Game of Trees Repos

Blob

Date:: Wed Mar 18 16:13:48 2020 UTC
Message:: write large objects to disk when resolving deltas; raise in-mem delta threshold
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3  * Copyright (c) 2020 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/syslimits.h>
21 #include <sys/time.h>
22 #include <sys/types.h>
23 #include <sys/uio.h>
24 #include <sys/mman.h>
25 
26 #include <stdint.h>
27 #include <errno.h>
28 #include <imsg.h>
29 #include <limits.h>
30 #include <signal.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <ctype.h>
35 #include <sha1.h>
36 #include <fcntl.h>
37 #include <zlib.h>
38 #include <err.h>
39 #include <assert.h>
40 #include <dirent.h>
41 
42 #include "got_error.h"
43 #include "got_object.h"
44 
45 #include "got_lib_sha1.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_inflate.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_object_idset.h"
51 #include "got_lib_privsep.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_delta_cache.h"
54 
55 #ifndef nitems
56 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
57 #endif
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_crc(uint32_t *crc, 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 		*crc = crc32(*crc, buf, r);
146 	}
147 
148 	return NULL;
149 }
150 
151 static const struct got_error *
152 read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len)
153 {
154 	uint8_t buf[8192];
155 	size_t n, r;
156 
157 	for (n = len; n > 0; n -= r) {
158 		r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
159 		if (r == 0) {
160 			if (feof(f))
161 				return NULL;
162 			return got_ferror(f, GOT_ERR_IO);
163 		}
164 		SHA1Update(ctx, buf, r);
165 	}
166 
167 	return NULL;
168 }
169 
170 static const struct got_error *
171 read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
172     FILE *tmpfile)
173 {
174 	const struct got_error *err = NULL;
175 	SHA1_CTX ctx;
176 	uint8_t *data = NULL;
177 	size_t datalen = 0;
178 	ssize_t n;
179 	char *header;
180 	size_t headerlen;
181 	const char *obj_label;
182 	size_t mapoff = obj->off;
183 
184 	err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
185 	    &obj->tslen, pack, obj->off);
186 	if (err)
187 		return err;
188 
189 	if (pack->map) {
190 		obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
191 		mapoff += obj->tslen;
192 	} else {
193 		/* XXX Seek back and get the CRC of on-disk type+size bytes. */
194 		if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
195 			return got_error_from_errno("lseek");
196 		err = read_crc(&obj->crc, pack->fd, obj->tslen);
197 		if (err)
198 			return err;
199 	}
200 
201 	switch (obj->type) {
202 	case GOT_OBJ_TYPE_BLOB:
203 	case GOT_OBJ_TYPE_COMMIT:
204 	case GOT_OBJ_TYPE_TREE:
205 	case GOT_OBJ_TYPE_TAG:
206 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
207 			if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
208 				err = got_error_from_errno("fseek");
209 				break;
210 			}
211 			if (pack->map) {
212 				err = got_inflate_to_file_mmap(&datalen,
213 				    &obj->len, &obj->crc, pack->map, mapoff,
214 				    pack->filesize - mapoff, tmpfile);
215 			} else {
216 				err = got_inflate_to_file_fd(&datalen,
217 				    &obj->len, &obj->crc, pack->fd, tmpfile);
218 			}
219 		} else {
220 			if (pack->map) {
221 				err = got_inflate_to_mem_mmap(&data, &datalen,
222 				    &obj->len, &obj->crc, pack->map, mapoff,
223 				    pack->filesize - mapoff);
224 			} else {
225 				err = got_inflate_to_mem_fd(&data, &datalen,
226 				    &obj->len, &obj->crc, obj->size, pack->fd);
227 			}
228 		}
229 		if (err)
230 			break;
231 		SHA1Init(&ctx);
232 		err = get_obj_type_label(&obj_label, obj->type);
233 		if (err) {
234 			free(data);
235 			break;
236 		}
237 		if (asprintf(&header, "%s %lld", obj_label, obj->size) == -1) {
238 			err = got_error_from_errno("asprintf");
239 			free(data);
240 			break;
241 		}
242 		headerlen = strlen(header) + 1;
243 		SHA1Update(&ctx, header, headerlen);
244 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
245 			err = read_file_sha1(&ctx, tmpfile, datalen);
246 			if (err)
247 				break;
248 		} else
249 			SHA1Update(&ctx, data, datalen);
250 		SHA1Final(obj->id.sha1, &ctx);
251 		free(header);
252 		free(data);
253 		break;
254 	case GOT_OBJ_TYPE_REF_DELTA:
