op public repos

Blob

Date:: Mon Apr 24 12:33:54 2023 UTC
Message:: when finding changed paths iterate tree entries in on-disk order for speed ok op, jamsek
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_qid.h"
52 #include "got_lib_object_cache.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_repository.h"
55 
56 #ifndef nitems
57 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
58 #endif
59 
60 const struct got_error *
61 got_object_type_label(const char **label, int obj_type)
62 {
63 	const struct got_error *err = NULL;
64 
65 	switch (obj_type) {
66 	case GOT_OBJ_TYPE_BLOB:
67 		*label = GOT_OBJ_LABEL_BLOB;
68 		break;
69 	case GOT_OBJ_TYPE_TREE:
70 		*label = GOT_OBJ_LABEL_TREE;
71 		break;
72 	case GOT_OBJ_TYPE_COMMIT:
73 		*label = GOT_OBJ_LABEL_COMMIT;
74 		break;
75 	case GOT_OBJ_TYPE_TAG:
76 		*label = GOT_OBJ_LABEL_TAG;
77 		break;
78 	default:
79 		*label = NULL;
80 		err = got_error(GOT_ERR_OBJ_TYPE);
81 		break;
82 	}
83 
84 	return err;
85 }
86 
87 void
88 got_object_close(struct got_object *obj)
89 {
90 	if (obj->refcnt > 0) {
91 		obj->refcnt--;
92 		if (obj->refcnt > 0)
93 			return;
94 	}
95 
96 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
97 		struct got_delta *delta;
98 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
99 			delta = STAILQ_FIRST(&obj->deltas.entries);
100 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
101 			free(delta);
102 		}
103 	}
104 	free(obj);
105 }
106 
107 const struct got_error *
108 got_object_raw_close(struct got_raw_object *obj)
109 {
110 	const struct got_error *err = NULL;
111 
112 	if (obj->refcnt > 0) {
113 		obj->refcnt--;
114 		if (obj->refcnt > 0)
115 			return NULL;
116 	}
117 
118 	if (obj->close_cb)
119 		obj->close_cb(obj);
120 
121 	if (obj->f == NULL) {
122 		if (obj->fd != -1) {
123 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
124 				err = got_error_from_errno("munmap");
125 			if (close(obj->fd) == -1 && err == NULL)
126 				err = got_error_from_errno("close");
127 		} else
128 			free(obj->data);
129 	} else {
130 		if (fclose(obj->f) == EOF && err == NULL)
131 			err = got_error_from_errno("fclose");
132 	}
133 	free(obj);
134 	return err;
135 }
136 
137 const struct got_error *
138 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
139 {
140 	const char *obj_labels[] = {
141 		GOT_OBJ_LABEL_COMMIT,
142 		GOT_OBJ_LABEL_TREE,
143 		GOT_OBJ_LABEL_BLOB,
144 		GOT_OBJ_LABEL_TAG,
145 	};
146 	const int obj_types[] = {
147 		GOT_OBJ_TYPE_COMMIT,
148 		GOT_OBJ_TYPE_TREE,
149 		GOT_OBJ_TYPE_BLOB,
150 		GOT_OBJ_TYPE_TAG,
151 	};
152 	int type = 0;
153 	size_t size = 0;
154 	size_t i;
155 	char *end;
156 
157 	*obj = NULL;
158 
159 	end = memchr(buf, '\0', len);
160 	if (end == NULL)
161 		return got_error(GOT_ERR_BAD_OBJ_HDR);
162 
163 	for (i = 0; i < nitems(obj_labels); i++) {
164 		const char *label = obj_labels[i];
165 		size_t label_len = strlen(label);
166 		const char *errstr;
167 
