?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
portable: configure: use config.h
[got-portable.git] / lib / pack.c
blobf0189ff480b44a73b0aa170458502a0c0deb3cd7
1 /*
2 * Copyright (c) 2018, 2019, 2020 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 #include <sys/uio.h>
21 #include <sys/mman.h>
22 #include <sys/resource.h>
23 #include <sys/socket.h>
24
25 #include <fcntl.h>
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdint.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <unistd.h>
33 #include <zlib.h>
34
35 #include "got_error.h"
36 #include "got_object.h"
37 #include "got_path.h"
38
39 #include "got_lib_hash.h"
40 #include "got_lib_delta.h"
41 #include "got_lib_delta_cache.h"
42 #include "got_lib_inflate.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_parse.h"
45 #include "got_lib_privsep.h"
46 #include "got_lib_pack.h"
47
48 #ifndef nitems
49 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
50 #endif
51
52 #ifndef MIN
53 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
54 #endif
55
56 static const struct got_error *
57 verify_fanout_table(uint32_t *fanout_table)
58 {
59 int i;
60
61 for (i = 0; i < 0xff - 1; i++) {
62 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
63 return got_error(GOT_ERR_BAD_PACKIDX);
64 }
65
66 return NULL;
67 }
68
69 const struct got_error *
70 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
71 {
72 const struct got_error *err = NULL;
73 enum got_hash_algorithm algo = GOT_HASH_SHA1;
74 struct got_packidx_v2_hdr *h;
75 struct got_hash ctx;
76 uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
77 size_t nobj, len_fanout, len_ids, offset, remain;
78 ssize_t n;
79 int i;
80
81 got_hash_init(&ctx, algo);
82
83 h = &p->hdr;
84 offset = 0;
85 remain = p->len;
86
87 if (remain < sizeof(*h->magic)) {
88 err = got_error(GOT_ERR_BAD_PACKIDX);
89 goto done;
90 }
91 if (p->map)
92 h->magic = (uint32_t *)(p->map + offset);
93 else {
94 h->magic = malloc(sizeof(*h->magic));
95 if (h->magic == NULL) {
96 err = got_error_from_errno("malloc");
97 goto done;
98 }
99 n = read(p->fd, h->magic, sizeof(*h->magic));
100 if (n < 0) {
101 err = got_error_from_errno("read");
102 goto done;
103 } else if (n != sizeof(*h->magic)) {
104 err = got_error(GOT_ERR_BAD_PACKIDX);
105 goto done;
106 }
107 }
108 if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
109 err = got_error(GOT_ERR_BAD_PACKIDX);
110 goto done;
111 }
112 offset += sizeof(*h->magic);
113 remain -= sizeof(*h->magic);
114
115 if (verify)
116 got_hash_update(&ctx, h->magic, sizeof(*h->magic));
117
118 if (remain < sizeof(*h->version)) {
119 err = got_error(GOT_ERR_BAD_PACKIDX);
120 goto done;
121 }
122 if (p->map)
123 h->version = (uint32_t *)(p->map + offset);
124 else {
125 h->version = malloc(sizeof(*h->version));
126 if (h->version == NULL) {
127 err = got_error_from_errno("malloc");
128 goto done;
129 }
130 n = read(p->fd, h->version, sizeof(*h->version));
131 if (n < 0) {
132 err = got_error_from_errno("read");
133 goto done;
134 } else if (n != sizeof(*h->version)) {
135 err = got_error(GOT_ERR_BAD_PACKIDX);
136 goto done;
137 }
138 }
139 if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
140 err = got_error(GOT_ERR_BAD_PACKIDX);
141 goto done;
142 }
143 offset += sizeof(*h->version);
144 remain -= sizeof(*h->version);
145
146 if (verify)
147 got_hash_update(&ctx, h->version, sizeof(*h->version));
148
149 len_fanout =
150 sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
151 if (remain < len_fanout) {
152 err = got_error(GOT_ERR_BAD_PACKIDX);
153 goto done;
154 }
155 if (p->map)
156 h->fanout_table = (uint32_t *)(p->map + offset);
157 else {
158 h->fanout_table = malloc(len_fanout);
159 if (h->fanout_table == NULL) {
160 err = got_error_from_errno("malloc");
161 goto done;
162 }
163 n = read(p->fd, h->fanout_table, len_fanout);
164 if (n < 0) {
165 err = got_error_from_errno("read");
166 goto done;
167 } else if (n != len_fanout) {
168 err = got_error(GOT_ERR_BAD_PACKIDX);
169 goto done;
170 }
171 }
172 err = verify_fanout_table(h->fanout_table);
173 if (err)
174 goto done;
175 if (verify)
176 got_hash_update(&ctx, h->fanout_table, len_fanout);
177 offset += len_fanout;
178 remain -= len_fanout;
179
180 nobj = be32toh(h->fanout_table[0xff]);
181 len_ids = nobj * sizeof(*h->sorted_ids);
182 if (len_ids <= nobj || len_ids > remain) {
183 err = got_error(GOT_ERR_BAD_PACKIDX);
184 goto done;
185 }
186 if (p->map)
187 h->sorted_ids =
188 (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
189 else {
190 h->sorted_ids = malloc(len_ids);
191 if (h->sorted_ids == NULL) {
192 err = got_error(GOT_ERR_BAD_PACKIDX);
193 goto done;
194 }
195 n = read(p->fd, h->sorted_ids, len_ids);
196 if (n < 0)
197 err = got_error_from_errno("read");
198 else if (n != len_ids) {
199 err = got_error(GOT_ERR_BAD_PACKIDX);
200 goto done;
201 }
202 }
203 if (verify)
204 got_hash_update(&ctx, h->sorted_ids, len_ids);
205 offset += len_ids;
206 remain -= len_ids;
207
208 if (remain < nobj * sizeof(*h->crc32)) {
209 err = got_error(GOT_ERR_BAD_PACKIDX);
210 goto done;
211 }
212 if (p->map)
213 h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
214 else {
215 h->crc32 = malloc(nobj * sizeof(*h->crc32));
216 if (h->crc32 == NULL) {
217 err = got_error_from_errno("malloc");
218 goto done;
219 }
220 n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
221 if (n < 0)
222 err = got_error_from_errno("read");
223 else if (n != nobj * sizeof(*h->crc32)) {
224 err = got_error(GOT_ERR_BAD_PACKIDX);
225 goto done;
226 }
227 }
228 if (verify)
229 got_hash_update(&ctx, h->crc32, nobj * sizeof(*h->crc32));
230 remain -= nobj * sizeof(*h->crc32);
231 offset += nobj * sizeof(*h->crc32);
232
233 if (remain < nobj * sizeof(*h->offsets)) {
234 err = got_error(GOT_ERR_BAD_PACKIDX);
235 goto done;
236 }
237 if (p->map)
238 h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
239 else {
240 h->offsets = malloc(nobj * sizeof(*h->offsets));
241 if (h->offsets == NULL) {
242 err = got_error_from_errno("malloc");
243 goto done;
244 }
245 n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
246 if (n < 0)
247 err = got_error_from_errno("read");
248 else if (n != nobj * sizeof(*h->offsets)) {
249 err = got_error(GOT_ERR_BAD_PACKIDX);
250 goto done;
251 }
252 }
253 if (verify)
254 got_hash_update(&ctx, h->offsets, nobj * sizeof(*h->offsets));
255 remain -= nobj * sizeof(*h->offsets);
256 offset += nobj * sizeof(*h->offsets);
257
258 /* Large file offsets are contained only in files > 2GB. */
259 if (verify || packfile_size > 0x7fffffff) {
260 for (i = 0; i < nobj; i++) {
261 uint32_t o = h->offsets[i];
262 if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
263 p->nlargeobj++;
264 }
265 }
266 if (p->nlargeobj == 0)
267 goto checksum;
268 else if (packfile_size <= 0x7fffffff) {
269 err = got_error(GOT_ERR_BAD_PACKIDX);
270 goto done;
271 }
272
273 if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
274 err = got_error(GOT_ERR_BAD_PACKIDX);
275 goto done;
276 }
277 if (p->map)
278 h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
279 else {
280 h->large_offsets = malloc(p->nlargeobj *
281 sizeof(*h->large_offsets));
282 if (h->large_offsets == NULL) {
283 err = got_error_from_errno("malloc");
284 goto done;
285 }
286 n = read(p->fd, h->large_offsets,
287 p->nlargeobj * sizeof(*h->large_offsets));
288 if (n < 0)