255 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
256 		if (pack->map) {
257 			if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
258 				err = got_error(GOT_ERR_BAD_PACKFILE);
259 				break;
260 			}
261 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
262 			    SHA1_DIGEST_LENGTH);
263 			obj->crc = crc32(obj->crc, pack->map + mapoff,
264 			    SHA1_DIGEST_LENGTH);
265 			mapoff += SHA1_DIGEST_LENGTH;
266 			err = got_inflate_to_mem_mmap(NULL, &datalen,
267 			    &obj->len, &obj->crc, pack->map, mapoff,
268 			    pack->filesize - mapoff);
269 			if (err)
270 				break;
271 		} else {
272 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
273 			    SHA1_DIGEST_LENGTH);
274 			if (n == -1) {
275 				err = got_error_from_errno("read");
276 				break;
277 			}
278 			if (n < sizeof(obj->id)) {
279 				err = got_error(GOT_ERR_BAD_PACKFILE);
280 				break;
281 			}
282 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
283 			    SHA1_DIGEST_LENGTH);
284 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
285 			    &obj->crc, obj->size, pack->fd);
286 			if (err)
287 				break;
288 		}
289 		obj->len += SHA1_DIGEST_LENGTH;
290 		break;
291 	case GOT_OBJ_TYPE_OFFSET_DELTA:
292 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
293 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
294 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
295 		    obj->tslen);
296 		if (err)
297 			break;
298 
299 		if (pack->map) {
300 			obj->crc = crc32(obj->crc, pack->map + mapoff,
301 			    obj->delta.ofs.base_offsetlen);
302 			mapoff += obj->delta.ofs.base_offsetlen;
303 			err = got_inflate_to_mem_mmap(NULL, &datalen,
304 			    &obj->len, &obj->crc, pack->map, mapoff,
305 			    pack->filesize - mapoff);
306 			if (err)
307 				break;
308 		} else {
309 			/*
310 			 * XXX Seek back and get the CRC of on-disk
311 			 * offset bytes.
312 			 */
313 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
314 			    == -1) {
315 				err = got_error_from_errno("lseek");
316 				break;
317 			}
318 			err = read_crc(&obj->crc, pack->fd,
319 			    obj->delta.ofs.base_offsetlen);
320 			if (err)
321 				break;
322 
323 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
324 			    &obj->crc, obj->size, pack->fd);
325 			if (err)
326 				break;
327 		}
328 		obj->len += obj->delta.ofs.base_offsetlen;
329 		break;
330 	default:
331 		err = got_error(GOT_ERR_OBJ_TYPE);
332 		break;
333 	}
334 
335 	return err;
336 }
337 
338 static const struct got_error *
339 hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
340 {
341 	ssize_t w;
342 
343 	SHA1Update(ctx, buf, len);
344 
345 	w = write(fd, buf, len);
346 	if (w == -1)
347 		return got_error_from_errno("write");
348 	if (w != len)
349 		return got_error(GOT_ERR_IO);
350 
351 	return NULL;
352 }
353 
354 static const struct got_error *
355 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
356     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
357     FILE *delta_accum_file)
358 {
359 	const struct got_error *err = NULL;
360 	struct got_delta_chain deltas;
361 	struct got_delta *delta;
362 	uint8_t *buf = NULL;
363 	size_t len = 0;
364 	SHA1_CTX ctx;
365 	char *header = NULL;
366 	size_t headerlen;
367 	uint64_t max_size;
368 	int base_obj_type;
369 	const char *obj_label;
370 
371 	deltas.nentries = 0;
372 	SIMPLEQ_INIT(&deltas.entries);
373 
374 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
375 	    obj->off, obj->tslen, obj->type, obj->size,
376 	    GOT_DELTA_CHAIN_RECURSION_MAX);
377 	if (err)
378 		goto done;
379 
380 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
381 	if (err)
382 		goto done;
383 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
384 		rewind(tmpfile);
385 		rewind(delta_base_file);
386 		rewind(delta_accum_file);
387 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
388 		    pack, tmpfile, delta_base_file, delta_accum_file);
389 		if (err)
390 			goto done;
391 	} else {
392 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
393 		    &deltas, pack);
394 	}
395 	if (err)
396 		goto done;
397 
398 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
399 	if (err)
400 		goto done;
401 	err = get_obj_type_label(&obj_label, base_obj_type);
402 	if (err)
403 		goto done;
404 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
