Got Mirror

Blob

Date:: Wed Oct 13 11:09:15 2021 UTC
Message:: use RB_TREE instead of STAILQ to manage packindex bloom filters; much faster
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 
25 #include <errno.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <stdint.h>
30 #include <sha1.h>
31 #include <zlib.h>
32 #include <ctype.h>
33 #include <limits.h>
34 #include <imsg.h>
35 #include <time.h>
36 #include <unistd.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_sha1.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_privsep.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 	const struct got_error *err = NULL;
81 
82 	*qid = malloc(sizeof(**qid));
83 	if (*qid == NULL)
84 		return got_error_from_errno("malloc");
85 
86 	(*qid)->id = malloc(sizeof(*((*qid)->id)));
87 	if ((*qid)->id == NULL) {
88 		err = got_error_from_errno("malloc");
89 		got_object_qid_free(*qid);
90 		*qid = NULL;
91 		return err;
92 	}
93 	(*qid)->data = NULL;
94 
95 	return NULL;
96 }
97 
98 const struct got_error *
99 got_object_id_str(char **outbuf, struct got_object_id *id)
100 {
101 	static const size_t len = SHA1_DIGEST_STRING_LENGTH;
102 
103 	*outbuf = malloc(len);
104 	if (*outbuf == NULL)
105 		return got_error_from_errno("malloc");
106 
107 	if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
108 		free(*outbuf);
109 		*outbuf = NULL;
110 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
111 	}
112 
113 	return NULL;
114 }
115 
116 void
117 got_object_close(struct got_object *obj)
118 {
119 	if (obj->refcnt > 0) {
120 		obj->refcnt--;
121 		if (obj->refcnt > 0)
122 			return;
123 	}
124 
125 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
126 		struct got_delta *delta;
127 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
128 			delta = STAILQ_FIRST(&obj->deltas.entries);
129 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
130 			free(delta);
131 		}
132 	}
133 	free(obj);
134 }
135 
136 void
137 got_object_qid_free(struct got_object_qid *qid)
138 {
139 	free(qid->id);
140 	free(qid);
141 }
142 
143 void
144 got_object_id_queue_free(struct got_object_id_queue *ids)
145 {
146 	struct got_object_qid *qid;
147 
148 	while (!STAILQ_EMPTY(ids)) {
149 		qid = STAILQ_FIRST(ids);
150 		STAILQ_REMOVE_HEAD(ids, entry);
151 		got_object_qid_free(qid);
152 	}
153 }
154 
155 const struct got_error *
156 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
157 {
158 	const char *obj_labels[] = {
159 		GOT_OBJ_LABEL_COMMIT,
160 		GOT_OBJ_LABEL_TREE,
161 		GOT_OBJ_LABEL_BLOB,
162 		GOT_OBJ_LABEL_TAG,
163 	};
164 	const int obj_types[] = {
165 		GOT_OBJ_TYPE_COMMIT,
166 		GOT_OBJ_TYPE_TREE,
167 		GOT_OBJ_TYPE_BLOB,
168 		GOT_OBJ_TYPE_TAG,
169 	};
170 	int type = 0;
171 	size_t size = 0, hdrlen = 0;
172 	size_t i;
173 
174 	*obj = NULL;
175 
176 	hdrlen = strnlen(buf, len) + 1 /* '\0' */;
177 	if (hdrlen > len)
178 		return got_error(GOT_ERR_BAD_OBJ_HDR);
179 
180 	for (i = 0; i < nitems(obj_labels); i++) {
181 		const char *label = obj_labels[i];
182 		size_t label_len = strlen(label);
183 		const char *errstr;
184 
185 		if (strncmp(buf, label, label_len) != 0)
186 			continue;
187 
188 		type = obj_types[i];
189 		if (len <= label_len)
190 			return got_error(GOT_ERR_BAD_OBJ_HDR);
191 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
192 		if (errstr != NULL)
193 			return got_error(GOT_ERR_BAD_OBJ_HDR);
194 		break;
195 	}
196 
197 	if (type == 0)
198 		return got_error(GOT_ERR_BAD_OBJ_HDR);
199 
200 	*obj = calloc(1, sizeof(**obj));
201 	if (*obj == NULL)
202 		return got_error_from_errno("calloc");
203 	(*obj)->type = type;
204 	(*obj)->hdrlen = hdrlen;
205 	(*obj)->size = size;
206 	return NULL;
207 }
208 
209 const struct got_error *
210 got_object_read_header(struct got_object **obj, int fd)
211 {
212 	const struct got_error *err;
213 	struct got_inflate_buf zb;
214 	uint8_t *buf;
215 	const size_t zbsize = 64;
216 	size_t outlen, totlen;
217 	int nbuf = 1;
218 
219 	*obj = NULL;
220 
221 	buf = malloc(zbsize);
222 	if (buf == NULL)
223 		return got_error_from_errno("malloc");
224 
225 	err = got_inflate_init(&zb, buf, zbsize, NULL);
