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 #include <sys/resource.h>
23 #include <sys/socket.h>
24 
25 #include <fcntl.h>
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdint.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <sha1.h>
33 #include <endian.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <imsg.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 
42 #include "got_lib_sha1.h"
43 #include "got_lib_delta.h"
44 #include "got_lib_delta_cache.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_parse.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_pack.h"
50 
51 #ifndef nitems
52 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
53 #endif
54 
55 #ifndef MIN
56 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #endif
58 
59 static const struct got_error *
60 verify_fanout_table(uint32_t *fanout_table)
61 {
62 	int i;
63 
64 	for (i = 0; i < 0xff - 1; i++) {
65 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
66 			return got_error(GOT_ERR_BAD_PACKIDX);
67 	}
68 
69 	return NULL;
70 }
71 
72 const struct got_error *
73 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
74 {
75 	const struct got_error *err = NULL;
76 	struct got_packidx_v2_hdr *h;
77 	SHA1_CTX ctx;
78 	uint8_t sha1[SHA1_DIGEST_LENGTH];
79 	size_t nobj, len_fanout, len_ids, offset, remain;
80 	ssize_t n;
81 	int i;
82 
83 	SHA1Init(&ctx);
84 
85 	h = &p->hdr;
86 	offset = 0;
87 	remain = p->len;
88 
89 	if (remain < sizeof(*h->magic)) {
90 		err = got_error(GOT_ERR_BAD_PACKIDX);
91 		goto done;
92 	}
93 	if (p->map)
94 		h->magic = (uint32_t *)(p->map + offset);
95 	else {
96 		h->magic = malloc(sizeof(*h->magic));
97 		if (h->magic == NULL) {
98 			err = got_error_from_errno("malloc");
99 			goto done;
100 		}
101 		n = read(p->fd, h->magic, sizeof(*h->magic));
102 		if (n < 0) {
103 			err = got_error_from_errno("read");
104 			goto done;
105 		} else if (n != sizeof(*h->magic)) {
106 			err = got_error(GOT_ERR_BAD_PACKIDX);
107 			goto done;
108 		}
109 	}
110 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
111 		err = got_error(GOT_ERR_BAD_PACKIDX);
112 		goto done;
113 	}
114 	offset += sizeof(*h->magic);
115 	remain -= sizeof(*h->magic);
116 
117 	if (verify)
118 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
119 
120 	if (remain < sizeof(*h->version)) {
121 		err = got_error(GOT_ERR_BAD_PACKIDX);
122 		goto done;
123 	}
124 	if (p->map)
125 		h->version = (uint32_t *)(p->map + offset);
126 	else {
127 		h->version = malloc(sizeof(*h->version));
128 		if (h->version == NULL) {
129 			err = got_error_from_errno("malloc");
130 			goto done;
131 		}
132 		n = read(p->fd, h->version, sizeof(*h->version));
133 		if (n < 0) {
134 			err = got_error_from_errno("read");
135 			goto done;
136 		} else if (n != sizeof(*h->version)) {
137 			err = got_error(GOT_ERR_BAD_PACKIDX);
138 			goto done;
139 		}
140 	}
141 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
142 		err = got_error(GOT_ERR_BAD_PACKIDX);
143 		goto done;
144 	}
145 	offset += sizeof(*h->version);
146 	remain -= sizeof(*h->version);
147 
148 	if (verify)
149 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
150 
151 	len_fanout =
152 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
153 	if (remain < len_fanout) {
154 		err = got_error(GOT_ERR_BAD_PACKIDX);
155 		goto done;
156 	}
157 	if (p->map)
158 		h->fanout_table = (uint32_t *)(p->map + offset);
159 	else {
160 		h->fanout_table = malloc(len_fanout);
161 		if (h->fanout_table == NULL) {
162 			err = got_error_from_errno("malloc");
163 			goto done;
164 		}
165 		n = read(p->fd, h->fanout_table, len_fanout);
166 		if (n < 0) {
167 			err = got_error_from_errno("read");
168 			goto done;
169 		} else if (n != len_fanout) {
170 			err = got_error(GOT_ERR_BAD_PACKIDX);
171 			goto done;
172 		}
173 	}
174 	err = verify_fanout_table(h->fanout_table);
175 	if (err)
176 		goto done;
177 	if (verify)
178 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
179 	offset += len_fanout;
180 	remain -= len_fanout;
181 
182 	nobj = be32toh(h->fanout_table[0xff]);
183 	len_ids = nobj * sizeof(*h->sorted_ids);
184 	if (len_ids <= nobj || len_ids > remain) {
185 		err = got_error(GOT_ERR_BAD_PACKIDX);
186 		goto done;
187 	}
188 	if (p->map)
189 		h->sorted_ids =
190 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
191 	else {
192 		h->sorted_ids = malloc(len_ids);
193 		if (h->sorted_ids == NULL) {
194 			err = got_error(GOT_ERR_BAD_PACKIDX);
195 			goto done;
196 		}
197 		n = read(p->fd, h->sorted_ids, len_ids);
198 		if (n < 0)
199 			err = got_error_from_errno("read");
200 		else if (n != len_ids) {
201 			err = got_error(GOT_ERR_BAD_PACKIDX);
202 			goto done;
203 		}
204 	}
205 	if (verify)
206 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
207 	offset += len_ids;
208 	remain -= len_ids;
209 
210 	if (remain < nobj * sizeof(*h->crc32)) {
211 		err = got_error(GOT_ERR_BAD_PACKIDX);
212 		goto done;
213 	}
214 	if (p->map)
215 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
216 	else {
217 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
218 		if (h->crc32 == NULL) {
219 			err = got_error_from_errno("malloc");
220 			goto done;
221 		}
222 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
223 		if (n < 0)
224 			err = got_error_from_errno("read");
225 		else if (n != nobj * sizeof(*h->crc32)) {
226 			err = got_error(GOT_ERR_BAD_PACKIDX);
227 			goto done;
228 		}
229 	}
230 	if (verify)
231 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
232 	remain -= nobj * sizeof(*h->crc32);
233 	offset += nobj * sizeof(*h->crc32);
234 
235 	if (remain < nobj * sizeof(*h->offsets)) {
236 		err = got_error(GOT_ERR_BAD_PACKIDX);
237 		goto done;
238 	}
239 	if (p->map)
240 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
241 	else {
242 		h->offsets = malloc(nobj * sizeof(*h->offsets));
243 		if (h->offsets == NULL) {
244 			err = got_error_from_errno("malloc");
245 			goto done;
246 		}
247 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
248 		if (n < 0)
249 			err = got_error_from_errno("read");
250 		else if (n != nobj * sizeof(*h->offsets)) {
251 			err = got_error(GOT_ERR_BAD_PACKIDX);
252 			goto done;
253 		}
254 	}
255 	if (verify)
256 		SHA1Update(&ctx, (uint8_t *)h->offsets,
257 		    nobj * sizeof(*h->offsets));
258 	remain -= nobj * sizeof(*h->offsets);
259 	offset += nobj * sizeof(*h->offsets);
260 
261 	/* Large file offsets are contained only in files > 2GB. */
262 	if (verify || packfile_size > 0x7fffffff) {
263 		for (i = 0; i < nobj; i++) {
264 			uint32_t o = h->offsets[i];
265 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
266 				p->nlargeobj++;
267 		}
268 	}
269 	if (p->nlargeobj == 0)
270 		goto checksum;
271 	else if (packfile_size <= 0x7fffffff) {
272 		err = got_error(GOT_ERR_BAD_PACKIDX);
