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.
17 a1fd68d8 2018-01-12 stsp #include <sys/types.h>
18 0a0a3048 2018-01-10 stsp #include <sys/stat.h>
19 a1fd68d8 2018-01-12 stsp #include <sys/queue.h>
20 876c234b 2018-09-10 stsp #include <sys/uio.h>
21 57b35b75 2018-06-22 stsp #include <sys/mman.h>
23 7e656b93 2018-03-17 stsp #include <fcntl.h>
24 a1fd68d8 2018-01-12 stsp #include <errno.h>
25 0a0a3048 2018-01-10 stsp #include <stdio.h>
26 a1fd68d8 2018-01-12 stsp #include <stdint.h>
27 0a0a3048 2018-01-10 stsp #include <stdlib.h>
28 0a0a3048 2018-01-10 stsp #include <string.h>
29 0a0a3048 2018-01-10 stsp #include <limits.h>
30 0a0a3048 2018-01-10 stsp #include <sha1.h>
31 0a0a3048 2018-01-10 stsp #include <endian.h>
32 81a12da5 2020-09-09 naddy #include <unistd.h>
33 a1fd68d8 2018-01-12 stsp #include <zlib.h>
34 876c234b 2018-09-10 stsp #include <imsg.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 718b3ab0 2018-03-17 stsp #include "got_lib_sha1.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 dd88155e 2019-06-29 stsp #include "got_lib_object_parse.h"
46 876c234b 2018-09-10 stsp #include "got_lib_privsep.h"
47 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
49 79b11c62 2018-03-09 stsp #ifndef nitems
50 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
54 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 0a0a3048 2018-01-10 stsp static const struct got_error *
58 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
62 0a0a3048 2018-01-10 stsp for (i = 0; i < 0xff - 1; i++) {
63 a1fd68d8 2018-01-12 stsp if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
64 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PACKIDX);
67 0a0a3048 2018-01-10 stsp return NULL;
70 1510f469 2018-09-09 stsp const struct got_error *
71 c3564dfa 2021-07-15 stsp got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
73 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
74 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
75 0ebaf008 2018-01-10 stsp SHA1_CTX ctx;
76 0ebaf008 2018-01-10 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
77 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
81 0ebaf008 2018-01-10 stsp SHA1Init(&ctx);
83 57b35b75 2018-06-22 stsp h = &p->hdr;
85 57b35b75 2018-06-22 stsp remain = p->len;
87 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
88 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
92 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
94 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
95 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
96 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
99 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
100 b1317e77 2019-09-22 stsp if (n < 0) {
101 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
103 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
104 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
108 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
109 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
112 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
113 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
115 0cb74cf4 2018-07-08 stsp if (verify)
116 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
118 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
119 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
122 fc79a48d 2018-07-09 stsp if (p->map)
123 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
125 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
126 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
127 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
130 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
131 c6368c2e 2019-10-11 stsp if (n < 0) {
132 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
134 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
135 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
139 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
140 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
143 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
144 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
146 0cb74cf4 2018-07-08 stsp if (verify)
147 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
149 57b35b75 2018-06-22 stsp len_fanout =
150 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
151 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
152 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
155 fc79a48d 2018-07-09 stsp if (p->map)
156 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
158 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
159 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
160 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
163 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
164 c6368c2e 2019-10-11 stsp if (n < 0) {
165 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
167 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
168 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
172 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
175 0cb74cf4 2018-07-08 stsp if (verify)
176 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
177 57b35b75 2018-06-22 stsp offset += len_fanout;
178 57b35b75 2018-06-22 stsp remain -= len_fanout;
180 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
181 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
182 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
183 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
186 fc79a48d 2018-07-09 stsp if (p->map)
187 fc79a48d 2018-07-09 stsp h->sorted_ids =
188 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
190 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
191 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
192 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
195 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
197 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
198 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
199 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
203 0cb74cf4 2018-07-08 stsp if (verify)
204 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
205 57b35b75 2018-06-22 stsp offset += len_ids;
206 57b35b75 2018-06-22 stsp remain -= len_ids;
208 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
209 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
212 fc79a48d 2018-07-09 stsp if (p->map)
213 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
215 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
216 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
217 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
220 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
222 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
223 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
224 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
228 0cb74cf4 2018-07-08 stsp if (verify)
229 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
230 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
231 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
233 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
234 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
237 fc79a48d 2018-07-09 stsp if (p->map)
238 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
240 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
241 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
242 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
245 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
247 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
248 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
249 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
253 0cb74cf4 2018-07-08 stsp if (verify)
254 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->offsets,
255 0cb74cf4 2018-07-08 stsp nobj * sizeof(*h->offsets));
256 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->offsets);
257 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->offsets);
259 0a0a3048 2018-01-10 stsp /* Large file offsets are contained only in files > 2GB. */
260 c3564dfa 2021-07-15 stsp if (verify || packfile_size > 0x7fffffff) {
261 c3564dfa 2021-07-15 stsp for (i = 0; i < nobj; i++) {
262 c3564dfa 2021-07-15 stsp uint32_t o = h->offsets[i];
263 c3564dfa 2021-07-15 stsp if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
