Got Mirror

Blob

Date:: Mon Oct 24 21:29:59 2022 UTC
Message:: add bounds check when reading a delta offset from a packed object ok stsp@
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3  * Copyright (c) 2020, 2022 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/queue.h>
19 #include <sys/stat.h>
20 #include <sys/time.h>
21 #include <sys/types.h>
22 #include <sys/uio.h>
23 #include <sys/mman.h>
24 
25 #include <stdint.h>
26 #include <errno.h>
27 #include <imsg.h>
28 #include <limits.h>
29 #include <signal.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <ctype.h>
34 #include <sha1.h>
35 #include <endian.h>
36 #include <fcntl.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <err.h>
40 #include <assert.h>
41 #include <dirent.h>
42 
43 #include "got_error.h"
44 #include "got_object.h"
45 
46 #include "got_lib_sha1.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_inflate.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_parse.h"
51 #include "got_lib_object_idset.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_ratelimit.h"
55 #include "got_lib_pack_index.h"
56 #include "got_lib_delta_cache.h"
57 
58 struct got_indexed_object {
59 	struct got_object_id id;
60 
61 	/*
62 	 * Has this object been fully resolved?
63 	 * If so, we know its ID, otherwise we don't and 'id' is invalid.
64 	 */
65 	int valid;
66 
67 	/* Offset of type+size field for this object in pack file. */
68 	off_t off;
69 
70 	/* Type+size values parsed from pack file. */
71 	uint8_t type;
72 	uint64_t size;
73 
74 	/* Length of on-disk type+size data. */
75 	size_t tslen;
76 
77 	/* Length of object data following type+size. */
78 	size_t len;
79 
80 	uint32_t crc;
81 
82 	union {
83 		struct {
84 			/* For ref deltas. */
85 			struct got_object_id ref_id;
86 		} ref;
87 		struct {
88 			/* For offset deltas. */
89 			off_t base_offset;
90 			size_t base_offsetlen;
91 		} ofs;
92 	} delta;
93 };
94 
95 static void
96 putbe32(char *b, uint32_t n)
97 {
98 	b[0] = n >> 24;
99 	b[1] = n >> 16;
100 	b[2] = n >> 8;
101 	b[3] = n >> 0;
102 }
103 
104 static const struct got_error *
105 get_obj_type_label(const char **label, int obj_type)
106 {
107 	const struct got_error *err = NULL;
108 
109 	switch (obj_type) {
110 	case GOT_OBJ_TYPE_BLOB:
111 		*label = GOT_OBJ_LABEL_BLOB;
112 		break;
113 	case GOT_OBJ_TYPE_TREE:
114 		*label = GOT_OBJ_LABEL_TREE;
115 		break;
116 	case GOT_OBJ_TYPE_COMMIT:
117 		*label = GOT_OBJ_LABEL_COMMIT;
118 		break;
119 	case GOT_OBJ_TYPE_TAG:
120 		*label = GOT_OBJ_LABEL_TAG;
121 		break;
122 	default:
123 		*label = NULL;
124 		err = got_error(GOT_ERR_OBJ_TYPE);
125 		break;
126 	}
127 
128 	return err;
129 }
130 
131 static const struct got_error *
132 read_checksum(uint32_t *crc, SHA1_CTX *sha1_ctx, int fd, size_t len)
133 {
134 	uint8_t buf[8192];
135 	size_t n;
136 	ssize_t r;
137 
138 	for (n = len; n > 0; n -= r){
139 		r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
140 		if (r == -1)
141 			return got_error_from_errno("read");
142 		if (r == 0)
143 			break;
144 		if (crc)
145 			*crc = crc32(*crc, buf, r);
146 		if (sha1_ctx)
147 			SHA1Update(sha1_ctx, buf, r);
148 	}
149 
150 	return NULL;
151 }
152 
153 static const struct got_error *
154 read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len)
155 {
156 	uint8_t buf[8192];
157 	size_t n, r;
158 
159 	for (n = len; n > 0; n -= r) {
160 		r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
161 		if (r == 0) {
162 			if (feof(f))
163 				return NULL;
164 			return got_ferror(f, GOT_ERR_IO);
165 		}
166 		SHA1Update(ctx, buf, r);
167 	}
168 
169 	return NULL;
170 }
171 
172 static const struct got_error *
173 read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
174     FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
175 {
176 	const struct got_error *err = NULL;
177 	SHA1_CTX ctx;
178 	uint8_t *data = NULL;
179 	size_t datalen = 0;
180 	ssize_t n;
181 	char *header;
182 	size_t headerlen;
183 	const char *obj_label;
184 	size_t mapoff = obj->off;
185 	struct got_inflate_checksum csum;
186 
187 	memset(&csum, 0, sizeof(csum));
188 	csum.input_sha1 = pack_sha1_ctx;