226 	if (err)
227 		return err;
228 
229 	totlen = 0;
230 	do {
231 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
232 		if (err)
233 			goto done;
234 		if (outlen == 0)
235 			break;
236 		totlen += outlen;
237 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
238 			uint8_t *newbuf;
239 			nbuf++;
240 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
241 			if (newbuf == NULL) {
242 				err = got_error_from_errno("recallocarray");
243 				goto done;
244 			}
245 			buf = newbuf;
246 			zb.outbuf = newbuf + totlen;
247 			zb.outlen = (nbuf * zbsize) - totlen;
248 		}
249 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
250 
251 	err = got_object_parse_header(obj, buf, totlen);
252 done:
253 	free(buf);
254 	got_inflate_end(&zb);
255 	return err;
256 }
257 
258 struct got_commit_object *
259 got_object_commit_alloc_partial(void)
260 {
261 	struct got_commit_object *commit;
262 
263 	commit = calloc(1, sizeof(*commit));
264 	if (commit == NULL)
265 		return NULL;
266 	commit->tree_id = malloc(sizeof(*commit->tree_id));
267 	if (commit->tree_id == NULL) {
268 		free(commit);
269 		return NULL;
270 	}
271 
272 	STAILQ_INIT(&commit->parent_ids);
273 
274 	return commit;
275 }
276 
277 const struct got_error *
278 got_object_commit_add_parent(struct got_commit_object *commit,
279     const char *id_str)
280 {
281 	const struct got_error *err = NULL;
282 	struct got_object_qid *qid;
283 
284 	err = got_object_qid_alloc_partial(&qid);
285 	if (err)
286 		return err;
287 
288 	if (!got_parse_sha1_digest(qid->id->sha1, id_str)) {
289 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
290 		got_object_qid_free(qid);
291 		return err;
292 	}
293 
294 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
295 	commit->nparents++;
296 
297 	return NULL;
298 }
299 
300 static const struct got_error *
301 parse_gmtoff(time_t *gmtoff, const char *tzstr)
302 {
303 	int sign = 1;
304 	const char *p = tzstr;
305 	time_t h, m;
306 
307 	*gmtoff = 0;
308 
309 	if (*p == '-')
310 		sign = -1;
311 	else if (*p != '+')
312 		return got_error(GOT_ERR_BAD_OBJ_DATA);
313 	p++;
314 	if (!isdigit(*p) && !isdigit(*(p + 1)))
315 		return got_error(GOT_ERR_BAD_OBJ_DATA);
316 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
317 
318 	p += 2;
319 	if (!isdigit(*p) && !isdigit(*(p + 1)))
320 		return got_error(GOT_ERR_BAD_OBJ_DATA);
321 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
322 
323 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
324 	return NULL;
325 }
326 
327 static const struct got_error *
328 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
329 {
330 	const struct got_error *err = NULL;
331 	const char *errstr;
332 	char *space, *tzstr;
333 
334 	/* Parse and strip off trailing timezone indicator string. */
335 	space = strrchr(committer, ' ');
336 	if (space == NULL)
337 		return got_error(GOT_ERR_BAD_OBJ_DATA);
338 	tzstr = strdup(space + 1);
339 	if (tzstr == NULL)
340 		return got_error_from_errno("strdup");
341 	err = parse_gmtoff(gmtoff, tzstr);
342 	free(tzstr);
343 	if (err) { 
344 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
345 			return err;
346 		/* Old versions of Git omitted the timestamp. */
347 		*time = 0;
348 		*gmtoff = 0;
349 		return NULL;
350 	}
351 	*space = '\0';
352 
353 	/* Timestamp is separated from committer name + email by space. */
354 	space = strrchr(committer, ' ');
355 	if (space == NULL)
356 		return got_error(GOT_ERR_BAD_OBJ_DATA);
357 
358 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
359 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
360 	if (errstr)
361 		return got_error(GOT_ERR_BAD_OBJ_DATA);
362 
363 	/* Strip off parsed time information, leaving just author and email. */
364 	*space = '\0';
365 
366 	return NULL;
367 }
368 
369 void
370 got_object_commit_close(struct got_commit_object *commit)
371 {
372 	if (commit->refcnt > 0) {
373 		commit->refcnt--;
374 		if (commit->refcnt > 0)
375 			return;
376 	}
377 
378 	got_object_id_queue_free(&commit->parent_ids);
379 	free(commit->tree_id);
380 	free(commit->author);
