Blame


1 0a0a3048 2018-01-10 stsp /*
2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3 0a0a3048 2018-01-10 stsp *
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.
7 0a0a3048 2018-01-10 stsp *
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.
15 0a0a3048 2018-01-10 stsp */
16 8cff5e95 2023-03-08 thomas #include "got_compat.h"
17 0a0a3048 2018-01-10 stsp
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>
25 0a0a3048 2018-01-10 stsp
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>
35 0a0a3048 2018-01-10 stsp
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"
39 0a0a3048 2018-01-10 stsp
40 be288a59 2023-02-23 thomas #include "got_lib_hash.h"
41 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
42 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
43 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
44 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
45 232c0ac1 2023-04-22 thomas #include "got_lib_object_qid.h"
46 dd88155e 2019-06-29 stsp #include "got_lib_object_parse.h"
47 876c234b 2018-09-10 stsp #include "got_lib_privsep.h"
48 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
49 79b11c62 2018-03-09 stsp
50 79b11c62 2018-03-09 stsp #ifndef nitems
51 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
52 79b11c62 2018-03-09 stsp #endif
53 1411938b 2018-02-12 stsp
54 a1fd68d8 2018-01-12 stsp #ifndef MIN
55 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
56 a1fd68d8 2018-01-12 stsp #endif
57 a1fd68d8 2018-01-12 stsp
58 0a0a3048 2018-01-10 stsp static const struct got_error *
59 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
60 0a0a3048 2018-01-10 stsp {
61 0a0a3048 2018-01-10 stsp int i;
62 0a0a3048 2018-01-10 stsp
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);
66 0a0a3048 2018-01-10 stsp }
67 0a0a3048 2018-01-10 stsp
68 0a0a3048 2018-01-10 stsp return NULL;
69 0a0a3048 2018-01-10 stsp }
70 0a0a3048 2018-01-10 stsp
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)
73 0a0a3048 2018-01-10 stsp {
74 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
75 b16893ba 2023-02-24 thomas enum got_hash_algorithm algo = GOT_HASH_SHA1;
76 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
77 b16893ba 2023-02-24 thomas struct got_hash ctx;
78 b16893ba 2023-02-24 thomas uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
79 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
80 817c5a18 2018-09-09 stsp ssize_t n;
81 5e6be232 2019-11-08 stsp int i;
82 0a0a3048 2018-01-10 stsp
83 b16893ba 2023-02-24 thomas got_hash_init(&ctx, algo);
84 0ebaf008 2018-01-10 stsp
85 57b35b75 2018-06-22 stsp h = &p->hdr;
86 57b35b75 2018-06-22 stsp offset = 0;
87 57b35b75 2018-06-22 stsp remain = p->len;
88 0a0a3048 2018-01-10 stsp
89 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
90 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
91 0a0a3048 2018-01-10 stsp goto done;
92 0a0a3048 2018-01-10 stsp }
93 fc79a48d 2018-07-09 stsp if (p->map)
94 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
95 fc79a48d 2018-07-09 stsp else {
96 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
97 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
98 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
99 fc79a48d 2018-07-09 stsp goto done;
100 fc79a48d 2018-07-09 stsp }
101 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
102 b1317e77 2019-09-22 stsp if (n < 0) {
103 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
104 b1317e77 2019-09-22 stsp goto done;
105 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
106 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
107 fc79a48d 2018-07-09 stsp goto done;
108 fc79a48d 2018-07-09 stsp }
109 fc79a48d 2018-07-09 stsp }
110 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
111 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
112 57b35b75 2018-06-22 stsp goto done;
113 57b35b75 2018-06-22 stsp }
114 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
115 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
116 0a0a3048 2018-01-10 stsp
117 0cb74cf4 2018-07-08 stsp if (verify)
118 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->magic, sizeof(*h->magic));
119 0ebaf008 2018-01-10 stsp
120 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
121 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
122 0a0a3048 2018-01-10 stsp goto done;
123 0a0a3048 2018-01-10 stsp }
124 fc79a48d 2018-07-09 stsp if (p->map)
125 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
126 fc79a48d 2018-07-09 stsp else {
127 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
128 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
129 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
130 fc79a48d 2018-07-09 stsp goto done;
131 fc79a48d 2018-07-09 stsp }
132 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
133 c6368c2e 2019-10-11 stsp if (n < 0) {
134 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
135 c6368c2e 2019-10-11 stsp goto done;
136 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
137 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
138 fc79a48d 2018-07-09 stsp goto done;
139 fc79a48d 2018-07-09 stsp }
140 fc79a48d 2018-07-09 stsp }
141 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
142 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
143 0a0a3048 2018-01-10 stsp goto done;
144 0a0a3048 2018-01-10 stsp }
145 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
146 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
147 0a0a3048 2018-01-10 stsp
148 0cb74cf4 2018-07-08 stsp if (verify)
149 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->version, sizeof(*h->version));
150 0ebaf008 2018-01-10 stsp
151 57b35b75 2018-06-22 stsp len_fanout =
152 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
153 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
154 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
155 0a0a3048 2018-01-10 stsp goto done;
156 0a0a3048 2018-01-10 stsp }
157 fc79a48d 2018-07-09 stsp if (p->map)
158 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
159 fc79a48d 2018-07-09 stsp else {
160 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
161 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
162 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
163 fc79a48d 2018-07-09 stsp goto done;
164 fc79a48d 2018-07-09 stsp }
165 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
166 c6368c2e 2019-10-11 stsp if (n < 0) {
167 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
168 c6368c2e 2019-10-11 stsp goto done;
169 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
170 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
171 fc79a48d 2018-07-09 stsp goto done;
172 fc79a48d 2018-07-09 stsp }
173 fc79a48d 2018-07-09 stsp }
174 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
175 0a0a3048 2018-01-10 stsp if (err)
176 0a0a3048 2018-01-10 stsp goto done;
177 0cb74cf4 2018-07-08 stsp if (verify)
178 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->fanout_table, len_fanout);
179 57b35b75 2018-06-22 stsp offset += len_fanout;
180 57b35b75 2018-06-22 stsp remain -= len_fanout;
181 0a0a3048 2018-01-10 stsp
182 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
183 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
184 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
185 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
186 0a0a3048 2018-01-10 stsp goto done;
187 0a0a3048 2018-01-10 stsp }
188 fc79a48d 2018-07-09 stsp if (p->map)
189 fc79a48d 2018-07-09 stsp h->sorted_ids =
190 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
191 fc79a48d 2018-07-09 stsp else {
192 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
193 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
194 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
195 fc79a48d 2018-07-09 stsp goto done;
196 fc79a48d 2018-07-09 stsp }
197 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
198 faaa1c0f 2018-09-15 stsp if (n < 0)
199 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
200 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
201 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
202 fc79a48d 2018-07-09 stsp goto done;
203 fc79a48d 2018-07-09 stsp }
204 fc79a48d 2018-07-09 stsp }
205 0cb74cf4 2018-07-08 stsp if (verify)
206 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->sorted_ids, len_ids);
207 57b35b75 2018-06-22 stsp offset += len_ids;
208 57b35b75 2018-06-22 stsp remain -= len_ids;
209 0a0a3048 2018-01-10 stsp
210 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
211 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
212 0a0a3048 2018-01-10 stsp goto done;
213 0a0a3048 2018-01-10 stsp }
214 fc79a48d 2018-07-09 stsp if (p->map)
215 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
216 fc79a48d 2018-07-09 stsp else {
217 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
218 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
219 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
220 fc79a48d 2018-07-09 stsp goto done;
221 fc79a48d 2018-07-09 stsp }
222 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
223 faaa1c0f 2018-09-15 stsp if (n < 0)
224 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
225 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
226 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
227 fc79a48d 2018-07-09 stsp goto done;
228 fc79a48d 2018-07-09 stsp }
229 fc79a48d 2018-07-09 stsp }
230 0cb74cf4 2018-07-08 stsp if (verify)
231 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->crc32, nobj * sizeof(*h->crc32));
232 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
233 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
234 0ebaf008 2018-01-10 stsp
235 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
236 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
237 0a0a3048 2018-01-10 stsp goto done;
238 0a0a3048 2018-01-10 stsp }
239 fc79a48d 2018-07-09 stsp if (p->map)
240 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
241 fc79a48d 2018-07-09 stsp else {
242 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
243 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
244 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
245 fc79a48d 2018-07-09 stsp goto done;
246 fc79a48d 2018-07-09 stsp }
247 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
248 faaa1c0f 2018-09-15 stsp if (n < 0)
249 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
250 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
251 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
252 fc79a48d 2018-07-09 stsp goto done;
253 fc79a48d 2018-07-09 stsp }
254 fc79a48d 2018-07-09 stsp }
255 0cb74cf4 2018-07-08 stsp if (verify)
256 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->offsets, nobj * sizeof(*h->offsets));
257 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->offsets);
258 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->offsets);
259 0ebaf008 2018-01-10 stsp
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++;
266 c3564dfa 2021-07-15 stsp }
267 5e6be232 2019-11-08 stsp }
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);
272 c3564dfa 2021-07-15 stsp goto done;
273 c3564dfa 2021-07-15 stsp }
274 0a0a3048 2018-01-10 stsp
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);
277 0a0a3048 2018-01-10 stsp goto done;
278 0a0a3048 2018-01-10 stsp }
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));
281 fc79a48d 2018-07-09 stsp else {
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");
286 fc79a48d 2018-07-09 stsp goto done;
287 fc79a48d 2018-07-09 stsp }
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));
290 faaa1c0f 2018-09-15 stsp if (n < 0)
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);
294 fc79a48d 2018-07-09 stsp goto done;
295 fc79a48d 2018-07-09 stsp }
296 fc79a48d 2018-07-09 stsp }
297 0cb74cf4 2018-07-08 stsp if (verify)
298 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->large_offsets,
299 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
300 5e6be232 2019-11-08 stsp remain -= p->nlargeobj * sizeof(*h->large_offsets);
301 5e6be232 2019-11-08 stsp offset += p->nlargeobj * sizeof(*h->large_offsets);
302 0ebaf008 2018-01-10 stsp
303 0a0a3048 2018-01-10 stsp checksum:
304 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->trailer)) {
305 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
306 0a0a3048 2018-01-10 stsp goto done;
307 0a0a3048 2018-01-10 stsp }
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));
311 fc79a48d 2018-07-09 stsp else {
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");
315 fc79a48d 2018-07-09 stsp goto done;
316 fc79a48d 2018-07-09 stsp }
317 fc79a48d 2018-07-09 stsp n = read(p->fd, h->trailer, sizeof(*h->trailer));
318 faaa1c0f 2018-09-15 stsp if (n < 0)
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);
322 fc79a48d 2018-07-09 stsp goto done;
323 fc79a48d 2018-07-09 stsp }
324 fc79a48d 2018-07-09 stsp }
325 0cb74cf4 2018-07-08 stsp if (verify) {
326 b16893ba 2023-02-24 thomas got_hash_update(&ctx, h->trailer->packfile_sha1,
327 b16893ba 2023-02-24 thomas SHA1_DIGEST_LENGTH);
328 b16893ba 2023-02-24 thomas got_hash_final(&ctx, hash);
329 1e73031b 2023-04-04 thomas if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1) != 0)
330 0cb74cf4 2018-07-08 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
331 0cb74cf4 2018-07-08 stsp }
332 0a0a3048 2018-01-10 stsp done:
333 817c5a18 2018-09-09 stsp return err;
334 817c5a18 2018-09-09 stsp }
335 817c5a18 2018-09-09 stsp
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)
339 817c5a18 2018-09-09 stsp {
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;
344 817c5a18 2018-09-09 stsp
345 817c5a18 2018-09-09 stsp *packidx = NULL;
346 1124fe40 2021-07-07 stsp
347 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath, relpath);
348 1124fe40 2021-07-07 stsp if (err)
349 1124fe40 2021-07-07 stsp return err;
350 1124fe40 2021-07-07 stsp
351 1124fe40 2021-07-07 stsp /*
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.