264 c3564dfa 2021-07-15 stsp p->nlargeobj++;
267 5e6be232 2019-11-08 stsp if (p->nlargeobj == 0)
268 0a0a3048 2018-01-10 stsp goto checksum;
269 c3564dfa 2021-07-15 stsp else if (packfile_size <= 0x7fffffff) {
270 c3564dfa 2021-07-15 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
274 5e6be232 2019-11-08 stsp if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
275 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
278 fc79a48d 2018-07-09 stsp if (p->map)
279 fc79a48d 2018-07-09 stsp h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
281 5e6be232 2019-11-08 stsp h->large_offsets = malloc(p->nlargeobj *
282 5e6be232 2019-11-08 stsp sizeof(*h->large_offsets));
283 de30857e 2019-08-23 stsp if (h->large_offsets == NULL) {
284 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
287 fc79a48d 2018-07-09 stsp n = read(p->fd, h->large_offsets,
288 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
290 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
291 5e6be232 2019-11-08 stsp else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
292 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
296 0cb74cf4 2018-07-08 stsp if (verify)
297 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t*)h->large_offsets,
298 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
299 5e6be232 2019-11-08 stsp remain -= p->nlargeobj * sizeof(*h->large_offsets);
300 5e6be232 2019-11-08 stsp offset += p->nlargeobj * sizeof(*h->large_offsets);
303 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->trailer)) {
304 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
307 fc79a48d 2018-07-09 stsp if (p->map)
308 fc79a48d 2018-07-09 stsp h->trailer =
309 fc79a48d 2018-07-09 stsp (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
311 fc79a48d 2018-07-09 stsp h->trailer = malloc(sizeof(*h->trailer));
312 fc79a48d 2018-07-09 stsp if (h->trailer == NULL) {
313 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
316 fc79a48d 2018-07-09 stsp n = read(p->fd, h->trailer, sizeof(*h->trailer));
318 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
319 faaa1c0f 2018-09-15 stsp else if (n != sizeof(*h->trailer)) {
320 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
324 0cb74cf4 2018-07-08 stsp if (verify) {
325 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
326 0cb74cf4 2018-07-08 stsp SHA1Final(sha1, &ctx);
327 0cb74cf4 2018-07-08 stsp if (memcmp(h->trailer->packidx_sha1, sha1,
328 0cb74cf4 2018-07-08 stsp SHA1_DIGEST_LENGTH) != 0)
329 0cb74cf4 2018-07-08 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
332 817c5a18 2018-09-09 stsp return err;
335 817c5a18 2018-09-09 stsp const struct got_error *
336 6d5a9006 2020-12-16 yzhong got_packidx_open(struct got_packidx **packidx,
337 6d5a9006 2020-12-16 yzhong int dir_fd, const char *relpath, int verify)
339 817c5a18 2018-09-09 stsp const struct got_error *err = NULL;
340 1124fe40 2021-07-07 stsp struct got_packidx *p = NULL;
341 1124fe40 2021-07-07 stsp char *pack_relpath;
342 c3564dfa 2021-07-15 stsp struct stat idx_sb, pack_sb;
344 817c5a18 2018-09-09 stsp *packidx = NULL;
346 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath, relpath);
348 1124fe40 2021-07-07 stsp return err;
351 1124fe40 2021-07-07 stsp * Ensure that a corresponding pack file exists.
352 1124fe40 2021-07-07 stsp * Some Git repositories have this problem. Git seems to ignore
353 1124fe40 2021-07-07 stsp * the existence of lonely pack index files but we do not.
355 c3564dfa 2021-07-15 stsp if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
356 1124fe40 2021-07-07 stsp if (errno == ENOENT) {
357 1124fe40 2021-07-07 stsp err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
358 1124fe40 2021-07-07 stsp "%s", relpath);
360 1124fe40 2021-07-07 stsp err = got_error_from_errno2("fstatat", pack_relpath);
364 817c5a18 2018-09-09 stsp p = calloc(1, sizeof(*p));
365 1124fe40 2021-07-07 stsp if (p == NULL) {
366 1124fe40 2021-07-07 stsp err = got_error_from_errno("calloc");
370 e7ae0baf 2021-12-31 stsp p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
371 6772cf22 2019-08-28 hiltjo if (p->fd == -1) {
372 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("openat", relpath);
376 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
377 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
380 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
381 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
382 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
386 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
387 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
388 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
392 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
393 817c5a18 2018-09-09 stsp p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
394 3a11398b 2019-02-21 stsp if (p->map == MAP_FAILED) {
395 3a11398b 2019-02-21 stsp if (errno != ENOMEM) {
396 638f9024 2019-05-13 stsp err = got_error_from_errno("mmap");
399 817c5a18 2018-09-09 stsp p->map = NULL; /* fall back to read(2) */
403 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
407 1124fe40 2021-07-07 stsp got_packidx_close(p);
409 0a0a3048 2018-01-10 stsp *packidx = p;
410 1124fe40 2021-07-07 stsp free(pack_relpath);
411 0a0a3048 2018-01-10 stsp return err;
414 57b35b75 2018-06-22 stsp const struct got_error *
415 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
417 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
419 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
420 fc79a48d 2018-07-09 stsp if (packidx->map) {
421 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
422 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
424 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
425 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
426 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
427 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
428 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
429 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
433 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
434 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
435 67fd6849 2022-02-13 stsp free(packidx->sorted_offsets);
436 67fd6849 2022-02-13 stsp free(packidx->sorted_large_offsets);
437 0a0a3048 2018-01-10 stsp free(packidx);
439 57b35b75 2018-06-22 stsp return err;
442 509c9973 2021-04-10 stsp const struct got_error *
443 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
445 509c9973 2021-04-10 stsp size_t size;
447 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
448 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
449 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
450 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
452 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
453 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
454 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
456 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
457 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
458 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
459 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
461 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
462 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
464 509c9973 2021-04-10 stsp return NULL;
468 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
470 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
471 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
472 a1fd68d8 2018-01-12 stsp uint64_t loffset;
473 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
474 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
475 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
477 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
478 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
480 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
484 c0df5966 2021-12-31 stsp got_packidx_get_object_idx(struct got_packidx *packidx,
485 c0df5966 2021-12-31 stsp struct got_object_id *id)
487 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
488 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
489 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
491 a1fd68d8 2018-01-12 stsp if (id0 > 0)
492 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
494 40aeb19c 2018-06-22 stsp while (left <= right) {
