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 0136599f 2022-10-20 stsp #include <endian.h>
36 0136599f 2022-10-20 stsp #include <fcntl.h>
37 0136599f 2022-10-20 stsp #include <unistd.h>
38 0136599f 2022-10-20 stsp #include <zlib.h>
39 0136599f 2022-10-20 stsp #include <err.h>
40 0136599f 2022-10-20 stsp #include <assert.h>
41 0136599f 2022-10-20 stsp #include <dirent.h>
42 0136599f 2022-10-20 stsp
43 0136599f 2022-10-20 stsp #include "got_error.h"
44 0136599f 2022-10-20 stsp #include "got_object.h"
45 0136599f 2022-10-20 stsp
46 0136599f 2022-10-20 stsp #include "got_lib_sha1.h"
47 0136599f 2022-10-20 stsp #include "got_lib_delta.h"
48 0136599f 2022-10-20 stsp #include "got_lib_inflate.h"
49 0136599f 2022-10-20 stsp #include "got_lib_object.h"
50 0136599f 2022-10-20 stsp #include "got_lib_object_parse.h"
51 0136599f 2022-10-20 stsp #include "got_lib_object_idset.h"
52 0136599f 2022-10-20 stsp #include "got_lib_privsep.h"
53 0136599f 2022-10-20 stsp #include "got_lib_pack.h"
54 0136599f 2022-10-20 stsp #include "got_lib_ratelimit.h"
55 0136599f 2022-10-20 stsp #include "got_lib_pack_index.h"
56 0136599f 2022-10-20 stsp #include "got_lib_delta_cache.h"
57 0136599f 2022-10-20 stsp
58 0136599f 2022-10-20 stsp struct got_indexed_object {
59 0136599f 2022-10-20 stsp struct got_object_id id;
60 0136599f 2022-10-20 stsp
61 0136599f 2022-10-20 stsp /*
62 0136599f 2022-10-20 stsp * Has this object been fully resolved?
63 0136599f 2022-10-20 stsp * If so, we know its ID, otherwise we don't and 'id' is invalid.
64 0136599f 2022-10-20 stsp */
65 0136599f 2022-10-20 stsp int valid;
66 0136599f 2022-10-20 stsp
67 0136599f 2022-10-20 stsp /* Offset of type+size field for this object in pack file. */
68 0136599f 2022-10-20 stsp off_t off;
69 0136599f 2022-10-20 stsp
70 0136599f 2022-10-20 stsp /* Type+size values parsed from pack file. */
71 0136599f 2022-10-20 stsp uint8_t type;
72 0136599f 2022-10-20 stsp uint64_t size;
73 0136599f 2022-10-20 stsp
74 0136599f 2022-10-20 stsp /* Length of on-disk type+size data. */
75 0136599f 2022-10-20 stsp size_t tslen;
76 0136599f 2022-10-20 stsp
77 0136599f 2022-10-20 stsp /* Length of object data following type+size. */
78 0136599f 2022-10-20 stsp size_t len;
79 0136599f 2022-10-20 stsp
80 0136599f 2022-10-20 stsp uint32_t crc;
81 0136599f 2022-10-20 stsp
82 0136599f 2022-10-20 stsp union {
83 0136599f 2022-10-20 stsp struct {
84 0136599f 2022-10-20 stsp /* For ref deltas. */
85 0136599f 2022-10-20 stsp struct got_object_id ref_id;
86 0136599f 2022-10-20 stsp } ref;
87 0136599f 2022-10-20 stsp struct {
88 0136599f 2022-10-20 stsp /* For offset deltas. */
89 0136599f 2022-10-20 stsp off_t base_offset;
90 0136599f 2022-10-20 stsp size_t base_offsetlen;
91 0136599f 2022-10-20 stsp } ofs;
92 0136599f 2022-10-20 stsp } delta;
93 0136599f 2022-10-20 stsp };
94 0136599f 2022-10-20 stsp
95 0136599f 2022-10-20 stsp static void
96 0136599f 2022-10-20 stsp putbe32(char *b, uint32_t n)
97 0136599f 2022-10-20 stsp {
98 0136599f 2022-10-20 stsp b[0] = n >> 24;
99 0136599f 2022-10-20 stsp b[1] = n >> 16;
100 0136599f 2022-10-20 stsp b[2] = n >> 8;
101 0136599f 2022-10-20 stsp b[3] = n >> 0;
102 0136599f 2022-10-20 stsp }
103 0136599f 2022-10-20 stsp
104 0136599f 2022-10-20 stsp static const struct got_error *
105 0136599f 2022-10-20 stsp get_obj_type_label(const char **label, int obj_type)
106 0136599f 2022-10-20 stsp {
107 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
108 0136599f 2022-10-20 stsp
109 0136599f 2022-10-20 stsp switch (obj_type) {
110 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_BLOB:
111 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_BLOB;
112 0136599f 2022-10-20 stsp break;
113 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TREE:
114 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_TREE;
115 0136599f 2022-10-20 stsp break;
116 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_COMMIT:
117 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_COMMIT;
118 0136599f 2022-10-20 stsp break;
119 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TAG:
120 0136599f 2022-10-20 stsp *label = GOT_OBJ_LABEL_TAG;
121 0136599f 2022-10-20 stsp break;
122 0136599f 2022-10-20 stsp default:
123 0136599f 2022-10-20 stsp *label = NULL;
124 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_OBJ_TYPE);
125 0136599f 2022-10-20 stsp break;
126 0136599f 2022-10-20 stsp }
127 0136599f 2022-10-20 stsp
128 0136599f 2022-10-20 stsp return err;
129 0136599f 2022-10-20 stsp }
130 0136599f 2022-10-20 stsp
131 0136599f 2022-10-20 stsp static const struct got_error *
132 0136599f 2022-10-20 stsp read_checksum(uint32_t *crc, SHA1_CTX *sha1_ctx, int fd, size_t len)
133 0136599f 2022-10-20 stsp {
134 0136599f 2022-10-20 stsp uint8_t buf[8192];
135 0136599f 2022-10-20 stsp size_t n;
136 0136599f 2022-10-20 stsp ssize_t r;
137 0136599f 2022-10-20 stsp
138 0136599f 2022-10-20 stsp for (n = len; n > 0; n -= r){
139 0136599f 2022-10-20 stsp r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
140 0136599f 2022-10-20 stsp if (r == -1)
141 0136599f 2022-10-20 stsp return got_error_from_errno("read");
142 0136599f 2022-10-20 stsp if (r == 0)
143 0136599f 2022-10-20 stsp break;
144 0136599f 2022-10-20 stsp if (crc)
145 0136599f 2022-10-20 stsp *crc = crc32(*crc, buf, r);
146 0136599f 2022-10-20 stsp if (sha1_ctx)
147 0136599f 2022-10-20 stsp SHA1Update(sha1_ctx, buf, r);
148 0136599f 2022-10-20 stsp }
149 0136599f 2022-10-20 stsp
150 0136599f 2022-10-20 stsp return NULL;
151 0136599f 2022-10-20 stsp }
152 0136599f 2022-10-20 stsp
153 0136599f 2022-10-20 stsp static const struct got_error *
154 0136599f 2022-10-20 stsp read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len)
155 0136599f 2022-10-20 stsp {
156 0136599f 2022-10-20 stsp uint8_t buf[8192];
157 0136599f 2022-10-20 stsp size_t n, r;
158 0136599f 2022-10-20 stsp
159 0136599f 2022-10-20 stsp for (n = len; n > 0; n -= r) {
160 0136599f 2022-10-20 stsp r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
161 0136599f 2022-10-20 stsp if (r == 0) {
