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 511a516b 2018-05-19 stsp #include "got_opentemp.h"
39 324d37e7 2019-05-11 stsp #include "got_path.h"
41 718b3ab0 2018-03-17 stsp #include "got_lib_sha1.h"
42 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
43 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
44 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
45 718b3ab0 2018-03-17 stsp #include "got_lib_object.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 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
76 0ebaf008 2018-01-10 stsp SHA1_CTX ctx;
77 0ebaf008 2018-01-10 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
78 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
82 0ebaf008 2018-01-10 stsp SHA1Init(&ctx);
84 57b35b75 2018-06-22 stsp h = &p->hdr;
86 57b35b75 2018-06-22 stsp remain = p->len;
88 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
89 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
93 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
95 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
96 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
97 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
100 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
101 b1317e77 2019-09-22 stsp if (n < 0) {
102 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
104 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
105 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
109 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
110 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
113 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
114 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
116 0cb74cf4 2018-07-08 stsp if (verify)
117 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
119 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
120 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
123 fc79a48d 2018-07-09 stsp if (p->map)
124 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
126 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
127 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
128 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
131 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
132 c6368c2e 2019-10-11 stsp if (n < 0) {
133 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
135 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
136 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
140 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
141 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
144 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
145 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
147 0cb74cf4 2018-07-08 stsp if (verify)
148 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
150 57b35b75 2018-06-22 stsp len_fanout =
151 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
152 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
153 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
156 fc79a48d 2018-07-09 stsp if (p->map)
157 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
159 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
160 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
161 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
164 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
165 c6368c2e 2019-10-11 stsp if (n < 0) {
166 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
168 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
169 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
173 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
176 0cb74cf4 2018-07-08 stsp if (verify)
177 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
178 57b35b75 2018-06-22 stsp offset += len_fanout;
179 57b35b75 2018-06-22 stsp remain -= len_fanout;
181 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
182 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
183 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
184 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
187 fc79a48d 2018-07-09 stsp if (p->map)
188 fc79a48d 2018-07-09 stsp h->sorted_ids =
189 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
191 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
192 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
193 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
196 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
198 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
199 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
200 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
204 0cb74cf4 2018-07-08 stsp if (verify)
205 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
206 57b35b75 2018-06-22 stsp offset += len_ids;
207 57b35b75 2018-06-22 stsp remain -= len_ids;
209 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
210 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
213 fc79a48d 2018-07-09 stsp if (p->map)
214 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
216 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
217 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
218 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
221 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
223 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
224 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
225 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
229 0cb74cf4 2018-07-08 stsp if (verify)
230 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
231 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
232 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
234 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
235 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
238 fc79a48d 2018-07-09 stsp if (p->map)
239 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
241 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
242 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
243 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
246 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
248 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
249 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
250 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
254 0cb74cf4 2018-07-08 stsp if (verify)
255 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->offsets,
256 0cb74cf4 2018-07-08 stsp 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 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t*)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 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
327 0cb74cf4 2018-07-08 stsp SHA1Final(sha1, &ctx);
328 0cb74cf4 2018-07-08 stsp if (memcmp(h->trailer->packidx_sha1, sha1,
