2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
4 0a0a3048 2018-01-10 stsp * Permission to use, copy, modify, and distribute this software for any
5 0a0a3048 2018-01-10 stsp * purpose with or without fee is hereby granted, provided that the above
6 0a0a3048 2018-01-10 stsp * copyright notice and this permission notice appear in all copies.
8 0a0a3048 2018-01-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 0a0a3048 2018-01-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 0a0a3048 2018-01-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 0a0a3048 2018-01-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 0a0a3048 2018-01-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 0a0a3048 2018-01-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 0a0a3048 2018-01-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 8cff5e95 2023-03-08 thomas #include "got_compat.h"
18 a1fd68d8 2018-01-12 stsp #include <sys/types.h>
19 0a0a3048 2018-01-10 stsp #include <sys/stat.h>
20 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
21 876c234b 2018-09-10 stsp #include <sys/uio.h>
22 57b35b75 2018-06-22 stsp #include <sys/mman.h>
23 68036464 2022-06-26 thomas #include <sys/resource.h>
24 68036464 2022-06-26 thomas #include <sys/socket.h>
26 7e656b93 2018-03-17 stsp #include <fcntl.h>
27 a1fd68d8 2018-01-12 stsp #include <errno.h>
28 0a0a3048 2018-01-10 stsp #include <stdio.h>
29 a1fd68d8 2018-01-12 stsp #include <stdint.h>
30 0a0a3048 2018-01-10 stsp #include <stdlib.h>
31 0a0a3048 2018-01-10 stsp #include <string.h>
32 0a0a3048 2018-01-10 stsp #include <limits.h>
33 81a12da5 2020-09-09 naddy #include <unistd.h>
34 a1fd68d8 2018-01-12 stsp #include <zlib.h>
36 0a0a3048 2018-01-10 stsp #include "got_error.h"
37 a1fd68d8 2018-01-12 stsp #include "got_object.h"
38 324d37e7 2019-05-11 stsp #include "got_path.h"
40 be288a59 2023-02-23 thomas #include "got_lib_hash.h"
41 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
42 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
43 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
44 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
45 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 b16893ba 2023-02-24 thomas enum got_hash_algorithm algo = GOT_HASH_SHA1;
75 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
76 b16893ba 2023-02-24 thomas struct got_hash ctx;
77 b16893ba 2023-02-24 thomas uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
78 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
82 b16893ba 2023-02-24 thomas got_hash_init(&ctx, algo);
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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->offsets, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, 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 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->trailer->packfile_sha1,
326 b16893ba 2023-02-24 thomas SHA1_DIGEST_LENGTH);
327 b16893ba 2023-02-24 thomas got_hash_final(&ctx, hash);
328 1e73031b 2023-04-04 thomas if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1) != 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 fc63f50d 2021-12-31 thomas 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 aa75acde 2022-10-25 thomas if (p->len > 0 && p->len <= SIZE_MAX) {
394 aa75acde 2022-10-25 thomas p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
395 aa75acde 2022-10-25 thomas if (p->map == MAP_FAILED) {
396 aa75acde 2022-10-25 thomas if (errno != ENOMEM) {
397 aa75acde 2022-10-25 thomas err = got_error_from_errno("mmap");
398 aa75acde 2022-10-25 thomas goto done;
400 aa75acde 2022-10-25 thomas p->map = NULL; /* fall back to read(2) */
405 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
409 1124fe40 2021-07-07 stsp got_packidx_close(p);
411 0a0a3048 2018-01-10 stsp *packidx = p;
412 1124fe40 2021-07-07 stsp free(pack_relpath);
413 0a0a3048 2018-01-10 stsp return err;
416 57b35b75 2018-06-22 stsp const struct got_error *
417 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
419 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
421 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
422 fc79a48d 2018-07-09 stsp if (packidx->map) {
423 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
424 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
426 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
427 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
428 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
429 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
435 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
436 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
437 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_offsets);
438 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_large_offsets);
439 0a0a3048 2018-01-10 stsp free(packidx);
441 57b35b75 2018-06-22 stsp return err;
444 509c9973 2021-04-10 stsp const struct got_error *
445 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
447 509c9973 2021-04-10 stsp size_t size;
449 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
450 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
451 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
452 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
454 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
455 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
456 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
458 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
459 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
460 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
461 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
463 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
464 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
466 509c9973 2021-04-10 stsp return NULL;
470 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
472 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
473 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
474 a1fd68d8 2018-01-12 stsp uint64_t loffset;
475 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
476 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
477 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
479 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
480 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
482 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
486 48b4f239 2021-12-31 thomas got_packidx_get_object_idx(struct got_packidx *packidx,
487 48b4f239 2021-12-31 thomas struct got_object_id *id)
489 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
490 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
491 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
493 a1fd68d8 2018-01-12 stsp if (id0 > 0)
494 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
496 40aeb19c 2018-06-22 stsp while (left <= right) {
497 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
498 40aeb19c 2018-06-22 stsp int i, cmp;
500 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
501 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
502 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
503 16dcbf91 2018-04-01 stsp if (cmp == 0)
505 40aeb19c 2018-06-22 stsp else if (cmp > 0)
506 40aeb19c 2018-06-22 stsp left = i + 1;
507 40aeb19c 2018-06-22 stsp else if (cmp < 0)
508 40aeb19c 2018-06-22 stsp right = i - 1;
514 f9c2e8e5 2022-02-13 thomas static int
515 f9c2e8e5 2022-02-13 thomas offset_cmp(const void *pa, const void *pb)
517 f9c2e8e5 2022-02-13 thomas const struct got_pack_offset_index *a, *b;
519 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_offset_index *)pa;
520 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_offset_index *)pb;
522 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
523 f9c2e8e5 2022-02-13 thomas return -1;
524 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
525 f9c2e8e5 2022-02-13 thomas return 1;
527 f9c2e8e5 2022-02-13 thomas return 0;
530 f9c2e8e5 2022-02-13 thomas static int
531 f9c2e8e5 2022-02-13 thomas large_offset_cmp(const void *pa, const void *pb)
533 f9c2e8e5 2022-02-13 thomas const struct got_pack_large_offset_index *a, *b;
535 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_large_offset_index *)pa;