495 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
496 40aeb19c 2018-06-22 stsp int i, cmp;
498 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
499 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
500 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
501 16dcbf91 2018-04-01 stsp if (cmp == 0)
503 40aeb19c 2018-06-22 stsp else if (cmp > 0)
504 40aeb19c 2018-06-22 stsp left = i + 1;
505 40aeb19c 2018-06-22 stsp else if (cmp < 0)
506 40aeb19c 2018-06-22 stsp right = i - 1;
513 67fd6849 2022-02-13 stsp offset_cmp(const void *pa, const void *pb)
515 67fd6849 2022-02-13 stsp const struct got_pack_offset_index *a, *b;
517 67fd6849 2022-02-13 stsp a = (const struct got_pack_offset_index *)pa;
518 67fd6849 2022-02-13 stsp b = (const struct got_pack_offset_index *)pb;
520 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
522 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
529 67fd6849 2022-02-13 stsp large_offset_cmp(const void *pa, const void *pb)
531 67fd6849 2022-02-13 stsp const struct got_pack_large_offset_index *a, *b;
533 67fd6849 2022-02-13 stsp a = (const struct got_pack_large_offset_index *)pa;
534 67fd6849 2022-02-13 stsp b = (const struct got_pack_large_offset_index *)pb;
536 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
538 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
544 67fd6849 2022-02-13 stsp static const struct got_error *
545 67fd6849 2022-02-13 stsp build_offset_index(struct got_packidx *p)
547 67fd6849 2022-02-13 stsp uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
548 67fd6849 2022-02-13 stsp unsigned int i, j, k;
550 67fd6849 2022-02-13 stsp p->sorted_offsets = calloc(nobj - p->nlargeobj,
551 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]));
552 67fd6849 2022-02-13 stsp if (p->sorted_offsets == NULL)
553 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
555 67fd6849 2022-02-13 stsp if (p->nlargeobj > 0) {
556 67fd6849 2022-02-13 stsp p->sorted_large_offsets = calloc(p->nlargeobj,
557 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]));
558 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets == NULL)
559 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
564 67fd6849 2022-02-13 stsp for (i = 0; i < nobj; i++) {
565 67fd6849 2022-02-13 stsp uint32_t offset = be32toh(p->hdr.offsets[i]);
566 67fd6849 2022-02-13 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
567 67fd6849 2022-02-13 stsp uint64_t loffset;
568 67fd6849 2022-02-13 stsp uint32_t idx;
569 67fd6849 2022-02-13 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
570 67fd6849 2022-02-13 stsp if (idx >= p->nlargeobj ||
571 67fd6849 2022-02-13 stsp p->nlargeobj == 0 ||
572 67fd6849 2022-02-13 stsp p->hdr.large_offsets == NULL)
573 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
574 67fd6849 2022-02-13 stsp loffset = be64toh(p->hdr.large_offsets[idx]);
575 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].offset = loffset;
576 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].idx = i;
579 67fd6849 2022-02-13 stsp p->sorted_offsets[k].offset = offset;
580 67fd6849 2022-02-13 stsp p->sorted_offsets[k].idx = i;
584 67fd6849 2022-02-13 stsp if (j != p->nlargeobj || k != nobj - p->nlargeobj)
585 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
587 67fd6849 2022-02-13 stsp qsort(p->sorted_offsets, nobj - p->nlargeobj,
588 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]), offset_cmp);
590 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets)
591 67fd6849 2022-02-13 stsp qsort(p->sorted_large_offsets, p->nlargeobj,
592 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
594 67fd6849 2022-02-13 stsp return NULL;
597 876c234b 2018-09-10 stsp const struct got_error *
598 67fd6849 2022-02-13 stsp got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
600 67fd6849 2022-02-13 stsp const struct got_error *err;
601 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
602 67fd6849 2022-02-13 stsp int i, left, right;
606 67fd6849 2022-02-13 stsp if (packidx->sorted_offsets == NULL) {
607 67fd6849 2022-02-13 stsp err = build_offset_index(packidx);
609 67fd6849 2022-02-13 stsp return err;
612 67fd6849 2022-02-13 stsp if (offset >= 0x7fffffff) {
613 67fd6849 2022-02-13 stsp uint64_t lo;
614 67fd6849 2022-02-13 stsp left = 0, right = packidx->nlargeobj - 1;
615 67fd6849 2022-02-13 stsp while (left <= right) {
616 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
617 67fd6849 2022-02-13 stsp lo = packidx->sorted_large_offsets[i].offset;
618 67fd6849 2022-02-13 stsp if (lo == offset) {
619 67fd6849 2022-02-13 stsp *idx = packidx->sorted_large_offsets[i].idx;
621 67fd6849 2022-02-13 stsp } else if (offset > lo)
622 67fd6849 2022-02-13 stsp left = i + 1;
623 67fd6849 2022-02-13 stsp else if (offset < lo)
624 67fd6849 2022-02-13 stsp right = i - 1;
627 67fd6849 2022-02-13 stsp uint32_t o;
628 67fd6849 2022-02-13 stsp left = 0, right = totobj - packidx->nlargeobj - 1;
629 67fd6849 2022-02-13 stsp while (left <= right) {
630 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
631 67fd6849 2022-02-13 stsp o = packidx->sorted_offsets[i].offset;
632 67fd6849 2022-02-13 stsp if (o == offset) {
633 67fd6849 2022-02-13 stsp *idx = packidx->sorted_offsets[i].idx;
635 67fd6849 2022-02-13 stsp } else if (offset > o)
636 67fd6849 2022-02-13 stsp left = i + 1;
637 67fd6849 2022-02-13 stsp else if (offset < o)
638 67fd6849 2022-02-13 stsp right = i - 1;
642 67fd6849 2022-02-13 stsp return NULL;
645 67fd6849 2022-02-13 stsp const struct got_error *
646 67fd6849 2022-02-13 stsp got_packidx_get_object_id(struct got_object_id *id,
647 67fd6849 2022-02-13 stsp struct got_packidx *packidx, int idx)
649 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
650 67fd6849 2022-02-13 stsp struct got_packidx_object_id *oid;
652 67fd6849 2022-02-13 stsp if (idx < 0 || idx >= totobj)
653 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_NO_OBJ);
655 67fd6849 2022-02-13 stsp oid = &packidx->hdr.sorted_ids[idx];
656 67fd6849 2022-02-13 stsp memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
657 67fd6849 2022-02-13 stsp return NULL;
660 67fd6849 2022-02-13 stsp const struct got_error *
661 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
662 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
664 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
665 4277420a 2019-06-29 stsp u_int8_t id0;
666 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
667 4277420a 2019-06-29 stsp char hex[3];
668 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
669 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
670 404bde06 2022-01-03 stsp uint32_t i = 0;
672 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
674 4277420a 2019-06-29 stsp if (prefix_len < 2)
675 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
677 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
678 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
679 4277420a 2019-06-29 stsp hex[2] = '\0';
680 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
681 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
683 404bde06 2022-01-03 stsp if (id0 > 0)
684 404bde06 2022-01-03 stsp i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
685 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
686 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
687 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
688 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
691 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
692 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
694 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
695 4277420a 2019-06-29 stsp if (cmp < 0) {
696 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
698 4277420a 2019-06-29 stsp } else if (cmp > 0)
701 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
704 dd88155e 2019-06-29 stsp memcpy(qid->id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
705 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
707 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
711 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
712 dd88155e 2019-06-29 stsp return err;
715 b4f37570 2021-06-19 stsp static const struct got_error *
716 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
718 96732e0b 2018-11-11 stsp const struct got_error *err = NULL;
720 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
