Gotweb

Blob

Date:: Sat Oct 16 11:09:49 2021 UTC
Message:: cache raw objects in order to speed up gotadmin pack
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 <zlib.h>
31 #include <ctype.h>
32 #include <limits.h>
33 #include <time.h>
34 #include <unistd.h>
35 
36 #include "got_compat.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 const struct got_error *
137 got_object_raw_close(struct got_raw_object *obj)
138 {
139 	const struct got_error *err = NULL;
140 
141 	if (obj->refcnt > 0) {
142 		obj->refcnt--;
143 		if (obj->refcnt > 0)
144 			return NULL;
145 	}
146 
147 	free(obj->read_buf);
148 	if (obj->f != NULL && fclose(obj->f) == EOF && err == NULL)
149 		err = got_error_from_errno("fclose");
150 	free(obj->data);
151 	free(obj);
152 	return err;
153 }
154 
155 void
156 got_object_qid_free(struct got_object_qid *qid)
157 {
158 	free(qid->id);
159 	free(qid);
160 }
161 
162 void
163 got_object_id_queue_free(struct got_object_id_queue *ids)
164 {
165 	struct got_object_qid *qid;
166 
167 	while (!STAILQ_EMPTY(ids)) {
168 		qid = STAILQ_FIRST(ids);
169 		STAILQ_REMOVE_HEAD(ids, entry);
170 		got_object_qid_free(qid);
171 	}
172 }
173 
174 const struct got_error *
175 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
176 {
177 	const char *obj_labels[] = {
178 		GOT_OBJ_LABEL_COMMIT,
179 		GOT_OBJ_LABEL_TREE,
180 		GOT_OBJ_LABEL_BLOB,
181 		GOT_OBJ_LABEL_TAG,
182 	};
183 	const int obj_types[] = {
184 		GOT_OBJ_TYPE_COMMIT,
185 		GOT_OBJ_TYPE_TREE,
186 		GOT_OBJ_TYPE_BLOB,
187 		GOT_OBJ_TYPE_TAG,
188 	};
189 	int type = 0;
190 	size_t size = 0, hdrlen = 0;
191 	size_t i;
192 
193 	*obj = NULL;
194 
195 	hdrlen = strnlen(buf, len) + 1 /* '\0' */;
196 	if (hdrlen > len)
197 		return got_error(GOT_ERR_BAD_OBJ_HDR);
198 
199 	for (i = 0; i < nitems(obj_labels); i++) {
200 		const char *label = obj_labels[i];
201 		size_t label_len = strlen(label);
202 		const char *errstr;
203 
204 		if (strncmp(buf, label, label_len) != 0)
205 			continue;
206 
207 		type = obj_types[i];
208 		if (len <= label_len)
209 			return got_error(GOT_ERR_BAD_OBJ_HDR);
210 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
211 		if (errstr != NULL)
212 			return got_error(GOT_ERR_BAD_OBJ_HDR);
213 		break;
214 	}
215 
216 	if (type == 0)
217 		return got_error(GOT_ERR_BAD_OBJ_HDR);
218 
219 	*obj = calloc(1, sizeof(**obj));
220 	if (*obj == NULL)
221 		return got_error_from_errno("calloc");
222 	(*obj)->type = type;
223 	(*obj)->hdrlen = hdrlen;
224 	(*obj)->size = size;
225 	return NULL;
226 }
227 
228 const struct got_error *
229 got_object_read_header(struct got_object **obj, int fd)
230 {
231 	const struct got_error *err;
232 	struct got_inflate_buf zb;
233 	uint8_t *buf;
234 	const size_t zbsize = 64;
235 	size_t outlen, totlen;
236 	int nbuf = 1;
237 
238 	*obj = NULL;
239 
240 	buf = malloc(zbsize);
241 	if (buf == NULL)
242 		return got_error_from_errno("malloc");
243 
244 	err = got_inflate_init(&zb, buf, zbsize, NULL);
245 	if (err)
246 		return err;
247 
248 	totlen = 0;
249 	do {
250 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
251 		if (err)
252 			goto done;
253 		if (outlen == 0)
254 			break;
255 		totlen += outlen;
256 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
257 			uint8_t *newbuf;
258 			nbuf++;
259 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
260 			if (newbuf == NULL) {
261 				err = got_error_from_errno("recallocarray");
262 				goto done;
263 			}
264 			buf = newbuf;
265 			zb.outbuf = newbuf + totlen;
266 			zb.outlen = (nbuf * zbsize) - totlen;
267 		}
268 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
269 
270 	err = got_object_parse_header(obj, buf, totlen);
271 done:
272 	free(buf);
273 	got_inflate_end(&zb);
274 	return err;
275 }
276 
277 struct got_commit_object *
278 got_object_commit_alloc_partial(void)
279 {
280 	struct got_commit_object *commit;
281 