536 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_large_offset_index *)pb;
538 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
539 f9c2e8e5 2022-02-13 thomas return -1;
540 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
541 f9c2e8e5 2022-02-13 thomas return 1;
543 f9c2e8e5 2022-02-13 thomas return 0;
546 f9c2e8e5 2022-02-13 thomas static const struct got_error *
547 f9c2e8e5 2022-02-13 thomas build_offset_index(struct got_packidx *p)
549 f9c2e8e5 2022-02-13 thomas uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
550 f9c2e8e5 2022-02-13 thomas unsigned int i, j, k;
552 f9c2e8e5 2022-02-13 thomas p->sorted_offsets = calloc(nobj - p->nlargeobj,
553 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]));
554 f9c2e8e5 2022-02-13 thomas if (p->sorted_offsets == NULL)
555 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
557 f9c2e8e5 2022-02-13 thomas if (p->nlargeobj > 0) {
558 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets = calloc(p->nlargeobj,
559 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]));
560 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets == NULL)
561 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
566 f9c2e8e5 2022-02-13 thomas for (i = 0; i < nobj; i++) {
567 f9c2e8e5 2022-02-13 thomas uint32_t offset = be32toh(p->hdr.offsets[i]);
568 f9c2e8e5 2022-02-13 thomas if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
569 f9c2e8e5 2022-02-13 thomas uint64_t loffset;
570 f9c2e8e5 2022-02-13 thomas uint32_t idx;
571 f9c2e8e5 2022-02-13 thomas idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
572 f9c2e8e5 2022-02-13 thomas if (idx >= p->nlargeobj ||
573 f9c2e8e5 2022-02-13 thomas p->nlargeobj == 0 ||
574 f9c2e8e5 2022-02-13 thomas p->hdr.large_offsets == NULL)
575 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
576 f9c2e8e5 2022-02-13 thomas loffset = be64toh(p->hdr.large_offsets[idx]);
577 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].offset = loffset;
578 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].idx = i;
581 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].offset = offset;
582 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].idx = i;
586 f9c2e8e5 2022-02-13 thomas if (j != p->nlargeobj || k != nobj - p->nlargeobj)
587 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
589 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_offsets, nobj - p->nlargeobj,
590 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]), offset_cmp);
592 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets)
593 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_large_offsets, p->nlargeobj,
594 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
596 f9c2e8e5 2022-02-13 thomas return NULL;
599 876c234b 2018-09-10 stsp const struct got_error *
600 f9c2e8e5 2022-02-13 thomas got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
602 f9c2e8e5 2022-02-13 thomas const struct got_error *err;
603 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
604 f9c2e8e5 2022-02-13 thomas int i, left, right;
606 f9c2e8e5 2022-02-13 thomas *idx = -1;
608 f9c2e8e5 2022-02-13 thomas if (packidx->sorted_offsets == NULL) {
609 f9c2e8e5 2022-02-13 thomas err = build_offset_index(packidx);
611 f9c2e8e5 2022-02-13 thomas return err;
614 f9c2e8e5 2022-02-13 thomas if (offset >= 0x7fffffff) {
615 f9c2e8e5 2022-02-13 thomas uint64_t lo;
616 f9c2e8e5 2022-02-13 thomas left = 0, right = packidx->nlargeobj - 1;
617 f9c2e8e5 2022-02-13 thomas while (left <= right) {
618 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
619 f9c2e8e5 2022-02-13 thomas lo = packidx->sorted_large_offsets[i].offset;
620 f9c2e8e5 2022-02-13 thomas if (lo == offset) {
621 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_large_offsets[i].idx;
623 f9c2e8e5 2022-02-13 thomas } else if (offset > lo)
624 f9c2e8e5 2022-02-13 thomas left = i + 1;
625 f9c2e8e5 2022-02-13 thomas else if (offset < lo)
626 f9c2e8e5 2022-02-13 thomas right = i - 1;
629 f9c2e8e5 2022-02-13 thomas uint32_t o;
630 f9c2e8e5 2022-02-13 thomas left = 0, right = totobj - packidx->nlargeobj - 1;
631 f9c2e8e5 2022-02-13 thomas while (left <= right) {
632 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
633 f9c2e8e5 2022-02-13 thomas o = packidx->sorted_offsets[i].offset;
634 f9c2e8e5 2022-02-13 thomas if (o == offset) {
635 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_offsets[i].idx;
637 f9c2e8e5 2022-02-13 thomas } else if (offset > o)
638 f9c2e8e5 2022-02-13 thomas left = i + 1;
639 f9c2e8e5 2022-02-13 thomas else if (offset < o)
640 f9c2e8e5 2022-02-13 thomas right = i - 1;
644 f9c2e8e5 2022-02-13 thomas return NULL;
647 f9c2e8e5 2022-02-13 thomas const struct got_error *
648 f9c2e8e5 2022-02-13 thomas got_packidx_get_object_id(struct got_object_id *id,
649 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx, int idx)
651 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
652 f9c2e8e5 2022-02-13 thomas struct got_packidx_object_id *oid;
654 f9c2e8e5 2022-02-13 thomas if (idx < 0 || idx >= totobj)
655 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_NO_OBJ);
657 f9c2e8e5 2022-02-13 thomas oid = &packidx->hdr.sorted_ids[idx];
658 f9c2e8e5 2022-02-13 thomas memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
659 f9c2e8e5 2022-02-13 thomas return NULL;
662 f9c2e8e5 2022-02-13 thomas const struct got_error *
663 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
664 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
666 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
667 4277420a 2019-06-29 stsp u_int8_t id0;
668 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
669 4277420a 2019-06-29 stsp char hex[3];
670 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
671 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
672 ec138a87 2022-01-03 thomas uint32_t i = 0;
674 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
676 4277420a 2019-06-29 stsp if (prefix_len < 2)
677 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
679 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
680 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
681 4277420a 2019-06-29 stsp hex[2] = '\0';
682 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
683 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
685 ec138a87 2022-01-03 thomas if (id0 > 0)
686 ec138a87 2022-01-03 thomas i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
687 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
688 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
689 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
690 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
693 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
694 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
696 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
697 4277420a 2019-06-29 stsp if (cmp < 0) {
698 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
700 4277420a 2019-06-29 stsp } else if (cmp > 0)
703 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
706 ec242592 2022-04-22 thomas memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
707 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
709 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
713 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
714 68036464 2022-06-26 thomas return err;
717 68036464 2022-06-26 thomas static void
718 68036464 2022-06-26 thomas set_max_datasize(void)
720 68036464 2022-06-26 thomas struct rlimit rl;
722 68036464 2022-06-26 thomas if (getrlimit(RLIMIT_DATA, &rl) != 0)
725 68036464 2022-06-26 thomas rl.rlim_cur = rl.rlim_max;
726 68036464 2022-06-26 thomas setrlimit(RLIMIT_DATA, &rl);
729 68036464 2022-06-26 thomas const struct got_error *
730 68036464 2022-06-26 thomas got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
732 68036464 2022-06-26 thomas const struct got_error *err = NULL;
733 68036464 2022-06-26 thomas int imsg_fds[2];
734 68036464 2022-06-26 thomas pid_t pid;
735 68036464 2022-06-26 thomas struct imsgbuf *ibuf;
737 68036464 2022-06-26 thomas ibuf = calloc(1, sizeof(*ibuf));
738 68036464 2022-06-26 thomas if (ibuf == NULL)
739 68036464 2022-06-26 thomas return got_error_from_errno("calloc");
741 68036464 2022-06-26 thomas pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