721 876c234b 2018-09-10 stsp return NULL;
723 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
725 96732e0b 2018-11-11 stsp return err;
726 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
727 08578a35 2021-01-22 stsp if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
728 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
729 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
730 876c234b 2018-09-10 stsp free(pack->privsep_child);
731 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
732 876c234b 2018-09-10 stsp return err;
735 d7464085 2018-07-09 stsp const struct got_error *
736 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
738 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
740 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
741 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
742 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
743 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
744 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
745 8b2180d4 2018-04-26 stsp pack->fd = -1;
746 4589e373 2018-03-17 stsp free(pack->path_packfile);
747 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
748 4589e373 2018-03-17 stsp pack->filesize = 0;
749 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
750 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
751 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
754 c7fe698a 2018-01-23 stsp return err;
757 668a20f6 2020-03-18 stsp const struct got_error *
758 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
759 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
761 a487c1d0 2018-01-14 stsp uint8_t t = 0;
762 a487c1d0 2018-01-14 stsp uint64_t s = 0;
763 a487c1d0 2018-01-14 stsp uint8_t sizeN;
764 d7464085 2018-07-09 stsp size_t mapoff = 0;
769 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
770 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
772 d7464085 2018-07-09 stsp if (pack->map) {
773 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
775 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
776 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
780 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
782 a487c1d0 2018-01-14 stsp return got_error(GOT_ERR_NO_SPACE);
784 d7464085 2018-07-09 stsp if (pack->map) {
785 3976db15 2022-01-10 stsp if (mapoff + sizeof(sizeN) >= pack->filesize)
786 3976db15 2022-01-10 stsp return got_error(GOT_ERR_BAD_PACKFILE);
787 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
788 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
790 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
792 638f9024 2019-05-13 stsp return got_error_from_errno("read");
793 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
794 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
796 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
798 a1fd68d8 2018-01-12 stsp if (i == 0) {
799 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
800 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
801 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
803 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
804 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
807 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
811 a487c1d0 2018-01-14 stsp return NULL;
814 a1fd68d8 2018-01-12 stsp static const struct got_error *
815 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
816 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
818 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
819 6ccb713b 2018-01-19 stsp if (*obj == NULL)
820 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
822 6ccb713b 2018-01-19 stsp (*obj)->type = type;
823 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
824 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
825 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
826 6ccb713b 2018-01-19 stsp (*obj)->size = size;
827 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
828 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
830 b107e67f 2018-01-19 stsp return NULL;
833 b107e67f 2018-01-19 stsp static const struct got_error *
834 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
835 d7464085 2018-07-09 stsp off_t delta_offset)
837 b107e67f 2018-01-19 stsp int64_t o = 0;
838 b107e67f 2018-01-19 stsp uint8_t offN;
841 a0de39f3 2019-08-09 stsp *offset = 0;
845 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
847 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
849 d7464085 2018-07-09 stsp if (pack->map) {
850 d7464085 2018-07-09 stsp size_t mapoff;
851 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
852 3976db15 2022-01-10 stsp if (mapoff + sizeof(offN) >= pack->filesize)
853 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
854 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
857 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
859 638f9024 2019-05-13 stsp return got_error_from_errno("read");
860 d7464085 2018-07-09 stsp if (n != sizeof(offN))
861 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
863 d7464085 2018-07-09 stsp *len += sizeof(offN);
865 b107e67f 2018-01-19 stsp if (i == 0)
866 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
870 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
873 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
875 b107e67f 2018-01-19 stsp *offset = o;
876 6ccb713b 2018-01-19 stsp return NULL;
879 668a20f6 2020-03-18 stsp const struct got_error *
880 c0df5966 2021-12-31 stsp got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
881 c0df5966 2021-12-31 stsp struct got_pack *pack, off_t offset, int tslen)
883 96f5e8b3 2018-01-23 stsp const struct got_error *err;
884 b107e67f 2018-01-19 stsp int64_t negoffset;
885 b107e67f 2018-01-19 stsp size_t negofflen;
889 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
890 d7464085 2018-07-09 stsp offset + tslen);
892 b107e67f 2018-01-19 stsp return err;
894 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
895 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
896 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
897 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
899 d7464085 2018-07-09 stsp *len = negofflen;
900 96f5e8b3 2018-01-23 stsp return NULL;
903 0e22967e 2018-02-11 stsp static const struct got_error *
904 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
905 42c69117 2019-11-10 stsp size_t delta_data_offset, struct got_pack *pack)
907 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
909 42c69117 2019-11-10 stsp if (pack->map) {
910 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
911 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
912 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
913 2e5a6fad 2020-03-18 stsp NULL, NULL, pack->map, delta_data_offset,
914 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
916 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
917 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
918 3ab5e33c 2020-03-18 stsp err = got_inflate_to_mem_fd(delta_buf, delta_len, NULL,
919 55fdd257 2020-03-18 stsp NULL, 0, pack->fd);
921 42c69117 2019-11-10 stsp return err;
924 96f5e8b3 2018-01-23 stsp static const struct got_error *
925 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
926 42c69117 2019-11-10 stsp int delta_type, size_t delta_size, size_t delta_data_offset)
928 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
930 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
931 42c69117 2019-11-10 stsp delta_data_offset);
932 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
933 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
934 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
935 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
936 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
937 bdd2fbb3 2018-02-11 stsp return NULL;
940 bdd2fbb3 2018-02-11 stsp static const struct got_error *
941 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
942 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
943 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
946 a3500804 2018-01-23 stsp const struct got_error *err;
947 c3703302 2018-01-23 stsp off_t base_offset;
948 c3703302 2018-01-23 stsp uint8_t base_type;
949 c3703302 2018-01-23 stsp uint64_t base_size;
950 c3703302 2018-01-23 stsp size_t base_tslen;
951 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