289 err = got_error_from_errno("read");
290 else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
291 err = got_error(GOT_ERR_BAD_PACKIDX);
292 goto done;
293 }
294 }
295 if (verify)
296 got_hash_update(&ctx, h->large_offsets,
297 p->nlargeobj * sizeof(*h->large_offsets));
298 remain -= p->nlargeobj * sizeof(*h->large_offsets);
299 offset += p->nlargeobj * sizeof(*h->large_offsets);
300
301 checksum:
302 if (remain < sizeof(*h->trailer)) {
303 err = got_error(GOT_ERR_BAD_PACKIDX);
304 goto done;
305 }
306 if (p->map)
307 h->trailer =
308 (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
309 else {
310 h->trailer = malloc(sizeof(*h->trailer));
311 if (h->trailer == NULL) {
312 err = got_error_from_errno("malloc");
313 goto done;
314 }
315 n = read(p->fd, h->trailer, sizeof(*h->trailer));
316 if (n < 0)
317 err = got_error_from_errno("read");
318 else if (n != sizeof(*h->trailer)) {
319 err = got_error(GOT_ERR_BAD_PACKIDX);
320 goto done;
321 }
322 }
323 if (verify) {
324 got_hash_update(&ctx, h->trailer->packfile_sha1,
325 SHA1_DIGEST_LENGTH);
326 got_hash_final(&ctx, hash);
327 if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1)
328 != 0)
329 err = got_error(GOT_ERR_PACKIDX_CSUM);
330 }
331 done:
332 return err;
333 }
334
335 const struct got_error *
336 got_packidx_open(struct got_packidx **packidx,
337 int dir_fd, const char *relpath, int verify)
338 {
339 const struct got_error *err = NULL;
340 struct got_packidx *p = NULL;
341 char *pack_relpath;
342 struct stat idx_sb, pack_sb;
343
344 *packidx = NULL;
345
346 err = got_packidx_get_packfile_path(&pack_relpath, relpath);
347 if (err)
348 return err;
349
350 /*
351 * Ensure that a corresponding pack file exists.
352 * Some Git repositories have this problem. Git seems to ignore
353 * the existence of lonely pack index files but we do not.
354 */
355 if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
356 if (errno == ENOENT) {
357 err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
358 "%s", relpath);
359 } else
360 err = got_error_from_errno2("fstatat", pack_relpath);
361 goto done;
362 }
363
364 p = calloc(1, sizeof(*p));
365 if (p == NULL) {
366 err = got_error_from_errno("calloc");
367 goto done;
368 }
369
370 p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
371 if (p->fd == -1) {
372 err = got_error_from_errno2("openat", relpath);
373 goto done;
374 }
375
376 if (fstat(p->fd, &idx_sb) != 0) {
377 err = got_error_from_errno2("fstat", relpath);
378 goto done;
379 }
380 p->len = idx_sb.st_size;
381 if (p->len < sizeof(p->hdr)) {
382 err = got_error(GOT_ERR_BAD_PACKIDX);
383 goto done;
384 }
385
386 p->path_packidx = strdup(relpath);
387 if (p->path_packidx == NULL) {
388 err = got_error_from_errno("strdup");
389 goto done;
390 }
391
392 #ifndef GOT_PACK_NO_MMAP
393 if (p->len > 0 && p->len <= SIZE_MAX) {
394 p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
395 if (p->map == MAP_FAILED) {
396 if (errno != ENOMEM) {
397 err = got_error_from_errno("mmap");
398 goto done;
399 }
400 p->map = NULL; /* fall back to read(2) */
401 }
402 }
403 #endif
404
405 err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
406 done:
407 if (err) {
408 if (p)
409 got_packidx_close(p);
410 } else
411 *packidx = p;
412 free(pack_relpath);
413 return err;
414 }
415
416 const struct got_error *
417 got_packidx_close(struct got_packidx *packidx)
418 {
419 const struct got_error *err = NULL;
420
421 free(packidx->path_packidx);
422 if (packidx->map) {
423 if (munmap(packidx->map, packidx->len) == -1)
424 err = got_error_from_errno("munmap");
425 } else {
426 free(packidx->hdr.magic);
427 free(packidx->hdr.version);
428 free(packidx->hdr.fanout_table);
429 free(packidx->hdr.sorted_ids);
430 free(packidx->hdr.crc32);
431 free(packidx->hdr.offsets);
432 free(packidx->hdr.large_offsets);
433 free(packidx->hdr.trailer);
434 }
435 if (close(packidx->fd) == -1 && err == NULL)
436 err = got_error_from_errno("close");
437 free(packidx->sorted_offsets);
438 free(packidx->sorted_large_offsets);
439 free(packidx);
440
441 return err;
442 }
443
444 const struct got_error *
445 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
446 {
447 size_t size;
448
449 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
450 size = strlen(path_packidx) + 2;
451 if (size < GOT_PACKFILE_NAMELEN + 1)
452 return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
453
454 *path_packfile = malloc(size);
455 if (*path_packfile == NULL)
456 return got_error_from_errno("malloc");
457
458 /* Copy up to and excluding ".idx". */
459 if (strlcpy(*path_packfile, path_packidx,
460 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
461 return got_error(GOT_ERR_NO_SPACE);
462
463 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
464 return got_error(GOT_ERR_NO_SPACE);
465
466 return NULL;
467 }
468
469 off_t
470 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
471 {
472 uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
473 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
474 uint64_t loffset;
475 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
476 if (idx < 0 || idx >= packidx->nlargeobj ||
477 packidx->hdr.large_offsets == NULL)
478 return -1;
479 loffset = be64toh(packidx->hdr.large_offsets[idx]);
480 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
481 }
482 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
483 }
484
485 int
486 got_packidx_get_object_idx(struct got_packidx *packidx,
487 struct got_object_id *id)
488 {
489 u_int8_t id0 = id->sha1[0];
490 uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
491 int left = 0, right = totobj - 1;
492
493 if (id0 > 0)
494 left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
495
496 while (left <= right) {
497 struct got_packidx_object_id *oid;
498 int i, cmp;
499
500 i = ((left + right) / 2);
501 oid = &packidx->hdr.sorted_ids[i];
502 cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
503 if (cmp == 0)
504 return i;
505 else if (cmp > 0)
506 left = i + 1;
507 else if (cmp < 0)
508 right = i - 1;
509 }
510
511 return -1;
512 }
513
514 static int
515 offset_cmp(const void *pa, const void *pb)
516 {
517 const struct got_pack_offset_index *a, *b;
518
519 a = (const struct got_pack_offset_index *)pa;
520 b = (const struct got_pack_offset_index *)pb;
521
522 if (a->offset < b->offset)
523 return -1;
524 else if (a->offset > b->offset)
525 return 1;
526
527 return 0;
528 }
529
530 static int
531 large_offset_cmp(const void *pa, const void *pb)
532 {
533 const struct got_pack_large_offset_index *a, *b;
534
535 a = (const struct got_pack_large_offset_index *)pa;
536 b = (const struct got_pack_large_offset_index *)pb;
537
538 if (a->offset < b->offset)
539 return -1;
540 else if (a->offset > b->offset)
541 return 1;
542
543 return 0;
544 }
545
546 static const struct got_error *
547 build_offset_index(struct got_packidx *p)
548 {
549 uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
550 unsigned int i, j, k;
551
552 p->sorted_offsets = calloc(nobj - p->nlargeobj,
553 sizeof(p->sorted_offsets[0]));
554 if (p->sorted_offsets == NULL)
555 return got_error_from_errno("calloc");
556
557 if (p->nlargeobj > 0) {
558 p->sorted_large_offsets = calloc(p->nlargeobj,
559 sizeof(p->sorted_large_offsets[0]));
560 if (p->sorted_large_offsets == NULL)
561 return got_error_from_errno("calloc");
562 }
563
564 j = 0;
565 k = 0;