189 	csum.input_crc = &obj->crc;
190 
191 	err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
192 	    &obj->tslen, pack, obj->off);
193 	if (err)
194 		return err;
195 
196 	if (pack->map) {
197 		obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
198 		SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
199 		mapoff += obj->tslen;
200 	} else {
201 		/* XXX Seek back and get the CRC of on-disk type+size bytes. */
202 		if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
203 			return got_error_from_errno("lseek");
204 		err = read_checksum(&obj->crc, pack_sha1_ctx,
205 		    pack->fd, obj->tslen);
206 		if (err)
207 			return err;
208 	}
209 
210 	switch (obj->type) {
211 	case GOT_OBJ_TYPE_BLOB:
212 	case GOT_OBJ_TYPE_COMMIT:
213 	case GOT_OBJ_TYPE_TREE:
214 	case GOT_OBJ_TYPE_TAG:
215 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
216 			if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
217 				err = got_error_from_errno("fseek");
218 				break;
219 			}
220 			if (pack->map) {
221 				err = got_inflate_to_file_mmap(&datalen,
222 				    &obj->len, &csum, pack->map, mapoff,
223 				    pack->filesize - mapoff, tmpfile);
224 			} else {
225 				err = got_inflate_to_file_fd(&datalen,
226 				    &obj->len, &csum, pack->fd, tmpfile);
227 			}
228 		} else {
229 			if (pack->map) {
230 				err = got_inflate_to_mem_mmap(&data, &datalen,
231 				    &obj->len, &csum, pack->map, mapoff,
232 				    pack->filesize - mapoff);
233 			} else {
234 				err = got_inflate_to_mem_fd(&data, &datalen,
235 				    &obj->len, &csum, obj->size, pack->fd);
236 			}
237 		}
238 		if (err)
239 			break;
240 		SHA1Init(&ctx);
241 		err = get_obj_type_label(&obj_label, obj->type);
242 		if (err) {
243 			free(data);
244 			break;
245 		}
246 		if (asprintf(&header, "%s %lld", obj_label,
247 		    (long long)obj->size) == -1) {
248 			err = got_error_from_errno("asprintf");
249 			free(data);
250 			break;
251 		}
252 		headerlen = strlen(header) + 1;
253 		SHA1Update(&ctx, header, headerlen);
254 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
255 			err = read_file_sha1(&ctx, tmpfile, datalen);
256 			if (err) {
257 				free(header);
258 				free(data);
259 				break;
260 			}
261 		} else
262 			SHA1Update(&ctx, data, datalen);
263 		SHA1Final(obj->id.sha1, &ctx);
264 		free(header);
265 		free(data);
266 		break;
267 	case GOT_OBJ_TYPE_REF_DELTA:
268 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
269 		if (pack->map) {
270 			if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
271 				err = got_error(GOT_ERR_BAD_PACKFILE);
272 				break;
273 			}
274 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
275 			    SHA1_DIGEST_LENGTH);
276 			obj->crc = crc32(obj->crc, pack->map + mapoff,
277 			    SHA1_DIGEST_LENGTH);
278 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
279 			    SHA1_DIGEST_LENGTH);
280 			mapoff += SHA1_DIGEST_LENGTH;
281 			err = got_inflate_to_mem_mmap(NULL, &datalen,
282 			    &obj->len, &csum, pack->map, mapoff,
283 			    pack->filesize - mapoff);
284 			if (err)
285 				break;
286 		} else {
287 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
288 			    SHA1_DIGEST_LENGTH);
289 			if (n == -1) {
290 				err = got_error_from_errno("read");
291 				break;
292 			}
293 			if (n < sizeof(obj->id)) {
294 				err = got_error(GOT_ERR_BAD_PACKFILE);
295 				break;
296 			}
297 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
298 			    SHA1_DIGEST_LENGTH);
299 			SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
300 			    SHA1_DIGEST_LENGTH);
301 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
302 			    &csum, obj->size, pack->fd);
303 			if (err)
304 				break;
305 		}
306 		obj->len += SHA1_DIGEST_LENGTH;
307 		break;
308 	case GOT_OBJ_TYPE_OFFSET_DELTA:
309 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
310 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
311 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
312 		    obj->tslen);
313 		if (err)
314 			break;
315 
316 		if (pack->map) {
317 			if (mapoff + obj->delta.ofs.base_offsetlen >=
318 			    pack->filesize) {
319 				err = got_error(GOT_ERR_BAD_PACKFILE);
