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 			if (mapoff + SHA1_DIGEST_LENGTH > SIZE_MAX) {
275 				err = got_error_fmt(GOT_ERR_RANGE,
276 				    "mapoff %lld would overflow size_t",
277 				    (long long)mapoff + SHA1_DIGEST_LENGTH);
278 				break;
279 			}
280 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
281 			    SHA1_DIGEST_LENGTH);
282 			obj->crc = crc32(obj->crc, pack->map + mapoff,
283 			    SHA1_DIGEST_LENGTH);
284 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
285 			    SHA1_DIGEST_LENGTH);
286 			mapoff += SHA1_DIGEST_LENGTH;
287 			err = got_inflate_to_mem_mmap(NULL, &datalen,
288 			    &obj->len, &csum, pack->map, mapoff,
289 			    pack->filesize - mapoff);
290 			if (err)
291 				break;
292 		} else {
293 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
294 			    SHA1_DIGEST_LENGTH);
295 			if (n == -1) {
296 				err = got_error_from_errno("read");
297 				break;
298 			}
299 			if (n < sizeof(obj->id)) {
300 				err = got_error(GOT_ERR_BAD_PACKFILE);
301 				break;
302 			}
303 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
304 			    SHA1_DIGEST_LENGTH);
305 			SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
306 			    SHA1_DIGEST_LENGTH);
307 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
308 			    &csum, obj->size, pack->fd);
309 			if (err)
310 				break;
311 		}
312 		obj->len += SHA1_DIGEST_LENGTH;
313 		break;
314 	case GOT_OBJ_TYPE_OFFSET_DELTA:
315 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
316 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
317 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
318 		    obj->tslen);
319 		if (err)
320 			break;
321 
322 		if (pack->map) {
323 			if (mapoff + obj->delta.ofs.base_offsetlen >=
324 			    pack->filesize) {
325 				err = got_error(GOT_ERR_BAD_PACKFILE);
326 				break;
327 			}
328 
329 			if (mapoff + obj->delta.ofs.base_offsetlen >
330 			    SIZE_MAX) {
331 				err = got_error_fmt(GOT_ERR_RANGE,
332 				    "mapoff %lld would overflow size_t",
333 				    (long long)mapoff
334 				    + obj->delta.ofs.base_offsetlen);
335 			}
336 
337 			obj->crc = crc32(obj->crc, pack->map + mapoff,
338 			    obj->delta.ofs.base_offsetlen);
339 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
340 			    obj->delta.ofs.base_offsetlen);
341 			mapoff += obj->delta.ofs.base_offsetlen;
342 			err = got_inflate_to_mem_mmap(NULL, &datalen,
343 			    &obj->len, &csum, pack->map, mapoff,
344 			    pack->filesize - mapoff);
345 			if (err)
346 				break;
347 		} else {
348 			/*
349 			 * XXX Seek back and get CRC and SHA1 of on-disk
350 			 * offset bytes.
351 			 */
352 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
353 			    == -1) {
354 				err = got_error_from_errno("lseek");
355 				break;
356 			}
357 			err = read_checksum(&obj->crc, pack_sha1_ctx,
358 			    pack->fd, obj->delta.ofs.base_offsetlen);
359 			if (err)
360 				break;
361 
362 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
363 			    &csum, obj->size, pack->fd);
364 			if (err)
365 				break;
366 		}
367 		obj->len += obj->delta.ofs.base_offsetlen;
368 		break;
369 	default:
370 		err = got_error(GOT_ERR_OBJ_TYPE);
371 		break;
372 	}
373 
374 	return err;
375 }
376 
377 const struct got_error *
378 got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
379 {
380 	ssize_t w;
381 
382 	SHA1Update(ctx, buf, len);
383 
384 	w = write(fd, buf, len);
385 	if (w == -1)
386 		return got_error_from_errno("write");
387 	if (w != len)
388 		return got_error(GOT_ERR_IO);
389 
390 	return NULL;
391 }
392 
393 static const struct got_error *
394 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
395     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
396     FILE *delta_accum_file)
397 {
398 	const struct got_error *err = NULL;
399 	struct got_delta_chain deltas;
400 	struct got_delta *delta;
401 	uint8_t *buf = NULL;
402 	size_t len = 0;
403 	SHA1_CTX ctx;
404 	char *header = NULL;
405 	size_t headerlen;
406 	uint64_t max_size;
407 	int base_obj_type;
408 	const char *obj_label;
409 
410 	deltas.nentries = 0;
411 	STAILQ_INIT(&deltas.entries);
412 
413 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
414 	    obj->off, obj->tslen, obj->type, obj->size,