329 0cb74cf4 2018-07-08 stsp SHA1_DIGEST_LENGTH) != 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 e7ae0baf 2021-12-31 stsp 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);
378 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
379 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
380 817c5a18 2018-09-09 stsp close(p->fd);
384 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
385 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
386 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
387 817c5a18 2018-09-09 stsp close(p->fd);
392 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
393 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
394 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
398 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
399 817c5a18 2018-09-09 stsp p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
400 3a11398b 2019-02-21 stsp if (p->map == MAP_FAILED) {
401 3a11398b 2019-02-21 stsp if (errno != ENOMEM) {
402 638f9024 2019-05-13 stsp err = got_error_from_errno("mmap");
405 817c5a18 2018-09-09 stsp p->map = NULL; /* fall back to read(2) */
409 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
413 1124fe40 2021-07-07 stsp got_packidx_close(p);
415 0a0a3048 2018-01-10 stsp *packidx = p;
416 1124fe40 2021-07-07 stsp free(pack_relpath);
417 0a0a3048 2018-01-10 stsp return err;
420 57b35b75 2018-06-22 stsp const struct got_error *
421 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
423 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
425 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
426 fc79a48d 2018-07-09 stsp if (packidx->map) {
427 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
428 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
434 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
435 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
436 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
437 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
439 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
440 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
441 0a0a3048 2018-01-10 stsp free(packidx);
443 57b35b75 2018-06-22 stsp return err;
446 509c9973 2021-04-10 stsp const struct got_error *
447 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
449 509c9973 2021-04-10 stsp size_t size;
451 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
452 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
453 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
454 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
456 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
457 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
458 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
460 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
461 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
462 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
463 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
465 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
466 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
468 509c9973 2021-04-10 stsp return NULL;
472 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
474 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
475 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
476 a1fd68d8 2018-01-12 stsp uint64_t loffset;
477 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
478 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
479 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
481 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
482 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
484 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
488 c0df5966 2021-12-31 stsp got_packidx_get_object_idx(struct got_packidx *packidx,
489 c0df5966 2021-12-31 stsp struct got_object_id *id)
491 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
492 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
493 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
495 a1fd68d8 2018-01-12 stsp if (id0 > 0)
496 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
498 40aeb19c 2018-06-22 stsp while (left <= right) {
499 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
500 40aeb19c 2018-06-22 stsp int i, cmp;
502 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
503 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
504 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
505 16dcbf91 2018-04-01 stsp if (cmp == 0)
507 40aeb19c 2018-06-22 stsp else if (cmp > 0)
508 40aeb19c 2018-06-22 stsp left = i + 1;
509 40aeb19c 2018-06-22 stsp else if (cmp < 0)
510 40aeb19c 2018-06-22 stsp right = i - 1;
516 876c234b 2018-09-10 stsp const struct got_error *
517 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
518 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
520 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
521 4277420a 2019-06-29 stsp u_int8_t id0;
522 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
523 4277420a 2019-06-29 stsp char hex[3];
524 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
525 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
526 404bde06 2022-01-03 stsp uint32_t i = 0;
528 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
530 4277420a 2019-06-29 stsp if (prefix_len < 2)
531 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
533 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
534 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
535 4277420a 2019-06-29 stsp hex[2] = '\0';
536 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
537 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
539 404bde06 2022-01-03 stsp if (id0 > 0)
540 404bde06 2022-01-03 stsp i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
541 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
542 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
543 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
544 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
547 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
548 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
550 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
551 4277420a 2019-06-29 stsp if (cmp < 0) {
552 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
554 4277420a 2019-06-29 stsp } else if (cmp > 0)
557 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
560 dd88155e 2019-06-29 stsp memcpy(qid->id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
561 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
563 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
567 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
568 dd88155e 2019-06-29 stsp return err;
571 b4f37570 2021-06-19 stsp static const struct got_error *