273 		goto done;
274 	}
275 
276 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
277 		err = got_error(GOT_ERR_BAD_PACKIDX);
278 		goto done;
279 	}
280 	if (p->map)
281 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
282 	else {
283 		h->large_offsets = malloc(p->nlargeobj *
284 		    sizeof(*h->large_offsets));
285 		if (h->large_offsets == NULL) {
286 			err = got_error_from_errno("malloc");
287 			goto done;
288 		}
289 		n = read(p->fd, h->large_offsets,
290 		    p->nlargeobj * sizeof(*h->large_offsets));
291 		if (n < 0)
292 			err = got_error_from_errno("read");
293 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
294 			err = got_error(GOT_ERR_BAD_PACKIDX);
295 			goto done;
296 		}
297 	}
298 	if (verify)
299 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
300 		    p->nlargeobj * sizeof(*h->large_offsets));
301 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
302 	offset += p->nlargeobj * sizeof(*h->large_offsets);
303 
304 checksum:
305 	if (remain < sizeof(*h->trailer)) {
306 		err = got_error(GOT_ERR_BAD_PACKIDX);
307 		goto done;
308 	}
309 	if (p->map)
310 		h->trailer =
311 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
312 	else {
313 		h->trailer = malloc(sizeof(*h->trailer));
314 		if (h->trailer == NULL) {
315 			err = got_error_from_errno("malloc");
316 			goto done;
317 		}
318 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
319 		if (n < 0)
320 			err = got_error_from_errno("read");
321 		else if (n != sizeof(*h->trailer)) {
322 			err = got_error(GOT_ERR_BAD_PACKIDX);
323 			goto done;
324 		}
325 	}
326 	if (verify) {
327 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
328 		SHA1Final(sha1, &ctx);
329 		if (memcmp(h->trailer->packidx_sha1, sha1,
330 		    SHA1_DIGEST_LENGTH) != 0)
331 			err = got_error(GOT_ERR_PACKIDX_CSUM);
332 	}
333 done:
334 	return err;
335 }
336 
337 const struct got_error *
338 got_packidx_open(struct got_packidx **packidx,
339     int dir_fd, const char *relpath, int verify)
340 {
341 	const struct got_error *err = NULL;
342 	struct got_packidx *p = NULL;
343 	char *pack_relpath;
344 	struct stat idx_sb, pack_sb;
345 
346 	*packidx = NULL;
347 
348 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
349 	if (err)
350 		return err;
351 
352 	/*
353 	 * Ensure that a corresponding pack file exists.
354 	 * Some Git repositories have this problem. Git seems to ignore
355 	 * the existence of lonely pack index files but we do not.
356 	 */
357 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
358 		if (errno == ENOENT) {
359 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
360 			    "%s", relpath);
361 		} else
362 			err = got_error_from_errno2("fstatat", pack_relpath);
363 		goto done;
364 	}
365 
366 	p = calloc(1, sizeof(*p));
367 	if (p == NULL) {
368 		err = got_error_from_errno("calloc");
369 		goto done;
370 	}
371 
372 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
373 	if (p->fd == -1) {
374 		err = got_error_from_errno2("openat", relpath);
375 		goto done;
376 	}
377 
378 	if (fstat(p->fd, &idx_sb) != 0) {
379 		err = got_error_from_errno2("fstat", relpath);
380 		goto done;
381 	}
382 	p->len = idx_sb.st_size;
383 	if (p->len < sizeof(p->hdr)) {
384 		err = got_error(GOT_ERR_BAD_PACKIDX);
385 		goto done;
386 	}
387 
388 	p->path_packidx = strdup(relpath);
389 	if (p->path_packidx == NULL) {
390 		err = got_error_from_errno("strdup");
391 		goto done;
392 	}
393 
394 #ifndef GOT_PACK_NO_MMAP
395 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
396 	if (p->map == MAP_FAILED) {
397 		if (errno != ENOMEM) {
398 			err = got_error_from_errno("mmap");
399 			goto done;
400 		}
401 		p->map = NULL; /* fall back to read(2) */
402 	}
403 #endif
404 
405 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
406 done:
407 	if (err) {
408 		if (p)
409 			got_packidx_close(p);
410 	} else
411 		*packidx = p;
412 	free(pack_relpath);
413 	return err;
414 }
415 
416 const struct got_error *
417 got_packidx_close(struct got_packidx *packidx)
418 {
419 	const struct got_error *err = NULL;
420 
421 	free(packidx->path_packidx);
422 	if (packidx->map) {
423 		if (munmap(packidx->map, packidx->len) == -1)
424 			err = got_error_from_errno("munmap");
425 	} else {
426 		free(packidx->hdr.magic);
427 		free(packidx->hdr.version);
428 		free(packidx->hdr.fanout_table);
429 		free(packidx->hdr.sorted_ids);
430 		free(packidx->hdr.crc32);
431 		free(packidx->hdr.offsets);
432 		free(packidx->hdr.large_offsets);
433 		free(packidx->hdr.trailer);
434 	}
435 	if (close(packidx->fd) == -1 && err == NULL)
436 		err = got_error_from_errno("close");
437 	free(packidx->sorted_offsets);
438 	free(packidx->sorted_large_offsets);
439 	free(packidx);
440 
441 	return err;
442 }
443 
444 const struct got_error *
445 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
446 {
447 	size_t size;
448 
449 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
450 	size = strlen(path_packidx) + 2;
451 	if (size < GOT_PACKFILE_NAMELEN + 1)
452 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
453 
454 	*path_packfile = malloc(size);
455 	if (*path_packfile == NULL)
456 		return got_error_from_errno("malloc");
457 
458 	/* Copy up to and excluding ".idx". */
459 	if (strlcpy(*path_packfile, path_packidx,
460 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
461 		return got_error(GOT_ERR_NO_SPACE);
462 
463 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
464 		return got_error(GOT_ERR_NO_SPACE);
465 
466 	return NULL;
467 }
468 
469 off_t
470 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
471 {
472 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
473 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
474 		uint64_t loffset;
475 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
476 		if (idx < 0 || idx >= packidx->nlargeobj ||
477 		    packidx->hdr.large_offsets == NULL)
478 			return -1;
479 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
480 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
481 	}
482 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
483 }
484 
485 int
486 got_packidx_get_object_idx(struct got_packidx *packidx,
487     struct got_object_id *id)
488 {
489 	u_int8_t id0 = id->sha1[0];
490 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
491 	int left = 0, right = totobj - 1;
492 
493 	if (id0 > 0)
494 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
495 
496 	while (left <= right) {
497 		struct got_packidx_object_id *oid;
498 		int i, cmp;
499 
500 		i = ((left + right) / 2);
501 		oid = &packidx->hdr.sorted_ids[i];
502 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
503 		if (cmp == 0)
504 			return i;
505 		else if (cmp > 0)
506 			left = i + 1;
507 		else if (cmp < 0)
508 			right = i - 1;
509 	}
510 
511 	return -1;
512 }
513 
514 static int