566 for (i = 0; i < nobj; i++) {
567 uint32_t offset = be32toh(p->hdr.offsets[i]);
568 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
569 uint64_t loffset;
570 uint32_t idx;
571 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
572 if (idx >= p->nlargeobj ||
573 p->nlargeobj == 0 ||
574 p->hdr.large_offsets == NULL)
575 return got_error(GOT_ERR_BAD_PACKIDX);
576 loffset = be64toh(p->hdr.large_offsets[idx]);
577 p->sorted_large_offsets[j].offset = loffset;
578 p->sorted_large_offsets[j].idx = i;
579 j++;
580 } else {
581 p->sorted_offsets[k].offset = offset;
582 p->sorted_offsets[k].idx = i;
583 k++;
584 }
585 }
586 if (j != p->nlargeobj || k != nobj - p->nlargeobj)
587 return got_error(GOT_ERR_BAD_PACKIDX);
588
589 qsort(p->sorted_offsets, nobj - p->nlargeobj,
590 sizeof(p->sorted_offsets[0]), offset_cmp);
591
592 if (p->sorted_large_offsets)
593 qsort(p->sorted_large_offsets, p->nlargeobj,
594 sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
595
596 return NULL;
597 }
598
599 const struct got_error *
600 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
601 {
602 const struct got_error *err;
603 uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
604 int i, left, right;
605
606 *idx = -1;
607
608 if (packidx->sorted_offsets == NULL) {
609 err = build_offset_index(packidx);
610 if (err)
611 return err;
612 }
613
614 if (offset >= 0x7fffffff) {
615 uint64_t lo;
616 left = 0, right = packidx->nlargeobj - 1;
617 while (left <= right) {
618 i = ((left + right) / 2);
619 lo = packidx->sorted_large_offsets[i].offset;
620 if (lo == offset) {
621 *idx = packidx->sorted_large_offsets[i].idx;
622 break;
623 } else if (offset > lo)
624 left = i + 1;
625 else if (offset < lo)
626 right = i - 1;
627 }
628 } else {
629 uint32_t o;
630 left = 0, right = totobj - packidx->nlargeobj - 1;
631 while (left <= right) {
632 i = ((left + right) / 2);
633 o = packidx->sorted_offsets[i].offset;
634 if (o == offset) {
635 *idx = packidx->sorted_offsets[i].idx;
636 break;
637 } else if (offset > o)
638 left = i + 1;
639 else if (offset < o)
640 right = i - 1;
641 }
642 }
643
644 return NULL;
645 }
646
647 const struct got_error *
648 got_packidx_get_object_id(struct got_object_id *id,
649 struct got_packidx *packidx, int idx)
650 {
651 uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
652 struct got_packidx_object_id *oid;
653
654 if (idx < 0 || idx >= totobj)
655 return got_error(GOT_ERR_NO_OBJ);
656
657 oid = &packidx->hdr.sorted_ids[idx];
658 memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
659 return NULL;
660 }
661
662 const struct got_error *
663 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
664 struct got_packidx *packidx, const char *id_str_prefix)
665 {
666 const struct got_error *err = NULL;
667 u_int8_t id0;
668 uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
669 char hex[3];
670 size_t prefix_len = strlen(id_str_prefix);
671 struct got_packidx_object_id *oid;
672 uint32_t i = 0;
673
674 STAILQ_INIT(matched_ids);
675
676 if (prefix_len < 2)
677 return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
678
679 hex[0] = id_str_prefix[0];
680 hex[1] = id_str_prefix[1];
681 hex[2] = '\0';
682 if (!got_parse_xdigit(&id0, hex))
683 return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
684
685 if (id0 > 0)
686 i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
687 oid = &packidx->hdr.sorted_ids[i];
688 while (i < totobj && oid->sha1[0] == id0) {
689 char id_str[SHA1_DIGEST_STRING_LENGTH];
690 struct got_object_qid *qid;
691 int cmp;
692
693 if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
694 return got_error(GOT_ERR_NO_SPACE);
695
696 cmp = strncmp(id_str, id_str_prefix, prefix_len);
697 if (cmp < 0) {
698 oid = &packidx->hdr.sorted_ids[++i];
699 continue;
700 } else if (cmp > 0)
701 break;
702
703 err = got_object_qid_alloc_partial(&qid);
704 if (err)
705 break;
706 memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
707 STAILQ_INSERT_TAIL(matched_ids, qid, entry);
708
709 oid = &packidx->hdr.sorted_ids[++i];
710 }
711
712 if (err)
713 got_object_id_queue_free(matched_ids);
714 return err;
715 }
716
717 static void
718 set_max_datasize(void)
719 {
720 struct rlimit rl;
721
722 if (getrlimit(RLIMIT_DATA, &rl) != 0)
723 return;
724
725 rl.rlim_cur = rl.rlim_max;
726 setrlimit(RLIMIT_DATA, &rl);
727 }
728
729 const struct got_error *
730 got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
731 {
732 const struct got_error *err = NULL;
733 int imsg_fds[2];
734 pid_t pid;
735 struct imsgbuf *ibuf;
736
737 ibuf = calloc(1, sizeof(*ibuf));
738 if (ibuf == NULL)
739 return got_error_from_errno("calloc");
740
741 pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
742 if (pack->privsep_child == NULL) {
743 err = got_error_from_errno("calloc");
744 free(ibuf);
745 return err;
746 }
747 pack->child_has_tempfiles = 0;
748 pack->child_has_delta_outfd = 0;
749
750 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
751 err = got_error_from_errno("socketpair");
752 goto done;
753 }
754
755 pid = fork();
756 if (pid == -1) {
757 err = got_error_from_errno("fork");
758 goto done;
759 } else if (pid == 0) {
760 set_max_datasize();
761 got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
762 pack->path_packfile);
763 /* not reached */
764 }
765
766 if (close(imsg_fds[1]) == -1)
767 return got_error_from_errno("close");
768 pack->privsep_child->imsg_fd = imsg_fds[0];
769 pack->privsep_child->pid = pid;
770 imsg_init(ibuf, imsg_fds[0]);
771 pack->privsep_child->ibuf = ibuf;
772
773 err = got_privsep_init_pack_child(ibuf, pack, packidx);
774 if (err) {
775 const struct got_error *child_err;
776 err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
777 child_err = got_privsep_wait_for_child(
778 pack->privsep_child->pid);
779 if (child_err && err == NULL)
780 err = child_err;
781 }
782 done:
783 if (err) {
784 free(ibuf);
785 free(pack->privsep_child);
786 pack->privsep_child = NULL;
787 }
788 return err;
789 }
790
791 static const struct got_error *
792 pack_stop_privsep_child(struct got_pack *pack)
793 {
794 const struct got_error *err = NULL, *close_err = NULL;
795
796 if (pack->privsep_child == NULL)
797 return NULL;
798
799 err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
800 if (err)
801 return err;
802 if (close(pack->privsep_child->imsg_fd) == -1)
803 close_err = got_error_from_errno("close");
804 err = got_privsep_wait_for_child(pack->privsep_child->pid);
805 if (close_err && err == NULL)
806 err = close_err;
807 imsg_clear(pack->privsep_child->ibuf);
808 free(pack->privsep_child->ibuf);
809 free(pack->privsep_child);
810 pack->privsep_child = NULL;
811 return err;
812 }
813
814 const struct got_error *
815 got_pack_close(struct got_pack *pack)
816 {
817 const struct got_error *err = NULL;
818
819 err = pack_stop_privsep_child(pack);
820 if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
821 err = got_error_from_errno("munmap");
822 if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
823 err = got_error_from_errno("close");
824 pack->fd = -1;
825 free(pack->path_packfile);
826 pack->path_packfile = NULL;
827 pack->filesize = 0;
828 if (pack->delta_cache) {
829 got_delta_cache_free(pack->delta_cache);
830 pack->delta_cache = NULL;
831 }
832
833 /*
834 * Leave accumfd and basefd alone. They are managed by the
835 * repository layer and can be reused.
