Blame


1 0136599f 2022-10-20 stsp /*
2 0136599f 2022-10-20 stsp * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3 0136599f 2022-10-20 stsp * Copyright (c) 2020, 2022 Stefan Sperling <stsp@openbsd.org>
4 0136599f 2022-10-20 stsp *
5 0136599f 2022-10-20 stsp * Permission to use, copy, modify, and distribute this software for any
6 0136599f 2022-10-20 stsp * purpose with or without fee is hereby granted, provided that the above
7 0136599f 2022-10-20 stsp * copyright notice and this permission notice appear in all copies.
8 0136599f 2022-10-20 stsp *
9 0136599f 2022-10-20 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 0136599f 2022-10-20 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 0136599f 2022-10-20 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 0136599f 2022-10-20 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 0136599f 2022-10-20 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 0136599f 2022-10-20 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 0136599f 2022-10-20 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 0136599f 2022-10-20 stsp */
17 0136599f 2022-10-20 stsp
18 0136599f 2022-10-20 stsp #include <sys/queue.h>
19 0136599f 2022-10-20 stsp #include <sys/stat.h>
20 0136599f 2022-10-20 stsp #include <sys/time.h>
21 0136599f 2022-10-20 stsp #include <sys/types.h>
22 0136599f 2022-10-20 stsp #include <sys/uio.h>
23 0136599f 2022-10-20 stsp #include <sys/mman.h>
24 0136599f 2022-10-20 stsp
25 0136599f 2022-10-20 stsp #include <stdint.h>
26 0136599f 2022-10-20 stsp #include <errno.h>
27 0136599f 2022-10-20 stsp #include <imsg.h>
28 0136599f 2022-10-20 stsp #include <limits.h>
29 0136599f 2022-10-20 stsp #include <signal.h>
30 0136599f 2022-10-20 stsp #include <stdio.h>
31 0136599f 2022-10-20 stsp #include <stdlib.h>
32 0136599f 2022-10-20 stsp #include <string.h>
33 0136599f 2022-10-20 stsp #include <ctype.h>
34 0136599f 2022-10-20 stsp #include <sha1.h>
35 69c6accf 2023-02-04 op #include <sha2.h>
36 0136599f 2022-10-20 stsp #include <endian.h>
37 0136599f 2022-10-20 stsp #include <fcntl.h>
38 0136599f 2022-10-20 stsp #include <unistd.h>
39 0136599f 2022-10-20 stsp #include <zlib.h>
40 0136599f 2022-10-20 stsp #include <err.h>
41 0136599f 2022-10-20 stsp #include <assert.h>
42 0136599f 2022-10-20 stsp #include <dirent.h>
43 0136599f 2022-10-20 stsp
44 0136599f 2022-10-20 stsp #include "got_error.h"
45 0136599f 2022-10-20 stsp #include "got_object.h"
46 0136599f 2022-10-20 stsp
47 1362b0e3 2023-02-04 op #include "got_lib_hash.h"
48 0136599f 2022-10-20 stsp #include "got_lib_delta.h"
49 0136599f 2022-10-20 stsp #include "got_lib_inflate.h"
50 0136599f 2022-10-20 stsp #include "got_lib_object.h"
51 0136599f 2022-10-20 stsp #include "got_lib_object_parse.h"
52 0136599f 2022-10-20 stsp #include "got_lib_object_idset.h"
53 0136599f 2022-10-20 stsp #include "got_lib_privsep.h"
54 0136599f 2022-10-20 stsp #include "got_lib_pack.h"
55 0136599f 2022-10-20 stsp #include "got_lib_ratelimit.h"
56 0136599f 2022-10-20 stsp #include "got_lib_pack_index.h"
57 0136599f 2022-10-20 stsp #include "got_lib_delta_cache.h"
58 0136599f 2022-10-20 stsp
59 0136599f 2022-10-20 stsp struct got_indexed_object {
60 0136599f 2022-10-20 stsp struct got_object_id id;
61 0136599f 2022-10-20 stsp
62 0136599f 2022-10-20 stsp /*
63 0136599f 2022-10-20 stsp * Has this object been fully resolved?
64 0136599f 2022-10-20 stsp * If so, we know its ID, otherwise we don't and 'id' is invalid.
65 0136599f 2022-10-20 stsp */
66 0136599f 2022-10-20 stsp int valid;
67 0136599f 2022-10-20 stsp
68 0136599f 2022-10-20 stsp /* Offset of type+size field for this object in pack file. */
69 0136599f 2022-10-20 stsp off_t off;
70 0136599f 2022-10-20 stsp
71 0136599f 2022-10-20 stsp /* Type+size values parsed from pack file. */
72 0136599f 2022-10-20 stsp uint8_t type;
73 0136599f 2022-10-20 stsp uint64_t size;
74 0136599f 2022-10-20 stsp
75 0136599f 2022-10-20 stsp /* Length of on-disk type+size data. */
76 0136599f 2022-10-20 stsp size_t tslen;
77 0136599f 2022-10-20 stsp
78 0136599f 2022-10-20 stsp /* Length of object data following type+size. */
79 0136599f 2022-10-20 stsp size_t len;
80 0136599f 2022-10-20 stsp
81 0136599f 2022-10-20 stsp uint32_t crc;
82 0136599f 2022-10-20 stsp
83 0136599f 2022-10-20 stsp union {
84 0136599f 2022-10-20 stsp struct {
85 0136599f 2022-10-20 stsp /* For ref deltas. */
86 0136599f 2022-10-20 stsp struct got_object_id ref_id;
87 0136599f 2022-10-20 stsp } ref;
88 0136599f 2022-10-20 stsp struct {
89 0136599f 2022-10-20 stsp /* For offset deltas. */
90 0136599f 2022-10-20 stsp off_t base_offset;
91 0136599f 2022-10-20 stsp size_t base_offsetlen;
92 0136599f 2022-10-20 stsp } ofs;
93 0136599f 2022-10-20 stsp } delta;
94 0136599f 2022-10-20 stsp };
95 0136599f 2022-10-20 stsp
96 0136599f 2022-10-20 stsp static void
97 0136599f 2022-10-20 stsp putbe32(char *b, uint32_t n)
98 0136599f 2022-10-20 stsp {
99 0136599f 2022-10-20 stsp b[0] = n >> 24;
100 0136599f 2022-10-20 stsp b[1] = n >> 16;
101 0136599f 2022-10-20 stsp b[2] = n >> 8;
102 0136599f 2022-10-20 stsp b[3] = n >> 0;
103 0136599f 2022-10-20 stsp }
104 0136599f 2022-10-20 stsp
105 0136599f 2022-10-20 stsp static const struct got_error *
106 0136599f 2022-10-20 stsp get_obj_type_label(const char **label, int obj_type)
107 0136599f 2022-10-20 stsp {
108 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
109 0136599f 2022-10-20 stsp
110 0136599f 2022-10-20 stsp switch (obj_type) {
111 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_BLOB:
112 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_BLOB;
113 0136599f 2022-10-20 stsp break;
114 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TREE:
115 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_TREE;
116 0136599f 2022-10-20 stsp break;
117 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_COMMIT:
118 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_COMMIT;