415 	    GOT_DELTA_CHAIN_RECURSION_MAX);
416 	if (err)
417 		goto done;
418 
419 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
420 	if (err)
421 		goto done;
422 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
423 		rewind(tmpfile);
424 		rewind(delta_base_file);
425 		rewind(delta_accum_file);
426 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
427 		    pack, tmpfile, delta_base_file, delta_accum_file);
428 		if (err)
429 			goto done;
430 	} else {
431 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
432 		    &deltas, pack);
433 	}
434 	if (err)
435 		goto done;
436 
437 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
438 	if (err)
439 		goto done;
440 	err = get_obj_type_label(&obj_label, base_obj_type);
441 	if (err)
442 		goto done;
443 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
444 		err = got_error_from_errno("asprintf");
445 		goto done;
446 	}
447 	headerlen = strlen(header) + 1;
448 	SHA1Init(&ctx);
449 	SHA1Update(&ctx, header, headerlen);
450 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
451 		err = read_file_sha1(&ctx, tmpfile, len);
452 		if (err)
453 			goto done;
454 	} else
455 		SHA1Update(&ctx, buf, len);
456 	SHA1Final(obj->id.sha1, &ctx);
457 done:
458 	free(buf);
459 	free(header);
460 	while (!STAILQ_EMPTY(&deltas.entries)) {
461 		delta = STAILQ_FIRST(&deltas.entries);
462 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
463 		free(delta);
464 	}
465 	return err;
466 }
467 
468 /* Determine the slot in the pack index a given object ID should use. */
469 static int
470 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
471 {
472 	u_int8_t id0 = sha1[0];
473 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
474 	int left = 0, right = nindexed - 1;
475 	int cmp = 0, i = 0;
476 
477 	if (id0 > 0)
478 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
479 
480 	while (left <= right) {
481 		struct got_packidx_object_id *oid;
482 
483 		i = ((left + right) / 2);
484 		oid = &packidx->hdr.sorted_ids[i];
485 
486 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
487 		if (cmp == 0)
488 			return -1; /* object already indexed */
489 		else if (cmp > 0)
490 			left = i + 1;
491 		else if (cmp < 0)
492 			right = i - 1;
493 	}
494 
495 	return left;
496 }
497 
498 #if 0
499 static void
500 print_packidx(struct got_packidx *packidx)
501 {
502 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
503 	int i;
504 
505 	fprintf(stderr, "object IDs:\n");
506 	for (i = 0; i < nindexed; i++) {
507 		char hex[SHA1_DIGEST_STRING_LENGTH];
508 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
509 		    hex, sizeof(hex));
510 		fprintf(stderr, "%s\n", hex);
511 	}
512 	fprintf(stderr, "\n");
513 
514 	fprintf(stderr, "object offsets:\n");
515 	for (i = 0; i < nindexed; i++) {
516 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
517 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
518 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
519 			fprintf(stderr, "%u -> %llu\n", offset,
520 			    be64toh(packidx->hdr.large_offsets[j]));
521 		} else
522 			fprintf(stderr, "%u\n", offset);
523 	}
524 	fprintf(stderr, "\n");
525 
526 	fprintf(stderr, "fanout table:");
527 	for (i = 0; i <= 0xff; i++)
528 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
529 	fprintf(stderr, "\n");
530 }
531 #endif
532 
533 static void
534 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
535     struct got_indexed_object *obj)
536 {
537 	int i;
538 
539 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
540 	    SHA1_DIGEST_LENGTH);
541 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
542 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
543 		packidx->hdr.offsets[idx] = htobe32(obj->off);
544 	else {
545 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
546 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
547 		packidx->hdr.large_offsets[packidx->nlargeobj] =
548 		    htobe64(obj->off);
549 		packidx->nlargeobj++;
550 	}
551 
552 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
553 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
554 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
555 	}
556 }
557 
558 static int
559 indexed_obj_cmp(const void *pa, const void *pb)
560 {