572 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
574 96732e0b 2018-11-11 stsp const struct got_error *err = NULL;
576 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
577 876c234b 2018-09-10 stsp return NULL;
579 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
581 96732e0b 2018-11-11 stsp return err;
582 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
583 08578a35 2021-01-22 stsp if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
584 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
585 cc2a8ef4 2021-06-19 tracey imsg_clear(pack->privsep_child->ibuf);
586 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
587 876c234b 2018-09-10 stsp free(pack->privsep_child);
588 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
589 876c234b 2018-09-10 stsp return err;
592 d7464085 2018-07-09 stsp const struct got_error *
593 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
595 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
597 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
598 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
599 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
600 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
601 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
602 8b2180d4 2018-04-26 stsp pack->fd = -1;
603 4589e373 2018-03-17 stsp free(pack->path_packfile);
604 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
605 4589e373 2018-03-17 stsp pack->filesize = 0;
606 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
607 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
608 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
611 c7fe698a 2018-01-23 stsp return err;
614 668a20f6 2020-03-18 stsp const struct got_error *
615 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
616 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
618 a487c1d0 2018-01-14 stsp uint8_t t = 0;
619 a487c1d0 2018-01-14 stsp uint64_t s = 0;
620 a487c1d0 2018-01-14 stsp uint8_t sizeN;
621 d7464085 2018-07-09 stsp size_t mapoff = 0;
626 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
627 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
629 d7464085 2018-07-09 stsp if (pack->map) {
630 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
632 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
633 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
637 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
639 a487c1d0 2018-01-14 stsp return got_error(GOT_ERR_NO_SPACE);
641 d7464085 2018-07-09 stsp if (pack->map) {
642 3976db15 2022-01-10 stsp if (mapoff + sizeof(sizeN) >= pack->filesize)
643 3976db15 2022-01-10 stsp return got_error(GOT_ERR_BAD_PACKFILE);
644 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
645 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
647 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
649 638f9024 2019-05-13 stsp return got_error_from_errno("read");
650 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
651 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
653 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
655 a1fd68d8 2018-01-12 stsp if (i == 0) {
656 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
657 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
658 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
660 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
661 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
664 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
668 a487c1d0 2018-01-14 stsp return NULL;
671 a1fd68d8 2018-01-12 stsp static const struct got_error *
672 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
673 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
675 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
676 6ccb713b 2018-01-19 stsp if (*obj == NULL)
677 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
679 6ccb713b 2018-01-19 stsp (*obj)->type = type;
680 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
681 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
682 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
683 6ccb713b 2018-01-19 stsp (*obj)->size = size;
684 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
685 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
687 b107e67f 2018-01-19 stsp return NULL;
690 b107e67f 2018-01-19 stsp static const struct got_error *
691 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
692 d7464085 2018-07-09 stsp off_t delta_offset)
694 b107e67f 2018-01-19 stsp int64_t o = 0;
695 b107e67f 2018-01-19 stsp uint8_t offN;
698 a0de39f3 2019-08-09 stsp *offset = 0;
702 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
704 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
706 d7464085 2018-07-09 stsp if (pack->map) {
707 d7464085 2018-07-09 stsp size_t mapoff;
708 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
709 3976db15 2022-01-10 stsp if (mapoff + sizeof(offN) >= pack->filesize)
710 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
711 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
714 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
716 638f9024 2019-05-13 stsp return got_error_from_errno("read");
717 d7464085 2018-07-09 stsp if (n != sizeof(offN))
718 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
720 d7464085 2018-07-09 stsp *len += sizeof(offN);
722 b107e67f 2018-01-19 stsp if (i == 0)
723 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
727 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
730 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
732 b107e67f 2018-01-19 stsp *offset = o;
733 6ccb713b 2018-01-19 stsp return NULL;
736 668a20f6 2020-03-18 stsp const struct got_error *
737 c0df5966 2021-12-31 stsp got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
738 c0df5966 2021-12-31 stsp struct got_pack *pack, off_t offset, int tslen)
740 96f5e8b3 2018-01-23 stsp const struct got_error *err;
741 b107e67f 2018-01-19 stsp int64_t negoffset;
742 b107e67f 2018-01-19 stsp size_t negofflen;
746 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
747 d7464085 2018-07-09 stsp offset + tslen);
749 b107e67f 2018-01-19 stsp return err;
751 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
752 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
753 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
754 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
756 d7464085 2018-07-09 stsp *len = negofflen;
757 96f5e8b3 2018-01-23 stsp return NULL;
760 0e22967e 2018-02-11 stsp static const struct got_error *
761 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
762 42c69117 2019-11-10 stsp size_t delta_data_offset, struct got_pack *pack)
764 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
766 42c69117 2019-11-10 stsp if (pack->map) {