282 	commit = calloc(1, sizeof(*commit));
283 	if (commit == NULL)
284 		return NULL;
285 	commit->tree_id = malloc(sizeof(*commit->tree_id));
286 	if (commit->tree_id == NULL) {
287 		free(commit);
288 		return NULL;
289 	}
290 
291 	STAILQ_INIT(&commit->parent_ids);
292 
293 	return commit;
294 }
295 
296 const struct got_error *
297 got_object_commit_add_parent(struct got_commit_object *commit,
298     const char *id_str)
299 {
300 	const struct got_error *err = NULL;
301 	struct got_object_qid *qid;
302 
303 	err = got_object_qid_alloc_partial(&qid);
304 	if (err)
305 		return err;
306 
307 	if (!got_parse_sha1_digest(qid->id->sha1, id_str)) {
308 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
309 		got_object_qid_free(qid);
310 		return err;
311 	}
312 
313 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
314 	commit->nparents++;
315 
316 	return NULL;
317 }
318 
319 static const struct got_error *
320 parse_gmtoff(time_t *gmtoff, const char *tzstr)
321 {
322 	int sign = 1;
323 	const char *p = tzstr;
324 	time_t h, m;
325 
326 	*gmtoff = 0;
327 
328 	if (*p == '-')
329 		sign = -1;
330 	else if (*p != '+')
331 		return got_error(GOT_ERR_BAD_OBJ_DATA);
332 	p++;
333 	if (!isdigit(*p) && !isdigit(*(p + 1)))
334 		return got_error(GOT_ERR_BAD_OBJ_DATA);
335 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
336 
337 	p += 2;
338 	if (!isdigit(*p) && !isdigit(*(p + 1)))
339 		return got_error(GOT_ERR_BAD_OBJ_DATA);
340 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
341 
342 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
343 	return NULL;
344 }
345 
346 static const struct got_error *
347 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
348 {
349 	const struct got_error *err = NULL;
350 	const char *errstr;
351 	char *space, *tzstr;
352 
353 	/* Parse and strip off trailing timezone indicator string. */
354 	space = strrchr(committer, ' ');
355 	if (space == NULL)
356 		return got_error(GOT_ERR_BAD_OBJ_DATA);
357 	tzstr = strdup(space + 1);
358 	if (tzstr == NULL)
359 		return got_error_from_errno("strdup");
360 	err = parse_gmtoff(gmtoff, tzstr);
361 	free(tzstr);
362 	if (err) { 
363 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
364 			return err;
365 		/* Old versions of Git omitted the timestamp. */
366 		*time = 0;
367 		*gmtoff = 0;
368 		return NULL;
369 	}
370 	*space = '\0';
371 
372 	/* Timestamp is separated from committer name + email by space. */
373 	space = strrchr(committer, ' ');
374 	if (space == NULL)
375 		return got_error(GOT_ERR_BAD_OBJ_DATA);
376 
377 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
378 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
379 	if (errstr)
380 		return got_error(GOT_ERR_BAD_OBJ_DATA);
381 
382 	/* Strip off parsed time information, leaving just author and email. */
383 	*space = '\0';
384 
385 	return NULL;
386 }
387 
388 void
389 got_object_commit_close(struct got_commit_object *commit)
390 {
391 	if (commit->refcnt > 0) {
392 		commit->refcnt--;
393 		if (commit->refcnt > 0)
394 			return;
395 	}
396 
397 	got_object_id_queue_free(&commit->parent_ids);
398 	free(commit->tree_id);
399 	free(commit->author);
400 	free(commit->committer);
401 	free(commit->logmsg);
402 	free(commit);
403 }
404 
405 struct got_object_id *
406 got_object_commit_get_tree_id(struct got_commit_object *commit)
407 {
408 	return commit->tree_id;
409 }
410 
411 int
412 got_object_commit_get_nparents(struct got_commit_object *commit)
413 {
414 	return commit->nparents;
415 }
416 
417 const struct got_object_id_queue *
418 got_object_commit_get_parent_ids(struct got_commit_object *commit)
419 {
420 	return &commit->parent_ids;
421 }
422 
423 const char *
424 got_object_commit_get_author(struct got_commit_object *commit)
425 {
426 	return commit->author;
427 }
428 
429 time_t
430 got_object_commit_get_author_time(struct got_commit_object *commit)
431 {
