2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
4 0a0a3048 2018-01-10 stsp * Permission to use, copy, modify, and distribute this software for any
5 0a0a3048 2018-01-10 stsp * purpose with or without fee is hereby granted, provided that the above
6 0a0a3048 2018-01-10 stsp * copyright notice and this permission notice appear in all copies.
8 0a0a3048 2018-01-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 0a0a3048 2018-01-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 0a0a3048 2018-01-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 0a0a3048 2018-01-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 0a0a3048 2018-01-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 0a0a3048 2018-01-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 0a0a3048 2018-01-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 8cff5e95 2023-03-08 thomas #include "got_compat.h"
18 a1fd68d8 2018-01-12 stsp #include <sys/types.h>
19 0a0a3048 2018-01-10 stsp #include <sys/stat.h>
20 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
21 876c234b 2018-09-10 stsp #include <sys/uio.h>
22 57b35b75 2018-06-22 stsp #include <sys/mman.h>
23 68036464 2022-06-26 thomas #include <sys/resource.h>
24 68036464 2022-06-26 thomas #include <sys/socket.h>
26 7e656b93 2018-03-17 stsp #include <fcntl.h>
27 a1fd68d8 2018-01-12 stsp #include <errno.h>
28 0a0a3048 2018-01-10 stsp #include <stdio.h>
29 a1fd68d8 2018-01-12 stsp #include <stdint.h>
30 0a0a3048 2018-01-10 stsp #include <stdlib.h>
31 0a0a3048 2018-01-10 stsp #include <string.h>
32 0a0a3048 2018-01-10 stsp #include <limits.h>
33 81a12da5 2020-09-09 naddy #include <unistd.h>
34 a1fd68d8 2018-01-12 stsp #include <zlib.h>
36 0a0a3048 2018-01-10 stsp #include "got_error.h"
37 a1fd68d8 2018-01-12 stsp #include "got_object.h"
38 324d37e7 2019-05-11 stsp #include "got_path.h"
40 be288a59 2023-02-23 thomas #include "got_lib_hash.h"
41 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
42 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
43 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
44 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
45 232c0ac1 2023-04-22 thomas #include "got_lib_object_qid.h"
46 dd88155e 2019-06-29 stsp #include "got_lib_object_parse.h"
47 876c234b 2018-09-10 stsp #include "got_lib_privsep.h"
48 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
50 79b11c62 2018-03-09 stsp #ifndef nitems
51 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
55 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
58 0a0a3048 2018-01-10 stsp static const struct got_error *
59 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
63 0a0a3048 2018-01-10 stsp for (i = 0; i < 0xff - 1; i++) {
64 a1fd68d8 2018-01-12 stsp if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
65 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PACKIDX);
68 0a0a3048 2018-01-10 stsp return NULL;
71 1510f469 2018-09-09 stsp const struct got_error *
72 c3564dfa 2021-07-15 stsp got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
74 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
75 b16893ba 2023-02-24 thomas enum got_hash_algorithm algo = GOT_HASH_SHA1;
76 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
77 b16893ba 2023-02-24 thomas struct got_hash ctx;
78 b16893ba 2023-02-24 thomas uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
79 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
83 b16893ba 2023-02-24 thomas got_hash_init(&ctx, algo);
85 57b35b75 2018-06-22 stsp h = &p->hdr;
87 57b35b75 2018-06-22 stsp remain = p->len;
89 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
90 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
94 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
96 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
97 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
98 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
101 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
102 b1317e77 2019-09-22 stsp if (n < 0) {
103 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
105 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
106 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
110 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
111 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
114 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
115 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
117 0cb74cf4 2018-07-08 stsp if (verify)
118 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->magic, sizeof(*h->magic));
120 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
121 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
124 fc79a48d 2018-07-09 stsp if (p->map)
125 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
127 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
128 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
129 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
132 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
133 c6368c2e 2019-10-11 stsp if (n < 0) {
134 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
136 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
137 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
141 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
142 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
145 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
146 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
148 0cb74cf4 2018-07-08 stsp if (verify)
149 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->version, sizeof(*h->version));
151 57b35b75 2018-06-22 stsp len_fanout =
152 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
153 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
154 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
157 fc79a48d 2018-07-09 stsp if (p->map)
158 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
160 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
161 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
162 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
165 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
166 c6368c2e 2019-10-11 stsp if (n < 0) {
167 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
169 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
170 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
174 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
177 0cb74cf4 2018-07-08 stsp if (verify)
178 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->fanout_table, len_fanout);
179 57b35b75 2018-06-22 stsp offset += len_fanout;
180 57b35b75 2018-06-22 stsp remain -= len_fanout;
182 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
183 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
184 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
185 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
188 fc79a48d 2018-07-09 stsp if (p->map)
189 fc79a48d 2018-07-09 stsp h->sorted_ids =
190 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
192 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
193 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
194 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
197 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
199 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
200 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
201 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
205 0cb74cf4 2018-07-08 stsp if (verify)
206 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->sorted_ids, len_ids);
207 57b35b75 2018-06-22 stsp offset += len_ids;
208 57b35b75 2018-06-22 stsp remain -= len_ids;
210 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
211 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
214 fc79a48d 2018-07-09 stsp if (p->map)
215 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
217 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
218 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
219 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
222 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
224 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
225 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
226 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
230 0cb74cf4 2018-07-08 stsp if (verify)
231 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->crc32, nobj * sizeof(*h->crc32));
232 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
233 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
235 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
236 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
239 fc79a48d 2018-07-09 stsp if (p->map)
240 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
242 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
243 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
244 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
247 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
249 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
250 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
251 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
255 0cb74cf4 2018-07-08 stsp if (verify)
256 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->offsets, nobj * sizeof(*h->offsets));
257 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->offsets);
258 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->offsets);
260 0a0a3048 2018-01-10 stsp /* Large file offsets are contained only in files > 2GB. */
261 c3564dfa 2021-07-15 stsp if (verify || packfile_size > 0x7fffffff) {
262 c3564dfa 2021-07-15 stsp for (i = 0; i < nobj; i++) {
263 c3564dfa 2021-07-15 stsp uint32_t o = h->offsets[i];
264 c3564dfa 2021-07-15 stsp if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
265 c3564dfa 2021-07-15 stsp p->nlargeobj++;
268 5e6be232 2019-11-08 stsp if (p->nlargeobj == 0)
269 0a0a3048 2018-01-10 stsp goto checksum;
270 c3564dfa 2021-07-15 stsp else if (packfile_size <= 0x7fffffff) {
271 c3564dfa 2021-07-15 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
275 5e6be232 2019-11-08 stsp if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
276 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
279 fc79a48d 2018-07-09 stsp if (p->map)
280 fc79a48d 2018-07-09 stsp h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
282 5e6be232 2019-11-08 stsp h->large_offsets = malloc(p->nlargeobj *
283 5e6be232 2019-11-08 stsp sizeof(*h->large_offsets));
284 de30857e 2019-08-23 stsp if (h->large_offsets == NULL) {
285 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
288 fc79a48d 2018-07-09 stsp n = read(p->fd, h->large_offsets,
289 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
291 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
292 5e6be232 2019-11-08 stsp else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
293 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
297 0cb74cf4 2018-07-08 stsp if (verify)
298 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->large_offsets,
299 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
300 5e6be232 2019-11-08 stsp remain -= p->nlargeobj * sizeof(*h->large_offsets);
301 5e6be232 2019-11-08 stsp offset += p->nlargeobj * sizeof(*h->large_offsets);
304 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->trailer)) {
305 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
308 fc79a48d 2018-07-09 stsp if (p->map)
309 fc79a48d 2018-07-09 stsp h->trailer =
310 fc79a48d 2018-07-09 stsp (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
312 fc79a48d 2018-07-09 stsp h->trailer = malloc(sizeof(*h->trailer));
313 fc79a48d 2018-07-09 stsp if (h->trailer == NULL) {
314 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
317 fc79a48d 2018-07-09 stsp n = read(p->fd, h->trailer, sizeof(*h->trailer));
319 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
320 faaa1c0f 2018-09-15 stsp else if (n != sizeof(*h->trailer)) {
321 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
325 0cb74cf4 2018-07-08 stsp if (verify) {
326 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->trailer->packfile_sha1,
327 b16893ba 2023-02-24 thomas SHA1_DIGEST_LENGTH);
328 b16893ba 2023-02-24 thomas got_hash_final(&ctx, hash);
329 1e73031b 2023-04-04 thomas if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1) != 0)
330 0cb74cf4 2018-07-08 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
333 817c5a18 2018-09-09 stsp return err;
336 817c5a18 2018-09-09 stsp const struct got_error *
337 6d5a9006 2020-12-16 yzhong got_packidx_open(struct got_packidx **packidx,
338 6d5a9006 2020-12-16 yzhong int dir_fd, const char *relpath, int verify)
340 817c5a18 2018-09-09 stsp const struct got_error *err = NULL;
341 1124fe40 2021-07-07 stsp struct got_packidx *p = NULL;
342 1124fe40 2021-07-07 stsp char *pack_relpath;
343 c3564dfa 2021-07-15 stsp struct stat idx_sb, pack_sb;
345 817c5a18 2018-09-09 stsp *packidx = NULL;
347 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath, relpath);
349 1124fe40 2021-07-07 stsp return err;
352 1124fe40 2021-07-07 stsp * Ensure that a corresponding pack file exists.
