1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/uio.h>
21 #include <sys/mman.h>
22 
23 #include <fcntl.h>
24 #include <errno.h>
25 #include <stdio.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <limits.h>
30 #include <sha1.h>
31 #include <endian.h>
32 #include <unistd.h>
33 #include <zlib.h>
34 #include <imsg.h>
35 
36 #include "got_error.h"
37 #include "got_object.h"
38 #include "got_path.h"
39 
40 #include "got_lib_sha1.h"
41 #include "got_lib_delta.h"
42 #include "got_lib_delta_cache.h"
43 #include "got_lib_inflate.h"
44 #include "got_lib_object.h"
45 #include "got_lib_object_parse.h"
46 #include "got_lib_privsep.h"
47 #include "got_lib_pack.h"
48 
49 #ifndef nitems
50 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
51 #endif
52 
53 #ifndef MIN
54 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
55 #endif
56 
57 static const struct got_error *
58 verify_fanout_table(uint32_t *fanout_table)
59 {
60 	int i;
61 
62 	for (i = 0; i < 0xff - 1; i++) {
63 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
64 			return got_error(GOT_ERR_BAD_PACKIDX);
65 	}
66 
67 	return NULL;
68 }
69 
70 const struct got_error *
71 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
72 {
73 	const struct got_error *err = NULL;
74 	struct got_packidx_v2_hdr *h;
75 	SHA1_CTX ctx;
76 	uint8_t sha1[SHA1_DIGEST_LENGTH];
77 	size_t nobj, len_fanout, len_ids, offset, remain;
78 	ssize_t n;
79 	int i;
80 
81 	SHA1Init(&ctx);
82 
83 	h = &p->hdr;
84 	offset = 0;
85 	remain = p->len;
86 
87 	if (remain < sizeof(*h->magic)) {
88 		err = got_error(GOT_ERR_BAD_PACKIDX);
89 		goto done;
90 	}
91 	if (p->map)
92 		h->magic = (uint32_t *)(p->map + offset);
93 	else {
94 		h->magic = malloc(sizeof(*h->magic));
95 		if (h->magic == NULL) {
96 			err = got_error_from_errno("malloc");
97 			goto done;
98 		}
99 		n = read(p->fd, h->magic, sizeof(*h->magic));
100 		if (n < 0) {
101 			err = got_error_from_errno("read");
102 			goto done;
103 		} else if (n != sizeof(*h->magic)) {
104 			err = got_error(GOT_ERR_BAD_PACKIDX);
105 			goto done;
106 		}
107 	}
108 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
109 		err = got_error(GOT_ERR_BAD_PACKIDX);
110 		goto done;
111 	}
112 	offset += sizeof(*h->magic);
113 	remain -= sizeof(*h->magic);
114 
115 	if (verify)
116 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
117 
118 	if (remain < sizeof(*h->version)) {
119 		err = got_error(GOT_ERR_BAD_PACKIDX);
120 		goto done;
121 	}
122 	if (p->map)
123 		h->version = (uint32_t *)(p->map + offset);
124 	else {
125 		h->version = malloc(sizeof(*h->version));
126 		if (h->version == NULL) {
127 			err = got_error_from_errno("malloc");
128 			goto done;
129 		}
130 		n = read(p->fd, h->version, sizeof(*h->version));
131 		if (n < 0) {
132 			err = got_error_from_errno("read");
133 			goto done;
134 		} else if (n != sizeof(*h->version)) {
135 			err = got_error(GOT_ERR_BAD_PACKIDX);
136 			goto done;
137 		}
138 	}
139 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
140 		err = got_error(GOT_ERR_BAD_PACKIDX);
141 		goto done;
142 	}
143 	offset += sizeof(*h->version);
144 	remain -= sizeof(*h->version);
145 
146 	if (verify)
147 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
148 
149 	len_fanout =
150 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
151 	if (remain < len_fanout) {
152 		err = got_error(GOT_ERR_BAD_PACKIDX);
153 		goto done;
154 	}
155 	if (p->map)
156 		h->fanout_table = (uint32_t *)(p->map + offset);
157 	else {
158 		h->fanout_table = malloc(len_fanout);
159 		if (h->fanout_table == NULL) {
160 			err = got_error_from_errno("malloc");
161 			goto done;
162 		}
163 		n = read(p->fd, h->fanout_table, len_fanout);
164 		if (n < 0) {
165 			err = got_error_from_errno("read");
166 			goto done;
167 		} else if (n != len_fanout) {
168 			err = got_error(GOT_ERR_BAD_PACKIDX);
169 			goto done;
170 		}
171 	}
172 	err = verify_fanout_table(h->fanout_table);
173 	if (err)
174 		goto done;
175 	if (verify)
176 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
177 	offset += len_fanout;
178 	remain -= len_fanout;
179 
180 	nobj = be32toh(h->fanout_table[0xff]);
181 	len_ids = nobj * sizeof(*h->sorted_ids);
182 	if (len_ids <= nobj || len_ids > remain) {
183 		err = got_error(GOT_ERR_BAD_PACKIDX);
184 		goto done;
185 	}
186 	if (p->map)
187 		h->sorted_ids =
188 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
189 	else {
190 		h->sorted_ids = malloc(len_ids);
191 		if (h->sorted_ids == NULL) {
192 			err = got_error(GOT_ERR_BAD_PACKIDX);
193 			goto done;
194 		}
195 		n = read(p->fd, h->sorted_ids, len_ids);
196 		if (n < 0)
197 			err = got_error_from_errno("read");
198 		else if (n != len_ids) {
199 			err = got_error(GOT_ERR_BAD_PACKIDX);
200 			goto done;
201 		}
202 	}
203 	if (verify)
204 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
205 	offset += len_ids;
206 	remain -= len_ids;
207 
208 	if (remain < nobj * sizeof(*h->crc32)) {
209 		err = got_error(GOT_ERR_BAD_PACKIDX);
210 		goto done;
211 	}
212 	if (p->map)
213 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
214 	else {
215 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
216 		if (h->crc32 == NULL) {
217 			err = got_error_from_errno("malloc");
218 			goto done;
219 		}
220 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
221 		if (n < 0)
222 			err = got_error_from_errno("read");
223 		else if (n != nobj * sizeof(*h->crc32)) {
224 			err = got_error(GOT_ERR_BAD_PACKIDX);
225 			goto done;
226 		}
227 	}
228 	if (verify)
229 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
230 	remain -= nobj * sizeof(*h->crc32);
231 	offset += nobj * sizeof(*h->crc32);
232 
233 	if (remain < nobj * sizeof(*h->offsets)) {
234 		err = got_error(GOT_ERR_BAD_PACKIDX);
235 		goto done;
236 	}
237 	if (p->map)
238 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
239 	else {
240 		h->offsets = malloc(nobj * sizeof(*h->offsets));
241 		if (h->offsets == NULL) {
242 			err = got_error_from_errno("malloc");
243 			goto done;
244 		}
245 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
246 		if (n < 0)
247 			err = got_error_from_errno("read");
248 		else if (n != nobj * sizeof(*h->offsets)) {
249 			err = got_error(GOT_ERR_BAD_PACKIDX);
250 			goto done;
251 		}
252 	}
253 	if (verify)
254 		SHA1Update(&ctx, (uint8_t *)h->offsets,
255 		    nobj * sizeof(*h->offsets));
256 	remain -= nobj * sizeof(*h->offsets);
257 	offset += nobj * sizeof(*h->offsets);
258 
259 	/* Large file offsets are contained only in files > 2GB. */
260 	if (verify || packfile_size > 0x7fffffff) {
261 		for (i = 0; i < nobj; i++) {
262 			uint32_t o = h->offsets[i];
