op public repos

Blob

Date:: Fri Jan 19 15:28:55 2018 UTC
Message:: Add helper function for opening packed plain objects
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 
21 #include <dirent.h>
22 #include <errno.h>
23 #include <stdio.h>
24 #include <stdint.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <limits.h>
28 #include <sha1.h>
29 #include <endian.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_object.h"
34 #include "got_repository.h"
35 #include "got_sha1.h"
36 #include "pack.h"
37 #include "path.h"
38 #include "delta.h"
39 
40 #define GOT_PACK_PREFIX		"pack-"
41 #define GOT_PACKFILE_SUFFIX	".pack"
42 #define GOT_PACKIDX_SUFFIX		".idx"
43 #define GOT_PACKFILE_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
44 				SHA1_DIGEST_STRING_LENGTH - 1 + \
45 				strlen(GOT_PACKFILE_SUFFIX))
46 #define GOT_PACKIDX_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
47 				SHA1_DIGEST_STRING_LENGTH - 1 + \
48 				strlen(GOT_PACKIDX_SUFFIX))
49 
50 #ifndef MIN
51 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
52 #endif
53 
54 static const struct got_error *
55 verify_fanout_table(uint32_t *fanout_table)
56 {
57 	int i;
58 
59 	for (i = 0; i < 0xff - 1; i++) {
60 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
61 			return got_error(GOT_ERR_BAD_PACKIDX);
62 	}
63 
64 	return NULL;
65 }
66 
67 static const struct got_error *
68 get_packfile_size(size_t *size, const char *path_idx)
69 {
70 	struct stat sb;
71 	char *path_pack;
72 	char base_path[PATH_MAX];
73 	char *dot;
74 
75 	if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
76 		return got_error(GOT_ERR_NO_SPACE);
77 
78 	dot = strrchr(base_path, '.');
79 	if (dot == NULL)
80 		return got_error(GOT_ERR_BAD_PATH);
81 	*dot = '\0';
82 	if (asprintf(&path_pack, "%s.pack", base_path) == -1)
83 		return got_error(GOT_ERR_NO_MEM);
84 
85 	if (stat(path_pack, &sb) != 0) {
86 		free(path_pack);
87 		return got_error_from_errno();
88 	}
89 
90 	free(path_pack);
91 	*size = sb.st_size;
92 	return 0;
93 }
94 
95 const struct got_error *
96 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
97 {
98 	struct got_packidx_v2_hdr *p;
99 	FILE *f;
100 	const struct got_error *err = NULL;
101 	size_t n, nobj, packfile_size;
102 	SHA1_CTX ctx;
103 	uint8_t sha1[SHA1_DIGEST_LENGTH];
104 
105 	SHA1Init(&ctx);
106 
107 	f = fopen(path, "rb");
108 	if (f == NULL)
109 		return got_error(GOT_ERR_BAD_PATH);
110 
111 	err = get_packfile_size(&packfile_size, path);
112 	if (err)
113 		return err;
114 
115 	p = calloc(1, sizeof(*p));
116 	if (p == NULL) {
117 		err = got_error(GOT_ERR_NO_MEM);
118 		goto done;
119 	}
120 
121 	n = fread(&p->magic, sizeof(p->magic), 1, f);
122 	if (n != 1) {
123 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
124 		goto done;
125 	}
126 
127 	if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
128 		err = got_error(GOT_ERR_BAD_PACKIDX);
129 		goto done;
130 	}
131 
132 	SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
133 
134 	n = fread(&p->version, sizeof(p->version), 1, f);
135 	if (n != 1) {
136 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
137 		goto done;
138 	}
139 
140 	if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
141 		err = got_error(GOT_ERR_BAD_PACKIDX);
142 		goto done;
143 	}
144 
145 	SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
146 
147 	n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
148 	if (n != 1) {
149 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
150 		goto done;
151 	}
152 
153 	err = verify_fanout_table(p->fanout_table);
154 	if (err)
155 		goto done;
156 
157 	SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
158 
159 	nobj = betoh32(p->fanout_table[0xff]);
160 
161 	p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
162 	if (p->sorted_ids == NULL) {
163 		err = got_error(GOT_ERR_NO_MEM);
164 		goto done;
165 	}
166 
167 	n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
168 	if (n != nobj) {
169 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
170 		goto done;
171 	}
172 