836 */
837
838 return err;
839 }
840
841 const struct got_error *
842 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
843 struct got_pack *pack, off_t offset)
844 {
845 uint8_t t = 0;
846 uint64_t s = 0;
847 uint8_t sizeN;
848 size_t mapoff = 0;
849 int i = 0;
850
851 *len = 0;
852
853 if (offset >= pack->filesize)
854 return got_error(GOT_ERR_PACK_OFFSET);
855
856 if (pack->map) {
857 if (offset > SIZE_MAX) {
858 return got_error_fmt(GOT_ERR_PACK_OFFSET,
859 "offset %lld overflows size_t",
860 (long long)offset);
861 }
862
863 mapoff = (size_t)offset;
864 } else {
865 if (lseek(pack->fd, offset, SEEK_SET) == -1)
866 return got_error_from_errno("lseek");
867 }
868
869 do {
870 /* We do not support size values which don't fit in 64 bit. */
871 if (i > 9)
872 return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
873 "packfile offset %lld", (long long)offset);
874
875 if (pack->map) {
876 if (mapoff + sizeof(sizeN) >= pack->filesize)
877 return got_error(GOT_ERR_BAD_PACKFILE);
878 sizeN = *(pack->map + mapoff);
879 mapoff += sizeof(sizeN);
880 } else {
881 ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
882 if (n < 0)
883 return got_error_from_errno("read");
884 if (n != sizeof(sizeN))
885 return got_error(GOT_ERR_BAD_PACKFILE);
886 }
887 *len += sizeof(sizeN);
888
889 if (i == 0) {
890 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
891 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
892 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
893 } else {
894 size_t shift = 4 + 7 * (i - 1);
895 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
896 }
897 i++;
898 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
899
900 *type = t;
901 *size = s;
902 return NULL;
903 }
904
905 static const struct got_error *
906 open_plain_object(struct got_object **obj, struct got_object_id *id,
907 uint8_t type, off_t offset, size_t size, int idx)
908 {
909 *obj = calloc(1, sizeof(**obj));
910 if (*obj == NULL)
911 return got_error_from_errno("calloc");
912
913 (*obj)->type = type;
914 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
915 (*obj)->pack_idx = idx;
916 (*obj)->hdrlen = 0;
917 (*obj)->size = size;
918 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
919 (*obj)->pack_offset = offset;
920
921 return NULL;
922 }
923
924 static const struct got_error *
925 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
926 off_t delta_offset)
927 {
928 int64_t o = 0;
929 uint8_t offN;
930 int i = 0;
931
932 *offset = 0;
933 *len = 0;
934
935 do {
936 /* We do not support offset values which don't fit in 64 bit. */
937 if (i > 8)
938 return got_error(GOT_ERR_NO_SPACE);
939
940 if (pack->map) {
941 size_t mapoff;
942
943 if (delta_offset + *len > SIZE_MAX) {
944 return got_error_fmt(GOT_ERR_PACK_OFFSET,
945 "mapoff %lld would overflow size_t",
946 (long long)delta_offset + *len);
947 }
948
949 mapoff = (size_t)delta_offset + *len;
950 if (mapoff + sizeof(offN) >= pack->filesize)
951 return got_error(GOT_ERR_PACK_OFFSET);
952 offN = *(pack->map + mapoff);
953 } else {
954 ssize_t n;
955 n = read(pack->fd, &offN, sizeof(offN));
956 if (n < 0)
957 return got_error_from_errno("read");
958 if (n != sizeof(offN))
959 return got_error(GOT_ERR_BAD_PACKFILE);
960 }
961 *len += sizeof(offN);
962
963 if (i == 0)
964 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
965 else {
966 o++;
967 o <<= 7;
968 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
969 }
970 i++;
971 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
972
973 *offset = o;
974 return NULL;
975 }
976
977 const struct got_error *
978 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
979 struct got_pack *pack, off_t offset, size_t tslen)
980 {
981 const struct got_error *err;
982 int64_t negoffset;
983 size_t negofflen;
984
985 *len = 0;
986
987 err = parse_negative_offset(&negoffset, &negofflen, pack,
988 offset + tslen);
989 if (err)
990 return err;
991
992 /* Compute the base object's offset (must be in the same pack file). */
993 *base_offset = (offset - negoffset);
994 if (*base_offset <= 0)
995 return got_error(GOT_ERR_BAD_PACKFILE);
996
997 *len = negofflen;
998 return NULL;
999 }
1000
1001 static const struct got_error *
1002 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1003 size_t *delta_compressed_len, size_t delta_data_offset,
1004 struct got_pack *pack)
1005 {
1006 const struct got_error *err = NULL;
1007 size_t consumed = 0;
1008
1009 if (pack->map) {
1010 if (delta_data_offset >= pack->filesize)
1011 return got_error(GOT_ERR_PACK_OFFSET);
1012 err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1013 &consumed, NULL, pack->map, delta_data_offset,
1014 pack->filesize - delta_data_offset);
1015 if (err)
1016 return err;
1017 } else {
1018 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1019 return got_error_from_errno("lseek");
1020 err = got_inflate_to_mem_fd(delta_buf, delta_len,
1021 &consumed, NULL, 0, pack->fd);
1022 if (err)
1023 return err;
1024 }
1025
1026 if (delta_compressed_len)
1027 *delta_compressed_len = consumed;
1028
1029 return NULL;
1030 }
1031
1032 static const struct got_error *
1033 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1034 int delta_type, size_t delta_size, off_t delta_data_offset)
1035 {
1036 struct got_delta *delta;
1037
1038 delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1039 delta_data_offset);
1040 if (delta == NULL)
1041 return got_error_from_errno("got_delta_open");
1042 /* delta is freed in got_object_close() */
1043 deltas->nentries++;
1044 STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1045 return NULL;
1046 }
1047
1048 static const struct got_error *
1049 resolve_offset_delta(struct got_delta_chain *deltas,
1050 struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1051 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1052 {
1053 const struct got_error *err;
1054 off_t base_offset;
1055 uint8_t base_type;
1056 uint64_t base_size;
1057 size_t base_tslen;
1058 off_t delta_data_offset;
1059 size_t consumed;
1060
1061 err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1062 delta_offset, tslen);
1063 if (err)
1064 return err;
1065
1066 delta_data_offset = delta_offset + tslen + consumed;
1067 if (delta_data_offset >= pack->filesize)
1068 return got_error(GOT_ERR_PACK_OFFSET);
1069
1070 if (pack->map == NULL) {