263 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
264 				p->nlargeobj++;
265 		}
266 	}
267 	if (p->nlargeobj == 0)
268 		goto checksum;
269 	else if (packfile_size <= 0x7fffffff) {
270 		err = got_error(GOT_ERR_BAD_PACKIDX);
271 		goto done;
272 	}
273 
274 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
275 		err = got_error(GOT_ERR_BAD_PACKIDX);
276 		goto done;
277 	}
278 	if (p->map)
279 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
280 	else {
281 		h->large_offsets = malloc(p->nlargeobj *
282 		    sizeof(*h->large_offsets));
283 		if (h->large_offsets == NULL) {
284 			err = got_error_from_errno("malloc");
285 			goto done;
286 		}
287 		n = read(p->fd, h->large_offsets,
288 		    p->nlargeobj * sizeof(*h->large_offsets));
289 		if (n < 0)
290 			err = got_error_from_errno("read");
291 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
292 			err = got_error(GOT_ERR_BAD_PACKIDX);
293 			goto done;
294 		}
295 	}
296 	if (verify)
297 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
298 		    p->nlargeobj * sizeof(*h->large_offsets));
299 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
300 	offset += p->nlargeobj * sizeof(*h->large_offsets);
301 
302 checksum:
303 	if (remain < sizeof(*h->trailer)) {
304 		err = got_error(GOT_ERR_BAD_PACKIDX);
305 		goto done;
306 	}
307 	if (p->map)
308 		h->trailer =
309 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
310 	else {
311 		h->trailer = malloc(sizeof(*h->trailer));
312 		if (h->trailer == NULL) {
313 			err = got_error_from_errno("malloc");
314 			goto done;
315 		}
316 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
317 		if (n < 0)
318 			err = got_error_from_errno("read");
319 		else if (n != sizeof(*h->trailer)) {
320 			err = got_error(GOT_ERR_BAD_PACKIDX);
321 			goto done;
322 		}
323 	}
324 	if (verify) {
325 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
326 		SHA1Final(sha1, &ctx);
327 		if (memcmp(h->trailer->packidx_sha1, sha1,
328 		    SHA1_DIGEST_LENGTH) != 0)
329 			err = got_error(GOT_ERR_PACKIDX_CSUM);
330 	}
331 done:
332 	return err;
333 }
334 
335 const struct got_error *
336 got_packidx_open(struct got_packidx **packidx,
337     int dir_fd, const char *relpath, int verify)
338 {
339 	const struct got_error *err = NULL;
340 	struct got_packidx *p = NULL;
341 	char *pack_relpath;
342 	struct stat idx_sb, pack_sb;
343 
344 	*packidx = NULL;
345 
346 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
347 	if (err)
348 		return err;
349 
350 	/*
351 	 * Ensure that a corresponding pack file exists.
352 	 * Some Git repositories have this problem. Git seems to ignore
353 	 * the existence of lonely pack index files but we do not.
354 	 */
355 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
356 		if (errno == ENOENT) {
357 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
358 			    "%s", relpath);
359 		} else
360 			err = got_error_from_errno2("fstatat", pack_relpath);
361 		goto done;
362 	}
363 
364 	p = calloc(1, sizeof(*p));
365 	if (p == NULL) {
366 		err = got_error_from_errno("calloc");
367 		goto done;
368 	}
369 
370 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
371 	if (p->fd == -1) {
372 		err = got_error_from_errno2("openat", relpath);
373 		goto done;
374 	}
375 
376 	if (fstat(p->fd, &idx_sb) != 0) {
377 		err = got_error_from_errno2("fstat", relpath);
378 		goto done;
379 	}
380 	p->len = idx_sb.st_size;
381 	if (p->len < sizeof(p->hdr)) {
382 		err = got_error(GOT_ERR_BAD_PACKIDX);
383 		goto done;
384 	}
385 
386 	p->path_packidx = strdup(relpath);
387 	if (p->path_packidx == NULL) {
388 		err = got_error_from_errno("strdup");
389 		goto done;
390 	}
391 
392 #ifndef GOT_PACK_NO_MMAP
393 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
394 	if (p->map == MAP_FAILED) {
395 		if (errno != ENOMEM) {
396 			err = got_error_from_errno("mmap");
397 			goto done;
398 		}
399 		p->map = NULL; /* fall back to read(2) */
400 	}
401 #endif
402 
403 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
404 done:
405 	if (err) {
406 		if (p)
407 			got_packidx_close(p);
408 	} else
409 		*packidx = p;
410 	free(pack_relpath);
411 	return err;
412 }
413 
414 const struct got_error *
415 got_packidx_close(struct got_packidx *packidx)
416 {
417 	const struct got_error *err = NULL;
418 
419 	free(packidx->path_packidx);
420 	if (packidx->map) {
421 		if (munmap(packidx->map, packidx->len) == -1)
422 			err = got_error_from_errno("munmap");
423 	} else {
424 		free(packidx->hdr.magic);
425 		free(packidx->hdr.version);
426 		free(packidx->hdr.fanout_table);
427 		free(packidx->hdr.sorted_ids);
428 		free(packidx->hdr.crc32);
429 		free(packidx->hdr.offsets);
430 		free(packidx->hdr.large_offsets);
431 		free(packidx->hdr.trailer);
432 	}
433 	if (close(packidx->fd) == -1 && err == NULL)
434 		err = got_error_from_errno("close");
435 	free(packidx->sorted_offsets);
436 	free(packidx->sorted_large_offsets);
437 	free(packidx);
438 
439 	return err;
440 }
441 
442 const struct got_error *
443 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
444 {
445 	size_t size;
446 
447 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
448 	size = strlen(path_packidx) + 2;
449 	if (size < GOT_PACKFILE_NAMELEN + 1)
450 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
451 
452 	*path_packfile = malloc(size);
453 	if (*path_packfile == NULL)
454 		return got_error_from_errno("malloc");
455 
456 	/* Copy up to and excluding ".idx". */
457 	if (strlcpy(*path_packfile, path_packidx,
458 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
459 		return got_error(GOT_ERR_NO_SPACE);
460 
461 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
462 		return got_error(GOT_ERR_NO_SPACE);
463 
464 	return NULL;
465 }
466 
467 off_t
468 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
469 {
470 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
471 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
472 		uint64_t loffset;
473 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
474 		if (idx < 0 || idx >= packidx->nlargeobj ||
475 		    packidx->hdr.large_offsets == NULL)
476 			return -1;
477 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
478 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
479 	}
480 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
481 }
482 
483 int
484 got_packidx_get_object_idx(struct got_packidx *packidx,
485     struct got_object_id *id)
486 {
487 	u_int8_t id0 = id->sha1[0];
488 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
489 	int left = 0, right = totobj - 1;
490 
491 	if (id0 > 0)
492 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