742 68036464 2022-06-26 thomas if (pack->privsep_child == NULL) {
743 68036464 2022-06-26 thomas err = got_error_from_errno("calloc");
744 68036464 2022-06-26 thomas free(ibuf);
745 68036464 2022-06-26 thomas return err;
747 68036464 2022-06-26 thomas pack->child_has_tempfiles = 0;
748 68036464 2022-06-26 thomas pack->child_has_delta_outfd = 0;
750 68036464 2022-06-26 thomas if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
751 68036464 2022-06-26 thomas err = got_error_from_errno("socketpair");
752 68036464 2022-06-26 thomas goto done;
755 68036464 2022-06-26 thomas pid = fork();
756 68036464 2022-06-26 thomas if (pid == -1) {
757 68036464 2022-06-26 thomas err = got_error_from_errno("fork");
758 68036464 2022-06-26 thomas goto done;
759 68036464 2022-06-26 thomas } else if (pid == 0) {
760 68036464 2022-06-26 thomas set_max_datasize();
761 68036464 2022-06-26 thomas got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
762 68036464 2022-06-26 thomas pack->path_packfile);
763 68036464 2022-06-26 thomas /* not reached */
766 68036464 2022-06-26 thomas if (close(imsg_fds[1]) == -1)
767 68036464 2022-06-26 thomas return got_error_from_errno("close");
768 68036464 2022-06-26 thomas pack->privsep_child->imsg_fd = imsg_fds[0];
769 68036464 2022-06-26 thomas pack->privsep_child->pid = pid;
770 68036464 2022-06-26 thomas imsg_init(ibuf, imsg_fds[0]);
771 68036464 2022-06-26 thomas pack->privsep_child->ibuf = ibuf;
773 68036464 2022-06-26 thomas err = got_privsep_init_pack_child(ibuf, pack, packidx);
774 68036464 2022-06-26 thomas if (err) {
775 68036464 2022-06-26 thomas const struct got_error *child_err;
776 68036464 2022-06-26 thomas err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
777 68036464 2022-06-26 thomas child_err = got_privsep_wait_for_child(
778 68036464 2022-06-26 thomas pack->privsep_child->pid);
779 68036464 2022-06-26 thomas if (child_err && err == NULL)
780 68036464 2022-06-26 thomas err = child_err;
783 68036464 2022-06-26 thomas if (err) {
784 68036464 2022-06-26 thomas free(ibuf);
785 68036464 2022-06-26 thomas free(pack->privsep_child);
786 68036464 2022-06-26 thomas pack->privsep_child = NULL;
788 dd88155e 2019-06-29 stsp return err;
791 b4f37570 2021-06-19 stsp static const struct got_error *
792 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
794 a6ca45d5 2022-10-24 thomas const struct got_error *err = NULL, *close_err = NULL;
796 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
797 876c234b 2018-09-10 stsp return NULL;
799 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
801 96732e0b 2018-11-11 stsp return err;
802 a6ca45d5 2022-10-24 thomas if (close(pack->privsep_child->imsg_fd) == -1)
803 a6ca45d5 2022-10-24 thomas close_err = got_error_from_errno("close");
804 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
805 a6ca45d5 2022-10-24 thomas if (close_err && err == NULL)
806 a6ca45d5 2022-10-24 thomas err = close_err;
807 a6ca45d5 2022-10-24 thomas imsg_clear(pack->privsep_child->ibuf);
808 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
809 876c234b 2018-09-10 stsp free(pack->privsep_child);
810 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
811 876c234b 2018-09-10 stsp return err;
814 d7464085 2018-07-09 stsp const struct got_error *
815 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
817 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
819 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
820 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
821 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
822 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
823 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
824 8b2180d4 2018-04-26 stsp pack->fd = -1;
825 4589e373 2018-03-17 stsp free(pack->path_packfile);
826 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
827 4589e373 2018-03-17 stsp pack->filesize = 0;
828 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
829 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
830 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
834 bfb5ee0b 2022-05-31 thomas * Leave accumfd and basefd alone. They are managed by the
835 bfb5ee0b 2022-05-31 thomas * repository layer and can be reused.
838 c7fe698a 2018-01-23 stsp return err;
841 668a20f6 2020-03-18 stsp const struct got_error *
842 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
843 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
845 a487c1d0 2018-01-14 stsp uint8_t t = 0;
846 a487c1d0 2018-01-14 stsp uint64_t s = 0;
847 a487c1d0 2018-01-14 stsp uint8_t sizeN;
848 d7464085 2018-07-09 stsp size_t mapoff = 0;
853 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
854 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
856 d7464085 2018-07-09 stsp if (pack->map) {
857 66e6097f 2022-10-27 thomas if (offset > SIZE_MAX) {
858 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
859 66e6097f 2022-10-27 thomas "offset %lld overflows size_t",
860 66e6097f 2022-10-27 thomas (long long)offset);
863 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
865 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
866 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
870 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
872 eac8a741 2022-10-20 thomas return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
873 04fb0dd8 2022-10-24 thomas "packfile offset %lld", (long long)offset);
875 d7464085 2018-07-09 stsp if (pack->map) {
876 1944573a 2022-01-10 thomas if (mapoff + sizeof(sizeN) >= pack->filesize)
877 1944573a 2022-01-10 thomas return got_error(GOT_ERR_BAD_PACKFILE);
878 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
879 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
881 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
883 638f9024 2019-05-13 stsp return got_error_from_errno("read");
884 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
885 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
887 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
889 a1fd68d8 2018-01-12 stsp if (i == 0) {
890 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
891 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
892 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
894 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
895 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
898 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
902 a487c1d0 2018-01-14 stsp return NULL;
905 a1fd68d8 2018-01-12 stsp static const struct got_error *
906 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
907 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
909 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
910 6ccb713b 2018-01-19 stsp if (*obj == NULL)
911 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
913 6ccb713b 2018-01-19 stsp (*obj)->type = type;
914 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
915 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
916 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
917 6ccb713b 2018-01-19 stsp (*obj)->size = size;
918 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
919 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
921 b107e67f 2018-01-19 stsp return NULL;
924 b107e67f 2018-01-19 stsp static const struct got_error *
925 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
926 d7464085 2018-07-09 stsp off_t delta_offset)
928 b107e67f 2018-01-19 stsp int64_t o = 0;
929 b107e67f 2018-01-19 stsp uint8_t offN;
932 a0de39f3 2019-08-09 stsp *offset = 0;
936 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
938 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
940 d7464085 2018-07-09 stsp if (pack->map) {
941 d7464085 2018-07-09 stsp size_t mapoff;
943 66e6097f 2022-10-27 thomas if (delta_offset + *len > SIZE_MAX) {
944 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
945 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
946 66e6097f 2022-10-27 thomas (long long)delta_offset + *len);
949 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
950 1944573a 2022-01-10 thomas if (mapoff + sizeof(offN) >= pack->filesize)
951 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
952 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