168 		if (len <= label_len || buf + label_len >= end ||
169 		    strncmp(buf, label, label_len) != 0)
170 			continue;
171 
172 		type = obj_types[i];
173 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
174 		if (errstr != NULL)
175 			return got_error(GOT_ERR_BAD_OBJ_HDR);
176 		break;
177 	}
178 
179 	if (type == 0)
180 		return got_error(GOT_ERR_BAD_OBJ_HDR);
181 
182 	*obj = calloc(1, sizeof(**obj));
183 	if (*obj == NULL)
184 		return got_error_from_errno("calloc");
185 	(*obj)->type = type;
186 	(*obj)->hdrlen = end - buf + 1;
187 	(*obj)->size = size;
188 	return NULL;
189 }
190 
191 const struct got_error *
192 got_object_read_header(struct got_object **obj, int fd)
193 {
194 	const struct got_error *err;
195 	struct got_inflate_buf zb;
196 	uint8_t *buf;
197 	const size_t zbsize = 64;
198 	size_t outlen, totlen;
199 	int nbuf = 1;
200 
201 	*obj = NULL;
202 
203 	buf = malloc(zbsize);
204 	if (buf == NULL)
205 		return got_error_from_errno("malloc");
206 	buf[0] = '\0';
207 
208 	err = got_inflate_init(&zb, buf, zbsize, NULL);
209 	if (err)
210 		return err;
211 
212 	totlen = 0;
213 	do {
214 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
215 		if (err)
216 			goto done;
217 		if (outlen == 0)
218 			break;
219 		totlen += outlen;
220 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
221 			uint8_t *newbuf;
222 			nbuf++;
223 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
224 			if (newbuf == NULL) {
225 				err = got_error_from_errno("recallocarray");
226 				goto done;
227 			}
228 			buf = newbuf;
229 			zb.outbuf = newbuf + totlen;
230 			zb.outlen = (nbuf * zbsize) - totlen;
231 		}
232 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
233 
234 	err = got_object_parse_header(obj, buf, totlen);
235 done:
236 	free(buf);
237 	got_inflate_end(&zb);
238 	return err;
239 }
240 
241 const struct got_error *
242 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
243     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
244     int infd)
245 {
246 	const struct got_error *err = NULL;
247 	struct got_object *obj;
248 	struct got_inflate_checksum csum;
249 	struct got_object_id id;
250 	struct got_hash ctx;
251 	size_t len, consumed;
252 	FILE *f = NULL;
253 
254 	*outbuf = NULL;
255 	*size = 0;
256 	*hdrlen = 0;
257 
258 	got_hash_init(&ctx, GOT_HASH_SHA1);
259 	memset(&csum, 0, sizeof(csum));
260 	csum.output_ctx = &ctx;
261 
262 	if (lseek(infd, SEEK_SET, 0) == -1)
263 		return got_error_from_errno("lseek");
264 
265 	err = got_object_read_header(&obj, infd);
266 	if (err)
267 		return err;
268 
269 	if (lseek(infd, SEEK_SET, 0) == -1)
270 		return got_error_from_errno("lseek");
271 
272 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
273 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
274 		    obj->size + obj->hdrlen, infd);
275 	} else {
276 		int fd;
277 		/*
278 		 * XXX This uses an extra file descriptor for no good reason.
279 		 * We should have got_inflate_fd_to_fd().