119 0136599f 2022-10-20 stsp break;
120 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TAG:
121 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_TAG;
122 0136599f 2022-10-20 stsp break;
123 0136599f 2022-10-20 stsp default:
124 0136599f 2022-10-20 stsp *label = NULL;
125 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_OBJ_TYPE);
126 0136599f 2022-10-20 stsp break;
127 0136599f 2022-10-20 stsp }
128 0136599f 2022-10-20 stsp
129 0136599f 2022-10-20 stsp return err;
130 0136599f 2022-10-20 stsp }
131 0136599f 2022-10-20 stsp
132 0136599f 2022-10-20 stsp static const struct got_error *
133 3093e2d7 2023-02-04 op read_checksum(uint32_t *crc, SHA1_CTX *hash_ctx, int fd, size_t len)
134 0136599f 2022-10-20 stsp {
135 0136599f 2022-10-20 stsp uint8_t buf[8192];
136 0136599f 2022-10-20 stsp size_t n;
137 0136599f 2022-10-20 stsp ssize_t r;
138 0136599f 2022-10-20 stsp
139 0136599f 2022-10-20 stsp for (n = len; n > 0; n -= r){
140 0136599f 2022-10-20 stsp r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
141 0136599f 2022-10-20 stsp if (r == -1)
142 0136599f 2022-10-20 stsp return got_error_from_errno("read");
143 0136599f 2022-10-20 stsp if (r == 0)
144 0136599f 2022-10-20 stsp break;
145 0136599f 2022-10-20 stsp if (crc)
146 0136599f 2022-10-20 stsp *crc = crc32(*crc, buf, r);
147 3093e2d7 2023-02-04 op if (hash_ctx)
148 3093e2d7 2023-02-04 op SHA1Update(hash_ctx, buf, r);
149 0136599f 2022-10-20 stsp }
150 0136599f 2022-10-20 stsp
151 0136599f 2022-10-20 stsp return NULL;
152 0136599f 2022-10-20 stsp }
153 0136599f 2022-10-20 stsp
154 0136599f 2022-10-20 stsp static const struct got_error *
155 3093e2d7 2023-02-04 op read_file_hash(SHA1_CTX *ctx, FILE *f, size_t len)
156 0136599f 2022-10-20 stsp {
157 0136599f 2022-10-20 stsp uint8_t buf[8192];
158 0136599f 2022-10-20 stsp size_t n, r;
159 0136599f 2022-10-20 stsp
160 0136599f 2022-10-20 stsp for (n = len; n > 0; n -= r) {
161 0136599f 2022-10-20 stsp r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
162 0136599f 2022-10-20 stsp if (r == 0) {
163 0136599f 2022-10-20 stsp if (feof(f))
164 0136599f 2022-10-20 stsp return NULL;
165 0136599f 2022-10-20 stsp return got_ferror(f, GOT_ERR_IO);
166 0136599f 2022-10-20 stsp }
167 0136599f 2022-10-20 stsp SHA1Update(ctx, buf, r);
168 0136599f 2022-10-20 stsp }
169 0136599f 2022-10-20 stsp
170 0136599f 2022-10-20 stsp return NULL;
171 0136599f 2022-10-20 stsp }
172 0136599f 2022-10-20 stsp
173 0136599f 2022-10-20 stsp static const struct got_error *
174 0136599f 2022-10-20 stsp read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
175 0136599f 2022-10-20 stsp FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
176 0136599f 2022-10-20 stsp {
177 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
178 0136599f 2022-10-20 stsp SHA1_CTX ctx;
179 0136599f 2022-10-20 stsp uint8_t *data = NULL;
180 0136599f 2022-10-20 stsp size_t datalen = 0;
181 0136599f 2022-10-20 stsp ssize_t n;
182 0136599f 2022-10-20 stsp char *header;
183 0136599f 2022-10-20 stsp size_t headerlen;
184 0136599f 2022-10-20 stsp const char *obj_label;
185 0136599f 2022-10-20 stsp size_t mapoff = obj->off;
186 0136599f 2022-10-20 stsp struct got_inflate_checksum csum;
187 0136599f 2022-10-20 stsp
188 0136599f 2022-10-20 stsp memset(&csum, 0, sizeof(csum));
189 0136599f 2022-10-20 stsp csum.input_sha1 = pack_sha1_ctx;
190 0136599f 2022-10-20 stsp csum.input_crc = &obj->crc;
191 0136599f 2022-10-20 stsp
192 0136599f 2022-10-20 stsp err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
193 0136599f 2022-10-20 stsp &obj->tslen, pack, obj->off);
194 0136599f 2022-10-20 stsp if (err)
195 0136599f 2022-10-20 stsp return err;
196 0136599f 2022-10-20 stsp
197 0136599f 2022-10-20 stsp if (pack->map) {
198 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
199 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
200 0136599f 2022-10-20 stsp mapoff += obj->tslen;
201 0136599f 2022-10-20 stsp } else {
202 0136599f 2022-10-20 stsp /* XXX Seek back and get the CRC of on-disk type+size bytes. */
203 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
204 0136599f 2022-10-20 stsp return got_error_from_errno("lseek");
205 0136599f 2022-10-20 stsp err = read_checksum(&obj->crc, pack_sha1_ctx,
206 0136599f 2022-10-20 stsp pack->fd, obj->tslen);
207 0136599f 2022-10-20 stsp if (err)
208 0136599f 2022-10-20 stsp return err;
209 0136599f 2022-10-20 stsp }
210 0136599f 2022-10-20 stsp
211 0136599f 2022-10-20 stsp switch (obj->type) {
212 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_BLOB:
213 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_COMMIT:
214 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TREE:
215 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TAG:
216 0136599f 2022-10-20 stsp if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
217 0136599f 2022-10-20 stsp if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
218 0136599f 2022-10-20 stsp err = got_error_from_errno("fseek");
219 0136599f 2022-10-20 stsp break;
220 0136599f 2022-10-20 stsp }
221 0136599f 2022-10-20 stsp if (pack->map) {
222 0136599f 2022-10-20 stsp err = got_inflate_to_file_mmap(&datalen,
223 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
224 0136599f 2022-10-20 stsp pack->filesize - mapoff, tmpfile);
225 0136599f 2022-10-20 stsp } else {
226 0136599f 2022-10-20 stsp err = got_inflate_to_file_fd(&datalen,
227 0136599f 2022-10-20 stsp &obj->len, &csum, pack->fd, tmpfile);
228 0136599f 2022-10-20 stsp }
229 0136599f 2022-10-20 stsp } else {
230 0136599f 2022-10-20 stsp if (pack->map) {
231 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(&data, &datalen,
232 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
233 0136599f 2022-10-20 stsp pack->filesize - mapoff);
234 0136599f 2022-10-20 stsp } else {
235 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(&data, &datalen,
236 0136599f 2022-10-20 stsp &obj->len, &csum, obj->size, pack->fd);