405 		err = got_error_from_errno("asprintf");
406 		goto done;
407 	}
408 	headerlen = strlen(header) + 1;
409 	SHA1Init(&ctx);
410 	SHA1Update(&ctx, header, headerlen);
411 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
412 		err = read_file_sha1(&ctx, tmpfile, len);
413 		if (err)
414 			goto done;
415 	} else
416 		SHA1Update(&ctx, buf, len);
417 	SHA1Final(obj->id.sha1, &ctx);
418 done:
419 	free(buf);
420 	free(header);
421 	while (!SIMPLEQ_EMPTY(&deltas.entries)) {
422 		delta = SIMPLEQ_FIRST(&deltas.entries);
423 		SIMPLEQ_REMOVE_HEAD(&deltas.entries, entry);
424 		free(delta);
425 	}
426 	return err;
427 }
428 
429 /* Determine the slot in the pack index a given object ID should use. */
430 static int
431 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
432 {
433 	u_int8_t id0 = sha1[0];
434 	uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
435 	int left = 0, right = nindexed - 1;
436 	int cmp = 0, i = 0;
437 
438 	if (id0 > 0)
439 		left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
440 
441 	while (left <= right) {
442 		struct got_packidx_object_id *oid;
443 
444 		i = ((left + right) / 2);
445 		oid = &packidx->hdr.sorted_ids[i];
446 
447 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
448 		if (cmp == 0)
449 			return -1; /* object already indexed */
450 		else if (cmp > 0)
451 			left = i + 1;
452 		else if (cmp < 0)
453 			right = i - 1;
454 	}
455 
456 	return left;
457 }
458 
459 #if 0
460 static void
461 print_packidx(struct got_packidx *packidx)
462 {
463 	uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
464 	int i;
465 
466 	fprintf(stderr, "object IDs:\n");
467 	for (i = 0; i < nindexed; i++) {
468 		char hex[SHA1_DIGEST_STRING_LENGTH];
469 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
470 		    hex, sizeof(hex));
471 		fprintf(stderr, "%s\n", hex);
472 	}
473 	fprintf(stderr, "\n");
474 
475 	fprintf(stderr, "object offsets:\n");
476 	for (i = 0; i < nindexed; i++) {
477 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
478 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
479 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
480 			fprintf(stderr, "%u -> %llu\n", offset, 
481 			    be64toh(packidx->hdr.large_offsets[j]));
482 		} else
483 			fprintf(stderr, "%u\n", offset);
484 	}
485 	fprintf(stderr, "\n");
486 
487 	fprintf(stderr, "fanout table:");
488 	for (i = 0; i <= 0xff; i++)
489 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
490 	fprintf(stderr, "\n");
491 }
492 #endif
493 
494 static void
495 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
496     struct got_indexed_object *obj)
497 {
498 	int i;
499 
500 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
501 	    SHA1_DIGEST_LENGTH);
502 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
503 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
504 		packidx->hdr.offsets[idx] = htobe32(obj->off);
505 	else {
506 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
507 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
508 		packidx->hdr.large_offsets[packidx->nlargeobj] =
509 		    htobe64(obj->off);
510 		packidx->nlargeobj++;
511 	}
512 
513 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
514 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
515 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
516 	}
517 }
518 
519 static int
520 indexed_obj_cmp(const void *pa, const void *pb)
521 {
522 	struct got_indexed_object *a, *b;
523 
524 	a = (struct got_indexed_object *)pa;
525 	b = (struct got_indexed_object *)pb;
526 	return got_object_id_cmp(&a->id, &b->id);
527 }
528 
529 static void
530 make_packidx(struct got_packidx *packidx, int nobj,
531     struct got_indexed_object *objects)
532 {
533 	struct got_indexed_object *obj;
534 	int i;
535 	uint32_t idx = 0;
536 
537 	qsort(objects, nobj, sizeof(struct got_indexed_object),
538 	    indexed_obj_cmp);
539 
540 	memset(packidx->hdr.fanout_table, 0,
541 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
542 	packidx->nlargeobj = 0;
543 
544 	for (i = 0; i < nobj; i++) {
545 		obj = &objects[i];
546 		if (obj->valid)
547 			add_indexed_object(packidx, idx++, obj);
548 	}
549 }
550 
551 static void
552 update_packidx(struct got_packidx *packidx, int nobj,
553     struct got_indexed_object *obj)
554 {
555 	uint32_t idx;
