op public repos

Blob

Date:: Thu Feb 23 18:51:38 2023 UTC
Message:: add some helper functions to compute hashes This adds a set of functions to abstract over SHA1Init, SHA1Update, SHA1Final, their respective SHA256 variants and how to compare digests. Replace all the SHA1*() usage with the new APIs. It's a preparatory step for sha256 handling. 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/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 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
76 }
77 
78 const struct got_error *
79 got_object_qid_alloc_partial(struct got_object_qid **qid)
80 {
81 	*qid = malloc(sizeof(**qid));
82 	if (*qid == NULL)
83 		return got_error_from_errno("malloc");
84 
85 	(*qid)->data = NULL;
86 	return NULL;
87 }
88 
89 const struct got_error *
90 got_object_id_str(char **outbuf, struct got_object_id *id)
91 {
92 	static const size_t len = GOT_OBJECT_ID_HEX_MAXLEN;
93 
94 	*outbuf = malloc(len);
95 	if (*outbuf == NULL)
96 		return got_error_from_errno("malloc");
97 
98 	if (got_object_id_hex(id, *outbuf, len) == NULL) {
99 		free(*outbuf);
100 		*outbuf = NULL;
101 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
102 	}
103 
104 	return NULL;
105 }
106 
107 char *
108 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
109 {
110 	return got_sha1_digest_to_str(id->sha1, buf, len);
111 }
112 
113 void
114 got_object_close(struct got_object *obj)
115 {
116 	if (obj->refcnt > 0) {
117 		obj->refcnt--;
118 		if (obj->refcnt > 0)
119 			return;
120 	}
121 
122 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
123 		struct got_delta *delta;
124 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
125 			delta = STAILQ_FIRST(&obj->deltas.entries);
126 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
127 			free(delta);
128 		}
129 	}
130 	free(obj);
131 }
132 
133 const struct got_error *
134 got_object_raw_close(struct got_raw_object *obj)
135 {
136 	const struct got_error *err = NULL;
137 
138 	if (obj->refcnt > 0) {
139 		obj->refcnt--;
140 		if (obj->refcnt > 0)
141 			return NULL;
142 	}
143 
144 	if (obj->close_cb)
145 		obj->close_cb(obj);
146 
147 	if (obj->f == NULL) {
148 		if (obj->fd != -1) {
149 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
150 				err = got_error_from_errno("munmap");
151 			if (close(obj->fd) == -1 && err == NULL)
152 				err = got_error_from_errno("close");
153 		} else
154 			free(obj->data);
155 	} else {
156 		if (fclose(obj->f) == EOF && err == NULL)
157 			err = got_error_from_errno("fclose");
158 	}
159 	free(obj);
160 	return err;
161 }
162 
163 void
164 got_object_qid_free(struct got_object_qid *qid)
165 {
166 	free(qid);
167 }
168 
169 void
170 got_object_id_queue_free(struct got_object_id_queue *ids)
171 {
172 	struct got_object_qid *qid;
173 
174 	while (!STAILQ_EMPTY(ids)) {
175 		qid = STAILQ_FIRST(ids);
176 		STAILQ_REMOVE_HEAD(ids, entry);
177 		got_object_qid_free(qid);
178 	}
179 }
180 
181 const struct got_error *
182 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
183 {
184 	const char *obj_labels[] = {
185 		GOT_OBJ_LABEL_COMMIT,
186 		GOT_OBJ_LABEL_TREE,
187 		GOT_OBJ_LABEL_BLOB,
188 		GOT_OBJ_LABEL_TAG,
189 	};
190 	const int obj_types[] = {
191 		GOT_OBJ_TYPE_COMMIT,
192 		GOT_OBJ_TYPE_TREE,
193 		GOT_OBJ_TYPE_BLOB,
194 		GOT_OBJ_TYPE_TAG,
195 	};
196 	int type = 0;
