Blame


1 142012ed 2022-10-20 thomas /*
2 142012ed 2022-10-20 thomas * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3 142012ed 2022-10-20 thomas * Copyright (c) 2020, 2022 Stefan Sperling <stsp@openbsd.org>
4 142012ed 2022-10-20 thomas *
5 142012ed 2022-10-20 thomas * Permission to use, copy, modify, and distribute this software for any
6 142012ed 2022-10-20 thomas * purpose with or without fee is hereby granted, provided that the above
7 142012ed 2022-10-20 thomas * copyright notice and this permission notice appear in all copies.
8 142012ed 2022-10-20 thomas *
9 142012ed 2022-10-20 thomas * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 142012ed 2022-10-20 thomas * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 142012ed 2022-10-20 thomas * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 142012ed 2022-10-20 thomas * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 142012ed 2022-10-20 thomas * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 142012ed 2022-10-20 thomas * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 142012ed 2022-10-20 thomas * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 142012ed 2022-10-20 thomas */
17 142012ed 2022-10-20 thomas
18 142012ed 2022-10-20 thomas #include <sys/queue.h>
19 142012ed 2022-10-20 thomas #include <sys/stat.h>
20 142012ed 2022-10-20 thomas #include <sys/time.h>
21 142012ed 2022-10-20 thomas #include <sys/types.h>
22 142012ed 2022-10-20 thomas #include <sys/uio.h>
23 142012ed 2022-10-20 thomas #include <sys/mman.h>
24 142012ed 2022-10-20 thomas
25 142012ed 2022-10-20 thomas #include <stdint.h>
26 142012ed 2022-10-20 thomas #include <errno.h>
27 142012ed 2022-10-20 thomas #include <imsg.h>
28 142012ed 2022-10-20 thomas #include <limits.h>
29 142012ed 2022-10-20 thomas #include <signal.h>
30 142012ed 2022-10-20 thomas #include <stdio.h>
31 142012ed 2022-10-20 thomas #include <stdlib.h>
32 142012ed 2022-10-20 thomas #include <string.h>
33 142012ed 2022-10-20 thomas #include <ctype.h>
34 588a8092 2023-02-23 thomas #include <sha1.h>
35 588a8092 2023-02-23 thomas #include <sha2.h>
36 588a8092 2023-02-23 thomas #include <endian.h>
37 142012ed 2022-10-20 thomas #include <fcntl.h>
38 142012ed 2022-10-20 thomas #include <unistd.h>
39 142012ed 2022-10-20 thomas #include <zlib.h>
40 142012ed 2022-10-20 thomas #include <err.h>
41 142012ed 2022-10-20 thomas #include <assert.h>
42 142012ed 2022-10-20 thomas #include <dirent.h>
43 142012ed 2022-10-20 thomas
44 142012ed 2022-10-20 thomas #include "got_error.h"
45 142012ed 2022-10-20 thomas #include "got_object.h"
46 142012ed 2022-10-20 thomas
47 142012ed 2022-10-20 thomas #include "got_lib_sha1.h"
48 142012ed 2022-10-20 thomas #include "got_lib_delta.h"
49 142012ed 2022-10-20 thomas #include "got_lib_inflate.h"
50 142012ed 2022-10-20 thomas #include "got_lib_object.h"
51 142012ed 2022-10-20 thomas #include "got_lib_object_parse.h"
52 142012ed 2022-10-20 thomas #include "got_lib_object_idset.h"
53 142012ed 2022-10-20 thomas #include "got_lib_privsep.h"
54 142012ed 2022-10-20 thomas #include "got_lib_pack.h"
55 142012ed 2022-10-20 thomas #include "got_lib_ratelimit.h"
56 142012ed 2022-10-20 thomas #include "got_lib_pack_index.h"
57 142012ed 2022-10-20 thomas #include "got_lib_delta_cache.h"
58 142012ed 2022-10-20 thomas
59 142012ed 2022-10-20 thomas struct got_indexed_object {
60 142012ed 2022-10-20 thomas struct got_object_id id;
61 142012ed 2022-10-20 thomas
62 142012ed 2022-10-20 thomas /*
63 142012ed 2022-10-20 thomas * Has this object been fully resolved?
64 142012ed 2022-10-20 thomas * If so, we know its ID, otherwise we don't and 'id' is invalid.
65 142012ed 2022-10-20 thomas */
66 142012ed 2022-10-20 thomas int valid;
67 142012ed 2022-10-20 thomas
68 142012ed 2022-10-20 thomas /* Offset of type+size field for this object in pack file. */
69 142012ed 2022-10-20 thomas off_t off;
70 142012ed 2022-10-20 thomas
71 142012ed 2022-10-20 thomas /* Type+size values parsed from pack file. */
72 142012ed 2022-10-20 thomas uint8_t type;
73 142012ed 2022-10-20 thomas uint64_t size;
74 142012ed 2022-10-20 thomas
75 142012ed 2022-10-20 thomas /* Length of on-disk type+size data. */
76 142012ed 2022-10-20 thomas size_t tslen;
77 142012ed 2022-10-20 thomas
78 142012ed 2022-10-20 thomas /* Length of object data following type+size. */
79 142012ed 2022-10-20 thomas size_t len;
80 142012ed 2022-10-20 thomas
81 142012ed 2022-10-20 thomas uint32_t crc;
82 142012ed 2022-10-20 thomas
83 142012ed 2022-10-20 thomas union {
84 142012ed 2022-10-20 thomas struct {
85 142012ed 2022-10-20 thomas /* For ref deltas. */
86 142012ed 2022-10-20 thomas struct got_object_id ref_id;
87 142012ed 2022-10-20 thomas } ref;
88 142012ed 2022-10-20 thomas struct {
89 142012ed 2022-10-20 thomas /* For offset deltas. */
90 142012ed 2022-10-20 thomas off_t base_offset;
91 142012ed 2022-10-20 thomas size_t base_offsetlen;
92 142012ed 2022-10-20 thomas } ofs;
93 142012ed 2022-10-20 thomas } delta;
94 142012ed 2022-10-20 thomas };
95 142012ed 2022-10-20 thomas
96 142012ed 2022-10-20 thomas static void
97 142012ed 2022-10-20 thomas putbe32(char *b, uint32_t n)
98 142012ed 2022-10-20 thomas {
99 142012ed 2022-10-20 thomas b[0] = n >> 24;
100 142012ed 2022-10-20 thomas b[1] = n >> 16;
101 142012ed 2022-10-20 thomas b[2] = n >> 8;
102 142012ed 2022-10-20 thomas b[3] = n >> 0;
103 142012ed 2022-10-20 thomas }
104 142012ed 2022-10-20 thomas
105 142012ed 2022-10-20 thomas static const struct got_error *
106 142012ed 2022-10-20 thomas get_obj_type_label(const char **label, int obj_type)
107 142012ed 2022-10-20 thomas {
108 142012ed 2022-10-20 thomas const struct got_error *err = NULL;
109 142012ed 2022-10-20 thomas
110 142012ed 2022-10-20 thomas switch (obj_type) {
111 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_BLOB:
112 142012ed 2022-10-20 thomas *label = GOT_OBJ_LABEL_BLOB;
113 142012ed 2022-10-20 thomas break;
114 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_TREE:
115 142012ed 2022-10-20 thomas *label = GOT_OBJ_LABEL_TREE;
116 142012ed 2022-10-20 thomas break;
117 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_COMMIT:
118 142012ed 2022-10-20 thomas *label = GOT_OBJ_LABEL_COMMIT;
119 142012ed 2022-10-20 thomas break;
120 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_TAG:
121 142012ed 2022-10-20 thomas *label = GOT_OBJ_LABEL_TAG;
122 142012ed 2022-10-20 thomas break;
123 142012ed 2022-10-20 thomas default:
124 142012ed 2022-10-20 thomas *label = NULL;
125 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_OBJ_TYPE);
126 142012ed 2022-10-20 thomas break;
127 142012ed 2022-10-20 thomas }
128 142012ed 2022-10-20 thomas
129 142012ed 2022-10-20 thomas return err;
130 142012ed 2022-10-20 thomas }
131 142012ed 2022-10-20 thomas
132 142012ed 2022-10-20 thomas static const struct got_error *
133 142012ed 2022-10-20 thomas read_checksum(uint32_t *crc, SHA1_CTX *sha1_ctx, int fd, size_t len)
134 142012ed 2022-10-20 thomas {
135 142012ed 2022-10-20 thomas uint8_t buf[8192];
136 142012ed 2022-10-20 thomas size_t n;
137 142012ed 2022-10-20 thomas ssize_t r;
138 142012ed 2022-10-20 thomas
139 142012ed 2022-10-20 thomas for (n = len; n > 0; n -= r){
140 142012ed 2022-10-20 thomas r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
141 142012ed 2022-10-20 thomas if (r == -1)
142 142012ed 2022-10-20 thomas return got_error_from_errno("read");
143 142012ed 2022-10-20 thomas if (r == 0)
144 142012ed 2022-10-20 thomas break;
145 142012ed 2022-10-20 thomas if (crc)
146 142012ed 2022-10-20 thomas *crc = crc32(*crc, buf, r);
147 142012ed 2022-10-20 thomas if (sha1_ctx)
148 142012ed 2022-10-20 thomas SHA1Update(sha1_ctx, buf, r);
149 142012ed 2022-10-20 thomas }
150 142012ed 2022-10-20 thomas
151 142012ed 2022-10-20 thomas return NULL;
152 142012ed 2022-10-20 thomas }
153 142012ed 2022-10-20 thomas
154 142012ed 2022-10-20 thomas static const struct got_error *
155 142012ed 2022-10-20 thomas read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len)
156 142012ed 2022-10-20 thomas {
157 142012ed 2022-10-20 thomas uint8_t buf[8192];
158 142012ed 2022-10-20 thomas size_t n, r;
159 142012ed 2022-10-20 thomas
160 142012ed 2022-10-20 thomas for (n = len; n > 0; n -= r) {
161 142012ed 2022-10-20 thomas r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
162 142012ed 2022-10-20 thomas if (r == 0) {
163 142012ed 2022-10-20 thomas if (feof(f))
164 142012ed 2022-10-20 thomas return NULL;
165 142012ed 2022-10-20 thomas return got_ferror(f, GOT_ERR_IO);
166 142012ed 2022-10-20 thomas }
167 142012ed 2022-10-20 thomas SHA1Update(ctx, buf, r);
168 142012ed 2022-10-20 thomas }
169 142012ed 2022-10-20 thomas
170 142012ed 2022-10-20 thomas return NULL;
171 142012ed 2022-10-20 thomas }
172 142012ed 2022-10-20 thomas
173 142012ed 2022-10-20 thomas static const struct got_error *
174 142012ed 2022-10-20 thomas read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
175 142012ed 2022-10-20 thomas FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
176 142012ed 2022-10-20 thomas {
177 142012ed 2022-10-20 thomas const struct got_error *err = NULL;
178 142012ed 2022-10-20 thomas SHA1_CTX ctx;
179 142012ed 2022-10-20 thomas uint8_t *data = NULL;
180 142012ed 2022-10-20 thomas size_t datalen = 0;
181 142012ed 2022-10-20 thomas ssize_t n;
182 142012ed 2022-10-20 thomas char *header;
183 142012ed 2022-10-20 thomas size_t headerlen;
184 142012ed 2022-10-20 thomas const char *obj_label;
185 142012ed 2022-10-20 thomas size_t mapoff = obj->off;
186 142012ed 2022-10-20 thomas struct got_inflate_checksum csum;
187 142012ed 2022-10-20 thomas
188 142012ed 2022-10-20 thomas memset(&csum, 0, sizeof(csum));
189 142012ed 2022-10-20 thomas csum.input_sha1 = pack_sha1_ctx;
190 142012ed 2022-10-20 thomas csum.input_crc = &obj->crc;
191 142012ed 2022-10-20 thomas
192 142012ed 2022-10-20 thomas err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
193 142012ed 2022-10-20 thomas &obj->tslen, pack, obj->off);
194 142012ed 2022-10-20 thomas if (err)
195 142012ed 2022-10-20 thomas return err;
196 142012ed 2022-10-20 thomas
197 142012ed 2022-10-20 thomas if (pack->map) {
198 142012ed 2022-10-20 thomas obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
199 142012ed 2022-10-20 thomas SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
200 142012ed 2022-10-20 thomas mapoff += obj->tslen;
201 142012ed 2022-10-20 thomas } else {
202 142012ed 2022-10-20 thomas /* XXX Seek back and get the CRC of on-disk type+size bytes. */
203 142012ed 2022-10-20 thomas if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
204 142012ed 2022-10-20 thomas return got_error_from_errno("lseek");
205 142012ed 2022-10-20 thomas err = read_checksum(&obj->crc, pack_sha1_ctx,
206 142012ed 2022-10-20 thomas pack->fd, obj->tslen);
207 142012ed 2022-10-20 thomas if (err)
208 142012ed 2022-10-20 thomas return err;
209 142012ed 2022-10-20 thomas }
210 142012ed 2022-10-20 thomas
211 142012ed 2022-10-20 thomas switch (obj->type) {
212 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_BLOB:
213 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_COMMIT:
214 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_TREE:
215 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_TAG:
216 142012ed 2022-10-20 thomas if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
217 142012ed 2022-10-20 thomas if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
218 142012ed 2022-10-20 thomas err = got_error_from_errno("fseek");
219 142012ed 2022-10-20 thomas break;
220 142012ed 2022-10-20 thomas }
221 142012ed 2022-10-20 thomas if (pack->map) {
222 142012ed 2022-10-20 thomas err = got_inflate_to_file_mmap(&datalen,
223 142012ed 2022-10-20 thomas &obj->len, &csum, pack->map, mapoff,
224 142012ed 2022-10-20 thomas pack->filesize - mapoff, tmpfile);
225 142012ed 2022-10-20 thomas } else {
226 142012ed 2022-10-20 thomas err = got_inflate_to_file_fd(&datalen,
227 142012ed 2022-10-20 thomas &obj->len, &csum, pack->fd, tmpfile);
228 142012ed 2022-10-20 thomas }
229 142012ed 2022-10-20 thomas } else {
230 142012ed 2022-10-20 thomas if (pack->map) {
231 142012ed 2022-10-20 thomas err = got_inflate_to_mem_mmap(&data, &datalen,
232 142012ed 2022-10-20 thomas &obj->len, &csum, pack->map, mapoff,
233 142012ed 2022-10-20 thomas pack->filesize - mapoff);
234 142012ed 2022-10-20 thomas } else {
235 142012ed 2022-10-20 thomas err = got_inflate_to_mem_fd(&data, &datalen,
236 142012ed 2022-10-20 thomas &obj->len, &csum, obj->size, pack->fd);
237 142012ed 2022-10-20 thomas }
238 142012ed 2022-10-20 thomas }
239 142012ed 2022-10-20 thomas if (err)
