op public repos

Blob

Date:: Sat Feb 4 14:04:28 2023 UTC
Message:: fixup sed
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 <sha2.h>
36 #include <endian.h>
37 #include <fcntl.h>
38 #include <unistd.h>
39 #include <zlib.h>
40 #include <err.h>
41 #include <assert.h>
42 #include <dirent.h>
43 
44 #include "got_error.h"
45 #include "got_object.h"
46 
47 #include "got_lib_hash.h"
48 #include "got_lib_delta.h"
49 #include "got_lib_inflate.h"
50 #include "got_lib_object.h"
51 #include "got_lib_object_parse.h"
52 #include "got_lib_object_idset.h"
53 #include "got_lib_privsep.h"
54 #include "got_lib_pack.h"
55 #include "got_lib_ratelimit.h"
56 #include "got_lib_pack_index.h"
57 #include "got_lib_delta_cache.h"
58 
59 struct got_indexed_object {
60 	struct got_object_id id;
61 
62 	/*
63 	 * Has this object been fully resolved?
64 	 * If so, we know its ID, otherwise we don't and 'id' is invalid.
65 	 */
66 	int valid;
67 
68 	/* Offset of type+size field for this object in pack file. */
69 	off_t off;
70 
71 	/* Type+size values parsed from pack file. */
72 	uint8_t type;
73 	uint64_t size;
74 
75 	/* Length of on-disk type+size data. */
76 	size_t tslen;
77 
78 	/* Length of object data following type+size. */
79 	size_t len;
80 
81 	uint32_t crc;
82 
83 	union {
84 		struct {
85 			/* For ref deltas. */
86 			struct got_object_id ref_id;
87 		} ref;
88 		struct {
89 			/* For offset deltas. */
90 			off_t base_offset;
91 			size_t base_offsetlen;
92 		} ofs;
93 	} delta;
94 };
95 
96 static void
97 putbe32(char *b, uint32_t n)
98 {
99 	b[0] = n >> 24;
100 	b[1] = n >> 16;
101 	b[2] = n >> 8;
102 	b[3] = n >> 0;
103 }
104 
105 static const struct got_error *
106 get_obj_type_label(const char **label, int obj_type)
107 {
108 	const struct got_error *err = NULL;
109 
110 	switch (obj_type) {
111 	case GOT_OBJ_TYPE_BLOB:
112 		*label = GOT_OBJ_LABEL_BLOB;
113 		break;
114 	case GOT_OBJ_TYPE_TREE:
115 		*label = GOT_OBJ_LABEL_TREE;
116 		break;
117 	case GOT_OBJ_TYPE_COMMIT:
118 		*label = GOT_OBJ_LABEL_COMMIT;
119 		break;
120 	case GOT_OBJ_TYPE_TAG:
121 		*label = GOT_OBJ_LABEL_TAG;
122 		break;
123 	default:
124 		*label = NULL;
125 		err = got_error(GOT_ERR_OBJ_TYPE);
126 		break;
127 	}
128 
129 	return err;
130 }
131 
132 static const struct got_error *
133 read_checksum(uint32_t *crc, SHA1_CTX *hash_ctx, int fd, size_t len)
134 {
135 	uint8_t buf[8192];
136 	size_t n;
137 	ssize_t r;
138 
139 	for (n = len; n > 0; n -= r){
140 		r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
141 		if (r == -1)
142 			return got_error_from_errno("read");
143 		if (r == 0)
144 			break;
145 		if (crc)
146 			*crc = crc32(*crc, buf, r);
147 		if (hash_ctx)
148 			SHA1Update(hash_ctx, buf, r);
149 	}
150 
151 	return NULL;
152 }
153 
154 static const struct got_error *
155 read_file_hash(SHA1_CTX *ctx, FILE *f, size_t len)
156 {
157 	uint8_t buf[8192];
158 	size_t n, r;
159 
160 	for (n = len; n > 0; n -= r) {
161 		r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
162 		if (r == 0) {
163 			if (feof(f))
164 				return NULL;
165 			return got_ferror(f, GOT_ERR_IO);
166 		}
167 		SHA1Update(ctx, buf, r);
168 	}
169 
170 	return NULL;
171 }
172 
173 static const struct got_error *
174 read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
175     FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
176 {
177 	const struct got_error *err = NULL;
178 	SHA1_CTX ctx;
179 	uint8_t *data = NULL;
180 	size_t datalen = 0;
181 	ssize_t n;
182 	char *header;
183 	size_t headerlen;
184 	const char *obj_label;
185 	size_t mapoff = obj->off;
186 	struct got_inflate_checksum csum;
187 
188 	memset(&csum, 0, sizeof(csum));
189 	csum.input_sha1 = pack_sha1_ctx;
190 	csum.input_crc = &obj->crc;
191 
192 	err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
193 	    &obj->tslen, pack, obj->off);
194 	if (err)
195 		return err;
196 
197 	if (pack->map) {
198 		obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
199 		SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
200 		mapoff += obj->tslen;
201 	} else {
202 		/* XXX Seek back and get the CRC of on-disk type+size bytes. */