955 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
957 638f9024 2019-05-13 stsp return got_error_from_errno("read");
958 d7464085 2018-07-09 stsp if (n != sizeof(offN))
959 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
961 d7464085 2018-07-09 stsp *len += sizeof(offN);
963 b107e67f 2018-01-19 stsp if (i == 0)
964 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
968 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
971 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
973 b107e67f 2018-01-19 stsp *offset = o;
974 6ccb713b 2018-01-19 stsp return NULL;
977 668a20f6 2020-03-18 stsp const struct got_error *
978 48b4f239 2021-12-31 thomas got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
979 2424a69d 2022-10-24 thomas struct got_pack *pack, off_t offset, size_t tslen)
981 96f5e8b3 2018-01-23 stsp const struct got_error *err;
982 b107e67f 2018-01-19 stsp int64_t negoffset;
983 b107e67f 2018-01-19 stsp size_t negofflen;
987 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
988 d7464085 2018-07-09 stsp offset + tslen);
990 b107e67f 2018-01-19 stsp return err;
992 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
993 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
994 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
995 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
997 d7464085 2018-07-09 stsp *len = negofflen;
998 96f5e8b3 2018-01-23 stsp return NULL;
1001 0e22967e 2018-02-11 stsp static const struct got_error *
1002 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1003 9249e7e3 2022-05-12 thomas size_t *delta_compressed_len, size_t delta_data_offset,
1004 9249e7e3 2022-05-12 thomas struct got_pack *pack)
1006 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
1007 9249e7e3 2022-05-12 thomas size_t consumed = 0;
1009 42c69117 2019-11-10 stsp if (pack->map) {
1010 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
1011 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1012 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1013 9249e7e3 2022-05-12 thomas &consumed, NULL, pack->map, delta_data_offset,
1014 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
1015 9249e7e3 2022-05-12 thomas if (err)
1016 9249e7e3 2022-05-12 thomas return err;
1018 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1019 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
1020 9249e7e3 2022-05-12 thomas err = got_inflate_to_mem_fd(delta_buf, delta_len,
1021 9249e7e3 2022-05-12 thomas &consumed, NULL, 0, pack->fd);
1022 9249e7e3 2022-05-12 thomas if (err)
1023 9249e7e3 2022-05-12 thomas return err;
1026 9249e7e3 2022-05-12 thomas if (delta_compressed_len)
1027 9249e7e3 2022-05-12 thomas *delta_compressed_len = consumed;
1029 9249e7e3 2022-05-12 thomas return NULL;
1032 96f5e8b3 2018-01-23 stsp static const struct got_error *
1033 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1034 e485d53c 2022-10-28 thomas int delta_type, size_t delta_size, off_t delta_data_offset)
1036 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
1038 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1039 42c69117 2019-11-10 stsp delta_data_offset);
1040 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
1041 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
1042 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
1043 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
1044 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1045 bdd2fbb3 2018-02-11 stsp return NULL;
1048 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1049 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1050 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1051 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1053 a3500804 2018-01-23 stsp const struct got_error *err;
1054 c3703302 2018-01-23 stsp off_t base_offset;
1055 c3703302 2018-01-23 stsp uint8_t base_type;
1056 c3703302 2018-01-23 stsp uint64_t base_size;
1057 c3703302 2018-01-23 stsp size_t base_tslen;
1058 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1059 42c69117 2019-11-10 stsp size_t consumed;
1061 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1062 d7464085 2018-07-09 stsp delta_offset, tslen);
1064 a3500804 2018-01-23 stsp return err;
1066 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
1067 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
1068 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1070 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1071 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1072 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
1073 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1076 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1077 e485d53c 2022-10-28 thomas delta_data_offset);
1079 1a35c1bc 2019-05-22 stsp return err;
1081 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
1082 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1083 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1085 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1086 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1088 1a35c1bc 2019-05-22 stsp return err;
1090 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1091 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
1094 c4330eff 2021-06-22 stsp const struct got_error *
1095 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1096 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
1098 c4330eff 2021-06-22 stsp if (pack->map) {
1099 66e6097f 2022-10-27 thomas size_t mapoff;
1101 66e6097f 2022-10-27 thomas if (delta_offset + tslen > SIZE_MAX) {
1102 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
1103 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
1104 66e6097f 2022-10-27 thomas (long long)delta_offset + tslen);
1107 66e6097f 2022-10-27 thomas mapoff = delta_offset + tslen;
1108 1944573a 2022-01-10 thomas if (mapoff + sizeof(*id) >= pack->filesize)
1109 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
1110 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
1112 c4330eff 2021-06-22 stsp ssize_t n;
1113 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1114 c4330eff 2021-06-22 stsp if (n < 0)
1115 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1116 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1117 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1120 c4330eff 2021-06-22 stsp return NULL;
1123 c3703302 2018-01-23 stsp static const struct got_error *
1124 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1125 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1126 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1128 6b9c9673 2018-01-23 stsp const struct got_error *err;
1129 6b9c9673 2018-01-23 stsp struct got_object_id id;
1131 6b9c9673 2018-01-23 stsp off_t base_offset;
1132 6b9c9673 2018-01-23 stsp uint8_t base_type;
1133 6b9c9673 2018-01-23 stsp uint64_t base_size;
1134 6b9c9673 2018-01-23 stsp size_t base_tslen;
1135 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1137 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1138 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1140 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1142 c4330eff 2021-06-22 stsp return err;
1143 d7464085 2018-07-09 stsp if (pack->map) {
1144 4907b168 2023-02-20 thomas delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1146 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1147 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1148 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1151 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1152 e485d53c 2022-10-28 thomas delta_data_offset);
1154 1a35c1bc 2019-05-22 stsp return err;
1156 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1157 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1158 1a35c1bc 2019-05-22 stsp if (idx == -1)
1159 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1161 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1162 e4a6dbc6 2022-07-24 thomas if (base_offset == -1)
1163 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1165 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1166 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1168 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1169 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1171 1a35c1bc 2019-05-22 stsp return err;