353 1124fe40 2021-07-07 stsp * Some Git repositories have this problem. Git seems to ignore
354 1124fe40 2021-07-07 stsp * the existence of lonely pack index files but we do not.
356 c3564dfa 2021-07-15 stsp if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
357 1124fe40 2021-07-07 stsp if (errno == ENOENT) {
358 1124fe40 2021-07-07 stsp err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
359 1124fe40 2021-07-07 stsp "%s", relpath);
361 1124fe40 2021-07-07 stsp err = got_error_from_errno2("fstatat", pack_relpath);
365 817c5a18 2018-09-09 stsp p = calloc(1, sizeof(*p));
366 1124fe40 2021-07-07 stsp if (p == NULL) {
367 1124fe40 2021-07-07 stsp err = got_error_from_errno("calloc");
371 fc63f50d 2021-12-31 thomas p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
372 6772cf22 2019-08-28 hiltjo if (p->fd == -1) {
373 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("openat", relpath);
377 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
378 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
381 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
382 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
383 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
387 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
388 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
389 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
393 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
394 aa75acde 2022-10-25 thomas if (p->len > 0 && p->len <= SIZE_MAX) {
395 aa75acde 2022-10-25 thomas p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
396 aa75acde 2022-10-25 thomas if (p->map == MAP_FAILED) {
397 aa75acde 2022-10-25 thomas if (errno != ENOMEM) {
398 aa75acde 2022-10-25 thomas err = got_error_from_errno("mmap");
399 aa75acde 2022-10-25 thomas goto done;
401 aa75acde 2022-10-25 thomas p->map = NULL; /* fall back to read(2) */
406 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
410 1124fe40 2021-07-07 stsp got_packidx_close(p);
412 0a0a3048 2018-01-10 stsp *packidx = p;
413 1124fe40 2021-07-07 stsp free(pack_relpath);
414 0a0a3048 2018-01-10 stsp return err;
417 57b35b75 2018-06-22 stsp const struct got_error *
418 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
420 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
422 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
423 fc79a48d 2018-07-09 stsp if (packidx->map) {
424 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
425 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
427 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
428 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
429 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
434 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
436 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
437 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
438 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_offsets);
439 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_large_offsets);
440 0a0a3048 2018-01-10 stsp free(packidx);
442 57b35b75 2018-06-22 stsp return err;
445 509c9973 2021-04-10 stsp const struct got_error *
446 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
448 509c9973 2021-04-10 stsp size_t size;
450 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
451 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
452 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
453 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
455 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
456 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
457 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
459 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
460 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
461 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
462 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
464 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
465 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
467 509c9973 2021-04-10 stsp return NULL;
471 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
473 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
474 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
475 a1fd68d8 2018-01-12 stsp uint64_t loffset;
476 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
477 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
478 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
480 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
481 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
483 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
487 48b4f239 2021-12-31 thomas got_packidx_get_object_idx(struct got_packidx *packidx,
488 48b4f239 2021-12-31 thomas struct got_object_id *id)
490 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
491 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
492 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
494 a1fd68d8 2018-01-12 stsp if (id0 > 0)
495 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
497 40aeb19c 2018-06-22 stsp while (left <= right) {
498 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
499 40aeb19c 2018-06-22 stsp int i, cmp;
501 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
502 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
503 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
504 16dcbf91 2018-04-01 stsp if (cmp == 0)
506 40aeb19c 2018-06-22 stsp else if (cmp > 0)
507 40aeb19c 2018-06-22 stsp left = i + 1;
508 40aeb19c 2018-06-22 stsp else if (cmp < 0)
509 40aeb19c 2018-06-22 stsp right = i - 1;
515 f9c2e8e5 2022-02-13 thomas static int
516 f9c2e8e5 2022-02-13 thomas offset_cmp(const void *pa, const void *pb)
518 f9c2e8e5 2022-02-13 thomas const struct got_pack_offset_index *a, *b;
520 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_offset_index *)pa;
521 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_offset_index *)pb;
523 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
524 f9c2e8e5 2022-02-13 thomas return -1;
525 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
526 f9c2e8e5 2022-02-13 thomas return 1;
528 f9c2e8e5 2022-02-13 thomas return 0;
531 f9c2e8e5 2022-02-13 thomas static int
532 f9c2e8e5 2022-02-13 thomas large_offset_cmp(const void *pa, const void *pb)
534 f9c2e8e5 2022-02-13 thomas const struct got_pack_large_offset_index *a, *b;
536 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_large_offset_index *)pa;
537 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_large_offset_index *)pb;
539 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
540 f9c2e8e5 2022-02-13 thomas return -1;
541 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
542 f9c2e8e5 2022-02-13 thomas return 1;
544 f9c2e8e5 2022-02-13 thomas return 0;
547 f9c2e8e5 2022-02-13 thomas static const struct got_error *
548 f9c2e8e5 2022-02-13 thomas build_offset_index(struct got_packidx *p)
550 f9c2e8e5 2022-02-13 thomas uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
551 f9c2e8e5 2022-02-13 thomas unsigned int i, j, k;
553 f9c2e8e5 2022-02-13 thomas p->sorted_offsets = calloc(nobj - p->nlargeobj,
554 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]));
555 f9c2e8e5 2022-02-13 thomas if (p->sorted_offsets == NULL)
556 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
558 f9c2e8e5 2022-02-13 thomas if (p->nlargeobj > 0) {
559 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets = calloc(p->nlargeobj,
560 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]));
561 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets == NULL)
562 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
567 f9c2e8e5 2022-02-13 thomas for (i = 0; i < nobj; i++) {
568 f9c2e8e5 2022-02-13 thomas uint32_t offset = be32toh(p->hdr.offsets[i]);
569 f9c2e8e5 2022-02-13 thomas if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
570 f9c2e8e5 2022-02-13 thomas uint64_t loffset;
571 f9c2e8e5 2022-02-13 thomas uint32_t idx;
572 f9c2e8e5 2022-02-13 thomas idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
573 f9c2e8e5 2022-02-13 thomas if (idx >= p->nlargeobj ||
574 f9c2e8e5 2022-02-13 thomas p->nlargeobj == 0 ||
575 f9c2e8e5 2022-02-13 thomas p->hdr.large_offsets == NULL)
576 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
577 f9c2e8e5 2022-02-13 thomas loffset = be64toh(p->hdr.large_offsets[idx]);
578 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].offset = loffset;
579 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].idx = i;
582 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].offset = offset;
583 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].idx = i;
587 f9c2e8e5 2022-02-13 thomas if (j != p->nlargeobj || k != nobj - p->nlargeobj)
588 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
590 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_offsets, nobj - p->nlargeobj,
591 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]), offset_cmp);
593 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets)
594 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_large_offsets, p->nlargeobj,
595 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