381 	free(commit->committer);
382 	free(commit->logmsg);
383 	free(commit);
384 }
385 
386 struct got_object_id *
387 got_object_commit_get_tree_id(struct got_commit_object *commit)
388 {
389 	return commit->tree_id;
390 }
391 
392 int
393 got_object_commit_get_nparents(struct got_commit_object *commit)
394 {
395 	return commit->nparents;
396 }
397 
398 const struct got_object_id_queue *
399 got_object_commit_get_parent_ids(struct got_commit_object *commit)
400 {
401 	return &commit->parent_ids;
402 }
403 
404 const char *
405 got_object_commit_get_author(struct got_commit_object *commit)
406 {
407 	return commit->author;
408 }
409 
410 time_t
411 got_object_commit_get_author_time(struct got_commit_object *commit)
412 {
413 	return commit->author_time;
414 }
415 
416 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
417 {
418 	return commit->author_gmtoff;
419 }
420 
421 const char *
422 got_object_commit_get_committer(struct got_commit_object *commit)
423 {
424 	return commit->committer;
425 }
426 
427 time_t
428 got_object_commit_get_committer_time(struct got_commit_object *commit)
429 {
430 	return commit->committer_time;
431 }
432 
433 time_t
434 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
435 {
436 	return commit->committer_gmtoff;
437 }
438 
439 const struct got_error *
440 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
441 {
442 	const struct got_error *err = NULL;
443 	const char *src;
444 	char *dst;
445 	size_t len;
446 
447 	len = strlen(commit->logmsg);
448 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
449 	if (*logmsg == NULL)
450 		return got_error_from_errno("malloc");
451 
452 	/*
453 	 * Strip out unusual headers. Headers are separated from the commit
454 	 * message body by a single empty line.
455 	 */
456 	src = commit->logmsg;
457 	dst = *logmsg;
458 	while (*src != '\0' && *src != '\n') {
459 		int copy_header = 1, eol = 0;
460 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
461 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
462 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
463 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
464 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
465 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
466 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
467 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
468 			copy_header = 0;
469 
470 		while (*src != '\0' && !eol) {
471 			if (copy_header) {
472 				*dst = *src;
473 				dst++;
474 			}
475 			if (*src == '\n')
476 				eol = 1;
477 			src++;
478 		}
479 	}
480 	*dst = '\0';
481 
482 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
483 		err = got_error(GOT_ERR_NO_SPACE);
484 		goto done;
485 	}
486 
487 	/* Trim redundant trailing whitespace. */
488 	len = strlen(*logmsg);
489 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
490 	    isspace((unsigned char)(*logmsg)[len - 1])) {
491 		(*logmsg)[len - 1] = '\0';
492 		len--;
493 	}
494 
495 	/* Append a trailing newline if missing. */
496 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
497 		(*logmsg)[len] = '\n';
498 		(*logmsg)[len + 1] = '\0';
499 	}
500 done:
501 	if (err) {
502 		free(*logmsg);
503 		*logmsg = NULL;
504 	}
505 	return err;
506 }
507 
508 const char *
509 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
510 {
511 	return commit->logmsg;
512 }
513 
514 const struct got_error *
515 got_object_parse_commit(struct got_commit_object **commit, char *buf,
516     size_t len)
517 {
518 	const struct got_error *err = NULL;
519 	char *s = buf;
520 	size_t label_len;
521 	ssize_t remain = (ssize_t)len;
522 
523 	if (remain == 0)
524 		return got_error(GOT_ERR_BAD_OBJ_DATA);
525 
526 	*commit = got_object_commit_alloc_partial();
527 	if (*commit == NULL)
528 		return got_error_from_errno("got_object_commit_alloc_partial");
529 
530 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
531 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
532 		remain -= label_len;
533 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
534 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
535 			goto done;