556 	uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
557 
558 	idx = find_object_idx(packidx, obj->id.sha1);
559 	if (idx == -1) {
560 		char hex[SHA1_DIGEST_STRING_LENGTH];
561 		got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
562 		return; /* object already indexed */
563 	}
564 
565 	memmove(&packidx->hdr.sorted_ids[idx + 1],
566 	    &packidx->hdr.sorted_ids[idx],
567 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
568 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
569 	    sizeof(uint32_t) * (nindexed - idx));
570 
571 	add_indexed_object(packidx, idx, obj);
572 }
573 
574 static const struct got_error *
575 index_pack(struct got_pack *pack, int idxfd, FILE *tmpfile,
576     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_hash,
577     struct imsgbuf *ibuf)
578 {
579 	const struct got_error *err;
580 	struct got_packfile_hdr hdr;
581 	struct got_packidx packidx;
582 	char buf[8];
583 	int nobj, nvalid, nloose, nresolved = 0, i;
584 	struct got_indexed_object *objects = NULL, *obj;
585 	SHA1_CTX ctx;
586 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
587 	ssize_t r, w;
588 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
589 	size_t mapoff = 0;
590 	int p_indexed = 0, last_p_indexed = -1;
591 	int p_resolved = 0, last_p_resolved = -1;
592 
593 	/* Check pack file header. */
594 	if (pack->map) {
595 		if (pack->filesize < sizeof(hdr))
596 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
597 			    "short packfile header");
598 		memcpy(&hdr, pack->map, sizeof(hdr));
599 		mapoff += sizeof(hdr);
600 	} else {
601 		r = read(pack->fd, &hdr, sizeof(hdr));
602 		if (r == -1)
603 			return got_error_from_errno("read");
604 		if (r < sizeof(hdr))
605 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
606 			    "short packfile header");
607 	}
608 
609 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
610 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
611 		    "bad packfile signature");
612 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
613 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
614 		    "bad packfile version");
615 	nobj = betoh32(hdr.nobjects);
616 	if (nobj == 0)
617 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
618 		    "bad packfile with zero objects");
619 
620 	/*
621 	 * Create an in-memory pack index which will grow as objects
622 	 * IDs in the pack file are discovered. Only fields used to
623 	 * read deltified objects will be needed by the pack.c library
624 	 * code, so setting up just a pack index header is sufficient.
625 	 */
626 	memset(&packidx, 0, sizeof(packidx));
627 	packidx.hdr.magic = malloc(sizeof(uint32_t));
628 	if (packidx.hdr.magic == NULL)
629 		return got_error_from_errno("calloc");
630 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
631 	packidx.hdr.version = malloc(sizeof(uint32_t));
632 	if (packidx.hdr.version == NULL) {
633 		err = got_error_from_errno("malloc");
634 		goto done;
635 	}
636 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
637 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
638 	    sizeof(uint32_t));
639 	if (packidx.hdr.fanout_table == NULL) {
640 		err = got_error_from_errno("calloc");
641 		goto done;
642 	}
643 	packidx.hdr.sorted_ids = calloc(nobj,
644 	    sizeof(struct got_packidx_object_id));
645 	if (packidx.hdr.sorted_ids == NULL) {
646 		err = got_error_from_errno("calloc");
647 		goto done;
648 	}
649 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
650 	if (packidx.hdr.crc32 == NULL) {
651 		err = got_error_from_errno("calloc");
652 		goto done;
653 	}
654 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
655 	if (packidx.hdr.offsets == NULL) {
656 		err = got_error_from_errno("calloc");
657 		goto done;
658 	}
659 	/* Large offsets table is empty for pack files < 2 GB. */
660 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
661 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
662 		if (packidx.hdr.large_offsets == NULL) {
663 			err = got_error_from_errno("calloc");
664 			goto done;
665 		}
666 	}
667 
668 	nvalid = 0;
669 	nloose = 0;
670 	objects = calloc(nobj, sizeof(struct got_indexed_object));
671 	if (objects == NULL)
672 		return got_error_from_errno("calloc");
673 
674 	/*
675 	 * First pass: locate all objects and identify un-deltified objects.