240 142012ed 2022-10-20 thomas break;
241 142012ed 2022-10-20 thomas SHA1Init(&ctx);
242 142012ed 2022-10-20 thomas err = get_obj_type_label(&obj_label, obj->type);
243 142012ed 2022-10-20 thomas if (err) {
244 142012ed 2022-10-20 thomas free(data);
245 142012ed 2022-10-20 thomas break;
246 142012ed 2022-10-20 thomas }
247 142012ed 2022-10-20 thomas if (asprintf(&header, "%s %lld", obj_label,
248 142012ed 2022-10-20 thomas (long long)obj->size) == -1) {
249 142012ed 2022-10-20 thomas err = got_error_from_errno("asprintf");
250 142012ed 2022-10-20 thomas free(data);
251 142012ed 2022-10-20 thomas break;
252 142012ed 2022-10-20 thomas }
253 142012ed 2022-10-20 thomas headerlen = strlen(header) + 1;
254 142012ed 2022-10-20 thomas SHA1Update(&ctx, header, headerlen);
255 142012ed 2022-10-20 thomas if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
256 142012ed 2022-10-20 thomas err = read_file_sha1(&ctx, tmpfile, datalen);
257 142012ed 2022-10-20 thomas if (err) {
258 142012ed 2022-10-20 thomas free(header);
259 142012ed 2022-10-20 thomas free(data);
260 142012ed 2022-10-20 thomas break;
261 142012ed 2022-10-20 thomas }
262 142012ed 2022-10-20 thomas } else
263 142012ed 2022-10-20 thomas SHA1Update(&ctx, data, datalen);
264 142012ed 2022-10-20 thomas SHA1Final(obj->id.sha1, &ctx);
265 142012ed 2022-10-20 thomas free(header);
266 142012ed 2022-10-20 thomas free(data);
267 142012ed 2022-10-20 thomas break;
268 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_REF_DELTA:
269 142012ed 2022-10-20 thomas memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
270 142012ed 2022-10-20 thomas if (pack->map) {
271 142012ed 2022-10-20 thomas if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
272 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_BAD_PACKFILE);
273 66e6097f 2022-10-27 thomas break;
274 66e6097f 2022-10-27 thomas }
275 66e6097f 2022-10-27 thomas if (mapoff + SHA1_DIGEST_LENGTH > SIZE_MAX) {
276 66e6097f 2022-10-27 thomas err = got_error_fmt(GOT_ERR_RANGE,
277 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
278 66e6097f 2022-10-27 thomas (long long)mapoff + SHA1_DIGEST_LENGTH);
279 142012ed 2022-10-20 thomas break;
280 142012ed 2022-10-20 thomas }
281 142012ed 2022-10-20 thomas memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
282 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
283 142012ed 2022-10-20 thomas obj->crc = crc32(obj->crc, pack->map + mapoff,
284 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
285 142012ed 2022-10-20 thomas SHA1Update(pack_sha1_ctx, pack->map + mapoff,
286 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
287 142012ed 2022-10-20 thomas mapoff += SHA1_DIGEST_LENGTH;
288 142012ed 2022-10-20 thomas err = got_inflate_to_mem_mmap(NULL, &datalen,
289 142012ed 2022-10-20 thomas &obj->len, &csum, pack->map, mapoff,
290 142012ed 2022-10-20 thomas pack->filesize - mapoff);
291 142012ed 2022-10-20 thomas if (err)
292 142012ed 2022-10-20 thomas break;
293 142012ed 2022-10-20 thomas } else {
294 142012ed 2022-10-20 thomas n = read(pack->fd, obj->delta.ref.ref_id.sha1,
295 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
296 142012ed 2022-10-20 thomas if (n == -1) {
297 142012ed 2022-10-20 thomas err = got_error_from_errno("read");
298 142012ed 2022-10-20 thomas break;
299 142012ed 2022-10-20 thomas }
300 142012ed 2022-10-20 thomas if (n < sizeof(obj->id)) {
301 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_BAD_PACKFILE);
302 142012ed 2022-10-20 thomas break;
303 142012ed 2022-10-20 thomas }
304 142012ed 2022-10-20 thomas obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
305 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
306 142012ed 2022-10-20 thomas SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
307 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
308 142012ed 2022-10-20 thomas err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
309 142012ed 2022-10-20 thomas &csum, obj->size, pack->fd);
310 142012ed 2022-10-20 thomas if (err)
311 142012ed 2022-10-20 thomas break;
312 142012ed 2022-10-20 thomas }
313 142012ed 2022-10-20 thomas obj->len += SHA1_DIGEST_LENGTH;
314 142012ed 2022-10-20 thomas break;
315 142012ed 2022-10-20 thomas case GOT_OBJ_TYPE_OFFSET_DELTA:
316 142012ed 2022-10-20 thomas memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
317 142012ed 2022-10-20 thomas err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
318 142012ed 2022-10-20 thomas &obj->delta.ofs.base_offsetlen, pack, obj->off,
319 142012ed 2022-10-20 thomas obj->tslen);
320 142012ed 2022-10-20 thomas if (err)
321 142012ed 2022-10-20 thomas break;
322 142012ed 2022-10-20 thomas
323 142012ed 2022-10-20 thomas if (pack->map) {
324 6cf83e6a 2022-10-25 thomas if (mapoff + obj->delta.ofs.base_offsetlen >=
325 6cf83e6a 2022-10-25 thomas pack->filesize) {
326 6cf83e6a 2022-10-25 thomas err = got_error(GOT_ERR_BAD_PACKFILE);
327 6cf83e6a 2022-10-25 thomas break;
328 6cf83e6a 2022-10-25 thomas }
329 6cf83e6a 2022-10-25 thomas
330 66e6097f 2022-10-27 thomas if (mapoff + obj->delta.ofs.base_offsetlen >
331 66e6097f 2022-10-27 thomas SIZE_MAX) {
332 66e6097f 2022-10-27 thomas err = got_error_fmt(GOT_ERR_RANGE,
333 66e6097f 2022-10-27 thomas "mapoff %lld would overflow size_t",
334 66e6097f 2022-10-27 thomas (long long)mapoff
335 66e6097f 2022-10-27 thomas + obj->delta.ofs.base_offsetlen);
336 66e6097f 2022-10-27 thomas }
337 66e6097f 2022-10-27 thomas
338 142012ed 2022-10-20 thomas obj->crc = crc32(obj->crc, pack->map + mapoff,
339 142012ed 2022-10-20 thomas obj->delta.ofs.base_offsetlen);
340 142012ed 2022-10-20 thomas SHA1Update(pack_sha1_ctx, pack->map + mapoff,
341 142012ed 2022-10-20 thomas obj->delta.ofs.base_offsetlen);
342 142012ed 2022-10-20 thomas mapoff += obj->delta.ofs.base_offsetlen;
343 142012ed 2022-10-20 thomas err = got_inflate_to_mem_mmap(NULL, &datalen,
344 142012ed 2022-10-20 thomas &obj->len, &csum, pack->map, mapoff,
345 142012ed 2022-10-20 thomas pack->filesize - mapoff);
346 142012ed 2022-10-20 thomas if (err)
347 142012ed 2022-10-20 thomas break;
348 142012ed 2022-10-20 thomas } else {
349 142012ed 2022-10-20 thomas /*
350 142012ed 2022-10-20 thomas * XXX Seek back and get CRC and SHA1 of on-disk
351 142012ed 2022-10-20 thomas * offset bytes.