355 1124fe40 2021-07-07 stsp */
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);
360 1124fe40 2021-07-07 stsp } else
361 1124fe40 2021-07-07 stsp err = got_error_from_errno2("fstatat", pack_relpath);
362 1124fe40 2021-07-07 stsp goto done;
363 1124fe40 2021-07-07 stsp }
364 817c5a18 2018-09-09 stsp
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");
368 1124fe40 2021-07-07 stsp goto done;
369 1124fe40 2021-07-07 stsp }
370 817c5a18 2018-09-09 stsp
371 fc63f50d 2021-12-31 thomas p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
372 6772cf22 2019-08-28 hiltjo if (p->fd == -1) {
373 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("openat", relpath);
374 1124fe40 2021-07-07 stsp goto done;
375 6772cf22 2019-08-28 hiltjo }
376 817c5a18 2018-09-09 stsp
377 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
378 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
379 1124fe40 2021-07-07 stsp goto done;
380 817c5a18 2018-09-09 stsp }
381 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
382 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
383 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
384 1124fe40 2021-07-07 stsp goto done;
385 817c5a18 2018-09-09 stsp }
386 817c5a18 2018-09-09 stsp
387 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
388 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
389 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
390 817c5a18 2018-09-09 stsp goto done;
391 817c5a18 2018-09-09 stsp }
392 817c5a18 2018-09-09 stsp
393 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
394 aa75acde 2022-10-25 thomas if (p->len > 0 && p->len <= SIZE_MAX) {
395 aa75acde 2022-10-25 thomas p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
396 aa75acde 2022-10-25 thomas if (p->map == MAP_FAILED) {
397 aa75acde 2022-10-25 thomas if (errno != ENOMEM) {
398 aa75acde 2022-10-25 thomas err = got_error_from_errno("mmap");
399 aa75acde 2022-10-25 thomas goto done;
400 aa75acde 2022-10-25 thomas }
401 aa75acde 2022-10-25 thomas p->map = NULL; /* fall back to read(2) */
402 3a11398b 2019-02-21 stsp }
403 3a11398b 2019-02-21 stsp }
404 817c5a18 2018-09-09 stsp #endif
405 817c5a18 2018-09-09 stsp
406 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
407 817c5a18 2018-09-09 stsp done:
408 1124fe40 2021-07-07 stsp if (err) {
409 1124fe40 2021-07-07 stsp if (p)
410 1124fe40 2021-07-07 stsp got_packidx_close(p);
411 1124fe40 2021-07-07 stsp } else
412 0a0a3048 2018-01-10 stsp *packidx = p;
413 1124fe40 2021-07-07 stsp free(pack_relpath);
414 0a0a3048 2018-01-10 stsp return err;
415 0a0a3048 2018-01-10 stsp }
416 0a0a3048 2018-01-10 stsp
417 57b35b75 2018-06-22 stsp const struct got_error *
418 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
419 0a0a3048 2018-01-10 stsp {
420 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
421 57b35b75 2018-06-22 stsp
422 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
423 fc79a48d 2018-07-09 stsp if (packidx->map) {
424 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
425 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
426 fc79a48d 2018-07-09 stsp } else {
427 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
428 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
429 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
434 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
435 57b35b75 2018-06-22 stsp }
436 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
437 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
438 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_offsets);
439 f9c2e8e5 2022-02-13 thomas free(packidx->sorted_large_offsets);
440 0a0a3048 2018-01-10 stsp free(packidx);
441 57b35b75 2018-06-22 stsp
442 57b35b75 2018-06-22 stsp return err;
443 a1fd68d8 2018-01-12 stsp }
444 509c9973 2021-04-10 stsp
445 509c9973 2021-04-10 stsp const struct got_error *
446 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
447 509c9973 2021-04-10 stsp {
448 509c9973 2021-04-10 stsp size_t size;
449 509c9973 2021-04-10 stsp
450 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
451 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
452 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
453 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
454 a1fd68d8 2018-01-12 stsp
455 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
456 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
457 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
458 509c9973 2021-04-10 stsp
459 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
460 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
461 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
462 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
463 509c9973 2021-04-10 stsp
464 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
465 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
466 509c9973 2021-04-10 stsp
467 509c9973 2021-04-10 stsp return NULL;
468 509c9973 2021-04-10 stsp }
469 509c9973 2021-04-10 stsp
470 02828bfd 2021-06-22 stsp off_t
471 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
472 a1fd68d8 2018-01-12 stsp {
473 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
474 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
475 a1fd68d8 2018-01-12 stsp uint64_t loffset;
476 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
477 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
478 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
479 a1fd68d8 2018-01-12 stsp return -1;
480 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
481 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
482 a1fd68d8 2018-01-12 stsp }
483 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
484 a1fd68d8 2018-01-12 stsp }
485 a1fd68d8 2018-01-12 stsp
486 1510f469 2018-09-09 stsp int
487 48b4f239 2021-12-31 thomas got_packidx_get_object_idx(struct got_packidx *packidx,
488 48b4f239 2021-12-31 thomas struct got_object_id *id)
489 a1fd68d8 2018-01-12 stsp {
490 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
491 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
492 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
493 a1fd68d8 2018-01-12 stsp
494 a1fd68d8 2018-01-12 stsp if (id0 > 0)
495 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
496 a1fd68d8 2018-01-12 stsp
497 40aeb19c 2018-06-22 stsp while (left <= right) {
498 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
499 40aeb19c 2018-06-22 stsp int i, cmp;
500 a1fd68d8 2018-01-12 stsp
501 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
502 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
503 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
504 16dcbf91 2018-04-01 stsp if (cmp == 0)
505 6c00b545 2018-01-17 stsp return i;
506 40aeb19c 2018-06-22 stsp else if (cmp > 0)
507 40aeb19c 2018-06-22 stsp left = i + 1;
508 40aeb19c 2018-06-22 stsp else if (cmp < 0)
509 40aeb19c 2018-06-22 stsp right = i - 1;
510 a1fd68d8 2018-01-12 stsp }
511 a1fd68d8 2018-01-12 stsp
512 a1fd68d8 2018-01-12 stsp return -1;
513 f9c2e8e5 2022-02-13 thomas }
514 f9c2e8e5 2022-02-13 thomas
515 f9c2e8e5 2022-02-13 thomas static int
516 f9c2e8e5 2022-02-13 thomas offset_cmp(const void *pa, const void *pb)
517 f9c2e8e5 2022-02-13 thomas {
518 f9c2e8e5 2022-02-13 thomas const struct got_pack_offset_index *a, *b;
519 f9c2e8e5 2022-02-13 thomas
520 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_offset_index *)pa;
521 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_offset_index *)pb;
522 f9c2e8e5 2022-02-13 thomas
523 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
524 f9c2e8e5 2022-02-13 thomas return -1;
525 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
526 f9c2e8e5 2022-02-13 thomas return 1;
527 f9c2e8e5 2022-02-13 thomas
528 f9c2e8e5 2022-02-13 thomas return 0;
529 876c234b 2018-09-10 stsp }
530 876c234b 2018-09-10 stsp
531 f9c2e8e5 2022-02-13 thomas static int
532 f9c2e8e5 2022-02-13 thomas large_offset_cmp(const void *pa, const void *pb)
533 f9c2e8e5 2022-02-13 thomas {
534 f9c2e8e5 2022-02-13 thomas const struct got_pack_large_offset_index *a, *b;
535 f9c2e8e5 2022-02-13 thomas
536 f9c2e8e5 2022-02-13 thomas a = (const struct got_pack_large_offset_index *)pa;
537 f9c2e8e5 2022-02-13 thomas b = (const struct got_pack_large_offset_index *)pb;
538 f9c2e8e5 2022-02-13 thomas
539 f9c2e8e5 2022-02-13 thomas if (a->offset < b->offset)
540 f9c2e8e5 2022-02-13 thomas return -1;
541 f9c2e8e5 2022-02-13 thomas else if (a->offset > b->offset)
542 f9c2e8e5 2022-02-13 thomas return 1;
543 f9c2e8e5 2022-02-13 thomas
544 f9c2e8e5 2022-02-13 thomas return 0;
545 f9c2e8e5 2022-02-13 thomas }
546 f9c2e8e5 2022-02-13 thomas
547 f9c2e8e5 2022-02-13 thomas static const struct got_error *
548 f9c2e8e5 2022-02-13 thomas build_offset_index(struct got_packidx *p)
549 f9c2e8e5 2022-02-13 thomas {
550 f9c2e8e5 2022-02-13 thomas uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
551 f9c2e8e5 2022-02-13 thomas unsigned int i, j, k;
552 f9c2e8e5 2022-02-13 thomas
553 f9c2e8e5 2022-02-13 thomas p->sorted_offsets = calloc(nobj - p->nlargeobj,
554 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]));
555 f9c2e8e5 2022-02-13 thomas if (p->sorted_offsets == NULL)
556 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
557 f9c2e8e5 2022-02-13 thomas
558 f9c2e8e5 2022-02-13 thomas if (p->nlargeobj > 0) {
559 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets = calloc(p->nlargeobj,
560 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]));
561 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets == NULL)
562 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("calloc");
563 f9c2e8e5 2022-02-13 thomas }
564 f9c2e8e5 2022-02-13 thomas
565 f9c2e8e5 2022-02-13 thomas j = 0;
566 f9c2e8e5 2022-02-13 thomas k = 0;
567 f9c2e8e5 2022-02-13 thomas for (i = 0; i < nobj; i++) {
568 f9c2e8e5 2022-02-13 thomas uint32_t offset = be32toh(p->hdr.offsets[i]);
569 f9c2e8e5 2022-02-13 thomas if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
570 f9c2e8e5 2022-02-13 thomas uint64_t loffset;
571 f9c2e8e5 2022-02-13 thomas uint32_t idx;
572 f9c2e8e5 2022-02-13 thomas idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
573 f9c2e8e5 2022-02-13 thomas if (idx >= p->nlargeobj ||
574 f9c2e8e5 2022-02-13 thomas p->nlargeobj == 0 ||
575 f9c2e8e5 2022-02-13 thomas p->hdr.large_offsets == NULL)
576 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
577 f9c2e8e5 2022-02-13 thomas loffset = be64toh(p->hdr.large_offsets[idx]);
578 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].offset = loffset;
579 f9c2e8e5 2022-02-13 thomas p->sorted_large_offsets[j].idx = i;
580 f9c2e8e5 2022-02-13 thomas j++;
581 f9c2e8e5 2022-02-13 thomas } else {
582 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].offset = offset;
583 f9c2e8e5 2022-02-13 thomas p->sorted_offsets[k].idx = i;
584 f9c2e8e5 2022-02-13 thomas k++;
585 f9c2e8e5 2022-02-13 thomas }
586 f9c2e8e5 2022-02-13 thomas }
587 f9c2e8e5 2022-02-13 thomas if (j != p->nlargeobj || k != nobj - p->nlargeobj)
588 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
589 f9c2e8e5 2022-02-13 thomas
590 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_offsets, nobj - p->nlargeobj,
591 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_offsets[0]), offset_cmp);
592 f9c2e8e5 2022-02-13 thomas
593 f9c2e8e5 2022-02-13 thomas if (p->sorted_large_offsets)
594 f9c2e8e5 2022-02-13 thomas qsort(p->sorted_large_offsets, p->nlargeobj,
595 f9c2e8e5 2022-02-13 thomas sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
596 f9c2e8e5 2022-02-13 thomas
597 f9c2e8e5 2022-02-13 thomas return NULL;
598 f9c2e8e5 2022-02-13 thomas }
599 f9c2e8e5 2022-02-13 thomas
600 876c234b 2018-09-10 stsp const struct got_error *
601 f9c2e8e5 2022-02-13 thomas got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
602 f9c2e8e5 2022-02-13 thomas {
