2 93658fb9 2020-03-18 stsp * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3 668a20f6 2020-03-18 stsp * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
5 93658fb9 2020-03-18 stsp * Permission to use, copy, modify, and distribute this software for any
6 93658fb9 2020-03-18 stsp * purpose with or without fee is hereby granted, provided that the above
7 93658fb9 2020-03-18 stsp * copyright notice and this permission notice appear in all copies.
9 93658fb9 2020-03-18 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 93658fb9 2020-03-18 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 93658fb9 2020-03-18 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 93658fb9 2020-03-18 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 93658fb9 2020-03-18 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 93658fb9 2020-03-18 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 93658fb9 2020-03-18 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 93658fb9 2020-03-18 stsp #include <sys/queue.h>
19 93658fb9 2020-03-18 stsp #include <sys/stat.h>
20 93658fb9 2020-03-18 stsp #include <sys/syslimits.h>
21 93658fb9 2020-03-18 stsp #include <sys/time.h>
22 93658fb9 2020-03-18 stsp #include <sys/types.h>
23 93658fb9 2020-03-18 stsp #include <sys/uio.h>
25 93658fb9 2020-03-18 stsp #include <stdint.h>
26 93658fb9 2020-03-18 stsp #include <errno.h>
27 93658fb9 2020-03-18 stsp #include <imsg.h>
28 93658fb9 2020-03-18 stsp #include <limits.h>
29 93658fb9 2020-03-18 stsp #include <signal.h>
30 93658fb9 2020-03-18 stsp #include <stdio.h>
31 93658fb9 2020-03-18 stsp #include <stdlib.h>
32 93658fb9 2020-03-18 stsp #include <string.h>
33 93658fb9 2020-03-18 stsp #include <ctype.h>
34 93658fb9 2020-03-18 stsp #include <sha1.h>
35 93658fb9 2020-03-18 stsp #include <fcntl.h>
36 93658fb9 2020-03-18 stsp #include <zlib.h>
37 93658fb9 2020-03-18 stsp #include <err.h>
38 93658fb9 2020-03-18 stsp #include <assert.h>
39 93658fb9 2020-03-18 stsp #include <dirent.h>
41 93658fb9 2020-03-18 stsp #include "got_error.h"
42 93658fb9 2020-03-18 stsp #include "got_object.h"
44 93658fb9 2020-03-18 stsp #include "got_lib_sha1.h"
45 93658fb9 2020-03-18 stsp #include "got_lib_delta.h"
46 93658fb9 2020-03-18 stsp #include "got_lib_inflate.h"
47 93658fb9 2020-03-18 stsp #include "got_lib_object.h"
48 93658fb9 2020-03-18 stsp #include "got_lib_object_parse.h"
49 93658fb9 2020-03-18 stsp #include "got_lib_object_idset.h"
50 93658fb9 2020-03-18 stsp #include "got_lib_privsep.h"
51 668a20f6 2020-03-18 stsp #include "got_lib_pack.h"
52 668a20f6 2020-03-18 stsp #include "got_lib_delta_cache.h"
54 668a20f6 2020-03-18 stsp struct got_indexed_object {
55 668a20f6 2020-03-18 stsp struct got_object_id id;
58 668a20f6 2020-03-18 stsp * Has this object been fully resolved?
59 668a20f6 2020-03-18 stsp * If so, we know its ID, otherwise we don't and 'id' is invalid.