237 0136599f 2022-10-20 stsp }
238 0136599f 2022-10-20 stsp }
239 0136599f 2022-10-20 stsp if (err)
240 0136599f 2022-10-20 stsp break;
241 0136599f 2022-10-20 stsp SHA1Init(&ctx);
242 0136599f 2022-10-20 stsp err = get_obj_type_label(&obj_label, obj->type);
243 0136599f 2022-10-20 stsp if (err) {
244 0136599f 2022-10-20 stsp free(data);
245 0136599f 2022-10-20 stsp break;
246 0136599f 2022-10-20 stsp }
247 0136599f 2022-10-20 stsp if (asprintf(&header, "%s %lld", obj_label,
248 0136599f 2022-10-20 stsp (long long)obj->size) == -1) {
249 0136599f 2022-10-20 stsp err = got_error_from_errno("asprintf");
250 0136599f 2022-10-20 stsp free(data);
251 0136599f 2022-10-20 stsp break;
252 0136599f 2022-10-20 stsp }
253 0136599f 2022-10-20 stsp headerlen = strlen(header) + 1;
254 0136599f 2022-10-20 stsp SHA1Update(&ctx, header, headerlen);
255 0136599f 2022-10-20 stsp if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
256 3093e2d7 2023-02-04 op err = read_file_hash(&ctx, tmpfile, datalen);
257 0136599f 2022-10-20 stsp if (err) {
258 0136599f 2022-10-20 stsp free(header);
259 0136599f 2022-10-20 stsp free(data);
260 0136599f 2022-10-20 stsp break;
261 0136599f 2022-10-20 stsp }
262 0136599f 2022-10-20 stsp } else
263 0136599f 2022-10-20 stsp SHA1Update(&ctx, data, datalen);
264 3093e2d7 2023-02-04 op SHA1Final(obj->id.hash, &ctx);
265 0136599f 2022-10-20 stsp free(header);
266 0136599f 2022-10-20 stsp free(data);
267 0136599f 2022-10-20 stsp break;
268 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_REF_DELTA:
269 3093e2d7 2023-02-04 op memset(obj->id.hash, 0xff, SHA1_DIGEST_LENGTH);
270 0136599f 2022-10-20 stsp if (pack->map) {
271 0136599f 2022-10-20 stsp if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
272 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
273 ad4cc361 2022-10-27 op break;
274 ad4cc361 2022-10-27 op }
275 ad4cc361 2022-10-27 op if (mapoff + SHA1_DIGEST_LENGTH > SIZE_MAX) {
276 ad4cc361 2022-10-27 op err = got_error_fmt(GOT_ERR_RANGE,
277 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
278 ad4cc361 2022-10-27 op (long long)mapoff + SHA1_DIGEST_LENGTH);
279 0136599f 2022-10-20 stsp break;
280 0136599f 2022-10-20 stsp }
281 3093e2d7 2023-02-04 op memcpy(obj->delta.ref.ref_id.hash, pack->map + mapoff,
282 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
283 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff,
284 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
285 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff,
286 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
287 0136599f 2022-10-20 stsp mapoff += SHA1_DIGEST_LENGTH;
288 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(NULL, &datalen,
289 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
290 0136599f 2022-10-20 stsp pack->filesize - mapoff);
291 0136599f 2022-10-20 stsp if (err)
292 0136599f 2022-10-20 stsp break;
293 0136599f 2022-10-20 stsp } else {
294 3093e2d7 2023-02-04 op n = read(pack->fd, obj->delta.ref.ref_id.hash,
295 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
296 0136599f 2022-10-20 stsp if (n == -1) {
297 0136599f 2022-10-20 stsp err = got_error_from_errno("read");
298 0136599f 2022-10-20 stsp break;
299 0136599f 2022-10-20 stsp }
300 0136599f 2022-10-20 stsp if (n < sizeof(obj->id)) {
301 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
302 0136599f 2022-10-20 stsp break;
303 0136599f 2022-10-20 stsp }
304 3093e2d7 2023-02-04 op obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.hash,
305 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
306 3093e2d7 2023-02-04 op SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.hash,
307 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
308 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
309 0136599f 2022-10-20 stsp &csum, obj->size, pack->fd);
310 0136599f 2022-10-20 stsp if (err)
311 0136599f 2022-10-20 stsp break;
312 0136599f 2022-10-20 stsp }
313 0136599f 2022-10-20 stsp obj->len += SHA1_DIGEST_LENGTH;
314 0136599f 2022-10-20 stsp break;
315 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
316 3093e2d7 2023-02-04 op memset(obj->id.hash, 0xff, SHA1_DIGEST_LENGTH);
317 0136599f 2022-10-20 stsp err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
318 0136599f 2022-10-20 stsp &obj->delta.ofs.base_offsetlen, pack, obj->off,
319 0136599f 2022-10-20 stsp obj->tslen);
320 0136599f 2022-10-20 stsp if (err)
321 0136599f 2022-10-20 stsp break;
322 0136599f 2022-10-20 stsp
323 0136599f 2022-10-20 stsp if (pack->map) {
324 5326bd73 2022-10-24 op if (mapoff + obj->delta.ofs.base_offsetlen >=
325 5326bd73 2022-10-24 op pack->filesize) {
326 5326bd73 2022-10-24 op err = got_error(GOT_ERR_BAD_PACKFILE);
327 5326bd73 2022-10-24 op break;
328 5326bd73 2022-10-24 op }
329 5326bd73 2022-10-24 op
330 ad4cc361 2022-10-27 op if (mapoff + obj->delta.ofs.base_offsetlen >
331 ad4cc361 2022-10-27 op SIZE_MAX) {
332 ad4cc361 2022-10-27 op err = got_error_fmt(GOT_ERR_RANGE,
333 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
334 ad4cc361 2022-10-27 op (long long)mapoff
335 ad4cc361 2022-10-27 op + obj->delta.ofs.base_offsetlen);
336 ad4cc361 2022-10-27 op }
337 ad4cc361 2022-10-27 op
338 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff,
339 0136599f 2022-10-20 stsp obj->delta.ofs.base_offsetlen);
340 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff,
341 0136599f 2022-10-20 stsp obj->delta.ofs.base_offsetlen);
342 0136599f 2022-10-20 stsp mapoff += obj->delta.ofs.base_offsetlen;
343 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(NULL, &datalen,
344 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
345 0136599f 2022-10-20 stsp pack->filesize - mapoff);
346 0136599f 2022-10-20 stsp if (err)
347 0136599f 2022-10-20 stsp break;
348 0136599f 2022-10-20 stsp } else {
349 0136599f 2022-10-20 stsp /*
350 0136599f 2022-10-20 stsp * XXX Seek back and get CRC and SHA1 of on-disk
351 0136599f 2022-10-20 stsp * offset bytes.