767 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
768 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
769 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
770 2e5a6fad 2020-03-18 stsp NULL, NULL, pack->map, delta_data_offset,
771 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
773 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
774 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
775 3ab5e33c 2020-03-18 stsp err = got_inflate_to_mem_fd(delta_buf, delta_len, NULL,
776 55fdd257 2020-03-18 stsp NULL, 0, pack->fd);
778 42c69117 2019-11-10 stsp return err;
781 96f5e8b3 2018-01-23 stsp static const struct got_error *
782 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
783 42c69117 2019-11-10 stsp int delta_type, size_t delta_size, size_t delta_data_offset)
785 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
787 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
788 42c69117 2019-11-10 stsp delta_data_offset);
789 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
790 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
791 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
792 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
793 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
794 bdd2fbb3 2018-02-11 stsp return NULL;
797 bdd2fbb3 2018-02-11 stsp static const struct got_error *
798 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
799 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
800 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
803 a3500804 2018-01-23 stsp const struct got_error *err;
804 c3703302 2018-01-23 stsp off_t base_offset;
805 c3703302 2018-01-23 stsp uint8_t base_type;
806 c3703302 2018-01-23 stsp uint64_t base_size;
807 c3703302 2018-01-23 stsp size_t base_tslen;
808 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
809 42c69117 2019-11-10 stsp size_t consumed;
811 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
812 d7464085 2018-07-09 stsp delta_offset, tslen);
814 a3500804 2018-01-23 stsp return err;
816 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
817 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
818 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
820 d7464085 2018-07-09 stsp if (pack->map == NULL) {
821 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
822 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
823 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
826 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
827 42c69117 2019-11-10 stsp delta_data_offset);
829 1a35c1bc 2019-05-22 stsp return err;
831 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
832 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
833 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
835 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
836 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
838 1a35c1bc 2019-05-22 stsp return err;
840 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
841 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
844 c4330eff 2021-06-22 stsp const struct got_error *
845 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
846 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
848 c4330eff 2021-06-22 stsp if (pack->map) {
849 c4330eff 2021-06-22 stsp size_t mapoff = delta_offset + tslen;
850 3976db15 2022-01-10 stsp if (mapoff + sizeof(*id) >= pack->filesize)
851 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
852 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
855 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
857 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
858 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
859 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
862 c4330eff 2021-06-22 stsp return NULL;
865 c3703302 2018-01-23 stsp static const struct got_error *
866 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
867 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
868 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
870 6b9c9673 2018-01-23 stsp const struct got_error *err;
871 6b9c9673 2018-01-23 stsp struct got_object_id id;
873 6b9c9673 2018-01-23 stsp off_t base_offset;
874 6b9c9673 2018-01-23 stsp uint8_t base_type;
875 6b9c9673 2018-01-23 stsp uint64_t base_size;
876 6b9c9673 2018-01-23 stsp size_t base_tslen;
877 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
879 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
880 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
882 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
884 c4330eff 2021-06-22 stsp return err;
885 d7464085 2018-07-09 stsp if (pack->map) {
886 c4330eff 2021-06-22 stsp delta_data_offset = delta_offset + tslen + sizeof(id);
888 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
889 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
890 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
893 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
894 42c69117 2019-11-10 stsp delta_data_offset);
896 1a35c1bc 2019-05-22 stsp return err;
898 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
899 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
900 1a35c1bc 2019-05-22 stsp if (idx == -1)
901 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
903 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
904 1a35c1bc 2019-05-22 stsp if (base_offset == (uint64_t)-1)
905 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
907 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
908 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
910 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
911 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
913 1a35c1bc 2019-05-22 stsp return err;
915 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
916 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
919 668a20f6 2020-03-18 stsp const struct got_error *
920 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
921 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
922 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
924 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
926 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
927 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
929 c3703302 2018-01-23 stsp switch (delta_type) {
930 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
931 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
932 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
933 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
934 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
935 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
936 42c69117 2019-11-10 stsp delta_size, 0);
938 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