197 	size_t size = 0;
198 	size_t i;
199 	char *end;
200 
201 	*obj = NULL;
202 
203 	end = memchr(buf, '\0', len);
204 	if (end == NULL)
205 		return got_error(GOT_ERR_BAD_OBJ_HDR);
206 
207 	for (i = 0; i < nitems(obj_labels); i++) {
208 		const char *label = obj_labels[i];
209 		size_t label_len = strlen(label);
210 		const char *errstr;
211 
212 		if (len <= label_len || buf + label_len >= end ||
213 		    strncmp(buf, label, label_len) != 0)
214 			continue;
215 
216 		type = obj_types[i];
217 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
218 		if (errstr != NULL)
219 			return got_error(GOT_ERR_BAD_OBJ_HDR);
220 		break;
221 	}
222 
223 	if (type == 0)
224 		return got_error(GOT_ERR_BAD_OBJ_HDR);
225 
226 	*obj = calloc(1, sizeof(**obj));
227 	if (*obj == NULL)
228 		return got_error_from_errno("calloc");
229 	(*obj)->type = type;
230 	(*obj)->hdrlen = end - buf + 1;
231 	(*obj)->size = size;
232 	return NULL;
233 }
234 
235 const struct got_error *
236 got_object_read_header(struct got_object **obj, int fd)
237 {
238 	const struct got_error *err;
239 	struct got_inflate_buf zb;
240 	uint8_t *buf;
241 	const size_t zbsize = 64;
242 	size_t outlen, totlen;
243 	int nbuf = 1;
244 
245 	*obj = NULL;
246 
247 	buf = malloc(zbsize);
248 	if (buf == NULL)
249 		return got_error_from_errno("malloc");
250 	buf[0] = '\0';
251 
252 	err = got_inflate_init(&zb, buf, zbsize, NULL);
253 	if (err)
254 		return err;
255 
256 	totlen = 0;
257 	do {
258 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
259 		if (err)
260 			goto done;
261 		if (outlen == 0)
262 			break;
263 		totlen += outlen;
264 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
265 			uint8_t *newbuf;
266 			nbuf++;
267 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
268 			if (newbuf == NULL) {
269 				err = got_error_from_errno("recallocarray");
270 				goto done;
271 			}
272 			buf = newbuf;
273 			zb.outbuf = newbuf + totlen;
274 			zb.outlen = (nbuf * zbsize) - totlen;
275 		}
276 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
277 
278 	err = got_object_parse_header(obj, buf, totlen);
279 done:
280 	free(buf);
281 	got_inflate_end(&zb);
282 	return err;
283 }
284 
285 const struct got_error *
286 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
287     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
288     int infd)
289 {
290 	const struct got_error *err = NULL;
291 	struct got_object *obj;
292 	struct got_inflate_checksum csum;
293 	struct got_object_id id;
294 	struct got_hash ctx;
295 	size_t len, consumed;
296 	FILE *f = NULL;
297 
298 	*outbuf = NULL;
299 	*size = 0;
300 	*hdrlen = 0;
301 
302 	got_hash_init(&ctx, GOT_HASH_SHA1);
303 	memset(&csum, 0, sizeof(csum));
304 	csum.output_ctx = &ctx;
305 
306 	if (lseek(infd, SEEK_SET, 0) == -1)
307 		return got_error_from_errno("lseek");
308 
309 	err = got_object_read_header(&obj, infd);
310 	if (err)
311 		return err;
312 
313 	if (lseek(infd, SEEK_SET, 0) == -1)
314 		return got_error_from_errno("lseek");
315 
316 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
317 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
318 		    obj->size + obj->hdrlen, infd);
319 	} else {
320 		int fd;
321 		/*
322 		 * XXX This uses an extra file descriptor for no good reason.
323 		 * We should have got_inflate_fd_to_fd().