352 142012ed 2022-10-20 thomas */
353 142012ed 2022-10-20 thomas if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
354 142012ed 2022-10-20 thomas == -1) {
355 142012ed 2022-10-20 thomas err = got_error_from_errno("lseek");
356 142012ed 2022-10-20 thomas break;
357 142012ed 2022-10-20 thomas }
358 142012ed 2022-10-20 thomas err = read_checksum(&obj->crc, pack_sha1_ctx,
359 142012ed 2022-10-20 thomas pack->fd, obj->delta.ofs.base_offsetlen);
360 142012ed 2022-10-20 thomas if (err)
361 142012ed 2022-10-20 thomas break;
362 142012ed 2022-10-20 thomas
363 142012ed 2022-10-20 thomas err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
364 142012ed 2022-10-20 thomas &csum, obj->size, pack->fd);
365 142012ed 2022-10-20 thomas if (err)
366 142012ed 2022-10-20 thomas break;
367 142012ed 2022-10-20 thomas }
368 142012ed 2022-10-20 thomas obj->len += obj->delta.ofs.base_offsetlen;
369 142012ed 2022-10-20 thomas break;
370 142012ed 2022-10-20 thomas default:
371 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_OBJ_TYPE);
372 142012ed 2022-10-20 thomas break;
373 142012ed 2022-10-20 thomas }
374 142012ed 2022-10-20 thomas
375 142012ed 2022-10-20 thomas return err;
376 142012ed 2022-10-20 thomas }
377 142012ed 2022-10-20 thomas
378 3efd8e31 2022-10-23 thomas const struct got_error *
379 3efd8e31 2022-10-23 thomas got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
380 142012ed 2022-10-20 thomas {
381 142012ed 2022-10-20 thomas ssize_t w;
382 142012ed 2022-10-20 thomas
383 142012ed 2022-10-20 thomas SHA1Update(ctx, buf, len);
384 142012ed 2022-10-20 thomas
385 142012ed 2022-10-20 thomas w = write(fd, buf, len);
386 142012ed 2022-10-20 thomas if (w == -1)
387 142012ed 2022-10-20 thomas return got_error_from_errno("write");
388 142012ed 2022-10-20 thomas if (w != len)
389 142012ed 2022-10-20 thomas return got_error(GOT_ERR_IO);
390 142012ed 2022-10-20 thomas
391 142012ed 2022-10-20 thomas return NULL;
392 142012ed 2022-10-20 thomas }
393 142012ed 2022-10-20 thomas
394 142012ed 2022-10-20 thomas static const struct got_error *
395 142012ed 2022-10-20 thomas resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
396 142012ed 2022-10-20 thomas struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
397 142012ed 2022-10-20 thomas FILE *delta_accum_file)
398 142012ed 2022-10-20 thomas {
399 142012ed 2022-10-20 thomas const struct got_error *err = NULL;
400 142012ed 2022-10-20 thomas struct got_delta_chain deltas;
401 142012ed 2022-10-20 thomas struct got_delta *delta;
402 142012ed 2022-10-20 thomas uint8_t *buf = NULL;
403 142012ed 2022-10-20 thomas size_t len = 0;
404 142012ed 2022-10-20 thomas SHA1_CTX ctx;
405 142012ed 2022-10-20 thomas char *header = NULL;
406 142012ed 2022-10-20 thomas size_t headerlen;
407 142012ed 2022-10-20 thomas uint64_t max_size;
408 142012ed 2022-10-20 thomas int base_obj_type;
409 142012ed 2022-10-20 thomas const char *obj_label;
410 142012ed 2022-10-20 thomas
411 142012ed 2022-10-20 thomas deltas.nentries = 0;
412 142012ed 2022-10-20 thomas STAILQ_INIT(&deltas.entries);
413 142012ed 2022-10-20 thomas
414 142012ed 2022-10-20 thomas err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
415 142012ed 2022-10-20 thomas obj->off, obj->tslen, obj->type, obj->size,
416 142012ed 2022-10-20 thomas GOT_DELTA_CHAIN_RECURSION_MAX);
417 142012ed 2022-10-20 thomas if (err)
418 142012ed 2022-10-20 thomas goto done;
419 142012ed 2022-10-20 thomas
420 142012ed 2022-10-20 thomas err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
421 142012ed 2022-10-20 thomas if (err)
422 142012ed 2022-10-20 thomas goto done;
423 142012ed 2022-10-20 thomas if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
424 142012ed 2022-10-20 thomas rewind(tmpfile);
425 142012ed 2022-10-20 thomas rewind(delta_base_file);
426 142012ed 2022-10-20 thomas rewind(delta_accum_file);
427 142012ed 2022-10-20 thomas err = got_pack_dump_delta_chain_to_file(&len, &deltas,
428 142012ed 2022-10-20 thomas pack, tmpfile, delta_base_file, delta_accum_file);
429 142012ed 2022-10-20 thomas if (err)
430 142012ed 2022-10-20 thomas goto done;
431 142012ed 2022-10-20 thomas } else {
432 142012ed 2022-10-20 thomas err = got_pack_dump_delta_chain_to_mem(&buf, &len,
433 142012ed 2022-10-20 thomas &deltas, pack);
434 142012ed 2022-10-20 thomas }
435 142012ed 2022-10-20 thomas if (err)
436 142012ed 2022-10-20 thomas goto done;
437 142012ed 2022-10-20 thomas
438 142012ed 2022-10-20 thomas err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
439 142012ed 2022-10-20 thomas if (err)
440 142012ed 2022-10-20 thomas goto done;
441 142012ed 2022-10-20 thomas err = get_obj_type_label(&obj_label, base_obj_type);
442 142012ed 2022-10-20 thomas if (err)
443 142012ed 2022-10-20 thomas goto done;
444 142012ed 2022-10-20 thomas if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
445 142012ed 2022-10-20 thomas err = got_error_from_errno("asprintf");
446 142012ed 2022-10-20 thomas goto done;
447 142012ed 2022-10-20 thomas }
448 142012ed 2022-10-20 thomas headerlen = strlen(header) + 1;
449 142012ed 2022-10-20 thomas SHA1Init(&ctx);
450 142012ed 2022-10-20 thomas SHA1Update(&ctx, header, headerlen);
451 142012ed 2022-10-20 thomas if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
452 142012ed 2022-10-20 thomas err = read_file_sha1(&ctx, tmpfile, len);
453 142012ed 2022-10-20 thomas if (err)
454 142012ed 2022-10-20 thomas goto done;
455 142012ed 2022-10-20 thomas } else
456 142012ed 2022-10-20 thomas SHA1Update(&ctx, buf, len);
457 142012ed 2022-10-20 thomas SHA1Final(obj->id.sha1, &ctx);
458 142012ed 2022-10-20 thomas done:
459 142012ed 2022-10-20 thomas free(buf);
460 142012ed 2022-10-20 thomas free(header);
461 142012ed 2022-10-20 thomas while (!STAILQ_EMPTY(&deltas.entries)) {
462 142012ed 2022-10-20 thomas delta = STAILQ_FIRST(&deltas.entries);
463 142012ed 2022-10-20 thomas STAILQ_REMOVE_HEAD(&deltas.entries, entry);
464 142012ed 2022-10-20 thomas free(delta);
465 142012ed 2022-10-20 thomas }
466 142012ed 2022-10-20 thomas return err;
467 142012ed 2022-10-20 thomas }
468 142012ed 2022-10-20 thomas
469 142012ed 2022-10-20 thomas /* Determine the slot in the pack index a given object ID should use. */
470 142012ed 2022-10-20 thomas static int
471 142012ed 2022-10-20 thomas find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
472 142012ed 2022-10-20 thomas {
473 142012ed 2022-10-20 thomas u_int8_t id0 = sha1[0];
474 142012ed 2022-10-20 thomas uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
475 142012ed 2022-10-20 thomas int left = 0, right = nindexed - 1;
476 142012ed 2022-10-20 thomas int cmp = 0, i = 0;
477 142012ed 2022-10-20 thomas
478 142012ed 2022-10-20 thomas if (id0 > 0)
479 142012ed 2022-10-20 thomas left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
480 142012ed 2022-10-20 thomas
481 142012ed 2022-10-20 thomas while (left <= right) {
482 142012ed 2022-10-20 thomas struct got_packidx_object_id *oid;
483 142012ed 2022-10-20 thomas
484 142012ed 2022-10-20 thomas i = ((left + right) / 2);
485 142012ed 2022-10-20 thomas oid = &packidx->hdr.sorted_ids[i];
486 142012ed 2022-10-20 thomas
487 142012ed 2022-10-20 thomas cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
488 142012ed 2022-10-20 thomas if (cmp == 0)
489 142012ed 2022-10-20 thomas return -1; /* object already indexed */
490 142012ed 2022-10-20 thomas else if (cmp > 0)