939 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
940 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
942 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
943 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
944 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
947 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
950 96f5e8b3 2018-01-23 stsp return err;
953 96f5e8b3 2018-01-23 stsp static const struct got_error *
954 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
955 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
956 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
958 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
959 96f5e8b3 2018-01-23 stsp int resolved_type;
961 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
962 b107e67f 2018-01-19 stsp if (*obj == NULL)
963 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
965 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
966 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
967 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
968 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
969 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
971 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
972 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
973 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
974 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
976 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
977 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
978 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
982 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
985 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
988 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
989 96f5e8b3 2018-01-23 stsp *obj = NULL;
991 96f5e8b3 2018-01-23 stsp return err;
994 2090a03d 2018-09-09 stsp const struct got_error *
995 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
996 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
998 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
999 a1fd68d8 2018-01-12 stsp off_t offset;
1000 6c00b545 2018-01-17 stsp uint8_t type;
1001 6c00b545 2018-01-17 stsp uint64_t size;
1002 3ee5fc21 2018-01-17 stsp size_t tslen;
1004 6c00b545 2018-01-17 stsp *obj = NULL;
1006 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1007 a1fd68d8 2018-01-12 stsp if (offset == (uint64_t)-1)
1008 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1010 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1011 668a20f6 2020-03-18 stsp pack, offset);
1013 35c73765 2018-09-09 stsp return err;
1015 6c00b545 2018-01-17 stsp switch (type) {
1016 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1017 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1018 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1019 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1020 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1021 5f25cc85 2019-11-26 stsp size, idx);
1023 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1024 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1025 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1026 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1029 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1033 3ee5fc21 2018-01-17 stsp return err;
1036 d582f26c 2020-03-18 stsp const struct got_error *
1037 c0df5966 2021-12-31 stsp got_pack_get_delta_chain_max_size(uint64_t *max_size,
1038 c0df5966 2021-12-31 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1040 22484865 2018-03-13 stsp struct got_delta *delta;
1041 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1043 22484865 2018-03-13 stsp *max_size = 0;
1044 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1045 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1046 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1047 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1048 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1049 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1050 22484865 2018-03-13 stsp const struct got_error *err;
1051 42c69117 2019-11-10 stsp uint8_t *delta_buf;
1052 42c69117 2019-11-10 stsp size_t delta_len;
1053 ab2f42e7 2019-11-10 stsp int cached = 1;
1055 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1056 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1057 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1058 ab2f42e7 2019-11-10 stsp cached = 0;
1059 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1060 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1062 ab2f42e7 2019-11-10 stsp return err;
1063 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1064 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1065 ab2f42e7 2019-11-10 stsp if (err == NULL)
1066 ab2f42e7 2019-11-10 stsp cached = 1;
1067 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1068 ab2f42e7 2019-11-10 stsp free(delta_buf);
1069 ab2f42e7 2019-11-10 stsp return err;
1072 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1073 42c69117 2019-11-10 stsp delta_buf, delta_len);
1074 ab2f42e7 2019-11-10 stsp if (!cached)
1075 ab2f42e7 2019-11-10 stsp free(delta_buf);
1077 22484865 2018-03-13 stsp return err;
1079 22484865 2018-03-13 stsp base_size = delta->size;
1080 22484865 2018-03-13 stsp if (base_size > *max_size)
1081 22484865 2018-03-13 stsp *max_size = base_size;
1082 22484865 2018-03-13 stsp if (result_size > *max_size)
1083 22484865 2018-03-13 stsp *max_size = result_size;
1086 9feb4ff2 2018-03-14 stsp return NULL;
1089 ac544f8c 2019-01-13 stsp const struct got_error *
1090 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1091 42c69117 2019-11-10 stsp struct got_pack *pack)
1093 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1094 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1096 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1099 668a20f6 2020-03-18 stsp const struct got_error *
1100 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1101 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1102 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1104 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1105 6691714a 2018-01-23 stsp struct got_delta *delta;
1106 42c69117 2019-11-10 stsp uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1107 42c69117 2019-11-10 stsp size_t base_bufsz = 0, accum_size = 0, delta_len;
1108 22484865 2018-03-13 stsp uint64_t max_size;
1109 6691714a 2018-01-23 stsp int n = 0;
1111 b29656e2 2018-03-16 stsp *result_size = 0;
1113 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1114 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1116 8628c62d 2018-03-15 stsp /* We process small enough files entirely in memory for speed. */