1173 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1174 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1177 668a20f6 2020-03-18 stsp const struct got_error *
1178 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1179 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1180 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1182 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1184 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1185 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1187 c3703302 2018-01-23 stsp switch (delta_type) {
1188 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1189 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1190 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1191 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1192 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1193 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1194 42c69117 2019-11-10 stsp delta_size, 0);
1196 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1197 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1198 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1200 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1201 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1202 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1205 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1208 96f5e8b3 2018-01-23 stsp return err;
1211 96f5e8b3 2018-01-23 stsp static const struct got_error *
1212 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1213 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1214 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1216 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1217 96f5e8b3 2018-01-23 stsp int resolved_type;
1219 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1220 b107e67f 2018-01-19 stsp if (*obj == NULL)
1221 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1223 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1224 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1225 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1226 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1227 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1229 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1230 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1231 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1232 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1234 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1235 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1236 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1238 96f5e8b3 2018-01-23 stsp goto done;
1240 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1242 96f5e8b3 2018-01-23 stsp goto done;
1243 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1245 96f5e8b3 2018-01-23 stsp if (err) {
1246 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1247 96f5e8b3 2018-01-23 stsp *obj = NULL;
1249 96f5e8b3 2018-01-23 stsp return err;
1252 2090a03d 2018-09-09 stsp const struct got_error *
1253 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1254 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1256 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1257 a1fd68d8 2018-01-12 stsp off_t offset;
1258 6c00b545 2018-01-17 stsp uint8_t type;
1259 6c00b545 2018-01-17 stsp uint64_t size;
1260 3ee5fc21 2018-01-17 stsp size_t tslen;
1262 6c00b545 2018-01-17 stsp *obj = NULL;
1264 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1265 e4a6dbc6 2022-07-24 thomas if (offset == -1)
1266 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1268 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1269 668a20f6 2020-03-18 stsp pack, offset);
1271 35c73765 2018-09-09 stsp return err;
1273 6c00b545 2018-01-17 stsp switch (type) {
1274 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1275 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1276 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1277 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1278 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1279 5f25cc85 2019-11-26 stsp size, idx);
1281 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1282 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1283 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1284 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1287 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1291 3ee5fc21 2018-01-17 stsp return err;
1294 d582f26c 2020-03-18 stsp const struct got_error *
1295 48b4f239 2021-12-31 thomas got_pack_get_delta_chain_max_size(uint64_t *max_size,
1296 48b4f239 2021-12-31 thomas struct got_delta_chain *deltas, struct got_pack *pack)
1298 22484865 2018-03-13 stsp struct got_delta *delta;
1299 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1301 22484865 2018-03-13 stsp *max_size = 0;
1302 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1303 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1304 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1305 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1306 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1307 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1308 22484865 2018-03-13 stsp const struct got_error *err;
1309 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1310 42c69117 2019-11-10 stsp size_t delta_len;
1311 ab2f42e7 2019-11-10 stsp int cached = 1;
1313 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1314 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1315 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1317 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1318 ab2f42e7 2019-11-10 stsp cached = 0;
1319 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1320 9249e7e3 2022-05-12 thomas NULL, delta->data_offset, pack);
1322 ab2f42e7 2019-11-10 stsp return err;
1324 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1325 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1326 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1327 ab2f42e7 2019-11-10 stsp if (err == NULL)
1328 ab2f42e7 2019-11-10 stsp cached = 1;
1329 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1330 ab2f42e7 2019-11-10 stsp free(delta_buf);
1331 ab2f42e7 2019-11-10 stsp return err;
1334 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1335 42c69117 2019-11-10 stsp delta_buf, delta_len);
1336 ab2f42e7 2019-11-10 stsp if (!cached)
1337 ab2f42e7 2019-11-10 stsp free(delta_buf);
1339 22484865 2018-03-13 stsp return err;
1341 22484865 2018-03-13 stsp base_size = delta->size;
1342 22484865 2018-03-13 stsp if (base_size > *max_size)
1343 22484865 2018-03-13 stsp *max_size = base_size;
1344 22484865 2018-03-13 stsp if (result_size > *max_size)
1345 22484865 2018-03-13 stsp *max_size = result_size;
1348 9feb4ff2 2018-03-14 stsp return NULL;
1351 ac544f8c 2019-01-13 stsp const struct got_error *
1352 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1353 42c69117 2019-11-10 stsp struct got_pack *pack)
1355 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1356 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1358 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1361 668a20f6 2020-03-18 stsp const struct got_error *
1362 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1363 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1364 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1366 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1367 6691714a 2018-01-23 stsp struct got_delta *delta;
1368 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1369 210941d5 2022-05-31 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1370 210941d5 2022-05-31 thomas /* We process small enough files entirely in memory for speed. */
1371 210941d5 2022-05-31 thomas const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1372 210941d5 2022-05-31 thomas uint64_t max_size = 0;
1373 6691714a 2018-01-23 stsp int n = 0;
1375 b29656e2 2018-03-16 stsp *result_size = 0;
1377 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1378 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1380 210941d5 2022-05-31 thomas if (fseeko(base_file, 0L, SEEK_SET) == -1)