162 0136599f 2022-10-20 stsp if (feof(f))
163 0136599f 2022-10-20 stsp return NULL;
164 0136599f 2022-10-20 stsp return got_ferror(f, GOT_ERR_IO);
165 0136599f 2022-10-20 stsp }
166 0136599f 2022-10-20 stsp SHA1Update(ctx, buf, r);
167 0136599f 2022-10-20 stsp }
168 0136599f 2022-10-20 stsp
169 0136599f 2022-10-20 stsp return NULL;
170 0136599f 2022-10-20 stsp }
171 0136599f 2022-10-20 stsp
172 0136599f 2022-10-20 stsp static const struct got_error *
173 0136599f 2022-10-20 stsp read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
174 0136599f 2022-10-20 stsp FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
175 0136599f 2022-10-20 stsp {
176 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
177 0136599f 2022-10-20 stsp SHA1_CTX ctx;
178 0136599f 2022-10-20 stsp uint8_t *data = NULL;
179 0136599f 2022-10-20 stsp size_t datalen = 0;
180 0136599f 2022-10-20 stsp ssize_t n;
181 0136599f 2022-10-20 stsp char *header;
182 0136599f 2022-10-20 stsp size_t headerlen;
183 0136599f 2022-10-20 stsp const char *obj_label;
184 0136599f 2022-10-20 stsp size_t mapoff = obj->off;
185 0136599f 2022-10-20 stsp struct got_inflate_checksum csum;
186 0136599f 2022-10-20 stsp
187 0136599f 2022-10-20 stsp memset(&csum, 0, sizeof(csum));
188 0136599f 2022-10-20 stsp csum.input_sha1 = pack_sha1_ctx;
189 0136599f 2022-10-20 stsp csum.input_crc = &obj->crc;
190 0136599f 2022-10-20 stsp
191 0136599f 2022-10-20 stsp err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
192 0136599f 2022-10-20 stsp &obj->tslen, pack, obj->off);
193 0136599f 2022-10-20 stsp if (err)
194 0136599f 2022-10-20 stsp return err;
195 0136599f 2022-10-20 stsp
196 0136599f 2022-10-20 stsp if (pack->map) {
197 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
198 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
199 0136599f 2022-10-20 stsp mapoff += obj->tslen;
200 0136599f 2022-10-20 stsp } else {
201 0136599f 2022-10-20 stsp /* XXX Seek back and get the CRC of on-disk type+size bytes. */
202 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
203 0136599f 2022-10-20 stsp return got_error_from_errno("lseek");
204 0136599f 2022-10-20 stsp err = read_checksum(&obj->crc, pack_sha1_ctx,
205 0136599f 2022-10-20 stsp pack->fd, obj->tslen);
206 0136599f 2022-10-20 stsp if (err)
207 0136599f 2022-10-20 stsp return err;
208 0136599f 2022-10-20 stsp }
209 0136599f 2022-10-20 stsp
210 0136599f 2022-10-20 stsp switch (obj->type) {
211 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_BLOB:
212 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_COMMIT:
213 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TREE:
214 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_TAG:
215 0136599f 2022-10-20 stsp if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
216 0136599f 2022-10-20 stsp if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
217 0136599f 2022-10-20 stsp err = got_error_from_errno("fseek");
218 0136599f 2022-10-20 stsp break;
219 0136599f 2022-10-20 stsp }
220 0136599f 2022-10-20 stsp if (pack->map) {
221 0136599f 2022-10-20 stsp err = got_inflate_to_file_mmap(&datalen,
222 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
223 0136599f 2022-10-20 stsp pack->filesize - mapoff, tmpfile);
224 0136599f 2022-10-20 stsp } else {
225 0136599f 2022-10-20 stsp err = got_inflate_to_file_fd(&datalen,
226 0136599f 2022-10-20 stsp &obj->len, &csum, pack->fd, tmpfile);
227 0136599f 2022-10-20 stsp }
228 0136599f 2022-10-20 stsp } else {
229 0136599f 2022-10-20 stsp if (pack->map) {
230 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(&data, &datalen,
231 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
232 0136599f 2022-10-20 stsp pack->filesize - mapoff);
233 0136599f 2022-10-20 stsp } else {
234 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(&data, &datalen,
235 0136599f 2022-10-20 stsp &obj->len, &csum, obj->size, pack->fd);
236 0136599f 2022-10-20 stsp }
237 0136599f 2022-10-20 stsp }
238 0136599f 2022-10-20 stsp if (err)
239 0136599f 2022-10-20 stsp break;
240 0136599f 2022-10-20 stsp SHA1Init(&ctx);
241 0136599f 2022-10-20 stsp err = get_obj_type_label(&obj_label, obj->type);
242 0136599f 2022-10-20 stsp if (err) {
243 0136599f 2022-10-20 stsp free(data);
244 0136599f 2022-10-20 stsp break;
245 0136599f 2022-10-20 stsp }
246 0136599f 2022-10-20 stsp if (asprintf(&header, "%s %lld", obj_label,
247 0136599f 2022-10-20 stsp (long long)obj->size) == -1) {
248 0136599f 2022-10-20 stsp err = got_error_from_errno("asprintf");
249 0136599f 2022-10-20 stsp free(data);
250 0136599f 2022-10-20 stsp break;
251 0136599f 2022-10-20 stsp }
252 0136599f 2022-10-20 stsp headerlen = strlen(header) + 1;
253 0136599f 2022-10-20 stsp SHA1Update(&ctx, header, headerlen);
254 0136599f 2022-10-20 stsp if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
255 0136599f 2022-10-20 stsp err = read_file_sha1(&ctx, tmpfile, datalen);
256 0136599f 2022-10-20 stsp if (err) {
257 0136599f 2022-10-20 stsp free(header);
258 0136599f 2022-10-20 stsp free(data);
259 0136599f 2022-10-20 stsp break;
260 0136599f 2022-10-20 stsp }
261 0136599f 2022-10-20 stsp } else
262 0136599f 2022-10-20 stsp SHA1Update(&ctx, data, datalen);
263 0136599f 2022-10-20 stsp SHA1Final(obj->id.sha1, &ctx);
264 0136599f 2022-10-20 stsp free(header);
265 0136599f 2022-10-20 stsp free(data);
266 0136599f 2022-10-20 stsp break;
267 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_REF_DELTA:
268 0136599f 2022-10-20 stsp memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
269 0136599f 2022-10-20 stsp if (pack->map) {
270 0136599f 2022-10-20 stsp if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
271 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
272 0136599f 2022-10-20 stsp break;
273 0136599f 2022-10-20 stsp }
274 0136599f 2022-10-20 stsp memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
275 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
276 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff,
277 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