603 f9c2e8e5 2022-02-13 thomas const struct got_error *err;
604 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
605 f9c2e8e5 2022-02-13 thomas int i, left, right;
606 f9c2e8e5 2022-02-13 thomas
607 f9c2e8e5 2022-02-13 thomas *idx = -1;
608 f9c2e8e5 2022-02-13 thomas
609 f9c2e8e5 2022-02-13 thomas if (packidx->sorted_offsets == NULL) {
610 f9c2e8e5 2022-02-13 thomas err = build_offset_index(packidx);
611 f9c2e8e5 2022-02-13 thomas if (err)
612 f9c2e8e5 2022-02-13 thomas return err;
613 f9c2e8e5 2022-02-13 thomas }
614 f9c2e8e5 2022-02-13 thomas
615 f9c2e8e5 2022-02-13 thomas if (offset >= 0x7fffffff) {
616 f9c2e8e5 2022-02-13 thomas uint64_t lo;
617 f9c2e8e5 2022-02-13 thomas left = 0, right = packidx->nlargeobj - 1;
618 f9c2e8e5 2022-02-13 thomas while (left <= right) {
619 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
620 f9c2e8e5 2022-02-13 thomas lo = packidx->sorted_large_offsets[i].offset;
621 f9c2e8e5 2022-02-13 thomas if (lo == offset) {
622 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_large_offsets[i].idx;
623 f9c2e8e5 2022-02-13 thomas break;
624 f9c2e8e5 2022-02-13 thomas } else if (offset > lo)
625 f9c2e8e5 2022-02-13 thomas left = i + 1;
626 f9c2e8e5 2022-02-13 thomas else if (offset < lo)
627 f9c2e8e5 2022-02-13 thomas right = i - 1;
628 f9c2e8e5 2022-02-13 thomas }
629 f9c2e8e5 2022-02-13 thomas } else {
630 f9c2e8e5 2022-02-13 thomas uint32_t o;
631 f9c2e8e5 2022-02-13 thomas left = 0, right = totobj - packidx->nlargeobj - 1;
632 f9c2e8e5 2022-02-13 thomas while (left <= right) {
633 f9c2e8e5 2022-02-13 thomas i = ((left + right) / 2);
634 f9c2e8e5 2022-02-13 thomas o = packidx->sorted_offsets[i].offset;
635 f9c2e8e5 2022-02-13 thomas if (o == offset) {
636 f9c2e8e5 2022-02-13 thomas *idx = packidx->sorted_offsets[i].idx;
637 f9c2e8e5 2022-02-13 thomas break;
638 f9c2e8e5 2022-02-13 thomas } else if (offset > o)
639 f9c2e8e5 2022-02-13 thomas left = i + 1;
640 f9c2e8e5 2022-02-13 thomas else if (offset < o)
641 f9c2e8e5 2022-02-13 thomas right = i - 1;
642 f9c2e8e5 2022-02-13 thomas }
643 f9c2e8e5 2022-02-13 thomas }
644 f9c2e8e5 2022-02-13 thomas
645 f9c2e8e5 2022-02-13 thomas return NULL;
646 f9c2e8e5 2022-02-13 thomas }
647 f9c2e8e5 2022-02-13 thomas
648 f9c2e8e5 2022-02-13 thomas const struct got_error *
649 f9c2e8e5 2022-02-13 thomas got_packidx_get_object_id(struct got_object_id *id,
650 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx, int idx)
651 f9c2e8e5 2022-02-13 thomas {
652 f9c2e8e5 2022-02-13 thomas uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
653 f9c2e8e5 2022-02-13 thomas struct got_packidx_object_id *oid;
654 f9c2e8e5 2022-02-13 thomas
655 f9c2e8e5 2022-02-13 thomas if (idx < 0 || idx >= totobj)
656 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_NO_OBJ);
657 f9c2e8e5 2022-02-13 thomas
658 f9c2e8e5 2022-02-13 thomas oid = &packidx->hdr.sorted_ids[idx];
659 f9c2e8e5 2022-02-13 thomas memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
660 f9c2e8e5 2022-02-13 thomas return NULL;
661 f9c2e8e5 2022-02-13 thomas }
662 f9c2e8e5 2022-02-13 thomas
663 f9c2e8e5 2022-02-13 thomas const struct got_error *
664 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
665 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
666 e09a504c 2019-06-28 stsp {
667 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
668 4277420a 2019-06-29 stsp u_int8_t id0;
669 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
670 4277420a 2019-06-29 stsp char hex[3];
671 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
672 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
673 ec138a87 2022-01-03 thomas uint32_t i = 0;
674 e09a504c 2019-06-28 stsp
675 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
676 4277420a 2019-06-29 stsp
677 4277420a 2019-06-29 stsp if (prefix_len < 2)
678 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
679 4277420a 2019-06-29 stsp
680 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
681 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
682 4277420a 2019-06-29 stsp hex[2] = '\0';
683 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
684 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
685 4277420a 2019-06-29 stsp
686 ec138a87 2022-01-03 thomas if (id0 > 0)
687 ec138a87 2022-01-03 thomas i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
688 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
689 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
690 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
691 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
692 4277420a 2019-06-29 stsp int cmp;
693 4277420a 2019-06-29 stsp
694 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
695 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
696 4277420a 2019-06-29 stsp
697 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
698 4277420a 2019-06-29 stsp if (cmp < 0) {
699 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
700 4277420a 2019-06-29 stsp continue;
701 4277420a 2019-06-29 stsp } else if (cmp > 0)
702 e09a504c 2019-06-28 stsp break;
703 4277420a 2019-06-29 stsp
704 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
705 dd88155e 2019-06-29 stsp if (err)
706 dd88155e 2019-06-29 stsp break;
707 ec242592 2022-04-22 thomas memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
708 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
709 4277420a 2019-06-29 stsp
710 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
711 e09a504c 2019-06-28 stsp }
712 e09a504c 2019-06-28 stsp
713 0adc7bcc 2019-06-29 stsp if (err)
714 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
715 68036464 2022-06-26 thomas return err;
716 68036464 2022-06-26 thomas }
717 68036464 2022-06-26 thomas
718 68036464 2022-06-26 thomas static void
719 68036464 2022-06-26 thomas set_max_datasize(void)
720 68036464 2022-06-26 thomas {
721 68036464 2022-06-26 thomas struct rlimit rl;
722 68036464 2022-06-26 thomas
723 68036464 2022-06-26 thomas if (getrlimit(RLIMIT_DATA, &rl) != 0)
724 68036464 2022-06-26 thomas return;
725 68036464 2022-06-26 thomas
726 68036464 2022-06-26 thomas rl.rlim_cur = rl.rlim_max;
727 68036464 2022-06-26 thomas setrlimit(RLIMIT_DATA, &rl);
728 68036464 2022-06-26 thomas }
729 68036464 2022-06-26 thomas
730 68036464 2022-06-26 thomas const struct got_error *
731 68036464 2022-06-26 thomas got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
732 68036464 2022-06-26 thomas {
733 68036464 2022-06-26 thomas const struct got_error *err = NULL;
734 68036464 2022-06-26 thomas int imsg_fds[2];
735 68036464 2022-06-26 thomas pid_t pid;
736 68036464 2022-06-26 thomas struct imsgbuf *ibuf;
737 68036464 2022-06-26 thomas
738 68036464 2022-06-26 thomas ibuf = calloc(1, sizeof(*ibuf));
739 68036464 2022-06-26 thomas if (ibuf == NULL)
740 68036464 2022-06-26 thomas return got_error_from_errno("calloc");
741 68036464 2022-06-26 thomas
742 68036464 2022-06-26 thomas pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
743 68036464 2022-06-26 thomas if (pack->privsep_child == NULL) {
744 68036464 2022-06-26 thomas err = got_error_from_errno("calloc");
745 68036464 2022-06-26 thomas free(ibuf);
746 68036464 2022-06-26 thomas return err;
747 68036464 2022-06-26 thomas }
748 68036464 2022-06-26 thomas pack->child_has_tempfiles = 0;
749 68036464 2022-06-26 thomas pack->child_has_delta_outfd = 0;
750 68036464 2022-06-26 thomas
751 68036464 2022-06-26 thomas if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
752 68036464 2022-06-26 thomas err = got_error_from_errno("socketpair");
753 68036464 2022-06-26 thomas goto done;
754 68036464 2022-06-26 thomas }
755 68036464 2022-06-26 thomas
756 68036464 2022-06-26 thomas pid = fork();
757 68036464 2022-06-26 thomas if (pid == -1) {
758 68036464 2022-06-26 thomas err = got_error_from_errno("fork");
759 68036464 2022-06-26 thomas goto done;
760 68036464 2022-06-26 thomas } else if (pid == 0) {
761 68036464 2022-06-26 thomas set_max_datasize();
762 68036464 2022-06-26 thomas got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
763 68036464 2022-06-26 thomas pack->path_packfile);
764 68036464 2022-06-26 thomas /* not reached */
765 68036464 2022-06-26 thomas }
766 68036464 2022-06-26 thomas
767 68036464 2022-06-26 thomas if (close(imsg_fds[1]) == -1)
768 68036464 2022-06-26 thomas return got_error_from_errno("close");
769 68036464 2022-06-26 thomas pack->privsep_child->imsg_fd = imsg_fds[0];
770 68036464 2022-06-26 thomas pack->privsep_child->pid = pid;
771 68036464 2022-06-26 thomas imsg_init(ibuf, imsg_fds[0]);
772 68036464 2022-06-26 thomas pack->privsep_child->ibuf = ibuf;
773 68036464 2022-06-26 thomas
774 68036464 2022-06-26 thomas err = got_privsep_init_pack_child(ibuf, pack, packidx);
775 68036464 2022-06-26 thomas if (err) {
776 68036464 2022-06-26 thomas const struct got_error *child_err;
777 68036464 2022-06-26 thomas err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
778 68036464 2022-06-26 thomas child_err = got_privsep_wait_for_child(
779 68036464 2022-06-26 thomas pack->privsep_child->pid);
780 68036464 2022-06-26 thomas if (child_err && err == NULL)
781 68036464 2022-06-26 thomas err = child_err;
782 68036464 2022-06-26 thomas }
783 68036464 2022-06-26 thomas done:
784 68036464 2022-06-26 thomas if (err) {
785 68036464 2022-06-26 thomas free(ibuf);
786 68036464 2022-06-26 thomas free(pack->privsep_child);
787 68036464 2022-06-26 thomas pack->privsep_child = NULL;
788 68036464 2022-06-26 thomas }
789 dd88155e 2019-06-29 stsp return err;
790 e09a504c 2019-06-28 stsp }
791 e09a504c 2019-06-28 stsp
792 b4f37570 2021-06-19 stsp static const struct got_error *
793 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
794 876c234b 2018-09-10 stsp {
795 a6ca45d5 2022-10-24 thomas const struct got_error *err = NULL, *close_err = NULL;
796 876c234b 2018-09-10 stsp
797 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
798 876c234b 2018-09-10 stsp return NULL;
799 876c234b 2018-09-10 stsp
800 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
801 96732e0b 2018-11-11 stsp if (err)
802 96732e0b 2018-11-11 stsp return err;
803 a6ca45d5 2022-10-24 thomas if (close(pack->privsep_child->imsg_fd) == -1)
804 a6ca45d5 2022-10-24 thomas close_err = got_error_from_errno("close");
805 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
806 a6ca45d5 2022-10-24 thomas if (close_err && err == NULL)
807 a6ca45d5 2022-10-24 thomas err = close_err;
808 a6ca45d5 2022-10-24 thomas imsg_clear(pack->privsep_child->ibuf);
809 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
810 876c234b 2018-09-10 stsp free(pack->privsep_child);
811 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
812 876c234b 2018-09-10 stsp return err;
813 7e656b93 2018-03-17 stsp }
814 7e656b93 2018-03-17 stsp
815 d7464085 2018-07-09 stsp const struct got_error *
816 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
817 7e656b93 2018-03-17 stsp {
818 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
819 d7464085 2018-07-09 stsp
820 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
821 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
822 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
823 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
824 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
825 8b2180d4 2018-04-26 stsp pack->fd = -1;
826 4589e373 2018-03-17 stsp free(pack->path_packfile);
827 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
828 4589e373 2018-03-17 stsp pack->filesize = 0;
829 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
830 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
831 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
832 ab2f42e7 2019-11-10 stsp }
833 7e656b93 2018-03-17 stsp
834 bfb5ee0b 2022-05-31 thomas /*
835 bfb5ee0b 2022-05-31 thomas * Leave accumfd and basefd alone. They are managed by the
836 bfb5ee0b 2022-05-31 thomas * repository layer and can be reused.