203 		if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
204 			return got_error_from_errno("lseek");
205 		err = read_checksum(&obj->crc, pack_sha1_ctx,
206 		    pack->fd, obj->tslen);
207 		if (err)
208 			return err;
209 	}
210 
211 	switch (obj->type) {
212 	case GOT_OBJ_TYPE_BLOB:
213 	case GOT_OBJ_TYPE_COMMIT:
214 	case GOT_OBJ_TYPE_TREE:
215 	case GOT_OBJ_TYPE_TAG:
216 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
217 			if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
218 				err = got_error_from_errno("fseek");
219 				break;
220 			}
221 			if (pack->map) {
222 				err = got_inflate_to_file_mmap(&datalen,
223 				    &obj->len, &csum, pack->map, mapoff,
224 				    pack->filesize - mapoff, tmpfile);
225 			} else {
226 				err = got_inflate_to_file_fd(&datalen,
227 				    &obj->len, &csum, pack->fd, tmpfile);
228 			}
229 		} else {
230 			if (pack->map) {
231 				err = got_inflate_to_mem_mmap(&data, &datalen,
232 				    &obj->len, &csum, pack->map, mapoff,
233 				    pack->filesize - mapoff);
234 			} else {
235 				err = got_inflate_to_mem_fd(&data, &datalen,
236 				    &obj->len, &csum, obj->size, pack->fd);
237 			}
238 		}
239 		if (err)
240 			break;
241 		SHA1Init(&ctx);
242 		err = get_obj_type_label(&obj_label, obj->type);
243 		if (err) {
244 			free(data);
245 			break;
246 		}
247 		if (asprintf(&header, "%s %lld", obj_label,
248 		    (long long)obj->size) == -1) {
249 			err = got_error_from_errno("asprintf");
250 			free(data);
251 			break;
252 		}
253 		headerlen = strlen(header) + 1;
254 		SHA1Update(&ctx, header, headerlen);
255 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
256 			err = read_file_hash(&ctx, tmpfile, datalen);
257 			if (err) {
258 				free(header);
259 				free(data);
260 				break;
261 			}
262 		} else
263 			SHA1Update(&ctx, data, datalen);
264 		SHA1Final(obj->id.hash, &ctx);
265 		free(header);
266 		free(data);
267 		break;
268 	case GOT_OBJ_TYPE_REF_DELTA:
269 		memset(obj->id.hash, 0xff, SHA1_DIGEST_LENGTH);
270 		if (pack->map) {
271 			if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
272 				err = got_error(GOT_ERR_BAD_PACKFILE);
273 				break;
274 			}
275 			if (mapoff + SHA1_DIGEST_LENGTH > SIZE_MAX) {
276 				err = got_error_fmt(GOT_ERR_RANGE,
277 				    "mapoff %lld would overflow size_t",
278 				    (long long)mapoff + SHA1_DIGEST_LENGTH);
279 				break;
280 			}
281 			memcpy(obj->delta.ref.ref_id.hash, pack->map + mapoff,
282 			    SHA1_DIGEST_LENGTH);
283 			obj->crc = crc32(obj->crc, pack->map + mapoff,
284 			    SHA1_DIGEST_LENGTH);
285 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
286 			    SHA1_DIGEST_LENGTH);
287 			mapoff += SHA1_DIGEST_LENGTH;
288 			err = got_inflate_to_mem_mmap(NULL, &datalen,
289 			    &obj->len, &csum, pack->map, mapoff,
290 			    pack->filesize - mapoff);
291 			if (err)
292 				break;
293 		} else {
294 			n = read(pack->fd, obj->delta.ref.ref_id.hash,
295 			    SHA1_DIGEST_LENGTH);
296 			if (n == -1) {
297 				err = got_error_from_errno("read");
298 				break;
299 			}
300 			if (n < sizeof(obj->id)) {
301 				err = got_error(GOT_ERR_BAD_PACKFILE);
302 				break;
303 			}
304 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.hash,
305 			    SHA1_DIGEST_LENGTH);
306 			SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.hash,
307 			    SHA1_DIGEST_LENGTH);
308 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
309 			    &csum, obj->size, pack->fd);
310 			if (err)
311 				break;
312 		}
313 		obj->len += SHA1_DIGEST_LENGTH;
314 		break;
315 	case GOT_OBJ_TYPE_OFFSET_DELTA:
316 		memset(obj->id.hash, 0xff, SHA1_DIGEST_LENGTH);
317 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
318 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
319 		    obj->tslen);
320 		if (err)
321 			break;
322 
323 		if (pack->map) {
324 			if (mapoff + obj->delta.ofs.base_offsetlen >=
325 			    pack->filesize) {
326 				err = got_error(GOT_ERR_BAD_PACKFILE);
327 				break;
328 			}
329 
330 			if (mapoff + obj->delta.ofs.base_offsetlen >
331 			    SIZE_MAX) {
332 				err = got_error_fmt(GOT_ERR_RANGE,
333 				    "mapoff %lld would overflow size_t",