320 				break;
321 			}
322 
323 			obj->crc = crc32(obj->crc, pack->map + mapoff,
324 			    obj->delta.ofs.base_offsetlen);
325 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
326 			    obj->delta.ofs.base_offsetlen);
327 			mapoff += obj->delta.ofs.base_offsetlen;
328 			err = got_inflate_to_mem_mmap(NULL, &datalen,
329 			    &obj->len, &csum, pack->map, mapoff,
330 			    pack->filesize - mapoff);
331 			if (err)
332 				break;
333 		} else {
334 			/*
335 			 * XXX Seek back and get CRC and SHA1 of on-disk
336 			 * offset bytes.
337 			 */
338 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
339 			    == -1) {
340 				err = got_error_from_errno("lseek");
341 				break;
342 			}
343 			err = read_checksum(&obj->crc, pack_sha1_ctx,
344 			    pack->fd, obj->delta.ofs.base_offsetlen);
345 			if (err)
346 				break;
347 
348 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
349 			    &csum, obj->size, pack->fd);
350 			if (err)
351 				break;
352 		}
353 		obj->len += obj->delta.ofs.base_offsetlen;
354 		break;
355 	default:
356 		err = got_error(GOT_ERR_OBJ_TYPE);
357 		break;
358 	}
359 
360 	return err;
361 }
362 
363 const struct got_error *
364 got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
365 {
366 	ssize_t w;
367 
368 	SHA1Update(ctx, buf, len);
369 
370 	w = write(fd, buf, len);
371 	if (w == -1)
372 		return got_error_from_errno("write");
373 	if (w != len)
374 		return got_error(GOT_ERR_IO);
375 
376 	return NULL;
377 }
378 
379 static const struct got_error *
380 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
381     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
382     FILE *delta_accum_file)
383 {
384 	const struct got_error *err = NULL;
385 	struct got_delta_chain deltas;
386 	struct got_delta *delta;
387 	uint8_t *buf = NULL;
388 	size_t len = 0;
389 	SHA1_CTX ctx;
390 	char *header = NULL;
391 	size_t headerlen;
392 	uint64_t max_size;
393 	int base_obj_type;
394 	const char *obj_label;
395 
396 	deltas.nentries = 0;
397 	STAILQ_INIT(&deltas.entries);
398 
399 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
400 	    obj->off, obj->tslen, obj->type, obj->size,
401 	    GOT_DELTA_CHAIN_RECURSION_MAX);
402 	if (err)
403 		goto done;
404 
405 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
406 	if (err)
407 		goto done;
408 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
409 		rewind(tmpfile);
410 		rewind(delta_base_file);
411 		rewind(delta_accum_file);
412 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
413 		    pack, tmpfile, delta_base_file, delta_accum_file);
414 		if (err)
415 			goto done;
416 	} else {
417 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
418 		    &deltas, pack);
419 	}
420 	if (err)
421 		goto done;
422 
423 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
424 	if (err)
425 		goto done;
426 	err = get_obj_type_label(&obj_label, base_obj_type);
427 	if (err)
428 		goto done;
429 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
430 		err = got_error_from_errno("asprintf");
431 		goto done;
432 	}
433 	headerlen = strlen(header) + 1;
434 	SHA1Init(&ctx);
435 	SHA1Update(&ctx, header, headerlen);
436 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
437 		err = read_file_sha1(&ctx, tmpfile, len);
438 		if (err)
439 			goto done;
440 	} else
441 		SHA1Update(&ctx, buf, len);
442 	SHA1Final(obj->id.sha1, &ctx);
443 done:
444 	free(buf);
445 	free(header);
446 	while (!STAILQ_EMPTY(&deltas.entries)) {
447 		delta = STAILQ_FIRST(&deltas.entries);
448 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
449 		free(delta);
450 	}
451 	return err;
452 }
453 
454 /* Determine the slot in the pack index a given object ID should use. */
455 static int
456 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
457 {
458 	u_int8_t id0 = sha1[0];
459 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
460 	int left = 0, right = nindexed - 1;
461 	int cmp = 0, i = 0;
462 
463 	if (id0 > 0)
464 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
465 
466 	while (left <= right) {
467 		struct got_packidx_object_id *oid;
468 
469 		i = ((left + right) / 2);