676 	 *
677 	 * When this pass has completed we will know offset, type, size, and
678 	 * CRC information for all objects in this pack file. We won't know
679 	 * any of the actual object IDs of deltified objects yet since we
680 	 * will not yet attempt to combine deltas.
681 	 */
682 	pass = 1;
683 	for (i = 0; i < nobj; i++) {
684 		/* Don't send too many progress privsep messages. */
685 		p_indexed = ((i + 1) * 100) / nobj;
686 		if (p_indexed != last_p_indexed) {
687 			err = got_privsep_send_index_pack_progress(ibuf,
688 			    nobj, i + 1, nloose, 0);
689 			if (err)
690 				goto done;
691 			last_p_indexed = p_indexed;
692 		}
693 
694 		obj = &objects[i];
695 		obj->crc = crc32(0L, NULL, 0);
696 
697 		/* Store offset to type+size information for this object. */
698 		if (pack->map) {
699 			obj->off = mapoff;
700 		} else {
701 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
702 			if (obj->off == -1) {
703 				err = got_error_from_errno("lseek");
704 				goto done;
705 			}
706 		}
707 
708 		err = read_packed_object(pack, obj, tmpfile);
709 		if (err)
710 			goto done;
711 
712 		if (pack->map) {
713 			mapoff += obj->tslen + obj->len;
714 		} else {
715 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
716 			    SEEK_SET) == -1) {
717 				err = got_error_from_errno("lseek");
718 				goto done;
719 			}
720 		}
721 
722 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
723 		    obj->type == GOT_OBJ_TYPE_TREE ||
724 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
725 		    obj->type == GOT_OBJ_TYPE_TAG) {
726 			obj->valid = 1;
727 			nloose++;
728 		} else {
729 			if (first_delta_idx == -1)
730 				first_delta_idx = i;
731 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
732 				have_ref_deltas = 1;
733 		}
734 	}
735 	nvalid = nloose;
736 
737 	if (first_delta_idx == -1)
738 		first_delta_idx = 0;
739 
740 	/* In order to resolve ref deltas we need an in-progress pack index. */
741 	if (have_ref_deltas)
742 		make_packidx(&packidx, nobj, objects);
743 
744 	/*
745 	 * Second pass: We can now resolve deltas to compute the IDs of
746 	 * objects which appear in deltified form. Because deltas can be
747 	 * chained this pass may require a couple of iterations until all
748 	 * IDs of deltified objects have been discovered.
749 	 */
750 	pass++;
751 	while (nvalid != nobj) {
752 		int n = 0;
753 		/*
754 		 * This loop will only run once unless the pack file
755 		 * contains ref deltas which refer to objects located
756 		 * later in the pack file, which is unusual.
757 		 * Offset deltas can always be resolved in one pass
758 		 * unless the packfile is corrupt.
759 		 */
760 		for (i = first_delta_idx; i < nobj; i++) {
761 			obj = &objects[i];
762 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
763 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
764 				continue;
765 
766 			if (obj->valid)
767 				continue;
768 
769 			if (pack->map == NULL && lseek(pack->fd,
770 			    obj->off + obj->tslen, SEEK_SET) == -1) {
771 				err = got_error_from_errno("lseek");
772 				goto done;
773 			}
774 
775 			err = resolve_deltified_object(pack, &packidx, obj,
776 			    tmpfile, delta_base_file, delta_accum_file);
777 			if (err) {
778 				if (err->code != GOT_ERR_NO_OBJ)
779 					goto done;
780 				/*
781 				 * We cannot resolve this object yet because
782 				 * a delta base is unknown. Try again later.