952 42c69117 2019-11-10 stsp size_t consumed;
954 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
955 d7464085 2018-07-09 stsp delta_offset, tslen);
957 a3500804 2018-01-23 stsp return err;
959 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
960 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
961 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
963 d7464085 2018-07-09 stsp if (pack->map == NULL) {
964 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
965 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
966 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
969 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
970 42c69117 2019-11-10 stsp delta_data_offset);
972 1a35c1bc 2019-05-22 stsp return err;
974 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
975 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
976 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
978 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
979 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
981 1a35c1bc 2019-05-22 stsp return err;
983 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
984 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
987 c4330eff 2021-06-22 stsp const struct got_error *
988 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
989 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
991 c4330eff 2021-06-22 stsp if (pack->map) {
992 c4330eff 2021-06-22 stsp size_t mapoff = delta_offset + tslen;
993 3976db15 2022-01-10 stsp if (mapoff + sizeof(*id) >= pack->filesize)
994 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
995 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
998 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1000 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1001 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1002 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1005 c4330eff 2021-06-22 stsp return NULL;
1008 c3703302 2018-01-23 stsp static const struct got_error *
1009 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1010 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1011 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1013 6b9c9673 2018-01-23 stsp const struct got_error *err;
1014 6b9c9673 2018-01-23 stsp struct got_object_id id;
1016 6b9c9673 2018-01-23 stsp off_t base_offset;
1017 6b9c9673 2018-01-23 stsp uint8_t base_type;
1018 6b9c9673 2018-01-23 stsp uint64_t base_size;
1019 6b9c9673 2018-01-23 stsp size_t base_tslen;
1020 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1022 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1023 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1025 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1027 c4330eff 2021-06-22 stsp return err;
1028 d7464085 2018-07-09 stsp if (pack->map) {
1029 c4330eff 2021-06-22 stsp delta_data_offset = delta_offset + tslen + sizeof(id);
1031 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1032 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1033 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1036 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1037 42c69117 2019-11-10 stsp delta_data_offset);
1039 1a35c1bc 2019-05-22 stsp return err;
1041 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1042 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1043 1a35c1bc 2019-05-22 stsp if (idx == -1)
1044 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1046 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1047 1a35c1bc 2019-05-22 stsp if (base_offset == (uint64_t)-1)
1048 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1050 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1051 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1053 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1054 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1056 1a35c1bc 2019-05-22 stsp return err;
1058 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1059 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1062 668a20f6 2020-03-18 stsp const struct got_error *
1063 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1064 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1065 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1067 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1069 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1070 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1072 c3703302 2018-01-23 stsp switch (delta_type) {
1073 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1074 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1075 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1076 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1077 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1078 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1079 42c69117 2019-11-10 stsp delta_size, 0);
1081 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1082 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1083 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1085 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1086 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1087 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1090 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1093 96f5e8b3 2018-01-23 stsp return err;
1096 96f5e8b3 2018-01-23 stsp static const struct got_error *
1097 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1098 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1099 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1101 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1102 96f5e8b3 2018-01-23 stsp int resolved_type;
1104 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1105 b107e67f 2018-01-19 stsp if (*obj == NULL)
1106 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1108 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1109 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1110 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1111 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1112 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1114 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1115 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1116 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1117 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1119 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1120 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1121 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1123 96f5e8b3 2018-01-23 stsp goto done;
1125 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1127 96f5e8b3 2018-01-23 stsp goto done;
1128 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1130 96f5e8b3 2018-01-23 stsp if (err) {
1131 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1132 96f5e8b3 2018-01-23 stsp *obj = NULL;
1134 96f5e8b3 2018-01-23 stsp return err;
1137 2090a03d 2018-09-09 stsp const struct got_error *
1138 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1139 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1141 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1142 a1fd68d8 2018-01-12 stsp off_t offset;
1143 6c00b545 2018-01-17 stsp uint8_t type;
1144 6c00b545 2018-01-17 stsp uint64_t size;
1145 3ee5fc21 2018-01-17 stsp size_t tslen;
1147 6c00b545 2018-01-17 stsp *obj = NULL;
1149 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1150 a1fd68d8 2018-01-12 stsp if (offset == (uint64_t)-1)
1151 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1153 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1154 668a20f6 2020-03-18 stsp pack, offset);
1156 35c73765 2018-09-09 stsp return err;
1158 6c00b545 2018-01-17 stsp switch (type) {
1159 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1160 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1161 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1162 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1163 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1164 5f25cc85 2019-11-26 stsp size, idx);
1166 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1167 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1168 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1169 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1172 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1176 3ee5fc21 2018-01-17 stsp return err;