173 	SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
174 	    nobj * sizeof(*p->sorted_ids));
175 
176 	p->crc32 = calloc(nobj, sizeof(*p->crc32));
177 	if (p->crc32 == NULL) {
178 		err = got_error(GOT_ERR_NO_MEM);
179 		goto done;
180 	}
181 
182 	n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
183 	if (n != nobj) {
184 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
185 		goto done;
186 	}
187 
188 	SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
189 
190 	p->offsets = calloc(nobj, sizeof(*p->offsets));
191 	if (p->offsets == NULL) {
192 		err = got_error(GOT_ERR_NO_MEM);
193 		goto done;
194 	}
195 
196 	n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
197 	if (n != nobj) {
198 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
199 		goto done;
200 	}
201 
202 	SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
203 
204 	/* Large file offsets are contained only in files > 2GB. */
205 	if (packfile_size <= 0x80000000)
206 		goto checksum;
207 
208 	p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
209 	if (p->large_offsets == NULL) {
210 		err = got_error(GOT_ERR_NO_MEM);
211 		goto done;
212 	}
213 
214 	n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
215 	if (n != nobj) {
216 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
217 		goto done;
218 	}
219 
220 	SHA1Update(&ctx, (uint8_t*)p->large_offsets,
221 	    nobj * sizeof(*p->large_offsets));
222 
223 checksum:
224 	n = fread(&p->trailer, sizeof(p->trailer), 1, f);
225 	if (n != 1) {
226 		err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
227 		goto done;
228 	}
229 
230 	SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
231 	SHA1Final(sha1, &ctx);
232 	if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
233 		err = got_error(GOT_ERR_PACKIDX_CSUM);
234 done:
235 	fclose(f);
236 	if (err)
237 		got_packidx_close(p);
238 	else
239 		*packidx = p;
240 	return err;
241 }
242 
243 void
244 got_packidx_close(struct got_packidx_v2_hdr *packidx)
245 {
246 	free(packidx->sorted_ids);
247 	free(packidx->offsets);
248 	free(packidx->crc32);
249 	free(packidx->large_offsets);
250 	free(packidx);
251 }
252 
253 static int
254 is_packidx_filename(const char *name, size_t len)
255 {
256 	if (len != GOT_PACKIDX_NAMELEN)
257 		return 0;
258 
259 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
260 		return 0;
261 
262 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
263 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
264 		return 0;
265 
266 	return 1;
267 }
268 
269 static off_t
270 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
271 {
272 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
273 	uint32_t offset = betoh32(packidx->offsets[idx]);
274 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
275 		uint64_t loffset;
276 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
277 		if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
278 			return -1;
279 		loffset = betoh64(packidx->large_offsets[idx]);
280 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
281 	}
282 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
283 }
284 
285 static int
286 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id)
287 {
288 	u_int8_t id0 = id->sha1[0];
289 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
290 	int i = 0;
291 
292 	if (id0 > 0)
293 		i = betoh32(packidx->fanout_table[id0 - 1]);
294 
295 	while (i < totobj) {
296 		struct got_object_id *oid = &packidx->sorted_ids[i];
297 		uint32_t offset;
298 		int cmp = got_object_id_cmp(id, oid);
299 
300 		if (cmp == 0)
301 			return i;
302 		else if (cmp > 0)
303 			break;
304 		i++;
305 	}
306 
307 	return -1;
308 }
309 
310 const struct got_error *
311 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
312 {
313 	const struct got_error *err = NULL;
314 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
315 	struct got_packfile_hdr hdr;
316 	size_t n;
317 
318 	n = fread(&hdr, sizeof(hdr), 1, f);
319 	if (n != 1)
320 		return got_ferror(f, GOT_ERR_BAD_PACKIDX);
321 
322 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