783 				 */
784 				continue;
785 			}
786 
787 			obj->valid = 1;
788 			n++;
789 			if (have_ref_deltas)
790 				update_packidx(&packidx, nobj, obj);
791 			/* Don't send too many progress privsep messages. */
792 			p_resolved = ((nresolved + n) * 100) / nobj;
793 			if (p_resolved != last_p_resolved) {
794 				err = got_privsep_send_index_pack_progress(ibuf,
795 				    nobj, nobj, nloose, nresolved + n);
796 				if (err)
797 					goto done;
798 				last_p_resolved = p_resolved;
799 			}
800 
801 		}
802 		if (pass++ > 3 && n == 0) {
803 			static char msg[64];
804 			snprintf(msg, sizeof(msg), "could not resolve "
805 			    "any of deltas; packfile could be corrupt");
806 			err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
807 			goto done;
808 			
809 		}
810 		nresolved += n;
811 		nvalid += nresolved;
812 	}
813 
814 	if (nloose + nresolved != nobj) {
815 		static char msg[64];
816 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
817 		    nloose + nresolved, nobj);
818 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
819 		goto done;
820 	}
821 
822 	err = got_privsep_send_index_pack_progress(ibuf, nobj, nobj,
823 	    nloose, nresolved);
824 	if (err)
825 		goto done;
826 
827 	make_packidx(&packidx, nobj, objects);
828 
829 	free(objects);
830 	objects = NULL;
831 
832 	SHA1Init(&ctx);
833 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
834 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
835 	err = hwrite(idxfd, buf, 8, &ctx);
836 	if (err)
837 		goto done;
838 	err = hwrite(idxfd, packidx.hdr.fanout_table,
839 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
840 	if (err)
841 		goto done;
842 	err = hwrite(idxfd, packidx.hdr.sorted_ids,
843 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
844 	if (err)
845 		goto done;
846 	err = hwrite(idxfd, packidx.hdr.crc32, nobj * sizeof(uint32_t), &ctx);
847 	if (err)
848 		goto done;
849 	err = hwrite(idxfd, packidx.hdr.offsets, nobj * sizeof(uint32_t),
850 	    &ctx);
851 	if (err)
852 		goto done;
853 	if (packidx.nlargeobj > 0) {
854 		err = hwrite(idxfd, packidx.hdr.large_offsets,
855 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
856 		if (err)
857 			goto done;
858 	}
859 	err = hwrite(idxfd, pack_hash, SHA1_DIGEST_LENGTH, &ctx);
860 	if (err)
861 		goto done;
862 
863 	SHA1Final(packidx_hash, &ctx);
864 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
865 	if (w == -1) {
866 		err = got_error_from_errno("write");
867 		goto done;
868 	}
869 	if (w != sizeof(packidx_hash)) {
870 		err = got_error(GOT_ERR_IO);
871 		goto done;
872 	}
873 done:
874 	free(objects);
875 	free(packidx.hdr.magic);
876 	free(packidx.hdr.version);
877 	free(packidx.hdr.fanout_table);
878 	free(packidx.hdr.sorted_ids);
879 	free(packidx.hdr.offsets);
880 	free(packidx.hdr.large_offsets);
881 	return err;
882 }
883 
884 int
885 main(int argc, char **argv)
886 {
887 	const struct got_error *err = NULL, *close_err;
888 	struct imsgbuf ibuf;
889 	struct imsg imsg;
890 	int idxfd = -1, tmpfd = -1, i;
891 	FILE *tmpfiles[3];
892 	struct got_pack pack;
893 	uint8_t pack_hash[SHA1_DIGEST_LENGTH];
894 	off_t packfile_size;
895 #if 0
896 	static int attached;
897 	while (!attached)
898 		sleep(1);
899 #endif
900 
901 	for (i = 0; i < nitems(tmpfiles); i++)
902 		tmpfiles[i] = NULL;
903 
904 	memset(&pack, 0, sizeof(pack));
905 	pack.fd = -1;
906 	pack.delta_cache = got_delta_cache_alloc(500,
907 	    GOT_DELTA_RESULT_SIZE_CACHED_MAX);
908 	if (pack.delta_cache == NULL) {