352 0136599f 2022-10-20 stsp */
353 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
354 0136599f 2022-10-20 stsp == -1) {
355 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
356 0136599f 2022-10-20 stsp break;
357 0136599f 2022-10-20 stsp }
358 0136599f 2022-10-20 stsp err = read_checksum(&obj->crc, pack_sha1_ctx,
359 0136599f 2022-10-20 stsp pack->fd, obj->delta.ofs.base_offsetlen);
360 0136599f 2022-10-20 stsp if (err)
361 0136599f 2022-10-20 stsp break;
362 0136599f 2022-10-20 stsp
363 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
364 0136599f 2022-10-20 stsp &csum, obj->size, pack->fd);
365 0136599f 2022-10-20 stsp if (err)
366 0136599f 2022-10-20 stsp break;
367 0136599f 2022-10-20 stsp }
368 0136599f 2022-10-20 stsp obj->len += obj->delta.ofs.base_offsetlen;
369 0136599f 2022-10-20 stsp break;
370 0136599f 2022-10-20 stsp default:
371 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_OBJ_TYPE);
372 0136599f 2022-10-20 stsp break;
373 0136599f 2022-10-20 stsp }
374 0136599f 2022-10-20 stsp
375 0136599f 2022-10-20 stsp return err;
376 0136599f 2022-10-20 stsp }
377 0136599f 2022-10-20 stsp
378 13b2bc37 2022-10-23 stsp const struct got_error *
379 13b2bc37 2022-10-23 stsp got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
380 0136599f 2022-10-20 stsp {
381 0136599f 2022-10-20 stsp ssize_t w;
382 0136599f 2022-10-20 stsp
383 0136599f 2022-10-20 stsp SHA1Update(ctx, buf, len);
384 0136599f 2022-10-20 stsp
385 0136599f 2022-10-20 stsp w = write(fd, buf, len);
386 0136599f 2022-10-20 stsp if (w == -1)
387 0136599f 2022-10-20 stsp return got_error_from_errno("write");
388 0136599f 2022-10-20 stsp if (w != len)
389 0136599f 2022-10-20 stsp return got_error(GOT_ERR_IO);
390 0136599f 2022-10-20 stsp
391 0136599f 2022-10-20 stsp return NULL;
392 0136599f 2022-10-20 stsp }
393 0136599f 2022-10-20 stsp
394 0136599f 2022-10-20 stsp static const struct got_error *
395 0136599f 2022-10-20 stsp resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
396 0136599f 2022-10-20 stsp struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
397 0136599f 2022-10-20 stsp FILE *delta_accum_file)
398 0136599f 2022-10-20 stsp {
399 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
400 0136599f 2022-10-20 stsp struct got_delta_chain deltas;
401 0136599f 2022-10-20 stsp struct got_delta *delta;
402 0136599f 2022-10-20 stsp uint8_t *buf = NULL;
403 0136599f 2022-10-20 stsp size_t len = 0;
404 0136599f 2022-10-20 stsp SHA1_CTX ctx;
405 0136599f 2022-10-20 stsp char *header = NULL;
406 0136599f 2022-10-20 stsp size_t headerlen;
407 0136599f 2022-10-20 stsp uint64_t max_size;
408 0136599f 2022-10-20 stsp int base_obj_type;
409 0136599f 2022-10-20 stsp const char *obj_label;
410 0136599f 2022-10-20 stsp
411 0136599f 2022-10-20 stsp deltas.nentries = 0;
412 0136599f 2022-10-20 stsp STAILQ_INIT(&deltas.entries);
413 0136599f 2022-10-20 stsp
414 0136599f 2022-10-20 stsp err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
415 0136599f 2022-10-20 stsp obj->off, obj->tslen, obj->type, obj->size,
416 0136599f 2022-10-20 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
417 0136599f 2022-10-20 stsp if (err)
418 0136599f 2022-10-20 stsp goto done;
419 0136599f 2022-10-20 stsp
420 0136599f 2022-10-20 stsp err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
421 0136599f 2022-10-20 stsp if (err)
422 0136599f 2022-10-20 stsp goto done;
423 0136599f 2022-10-20 stsp if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
424 0136599f 2022-10-20 stsp rewind(tmpfile);
425 0136599f 2022-10-20 stsp rewind(delta_base_file);
426 0136599f 2022-10-20 stsp rewind(delta_accum_file);
427 0136599f 2022-10-20 stsp err = got_pack_dump_delta_chain_to_file(&len, &deltas,
428 0136599f 2022-10-20 stsp pack, tmpfile, delta_base_file, delta_accum_file);
429 0136599f 2022-10-20 stsp if (err)
430 0136599f 2022-10-20 stsp goto done;
431 0136599f 2022-10-20 stsp } else {
432 0136599f 2022-10-20 stsp err = got_pack_dump_delta_chain_to_mem(&buf, &len,
433 0136599f 2022-10-20 stsp &deltas, pack);
434 0136599f 2022-10-20 stsp }
435 0136599f 2022-10-20 stsp if (err)
436 0136599f 2022-10-20 stsp goto done;
437 0136599f 2022-10-20 stsp
438 0136599f 2022-10-20 stsp err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
439 0136599f 2022-10-20 stsp if (err)
440 0136599f 2022-10-20 stsp goto done;
441 0136599f 2022-10-20 stsp err = get_obj_type_label(&obj_label, base_obj_type);
442 0136599f 2022-10-20 stsp if (err)
443 0136599f 2022-10-20 stsp goto done;
444 0136599f 2022-10-20 stsp if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
445 0136599f 2022-10-20 stsp err = got_error_from_errno("asprintf");
446 0136599f 2022-10-20 stsp goto done;
447 0136599f 2022-10-20 stsp }
448 0136599f 2022-10-20 stsp headerlen = strlen(header) + 1;
449 0136599f 2022-10-20 stsp SHA1Init(&ctx);
450 0136599f 2022-10-20 stsp SHA1Update(&ctx, header, headerlen);
451 0136599f 2022-10-20 stsp if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
452 3093e2d7 2023-02-04 op err = read_file_hash(&ctx, tmpfile, len);
453 0136599f 2022-10-20 stsp if (err)
454 0136599f 2022-10-20 stsp goto done;
455 0136599f 2022-10-20 stsp } else
456 0136599f 2022-10-20 stsp SHA1Update(&ctx, buf, len);
457 3093e2d7 2023-02-04 op SHA1Final(obj->id.hash, &ctx);
458 0136599f 2022-10-20 stsp done:
459 0136599f 2022-10-20 stsp free(buf);
460 0136599f 2022-10-20 stsp free(header);
461 0136599f 2022-10-20 stsp while (!STAILQ_EMPTY(&deltas.entries)) {
462 0136599f 2022-10-20 stsp delta = STAILQ_FIRST(&deltas.entries);
463 0136599f 2022-10-20 stsp STAILQ_REMOVE_HEAD(&deltas.entries, entry);
464 0136599f 2022-10-20 stsp free(delta);
465 0136599f 2022-10-20 stsp }
466 0136599f 2022-10-20 stsp return err;
467 0136599f 2022-10-20 stsp }
468 0136599f 2022-10-20 stsp
469 0136599f 2022-10-20 stsp /* Determine the slot in the pack index a given object ID should use. */
470 0136599f 2022-10-20 stsp static int
471 3093e2d7 2023-02-04 op find_object_idx(struct got_packidx *packidx, uint8_t *hash)
472 0136599f 2022-10-20 stsp {
473 3093e2d7 2023-02-04 op u_int8_t id0 = hash[0];
474 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
475 0136599f 2022-10-20 stsp int left = 0, right = nindexed - 1;
476 0136599f 2022-10-20 stsp int cmp = 0, i = 0;
477 0136599f 2022-10-20 stsp
478 0136599f 2022-10-20 stsp if (id0 > 0)
479 0136599f 2022-10-20 stsp left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
480 0136599f 2022-10-20 stsp
481 0136599f 2022-10-20 stsp while (left <= right) {
482 0136599f 2022-10-20 stsp struct got_packidx_object_id *oid;
483 0136599f 2022-10-20 stsp
484 0136599f 2022-10-20 stsp i = ((left + right) / 2);
485 0136599f 2022-10-20 stsp oid = &packidx->hdr.sorted_ids[i];
486 0136599f 2022-10-20 stsp
487 3093e2d7 2023-02-04 op cmp = memcmp(hash, oid->hash, SHA1_DIGEST_LENGTH);
488 0136599f 2022-10-20 stsp if (cmp == 0)
489 0136599f 2022-10-20 stsp return -1; /* object already indexed */
490 0136599f 2022-10-20 stsp else if (cmp > 0)
491 0136599f 2022-10-20 stsp left = i + 1;