432 	return commit->author_time;
433 }
434 
435 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
436 {
437 	return commit->author_gmtoff;
438 }
439 
440 const char *
441 got_object_commit_get_committer(struct got_commit_object *commit)
442 {
443 	return commit->committer;
444 }
445 
446 time_t
447 got_object_commit_get_committer_time(struct got_commit_object *commit)
448 {
449 	return commit->committer_time;
450 }
451 
452 time_t
453 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
454 {
455 	return commit->committer_gmtoff;
456 }
457 
458 const struct got_error *
459 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
460 {
461 	const struct got_error *err = NULL;
462 	const char *src;
463 	char *dst;
464 	size_t len;
465 
466 	len = strlen(commit->logmsg);
467 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
468 	if (*logmsg == NULL)
469 		return got_error_from_errno("malloc");
470 
471 	/*
472 	 * Strip out unusual headers. Headers are separated from the commit
473 	 * message body by a single empty line.
474 	 */
475 	src = commit->logmsg;
476 	dst = *logmsg;
477 	while (*src != '\0' && *src != '\n') {
478 		int copy_header = 1, eol = 0;
479 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
480 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
481 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
482 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
483 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
484 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
485 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
486 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
487 			copy_header = 0;
488 
489 		while (*src != '\0' && !eol) {
490 			if (copy_header) {
491 				*dst = *src;
492 				dst++;
493 			}
494 			if (*src == '\n')
495 				eol = 1;
496 			src++;
497 		}
498 	}
499 	*dst = '\0';
500 
501 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
502 		err = got_error(GOT_ERR_NO_SPACE);
503 		goto done;
504 	}
505 
506 	/* Trim redundant trailing whitespace. */
507 	len = strlen(*logmsg);
508 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
509 	    isspace((unsigned char)(*logmsg)[len - 1])) {
510 		(*logmsg)[len - 1] = '\0';
511 		len--;
512 	}
513 
514 	/* Append a trailing newline if missing. */
515 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
516 		(*logmsg)[len] = '\n';
517 		(*logmsg)[len + 1] = '\0';
518 	}
519 done:
520 	if (err) {
521 		free(*logmsg);
522 		*logmsg = NULL;
523 	}
524 	return err;
525 }
526 
527 const char *
528 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
529 {
530 	return commit->logmsg;
531 }
532 
533 const struct got_error *
534 got_object_parse_commit(struct got_commit_object **commit, char *buf,
535     size_t len)
536 {
537 	const struct got_error *err = NULL;
538 	char *s = buf;
539 	size_t label_len;
540 	ssize_t remain = (ssize_t)len;
541 
542 	if (remain == 0)
543 		return got_error(GOT_ERR_BAD_OBJ_DATA);
544 
545 	*commit = got_object_commit_alloc_partial();
546 	if (*commit == NULL)
547 		return got_error_from_errno("got_object_commit_alloc_partial");
548 
549 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
550 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, 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 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
558 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
559 			goto done;
560 		}
561 		remain -= SHA1_DIGEST_STRING_LENGTH;
562 		s += SHA1_DIGEST_STRING_LENGTH;
563 	} else {
564 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
565 		goto done;
566 	}
567 
568 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
569 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
570 		remain -= label_len;
571 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
572 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
573 			goto done;
574 		}
575 		s += label_len;