1071 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1072 if (delta_data_offset == -1)
1073 return got_error_from_errno("lseek");
1074 }
1075
1076 err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1077 delta_data_offset);
1078 if (err)
1079 return err;
1080
1081 /* An offset delta must be in the same packfile. */
1082 if (base_offset >= pack->filesize)
1083 return got_error(GOT_ERR_PACK_OFFSET);
1084
1085 err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1086 &base_tslen, pack, base_offset);
1087 if (err)
1088 return err;
1089
1090 return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1091 base_tslen, base_type, base_size, recursion - 1);
1092 }
1093
1094 const struct got_error *
1095 got_pack_parse_ref_delta(struct got_object_id *id,
1096 struct got_pack *pack, off_t delta_offset, int tslen)
1097 {
1098 if (pack->map) {
1099 size_t mapoff;
1100
1101 if (delta_offset + tslen > SIZE_MAX) {
1102 return got_error_fmt(GOT_ERR_PACK_OFFSET,
1103 "mapoff %lld would overflow size_t",
1104 (long long)delta_offset + tslen);
1105 }
1106
1107 mapoff = delta_offset + tslen;
1108 if (mapoff + sizeof(*id) >= pack->filesize)
1109 return got_error(GOT_ERR_PACK_OFFSET);
1110 memcpy(id, pack->map + mapoff, sizeof(*id));
1111 } else {
1112 ssize_t n;
1113 n = read(pack->fd, id, sizeof(*id));
1114 if (n < 0)
1115 return got_error_from_errno("read");
1116 if (n != sizeof(*id))
1117 return got_error(GOT_ERR_BAD_PACKFILE);
1118 }
1119
1120 return NULL;
1121 }
1122
1123 static const struct got_error *
1124 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1125 struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1126 size_t delta_size, unsigned int recursion)
1127 {
1128 const struct got_error *err;
1129 struct got_object_id id;
1130 int idx;
1131 off_t base_offset;
1132 uint8_t base_type;
1133 uint64_t base_size;
1134 size_t base_tslen;
1135 off_t delta_data_offset;
1136
1137 if (delta_offset + tslen >= pack->filesize)
1138 return got_error(GOT_ERR_PACK_OFFSET);
1139
1140 err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1141 if (err)
1142 return err;
1143 if (pack->map) {
1144 delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1145 } else {
1146 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1147 if (delta_data_offset == -1)
1148 return got_error_from_errno("lseek");
1149 }
1150
1151 err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1152 delta_data_offset);
1153 if (err)
1154 return err;
1155
1156 /* Delta base must be in the same pack file. */
1157 idx = got_packidx_get_object_idx(packidx, &id);
1158 if (idx == -1)
1159 return got_error(GOT_ERR_NO_OBJ);
1160
1161 base_offset = got_packidx_get_object_offset(packidx, idx);
1162 if (base_offset == -1)
1163 return got_error(GOT_ERR_BAD_PACKIDX);
1164
1165 if (base_offset >= pack->filesize)
1166 return got_error(GOT_ERR_PACK_OFFSET);
1167
1168 err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1169 &base_tslen, pack, base_offset);
1170 if (err)
1171 return err;
1172
1173 return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1174 base_tslen, base_type, base_size, recursion - 1);
1175 }
1176
1177 const struct got_error *
1178 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1179 struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1180 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1181 {
1182 const struct got_error *err = NULL;
1183
1184 if (--recursion == 0)
1185 return got_error(GOT_ERR_RECURSION);
1186
1187 switch (delta_type) {
1188 case GOT_OBJ_TYPE_COMMIT:
1189 case GOT_OBJ_TYPE_TREE:
1190 case GOT_OBJ_TYPE_BLOB:
1191 case GOT_OBJ_TYPE_TAG:
1192 /* Plain types are the final delta base. Recursion ends. */
1193 err = add_delta(deltas, delta_offset, tslen, delta_type,
1194 delta_size, 0);
1195 break;
1196 case GOT_OBJ_TYPE_OFFSET_DELTA:
1197 err = resolve_offset_delta(deltas, packidx, pack,
1198 delta_offset, tslen, delta_type, delta_size, recursion - 1);
1199 break;
1200 case GOT_OBJ_TYPE_REF_DELTA:
1201 err = resolve_ref_delta(deltas, packidx, pack,
1202 delta_offset, tslen, delta_type, delta_size, recursion - 1);
1203 break;
1204 default:
1205 return got_error(GOT_ERR_OBJ_TYPE);
1206 }
1207
1208 return err;
1209 }
1210
1211 static const struct got_error *
1212 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1213 struct got_pack *pack, struct got_object_id *id, off_t offset,
1214 size_t tslen, int delta_type, size_t delta_size, int idx)
1215 {
1216 const struct got_error *err = NULL;
1217 int resolved_type;
1218
1219 *obj = calloc(1, sizeof(**obj));
1220 if (*obj == NULL)
1221 return got_error_from_errno("calloc");
1222
1223 (*obj)->flags = 0;
1224 (*obj)->hdrlen = 0;
1225 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1226 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1227 (*obj)->pack_offset = offset + tslen;
1228
1229 STAILQ_INIT(&(*obj)->deltas.entries);
1230 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1231 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1232 (*obj)->pack_idx = idx;
1233
1234 err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1235 offset, tslen, delta_type, delta_size,
1236 GOT_DELTA_CHAIN_RECURSION_MAX);
1237 if (err)
1238 goto done;
1239
1240 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1241 if (err)
1242 goto done;
1243 (*obj)->type = resolved_type;
1244 done:
1245 if (err) {
1246 got_object_close(*obj);
1247 *obj = NULL;
1248 }
1249 return err;
1250 }
1251
1252 const struct got_error *
1253 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1254 struct got_packidx *packidx, int idx, struct got_object_id *id)
1255 {
1256 const struct got_error *err = NULL;
1257 off_t offset;
1258 uint8_t type;
1259 uint64_t size;
1260 size_t tslen;
1261
1262 *obj = NULL;
1263
1264 offset = got_packidx_get_object_offset(packidx, idx);
1265 if (offset == -1)
1266 return got_error(GOT_ERR_BAD_PACKIDX);
1267
1268 err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1269 pack, offset);
1270 if (err)
1271 return err;
1272
1273 switch (type) {
1274 case GOT_OBJ_TYPE_COMMIT:
1275 case GOT_OBJ_TYPE_TREE:
1276 case GOT_OBJ_TYPE_BLOB:
1277 case GOT_OBJ_TYPE_TAG:
1278 err = open_plain_object(obj, id, type, offset + tslen,
1279 size, idx);
1280 break;
1281 case GOT_OBJ_TYPE_OFFSET_DELTA:
1282 case GOT_OBJ_TYPE_REF_DELTA:
1283 err = open_delta_object(obj, packidx, pack, id, offset,