1381 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1382 210941d5 2022-05-31 thomas if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1383 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1385 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1386 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1387 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1388 210941d5 2022-05-31 thomas uint64_t base_size, result_size = 0;
1389 ab2f42e7 2019-11-10 stsp int cached = 1;
1390 a6b158cc 2018-02-11 stsp if (n == 0) {
1391 34fca9c3 2018-11-11 stsp size_t mapoff;
1392 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1394 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1395 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1396 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1397 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1398 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1399 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1400 72eb3431 2018-04-01 stsp goto done;
1403 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1404 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1405 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1406 0c048b15 2018-04-27 stsp goto done;
1408 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1409 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1411 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1412 d7464085 2018-07-09 stsp goto done;
1415 210941d5 2022-05-31 thomas if (delta->size > max_size)
1416 210941d5 2022-05-31 thomas max_size = delta->size;
1417 210941d5 2022-05-31 thomas if (max_size > max_bufsize) {
1418 d7464085 2018-07-09 stsp if (pack->map) {
1419 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1420 66e6097f 2022-10-27 thomas return got_error_fmt(
1421 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1422 66e6097f 2022-10-27 thomas "delta offset %lld "
1423 66e6097f 2022-10-27 thomas "overflows size_t",
1424 66e6097f 2022-10-27 thomas (long long)
1425 66e6097f 2022-10-27 thomas delta_data_offset);
1428 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1429 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1430 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1431 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1432 4788f1ce 2020-03-18 stsp base_file);
1434 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1435 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1436 4788f1ce 2020-03-18 stsp base_file);
1438 210941d5 2022-05-31 thomas accum_buf = malloc(max_size);
1439 210941d5 2022-05-31 thomas if (accum_buf == NULL) {
1440 210941d5 2022-05-31 thomas err = got_error_from_errno("malloc");
1441 210941d5 2022-05-31 thomas goto done;
1443 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1444 d7464085 2018-07-09 stsp if (pack->map) {
1445 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1446 66e6097f 2022-10-27 thomas return got_error_fmt(
1447 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1448 66e6097f 2022-10-27 thomas "delta offset %lld "
1449 66e6097f 2022-10-27 thomas "overflows size_t",
1450 66e6097f 2022-10-27 thomas (long long)
1451 66e6097f 2022-10-27 thomas delta_data_offset);
1454 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1455 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1456 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1457 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1458 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1460 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1461 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1462 55fdd257 2020-03-18 stsp pack->fd);
1465 a6b158cc 2018-02-11 stsp goto done;
1467 210941d5 2022-05-31 thomas if (base_buf == NULL)
1468 8628c62d 2018-03-15 stsp rewind(base_file);
1472 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1473 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1474 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1476 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1477 ab2f42e7 2019-11-10 stsp cached = 0;
1478 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1479 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1481 ab2f42e7 2019-11-10 stsp goto done;
1483 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1484 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1485 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1486 ab2f42e7 2019-11-10 stsp if (err == NULL)
1487 ab2f42e7 2019-11-10 stsp cached = 1;
1488 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1489 ab2f42e7 2019-11-10 stsp free(delta_buf);
1490 ab2f42e7 2019-11-10 stsp goto done;
1494 210941d5 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1495 210941d5 2022-05-31 thomas delta_buf, delta_len);
1496 b1fad72f 2022-11-08 thomas if (err) {
1497 b1fad72f 2022-11-08 thomas if (!cached)
1498 b1fad72f 2022-11-08 thomas free(delta_buf);
1499 210941d5 2022-05-31 thomas goto done;
1501 210941d5 2022-05-31 thomas if (base_size > max_size)
1502 210941d5 2022-05-31 thomas max_size = base_size;
1503 210941d5 2022-05-31 thomas if (result_size > max_size)
1504 210941d5 2022-05-31 thomas max_size = result_size;
1506 210941d5 2022-05-31 thomas if (base_buf && max_size > max_bufsize) {
1507 210941d5 2022-05-31 thomas /* Switch from buffers to temporary files. */
1508 210941d5 2022-05-31 thomas size_t w = fwrite(base_buf, 1, base_bufsz,
1509 210941d5 2022-05-31 thomas base_file);
1510 210941d5 2022-05-31 thomas if (w != base_bufsz) {
1511 210941d5 2022-05-31 thomas err = got_ferror(outfile, GOT_ERR_IO);
1512 b1fad72f 2022-11-08 thomas if (!cached)
1513 b1fad72f 2022-11-08 thomas free(delta_buf);
1514 210941d5 2022-05-31 thomas goto done;
1516 210941d5 2022-05-31 thomas free(base_buf);
1517 210941d5 2022-05-31 thomas base_buf = NULL;
1518 210941d5 2022-05-31 thomas free(accum_buf);
1519 210941d5 2022-05-31 thomas accum_buf = NULL;
1522 210941d5 2022-05-31 thomas if (base_buf && max_size > base_bufsz) {
1523 210941d5 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1524 210941d5 2022-05-31 thomas if (p == NULL) {
1525 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1526 b1fad72f 2022-11-08 thomas if (!cached)
1527 b1fad72f 2022-11-08 thomas free(delta_buf);
1528 210941d5 2022-05-31 thomas goto done;
1530 210941d5 2022-05-31 thomas base_buf = p;
1531 210941d5 2022-05-31 thomas base_bufsz = max_size;
1534 210941d5 2022-05-31 thomas if (accum_buf && max_size > accum_bufsz) {
1535 210941d5 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1536 210941d5 2022-05-31 thomas if (p == NULL) {
1537 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1538 b1fad72f 2022-11-08 thomas if (!cached)
1539 b1fad72f 2022-11-08 thomas free(delta_buf);
1540 210941d5 2022-05-31 thomas goto done;
1542 210941d5 2022-05-31 thomas accum_buf = p;
1543 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1546 8628c62d 2018-03-15 stsp if (base_buf) {
1547 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1548 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1549 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1552 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1553 42c69117 2019-11-10 stsp delta_len,
1554 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1555 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1556 b29656e2 2018-03-16 stsp &accum_size);
1558 ab2f42e7 2019-11-10 stsp if (!cached)
1559 ab2f42e7 2019-11-10 stsp free(delta_buf);
1561 6691714a 2018-01-23 stsp goto done;
1563 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1564 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1565 8628c62d 2018-03-15 stsp if (base_buf) {
1566 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1567 210941d5 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1568 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1569 210941d5 2022-05-31 thomas accum_bufsz = base_bufsz;
1570 8628c62d 2018-03-15 stsp base_buf = tmp;
1571 210941d5 2022-05-31 thomas base_bufsz = tmp_size;
1573 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1574 8628c62d 2018-03-15 stsp accum_file = base_file;
1575 8628c62d 2018-03-15 stsp base_file = tmp;
1576 8628c62d 2018-03-15 stsp rewind(base_file);
1577 8628c62d 2018-03-15 stsp rewind(accum_file);