324 		 */
325 		fd = dup(infd);
326 		if (fd == -1)
327 			return got_error_from_errno("dup");
328 		f = fdopen(fd, "r");
329 		if (f == NULL) {
330 			err = got_error_from_errno("fdopen");
331 			abort();
332 			close(fd);
333 			goto done;
334 		}
335 		err = got_inflate_to_fd(&len, f, &csum, outfd);
336 	}
337 	if (err)
338 		goto done;
339 
340 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
341 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
342 		goto done;
343 	}
344 
345 	got_hash_final_object_id(&ctx, &id);
346 	if (got_object_id_cmp(expected_id, &id) != 0) {
347 		err = got_error_checksum(expected_id);
348 		goto done;
349 	}
350 
351 	*size = obj->size;
352 	*hdrlen = obj->hdrlen;
353 done:
354 	got_object_close(obj);
355 	if (f && fclose(f) == EOF && err == NULL)
356 		err = got_error_from_errno("fclose");
357 	return err;
358 }
359 
360 struct got_commit_object *
361 got_object_commit_alloc_partial(void)
362 {
363 	struct got_commit_object *commit;
364 
365 	commit = calloc(1, sizeof(*commit));
366 	if (commit == NULL)
367 		return NULL;
368 	commit->tree_id = malloc(sizeof(*commit->tree_id));
369 	if (commit->tree_id == NULL) {
370 		free(commit);
371 		return NULL;
372 	}
373 
374 	STAILQ_INIT(&commit->parent_ids);
375 
376 	return commit;
377 }
378 
379 const struct got_error *
380 got_object_commit_add_parent(struct got_commit_object *commit,
381     const char *id_str)
382 {
383 	const struct got_error *err = NULL;
384 	struct got_object_qid *qid;
385 
386 	err = got_object_qid_alloc_partial(&qid);
387 	if (err)
388 		return err;
389 
390 	if (!got_parse_object_id(&qid->id, id_str, GOT_HASH_SHA1)) {
391 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
392 		got_object_qid_free(qid);
393 		return err;
394 	}
395 
396 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
397 	commit->nparents++;
398 
399 	return NULL;
400 }
401 
402 static const struct got_error *
403 parse_gmtoff(time_t *gmtoff, const char *tzstr)
404 {
405 	int sign = 1;
406 	const char *p = tzstr;
407 	time_t h, m;
408 
409 	*gmtoff = 0;
410 
411 	if (*p == '-')
412 		sign = -1;
413 	else if (*p != '+')
414 		return got_error(GOT_ERR_BAD_OBJ_DATA);
415 	p++;
416 	if (!isdigit((unsigned char)*p) &&
417 	    !isdigit((unsigned char)*(p + 1)))
418 		return got_error(GOT_ERR_BAD_OBJ_DATA);
419 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
420 
421 	p += 2;
422 	if (!isdigit((unsigned char)*p) &&
423 	    !isdigit((unsigned char)*(p + 1)))
424 		return got_error(GOT_ERR_BAD_OBJ_DATA);
425 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
426 
427 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
428 	return NULL;
429 }
430 
431 static const struct got_error *
432 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
433 {
434 	const struct got_error *err = NULL;
435 	const char *errstr;
436 	char *space, *tzstr;
437 
438 	/* Parse and strip off trailing timezone indicator string. */
439 	space = strrchr(committer, ' ');
440 	if (space == NULL)
441 		return got_error(GOT_ERR_BAD_OBJ_DATA);
442 	tzstr = strdup(space + 1);
443 	if (tzstr == NULL)
444 		return got_error_from_errno("strdup");
445 	err = parse_gmtoff(gmtoff, tzstr);
446 	free(tzstr);
447 	if (err) {
448 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
449 			return err;
450 		/* Old versions of Git omitted the timestamp. */
451 		*time = 0;
452 		*gmtoff = 0;
453 		return NULL;
454 	}
455 	*space = '\0';
456 
457 	/* Timestamp is separated from committer name + email by space. */
458 	space = strrchr(committer, ' ');
459 	if (space == NULL)
460 		return got_error(GOT_ERR_BAD_OBJ_DATA);
461 
462 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
463 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
464 	if (errstr)
465 		return got_error(GOT_ERR_BAD_OBJ_DATA);
466 
467 	/* Strip off parsed time information, leaving just author and email. */
468 	*space = '\0';
469 
470 	return NULL;
471 }
472 
473 void
474 got_object_commit_close(struct got_commit_object *commit)
475 {
476 	if (commit->refcnt > 0) {
477 		commit->refcnt--;
478 		if (commit->refcnt > 0)
479 			return;