515 offset_cmp(const void *pa, const void *pb)
516 {
517 	const struct got_pack_offset_index *a, *b;
518 
519 	a = (const struct got_pack_offset_index *)pa;
520 	b = (const struct got_pack_offset_index *)pb;
521 
522 	if (a->offset < b->offset)
523 		return -1;
524 	else if (a->offset > b->offset)
525 		return 1;
526 
527 	return 0;
528 }
529 
530 static int
531 large_offset_cmp(const void *pa, const void *pb)
532 {
533 	const struct got_pack_large_offset_index *a, *b;
534 
535 	a = (const struct got_pack_large_offset_index *)pa;
536 	b = (const struct got_pack_large_offset_index *)pb;
537 
538 	if (a->offset < b->offset)
539 		return -1;
540 	else if (a->offset > b->offset)
541 		return 1;
542 
543 	return 0;
544 }
545 
546 static const struct got_error *
547 build_offset_index(struct got_packidx *p)
548 {
549 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
550 	unsigned int i, j, k;
551 
552 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
553 	    sizeof(p->sorted_offsets[0]));
554 	if (p->sorted_offsets == NULL)
555 		return got_error_from_errno("calloc");
556 
557 	if (p->nlargeobj > 0) {
558 		p->sorted_large_offsets = calloc(p->nlargeobj,
559 		    sizeof(p->sorted_large_offsets[0]));
560 		if (p->sorted_large_offsets == NULL)
561 			return got_error_from_errno("calloc");
562 	}
563 
564 	j = 0;
565 	k = 0;
566 	for (i = 0; i < nobj; i++) {
567 		uint32_t offset = be32toh(p->hdr.offsets[i]);
568 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
569 			uint64_t loffset;
570 			uint32_t idx;
571 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
572 			if (idx >= p->nlargeobj ||
573 			    p->nlargeobj == 0 ||
574 			    p->hdr.large_offsets == NULL)
575 				return got_error(GOT_ERR_BAD_PACKIDX);
576 			loffset = be64toh(p->hdr.large_offsets[idx]);
577 			p->sorted_large_offsets[j].offset = loffset;
578 			p->sorted_large_offsets[j].idx = i;
579 			j++;
580 		} else {
581 			p->sorted_offsets[k].offset = offset;
582 			p->sorted_offsets[k].idx = i;
583 			k++;
584 		}
585 	}
586 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
587 		return got_error(GOT_ERR_BAD_PACKIDX);
588 
589 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
590 	    sizeof(p->sorted_offsets[0]), offset_cmp);
591 
592 	if (p->sorted_large_offsets)
593 		qsort(p->sorted_large_offsets, p->nlargeobj,
594 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
595 
596 	return NULL;
597 }
598 
599 const struct got_error *
600 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
601 {
602 	const struct got_error *err;
603 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
604 	int i, left, right;
605 
606 	*idx = -1;
607 
608 	if (packidx->sorted_offsets == NULL) {
609 		err = build_offset_index(packidx);
610 		if (err)
611 			return err;
612 	}
613 
614 	if (offset >= 0x7fffffff) {
615 		uint64_t lo;
616 		left = 0, right = packidx->nlargeobj - 1;
617 		while (left <= right) {
618 			i = ((left + right) / 2);
619 			lo = packidx->sorted_large_offsets[i].offset;
620 			if (lo == offset) {
621 				*idx = packidx->sorted_large_offsets[i].idx;
622 				break;
623 			} else if (offset > lo)
624 				left = i + 1;
625 			else if (offset < lo)
626 				right = i - 1;
627 		}
628 	} else {
629 		uint32_t o;
630 		left = 0, right = totobj - packidx->nlargeobj - 1;
631 		while (left <= right) {
632 			i = ((left + right) / 2);
633 			o = packidx->sorted_offsets[i].offset;
634 			if (o == offset) {
635 				*idx = packidx->sorted_offsets[i].idx;
636 				break;
637 			} else if (offset > o)
638 				left = i + 1;
639 			else if (offset < o)
640 				right = i - 1;
641 		}
642 	}
643 
644 	return NULL;
645 }
646 
647 const struct got_error *
648 got_packidx_get_object_id(struct got_object_id *id,
649     struct got_packidx *packidx, int idx)
650 {
651 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
652 	struct got_packidx_object_id *oid;
653 
654 	if (idx < 0 || idx >= totobj)
655 		return got_error(GOT_ERR_NO_OBJ);
656 
657 	oid = &packidx->hdr.sorted_ids[idx];
658 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
659 	return NULL;
660 }
661 
662 const struct got_error *
663 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
664     struct got_packidx *packidx, const char *id_str_prefix)
665 {
666 	const struct got_error *err = NULL;
667 	u_int8_t id0;
668 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
669 	char hex[3];
670 	size_t prefix_len = strlen(id_str_prefix);
671 	struct got_packidx_object_id *oid;
672 	uint32_t i = 0;
673 
674 	STAILQ_INIT(matched_ids);
675 
676 	if (prefix_len < 2)
677 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
678 
679 	hex[0] = id_str_prefix[0];
680 	hex[1] = id_str_prefix[1];
681 	hex[2] = '\0';
682 	if (!got_parse_xdigit(&id0, hex))
683 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
684 
685 	if (id0 > 0)
686 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
687 	oid = &packidx->hdr.sorted_ids[i];
688 	while (i < totobj && oid->sha1[0] == id0) {
689 		char id_str[SHA1_DIGEST_STRING_LENGTH];
690 		struct got_object_qid *qid;
691 		int cmp;
692 
693 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
694 		        return got_error(GOT_ERR_NO_SPACE);
695 
696 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
697 		if (cmp < 0) {
698 			oid = &packidx->hdr.sorted_ids[++i];
699 			continue;
700 		} else if (cmp > 0)
701 			break;
702 
703 		err = got_object_qid_alloc_partial(&qid);
704 		if (err)
705 			break;
706 		memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
707 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
708 
709 		oid = &packidx->hdr.sorted_ids[++i];
710 	}
711 
712 	if (err)
713 		got_object_id_queue_free(matched_ids);
714 	return err;
715 }
716 
717 static void
718 set_max_datasize(void)
719 {
720 	struct rlimit rl;
721 
722 	if (getrlimit(RLIMIT_DATA, &rl) != 0)
723 		return;
724 
725 	rl.rlim_cur = rl.rlim_max;
726 	setrlimit(RLIMIT_DATA, &rl);
727 }
728 
729 const struct got_error *
730 got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
731 {
732 	const struct got_error *err = NULL;
733 	int imsg_fds[2];
734 	pid_t pid;
735 	struct imsgbuf *ibuf;
736 
737 	ibuf = calloc(1, sizeof(*ibuf));
738 	if (ibuf == NULL)
739 		return got_error_from_errno("calloc");
740 
741 	pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
742 	if (pack->privsep_child == NULL) {
743 		err = got_error_from_errno("calloc");
744 		free(ibuf);
745 		return err;
746 	}
747 	pack->child_has_tempfiles = 0;
748 	pack->child_has_delta_outfd = 0;
749 
750 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
751 		err = got_error_from_errno("socketpair");
752 		goto done;
753 	}
754 
755 	pid = fork();
756 	if (pid == -1) {
757 		err = got_error_from_errno("fork");
758 		goto done;
759 	} else if (pid == 0) {
760 		set_max_datasize();