576 		err = got_object_commit_add_parent(*commit, s);
577 		if (err)
578 			goto done;
579 
580 		remain -= SHA1_DIGEST_STRING_LENGTH;
581 		s += SHA1_DIGEST_STRING_LENGTH;
582 	}
583 
584 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
585 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
586 		char *p;
587 		size_t slen;
588 
589 		remain -= label_len;
590 		if (remain <= 0) {
591 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
592 			goto done;
593 		}
594 		s += label_len;
595 		p = memchr(s, '\n', remain);
596 		if (p == NULL) {
597 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
598 			goto done;
599 		}
600 		*p = '\0';
601 		slen = strlen(s);
602 		err = parse_commit_time(&(*commit)->author_time,
603 		    &(*commit)->author_gmtoff, s);
604 		if (err)
605 			goto done;
606 		(*commit)->author = strdup(s);
607 		if ((*commit)->author == NULL) {
608 			err = got_error_from_errno("strdup");
609 			goto done;
610 		}
611 		s += slen + 1;
612 		remain -= slen + 1;
613 	}
614 
615 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
616 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
617 		char *p;
618 		size_t slen;
619 
620 		remain -= label_len;
621 		if (remain <= 0) {
622 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
623 			goto done;
624 		}
625 		s += label_len;
626 		p = memchr(s, '\n', remain);
627 		if (p == NULL) {
628 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
629 			goto done;
630 		}
631 		*p = '\0';
632 		slen = strlen(s);
633 		err = parse_commit_time(&(*commit)->committer_time,
634 		    &(*commit)->committer_gmtoff, s);
635 		if (err)
636 			goto done;
637 		(*commit)->committer = strdup(s);
638 		if ((*commit)->committer == NULL) {
639 			err = got_error_from_errno("strdup");
640 			goto done;
641 		}
642 		s += slen + 1;
643 		remain -= slen + 1;
644 	}
645 
646 	(*commit)->logmsg = strndup(s, remain);
647 	if ((*commit)->logmsg == NULL) {
648 		err = got_error_from_errno("strndup");
649 		goto done;
650 	}
651 done:
652 	if (err) {
653 		got_object_commit_close(*commit);
654 		*commit = NULL;
655 	}
656 	return err;
657 }
658 
659 void
660 got_object_tree_close(struct got_tree_object *tree)
661 {
662 	if (tree->refcnt > 0) {
663 		tree->refcnt--;
664 		if (tree->refcnt > 0)
665 			return;
666 	}
667 
668 	free(tree->entries);
669 	free(tree);
670 }
671 
672 static const struct got_error *
673 parse_tree_entry(struct got_parsed_tree_entry **pte, const char **name,
674     size_t *elen, char *buf,
675     size_t maxlen)
676 {
677 	char *p, *space;
678 	const struct got_error *err = NULL;
679 
680 	*name = NULL;
681 	*elen = 0;
682 
683 	*pte = malloc(sizeof(**pte));
684 	if (*pte == NULL)
685 		return got_error_from_errno("malloc");
686 
687 	*elen = strnlen(buf, maxlen) + 1;
688 	if (*elen > maxlen) {
689 		free(*pte);
690 		*pte = NULL;
691 		return got_error(GOT_ERR_BAD_OBJ_DATA);
692 	}
693 
694 	space = memchr(buf, ' ', *elen);
695 	if (space == NULL || space <= buf) {
696 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
697 		free(*pte);
698 		*pte = NULL;
699 		return err;
700 	}
701 	(*pte)->mode = 0;
702 	p = buf;
703 	while (p < space) {
704 		if (*p < '0' && *p > '7') {
705 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
706 			goto done;
707 		}
708 		(*pte)->mode <<= 3;
709 		(*pte)->mode |= *p - '0';
710 		p++;
711 	}
712 
713 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH) {
714 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
715 		goto done;
716 	}
717 	*name = space + 1;
718 	buf += *elen;
719 	(*pte)->id = buf;
720 	*elen += SHA1_DIGEST_LENGTH;
721 done:
722 	if (err) {
723 		free(*pte);
724 		*pte = NULL;
725 	}
726 	return err;
727 }
728 
729 const struct got_error *
730 got_object_parse_tree(struct got_pathlist_head *entries, int *nentries,
731     uint8_t *buf, size_t len)
732 {
733 	const struct got_error *err = NULL;