493 
494 	while (left <= right) {
495 		struct got_packidx_object_id *oid;
496 		int i, cmp;
497 
498 		i = ((left + right) / 2);
499 		oid = &packidx->hdr.sorted_ids[i];
500 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
501 		if (cmp == 0)
502 			return i;
503 		else if (cmp > 0)
504 			left = i + 1;
505 		else if (cmp < 0)
506 			right = i - 1;
507 	}
508 
509 	return -1;
510 }
511 
512 static int
513 offset_cmp(const void *pa, const void *pb)
514 {
515 	const struct got_pack_offset_index *a, *b;
516 
517 	a = (const struct got_pack_offset_index *)pa;
518 	b = (const struct got_pack_offset_index *)pb;
519 
520 	if (a->offset < b->offset)
521 		return -1;
522 	else if (a->offset > b->offset)
523 		return 1;
524 
525 	return 0;
526 }
527 
528 static int
529 large_offset_cmp(const void *pa, const void *pb)
530 {
531 	const struct got_pack_large_offset_index *a, *b;
532 
533 	a = (const struct got_pack_large_offset_index *)pa;
534 	b = (const struct got_pack_large_offset_index *)pb;
535 
536 	if (a->offset < b->offset)
537 		return -1;
538 	else if (a->offset > b->offset)
539 		return 1;
540 
541 	return 0;
542 }
543 
544 static const struct got_error *
545 build_offset_index(struct got_packidx *p)
546 {
547 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
548 	unsigned int i, j, k;
549 
550 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
551 	    sizeof(p->sorted_offsets[0]));
552 	if (p->sorted_offsets == NULL)
553 		return got_error_from_errno("calloc");
554 
555 	if (p->nlargeobj > 0) {
556 		p->sorted_large_offsets = calloc(p->nlargeobj,
557 		    sizeof(p->sorted_large_offsets[0]));
558 		if (p->sorted_large_offsets == NULL)
559 			return got_error_from_errno("calloc");
560 	}
561 
562 	j = 0;
563 	k = 0;
564 	for (i = 0; i < nobj; i++) {
565 		uint32_t offset = be32toh(p->hdr.offsets[i]);
566 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
567 			uint64_t loffset;
568 			uint32_t idx;
569 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
570 			if (idx >= p->nlargeobj ||
571 			    p->nlargeobj == 0 ||
572 			    p->hdr.large_offsets == NULL)
573 				return got_error(GOT_ERR_BAD_PACKIDX);
574 			loffset = be64toh(p->hdr.large_offsets[idx]);
575 			p->sorted_large_offsets[j].offset = loffset;
576 			p->sorted_large_offsets[j].idx = i;
577 			j++;
578 		} else {
579 			p->sorted_offsets[k].offset = offset;
580 			p->sorted_offsets[k].idx = i;
581 			k++;
582 		}
583 	}
584 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
585 		return got_error(GOT_ERR_BAD_PACKIDX);
586 
587 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
588 	    sizeof(p->sorted_offsets[0]), offset_cmp);
589 
590 	if (p->sorted_large_offsets)
591 		qsort(p->sorted_large_offsets, p->nlargeobj,
592 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
593 
594 	return NULL;
595 }
596 
597 const struct got_error *
598 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
599 {
600 	const struct got_error *err;
601 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
602 	int i, left, right;
603 
604 	*idx = -1;
605 
606 	if (packidx->sorted_offsets == NULL) {
607 		err = build_offset_index(packidx);
608 		if (err)
609 			return err;
610 	}
611 
612 	if (offset >= 0x7fffffff) {
613 		uint64_t lo;
614 		left = 0, right = packidx->nlargeobj - 1;
615 		while (left <= right) {
616 			i = ((left + right) / 2);
617 			lo = packidx->sorted_large_offsets[i].offset;
618 			if (lo == offset) {
619 				*idx = packidx->sorted_large_offsets[i].idx;
620 				break;
621 			} else if (offset > lo)
622 				left = i + 1;
623 			else if (offset < lo)
624 				right = i - 1;
625 		}
626 	} else {
627 		uint32_t o;
628 		left = 0, right = totobj - packidx->nlargeobj - 1;
629 		while (left <= right) {
630 			i = ((left + right) / 2);
631 			o = packidx->sorted_offsets[i].offset;
632 			if (o == offset) {
633 				*idx = packidx->sorted_offsets[i].idx;
634 				break;
635 			} else if (offset > o)
636 				left = i + 1;
637 			else if (offset < o)
638 				right = i - 1;
639 		}
640 	}
641 
642 	return NULL;
643 }
644 
645 const struct got_error *
646 got_packidx_get_object_id(struct got_object_id *id,
647     struct got_packidx *packidx, int idx)
648 {
649 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
650 	struct got_packidx_object_id *oid;
651 
652 	if (idx < 0 || idx >= totobj)
653 		return got_error(GOT_ERR_NO_OBJ);
654 
655 	oid = &packidx->hdr.sorted_ids[idx];
656 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
657 	return NULL;
658 }
659 
660 const struct got_error *
661 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
662     struct got_packidx *packidx, const char *id_str_prefix)
663 {
664 	const struct got_error *err = NULL;
665 	u_int8_t id0;
666 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
667 	char hex[3];
668 	size_t prefix_len = strlen(id_str_prefix);
669 	struct got_packidx_object_id *oid;
670 	uint32_t i = 0;
671 
672 	STAILQ_INIT(matched_ids);
673 
674 	if (prefix_len < 2)
675 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
676 
677 	hex[0] = id_str_prefix[0];
678 	hex[1] = id_str_prefix[1];
679 	hex[2] = '\0';
680 	if (!got_parse_xdigit(&id0, hex))
681 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
682 
683 	if (id0 > 0)
684 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
685 	oid = &packidx->hdr.sorted_ids[i];
686 	while (i < totobj && oid->sha1[0] == id0) {
687 		char id_str[SHA1_DIGEST_STRING_LENGTH];
688 		struct got_object_qid *qid;
689 		int cmp;
690 
691 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
692 		        return got_error(GOT_ERR_NO_SPACE);
693 
694 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
695 		if (cmp < 0) {
696 			oid = &packidx->hdr.sorted_ids[++i];
697 			continue;
698 		} else if (cmp > 0)
699 			break;
700 
701 		err = got_object_qid_alloc_partial(&qid);
702 		if (err)
703 			break;
704 		memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
705 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
706 
707 		oid = &packidx->hdr.sorted_ids[++i];
708 	}
709 
710 	if (err)
711 		got_object_id_queue_free(matched_ids);
712 	return err;
713 }
714 
715 static const struct got_error *
716 pack_stop_privsep_child(struct got_pack *pack)
717 {
718 	const struct got_error *err = NULL;
719 
720 	if (pack->privsep_child == NULL)
721 		return NULL;
722 
723 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
724 	if (err)
725 		return err;
726 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
727 	if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
728 		err = got_error_from_errno("close");
729 	free(pack->privsep_child->ibuf);