761 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
762 		    pack->path_packfile);
763 		/* not reached */
764 	}
765 
766 	if (close(imsg_fds[1]) == -1)
767 		return got_error_from_errno("close");
768 	pack->privsep_child->imsg_fd = imsg_fds[0];
769 	pack->privsep_child->pid = pid;
770 	imsg_init(ibuf, imsg_fds[0]);
771 	pack->privsep_child->ibuf = ibuf;
772 
773 	err = got_privsep_init_pack_child(ibuf, pack, packidx);
774 	if (err) {
775 		const struct got_error *child_err;
776 		err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
777 		child_err = got_privsep_wait_for_child(
778 		    pack->privsep_child->pid);
779 		if (child_err && err == NULL)
780 			err = child_err;
781 	}
782 done:
783 	if (err) {
784 		free(ibuf);
785 		free(pack->privsep_child);
786 		pack->privsep_child = NULL;
787 	}
788 	return err;
789 }
790 
791 static const struct got_error *
792 pack_stop_privsep_child(struct got_pack *pack)
793 {
794 	const struct got_error *err = NULL;
795 
796 	if (pack->privsep_child == NULL)
797 		return NULL;
798 
799 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
800 	if (err)
801 		return err;
802 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
803 	if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
804 		err = got_error_from_errno("close");
805 	free(pack->privsep_child->ibuf);
806 	free(pack->privsep_child);
807 	pack->privsep_child = NULL;
808 	return err;
809 }
810 
811 const struct got_error *
812 got_pack_close(struct got_pack *pack)
813 {
814 	const struct got_error *err = NULL;
815 
816 	err = pack_stop_privsep_child(pack);
817 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
818 		err = got_error_from_errno("munmap");
819 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
820 		err = got_error_from_errno("close");
821 	pack->fd = -1;
822 	free(pack->path_packfile);
823 	pack->path_packfile = NULL;
824 	pack->filesize = 0;
825 	if (pack->delta_cache) {
826 		got_delta_cache_free(pack->delta_cache);
827 		pack->delta_cache = NULL;
828 	}
829 
830 	/*
831 	 * Leave accumfd and basefd alone. They are managed by the
832 	 * repository layer and can be reused.
833 	 */
834 
835 	return err;
836 }
837 
838 const struct got_error *
839 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
840     struct got_pack *pack, off_t offset)
841 {
842 	uint8_t t = 0;
843 	uint64_t s = 0;
844 	uint8_t sizeN;
845 	size_t mapoff = 0;
846 	int i = 0;
847 
848 	*len = 0;
849 
850 	if (offset >= pack->filesize)
851 		return got_error(GOT_ERR_PACK_OFFSET);
852 
853 	if (pack->map) {
854 		mapoff = (size_t)offset;
855 	} else {
856 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
857 			return got_error_from_errno("lseek");
858 	}
859 
860 	do {
861 		/* We do not support size values which don't fit in 64 bit. */
862 		if (i > 9)
863 			return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
864 			    "packfile offset %llu", offset);
865 
866 		if (pack->map) {
867 			if (mapoff + sizeof(sizeN) >= pack->filesize)
868 				return got_error(GOT_ERR_BAD_PACKFILE);
869 			sizeN = *(pack->map + mapoff);
870 			mapoff += sizeof(sizeN);
871 		} else {
872 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
873 			if (n < 0)
874 				return got_error_from_errno("read");
875 			if (n != sizeof(sizeN))
876 				return got_error(GOT_ERR_BAD_PACKFILE);
877 		}
878 		*len += sizeof(sizeN);
879 
880 		if (i == 0) {
881 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
882 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
883 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
884 		} else {
885 			size_t shift = 4 + 7 * (i - 1);
886 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
887 		}
888 		i++;
889 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
890 
891 	*type = t;
892 	*size = s;
893 	return NULL;
894 }
895 
896 static const struct got_error *
897 open_plain_object(struct got_object **obj, struct got_object_id *id,
898     uint8_t type, off_t offset, size_t size, int idx)
899 {
900 	*obj = calloc(1, sizeof(**obj));
901 	if (*obj == NULL)
902 		return got_error_from_errno("calloc");
903 
904 	(*obj)->type = type;
905 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
906 	(*obj)->pack_idx = idx;
907 	(*obj)->hdrlen = 0;
908 	(*obj)->size = size;
909 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
910 	(*obj)->pack_offset = offset;
911 
912 	return NULL;
913 }
914 
915 static const struct got_error *
916 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
917     off_t delta_offset)
918 {
919 	int64_t o = 0;
920 	uint8_t offN;
921 	int i = 0;
922 
923 	*offset = 0;
924 	*len = 0;
925 
926 	do {
927 		/* We do not support offset values which don't fit in 64 bit. */
928 		if (i > 8)
929 			return got_error(GOT_ERR_NO_SPACE);
930 
931 		if (pack->map) {
932 			size_t mapoff;
933 			mapoff = (size_t)delta_offset + *len;
934 			if (mapoff + sizeof(offN) >= pack->filesize)
935 				return got_error(GOT_ERR_PACK_OFFSET);
936 			offN = *(pack->map + mapoff);
937 		} else {
938 			ssize_t n;
939 			n = read(pack->fd, &offN, sizeof(offN));
940 			if (n < 0)
941 				return got_error_from_errno("read");
942 			if (n != sizeof(offN))
943 				return got_error(GOT_ERR_BAD_PACKFILE);
944 		}
945 		*len += sizeof(offN);
946 
947 		if (i == 0)
948 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
949 		else {
950 			o++;
951 			o <<= 7;
952 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
953 		}
954 		i++;
955 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
956 
957 	*offset = o;
958 	return NULL;
959 }
960 
961 const struct got_error *
962 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
963     struct got_pack *pack, off_t offset, int tslen)
964 {
965 	const struct got_error *err;
966 	int64_t negoffset;
967 	size_t negofflen;
968 
969 	*len = 0;
970 
971 	err = parse_negative_offset(&negoffset, &negofflen, pack,
972 	    offset + tslen);
973 	if (err)
974 		return err;
975 
976 	/* Compute the base object's offset (must be in the same pack file). */
977 	*base_offset = (offset - negoffset);
978 	if (*base_offset <= 0)
979 		return got_error(GOT_ERR_BAD_PACKFILE);
980 
981 	*len = negofflen;
982 	return NULL;
983 }
984 
985 static const struct got_error *
986 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
987     size_t *delta_compressed_len, size_t delta_data_offset,
988     struct got_pack *pack)
989 {
990 	const struct got_error *err = NULL;
991 	size_t consumed = 0;
992 
993 	if (pack->map) {
994 		if (delta_data_offset >= pack->filesize)
995 			return got_error(GOT_ERR_PACK_OFFSET);
996 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
997 		    &consumed, NULL, pack->map, delta_data_offset,
998 		    pack->filesize - delta_data_offset);
999 		if (err)
1000 			return err;
1001 	} else {
1002 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1003 			return got_error_from_errno("lseek");