1117 d582f26c 2020-03-18 stsp err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1119 22484865 2018-03-13 stsp return err;
1120 8628c62d 2018-03-15 stsp if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1121 8628c62d 2018-03-15 stsp accum_buf = malloc(max_size);
1122 8628c62d 2018-03-15 stsp if (accum_buf == NULL)
1123 638f9024 2019-05-13 stsp return got_error_from_errno("malloc");
1124 3840f4c9 2018-09-12 stsp base_file = NULL;
1125 3840f4c9 2018-09-12 stsp accum_file = NULL;
1127 3bfc2334 2022-01-11 stsp if (fseeko(base_file, 0L, SEEK_SET) == -1)
1128 3bfc2334 2022-01-11 stsp return got_error_from_errno("fseeko");
1129 3bfc2334 2022-01-11 stsp if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1130 3bfc2334 2022-01-11 stsp return got_error_from_errno("fseeko");
1133 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1134 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1135 ab2f42e7 2019-11-10 stsp int cached = 1;
1136 a6b158cc 2018-02-11 stsp if (n == 0) {
1137 34fca9c3 2018-11-11 stsp size_t mapoff;
1138 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1140 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1141 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1142 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1143 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1144 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1145 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1146 72eb3431 2018-04-01 stsp goto done;
1149 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1150 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1151 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1152 0c048b15 2018-04-27 stsp goto done;
1154 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1155 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1157 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1158 d7464085 2018-07-09 stsp goto done;
1161 d7464085 2018-07-09 stsp if (base_file) {
1162 d7464085 2018-07-09 stsp if (pack->map) {
1163 d7464085 2018-07-09 stsp mapoff = (size_t)delta_data_offset;
1164 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1165 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1166 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1167 4788f1ce 2020-03-18 stsp base_file);
1169 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1170 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1171 4788f1ce 2020-03-18 stsp base_file);
1173 d7464085 2018-07-09 stsp if (pack->map) {
1174 d7464085 2018-07-09 stsp mapoff = (size_t)delta_data_offset;
1175 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1176 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1177 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1178 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1180 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1181 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1182 55fdd257 2020-03-18 stsp pack->fd);
1185 a6b158cc 2018-02-11 stsp goto done;
1187 8628c62d 2018-03-15 stsp if (base_file)
1188 8628c62d 2018-03-15 stsp rewind(base_file);
1192 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1193 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1194 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1195 ab2f42e7 2019-11-10 stsp cached = 0;
1196 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1197 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1199 ab2f42e7 2019-11-10 stsp goto done;
1200 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1201 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1202 ab2f42e7 2019-11-10 stsp if (err == NULL)
1203 ab2f42e7 2019-11-10 stsp cached = 1;
1204 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1205 ab2f42e7 2019-11-10 stsp free(delta_buf);
1206 ab2f42e7 2019-11-10 stsp goto done;
1209 8628c62d 2018-03-15 stsp if (base_buf) {
1210 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1211 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1212 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1215 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1216 42c69117 2019-11-10 stsp delta_len,
1217 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1218 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1219 b29656e2 2018-03-16 stsp &accum_size);
1221 ab2f42e7 2019-11-10 stsp if (!cached)
1222 ab2f42e7 2019-11-10 stsp free(delta_buf);
1224 6691714a 2018-01-23 stsp goto done;
1226 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1227 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1228 8628c62d 2018-03-15 stsp if (base_buf) {
1229 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1231 34fca9c3 2018-11-11 stsp * Base buffer switches roles with accumulation
1232 34fca9c3 2018-11-11 stsp * buffer. Ensure it can hold the largest
1233 34fca9c3 2018-11-11 stsp * result in the delta chain. The initial
1234 34fca9c3 2018-11-11 stsp * allocation might have been smaller.
1236 34fca9c3 2018-11-11 stsp if (base_bufsz < max_size) {
1237 34fca9c3 2018-11-11 stsp uint8_t *p;
1238 34fca9c3 2018-11-11 stsp p = reallocarray(base_buf, 1, max_size);
1239 34fca9c3 2018-11-11 stsp if (p == NULL) {
1240 638f9024 2019-05-13 stsp err = got_error_from_errno(
1241 230a42bd 2019-05-11 jcs "reallocarray");
1242 34fca9c3 2018-11-11 stsp goto done;
1244 34fca9c3 2018-11-11 stsp base_buf = p;
1245 34fca9c3 2018-11-11 stsp base_bufsz = max_size;
1247 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1248 8628c62d 2018-03-15 stsp base_buf = tmp;
1250 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1251 8628c62d 2018-03-15 stsp accum_file = base_file;
1252 8628c62d 2018-03-15 stsp base_file = tmp;
1253 8628c62d 2018-03-15 stsp rewind(base_file);
1254 8628c62d 2018-03-15 stsp rewind(accum_file);
1260 8628c62d 2018-03-15 stsp free(base_buf);
1261 8628c62d 2018-03-15 stsp if (accum_buf) {
1262 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1263 8628c62d 2018-03-15 stsp free(accum_buf);
1264 8628c62d 2018-03-15 stsp if (len != accum_size)
1265 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1267 6691714a 2018-01-23 stsp rewind(outfile);
1268 b29656e2 2018-03-16 stsp if (err == NULL)
1269 b29656e2 2018-03-16 stsp *result_size = accum_size;
1270 e0ab43e7 2018-03-16 stsp return err;
1273 668a20f6 2020-03-18 stsp const struct got_error *
1274 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1275 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1277 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1278 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1279 42c69117 2019-11-10 stsp uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1280 42c69117 2019-11-10 stsp size_t base_bufsz = 0, accum_size = 0, delta_len;
1281 e0ab43e7 2018-03-16 stsp uint64_t max_size;