280 		 */
281 		fd = dup(infd);
282 		if (fd == -1)
283 			return got_error_from_errno("dup");
284 		f = fdopen(fd, "r");
285 		if (f == NULL) {
286 			err = got_error_from_errno("fdopen");
287 			abort();
288 			close(fd);
289 			goto done;
290 		}
291 		err = got_inflate_to_fd(&len, f, &csum, outfd);
292 	}
293 	if (err)
294 		goto done;
295 
296 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
297 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
298 		goto done;
299 	}
300 
301 	got_hash_final_object_id(&ctx, &id);
302 	if (got_object_id_cmp(expected_id, &id) != 0) {
303 		err = got_error_checksum(expected_id);
304 		goto done;
305 	}
306 
307 	*size = obj->size;
308 	*hdrlen = obj->hdrlen;
309 done:
310 	got_object_close(obj);
311 	if (f && fclose(f) == EOF && err == NULL)
312 		err = got_error_from_errno("fclose");
313 	return err;
314 }
315 
316 struct got_commit_object *
317 got_object_commit_alloc_partial(void)
318 {
319 	struct got_commit_object *commit;
320 
321 	commit = calloc(1, sizeof(*commit));
322 	if (commit == NULL)
323 		return NULL;
324 	commit->tree_id = malloc(sizeof(*commit->tree_id));
325 	if (commit->tree_id == NULL) {
326 		free(commit);
327 		return NULL;
328 	}
329 
330 	STAILQ_INIT(&commit->parent_ids);
331 
332 	return commit;
333 }
334 
335 const struct got_error *
336 got_object_commit_add_parent(struct got_commit_object *commit,
337     const char *id_str)
338 {
339 	const struct got_error *err = NULL;
340 	struct got_object_qid *qid;
341 
342 	err = got_object_qid_alloc_partial(&qid);
343 	if (err)
344 		return err;
345 
346 	if (!got_parse_object_id(&qid->id, id_str, GOT_HASH_SHA1)) {
347 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
348 		got_object_qid_free(qid);
349 		return err;
350 	}
351 
352 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
353 	commit->nparents++;
354 
355 	return NULL;
356 }
357 
358 static const struct got_error *
359 parse_gmtoff(time_t *gmtoff, const char *tzstr)
360 {
361 	int sign = 1;
362 	const char *p = tzstr;
363 	time_t h, m;
364 
365 	*gmtoff = 0;
366 
367 	if (*p == '-')
368 		sign = -1;
369 	else if (*p != '+')
370 		return got_error(GOT_ERR_BAD_OBJ_DATA);
371 	p++;
372 	if (!isdigit((unsigned char)*p) &&
373 	    !isdigit((unsigned char)*(p + 1)))
374 		return got_error(GOT_ERR_BAD_OBJ_DATA);
375 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
376 
377 	p += 2;
378 	if (!isdigit((unsigned char)*p) &&
379 	    !isdigit((unsigned char)*(p + 1)))
380 		return got_error(GOT_ERR_BAD_OBJ_DATA);
381 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
382 
383 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
384 	return NULL;
385 }
386 
387 static const struct got_error *
388 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
389 {
390 	const struct got_error *err = NULL;
391 	const char *errstr;
392 	char *space, *tzstr;
393 
394 	/* Parse and strip off trailing timezone indicator string. */
395 	space = strrchr(committer, ' ');
396 	if (space == NULL)
397 		return got_error(GOT_ERR_BAD_OBJ_DATA);
398 	tzstr = strdup(space + 1);
399 	if (tzstr == NULL)
400 		return got_error_from_errno("strdup");
401 	err = parse_gmtoff(gmtoff, tzstr);
402 	free(tzstr);
403 	if (err) {
404 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
405 			return err;
406 		/* Old versions of Git omitted the timestamp. */
407 		*time = 0;
408 		*gmtoff = 0;
409 		return NULL;
410 	}
411 	*space = '\0';
412 
413 	/* Timestamp is separated from committer name + email by space. */
414 	space = strrchr(committer, ' ');
415 	if (space == NULL)
416 		return got_error(GOT_ERR_BAD_OBJ_DATA);
417 
418 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
419 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
420 	if (errstr)
421 		return got_error(GOT_ERR_BAD_OBJ_DATA);
422 
423 	/* Strip off parsed time information, leaving just author and email. */
424 	*space = '\0';
425 
426 	return NULL;