278 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff,
279 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
280 0136599f 2022-10-20 stsp mapoff += SHA1_DIGEST_LENGTH;
281 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(NULL, &datalen,
282 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
283 0136599f 2022-10-20 stsp pack->filesize - mapoff);
284 0136599f 2022-10-20 stsp if (err)
285 0136599f 2022-10-20 stsp break;
286 0136599f 2022-10-20 stsp } else {
287 0136599f 2022-10-20 stsp n = read(pack->fd, obj->delta.ref.ref_id.sha1,
288 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
289 0136599f 2022-10-20 stsp if (n == -1) {
290 0136599f 2022-10-20 stsp err = got_error_from_errno("read");
291 0136599f 2022-10-20 stsp break;
292 0136599f 2022-10-20 stsp }
293 0136599f 2022-10-20 stsp if (n < sizeof(obj->id)) {
294 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
295 0136599f 2022-10-20 stsp break;
296 0136599f 2022-10-20 stsp }
297 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
298 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
299 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
300 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
301 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
302 0136599f 2022-10-20 stsp &csum, obj->size, pack->fd);
303 0136599f 2022-10-20 stsp if (err)
304 0136599f 2022-10-20 stsp break;
305 0136599f 2022-10-20 stsp }
306 0136599f 2022-10-20 stsp obj->len += SHA1_DIGEST_LENGTH;
307 0136599f 2022-10-20 stsp break;
308 0136599f 2022-10-20 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
309 0136599f 2022-10-20 stsp memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
310 0136599f 2022-10-20 stsp err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
311 0136599f 2022-10-20 stsp &obj->delta.ofs.base_offsetlen, pack, obj->off,
312 0136599f 2022-10-20 stsp obj->tslen);
313 0136599f 2022-10-20 stsp if (err)
314 0136599f 2022-10-20 stsp break;
315 0136599f 2022-10-20 stsp
316 0136599f 2022-10-20 stsp if (pack->map) {
317 0136599f 2022-10-20 stsp obj->crc = crc32(obj->crc, pack->map + mapoff,
318 0136599f 2022-10-20 stsp obj->delta.ofs.base_offsetlen);
319 0136599f 2022-10-20 stsp SHA1Update(pack_sha1_ctx, pack->map + mapoff,
320 0136599f 2022-10-20 stsp obj->delta.ofs.base_offsetlen);
321 0136599f 2022-10-20 stsp mapoff += obj->delta.ofs.base_offsetlen;
322 0136599f 2022-10-20 stsp err = got_inflate_to_mem_mmap(NULL, &datalen,
323 0136599f 2022-10-20 stsp &obj->len, &csum, pack->map, mapoff,
324 0136599f 2022-10-20 stsp pack->filesize - mapoff);
325 0136599f 2022-10-20 stsp if (err)
326 0136599f 2022-10-20 stsp break;
327 0136599f 2022-10-20 stsp } else {
328 0136599f 2022-10-20 stsp /*
329 0136599f 2022-10-20 stsp * XXX Seek back and get CRC and SHA1 of on-disk
330 0136599f 2022-10-20 stsp * offset bytes.
331 0136599f 2022-10-20 stsp */
332 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
333 0136599f 2022-10-20 stsp == -1) {
334 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
335 0136599f 2022-10-20 stsp break;
336 0136599f 2022-10-20 stsp }
337 0136599f 2022-10-20 stsp err = read_checksum(&obj->crc, pack_sha1_ctx,
338 0136599f 2022-10-20 stsp pack->fd, obj->delta.ofs.base_offsetlen);
339 0136599f 2022-10-20 stsp if (err)
340 0136599f 2022-10-20 stsp break;
341 0136599f 2022-10-20 stsp
342 0136599f 2022-10-20 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
343 0136599f 2022-10-20 stsp &csum, obj->size, pack->fd);
344 0136599f 2022-10-20 stsp if (err)
345 0136599f 2022-10-20 stsp break;
346 0136599f 2022-10-20 stsp }
347 0136599f 2022-10-20 stsp obj->len += obj->delta.ofs.base_offsetlen;
348 0136599f 2022-10-20 stsp break;
349 0136599f 2022-10-20 stsp default:
350 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_OBJ_TYPE);
351 0136599f 2022-10-20 stsp break;
352 0136599f 2022-10-20 stsp }
353 0136599f 2022-10-20 stsp
354 0136599f 2022-10-20 stsp return err;
355 0136599f 2022-10-20 stsp }
356 0136599f 2022-10-20 stsp
357 13b2bc37 2022-10-23 stsp const struct got_error *
358 13b2bc37 2022-10-23 stsp got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
359 0136599f 2022-10-20 stsp {
360 0136599f 2022-10-20 stsp ssize_t w;
361 0136599f 2022-10-20 stsp
362 0136599f 2022-10-20 stsp SHA1Update(ctx, buf, len);
363 0136599f 2022-10-20 stsp
364 0136599f 2022-10-20 stsp w = write(fd, buf, len);
365 0136599f 2022-10-20 stsp if (w == -1)
366 0136599f 2022-10-20 stsp return got_error_from_errno("write");
367 0136599f 2022-10-20 stsp if (w != len)
368 0136599f 2022-10-20 stsp return got_error(GOT_ERR_IO);
369 0136599f 2022-10-20 stsp
370 0136599f 2022-10-20 stsp return NULL;
371 0136599f 2022-10-20 stsp }
372 0136599f 2022-10-20 stsp
373 0136599f 2022-10-20 stsp static const struct got_error *
374 0136599f 2022-10-20 stsp resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
375 0136599f 2022-10-20 stsp struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
376 0136599f 2022-10-20 stsp FILE *delta_accum_file)
377 0136599f 2022-10-20 stsp {
378 0136599f 2022-10-20 stsp const struct got_error *err = NULL;
379 0136599f 2022-10-20 stsp struct got_delta_chain deltas;
380 0136599f 2022-10-20 stsp struct got_delta *delta;
381 0136599f 2022-10-20 stsp uint8_t *buf = NULL;
382 0136599f 2022-10-20 stsp size_t len = 0;
383 0136599f 2022-10-20 stsp SHA1_CTX ctx;
384 0136599f 2022-10-20 stsp char *header = NULL;
385 0136599f 2022-10-20 stsp size_t headerlen;
386 0136599f 2022-10-20 stsp uint64_t max_size;
387 0136599f 2022-10-20 stsp int base_obj_type;
388 0136599f 2022-10-20 stsp const char *obj_label;
389 0136599f 2022-10-20 stsp
390 0136599f 2022-10-20 stsp deltas.nentries = 0;
391 0136599f 2022-10-20 stsp STAILQ_INIT(&deltas.entries);
392 0136599f 2022-10-20 stsp
393 0136599f 2022-10-20 stsp err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
394 0136599f 2022-10-20 stsp obj->off, obj->tslen, obj->type, obj->size,
395 0136599f 2022-10-20 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
396 0136599f 2022-10-20 stsp if (err)
397 0136599f 2022-10-20 stsp goto done;