909 		err = got_error_from_errno("got_delta_cache_alloc");
910 		goto done;
911 	}
912 
913 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
914 #ifndef PROFILE
915 	/* revoke access to most system calls */
916 	if (pledge("stdio recvfd", NULL) == -1) {
917 		err = got_error_from_errno("pledge");
918 		got_privsep_send_error(&ibuf, err);
919 		return 1;
920 	}
921 #endif
922 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
923 	if (err)
924 		goto done;
925 	if (imsg.hdr.type == GOT_IMSG_STOP)
926 		goto done;
927 	if (imsg.hdr.type != GOT_IMSG_IDXPACK_REQUEST) {
928 		err = got_error(GOT_ERR_PRIVSEP_MSG);
929 		goto done;
930 	}
931 	if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(pack_hash)) {
932 		err = got_error(GOT_ERR_PRIVSEP_LEN);
933 		goto done;
934 	}
935 	memcpy(pack_hash, imsg.data, sizeof(pack_hash));
936 	pack.fd = imsg.fd;
937 
938 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
939 	if (err)
940 		goto done;
941 	if (imsg.hdr.type == GOT_IMSG_STOP)
942 		goto done;
943 	if (imsg.hdr.type != GOT_IMSG_IDXPACK_OUTFD) {
944 		err = got_error(GOT_ERR_PRIVSEP_MSG);
945 		goto done;
946 	}
947 	if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
948 		err = got_error(GOT_ERR_PRIVSEP_LEN);
949 		goto done;
950 	}
951 	idxfd = imsg.fd;
952 
953 	for (i = 0; i < nitems(tmpfiles); i++) {
954 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
955 		if (err)
956 			goto done;
957 		if (imsg.hdr.type == GOT_IMSG_STOP)
958 			goto done;
959 		if (imsg.hdr.type != GOT_IMSG_TMPFD) {
960 			err = got_error(GOT_ERR_PRIVSEP_MSG);
961 			goto done;
962 		}
963 		if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
964 			err = got_error(GOT_ERR_PRIVSEP_LEN);
965 			goto done;
966 		}
967 		tmpfd = imsg.fd;
968 		tmpfiles[i] = fdopen(tmpfd, "w+");
969 		if (tmpfiles[i] == NULL) {
970 			err = got_error_from_errno("fdopen");
971 			goto done;
972 		}
973 		tmpfd = -1;
974 	}
975 
976 	if (lseek(pack.fd, 0, SEEK_END) == -1) {
977 		err = got_error_from_errno("lseek");
978 		goto done;
979 	}
980 	packfile_size = lseek(pack.fd, 0, SEEK_CUR);
981 	if (packfile_size == -1) {
982 		err = got_error_from_errno("lseek");
983 		goto done;
984 	}
985 	pack.filesize = packfile_size; /* XXX off_t vs size_t */
986 
987 	if (lseek(pack.fd, 0, SEEK_SET) == -1) {
988 		err = got_error_from_errno("lseek");
989 		goto done;
990 	}
991 
992 #ifndef GOT_PACK_NO_MMAP
993 	pack.map = mmap(NULL, pack.filesize, PROT_READ, MAP_PRIVATE,
994 	    pack.fd, 0);
995 	if (pack.map == MAP_FAILED)
996 		pack.map = NULL; /* fall back to read(2) */
997 #endif
998 	err = index_pack(&pack, idxfd, tmpfiles[0], tmpfiles[1], tmpfiles[2],
999 	    pack_hash, &ibuf);
1000 done:
1001 	close_err = got_pack_close(&pack);
1002 	if (close_err && err == NULL)
1003 		err = close_err;
1004 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
1005 		err = got_error_from_errno("close");
1006 	if (tmpfd != -1 && close(tmpfd) == -1 && err == NULL)
1007 		err = got_error_from_errno("close");
1008 	for (i = 0; i < nitems(tmpfiles); i++) {
1009 		if (tmpfiles[i] != NULL && fclose(tmpfiles[i]) == EOF &&
1010 		    err == NULL)
1011 			err = got_error_from_errno("close");
1012 	}
1013 
1014 	if (err == NULL)
1015 		err = got_privsep_send_index_pack_done(&ibuf);
1016 	if (err) {
1017 		got_privsep_send_error(&ibuf, err);
1018 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
1019 		got_privsep_send_error(&ibuf, err);
1020 		exit(1);
1021 	}
1022 
1023 	exit(0);
1024 }