597 f9c2e8e5 2022-02-13 thomas return NULL;
600 876c234b 2018-09-10 stsp const struct got_error *
601 f9c2e8e5 2022-02-13 thomas got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
603 f9c2e8e5 2022-02-13 thomas const struct got_error *err;
604 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
605 f9c2e8e5 2022-02-13 thomas int i, left, right;
607 f9c2e8e5 2022-02-13 thomas *idx = -1;
609 f9c2e8e5 2022-02-13 thomas if (packidx->sorted_offsets == NULL) {
610 f9c2e8e5 2022-02-13 thomas err = build_offset_index(packidx);
612 f9c2e8e5 2022-02-13 thomas return err;
615 f9c2e8e5 2022-02-13 thomas if (offset >= 0x7fffffff) {
616 f9c2e8e5 2022-02-13 thomas uint64_t lo;
617 f9c2e8e5 2022-02-13 thomas left = 0, right = packidx->nlargeobj - 1;
618 f9c2e8e5 2022-02-13 thomas while (left <= right) {
619 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
620 f9c2e8e5 2022-02-13 thomas lo = packidx->sorted_large_offsets[i].offset;
621 f9c2e8e5 2022-02-13 thomas if (lo == offset) {
622 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_large_offsets[i].idx;
624 f9c2e8e5 2022-02-13 thomas } else if (offset > lo)
625 f9c2e8e5 2022-02-13 thomas left = i + 1;
626 f9c2e8e5 2022-02-13 thomas else if (offset < lo)
627 f9c2e8e5 2022-02-13 thomas right = i - 1;
630 f9c2e8e5 2022-02-13 thomas uint32_t o;
631 f9c2e8e5 2022-02-13 thomas left = 0, right = totobj - packidx->nlargeobj - 1;
632 f9c2e8e5 2022-02-13 thomas while (left <= right) {
633 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
634 f9c2e8e5 2022-02-13 thomas o = packidx->sorted_offsets[i].offset;
635 f9c2e8e5 2022-02-13 thomas if (o == offset) {
636 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_offsets[i].idx;
638 f9c2e8e5 2022-02-13 thomas } else if (offset > o)
639 f9c2e8e5 2022-02-13 thomas left = i + 1;
640 f9c2e8e5 2022-02-13 thomas else if (offset < o)
641 f9c2e8e5 2022-02-13 thomas right = i - 1;
645 f9c2e8e5 2022-02-13 thomas return NULL;
648 f9c2e8e5 2022-02-13 thomas const struct got_error *
649 f9c2e8e5 2022-02-13 thomas got_packidx_get_object_id(struct got_object_id *id,
650 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx, int idx)
652 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
653 f9c2e8e5 2022-02-13 thomas struct got_packidx_object_id *oid;
655 f9c2e8e5 2022-02-13 thomas if (idx < 0 || idx >= totobj)
656 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_NO_OBJ);
658 f9c2e8e5 2022-02-13 thomas oid = &packidx->hdr.sorted_ids[idx];
659 f9c2e8e5 2022-02-13 thomas memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
660 f9c2e8e5 2022-02-13 thomas return NULL;
663 f9c2e8e5 2022-02-13 thomas const struct got_error *
664 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
665 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
667 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
668 4277420a 2019-06-29 stsp u_int8_t id0;
669 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
670 4277420a 2019-06-29 stsp char hex[3];
671 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
672 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
673 ec138a87 2022-01-03 thomas uint32_t i = 0;
675 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
677 4277420a 2019-06-29 stsp if (prefix_len < 2)
678 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
680 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
681 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
682 4277420a 2019-06-29 stsp hex[2] = '\0';
683 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
684 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
686 ec138a87 2022-01-03 thomas if (id0 > 0)
687 ec138a87 2022-01-03 thomas i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
688 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
689 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
690 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
691 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
694 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
695 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
697 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
698 4277420a 2019-06-29 stsp if (cmp < 0) {
699 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
701 4277420a 2019-06-29 stsp } else if (cmp > 0)
704 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
707 ec242592 2022-04-22 thomas memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
708 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
710 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
714 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
715 68036464 2022-06-26 thomas return err;
718 68036464 2022-06-26 thomas static void
719 68036464 2022-06-26 thomas set_max_datasize(void)
721 68036464 2022-06-26 thomas struct rlimit rl;
723 68036464 2022-06-26 thomas if (getrlimit(RLIMIT_DATA, &rl) != 0)
726 68036464 2022-06-26 thomas rl.rlim_cur = rl.rlim_max;
727 68036464 2022-06-26 thomas setrlimit(RLIMIT_DATA, &rl);
730 68036464 2022-06-26 thomas const struct got_error *
731 68036464 2022-06-26 thomas got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
733 68036464 2022-06-26 thomas const struct got_error *err = NULL;
734 68036464 2022-06-26 thomas int imsg_fds[2];
735 68036464 2022-06-26 thomas pid_t pid;
736 68036464 2022-06-26 thomas struct imsgbuf *ibuf;
738 68036464 2022-06-26 thomas ibuf = calloc(1, sizeof(*ibuf));
739 68036464 2022-06-26 thomas if (ibuf == NULL)
740 68036464 2022-06-26 thomas return got_error_from_errno("calloc");
742 68036464 2022-06-26 thomas pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
743 68036464 2022-06-26 thomas if (pack->privsep_child == NULL) {
744 68036464 2022-06-26 thomas err = got_error_from_errno("calloc");
745 68036464 2022-06-26 thomas free(ibuf);
746 68036464 2022-06-26 thomas return err;
748 68036464 2022-06-26 thomas pack->child_has_tempfiles = 0;
749 68036464 2022-06-26 thomas pack->child_has_delta_outfd = 0;
751 68036464 2022-06-26 thomas if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
752 68036464 2022-06-26 thomas err = got_error_from_errno("socketpair");
753 68036464 2022-06-26 thomas goto done;
756 68036464 2022-06-26 thomas pid = fork();
757 68036464 2022-06-26 thomas if (pid == -1) {
758 68036464 2022-06-26 thomas err = got_error_from_errno("fork");
759 68036464 2022-06-26 thomas goto done;
760 68036464 2022-06-26 thomas } else if (pid == 0) {
761 68036464 2022-06-26 thomas set_max_datasize();
762 68036464 2022-06-26 thomas got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
763 68036464 2022-06-26 thomas pack->path_packfile);
764 68036464 2022-06-26 thomas /* not reached */
767 68036464 2022-06-26 thomas if (close(imsg_fds[1]) == -1)
768 68036464 2022-06-26 thomas return got_error_from_errno("close");
769 68036464 2022-06-26 thomas pack->privsep_child->imsg_fd = imsg_fds[0];
770 68036464 2022-06-26 thomas pack->privsep_child->pid = pid;
771 68036464 2022-06-26 thomas imsg_init(ibuf, imsg_fds[0]);
772 68036464 2022-06-26 thomas pack->privsep_child->ibuf = ibuf;
774 68036464 2022-06-26 thomas err = got_privsep_init_pack_child(ibuf, pack, packidx);
775 68036464 2022-06-26 thomas if (err) {
776 68036464 2022-06-26 thomas const struct got_error *child_err;
777 68036464 2022-06-26 thomas err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
778 68036464 2022-06-26 thomas child_err = got_privsep_wait_for_child(
779 68036464 2022-06-26 thomas pack->privsep_child->pid);
780 68036464 2022-06-26 thomas if (child_err && err == NULL)
781 68036464 2022-06-26 thomas err = child_err;
784 68036464 2022-06-26 thomas if (err) {
785 68036464 2022-06-26 thomas free(ibuf);
786 68036464 2022-06-26 thomas free(pack->privsep_child);
787 68036464 2022-06-26 thomas pack->privsep_child = NULL;
789 dd88155e 2019-06-29 stsp return err;
792 b4f37570 2021-06-19 stsp static const struct got_error *
793 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
795 a6ca45d5 2022-10-24 thomas const struct got_error *err = NULL, *close_err = NULL;
797 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
798 876c234b 2018-09-10 stsp return NULL;
800 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
802 96732e0b 2018-11-11 stsp return err;
803 a6ca45d5 2022-10-24 thomas if (close(pack->privsep_child->imsg_fd) == -1)
804 a6ca45d5 2022-10-24 thomas close_err = got_error_from_errno("close");
805 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
806 a6ca45d5 2022-10-24 thomas if (close_err && err == NULL)
807 a6ca45d5 2022-10-24 thomas err = close_err;
808 a6ca45d5 2022-10-24 thomas imsg_clear(pack->privsep_child->ibuf);
809 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
810 876c234b 2018-09-10 stsp free(pack->privsep_child);
811 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
812 876c234b 2018-09-10 stsp return err;
815 d7464085 2018-07-09 stsp const struct got_error *
816 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
818 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
820 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
821 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
822 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
823 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