398 0136599f 2022-10-20 stsp
399 0136599f 2022-10-20 stsp err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
400 0136599f 2022-10-20 stsp if (err)
401 0136599f 2022-10-20 stsp goto done;
402 0136599f 2022-10-20 stsp if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
403 0136599f 2022-10-20 stsp rewind(tmpfile);
404 0136599f 2022-10-20 stsp rewind(delta_base_file);
405 0136599f 2022-10-20 stsp rewind(delta_accum_file);
406 0136599f 2022-10-20 stsp err = got_pack_dump_delta_chain_to_file(&len, &deltas,
407 0136599f 2022-10-20 stsp pack, tmpfile, delta_base_file, delta_accum_file);
408 0136599f 2022-10-20 stsp if (err)
409 0136599f 2022-10-20 stsp goto done;
410 0136599f 2022-10-20 stsp } else {
411 0136599f 2022-10-20 stsp err = got_pack_dump_delta_chain_to_mem(&buf, &len,
412 0136599f 2022-10-20 stsp &deltas, pack);
413 0136599f 2022-10-20 stsp }
414 0136599f 2022-10-20 stsp if (err)
415 0136599f 2022-10-20 stsp goto done;
416 0136599f 2022-10-20 stsp
417 0136599f 2022-10-20 stsp err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
418 0136599f 2022-10-20 stsp if (err)
419 0136599f 2022-10-20 stsp goto done;
420 0136599f 2022-10-20 stsp err = get_obj_type_label(&obj_label, base_obj_type);
421 0136599f 2022-10-20 stsp if (err)
422 0136599f 2022-10-20 stsp goto done;
423 0136599f 2022-10-20 stsp if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
424 0136599f 2022-10-20 stsp err = got_error_from_errno("asprintf");
425 0136599f 2022-10-20 stsp goto done;
426 0136599f 2022-10-20 stsp }
427 0136599f 2022-10-20 stsp headerlen = strlen(header) + 1;
428 0136599f 2022-10-20 stsp SHA1Init(&ctx);
429 0136599f 2022-10-20 stsp SHA1Update(&ctx, header, headerlen);
430 0136599f 2022-10-20 stsp if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
431 0136599f 2022-10-20 stsp err = read_file_sha1(&ctx, tmpfile, len);
432 0136599f 2022-10-20 stsp if (err)
433 0136599f 2022-10-20 stsp goto done;
434 0136599f 2022-10-20 stsp } else
435 0136599f 2022-10-20 stsp SHA1Update(&ctx, buf, len);
436 0136599f 2022-10-20 stsp SHA1Final(obj->id.sha1, &ctx);
437 0136599f 2022-10-20 stsp done:
438 0136599f 2022-10-20 stsp free(buf);
439 0136599f 2022-10-20 stsp free(header);
440 0136599f 2022-10-20 stsp while (!STAILQ_EMPTY(&deltas.entries)) {
441 0136599f 2022-10-20 stsp delta = STAILQ_FIRST(&deltas.entries);
442 0136599f 2022-10-20 stsp STAILQ_REMOVE_HEAD(&deltas.entries, entry);
443 0136599f 2022-10-20 stsp free(delta);
444 0136599f 2022-10-20 stsp }
445 0136599f 2022-10-20 stsp return err;
446 0136599f 2022-10-20 stsp }
447 0136599f 2022-10-20 stsp
448 0136599f 2022-10-20 stsp /* Determine the slot in the pack index a given object ID should use. */
449 0136599f 2022-10-20 stsp static int
450 0136599f 2022-10-20 stsp find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
451 0136599f 2022-10-20 stsp {
452 0136599f 2022-10-20 stsp u_int8_t id0 = sha1[0];
453 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
454 0136599f 2022-10-20 stsp int left = 0, right = nindexed - 1;
455 0136599f 2022-10-20 stsp int cmp = 0, i = 0;
456 0136599f 2022-10-20 stsp
457 0136599f 2022-10-20 stsp if (id0 > 0)
458 0136599f 2022-10-20 stsp left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
459 0136599f 2022-10-20 stsp
460 0136599f 2022-10-20 stsp while (left <= right) {
461 0136599f 2022-10-20 stsp struct got_packidx_object_id *oid;
462 0136599f 2022-10-20 stsp
463 0136599f 2022-10-20 stsp i = ((left + right) / 2);
464 0136599f 2022-10-20 stsp oid = &packidx->hdr.sorted_ids[i];
465 0136599f 2022-10-20 stsp
466 0136599f 2022-10-20 stsp cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
467 0136599f 2022-10-20 stsp if (cmp == 0)
468 0136599f 2022-10-20 stsp return -1; /* object already indexed */
469 0136599f 2022-10-20 stsp else if (cmp > 0)
470 0136599f 2022-10-20 stsp left = i + 1;
471 0136599f 2022-10-20 stsp else if (cmp < 0)
472 0136599f 2022-10-20 stsp right = i - 1;
473 0136599f 2022-10-20 stsp }
474 0136599f 2022-10-20 stsp
475 0136599f 2022-10-20 stsp return left;
476 0136599f 2022-10-20 stsp }
477 0136599f 2022-10-20 stsp
478 0136599f 2022-10-20 stsp #if 0
479 0136599f 2022-10-20 stsp static void
480 0136599f 2022-10-20 stsp print_packidx(struct got_packidx *packidx)
481 0136599f 2022-10-20 stsp {
482 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
483 0136599f 2022-10-20 stsp int i;
484 0136599f 2022-10-20 stsp
485 0136599f 2022-10-20 stsp fprintf(stderr, "object IDs:\n");
486 0136599f 2022-10-20 stsp for (i = 0; i < nindexed; i++) {
487 0136599f 2022-10-20 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
488 0136599f 2022-10-20 stsp got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
489 0136599f 2022-10-20 stsp hex, sizeof(hex));
490 0136599f 2022-10-20 stsp fprintf(stderr, "%s\n", hex);
491 0136599f 2022-10-20 stsp }
492 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
493 0136599f 2022-10-20 stsp
494 0136599f 2022-10-20 stsp fprintf(stderr, "object offsets:\n");
495 0136599f 2022-10-20 stsp for (i = 0; i < nindexed; i++) {
496 0136599f 2022-10-20 stsp uint32_t offset = be32toh(packidx->hdr.offsets[i]);
497 0136599f 2022-10-20 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
498 0136599f 2022-10-20 stsp int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
499 0136599f 2022-10-20 stsp fprintf(stderr, "%u -> %llu\n", offset,
500 0136599f 2022-10-20 stsp be64toh(packidx->hdr.large_offsets[j]));
501 0136599f 2022-10-20 stsp } else
502 0136599f 2022-10-20 stsp fprintf(stderr, "%u\n", offset);
503 0136599f 2022-10-20 stsp }
504 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
505 0136599f 2022-10-20 stsp
506 0136599f 2022-10-20 stsp fprintf(stderr, "fanout table:");
507 0136599f 2022-10-20 stsp for (i = 0; i <= 0xff; i++)
508 0136599f 2022-10-20 stsp fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
509 0136599f 2022-10-20 stsp fprintf(stderr, "\n");
510 0136599f 2022-10-20 stsp }
511 0136599f 2022-10-20 stsp #endif
512 0136599f 2022-10-20 stsp
513 0136599f 2022-10-20 stsp static void