1179 d582f26c 2020-03-18 stsp const struct got_error *
1180 c0df5966 2021-12-31 stsp got_pack_get_delta_chain_max_size(uint64_t *max_size,
1181 c0df5966 2021-12-31 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1183 22484865 2018-03-13 stsp struct got_delta *delta;
1184 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1186 22484865 2018-03-13 stsp *max_size = 0;
1187 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1188 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1189 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1190 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1191 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1192 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1193 22484865 2018-03-13 stsp const struct got_error *err;
1194 42c69117 2019-11-10 stsp uint8_t *delta_buf;
1195 42c69117 2019-11-10 stsp size_t delta_len;
1196 ab2f42e7 2019-11-10 stsp int cached = 1;
1198 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1199 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1200 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1201 ab2f42e7 2019-11-10 stsp cached = 0;
1202 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1203 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1205 ab2f42e7 2019-11-10 stsp return err;
1206 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1207 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1208 ab2f42e7 2019-11-10 stsp if (err == NULL)
1209 ab2f42e7 2019-11-10 stsp cached = 1;
1210 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1211 ab2f42e7 2019-11-10 stsp free(delta_buf);
1212 ab2f42e7 2019-11-10 stsp return err;
1215 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1216 42c69117 2019-11-10 stsp delta_buf, delta_len);
1217 ab2f42e7 2019-11-10 stsp if (!cached)
1218 ab2f42e7 2019-11-10 stsp free(delta_buf);
1220 22484865 2018-03-13 stsp return err;
1222 22484865 2018-03-13 stsp base_size = delta->size;
1223 22484865 2018-03-13 stsp if (base_size > *max_size)
1224 22484865 2018-03-13 stsp *max_size = base_size;
1225 22484865 2018-03-13 stsp if (result_size > *max_size)
1226 22484865 2018-03-13 stsp *max_size = result_size;
1229 9feb4ff2 2018-03-14 stsp return NULL;
1232 ac544f8c 2019-01-13 stsp const struct got_error *
1233 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1234 42c69117 2019-11-10 stsp struct got_pack *pack)
1236 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1237 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1239 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1242 668a20f6 2020-03-18 stsp const struct got_error *
1243 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1244 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1245 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1247 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1248 6691714a 2018-01-23 stsp struct got_delta *delta;
1249 42c69117 2019-11-10 stsp uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1250 42c69117 2019-11-10 stsp size_t base_bufsz = 0, accum_size = 0, delta_len;
1251 22484865 2018-03-13 stsp uint64_t max_size;
1252 6691714a 2018-01-23 stsp int n = 0;
1254 b29656e2 2018-03-16 stsp *result_size = 0;
1256 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1257 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1259 8628c62d 2018-03-15 stsp /* We process small enough files entirely in memory for speed. */
1260 d582f26c 2020-03-18 stsp err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1262 22484865 2018-03-13 stsp return err;
1263 8628c62d 2018-03-15 stsp if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1264 8628c62d 2018-03-15 stsp accum_buf = malloc(max_size);
1265 8628c62d 2018-03-15 stsp if (accum_buf == NULL)
1266 638f9024 2019-05-13 stsp return got_error_from_errno("malloc");
1267 3840f4c9 2018-09-12 stsp base_file = NULL;
1268 3840f4c9 2018-09-12 stsp accum_file = NULL;
1270 3bfc2334 2022-01-11 stsp if (fseeko(base_file, 0L, SEEK_SET) == -1)
1271 3bfc2334 2022-01-11 stsp return got_error_from_errno("fseeko");
1272 3bfc2334 2022-01-11 stsp if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1273 3bfc2334 2022-01-11 stsp return got_error_from_errno("fseeko");
1276 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1277 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1278 ab2f42e7 2019-11-10 stsp int cached = 1;
1279 a6b158cc 2018-02-11 stsp if (n == 0) {
1280 34fca9c3 2018-11-11 stsp size_t mapoff;
1281 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1283 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1284 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1285 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1286 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1287 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1288 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1289 72eb3431 2018-04-01 stsp goto done;
1292 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1293 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1294 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1295 0c048b15 2018-04-27 stsp goto done;
1297 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1298 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1300 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1301 d7464085 2018-07-09 stsp goto done;
1304 d7464085 2018-07-09 stsp if (base_file) {
1305 d7464085 2018-07-09 stsp if (pack->map) {
1306 d7464085 2018-07-09 stsp mapoff = (size_t)delta_data_offset;
1307 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1308 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1309 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1310 4788f1ce 2020-03-18 stsp base_file);
1312 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1313 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1314 4788f1ce 2020-03-18 stsp base_file);
1316 d7464085 2018-07-09 stsp if (pack->map) {
1317 d7464085 2018-07-09 stsp mapoff = (size_t)delta_data_offset;
1318 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1319 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1320 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1321 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1323 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1324 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1325 55fdd257 2020-03-18 stsp pack->fd);
1328 a6b158cc 2018-02-11 stsp goto done;
1330 8628c62d 2018-03-15 stsp if (base_file)
1331 8628c62d 2018-03-15 stsp rewind(base_file);
1335 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1336 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1337 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1338 ab2f42e7 2019-11-10 stsp cached = 0;
1339 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1340 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1342 ab2f42e7 2019-11-10 stsp goto done;
1343 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1344 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1345 ab2f42e7 2019-11-10 stsp if (err == NULL)
1346 ab2f42e7 2019-11-10 stsp cached = 1;
1347 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1348 ab2f42e7 2019-11-10 stsp free(delta_buf);
1349 ab2f42e7 2019-11-10 stsp goto done;
1352 8628c62d 2018-03-15 stsp if (base_buf) {
1353 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1354 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1355 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1358 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1359 42c69117 2019-11-10 stsp delta_len,
1360 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1361 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1362 b29656e2 2018-03-16 stsp &accum_size);
1364 ab2f42e7 2019-11-10 stsp if (!cached)
1365 ab2f42e7 2019-11-10 stsp free(delta_buf);
1367 6691714a 2018-01-23 stsp goto done;
1369 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1370 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1371 8628c62d 2018-03-15 stsp if (base_buf) {
1372 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1374 34fca9c3 2018-11-11 stsp * Base buffer switches roles with accumulation
1375 34fca9c3 2018-11-11 stsp * buffer. Ensure it can hold the largest
1376 34fca9c3 2018-11-11 stsp * result in the delta chain. The initial
1377 34fca9c3 2018-11-11 stsp * allocation might have been smaller.