837 bfb5ee0b 2022-05-31 thomas */
838 bfb5ee0b 2022-05-31 thomas
839 c7fe698a 2018-01-23 stsp return err;
840 c7fe698a 2018-01-23 stsp }
841 c7fe698a 2018-01-23 stsp
842 668a20f6 2020-03-18 stsp const struct got_error *
843 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
844 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
845 a487c1d0 2018-01-14 stsp {
846 a487c1d0 2018-01-14 stsp uint8_t t = 0;
847 a487c1d0 2018-01-14 stsp uint64_t s = 0;
848 a487c1d0 2018-01-14 stsp uint8_t sizeN;
849 d7464085 2018-07-09 stsp size_t mapoff = 0;
850 a487c1d0 2018-01-14 stsp int i = 0;
851 d7464085 2018-07-09 stsp
852 d7464085 2018-07-09 stsp *len = 0;
853 d7464085 2018-07-09 stsp
854 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
855 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
856 d7464085 2018-07-09 stsp
857 d7464085 2018-07-09 stsp if (pack->map) {
858 66e6097f 2022-10-27 thomas if (offset > SIZE_MAX) {
859 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
860 66e6097f 2022-10-27 thomas "offset %lld overflows size_t",
861 66e6097f 2022-10-27 thomas (long long)offset);
862 66e6097f 2022-10-27 thomas }
863 66e6097f 2022-10-27 thomas
864 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
865 d7464085 2018-07-09 stsp } else {
866 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
867 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
868 d7464085 2018-07-09 stsp }
869 a487c1d0 2018-01-14 stsp
870 a1fd68d8 2018-01-12 stsp do {
871 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
872 a487c1d0 2018-01-14 stsp if (i > 9)
873 eac8a741 2022-10-20 thomas return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
874 04fb0dd8 2022-10-24 thomas "packfile offset %lld", (long long)offset);
875 a1fd68d8 2018-01-12 stsp
876 d7464085 2018-07-09 stsp if (pack->map) {
877 1944573a 2022-01-10 thomas if (mapoff + sizeof(sizeN) >= pack->filesize)
878 1944573a 2022-01-10 thomas return got_error(GOT_ERR_BAD_PACKFILE);
879 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
880 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
881 d7464085 2018-07-09 stsp } else {
882 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
883 d7464085 2018-07-09 stsp if (n < 0)
884 638f9024 2019-05-13 stsp return got_error_from_errno("read");
885 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
886 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
887 d7464085 2018-07-09 stsp }
888 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
889 8251fdbc 2018-01-12 stsp
890 a1fd68d8 2018-01-12 stsp if (i == 0) {
891 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
892 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
893 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
894 a1fd68d8 2018-01-12 stsp } else {
895 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
896 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
897 a1fd68d8 2018-01-12 stsp }
898 a1fd68d8 2018-01-12 stsp i++;
899 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
900 a1fd68d8 2018-01-12 stsp
901 a487c1d0 2018-01-14 stsp *type = t;
902 a487c1d0 2018-01-14 stsp *size = s;
903 a487c1d0 2018-01-14 stsp return NULL;
904 0a0a3048 2018-01-10 stsp }
905 c54542a0 2018-01-13 stsp
906 a1fd68d8 2018-01-12 stsp static const struct got_error *
907 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
908 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
909 6ccb713b 2018-01-19 stsp {
910 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
911 6ccb713b 2018-01-19 stsp if (*obj == NULL)
912 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
913 6ccb713b 2018-01-19 stsp
914 6ccb713b 2018-01-19 stsp (*obj)->type = type;
915 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
916 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
917 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
918 6ccb713b 2018-01-19 stsp (*obj)->size = size;
919 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
920 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
921 b107e67f 2018-01-19 stsp
922 b107e67f 2018-01-19 stsp return NULL;
923 b107e67f 2018-01-19 stsp }
924 b107e67f 2018-01-19 stsp
925 b107e67f 2018-01-19 stsp static const struct got_error *
926 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
927 d7464085 2018-07-09 stsp off_t delta_offset)
928 b107e67f 2018-01-19 stsp {
929 b107e67f 2018-01-19 stsp int64_t o = 0;
930 b107e67f 2018-01-19 stsp uint8_t offN;
931 b107e67f 2018-01-19 stsp int i = 0;
932 b107e67f 2018-01-19 stsp
933 a0de39f3 2019-08-09 stsp *offset = 0;
934 d7464085 2018-07-09 stsp *len = 0;
935 d7464085 2018-07-09 stsp
936 b107e67f 2018-01-19 stsp do {
937 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
938 b107e67f 2018-01-19 stsp if (i > 8)
939 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
940 b107e67f 2018-01-19 stsp
941 d7464085 2018-07-09 stsp if (pack->map) {
942 d7464085 2018-07-09 stsp size_t mapoff;
943 66e6097f 2022-10-27 thomas
944 66e6097f 2022-10-27 thomas if (delta_offset + *len > SIZE_MAX) {
945 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
946 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
947 66e6097f 2022-10-27 thomas (long long)delta_offset + *len);
948 66e6097f 2022-10-27 thomas }
949 66e6097f 2022-10-27 thomas
950 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
951 1944573a 2022-01-10 thomas if (mapoff + sizeof(offN) >= pack->filesize)
952 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
953 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
954 d7464085 2018-07-09 stsp } else {
955 d7464085 2018-07-09 stsp ssize_t n;
956 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
957 d7464085 2018-07-09 stsp if (n < 0)
958 638f9024 2019-05-13 stsp return got_error_from_errno("read");
959 d7464085 2018-07-09 stsp if (n != sizeof(offN))
960 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
961 d7464085 2018-07-09 stsp }
962 d7464085 2018-07-09 stsp *len += sizeof(offN);
963 b107e67f 2018-01-19 stsp
964 b107e67f 2018-01-19 stsp if (i == 0)
965 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
966 b107e67f 2018-01-19 stsp else {
967 cecc778e 2018-01-23 stsp o++;
968 b107e67f 2018-01-19 stsp o <<= 7;
969 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
970 b107e67f 2018-01-19 stsp }
971 b107e67f 2018-01-19 stsp i++;
972 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
973 6ccb713b 2018-01-19 stsp
974 b107e67f 2018-01-19 stsp *offset = o;
975 6ccb713b 2018-01-19 stsp return NULL;
976 6ccb713b 2018-01-19 stsp }
977 6ccb713b 2018-01-19 stsp
978 668a20f6 2020-03-18 stsp const struct got_error *
979 48b4f239 2021-12-31 thomas got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
980 2424a69d 2022-10-24 thomas struct got_pack *pack, off_t offset, size_t tslen)
981 b107e67f 2018-01-19 stsp {
982 96f5e8b3 2018-01-23 stsp const struct got_error *err;
983 b107e67f 2018-01-19 stsp int64_t negoffset;
984 b107e67f 2018-01-19 stsp size_t negofflen;
985 b107e67f 2018-01-19 stsp
986 d7464085 2018-07-09 stsp *len = 0;
987 d7464085 2018-07-09 stsp
988 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
989 d7464085 2018-07-09 stsp offset + tslen);
990 b107e67f 2018-01-19 stsp if (err)
991 b107e67f 2018-01-19 stsp return err;
992 b107e67f 2018-01-19 stsp
993 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
994 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
995 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
996 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
997 b107e67f 2018-01-19 stsp
998 d7464085 2018-07-09 stsp *len = negofflen;
999 96f5e8b3 2018-01-23 stsp return NULL;
1000 96f5e8b3 2018-01-23 stsp }
1001 96f5e8b3 2018-01-23 stsp
1002 0e22967e 2018-02-11 stsp static const struct got_error *
1003 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1004 9249e7e3 2022-05-12 thomas size_t *delta_compressed_len, size_t delta_data_offset,
1005 9249e7e3 2022-05-12 thomas struct got_pack *pack)
1006 42c69117 2019-11-10 stsp {
1007 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
1008 9249e7e3 2022-05-12 thomas size_t consumed = 0;
1009 42c69117 2019-11-10 stsp
1010 42c69117 2019-11-10 stsp if (pack->map) {
1011 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
1012 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1013 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1014 9249e7e3 2022-05-12 thomas &consumed, NULL, pack->map, delta_data_offset,
1015 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
1016 9249e7e3 2022-05-12 thomas if (err)
1017 9249e7e3 2022-05-12 thomas return err;
1018 42c69117 2019-11-10 stsp } else {
1019 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1020 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
1021 9249e7e3 2022-05-12 thomas err = got_inflate_to_mem_fd(delta_buf, delta_len,
1022 9249e7e3 2022-05-12 thomas &consumed, NULL, 0, pack->fd);
1023 9249e7e3 2022-05-12 thomas if (err)
1024 9249e7e3 2022-05-12 thomas return err;
1025 42c69117 2019-11-10 stsp }
1026 9249e7e3 2022-05-12 thomas
1027 9249e7e3 2022-05-12 thomas if (delta_compressed_len)
1028 9249e7e3 2022-05-12 thomas *delta_compressed_len = consumed;
1029 9249e7e3 2022-05-12 thomas
1030 9249e7e3 2022-05-12 thomas return NULL;
1031 42c69117 2019-11-10 stsp }
1032 a3500804 2018-01-23 stsp
1033 96f5e8b3 2018-01-23 stsp static const struct got_error *
1034 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1035 e485d53c 2022-10-28 thomas int delta_type, size_t delta_size, off_t delta_data_offset)
1036 bdd2fbb3 2018-02-11 stsp {
1037 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
1038 bdd2fbb3 2018-02-11 stsp
1039 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1040 42c69117 2019-11-10 stsp delta_data_offset);
1041 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
1042 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
1043 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
1044 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
1045 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1046 bdd2fbb3 2018-02-11 stsp return NULL;
1047 bdd2fbb3 2018-02-11 stsp }
1048 bdd2fbb3 2018-02-11 stsp
1049 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1050 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1051 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1052 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1053 a3500804 2018-01-23 stsp {
1054 a3500804 2018-01-23 stsp const struct got_error *err;
1055 c3703302 2018-01-23 stsp off_t base_offset;
1056 c3703302 2018-01-23 stsp uint8_t base_type;
1057 c3703302 2018-01-23 stsp uint64_t base_size;
1058 c3703302 2018-01-23 stsp size_t base_tslen;
1059 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1060 42c69117 2019-11-10 stsp size_t consumed;
1061 a3500804 2018-01-23 stsp
1062 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1063 d7464085 2018-07-09 stsp delta_offset, tslen);
1064 a3500804 2018-01-23 stsp if (err)
1065 a3500804 2018-01-23 stsp return err;
1066 a3500804 2018-01-23 stsp
1067 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
1068 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
1069 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1070 a53d2f13 2018-03-17 stsp
1071 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1072 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1073 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
1074 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1075 d7464085 2018-07-09 stsp }
1076 bdd2fbb3 2018-02-11 stsp
1077 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1078 e485d53c 2022-10-28 thomas delta_data_offset);
1079 bdd2fbb3 2018-02-11 stsp if (err)
1080 1a35c1bc 2019-05-22 stsp return err;
1081 bdd2fbb3 2018-02-11 stsp
1082 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
1083 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1084 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1085 b107e67f 2018-01-19 stsp
1086 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1087 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1088 b107e67f 2018-01-19 stsp if (err)
1089 1a35c1bc 2019-05-22 stsp return err;
1090 b107e67f 2018-01-19 stsp
1091 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1092 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
1093 c3703302 2018-01-23 stsp }
1094 c4330eff 2021-06-22 stsp
1095 c4330eff 2021-06-22 stsp const struct got_error *
1096 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1097 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
1098 c4330eff 2021-06-22 stsp {
1099 c4330eff 2021-06-22 stsp if (pack->map) {
1100 66e6097f 2022-10-27 thomas size_t mapoff;
1101 66e6097f 2022-10-27 thomas
1102 66e6097f 2022-10-27 thomas if (delta_offset + tslen > SIZE_MAX) {
1103 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_PACK_OFFSET,
1104 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