514 0136599f 2022-10-20 stsp add_indexed_object(struct got_packidx *packidx, uint32_t idx,
515 0136599f 2022-10-20 stsp struct got_indexed_object *obj)
516 0136599f 2022-10-20 stsp {
517 0136599f 2022-10-20 stsp int i;
518 0136599f 2022-10-20 stsp
519 0136599f 2022-10-20 stsp memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
520 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
521 0136599f 2022-10-20 stsp packidx->hdr.crc32[idx] = htobe32(obj->crc);
522 0136599f 2022-10-20 stsp if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
523 0136599f 2022-10-20 stsp packidx->hdr.offsets[idx] = htobe32(obj->off);
524 0136599f 2022-10-20 stsp else {
525 0136599f 2022-10-20 stsp packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
526 0136599f 2022-10-20 stsp GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
527 0136599f 2022-10-20 stsp packidx->hdr.large_offsets[packidx->nlargeobj] =
528 0136599f 2022-10-20 stsp htobe64(obj->off);
529 0136599f 2022-10-20 stsp packidx->nlargeobj++;
530 0136599f 2022-10-20 stsp }
531 0136599f 2022-10-20 stsp
532 0136599f 2022-10-20 stsp for (i = obj->id.sha1[0]; i <= 0xff; i++) {
533 0136599f 2022-10-20 stsp uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
534 0136599f 2022-10-20 stsp packidx->hdr.fanout_table[i] = htobe32(n + 1);
535 0136599f 2022-10-20 stsp }
536 0136599f 2022-10-20 stsp }
537 0136599f 2022-10-20 stsp
538 0136599f 2022-10-20 stsp static int
539 0136599f 2022-10-20 stsp indexed_obj_cmp(const void *pa, const void *pb)
540 0136599f 2022-10-20 stsp {
541 0136599f 2022-10-20 stsp struct got_indexed_object *a, *b;
542 0136599f 2022-10-20 stsp
543 0136599f 2022-10-20 stsp a = (struct got_indexed_object *)pa;
544 0136599f 2022-10-20 stsp b = (struct got_indexed_object *)pb;
545 0136599f 2022-10-20 stsp return got_object_id_cmp(&a->id, &b->id);
546 0136599f 2022-10-20 stsp }
547 0136599f 2022-10-20 stsp
548 0136599f 2022-10-20 stsp static void
549 9316cc27 2022-10-20 stsp make_packidx(struct got_packidx *packidx, uint32_t nobj,
550 0136599f 2022-10-20 stsp struct got_indexed_object *objects)
551 0136599f 2022-10-20 stsp {
552 0136599f 2022-10-20 stsp struct got_indexed_object *obj;
553 0136599f 2022-10-20 stsp int i;
554 0136599f 2022-10-20 stsp uint32_t idx = 0;
555 0136599f 2022-10-20 stsp
556 0136599f 2022-10-20 stsp qsort(objects, nobj, sizeof(struct got_indexed_object),
557 0136599f 2022-10-20 stsp indexed_obj_cmp);
558 0136599f 2022-10-20 stsp
559 0136599f 2022-10-20 stsp memset(packidx->hdr.fanout_table, 0,
560 0136599f 2022-10-20 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
561 0136599f 2022-10-20 stsp packidx->nlargeobj = 0;
562 0136599f 2022-10-20 stsp
563 0136599f 2022-10-20 stsp for (i = 0; i < nobj; i++) {
564 0136599f 2022-10-20 stsp obj = &objects[i];
565 0136599f 2022-10-20 stsp if (obj->valid)
566 0136599f 2022-10-20 stsp add_indexed_object(packidx, idx++, obj);
567 0136599f 2022-10-20 stsp }
568 0136599f 2022-10-20 stsp }
569 0136599f 2022-10-20 stsp
570 0136599f 2022-10-20 stsp static void
571 9316cc27 2022-10-20 stsp update_packidx(struct got_packidx *packidx, uint32_t nobj,
572 0136599f 2022-10-20 stsp struct got_indexed_object *obj)
573 0136599f 2022-10-20 stsp {
574 0136599f 2022-10-20 stsp int idx;
575 0136599f 2022-10-20 stsp uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
576 0136599f 2022-10-20 stsp
577 0136599f 2022-10-20 stsp idx = find_object_idx(packidx, obj->id.sha1);
578 0136599f 2022-10-20 stsp if (idx == -1) {
579 0136599f 2022-10-20 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
580 0136599f 2022-10-20 stsp got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
581 0136599f 2022-10-20 stsp return; /* object already indexed */
582 0136599f 2022-10-20 stsp }
583 0136599f 2022-10-20 stsp
584 0136599f 2022-10-20 stsp memmove(&packidx->hdr.sorted_ids[idx + 1],
585 0136599f 2022-10-20 stsp &packidx->hdr.sorted_ids[idx],
586 0136599f 2022-10-20 stsp sizeof(struct got_packidx_object_id) * (nindexed - idx));
587 0136599f 2022-10-20 stsp memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
588 0136599f 2022-10-20 stsp sizeof(uint32_t) * (nindexed - idx));
589 0136599f 2022-10-20 stsp
590 0136599f 2022-10-20 stsp add_indexed_object(packidx, idx, obj);
591 0136599f 2022-10-20 stsp }
592 0136599f 2022-10-20 stsp
593 0136599f 2022-10-20 stsp static const struct got_error *
594 9316cc27 2022-10-20 stsp report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
595 9316cc27 2022-10-20 stsp uint32_t nobj_resolved, struct got_ratelimit *rl,
596 0136599f 2022-10-20 stsp got_pack_index_progress_cb progress_cb, void *progress_arg)
597 0136599f 2022-10-20 stsp {
598 0136599f 2022-10-20 stsp const struct got_error *err;
599 0136599f 2022-10-20 stsp int elapsed = 0;
600 0136599f 2022-10-20 stsp
601 0136599f 2022-10-20 stsp if (rl) {
602 0136599f 2022-10-20 stsp err = got_ratelimit_check(&elapsed, rl);
603 0136599f 2022-10-20 stsp if (err || !elapsed)
604 0136599f 2022-10-20 stsp return err;
605 0136599f 2022-10-20 stsp }
606 0136599f 2022-10-20 stsp
607 0136599f 2022-10-20 stsp return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
608 0136599f 2022-10-20 stsp nobj_resolved);
609 0136599f 2022-10-20 stsp }
610 0136599f 2022-10-20 stsp
611 0136599f 2022-10-20 stsp const struct got_error *
612 0136599f 2022-10-20 stsp got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
613 0136599f 2022-10-20 stsp FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
614 713d6e11 2022-10-20 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
615 713d6e11 2022-10-20 stsp struct got_ratelimit *rl)
616 0136599f 2022-10-20 stsp {
617 0136599f 2022-10-20 stsp const struct got_error *err;
618 0136599f 2022-10-20 stsp struct got_packfile_hdr hdr;
619 0136599f 2022-10-20 stsp struct got_packidx packidx;
620 0136599f 2022-10-20 stsp char buf[8];
621 0136599f 2022-10-20 stsp char pack_sha1[SHA1_DIGEST_LENGTH];
622 9316cc27 2022-10-20 stsp uint32_t nobj, nvalid, nloose, nresolved = 0, i;
623 0136599f 2022-10-20 stsp struct got_indexed_object *objects = NULL, *obj;