491 142012ed 2022-10-20 thomas left = i + 1;
492 142012ed 2022-10-20 thomas else if (cmp < 0)
493 142012ed 2022-10-20 thomas right = i - 1;
494 142012ed 2022-10-20 thomas }
495 142012ed 2022-10-20 thomas
496 142012ed 2022-10-20 thomas return left;
497 142012ed 2022-10-20 thomas }
498 142012ed 2022-10-20 thomas
499 142012ed 2022-10-20 thomas #if 0
500 142012ed 2022-10-20 thomas static void
501 142012ed 2022-10-20 thomas print_packidx(struct got_packidx *packidx)
502 142012ed 2022-10-20 thomas {
503 142012ed 2022-10-20 thomas uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
504 142012ed 2022-10-20 thomas int i;
505 142012ed 2022-10-20 thomas
506 142012ed 2022-10-20 thomas fprintf(stderr, "object IDs:\n");
507 142012ed 2022-10-20 thomas for (i = 0; i < nindexed; i++) {
508 142012ed 2022-10-20 thomas char hex[SHA1_DIGEST_STRING_LENGTH];
509 142012ed 2022-10-20 thomas got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
510 142012ed 2022-10-20 thomas hex, sizeof(hex));
511 142012ed 2022-10-20 thomas fprintf(stderr, "%s\n", hex);
512 142012ed 2022-10-20 thomas }
513 142012ed 2022-10-20 thomas fprintf(stderr, "\n");
514 142012ed 2022-10-20 thomas
515 142012ed 2022-10-20 thomas fprintf(stderr, "object offsets:\n");
516 142012ed 2022-10-20 thomas for (i = 0; i < nindexed; i++) {
517 142012ed 2022-10-20 thomas uint32_t offset = be32toh(packidx->hdr.offsets[i]);
518 142012ed 2022-10-20 thomas if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
519 142012ed 2022-10-20 thomas int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
520 142012ed 2022-10-20 thomas fprintf(stderr, "%u -> %llu\n", offset,
521 142012ed 2022-10-20 thomas be64toh(packidx->hdr.large_offsets[j]));
522 142012ed 2022-10-20 thomas } else
523 142012ed 2022-10-20 thomas fprintf(stderr, "%u\n", offset);
524 142012ed 2022-10-20 thomas }
525 142012ed 2022-10-20 thomas fprintf(stderr, "\n");
526 142012ed 2022-10-20 thomas
527 142012ed 2022-10-20 thomas fprintf(stderr, "fanout table:");
528 142012ed 2022-10-20 thomas for (i = 0; i <= 0xff; i++)
529 142012ed 2022-10-20 thomas fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
530 142012ed 2022-10-20 thomas fprintf(stderr, "\n");
531 142012ed 2022-10-20 thomas }
532 142012ed 2022-10-20 thomas #endif
533 142012ed 2022-10-20 thomas
534 142012ed 2022-10-20 thomas static void
535 142012ed 2022-10-20 thomas add_indexed_object(struct got_packidx *packidx, uint32_t idx,
536 142012ed 2022-10-20 thomas struct got_indexed_object *obj)
537 142012ed 2022-10-20 thomas {
538 142012ed 2022-10-20 thomas int i;
539 142012ed 2022-10-20 thomas
540 142012ed 2022-10-20 thomas memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
541 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
542 142012ed 2022-10-20 thomas packidx->hdr.crc32[idx] = htobe32(obj->crc);
543 142012ed 2022-10-20 thomas if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
544 142012ed 2022-10-20 thomas packidx->hdr.offsets[idx] = htobe32(obj->off);
545 142012ed 2022-10-20 thomas else {
546 142012ed 2022-10-20 thomas packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
547 142012ed 2022-10-20 thomas GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
548 142012ed 2022-10-20 thomas packidx->hdr.large_offsets[packidx->nlargeobj] =
549 142012ed 2022-10-20 thomas htobe64(obj->off);
550 142012ed 2022-10-20 thomas packidx->nlargeobj++;
551 142012ed 2022-10-20 thomas }
552 142012ed 2022-10-20 thomas
553 142012ed 2022-10-20 thomas for (i = obj->id.sha1[0]; i <= 0xff; i++) {
554 142012ed 2022-10-20 thomas uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
555 142012ed 2022-10-20 thomas packidx->hdr.fanout_table[i] = htobe32(n + 1);
556 142012ed 2022-10-20 thomas }
557 142012ed 2022-10-20 thomas }
558 142012ed 2022-10-20 thomas
559 142012ed 2022-10-20 thomas static int
560 142012ed 2022-10-20 thomas indexed_obj_cmp(const void *pa, const void *pb)
561 142012ed 2022-10-20 thomas {
562 142012ed 2022-10-20 thomas struct got_indexed_object *a, *b;
563 142012ed 2022-10-20 thomas
564 142012ed 2022-10-20 thomas a = (struct got_indexed_object *)pa;
565 142012ed 2022-10-20 thomas b = (struct got_indexed_object *)pb;
566 142012ed 2022-10-20 thomas return got_object_id_cmp(&a->id, &b->id);
567 142012ed 2022-10-20 thomas }
568 142012ed 2022-10-20 thomas
569 142012ed 2022-10-20 thomas static void
570 eebe1fbb 2022-10-20 thomas make_packidx(struct got_packidx *packidx, uint32_t nobj,
571 142012ed 2022-10-20 thomas struct got_indexed_object *objects)
572 142012ed 2022-10-20 thomas {
573 142012ed 2022-10-20 thomas struct got_indexed_object *obj;
574 142012ed 2022-10-20 thomas int i;
575 142012ed 2022-10-20 thomas uint32_t idx = 0;
576 142012ed 2022-10-20 thomas
577 142012ed 2022-10-20 thomas qsort(objects, nobj, sizeof(struct got_indexed_object),
578 142012ed 2022-10-20 thomas indexed_obj_cmp);
579 142012ed 2022-10-20 thomas
580 142012ed 2022-10-20 thomas memset(packidx->hdr.fanout_table, 0,
581 142012ed 2022-10-20 thomas GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
582 142012ed 2022-10-20 thomas packidx->nlargeobj = 0;
583 142012ed 2022-10-20 thomas
584 142012ed 2022-10-20 thomas for (i = 0; i < nobj; i++) {
585 142012ed 2022-10-20 thomas obj = &objects[i];
586 142012ed 2022-10-20 thomas if (obj->valid)
587 142012ed 2022-10-20 thomas add_indexed_object(packidx, idx++, obj);
588 142012ed 2022-10-20 thomas }
589 142012ed 2022-10-20 thomas }
590 142012ed 2022-10-20 thomas
591 142012ed 2022-10-20 thomas static void
592 eebe1fbb 2022-10-20 thomas update_packidx(struct got_packidx *packidx, uint32_t nobj,
593 142012ed 2022-10-20 thomas struct got_indexed_object *obj)
594 142012ed 2022-10-20 thomas {
595 142012ed 2022-10-20 thomas int idx;
596 142012ed 2022-10-20 thomas uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
597 142012ed 2022-10-20 thomas
598 142012ed 2022-10-20 thomas idx = find_object_idx(packidx, obj->id.sha1);
599 aac3e701 2023-02-03 thomas if (idx == -1)
600 142012ed 2022-10-20 thomas return; /* object already indexed */
601 142012ed 2022-10-20 thomas
602 142012ed 2022-10-20 thomas memmove(&packidx->hdr.sorted_ids[idx + 1],
603 142012ed 2022-10-20 thomas &packidx->hdr.sorted_ids[idx],
604 142012ed 2022-10-20 thomas sizeof(struct got_packidx_object_id) * (nindexed - idx));
605 142012ed 2022-10-20 thomas memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
606 142012ed 2022-10-20 thomas sizeof(uint32_t) * (nindexed - idx));
607 142012ed 2022-10-20 thomas
608 142012ed 2022-10-20 thomas add_indexed_object(packidx, idx, obj);
609 142012ed 2022-10-20 thomas }
610 142012ed 2022-10-20 thomas
611 142012ed 2022-10-20 thomas static const struct got_error *
612 eebe1fbb 2022-10-20 thomas report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
613 eebe1fbb 2022-10-20 thomas uint32_t nobj_resolved, struct got_ratelimit *rl,
614 142012ed 2022-10-20 thomas got_pack_index_progress_cb progress_cb, void *progress_arg)
615 142012ed 2022-10-20 thomas {
616 142012ed 2022-10-20 thomas const struct got_error *err;
617 142012ed 2022-10-20 thomas int elapsed = 0;
618 142012ed 2022-10-20 thomas
619 142012ed 2022-10-20 thomas if (rl) {
620 142012ed 2022-10-20 thomas err = got_ratelimit_check(&elapsed, rl);
621 142012ed 2022-10-20 thomas if (err || !elapsed)
622 142012ed 2022-10-20 thomas return err;
623 142012ed 2022-10-20 thomas }
624 142012ed 2022-10-20 thomas