427 }
428 
429 void
430 got_object_commit_close(struct got_commit_object *commit)
431 {
432 	if (commit->refcnt > 0) {
433 		commit->refcnt--;
434 		if (commit->refcnt > 0)
435 			return;
436 	}
437 
438 	got_object_id_queue_free(&commit->parent_ids);
439 	free(commit->tree_id);
440 	free(commit->author);
441 	free(commit->committer);
442 	free(commit->logmsg);
443 	free(commit);
444 }
445 
446 struct got_object_id *
447 got_object_commit_get_tree_id(struct got_commit_object *commit)
448 {
449 	return commit->tree_id;
450 }
451 
452 int
453 got_object_commit_get_nparents(struct got_commit_object *commit)
454 {
455 	return commit->nparents;
456 }
457 
458 const struct got_object_id_queue *
459 got_object_commit_get_parent_ids(struct got_commit_object *commit)
460 {
461 	return &commit->parent_ids;
462 }
463 
464 const char *
465 got_object_commit_get_author(struct got_commit_object *commit)
466 {
467 	return commit->author;
468 }
469 
470 time_t
471 got_object_commit_get_author_time(struct got_commit_object *commit)
472 {
473 	return commit->author_time;
474 }
475 
476 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
477 {
478 	return commit->author_gmtoff;
479 }
480 
481 const char *
482 got_object_commit_get_committer(struct got_commit_object *commit)
483 {
484 	return commit->committer;
485 }
486 
487 time_t
488 got_object_commit_get_committer_time(struct got_commit_object *commit)
489 {
490 	return commit->committer_time;
491 }
492 
493 time_t
494 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
495 {
496 	return commit->committer_gmtoff;
497 }
498 
499 const struct got_error *
500 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
501 {
502 	const struct got_error *err = NULL;
503 	const char *src;
504 	char *dst;
505 	size_t len;
506 
507 	len = strlen(commit->logmsg);
508 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
509 	if (*logmsg == NULL)
510 		return got_error_from_errno("malloc");
511 
512 	/*
513 	 * Strip out unusual headers. Headers are separated from the commit
514 	 * message body by a single empty line.
515 	 */
516 	src = commit->logmsg;
517 	dst = *logmsg;
518 	while (*src != '\0' && *src != '\n') {
519 		int copy_header = 1, eol = 0;
520 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
521 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
522 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
523 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
524 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
525 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
526 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
527 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
528 			copy_header = 0;
529 
530 		while (*src != '\0' && !eol) {
531 			if (copy_header) {
532 				*dst = *src;
533 				dst++;
534 			}
535 			if (*src == '\n')
536 				eol = 1;
537 			src++;
538 		}
539 	}
540 	*dst = '\0';
541 
542 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
543 		err = got_error(GOT_ERR_NO_SPACE);
544 		goto done;
545 	}
546 
547 	/* Trim redundant trailing whitespace. */
548 	len = strlen(*logmsg);
549 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
550 	    isspace((unsigned char)(*logmsg)[len - 1])) {
551 		(*logmsg)[len - 1] = '\0';
552 		len--;
553 	}
554 
555 	/* Append a trailing newline if missing. */
556 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
557 		(*logmsg)[len] = '\n';
558 		(*logmsg)[len + 1] = '\0';
559 	}
560 done:
561 	if (err) {
562 		free(*logmsg);
563 		*logmsg = NULL;
564 	}
565 	return err;
566 }
567 
568 const char *
569 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
570 {
571 	return commit->logmsg;
572 }
573 
574 const struct got_error *
575 got_object_parse_commit(struct got_commit_object **commit, char *buf,
576     size_t len)
577 {
578 	const struct got_error *err = NULL;
579 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
580 	char *s = buf;