470 		oid = &packidx->hdr.sorted_ids[i];
471 
472 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
473 		if (cmp == 0)
474 			return -1; /* object already indexed */
475 		else if (cmp > 0)
476 			left = i + 1;
477 		else if (cmp < 0)
478 			right = i - 1;
479 	}
480 
481 	return left;
482 }
483 
484 #if 0
485 static void
486 print_packidx(struct got_packidx *packidx)
487 {
488 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
489 	int i;
490 
491 	fprintf(stderr, "object IDs:\n");
492 	for (i = 0; i < nindexed; i++) {
493 		char hex[SHA1_DIGEST_STRING_LENGTH];
494 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
495 		    hex, sizeof(hex));
496 		fprintf(stderr, "%s\n", hex);
497 	}
498 	fprintf(stderr, "\n");
499 
500 	fprintf(stderr, "object offsets:\n");
501 	for (i = 0; i < nindexed; i++) {
502 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
503 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
504 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
505 			fprintf(stderr, "%u -> %llu\n", offset,
506 			    be64toh(packidx->hdr.large_offsets[j]));
507 		} else
508 			fprintf(stderr, "%u\n", offset);
509 	}
510 	fprintf(stderr, "\n");
511 
512 	fprintf(stderr, "fanout table:");
513 	for (i = 0; i <= 0xff; i++)
514 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
515 	fprintf(stderr, "\n");
516 }
517 #endif
518 
519 static void
520 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
521     struct got_indexed_object *obj)
522 {
523 	int i;
524 
525 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
526 	    SHA1_DIGEST_LENGTH);
527 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
528 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
529 		packidx->hdr.offsets[idx] = htobe32(obj->off);
530 	else {
531 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
532 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
533 		packidx->hdr.large_offsets[packidx->nlargeobj] =
534 		    htobe64(obj->off);
535 		packidx->nlargeobj++;
536 	}
537 
538 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
539 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
540 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
541 	}
542 }
543 
544 static int
545 indexed_obj_cmp(const void *pa, const void *pb)
546 {
547 	struct got_indexed_object *a, *b;
548 
549 	a = (struct got_indexed_object *)pa;
550 	b = (struct got_indexed_object *)pb;
551 	return got_object_id_cmp(&a->id, &b->id);
552 }
553 
554 static void
555 make_packidx(struct got_packidx *packidx, uint32_t nobj,
556     struct got_indexed_object *objects)
557 {
558 	struct got_indexed_object *obj;
559 	int i;
560 	uint32_t idx = 0;
561 
562 	qsort(objects, nobj, sizeof(struct got_indexed_object),
563 	    indexed_obj_cmp);
564 
565 	memset(packidx->hdr.fanout_table, 0,
566 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
567 	packidx->nlargeobj = 0;
568 
569 	for (i = 0; i < nobj; i++) {
570 		obj = &objects[i];
571 		if (obj->valid)
572 			add_indexed_object(packidx, idx++, obj);
573 	}
574 }
575 
576 static void
577 update_packidx(struct got_packidx *packidx, uint32_t nobj,
578     struct got_indexed_object *obj)
579 {
580 	int idx;
581 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
582 
583 	idx = find_object_idx(packidx, obj->id.sha1);
584 	if (idx == -1) {
585 		char hex[SHA1_DIGEST_STRING_LENGTH];
586 		got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
587 		return; /* object already indexed */
588 	}
589 
590 	memmove(&packidx->hdr.sorted_ids[idx + 1],
591 	    &packidx->hdr.sorted_ids[idx],
592 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
593 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
594 	    sizeof(uint32_t) * (nindexed - idx));
595 
596 	add_indexed_object(packidx, idx, obj);
597 }
598 
599 static const struct got_error *
600 report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
601     uint32_t nobj_resolved, struct got_ratelimit *rl,
602     got_pack_index_progress_cb progress_cb, void *progress_arg)
603 {
604 	const struct got_error *err;
605 	int elapsed = 0;
606 
607 	if (rl) {
608 		err = got_ratelimit_check(&elapsed, rl);
609 		if (err || !elapsed)