730 	free(pack->privsep_child);
731 	pack->privsep_child = NULL;
732 	return err;
733 }
734 
735 const struct got_error *
736 got_pack_close(struct got_pack *pack)
737 {
738 	const struct got_error *err = NULL;
739 
740 	err = pack_stop_privsep_child(pack);
741 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
742 		err = got_error_from_errno("munmap");
743 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
744 		err = got_error_from_errno("close");
745 	pack->fd = -1;
746 	free(pack->path_packfile);
747 	pack->path_packfile = NULL;
748 	pack->filesize = 0;
749 	if (pack->delta_cache) {
750 		got_delta_cache_free(pack->delta_cache);
751 		pack->delta_cache = NULL;
752 	}
753 
754 	return err;
755 }
756 
757 const struct got_error *
758 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
759     struct got_pack *pack, off_t offset)
760 {
761 	uint8_t t = 0;
762 	uint64_t s = 0;
763 	uint8_t sizeN;
764 	size_t mapoff = 0;
765 	int i = 0;
766 
767 	*len = 0;
768 
769 	if (offset >= pack->filesize)
770 		return got_error(GOT_ERR_PACK_OFFSET);
771 
772 	if (pack->map) {
773 		mapoff = (size_t)offset;
774 	} else {
775 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
776 			return got_error_from_errno("lseek");
777 	}
778 
779 	do {
780 		/* We do not support size values which don't fit in 64 bit. */
781 		if (i > 9)
782 			return got_error(GOT_ERR_NO_SPACE);
783 
784 		if (pack->map) {
785 			if (mapoff + sizeof(sizeN) >= pack->filesize)
786 				return got_error(GOT_ERR_BAD_PACKFILE);
787 			sizeN = *(pack->map + mapoff);
788 			mapoff += sizeof(sizeN);
789 		} else {
790 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
791 			if (n < 0)
792 				return got_error_from_errno("read");
793 			if (n != sizeof(sizeN))
794 				return got_error(GOT_ERR_BAD_PACKFILE);
795 		}
796 		*len += sizeof(sizeN);
797 
798 		if (i == 0) {
799 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
800 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
801 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
802 		} else {
803 			size_t shift = 4 + 7 * (i - 1);
804 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
805 		}
806 		i++;
807 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
808 
809 	*type = t;
810 	*size = s;
811 	return NULL;
812 }
813 
814 static const struct got_error *
815 open_plain_object(struct got_object **obj, struct got_object_id *id,
816     uint8_t type, off_t offset, size_t size, int idx)
817 {
818 	*obj = calloc(1, sizeof(**obj));
819 	if (*obj == NULL)
820 		return got_error_from_errno("calloc");
821 
822 	(*obj)->type = type;
823 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
824 	(*obj)->pack_idx = idx;
825 	(*obj)->hdrlen = 0;
826 	(*obj)->size = size;
827 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
828 	(*obj)->pack_offset = offset;
829 
830 	return NULL;
831 }
832 
833 static const struct got_error *
834 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
835     off_t delta_offset)
836 {
837 	int64_t o = 0;
838 	uint8_t offN;
839 	int i = 0;
840 
841 	*offset = 0;
842 	*len = 0;
843 
844 	do {
845 		/* We do not support offset values which don't fit in 64 bit. */
846 		if (i > 8)
847 			return got_error(GOT_ERR_NO_SPACE);
848 
849 		if (pack->map) {
850 			size_t mapoff;
851 			mapoff = (size_t)delta_offset + *len;
852 			if (mapoff + sizeof(offN) >= pack->filesize)
853 				return got_error(GOT_ERR_PACK_OFFSET);
854 			offN = *(pack->map + mapoff);
855 		} else {
856 			ssize_t n;
857 			n = read(pack->fd, &offN, sizeof(offN));
858 			if (n < 0)
859 				return got_error_from_errno("read");
860 			if (n != sizeof(offN))
861 				return got_error(GOT_ERR_BAD_PACKFILE);
862 		}
863 		*len += sizeof(offN);
864 
865 		if (i == 0)
866 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
867 		else {
868 			o++;
869 			o <<= 7;
870 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
871 		}
872 		i++;
873 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
874 
875 	*offset = o;
876 	return NULL;
877 }
878 
879 const struct got_error *
880 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
881     struct got_pack *pack, off_t offset, int tslen)
882 {
883 	const struct got_error *err;
884 	int64_t negoffset;
885 	size_t negofflen;
886 
887 	*len = 0;
888 
889 	err = parse_negative_offset(&negoffset, &negofflen, pack,
890 	    offset + tslen);
891 	if (err)
892 		return err;
893 
894 	/* Compute the base object's offset (must be in the same pack file). */
895 	*base_offset = (offset - negoffset);
896 	if (*base_offset <= 0)
897 		return got_error(GOT_ERR_BAD_PACKFILE);
898 
899 	*len = negofflen;
900 	return NULL;
901 }
902 
903 static const struct got_error *
904 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
905     size_t *delta_compressed_len, size_t delta_data_offset,
906     struct got_pack *pack)
907 {
908 	const struct got_error *err = NULL;
909 	size_t consumed = 0;
910 
911 	if (pack->map) {
912 		if (delta_data_offset >= pack->filesize)
913 			return got_error(GOT_ERR_PACK_OFFSET);
914 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
915 		    &consumed, NULL, pack->map, delta_data_offset,
916 		    pack->filesize - delta_data_offset);
917 		if (err)
918 			return err;
919 	} else {
920 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
921 			return got_error_from_errno("lseek");
922 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
923 		    &consumed, NULL, 0, pack->fd);
924 		if (err)
925 			return err;
926 	}
927 
928 	if (delta_compressed_len)
929 		*delta_compressed_len = consumed;
930 
931 	return NULL;
932 }
933 
934 static const struct got_error *
935 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
936     int delta_type, size_t delta_size, size_t delta_data_offset)
937 {
938 	struct got_delta *delta;
939 
940 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
941 	    delta_data_offset);
942 	if (delta == NULL)
943 		return got_error_from_errno("got_delta_open");
944 	/* delta is freed in got_object_close() */
945 	deltas->nentries++;
946 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
947 	return NULL;
948 }
949 
950 static const struct got_error *
951 resolve_offset_delta(struct got_delta_chain *deltas,
952     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
953     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
954 
955 {
956 	const struct got_error *err;
957 	off_t base_offset;
958 	uint8_t base_type;
959 	uint64_t base_size;
960 	size_t base_tslen;
961 	off_t delta_data_offset;
962 	size_t consumed;
963 
964 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
965 	    delta_offset, tslen);
966 	if (err)