63 668a20f6 2020-03-18 stsp /* Offset of type+size field for this object in pack file. */
66 668a20f6 2020-03-18 stsp /* Type+size values parsed from pack file. */
67 668a20f6 2020-03-18 stsp uint8_t type;
68 668a20f6 2020-03-18 stsp uint64_t size;
70 668a20f6 2020-03-18 stsp /* Length of on-disk type+size data. */
71 668a20f6 2020-03-18 stsp size_t tslen;
73 668a20f6 2020-03-18 stsp /* Length of object data following type+size. */
76 668a20f6 2020-03-18 stsp uint32_t crc;
80 836f2c92 2020-03-18 stsp /* For ref deltas. */
81 836f2c92 2020-03-18 stsp struct got_object_id ref_id;
84 836f2c92 2020-03-18 stsp /* For offset deltas. */
85 836f2c92 2020-03-18 stsp off_t base_offset;
86 836f2c92 2020-03-18 stsp size_t base_offsetlen;
91 93658fb9 2020-03-18 stsp #define PUTBE32(b, n)\
93 93658fb9 2020-03-18 stsp (b)[0] = (n) >> 24; \
94 93658fb9 2020-03-18 stsp (b)[1] = (n) >> 16; \
95 93658fb9 2020-03-18 stsp (b)[2] = (n) >> 8; \
96 93658fb9 2020-03-18 stsp (b)[3] = (n) >> 0; \
99 93658fb9 2020-03-18 stsp #define PUTBE64(b, n)\
101 93658fb9 2020-03-18 stsp (b)[0] = (n) >> 56; \
102 93658fb9 2020-03-18 stsp (b)[1] = (n) >> 48; \
103 93658fb9 2020-03-18 stsp (b)[2] = (n) >> 40; \
104 93658fb9 2020-03-18 stsp (b)[3] = (n) >> 32; \
105 93658fb9 2020-03-18 stsp (b)[4] = (n) >> 24; \
106 93658fb9 2020-03-18 stsp (b)[5] = (n) >> 16; \
107 93658fb9 2020-03-18 stsp (b)[6] = (n) >> 8; \
108 93658fb9 2020-03-18 stsp (b)[7] = (n) >> 0; \
111 668a20f6 2020-03-18 stsp static const struct got_error *
112 668a20f6 2020-03-18 stsp get_obj_type_label(const char **label, int obj_type)
114 668a20f6 2020-03-18 stsp const struct got_error *err = NULL;
116 668a20f6 2020-03-18 stsp switch (obj_type) {
117 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_BLOB:
118 668a20f6 2020-03-18 stsp *label = GOT_OBJ_LABEL_BLOB;
120 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_TREE:
121 668a20f6 2020-03-18 stsp *label = GOT_OBJ_LABEL_TREE;
123 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_COMMIT:
124 668a20f6 2020-03-18 stsp *label = GOT_OBJ_LABEL_COMMIT;
126 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_TAG:
127 668a20f6 2020-03-18 stsp *label = GOT_OBJ_LABEL_TAG;
130 668a20f6 2020-03-18 stsp *label = NULL;
131 668a20f6 2020-03-18 stsp err = got_error(GOT_ERR_OBJ_TYPE);
135 668a20f6 2020-03-18 stsp return err;
138 1e87a3c3 2020-03-18 stsp static const struct got_error *
139 1e87a3c3 2020-03-18 stsp read_crc(uint32_t *crc, int fd, size_t len)
141 1e87a3c3 2020-03-18 stsp uint8_t buf[8192];
145 1e87a3c3 2020-03-18 stsp for (n = len; n > 0; n -= r){
146 1e87a3c3 2020-03-18 stsp r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
147 1e87a3c3 2020-03-18 stsp if (r == -1)
148 1e87a3c3 2020-03-18 stsp return got_error_from_errno("read");
149 1e87a3c3 2020-03-18 stsp if (r == 0)
151 1e87a3c3 2020-03-18 stsp *crc = crc32(*crc, buf, r);
154 1e87a3c3 2020-03-18 stsp return NULL;
157 668a20f6 2020-03-18 stsp static const struct got_error *
158 668a20f6 2020-03-18 stsp read_packed_object(struct got_pack *pack, struct got_indexed_object *obj)
160 668a20f6 2020-03-18 stsp const struct got_error *err = NULL;
161 668a20f6 2020-03-18 stsp SHA1_CTX ctx;
162 668a20f6 2020-03-18 stsp uint8_t *data;
163 668a20f6 2020-03-18 stsp size_t datalen;
165 668a20f6 2020-03-18 stsp char *header;
166 668a20f6 2020-03-18 stsp size_t headerlen;
167 668a20f6 2020-03-18 stsp const char *obj_label;
169 cbc66309 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
170 cbc66309 2020-03-18 stsp &obj->tslen, pack, obj->off);
172 668a20f6 2020-03-18 stsp return err;
174 1e87a3c3 2020-03-18 stsp /* XXX Seek back and get the CRC of on-disk type+size bytes. */
175 1e87a3c3 2020-03-18 stsp if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
176 1e87a3c3 2020-03-18 stsp return got_error_from_errno("lseek");
177 1e87a3c3 2020-03-18 stsp err = read_crc(&obj->crc, pack->fd, obj->tslen);