334 				    (long long)mapoff
335 				    + obj->delta.ofs.base_offsetlen);
336 			}
337 
338 			obj->crc = crc32(obj->crc, pack->map + mapoff,
339 			    obj->delta.ofs.base_offsetlen);
340 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
341 			    obj->delta.ofs.base_offsetlen);
342 			mapoff += obj->delta.ofs.base_offsetlen;
343 			err = got_inflate_to_mem_mmap(NULL, &datalen,
344 			    &obj->len, &csum, pack->map, mapoff,
345 			    pack->filesize - mapoff);
346 			if (err)
347 				break;
348 		} else {
349 			/*
350 			 * XXX Seek back and get CRC and SHA1 of on-disk
351 			 * offset bytes.
352 			 */
353 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
354 			    == -1) {
355 				err = got_error_from_errno("lseek");
356 				break;
357 			}
358 			err = read_checksum(&obj->crc, pack_sha1_ctx,
359 			    pack->fd, obj->delta.ofs.base_offsetlen);
360 			if (err)
361 				break;
362 
363 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
364 			    &csum, obj->size, pack->fd);
365 			if (err)
366 				break;
367 		}
368 		obj->len += obj->delta.ofs.base_offsetlen;
369 		break;
370 	default:
371 		err = got_error(GOT_ERR_OBJ_TYPE);
372 		break;
373 	}
374 
375 	return err;
376 }
377 
378 const struct got_error *
379 got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
380 {
381 	ssize_t w;
382 
383 	SHA1Update(ctx, buf, len);
384 
385 	w = write(fd, buf, len);
386 	if (w == -1)
387 		return got_error_from_errno("write");
388 	if (w != len)
389 		return got_error(GOT_ERR_IO);
390 
391 	return NULL;
392 }
393 
394 static const struct got_error *
395 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
396     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
397     FILE *delta_accum_file)
398 {
399 	const struct got_error *err = NULL;
400 	struct got_delta_chain deltas;
401 	struct got_delta *delta;
402 	uint8_t *buf = NULL;
403 	size_t len = 0;
404 	SHA1_CTX ctx;
405 	char *header = NULL;
406 	size_t headerlen;
407 	uint64_t max_size;
408 	int base_obj_type;
409 	const char *obj_label;
410 
411 	deltas.nentries = 0;
412 	STAILQ_INIT(&deltas.entries);
413 
414 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
415 	    obj->off, obj->tslen, obj->type, obj->size,
416 	    GOT_DELTA_CHAIN_RECURSION_MAX);
417 	if (err)
418 		goto done;
419 
420 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
421 	if (err)
422 		goto done;
423 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
424 		rewind(tmpfile);
425 		rewind(delta_base_file);
426 		rewind(delta_accum_file);
427 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
428 		    pack, tmpfile, delta_base_file, delta_accum_file);
429 		if (err)
430 			goto done;
431 	} else {
432 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
433 		    &deltas, pack);
434 	}
435 	if (err)
436 		goto done;
437 
438 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
439 	if (err)
440 		goto done;
441 	err = get_obj_type_label(&obj_label, base_obj_type);
442 	if (err)
443 		goto done;
444 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
445 		err = got_error_from_errno("asprintf");
446 		goto done;
447 	}
448 	headerlen = strlen(header) + 1;
449 	SHA1Init(&ctx);
450 	SHA1Update(&ctx, header, headerlen);
451 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
452 		err = read_file_hash(&ctx, tmpfile, len);
453 		if (err)
454 			goto done;
455 	} else
456 		SHA1Update(&ctx, buf, len);
457 	SHA1Final(obj->id.hash, &ctx);
458 done:
459 	free(buf);
460 	free(header);
461 	while (!STAILQ_EMPTY(&deltas.entries)) {
462 		delta = STAILQ_FIRST(&deltas.entries);
463 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
464 		free(delta);
465 	}
466 	return err;
467 }
468 
469 /* Determine the slot in the pack index a given object ID should use. */
470 static int
471 find_object_idx(struct got_packidx *packidx, uint8_t *hash)
472 {
473 	u_int8_t id0 = hash[0];
474 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
475 	int left = 0, right = nindexed - 1;
476 	int cmp = 0, i = 0;
477 
478 	if (id0 > 0)
479 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
480 
481 	while (left <= right) {