967 		return err;
968 
969 	delta_data_offset = delta_offset + tslen + consumed;
970 	if (delta_data_offset >= pack->filesize)
971 		return got_error(GOT_ERR_PACK_OFFSET);
972 
973 	if (pack->map == NULL) {
974 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
975 		if (delta_data_offset == -1)
976 			return got_error_from_errno("lseek");
977 	}
978 
979 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
980 	    delta_data_offset);
981 	if (err)
982 		return err;
983 
984 	/* An offset delta must be in the same packfile. */
985 	if (base_offset >= pack->filesize)
986 		return got_error(GOT_ERR_PACK_OFFSET);
987 
988 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
989 	    &base_tslen, pack, base_offset);
990 	if (err)
991 		return err;
992 
993 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
994 	    base_tslen, base_type, base_size, recursion - 1);
995 }
996 
997 const struct got_error *
998 got_pack_parse_ref_delta(struct got_object_id *id,
999     struct got_pack *pack, off_t delta_offset, int tslen)
1000 {
1001 	if (pack->map) {
1002 		size_t mapoff = delta_offset + tslen;
1003 		if (mapoff + sizeof(*id) >= pack->filesize)
1004 			return got_error(GOT_ERR_PACK_OFFSET);
1005 		memcpy(id, pack->map + mapoff, sizeof(*id));
1006 	} else {
1007 		ssize_t n;
1008 		n = read(pack->fd, id, sizeof(*id));
1009 		if (n < 0)
1010 			return got_error_from_errno("read");
1011 		if (n != sizeof(*id))
1012 			return got_error(GOT_ERR_BAD_PACKFILE);
1013 	}
1014 
1015 	return NULL;
1016 }
1017 
1018 static const struct got_error *
1019 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1020     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1021     size_t delta_size, unsigned int recursion)
1022 {
1023 	const struct got_error *err;
1024 	struct got_object_id id;
1025 	int idx;
1026 	off_t base_offset;
1027 	uint8_t base_type;
1028 	uint64_t base_size;
1029 	size_t base_tslen;
1030 	off_t delta_data_offset;
1031 
1032 	if (delta_offset + tslen >= pack->filesize)
1033 		return got_error(GOT_ERR_PACK_OFFSET);
1034 
1035 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1036 	if (err)
1037 		return err;
1038 	if (pack->map) {
1039 		delta_data_offset = delta_offset + tslen + sizeof(id);
1040 	} else {
1041 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1042 		if (delta_data_offset == -1)
1043 			return got_error_from_errno("lseek");
1044 	}
1045 
1046 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1047 	    delta_data_offset);
1048 	if (err)
1049 		return err;
1050 
1051 	/* Delta base must be in the same pack file. */
1052 	idx = got_packidx_get_object_idx(packidx, &id);
1053 	if (idx == -1)
1054 		return got_error(GOT_ERR_NO_OBJ);
1055 
1056 	base_offset = got_packidx_get_object_offset(packidx, idx);
1057 	if (base_offset == (uint64_t)-1)
1058 		return got_error(GOT_ERR_BAD_PACKIDX);
1059 
1060 	if (base_offset >= pack->filesize)
1061 		return got_error(GOT_ERR_PACK_OFFSET);
1062 
1063 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1064 	    &base_tslen, pack, base_offset);
1065 	if (err)
1066 		return err;
1067 
1068 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1069 	    base_tslen, base_type, base_size, recursion - 1);
1070 }
1071 
1072 const struct got_error *
1073 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1074     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1075     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1076 {
1077 	const struct got_error *err = NULL;
1078 
1079 	if (--recursion == 0)
1080 		return got_error(GOT_ERR_RECURSION);
1081 
1082 	switch (delta_type) {
1083 	case GOT_OBJ_TYPE_COMMIT:
1084 	case GOT_OBJ_TYPE_TREE:
1085 	case GOT_OBJ_TYPE_BLOB:
1086 	case GOT_OBJ_TYPE_TAG:
1087 		/* Plain types are the final delta base. Recursion ends. */
1088 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1089 		    delta_size, 0);
1090 		break;
1091 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1092 		err = resolve_offset_delta(deltas, packidx, pack,
1093 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1094 		break;
1095 	case GOT_OBJ_TYPE_REF_DELTA:
1096 		err = resolve_ref_delta(deltas, packidx, pack,
1097 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1098 		break;
1099 	default:
1100 		return got_error(GOT_ERR_OBJ_TYPE);
1101 	}
1102 
1103 	return err;
1104 }
1105 
1106 static const struct got_error *
1107 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1108     struct got_pack *pack, struct got_object_id *id, off_t offset,
1109     size_t tslen, int delta_type, size_t delta_size, int idx)
1110 {
1111 	const struct got_error *err = NULL;
1112 	int resolved_type;
1113 
1114 	*obj = calloc(1, sizeof(**obj));
1115 	if (*obj == NULL)
1116 		return got_error_from_errno("calloc");
1117 
1118 	(*obj)->flags = 0;
1119 	(*obj)->hdrlen = 0;
1120 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1121 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1122 	(*obj)->pack_offset = offset + tslen;
1123 
1124 	STAILQ_INIT(&(*obj)->deltas.entries);
1125 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1126 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1127 	(*obj)->pack_idx = idx;
1128 
1129 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1130 	    offset, tslen, delta_type, delta_size,
1131 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1132 	if (err)
1133 		goto done;
1134 
1135 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1136 	if (err)
1137 		goto done;
1138 	(*obj)->type = resolved_type;
1139 done:
1140 	if (err) {
1141 		got_object_close(*obj);
1142 		*obj = NULL;
1143 	}
1144 	return err;
1145 }
1146 
1147 const struct got_error *
1148 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1149     struct got_packidx *packidx, int idx, struct got_object_id *id)
1150 {
1151 	const struct got_error *err = NULL;
1152 	off_t offset;
1153 	uint8_t type;
1154 	uint64_t size;
1155 	size_t tslen;
1156 
1157 	*obj = NULL;
1158 
1159 	offset = got_packidx_get_object_offset(packidx, idx);
1160 	if (offset == (uint64_t)-1)
1161 		return got_error(GOT_ERR_BAD_PACKIDX);
1162 
1163 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1164 	    pack, offset);
1165 	if (err)
1166 		return err;
1167 
1168 	switch (type) {
1169 	case GOT_OBJ_TYPE_COMMIT:
1170 	case GOT_OBJ_TYPE_TREE:
1171 	case GOT_OBJ_TYPE_BLOB:
1172 	case GOT_OBJ_TYPE_TAG:
1173 		err = open_plain_object(obj, id, type, offset + tslen,
1174 		    size, idx);
1175 		break;
1176 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1177 	case GOT_OBJ_TYPE_REF_DELTA:
1178 		err = open_delta_object(obj, packidx, pack, id, offset,
1179 		    tslen, type, size, idx);