536 		}
537 		s += label_len;
538 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
539 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
540 			goto done;
541 		}
542 		remain -= SHA1_DIGEST_STRING_LENGTH;
543 		s += SHA1_DIGEST_STRING_LENGTH;
544 	} else {
545 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
546 		goto done;
547 	}
548 
549 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
550 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
551 		remain -= label_len;
552 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
553 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
554 			goto done;
555 		}
556 		s += label_len;
557 		err = got_object_commit_add_parent(*commit, s);
558 		if (err)
559 			goto done;
560 
561 		remain -= SHA1_DIGEST_STRING_LENGTH;
562 		s += SHA1_DIGEST_STRING_LENGTH;
563 	}
564 
565 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
566 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
567 		char *p;
568 		size_t slen;
569 
570 		remain -= label_len;
571 		if (remain <= 0) {
572 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
573 			goto done;
574 		}
575 		s += label_len;
576 		p = memchr(s, '\n', remain);
577 		if (p == NULL) {
578 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
579 			goto done;
580 		}
581 		*p = '\0';
582 		slen = strlen(s);
583 		err = parse_commit_time(&(*commit)->author_time,
584 		    &(*commit)->author_gmtoff, s);
585 		if (err)
586 			goto done;
587 		(*commit)->author = strdup(s);
588 		if ((*commit)->author == NULL) {
589 			err = got_error_from_errno("strdup");
590 			goto done;
591 		}
592 		s += slen + 1;
593 		remain -= slen + 1;
594 	}
595 
596 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
597 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
598 		char *p;
599 		size_t slen;
600 
601 		remain -= label_len;
602 		if (remain <= 0) {
603 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
604 			goto done;
605 		}
606 		s += label_len;
607 		p = memchr(s, '\n', remain);
608 		if (p == NULL) {
609 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
610 			goto done;
611 		}
612 		*p = '\0';
613 		slen = strlen(s);
614 		err = parse_commit_time(&(*commit)->committer_time,
615 		    &(*commit)->committer_gmtoff, s);
616 		if (err)
617 			goto done;
618 		(*commit)->committer = strdup(s);
619 		if ((*commit)->committer == NULL) {
620 			err = got_error_from_errno("strdup");
621 			goto done;
622 		}
623 		s += slen + 1;
624 		remain -= slen + 1;
625 	}
626 
627 	(*commit)->logmsg = strndup(s, remain);
628 	if ((*commit)->logmsg == NULL) {
629 		err = got_error_from_errno("strndup");
630 		goto done;
631 	}
632 done:
633 	if (err) {
634 		got_object_commit_close(*commit);
635 		*commit = NULL;
636 	}
637 	return err;
638 }
639 
640 void
641 got_object_tree_close(struct got_tree_object *tree)
642 {
643 	if (tree->refcnt > 0) {
644 		tree->refcnt--;
645 		if (tree->refcnt > 0)
646 			return;
647 	}
648 
649 	free(tree->entries);
650 	free(tree);
651 }
652 
653 static const struct got_error *
654 parse_tree_entry(struct got_parsed_tree_entry **pte, const char **name,
655     size_t *elen, char *buf,
656     size_t maxlen)
657 {
658 	char *p, *space;
659 	const struct got_error *err = NULL;
660 
661 	*name = NULL;
662 	*elen = 0;
663 
664 	*pte = malloc(sizeof(**pte));
665 	if (*pte == NULL)
666 		return got_error_from_errno("malloc");
667 
668 	*elen = strnlen(buf, maxlen) + 1;
669 	if (*elen > maxlen) {
670 		free(*pte);
671 		*pte = NULL;
672 		return got_error(GOT_ERR_BAD_OBJ_DATA);
673 	}
674 
675 	space = memchr(buf, ' ', *elen);
676 	if (space == NULL || space <= buf) {
677 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
678 		free(*pte);
679 		*pte = NULL;
680 		return err;
681 	}
682 	(*pte)->mode = 0;
683 	p = buf;
684 	while (p < space) {
685 		if (*p < '0' && *p > '7') {
686 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
687 			goto done;
688 		}
689 		(*pte)->mode <<= 3;
690 		(*pte)->mode |= *p - '0';
691 		p++;
692 	}
693 
694 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH) {