482 		struct got_packidx_object_id *oid;
483 
484 		i = ((left + right) / 2);
485 		oid = &packidx->hdr.sorted_ids[i];
486 
487 		cmp = memcmp(hash, oid->hash, SHA1_DIGEST_LENGTH);
488 		if (cmp == 0)
489 			return -1; /* object already indexed */
490 		else if (cmp > 0)
491 			left = i + 1;
492 		else if (cmp < 0)
493 			right = i - 1;
494 	}
495 
496 	return left;
497 }
498 
499 #if 0
500 static void
501 print_packidx(struct got_packidx *packidx)
502 {
503 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
504 	int i;
505 
506 	fprintf(stderr, "object IDs:\n");
507 	for (i = 0; i < nindexed; i++) {
508 		char hex[SHA1_DIGEST_STRING_LENGTH];
509 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].hash,
510 		    hex, sizeof(hex));
511 		fprintf(stderr, "%s\n", hex);
512 	}
513 	fprintf(stderr, "\n");
514 
515 	fprintf(stderr, "object offsets:\n");
516 	for (i = 0; i < nindexed; i++) {
517 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
518 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
519 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
520 			fprintf(stderr, "%u -> %llu\n", offset,
521 			    be64toh(packidx->hdr.large_offsets[j]));
522 		} else
523 			fprintf(stderr, "%u\n", offset);
524 	}
525 	fprintf(stderr, "\n");
526 
527 	fprintf(stderr, "fanout table:");
528 	for (i = 0; i <= 0xff; i++)
529 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
530 	fprintf(stderr, "\n");
531 }
532 #endif
533 
534 static void
535 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
536     struct got_indexed_object *obj)
537 {
538 	int i;
539 
540 	memcpy(packidx->hdr.sorted_ids[idx].hash, obj->id.hash,
541 	    SHA1_DIGEST_LENGTH);
542 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
543 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
544 		packidx->hdr.offsets[idx] = htobe32(obj->off);
545 	else {
546 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
547 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
548 		packidx->hdr.large_offsets[packidx->nlargeobj] =
549 		    htobe64(obj->off);
550 		packidx->nlargeobj++;
551 	}
552 
553 	for (i = obj->id.hash[0]; i <= 0xff; i++) {
554 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
555 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
556 	}
557 }
558 
559 static int
560 indexed_obj_cmp(const void *pa, const void *pb)
561 {
562 	struct got_indexed_object *a, *b;
563 
564 	a = (struct got_indexed_object *)pa;
565 	b = (struct got_indexed_object *)pb;
566 	return got_object_id_cmp(&a->id, &b->id);
567 }
568 
569 static void
570 make_packidx(struct got_packidx *packidx, uint32_t nobj,
571     struct got_indexed_object *objects)
572 {
573 	struct got_indexed_object *obj;
574 	int i;
575 	uint32_t idx = 0;
576 
577 	qsort(objects, nobj, sizeof(struct got_indexed_object),
578 	    indexed_obj_cmp);
579 
580 	memset(packidx->hdr.fanout_table, 0,
581 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
582 	packidx->nlargeobj = 0;
583 
584 	for (i = 0; i < nobj; i++) {
585 		obj = &objects[i];
586 		if (obj->valid)
587 			add_indexed_object(packidx, idx++, obj);
588 	}
589 }
590 
591 static void
592 update_packidx(struct got_packidx *packidx, uint32_t nobj,
593     struct got_indexed_object *obj)
594 {
595 	int idx;
596 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
597 
598 	idx = find_object_idx(packidx, obj->id.hash);
599 	if (idx == -1)
600 		return; /* object already indexed */
601 
602 	memmove(&packidx->hdr.sorted_ids[idx + 1],
603 	    &packidx->hdr.sorted_ids[idx],
604 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
605 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
606 	    sizeof(uint32_t) * (nindexed - idx));
607 
608 	add_indexed_object(packidx, idx, obj);
609 }
610 
611 static const struct got_error *
612 report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
613     uint32_t nobj_resolved, struct got_ratelimit *rl,
614     got_pack_index_progress_cb progress_cb, void *progress_arg)
615 {
616 	const struct got_error *err;
617 	int elapsed = 0;
618 
619 	if (rl) {
620 		err = got_ratelimit_check(&elapsed, rl);
621 		if (err || !elapsed)
622 			return err;