610 			return err;
611 	}
612 
613 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
614 	    nobj_resolved);
615 }
616 
617 const struct got_error *
618 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
619     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
620     got_pack_index_progress_cb progress_cb, void *progress_arg,
621     struct got_ratelimit *rl)
622 {
623 	const struct got_error *err;
624 	struct got_packfile_hdr hdr;
625 	struct got_packidx packidx;
626 	char buf[8];
627 	char pack_sha1[SHA1_DIGEST_LENGTH];
628 	uint32_t nobj, nvalid, nloose, nresolved = 0, i;
629 	struct got_indexed_object *objects = NULL, *obj;
630 	SHA1_CTX ctx;
631 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
632 	ssize_t r, w;
633 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
634 	size_t mapoff = 0;
635 	int p_indexed = 0, last_p_indexed = -1;
636 	int p_resolved = 0, last_p_resolved = -1;
637 
638 	/* Require that pack file header and SHA1 trailer are present. */
639 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
640 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
641 		    "short pack file");
642 
643 	if (pack->map) {
644 		memcpy(&hdr, pack->map, sizeof(hdr));
645 		mapoff += sizeof(hdr);
646 	} else {
647 		r = read(pack->fd, &hdr, sizeof(hdr));
648 		if (r == -1)
649 			return got_error_from_errno("read");
650 		if (r < sizeof(hdr))
651 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
652 			    "short pack file");
653 	}
654 
655 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
656 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
657 		    "bad packfile signature");
658 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
659 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
660 		    "bad packfile version");
661 	nobj = be32toh(hdr.nobjects);
662 	if (nobj == 0)
663 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
664 		    "bad packfile with zero objects");
665 
666 	/* We compute the SHA1 of pack file contents and verify later on. */
667 	SHA1Init(&ctx);
668 	SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
669 
670 	/*
671 	 * Create an in-memory pack index which will grow as objects
672 	 * IDs in the pack file are discovered. Only fields used to
673 	 * read deltified objects will be needed by the pack.c library
674 	 * code, so setting up just a pack index header is sufficient.
675 	 */
676 	memset(&packidx, 0, sizeof(packidx));
677 	packidx.hdr.magic = malloc(sizeof(uint32_t));
678 	if (packidx.hdr.magic == NULL)
679 		return got_error_from_errno("malloc");
680 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
681 	packidx.hdr.version = malloc(sizeof(uint32_t));
682 	if (packidx.hdr.version == NULL) {
683 		err = got_error_from_errno("malloc");
684 		goto done;
685 	}
686 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
687 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
688 	    sizeof(uint32_t));
689 	if (packidx.hdr.fanout_table == NULL) {
690 		err = got_error_from_errno("calloc");
691 		goto done;
692 	}
693 	packidx.hdr.sorted_ids = calloc(nobj,
694 	    sizeof(struct got_packidx_object_id));
695 	if (packidx.hdr.sorted_ids == NULL) {
696 		err = got_error_from_errno("calloc");
697 		goto done;
698 	}
699 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
700 	if (packidx.hdr.crc32 == NULL) {
701 		err = got_error_from_errno("calloc");
702 		goto done;
703 	}
704 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
705 	if (packidx.hdr.offsets == NULL) {
706 		err = got_error_from_errno("calloc");
707 		goto done;
708 	}
709 	/* Large offsets table is empty for pack files < 2 GB. */
710 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
711 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
712 		if (packidx.hdr.large_offsets == NULL) {
713 			err = got_error_from_errno("calloc");
714 			goto done;
715 		}
716 	}
717 
718 	nvalid = 0;
719 	nloose = 0;
720 	objects = calloc(nobj, sizeof(struct got_indexed_object));
721 	if (objects == NULL)
722 		return got_error_from_errno("calloc");
723 
724 	/*
725 	 * First pass: locate all objects and identify un-deltified objects.