480 	}
481 
482 	got_object_id_queue_free(&commit->parent_ids);
483 	free(commit->tree_id);
484 	free(commit->author);
485 	free(commit->committer);
486 	free(commit->logmsg);
487 	free(commit);
488 }
489 
490 struct got_object_id *
491 got_object_commit_get_tree_id(struct got_commit_object *commit)
492 {
493 	return commit->tree_id;
494 }
495 
496 int
497 got_object_commit_get_nparents(struct got_commit_object *commit)
498 {
499 	return commit->nparents;
500 }
501 
502 const struct got_object_id_queue *
503 got_object_commit_get_parent_ids(struct got_commit_object *commit)
504 {
505 	return &commit->parent_ids;
506 }
507 
508 const char *
509 got_object_commit_get_author(struct got_commit_object *commit)
510 {
511 	return commit->author;
512 }
513 
514 time_t
515 got_object_commit_get_author_time(struct got_commit_object *commit)
516 {
517 	return commit->author_time;
518 }
519 
520 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
521 {
522 	return commit->author_gmtoff;
523 }
524 
525 const char *
526 got_object_commit_get_committer(struct got_commit_object *commit)
527 {
528 	return commit->committer;
529 }
530 
531 time_t
532 got_object_commit_get_committer_time(struct got_commit_object *commit)
533 {
534 	return commit->committer_time;
535 }
536 
537 time_t
538 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
539 {
540 	return commit->committer_gmtoff;
541 }
542 
543 const struct got_error *
544 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
545 {
546 	const struct got_error *err = NULL;
547 	const char *src;
548 	char *dst;
549 	size_t len;
550 
551 	len = strlen(commit->logmsg);
552 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
553 	if (*logmsg == NULL)
554 		return got_error_from_errno("malloc");
555 
556 	/*
557 	 * Strip out unusual headers. Headers are separated from the commit
558 	 * message body by a single empty line.
559 	 */
560 	src = commit->logmsg;
561 	dst = *logmsg;
562 	while (*src != '\0' && *src != '\n') {
563 		int copy_header = 1, eol = 0;
564 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
565 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
566 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
567 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
568 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
569 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
570 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
571 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
572 			copy_header = 0;
573 
574 		while (*src != '\0' && !eol) {
575 			if (copy_header) {
576 				*dst = *src;
577 				dst++;
578 			}
579 			if (*src == '\n')
580 				eol = 1;
581 			src++;
582 		}
583 	}
584 	*dst = '\0';
585 
586 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
587 		err = got_error(GOT_ERR_NO_SPACE);
588 		goto done;
589 	}
590 
591 	/* Trim redundant trailing whitespace. */
592 	len = strlen(*logmsg);
593 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
594 	    isspace((unsigned char)(*logmsg)[len - 1])) {
595 		(*logmsg)[len - 1] = '\0';
596 		len--;
597 	}
598 
599 	/* Append a trailing newline if missing. */
600 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
601 		(*logmsg)[len] = '\n';
602 		(*logmsg)[len + 1] = '\0';
603 	}
604 done:
605 	if (err) {
606 		free(*logmsg);
607 		*logmsg = NULL;
608 	}
609 	return err;
610 }
611 
612 const char *
613 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
614 {
615 	return commit->logmsg;
616 }
617 
618 const struct got_error *
619 got_object_parse_commit(struct got_commit_object **commit, char *buf,
620     size_t len)
621 {
622 	const struct got_error *err = NULL;
623 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
624 	char *s = buf;
625 	size_t label_len;
626 	ssize_t remain = (ssize_t)len;
627 
628 	if (remain == 0)
629 		return got_error(GOT_ERR_BAD_OBJ_DATA);
630 
631 	*commit = got_object_commit_alloc_partial();
632 	if (*commit == NULL)
633 		return got_error_from_errno("got_object_commit_alloc_partial");
634 
635 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
636 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