1583 8628c62d 2018-03-15 stsp free(base_buf);
1584 8628c62d 2018-03-15 stsp if (accum_buf) {
1585 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1586 8628c62d 2018-03-15 stsp free(accum_buf);
1587 8628c62d 2018-03-15 stsp if (len != accum_size)
1588 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1590 6691714a 2018-01-23 stsp rewind(outfile);
1591 b29656e2 2018-03-16 stsp if (err == NULL)
1592 b29656e2 2018-03-16 stsp *result_size = accum_size;
1593 e0ab43e7 2018-03-16 stsp return err;
1596 668a20f6 2020-03-18 stsp const struct got_error *
1597 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1598 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1600 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1601 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1602 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1603 6c85883b 2022-05-31 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1604 6c85883b 2022-05-31 thomas uint64_t max_size = 0;
1605 e0ab43e7 2018-03-16 stsp int n = 0;
1607 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1608 e0ab43e7 2018-03-16 stsp *outlen = 0;
1610 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1611 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1613 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1614 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1615 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1616 6c85883b 2022-05-31 thomas uint64_t base_size, result_size = 0;
1617 ab2f42e7 2019-11-10 stsp int cached = 1;
1618 e0ab43e7 2018-03-16 stsp if (n == 0) {
1619 66e6097f 2022-10-27 thomas off_t delta_data_offset;
1621 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1622 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1623 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1624 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1625 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1626 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1627 72eb3431 2018-04-01 stsp goto done;
1630 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1631 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1632 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1633 0c048b15 2018-04-27 stsp goto done;
1636 6c85883b 2022-05-31 thomas if (delta->size > max_size)
1637 6c85883b 2022-05-31 thomas max_size = delta->size;
1639 d7464085 2018-07-09 stsp if (pack->map) {
1640 66e6097f 2022-10-27 thomas size_t mapoff;
1642 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1643 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1644 66e6097f 2022-10-27 thomas "delta %lld offset would "
1645 66e6097f 2022-10-27 thomas "overflow size_t",
1646 66e6097f 2022-10-27 thomas (long long)delta_data_offset);
1649 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1650 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1651 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1652 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1654 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1656 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1657 d7464085 2018-07-09 stsp goto done;
1659 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1660 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1661 55fdd257 2020-03-18 stsp pack->fd);
1664 d7464085 2018-07-09 stsp goto done;
1669 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1670 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1671 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1673 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1674 ab2f42e7 2019-11-10 stsp cached = 0;
1675 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1676 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1678 ab2f42e7 2019-11-10 stsp goto done;
1680 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1681 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1682 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1683 ab2f42e7 2019-11-10 stsp if (err == NULL)
1684 ab2f42e7 2019-11-10 stsp cached = 1;
1685 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1686 ab2f42e7 2019-11-10 stsp free(delta_buf);
1687 ab2f42e7 2019-11-10 stsp goto done;
1691 6c85883b 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1692 6c85883b 2022-05-31 thomas delta_buf, delta_len);
1693 b1fad72f 2022-11-08 thomas if (err) {
1694 b1fad72f 2022-11-08 thomas if (!cached)
1695 b1fad72f 2022-11-08 thomas free(delta_buf);
1696 6c85883b 2022-05-31 thomas goto done;
1698 6c85883b 2022-05-31 thomas if (base_size > max_size)
1699 6c85883b 2022-05-31 thomas max_size = base_size;
1700 6c85883b 2022-05-31 thomas if (result_size > max_size)
1701 6c85883b 2022-05-31 thomas max_size = result_size;
1703 6c85883b 2022-05-31 thomas if (max_size > base_bufsz) {
1704 6c85883b 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1705 6c85883b 2022-05-31 thomas if (p == NULL) {
1706 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1707 b1fad72f 2022-11-08 thomas if (!cached)
1708 b1fad72f 2022-11-08 thomas free(delta_buf);
1709 6c85883b 2022-05-31 thomas goto done;
1711 6c85883b 2022-05-31 thomas base_buf = p;
1712 6c85883b 2022-05-31 thomas base_bufsz = max_size;
1715 6c85883b 2022-05-31 thomas if (max_size > accum_bufsz) {
1716 6c85883b 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1717 6c85883b 2022-05-31 thomas if (p == NULL) {
1718 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1719 b1fad72f 2022-11-08 thomas if (!cached)
1720 b1fad72f 2022-11-08 thomas free(delta_buf);
1721 6c85883b 2022-05-31 thomas goto done;
1723 6c85883b 2022-05-31 thomas accum_buf = p;
1724 6c85883b 2022-05-31 thomas accum_bufsz = max_size;
1727 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1728 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1729 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1730 ab2f42e7 2019-11-10 stsp if (!cached)
1731 ab2f42e7 2019-11-10 stsp free(delta_buf);
1734 e0ab43e7 2018-03-16 stsp goto done;
1736 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1737 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1738 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1739 6c85883b 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1740 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1741 6c85883b 2022-05-31 thomas accum_bufsz = base_bufsz;
1742 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1743 6c85883b 2022-05-31 thomas base_bufsz = tmp_size;
1748 e0ab43e7 2018-03-16 stsp free(base_buf);
1749 e0ab43e7 2018-03-16 stsp if (err) {
1750 e0ab43e7 2018-03-16 stsp free(accum_buf);
1751 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1752 e0ab43e7 2018-03-16 stsp *outlen = 0;
1754 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1755 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1757 efd2a263 2018-01-19 stsp return err;
1760 3ee5fc21 2018-01-17 stsp const struct got_error *
1761 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1762 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1764 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1766 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1767 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1769 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1770 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1771 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1773 d7464085 2018-07-09 stsp if (pack->map) {
1774 66e6097f 2022-10-27 thomas size_t mapoff;
1776 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1777 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1778 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1779 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1782 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1783 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1784 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1787 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1788 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1789 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1790 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1793 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1794 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1796 a1fd68d8 2018-01-12 stsp return err;
1799 e0ab43e7 2018-03-16 stsp const struct got_error *