734 	size_t remain = len;
735 
736 	*nentries = 0;
737 	if (remain == 0)
738 		return NULL; /* tree is empty */
739 
740 	while (remain > 0) {
741 		struct got_parsed_tree_entry *pte;
742 		struct got_pathlist_entry *new = NULL;
743 		const char *name;
744 		size_t elen;
745 
746 		err = parse_tree_entry(&pte, &name, &elen, buf, remain);
747 		if (err)
748 			goto done;
749 		err = got_pathlist_insert(&new, entries, name, pte);
750 		if (err)
751 			goto done;
752 		if (new == NULL) {
753 			err = got_error(GOT_ERR_TREE_DUP_ENTRY);
754 			goto done;
755 		}
756 		buf += elen;
757 		remain -= elen;
758 		(*nentries)++;
759 	}
760 
761 	if (remain != 0) {
762 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
763 		goto done;
764 	}
765 done:
766 	if (err) {
767 		got_object_parsed_tree_entries_free(entries);
768 		*nentries = 0;
769 	}
770 	return err;
771 }
772 
773 void
774 got_object_parsed_tree_entries_free(struct got_pathlist_head *entries)
775 {
776 	struct got_pathlist_entry *pe;
777 
778 	TAILQ_FOREACH(pe, entries, entry) {
779 		struct got_parsed_tree_entry *pte = pe->data;
780 		free(pte);
781 	}
782 	got_pathlist_free(entries);
783 }
784 
785 void
786 got_object_tag_close(struct got_tag_object *tag)
787 {
788 	if (tag->refcnt > 0) {
789 		tag->refcnt--;
790 		if (tag->refcnt > 0)
791 			return;
792 	}
793 
794 	free(tag->tag);
795 	free(tag->tagger);
796 	free(tag->tagmsg);
797 	free(tag);
798 }
799 
800 const struct got_error *
801 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
802 {
803 	const struct got_error *err = NULL;
804 	size_t remain = len;
805 	char *s = buf;
806 	size_t label_len;
807 
808 	if (remain == 0)
809 		return got_error(GOT_ERR_BAD_OBJ_DATA);
810 
811 	*tag = calloc(1, sizeof(**tag));
812 	if (*tag == NULL)
813 		return got_error_from_errno("calloc");
814 
815 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
816 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
817 		remain -= label_len;
818 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
819 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
820 			goto done;
821 		}
822 		s += label_len;
823 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
824 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
825 			goto done;
826 		}
827 		remain -= SHA1_DIGEST_STRING_LENGTH;
828 		s += SHA1_DIGEST_STRING_LENGTH;
829 	} else {
830 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
831 		goto done;
832 	}
833 
834 	if (remain <= 0) {
835 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
836 		goto done;
837 	}
838 
839 	label_len = strlen(GOT_TAG_LABEL_TYPE);
840 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
841 		remain -= label_len;
842 		if (remain <= 0) {
843 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
844 			goto done;
845 		}
846 		s += label_len;
847 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
848 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
849 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
850 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
851 			s += label_len;
852 			remain -= label_len;
853 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
854 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
855 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
856 			label_len = strlen(GOT_OBJ_LABEL_TREE);
857 			s += label_len;
858 			remain -= label_len;
859 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
860 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
861 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
862 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
863 			s += label_len;
864 			remain -= label_len;
865 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
866 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
867 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
868 			label_len = strlen(GOT_OBJ_LABEL_TAG);