581 	size_t label_len;
582 	ssize_t remain = (ssize_t)len;
583 
584 	if (remain == 0)
585 		return got_error(GOT_ERR_BAD_OBJ_DATA);
586 
587 	*commit = got_object_commit_alloc_partial();
588 	if (*commit == NULL)
589 		return got_error_from_errno("got_object_commit_alloc_partial");
590 
591 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
592 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
593 		remain -= label_len;
594 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
595 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
596 			goto done;
597 		}
598 		s += label_len;
599 		if (!got_parse_object_id((*commit)->tree_id, s, algo)) {
600 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
601 			goto done;
602 		}
603 		remain -= SHA1_DIGEST_STRING_LENGTH;
604 		s += SHA1_DIGEST_STRING_LENGTH;
605 	} else {
606 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
607 		goto done;
608 	}
609 
610 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
611 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
612 		remain -= label_len;
613 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
614 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
615 			goto done;
616 		}
617 		s += label_len;
618 		err = got_object_commit_add_parent(*commit, s);
619 		if (err)
620 			goto done;
621 
622 		remain -= SHA1_DIGEST_STRING_LENGTH;
623 		s += SHA1_DIGEST_STRING_LENGTH;
624 	}
625 
626 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
627 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
628 		char *p;
629 		size_t slen;
630 
631 		remain -= label_len;
632 		if (remain <= 0) {
633 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
634 			goto done;
635 		}
636 		s += label_len;
637 		p = memchr(s, '\n', remain);
638 		if (p == NULL) {
639 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
640 			goto done;
641 		}
642 		*p = '\0';
643 		slen = strlen(s);
644 		err = parse_commit_time(&(*commit)->author_time,
645 		    &(*commit)->author_gmtoff, s);
646 		if (err)
647 			goto done;
648 		(*commit)->author = strdup(s);
649 		if ((*commit)->author == NULL) {
650 			err = got_error_from_errno("strdup");
651 			goto done;
652 		}
653 		s += slen + 1;
654 		remain -= slen + 1;
655 	}
656 
657 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
658 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
659 		char *p;
660 		size_t slen;
661 
662 		remain -= label_len;
663 		if (remain <= 0) {
664 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
665 			goto done;
666 		}
667 		s += label_len;
668 		p = memchr(s, '\n', remain);
669 		if (p == NULL) {
670 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
671 			goto done;
672 		}
673 		*p = '\0';
674 		slen = strlen(s);
675 		err = parse_commit_time(&(*commit)->committer_time,
676 		    &(*commit)->committer_gmtoff, s);
677 		if (err)
678 			goto done;
679 		(*commit)->committer = strdup(s);
680 		if ((*commit)->committer == NULL) {
681 			err = got_error_from_errno("strdup");
682 			goto done;
683 		}
684 		s += slen + 1;
685 		remain -= slen + 1;
686 	}
687 
688 	(*commit)->logmsg = strndup(s, remain);
689 	if ((*commit)->logmsg == NULL) {
690 		err = got_error_from_errno("strndup");
691 		goto done;
692 	}
693 done:
694 	if (err) {
695 		got_object_commit_close(*commit);
696 		*commit = NULL;
697 	}
698 	return err;
699 }
700 
701 const struct got_error *
702 got_object_read_commit(struct got_commit_object **commit, int fd,
703     struct got_object_id *expected_id, size_t expected_size)
704 {
705 	struct got_object *obj = NULL;
706 	const struct got_error *err = NULL;
707 	size_t len;
708 	uint8_t *p;
709 	struct got_inflate_checksum csum;
710 	struct got_hash ctx;
711 	struct got_object_id id;
712 
713 	got_hash_init(&ctx, GOT_HASH_SHA1);
714 	memset(&csum, 0, sizeof(csum));
715 	csum.output_ctx = &ctx;
716 
717 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
718 	if (err)
