain -= SHA1_DIGEST_STRING_LENGTH;
647 		s += SHA1_DIGEST_STRING_LENGTH;
648 	} else {
649 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
650 		goto done;
651 	}
652 
653 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
654 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
655 		remain -= label_len;
656 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
657 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
658 			goto done;
659 		}
660 		s += label_len;
661 		err = got_object_commit_add_parent(*commit, s);
662 		if (err)
663 			goto done;
664 
665 		remain -= SHA1_DIGEST_STRING_LENGTH;
666 		s += SHA1_DIGEST_STRING_LENGTH;
667 	}
668 
669 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
670 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
671 		char *p;
672 		size_t slen;
673 
674 		remain -= label_len;
675 		if (remain <= 0) {
676 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
677 			goto done;
678 		}
679 		s += label_len;
680 		p = memchr(s, '\n', remain);
681 		if (p == NULL) {
682 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
683 			goto done;
684 		}
685 		*p = '\0';
686 		slen = strlen(s);
687 		err = parse_commit_time(&(*commit)->author_time,
688 		    &(*commit)->author_gmtoff, s);
689 		if (err)
690 			goto done;
691 		(*commit)->author = strdup(s);
692 		if ((*commit)->author == NULL) {
693 			err = got_error_from_errno("strdup");
694 			goto done;
695 		}
696 		s += slen + 1;
697 		remain -= slen + 1;
698 	}
699 
700 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
701 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
702 		char *p;
703 		size_t slen;
704 
705 		remain -= label_len;
706 		if (remain <= 0) {
707 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
708 			goto done;
709 		}
710 		s += label_len;
711 		p = memchr(s, '\n', remain);
712 		if (p == NULL) {
713 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
714 			goto done;
715 		}
716 		*p = '\0';
717 		slen = strlen(s);
718 		err = parse_commit_time(&(*commit)->committer_time,
719 		    &(*commit)->committer_gmtoff, s);
720 		if (err)
721 			goto done;
722 		(*commit)->committer = strdup(s);
723 		if ((*commit)->committer == NULL) {
724 			err = got_error_from_errno("strdup");
725 			goto done;
726 		}
727 		s += slen + 1;
728 		remain -= slen + 1;
729 	}
730 
731 	(*commit)->logmsg = strndup(s, remain);
732 	if ((*commit)->logmsg == NULL) {
733 		err = got_error_from_errno("strndup");
734 		goto done;
735 	}
736 done:
737 	if (err) {
738 		got_object_commit_close(*commit);
739 		*commit = NULL;
740 	}
741 	return err;
742 }
743 
744 const struct got_error *
745 got_object_read_commit(struct got_commit_object **commit, int fd,
746     struct got_object_id *expected_id, size_t expected_size)
747 {
748 	struct got_object *obj = NULL;
749 	const struct got_error *err = NULL;
750 	size_t len;
751 	uint8_t *p;
752 	struct got_inflate_checksum csum;
753 	SHA1_CTX sha1_ctx;
754 	struct got_object_id id;
755 
756 	SHA1Init(&sha1_ctx);
757 	memset(&csum, 0, sizeof(csum));
758 	csum.output_sha1 = &sha1_ctx;
759 
760 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
761 	if (err)
762 		return err;
763 
764 	SHA1Final(id.sha1, &sha1_ctx);
765 	if (got_object_id_cmp(expected_id, &id) != 0) {
766 		err = got_error_checksum(expected_id);
767 		goto done;
768 	}
769 
770 	err = got_object_parse_header(&obj, p, len);
771 	if (err)
772 		goto done;
773 
774 	if (len < obj->hdrlen + obj->size) {
775 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
776 		goto done;
777 	}
778 
779 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
780 		err = got_error(GOT_ERR_OBJ_TYPE);
781 		goto done;
782 	}
783 
784 	/* Skip object header. */
785 	len -= obj->hdrlen;
786 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
787 done:
788 	free(p);
789 	if (obj)
790 		got_object_close(obj);
791 	return err;
792 }
793 
794 void
795 got_object_tree_close(struct got_tree_object *tree)
796 {
797 	if (tree->refcnt > 0) {
798 		tree->refcnt--;