1180 		break;
1181 	default:
1182 		err = got_error(GOT_ERR_OBJ_TYPE);
1183 		break;
1184 	}
1185 
1186 	return err;
1187 }
1188 
1189 const struct got_error *
1190 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1191     struct got_delta_chain *deltas, struct got_pack *pack)
1192 {
1193 	struct got_delta *delta;
1194 	uint64_t base_size = 0, result_size = 0;
1195 
1196 	*max_size = 0;
1197 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1198 		/* Plain object types are the delta base. */
1199 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1200 		    delta->type != GOT_OBJ_TYPE_TREE &&
1201 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1202 		    delta->type != GOT_OBJ_TYPE_TAG) {
1203 			const struct got_error *err;
1204 			uint8_t *delta_buf;
1205 			size_t delta_len;
1206 			int cached = 1;
1207 
1208 			got_delta_cache_get(&delta_buf, &delta_len,
1209 			    pack->delta_cache, delta->data_offset);
1210 			if (delta_buf == NULL) {
1211 				cached = 0;
1212 				err = read_delta_data(&delta_buf, &delta_len,
1213 				    NULL, delta->data_offset, pack);
1214 				if (err)
1215 					return err;
1216 				err = got_delta_cache_add(pack->delta_cache,
1217 				    delta->data_offset, delta_buf, delta_len);
1218 				if (err == NULL)
1219 					cached = 1;
1220 				else if (err->code != GOT_ERR_NO_SPACE) {
1221 					free(delta_buf);
1222 					return err;
1223 				}
1224 			}
1225 			err = got_delta_get_sizes(&base_size, &result_size,
1226 			    delta_buf, delta_len);
1227 			if (!cached)
1228 				free(delta_buf);
1229 			if (err)
1230 				return err;
1231 		} else
1232 			base_size = delta->size;
1233 		if (base_size > *max_size)
1234 			*max_size = base_size;
1235 		if (result_size > *max_size)
1236 			*max_size = result_size;
1237 	}
1238 
1239 	return NULL;
1240 }
1241 
1242 const struct got_error *
1243 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1244     struct got_pack *pack)
1245 {
1246 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1247 		return got_error(GOT_ERR_OBJ_TYPE);
1248 
1249 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1250 }
1251 
1252 const struct got_error *
1253 got_pack_dump_delta_chain_to_file(size_t *result_size,
1254     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1255     FILE *base_file, FILE *accum_file)
1256 {
1257 	const struct got_error *err = NULL;
1258 	struct got_delta *delta;
1259 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1260 	size_t base_bufsz = 0, accum_size = 0, delta_len;
1261 	uint64_t max_size;
1262 	int n = 0;
1263 
1264 	*result_size = 0;
1265 
1266 	if (STAILQ_EMPTY(&deltas->entries))
1267 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1268 
1269 	/* We process small enough files entirely in memory for speed. */
1270 	err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1271 	if (err)
1272 		return err;
1273 	if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1274 		accum_buf = malloc(max_size);
1275 		if (accum_buf == NULL)
1276 			return got_error_from_errno("malloc");
1277 		base_file = NULL;
1278 		accum_file = NULL;
1279 	} else {
1280 		if (fseeko(base_file, 0L, SEEK_SET) == -1)
1281 			return got_error_from_errno("fseeko");
1282 		if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1283 			return got_error_from_errno("fseeko");
1284 	}
1285 
1286 	/* Deltas are ordered in ascending order. */
1287 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1288 		int cached = 1;
1289 		if (n == 0) {
1290 			size_t mapoff;
1291 			off_t delta_data_offset;
1292 
1293 			/* Plain object types are the delta base. */
1294 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1295 			    delta->type != GOT_OBJ_TYPE_TREE &&
1296 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1297 			    delta->type != GOT_OBJ_TYPE_TAG) {
1298 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1299 				goto done;
1300 			}
1301 
1302 			delta_data_offset = delta->offset + delta->tslen;
1303 			if (delta_data_offset >= pack->filesize) {
1304 				err = got_error(GOT_ERR_PACK_OFFSET);
1305 				goto done;
1306 			}
1307 			if (pack->map == NULL) {
1308 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1309 				    == -1) {
1310 					err = got_error_from_errno("lseek");
1311 					goto done;
1312 				}
1313 			}
1314 			if (base_file) {
1315 				if (pack->map) {
1316 					mapoff = (size_t)delta_data_offset;
1317 					err = got_inflate_to_file_mmap(
1318 					    &base_bufsz, NULL, NULL, pack->map,
1319 					    mapoff, pack->filesize - mapoff,
1320 					    base_file);
1321 				} else
1322 					err = got_inflate_to_file_fd(
1323 					    &base_bufsz, NULL, NULL, pack->fd,
1324 					    base_file);
1325 			} else {
1326 				if (pack->map) {
1327 					mapoff = (size_t)delta_data_offset;
1328 					err = got_inflate_to_mem_mmap(&base_buf,
1329 					    &base_bufsz, NULL, NULL,
1330 					    pack->map, mapoff,
1331 					    pack->filesize - mapoff);
1332 				} else
1333 					err = got_inflate_to_mem_fd(&base_buf,
1334 					    &base_bufsz, NULL, NULL, max_size,
1335 					    pack->fd);
1336 			}
1337 			if (err)
1338 				goto done;
1339 			n++;
1340 			if (base_file)
1341 				rewind(base_file);
1342 			continue;
1343 		}
1344 
1345 		got_delta_cache_get(&delta_buf, &delta_len,
1346 		    pack->delta_cache, delta->data_offset);
1347 		if (delta_buf == NULL) {
1348 			cached = 0;
1349 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1350 			    delta->data_offset, pack);
1351 			if (err)
1352 				goto done;
1353 			err = got_delta_cache_add(pack->delta_cache,
1354 			    delta->data_offset, delta_buf, delta_len);
1355 			if (err == NULL)
1356 				cached = 1;
1357 			else if (err->code != GOT_ERR_NO_SPACE) {
1358 				free(delta_buf);
1359 				goto done;
1360 			}
1361 		}
1362 		if (base_buf) {
1363 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1364 			    delta_buf, delta_len, accum_buf,
1365 			    &accum_size, max_size);
1366 			n++;
1367 		} else {
1368 			err = got_delta_apply(base_file, delta_buf,
1369 			    delta_len,
1370 			    /* Final delta application writes to output file. */
1371 			    ++n < deltas->nentries ? accum_file : outfile,
1372 			    &accum_size);
1373 		}
1374 		if (!cached)
1375 			free(delta_buf);
1376 		if (err)
1377 			goto done;
1378 
1379 		if (n < deltas->nentries) {
1380 			/* Accumulated delta becomes the new base. */
1381 			if (base_buf) {
1382 				uint8_t *tmp = accum_buf;
1383 				/*
1384 				 * Base buffer switches roles with accumulation
1385 				 * buffer. Ensure it can hold the largest
1386 				 * result in the delta chain. The initial
1387 				 * allocation might have been smaller.