637 		remain -= label_len;
638 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
639 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
640 			goto done;
641 		}
642 		s += label_len;
643 		if (!got_parse_object_id((*commit)->tree_id, s, algo)) {
644 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
645 			goto done;
646 		}
647 		remain -= SHA1_DIGEST_STRING_LENGTH;
648 		s += SHA1_DIGEST_STRING_LENGTH;
649 	} else {
650 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
651 		goto done;
652 	}
653 
654 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
655 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
656 		remain -= label_len;
657 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
658 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
659 			goto done;
660 		}
661 		s += label_len;
662 		err = got_object_commit_add_parent(*commit, s);
663 		if (err)
664 			goto done;
665 
666 		remain -= SHA1_DIGEST_STRING_LENGTH;
667 		s += SHA1_DIGEST_STRING_LENGTH;
668 	}
669 
670 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
671 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
672 		char *p;
673 		size_t slen;
674 
675 		remain -= label_len;
676 		if (remain <= 0) {
677 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
678 			goto done;
679 		}
680 		s += label_len;
681 		p = memchr(s, '\n', remain);
682 		if (p == NULL) {
683 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
684 			goto done;
685 		}
686 		*p = '\0';
687 		slen = strlen(s);
688 		err = parse_commit_time(&(*commit)->author_time,
689 		    &(*commit)->author_gmtoff, s);
690 		if (err)
691 			goto done;
692 		(*commit)->author = strdup(s);
693 		if ((*commit)->author == NULL) {
694 			err = got_error_from_errno("strdup");
695 			goto done;
696 		}
697 		s += slen + 1;
698 		remain -= slen + 1;
699 	}
700 
701 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
702 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
703 		char *p;
704 		size_t slen;
705 
706 		remain -= label_len;
707 		if (remain <= 0) {
708 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
709 			goto done;
710 		}
711 		s += label_len;
712 		p = memchr(s, '\n', remain);
713 		if (p == NULL) {
714 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
715 			goto done;
716 		}
717 		*p = '\0';
718 		slen = strlen(s);
719 		err = parse_commit_time(&(*commit)->committer_time,
720 		    &(*commit)->committer_gmtoff, s);
721 		if (err)
722 			goto done;
723 		(*commit)->committer = strdup(s);
724 		if ((*commit)->committer == NULL) {
725 			err = got_error_from_errno("strdup");
726 			goto done;
727 		}
728 		s += slen + 1;
729 		remain -= slen + 1;
730 	}
731 
732 	(*commit)->logmsg = strndup(s, remain);
733 	if ((*commit)->logmsg == NULL) {
734 		err = got_error_from_errno("strndup");
735 		goto done;
736 	}
737 done:
738 	if (err) {
739 		got_object_commit_close(*commit);
740 		*commit = NULL;
741 	}
742 	return err;
743 }
744 
745 const struct got_error *
746 got_object_read_commit(struct got_commit_object **commit, int fd,
747     struct got_object_id *expected_id, size_t expected_size)
748 {
749 	struct got_object *obj = NULL;
750 	const struct got_error *err = NULL;
751 	size_t len;
752 	uint8_t *p;
753 	struct got_inflate_checksum csum;
754 	struct got_hash ctx;
755 	struct got_object_id id;
756 
757 	got_hash_init(&ctx, GOT_HASH_SHA1);
758 	memset(&csum, 0, sizeof(csum));
759 	csum.output_ctx = &ctx;
760 
761 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
762 	if (err)
763 		return err;
764 
765 	got_hash_final_object_id(&ctx, &id);
766 	if (got_object_id_cmp(expected_id, &id) != 0) {
767 		err = got_error_checksum(expected_id);
768 		goto done;
769 	}
770 
771 	err = got_object_parse_header(&obj, p, len);
772 	if (err)
773 		goto done;
774 
775 	if (len < obj->hdrlen + obj->size) {
776 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
777 		goto done;
778 	}
779 
780 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
781 		err = got_error(GOT_ERR_OBJ_TYPE);
782 		goto done;
783 	}
784 
785 	/* Skip object header. */