1282 e0ab43e7 2018-03-16 stsp int n = 0;
1284 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1285 e0ab43e7 2018-03-16 stsp *outlen = 0;
1287 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1288 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1290 d582f26c 2020-03-18 stsp err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1292 e0ab43e7 2018-03-16 stsp return err;
1293 e0ab43e7 2018-03-16 stsp accum_buf = malloc(max_size);
1294 e0ab43e7 2018-03-16 stsp if (accum_buf == NULL)
1295 638f9024 2019-05-13 stsp return got_error_from_errno("malloc");
1297 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1298 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1299 ab2f42e7 2019-11-10 stsp int cached = 1;
1300 e0ab43e7 2018-03-16 stsp if (n == 0) {
1301 0c048b15 2018-04-27 stsp size_t delta_data_offset;
1303 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1304 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1305 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1306 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1307 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1308 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1309 72eb3431 2018-04-01 stsp goto done;
1312 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1313 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1314 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1315 0c048b15 2018-04-27 stsp goto done;
1317 d7464085 2018-07-09 stsp if (pack->map) {
1318 d7464085 2018-07-09 stsp size_t mapoff = (size_t)delta_data_offset;
1319 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1320 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1321 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1323 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1325 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1326 d7464085 2018-07-09 stsp goto done;
1328 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1329 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1330 55fdd257 2020-03-18 stsp pack->fd);
1333 d7464085 2018-07-09 stsp goto done;
1338 ab2f42e7 2019-11-10 stsp got_delta_cache_get(&delta_buf, &delta_len,
1339 ab2f42e7 2019-11-10 stsp pack->delta_cache, delta->data_offset);
1340 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1341 ab2f42e7 2019-11-10 stsp cached = 0;
1342 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1343 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1345 ab2f42e7 2019-11-10 stsp goto done;
1346 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1347 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1348 ab2f42e7 2019-11-10 stsp if (err == NULL)
1349 ab2f42e7 2019-11-10 stsp cached = 1;
1350 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1351 ab2f42e7 2019-11-10 stsp free(delta_buf);
1352 ab2f42e7 2019-11-10 stsp goto done;
1355 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1356 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1357 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1358 ab2f42e7 2019-11-10 stsp if (!cached)
1359 ab2f42e7 2019-11-10 stsp free(delta_buf);
1362 e0ab43e7 2018-03-16 stsp goto done;
1364 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1365 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1366 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1368 34fca9c3 2018-11-11 stsp * Base buffer switches roles with accumulation buffer.
1369 34fca9c3 2018-11-11 stsp * Ensure it can hold the largest result in the delta
1370 34fca9c3 2018-11-11 stsp * chain. Initial allocation might have been smaller.
1372 34fca9c3 2018-11-11 stsp if (base_bufsz < max_size) {
1373 34fca9c3 2018-11-11 stsp uint8_t *p;
1374 34fca9c3 2018-11-11 stsp p = reallocarray(base_buf, 1, max_size);
1375 34fca9c3 2018-11-11 stsp if (p == NULL) {
1376 638f9024 2019-05-13 stsp err = got_error_from_errno(
1377 230a42bd 2019-05-11 jcs "reallocarray");
1378 34fca9c3 2018-11-11 stsp goto done;
1380 34fca9c3 2018-11-11 stsp base_buf = p;
1381 34fca9c3 2018-11-11 stsp base_bufsz = max_size;
1383 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1384 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1389 e0ab43e7 2018-03-16 stsp free(base_buf);
1390 e0ab43e7 2018-03-16 stsp if (err) {
1391 e0ab43e7 2018-03-16 stsp free(accum_buf);
1392 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1393 e0ab43e7 2018-03-16 stsp *outlen = 0;
1395 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1396 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1398 efd2a263 2018-01-19 stsp return err;
1401 3ee5fc21 2018-01-17 stsp const struct got_error *
1402 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1403 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1405 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1407 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1408 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1410 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1411 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1412 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1414 d7464085 2018-07-09 stsp if (pack->map) {
1415 d7464085 2018-07-09 stsp size_t mapoff = (size_t)obj->pack_offset;
1416 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1417 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1420 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1421 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1422 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1423 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1426 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1427 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1429 a1fd68d8 2018-01-12 stsp return err;
1432 e0ab43e7 2018-03-16 stsp const struct got_error *
1433 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1434 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1436 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1438 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1439 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1441 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1442 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1443 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1444 d7464085 2018-07-09 stsp if (pack->map) {
1445 d7464085 2018-07-09 stsp size_t mapoff = (size_t)obj->pack_offset;
1446 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1447 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1449 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1450 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1451 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1452 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1455 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1458 e0ab43e7 2018-03-16 stsp return err;