824 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
825 8b2180d4 2018-04-26 stsp pack->fd = -1;
826 4589e373 2018-03-17 stsp free(pack->path_packfile);
827 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
828 4589e373 2018-03-17 stsp pack->filesize = 0;
829 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
830 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
831 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
835 bfb5ee0b 2022-05-31 thomas * Leave accumfd and basefd alone. They are managed by the
836 bfb5ee0b 2022-05-31 thomas * repository layer and can be reused.
839 c7fe698a 2018-01-23 stsp return err;
842 668a20f6 2020-03-18 stsp const struct got_error *
843 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
844 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
846 a487c1d0 2018-01-14 stsp uint8_t t = 0;
847 a487c1d0 2018-01-14 stsp uint64_t s = 0;
848 a487c1d0 2018-01-14 stsp uint8_t sizeN;
849 d7464085 2018-07-09 stsp size_t mapoff = 0;
854 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
855 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
857 d7464085 2018-07-09 stsp if (pack->map) {
858 66e6097f 2022-10-27 thomas if (offset > SIZE_MAX) {
859 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
860 66e6097f 2022-10-27 thomas "offset %lld overflows size_t",
861 66e6097f 2022-10-27 thomas (long long)offset);
864 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
866 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
867 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
871 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
873 eac8a741 2022-10-20 thomas return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
874 04fb0dd8 2022-10-24 thomas "packfile offset %lld", (long long)offset);
876 d7464085 2018-07-09 stsp if (pack->map) {
877 1944573a 2022-01-10 thomas if (mapoff + sizeof(sizeN) >= pack->filesize)
878 1944573a 2022-01-10 thomas return got_error(GOT_ERR_BAD_PACKFILE);
879 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
880 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
882 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
884 638f9024 2019-05-13 stsp return got_error_from_errno("read");
885 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
886 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
888 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
890 a1fd68d8 2018-01-12 stsp if (i == 0) {
891 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
892 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
893 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
895 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
896 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
899 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
903 a487c1d0 2018-01-14 stsp return NULL;
906 a1fd68d8 2018-01-12 stsp static const struct got_error *
907 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
908 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
910 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
911 6ccb713b 2018-01-19 stsp if (*obj == NULL)
912 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
914 6ccb713b 2018-01-19 stsp (*obj)->type = type;
915 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
916 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
917 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
918 6ccb713b 2018-01-19 stsp (*obj)->size = size;
919 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
920 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
922 b107e67f 2018-01-19 stsp return NULL;
925 b107e67f 2018-01-19 stsp static const struct got_error *
926 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
927 d7464085 2018-07-09 stsp off_t delta_offset)
929 b107e67f 2018-01-19 stsp int64_t o = 0;
930 b107e67f 2018-01-19 stsp uint8_t offN;
933 a0de39f3 2019-08-09 stsp *offset = 0;
937 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
939 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
941 d7464085 2018-07-09 stsp if (pack->map) {
942 d7464085 2018-07-09 stsp size_t mapoff;
944 66e6097f 2022-10-27 thomas if (delta_offset + *len > SIZE_MAX) {
945 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
946 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
947 66e6097f 2022-10-27 thomas (long long)delta_offset + *len);
950 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
951 1944573a 2022-01-10 thomas if (mapoff + sizeof(offN) >= pack->filesize)
952 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
953 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
956 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
958 638f9024 2019-05-13 stsp return got_error_from_errno("read");
959 d7464085 2018-07-09 stsp if (n != sizeof(offN))
960 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
962 d7464085 2018-07-09 stsp *len += sizeof(offN);
964 b107e67f 2018-01-19 stsp if (i == 0)
965 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
969 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
972 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
974 b107e67f 2018-01-19 stsp *offset = o;
975 6ccb713b 2018-01-19 stsp return NULL;
978 668a20f6 2020-03-18 stsp const struct got_error *
979 48b4f239 2021-12-31 thomas got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
980 2424a69d 2022-10-24 thomas struct got_pack *pack, off_t offset, size_t tslen)
982 96f5e8b3 2018-01-23 stsp const struct got_error *err;
983 b107e67f 2018-01-19 stsp int64_t negoffset;
984 b107e67f 2018-01-19 stsp size_t negofflen;
988 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
989 d7464085 2018-07-09 stsp offset + tslen);
991 b107e67f 2018-01-19 stsp return err;
993 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
994 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
995 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
996 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
998 d7464085 2018-07-09 stsp *len = negofflen;
999 96f5e8b3 2018-01-23 stsp return NULL;
1002 0e22967e 2018-02-11 stsp static const struct got_error *
1003 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1004 9249e7e3 2022-05-12 thomas size_t *delta_compressed_len, size_t delta_data_offset,
1005 9249e7e3 2022-05-12 thomas struct got_pack *pack)
1007 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
1008 9249e7e3 2022-05-12 thomas size_t consumed = 0;
1010 42c69117 2019-11-10 stsp if (pack->map) {
1011 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
1012 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1013 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1014 9249e7e3 2022-05-12 thomas &consumed, NULL, pack->map, delta_data_offset,
1015 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
1016 9249e7e3 2022-05-12 thomas if (err)
1017 9249e7e3 2022-05-12 thomas return err;
1019 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1020 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
1021 9249e7e3 2022-05-12 thomas err = got_inflate_to_mem_fd(delta_buf, delta_len,
1022 9249e7e3 2022-05-12 thomas &consumed, NULL, 0, pack->fd);
1023 9249e7e3 2022-05-12 thomas if (err)
1024 9249e7e3 2022-05-12 thomas return err;
1027 9249e7e3 2022-05-12 thomas if (delta_compressed_len)
1028 9249e7e3 2022-05-12 thomas *delta_compressed_len = consumed;
1030 9249e7e3 2022-05-12 thomas return NULL;
1033 96f5e8b3 2018-01-23 stsp static const struct got_error *
1034 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1035 e485d53c 2022-10-28 thomas int delta_type, size_t delta_size, off_t delta_data_offset)
1037 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
1039 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1040 42c69117 2019-11-10 stsp delta_data_offset);
1041 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
1042 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
1043 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
1044 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
1045 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1046 bdd2fbb3 2018-02-11 stsp return NULL;
1049 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1050 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1051 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1052 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1054 a3500804 2018-01-23 stsp const struct got_error *err;
1055 c3703302 2018-01-23 stsp off_t base_offset;
1056 c3703302 2018-01-23 stsp uint8_t base_type;
1057 c3703302 2018-01-23 stsp uint64_t base_size;
1058 c3703302 2018-01-23 stsp size_t base_tslen;
1059 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1060 42c69117 2019-11-10 stsp size_t consumed;
1062 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1063 d7464085 2018-07-09 stsp delta_offset, tslen);
1065 a3500804 2018-01-23 stsp return err;
1067 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
1068 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
1069 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1071 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1072 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1073 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