786 	len -= obj->hdrlen;
787 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
788 done:
789 	free(p);
790 	if (obj)
791 		got_object_close(obj);
792 	return err;
793 }
794 
795 void
796 got_object_tree_close(struct got_tree_object *tree)
797 {
798 	if (tree->refcnt > 0) {
799 		tree->refcnt--;
800 		if (tree->refcnt > 0)
801 			return;
802 	}
803 
804 	free(tree->entries);
805 	free(tree);
806 }
807 
808 static const struct got_error *
809 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
810     size_t maxlen)
811 {
812 	char *p, *space;
813 
814 	*elen = 0;
815 
816 	*elen = strnlen(buf, maxlen) + 1;
817 	if (*elen > maxlen)
818 		return got_error(GOT_ERR_BAD_OBJ_DATA);
819 
820 	space = memchr(buf, ' ', *elen);
821 	if (space == NULL || space <= buf)
822 		return got_error(GOT_ERR_BAD_OBJ_DATA);
823 
824 	pte->mode = 0;
825 	p = buf;
826 	while (p < space) {
827 		if (*p < '0' || *p > '7')
828 			return got_error(GOT_ERR_BAD_OBJ_DATA);
829 		pte->mode <<= 3;
830 		pte->mode |= *p - '0';
831 		p++;
832 	}
833 
834 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
835 		return got_error(GOT_ERR_BAD_OBJ_DATA);
836 
837 	pte->name = space + 1;
838 	pte->namelen = strlen(pte->name);
839 	buf += *elen;
840 	pte->id = buf;
841 	*elen += SHA1_DIGEST_LENGTH;
842 	return NULL;
843 }
844 
845 static int
846 pte_cmp(const void *pa, const void *pb)
847 {
848 	const struct got_parsed_tree_entry *a = pa, *b = pb;
849 
850 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
851 }
852 
853 const struct got_error *
854 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
855     size_t *nentries_alloc, uint8_t *buf, size_t len)
856 {
857 	const struct got_error *err = NULL;
858 	size_t remain = len;
859 	const size_t nalloc = 16;
860 	struct got_parsed_tree_entry *pte;
861 	int i;
862 
863 	*nentries = 0;
864 	if (remain == 0)
865 		return NULL; /* tree is empty */
866 
867 	while (remain > 0) {
868 		size_t elen;
869 
870 		if (*nentries >= *nentries_alloc) {
871 			pte = recallocarray(*entries, *nentries_alloc,
872 			    *nentries_alloc + nalloc, sizeof(**entries));
873 			if (pte == NULL) {
874 				err = got_error_from_errno("recallocarray");
875 				goto done;
876 			}
877 			*entries = pte;
878 			*nentries_alloc += nalloc;
879 		}
880 
881 		pte = &(*entries)[*nentries];
882 		err = parse_tree_entry(pte, &elen, buf, remain);
883 		if (err)
884 			goto done;
885 		buf += elen;
886 		remain -= elen;
887 		(*nentries)++;
888 	}
889 
890 	if (remain != 0) {
891 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
892 		goto done;
893 	}
894 
895 	if (*nentries > 1) {
896 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
897 
898 		for (i = 0; i < *nentries - 1; i++) {
899 			struct got_parsed_tree_entry *prev = &(*entries)[i];
900 			pte = &(*entries)[i + 1];
901 			if (got_path_cmp(prev->name, pte->name,
902 			    prev->namelen, pte->namelen) == 0) {
903 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
904 				break;
905 			}
906 		}
907 	}
908 done:
909 	if (err)
910 		*nentries = 0;
911 	return err;
912 }
913 
914 const struct got_error *
915 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
916     size_t *nentries_alloc, uint8_t **p, int fd,
917     struct got_object_id *expected_id)
918 {
919 	const struct got_error *err = NULL;
920 	struct got_object *obj = NULL;
921 	size_t len;
922 	struct got_inflate_checksum csum;
923 	struct got_hash ctx;
924 	struct got_object_id id;
925 
926 	got_hash_init(&ctx, GOT_HASH_SHA1);
927 	memset(&csum, 0, sizeof(csum));
928 	csum.output_ctx = &ctx;
929 
930 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
931 	if (err)
932 		return err;
933 
934 	got_hash_final_object_id(&ctx, &id);
935 	if (got_object_id_cmp(expected_id, &id) != 0) {
936 		err = got_error_checksum(expected_id);
937 		goto done;
938 	}
939 
940 	err = got_object_parse_header(&obj, *p, len);
941 	if (err)
942 		goto done;
943 
944 	if (len < obj->hdrlen + obj->size) {
945 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
946 		goto done;
947 	}
948 
949 	/* Skip object header. */
950 	len -= obj->hdrlen;