1105 66e6097f 2022-10-27 thomas (long long)delta_offset + tslen);
1106 66e6097f 2022-10-27 thomas }
1107 66e6097f 2022-10-27 thomas
1108 66e6097f 2022-10-27 thomas mapoff = delta_offset + tslen;
1109 1944573a 2022-01-10 thomas if (mapoff + sizeof(*id) >= pack->filesize)
1110 1944573a 2022-01-10 thomas return got_error(GOT_ERR_PACK_OFFSET);
1111 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
1112 c4330eff 2021-06-22 stsp } else {
1113 c4330eff 2021-06-22 stsp ssize_t n;
1114 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1115 c4330eff 2021-06-22 stsp if (n < 0)
1116 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1117 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1118 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1119 c4330eff 2021-06-22 stsp }
1120 c3703302 2018-01-23 stsp
1121 c4330eff 2021-06-22 stsp return NULL;
1122 c4330eff 2021-06-22 stsp }
1123 c4330eff 2021-06-22 stsp
1124 c3703302 2018-01-23 stsp static const struct got_error *
1125 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1126 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1127 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1128 c3703302 2018-01-23 stsp {
1129 6b9c9673 2018-01-23 stsp const struct got_error *err;
1130 6b9c9673 2018-01-23 stsp struct got_object_id id;
1131 6b9c9673 2018-01-23 stsp int idx;
1132 6b9c9673 2018-01-23 stsp off_t base_offset;
1133 6b9c9673 2018-01-23 stsp uint8_t base_type;
1134 6b9c9673 2018-01-23 stsp uint64_t base_size;
1135 6b9c9673 2018-01-23 stsp size_t base_tslen;
1136 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1137 6b9c9673 2018-01-23 stsp
1138 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1139 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1140 a53d2f13 2018-03-17 stsp
1141 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1142 c4330eff 2021-06-22 stsp if (err)
1143 c4330eff 2021-06-22 stsp return err;
1144 d7464085 2018-07-09 stsp if (pack->map) {
1145 4907b168 2023-02-20 thomas delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1146 d7464085 2018-07-09 stsp } else {
1147 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1148 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1149 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1150 d7464085 2018-07-09 stsp }
1151 d7464085 2018-07-09 stsp
1152 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1153 e485d53c 2022-10-28 thomas delta_data_offset);
1154 bdd2fbb3 2018-02-11 stsp if (err)
1155 1a35c1bc 2019-05-22 stsp return err;
1156 bdd2fbb3 2018-02-11 stsp
1157 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1158 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1159 1a35c1bc 2019-05-22 stsp if (idx == -1)
1160 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1161 6b9c9673 2018-01-23 stsp
1162 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1163 e4a6dbc6 2022-07-24 thomas if (base_offset == -1)
1164 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1165 6b9c9673 2018-01-23 stsp
1166 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1167 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1168 6b9c9673 2018-01-23 stsp
1169 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1170 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1171 6b9c9673 2018-01-23 stsp if (err)
1172 1a35c1bc 2019-05-22 stsp return err;
1173 6b9c9673 2018-01-23 stsp
1174 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1175 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1176 6b9c9673 2018-01-23 stsp }
1177 6b9c9673 2018-01-23 stsp
1178 668a20f6 2020-03-18 stsp const struct got_error *
1179 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1180 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1181 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1182 6b9c9673 2018-01-23 stsp {
1183 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1184 c3703302 2018-01-23 stsp
1185 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1186 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1187 5b7e13a7 2018-04-02 stsp
1188 c3703302 2018-01-23 stsp switch (delta_type) {
1189 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1190 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1191 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1192 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1193 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1194 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1195 42c69117 2019-11-10 stsp delta_size, 0);
1196 4e8cda55 2018-01-19 stsp break;
1197 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1198 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1199 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1200 96f5e8b3 2018-01-23 stsp break;
1201 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1202 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1203 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1204 6b9c9673 2018-01-23 stsp break;
1205 4e8cda55 2018-01-19 stsp default:
1206 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1207 4e8cda55 2018-01-19 stsp }
1208 96f5e8b3 2018-01-23 stsp
1209 96f5e8b3 2018-01-23 stsp return err;
1210 96f5e8b3 2018-01-23 stsp }
1211 96f5e8b3 2018-01-23 stsp
1212 96f5e8b3 2018-01-23 stsp static const struct got_error *
1213 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1214 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1215 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1216 96f5e8b3 2018-01-23 stsp {
1217 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1218 96f5e8b3 2018-01-23 stsp int resolved_type;
1219 4e8cda55 2018-01-19 stsp
1220 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1221 b107e67f 2018-01-19 stsp if (*obj == NULL)
1222 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1223 b107e67f 2018-01-19 stsp
1224 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1225 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1226 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1227 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1228 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1229 1a35c1bc 2019-05-22 stsp
1230 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1231 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1232 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1233 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1234 96f5e8b3 2018-01-23 stsp
1235 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1236 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1237 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1238 96f5e8b3 2018-01-23 stsp if (err)
1239 96f5e8b3 2018-01-23 stsp goto done;
1240 96f5e8b3 2018-01-23 stsp
1241 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1242 96f5e8b3 2018-01-23 stsp if (err)
1243 96f5e8b3 2018-01-23 stsp goto done;
1244 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1245 96f5e8b3 2018-01-23 stsp done:
1246 96f5e8b3 2018-01-23 stsp if (err) {
1247 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1248 96f5e8b3 2018-01-23 stsp *obj = NULL;
1249 96f5e8b3 2018-01-23 stsp }
1250 96f5e8b3 2018-01-23 stsp return err;
1251 b107e67f 2018-01-19 stsp }
1252 b107e67f 2018-01-19 stsp
1253 2090a03d 2018-09-09 stsp const struct got_error *
1254 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1255 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1256 a1fd68d8 2018-01-12 stsp {
1257 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1258 a1fd68d8 2018-01-12 stsp off_t offset;
1259 6c00b545 2018-01-17 stsp uint8_t type;
1260 6c00b545 2018-01-17 stsp uint64_t size;
1261 3ee5fc21 2018-01-17 stsp size_t tslen;
1262 a1fd68d8 2018-01-12 stsp
1263 6c00b545 2018-01-17 stsp *obj = NULL;
1264 a1fd68d8 2018-01-12 stsp
1265 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1266 e4a6dbc6 2022-07-24 thomas if (offset == -1)
1267 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1268 a1fd68d8 2018-01-12 stsp
1269 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1270 668a20f6 2020-03-18 stsp pack, offset);
1271 a1fd68d8 2018-01-12 stsp if (err)
1272 35c73765 2018-09-09 stsp return err;
1273 a1fd68d8 2018-01-12 stsp
1274 6c00b545 2018-01-17 stsp switch (type) {
1275 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1276 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1277 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1278 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1279 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1280 5f25cc85 2019-11-26 stsp size, idx);
1281 6c00b545 2018-01-17 stsp break;
1282 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1283 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1284 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1285 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1286 b107e67f 2018-01-19 stsp break;
1287 6c00b545 2018-01-17 stsp default:
1288 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1289 35c73765 2018-09-09 stsp break;
1290 35c73765 2018-09-09 stsp }
1291 35c73765 2018-09-09 stsp
1292 3ee5fc21 2018-01-17 stsp return err;
1293 3ee5fc21 2018-01-17 stsp }
1294 efd2a263 2018-01-19 stsp
1295 d582f26c 2020-03-18 stsp const struct got_error *
1296 48b4f239 2021-12-31 thomas got_pack_get_delta_chain_max_size(uint64_t *max_size,
1297 48b4f239 2021-12-31 thomas struct got_delta_chain *deltas, struct got_pack *pack)
1298 22484865 2018-03-13 stsp {
1299 22484865 2018-03-13 stsp struct got_delta *delta;
1300 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1301 22484865 2018-03-13 stsp
1302 22484865 2018-03-13 stsp *max_size = 0;
1303 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1304 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1305 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1306 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1307 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1308 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1309 22484865 2018-03-13 stsp const struct got_error *err;
1310 723ed5ad 2022-10-18 thomas uint8_t *delta_buf = NULL;
1311 42c69117 2019-11-10 stsp size_t delta_len;
1312 ab2f42e7 2019-11-10 stsp int cached = 1;
1313 42c69117 2019-11-10 stsp
1314 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1315 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1316 f7a026c4 2023-04-28 thomas NULL, NULL, pack->delta_cache,
1317 f7a026c4 2023-04-28 thomas delta->data_offset);
1318 723ed5ad 2022-10-18 thomas }
1319 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1320 ab2f42e7 2019-11-10 stsp cached = 0;
1321 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1322 9249e7e3 2022-05-12 thomas NULL, delta->data_offset, pack);
1323 ab2f42e7 2019-11-10 stsp if (err)
1324 ab2f42e7 2019-11-10 stsp return err;
1325 723ed5ad 2022-10-18 thomas }
1326 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1327 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1328 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1329 ab2f42e7 2019-11-10 stsp if (err == NULL)
1330 ab2f42e7 2019-11-10 stsp cached = 1;
1331 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1332 ab2f42e7 2019-11-10 stsp free(delta_buf);
1333 ab2f42e7 2019-11-10 stsp return err;
1334 ab2f42e7 2019-11-10 stsp }
1335 ab2f42e7 2019-11-10 stsp }
1336 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1337 42c69117 2019-11-10 stsp delta_buf, delta_len);
1338 ab2f42e7 2019-11-10 stsp if (!cached)
1339 ab2f42e7 2019-11-10 stsp free(delta_buf);
1340 22484865 2018-03-13 stsp if (err)
1341 22484865 2018-03-13 stsp return err;
1342 22484865 2018-03-13 stsp } else
1343 22484865 2018-03-13 stsp base_size = delta->size;
1344 22484865 2018-03-13 stsp if (base_size > *max_size)
1345 22484865 2018-03-13 stsp *max_size = base_size;
1346 22484865 2018-03-13 stsp if (result_size > *max_size)
1347 22484865 2018-03-13 stsp *max_size = result_size;
1348 22484865 2018-03-13 stsp }
1349 bd1223b9 2018-03-14 stsp
1350 9feb4ff2 2018-03-14 stsp return NULL;
1351 ac544f8c 2019-01-13 stsp }
1352 ac544f8c 2019-01-13 stsp
1353 ac544f8c 2019-01-13 stsp const struct got_error *
1354 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1355 42c69117 2019-11-10 stsp struct got_pack *pack)
1356 ac544f8c 2019-01-13 stsp {
1357 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1358 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1359 85a703fa 2019-01-13 stsp
1360 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1361 22484865 2018-03-13 stsp }
1362 22484865 2018-03-13 stsp
1363 668a20f6 2020-03-18 stsp const struct got_error *
1364 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1365 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1366 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1367 efd2a263 2018-01-19 stsp {
1368 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1369 6691714a 2018-01-23 stsp struct got_delta *delta;