1074 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1077 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1078 e485d53c 2022-10-28 thomas delta_data_offset);
1080 1a35c1bc 2019-05-22 stsp return err;
1082 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
1083 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1084 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1086 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1087 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1089 1a35c1bc 2019-05-22 stsp return err;
1091 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1092 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
1095 c4330eff 2021-06-22 stsp const struct got_error *
1096 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1097 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
1099 c4330eff 2021-06-22 stsp if (pack->map) {
1100 66e6097f 2022-10-27 thomas size_t mapoff;
1102 66e6097f 2022-10-27 thomas if (delta_offset + tslen > SIZE_MAX) {
1103 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
1104 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
1105 66e6097f 2022-10-27 thomas (long long)delta_offset + tslen);
1108 66e6097f 2022-10-27 thomas mapoff = delta_offset + tslen;
1109 1944573a 2022-01-10 thomas if (mapoff + sizeof(*id) >= pack->filesize)
1110 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
1111 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
1113 c4330eff 2021-06-22 stsp ssize_t n;
1114 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1115 c4330eff 2021-06-22 stsp if (n < 0)
1116 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1117 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1118 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1121 c4330eff 2021-06-22 stsp return NULL;
1124 c3703302 2018-01-23 stsp static const struct got_error *
1125 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1126 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1127 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1129 6b9c9673 2018-01-23 stsp const struct got_error *err;
1130 6b9c9673 2018-01-23 stsp struct got_object_id id;
1132 6b9c9673 2018-01-23 stsp off_t base_offset;
1133 6b9c9673 2018-01-23 stsp uint8_t base_type;
1134 6b9c9673 2018-01-23 stsp uint64_t base_size;
1135 6b9c9673 2018-01-23 stsp size_t base_tslen;
1136 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1138 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1139 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1141 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1143 c4330eff 2021-06-22 stsp return err;
1144 d7464085 2018-07-09 stsp if (pack->map) {
1145 4907b168 2023-02-20 thomas delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1147 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1148 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1149 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1152 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1153 e485d53c 2022-10-28 thomas delta_data_offset);
1155 1a35c1bc 2019-05-22 stsp return err;
1157 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1158 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1159 1a35c1bc 2019-05-22 stsp if (idx == -1)
1160 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1162 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1163 e4a6dbc6 2022-07-24 thomas if (base_offset == -1)
1164 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1166 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1167 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1169 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1170 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1172 1a35c1bc 2019-05-22 stsp return err;
1174 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1175 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1178 668a20f6 2020-03-18 stsp const struct got_error *
1179 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1180 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1181 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1183 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1185 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1186 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1188 c3703302 2018-01-23 stsp switch (delta_type) {
1189 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1190 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1191 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1192 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1193 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1194 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1195 42c69117 2019-11-10 stsp delta_size, 0);
1197 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1198 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1199 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1201 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1202 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1203 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1206 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1209 96f5e8b3 2018-01-23 stsp return err;
1212 96f5e8b3 2018-01-23 stsp static const struct got_error *
1213 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1214 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1215 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1217 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1218 96f5e8b3 2018-01-23 stsp int resolved_type;
1220 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1221 b107e67f 2018-01-19 stsp if (*obj == NULL)
1222 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1224 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1225 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1226 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1227 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1228 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1230 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1231 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1232 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1233 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1235 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1236 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1237 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1239 96f5e8b3 2018-01-23 stsp goto done;
1241 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1243 96f5e8b3 2018-01-23 stsp goto done;
1244 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1246 96f5e8b3 2018-01-23 stsp if (err) {
1247 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1248 96f5e8b3 2018-01-23 stsp *obj = NULL;
1250 96f5e8b3 2018-01-23 stsp return err;
1253 2090a03d 2018-09-09 stsp const struct got_error *
1254 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1255 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1257 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1258 a1fd68d8 2018-01-12 stsp off_t offset;
1259 6c00b545 2018-01-17 stsp uint8_t type;
1260 6c00b545 2018-01-17 stsp uint64_t size;
1261 3ee5fc21 2018-01-17 stsp size_t tslen;
1263 6c00b545 2018-01-17 stsp *obj = NULL;
1265 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1266 e4a6dbc6 2022-07-24 thomas if (offset == -1)
1267 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1269 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1270 668a20f6 2020-03-18 stsp pack, offset);
1272 35c73765 2018-09-09 stsp return err;
1274 6c00b545 2018-01-17 stsp switch (type) {
1275 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1276 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1277 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1278 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1279 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1280 5f25cc85 2019-11-26 stsp size, idx);
1282 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1283 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1284 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1285 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1288 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1292 3ee5fc21 2018-01-17 stsp return err;
1295 d582f26c 2020-03-18 stsp const struct got_error *
1296 48b4f239 2021-12-31 thomas got_pack_get_delta_chain_max_size(uint64_t *max_size,
1297 48b4f239 2021-12-31 thomas struct got_delta_chain *deltas, struct got_pack *pack)
1299 22484865 2018-03-13 stsp struct got_delta *delta;
1300 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1302 22484865 2018-03-13 stsp *max_size = 0;
1303 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1304 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1305 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1306 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1307 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1308 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1309 22484865 2018-03-13 stsp const struct got_error *err;
1310 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1311 42c69117 2019-11-10 stsp size_t delta_len;
1312 ab2f42e7 2019-11-10 stsp int cached = 1;