561 	struct got_indexed_object *a, *b;
562 
563 	a = (struct got_indexed_object *)pa;
564 	b = (struct got_indexed_object *)pb;
565 	return got_object_id_cmp(&a->id, &b->id);
566 }
567 
568 static void
569 make_packidx(struct got_packidx *packidx, uint32_t nobj,
570     struct got_indexed_object *objects)
571 {
572 	struct got_indexed_object *obj;
573 	int i;
574 	uint32_t idx = 0;
575 
576 	qsort(objects, nobj, sizeof(struct got_indexed_object),
577 	    indexed_obj_cmp);
578 
579 	memset(packidx->hdr.fanout_table, 0,
580 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
581 	packidx->nlargeobj = 0;
582 
583 	for (i = 0; i < nobj; i++) {
584 		obj = &objects[i];
585 		if (obj->valid)
586 			add_indexed_object(packidx, idx++, obj);
587 	}
588 }
589 
590 static void
591 update_packidx(struct got_packidx *packidx, uint32_t nobj,
592     struct got_indexed_object *obj)
593 {
594 	int idx;
595 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
596 
597 	idx = find_object_idx(packidx, obj->id.sha1);
598 	if (idx == -1) {
599 		char hex[SHA1_DIGEST_STRING_LENGTH];
600 		got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
601 		return; /* object already indexed */
602 	}
603 
604 	memmove(&packidx->hdr.sorted_ids[idx + 1],
605 	    &packidx->hdr.sorted_ids[idx],
606 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
607 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
608 	    sizeof(uint32_t) * (nindexed - idx));
609 
610 	add_indexed_object(packidx, idx, obj);
611 }
612 
613 static const struct got_error *
614 report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
615     uint32_t nobj_resolved, struct got_ratelimit *rl,
616     got_pack_index_progress_cb progress_cb, void *progress_arg)
617 {
618 	const struct got_error *err;
619 	int elapsed = 0;
620 
621 	if (rl) {
622 		err = got_ratelimit_check(&elapsed, rl);
623 		if (err || !elapsed)
624 			return err;
625 	}
626 
627 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
628 	    nobj_resolved);
629 }
630 
631 const struct got_error *
632 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
633     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
634     got_pack_index_progress_cb progress_cb, void *progress_arg,
635     struct got_ratelimit *rl)
636 {
637 	const struct got_error *err;
638 	struct got_packfile_hdr hdr;
639 	struct got_packidx packidx;
640 	char buf[8];
641 	char pack_sha1[SHA1_DIGEST_LENGTH];
642 	uint32_t nobj, nvalid, nloose, nresolved = 0, i;
643 	struct got_indexed_object *objects = NULL, *obj;
644 	SHA1_CTX ctx;
645 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
646 	ssize_t r, w;
647 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
648 	size_t mapoff = 0;
649 	int p_indexed = 0, last_p_indexed = -1;
650 	int p_resolved = 0, last_p_resolved = -1;
651 
652 	/* Require that pack file header and SHA1 trailer are present. */
653 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
654 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
655 		    "short pack file");
656 
657 	if (pack->map) {
658 		memcpy(&hdr, pack->map, sizeof(hdr));
659 		mapoff += sizeof(hdr);
660 	} else {
661 		r = read(pack->fd, &hdr, sizeof(hdr));
662 		if (r == -1)
663 			return got_error_from_errno("read");
664 		if (r < sizeof(hdr))
665 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
666 			    "short pack file");
667 	}
668 
669 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
670 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
671 		    "bad packfile signature");
672 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
673 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
674 		    "bad packfile version");
675 	nobj = be32toh(hdr.nobjects);
676 	if (nobj == 0)
677 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
678 		    "bad packfile with zero objects");
679 
680 	/* We compute the SHA1 of pack file contents and verify later on. */
681 	SHA1Init(&ctx);
682 	SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
683 
684 	/*
685 	 * Create an in-memory pack index which will grow as objects
686 	 * IDs in the pack file are discovered. Only fields used to
687 	 * read deltified objects will be needed by the pack.c library
688 	 * code, so setting up just a pack index header is sufficient.