1370 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1371 f7a026c4 2023-04-28 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0;
1372 210941d5 2022-05-31 thomas /* We process small enough files entirely in memory for speed. */
1373 210941d5 2022-05-31 thomas const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1374 210941d5 2022-05-31 thomas uint64_t max_size = 0;
1375 6691714a 2018-01-23 stsp int n = 0;
1376 b29656e2 2018-03-16 stsp
1377 b29656e2 2018-03-16 stsp *result_size = 0;
1378 3ee5fc21 2018-01-17 stsp
1379 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1380 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1381 f7a026c4 2023-04-28 thomas
1382 f7a026c4 2023-04-28 thomas if (pack->delta_cache) {
1383 f7a026c4 2023-04-28 thomas uint8_t *delta_buf = NULL, *fulltext = NULL;
1384 f7a026c4 2023-04-28 thomas size_t delta_len, fulltext_len;
1385 f7a026c4 2023-04-28 thomas
1386 f7a026c4 2023-04-28 thomas delta = STAILQ_LAST(&deltas->entries, got_delta, entry);
1387 f7a026c4 2023-04-28 thomas got_delta_cache_get(&delta_buf, &delta_len,
1388 f7a026c4 2023-04-28 thomas &fulltext, &fulltext_len,
1389 f7a026c4 2023-04-28 thomas pack->delta_cache, delta->data_offset);
1390 f7a026c4 2023-04-28 thomas if (fulltext) {
1391 d0d910f2 2023-04-28 thomas size_t w;
1392 d0d910f2 2023-04-28 thomas
1393 d0d910f2 2023-04-28 thomas w = fwrite(fulltext, 1, fulltext_len, outfile);
1394 f7a026c4 2023-04-28 thomas if (w != fulltext_len)
1395 f7a026c4 2023-04-28 thomas return got_ferror(outfile, GOT_ERR_IO);
1396 5972d1ed 2023-04-28 thomas if (fflush(outfile) != 0)
1397 5972d1ed 2023-04-28 thomas return got_error_from_errno("fflush");
1398 f7a026c4 2023-04-28 thomas *result_size = fulltext_len;
1399 f7a026c4 2023-04-28 thomas return NULL;
1400 f7a026c4 2023-04-28 thomas }
1401 f7a026c4 2023-04-28 thomas }
1402 efd2a263 2018-01-19 stsp
1403 210941d5 2022-05-31 thomas if (fseeko(base_file, 0L, SEEK_SET) == -1)
1404 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1405 210941d5 2022-05-31 thomas if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1406 210941d5 2022-05-31 thomas return got_error_from_errno("fseeko");
1407 efd2a263 2018-01-19 stsp
1408 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1409 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1410 f7a026c4 2023-04-28 thomas uint8_t *delta_buf = NULL, *fulltext = NULL;
1411 f7a026c4 2023-04-28 thomas size_t delta_len, fulltext_len;
1412 210941d5 2022-05-31 thomas uint64_t base_size, result_size = 0;
1413 ab2f42e7 2019-11-10 stsp int cached = 1;
1414 a6b158cc 2018-02-11 stsp if (n == 0) {
1415 34fca9c3 2018-11-11 stsp size_t mapoff;
1416 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1417 9db65d41 2018-03-14 stsp
1418 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1419 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1420 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1421 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1422 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1423 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1424 72eb3431 2018-04-01 stsp goto done;
1425 72eb3431 2018-04-01 stsp }
1426 72eb3431 2018-04-01 stsp
1427 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1428 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1429 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1430 0c048b15 2018-04-27 stsp goto done;
1431 0c048b15 2018-04-27 stsp }
1432 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1433 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1434 d7464085 2018-07-09 stsp == -1) {
1435 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1436 d7464085 2018-07-09 stsp goto done;
1437 d7464085 2018-07-09 stsp }
1438 bdd2fbb3 2018-02-11 stsp }
1439 210941d5 2022-05-31 thomas if (delta->size > max_size)
1440 210941d5 2022-05-31 thomas max_size = delta->size;
1441 210941d5 2022-05-31 thomas if (max_size > max_bufsize) {
1442 d7464085 2018-07-09 stsp if (pack->map) {
1443 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1444 66e6097f 2022-10-27 thomas return got_error_fmt(
1445 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1446 66e6097f 2022-10-27 thomas "delta offset %lld "
1447 66e6097f 2022-10-27 thomas "overflows size_t",
1448 66e6097f 2022-10-27 thomas (long long)
1449 66e6097f 2022-10-27 thomas delta_data_offset);
1450 66e6097f 2022-10-27 thomas }
1451 66e6097f 2022-10-27 thomas
1452 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1453 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1454 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1455 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1456 4788f1ce 2020-03-18 stsp base_file);
1457 d7464085 2018-07-09 stsp } else
1458 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1459 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1460 4788f1ce 2020-03-18 stsp base_file);
1461 d7464085 2018-07-09 stsp } else {
1462 210941d5 2022-05-31 thomas accum_buf = malloc(max_size);
1463 210941d5 2022-05-31 thomas if (accum_buf == NULL) {
1464 210941d5 2022-05-31 thomas err = got_error_from_errno("malloc");
1465 210941d5 2022-05-31 thomas goto done;
1466 210941d5 2022-05-31 thomas }
1467 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1468 d7464085 2018-07-09 stsp if (pack->map) {
1469 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1470 66e6097f 2022-10-27 thomas return got_error_fmt(
1471 66e6097f 2022-10-27 thomas GOT_ERR_RANGE,
1472 66e6097f 2022-10-27 thomas "delta offset %lld "
1473 66e6097f 2022-10-27 thomas "overflows size_t",
1474 66e6097f 2022-10-27 thomas (long long)
1475 66e6097f 2022-10-27 thomas delta_data_offset);
1476 66e6097f 2022-10-27 thomas }
1477 66e6097f 2022-10-27 thomas
1478 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1479 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1480 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1481 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1482 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1483 d7464085 2018-07-09 stsp } else
1484 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1485 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1486 55fdd257 2020-03-18 stsp pack->fd);
1487 8628c62d 2018-03-15 stsp }
1488 a6b158cc 2018-02-11 stsp if (err)
1489 a6b158cc 2018-02-11 stsp goto done;
1490 a6b158cc 2018-02-11 stsp n++;
1491 210941d5 2022-05-31 thomas if (base_buf == NULL)
1492 8628c62d 2018-03-15 stsp rewind(base_file);
1493 ce833012 2023-04-28 thomas else if (pack->delta_cache && fulltext == NULL) {
1494 ce833012 2023-04-28 thomas err = got_delta_cache_add(pack->delta_cache,
1495 ce833012 2023-04-28 thomas delta_data_offset, NULL, 0);
1496 ce833012 2023-04-28 thomas if (err) {
1497 ce833012 2023-04-28 thomas if (err->code != GOT_ERR_NO_SPACE)
1498 ce833012 2023-04-28 thomas goto done;
1499 ce833012 2023-04-28 thomas err = NULL;
1500 ce833012 2023-04-28 thomas } else {
1501 ce833012 2023-04-28 thomas err = got_delta_cache_add_fulltext(
1502 ce833012 2023-04-28 thomas pack->delta_cache,
1503 ce833012 2023-04-28 thomas delta_data_offset,
1504 ce833012 2023-04-28 thomas base_buf, base_bufsz);
1505 ce833012 2023-04-28 thomas if (err &&
1506 ce833012 2023-04-28 thomas err->code != GOT_ERR_NO_SPACE)
1507 ce833012 2023-04-28 thomas goto done;
1508 ce833012 2023-04-28 thomas err = NULL;
1509 ce833012 2023-04-28 thomas }
1510 ce833012 2023-04-28 thomas }
1511 a6b158cc 2018-02-11 stsp continue;
1512 9db65d41 2018-03-14 stsp }
1513 885d3e02 2018-01-27 stsp
1514 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1515 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1516 f7a026c4 2023-04-28 thomas &fulltext, &fulltext_len,
1517 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1518 723ed5ad 2022-10-18 thomas }
1519 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1520 ab2f42e7 2019-11-10 stsp cached = 0;
1521 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1522 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1523 ab2f42e7 2019-11-10 stsp if (err)
1524 ab2f42e7 2019-11-10 stsp goto done;
1525 723ed5ad 2022-10-18 thomas }
1526 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1527 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1528 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1529 ab2f42e7 2019-11-10 stsp if (err == NULL)
1530 ab2f42e7 2019-11-10 stsp cached = 1;
1531 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1532 ab2f42e7 2019-11-10 stsp free(delta_buf);
1533 ab2f42e7 2019-11-10 stsp goto done;
1534 ab2f42e7 2019-11-10 stsp }
1535 ab2f42e7 2019-11-10 stsp }
1536 210941d5 2022-05-31 thomas
1537 210941d5 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1538 210941d5 2022-05-31 thomas delta_buf, delta_len);
1539 b1fad72f 2022-11-08 thomas if (err) {
1540 b1fad72f 2022-11-08 thomas if (!cached)
1541 b1fad72f 2022-11-08 thomas free(delta_buf);
1542 210941d5 2022-05-31 thomas goto done;
1543 b1fad72f 2022-11-08 thomas }
1544 210941d5 2022-05-31 thomas if (base_size > max_size)
1545 210941d5 2022-05-31 thomas max_size = base_size;
1546 210941d5 2022-05-31 thomas if (result_size > max_size)
1547 210941d5 2022-05-31 thomas max_size = result_size;
1548 f7a026c4 2023-04-28 thomas if (fulltext_len > max_size)
1549 f7a026c4 2023-04-28 thomas max_size = fulltext_len;
1550 210941d5 2022-05-31 thomas
1551 210941d5 2022-05-31 thomas if (base_buf && max_size > max_bufsize) {
1552 210941d5 2022-05-31 thomas /* Switch from buffers to temporary files. */
1553 210941d5 2022-05-31 thomas size_t w = fwrite(base_buf, 1, base_bufsz,
1554 210941d5 2022-05-31 thomas base_file);
1555 210941d5 2022-05-31 thomas if (w != base_bufsz) {
1556 210941d5 2022-05-31 thomas err = got_ferror(outfile, GOT_ERR_IO);
1557 b1fad72f 2022-11-08 thomas if (!cached)
1558 b1fad72f 2022-11-08 thomas free(delta_buf);
1559 210941d5 2022-05-31 thomas goto done;
1560 210941d5 2022-05-31 thomas }
1561 210941d5 2022-05-31 thomas free(base_buf);
1562 210941d5 2022-05-31 thomas base_buf = NULL;
1563 210941d5 2022-05-31 thomas free(accum_buf);
1564 210941d5 2022-05-31 thomas accum_buf = NULL;
1565 210941d5 2022-05-31 thomas }
1566 210941d5 2022-05-31 thomas
1567 210941d5 2022-05-31 thomas if (base_buf && max_size > base_bufsz) {
1568 210941d5 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1569 210941d5 2022-05-31 thomas if (p == NULL) {
1570 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1571 b1fad72f 2022-11-08 thomas if (!cached)
1572 b1fad72f 2022-11-08 thomas free(delta_buf);
1573 210941d5 2022-05-31 thomas goto done;
1574 210941d5 2022-05-31 thomas }
1575 210941d5 2022-05-31 thomas base_buf = p;
1576 210941d5 2022-05-31 thomas base_bufsz = max_size;
1577 210941d5 2022-05-31 thomas }
1578 210941d5 2022-05-31 thomas
1579 210941d5 2022-05-31 thomas if (accum_buf && max_size > accum_bufsz) {
1580 210941d5 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1581 210941d5 2022-05-31 thomas if (p == NULL) {
1582 210941d5 2022-05-31 thomas err = got_error_from_errno("realloc");
1583 b1fad72f 2022-11-08 thomas if (!cached)
1584 b1fad72f 2022-11-08 thomas free(delta_buf);
1585 210941d5 2022-05-31 thomas goto done;
1586 210941d5 2022-05-31 thomas }
1587 210941d5 2022-05-31 thomas accum_buf = p;
1588 210941d5 2022-05-31 thomas accum_bufsz = max_size;
1589 210941d5 2022-05-31 thomas }
1590 210941d5 2022-05-31 thomas
1591 8628c62d 2018-03-15 stsp if (base_buf) {
1592 f7a026c4 2023-04-28 thomas if (fulltext) {
1593 f7a026c4 2023-04-28 thomas memcpy(accum_buf, fulltext, fulltext_len);
1594 f7a026c4 2023-04-28 thomas accum_size = fulltext_len;
1595 f7a026c4 2023-04-28 thomas err = NULL;
1596 f7a026c4 2023-04-28 thomas } else {
1597 d0d910f2 2023-04-28 thomas err = got_delta_apply_in_mem(base_buf,
1598 d0d910f2 2023-04-28 thomas base_bufsz, delta_buf, delta_len,
1599 d0d910f2 2023-04-28 thomas accum_buf, &accum_size, max_size);
1600 f7a026c4 2023-04-28 thomas }
1601 8628c62d 2018-03-15 stsp n++;
1602 f7a026c4 2023-04-28 thomas if (!cached)
1603 f7a026c4 2023-04-28 thomas free(delta_buf);
1604 f7a026c4 2023-04-28 thomas if (err)
1605 f7a026c4 2023-04-28 thomas goto done;
1606 f7a026c4 2023-04-28 thomas if (fulltext == NULL) {
1607 f7a026c4 2023-04-28 thomas err = got_delta_cache_add_fulltext(
1608 f7a026c4 2023-04-28 thomas pack->delta_cache, delta->data_offset,
1609 f7a026c4 2023-04-28 thomas accum_buf, accum_size);
1610 f7a026c4 2023-04-28 thomas if (err) {
1611 f7a026c4 2023-04-28 thomas if (err->code != GOT_ERR_NO_SPACE)
1612 f7a026c4 2023-04-28 thomas goto done;
1613 f7a026c4 2023-04-28 thomas err = NULL;
1614 f7a026c4 2023-04-28 thomas }
1615 f7a026c4 2023-04-28 thomas }
1616 8628c62d 2018-03-15 stsp } else {
1617 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1618 42c69117 2019-11-10 stsp delta_len,