1314 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1315 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1316 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1318 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1319 ab2f42e7 2019-11-10 stsp cached = 0;
1320 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1321 9249e7e3 2022-05-12 thomas NULL, delta->data_offset, pack);
1323 ab2f42e7 2019-11-10 stsp return err;
1325 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1326 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1327 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1328 ab2f42e7 2019-11-10 stsp if (err == NULL)
1329 ab2f42e7 2019-11-10 stsp cached = 1;
1330 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1331 ab2f42e7 2019-11-10 stsp free(delta_buf);
1332 ab2f42e7 2019-11-10 stsp return err;
1335 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1336 42c69117 2019-11-10 stsp delta_buf, delta_len);
1337 ab2f42e7 2019-11-10 stsp if (!cached)
1338 ab2f42e7 2019-11-10 stsp free(delta_buf);
1340 22484865 2018-03-13 stsp return err;
1342 22484865 2018-03-13 stsp base_size = delta->size;
1343 22484865 2018-03-13 stsp if (base_size > *max_size)
1344 22484865 2018-03-13 stsp *max_size = base_size;
1345 22484865 2018-03-13 stsp if (result_size > *max_size)
1346 22484865 2018-03-13 stsp *max_size = result_size;
1349 9feb4ff2 2018-03-14 stsp return NULL;
1352 ac544f8c 2019-01-13 stsp const struct got_error *
1353 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1354 42c69117 2019-11-10 stsp struct got_pack *pack)
1356 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1357 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1359 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1362 668a20f6 2020-03-18 stsp const struct got_error *
1363 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1364 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1365 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1367 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1368 6691714a 2018-01-23 stsp struct got_delta *delta;
1369 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1370 210941d5 2022-05-31 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1371 210941d5 2022-05-31 thomas /* We process small enough files entirely in memory for speed. */
1372 210941d5 2022-05-31 thomas const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1373 210941d5 2022-05-31 thomas uint64_t max_size = 0;
1374 6691714a 2018-01-23 stsp int n = 0;
1376 b29656e2 2018-03-16 stsp *result_size = 0;
1378 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1379 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1381 210941d5 2022-05-31 thomas if (fseeko(base_file, 0L, SEEK_SET) == -1)
1382 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1383 210941d5 2022-05-31 thomas if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1384 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1386 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1387 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1388 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1389 210941d5 2022-05-31 thomas uint64_t base_size, result_size = 0;
1390 ab2f42e7 2019-11-10 stsp int cached = 1;
1391 a6b158cc 2018-02-11 stsp if (n == 0) {
1392 34fca9c3 2018-11-11 stsp size_t mapoff;
1393 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1395 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1396 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1397 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1398 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1399 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1400 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1401 72eb3431 2018-04-01 stsp goto done;
1404 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1405 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1406 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1407 0c048b15 2018-04-27 stsp goto done;
1409 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1410 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1412 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1413 d7464085 2018-07-09 stsp goto done;
1416 210941d5 2022-05-31 thomas if (delta->size > max_size)
1417 210941d5 2022-05-31 thomas max_size = delta->size;
1418 210941d5 2022-05-31 thomas if (max_size > max_bufsize) {
1419 d7464085 2018-07-09 stsp if (pack->map) {
1420 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1421 66e6097f 2022-10-27 thomas return got_error_fmt(
1422 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1423 66e6097f 2022-10-27 thomas "delta offset %lld "
1424 66e6097f 2022-10-27 thomas "overflows size_t",
1425 66e6097f 2022-10-27 thomas (long long)
1426 66e6097f 2022-10-27 thomas delta_data_offset);
1429 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1430 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1431 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1432 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1433 4788f1ce 2020-03-18 stsp base_file);
1435 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1436 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1437 4788f1ce 2020-03-18 stsp base_file);
1439 210941d5 2022-05-31 thomas accum_buf = malloc(max_size);
1440 210941d5 2022-05-31 thomas if (accum_buf == NULL) {
1441 210941d5 2022-05-31 thomas err = got_error_from_errno("malloc");
1442 210941d5 2022-05-31 thomas goto done;
1444 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1445 d7464085 2018-07-09 stsp if (pack->map) {
1446 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1447 66e6097f 2022-10-27 thomas return got_error_fmt(
1448 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1449 66e6097f 2022-10-27 thomas "delta offset %lld "
1450 66e6097f 2022-10-27 thomas "overflows size_t",
1451 66e6097f 2022-10-27 thomas (long long)
1452 66e6097f 2022-10-27 thomas delta_data_offset);
1455 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1456 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1457 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1458 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1459 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1461 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1462 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1463 55fdd257 2020-03-18 stsp pack->fd);
1466 a6b158cc 2018-02-11 stsp goto done;
1468 210941d5 2022-05-31 thomas if (base_buf == NULL)
1469 8628c62d 2018-03-15 stsp rewind(base_file);
1473 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1474 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1475 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1477 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1478 ab2f42e7 2019-11-10 stsp cached = 0;
1479 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1480 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1482 ab2f42e7 2019-11-10 stsp goto done;
1484 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1485 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1486 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1487 ab2f42e7 2019-11-10 stsp if (err == NULL)
1488 ab2f42e7 2019-11-10 stsp cached = 1;
1489 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1490 ab2f42e7 2019-11-10 stsp free(delta_buf);
1491 ab2f42e7 2019-11-10 stsp goto done;
1495 210941d5 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1496 210941d5 2022-05-31 thomas delta_buf, delta_len);
1497 b1fad72f 2022-11-08 thomas if (err) {
1498 b1fad72f 2022-11-08 thomas if (!cached)
1499 b1fad72f 2022-11-08 thomas free(delta_buf);
1500 210941d5 2022-05-31 thomas goto done;
1502 210941d5 2022-05-31 thomas if (base_size > max_size)
1503 210941d5 2022-05-31 thomas max_size = base_size;
1504 210941d5 2022-05-31 thomas if (result_size > max_size)
1505 210941d5 2022-05-31 thomas max_size = result_size;
1507 210941d5 2022-05-31 thomas if (base_buf && max_size > max_bufsize) {
1508 210941d5 2022-05-31 thomas /* Switch from buffers to temporary files. */
1509 210941d5 2022-05-31 thomas size_t w = fwrite(base_buf, 1, base_bufsz,
1510 210941d5 2022-05-31 thomas base_file);
1511 210941d5 2022-05-31 thomas if (w != base_bufsz) {
1512 210941d5 2022-05-31 thomas err = got_ferror(outfile, GOT_ERR_IO);
1513 b1fad72f 2022-11-08 thomas if (!cached)
1514 b1fad72f 2022-11-08 thomas free(delta_buf);
1515 210941d5 2022-05-31 thomas goto done;
1517 210941d5 2022-05-31 thomas free(base_buf);
1518 210941d5 2022-05-31 thomas base_buf = NULL;
1519 210941d5 2022-05-31 thomas free(accum_buf);
1520 210941d5 2022-05-31 thomas accum_buf = NULL;
1523 210941d5 2022-05-31 thomas if (base_buf && max_size > base_bufsz) {
1524 210941d5 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1525 210941d5 2022-05-31 thomas if (p == NULL) {
1526 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1527 b1fad72f 2022-11-08 thomas if (!cached)
1528 b1fad72f 2022-11-08 thomas free(delta_buf);
1529 210941d5 2022-05-31 thomas goto done;
1531 210941d5 2022-05-31 thomas base_buf = p;
1532 210941d5 2022-05-31 thomas base_bufsz = max_size;