623 	}
624 
625 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
626 	    nobj_resolved);
627 }
628 
629 const struct got_error *
630 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
631     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_hash_expected,
632     got_pack_index_progress_cb progress_cb, void *progress_arg,
633     struct got_ratelimit *rl)
634 {
635 	const struct got_error *err;
636 	struct got_packfile_hdr hdr;
637 	struct got_packidx packidx;
638 	char buf[8];
639 	char pack_hash[SHA1_DIGEST_LENGTH];
640 	uint32_t nobj, nvalid, nloose, nresolved = 0, i;
641 	struct got_indexed_object *objects = NULL, *obj;
642 	SHA1_CTX ctx;
643 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
644 	ssize_t r, w;
645 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
646 	size_t mapoff = 0;
647 	int p_indexed = 0, last_p_indexed = -1;
648 	int p_resolved = 0, last_p_resolved = -1;
649 
650 	/* Require that pack file header and SHA1 trailer are present. */
651 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
652 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
653 		    "short pack file");
654 
655 	if (pack->map) {
656 		memcpy(&hdr, pack->map, sizeof(hdr));
657 		mapoff += sizeof(hdr);
658 	} else {
659 		r = read(pack->fd, &hdr, sizeof(hdr));
660 		if (r == -1)
661 			return got_error_from_errno("read");
662 		if (r < sizeof(hdr))
663 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
664 			    "short pack file");
665 	}
666 
667 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
668 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
669 		    "bad packfile signature");
670 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
671 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
672 		    "bad packfile version");
673 	nobj = be32toh(hdr.nobjects);
674 	if (nobj == 0)
675 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
676 		    "bad packfile with zero objects");
677 
678 	/* We compute the SHA1 of pack file contents and verify later on. */
679 	SHA1Init(&ctx);
680 	SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
681 
682 	/*
683 	 * Create an in-memory pack index which will grow as objects
684 	 * IDs in the pack file are discovered. Only fields used to
685 	 * read deltified objects will be needed by the pack.c library
686 	 * code, so setting up just a pack index header is sufficient.
687 	 */
688 	memset(&packidx, 0, sizeof(packidx));
689 	packidx.hdr.magic = malloc(sizeof(uint32_t));
690 	if (packidx.hdr.magic == NULL)
691 		return got_error_from_errno("malloc");
692 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
693 	packidx.hdr.version = malloc(sizeof(uint32_t));
694 	if (packidx.hdr.version == NULL) {
695 		err = got_error_from_errno("malloc");
696 		goto done;
697 	}
698 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
699 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
700 	    sizeof(uint32_t));
701 	if (packidx.hdr.fanout_table == NULL) {
702 		err = got_error_from_errno("calloc");
703 		goto done;
704 	}
705 	packidx.hdr.sorted_ids = calloc(nobj,
706 	    sizeof(struct got_packidx_object_id));
707 	if (packidx.hdr.sorted_ids == NULL) {
708 		err = got_error_from_errno("calloc");
709 		goto done;
710 	}
711 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
712 	if (packidx.hdr.crc32 == NULL) {
713 		err = got_error_from_errno("calloc");
714 		goto done;
715 	}
716 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
717 	if (packidx.hdr.offsets == NULL) {
718 		err = got_error_from_errno("calloc");
719 		goto done;
720 	}
721 	/* Large offsets table is empty for pack files < 2 GB. */
722 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
723 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
724 		if (packidx.hdr.large_offsets == NULL) {
725 			err = got_error_from_errno("calloc");
726 			goto done;
727 		}
728 	}
729 
730 	nvalid = 0;
731 	nloose = 0;
732 	objects = calloc(nobj, sizeof(struct got_indexed_object));
733 	if (objects == NULL)
734 		return got_error_from_errno("calloc");
735 
736 	/*
737 	 * First pass: locate all objects and identify un-deltified objects.