689 	 */
690 	memset(&packidx, 0, sizeof(packidx));
691 	packidx.hdr.magic = malloc(sizeof(uint32_t));
692 	if (packidx.hdr.magic == NULL)
693 		return got_error_from_errno("malloc");
694 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
695 	packidx.hdr.version = malloc(sizeof(uint32_t));
696 	if (packidx.hdr.version == NULL) {
697 		err = got_error_from_errno("malloc");
698 		goto done;
699 	}
700 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
701 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
702 	    sizeof(uint32_t));
703 	if (packidx.hdr.fanout_table == NULL) {
704 		err = got_error_from_errno("calloc");
705 		goto done;
706 	}
707 	packidx.hdr.sorted_ids = calloc(nobj,
708 	    sizeof(struct got_packidx_object_id));
709 	if (packidx.hdr.sorted_ids == NULL) {
710 		err = got_error_from_errno("calloc");
711 		goto done;
712 	}
713 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
714 	if (packidx.hdr.crc32 == NULL) {
715 		err = got_error_from_errno("calloc");
716 		goto done;
717 	}
718 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
719 	if (packidx.hdr.offsets == NULL) {
720 		err = got_error_from_errno("calloc");
721 		goto done;
722 	}
723 	/* Large offsets table is empty for pack files < 2 GB. */
724 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
725 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
726 		if (packidx.hdr.large_offsets == NULL) {
727 			err = got_error_from_errno("calloc");
728 			goto done;
729 		}
730 	}
731 
732 	nvalid = 0;
733 	nloose = 0;
734 	objects = calloc(nobj, sizeof(struct got_indexed_object));
735 	if (objects == NULL)
736 		return got_error_from_errno("calloc");
737 
738 	/*
739 	 * First pass: locate all objects and identify un-deltified objects.
740 	 *
741 	 * When this pass has completed we will know offset, type, size, and
742 	 * CRC information for all objects in this pack file. We won't know
743 	 * any of the actual object IDs of deltified objects yet since we
744 	 * will not yet attempt to combine deltas.
745 	 */
746 	pass = 1;
747 	for (i = 0; i < nobj; i++) {
748 		/* Don't send too many progress privsep messages. */
749 		p_indexed = ((i + 1) * 100) / nobj;
750 		if (p_indexed != last_p_indexed) {
751 			err = report_progress(nobj, i + 1, nloose, 0,
752 			    rl, progress_cb, progress_arg);
753 			if (err)
754 				goto done;
755 			last_p_indexed = p_indexed;
756 		}
757 
758 		obj = &objects[i];
759 		obj->crc = crc32(0L, NULL, 0);
760 
761 		/* Store offset to type+size information for this object. */
762 		if (pack->map) {
763 			obj->off = mapoff;
764 		} else {
765 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
766 			if (obj->off == -1) {
767 				err = got_error_from_errno("lseek");
768 				goto done;
769 			}
770 		}
771 
772 		err = read_packed_object(pack, obj, tmpfile, &ctx);
773 		if (err)
774 			goto done;
775 
776 		if (pack->map) {
777 			mapoff += obj->tslen + obj->len;
778 		} else {
779 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
780 			    SEEK_SET) == -1) {
781 				err = got_error_from_errno("lseek");
782 				goto done;
783 			}
784 		}
785 
786 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
787 		    obj->type == GOT_OBJ_TYPE_TREE ||
788 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
789 		    obj->type == GOT_OBJ_TYPE_TAG) {
790 			obj->valid = 1;
791 			nloose++;
792 		} else {
793 			if (first_delta_idx == -1)
794 				first_delta_idx = i;
795 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
796 				have_ref_deltas = 1;
797 		}
798 	}
799 	nvalid = nloose;
800 
801 	/*
802 	 * Having done a full pass over the pack file and can now
803 	 * verify its checksum.
804 	 */
805 	SHA1Final(pack_sha1, &ctx);
806 
807 	if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
808 		err = got_error(GOT_ERR_PACKFILE_CSUM);
809 		goto done;
810 	}
811 
812 	/* Verify the SHA1 checksum stored at the end of the pack file. */
813 	if (pack->map) {
814 		if (pack->filesize > SIZE_MAX) {
815 			err = got_error_fmt(GOT_ERR_RANGE,
816 			    "filesize %lld overflows size_t",
817 			    (long long)pack->filesize);
818 			goto done;
819 		}
820 
821 		memcpy(pack_sha1_expected, pack->map +
822 		    pack->filesize - SHA1_DIGEST_LENGTH,
823 		    SHA1_DIGEST_LENGTH);
824 	} else {
825 		ssize_t n;
826 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
827 			err = got_error_from_errno("lseek");
828 			goto done;
829 		}
830 		n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
831 		if (n == -1) {
832 			err = got_error_from_errno("read");
833 			goto done;
834 		}
835 		if (n != SHA1_DIGEST_LENGTH) {
836 			err = got_error(GOT_ERR_IO);
837 			goto done;
838 		}
839 	}
840 	if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
841 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
842 		    "bad checksum in pack file trailer");
843 		goto done;
844 	}
845 
846 	if (first_delta_idx == -1)
847 		first_delta_idx = 0;
848 
849 	/* In order to resolve ref deltas we need an in-progress pack index. */
850 	if (have_ref_deltas)
851 		make_packidx(&packidx, nobj, objects);
852 
853 	/*
854 	 * Second pass: We can now resolve deltas to compute the IDs of
855 	 * objects which appear in deltified form. Because deltas can be
856 	 * chained this pass may require a couple of iterations until all
857 	 * IDs of deltified objects have been discovered.