1284 tslen, type, size, idx);
1285 break;
1286 default:
1287 err = got_error(GOT_ERR_OBJ_TYPE);
1288 break;
1289 }
1290
1291 return err;
1292 }
1293
1294 const struct got_error *
1295 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1296 struct got_delta_chain *deltas, struct got_pack *pack)
1297 {
1298 struct got_delta *delta;
1299 uint64_t base_size = 0, result_size = 0;
1300
1301 *max_size = 0;
1302 STAILQ_FOREACH(delta, &deltas->entries, entry) {
1303 /* Plain object types are the delta base. */
1304 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1305 delta->type != GOT_OBJ_TYPE_TREE &&
1306 delta->type != GOT_OBJ_TYPE_BLOB &&
1307 delta->type != GOT_OBJ_TYPE_TAG) {
1308 const struct got_error *err;
1309 uint8_t *delta_buf = NULL;
1310 size_t delta_len;
1311 int cached = 1;
1312
1313 if (pack->delta_cache) {
1314 got_delta_cache_get(&delta_buf, &delta_len,
1315 pack->delta_cache, delta->data_offset);
1316 }
1317 if (delta_buf == NULL) {
1318 cached = 0;
1319 err = read_delta_data(&delta_buf, &delta_len,
1320 NULL, delta->data_offset, pack);
1321 if (err)
1322 return err;
1323 }
1324 if (pack->delta_cache && !cached) {
1325 err = got_delta_cache_add(pack->delta_cache,
1326 delta->data_offset, delta_buf, delta_len);
1327 if (err == NULL)
1328 cached = 1;
1329 else if (err->code != GOT_ERR_NO_SPACE) {
1330 free(delta_buf);
1331 return err;
1332 }
1333 }
1334 err = got_delta_get_sizes(&base_size, &result_size,
1335 delta_buf, delta_len);
1336 if (!cached)
1337 free(delta_buf);
1338 if (err)
1339 return err;
1340 } else
1341 base_size = delta->size;
1342 if (base_size > *max_size)
1343 *max_size = base_size;
1344 if (result_size > *max_size)
1345 *max_size = result_size;
1346 }
1347
1348 return NULL;
1349 }
1350
1351 const struct got_error *
1352 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1353 struct got_pack *pack)
1354 {
1355 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1356 return got_error(GOT_ERR_OBJ_TYPE);
1357
1358 return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1359 }
1360
1361 const struct got_error *
1362 got_pack_dump_delta_chain_to_file(size_t *result_size,
1363 struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1364 FILE *base_file, FILE *accum_file)
1365 {
1366 const struct got_error *err = NULL;
1367 struct got_delta *delta;
1368 uint8_t *base_buf = NULL, *accum_buf = NULL;
1369 size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1370 /* We process small enough files entirely in memory for speed. */
1371 const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1372 uint64_t max_size = 0;
1373 int n = 0;
1374
1375 *result_size = 0;
1376
1377 if (STAILQ_EMPTY(&deltas->entries))
1378 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1379
1380 if (fseeko(base_file, 0L, SEEK_SET) == -1)
1381 return got_error_from_errno("fseeko");
1382 if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1383 return got_error_from_errno("fseeko");
1384
1385 /* Deltas are ordered in ascending order. */
1386 STAILQ_FOREACH(delta, &deltas->entries, entry) {
1387 uint8_t *delta_buf = NULL;
1388 uint64_t base_size, result_size = 0;
1389 int cached = 1;
1390 if (n == 0) {
1391 size_t mapoff;
1392 off_t delta_data_offset;
1393
1394 /* Plain object types are the delta base. */
1395 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1396 delta->type != GOT_OBJ_TYPE_TREE &&
1397 delta->type != GOT_OBJ_TYPE_BLOB &&
1398 delta->type != GOT_OBJ_TYPE_TAG) {
1399 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1400 goto done;
1401 }
1402
1403 delta_data_offset = delta->offset + delta->tslen;
1404 if (delta_data_offset >= pack->filesize) {
1405 err = got_error(GOT_ERR_PACK_OFFSET);
1406 goto done;
1407 }
1408 if (pack->map == NULL) {
1409 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1410 == -1) {
1411 err = got_error_from_errno("lseek");
1412 goto done;
1413 }
1414 }
1415 if (delta->size > max_size)
1416 max_size = delta->size;
1417 if (max_size > max_bufsize) {
1418 if (pack->map) {
1419 if (delta_data_offset > SIZE_MAX) {
1420 return got_error_fmt(
1421 GOT_ERR_RANGE,
1422 "delta offset %lld "
1423 "overflows size_t",
1424 (long long)
1425 delta_data_offset);
1426 }
1427
1428 mapoff = delta_data_offset;
1429 err = got_inflate_to_file_mmap(
1430 &base_bufsz, NULL, NULL, pack->map,
1431 mapoff, pack->filesize - mapoff,
1432 base_file);
1433 } else
1434 err = got_inflate_to_file_fd(
1435 &base_bufsz, NULL, NULL, pack->fd,
1436 base_file);
1437 } else {
1438 accum_buf = malloc(max_size);
1439 if (accum_buf == NULL) {
1440 err = got_error_from_errno("malloc");
1441 goto done;
1442 }
1443 accum_bufsz = max_size;
1444 if (pack->map) {
1445 if (delta_data_offset > SIZE_MAX) {
1446 return got_error_fmt(
1447 GOT_ERR_RANGE,
1448 "delta offset %lld "
1449 "overflows size_t",
1450 (long long)
1451 delta_data_offset);
1452 }
1453
1454 mapoff = delta_data_offset;
1455 err = got_inflate_to_mem_mmap(&base_buf,
1456 &base_bufsz, NULL, NULL,
1457 pack->map, mapoff,
1458 pack->filesize - mapoff);
1459 } else
1460 err = got_inflate_to_mem_fd(&base_buf,
1461 &base_bufsz, NULL, NULL, max_size,
1462 pack->fd);
1463 }
1464 if (err)
1465 goto done;
1466 n++;
1467 if (base_buf == NULL)
1468 rewind(base_file);
1469 continue;
1470 }
1471
1472 if (pack->delta_cache) {
1473 got_delta_cache_get(&delta_buf, &delta_len,
1474 pack->delta_cache, delta->data_offset);
1475 }
1476 if (delta_buf == NULL) {
1477 cached = 0;
1478 err = read_delta_data(&delta_buf, &delta_len, NULL,
1479 delta->data_offset, pack);
1480 if (err)
1481 goto done;
1482 }
1483 if (pack->delta_cache && !cached) {
1484 err = got_delta_cache_add(pack->delta_cache,
1485 delta->data_offset, delta_buf, delta_len);
1486 if (err == NULL)
1487 cached = 1;
1488 else if (err->code != GOT_ERR_NO_SPACE) {
1489 free(delta_buf);
1490 goto done;
1491 }
1492 }
1493
1494 err = got_delta_get_sizes(&base_size, &result_size,
1495 delta_buf, delta_len);
1496 if (err) {
1497 if (!cached)
1498 free(delta_buf);
1499 goto done;
1500 }
1501 if (base_size > max_size)
1502 max_size = base_size;
1503 if (result_size > max_size)
1504 max_size = result_size;
1505
1506 if (base_buf && max_size > max_bufsize) {
1507 /* Switch from buffers to temporary files. */
1508 size_t w = fwrite(base_buf, 1, base_bufsz,
1509 base_file);
1510 if (w != base_bufsz) {