1619 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1620 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1621 b29656e2 2018-03-16 stsp &accum_size);
1622 f7a026c4 2023-04-28 thomas if (!cached)
1623 f7a026c4 2023-04-28 thomas free(delta_buf);
1624 f7a026c4 2023-04-28 thomas if (err)
1625 f7a026c4 2023-04-28 thomas goto done;
1626 8628c62d 2018-03-15 stsp }
1627 6691714a 2018-01-23 stsp
1628 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1629 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1630 8628c62d 2018-03-15 stsp if (base_buf) {
1631 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1632 210941d5 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1633 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1634 210941d5 2022-05-31 thomas accum_bufsz = base_bufsz;
1635 8628c62d 2018-03-15 stsp base_buf = tmp;
1636 210941d5 2022-05-31 thomas base_bufsz = tmp_size;
1637 8628c62d 2018-03-15 stsp } else {
1638 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1639 8628c62d 2018-03-15 stsp accum_file = base_file;
1640 8628c62d 2018-03-15 stsp base_file = tmp;
1641 8628c62d 2018-03-15 stsp rewind(base_file);
1642 8628c62d 2018-03-15 stsp rewind(accum_file);
1643 8628c62d 2018-03-15 stsp }
1644 6691714a 2018-01-23 stsp }
1645 6691714a 2018-01-23 stsp }
1646 6691714a 2018-01-23 stsp
1647 6691714a 2018-01-23 stsp done:
1648 8628c62d 2018-03-15 stsp free(base_buf);
1649 8628c62d 2018-03-15 stsp if (accum_buf) {
1650 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1651 8628c62d 2018-03-15 stsp free(accum_buf);
1652 8628c62d 2018-03-15 stsp if (len != accum_size)
1653 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1654 8628c62d 2018-03-15 stsp }
1655 6691714a 2018-01-23 stsp rewind(outfile);
1656 b29656e2 2018-03-16 stsp if (err == NULL)
1657 b29656e2 2018-03-16 stsp *result_size = accum_size;
1658 e0ab43e7 2018-03-16 stsp return err;
1659 e0ab43e7 2018-03-16 stsp }
1660 e0ab43e7 2018-03-16 stsp
1661 668a20f6 2020-03-18 stsp const struct got_error *
1662 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1663 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1664 e0ab43e7 2018-03-16 stsp {
1665 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1666 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1667 723ed5ad 2022-10-18 thomas uint8_t *base_buf = NULL, *accum_buf = NULL;
1668 f7a026c4 2023-04-28 thomas size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0;
1669 6c85883b 2022-05-31 thomas uint64_t max_size = 0;
1670 e0ab43e7 2018-03-16 stsp int n = 0;
1671 e0ab43e7 2018-03-16 stsp
1672 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1673 e0ab43e7 2018-03-16 stsp *outlen = 0;
1674 e0ab43e7 2018-03-16 stsp
1675 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1676 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1677 f7a026c4 2023-04-28 thomas
1678 f7a026c4 2023-04-28 thomas if (pack->delta_cache) {
1679 f7a026c4 2023-04-28 thomas uint8_t *delta_buf = NULL, *fulltext = NULL;
1680 f7a026c4 2023-04-28 thomas size_t delta_len, fulltext_len;
1681 f7a026c4 2023-04-28 thomas
1682 f7a026c4 2023-04-28 thomas delta = STAILQ_LAST(&deltas->entries, got_delta, entry);
1683 f7a026c4 2023-04-28 thomas got_delta_cache_get(&delta_buf, &delta_len,
1684 f7a026c4 2023-04-28 thomas &fulltext, &fulltext_len,
1685 f7a026c4 2023-04-28 thomas pack->delta_cache, delta->data_offset);
1686 f7a026c4 2023-04-28 thomas if (fulltext) {
1687 f7a026c4 2023-04-28 thomas *outbuf = malloc(fulltext_len);
1688 f7a026c4 2023-04-28 thomas if (*outbuf == NULL)
1689 f7a026c4 2023-04-28 thomas return got_error_from_errno("malloc");
1690 f7a026c4 2023-04-28 thomas memcpy(*outbuf, fulltext, fulltext_len);
1691 f7a026c4 2023-04-28 thomas *outlen = fulltext_len;
1692 f7a026c4 2023-04-28 thomas return NULL;
1693 f7a026c4 2023-04-28 thomas }
1694 f7a026c4 2023-04-28 thomas }
1695 e0ab43e7 2018-03-16 stsp
1696 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1697 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1698 f7a026c4 2023-04-28 thomas uint8_t *delta_buf = NULL, *fulltext = NULL;
1699 f7a026c4 2023-04-28 thomas size_t delta_len, fulltext_len = 0;
1700 6c85883b 2022-05-31 thomas uint64_t base_size, result_size = 0;
1701 ab2f42e7 2019-11-10 stsp int cached = 1;
1702 e0ab43e7 2018-03-16 stsp if (n == 0) {
1703 66e6097f 2022-10-27 thomas off_t delta_data_offset;
1704 e0ab43e7 2018-03-16 stsp
1705 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1706 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1707 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1708 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1709 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1710 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1711 72eb3431 2018-04-01 stsp goto done;
1712 72eb3431 2018-04-01 stsp }
1713 72eb3431 2018-04-01 stsp
1714 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1715 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1716 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1717 0c048b15 2018-04-27 stsp goto done;
1718 0c048b15 2018-04-27 stsp }
1719 6c85883b 2022-05-31 thomas
1720 f7a026c4 2023-04-28 thomas if (pack->delta_cache) {
1721 f7a026c4 2023-04-28 thomas got_delta_cache_get(&delta_buf, &delta_len,
1722 f7a026c4 2023-04-28 thomas &fulltext, &fulltext_len,
1723 f7a026c4 2023-04-28 thomas pack->delta_cache, delta_data_offset);
1724 f7a026c4 2023-04-28 thomas }
1725 f7a026c4 2023-04-28 thomas
1726 6c85883b 2022-05-31 thomas if (delta->size > max_size)
1727 6c85883b 2022-05-31 thomas max_size = delta->size;
1728 f7a026c4 2023-04-28 thomas if (delta->size > fulltext_len)
1729 f7a026c4 2023-04-28 thomas max_size = fulltext_len;
1730 6c85883b 2022-05-31 thomas
1731 f7a026c4 2023-04-28 thomas if (fulltext) {
1732 f7a026c4 2023-04-28 thomas base_buf = malloc(fulltext_len);
1733 f7a026c4 2023-04-28 thomas if (base_buf == NULL) {
1734 f7a026c4 2023-04-28 thomas err = got_error_from_errno("malloc");
1735 f7a026c4 2023-04-28 thomas goto done;
1736 f7a026c4 2023-04-28 thomas }
1737 f7a026c4 2023-04-28 thomas memcpy(base_buf, fulltext, fulltext_len);
1738 f7a026c4 2023-04-28 thomas base_bufsz = fulltext_len;
1739 f7a026c4 2023-04-28 thomas } else if (pack->map) {
1740 66e6097f 2022-10-27 thomas size_t mapoff;
1741 66e6097f 2022-10-27 thomas
1742 66e6097f 2022-10-27 thomas if (delta_data_offset > SIZE_MAX) {
1743 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1744 66e6097f 2022-10-27 thomas "delta %lld offset would "
1745 66e6097f 2022-10-27 thomas "overflow size_t",
1746 66e6097f 2022-10-27 thomas (long long)delta_data_offset);
1747 66e6097f 2022-10-27 thomas }
1748 66e6097f 2022-10-27 thomas
1749 66e6097f 2022-10-27 thomas mapoff = delta_data_offset;
1750 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1751 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1752 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1753 d7464085 2018-07-09 stsp } else {
1754 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1755 d7464085 2018-07-09 stsp == -1) {
1756 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1757 d7464085 2018-07-09 stsp goto done;
1758 d7464085 2018-07-09 stsp }
1759 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1760 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1761 55fdd257 2020-03-18 stsp pack->fd);
1762 e0ab43e7 2018-03-16 stsp }
1763 d7464085 2018-07-09 stsp if (err)
1764 d7464085 2018-07-09 stsp goto done;
1765 e0ab43e7 2018-03-16 stsp n++;
1766 f7a026c4 2023-04-28 thomas
1767 f7a026c4 2023-04-28 thomas if (pack->delta_cache && fulltext == NULL) {
1768 f7a026c4 2023-04-28 thomas err = got_delta_cache_add(pack->delta_cache,
1769 f7a026c4 2023-04-28 thomas delta_data_offset, NULL, 0);
1770 f7a026c4 2023-04-28 thomas if (err) {
1771 f7a026c4 2023-04-28 thomas if (err->code != GOT_ERR_NO_SPACE)
1772 f7a026c4 2023-04-28 thomas goto done;
1773 f7a026c4 2023-04-28 thomas err = NULL;
1774 f7a026c4 2023-04-28 thomas } else {
1775 f7a026c4 2023-04-28 thomas err = got_delta_cache_add_fulltext(
1776 f7a026c4 2023-04-28 thomas pack->delta_cache,
1777 f7a026c4 2023-04-28 thomas delta_data_offset,
1778 ac488a03 2023-04-28 thomas base_buf, base_bufsz);
1779 d0d910f2 2023-04-28 thomas if (err &&
1780 d0d910f2 2023-04-28 thomas err->code != GOT_ERR_NO_SPACE)
1781 f7a026c4 2023-04-28 thomas goto done;
1782 f7a026c4 2023-04-28 thomas err = NULL;
1783 f7a026c4 2023-04-28 thomas }
1784 f7a026c4 2023-04-28 thomas }
1785 e0ab43e7 2018-03-16 stsp continue;
1786 e0ab43e7 2018-03-16 stsp }
1787 e0ab43e7 2018-03-16 stsp
1788 723ed5ad 2022-10-18 thomas if (pack->delta_cache) {
1789 723ed5ad 2022-10-18 thomas got_delta_cache_get(&delta_buf, &delta_len,
1790 f7a026c4 2023-04-28 thomas &fulltext, &fulltext_len,
1791 723ed5ad 2022-10-18 thomas pack->delta_cache, delta->data_offset);
1792 723ed5ad 2022-10-18 thomas }
1793 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1794 ab2f42e7 2019-11-10 stsp cached = 0;
1795 9249e7e3 2022-05-12 thomas err = read_delta_data(&delta_buf, &delta_len, NULL,
1796 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1797 ab2f42e7 2019-11-10 stsp if (err)
1798 ab2f42e7 2019-11-10 stsp goto done;
1799 723ed5ad 2022-10-18 thomas }
1800 723ed5ad 2022-10-18 thomas if (pack->delta_cache && !cached) {
1801 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1802 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1803 ab2f42e7 2019-11-10 stsp if (err == NULL)
1804 ab2f42e7 2019-11-10 stsp cached = 1;
1805 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1806 ab2f42e7 2019-11-10 stsp free(delta_buf);
1807 ab2f42e7 2019-11-10 stsp goto done;
1808 ab2f42e7 2019-11-10 stsp }
1809 ab2f42e7 2019-11-10 stsp }
1810 6c85883b 2022-05-31 thomas
1811 6c85883b 2022-05-31 thomas err = got_delta_get_sizes(&base_size, &result_size,
1812 6c85883b 2022-05-31 thomas delta_buf, delta_len);
1813 b1fad72f 2022-11-08 thomas if (err) {
1814 b1fad72f 2022-11-08 thomas if (!cached)
1815 b1fad72f 2022-11-08 thomas free(delta_buf);
1816 6c85883b 2022-05-31 thomas goto done;
1817 b1fad72f 2022-11-08 thomas }
1818 6c85883b 2022-05-31 thomas if (base_size > max_size)
1819 6c85883b 2022-05-31 thomas max_size = base_size;
1820 6c85883b 2022-05-31 thomas if (result_size > max_size)
1821 6c85883b 2022-05-31 thomas max_size = result_size;
1822 f7a026c4 2023-04-28 thomas if (fulltext_len > max_size)
1823 f7a026c4 2023-04-28 thomas max_size = fulltext_len;
1824 6c85883b 2022-05-31 thomas
1825 6c85883b 2022-05-31 thomas if (max_size > base_bufsz) {
1826 6c85883b 2022-05-31 thomas uint8_t *p = realloc(base_buf, max_size);
1827 6c85883b 2022-05-31 thomas if (p == NULL) {
1828 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1829 b1fad72f 2022-11-08 thomas if (!cached)
1830 b1fad72f 2022-11-08 thomas free(delta_buf);
1831 6c85883b 2022-05-31 thomas goto done;
1832 6c85883b 2022-05-31 thomas }
1833 6c85883b 2022-05-31 thomas base_buf = p;
1834 6c85883b 2022-05-31 thomas base_bufsz = max_size;
1835 6c85883b 2022-05-31 thomas }
1836 6c85883b 2022-05-31 thomas
1837 6c85883b 2022-05-31 thomas if (max_size > accum_bufsz) {
1838 6c85883b 2022-05-31 thomas uint8_t *p = realloc(accum_buf, max_size);
1839 6c85883b 2022-05-31 thomas if (p == NULL) {
1840 6c85883b 2022-05-31 thomas err = got_error_from_errno("realloc");
1841 b1fad72f 2022-11-08 thomas if (!cached)
1842 b1fad72f 2022-11-08 thomas free(delta_buf);
1843 6c85883b 2022-05-31 thomas goto done;
1844 6c85883b 2022-05-31 thomas }
1845 6c85883b 2022-05-31 thomas accum_buf = p;
1846 6c85883b 2022-05-31 thomas accum_bufsz = max_size;
1847 6c85883b 2022-05-31 thomas }
1848 6c85883b 2022-05-31 thomas
1849 f7a026c4 2023-04-28 thomas if (fulltext) {
1850 f7a026c4 2023-04-28 thomas memcpy(accum_buf, fulltext, fulltext_len);
1851 f7a026c4 2023-04-28 thomas accum_size = fulltext_len;
1852 f7a026c4 2023-04-28 thomas err = NULL;
1853 f7a026c4 2023-04-28 thomas } else {
1854 f7a026c4 2023-04-28 thomas err = got_delta_apply_in_mem(base_buf, base_bufsz,
1855 f7a026c4 2023-04-28 thomas delta_buf, delta_len, accum_buf,
1856 f7a026c4 2023-04-28 thomas &accum_size, max_size);
1857 f7a026c4 2023-04-28 thomas }
1858 ab2f42e7 2019-11-10 stsp if (!cached)
1859 ab2f42e7 2019-11-10 stsp free(delta_buf);
1860 e0ab43e7 2018-03-16 stsp n++;
1861 e0ab43e7 2018-03-16 stsp if (err)
1862 e0ab43e7 2018-03-16 stsp goto done;
1863 e0ab43e7 2018-03-16 stsp
1864 f7a026c4 2023-04-28 thomas if (fulltext == NULL) {
1865 f7a026c4 2023-04-28 thomas err = got_delta_cache_add_fulltext(pack->delta_cache,
1866 f7a026c4 2023-04-28 thomas delta->data_offset, accum_buf, accum_size);
1867 f7a026c4 2023-04-28 thomas if (err) {
1868 f7a026c4 2023-04-28 thomas if (err->code != GOT_ERR_NO_SPACE)
1869 f7a026c4 2023-04-28 thomas goto done;
1870 f7a026c4 2023-04-28 thomas err = NULL;
1871 f7a026c4 2023-04-28 thomas }
1872 f7a026c4 2023-04-28 thomas }
1873 f7a026c4 2023-04-28 thomas