719 		return err;
720 
721 	got_hash_final_object_id(&ctx, &id);
722 	if (got_object_id_cmp(expected_id, &id) != 0) {
723 		err = got_error_checksum(expected_id);
724 		goto done;
725 	}
726 
727 	err = got_object_parse_header(&obj, p, len);
728 	if (err)
729 		goto done;
730 
731 	if (len < obj->hdrlen + obj->size) {
732 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
733 		goto done;
734 	}
735 
736 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
737 		err = got_error(GOT_ERR_OBJ_TYPE);
738 		goto done;
739 	}
740 
741 	/* Skip object header. */
742 	len -= obj->hdrlen;
743 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
744 done:
745 	free(p);
746 	if (obj)
747 		got_object_close(obj);
748 	return err;
749 }
750 
751 void
752 got_object_tree_close(struct got_tree_object *tree)
753 {
754 	if (tree->refcnt > 0) {
755 		tree->refcnt--;
756 		if (tree->refcnt > 0)
757 			return;
758 	}
759 
760 	free(tree->entries);
761 	free(tree);
762 }
763 
764 const struct got_error *
765 got_object_parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen,
766     char *buf, size_t maxlen)
767 {
768 	char *p, *space;
769 
770 	*elen = 0;
771 
772 	*elen = strnlen(buf, maxlen) + 1;
773 	if (*elen > maxlen)
774 		return got_error(GOT_ERR_BAD_OBJ_DATA);
775 
776 	space = memchr(buf, ' ', *elen);
777 	if (space == NULL || space <= buf)
778 		return got_error(GOT_ERR_BAD_OBJ_DATA);
779 
780 	pte->mode = 0;
781 	p = buf;
782 	while (p < space) {
783 		if (*p < '0' || *p > '7')
784 			return got_error(GOT_ERR_BAD_OBJ_DATA);
785 		pte->mode <<= 3;
786 		pte->mode |= *p - '0';
787 		p++;
788 	}
789 
790 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
791 		return got_error(GOT_ERR_BAD_OBJ_DATA);
792 
793 	pte->name = space + 1;
794 	pte->namelen = strlen(pte->name);
795 	buf += *elen;
796 	pte->id = buf;
797 	*elen += SHA1_DIGEST_LENGTH;
798 	return NULL;
799 }
800 
801 static int
802 pte_cmp(const void *pa, const void *pb)
803 {
804 	const struct got_parsed_tree_entry *a = pa, *b = pb;
805 
806 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
807 }
808 
809 const struct got_error *
810 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
811     size_t *nentries_alloc, uint8_t *buf, size_t len)
812 {
813 	const struct got_error *err = NULL;
814 	size_t remain = len;
815 	const size_t nalloc = 16;
816 	struct got_parsed_tree_entry *pte;
817 	int i;
818 
819 	*nentries = 0;
820 	if (remain == 0)
821 		return NULL; /* tree is empty */
822 
823 	while (remain > 0) {
824 		size_t elen;
825 
826 		if (*nentries >= *nentries_alloc) {
827 			pte = recallocarray(*entries, *nentries_alloc,
828 			    *nentries_alloc + nalloc, sizeof(**entries));
829 			if (pte == NULL) {
830 				err = got_error_from_errno("recallocarray");
831 				goto done;
832 			}
833 			*entries = pte;
834 			*nentries_alloc += nalloc;
835 		}
836 
837 		pte = &(*entries)[*nentries];
838 		err = got_object_parse_tree_entry(pte, &elen, buf, remain);
839 		if (err)
840 			goto done;
841 		buf += elen;
842 		remain -= elen;
843 		(*nentries)++;
844 	}
845 
846 	if (remain != 0) {
847 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
848 		goto done;
849 	}
850 
851 	if (*nentries > 1) {
852 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
853 
854 		for (i = 0; i < *nentries - 1; i++) {
855 			struct got_parsed_tree_entry *prev = &(*entries)[i];
856 			pte = &(*entries)[i + 1];
857 			if (got_path_cmp(prev->name, pte->name,
858 			    prev->namelen, pte->namelen) == 0) {
859 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
860 				break;
861 			}
862 		}
863 	}
864 done:
865 	if (err)
866 		*nentries = 0;
867 	return err;
868 }
869 
870 const struct got_error *
871 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
872     size_t *nentries_alloc, uint8_t **p, int fd,
873     struct got_object_id *expected_id)
874 {
875 	const struct got_error *err = NULL;
876 	struct got_object *obj = NULL;