726 	 *
727 	 * When this pass has completed we will know offset, type, size, and
728 	 * CRC information for all objects in this pack file. We won't know
729 	 * any of the actual object IDs of deltified objects yet since we
730 	 * will not yet attempt to combine deltas.
731 	 */
732 	pass = 1;
733 	for (i = 0; i < nobj; i++) {
734 		/* Don't send too many progress privsep messages. */
735 		p_indexed = ((i + 1) * 100) / nobj;
736 		if (p_indexed != last_p_indexed) {
737 			err = report_progress(nobj, i + 1, nloose, 0,
738 			    rl, progress_cb, progress_arg);
739 			if (err)
740 				goto done;
741 			last_p_indexed = p_indexed;
742 		}
743 
744 		obj = &objects[i];
745 		obj->crc = crc32(0L, NULL, 0);
746 
747 		/* Store offset to type+size information for this object. */
748 		if (pack->map) {
749 			obj->off = mapoff;
750 		} else {
751 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
752 			if (obj->off == -1) {
753 				err = got_error_from_errno("lseek");
754 				goto done;
755 			}
756 		}
757 
758 		err = read_packed_object(pack, obj, tmpfile, &ctx);
759 		if (err)
760 			goto done;
761 
762 		if (pack->map) {
763 			mapoff += obj->tslen + obj->len;
764 		} else {
765 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
766 			    SEEK_SET) == -1) {
767 				err = got_error_from_errno("lseek");
768 				goto done;
769 			}
770 		}
771 
772 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
773 		    obj->type == GOT_OBJ_TYPE_TREE ||
774 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
775 		    obj->type == GOT_OBJ_TYPE_TAG) {
776 			obj->valid = 1;
777 			nloose++;
778 		} else {
779 			if (first_delta_idx == -1)
780 				first_delta_idx = i;
781 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
782 				have_ref_deltas = 1;
783 		}
784 	}
785 	nvalid = nloose;
786 
787 	/*
788 	 * Having done a full pass over the pack file and can now
789 	 * verify its checksum.
790 	 */
791 	SHA1Final(pack_sha1, &ctx);
792 
793 	if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
794 		err = got_error(GOT_ERR_PACKFILE_CSUM);
795 		goto done;
796 	}
797 
798 	/* Verify the SHA1 checksum stored at the end of the pack file. */
799 	if (pack->map) {
800 		memcpy(pack_sha1_expected, pack->map +
801 		    pack->filesize - SHA1_DIGEST_LENGTH,
802 		    SHA1_DIGEST_LENGTH);
803 	} else {
804 		ssize_t n;
805 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
806 			err = got_error_from_errno("lseek");
807 			goto done;
808 		}
809 		n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
810 		if (n == -1) {
811 			err = got_error_from_errno("read");
812 			goto done;
813 		}
814 		if (n != SHA1_DIGEST_LENGTH) {
815 			err = got_error(GOT_ERR_IO);
816 			goto done;
817 		}
818 	}
819 	if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
820 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
821 		    "bad checksum in pack file trailer");
822 		goto done;
823 	}
824 
825 	if (first_delta_idx == -1)
826 		first_delta_idx = 0;
827 
828 	/* In order to resolve ref deltas we need an in-progress pack index. */
829 	if (have_ref_deltas)
830 		make_packidx(&packidx, nobj, objects);
831 
832 	/*
833 	 * Second pass: We can now resolve deltas to compute the IDs of
834 	 * objects which appear in deltified form. Because deltas can be
835 	 * chained this pass may require a couple of iterations until all
836 	 * IDs of deltified objects have been discovered.