625 142012ed 2022-10-20 thomas return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
626 142012ed 2022-10-20 thomas nobj_resolved);
627 142012ed 2022-10-20 thomas }
628 142012ed 2022-10-20 thomas
629 142012ed 2022-10-20 thomas const struct got_error *
630 142012ed 2022-10-20 thomas got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
631 142012ed 2022-10-20 thomas FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
632 aecd2225 2022-10-20 thomas got_pack_index_progress_cb progress_cb, void *progress_arg,
633 aecd2225 2022-10-20 thomas struct got_ratelimit *rl)
634 142012ed 2022-10-20 thomas {
635 142012ed 2022-10-20 thomas const struct got_error *err;
636 142012ed 2022-10-20 thomas struct got_packfile_hdr hdr;
637 142012ed 2022-10-20 thomas struct got_packidx packidx;
638 142012ed 2022-10-20 thomas char buf[8];
639 142012ed 2022-10-20 thomas char pack_sha1[SHA1_DIGEST_LENGTH];
640 eebe1fbb 2022-10-20 thomas uint32_t nobj, nvalid, nloose, nresolved = 0, i;
641 142012ed 2022-10-20 thomas struct got_indexed_object *objects = NULL, *obj;
642 142012ed 2022-10-20 thomas SHA1_CTX ctx;
643 142012ed 2022-10-20 thomas uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
644 142012ed 2022-10-20 thomas ssize_t r, w;
645 142012ed 2022-10-20 thomas int pass, have_ref_deltas = 0, first_delta_idx = -1;
646 142012ed 2022-10-20 thomas size_t mapoff = 0;
647 142012ed 2022-10-20 thomas int p_indexed = 0, last_p_indexed = -1;
648 142012ed 2022-10-20 thomas int p_resolved = 0, last_p_resolved = -1;
649 142012ed 2022-10-20 thomas
650 142012ed 2022-10-20 thomas /* Require that pack file header and SHA1 trailer are present. */
651 142012ed 2022-10-20 thomas if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
652 142012ed 2022-10-20 thomas return got_error_msg(GOT_ERR_BAD_PACKFILE,
653 142012ed 2022-10-20 thomas "short pack file");
654 142012ed 2022-10-20 thomas
655 142012ed 2022-10-20 thomas if (pack->map) {
656 142012ed 2022-10-20 thomas memcpy(&hdr, pack->map, sizeof(hdr));
657 142012ed 2022-10-20 thomas mapoff += sizeof(hdr);
658 142012ed 2022-10-20 thomas } else {
659 142012ed 2022-10-20 thomas r = read(pack->fd, &hdr, sizeof(hdr));
660 142012ed 2022-10-20 thomas if (r == -1)
661 142012ed 2022-10-20 thomas return got_error_from_errno("read");
662 142012ed 2022-10-20 thomas if (r < sizeof(hdr))
663 142012ed 2022-10-20 thomas return got_error_msg(GOT_ERR_BAD_PACKFILE,
664 142012ed 2022-10-20 thomas "short pack file");
665 142012ed 2022-10-20 thomas }
666 142012ed 2022-10-20 thomas
667 142012ed 2022-10-20 thomas if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
668 142012ed 2022-10-20 thomas return got_error_msg(GOT_ERR_BAD_PACKFILE,
669 142012ed 2022-10-20 thomas "bad packfile signature");
670 142012ed 2022-10-20 thomas if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
671 142012ed 2022-10-20 thomas return got_error_msg(GOT_ERR_BAD_PACKFILE,
672 142012ed 2022-10-20 thomas "bad packfile version");
673 142012ed 2022-10-20 thomas nobj = be32toh(hdr.nobjects);
674 142012ed 2022-10-20 thomas if (nobj == 0)
675 142012ed 2022-10-20 thomas return got_error_msg(GOT_ERR_BAD_PACKFILE,
676 142012ed 2022-10-20 thomas "bad packfile with zero objects");
677 142012ed 2022-10-20 thomas
678 142012ed 2022-10-20 thomas /* We compute the SHA1 of pack file contents and verify later on. */
679 142012ed 2022-10-20 thomas SHA1Init(&ctx);
680 142012ed 2022-10-20 thomas SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
681 142012ed 2022-10-20 thomas
682 142012ed 2022-10-20 thomas /*
683 142012ed 2022-10-20 thomas * Create an in-memory pack index which will grow as objects
684 142012ed 2022-10-20 thomas * IDs in the pack file are discovered. Only fields used to
685 142012ed 2022-10-20 thomas * read deltified objects will be needed by the pack.c library
686 142012ed 2022-10-20 thomas * code, so setting up just a pack index header is sufficient.
687 142012ed 2022-10-20 thomas */
688 142012ed 2022-10-20 thomas memset(&packidx, 0, sizeof(packidx));
689 142012ed 2022-10-20 thomas packidx.hdr.magic = malloc(sizeof(uint32_t));
690 142012ed 2022-10-20 thomas if (packidx.hdr.magic == NULL)
691 142012ed 2022-10-20 thomas return got_error_from_errno("malloc");
692 142012ed 2022-10-20 thomas *packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
693 142012ed 2022-10-20 thomas packidx.hdr.version = malloc(sizeof(uint32_t));
694 142012ed 2022-10-20 thomas if (packidx.hdr.version == NULL) {
695 142012ed 2022-10-20 thomas err = got_error_from_errno("malloc");
696 142012ed 2022-10-20 thomas goto done;
697 142012ed 2022-10-20 thomas }
698 142012ed 2022-10-20 thomas *packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
699 142012ed 2022-10-20 thomas packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
700 142012ed 2022-10-20 thomas sizeof(uint32_t));
701 142012ed 2022-10-20 thomas if (packidx.hdr.fanout_table == NULL) {
702 142012ed 2022-10-20 thomas err = got_error_from_errno("calloc");
703 142012ed 2022-10-20 thomas goto done;
704 142012ed 2022-10-20 thomas }
705 142012ed 2022-10-20 thomas packidx.hdr.sorted_ids = calloc(nobj,
706 142012ed 2022-10-20 thomas sizeof(struct got_packidx_object_id));
707 142012ed 2022-10-20 thomas if (packidx.hdr.sorted_ids == NULL) {
708 142012ed 2022-10-20 thomas err = got_error_from_errno("calloc");
709 142012ed 2022-10-20 thomas goto done;
710 142012ed 2022-10-20 thomas }
711 142012ed 2022-10-20 thomas packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
712 142012ed 2022-10-20 thomas if (packidx.hdr.crc32 == NULL) {
713 142012ed 2022-10-20 thomas err = got_error_from_errno("calloc");
714 142012ed 2022-10-20 thomas goto done;
715 142012ed 2022-10-20 thomas }
716 142012ed 2022-10-20 thomas packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
717 142012ed 2022-10-20 thomas if (packidx.hdr.offsets == NULL) {
718 142012ed 2022-10-20 thomas err = got_error_from_errno("calloc");
719 142012ed 2022-10-20 thomas goto done;
720 142012ed 2022-10-20 thomas }
721 142012ed 2022-10-20 thomas /* Large offsets table is empty for pack files < 2 GB. */
722 142012ed 2022-10-20 thomas if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
723 142012ed 2022-10-20 thomas packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
724 142012ed 2022-10-20 thomas if (packidx.hdr.large_offsets == NULL) {
725 142012ed 2022-10-20 thomas err = got_error_from_errno("calloc");
726 142012ed 2022-10-20 thomas goto done;
727 142012ed 2022-10-20 thomas }
728 142012ed 2022-10-20 thomas }
729 142012ed 2022-10-20 thomas
730 142012ed 2022-10-20 thomas nvalid = 0;
731 142012ed 2022-10-20 thomas nloose = 0;
732 142012ed 2022-10-20 thomas objects = calloc(nobj, sizeof(struct got_indexed_object));
733 142012ed 2022-10-20 thomas if (objects == NULL)
734 142012ed 2022-10-20 thomas return got_error_from_errno("calloc");
735 142012ed 2022-10-20 thomas
736 142012ed 2022-10-20 thomas /*
737 142012ed 2022-10-20 thomas * First pass: locate all objects and identify un-deltified objects.
738 142012ed 2022-10-20 thomas *
739 142012ed 2022-10-20 thomas * When this pass has completed we will know offset, type, size, and
740 142012ed 2022-10-20 thomas * CRC information for all objects in this pack file. We won't know
741 142012ed 2022-10-20 thomas * any of the actual object IDs of deltified objects yet since we
742 142012ed 2022-10-20 thomas * will not yet attempt to combine deltas.