492 0136599f 2022-10-20 stsp else if (cmp < 0)
493 0136599f 2022-10-20 stsp right = i - 1;
494 0136599f 2022-10-20 stsp }
495 0136599f 2022-10-20 stsp
496 0136599f 2022-10-20 stsp return left;
497 0136599f 2022-10-20 stsp }
498 0136599f 2022-10-20 stsp
499 0136599f 2022-10-20 stsp #if 0
500 0136599f 2022-10-20 stsp static void
501 0136599f 2022-10-20 stsp print_packidx(struct got_packidx *packidx)
502 0136599f 2022-10-20 stsp {
503 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
504 0136599f 2022-10-20 stsp int i;
505 0136599f 2022-10-20 stsp
506 0136599f 2022-10-20 stsp fprintf(stderr, "object IDs:\n");
507 0136599f 2022-10-20 stsp for (i = 0; i < nindexed; i++) {
508 0136599f 2022-10-20 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
509 3093e2d7 2023-02-04 op got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].hash,
510 0136599f 2022-10-20 stsp hex, sizeof(hex));
511 0136599f 2022-10-20 stsp fprintf(stderr, "%s\n", hex);
512 0136599f 2022-10-20 stsp }
513 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
514 0136599f 2022-10-20 stsp
515 0136599f 2022-10-20 stsp fprintf(stderr, "object offsets:\n");
516 0136599f 2022-10-20 stsp for (i = 0; i < nindexed; i++) {
517 0136599f 2022-10-20 stsp uint32_t offset = be32toh(packidx->hdr.offsets[i]);
518 0136599f 2022-10-20 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
519 0136599f 2022-10-20 stsp int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
520 0136599f 2022-10-20 stsp fprintf(stderr, "%u -> %llu\n", offset,
521 0136599f 2022-10-20 stsp be64toh(packidx->hdr.large_offsets[j]));
522 0136599f 2022-10-20 stsp } else
523 0136599f 2022-10-20 stsp fprintf(stderr, "%u\n", offset);
524 0136599f 2022-10-20 stsp }
525 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
526 0136599f 2022-10-20 stsp
527 0136599f 2022-10-20 stsp fprintf(stderr, "fanout table:");
528 0136599f 2022-10-20 stsp for (i = 0; i <= 0xff; i++)
529 0136599f 2022-10-20 stsp fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
530 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
531 0136599f 2022-10-20 stsp }
532 0136599f 2022-10-20 stsp #endif
533 0136599f 2022-10-20 stsp
534 0136599f 2022-10-20 stsp static void
535 0136599f 2022-10-20 stsp add_indexed_object(struct got_packidx *packidx, uint32_t idx,
536 0136599f 2022-10-20 stsp struct got_indexed_object *obj)
537 0136599f 2022-10-20 stsp {
538 0136599f 2022-10-20 stsp int i;
539 0136599f 2022-10-20 stsp
540 3093e2d7 2023-02-04 op memcpy(packidx->hdr.sorted_ids[idx].hash, obj->id.hash,
541 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
542 0136599f 2022-10-20 stsp packidx->hdr.crc32[idx] = htobe32(obj->crc);
543 0136599f 2022-10-20 stsp if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
544 0136599f 2022-10-20 stsp packidx->hdr.offsets[idx] = htobe32(obj->off);
545 0136599f 2022-10-20 stsp else {
546 0136599f 2022-10-20 stsp packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
547 0136599f 2022-10-20 stsp GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
548 0136599f 2022-10-20 stsp packidx->hdr.large_offsets[packidx->nlargeobj] =
549 0136599f 2022-10-20 stsp htobe64(obj->off);
550 0136599f 2022-10-20 stsp packidx->nlargeobj++;
551 0136599f 2022-10-20 stsp }
552 0136599f 2022-10-20 stsp
553 3093e2d7 2023-02-04 op for (i = obj->id.hash[0]; i <= 0xff; i++) {
554 0136599f 2022-10-20 stsp uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
555 0136599f 2022-10-20 stsp packidx->hdr.fanout_table[i] = htobe32(n + 1);
556 0136599f 2022-10-20 stsp }
557 0136599f 2022-10-20 stsp }
558 0136599f 2022-10-20 stsp
559 0136599f 2022-10-20 stsp static int
560 0136599f 2022-10-20 stsp indexed_obj_cmp(const void *pa, const void *pb)
561 0136599f 2022-10-20 stsp {
562 0136599f 2022-10-20 stsp struct got_indexed_object *a, *b;
563 0136599f 2022-10-20 stsp
564 0136599f 2022-10-20 stsp a = (struct got_indexed_object *)pa;
565 0136599f 2022-10-20 stsp b = (struct got_indexed_object *)pb;
566 0136599f 2022-10-20 stsp return got_object_id_cmp(&a->id, &b->id);
567 0136599f 2022-10-20 stsp }
568 0136599f 2022-10-20 stsp
569 0136599f 2022-10-20 stsp static void
570 9316cc27 2022-10-20 stsp make_packidx(struct got_packidx *packidx, uint32_t nobj,
571 0136599f 2022-10-20 stsp struct got_indexed_object *objects)
572 0136599f 2022-10-20 stsp {
573 0136599f 2022-10-20 stsp struct got_indexed_object *obj;
574 0136599f 2022-10-20 stsp int i;
575 0136599f 2022-10-20 stsp uint32_t idx = 0;
576 0136599f 2022-10-20 stsp
577 0136599f 2022-10-20 stsp qsort(objects, nobj, sizeof(struct got_indexed_object),
578 0136599f 2022-10-20 stsp indexed_obj_cmp);
579 0136599f 2022-10-20 stsp
580 0136599f 2022-10-20 stsp memset(packidx->hdr.fanout_table, 0,
581 0136599f 2022-10-20 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
582 0136599f 2022-10-20 stsp packidx->nlargeobj = 0;
583 0136599f 2022-10-20 stsp
584 0136599f 2022-10-20 stsp for (i = 0; i < nobj; i++) {
585 0136599f 2022-10-20 stsp obj = &objects[i];
586 0136599f 2022-10-20 stsp if (obj->valid)
587 0136599f 2022-10-20 stsp add_indexed_object(packidx, idx++, obj);
588 0136599f 2022-10-20 stsp }
589 0136599f 2022-10-20 stsp }
590 0136599f 2022-10-20 stsp
591 0136599f 2022-10-20 stsp static void
592 9316cc27 2022-10-20 stsp update_packidx(struct got_packidx *packidx, uint32_t nobj,
593 0136599f 2022-10-20 stsp struct got_indexed_object *obj)
594 0136599f 2022-10-20 stsp {
595 0136599f 2022-10-20 stsp int idx;
596 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
597 0136599f 2022-10-20 stsp
598 0205a93b 2023-02-04 op idx = find_object_idx(packidx, obj->id.hash);
599 09cbb981 2023-02-03 op if (idx == -1)
600 0136599f 2022-10-20 stsp return; /* object already indexed */
601 0136599f 2022-10-20 stsp
602 0136599f 2022-10-20 stsp memmove(&packidx->hdr.sorted_ids[idx + 1],
603 0136599f 2022-10-20 stsp &packidx->hdr.sorted_ids[idx],
604 0136599f 2022-10-20 stsp sizeof(struct got_packidx_object_id) * (nindexed - idx));
605 0136599f 2022-10-20 stsp memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
606 0136599f 2022-10-20 stsp sizeof(uint32_t) * (nindexed - idx));
607 0136599f 2022-10-20 stsp
608 0136599f 2022-10-20 stsp add_indexed_object(packidx, idx, obj);
609 0136599f 2022-10-20 stsp }
610 0136599f 2022-10-20 stsp
611 0136599f 2022-10-20 stsp static const struct got_error *
612 9316cc27 2022-10-20 stsp report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
613 9316cc27 2022-10-20 stsp uint32_t nobj_resolved, struct got_ratelimit *rl,
614 0136599f 2022-10-20 stsp got_pack_index_progress_cb progress_cb, void *progress_arg)
615 0136599f 2022-10-20 stsp {
616 0136599f 2022-10-20 stsp const struct got_error *err;
617 0136599f 2022-10-20 stsp int elapsed = 0;
618 0136599f 2022-10-20 stsp
619 0136599f 2022-10-20 stsp if (rl) {
620 0136599f 2022-10-20 stsp err = got_ratelimit_check(&elapsed, rl);
621 0136599f 2022-10-20 stsp if (err || !elapsed)
622 0136599f 2022-10-20 stsp return err;
623 0136599f 2022-10-20 stsp }
624 0136599f 2022-10-20 stsp