1004 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1005 		    &consumed, NULL, 0, pack->fd);
1006 		if (err)
1007 			return err;
1008 	}
1009 
1010 	if (delta_compressed_len)
1011 		*delta_compressed_len = consumed;
1012 
1013 	return NULL;
1014 }
1015 
1016 static const struct got_error *
1017 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1018     int delta_type, size_t delta_size, size_t delta_data_offset)
1019 {
1020 	struct got_delta *delta;
1021 
1022 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1023 	    delta_data_offset);
1024 	if (delta == NULL)
1025 		return got_error_from_errno("got_delta_open");
1026 	/* delta is freed in got_object_close() */
1027 	deltas->nentries++;
1028 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1029 	return NULL;
1030 }
1031 
1032 static const struct got_error *
1033 resolve_offset_delta(struct got_delta_chain *deltas,
1034     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1035     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1036 
1037 {
1038 	const struct got_error *err;
1039 	off_t base_offset;
1040 	uint8_t base_type;
1041 	uint64_t base_size;
1042 	size_t base_tslen;
1043 	off_t delta_data_offset;
1044 	size_t consumed;
1045 
1046 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1047 	    delta_offset, tslen);
1048 	if (err)
1049 		return err;
1050 
1051 	delta_data_offset = delta_offset + tslen + consumed;
1052 	if (delta_data_offset >= pack->filesize)
1053 		return got_error(GOT_ERR_PACK_OFFSET);
1054 
1055 	if (pack->map == NULL) {
1056 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1057 		if (delta_data_offset == -1)
1058 			return got_error_from_errno("lseek");
1059 	}
1060 
1061 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1062 	    delta_data_offset);
1063 	if (err)
1064 		return err;
1065 
1066 	/* An offset delta must be in the same packfile. */
1067 	if (base_offset >= pack->filesize)
1068 		return got_error(GOT_ERR_PACK_OFFSET);
1069 
1070 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1071 	    &base_tslen, pack, base_offset);
1072 	if (err)
1073 		return err;
1074 
1075 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1076 	    base_tslen, base_type, base_size, recursion - 1);
1077 }
1078 
1079 const struct got_error *
1080 got_pack_parse_ref_delta(struct got_object_id *id,
1081     struct got_pack *pack, off_t delta_offset, int tslen)
1082 {
1083 	if (pack->map) {
1084 		size_t mapoff = delta_offset + tslen;
1085 		if (mapoff + sizeof(*id) >= pack->filesize)
1086 			return got_error(GOT_ERR_PACK_OFFSET);
1087 		memcpy(id, pack->map + mapoff, sizeof(*id));
1088 	} else {
1089 		ssize_t n;
1090 		n = read(pack->fd, id, sizeof(*id));
1091 		if (n < 0)
1092 			return got_error_from_errno("read");
1093 		if (n != sizeof(*id))
1094 			return got_error(GOT_ERR_BAD_PACKFILE);
1095 	}
1096 
1097 	return NULL;
1098 }
1099 
1100 static const struct got_error *
1101 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1102     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1103     size_t delta_size, unsigned int recursion)
1104 {
1105 	const struct got_error *err;
1106 	struct got_object_id id;
1107 	int idx;
1108 	off_t base_offset;
1109 	uint8_t base_type;
1110 	uint64_t base_size;
1111 	size_t base_tslen;
1112 	off_t delta_data_offset;
1113 
1114 	if (delta_offset + tslen >= pack->filesize)
1115 		return got_error(GOT_ERR_PACK_OFFSET);
1116 
1117 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1118 	if (err)
1119 		return err;
1120 	if (pack->map) {
1121 		delta_data_offset = delta_offset + tslen + sizeof(id);
1122 	} else {
1123 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1124 		if (delta_data_offset == -1)
1125 			return got_error_from_errno("lseek");
1126 	}
1127 
1128 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1129 	    delta_data_offset);
1130 	if (err)
1131 		return err;
1132 
1133 	/* Delta base must be in the same pack file. */
1134 	idx = got_packidx_get_object_idx(packidx, &id);
1135 	if (idx == -1)
1136 		return got_error(GOT_ERR_NO_OBJ);
1137 
1138 	base_offset = got_packidx_get_object_offset(packidx, idx);
1139 	if (base_offset == -1)
1140 		return got_error(GOT_ERR_BAD_PACKIDX);
1141 
1142 	if (base_offset >= pack->filesize)
1143 		return got_error(GOT_ERR_PACK_OFFSET);
1144 
1145 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1146 	    &base_tslen, pack, base_offset);
1147 	if (err)
1148 		return err;
1149 
1150 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1151 	    base_tslen, base_type, base_size, recursion - 1);
1152 }
1153 
1154 const struct got_error *
1155 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1156     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1157     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1158 {
1159 	const struct got_error *err = NULL;
1160 
1161 	if (--recursion == 0)
1162 		return got_error(GOT_ERR_RECURSION);
1163 
1164 	switch (delta_type) {
1165 	case GOT_OBJ_TYPE_COMMIT:
1166 	case GOT_OBJ_TYPE_TREE:
1167 	case GOT_OBJ_TYPE_BLOB:
1168 	case GOT_OBJ_TYPE_TAG:
1169 		/* Plain types are the final delta base. Recursion ends. */
1170 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1171 		    delta_size, 0);
1172 		break;
1173 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1174 		err = resolve_offset_delta(deltas, packidx, pack,
1175 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1176 		break;
1177 	case GOT_OBJ_TYPE_REF_DELTA:
1178 		err = resolve_ref_delta(deltas, packidx, pack,
1179 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1180 		break;
1181 	default:
1182 		return got_error(GOT_ERR_OBJ_TYPE);
1183 	}
1184 
1185 	return err;
1186 }
1187 
1188 static const struct got_error *
1189 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1190     struct got_pack *pack, struct got_object_id *id, off_t offset,
1191     size_t tslen, int delta_type, size_t delta_size, int idx)
1192 {
1193 	const struct got_error *err = NULL;
1194 	int resolved_type;
1195 
1196 	*obj = calloc(1, sizeof(**obj));
1197 	if (*obj == NULL)
1198 		return got_error_from_errno("calloc");
1199 
1200 	(*obj)->flags = 0;
1201 	(*obj)->hdrlen = 0;
1202 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1203 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1204 	(*obj)->pack_offset = offset + tslen;
1205 
1206 	STAILQ_INIT(&(*obj)->deltas.entries);
1207 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1208 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1209 	(*obj)->pack_idx = idx;
1210 
1211 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1212 	    offset, tslen, delta_type, delta_size,
1213 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1214 	if (err)
1215 		goto done;
1216 
1217 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1218 	if (err)