951 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
952 	    *p + obj->hdrlen, len);
953 done:
954 	if (obj)
955 		got_object_close(obj);
956 	return err;
957 }
958 
959 void
960 got_object_tag_close(struct got_tag_object *tag)
961 {
962 	if (tag->refcnt > 0) {
963 		tag->refcnt--;
964 		if (tag->refcnt > 0)
965 			return;
966 	}
967 
968 	free(tag->tag);
969 	free(tag->tagger);
970 	free(tag->tagmsg);
971 	free(tag);
972 }
973 
974 const struct got_error *
975 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
976 {
977 	const struct got_error *err = NULL;
978 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
979 	size_t remain = len;
980 	char *s = buf;
981 	size_t label_len;
982 
983 	if (remain == 0)
984 		return got_error(GOT_ERR_BAD_OBJ_DATA);
985 
986 	*tag = calloc(1, sizeof(**tag));
987 	if (*tag == NULL)
988 		return got_error_from_errno("calloc");
989 
990 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
991 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
992 		remain -= label_len;
993 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
994 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
995 			goto done;
996 		}
997 		s += label_len;
998 		if (!got_parse_object_id(&(*tag)->id, s, algo)) {
999 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1000 			goto done;
1001 		}
1002 		remain -= SHA1_DIGEST_STRING_LENGTH;
1003 		s += SHA1_DIGEST_STRING_LENGTH;
1004 	} else {
1005 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1006 		goto done;
1007 	}
1008 
1009 	if (remain <= 0) {
1010 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1011 		goto done;
1012 	}
1013 
1014 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1015 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1016 		remain -= label_len;
1017 		if (remain <= 0) {
1018 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1019 			goto done;
1020 		}
1021 		s += label_len;
1022 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1023 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1024 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1025 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1026 			s += label_len;
1027 			remain -= label_len;
1028 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1029 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1030 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1031 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1032 			s += label_len;
1033 			remain -= label_len;
1034 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1035 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1036 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1037 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1038 			s += label_len;
1039 			remain -= label_len;
1040 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1041 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1042 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1043 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1044 			s += label_len;
1045 			remain -= label_len;
1046 		} else {
1047 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1048 			goto done;
1049 		}
1050 
1051 		if (remain <= 0 || *s != '\n') {
1052 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1053 			goto done;
1054 		}
1055 		s++;
1056 		remain--;
1057 		if (remain <= 0) {
1058 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1059 			goto done;
1060 		}
1061 	} else {
1062 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1063 		goto done;
1064 	}
1065 
1066 	label_len = strlen(GOT_TAG_LABEL_TAG);
1067 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1068 		char *p;
1069 		size_t slen;
1070 		remain -= label_len;
1071 		if (remain <= 0) {
1072 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1073 			goto done;
1074 		}
1075 		s += label_len;
1076 		p = memchr(s, '\n', remain);
1077 		if (p == NULL) {
1078 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1079 			goto done;
1080 		}
1081 		*p = '\0';
1082 		slen = strlen(s);
1083 		(*tag)->tag = strndup(s, slen);
1084 		if ((*tag)->tag == NULL) {
1085 			err = got_error_from_errno("strndup");