624 0136599f 2022-10-20 stsp SHA1_CTX ctx;
625 0136599f 2022-10-20 stsp uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
626 0136599f 2022-10-20 stsp ssize_t r, w;
627 0136599f 2022-10-20 stsp int pass, have_ref_deltas = 0, first_delta_idx = -1;
628 0136599f 2022-10-20 stsp size_t mapoff = 0;
629 0136599f 2022-10-20 stsp int p_indexed = 0, last_p_indexed = -1;
630 0136599f 2022-10-20 stsp int p_resolved = 0, last_p_resolved = -1;
631 0136599f 2022-10-20 stsp
632 0136599f 2022-10-20 stsp /* Require that pack file header and SHA1 trailer are present. */
633 0136599f 2022-10-20 stsp if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
634 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
635 0136599f 2022-10-20 stsp "short pack file");
636 0136599f 2022-10-20 stsp
637 0136599f 2022-10-20 stsp if (pack->map) {
638 0136599f 2022-10-20 stsp memcpy(&hdr, pack->map, sizeof(hdr));
639 0136599f 2022-10-20 stsp mapoff += sizeof(hdr);
640 0136599f 2022-10-20 stsp } else {
641 0136599f 2022-10-20 stsp r = read(pack->fd, &hdr, sizeof(hdr));
642 0136599f 2022-10-20 stsp if (r == -1)
643 0136599f 2022-10-20 stsp return got_error_from_errno("read");
644 0136599f 2022-10-20 stsp if (r < sizeof(hdr))
645 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
646 0136599f 2022-10-20 stsp "short pack file");
647 0136599f 2022-10-20 stsp }
648 0136599f 2022-10-20 stsp
649 0136599f 2022-10-20 stsp if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
650 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
651 0136599f 2022-10-20 stsp "bad packfile signature");
652 0136599f 2022-10-20 stsp if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
653 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
654 0136599f 2022-10-20 stsp "bad packfile version");
655 0136599f 2022-10-20 stsp nobj = be32toh(hdr.nobjects);
656 0136599f 2022-10-20 stsp if (nobj == 0)
657 0136599f 2022-10-20 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
658 0136599f 2022-10-20 stsp "bad packfile with zero objects");
659 0136599f 2022-10-20 stsp
660 0136599f 2022-10-20 stsp /* We compute the SHA1 of pack file contents and verify later on. */
661 0136599f 2022-10-20 stsp SHA1Init(&ctx);
662 0136599f 2022-10-20 stsp SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
663 0136599f 2022-10-20 stsp
664 0136599f 2022-10-20 stsp /*
665 0136599f 2022-10-20 stsp * Create an in-memory pack index which will grow as objects
666 0136599f 2022-10-20 stsp * IDs in the pack file are discovered. Only fields used to
667 0136599f 2022-10-20 stsp * read deltified objects will be needed by the pack.c library
668 0136599f 2022-10-20 stsp * code, so setting up just a pack index header is sufficient.
669 0136599f 2022-10-20 stsp */
670 0136599f 2022-10-20 stsp memset(&packidx, 0, sizeof(packidx));
671 0136599f 2022-10-20 stsp packidx.hdr.magic = malloc(sizeof(uint32_t));
672 0136599f 2022-10-20 stsp if (packidx.hdr.magic == NULL)
673 0136599f 2022-10-20 stsp return got_error_from_errno("malloc");
674 0136599f 2022-10-20 stsp *packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
675 0136599f 2022-10-20 stsp packidx.hdr.version = malloc(sizeof(uint32_t));
676 0136599f 2022-10-20 stsp if (packidx.hdr.version == NULL) {
677 0136599f 2022-10-20 stsp err = got_error_from_errno("malloc");
678 0136599f 2022-10-20 stsp goto done;
679 0136599f 2022-10-20 stsp }
680 0136599f 2022-10-20 stsp *packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
681 0136599f 2022-10-20 stsp packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
682 0136599f 2022-10-20 stsp sizeof(uint32_t));
683 0136599f 2022-10-20 stsp if (packidx.hdr.fanout_table == NULL) {
684 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
685 0136599f 2022-10-20 stsp goto done;
686 0136599f 2022-10-20 stsp }
687 0136599f 2022-10-20 stsp packidx.hdr.sorted_ids = calloc(nobj,
688 0136599f 2022-10-20 stsp sizeof(struct got_packidx_object_id));
689 0136599f 2022-10-20 stsp if (packidx.hdr.sorted_ids == NULL) {
690 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
691 0136599f 2022-10-20 stsp goto done;
692 0136599f 2022-10-20 stsp }
693 0136599f 2022-10-20 stsp packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
694 0136599f 2022-10-20 stsp if (packidx.hdr.crc32 == NULL) {
695 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
696 0136599f 2022-10-20 stsp goto done;
697 0136599f 2022-10-20 stsp }
698 0136599f 2022-10-20 stsp packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
699 0136599f 2022-10-20 stsp if (packidx.hdr.offsets == NULL) {
700 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
701 0136599f 2022-10-20 stsp goto done;
702 0136599f 2022-10-20 stsp }
703 0136599f 2022-10-20 stsp /* Large offsets table is empty for pack files < 2 GB. */
704 0136599f 2022-10-20 stsp if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
705 0136599f 2022-10-20 stsp packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
706 0136599f 2022-10-20 stsp if (packidx.hdr.large_offsets == NULL) {
707 0136599f 2022-10-20 stsp err = got_error_from_errno("calloc");
708 0136599f 2022-10-20 stsp goto done;
709 0136599f 2022-10-20 stsp }
710 0136599f 2022-10-20 stsp }
711 0136599f 2022-10-20 stsp
712 0136599f 2022-10-20 stsp nvalid = 0;
713 0136599f 2022-10-20 stsp nloose = 0;
714 0136599f 2022-10-20 stsp objects = calloc(nobj, sizeof(struct got_indexed_object));
715 0136599f 2022-10-20 stsp if (objects == NULL)
716 0136599f 2022-10-20 stsp return got_error_from_errno("calloc");
717 0136599f 2022-10-20 stsp
718 0136599f 2022-10-20 stsp /*
719 0136599f 2022-10-20 stsp * First pass: locate all objects and identify un-deltified objects.
720 0136599f 2022-10-20 stsp *
721 0136599f 2022-10-20 stsp * When this pass has completed we will know offset, type, size, and
722 0136599f 2022-10-20 stsp * CRC information for all objects in this pack file. We won't know
723 0136599f 2022-10-20 stsp * any of the actual object IDs of deltified objects yet since we
724 0136599f 2022-10-20 stsp * will not yet attempt to combine deltas.