799 		if (tree->refcnt > 0)
800 			return;
801 	}
802 
803 	free(tree->entries);
804 	free(tree);
805 }
806 
807 static const struct got_error *
808 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
809     size_t maxlen)
810 {
811 	char *p, *space;
812 
813 	*elen = 0;
814 
815 	*elen = strnlen(buf, maxlen) + 1;
816 	if (*elen > maxlen)
817 		return got_error(GOT_ERR_BAD_OBJ_DATA);
818 
819 	space = memchr(buf, ' ', *elen);
820 	if (space == NULL || space <= buf)
821 		return got_error(GOT_ERR_BAD_OBJ_DATA);
822 
823 	pte->mode = 0;
824 	p = buf;
825 	while (p < space) {
826 		if (*p < '0' || *p > '7')
827 			return got_error(GOT_ERR_BAD_OBJ_DATA);
828 		pte->mode <<= 3;
829 		pte->mode |= *p - '0';
830 		p++;
831 	}
832 
833 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
834 		return got_error(GOT_ERR_BAD_OBJ_DATA);
835 
836 	pte->name = space + 1;
837 	pte->namelen = strlen(pte->name);
838 	buf += *elen;
839 	pte->id = buf;
840 	*elen += SHA1_DIGEST_LENGTH;
841 	return NULL;
842 }
843 
844 static int
845 pte_cmp(const void *pa, const void *pb)
846 {
847 	const struct got_parsed_tree_entry *a = pa, *b = pb;
848 
849 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
850 }
851 
852 const struct got_error *
853 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
854     size_t *nentries_alloc, uint8_t *buf, size_t len)
855 {
856 	const struct got_error *err = NULL;
857 	size_t remain = len;
858 	const size_t nalloc = 16;
859 	struct got_parsed_tree_entry *pte;
860 	int i;
861 
862 	*nentries = 0;
863 	if (remain == 0)
864 		return NULL; /* tree is empty */
865 
866 	while (remain > 0) {
867 		size_t elen;
868 
869 		if (*nentries >= *nentries_alloc) {
870 			pte = recallocarray(*entries, *nentries_alloc,
871 			    *nentries_alloc + nalloc, sizeof(**entries));
872 			if (pte == NULL) {
873 				err = got_error_from_errno("recallocarray");
874 				goto done;
875 			}
876 			*entries = pte;
877 			*nentries_alloc += nalloc;
878 		}
879 
880 		pte = &(*entries)[*nentries];
881 		err = parse_tree_entry(pte, &elen, buf, remain);
882 		if (err)
883 			goto done;
884 		buf += elen;
885 		remain -= elen;
886 		(*nentries)++;
887 	}
888 
889 	if (remain != 0) {
890 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
891 		goto done;
892 	}
893 
894 	if (*nentries > 1) {
895 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
896 
897 		for (i = 0; i < *nentries - 1; i++) {
898 			struct got_parsed_tree_entry *prev = &(*entries)[i];
899 			pte = &(*entries)[i + 1];
900 			if (got_path_cmp(prev->name, pte->name,
901 			    prev->namelen, pte->namelen) == 0) {
902 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
903 				break;
904 			}
905 		}
906 	}
907 done:
908 	if (err)
909 		*nentries = 0;
910 	return err;
911 }
912 
913 const struct got_error *
914 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
915     size_t *nentries_alloc, uint8_t **p, int fd,
916     struct got_object_id *expected_id)
917 {
918 	const struct got_error *err = NULL;
919 	struct got_object *obj = NULL;
920 	size_t len;
921 	struct got_inflate_checksum csum;
922 	SHA1_CTX sha1_ctx;
923 	struct got_object_id id;
924 
925 	SHA1Init(&sha1_ctx);
926 	memset(&csum, 0, sizeof(csum));
927 	csum.output_sha1 = &sha1_ctx;
928 
929 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
930 	if (err)
931 		return err;
932 
933 	SHA1Final(id.sha1, &sha1_ctx);
934 	if (got_object_id_cmp(expected_id, &id) != 0) {
935 		err = got_error_checksum(expected_id);
936 		goto done;
937 	}
938 
939 	err = got_object_parse_header(&obj, *p, len);
940 	if (err)
941 		goto done;
942 
943 	if (len < obj->hdrlen + obj->size) {
944 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
945 		goto done;
946 	}
947 
948 	/* Skip object header. */
949 	len -= obj->hdrlen;
950 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
951 	    *p + obj->hdrlen, len);
952 done:
953 	if (obj)
954 		got_object_close(obj);
955 	return err;
956 }
957 
958 void
959 got_object_tag_close(struct got_tag_object *tag)