858 	 */
859 	pass++;
860 	while (nvalid != nobj) {
861 		int n = 0;
862 		/*
863 		 * This loop will only run once unless the pack file
864 		 * contains ref deltas which refer to objects located
865 		 * later in the pack file, which is unusual.
866 		 * Offset deltas can always be resolved in one pass
867 		 * unless the packfile is corrupt.
868 		 */
869 		for (i = first_delta_idx; i < nobj; i++) {
870 			obj = &objects[i];
871 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
872 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
873 				continue;
874 
875 			if (obj->valid)
876 				continue;
877 
878 			if (pack->map == NULL && lseek(pack->fd,
879 			    obj->off + obj->tslen, SEEK_SET) == -1) {
880 				err = got_error_from_errno("lseek");
881 				goto done;
882 			}
883 
884 			err = resolve_deltified_object(pack, &packidx, obj,
885 			    tmpfile, delta_base_file, delta_accum_file);
886 			if (err) {
887 				if (err->code != GOT_ERR_NO_OBJ)
888 					goto done;
889 				/*
890 				 * We cannot resolve this object yet because
891 				 * a delta base is unknown. Try again later.
892 				 */
893 				continue;
894 			}
895 
896 			obj->valid = 1;
897 			n++;
898 			if (have_ref_deltas)
899 				update_packidx(&packidx, nobj, obj);
900 			/* Don't send too many progress privsep messages. */
901 			p_resolved = ((nresolved + n) * 100) / nobj;
902 			if (p_resolved != last_p_resolved) {
903 				err = report_progress(nobj, nobj,
904 				    nloose, nresolved + n, rl,
905 				    progress_cb, progress_arg);
906 				if (err)
907 					goto done;
908 				last_p_resolved = p_resolved;
909 			}
910 
911 		}
912 		if (pass++ > 3 && n == 0) {
913 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
914 			    "could not resolve any of deltas; packfile could "
915 			    "be corrupt");
916 			goto done;
917 		}
918 		nresolved += n;
919 		nvalid += nresolved;
920 	}
921 
922 	if (nloose + nresolved != nobj) {
923 		static char msg[64];
924 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
925 		    nloose + nresolved, nobj);
926 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
927 		goto done;
928 	}
929 
930 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
931 	    progress_cb, progress_arg);
932 	if (err)
933 		goto done;
934 
935 	make_packidx(&packidx, nobj, objects);
936 
937 	free(objects);
938 	objects = NULL;
939 
940 	SHA1Init(&ctx);
941 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
942 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
943 	err = got_pack_hwrite(idxfd, buf, 8, &ctx);
944 	if (err)
945 		goto done;
946 	err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
947 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
948 	if (err)
949 		goto done;
950 	err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
951 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
952 	if (err)
953 		goto done;
954 	err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
955 	    nobj * sizeof(uint32_t), &ctx);
956 	if (err)
957 		goto done;
958 	err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
959 	    nobj * sizeof(uint32_t), &ctx);
960 	if (err)
961 		goto done;
962 	if (packidx.nlargeobj > 0) {
963 		err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
964 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
965 		if (err)
966 			goto done;
967 	}
968 	err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
969 	if (err)
970 		goto done;
971 
972 	SHA1Final(packidx_hash, &ctx);
973 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
974 	if (w == -1) {
975 		err = got_error_from_errno("write");
976 		goto done;
977 	}
978 	if (w != sizeof(packidx_hash)) {
979 		err = got_error(GOT_ERR_IO);
980 		goto done;
981 	}
982 done:
983 	free(objects);
984 	free(packidx.hdr.magic);
985 	free(packidx.hdr.version);
986 	free(packidx.hdr.fanout_table);
987 	free(packidx.hdr.sorted_ids);
988 	free(packidx.hdr.offsets);
989 	free(packidx.hdr.large_offsets);
990 	return err;
991 }