179 1e87a3c3 2020-03-18 stsp return err;
181 668a20f6 2020-03-18 stsp switch (obj->type) {
182 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_BLOB:
183 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_COMMIT:
184 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_TREE:
185 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_TAG:
186 668a20f6 2020-03-18 stsp /* XXX TODO reading large objects into memory is bad! */
187 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(&data, &datalen, &obj->len,
188 55fdd257 2020-03-18 stsp &obj->crc, obj->size, pack->fd);
191 668a20f6 2020-03-18 stsp SHA1Init(&ctx);
192 668a20f6 2020-03-18 stsp err = get_obj_type_label(&obj_label, obj->type);
194 7cd14ea0 2020-03-18 stsp free(data);
197 668a20f6 2020-03-18 stsp if (asprintf(&header, "%s %lld", obj_label, obj->size) == -1) {
198 668a20f6 2020-03-18 stsp err = got_error_from_errno("asprintf");
199 668a20f6 2020-03-18 stsp free(data);
202 668a20f6 2020-03-18 stsp headerlen = strlen(header) + 1;
203 668a20f6 2020-03-18 stsp SHA1Update(&ctx, header, headerlen);
204 668a20f6 2020-03-18 stsp SHA1Update(&ctx, data, datalen);
205 668a20f6 2020-03-18 stsp SHA1Final(obj->id.sha1, &ctx);
206 668a20f6 2020-03-18 stsp free(header);
207 668a20f6 2020-03-18 stsp free(data);
209 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_REF_DELTA:
210 668a20f6 2020-03-18 stsp memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
211 836f2c92 2020-03-18 stsp n = read(pack->fd, &obj->delta.ref.ref_id.sha1,
212 836f2c92 2020-03-18 stsp SHA1_DIGEST_LENGTH);
213 668a20f6 2020-03-18 stsp if (n == -1) {
214 668a20f6 2020-03-18 stsp err = got_error_from_errno("read");
217 668a20f6 2020-03-18 stsp if (n < sizeof(obj->id)) {
218 668a20f6 2020-03-18 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
221 836f2c92 2020-03-18 stsp obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
222 1e87a3c3 2020-03-18 stsp SHA1_DIGEST_LENGTH);
223 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
224 55fdd257 2020-03-18 stsp &obj->crc, obj->size, pack->fd);
227 668a20f6 2020-03-18 stsp obj->len += SHA1_DIGEST_LENGTH;
229 668a20f6 2020-03-18 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
230 668a20f6 2020-03-18 stsp memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
231 836f2c92 2020-03-18 stsp err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
232 836f2c92 2020-03-18 stsp &obj->delta.ofs.base_offsetlen, pack, obj->off,
233 836f2c92 2020-03-18 stsp obj->tslen);
237 1e87a3c3 2020-03-18 stsp /* XXX Seek back and get the CRC of on-disk offset bytes. */
238 1e87a3c3 2020-03-18 stsp if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET) == -1) {
239 1e87a3c3 2020-03-18 stsp err = got_error_from_errno("lseek");
242 836f2c92 2020-03-18 stsp err = read_crc(&obj->crc, pack->fd,
243 836f2c92 2020-03-18 stsp obj->delta.ofs.base_offsetlen);
247 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
248 55fdd257 2020-03-18 stsp &obj->crc, obj->size, pack->fd);
251 836f2c92 2020-03-18 stsp obj->len += obj->delta.ofs.base_offsetlen;
254 668a20f6 2020-03-18 stsp err = got_error(GOT_ERR_OBJ_TYPE);
258 668a20f6 2020-03-18 stsp return err;
261 668a20f6 2020-03-18 stsp static const struct got_error *
262 668a20f6 2020-03-18 stsp hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
266 668a20f6 2020-03-18 stsp SHA1Update(ctx, buf, len);
268 668a20f6 2020-03-18 stsp w = write(fd, buf, len);
269 668a20f6 2020-03-18 stsp if (w == -1)
270 668a20f6 2020-03-18 stsp return got_error_from_errno("write");
271 668a20f6 2020-03-18 stsp if (w != len)
272 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_IO);
274 668a20f6 2020-03-18 stsp return NULL;
277 668a20f6 2020-03-18 stsp static const struct got_error *
278 668a20f6 2020-03-18 stsp resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