743 142012ed 2022-10-20 thomas */
744 142012ed 2022-10-20 thomas pass = 1;
745 142012ed 2022-10-20 thomas for (i = 0; i < nobj; i++) {
746 142012ed 2022-10-20 thomas /* Don't send too many progress privsep messages. */
747 142012ed 2022-10-20 thomas p_indexed = ((i + 1) * 100) / nobj;
748 142012ed 2022-10-20 thomas if (p_indexed != last_p_indexed) {
749 142012ed 2022-10-20 thomas err = report_progress(nobj, i + 1, nloose, 0,
750 aecd2225 2022-10-20 thomas rl, progress_cb, progress_arg);
751 142012ed 2022-10-20 thomas if (err)
752 142012ed 2022-10-20 thomas goto done;
753 142012ed 2022-10-20 thomas last_p_indexed = p_indexed;
754 142012ed 2022-10-20 thomas }
755 142012ed 2022-10-20 thomas
756 142012ed 2022-10-20 thomas obj = &objects[i];
757 142012ed 2022-10-20 thomas obj->crc = crc32(0L, NULL, 0);
758 142012ed 2022-10-20 thomas
759 142012ed 2022-10-20 thomas /* Store offset to type+size information for this object. */
760 142012ed 2022-10-20 thomas if (pack->map) {
761 142012ed 2022-10-20 thomas obj->off = mapoff;
762 142012ed 2022-10-20 thomas } else {
763 142012ed 2022-10-20 thomas obj->off = lseek(pack->fd, 0, SEEK_CUR);
764 142012ed 2022-10-20 thomas if (obj->off == -1) {
765 142012ed 2022-10-20 thomas err = got_error_from_errno("lseek");
766 142012ed 2022-10-20 thomas goto done;
767 142012ed 2022-10-20 thomas }
768 142012ed 2022-10-20 thomas }
769 142012ed 2022-10-20 thomas
770 142012ed 2022-10-20 thomas err = read_packed_object(pack, obj, tmpfile, &ctx);
771 142012ed 2022-10-20 thomas if (err)
772 142012ed 2022-10-20 thomas goto done;
773 142012ed 2022-10-20 thomas
774 142012ed 2022-10-20 thomas if (pack->map) {
775 142012ed 2022-10-20 thomas mapoff += obj->tslen + obj->len;
776 142012ed 2022-10-20 thomas } else {
777 142012ed 2022-10-20 thomas if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
778 142012ed 2022-10-20 thomas SEEK_SET) == -1) {
779 142012ed 2022-10-20 thomas err = got_error_from_errno("lseek");
780 142012ed 2022-10-20 thomas goto done;
781 142012ed 2022-10-20 thomas }
782 142012ed 2022-10-20 thomas }
783 142012ed 2022-10-20 thomas
784 142012ed 2022-10-20 thomas if (obj->type == GOT_OBJ_TYPE_BLOB ||
785 142012ed 2022-10-20 thomas obj->type == GOT_OBJ_TYPE_TREE ||
786 142012ed 2022-10-20 thomas obj->type == GOT_OBJ_TYPE_COMMIT ||
787 142012ed 2022-10-20 thomas obj->type == GOT_OBJ_TYPE_TAG) {
788 142012ed 2022-10-20 thomas obj->valid = 1;
789 142012ed 2022-10-20 thomas nloose++;
790 142012ed 2022-10-20 thomas } else {
791 142012ed 2022-10-20 thomas if (first_delta_idx == -1)
792 142012ed 2022-10-20 thomas first_delta_idx = i;
793 142012ed 2022-10-20 thomas if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
794 142012ed 2022-10-20 thomas have_ref_deltas = 1;
795 142012ed 2022-10-20 thomas }
796 142012ed 2022-10-20 thomas }
797 142012ed 2022-10-20 thomas nvalid = nloose;
798 142012ed 2022-10-20 thomas
799 142012ed 2022-10-20 thomas /*
800 142012ed 2022-10-20 thomas * Having done a full pass over the pack file and can now
801 142012ed 2022-10-20 thomas * verify its checksum.
802 142012ed 2022-10-20 thomas */
803 142012ed 2022-10-20 thomas SHA1Final(pack_sha1, &ctx);
804 3efd8e31 2022-10-23 thomas
805 142012ed 2022-10-20 thomas if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
806 3efd8e31 2022-10-23 thomas err = got_error(GOT_ERR_PACKFILE_CSUM);
807 142012ed 2022-10-20 thomas goto done;
808 142012ed 2022-10-20 thomas }
809 142012ed 2022-10-20 thomas
810 142012ed 2022-10-20 thomas /* Verify the SHA1 checksum stored at the end of the pack file. */
811 142012ed 2022-10-20 thomas if (pack->map) {
812 66e6097f 2022-10-27 thomas if (pack->filesize > SIZE_MAX) {
813 66e6097f 2022-10-27 thomas err = got_error_fmt(GOT_ERR_RANGE,
814 66e6097f 2022-10-27 thomas "filesize %lld overflows size_t",
815 66e6097f 2022-10-27 thomas (long long)pack->filesize);
816 66e6097f 2022-10-27 thomas goto done;
817 66e6097f 2022-10-27 thomas }
818 66e6097f 2022-10-27 thomas
819 142012ed 2022-10-20 thomas memcpy(pack_sha1_expected, pack->map +
820 142012ed 2022-10-20 thomas pack->filesize - SHA1_DIGEST_LENGTH,
821 142012ed 2022-10-20 thomas SHA1_DIGEST_LENGTH);
822 142012ed 2022-10-20 thomas } else {
823 142012ed 2022-10-20 thomas ssize_t n;
824 142012ed 2022-10-20 thomas if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
825 142012ed 2022-10-20 thomas err = got_error_from_errno("lseek");
826 142012ed 2022-10-20 thomas goto done;
827 142012ed 2022-10-20 thomas }
828 142012ed 2022-10-20 thomas n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
829 142012ed 2022-10-20 thomas if (n == -1) {
830 142012ed 2022-10-20 thomas err = got_error_from_errno("read");
831 142012ed 2022-10-20 thomas goto done;
832 142012ed 2022-10-20 thomas }
833 142012ed 2022-10-20 thomas if (n != SHA1_DIGEST_LENGTH) {
834 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_IO);
835 142012ed 2022-10-20 thomas goto done;
836 142012ed 2022-10-20 thomas }
837 142012ed 2022-10-20 thomas }
838 142012ed 2022-10-20 thomas if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
839 142012ed 2022-10-20 thomas err = got_error_msg(GOT_ERR_BAD_PACKFILE,
840 142012ed 2022-10-20 thomas "bad checksum in pack file trailer");
841 142012ed 2022-10-20 thomas goto done;
842 142012ed 2022-10-20 thomas }
843 142012ed 2022-10-20 thomas
844 142012ed 2022-10-20 thomas if (first_delta_idx == -1)
845 142012ed 2022-10-20 thomas first_delta_idx = 0;
846 142012ed 2022-10-20 thomas
847 142012ed 2022-10-20 thomas /* In order to resolve ref deltas we need an in-progress pack index. */
848 142012ed 2022-10-20 thomas if (have_ref_deltas)
849 142012ed 2022-10-20 thomas make_packidx(&packidx, nobj, objects);
850 142012ed 2022-10-20 thomas
851 142012ed 2022-10-20 thomas /*
852 142012ed 2022-10-20 thomas * Second pass: We can now resolve deltas to compute the IDs of
853 142012ed 2022-10-20 thomas * objects which appear in deltified form. Because deltas can be
854 142012ed 2022-10-20 thomas * chained this pass may require a couple of iterations until all
855 142012ed 2022-10-20 thomas * IDs of deltified objects have been discovered.
856 142012ed 2022-10-20 thomas */
857 142012ed 2022-10-20 thomas pass++;
858 142012ed 2022-10-20 thomas while (nvalid != nobj) {
859 142012ed 2022-10-20 thomas int n = 0;
860 142012ed 2022-10-20 thomas /*
861 142012ed 2022-10-20 thomas * This loop will only run once unless the pack file
862 142012ed 2022-10-20 thomas * contains ref deltas which refer to objects located
863 142012ed 2022-10-20 thomas * later in the pack file, which is unusual.