1086 			goto done;
1087 		}
1088 		s += slen + 1;
1089 		remain -= slen + 1;
1090 		if (remain <= 0) {
1091 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1092 			goto done;
1093 		}
1094 	} else {
1095 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1096 		goto done;
1097 	}
1098 
1099 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1100 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1101 		char *p;
1102 		size_t slen;
1103 
1104 		remain -= label_len;
1105 		if (remain <= 0) {
1106 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1107 			goto done;
1108 		}
1109 		s += label_len;
1110 		p = memchr(s, '\n', remain);
1111 		if (p == NULL) {
1112 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1113 			goto done;
1114 		}
1115 		*p = '\0';
1116 		slen = strlen(s);
1117 		err = parse_commit_time(&(*tag)->tagger_time,
1118 		    &(*tag)->tagger_gmtoff, s);
1119 		if (err)
1120 			goto done;
1121 		(*tag)->tagger = strdup(s);
1122 		if ((*tag)->tagger == NULL) {
1123 			err = got_error_from_errno("strdup");
1124 			goto done;
1125 		}
1126 		s += slen + 1;
1127 		remain -= slen + 1;
1128 		if (remain < 0) {
1129 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1130 			goto done;
1131 		}
1132 	} else {
1133 		/* Some old tags in the Linux git repo have no tagger. */
1134 		(*tag)->tagger = strdup("");
1135 		if ((*tag)->tagger == NULL) {
1136 			err = got_error_from_errno("strdup");
1137 			goto done;
1138 		}
1139 	}
1140 
1141 	(*tag)->tagmsg = strndup(s, remain);
1142 	if ((*tag)->tagmsg == NULL) {
1143 		err = got_error_from_errno("strndup");
1144 		goto done;
1145 	}
1146 done:
1147 	if (err) {
1148 		got_object_tag_close(*tag);
1149 		*tag = NULL;
1150 	}
1151 	return err;
1152 }
1153 
1154 const struct got_error *
1155 got_object_read_tag(struct got_tag_object **tag, int fd,
1156     struct got_object_id *expected_id, size_t expected_size)
1157 {
1158 	const struct got_error *err = NULL;
1159 	struct got_object *obj = NULL;
1160 	size_t len;
1161 	uint8_t *p;
1162 	struct got_inflate_checksum csum;
1163 	struct got_hash ctx;
1164 	struct got_object_id id;
1165 
1166 	got_hash_init(&ctx, GOT_HASH_SHA1);
1167 	memset(&csum, 0, sizeof(csum));
1168 	csum.output_ctx = &ctx;
1169 
1170 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1171 	    expected_size, fd);
1172 	if (err)
1173 		return err;
1174 
1175 	got_hash_final_object_id(&ctx, &id);
1176 	if (got_object_id_cmp(expected_id, &id) != 0) {
1177 		err = got_error_checksum(expected_id);
1178 		goto done;
1179 	}
1180 
1181 	err = got_object_parse_header(&obj, p, len);
1182 	if (err)
1183 		goto done;
1184 
1185 	if (len < obj->hdrlen + obj->size) {
1186 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1187 		goto done;
1188 	}
1189 
1190 	/* Skip object header. */
1191 	len -= obj->hdrlen;
1192 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1193 done:
1194 	free(p);
1195 	if (obj)
1196 		got_object_close(obj);
1197 	return err;
1198 }
1199 
1200 const struct got_error *
1201 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1202 {
1203 	const struct got_error *err = NULL;
1204 	static const size_t blocksize = 512;
1205 	size_t n, total, remain;
1206 	uint8_t *buf;
1207 
1208 	*outbuf = NULL;
1209 	*outlen = 0;
1210 
1211 	buf = malloc(blocksize);
1212 	if (buf == NULL)
1213 		return got_error_from_errno("malloc");
1214 
1215 	remain = blocksize;
1216 	total = 0;
1217 	for (;;) {
1218 		if (remain == 0) {
1219 			uint8_t *newbuf;
1220 			newbuf = reallocarray(buf, 1, total + blocksize);
1221 			if (newbuf == NULL) {
1222 				err = got_error_from_errno("reallocarray");
1223 				goto done;
1224 			}
1225 			buf = newbuf;
1226 			remain += blocksize;
1227 		}
1228 		n = fread(buf + total, 1, remain, f);
1229 		if (n == 0) {
1230 			if (ferror(f)) {
1231 				err = got_ferror(f, GOT_ERR_IO);
1232 				goto done;
1233 			}
1234 			break; /* EOF */
1235 		}
1236 		remain -= n;
1237 		total += n;
1238 	};
1239 
1240 done:
1241 	if (err == NULL) {
1242 		*outbuf = buf;
1243 		*outlen = total;
1244 	} else
1245 		free(buf);
1246 	return err;
1247 }