279 668a20f6 2020-03-18 stsp struct got_indexed_object *obj)
281 668a20f6 2020-03-18 stsp const struct got_error *err = NULL;
282 668a20f6 2020-03-18 stsp struct got_delta_chain deltas;
283 668a20f6 2020-03-18 stsp struct got_delta *delta;
284 668a20f6 2020-03-18 stsp uint8_t *buf = NULL;
285 668a20f6 2020-03-18 stsp size_t len;
286 668a20f6 2020-03-18 stsp SHA1_CTX ctx;
287 668a20f6 2020-03-18 stsp char *header;
288 668a20f6 2020-03-18 stsp size_t headerlen;
289 668a20f6 2020-03-18 stsp int base_obj_type;
290 668a20f6 2020-03-18 stsp const char *obj_label;
292 668a20f6 2020-03-18 stsp deltas.nentries = 0;
293 668a20f6 2020-03-18 stsp SIMPLEQ_INIT(&deltas.entries);
295 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
296 668a20f6 2020-03-18 stsp obj->off, obj->tslen, obj->type, obj->size,
297 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
301 668a20f6 2020-03-18 stsp /* XXX TODO reading large objects into memory is bad! */
302 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(&buf, &len, &deltas, pack);
306 668a20f6 2020-03-18 stsp SHA1Init(&ctx);
308 668a20f6 2020-03-18 stsp err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
311 668a20f6 2020-03-18 stsp err = get_obj_type_label(&obj_label, base_obj_type);
314 668a20f6 2020-03-18 stsp if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
315 668a20f6 2020-03-18 stsp err = got_error_from_errno("asprintf");
318 668a20f6 2020-03-18 stsp headerlen = strlen(header) + 1;
319 668a20f6 2020-03-18 stsp SHA1Update(&ctx, header, headerlen);
320 668a20f6 2020-03-18 stsp SHA1Update(&ctx, buf, len);
321 668a20f6 2020-03-18 stsp SHA1Final(obj->id.sha1, &ctx);
324 668a20f6 2020-03-18 stsp while (!SIMPLEQ_EMPTY(&deltas.entries)) {
325 668a20f6 2020-03-18 stsp delta = SIMPLEQ_FIRST(&deltas.entries);
326 668a20f6 2020-03-18 stsp SIMPLEQ_REMOVE_HEAD(&deltas.entries, entry);
327 668a20f6 2020-03-18 stsp free(delta);
329 668a20f6 2020-03-18 stsp return err;
332 668a20f6 2020-03-18 stsp /* Determine the slot in the pack index a given object ID should use. */
334 668a20f6 2020-03-18 stsp find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
336 668a20f6 2020-03-18 stsp u_int8_t id0 = sha1[0];
337 668a20f6 2020-03-18 stsp uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
338 668a20f6 2020-03-18 stsp int left = 0, right = nindexed - 1;
339 668a20f6 2020-03-18 stsp int cmp = 0, i = 0;
341 668a20f6 2020-03-18 stsp if (id0 > 0)
342 668a20f6 2020-03-18 stsp left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
344 668a20f6 2020-03-18 stsp while (left <= right) {
345 668a20f6 2020-03-18 stsp struct got_packidx_object_id *oid;
347 668a20f6 2020-03-18 stsp i = ((left + right) / 2);
348 668a20f6 2020-03-18 stsp oid = &packidx->hdr.sorted_ids[i];
350 668a20f6 2020-03-18 stsp cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
351 668a20f6 2020-03-18 stsp if (cmp == 0)
352 668a20f6 2020-03-18 stsp return -1; /* object already indexed */
353 668a20f6 2020-03-18 stsp else if (cmp > 0)
354 668a20f6 2020-03-18 stsp left = i + 1;
355 668a20f6 2020-03-18 stsp else if (cmp < 0)
356 668a20f6 2020-03-18 stsp right = i - 1;
359 668a20f6 2020-03-18 stsp return left;
363 668a20f6 2020-03-18 stsp static void
364 668a20f6 2020-03-18 stsp print_packidx(struct got_packidx *packidx)
366 668a20f6 2020-03-18 stsp uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
369 668a20f6 2020-03-18 stsp fprintf(stderr, "object IDs:\n");
370 668a20f6 2020-03-18 stsp for (i = 0; i < nindexed; i++) {
371 668a20f6 2020-03-18 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
372 668a20f6 2020-03-18 stsp got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
373 668a20f6 2020-03-18 stsp hex, sizeof(hex));
374 668a20f6 2020-03-18 stsp fprintf(stderr, "%s\n", hex);
376 668a20f6 2020-03-18 stsp fprintf(stderr, "\n");
378 668a20f6 2020-03-18 stsp fprintf(stderr, "object offsets:\n");