625 0136599f 2022-10-20 stsp return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
626 0136599f 2022-10-20 stsp nobj_resolved);
627 0136599f 2022-10-20 stsp }
628 0136599f 2022-10-20 stsp
629 0136599f 2022-10-20 stsp const struct got_error *
630 0136599f 2022-10-20 stsp got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
631 3093e2d7 2023-02-04 op FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_hash_expected,
632 713d6e11 2022-10-20 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
633 713d6e11 2022-10-20 stsp struct got_ratelimit *rl)
634 0136599f 2022-10-20 stsp {
635 0136599f 2022-10-20 stsp const struct got_error *err;
636 0136599f 2022-10-20 stsp struct got_packfile_hdr hdr;
637 0136599f 2022-10-20 stsp struct got_packidx packidx;
638 0136599f 2022-10-20 stsp char buf[8];
639 3093e2d7 2023-02-04 op char pack_hash[SHA1_DIGEST_LENGTH];
640 9316cc27 2022-10-20 stsp uint32_t nobj, nvalid, nloose, nresolved = 0, i;
641 0136599f 2022-10-20 stsp struct got_indexed_object *objects = NULL, *obj;
642 0136599f 2022-10-20 stsp SHA1_CTX ctx;
643 0136599f 2022-10-20 stsp uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
644 0136599f 2022-10-20 stsp ssize_t r, w;
645 0136599f 2022-10-20 stsp int pass, have_ref_deltas = 0, first_delta_idx = -1;
646 0136599f 2022-10-20 stsp size_t mapoff = 0;
647 0136599f 2022-10-20 stsp int p_indexed = 0, last_p_indexed = -1;
648 0136599f 2022-10-20 stsp int p_resolved = 0, last_p_resolved = -1;
649 0136599f 2022-10-20 stsp
650 0136599f 2022-10-20 stsp /* Require that pack file header and SHA1 trailer are present. */
651 0136599f 2022-10-20 stsp if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
652 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
653 0136599f 2022-10-20 stsp "short pack file");
654 0136599f 2022-10-20 stsp
655 0136599f 2022-10-20 stsp if (pack->map) {
656 0136599f 2022-10-20 stsp memcpy(&hdr, pack->map, sizeof(hdr));
657 0136599f 2022-10-20 stsp mapoff += sizeof(hdr);
658 0136599f 2022-10-20 stsp } else {
659 0136599f 2022-10-20 stsp r = read(pack->fd, &hdr, sizeof(hdr));
660 0136599f 2022-10-20 stsp if (r == -1)
661 0136599f 2022-10-20 stsp return got_error_from_errno("read");
662 0136599f 2022-10-20 stsp if (r < sizeof(hdr))
663 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
664 0136599f 2022-10-20 stsp "short pack file");
665 0136599f 2022-10-20 stsp }
666 0136599f 2022-10-20 stsp
667 0136599f 2022-10-20 stsp if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
668 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
669 0136599f 2022-10-20 stsp "bad packfile signature");
670 0136599f 2022-10-20 stsp if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
671 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
672 0136599f 2022-10-20 stsp "bad packfile version");
673 0136599f 2022-10-20 stsp nobj = be32toh(hdr.nobjects);
674 0136599f 2022-10-20 stsp if (nobj == 0)
675 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
676 0136599f 2022-10-20 stsp "bad packfile with zero objects");
677 0136599f 2022-10-20 stsp
678 0136599f 2022-10-20 stsp /* We compute the SHA1 of pack file contents and verify later on. */
679 0136599f 2022-10-20 stsp SHA1Init(&ctx);
680 0136599f 2022-10-20 stsp SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
681 0136599f 2022-10-20 stsp
682 0136599f 2022-10-20 stsp /*
683 0136599f 2022-10-20 stsp * Create an in-memory pack index which will grow as objects
684 0136599f 2022-10-20 stsp * IDs in the pack file are discovered. Only fields used to
685 0136599f 2022-10-20 stsp * read deltified objects will be needed by the pack.c library
686 0136599f 2022-10-20 stsp * code, so setting up just a pack index header is sufficient.
687 0136599f 2022-10-20 stsp */
688 0136599f 2022-10-20 stsp memset(&packidx, 0, sizeof(packidx));
689 0136599f 2022-10-20 stsp packidx.hdr.magic = malloc(sizeof(uint32_t));
690 0136599f 2022-10-20 stsp if (packidx.hdr.magic == NULL)
691 0136599f 2022-10-20 stsp return got_error_from_errno("malloc");
692 0136599f 2022-10-20 stsp *packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
693 0136599f 2022-10-20 stsp packidx.hdr.version = malloc(sizeof(uint32_t));
694 0136599f 2022-10-20 stsp if (packidx.hdr.version == NULL) {
695 0136599f 2022-10-20 stsp err = got_error_from_errno("malloc");
696 0136599f 2022-10-20 stsp goto done;
697 0136599f 2022-10-20 stsp }
698 0136599f 2022-10-20 stsp *packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
699 0136599f 2022-10-20 stsp packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
700 0136599f 2022-10-20 stsp sizeof(uint32_t));
701 0136599f 2022-10-20 stsp if (packidx.hdr.fanout_table == NULL) {
702 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
703 0136599f 2022-10-20 stsp goto done;
704 0136599f 2022-10-20 stsp }
705 0136599f 2022-10-20 stsp packidx.hdr.sorted_ids = calloc(nobj,
706 0136599f 2022-10-20 stsp sizeof(struct got_packidx_object_id));
707 0136599f 2022-10-20 stsp if (packidx.hdr.sorted_ids == NULL) {
708 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
709 0136599f 2022-10-20 stsp goto done;
710 0136599f 2022-10-20 stsp }
711 0136599f 2022-10-20 stsp packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
712 0136599f 2022-10-20 stsp if (packidx.hdr.crc32 == NULL) {
713 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
714 0136599f 2022-10-20 stsp goto done;
715 0136599f 2022-10-20 stsp }
716 0136599f 2022-10-20 stsp packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
717 0136599f 2022-10-20 stsp if (packidx.hdr.offsets == NULL) {
718 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
719 0136599f 2022-10-20 stsp goto done;
720 0136599f 2022-10-20 stsp }
721 0136599f 2022-10-20 stsp /* Large offsets table is empty for pack files < 2 GB. */
722 0136599f 2022-10-20 stsp if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
723 0136599f 2022-10-20 stsp packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
724 0136599f 2022-10-20 stsp if (packidx.hdr.large_offsets == NULL) {
725 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
726 0136599f 2022-10-20 stsp goto done;
727 0136599f 2022-10-20 stsp }
728 0136599f 2022-10-20 stsp }
729 0136599f 2022-10-20 stsp
730 0136599f 2022-10-20 stsp nvalid = 0;
731 0136599f 2022-10-20 stsp nloose = 0;
732 0136599f 2022-10-20 stsp objects = calloc(nobj, sizeof(struct got_indexed_object));
733 0136599f 2022-10-20 stsp if (objects == NULL)
734 0136599f 2022-10-20 stsp return got_error_from_errno("calloc");
735 0136599f 2022-10-20 stsp
736 0136599f 2022-10-20 stsp /*
737 0136599f 2022-10-20 stsp * First pass: locate all objects and identify un-deltified objects.
738 0136599f 2022-10-20 stsp *
739 0136599f 2022-10-20 stsp * When this pass has completed we will know offset, type, size, and
740 0136599f 2022-10-20 stsp * CRC information for all objects in this pack file. We won't know
741 0136599f 2022-10-20 stsp * any of the actual object IDs of deltified objects yet since we
742 0136599f 2022-10-20 stsp * will not yet attempt to combine deltas.