1219 		goto done;
1220 	(*obj)->type = resolved_type;
1221 done:
1222 	if (err) {
1223 		got_object_close(*obj);
1224 		*obj = NULL;
1225 	}
1226 	return err;
1227 }
1228 
1229 const struct got_error *
1230 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1231     struct got_packidx *packidx, int idx, struct got_object_id *id)
1232 {
1233 	const struct got_error *err = NULL;
1234 	off_t offset;
1235 	uint8_t type;
1236 	uint64_t size;
1237 	size_t tslen;
1238 
1239 	*obj = NULL;
1240 
1241 	offset = got_packidx_get_object_offset(packidx, idx);
1242 	if (offset == -1)
1243 		return got_error(GOT_ERR_BAD_PACKIDX);
1244 
1245 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1246 	    pack, offset);
1247 	if (err)
1248 		return err;
1249 
1250 	switch (type) {
1251 	case GOT_OBJ_TYPE_COMMIT:
1252 	case GOT_OBJ_TYPE_TREE:
1253 	case GOT_OBJ_TYPE_BLOB:
1254 	case GOT_OBJ_TYPE_TAG:
1255 		err = open_plain_object(obj, id, type, offset + tslen,
1256 		    size, idx);
1257 		break;
1258 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1259 	case GOT_OBJ_TYPE_REF_DELTA:
1260 		err = open_delta_object(obj, packidx, pack, id, offset,
1261 		    tslen, type, size, idx);
1262 		break;
1263 	default:
1264 		err = got_error(GOT_ERR_OBJ_TYPE);
1265 		break;
1266 	}
1267 
1268 	return err;
1269 }
1270 
1271 const struct got_error *
1272 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1273     struct got_delta_chain *deltas, struct got_pack *pack)
1274 {
1275 	struct got_delta *delta;
1276 	uint64_t base_size = 0, result_size = 0;
1277 
1278 	*max_size = 0;
1279 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1280 		/* Plain object types are the delta base. */
1281 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1282 		    delta->type != GOT_OBJ_TYPE_TREE &&
1283 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1284 		    delta->type != GOT_OBJ_TYPE_TAG) {
1285 			const struct got_error *err;
1286 			uint8_t *delta_buf = NULL;
1287 			size_t delta_len;
1288 			int cached = 1;
1289 
1290 			if (pack->delta_cache) {
1291 				got_delta_cache_get(&delta_buf, &delta_len,
1292 				    pack->delta_cache, delta->data_offset);
1293 			}
1294 			if (delta_buf == NULL) {
1295 				cached = 0;
1296 				err = read_delta_data(&delta_buf, &delta_len,
1297 				    NULL, delta->data_offset, pack);
1298 				if (err)
1299 					return err;
1300 			}
1301 			if (pack->delta_cache && !cached) {
1302 				err = got_delta_cache_add(pack->delta_cache,
1303 				    delta->data_offset, delta_buf, delta_len);
1304 				if (err == NULL)
1305 					cached = 1;
1306 				else if (err->code != GOT_ERR_NO_SPACE) {
1307 					free(delta_buf);
1308 					return err;
1309 				}
1310 			}
1311 			err = got_delta_get_sizes(&base_size, &result_size,
1312 			    delta_buf, delta_len);
1313 			if (!cached)
1314 				free(delta_buf);
1315 			if (err)
1316 				return err;
1317 		} else
1318 			base_size = delta->size;
1319 		if (base_size > *max_size)
1320 			*max_size = base_size;
1321 		if (result_size > *max_size)
1322 			*max_size = result_size;
1323 	}
1324 
1325 	return NULL;
1326 }
1327 
1328 const struct got_error *
1329 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1330     struct got_pack *pack)
1331 {
1332 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1333 		return got_error(GOT_ERR_OBJ_TYPE);
1334 
1335 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1336 }
1337 
1338 const struct got_error *
1339 got_pack_dump_delta_chain_to_file(size_t *result_size,
1340     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1341     FILE *base_file, FILE *accum_file)
1342 {
1343 	const struct got_error *err = NULL;
1344 	struct got_delta *delta;
1345 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1346 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1347 	/* We process small enough files entirely in memory for speed. */
1348 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1349 	uint64_t max_size = 0;
1350 	int n = 0;
1351 
1352 	*result_size = 0;
1353 
1354 	if (STAILQ_EMPTY(&deltas->entries))
1355 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1356 
1357 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1358 		return got_error_from_errno("fseeko");
1359 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1360 		return got_error_from_errno("fseeko");
1361 
1362 	/* Deltas are ordered in ascending order. */
1363 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1364 		uint8_t *delta_buf = NULL;
1365 		uint64_t base_size, result_size = 0;
1366 		int cached = 1;
1367 		if (n == 0) {
1368 			size_t mapoff;
1369 			off_t delta_data_offset;
1370 
1371 			/* Plain object types are the delta base. */
1372 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1373 			    delta->type != GOT_OBJ_TYPE_TREE &&
1374 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1375 			    delta->type != GOT_OBJ_TYPE_TAG) {
1376 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1377 				goto done;
1378 			}
1379 
1380 			delta_data_offset = delta->offset + delta->tslen;
1381 			if (delta_data_offset >= pack->filesize) {
1382 				err = got_error(GOT_ERR_PACK_OFFSET);
1383 				goto done;
1384 			}
1385 			if (pack->map == NULL) {
1386 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1387 				    == -1) {
1388 					err = got_error_from_errno("lseek");
1389 					goto done;
1390 				}
1391 			}
1392 			if (delta->size > max_size)
1393 				max_size = delta->size;
1394 			if (max_size > max_bufsize) {
1395 				if (pack->map) {
1396 					mapoff = (size_t)delta_data_offset;
1397 					err = got_inflate_to_file_mmap(
1398 					    &base_bufsz, NULL, NULL, pack->map,
1399 					    mapoff, pack->filesize - mapoff,
1400 					    base_file);
1401 				} else
1402 					err = got_inflate_to_file_fd(
1403 					    &base_bufsz, NULL, NULL, pack->fd,
1404 					    base_file);
1405 			} else {
1406 				accum_buf = malloc(max_size);
1407 				if (accum_buf == NULL) {
1408 					err = got_error_from_errno("malloc");
1409 					goto done;
1410 				}
1411 				accum_bufsz = max_size;
1412 				if (pack->map) {
1413 					mapoff = (size_t)delta_data_offset;
1414 					err = got_inflate_to_mem_mmap(&base_buf,
1415 					    &base_bufsz, NULL, NULL,
1416 					    pack->map, mapoff,
1417 					    pack->filesize - mapoff);
1418 				} else
1419 					err = got_inflate_to_mem_fd(&base_buf,
1420 					    &base_bufsz, NULL, NULL, max_size,
1421 					    pack->fd);
1422 			}
1423 			if (err)
1424 				goto done;
1425 			n++;
1426 			if (base_buf == NULL)
1427 				rewind(base_file);
1428 			continue;
1429 		}
1430 
1431 		if (pack->delta_cache) {
1432 			got_delta_cache_get(&delta_buf, &delta_len,
1433 			    pack->delta_cache, delta->data_offset);
1434 		}
1435 		if (delta_buf == NULL) {
1436 			cached = 0;