323 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
324 	    betoh32(hdr.nobjects) != totobj)
325 		err = got_error(GOT_ERR_BAD_PACKFILE);
326 
327 	return err;
328 }
329 
330 static const struct got_error *
331 decode_type_and_size(uint8_t *type, uint64_t *size, size_t *len, FILE *packfile)
332 {
333 	uint8_t t = 0;
334 	uint64_t s = 0;
335 	uint8_t sizeN;
336 	size_t n;
337 	int i = 0;
338 
339 	do {
340 		/* We do not support size values which don't fit in 64 bit. */
341 		if (i > 9)
342 			return got_error(GOT_ERR_NO_SPACE);
343 
344 		n = fread(&sizeN, sizeof(sizeN), 1, packfile);
345 		if (n != 1)
346 			return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
347 
348 		if (i == 0) {
349 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
350 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
351 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
352 		} else {
353 			size_t shift = 4 + 7 * (i - 1);
354 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
355 		}
356 		i++;
357 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
358 
359 	*type = t;
360 	*size = s;
361 	*len = i * sizeof(sizeN);
362 	return NULL;
363 }
364 
365 static const struct got_error *
366 open_packed_plain_object(struct got_object **obj, const char *path_packfile,
367     struct got_object_id *id, uint8_t type, off_t offset, size_t size)
368 {
369 	*obj = calloc(1, sizeof(**obj));
370 	if (*obj == NULL)
371 		return got_error(GOT_ERR_NO_MEM);
372 
373 	(*obj)->path_packfile = strdup(path_packfile);
374 	if ((*obj)->path_packfile == NULL) {
375 		free(*obj);
376 		*obj = NULL;
377 		return got_error(GOT_ERR_NO_MEM);
378 	}
379 
380 	(*obj)->type = type;
381 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
382 	(*obj)->hdrlen = 0;
383 	(*obj)->size = size;
384 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
385 	(*obj)->pack_offset = offset;
386 
387 	return NULL;
388 }
389 
390 static const struct got_error *
391 open_packed_object(struct got_object **obj, struct got_repository *repo,
392     const char *path_packdir, struct got_packidx_v2_hdr *packidx,
393     struct got_object_id *id)
394 {
395 	const struct got_error *err = NULL;
396 	int idx = get_object_idx(packidx, id);
397 	off_t offset;
398 	char hex[SHA1_DIGEST_STRING_LENGTH];
399 	char *sha1str;
400 	char *path_packfile;
401 	FILE *packfile;
402 	uint8_t type;
403 	uint64_t size;
404 	size_t tslen;
405 
406 	*obj = NULL;
407 	if (idx == -1) /* object not found in pack index */
408 		return NULL;
409 
410 	offset = get_object_offset(packidx, idx);
411 	if (offset == (uint64_t)-1)
412 		return got_error(GOT_ERR_BAD_PACKIDX);
413 
414 	sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
415 	    hex, sizeof(hex));
416 	if (sha1str == NULL)
417 		return got_error(GOT_ERR_PACKIDX_CSUM);
418 
419 	if (asprintf(&path_packfile, "%s/%s%s%s", path_packdir,
420 	    GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1)
421 		return got_error(GOT_ERR_NO_MEM);
422 
423 	packfile = fopen(path_packfile, "rb");
424 	if (packfile == NULL) {
425 		err = got_error_from_errno();
426 		goto done;
427 	}
428 
429 	err = read_packfile_hdr(packfile, packidx);
430 	if (err)
431 		goto done;
432 
433 	if (fseeko(packfile, offset, SEEK_SET) != 0) {
434 		err = got_error_from_errno();
435 		goto done;
436 	}
437 
438 	err = decode_type_and_size(&type, &size, &tslen, packfile);
439 	if (err)
440 		goto done;
441 
442 	switch (type) {
443 	case GOT_OBJ_TYPE_COMMIT:
444 	case GOT_OBJ_TYPE_TREE:
445 	case GOT_OBJ_TYPE_BLOB:
446 		err = open_packed_plain_object(obj, path_packfile, id, type,
447 		    offset + tslen, size);
448 		break;
449 
450 	case GOT_OBJ_TYPE_REF_DELTA:
451 	case GOT_OBJ_TYPE_TAG:
452 	case GOT_OBJ_TYPE_OFFSET_DELTA:
453 	default:
454 		err = got_error(GOT_ERR_NOT_IMPL);
455 		goto done;
456 	}
457 done:
458 	free(path_packfile);
459 	if (err)
460 		free(*obj);
461 	if (packfile && fclose(packfile) == -1 && err == 0)
462 		err = got_error_from_errno();
463 	return err;
464 }
465 
466 const struct got_error *