738 	 *
739 	 * When this pass has completed we will know offset, type, size, and
740 	 * CRC information for all objects in this pack file. We won't know
741 	 * any of the actual object IDs of deltified objects yet since we
742 	 * will not yet attempt to combine deltas.
743 	 */
744 	pass = 1;
745 	for (i = 0; i < nobj; i++) {
746 		/* Don't send too many progress privsep messages. */
747 		p_indexed = ((i + 1) * 100) / nobj;
748 		if (p_indexed != last_p_indexed) {
749 			err = report_progress(nobj, i + 1, nloose, 0,
750 			    rl, progress_cb, progress_arg);
751 			if (err)
752 				goto done;
753 			last_p_indexed = p_indexed;
754 		}
755 
756 		obj = &objects[i];
757 		obj->crc = crc32(0L, NULL, 0);
758 
759 		/* Store offset to type+size information for this object. */
760 		if (pack->map) {
761 			obj->off = mapoff;
762 		} else {
763 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
764 			if (obj->off == -1) {
765 				err = got_error_from_errno("lseek");
766 				goto done;
767 			}
768 		}
769 
770 		err = read_packed_object(pack, obj, tmpfile, &ctx);
771 		if (err)
772 			goto done;
773 
774 		if (pack->map) {
775 			mapoff += obj->tslen + obj->len;
776 		} else {
777 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
778 			    SEEK_SET) == -1) {
779 				err = got_error_from_errno("lseek");
780 				goto done;
781 			}
782 		}
783 
784 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
785 		    obj->type == GOT_OBJ_TYPE_TREE ||
786 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
787 		    obj->type == GOT_OBJ_TYPE_TAG) {
788 			obj->valid = 1;
789 			nloose++;
790 		} else {
791 			if (first_delta_idx == -1)
792 				first_delta_idx = i;
793 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
794 				have_ref_deltas = 1;
795 		}
796 	}
797 	nvalid = nloose;
798 
799 	/*
800 	 * Having done a full pass over the pack file and can now
801 	 * verify its checksum.
802 	 */
803 	SHA1Final(pack_hash, &ctx);
804 
805 	if (memcmp(pack_hash_expected, pack_hash, SHA1_DIGEST_LENGTH) != 0) {
806 		err = got_error(GOT_ERR_PACKFILE_CSUM);
807 		goto done;
808 	}
809 
810 	/* Verify the SHA1 checksum stored at the end of the pack file. */
811 	if (pack->map) {
812 		if (pack->filesize > SIZE_MAX) {
813 			err = got_error_fmt(GOT_ERR_RANGE,
814 			    "filesize %lld overflows size_t",
815 			    (long long)pack->filesize);
816 			goto done;
817 		}
818 
819 		memcpy(pack_hash_expected, pack->map +
820 		    pack->filesize - SHA1_DIGEST_LENGTH,
821 		    SHA1_DIGEST_LENGTH);
822 	} else {
823 		ssize_t n;
824 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
825 			err = got_error_from_errno("lseek");
826 			goto done;
827 		}
828 		n = read(pack->fd, pack_hash_expected, SHA1_DIGEST_LENGTH);
829 		if (n == -1) {
830 			err = got_error_from_errno("read");
831 			goto done;
832 		}
833 		if (n != SHA1_DIGEST_LENGTH) {
834 			err = got_error(GOT_ERR_IO);
835 			goto done;
836 		}
837 	}
838 	if (memcmp(pack_hash, pack_hash_expected, SHA1_DIGEST_LENGTH) != 0) {
839 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
840 		    "bad checksum in pack file trailer");
841 		goto done;
842 	}
843 
844 	if (first_delta_idx == -1)
845 		first_delta_idx = 0;
846 
847 	/* In order to resolve ref deltas we need an in-progress pack index. */
848 	if (have_ref_deltas)
849 		make_packidx(&packidx, nobj, objects);
850 
851 	/*
852 	 * Second pass: We can now resolve deltas to compute the IDs of
853 	 * objects which appear in deltified form. Because deltas can be
854 	 * chained this pass may require a couple of iterations until all
855 	 * IDs of deltified objects have been discovered.