1535 210941d5 2022-05-31 thomas if (accum_buf && max_size > accum_bufsz) {
1536 210941d5 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1537 210941d5 2022-05-31 thomas if (p == NULL) {
1538 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1539 b1fad72f 2022-11-08 thomas if (!cached)
1540 b1fad72f 2022-11-08 thomas free(delta_buf);
1541 210941d5 2022-05-31 thomas goto done;
1543 210941d5 2022-05-31 thomas accum_buf = p;
1544 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1547 8628c62d 2018-03-15 stsp if (base_buf) {
1548 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1549 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1550 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1553 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1554 42c69117 2019-11-10 stsp delta_len,
1555 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1556 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1557 b29656e2 2018-03-16 stsp &accum_size);
1559 ab2f42e7 2019-11-10 stsp if (!cached)
1560 ab2f42e7 2019-11-10 stsp free(delta_buf);
1562 6691714a 2018-01-23 stsp goto done;
1564 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1565 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1566 8628c62d 2018-03-15 stsp if (base_buf) {
1567 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1568 210941d5 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1569 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1570 210941d5 2022-05-31 thomas accum_bufsz = base_bufsz;
1571 8628c62d 2018-03-15 stsp base_buf = tmp;
1572 210941d5 2022-05-31 thomas base_bufsz = tmp_size;
1574 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1575 8628c62d 2018-03-15 stsp accum_file = base_file;
1576 8628c62d 2018-03-15 stsp base_file = tmp;
1577 8628c62d 2018-03-15 stsp rewind(base_file);
1578 8628c62d 2018-03-15 stsp rewind(accum_file);
1584 8628c62d 2018-03-15 stsp free(base_buf);
1585 8628c62d 2018-03-15 stsp if (accum_buf) {
1586 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1587 8628c62d 2018-03-15 stsp free(accum_buf);
1588 8628c62d 2018-03-15 stsp if (len != accum_size)
1589 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1591 6691714a 2018-01-23 stsp rewind(outfile);
1592 b29656e2 2018-03-16 stsp if (err == NULL)
1593 b29656e2 2018-03-16 stsp *result_size = accum_size;
1594 e0ab43e7 2018-03-16 stsp return err;
1597 668a20f6 2020-03-18 stsp const struct got_error *
1598 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1599 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1601 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1602 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1603 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1604 6c85883b 2022-05-31 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1605 6c85883b 2022-05-31 thomas uint64_t max_size = 0;
1606 e0ab43e7 2018-03-16 stsp int n = 0;
1608 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1609 e0ab43e7 2018-03-16 stsp *outlen = 0;
1611 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1612 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1614 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1615 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1616 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1617 6c85883b 2022-05-31 thomas uint64_t base_size, result_size = 0;
1618 ab2f42e7 2019-11-10 stsp int cached = 1;
1619 e0ab43e7 2018-03-16 stsp if (n == 0) {
1620 66e6097f 2022-10-27 thomas off_t delta_data_offset;
1622 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1623 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1624 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1625 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1626 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1627 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1628 72eb3431 2018-04-01 stsp goto done;
1631 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1632 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1633 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1634 0c048b15 2018-04-27 stsp goto done;
1637 6c85883b 2022-05-31 thomas if (delta->size > max_size)
1638 6c85883b 2022-05-31 thomas max_size = delta->size;
1640 d7464085 2018-07-09 stsp if (pack->map) {
1641 66e6097f 2022-10-27 thomas size_t mapoff;
1643 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1644 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1645 66e6097f 2022-10-27 thomas "delta %lld offset would "
1646 66e6097f 2022-10-27 thomas "overflow size_t",
1647 66e6097f 2022-10-27 thomas (long long)delta_data_offset);
1650 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1651 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1652 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1653 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1655 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1657 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1658 d7464085 2018-07-09 stsp goto done;
1660 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1661 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1662 55fdd257 2020-03-18 stsp pack->fd);
1665 d7464085 2018-07-09 stsp goto done;
1670 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1671 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1672 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1674 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1675 ab2f42e7 2019-11-10 stsp cached = 0;
1676 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1677 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1679 ab2f42e7 2019-11-10 stsp goto done;
1681 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1682 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1683 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1684 ab2f42e7 2019-11-10 stsp if (err == NULL)
1685 ab2f42e7 2019-11-10 stsp cached = 1;
1686 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1687 ab2f42e7 2019-11-10 stsp free(delta_buf);
1688 ab2f42e7 2019-11-10 stsp goto done;
1692 6c85883b 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1693 6c85883b 2022-05-31 thomas delta_buf, delta_len);
1694 b1fad72f 2022-11-08 thomas if (err) {
1695 b1fad72f 2022-11-08 thomas if (!cached)
1696 b1fad72f 2022-11-08 thomas free(delta_buf);
1697 6c85883b 2022-05-31 thomas goto done;
1699 6c85883b 2022-05-31 thomas if (base_size > max_size)
1700 6c85883b 2022-05-31 thomas max_size = base_size;
1701 6c85883b 2022-05-31 thomas if (result_size > max_size)
1702 6c85883b 2022-05-31 thomas max_size = result_size;
1704 6c85883b 2022-05-31 thomas if (max_size > base_bufsz) {
1705 6c85883b 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1706 6c85883b 2022-05-31 thomas if (p == NULL) {
1707 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1708 b1fad72f 2022-11-08 thomas if (!cached)
1709 b1fad72f 2022-11-08 thomas free(delta_buf);
1710 6c85883b 2022-05-31 thomas goto done;
1712 6c85883b 2022-05-31 thomas base_buf = p;
1713 6c85883b 2022-05-31 thomas base_bufsz = max_size;
1716 6c85883b 2022-05-31 thomas if (max_size > accum_bufsz) {
1717 6c85883b 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1718 6c85883b 2022-05-31 thomas if (p == NULL) {
1719 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1720 b1fad72f 2022-11-08 thomas if (!cached)
1721 b1fad72f 2022-11-08 thomas free(delta_buf);
1722 6c85883b 2022-05-31 thomas goto done;
1724 6c85883b 2022-05-31 thomas accum_buf = p;
1725 6c85883b 2022-05-31 thomas accum_bufsz = max_size;
1728 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1729 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1730 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1731 ab2f42e7 2019-11-10 stsp if (!cached)
1732 ab2f42e7 2019-11-10 stsp free(delta_buf);
1735 e0ab43e7 2018-03-16 stsp goto done;
1737 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1738 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1739 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1740 6c85883b 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1741 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1742 6c85883b 2022-05-31 thomas accum_bufsz = base_bufsz;
1743 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1744 6c85883b 2022-05-31 thomas base_bufsz = tmp_size;
1749 e0ab43e7 2018-03-16 stsp free(base_buf);
1750 e0ab43e7 2018-03-16 stsp if (err) {
1751 e0ab43e7 2018-03-16 stsp free(accum_buf);
1752 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1753 e0ab43e7 2018-03-16 stsp *outlen = 0;
1755 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1756 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1758 efd2a263 2018-01-19 stsp return err;
1761 3ee5fc21 2018-01-17 stsp const struct got_error *
1762 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1763 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1765 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1767 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1768 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1770 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1771 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1772 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1774 d7464085 2018-07-09 stsp if (pack->map) {