695 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
696 		goto done;
697 	}
698 	*name = space + 1;
699 	buf += *elen;
700 	(*pte)->id = buf;
701 	*elen += SHA1_DIGEST_LENGTH;
702 done:
703 	if (err) {
704 		free(*pte);
705 		*pte = NULL;
706 	}
707 	return err;
708 }
709 
710 const struct got_error *
711 got_object_parse_tree(struct got_pathlist_head *entries, int *nentries,
712     uint8_t *buf, size_t len)
713 {
714 	const struct got_error *err = NULL;
715 	size_t remain = len;
716 
717 	*nentries = 0;
718 	if (remain == 0)
719 		return NULL; /* tree is empty */
720 
721 	while (remain > 0) {
722 		struct got_parsed_tree_entry *pte;
723 		struct got_pathlist_entry *new = NULL;
724 		const char *name;
725 		size_t elen;
726 
727 		err = parse_tree_entry(&pte, &name, &elen, buf, remain);
728 		if (err)
729 			goto done;
730 		err = got_pathlist_insert(&new, entries, name, pte);
731 		if (err)
732 			goto done;
733 		if (new == NULL) {
734 			err = got_error(GOT_ERR_TREE_DUP_ENTRY);
735 			goto done;
736 		}
737 		buf += elen;
738 		remain -= elen;
739 		(*nentries)++;
740 	}
741 
742 	if (remain != 0) {
743 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
744 		goto done;
745 	}
746 done:
747 	if (err) {
748 		got_object_parsed_tree_entries_free(entries);
749 		*nentries = 0;
750 	}
751 	return err;
752 }
753 
754 void
755 got_object_parsed_tree_entries_free(struct got_pathlist_head *entries)
756 {
757 	struct got_pathlist_entry *pe;
758 
759 	TAILQ_FOREACH(pe, entries, entry) {
760 		struct got_parsed_tree_entry *pte = pe->data;
761 		free(pte);
762 	}
763 	got_pathlist_free(entries);
764 }
765 
766 void
767 got_object_tag_close(struct got_tag_object *tag)
768 {
769 	if (tag->refcnt > 0) {
770 		tag->refcnt--;
771 		if (tag->refcnt > 0)
772 			return;
773 	}
774 
775 	free(tag->tag);
776 	free(tag->tagger);
777 	free(tag->tagmsg);
778 	free(tag);
779 }
780 
781 const struct got_error *
782 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
783 {
784 	const struct got_error *err = NULL;
785 	size_t remain = len;
786 	char *s = buf;
787 	size_t label_len;
788 
789 	if (remain == 0)
790 		return got_error(GOT_ERR_BAD_OBJ_DATA);
791 
792 	*tag = calloc(1, sizeof(**tag));
793 	if (*tag == NULL)
794 		return got_error_from_errno("calloc");
795 
796 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
797 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
798 		remain -= label_len;
799 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
800 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
801 			goto done;
802 		}
803 		s += label_len;
804 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
805 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
806 			goto done;
807 		}
808 		remain -= SHA1_DIGEST_STRING_LENGTH;
809 		s += SHA1_DIGEST_STRING_LENGTH;
810 	} else {
811 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
812 		goto done;
813 	}
814 
815 	if (remain <= 0) {
816 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
817 		goto done;
818 	}
819 
820 	label_len = strlen(GOT_TAG_LABEL_TYPE);
821 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
822 		remain -= label_len;
823 		if (remain <= 0) {
824 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
825 			goto done;
826 		}
827 		s += label_len;
828 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
829 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
830 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
831 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
832 			s += label_len;
833 			remain -= label_len;
834 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
835 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
836 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
837 			label_len = strlen(GOT_OBJ_LABEL_TREE);
838 			s += label_len;
839 			remain -= label_len;
840 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
841 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
842 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
843 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
844 			s += label_len;
845 			remain -= label_len;