864 142012ed 2022-10-20 thomas * Offset deltas can always be resolved in one pass
865 142012ed 2022-10-20 thomas * unless the packfile is corrupt.
866 142012ed 2022-10-20 thomas */
867 142012ed 2022-10-20 thomas for (i = first_delta_idx; i < nobj; i++) {
868 142012ed 2022-10-20 thomas obj = &objects[i];
869 142012ed 2022-10-20 thomas if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
870 142012ed 2022-10-20 thomas obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
871 142012ed 2022-10-20 thomas continue;
872 142012ed 2022-10-20 thomas
873 142012ed 2022-10-20 thomas if (obj->valid)
874 142012ed 2022-10-20 thomas continue;
875 142012ed 2022-10-20 thomas
876 142012ed 2022-10-20 thomas if (pack->map == NULL && lseek(pack->fd,
877 142012ed 2022-10-20 thomas obj->off + obj->tslen, SEEK_SET) == -1) {
878 142012ed 2022-10-20 thomas err = got_error_from_errno("lseek");
879 142012ed 2022-10-20 thomas goto done;
880 142012ed 2022-10-20 thomas }
881 142012ed 2022-10-20 thomas
882 142012ed 2022-10-20 thomas err = resolve_deltified_object(pack, &packidx, obj,
883 142012ed 2022-10-20 thomas tmpfile, delta_base_file, delta_accum_file);
884 142012ed 2022-10-20 thomas if (err) {
885 142012ed 2022-10-20 thomas if (err->code != GOT_ERR_NO_OBJ)
886 142012ed 2022-10-20 thomas goto done;
887 142012ed 2022-10-20 thomas /*
888 142012ed 2022-10-20 thomas * We cannot resolve this object yet because
889 142012ed 2022-10-20 thomas * a delta base is unknown. Try again later.
890 142012ed 2022-10-20 thomas */
891 142012ed 2022-10-20 thomas continue;
892 142012ed 2022-10-20 thomas }
893 142012ed 2022-10-20 thomas
894 142012ed 2022-10-20 thomas obj->valid = 1;
895 142012ed 2022-10-20 thomas n++;
896 142012ed 2022-10-20 thomas if (have_ref_deltas)
897 142012ed 2022-10-20 thomas update_packidx(&packidx, nobj, obj);
898 142012ed 2022-10-20 thomas /* Don't send too many progress privsep messages. */
899 142012ed 2022-10-20 thomas p_resolved = ((nresolved + n) * 100) / nobj;
900 142012ed 2022-10-20 thomas if (p_resolved != last_p_resolved) {
901 142012ed 2022-10-20 thomas err = report_progress(nobj, nobj,
902 aecd2225 2022-10-20 thomas nloose, nresolved + n, rl,
903 142012ed 2022-10-20 thomas progress_cb, progress_arg);
904 142012ed 2022-10-20 thomas if (err)
905 142012ed 2022-10-20 thomas goto done;
906 142012ed 2022-10-20 thomas last_p_resolved = p_resolved;
907 142012ed 2022-10-20 thomas }
908 142012ed 2022-10-20 thomas
909 142012ed 2022-10-20 thomas }
910 142012ed 2022-10-20 thomas if (pass++ > 3 && n == 0) {
911 142012ed 2022-10-20 thomas err = got_error_msg(GOT_ERR_BAD_PACKFILE,
912 142012ed 2022-10-20 thomas "could not resolve any of deltas; packfile could "
913 142012ed 2022-10-20 thomas "be corrupt");
914 142012ed 2022-10-20 thomas goto done;
915 142012ed 2022-10-20 thomas }
916 142012ed 2022-10-20 thomas nresolved += n;
917 88056de1 2023-01-19 thomas nvalid += n;
918 142012ed 2022-10-20 thomas }
919 142012ed 2022-10-20 thomas
920 142012ed 2022-10-20 thomas if (nloose + nresolved != nobj) {
921 142012ed 2022-10-20 thomas static char msg[64];
922 142012ed 2022-10-20 thomas snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
923 142012ed 2022-10-20 thomas nloose + nresolved, nobj);
924 142012ed 2022-10-20 thomas err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
925 142012ed 2022-10-20 thomas goto done;
926 142012ed 2022-10-20 thomas }
927 142012ed 2022-10-20 thomas
928 142012ed 2022-10-20 thomas err = report_progress(nobj, nobj, nloose, nresolved, NULL,
929 142012ed 2022-10-20 thomas progress_cb, progress_arg);
930 142012ed 2022-10-20 thomas if (err)
931 142012ed 2022-10-20 thomas goto done;
932 142012ed 2022-10-20 thomas
933 142012ed 2022-10-20 thomas make_packidx(&packidx, nobj, objects);
934 142012ed 2022-10-20 thomas
935 142012ed 2022-10-20 thomas free(objects);
936 142012ed 2022-10-20 thomas objects = NULL;
937 142012ed 2022-10-20 thomas
938 142012ed 2022-10-20 thomas SHA1Init(&ctx);
939 142012ed 2022-10-20 thomas putbe32(buf, GOT_PACKIDX_V2_MAGIC);
940 142012ed 2022-10-20 thomas putbe32(buf + 4, GOT_PACKIDX_VERSION);
941 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, buf, 8, &ctx);
942 142012ed 2022-10-20 thomas if (err)
943 142012ed 2022-10-20 thomas goto done;
944 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
945 142012ed 2022-10-20 thomas GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
946 142012ed 2022-10-20 thomas if (err)
947 142012ed 2022-10-20 thomas goto done;
948 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
949 142012ed 2022-10-20 thomas nobj * SHA1_DIGEST_LENGTH, &ctx);
950 142012ed 2022-10-20 thomas if (err)
951 142012ed 2022-10-20 thomas goto done;
952 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
953 3efd8e31 2022-10-23 thomas nobj * sizeof(uint32_t), &ctx);
954 142012ed 2022-10-20 thomas if (err)
955 142012ed 2022-10-20 thomas goto done;
956 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
957 3efd8e31 2022-10-23 thomas nobj * sizeof(uint32_t), &ctx);
958 142012ed 2022-10-20 thomas if (err)
959 142012ed 2022-10-20 thomas goto done;
960 142012ed 2022-10-20 thomas if (packidx.nlargeobj > 0) {
961 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
962 142012ed 2022-10-20 thomas packidx.nlargeobj * sizeof(uint64_t), &ctx);
963 142012ed 2022-10-20 thomas if (err)
964 142012ed 2022-10-20 thomas goto done;
965 142012ed 2022-10-20 thomas }
966 3efd8e31 2022-10-23 thomas err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
967 142012ed 2022-10-20 thomas if (err)
968 142012ed 2022-10-20 thomas goto done;
969 142012ed 2022-10-20 thomas
970 142012ed 2022-10-20 thomas SHA1Final(packidx_hash, &ctx);
971 142012ed 2022-10-20 thomas w = write(idxfd, packidx_hash, sizeof(packidx_hash));
972 142012ed 2022-10-20 thomas if (w == -1) {
973 142012ed 2022-10-20 thomas err = got_error_from_errno("write");
974 142012ed 2022-10-20 thomas goto done;
975 142012ed 2022-10-20 thomas }
976 142012ed 2022-10-20 thomas if (w != sizeof(packidx_hash)) {
977 142012ed 2022-10-20 thomas err = got_error(GOT_ERR_IO);
978 142012ed 2022-10-20 thomas goto done;
979 142012ed 2022-10-20 thomas }
980 142012ed 2022-10-20 thomas done:
981 142012ed 2022-10-20 thomas free(objects);
982 142012ed 2022-10-20 thomas free(packidx.hdr.magic);
983 142012ed 2022-10-20 thomas free(packidx.hdr.version);
984 142012ed 2022-10-20 thomas free(packidx.hdr.fanout_table);
985 142012ed 2022-10-20 thomas free(packidx.hdr.sorted_ids);
986 142012ed 2022-10-20 thomas free(packidx.hdr.offsets);
987 142012ed 2022-10-20 thomas free(packidx.hdr.large_offsets);
988 142012ed 2022-10-20 thomas return err;
989 142012ed 2022-10-20 thomas }