1388 				 */
1389 				if (base_bufsz < max_size) {
1390 					uint8_t *p;
1391 					p = reallocarray(base_buf, 1, max_size);
1392 					if (p == NULL) {
1393 						err = got_error_from_errno(
1394 						    "reallocarray");
1395 						goto done;
1396 					}
1397 					base_buf = p;
1398 					base_bufsz = max_size;
1399 				}
1400 				accum_buf = base_buf;
1401 				base_buf = tmp;
1402 			} else {
1403 				FILE *tmp = accum_file;
1404 				accum_file = base_file;
1405 				base_file = tmp;
1406 				rewind(base_file);
1407 				rewind(accum_file);
1408 			}
1409 		}
1410 	}
1411 
1412 done:
1413 	free(base_buf);
1414 	if (accum_buf) {
1415 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1416 		free(accum_buf);
1417 		if (len != accum_size)
1418 			err = got_ferror(outfile, GOT_ERR_IO);
1419 	}
1420 	rewind(outfile);
1421 	if (err == NULL)
1422 		*result_size = accum_size;
1423 	return err;
1424 }
1425 
1426 const struct got_error *
1427 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1428     struct got_delta_chain *deltas, struct got_pack *pack)
1429 {
1430 	const struct got_error *err = NULL;
1431 	struct got_delta *delta;
1432 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1433 	size_t base_bufsz = 0, accum_size = 0, delta_len;
1434 	uint64_t max_size;
1435 	int n = 0;
1436 
1437 	*outbuf = NULL;
1438 	*outlen = 0;
1439 
1440 	if (STAILQ_EMPTY(&deltas->entries))
1441 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1442 
1443 	err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1444 	if (err)
1445 		return err;
1446 	accum_buf = malloc(max_size);
1447 	if (accum_buf == NULL)
1448 		return got_error_from_errno("malloc");
1449 
1450 	/* Deltas are ordered in ascending order. */
1451 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1452 		int cached = 1;
1453 		if (n == 0) {
1454 			size_t delta_data_offset;
1455 
1456 			/* Plain object types are the delta base. */
1457 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1458 			    delta->type != GOT_OBJ_TYPE_TREE &&
1459 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1460 			    delta->type != GOT_OBJ_TYPE_TAG) {
1461 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1462 				goto done;
1463 			}
1464 
1465 			delta_data_offset = delta->offset + delta->tslen;
1466 			if (delta_data_offset >= pack->filesize) {
1467 				err = got_error(GOT_ERR_PACK_OFFSET);
1468 				goto done;
1469 			}
1470 			if (pack->map) {
1471 				size_t mapoff = (size_t)delta_data_offset;
1472 				err = got_inflate_to_mem_mmap(&base_buf,
1473 				    &base_bufsz, NULL, NULL, pack->map,
1474 				    mapoff, pack->filesize - mapoff);
1475 			} else {
1476 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1477 				    == -1) {
1478 					err = got_error_from_errno("lseek");
1479 					goto done;
1480 				}
1481 				err = got_inflate_to_mem_fd(&base_buf,
1482 				    &base_bufsz, NULL, NULL, max_size,
1483 				    pack->fd);
1484 			}
1485 			if (err)
1486 				goto done;
1487 			n++;
1488 			continue;
1489 		}
1490 
1491 		got_delta_cache_get(&delta_buf, &delta_len,
1492 		    pack->delta_cache, delta->data_offset);
1493 		if (delta_buf == NULL) {
1494 			cached = 0;
1495 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1496 			    delta->data_offset, pack);
1497 			if (err)
1498 				goto done;
1499 			err = got_delta_cache_add(pack->delta_cache,
1500 			    delta->data_offset, delta_buf, delta_len);
1501 			if (err == NULL)
1502 				cached = 1;
1503 			else if (err->code != GOT_ERR_NO_SPACE) {
1504 				free(delta_buf);
1505 				goto done;
1506 			}
1507 		}
1508 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1509 		    delta_buf, delta_len, accum_buf,
1510 		    &accum_size, max_size);
1511 		if (!cached)
1512 			free(delta_buf);
1513 		n++;
1514 		if (err)
1515 			goto done;
1516 
1517 		if (n < deltas->nentries) {
1518 			/* Accumulated delta becomes the new base. */
1519 			uint8_t *tmp = accum_buf;
1520 			/*
1521 			 * Base buffer switches roles with accumulation buffer.
1522 			 * Ensure it can hold the largest result in the delta
1523 			 * chain. Initial allocation might have been smaller.