960 {
961 	if (tag->refcnt > 0) {
962 		tag->refcnt--;
963 		if (tag->refcnt > 0)
964 			return;
965 	}
966 
967 	free(tag->tag);
968 	free(tag->tagger);
969 	free(tag->tagmsg);
970 	free(tag);
971 }
972 
973 const struct got_error *
974 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
975 {
976 	const struct got_error *err = NULL;
977 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
978 	size_t remain = len;
979 	char *s = buf;
980 	size_t label_len;
981 
982 	if (remain == 0)
983 		return got_error(GOT_ERR_BAD_OBJ_DATA);
984 
985 	*tag = calloc(1, sizeof(**tag));
986 	if (*tag == NULL)
987 		return got_error_from_errno("calloc");
988 
989 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
990 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
991 		remain -= label_len;
992 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
993 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
994 			goto done;
995 		}
996 		s += label_len;
997 		if (!got_parse_object_id(&(*tag)->id, s, algo)) {
998 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
999 			goto done;
1000 		}
1001 		remain -= SHA1_DIGEST_STRING_LENGTH;
1002 		s += SHA1_DIGEST_STRING_LENGTH;
1003 	} else {
1004 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1005 		goto done;
1006 	}
1007 
1008 	if (remain <= 0) {
1009 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1010 		goto done;
1011 	}
1012 
1013 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1014 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1015 		remain -= label_len;
1016 		if (remain <= 0) {
1017 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1018 			goto done;
1019 		}
1020 		s += label_len;
1021 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1022 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1023 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1024 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1025 			s += label_len;
1026 			remain -= label_len;
1027 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1028 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1029 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1030 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1031 			s += label_len;
1032 			remain -= label_len;
1033 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1034 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1035 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1036 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1037 			s += label_len;
1038 			remain -= label_len;
1039 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1040 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1041 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1042 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1043 			s += label_len;
1044 			remain -= label_len;
1045 		} else {
1046 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1047 			goto done;
1048 		}
1049 
1050 		if (remain <= 0 || *s != '\n') {
1051 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1052 			goto done;
1053 		}
1054 		s++;
1055 		remain--;
1056 		if (remain <= 0) {
1057 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1058 			goto done;
1059 		}
1060 	} else {
1061 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1062 		goto done;
1063 	}
1064 
1065 	label_len = strlen(GOT_TAG_LABEL_TAG);
1066 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1067 		char *p;
1068 		size_t slen;
1069 		remain -= label_len;
1070 		if (remain <= 0) {
1071 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1072 			goto done;
1073 		}
1074 		s += label_len;
1075 		p = memchr(s, '\n', remain);
1076 		if (p == NULL) {
1077 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1078 			goto done;
1079 		}
1080 		*p = '\0';
1081 		slen = strlen(s);
1082 		(*tag)->tag = strndup(s, slen);
1083 		if ((*tag)->tag == NULL) {
1084 			err = got_error_from_errno("strndup");
1085 			goto done;
1086 		}
1087 		s += slen + 1;
1088 		remain -= slen + 1;