467 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
468     struct got_repository *repo)
469 {
470 	const struct got_error *err = NULL;
471 	DIR *packdir = NULL;
472 	struct dirent *dent;
473 	char *path_packdir = got_repo_get_path_objects_pack(repo);
474 
475 	if (path_packdir == NULL) {
476 		err = got_error(GOT_ERR_NO_MEM);
477 		goto done;
478 	}
479 
480 	packdir = opendir(path_packdir);
481 	if (packdir == NULL) {
482 		err = got_error_from_errno();
483 		goto done;
484 	}
485 
486 	while ((dent = readdir(packdir)) != NULL) {
487 		struct got_packidx_v2_hdr *packidx;
488 		char *path_packidx, *path_object;
489 
490 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
491 			continue;
492 
493 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
494 		    dent->d_name) == -1) {
495 			err = got_error(GOT_ERR_NO_MEM);
496 			goto done;
497 		}
498 
499 		err = got_packidx_open(&packidx, path_packidx);
500 		free(path_packidx);
501 		if (err)
502 			goto done;
503 
504 		err = open_packed_object(obj, repo, path_packdir, packidx, id);
505 		got_packidx_close(packidx);
506 		if (err)
507 			goto done;
508 		if (*obj != NULL)
509 			break;
510 	}
511 
512 done:
513 	free(path_packdir);
514 	if (packdir && closedir(packdir) != 0 && err == 0)
515 		err = got_error_from_errno();
516 	return err;
517 }
518 
519 static const struct got_error *
520 dump_plain_object(FILE *infile, uint8_t type, size_t size, FILE *outfile)
521 {
522 	size_t n;
523 
524 	while (size > 0) {
525 		uint8_t data[2048];
526 		size_t len = MIN(size, sizeof(data));
527 
528 		n = fread(data, len, 1, infile);
529 		if (n != 1)
530 			return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
531 
532 		n = fwrite(data, len, 1, outfile);
533 		if (n != 1)
534 			return got_ferror(outfile, GOT_ERR_IO);
535 
536 		size -= len;
537 	}
538 
539 	rewind(outfile);
540 	return NULL;
541 }
542 
543 static const struct got_error *
544 dump_ref_delta_object(struct got_repository *repo, FILE *infile, uint8_t type,
545     size_t size, FILE *outfile)
546 {
547 	const struct got_error *err = NULL;
548 	struct got_object_id base_id;
549 	struct got_object *base_obj;
550 	int n;
551 
552 	if (size < sizeof(base_id))
553 		return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
554 
555 	n = fread(&base_id, sizeof(base_id), 1, infile);
556 	if (n != 1)
557 		return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
558 
559 	size -= sizeof(base_id);
560 	if (size <= 0)
561 		return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
562 
563 	err = got_object_open(&base_obj, repo, &base_id);
564 	if (err)
565 		return err;
566 
567 	err = got_delta_apply(repo, infile, size, base_obj, outfile);
568 	got_object_close(base_obj);
569 	return err;
570 }
571 
572 const struct got_error *
573 got_packfile_extract_object(FILE **f, struct got_object *obj,
574     struct got_repository *repo)
575 {
576 	const struct got_error *err = NULL;
577 	FILE *packfile = NULL;
578 
579 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
580 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
581 
582 	*f = got_opentemp();
583 	if (*f == NULL) {
584 		err = got_error(GOT_ERR_FILE_OPEN);
585 		goto done;
586 	}
587 
588 	packfile = fopen(obj->path_packfile, "rb");
589 	if (packfile == NULL) {
590 		err = got_error_from_errno();
591 		goto done;
592 	}
593 
594 	if (fseeko(packfile, obj->pack_offset, SEEK_SET) != 0) {
595 		err = got_error_from_errno();
596 		goto done;
597 	}
598 
599 	switch (obj->type) {
600 	case GOT_OBJ_TYPE_COMMIT:
601 	case GOT_OBJ_TYPE_TREE:
602 	case GOT_OBJ_TYPE_BLOB:
603 		err = dump_plain_object(packfile, obj->type, obj->size, *f);
604 		break;
605 	case GOT_OBJ_TYPE_REF_DELTA:
606 		err = dump_ref_delta_object(repo, packfile, obj->type,
607 		    obj->size, *f);
608 		break;
609 	case GOT_OBJ_TYPE_TAG:
610 	case GOT_OBJ_TYPE_OFFSET_DELTA:
611 	default:
612 		err = got_error(GOT_ERR_NOT_IMPL);
613 		goto done;
614 	}
615 done:
616 	if (packfile)
617 		fclose(packfile);
618 	if (err && *f)
619 		fclose(*f);
620 	return err;
621 }