1524 			 */
1525 			if (base_bufsz < max_size) {
1526 				uint8_t *p;
1527 				p = reallocarray(base_buf, 1, max_size);
1528 				if (p == NULL) {
1529 					err = got_error_from_errno(
1530 					    "reallocarray");
1531 					goto done;
1532 				}
1533 				base_buf = p;
1534 				base_bufsz = max_size;
1535 			}
1536 			accum_buf = base_buf;
1537 			base_buf = tmp;
1538 		}
1539 	}
1540 
1541 done:
1542 	free(base_buf);
1543 	if (err) {
1544 		free(accum_buf);
1545 		*outbuf = NULL;
1546 		*outlen = 0;
1547 	} else {
1548 		*outbuf = accum_buf;
1549 		*outlen = accum_size;
1550 	}
1551 	return err;
1552 }
1553 
1554 const struct got_error *
1555 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1556     FILE *outfile, FILE *base_file, FILE *accum_file)
1557 {
1558 	const struct got_error *err = NULL;
1559 
1560 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1561 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1562 
1563 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1564 		if (obj->pack_offset >= pack->filesize)
1565 			return got_error(GOT_ERR_PACK_OFFSET);
1566 
1567 		if (pack->map) {
1568 			size_t mapoff = (size_t)obj->pack_offset;
1569 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1570 			    pack->map, mapoff, pack->filesize - mapoff,
1571 			    outfile);
1572 		} else {
1573 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1574 				return got_error_from_errno("lseek");
1575 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1576 			    pack->fd, outfile);
1577 		}
1578 	} else
1579 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1580 		    &obj->deltas, pack, outfile, base_file, accum_file);
1581 
1582 	return err;
1583 }
1584 
1585 const struct got_error *
1586 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1587     struct got_object *obj, struct got_pack *pack)
1588 {
1589 	const struct got_error *err = NULL;
1590 
1591 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1592 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1593 
1594 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1595 		if (obj->pack_offset >= pack->filesize)
1596 			return got_error(GOT_ERR_PACK_OFFSET);
1597 		if (pack->map) {
1598 			size_t mapoff = (size_t)obj->pack_offset;
1599 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1600 			    pack->map, mapoff, pack->filesize - mapoff);
1601 		} else {
1602 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1603 				return got_error_from_errno("lseek");
1604 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1605 			    obj->size, pack->fd);
1606 		}
1607 	} else
1608 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1609 		    pack);
1610 
1611 	return err;
1612 }
1613 
1614 static const struct got_error *
1615 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1616     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1617     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1618 {
1619 	const struct got_error *err = NULL;
1620 
1621 	/* Validate decompression and obtain the decompressed size. */
1622 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1623 	    delta_data_offset, pack);
1624 	if (err)
1625 		return err;
1626 
1627 	/* Read delta base/result sizes from head of delta stream. */
1628 	err = got_delta_get_sizes(base_size, result_size,
1629 	    *delta_buf, *delta_len);
1630 	if (err)
1631 		goto done;
1632 
1633 	/* Discard decompressed delta and read it again in compressed form. */
1634 	free(*delta_buf);
1635 	*delta_buf = malloc(*delta_len_compressed);
1636 	if (*delta_buf == NULL) {
1637 		err = got_error_from_errno("malloc");
1638 		goto done;
1639 	}
1640 	if (pack->map) {
1641 		if (delta_data_offset >= pack->filesize)
1642 			err = got_error(GOT_ERR_PACK_OFFSET);
1643 		memcpy(*delta_buf, pack->map + delta_data_offset,
1644 		    *delta_len_compressed);
1645 	} else {
1646 		ssize_t n;
1647 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1648 			err = got_error_from_errno("lseek");
1649 			goto done;
1650 		}
1651 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1652 		if (n < 0) {
1653 			err = got_error_from_errno("read");
1654 			goto done;
1655 		} else if (n != *delta_len_compressed) {
1656 			err = got_error(GOT_ERR_IO);
1657 			goto done;
1658 		}
1659 	}
1660 done:
1661 	if (err) {
1662 		free(*delta_buf);
1663 		*delta_buf = NULL;
1664 		*delta_len = 0;
1665 		*delta_len_compressed = 0;
1666 		*base_size = 0;
1667 		*result_size = 0;
1668 	}
1669 	return err;
1670 }
1671 
1672 const struct got_error *
1673 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1674     size_t *delta_compressed_size, off_t *delta_offset, off_t *base_offset,
1675     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1676     struct got_pack *pack, struct got_packidx *packidx, int idx)
1677 {
1678 	const struct got_error *err = NULL;
1679 	off_t offset;
1680 	uint8_t type;
1681 	uint64_t size;
1682 	size_t tslen, delta_hdrlen;
1683 	off_t delta_data_offset;
1684 
1685 	*delta_buf = NULL;
1686 	*delta_size = 0;
1687 	*delta_compressed_size = 0;
1688 	*delta_offset = 0;
1689 	*base_offset = 0;
1690 	*base_size = 0;
1691 	*result_size = 0;
1692 
1693 	offset = got_packidx_get_object_offset(packidx, idx);
1694 	if (offset == (uint64_t)-1)
1695 		return got_error(GOT_ERR_BAD_PACKIDX);
1696 
1697 	if (offset >= pack->filesize)
1698 		return got_error(GOT_ERR_PACK_OFFSET);
1699 
1700 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1701 	    pack, offset);
1702 	if (err)
1703 		return err;
1704 
1705 	if (tslen + size < tslen || offset + size < size ||
1706 	    tslen + offset < tslen)
1707 		return got_error(GOT_ERR_PACK_OFFSET);
1708 
1709 	switch (type) {
1710 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1711 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1712 		    pack, offset, tslen);
1713 		if (err)
1714 			return err;
1715 		break;
1716 	case GOT_OBJ_TYPE_REF_DELTA:
1717 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1718 		if (err)
1719 			return err;
1720 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1721 		break;
1722 	default:
1723 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1724 		    "non-delta object type %d found at offset %llu",
1725 		    type, offset);
1726 	}
1727 
1728 	if (tslen + delta_hdrlen < delta_hdrlen ||
1729 	    offset + delta_hdrlen < delta_hdrlen)
1730 		return got_error(GOT_ERR_BAD_DELTA);
1731 
1732 	delta_data_offset = offset + tslen + delta_hdrlen;
1733 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1734 	    base_size, result_size, delta_data_offset, pack, packidx);
1735 	if (err)
1736 		return err;
1737 
1738 	if (*delta_size != size) {
1739 		err = got_error(GOT_ERR_BAD_DELTA);
1740 		goto done;
1741 	}
1742 
1743 	*delta_offset = offset;
1744 done:
1745 	if (err) {
1746 		free(*delta_buf);
1747 		*delta_buf = NULL;
1748 		*delta_size = 0;
1749 		*delta_compressed_size = 0;
1750 		*delta_offset = 0;
1751 		*base_offset = 0;
1752 		*base_size = 0;
1753 		*result_size = 0;
1754 	}
1755 	return err;
1756 }