725 0136599f 2022-10-20 stsp */
726 0136599f 2022-10-20 stsp pass = 1;
727 0136599f 2022-10-20 stsp for (i = 0; i < nobj; i++) {
728 0136599f 2022-10-20 stsp /* Don't send too many progress privsep messages. */
729 0136599f 2022-10-20 stsp p_indexed = ((i + 1) * 100) / nobj;
730 0136599f 2022-10-20 stsp if (p_indexed != last_p_indexed) {
731 0136599f 2022-10-20 stsp err = report_progress(nobj, i + 1, nloose, 0,
732 713d6e11 2022-10-20 stsp rl, progress_cb, progress_arg);
733 0136599f 2022-10-20 stsp if (err)
734 0136599f 2022-10-20 stsp goto done;
735 0136599f 2022-10-20 stsp last_p_indexed = p_indexed;
736 0136599f 2022-10-20 stsp }
737 0136599f 2022-10-20 stsp
738 0136599f 2022-10-20 stsp obj = &objects[i];
739 0136599f 2022-10-20 stsp obj->crc = crc32(0L, NULL, 0);
740 0136599f 2022-10-20 stsp
741 0136599f 2022-10-20 stsp /* Store offset to type+size information for this object. */
742 0136599f 2022-10-20 stsp if (pack->map) {
743 0136599f 2022-10-20 stsp obj->off = mapoff;
744 0136599f 2022-10-20 stsp } else {
745 0136599f 2022-10-20 stsp obj->off = lseek(pack->fd, 0, SEEK_CUR);
746 0136599f 2022-10-20 stsp if (obj->off == -1) {
747 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
748 0136599f 2022-10-20 stsp goto done;
749 0136599f 2022-10-20 stsp }
750 0136599f 2022-10-20 stsp }
751 0136599f 2022-10-20 stsp
752 0136599f 2022-10-20 stsp err = read_packed_object(pack, obj, tmpfile, &ctx);
753 0136599f 2022-10-20 stsp if (err)
754 0136599f 2022-10-20 stsp goto done;
755 0136599f 2022-10-20 stsp
756 0136599f 2022-10-20 stsp if (pack->map) {
757 0136599f 2022-10-20 stsp mapoff += obj->tslen + obj->len;
758 0136599f 2022-10-20 stsp } else {
759 0136599f 2022-10-20 stsp if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
760 0136599f 2022-10-20 stsp SEEK_SET) == -1) {
761 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
762 0136599f 2022-10-20 stsp goto done;
763 0136599f 2022-10-20 stsp }
764 0136599f 2022-10-20 stsp }
765 0136599f 2022-10-20 stsp
766 0136599f 2022-10-20 stsp if (obj->type == GOT_OBJ_TYPE_BLOB ||
767 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_TREE ||
768 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_COMMIT ||
769 0136599f 2022-10-20 stsp obj->type == GOT_OBJ_TYPE_TAG) {
770 0136599f 2022-10-20 stsp obj->valid = 1;
771 0136599f 2022-10-20 stsp nloose++;
772 0136599f 2022-10-20 stsp } else {
773 0136599f 2022-10-20 stsp if (first_delta_idx == -1)
774 0136599f 2022-10-20 stsp first_delta_idx = i;
775 0136599f 2022-10-20 stsp if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
776 0136599f 2022-10-20 stsp have_ref_deltas = 1;
777 0136599f 2022-10-20 stsp }
778 0136599f 2022-10-20 stsp }
779 0136599f 2022-10-20 stsp nvalid = nloose;
780 0136599f 2022-10-20 stsp
781 0136599f 2022-10-20 stsp /*
782 0136599f 2022-10-20 stsp * Having done a full pass over the pack file and can now
783 0136599f 2022-10-20 stsp * verify its checksum.
784 0136599f 2022-10-20 stsp */
785 0136599f 2022-10-20 stsp SHA1Final(pack_sha1, &ctx);
786 13b2bc37 2022-10-23 stsp
787 0136599f 2022-10-20 stsp if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
788 13b2bc37 2022-10-23 stsp err = got_error(GOT_ERR_PACKFILE_CSUM);
789 0136599f 2022-10-20 stsp goto done;
790 0136599f 2022-10-20 stsp }
791 0136599f 2022-10-20 stsp
792 0136599f 2022-10-20 stsp /* Verify the SHA1 checksum stored at the end of the pack file. */
793 0136599f 2022-10-20 stsp if (pack->map) {
794 0136599f 2022-10-20 stsp memcpy(pack_sha1_expected, pack->map +
795 0136599f 2022-10-20 stsp pack->filesize - SHA1_DIGEST_LENGTH,
796 0136599f 2022-10-20 stsp SHA1_DIGEST_LENGTH);
797 0136599f 2022-10-20 stsp } else {
798 0136599f 2022-10-20 stsp ssize_t n;
799 0136599f 2022-10-20 stsp if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
800 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
801 0136599f 2022-10-20 stsp goto done;
802 0136599f 2022-10-20 stsp }
803 0136599f 2022-10-20 stsp n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
804 0136599f 2022-10-20 stsp if (n == -1) {
805 0136599f 2022-10-20 stsp err = got_error_from_errno("read");
806 0136599f 2022-10-20 stsp goto done;
807 0136599f 2022-10-20 stsp }
808 0136599f 2022-10-20 stsp if (n != SHA1_DIGEST_LENGTH) {
809 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_IO);
810 0136599f 2022-10-20 stsp goto done;
811 0136599f 2022-10-20 stsp }
812 0136599f 2022-10-20 stsp }
813 0136599f 2022-10-20 stsp if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
814 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE,
815 0136599f 2022-10-20 stsp "bad checksum in pack file trailer");
816 0136599f 2022-10-20 stsp goto done;
817 0136599f 2022-10-20 stsp }
818 0136599f 2022-10-20 stsp
819 0136599f 2022-10-20 stsp if (first_delta_idx == -1)
820 0136599f 2022-10-20 stsp first_delta_idx = 0;
821 0136599f 2022-10-20 stsp
822 0136599f 2022-10-20 stsp /* In order to resolve ref deltas we need an in-progress pack index. */
823 0136599f 2022-10-20 stsp if (have_ref_deltas)
824 0136599f 2022-10-20 stsp make_packidx(&packidx, nobj, objects);
825 0136599f 2022-10-20 stsp
826 0136599f 2022-10-20 stsp /*
827 0136599f 2022-10-20 stsp * Second pass: We can now resolve deltas to compute the IDs of
828 0136599f 2022-10-20 stsp * objects which appear in deltified form. Because deltas can be
829 0136599f 2022-10-20 stsp * chained this pass may require a couple of iterations until all
830 0136599f 2022-10-20 stsp * IDs of deltified objects have been discovered.
831 0136599f 2022-10-20 stsp */
832 0136599f 2022-10-20 stsp pass++;
833 0136599f 2022-10-20 stsp while (nvalid != nobj) {
834 0136599f 2022-10-20 stsp int n = 0;
835 0136599f 2022-10-20 stsp /*
836 0136599f 2022-10-20 stsp * This loop will only run once unless the pack file
837 0136599f 2022-10-20 stsp * contains ref deltas which refer to objects located
838 0136599f 2022-10-20 stsp * later in the pack file, which is unusual.
839 0136599f 2022-10-20 stsp * Offset deltas can always be resolved in one pass
840 0136599f 2022-10-20 stsp * unless the packfile is corrupt.