877 	size_t len;
878 	struct got_inflate_checksum csum;
879 	struct got_hash ctx;
880 	struct got_object_id id;
881 
882 	got_hash_init(&ctx, GOT_HASH_SHA1);
883 	memset(&csum, 0, sizeof(csum));
884 	csum.output_ctx = &ctx;
885 
886 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
887 	if (err)
888 		return err;
889 
890 	got_hash_final_object_id(&ctx, &id);
891 	if (got_object_id_cmp(expected_id, &id) != 0) {
892 		err = got_error_checksum(expected_id);
893 		goto done;
894 	}
895 
896 	err = got_object_parse_header(&obj, *p, len);
897 	if (err)
898 		goto done;
899 
900 	if (len < obj->hdrlen + obj->size) {
901 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
902 		goto done;
903 	}
904 
905 	/* Skip object header. */
906 	len -= obj->hdrlen;
907 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
908 	    *p + obj->hdrlen, len);
909 done:
910 	if (obj)
911 		got_object_close(obj);
912 	return err;
913 }
914 
915 void
916 got_object_tag_close(struct got_tag_object *tag)
917 {
918 	if (tag->refcnt > 0) {
919 		tag->refcnt--;
920 		if (tag->refcnt > 0)
921 			return;
922 	}
923 
924 	free(tag->tag);
925 	free(tag->tagger);
926 	free(tag->tagmsg);
927 	free(tag);
928 }
929 
930 const struct got_error *
931 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
932 {
933 	const struct got_error *err = NULL;
934 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
935 	size_t remain = len;
936 	char *s = buf;
937 	size_t label_len;
938 
939 	if (remain == 0)
940 		return got_error(GOT_ERR_BAD_OBJ_DATA);
941 
942 	*tag = calloc(1, sizeof(**tag));
943 	if (*tag == NULL)
944 		return got_error_from_errno("calloc");
945 
946 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
947 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
948 		remain -= label_len;
949 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
950 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
951 			goto done;
952 		}
953 		s += label_len;
954 		if (!got_parse_object_id(&(*tag)->id, s, algo)) {
955 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
956 			goto done;
957 		}
958 		remain -= SHA1_DIGEST_STRING_LENGTH;
959 		s += SHA1_DIGEST_STRING_LENGTH;
960 	} else {
961 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
962 		goto done;
963 	}
964 
965 	if (remain <= 0) {
966 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
967 		goto done;
968 	}
969 
970 	label_len = strlen(GOT_TAG_LABEL_TYPE);
971 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
972 		remain -= label_len;
973 		if (remain <= 0) {
974 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
975 			goto done;
976 		}
977 		s += label_len;
978 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
979 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
980 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
981 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
982 			s += label_len;
983 			remain -= label_len;
984 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
985 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
986 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
987 			label_len = strlen(GOT_OBJ_LABEL_TREE);
988 			s += label_len;
989 			remain -= label_len;
990 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
991 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
992 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
993 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
994 			s += label_len;
995 			remain -= label_len;
996 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
997 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
998 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
999 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1000 			s += label_len;
1001 			remain -= label_len;
1002 		} else {
1003 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1004 			goto done;
1005 		}
1006 
1007 		if (remain <= 0 || *s != '\n') {