1800 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1801 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1803 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1805 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1806 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1808 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1809 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1810 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1811 d7464085 2018-07-09 stsp if (pack->map) {
1812 66e6097f 2022-10-27 thomas size_t mapoff;
1814 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1815 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1816 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1817 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1820 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1821 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1822 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1824 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1825 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1826 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1827 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1830 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1833 e0ab43e7 2018-03-16 stsp return err;
1836 9249e7e3 2022-05-12 thomas static const struct got_error *
1837 9249e7e3 2022-05-12 thomas read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1838 9249e7e3 2022-05-12 thomas size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1839 9249e7e3 2022-05-12 thomas off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1841 9249e7e3 2022-05-12 thomas const struct got_error *err = NULL;
1843 9249e7e3 2022-05-12 thomas /* Validate decompression and obtain the decompressed size. */
1844 9249e7e3 2022-05-12 thomas err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1845 9249e7e3 2022-05-12 thomas delta_data_offset, pack);
1846 9249e7e3 2022-05-12 thomas if (err)
1847 9249e7e3 2022-05-12 thomas return err;
1849 9249e7e3 2022-05-12 thomas /* Read delta base/result sizes from head of delta stream. */
1850 9249e7e3 2022-05-12 thomas err = got_delta_get_sizes(base_size, result_size,
1851 9249e7e3 2022-05-12 thomas *delta_buf, *delta_len);
1852 9249e7e3 2022-05-12 thomas if (err)
1853 9249e7e3 2022-05-12 thomas goto done;
1855 9249e7e3 2022-05-12 thomas /* Discard decompressed delta and read it again in compressed form. */
1856 9249e7e3 2022-05-12 thomas free(*delta_buf);
1857 9249e7e3 2022-05-12 thomas *delta_buf = malloc(*delta_len_compressed);
1858 9249e7e3 2022-05-12 thomas if (*delta_buf == NULL) {
1859 9249e7e3 2022-05-12 thomas err = got_error_from_errno("malloc");
1860 9249e7e3 2022-05-12 thomas goto done;
1862 9249e7e3 2022-05-12 thomas if (pack->map) {
1863 30d415a5 2022-10-25 thomas if (delta_data_offset >= pack->filesize) {
1864 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_PACK_OFFSET);
1865 30d415a5 2022-10-25 thomas goto done;
1867 9249e7e3 2022-05-12 thomas memcpy(*delta_buf, pack->map + delta_data_offset,
1868 9249e7e3 2022-05-12 thomas *delta_len_compressed);
1869 9249e7e3 2022-05-12 thomas } else {
1870 9249e7e3 2022-05-12 thomas ssize_t n;
1871 9249e7e3 2022-05-12 thomas if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1872 9249e7e3 2022-05-12 thomas err = got_error_from_errno("lseek");
1873 9249e7e3 2022-05-12 thomas goto done;
1875 9249e7e3 2022-05-12 thomas n = read(pack->fd, *delta_buf, *delta_len_compressed);
1876 9249e7e3 2022-05-12 thomas if (n < 0) {
1877 9249e7e3 2022-05-12 thomas err = got_error_from_errno("read");
1878 9249e7e3 2022-05-12 thomas goto done;
1879 9249e7e3 2022-05-12 thomas } else if (n != *delta_len_compressed) {
1880 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_IO);
1881 9249e7e3 2022-05-12 thomas goto done;
1885 9249e7e3 2022-05-12 thomas if (err) {
1886 9249e7e3 2022-05-12 thomas free(*delta_buf);
1887 9249e7e3 2022-05-12 thomas *delta_buf = NULL;
1888 9249e7e3 2022-05-12 thomas *delta_len = 0;
1889 9249e7e3 2022-05-12 thomas *delta_len_compressed = 0;
1890 9249e7e3 2022-05-12 thomas *base_size = 0;
1891 9249e7e3 2022-05-12 thomas *result_size = 0;
1893 9249e7e3 2022-05-12 thomas return err;
1896 f9c2e8e5 2022-02-13 thomas const struct got_error *
1897 f9c2e8e5 2022-02-13 thomas got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1898 c44c7d6e 2022-12-04 thomas size_t *delta_compressed_size, off_t *delta_offset,
1899 c44c7d6e 2022-12-04 thomas off_t *delta_data_offset, off_t *base_offset,
1900 9249e7e3 2022-05-12 thomas struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1901 9249e7e3 2022-05-12 thomas struct got_pack *pack, struct got_packidx *packidx, int idx)
1903 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
1904 f9c2e8e5 2022-02-13 thomas off_t offset;
1905 f9c2e8e5 2022-02-13 thomas uint8_t type;
1906 f9c2e8e5 2022-02-13 thomas uint64_t size;
1907 f9c2e8e5 2022-02-13 thomas size_t tslen, delta_hdrlen;
1909 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
1910 f9c2e8e5 2022-02-13 thomas *delta_size = 0;
1911 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
1912 f9c2e8e5 2022-02-13 thomas *delta_offset = 0;
1913 c44c7d6e 2022-12-04 thomas *delta_data_offset = 0;
1914 f9c2e8e5 2022-02-13 thomas *base_offset = 0;
1915 f9c2e8e5 2022-02-13 thomas *base_size = 0;
1916 f9c2e8e5 2022-02-13 thomas *result_size = 0;
1918 f9c2e8e5 2022-02-13 thomas offset = got_packidx_get_object_offset(packidx, idx);
1919 e4a6dbc6 2022-07-24 thomas if (offset == -1)
1920 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
1922 f9c2e8e5 2022-02-13 thomas if (offset >= pack->filesize)
1923 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
1925 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1926 f9c2e8e5 2022-02-13 thomas pack, offset);
1927 f9c2e8e5 2022-02-13 thomas if (err)
1928 f9c2e8e5 2022-02-13 thomas return err;
1930 f9c2e8e5 2022-02-13 thomas if (tslen + size < tslen || offset + size < size ||
1931 f9c2e8e5 2022-02-13 thomas tslen + offset < tslen)
1932 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
1934 f9c2e8e5 2022-02-13 thomas switch (type) {
1935 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_OFFSET_DELTA:
1936 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1937 f9c2e8e5 2022-02-13 thomas pack, offset, tslen);
1938 f9c2e8e5 2022-02-13 thomas if (err)
1939 f9c2e8e5 2022-02-13 thomas return err;
1941 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_REF_DELTA:
1942 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1943 f9c2e8e5 2022-02-13 thomas if (err)
1944 f9c2e8e5 2022-02-13 thomas return err;
1945 f9c2e8e5 2022-02-13 thomas delta_hdrlen = SHA1_DIGEST_LENGTH;
1947 f9c2e8e5 2022-02-13 thomas default:
1948 f9c2e8e5 2022-02-13 thomas return got_error_fmt(GOT_ERR_OBJ_TYPE,
1949 e4a6dbc6 2022-07-24 thomas "non-delta object type %d found at offset %lld",
1950 e4a6dbc6 2022-07-24 thomas type, (long long)offset);
1953 f9c2e8e5 2022-02-13 thomas if (tslen + delta_hdrlen < delta_hdrlen ||
1954 f9c2e8e5 2022-02-13 thomas offset + delta_hdrlen < delta_hdrlen)
1955 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_DELTA);
1957 c44c7d6e 2022-12-04 thomas *delta_data_offset = offset + tslen + delta_hdrlen;
1958 9249e7e3 2022-05-12 thomas err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1959 c44c7d6e 2022-12-04 thomas base_size, result_size, *delta_data_offset, pack, packidx);
1960 f9c2e8e5 2022-02-13 thomas if (err)
1961 f9c2e8e5 2022-02-13 thomas return err;
1963 f9c2e8e5 2022-02-13 thomas if (*delta_size != size) {
1964 f9c2e8e5 2022-02-13 thomas err = got_error(GOT_ERR_BAD_DELTA);
1965 f9c2e8e5 2022-02-13 thomas goto done;
1968 f9c2e8e5 2022-02-13 thomas *delta_offset = offset;
1970 f9c2e8e5 2022-02-13 thomas if (err) {
1971 f9c2e8e5 2022-02-13 thomas free(*delta_buf);
1972 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
1973 9249e7e3 2022-05-12 thomas *delta_size = 0;
1974 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
1975 9249e7e3 2022-05-12 thomas *delta_offset = 0;
1976 9249e7e3 2022-05-12 thomas *base_offset = 0;
1977 9249e7e3 2022-05-12 thomas *base_size = 0;
1978 9249e7e3 2022-05-12 thomas *result_size = 0;
1980 f9c2e8e5 2022-02-13 thomas return err;