841 0136599f 2022-10-20 stsp */
842 0136599f 2022-10-20 stsp for (i = first_delta_idx; i < nobj; i++) {
843 0136599f 2022-10-20 stsp obj = &objects[i];
844 0136599f 2022-10-20 stsp if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
845 0136599f 2022-10-20 stsp obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
846 0136599f 2022-10-20 stsp continue;
847 0136599f 2022-10-20 stsp
848 0136599f 2022-10-20 stsp if (obj->valid)
849 0136599f 2022-10-20 stsp continue;
850 0136599f 2022-10-20 stsp
851 0136599f 2022-10-20 stsp if (pack->map == NULL && lseek(pack->fd,
852 0136599f 2022-10-20 stsp obj->off + obj->tslen, SEEK_SET) == -1) {
853 0136599f 2022-10-20 stsp err = got_error_from_errno("lseek");
854 0136599f 2022-10-20 stsp goto done;
855 0136599f 2022-10-20 stsp }
856 0136599f 2022-10-20 stsp
857 0136599f 2022-10-20 stsp err = resolve_deltified_object(pack, &packidx, obj,
858 0136599f 2022-10-20 stsp tmpfile, delta_base_file, delta_accum_file);
859 0136599f 2022-10-20 stsp if (err) {
860 0136599f 2022-10-20 stsp if (err->code != GOT_ERR_NO_OBJ)
861 0136599f 2022-10-20 stsp goto done;
862 0136599f 2022-10-20 stsp /*
863 0136599f 2022-10-20 stsp * We cannot resolve this object yet because
864 0136599f 2022-10-20 stsp * a delta base is unknown. Try again later.
865 0136599f 2022-10-20 stsp */
866 0136599f 2022-10-20 stsp continue;
867 0136599f 2022-10-20 stsp }
868 0136599f 2022-10-20 stsp
869 0136599f 2022-10-20 stsp obj->valid = 1;
870 0136599f 2022-10-20 stsp n++;
871 0136599f 2022-10-20 stsp if (have_ref_deltas)
872 0136599f 2022-10-20 stsp update_packidx(&packidx, nobj, obj);
873 0136599f 2022-10-20 stsp /* Don't send too many progress privsep messages. */
874 0136599f 2022-10-20 stsp p_resolved = ((nresolved + n) * 100) / nobj;
875 0136599f 2022-10-20 stsp if (p_resolved != last_p_resolved) {
876 0136599f 2022-10-20 stsp err = report_progress(nobj, nobj,
877 713d6e11 2022-10-20 stsp nloose, nresolved + n, rl,
878 0136599f 2022-10-20 stsp progress_cb, progress_arg);
879 0136599f 2022-10-20 stsp if (err)
880 0136599f 2022-10-20 stsp goto done;
881 0136599f 2022-10-20 stsp last_p_resolved = p_resolved;
882 0136599f 2022-10-20 stsp }
883 0136599f 2022-10-20 stsp
884 0136599f 2022-10-20 stsp }
885 0136599f 2022-10-20 stsp if (pass++ > 3 && n == 0) {
886 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE,
887 0136599f 2022-10-20 stsp "could not resolve any of deltas; packfile could "
888 0136599f 2022-10-20 stsp "be corrupt");
889 0136599f 2022-10-20 stsp goto done;
890 0136599f 2022-10-20 stsp }
891 0136599f 2022-10-20 stsp nresolved += n;
892 0136599f 2022-10-20 stsp nvalid += nresolved;
893 0136599f 2022-10-20 stsp }
894 0136599f 2022-10-20 stsp
895 0136599f 2022-10-20 stsp if (nloose + nresolved != nobj) {
896 0136599f 2022-10-20 stsp static char msg[64];
897 0136599f 2022-10-20 stsp snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
898 0136599f 2022-10-20 stsp nloose + nresolved, nobj);
899 0136599f 2022-10-20 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
900 0136599f 2022-10-20 stsp goto done;
901 0136599f 2022-10-20 stsp }
902 0136599f 2022-10-20 stsp
903 0136599f 2022-10-20 stsp err = report_progress(nobj, nobj, nloose, nresolved, NULL,
904 0136599f 2022-10-20 stsp progress_cb, progress_arg);
905 0136599f 2022-10-20 stsp if (err)
906 0136599f 2022-10-20 stsp goto done;
907 0136599f 2022-10-20 stsp
908 0136599f 2022-10-20 stsp make_packidx(&packidx, nobj, objects);
909 0136599f 2022-10-20 stsp
910 0136599f 2022-10-20 stsp free(objects);
911 0136599f 2022-10-20 stsp objects = NULL;
912 0136599f 2022-10-20 stsp
913 0136599f 2022-10-20 stsp SHA1Init(&ctx);
914 0136599f 2022-10-20 stsp putbe32(buf, GOT_PACKIDX_V2_MAGIC);
915 0136599f 2022-10-20 stsp putbe32(buf + 4, GOT_PACKIDX_VERSION);
916 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, buf, 8, &ctx);
917 0136599f 2022-10-20 stsp if (err)
918 0136599f 2022-10-20 stsp goto done;
919 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
920 0136599f 2022-10-20 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
921 0136599f 2022-10-20 stsp if (err)
922 0136599f 2022-10-20 stsp goto done;
923 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
924 0136599f 2022-10-20 stsp nobj * SHA1_DIGEST_LENGTH, &ctx);
925 0136599f 2022-10-20 stsp if (err)
926 0136599f 2022-10-20 stsp goto done;
927 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
928 13b2bc37 2022-10-23 stsp nobj * sizeof(uint32_t), &ctx);
929 0136599f 2022-10-20 stsp if (err)
930 0136599f 2022-10-20 stsp goto done;
931 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
932 13b2bc37 2022-10-23 stsp nobj * sizeof(uint32_t), &ctx);
933 0136599f 2022-10-20 stsp if (err)
934 0136599f 2022-10-20 stsp goto done;
935 0136599f 2022-10-20 stsp if (packidx.nlargeobj > 0) {
936 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
937 0136599f 2022-10-20 stsp packidx.nlargeobj * sizeof(uint64_t), &ctx);
938 0136599f 2022-10-20 stsp if (err)
939 0136599f 2022-10-20 stsp goto done;
940 0136599f 2022-10-20 stsp }
941 13b2bc37 2022-10-23 stsp err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
942 0136599f 2022-10-20 stsp if (err)
943 0136599f 2022-10-20 stsp goto done;
944 0136599f 2022-10-20 stsp
945 0136599f 2022-10-20 stsp SHA1Final(packidx_hash, &ctx);
946 0136599f 2022-10-20 stsp w = write(idxfd, packidx_hash, sizeof(packidx_hash));
947 0136599f 2022-10-20 stsp if (w == -1) {
948 0136599f 2022-10-20 stsp err = got_error_from_errno("write");
949 0136599f 2022-10-20 stsp goto done;
950 0136599f 2022-10-20 stsp }
951 0136599f 2022-10-20 stsp if (w != sizeof(packidx_hash)) {
952 0136599f 2022-10-20 stsp err = got_error(GOT_ERR_IO);
953 0136599f 2022-10-20 stsp goto done;
954 0136599f 2022-10-20 stsp }
955 0136599f 2022-10-20 stsp done:
956 0136599f 2022-10-20 stsp free(objects);
957 0136599f 2022-10-20 stsp free(packidx.hdr.magic);
958 0136599f 2022-10-20 stsp free(packidx.hdr.version);
959 0136599f 2022-10-20 stsp free(packidx.hdr.fanout_table);
960 0136599f 2022-10-20 stsp free(packidx.hdr.sorted_ids);
961 0136599f 2022-10-20 stsp free(packidx.hdr.offsets);
962 0136599f 2022-10-20 stsp free(packidx.hdr.large_offsets);
963 0136599f 2022-10-20 stsp return err;
964 0136599f 2022-10-20 stsp }