1511 err = got_ferror(outfile, GOT_ERR_IO);
1512 if (!cached)
1513 free(delta_buf);
1514 goto done;
1515 }
1516 free(base_buf);
1517 base_buf = NULL;
1518 free(accum_buf);
1519 accum_buf = NULL;
1520 }
1521
1522 if (base_buf && max_size > base_bufsz) {
1523 uint8_t *p = realloc(base_buf, max_size);
1524 if (p == NULL) {
1525 err = got_error_from_errno("realloc");
1526 if (!cached)
1527 free(delta_buf);
1528 goto done;
1529 }
1530 base_buf = p;
1531 base_bufsz = max_size;
1532 }
1533
1534 if (accum_buf && max_size > accum_bufsz) {
1535 uint8_t *p = realloc(accum_buf, max_size);
1536 if (p == NULL) {
1537 err = got_error_from_errno("realloc");
1538 if (!cached)
1539 free(delta_buf);
1540 goto done;
1541 }
1542 accum_buf = p;
1543 accum_bufsz = max_size;
1544 }
1545
1546 if (base_buf) {
1547 err = got_delta_apply_in_mem(base_buf, base_bufsz,
1548 delta_buf, delta_len, accum_buf,
1549 &accum_size, max_size);
1550 n++;
1551 } else {
1552 err = got_delta_apply(base_file, delta_buf,
1553 delta_len,
1554 /* Final delta application writes to output file. */
1555 ++n < deltas->nentries ? accum_file : outfile,
1556 &accum_size);
1557 }
1558 if (!cached)
1559 free(delta_buf);
1560 if (err)
1561 goto done;
1562
1563 if (n < deltas->nentries) {
1564 /* Accumulated delta becomes the new base. */
1565 if (base_buf) {
1566 uint8_t *tmp = accum_buf;
1567 size_t tmp_size = accum_bufsz;
1568 accum_buf = base_buf;
1569 accum_bufsz = base_bufsz;
1570 base_buf = tmp;
1571 base_bufsz = tmp_size;
1572 } else {
1573 FILE *tmp = accum_file;
1574 accum_file = base_file;
1575 base_file = tmp;
1576 rewind(base_file);
1577 rewind(accum_file);
1578 }
1579 }
1580 }
1581
1582 done:
1583 free(base_buf);
1584 if (accum_buf) {
1585 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1586 free(accum_buf);
1587 if (len != accum_size)
1588 err = got_ferror(outfile, GOT_ERR_IO);
1589 }
1590 rewind(outfile);
1591 if (err == NULL)
1592 *result_size = accum_size;
1593 return err;
1594 }
1595
1596 const struct got_error *
1597 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1598 struct got_delta_chain *deltas, struct got_pack *pack)
1599 {
1600 const struct got_error *err = NULL;
1601 struct got_delta *delta;
1602 uint8_t *base_buf = NULL, *accum_buf = NULL;
1603 size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1604 uint64_t max_size = 0;
1605 int n = 0;
1606
1607 *outbuf = NULL;
1608 *outlen = 0;
1609
1610 if (STAILQ_EMPTY(&deltas->entries))
1611 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1612
1613 /* Deltas are ordered in ascending order. */
1614 STAILQ_FOREACH(delta, &deltas->entries, entry) {
1615 uint8_t *delta_buf = NULL;
1616 uint64_t base_size, result_size = 0;
1617 int cached = 1;
1618 if (n == 0) {
1619 off_t delta_data_offset;
1620
1621 /* Plain object types are the delta base. */
1622 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1623 delta->type != GOT_OBJ_TYPE_TREE &&
1624 delta->type != GOT_OBJ_TYPE_BLOB &&
1625 delta->type != GOT_OBJ_TYPE_TAG) {
1626 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1627 goto done;
1628 }
1629
1630 delta_data_offset = delta->offset + delta->tslen;
1631 if (delta_data_offset >= pack->filesize) {
1632 err = got_error(GOT_ERR_PACK_OFFSET);
1633 goto done;
1634 }
1635
1636 if (delta->size > max_size)
1637 max_size = delta->size;
1638
1639 if (pack->map) {
1640 size_t mapoff;
1641
1642 if (delta_data_offset > SIZE_MAX) {
1643 return got_error_fmt(GOT_ERR_RANGE,
1644 "delta %lld offset would "
1645 "overflow size_t",
1646 (long long)delta_data_offset);
1647 }
1648
1649 mapoff = delta_data_offset;
1650 err = got_inflate_to_mem_mmap(&base_buf,
1651 &base_bufsz, NULL, NULL, pack->map,
1652 mapoff, pack->filesize - mapoff);
1653 } else {
1654 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1655 == -1) {
1656 err = got_error_from_errno("lseek");
1657 goto done;
1658 }
1659 err = got_inflate_to_mem_fd(&base_buf,
1660 &base_bufsz, NULL, NULL, max_size,
1661 pack->fd);
1662 }
1663 if (err)
1664 goto done;
1665 n++;
1666 continue;
1667 }
1668
1669 if (pack->delta_cache) {
1670 got_delta_cache_get(&delta_buf, &delta_len,
1671 pack->delta_cache, delta->data_offset);
1672 }
1673 if (delta_buf == NULL) {
1674 cached = 0;
1675 err = read_delta_data(&delta_buf, &delta_len, NULL,
1676 delta->data_offset, pack);
1677 if (err)
1678 goto done;
1679 }
1680 if (pack->delta_cache && !cached) {
1681 err = got_delta_cache_add(pack->delta_cache,
1682 delta->data_offset, delta_buf, delta_len);
1683 if (err == NULL)
1684 cached = 1;
1685 else if (err->code != GOT_ERR_NO_SPACE) {
1686 free(delta_buf);
1687 goto done;
1688 }
1689 }
1690
1691 err = got_delta_get_sizes(&base_size, &result_size,
1692 delta_buf, delta_len);
1693 if (err) {
1694 if (!cached)
1695 free(delta_buf);
1696 goto done;
1697 }
1698 if (base_size > max_size)
1699 max_size = base_size;
1700 if (result_size > max_size)
1701 max_size = result_size;
1702
1703 if (max_size > base_bufsz) {
1704 uint8_t *p = realloc(base_buf, max_size);
1705 if (p == NULL) {
1706 err = got_error_from_errno("realloc");
1707 if (!cached)
1708 free(delta_buf);
1709 goto done;
1710 }
1711 base_buf = p;
1712 base_bufsz = max_size;
1713 }
1714
1715 if (max_size > accum_bufsz) {
1716 uint8_t *p = realloc(accum_buf, max_size);
1717 if (p == NULL) {
1718 err = got_error_from_errno("realloc");
1719 if (!cached)
1720 free(delta_buf);
1721 goto done;
1722 }
1723 accum_buf = p;
1724 accum_bufsz = max_size;
1725 }
1726
1727 err = got_delta_apply_in_mem(base_buf, base_bufsz,
1728 delta_buf, delta_len, accum_buf,
1729 &accum_size, max_size);
1730 if (!cached)
1731 free(delta_buf);
1732 n++;
1733 if (err)
1734 goto done;
1735
1736 if (n < deltas->nentries) {
1737 /* Accumulated delta becomes the new base. */
1738 uint8_t *tmp = accum_buf;
1739 size_t tmp_size = accum_bufsz;
1740 accum_buf = base_buf;
1741 accum_bufsz = base_bufsz;
1742 base_buf = tmp;
1743 base_bufsz = tmp_size;
1744 }
1745 }
1746
1747 done:
1748 free(base_buf);
1749 if (err) {
1750 free(accum_buf);
1751 *outbuf = NULL;
1752 *outlen = 0;
1753 } else {
1754 *outbuf = accum_buf;
1755 *outlen = accum_size;
1756 }
1757 return err;
1758 }
1759
1760 const struct got_error *
1761 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1762 FILE *outfile, FILE *base_file, FILE *accum_file)
1763 {
1764 const struct got_error *err = NULL;