846 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
847 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
848 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
849 			label_len = strlen(GOT_OBJ_LABEL_TAG);
850 			s += label_len;
851 			remain -= label_len;
852 		} else {
853 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
854 			goto done;
855 		}
856 
857 		if (remain <= 0 || *s != '\n') {
858 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
859 			goto done;
860 		}
861 		s++;
862 		remain--;
863 		if (remain <= 0) {
864 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
865 			goto done;
866 		}
867 	} else {
868 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
869 		goto done;
870 	}
871 
872 	label_len = strlen(GOT_TAG_LABEL_TAG);
873 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
874 		char *p;
875 		size_t slen;
876 		remain -= label_len;
877 		if (remain <= 0) {
878 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
879 			goto done;
880 		}
881 		s += label_len;
882 		p = memchr(s, '\n', remain);
883 		if (p == NULL) {
884 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
885 			goto done;
886 		}
887 		*p = '\0';
888 		slen = strlen(s);
889 		(*tag)->tag = strndup(s, slen);
890 		if ((*tag)->tag == NULL) {
891 			err = got_error_from_errno("strndup");
892 			goto done;
893 		}
894 		s += slen + 1;
895 		remain -= slen + 1;
896 		if (remain <= 0) {
897 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
898 			goto done;
899 		}
900 	} else {
901 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
902 		goto done;
903 	}
904 
905 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
906 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
907 		char *p;
908 		size_t slen;
909 
910 		remain -= label_len;
911 		if (remain <= 0) {
912 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
913 			goto done;
914 		}
915 		s += label_len;
916 		p = memchr(s, '\n', remain);
917 		if (p == NULL) {
918 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
919 			goto done;
920 		}
921 		*p = '\0';
922 		slen = strlen(s);
923 		err = parse_commit_time(&(*tag)->tagger_time,
924 		    &(*tag)->tagger_gmtoff, s);
925 		if (err)
926 			goto done;
927 		(*tag)->tagger = strdup(s);
928 		if ((*tag)->tagger == NULL) {
929 			err = got_error_from_errno("strdup");
930 			goto done;
931 		}
932 		s += slen + 1;
933 		remain -= slen + 1;
934 		if (remain < 0) {
935 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
936 			goto done;
937 		}
938 	} else {
939 		/* Some old tags in the Linux git repo have no tagger. */
940 		(*tag)->tagger = strdup("");
941 		if ((*tag)->tagger == NULL) {
942 			err = got_error_from_errno("strdup");
943 			goto done;
944 		}
945 	}
946 
947 	(*tag)->tagmsg = strndup(s, remain);
948 	if ((*tag)->tagmsg == NULL) {
949 		err = got_error_from_errno("strndup");
950 		goto done;
951 	}
952 done:
953 	if (err) {
954 		got_object_tag_close(*tag);
955 		*tag = NULL;
956 	}
957 	return err;
958 }
959 
960 const struct got_error *
961 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
962 {
963 	const struct got_error *err = NULL;
964 	static const size_t blocksize = 512;
965 	size_t n, total, remain;
966 	uint8_t *buf;
967 
968 	*outbuf = NULL;
969 	*outlen = 0;
970 
971 	buf = malloc(blocksize);
972 	if (buf == NULL)
973 		return got_error_from_errno("malloc");
974 
975 	remain = blocksize;
976 	total = 0;
977 	for (;;) {
978 		if (remain == 0) {
979 			uint8_t *newbuf;
980 			newbuf = reallocarray(buf, 1, total + blocksize);
981 			if (newbuf == NULL) {
982 				err = got_error_from_errno("reallocarray");
983 				goto done;
984 			}
985 			buf = newbuf;
986 			remain += blocksize;
987 		}
988 		n = fread(buf + total, 1, remain, f);
989 		if (n == 0) {
990 			if (ferror(f)) {
991 				err = got_ferror(f, GOT_ERR_IO);
992 				goto done;
993 			}
994 			break; /* EOF */
995 		}
996 		remain -= n;
997 		total += n;
998 	};
999 
1000 done:
1001 	if (err == NULL) {
1002 		*outbuf = buf;
1003 		*outlen = total;
1004 	} else
1005 		free(buf);
1006 	return err;
1007 }