1008 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1009 			goto done;
1010 		}
1011 		s++;
1012 		remain--;
1013 		if (remain <= 0) {
1014 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1015 			goto done;
1016 		}
1017 	} else {
1018 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1019 		goto done;
1020 	}
1021 
1022 	label_len = strlen(GOT_TAG_LABEL_TAG);
1023 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1024 		char *p;
1025 		size_t slen;
1026 		remain -= label_len;
1027 		if (remain <= 0) {
1028 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1029 			goto done;
1030 		}
1031 		s += label_len;
1032 		p = memchr(s, '\n', remain);
1033 		if (p == NULL) {
1034 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1035 			goto done;
1036 		}
1037 		*p = '\0';
1038 		slen = strlen(s);
1039 		(*tag)->tag = strndup(s, slen);
1040 		if ((*tag)->tag == NULL) {
1041 			err = got_error_from_errno("strndup");
1042 			goto done;
1043 		}
1044 		s += slen + 1;
1045 		remain -= slen + 1;
1046 		if (remain <= 0) {
1047 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1048 			goto done;
1049 		}
1050 	} else {
1051 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1052 		goto done;
1053 	}
1054 
1055 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1056 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1057 		char *p;
1058 		size_t slen;
1059 
1060 		remain -= label_len;
1061 		if (remain <= 0) {
1062 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1063 			goto done;
1064 		}
1065 		s += label_len;
1066 		p = memchr(s, '\n', remain);
1067 		if (p == NULL) {
1068 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1069 			goto done;
1070 		}
1071 		*p = '\0';
1072 		slen = strlen(s);
1073 		err = parse_commit_time(&(*tag)->tagger_time,
1074 		    &(*tag)->tagger_gmtoff, s);
1075 		if (err)
1076 			goto done;
1077 		(*tag)->tagger = strdup(s);
1078 		if ((*tag)->tagger == NULL) {
1079 			err = got_error_from_errno("strdup");
1080 			goto done;
1081 		}
1082 		s += slen + 1;
1083 		remain -= slen + 1;
1084 		if (remain < 0) {
1085 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1086 			goto done;
1087 		}
1088 	} else {
1089 		/* Some old tags in the Linux git repo have no tagger. */
1090 		(*tag)->tagger = strdup("");
1091 		if ((*tag)->tagger == NULL) {
1092 			err = got_error_from_errno("strdup");
1093 			goto done;
1094 		}
1095 	}
1096 
1097 	(*tag)->tagmsg = strndup(s, remain);
1098 	if ((*tag)->tagmsg == NULL) {
1099 		err = got_error_from_errno("strndup");
1100 		goto done;
1101 	}
1102 done:
1103 	if (err) {
1104 		got_object_tag_close(*tag);
1105 		*tag = NULL;
1106 	}
1107 	return err;
1108 }
1109 
1110 const struct got_error *
1111 got_object_read_tag(struct got_tag_object **tag, int fd,
1112     struct got_object_id *expected_id, size_t expected_size)
1113 {
1114 	const struct got_error *err = NULL;
1115 	struct got_object *obj = NULL;
1116 	size_t len;
1117 	uint8_t *p;
1118 	struct got_inflate_checksum csum;
1119 	struct got_hash ctx;
1120 	struct got_object_id id;
1121 
1122 	got_hash_init(&ctx, GOT_HASH_SHA1);
1123 	memset(&csum, 0, sizeof(csum));
1124 	csum.output_ctx = &ctx;
1125 
1126 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1127 	    expected_size, fd);
1128 	if (err)
1129 		return err;
1130 
1131 	got_hash_final_object_id(&ctx, &id);
1132 	if (got_object_id_cmp(expected_id, &id) != 0) {
1133 		err = got_error_checksum(expected_id);
1134 		goto done;
1135 	}
1136 
1137 	err = got_object_parse_header(&obj, p, len);
1138 	if (err)
1139 		goto done;
1140 
1141 	if (len < obj->hdrlen + obj->size) {
1142 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1143 		goto done;
1144 	}
1145 
1146 	/* Skip object header. */
1147 	len -= obj->hdrlen;
1148 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1149 done:
1150 	free(p);
1151 	if (obj)
1152 		got_object_close(obj);
1153 	return err;
1154 }