1379 34fca9c3 2018-11-11 stsp if (base_bufsz < max_size) {
1380 34fca9c3 2018-11-11 stsp uint8_t *p;
1381 34fca9c3 2018-11-11 stsp p = reallocarray(base_buf, 1, max_size);
1382 34fca9c3 2018-11-11 stsp if (p == NULL) {
1383 638f9024 2019-05-13 stsp err = got_error_from_errno(
1384 230a42bd 2019-05-11 jcs "reallocarray");
1385 34fca9c3 2018-11-11 stsp goto done;
1387 34fca9c3 2018-11-11 stsp base_buf = p;
1388 34fca9c3 2018-11-11 stsp base_bufsz = max_size;
1390 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1391 8628c62d 2018-03-15 stsp base_buf = tmp;
1393 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1394 8628c62d 2018-03-15 stsp accum_file = base_file;
1395 8628c62d 2018-03-15 stsp base_file = tmp;
1396 8628c62d 2018-03-15 stsp rewind(base_file);
1397 8628c62d 2018-03-15 stsp rewind(accum_file);
1403 8628c62d 2018-03-15 stsp free(base_buf);
1404 8628c62d 2018-03-15 stsp if (accum_buf) {
1405 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1406 8628c62d 2018-03-15 stsp free(accum_buf);
1407 8628c62d 2018-03-15 stsp if (len != accum_size)
1408 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1410 6691714a 2018-01-23 stsp rewind(outfile);
1411 b29656e2 2018-03-16 stsp if (err == NULL)
1412 b29656e2 2018-03-16 stsp *result_size = accum_size;
1413 e0ab43e7 2018-03-16 stsp return err;
1416 668a20f6 2020-03-18 stsp const struct got_error *
1417 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1418 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1420 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1421 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1422 42c69117 2019-11-10 stsp uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1423 42c69117 2019-11-10 stsp size_t base_bufsz = 0, accum_size = 0, delta_len;
1424 e0ab43e7 2018-03-16 stsp uint64_t max_size;
1425 e0ab43e7 2018-03-16 stsp int n = 0;
1427 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1428 e0ab43e7 2018-03-16 stsp *outlen = 0;
1430 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1431 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1433 d582f26c 2020-03-18 stsp err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1435 e0ab43e7 2018-03-16 stsp return err;
1436 e0ab43e7 2018-03-16 stsp accum_buf = malloc(max_size);
1437 e0ab43e7 2018-03-16 stsp if (accum_buf == NULL)
1438 638f9024 2019-05-13 stsp return got_error_from_errno("malloc");
1440 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1441 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1442 ab2f42e7 2019-11-10 stsp int cached = 1;
1443 e0ab43e7 2018-03-16 stsp if (n == 0) {
1444 0c048b15 2018-04-27 stsp size_t delta_data_offset;
1446 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1447 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1448 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1449 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1450 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1451 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1452 72eb3431 2018-04-01 stsp goto done;
1455 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1456 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1457 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1458 0c048b15 2018-04-27 stsp goto done;
1460 d7464085 2018-07-09 stsp if (pack->map) {
1461 d7464085 2018-07-09 stsp size_t mapoff = (size_t)delta_data_offset;
1462 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1463 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1464 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1466 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1468 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1469 d7464085 2018-07-09 stsp goto done;
1471 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1472 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1473 55fdd257 2020-03-18 stsp pack->fd);
1476 d7464085 2018-07-09 stsp goto done;
1481 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1482 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1483 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1484 ab2f42e7 2019-11-10 stsp cached = 0;
1485 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1486 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1488 ab2f42e7 2019-11-10 stsp goto done;
1489 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1490 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1491 ab2f42e7 2019-11-10 stsp if (err == NULL)
1492 ab2f42e7 2019-11-10 stsp cached = 1;
1493 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1494 ab2f42e7 2019-11-10 stsp free(delta_buf);
1495 ab2f42e7 2019-11-10 stsp goto done;
1498 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1499 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1500 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1501 ab2f42e7 2019-11-10 stsp if (!cached)
1502 ab2f42e7 2019-11-10 stsp free(delta_buf);
1505 e0ab43e7 2018-03-16 stsp goto done;
1507 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1508 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1509 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1511 34fca9c3 2018-11-11 stsp * Base buffer switches roles with accumulation buffer.
1512 34fca9c3 2018-11-11 stsp * Ensure it can hold the largest result in the delta
1513 34fca9c3 2018-11-11 stsp * chain. Initial allocation might have been smaller.