856 	 */
857 	pass++;
858 	while (nvalid != nobj) {
859 		int n = 0;
860 		/*
861 		 * This loop will only run once unless the pack file
862 		 * contains ref deltas which refer to objects located
863 		 * later in the pack file, which is unusual.
864 		 * Offset deltas can always be resolved in one pass
865 		 * unless the packfile is corrupt.
866 		 */
867 		for (i = first_delta_idx; i < nobj; i++) {
868 			obj = &objects[i];
869 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
870 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
871 				continue;
872 
873 			if (obj->valid)
874 				continue;
875 
876 			if (pack->map == NULL && lseek(pack->fd,
877 			    obj->off + obj->tslen, SEEK_SET) == -1) {
878 				err = got_error_from_errno("lseek");
879 				goto done;
880 			}
881 
882 			err = resolve_deltified_object(pack, &packidx, obj,
883 			    tmpfile, delta_base_file, delta_accum_file);
884 			if (err) {
885 				if (err->code != GOT_ERR_NO_OBJ)
886 					goto done;
887 				/*
888 				 * We cannot resolve this object yet because
889 				 * a delta base is unknown. Try again later.
890 				 */
891 				continue;
892 			}
893 
894 			obj->valid = 1;
895 			n++;
896 			if (have_ref_deltas)
897 				update_packidx(&packidx, nobj, obj);
898 			/* Don't send too many progress privsep messages. */
899 			p_resolved = ((nresolved + n) * 100) / nobj;
900 			if (p_resolved != last_p_resolved) {
901 				err = report_progress(nobj, nobj,
902 				    nloose, nresolved + n, rl,
903 				    progress_cb, progress_arg);
904 				if (err)
905 					goto done;
906 				last_p_resolved = p_resolved;
907 			}
908 
909 		}
910 		if (pass++ > 3 && n == 0) {
911 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
912 			    "could not resolve any of deltas; packfile could "
913 			    "be corrupt");
914 			goto done;
915 		}
916 		nresolved += n;
917 		nvalid += n;
918 	}
919 
920 	if (nloose + nresolved != nobj) {
921 		static char msg[64];
922 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
923 		    nloose + nresolved, nobj);
924 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
925 		goto done;
926 	}
927 
928 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
929 	    progress_cb, progress_arg);
930 	if (err)
931 		goto done;
932 
933 	make_packidx(&packidx, nobj, objects);
934 
935 	free(objects);
936 	objects = NULL;
937 
938 	SHA1Init(&ctx);
939 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
940 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
941 	err = got_pack_hwrite(idxfd, buf, 8, &ctx);
942 	if (err)
943 		goto done;
944 	err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
945 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
946 	if (err)
947 		goto done;
948 	err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
949 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
950 	if (err)
951 		goto done;
952 	err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
953 	    nobj * sizeof(uint32_t), &ctx);
954 	if (err)
955 		goto done;
956 	err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
957 	    nobj * sizeof(uint32_t), &ctx);
958 	if (err)
959 		goto done;
960 	if (packidx.nlargeobj > 0) {
961 		err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
962 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
963 		if (err)
964 			goto done;
965 	}
966 	err = got_pack_hwrite(idxfd, pack_hash, SHA1_DIGEST_LENGTH, &ctx);
967 	if (err)
968 		goto done;
969 
970 	SHA1Final(packidx_hash, &ctx);
971 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
972 	if (w == -1) {
973 		err = got_error_from_errno("write");
974 		goto done;
975 	}
976 	if (w != sizeof(packidx_hash)) {
977 		err = got_error(GOT_ERR_IO);
978 		goto done;
979 	}
980 done:
981 	free(objects);
982 	free(packidx.hdr.magic);
983 	free(packidx.hdr.version);
984 	free(packidx.hdr.fanout_table);
985 	free(packidx.hdr.sorted_ids);
986 	free(packidx.hdr.offsets);
987 	free(packidx.hdr.large_offsets);
988 	return err;
989 }