837 	 */
838 	pass++;
839 	while (nvalid != nobj) {
840 		int n = 0;
841 		/*
842 		 * This loop will only run once unless the pack file
843 		 * contains ref deltas which refer to objects located
844 		 * later in the pack file, which is unusual.
845 		 * Offset deltas can always be resolved in one pass
846 		 * unless the packfile is corrupt.
847 		 */
848 		for (i = first_delta_idx; i < nobj; i++) {
849 			obj = &objects[i];
850 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
851 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
852 				continue;
853 
854 			if (obj->valid)
855 				continue;
856 
857 			if (pack->map == NULL && lseek(pack->fd,
858 			    obj->off + obj->tslen, SEEK_SET) == -1) {
859 				err = got_error_from_errno("lseek");
860 				goto done;
861 			}
862 
863 			err = resolve_deltified_object(pack, &packidx, obj,
864 			    tmpfile, delta_base_file, delta_accum_file);
865 			if (err) {
866 				if (err->code != GOT_ERR_NO_OBJ)
867 					goto done;
868 				/*
869 				 * We cannot resolve this object yet because
870 				 * a delta base is unknown. Try again later.
871 				 */
872 				continue;
873 			}
874 
875 			obj->valid = 1;
876 			n++;
877 			if (have_ref_deltas)
878 				update_packidx(&packidx, nobj, obj);
879 			/* Don't send too many progress privsep messages. */
880 			p_resolved = ((nresolved + n) * 100) / nobj;
881 			if (p_resolved != last_p_resolved) {
882 				err = report_progress(nobj, nobj,
883 				    nloose, nresolved + n, rl,
884 				    progress_cb, progress_arg);
885 				if (err)
886 					goto done;
887 				last_p_resolved = p_resolved;
888 			}
889 
890 		}
891 		if (pass++ > 3 && n == 0) {
892 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
893 			    "could not resolve any of deltas; packfile could "
894 			    "be corrupt");
895 			goto done;
896 		}
897 		nresolved += n;
898 		nvalid += nresolved;
899 	}
900 
901 	if (nloose + nresolved != nobj) {
902 		static char msg[64];
903 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
904 		    nloose + nresolved, nobj);
905 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
906 		goto done;
907 	}
908 
909 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
910 	    progress_cb, progress_arg);
911 	if (err)
912 		goto done;
913 
914 	make_packidx(&packidx, nobj, objects);
915 
916 	free(objects);
917 	objects = NULL;
918 
919 	SHA1Init(&ctx);
920 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
921 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
922 	err = got_pack_hwrite(idxfd, buf, 8, &ctx);
923 	if (err)
924 		goto done;
925 	err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
926 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
927 	if (err)
928 		goto done;
929 	err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
930 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
931 	if (err)
932 		goto done;
933 	err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
934 	    nobj * sizeof(uint32_t), &ctx);
935 	if (err)
936 		goto done;
937 	err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
938 	    nobj * sizeof(uint32_t), &ctx);
939 	if (err)
940 		goto done;
941 	if (packidx.nlargeobj > 0) {
942 		err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
943 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
944 		if (err)
945 			goto done;
946 	}
947 	err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
948 	if (err)
949 		goto done;
950 
951 	SHA1Final(packidx_hash, &ctx);
952 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
953 	if (w == -1) {
954 		err = got_error_from_errno("write");
955 		goto done;
956 	}
957 	if (w != sizeof(packidx_hash)) {
958 		err = got_error(GOT_ERR_IO);
959 		goto done;
960 	}
961 done:
962 	free(objects);
963 	free(packidx.hdr.magic);
964 	free(packidx.hdr.version);
965 	free(packidx.hdr.fanout_table);
966 	free(packidx.hdr.sorted_ids);
967 	free(packidx.hdr.offsets);
968 	free(packidx.hdr.large_offsets);
969 	return err;
970 }