1515 34fca9c3 2018-11-11 stsp if (base_bufsz < max_size) {
1516 34fca9c3 2018-11-11 stsp uint8_t *p;
1517 34fca9c3 2018-11-11 stsp p = reallocarray(base_buf, 1, max_size);
1518 34fca9c3 2018-11-11 stsp if (p == NULL) {
1519 638f9024 2019-05-13 stsp err = got_error_from_errno(
1520 230a42bd 2019-05-11 jcs "reallocarray");
1521 34fca9c3 2018-11-11 stsp goto done;
1523 34fca9c3 2018-11-11 stsp base_buf = p;
1524 34fca9c3 2018-11-11 stsp base_bufsz = max_size;
1526 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1527 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1532 e0ab43e7 2018-03-16 stsp free(base_buf);
1533 e0ab43e7 2018-03-16 stsp if (err) {
1534 e0ab43e7 2018-03-16 stsp free(accum_buf);
1535 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1536 e0ab43e7 2018-03-16 stsp *outlen = 0;
1538 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1539 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1541 efd2a263 2018-01-19 stsp return err;
1544 3ee5fc21 2018-01-17 stsp const struct got_error *
1545 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1546 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1548 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1550 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1551 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1553 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1554 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1555 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1557 d7464085 2018-07-09 stsp if (pack->map) {
1558 d7464085 2018-07-09 stsp size_t mapoff = (size_t)obj->pack_offset;
1559 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1560 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1563 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1564 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1565 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1566 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1569 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1570 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1572 a1fd68d8 2018-01-12 stsp return err;
1575 e0ab43e7 2018-03-16 stsp const struct got_error *
1576 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1577 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1579 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1581 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1582 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1584 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1585 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1586 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1587 d7464085 2018-07-09 stsp if (pack->map) {
1588 d7464085 2018-07-09 stsp size_t mapoff = (size_t)obj->pack_offset;
1589 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1590 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1592 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1593 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1594 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1595 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1598 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1601 e0ab43e7 2018-03-16 stsp return err;
1604 67fd6849 2022-02-13 stsp const struct got_error *
1605 67fd6849 2022-02-13 stsp got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1606 67fd6849 2022-02-13 stsp off_t *delta_offset, off_t *base_offset, struct got_object_id *base_id,
1607 67fd6849 2022-02-13 stsp uint64_t *base_size, uint64_t *result_size, struct got_pack *pack,
1608 67fd6849 2022-02-13 stsp struct got_packidx *packidx, int idx)
1610 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
1611 67fd6849 2022-02-13 stsp off_t offset;
1612 67fd6849 2022-02-13 stsp uint8_t type;
1613 67fd6849 2022-02-13 stsp uint64_t size;
1614 67fd6849 2022-02-13 stsp size_t tslen, delta_hdrlen;
1616 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1617 67fd6849 2022-02-13 stsp *delta_size = 0;
1618 67fd6849 2022-02-13 stsp *delta_offset = 0;
1619 67fd6849 2022-02-13 stsp *base_offset = 0;
1620 67fd6849 2022-02-13 stsp *base_size = 0;
1621 67fd6849 2022-02-13 stsp *result_size = 0;
1623 67fd6849 2022-02-13 stsp offset = got_packidx_get_object_offset(packidx, idx);
1624 67fd6849 2022-02-13 stsp if (offset == (uint64_t)-1)
1625 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1627 67fd6849 2022-02-13 stsp if (offset >= pack->filesize)
1628 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1630 67fd6849 2022-02-13 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1631 67fd6849 2022-02-13 stsp pack, offset);
1633 67fd6849 2022-02-13 stsp return err;
1635 67fd6849 2022-02-13 stsp if (tslen + size < tslen || offset + size < size ||
1636 67fd6849 2022-02-13 stsp tslen + offset < tslen)
1637 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1639 67fd6849 2022-02-13 stsp switch (type) {
1640 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1641 67fd6849 2022-02-13 stsp err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1642 67fd6849 2022-02-13 stsp pack, offset, tslen);
1644 67fd6849 2022-02-13 stsp return err;
1646 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_REF_DELTA:
1647 67fd6849 2022-02-13 stsp err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1649 67fd6849 2022-02-13 stsp return err;
1650 67fd6849 2022-02-13 stsp delta_hdrlen = SHA1_DIGEST_LENGTH;
1653 67fd6849 2022-02-13 stsp return got_error_fmt(GOT_ERR_OBJ_TYPE,
1654 67fd6849 2022-02-13 stsp "non-delta object type %d found at offset %llu",
1655 67fd6849 2022-02-13 stsp type, offset);
1658 67fd6849 2022-02-13 stsp if (tslen + delta_hdrlen < delta_hdrlen ||
1659 67fd6849 2022-02-13 stsp offset + delta_hdrlen < delta_hdrlen)
1660 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_DELTA);
1662 67fd6849 2022-02-13 stsp err = read_delta_data(delta_buf, delta_size,
1663 67fd6849 2022-02-13 stsp offset + tslen + delta_hdrlen, pack);
1665 67fd6849 2022-02-13 stsp return err;
1667 67fd6849 2022-02-13 stsp if (*delta_size != size) {
1668 67fd6849 2022-02-13 stsp err = got_error(GOT_ERR_BAD_DELTA);
1669 67fd6849 2022-02-13 stsp goto done;
1672 67fd6849 2022-02-13 stsp err = got_delta_get_sizes(base_size, result_size, *delta_buf, size);
1674 67fd6849 2022-02-13 stsp goto done;
1676 67fd6849 2022-02-13 stsp *delta_offset = offset;
1678 67fd6849 2022-02-13 stsp if (err) {
1679 67fd6849 2022-02-13 stsp free(*delta_buf);
1680 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1682 67fd6849 2022-02-13 stsp return err;