1437 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1438 			    delta->data_offset, pack);
1439 			if (err)
1440 				goto done;
1441 		}
1442 		if (pack->delta_cache && !cached) {
1443 			err = got_delta_cache_add(pack->delta_cache,
1444 			    delta->data_offset, delta_buf, delta_len);
1445 			if (err == NULL)
1446 				cached = 1;
1447 			else if (err->code != GOT_ERR_NO_SPACE) {
1448 				free(delta_buf);
1449 				goto done;
1450 			}
1451 		}
1452 
1453 		err = got_delta_get_sizes(&base_size, &result_size,
1454 		    delta_buf, delta_len);
1455 		if (err)
1456 			goto done;
1457 		if (base_size > max_size)
1458 			max_size = base_size;
1459 		if (result_size > max_size)
1460 			max_size = result_size;
1461 
1462 		if (base_buf && max_size > max_bufsize) {
1463 			/* Switch from buffers to temporary files. */
1464 			size_t w = fwrite(base_buf, 1, base_bufsz,
1465 			    base_file);
1466 			if (w != base_bufsz) {
1467 				err = got_ferror(outfile, GOT_ERR_IO);
1468 				goto done;
1469 			}
1470 			free(base_buf);
1471 			base_buf = NULL;
1472 			free(accum_buf);
1473 			accum_buf = NULL;
1474 		}
1475 
1476 		if (base_buf && max_size > base_bufsz) {
1477 			uint8_t *p = realloc(base_buf, max_size);
1478 			if (p == NULL) {
1479 				err = got_error_from_errno("realloc");
1480 				goto done;
1481 			}
1482 			base_buf = p;
1483 			base_bufsz = max_size;
1484 		}
1485 
1486 		if (accum_buf && max_size > accum_bufsz) {
1487 			uint8_t *p = realloc(accum_buf, max_size);
1488 			if (p == NULL) {
1489 				err = got_error_from_errno("realloc");
1490 				goto done;
1491 			}
1492 			accum_buf = p;
1493 			accum_bufsz = max_size;
1494 		}
1495 
1496 		if (base_buf) {
1497 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1498 			    delta_buf, delta_len, accum_buf,
1499 			    &accum_size, max_size);
1500 			n++;
1501 		} else {
1502 			err = got_delta_apply(base_file, delta_buf,
1503 			    delta_len,
1504 			    /* Final delta application writes to output file. */
1505 			    ++n < deltas->nentries ? accum_file : outfile,
1506 			    &accum_size);
1507 		}
1508 		if (!cached)
1509 			free(delta_buf);
1510 		if (err)
1511 			goto done;
1512 
1513 		if (n < deltas->nentries) {
1514 			/* Accumulated delta becomes the new base. */
1515 			if (base_buf) {
1516 				uint8_t *tmp = accum_buf;
1517 				size_t tmp_size = accum_bufsz;
1518 				accum_buf = base_buf;
1519 				accum_bufsz = base_bufsz;
1520 				base_buf = tmp;
1521 				base_bufsz = tmp_size;
1522 			} else {
1523 				FILE *tmp = accum_file;
1524 				accum_file = base_file;
1525 				base_file = tmp;
1526 				rewind(base_file);
1527 				rewind(accum_file);
1528 			}
1529 		}
1530 	}
1531 
1532 done:
1533 	free(base_buf);
1534 	if (accum_buf) {
1535 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1536 		free(accum_buf);
1537 		if (len != accum_size)
1538 			err = got_ferror(outfile, GOT_ERR_IO);
1539 	}
1540 	rewind(outfile);
1541 	if (err == NULL)
1542 		*result_size = accum_size;
1543 	return err;
1544 }
1545 
1546 const struct got_error *
1547 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1548     struct got_delta_chain *deltas, struct got_pack *pack)
1549 {
1550 	const struct got_error *err = NULL;
1551 	struct got_delta *delta;
1552 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1553 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1554 	uint64_t max_size = 0;
1555 	int n = 0;
1556 
1557 	*outbuf = NULL;
1558 	*outlen = 0;
1559 
1560 	if (STAILQ_EMPTY(&deltas->entries))
1561 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1562 
1563 	/* Deltas are ordered in ascending order. */
1564 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1565 		uint8_t *delta_buf = NULL;
1566 		uint64_t base_size, result_size = 0;
1567 		int cached = 1;
1568 		if (n == 0) {
1569 			size_t delta_data_offset;
1570 
1571 			/* Plain object types are the delta base. */
1572 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1573 			    delta->type != GOT_OBJ_TYPE_TREE &&
1574 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1575 			    delta->type != GOT_OBJ_TYPE_TAG) {
1576 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1577 				goto done;
1578 			}
1579 
1580 			delta_data_offset = delta->offset + delta->tslen;
1581 			if (delta_data_offset >= pack->filesize) {
1582 				err = got_error(GOT_ERR_PACK_OFFSET);
1583 				goto done;
1584 			}
1585 
1586 			if (delta->size > max_size)
1587 				max_size = delta->size;
1588 
1589 			if (pack->map) {
1590 				size_t mapoff = (size_t)delta_data_offset;
1591 				err = got_inflate_to_mem_mmap(&base_buf,
1592 				    &base_bufsz, NULL, NULL, pack->map,
1593 				    mapoff, pack->filesize - mapoff);
1594 			} else {
1595 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1596 				    == -1) {
1597 					err = got_error_from_errno("lseek");
1598 					goto done;
1599 				}
1600 				err = got_inflate_to_mem_fd(&base_buf,
1601 				    &base_bufsz, NULL, NULL, max_size,
1602 				    pack->fd);
1603 			}
1604 			if (err)
1605 				goto done;
1606 			n++;
1607 			continue;
1608 		}
1609 
1610 		if (pack->delta_cache) {
1611 			got_delta_cache_get(&delta_buf, &delta_len,
1612 			    pack->delta_cache, delta->data_offset);
1613 		}
1614 		if (delta_buf == NULL) {
1615 			cached = 0;
1616 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1617 			    delta->data_offset, pack);
1618 			if (err)
1619 				goto done;
1620 		}
1621 		if (pack->delta_cache && !cached) {
1622 			err = got_delta_cache_add(pack->delta_cache,
1623 			    delta->data_offset, delta_buf, delta_len);
1624 			if (err == NULL)
1625 				cached = 1;
1626 			else if (err->code != GOT_ERR_NO_SPACE) {
1627 				free(delta_buf);
1628 				goto done;
1629 			}
1630 		}
1631 
1632 		err = got_delta_get_sizes(&base_size, &result_size,
1633 		    delta_buf, delta_len);
1634 		if (err)
1635 			goto done;
1636 		if (base_size > max_size)
1637 			max_size = base_size;
1638 		if (result_size > max_size)
1639 			max_size = result_size;
1640 
1641 		if (max_size > base_bufsz) {
1642 			uint8_t *p = realloc(base_buf, max_size);
1643 			if (p == NULL) {
1644 				err = got_error_from_errno("realloc");
1645 				goto done;
1646 			}
1647 			base_buf = p;
1648 			base_bufsz = max_size;
1649 		}
1650 
1651 		if (max_size > accum_bufsz) {
1652 			uint8_t *p = realloc(accum_buf, max_size);
1653 			if (p == NULL) {
1654 				err = got_error_from_errno("realloc");
1655 				goto done;
1656 			}
1657 			accum_buf = p;
1658 			accum_bufsz = max_size;
1659 		}
1660 
1661 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1662 		    delta_buf, delta_len, accum_buf,
1663 		    &accum_size, max_size);
1664 		if (!cached)
1665 			free(delta_buf);
1666 		n++;
1667 		if (err)
1668 			goto done;
1669 
1670 		if (n < deltas->nentries) {
1671 			/* Accumulated delta becomes the new base. */
1672 			uint8_t *tmp = accum_buf;