743 0136599f 2022-10-20 stsp */
744 0136599f 2022-10-20 stsp pass = 1;
745 0136599f 2022-10-20 stsp for (i = 0; i < nobj; i++) {
746 0136599f 2022-10-20 stsp /* Don't send too many progress privsep messages. */
747 0136599f 2022-10-20 stsp p_indexed = ((i + 1) * 100) / nobj;
748 0136599f 2022-10-20 stsp if (p_indexed != last_p_indexed) {
749 0136599f 2022-10-20 stsp err = report_progress(nobj, i + 1, nloose, 0,
750 713d6e11 2022-10-20 stsp rl, progress_cb, progress_arg);
751 0136599f 2022-10-20 stsp if (err)
752 0136599f 2022-10-20 stsp goto done;
753 0136599f 2022-10-20 stsp last_p_indexed = p_indexed;
754 0136599f 2022-10-20 stsp }
755 0136599f 2022-10-20 stsp
756 0136599f 2022-10-20 stsp obj = &objects[i];
757 0136599f 2022-10-20 stsp obj->crc = crc32(0L, NULL, 0);
758 0136599f 2022-10-20 stsp
759 0136599f 2022-10-20 stsp /* Store offset to type+size information for this object. */
760 0136599f 2022-10-20 stsp if (pack->map) {
761 0136599f 2022-10-20 stsp obj->off = mapoff;
762 0136599f 2022-10-20 stsp } else {
763 0136599f 2022-10-20 stsp obj->off = lseek(pack->fd, 0, SEEK_CUR);
764 0136599f 2022-10-20 stsp if (obj->off == -1) {
765 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
766 0136599f 2022-10-20 stsp goto done;
767 0136599f 2022-10-20 stsp }
768 0136599f 2022-10-20 stsp }
769 0136599f 2022-10-20 stsp
770 0136599f 2022-10-20 stsp err = read_packed_object(pack, obj, tmpfile, &ctx);
771 0136599f 2022-10-20 stsp if (err)
772 0136599f 2022-10-20 stsp goto done;
773 0136599f 2022-10-20 stsp
774 0136599f 2022-10-20 stsp if (pack->map) {
775 0136599f 2022-10-20 stsp mapoff += obj->tslen + obj->len;
776 0136599f 2022-10-20 stsp } else {
777 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
778 0136599f 2022-10-20 stsp SEEK_SET) == -1) {
779 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
780 0136599f 2022-10-20 stsp goto done;
781 0136599f 2022-10-20 stsp }
782 0136599f 2022-10-20 stsp }
783 0136599f 2022-10-20 stsp
784 0136599f 2022-10-20 stsp if (obj->type == GOT_OBJ_TYPE_BLOB ||
785 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_TREE ||
786 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_COMMIT ||
787 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_TAG) {
788 0136599f 2022-10-20 stsp obj->valid = 1;
789 0136599f 2022-10-20 stsp nloose++;
790 0136599f 2022-10-20 stsp } else {
791 0136599f 2022-10-20 stsp if (first_delta_idx == -1)
792 0136599f 2022-10-20 stsp first_delta_idx = i;
793 0136599f 2022-10-20 stsp if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
794 0136599f 2022-10-20 stsp have_ref_deltas = 1;
795 0136599f 2022-10-20 stsp }
796 0136599f 2022-10-20 stsp }
797 0136599f 2022-10-20 stsp nvalid = nloose;
798 0136599f 2022-10-20 stsp
799 0136599f 2022-10-20 stsp /*
800 0136599f 2022-10-20 stsp * Having done a full pass over the pack file and can now
801 0136599f 2022-10-20 stsp * verify its checksum.
802 0136599f 2022-10-20 stsp */
803 3093e2d7 2023-02-04 op SHA1Final(pack_hash, &ctx);
804 13b2bc37 2022-10-23 stsp
805 3093e2d7 2023-02-04 op if (memcmp(pack_hash_expected, pack_hash, SHA1_DIGEST_LENGTH) != 0) {
806 13b2bc37 2022-10-23 stsp err = got_error(GOT_ERR_PACKFILE_CSUM);
807 0136599f 2022-10-20 stsp goto done;
808 0136599f 2022-10-20 stsp }
809 0136599f 2022-10-20 stsp
810 0136599f 2022-10-20 stsp /* Verify the SHA1 checksum stored at the end of the pack file. */
811 0136599f 2022-10-20 stsp if (pack->map) {
812 ad4cc361 2022-10-27 op if (pack->filesize > SIZE_MAX) {
813 ad4cc361 2022-10-27 op err = got_error_fmt(GOT_ERR_RANGE,
814 ad4cc361 2022-10-27 op "filesize %lld overflows size_t",
815 ad4cc361 2022-10-27 op (long long)pack->filesize);
816 ad4cc361 2022-10-27 op goto done;
817 ad4cc361 2022-10-27 op }
818 ad4cc361 2022-10-27 op
819 3093e2d7 2023-02-04 op memcpy(pack_hash_expected, pack->map +
820 0136599f 2022-10-20 stsp pack->filesize - SHA1_DIGEST_LENGTH,
821 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
822 0136599f 2022-10-20 stsp } else {
823 0136599f 2022-10-20 stsp ssize_t n;
824 0136599f 2022-10-20 stsp if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
825 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
826 0136599f 2022-10-20 stsp goto done;
827 0136599f 2022-10-20 stsp }
828 3093e2d7 2023-02-04 op n = read(pack->fd, pack_hash_expected, SHA1_DIGEST_LENGTH);
829 0136599f 2022-10-20 stsp if (n == -1) {
830 0136599f 2022-10-20 stsp err = got_error_from_errno("read");
831 0136599f 2022-10-20 stsp goto done;
832 0136599f 2022-10-20 stsp }
833 0136599f 2022-10-20 stsp if (n != SHA1_DIGEST_LENGTH) {
834 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_IO);
835 0136599f 2022-10-20 stsp goto done;
836 0136599f 2022-10-20 stsp }
837 0136599f 2022-10-20 stsp }
838 3093e2d7 2023-02-04 op if (memcmp(pack_hash, pack_hash_expected, SHA1_DIGEST_LENGTH) != 0) {
839 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE,
840 0136599f 2022-10-20 stsp "bad checksum in pack file trailer");
841 0136599f 2022-10-20 stsp goto done;
842 0136599f 2022-10-20 stsp }
843 0136599f 2022-10-20 stsp
844 0136599f 2022-10-20 stsp if (first_delta_idx == -1)
845 0136599f 2022-10-20 stsp first_delta_idx = 0;
846 0136599f 2022-10-20 stsp
847 0136599f 2022-10-20 stsp /* In order to resolve ref deltas we need an in-progress pack index. */
848 0136599f 2022-10-20 stsp if (have_ref_deltas)
849 0136599f 2022-10-20 stsp make_packidx(&packidx, nobj, objects);
850 0136599f 2022-10-20 stsp
851 0136599f 2022-10-20 stsp /*
852 0136599f 2022-10-20 stsp * Second pass: We can now resolve deltas to compute the IDs of
853 0136599f 2022-10-20 stsp * objects which appear in deltified form. Because deltas can be
854 0136599f 2022-10-20 stsp * chained this pass may require a couple of iterations until all
855 0136599f 2022-10-20 stsp * IDs of deltified objects have been discovered.
856 0136599f 2022-10-20 stsp */
857 0136599f 2022-10-20 stsp pass++;
858 0136599f 2022-10-20 stsp while (nvalid != nobj) {
859 0136599f 2022-10-20 stsp int n = 0;
860 0136599f 2022-10-20 stsp /*
861 0136599f 2022-10-20 stsp * This loop will only run once unless the pack file
862 0136599f 2022-10-20 stsp * contains ref deltas which refer to objects located
863 0136599f 2022-10-20 stsp * later in the pack file, which is unusual.
864 0136599f 2022-10-20 stsp * Offset deltas can always be resolved in one pass
865 0136599f 2022-10-20 stsp * unless the packfile is corrupt.