1874 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1875 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1876 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1877 6c85883b 2022-05-31 thomas size_t tmp_size = accum_bufsz;
1878 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1879 6c85883b 2022-05-31 thomas accum_bufsz = base_bufsz;
1880 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1881 6c85883b 2022-05-31 thomas base_bufsz = tmp_size;
1882 e0ab43e7 2018-03-16 stsp }
1883 e0ab43e7 2018-03-16 stsp }
1884 e0ab43e7 2018-03-16 stsp
1885 e0ab43e7 2018-03-16 stsp done:
1886 e0ab43e7 2018-03-16 stsp free(base_buf);
1887 e0ab43e7 2018-03-16 stsp if (err) {
1888 e0ab43e7 2018-03-16 stsp free(accum_buf);
1889 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1890 e0ab43e7 2018-03-16 stsp *outlen = 0;
1891 e0ab43e7 2018-03-16 stsp } else {
1892 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1893 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1894 e0ab43e7 2018-03-16 stsp }
1895 efd2a263 2018-01-19 stsp return err;
1896 efd2a263 2018-01-19 stsp }
1897 efd2a263 2018-01-19 stsp
1898 3ee5fc21 2018-01-17 stsp const struct got_error *
1899 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1900 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1901 3ee5fc21 2018-01-17 stsp {
1902 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1903 3ee5fc21 2018-01-17 stsp
1904 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1905 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1906 2ce68b2f 2018-09-09 stsp
1907 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1908 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1909 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1910 72eb3431 2018-04-01 stsp
1911 d7464085 2018-07-09 stsp if (pack->map) {
1912 66e6097f 2022-10-27 thomas size_t mapoff;
1913 66e6097f 2022-10-27 thomas
1914 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1915 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1916 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1917 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1918 66e6097f 2022-10-27 thomas }
1919 66e6097f 2022-10-27 thomas
1920 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1921 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1922 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1923 4788f1ce 2020-03-18 stsp outfile);
1924 d7464085 2018-07-09 stsp } else {
1925 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1926 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1927 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1928 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1929 d7464085 2018-07-09 stsp }
1930 040bf4a1 2018-04-01 stsp } else
1931 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1932 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1933 24140570 2018-09-09 stsp
1934 a1fd68d8 2018-01-12 stsp return err;
1935 a1fd68d8 2018-01-12 stsp }
1936 e0ab43e7 2018-03-16 stsp
1937 e0ab43e7 2018-03-16 stsp const struct got_error *
1938 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1939 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1940 e0ab43e7 2018-03-16 stsp {
1941 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1942 e0ab43e7 2018-03-16 stsp
1943 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1944 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1945 72eb3431 2018-04-01 stsp
1946 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1947 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1948 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1949 d7464085 2018-07-09 stsp if (pack->map) {
1950 66e6097f 2022-10-27 thomas size_t mapoff;
1951 66e6097f 2022-10-27 thomas
1952 66e6097f 2022-10-27 thomas if (obj->pack_offset > SIZE_MAX) {
1953 66e6097f 2022-10-27 thomas return got_error_fmt(GOT_ERR_RANGE,
1954 66e6097f 2022-10-27 thomas "pack offset %lld would overflow size_t",
1955 66e6097f 2022-10-27 thomas (long long)obj->pack_offset);
1956 66e6097f 2022-10-27 thomas }
1957 66e6097f 2022-10-27 thomas
1958 66e6097f 2022-10-27 thomas mapoff = obj->pack_offset;
1959 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1960 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1961 d7464085 2018-07-09 stsp } else {
1962 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1963 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1964 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1965 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1966 e0ab43e7 2018-03-16 stsp }
1967 e0ab43e7 2018-03-16 stsp } else
1968 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1969 668a20f6 2020-03-18 stsp pack);
1970 7e212e3d 2018-09-09 stsp
1971 e0ab43e7 2018-03-16 stsp return err;
1972 e0ab43e7 2018-03-16 stsp }
1973 f9c2e8e5 2022-02-13 thomas
1974 9249e7e3 2022-05-12 thomas static const struct got_error *
1975 9249e7e3 2022-05-12 thomas read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1976 9249e7e3 2022-05-12 thomas size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1977 9249e7e3 2022-05-12 thomas off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1978 9249e7e3 2022-05-12 thomas {
1979 9249e7e3 2022-05-12 thomas const struct got_error *err = NULL;
1980 9249e7e3 2022-05-12 thomas
1981 9249e7e3 2022-05-12 thomas /* Validate decompression and obtain the decompressed size. */
1982 9249e7e3 2022-05-12 thomas err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1983 9249e7e3 2022-05-12 thomas delta_data_offset, pack);
1984 9249e7e3 2022-05-12 thomas if (err)
1985 9249e7e3 2022-05-12 thomas return err;
1986 9249e7e3 2022-05-12 thomas
1987 9249e7e3 2022-05-12 thomas /* Read delta base/result sizes from head of delta stream. */
1988 9249e7e3 2022-05-12 thomas err = got_delta_get_sizes(base_size, result_size,
1989 9249e7e3 2022-05-12 thomas *delta_buf, *delta_len);
1990 9249e7e3 2022-05-12 thomas if (err)
1991 9249e7e3 2022-05-12 thomas goto done;
1992 9249e7e3 2022-05-12 thomas
1993 9249e7e3 2022-05-12 thomas /* Discard decompressed delta and read it again in compressed form. */
1994 9249e7e3 2022-05-12 thomas free(*delta_buf);
1995 9249e7e3 2022-05-12 thomas *delta_buf = malloc(*delta_len_compressed);
1996 9249e7e3 2022-05-12 thomas if (*delta_buf == NULL) {
1997 9249e7e3 2022-05-12 thomas err = got_error_from_errno("malloc");
1998 9249e7e3 2022-05-12 thomas goto done;
1999 9249e7e3 2022-05-12 thomas }
2000 9249e7e3 2022-05-12 thomas if (pack->map) {
2001 30d415a5 2022-10-25 thomas if (delta_data_offset >= pack->filesize) {
2002 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_PACK_OFFSET);
2003 30d415a5 2022-10-25 thomas goto done;
2004 30d415a5 2022-10-25 thomas }
2005 9249e7e3 2022-05-12 thomas memcpy(*delta_buf, pack->map + delta_data_offset,
2006 9249e7e3 2022-05-12 thomas *delta_len_compressed);
2007 9249e7e3 2022-05-12 thomas } else {
2008 9249e7e3 2022-05-12 thomas ssize_t n;
2009 9249e7e3 2022-05-12 thomas if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
2010 9249e7e3 2022-05-12 thomas err = got_error_from_errno("lseek");
2011 9249e7e3 2022-05-12 thomas goto done;
2012 9249e7e3 2022-05-12 thomas }
2013 9249e7e3 2022-05-12 thomas n = read(pack->fd, *delta_buf, *delta_len_compressed);
2014 9249e7e3 2022-05-12 thomas if (n < 0) {
2015 9249e7e3 2022-05-12 thomas err = got_error_from_errno("read");
2016 9249e7e3 2022-05-12 thomas goto done;
2017 9249e7e3 2022-05-12 thomas } else if (n != *delta_len_compressed) {
2018 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_IO);
2019 9249e7e3 2022-05-12 thomas goto done;
2020 9249e7e3 2022-05-12 thomas }
2021 9249e7e3 2022-05-12 thomas }
2022 9249e7e3 2022-05-12 thomas done:
2023 9249e7e3 2022-05-12 thomas if (err) {
2024 9249e7e3 2022-05-12 thomas free(*delta_buf);
2025 9249e7e3 2022-05-12 thomas *delta_buf = NULL;
2026 9249e7e3 2022-05-12 thomas *delta_len = 0;
2027 9249e7e3 2022-05-12 thomas *delta_len_compressed = 0;
2028 9249e7e3 2022-05-12 thomas *base_size = 0;
2029 9249e7e3 2022-05-12 thomas *result_size = 0;
2030 9249e7e3 2022-05-12 thomas }
2031 9249e7e3 2022-05-12 thomas return err;
2032 9249e7e3 2022-05-12 thomas }
2033 9249e7e3 2022-05-12 thomas
2034 f9c2e8e5 2022-02-13 thomas const struct got_error *
2035 f9c2e8e5 2022-02-13 thomas got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
2036 c44c7d6e 2022-12-04 thomas size_t *delta_compressed_size, off_t *delta_offset,
2037 c44c7d6e 2022-12-04 thomas off_t *delta_data_offset, off_t *base_offset,
2038 9249e7e3 2022-05-12 thomas struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
2039 9249e7e3 2022-05-12 thomas struct got_pack *pack, struct got_packidx *packidx, int idx)
2040 f9c2e8e5 2022-02-13 thomas {
2041 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
2042 f9c2e8e5 2022-02-13 thomas off_t offset;
2043 f9c2e8e5 2022-02-13 thomas uint8_t type;
2044 f9c2e8e5 2022-02-13 thomas uint64_t size;
2045 f9c2e8e5 2022-02-13 thomas size_t tslen, delta_hdrlen;
2046 f9c2e8e5 2022-02-13 thomas
2047 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
2048 f9c2e8e5 2022-02-13 thomas *delta_size = 0;
2049 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
2050 f9c2e8e5 2022-02-13 thomas *delta_offset = 0;
2051 c44c7d6e 2022-12-04 thomas *delta_data_offset = 0;
2052 f9c2e8e5 2022-02-13 thomas *base_offset = 0;
2053 f9c2e8e5 2022-02-13 thomas *base_size = 0;
2054 f9c2e8e5 2022-02-13 thomas *result_size = 0;
2055 f9c2e8e5 2022-02-13 thomas
2056 f9c2e8e5 2022-02-13 thomas offset = got_packidx_get_object_offset(packidx, idx);
2057 e4a6dbc6 2022-07-24 thomas if (offset == -1)
2058 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKIDX);
2059 f9c2e8e5 2022-02-13 thomas
2060 f9c2e8e5 2022-02-13 thomas if (offset >= pack->filesize)
2061 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
2062 f9c2e8e5 2022-02-13 thomas
2063 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
2064 f9c2e8e5 2022-02-13 thomas pack, offset);
2065 f9c2e8e5 2022-02-13 thomas if (err)
2066 f9c2e8e5 2022-02-13 thomas return err;
2067 f9c2e8e5 2022-02-13 thomas
2068 f9c2e8e5 2022-02-13 thomas if (tslen + size < tslen || offset + size < size ||
2069 f9c2e8e5 2022-02-13 thomas tslen + offset < tslen)
2070 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_PACK_OFFSET);
2071 f9c2e8e5 2022-02-13 thomas
2072 f9c2e8e5 2022-02-13 thomas switch (type) {
2073 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_OFFSET_DELTA:
2074 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
2075 f9c2e8e5 2022-02-13 thomas pack, offset, tslen);
2076 f9c2e8e5 2022-02-13 thomas if (err)
2077 f9c2e8e5 2022-02-13 thomas return err;
2078 f9c2e8e5 2022-02-13 thomas break;
2079 f9c2e8e5 2022-02-13 thomas case GOT_OBJ_TYPE_REF_DELTA:
2080 f9c2e8e5 2022-02-13 thomas err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
2081 f9c2e8e5 2022-02-13 thomas if (err)
2082 f9c2e8e5 2022-02-13 thomas return err;
2083 f9c2e8e5 2022-02-13 thomas delta_hdrlen = SHA1_DIGEST_LENGTH;
2084 f9c2e8e5 2022-02-13 thomas break;
2085 f9c2e8e5 2022-02-13 thomas default:
2086 f9c2e8e5 2022-02-13 thomas return got_error_fmt(GOT_ERR_OBJ_TYPE,
2087 e4a6dbc6 2022-07-24 thomas "non-delta object type %d found at offset %lld",
2088 e4a6dbc6 2022-07-24 thomas type, (long long)offset);
2089 f9c2e8e5 2022-02-13 thomas }
2090 f9c2e8e5 2022-02-13 thomas
2091 f9c2e8e5 2022-02-13 thomas if (tslen + delta_hdrlen < delta_hdrlen ||
2092 f9c2e8e5 2022-02-13 thomas offset + delta_hdrlen < delta_hdrlen)
2093 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_DELTA);
2094 f9c2e8e5 2022-02-13 thomas
2095 c44c7d6e 2022-12-04 thomas *delta_data_offset = offset + tslen + delta_hdrlen;
2096 9249e7e3 2022-05-12 thomas err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
2097 c44c7d6e 2022-12-04 thomas base_size, result_size, *delta_data_offset, pack, packidx);
2098 f9c2e8e5 2022-02-13 thomas if (err)
2099 f9c2e8e5 2022-02-13 thomas return err;
2100 f9c2e8e5 2022-02-13 thomas
2101 f9c2e8e5 2022-02-13 thomas if (*delta_size != size) {
2102 f9c2e8e5 2022-02-13 thomas err = got_error(GOT_ERR_BAD_DELTA);
2103 f9c2e8e5 2022-02-13 thomas goto done;
2104 f9c2e8e5 2022-02-13 thomas }
2105 f9c2e8e5 2022-02-13 thomas
2106 f9c2e8e5 2022-02-13 thomas *delta_offset = offset;
2107 f9c2e8e5 2022-02-13 thomas done:
2108 f9c2e8e5 2022-02-13 thomas if (err) {
2109 f9c2e8e5 2022-02-13 thomas free(*delta_buf);
2110 f9c2e8e5 2022-02-13 thomas *delta_buf = NULL;
2111 9249e7e3 2022-05-12 thomas *delta_size = 0;
2112 9249e7e3 2022-05-12 thomas *delta_compressed_size = 0;
2113 9249e7e3 2022-05-12 thomas *delta_offset = 0;
2114 9249e7e3 2022-05-12 thomas *base_offset = 0;
2115 9249e7e3 2022-05-12 thomas *base_size = 0;
2116 9249e7e3 2022-05-12 thomas *result_size = 0;
2117 f9c2e8e5 2022-02-13 thomas }
2118 f9c2e8e5 2022-02-13 thomas return err;
2119 f9c2e8e5 2022-02-13 thomas }