1673 			size_t tmp_size = accum_bufsz;
1674 			accum_buf = base_buf;
1675 			accum_bufsz = base_bufsz;
1676 			base_buf = tmp;
1677 			base_bufsz = tmp_size;
1678 		}
1679 	}
1680 
1681 done:
1682 	free(base_buf);
1683 	if (err) {
1684 		free(accum_buf);
1685 		*outbuf = NULL;
1686 		*outlen = 0;
1687 	} else {
1688 		*outbuf = accum_buf;
1689 		*outlen = accum_size;
1690 	}
1691 	return err;
1692 }
1693 
1694 const struct got_error *
1695 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1696     FILE *outfile, FILE *base_file, FILE *accum_file)
1697 {
1698 	const struct got_error *err = NULL;
1699 
1700 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1701 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1702 
1703 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1704 		if (obj->pack_offset >= pack->filesize)
1705 			return got_error(GOT_ERR_PACK_OFFSET);
1706 
1707 		if (pack->map) {
1708 			size_t mapoff = (size_t)obj->pack_offset;
1709 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1710 			    pack->map, mapoff, pack->filesize - mapoff,
1711 			    outfile);
1712 		} else {
1713 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1714 				return got_error_from_errno("lseek");
1715 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1716 			    pack->fd, outfile);
1717 		}
1718 	} else
1719 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1720 		    &obj->deltas, pack, outfile, base_file, accum_file);
1721 
1722 	return err;
1723 }
1724 
1725 const struct got_error *
1726 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1727     struct got_object *obj, struct got_pack *pack)
1728 {
1729 	const struct got_error *err = NULL;
1730 
1731 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1732 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1733 
1734 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1735 		if (obj->pack_offset >= pack->filesize)
1736 			return got_error(GOT_ERR_PACK_OFFSET);
1737 		if (pack->map) {
1738 			size_t mapoff = (size_t)obj->pack_offset;
1739 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1740 			    pack->map, mapoff, pack->filesize - mapoff);
1741 		} else {
1742 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1743 				return got_error_from_errno("lseek");
1744 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1745 			    obj->size, pack->fd);
1746 		}
1747 	} else
1748 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1749 		    pack);
1750 
1751 	return err;
1752 }
1753 
1754 static const struct got_error *
1755 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1756     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1757     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1758 {
1759 	const struct got_error *err = NULL;
1760 
1761 	/* Validate decompression and obtain the decompressed size. */
1762 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1763 	    delta_data_offset, pack);
1764 	if (err)
1765 		return err;
1766 
1767 	/* Read delta base/result sizes from head of delta stream. */
1768 	err = got_delta_get_sizes(base_size, result_size,
1769 	    *delta_buf, *delta_len);
1770 	if (err)
1771 		goto done;
1772 
1773 	/* Discard decompressed delta and read it again in compressed form. */
1774 	free(*delta_buf);
1775 	*delta_buf = malloc(*delta_len_compressed);
1776 	if (*delta_buf == NULL) {
1777 		err = got_error_from_errno("malloc");
1778 		goto done;
1779 	}
1780 	if (pack->map) {
1781 		if (delta_data_offset >= pack->filesize)
1782 			err = got_error(GOT_ERR_PACK_OFFSET);
1783 		memcpy(*delta_buf, pack->map + delta_data_offset,
1784 		    *delta_len_compressed);
1785 	} else {
1786 		ssize_t n;
1787 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1788 			err = got_error_from_errno("lseek");
1789 			goto done;
1790 		}
1791 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1792 		if (n < 0) {
1793 			err = got_error_from_errno("read");
1794 			goto done;
1795 		} else if (n != *delta_len_compressed) {
1796 			err = got_error(GOT_ERR_IO);
1797 			goto done;
1798 		}
1799 	}
1800 done:
1801 	if (err) {
1802 		free(*delta_buf);
1803 		*delta_buf = NULL;
1804 		*delta_len = 0;
1805 		*delta_len_compressed = 0;
1806 		*base_size = 0;
1807 		*result_size = 0;
1808 	}
1809 	return err;
1810 }
1811 
1812 const struct got_error *
1813 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1814     size_t *delta_compressed_size, off_t *delta_offset, off_t *base_offset,
1815     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1816     struct got_pack *pack, struct got_packidx *packidx, int idx)
1817 {
1818 	const struct got_error *err = NULL;
1819 	off_t offset;
1820 	uint8_t type;
1821 	uint64_t size;
1822 	size_t tslen, delta_hdrlen;
1823 	off_t delta_data_offset;
1824 
1825 	*delta_buf = NULL;
1826 	*delta_size = 0;
1827 	*delta_compressed_size = 0;
1828 	*delta_offset = 0;
1829 	*base_offset = 0;
1830 	*base_size = 0;
1831 	*result_size = 0;
1832 
1833 	offset = got_packidx_get_object_offset(packidx, idx);
1834 	if (offset == -1)
1835 		return got_error(GOT_ERR_BAD_PACKIDX);
1836 
1837 	if (offset >= pack->filesize)
1838 		return got_error(GOT_ERR_PACK_OFFSET);
1839 
1840 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1841 	    pack, offset);
1842 	if (err)
1843 		return err;
1844 
1845 	if (tslen + size < tslen || offset + size < size ||
1846 	    tslen + offset < tslen)
1847 		return got_error(GOT_ERR_PACK_OFFSET);
1848 
1849 	switch (type) {
1850 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1851 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1852 		    pack, offset, tslen);
1853 		if (err)
1854 			return err;
1855 		break;
1856 	case GOT_OBJ_TYPE_REF_DELTA:
1857 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1858 		if (err)
1859 			return err;
1860 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1861 		break;
1862 	default:
1863 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1864 		    "non-delta object type %d found at offset %lld",
1865 		    type, (long long)offset);
1866 	}
1867 
1868 	if (tslen + delta_hdrlen < delta_hdrlen ||
1869 	    offset + delta_hdrlen < delta_hdrlen)
1870 		return got_error(GOT_ERR_BAD_DELTA);
1871 
1872 	delta_data_offset = offset + tslen + delta_hdrlen;
1873 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1874 	    base_size, result_size, delta_data_offset, pack, packidx);
1875 	if (err)
1876 		return err;
1877 
1878 	if (*delta_size != size) {
1879 		err = got_error(GOT_ERR_BAD_DELTA);
1880 		goto done;
1881 	}
1882 
1883 	*delta_offset = offset;
1884 done:
1885 	if (err) {
1886 		free(*delta_buf);
1887 		*delta_buf = NULL;
1888 		*delta_size = 0;
1889 		*delta_compressed_size = 0;
1890 		*delta_offset = 0;
1891 		*base_offset = 0;
1892 		*base_size = 0;
1893 		*result_size = 0;
1894 	}
1895 	return err;
1896 }