866 0136599f 2022-10-20 stsp */
867 0136599f 2022-10-20 stsp for (i = first_delta_idx; i < nobj; i++) {
868 0136599f 2022-10-20 stsp obj = &objects[i];
869 0136599f 2022-10-20 stsp if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
870 0136599f 2022-10-20 stsp obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
871 0136599f 2022-10-20 stsp continue;
872 0136599f 2022-10-20 stsp
873 0136599f 2022-10-20 stsp if (obj->valid)
874 0136599f 2022-10-20 stsp continue;
875 0136599f 2022-10-20 stsp
876 0136599f 2022-10-20 stsp if (pack->map == NULL && lseek(pack->fd,
877 0136599f 2022-10-20 stsp obj->off + obj->tslen, SEEK_SET) == -1) {
878 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
879 0136599f 2022-10-20 stsp goto done;
880 0136599f 2022-10-20 stsp }
881 0136599f 2022-10-20 stsp
882 0136599f 2022-10-20 stsp err = resolve_deltified_object(pack, &packidx, obj,
883 0136599f 2022-10-20 stsp tmpfile, delta_base_file, delta_accum_file);
884 0136599f 2022-10-20 stsp if (err) {
885 0136599f 2022-10-20 stsp if (err->code != GOT_ERR_NO_OBJ)
886 0136599f 2022-10-20 stsp goto done;
887 0136599f 2022-10-20 stsp /*
888 0136599f 2022-10-20 stsp * We cannot resolve this object yet because
889 0136599f 2022-10-20 stsp * a delta base is unknown. Try again later.
890 0136599f 2022-10-20 stsp */
891 0136599f 2022-10-20 stsp continue;
892 0136599f 2022-10-20 stsp }
893 0136599f 2022-10-20 stsp
894 0136599f 2022-10-20 stsp obj->valid = 1;
895 0136599f 2022-10-20 stsp n++;
896 0136599f 2022-10-20 stsp if (have_ref_deltas)
897 0136599f 2022-10-20 stsp update_packidx(&packidx, nobj, obj);
898 0136599f 2022-10-20 stsp /* Don't send too many progress privsep messages. */
899 0136599f 2022-10-20 stsp p_resolved = ((nresolved + n) * 100) / nobj;
900 0136599f 2022-10-20 stsp if (p_resolved != last_p_resolved) {
901 0136599f 2022-10-20 stsp err = report_progress(nobj, nobj,
902 713d6e11 2022-10-20 stsp nloose, nresolved + n, rl,
903 0136599f 2022-10-20 stsp progress_cb, progress_arg);
904 0136599f 2022-10-20 stsp if (err)
905 0136599f 2022-10-20 stsp goto done;
906 0136599f 2022-10-20 stsp last_p_resolved = p_resolved;
907 0136599f 2022-10-20 stsp }
908 0136599f 2022-10-20 stsp
909 0136599f 2022-10-20 stsp }
910 0136599f 2022-10-20 stsp if (pass++ > 3 && n == 0) {
911 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE,
912 0136599f 2022-10-20 stsp "could not resolve any of deltas; packfile could "
913 0136599f 2022-10-20 stsp "be corrupt");
914 0136599f 2022-10-20 stsp goto done;
915 0136599f 2022-10-20 stsp }
916 0136599f 2022-10-20 stsp nresolved += n;
917 cf534ab8 2023-01-17 stsp nvalid += n;
918 0136599f 2022-10-20 stsp }
919 0136599f 2022-10-20 stsp
920 0136599f 2022-10-20 stsp if (nloose + nresolved != nobj) {
921 0136599f 2022-10-20 stsp static char msg[64];
922 0136599f 2022-10-20 stsp snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
923 0136599f 2022-10-20 stsp nloose + nresolved, nobj);
924 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
925 0136599f 2022-10-20 stsp goto done;
926 0136599f 2022-10-20 stsp }
927 0136599f 2022-10-20 stsp
928 0136599f 2022-10-20 stsp err = report_progress(nobj, nobj, nloose, nresolved, NULL,
929 0136599f 2022-10-20 stsp progress_cb, progress_arg);
930 0136599f 2022-10-20 stsp if (err)
931 0136599f 2022-10-20 stsp goto done;
932 0136599f 2022-10-20 stsp
933 0136599f 2022-10-20 stsp make_packidx(&packidx, nobj, objects);
934 0136599f 2022-10-20 stsp
935 0136599f 2022-10-20 stsp free(objects);
936 0136599f 2022-10-20 stsp objects = NULL;
937 0136599f 2022-10-20 stsp
938 0136599f 2022-10-20 stsp SHA1Init(&ctx);
939 0136599f 2022-10-20 stsp putbe32(buf, GOT_PACKIDX_V2_MAGIC);
940 0136599f 2022-10-20 stsp putbe32(buf + 4, GOT_PACKIDX_VERSION);
941 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, buf, 8, &ctx);
942 0136599f 2022-10-20 stsp if (err)
943 0136599f 2022-10-20 stsp goto done;
944 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
945 0136599f 2022-10-20 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
946 0136599f 2022-10-20 stsp if (err)
947 0136599f 2022-10-20 stsp goto done;
948 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
949 0136599f 2022-10-20 stsp nobj * SHA1_DIGEST_LENGTH, &ctx);
950 0136599f 2022-10-20 stsp if (err)
951 0136599f 2022-10-20 stsp goto done;
952 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
953 13b2bc37 2022-10-23 stsp nobj * sizeof(uint32_t), &ctx);
954 0136599f 2022-10-20 stsp if (err)
955 0136599f 2022-10-20 stsp goto done;
956 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
957 13b2bc37 2022-10-23 stsp nobj * sizeof(uint32_t), &ctx);
958 0136599f 2022-10-20 stsp if (err)
959 0136599f 2022-10-20 stsp goto done;
960 0136599f 2022-10-20 stsp if (packidx.nlargeobj > 0) {
961 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
962 0136599f 2022-10-20 stsp packidx.nlargeobj * sizeof(uint64_t), &ctx);
963 0136599f 2022-10-20 stsp if (err)
964 0136599f 2022-10-20 stsp goto done;
965 0136599f 2022-10-20 stsp }
966 3093e2d7 2023-02-04 op err = got_pack_hwrite(idxfd, pack_hash, SHA1_DIGEST_LENGTH, &ctx);
967 0136599f 2022-10-20 stsp if (err)
968 0136599f 2022-10-20 stsp goto done;
969 0136599f 2022-10-20 stsp
970 0136599f 2022-10-20 stsp SHA1Final(packidx_hash, &ctx);
971 0136599f 2022-10-20 stsp w = write(idxfd, packidx_hash, sizeof(packidx_hash));
972 0136599f 2022-10-20 stsp if (w == -1) {
973 0136599f 2022-10-20 stsp err = got_error_from_errno("write");
974 0136599f 2022-10-20 stsp goto done;
975 0136599f 2022-10-20 stsp }
976 0136599f 2022-10-20 stsp if (w != sizeof(packidx_hash)) {
977 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_IO);
978 0136599f 2022-10-20 stsp goto done;
979 0136599f 2022-10-20 stsp }
980 0136599f 2022-10-20 stsp done:
981 0136599f 2022-10-20 stsp free(objects);
982 0136599f 2022-10-20 stsp free(packidx.hdr.magic);
983 0136599f 2022-10-20 stsp free(packidx.hdr.version);
984 0136599f 2022-10-20 stsp free(packidx.hdr.fanout_table);
985 0136599f 2022-10-20 stsp free(packidx.hdr.sorted_ids);
986 0136599f 2022-10-20 stsp free(packidx.hdr.offsets);
987 0136599f 2022-10-20 stsp free(packidx.hdr.large_offsets);
988 0136599f 2022-10-20 stsp return err;
989 0136599f 2022-10-20 stsp }