1765
1766 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1767 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1768
1769 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1770 if (obj->pack_offset >= pack->filesize)
1771 return got_error(GOT_ERR_PACK_OFFSET);
1772
1773 if (pack->map) {
1774 size_t mapoff;
1775
1776 if (obj->pack_offset > SIZE_MAX) {
1777 return got_error_fmt(GOT_ERR_RANGE,
1778 "pack offset %lld would overflow size_t",
1779 (long long)obj->pack_offset);
1780 }
1781
1782 mapoff = obj->pack_offset;
1783 err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1784 pack->map, mapoff, pack->filesize - mapoff,
1785 outfile);
1786 } else {
1787 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1788 return got_error_from_errno("lseek");
1789 err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1790 pack->fd, outfile);
1791 }
1792 } else
1793 err = got_pack_dump_delta_chain_to_file(&obj->size,
1794 &obj->deltas, pack, outfile, base_file, accum_file);
1795
1796 return err;
1797 }
1798
1799 const struct got_error *
1800 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1801 struct got_object *obj, struct got_pack *pack)
1802 {
1803 const struct got_error *err = NULL;
1804
1805 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1806 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1807
1808 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1809 if (obj->pack_offset >= pack->filesize)
1810 return got_error(GOT_ERR_PACK_OFFSET);
1811 if (pack->map) {
1812 size_t mapoff;
1813
1814 if (obj->pack_offset > SIZE_MAX) {
1815 return got_error_fmt(GOT_ERR_RANGE,
1816 "pack offset %lld would overflow size_t",
1817 (long long)obj->pack_offset);
1818 }
1819
1820 mapoff = obj->pack_offset;
1821 err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1822 pack->map, mapoff, pack->filesize - mapoff);
1823 } else {
1824 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1825 return got_error_from_errno("lseek");
1826 err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1827 obj->size, pack->fd);
1828 }
1829 } else
1830 err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1831 pack);
1832
1833 return err;
1834 }
1835
1836 static const struct got_error *
1837 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1838 size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1839 off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1840 {
1841 const struct got_error *err = NULL;
1842
1843 /* Validate decompression and obtain the decompressed size. */
1844 err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1845 delta_data_offset, pack);
1846 if (err)
1847 return err;
1848
1849 /* Read delta base/result sizes from head of delta stream. */
1850 err = got_delta_get_sizes(base_size, result_size,
1851 *delta_buf, *delta_len);
1852 if (err)
1853 goto done;
1854
1855 /* Discard decompressed delta and read it again in compressed form. */
1856 free(*delta_buf);
1857 *delta_buf = malloc(*delta_len_compressed);
1858 if (*delta_buf == NULL) {
1859 err = got_error_from_errno("malloc");
1860 goto done;
1861 }
1862 if (pack->map) {
1863 if (delta_data_offset >= pack->filesize) {
1864 err = got_error(GOT_ERR_PACK_OFFSET);
1865 goto done;
1866 }
1867 memcpy(*delta_buf, pack->map + delta_data_offset,
1868 *delta_len_compressed);
1869 } else {
1870 ssize_t n;
1871 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1872 err = got_error_from_errno("lseek");
1873 goto done;
1874 }
1875 n = read(pack->fd, *delta_buf, *delta_len_compressed);
1876 if (n < 0) {
1877 err = got_error_from_errno("read");
1878 goto done;
1879 } else if (n != *delta_len_compressed) {
1880 err = got_error(GOT_ERR_IO);
1881 goto done;
1882 }
1883 }
1884 done:
1885 if (err) {
1886 free(*delta_buf);
1887 *delta_buf = NULL;
1888 *delta_len = 0;
1889 *delta_len_compressed = 0;
1890 *base_size = 0;
1891 *result_size = 0;
1892 }
1893 return err;
1894 }
1895
1896 const struct got_error *
1897 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1898 size_t *delta_compressed_size, off_t *delta_offset,
1899 off_t *delta_data_offset, off_t *base_offset,
1900 struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1901 struct got_pack *pack, struct got_packidx *packidx, int idx)
1902 {
1903 const struct got_error *err = NULL;
1904 off_t offset;
1905 uint8_t type;
1906 uint64_t size;
1907 size_t tslen, delta_hdrlen;
1908
1909 *delta_buf = NULL;
1910 *delta_size = 0;
1911 *delta_compressed_size = 0;
1912 *delta_offset = 0;
1913 *delta_data_offset = 0;
1914 *base_offset = 0;
1915 *base_size = 0;
1916 *result_size = 0;
1917
1918 offset = got_packidx_get_object_offset(packidx, idx);
1919 if (offset == -1)
1920 return got_error(GOT_ERR_BAD_PACKIDX);
1921
1922 if (offset >= pack->filesize)
1923 return got_error(GOT_ERR_PACK_OFFSET);
1924
1925 err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1926 pack, offset);
1927 if (err)
1928 return err;
1929
1930 if (tslen + size < tslen || offset + size < size ||
1931 tslen + offset < tslen)
1932 return got_error(GOT_ERR_PACK_OFFSET);
1933
1934 switch (type) {
1935 case GOT_OBJ_TYPE_OFFSET_DELTA:
1936 err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1937 pack, offset, tslen);
1938 if (err)
1939 return err;
1940 break;
1941 case GOT_OBJ_TYPE_REF_DELTA:
1942 err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1943 if (err)
1944 return err;
1945 delta_hdrlen = SHA1_DIGEST_LENGTH;
1946 break;
1947 default:
1948 return got_error_fmt(GOT_ERR_OBJ_TYPE,
1949 "non-delta object type %d found at offset %lld",
1950 type, (long long)offset);
1951 }
1952
1953 if (tslen + delta_hdrlen < delta_hdrlen ||
1954 offset + delta_hdrlen < delta_hdrlen)
1955 return got_error(GOT_ERR_BAD_DELTA);
1956
1957 *delta_data_offset = offset + tslen + delta_hdrlen;
1958 err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1959 base_size, result_size, *delta_data_offset, pack, packidx);
1960 if (err)
1961 return err;
1962
1963 if (*delta_size != size) {
1964 err = got_error(GOT_ERR_BAD_DELTA);
1965 goto done;
1966 }
1967
1968 *delta_offset = offset;
1969 done:
1970 if (err) {
1971 free(*delta_buf);
1972 *delta_buf = NULL;
1973 *delta_size = 0;
1974 *delta_compressed_size = 0;
1975 *delta_offset = 0;
1976 *base_offset = 0;
1977 *base_size = 0;
1978 *result_size = 0;
1979 }
1980 return err;
1981 }
portable version of Game of Trees