869 			s += label_len;
870 			remain -= label_len;
871 		} else {
872 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
873 			goto done;
874 		}
875 
876 		if (remain <= 0 || *s != '\n') {
877 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
878 			goto done;
879 		}
880 		s++;
881 		remain--;
882 		if (remain <= 0) {
883 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
884 			goto done;
885 		}
886 	} else {
887 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
888 		goto done;
889 	}
890 
891 	label_len = strlen(GOT_TAG_LABEL_TAG);
892 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
893 		char *p;
894 		size_t slen;
895 		remain -= label_len;
896 		if (remain <= 0) {
897 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
898 			goto done;
899 		}
900 		s += label_len;
901 		p = memchr(s, '\n', remain);
902 		if (p == NULL) {
903 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
904 			goto done;
905 		}
906 		*p = '\0';
907 		slen = strlen(s);
908 		(*tag)->tag = strndup(s, slen);
909 		if ((*tag)->tag == NULL) {
910 			err = got_error_from_errno("strndup");
911 			goto done;
912 		}
913 		s += slen + 1;
914 		remain -= slen + 1;
915 		if (remain <= 0) {
916 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
917 			goto done;
918 		}
919 	} else {
920 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
921 		goto done;
922 	}
923 
924 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
925 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
926 		char *p;
927 		size_t slen;
928 
929 		remain -= label_len;
930 		if (remain <= 0) {
931 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
932 			goto done;
933 		}
934 		s += label_len;
935 		p = memchr(s, '\n', remain);
936 		if (p == NULL) {
937 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
938 			goto done;
939 		}
940 		*p = '\0';
941 		slen = strlen(s);
942 		err = parse_commit_time(&(*tag)->tagger_time,
943 		    &(*tag)->tagger_gmtoff, s);
944 		if (err)
945 			goto done;
946 		(*tag)->tagger = strdup(s);
947 		if ((*tag)->tagger == NULL) {
948 			err = got_error_from_errno("strdup");
949 			goto done;
950 		}
951 		s += slen + 1;
952 		remain -= slen + 1;
953 		if (remain < 0) {
954 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
955 			goto done;
956 		}
957 	} else {
958 		/* Some old tags in the Linux git repo have no tagger. */
959 		(*tag)->tagger = strdup("");
960 		if ((*tag)->tagger == NULL) {
961 			err = got_error_from_errno("strdup");
962 			goto done;
963 		}
964 	}
965 
966 	(*tag)->tagmsg = strndup(s, remain);
967 	if ((*tag)->tagmsg == NULL) {
968 		err = got_error_from_errno("strndup");
969 		goto done;
970 	}
971 done:
972 	if (err) {
973 		got_object_tag_close(*tag);
974 		*tag = NULL;
975 	}
976 	return err;
977 }
978 
979 const struct got_error *
980 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
981 {
982 	const struct got_error *err = NULL;
983 	static const size_t blocksize = 512;
984 	size_t n, total, remain;
985 	uint8_t *buf;
986 
987 	*outbuf = NULL;
988 	*outlen = 0;
989 
990 	buf = malloc(blocksize);
991 	if (buf == NULL)
992 		return got_error_from_errno("malloc");
993 
994 	remain = blocksize;
995 	total = 0;
996 	for (;;) {
997 		if (remain == 0) {
998 			uint8_t *newbuf;
999 			newbuf = reallocarray(buf, 1, total + blocksize);
1000 			if (newbuf == NULL) {
1001 				err = got_error_from_errno("reallocarray");
1002 				goto done;
1003 			}
1004 			buf = newbuf;
1005 			remain += blocksize;
1006 		}
1007 		n = fread(buf + total, 1, remain, f);
1008 		if (n == 0) {
1009 			if (ferror(f)) {
1010 				err = got_ferror(f, GOT_ERR_IO);
1011 				goto done;
1012 			}
1013 			break; /* EOF */
1014 		}
1015 		remain -= n;
1016 		total += n;
1017 	};
1018 
1019 done:
1020 	if (err == NULL) {
1021 		*outbuf = buf;
1022 		*outlen = total;
1023 	} else
1024 		free(buf);
1025 	return err;
1026 }