1775 66e6097f 2022-10-27 thomas size_t mapoff;
1777 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1778 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1779 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1780 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1783 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1784 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1785 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1788 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1789 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1790 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1791 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1794 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1795 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1797 a1fd68d8 2018-01-12 stsp return err;
1800 e0ab43e7 2018-03-16 stsp const struct got_error *
1801 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1802 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1804 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1806 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1807 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1809 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1810 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1811 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1812 d7464085 2018-07-09 stsp if (pack->map) {
1813 66e6097f 2022-10-27 thomas size_t mapoff;
1815 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1816 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1817 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1818 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1821 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1822 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1823 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1825 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1826 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1827 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1828 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1831 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1834 e0ab43e7 2018-03-16 stsp return err;
1837 9249e7e3 2022-05-12 thomas static const struct got_error *
1838 9249e7e3 2022-05-12 thomas read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1839 9249e7e3 2022-05-12 thomas size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1840 9249e7e3 2022-05-12 thomas off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1842 9249e7e3 2022-05-12 thomas const struct got_error *err = NULL;
1844 9249e7e3 2022-05-12 thomas /* Validate decompression and obtain the decompressed size. */
1845 9249e7e3 2022-05-12 thomas err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1846 9249e7e3 2022-05-12 thomas delta_data_offset, pack);
1847 9249e7e3 2022-05-12 thomas if (err)
1848 9249e7e3 2022-05-12 thomas return err;
1850 9249e7e3 2022-05-12 thomas /* Read delta base/result sizes from head of delta stream. */
1851 9249e7e3 2022-05-12 thomas err = got_delta_get_sizes(base_size, result_size,
1852 9249e7e3 2022-05-12 thomas *delta_buf, *delta_len);
1853 9249e7e3 2022-05-12 thomas if (err)
1854 9249e7e3 2022-05-12 thomas goto done;
1856 9249e7e3 2022-05-12 thomas /* Discard decompressed delta and read it again in compressed form. */
1857 9249e7e3 2022-05-12 thomas free(*delta_buf);
1858 9249e7e3 2022-05-12 thomas *delta_buf = malloc(*delta_len_compressed);
1859 9249e7e3 2022-05-12 thomas if (*delta_buf == NULL) {
1860 9249e7e3 2022-05-12 thomas err = got_error_from_errno("malloc");
1861 9249e7e3 2022-05-12 thomas goto done;
1863 9249e7e3 2022-05-12 thomas if (pack->map) {
1864 30d415a5 2022-10-25 thomas if (delta_data_offset >= pack->filesize) {
1865 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_PACK_OFFSET);
1866 30d415a5 2022-10-25 thomas goto done;
1868 9249e7e3 2022-05-12 thomas memcpy(*delta_buf, pack->map + delta_data_offset,
1869 9249e7e3 2022-05-12 thomas *delta_len_compressed);
1870 9249e7e3 2022-05-12 thomas } else {
1871 9249e7e3 2022-05-12 thomas ssize_t n;
1872 9249e7e3 2022-05-12 thomas if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1873 9249e7e3 2022-05-12 thomas err = got_error_from_errno("lseek");
1874 9249e7e3 2022-05-12 thomas goto done;
1876 9249e7e3 2022-05-12 thomas n = read(pack->fd, *delta_buf, *delta_len_compressed);
1877 9249e7e3 2022-05-12 thomas if (n < 0) {
1878 9249e7e3 2022-05-12 thomas err = got_error_from_errno("read");
1879 9249e7e3 2022-05-12 thomas goto done;
1880 9249e7e3 2022-05-12 thomas } else if (n != *delta_len_compressed) {
1881 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_IO);
1882 9249e7e3 2022-05-12 thomas goto done;
1886 9249e7e3 2022-05-12 thomas if (err) {
1887 9249e7e3 2022-05-12 thomas free(*delta_buf);
1888 9249e7e3 2022-05-12 thomas *delta_buf = NULL;
1889 9249e7e3 2022-05-12 thomas *delta_len = 0;
1890 9249e7e3 2022-05-12 thomas *delta_len_compressed = 0;
1891 9249e7e3 2022-05-12 thomas *base_size = 0;
1892 9249e7e3 2022-05-12 thomas *result_size = 0;
1894 9249e7e3 2022-05-12 thomas return err;
1897 f9c2e8e5 2022-02-13 thomas const struct got_error *
1898 f9c2e8e5 2022-02-13 thomas got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1899 c44c7d6e 2022-12-04 thomas size_t *delta_compressed_size, off_t *delta_offset,
1900 c44c7d6e 2022-12-04 thomas off_t *delta_data_offset, off_t *base_offset,
1901 9249e7e3 2022-05-12 thomas struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1902 9249e7e3 2022-05-12 thomas struct got_pack *pack, struct got_packidx *packidx, int idx)
1904 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
1905 f9c2e8e5 2022-02-13 thomas off_t offset;
1906 f9c2e8e5 2022-02-13 thomas uint8_t type;
1907 f9c2e8e5 2022-02-13 thomas uint64_t size;
1908 f9c2e8e5 2022-02-13 thomas size_t tslen, delta_hdrlen;
1910 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
1911 f9c2e8e5 2022-02-13 thomas *delta_size = 0;
1912 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
1913 f9c2e8e5 2022-02-13 thomas *delta_offset = 0;
1914 c44c7d6e 2022-12-04 thomas *delta_data_offset = 0;
1915 f9c2e8e5 2022-02-13 thomas *base_offset = 0;
1916 f9c2e8e5 2022-02-13 thomas *base_size = 0;
1917 f9c2e8e5 2022-02-13 thomas *result_size = 0;
1919 f9c2e8e5 2022-02-13 thomas offset = got_packidx_get_object_offset(packidx, idx);
1920 e4a6dbc6 2022-07-24 thomas if (offset == -1)
1921 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
1923 f9c2e8e5 2022-02-13 thomas if (offset >= pack->filesize)
1924 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
1926 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1927 f9c2e8e5 2022-02-13 thomas pack, offset);
1928 f9c2e8e5 2022-02-13 thomas if (err)
1929 f9c2e8e5 2022-02-13 thomas return err;
1931 f9c2e8e5 2022-02-13 thomas if (tslen + size < tslen || offset + size < size ||
1932 f9c2e8e5 2022-02-13 thomas tslen + offset < tslen)
1933 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
1935 f9c2e8e5 2022-02-13 thomas switch (type) {
1936 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_OFFSET_DELTA:
1937 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1938 f9c2e8e5 2022-02-13 thomas pack, offset, tslen);
1939 f9c2e8e5 2022-02-13 thomas if (err)
1940 f9c2e8e5 2022-02-13 thomas return err;
1942 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_REF_DELTA:
1943 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1944 f9c2e8e5 2022-02-13 thomas if (err)
1945 f9c2e8e5 2022-02-13 thomas return err;
1946 f9c2e8e5 2022-02-13 thomas delta_hdrlen = SHA1_DIGEST_LENGTH;
1948 f9c2e8e5 2022-02-13 thomas default:
1949 f9c2e8e5 2022-02-13 thomas return got_error_fmt(GOT_ERR_OBJ_TYPE,
1950 e4a6dbc6 2022-07-24 thomas "non-delta object type %d found at offset %lld",
1951 e4a6dbc6 2022-07-24 thomas type, (long long)offset);
1954 f9c2e8e5 2022-02-13 thomas if (tslen + delta_hdrlen < delta_hdrlen ||
1955 f9c2e8e5 2022-02-13 thomas offset + delta_hdrlen < delta_hdrlen)
1956 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_DELTA);
1958 c44c7d6e 2022-12-04 thomas *delta_data_offset = offset + tslen + delta_hdrlen;
1959 9249e7e3 2022-05-12 thomas err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1960 c44c7d6e 2022-12-04 thomas base_size, result_size, *delta_data_offset, pack, packidx);
1961 f9c2e8e5 2022-02-13 thomas if (err)
1962 f9c2e8e5 2022-02-13 thomas return err;
1964 f9c2e8e5 2022-02-13 thomas if (*delta_size != size) {
1965 f9c2e8e5 2022-02-13 thomas err = got_error(GOT_ERR_BAD_DELTA);
1966 f9c2e8e5 2022-02-13 thomas goto done;
1969 f9c2e8e5 2022-02-13 thomas *delta_offset = offset;
1971 f9c2e8e5 2022-02-13 thomas if (err) {
1972 f9c2e8e5 2022-02-13 thomas free(*delta_buf);
1973 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
1974 9249e7e3 2022-05-12 thomas *delta_size = 0;
1975 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
1976 9249e7e3 2022-05-12 thomas *delta_offset = 0;
1977 9249e7e3 2022-05-12 thomas *base_offset = 0;
1978 9249e7e3 2022-05-12 thomas *base_size = 0;
1979 9249e7e3 2022-05-12 thomas *result_size = 0;
1981 f9c2e8e5 2022-02-13 thomas return err;