1089 		if (remain <= 0) {
1090 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1091 			goto done;
1092 		}
1093 	} else {
1094 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1095 		goto done;
1096 	}
1097 
1098 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1099 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1100 		char *p;
1101 		size_t slen;
1102 
1103 		remain -= label_len;
1104 		if (remain <= 0) {
1105 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1106 			goto done;
1107 		}
1108 		s += label_len;
1109 		p = memchr(s, '\n', remain);
1110 		if (p == NULL) {
1111 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1112 			goto done;
1113 		}
1114 		*p = '\0';
1115 		slen = strlen(s);
1116 		err = parse_commit_time(&(*tag)->tagger_time,
1117 		    &(*tag)->tagger_gmtoff, s);
1118 		if (err)
1119 			goto done;
1120 		(*tag)->tagger = strdup(s);
1121 		if ((*tag)->tagger == NULL) {
1122 			err = got_error_from_errno("strdup");
1123 			goto done;
1124 		}
1125 		s += slen + 1;
1126 		remain -= slen + 1;
1127 		if (remain < 0) {
1128 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1129 			goto done;
1130 		}
1131 	} else {
1132 		/* Some old tags in the Linux git repo have no tagger. */
1133 		(*tag)->tagger = strdup("");
1134 		if ((*tag)->tagger == NULL) {
1135 			err = got_error_from_errno("strdup");
1136 			goto done;
1137 		}
1138 	}
1139 
1140 	(*tag)->tagmsg = strndup(s, remain);
1141 	if ((*tag)->tagmsg == NULL) {
1142 		err = got_error_from_errno("strndup");
1143 		goto done;
1144 	}
1145 done:
1146 	if (err) {
1147 		got_object_tag_close(*tag);
1148 		*tag = NULL;
1149 	}
1150 	return err;
1151 }
1152 
1153 const struct got_error *
1154 got_object_read_tag(struct got_tag_object **tag, int fd,
1155     struct got_object_id *expected_id, size_t expected_size)
1156 {
1157 	const struct got_error *err = NULL;
1158 	struct got_object *obj = NULL;
1159 	size_t len;
1160 	uint8_t *p;
1161 	struct got_inflate_checksum csum;
1162 	SHA1_CTX sha1_ctx;
1163 	struct got_object_id id;
1164 
1165 	SHA1Init(&sha1_ctx);
1166 	memset(&csum, 0, sizeof(csum));
1167 	csum.output_sha1 = &sha1_ctx;
1168 
1169 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1170 	    expected_size, fd);
1171 	if (err)
1172 		return err;
1173 
1174 	SHA1Final(id.sha1, &sha1_ctx);
1175 	if (got_object_id_cmp(expected_id, &id) != 0) {
1176 		err = got_error_checksum(expected_id);
1177 		goto done;
1178 	}
1179 
1180 	err = got_object_parse_header(&obj, p, len);
1181 	if (err)
1182 		goto done;
1183 
1184 	if (len < obj->hdrlen + obj->size) {
1185 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1186 		goto done;
1187 	}
1188 
1189 	/* Skip object header. */
1190 	len -= obj->hdrlen;
1191 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1192 done:
1193 	free(p);
1194 	if (obj)
1195 		got_object_close(obj);
1196 	return err;
1197 }
1198 
1199 const struct got_error *
1200 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1201 {
1202 	const struct got_error *err = NULL;
1203 	static const size_t blocksize = 512;
1204 	size_t n, total, remain;
1205 	uint8_t *buf;
1206 
1207 	*outbuf = NULL;
1208 	*outlen = 0;
1209 
1210 	buf = malloc(blocksize);
1211 	if (buf == NULL)
1212 		return got_error_from_errno("malloc");
1213 
1214 	remain = blocksize;
1215 	total = 0;
1216 	for (;;) {
1217 		if (remain == 0) {
1218 			uint8_t *newbuf;
1219 			newbuf = reallocarray(buf, 1, total + blocksize);
1220 			if (newbuf == NULL) {
1221 				err = got_error_from_errno("reallocarray");
1222 				goto done;
1223 			}
1224 			buf = newbuf;
1225 			remain += blocksize;
1226 		}
1227 		n = fread(buf + total, 1, remain, f);
1228 		if (n == 0) {
1229 			if (ferror(f)) {
1230 				err = got_ferror(f, GOT_ERR_IO);
1231 				goto done;
1232 			}
1233 			break; /* EOF */
1234 		}
1235 		remain -= n;
1236 		total += n;
1237 	};
1238 
1239 done:
1240 	if (err == NULL) {
1241 		*outbuf = buf;
1242 		*outlen = total;
1243 	} else
1244 		free(buf);
1245 	return err;
1246 }