379 668a20f6 2020-03-18 stsp for (i = 0; i < nindexed; i++) {
380 668a20f6 2020-03-18 stsp uint32_t offset = be32toh(packidx->hdr.offsets[i]);
381 668a20f6 2020-03-18 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
382 668a20f6 2020-03-18 stsp int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
383 668a20f6 2020-03-18 stsp fprintf(stderr, "%u -> %llu\n", offset,
384 668a20f6 2020-03-18 stsp be64toh(packidx->hdr.large_offsets[j]));
386 668a20f6 2020-03-18 stsp fprintf(stderr, "%u\n", offset);
388 668a20f6 2020-03-18 stsp fprintf(stderr, "\n");
390 668a20f6 2020-03-18 stsp fprintf(stderr, "fanout table:");
391 668a20f6 2020-03-18 stsp for (i = 0; i <= 0xff; i++)
392 668a20f6 2020-03-18 stsp fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
393 668a20f6 2020-03-18 stsp fprintf(stderr, "\n");
397 668a20f6 2020-03-18 stsp static void
398 950de2cd 2020-03-18 stsp add_indexed_object(struct got_packidx *packidx, uint32_t idx,
399 668a20f6 2020-03-18 stsp struct got_indexed_object *obj)
403 668a20f6 2020-03-18 stsp memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
404 668a20f6 2020-03-18 stsp SHA1_DIGEST_LENGTH);
405 668a20f6 2020-03-18 stsp if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
406 668a20f6 2020-03-18 stsp packidx->hdr.offsets[idx] = htobe32(obj->off);
408 7bad1537 2020-03-18 stsp packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
409 668a20f6 2020-03-18 stsp GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
410 7bad1537 2020-03-18 stsp packidx->hdr.large_offsets[packidx->nlargeobj] =
411 7bad1537 2020-03-18 stsp htobe64(obj->off);
412 7bad1537 2020-03-18 stsp packidx->nlargeobj++;
415 668a20f6 2020-03-18 stsp for (i = obj->id.sha1[0]; i <= 0xff; i++) {
416 950de2cd 2020-03-18 stsp uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
417 668a20f6 2020-03-18 stsp packidx->hdr.fanout_table[i] = htobe32(n + 1);
422 8bb2b40c 2020-03-18 stsp indexed_obj_cmp(const void *pa, const void *pb)
424 8bb2b40c 2020-03-18 stsp struct got_indexed_object *a, *b;
426 8bb2b40c 2020-03-18 stsp a = *(struct got_indexed_object **)pa;
427 8bb2b40c 2020-03-18 stsp b = *(struct got_indexed_object **)pb;
428 8bb2b40c 2020-03-18 stsp return got_object_id_cmp(&a->id, &b->id);
431 950de2cd 2020-03-18 stsp static void
432 0fd91daf 2020-03-18 stsp make_packidx(struct got_packidx *packidx, int nobj,
433 950de2cd 2020-03-18 stsp struct got_indexed_object **objects)
435 950de2cd 2020-03-18 stsp struct got_indexed_object *obj;
437 950de2cd 2020-03-18 stsp uint32_t idx = 0;
439 950de2cd 2020-03-18 stsp mergesort(objects, nobj, sizeof(struct got_indexed_object *),
440 950de2cd 2020-03-18 stsp indexed_obj_cmp);
442 0fd91daf 2020-03-18 stsp memset(packidx->hdr.fanout_table, 0,
443 0fd91daf 2020-03-18 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
444 0fd91daf 2020-03-18 stsp packidx->nlargeobj = 0;
446 950de2cd 2020-03-18 stsp for (i = 0; i < nobj; i++) {
447 950de2cd 2020-03-18 stsp obj = objects[i];
448 0fd91daf 2020-03-18 stsp if (obj->valid)
449 0fd91daf 2020-03-18 stsp add_indexed_object(packidx, idx++, obj);
453 950de2cd 2020-03-18 stsp static void
454 950de2cd 2020-03-18 stsp update_packidx(struct got_packidx *packidx, int nobj,
455 950de2cd 2020-03-18 stsp struct got_indexed_object *obj)
457 950de2cd 2020-03-18 stsp uint32_t idx;
458 950de2cd 2020-03-18 stsp uint32_t nindexed = betoh32(packidx->hdr.fanout_table[0xff]);
460 950de2cd 2020-03-18 stsp idx = find_object_idx(packidx, obj->id.sha1);
461 950de2cd 2020-03-18 stsp if (idx == -1) {
462 950de2cd 2020-03-18 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
463 950de2cd 2020-03-18 stsp got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
464 950de2cd 2020-03-18 stsp return; /* object already indexed */
467 950de2cd 2020-03-18 stsp memmove(&packidx->hdr.sorted_ids[idx + 1],
468 950de2cd 2020-03-18 stsp &packidx->hdr.sorted_ids[idx],
469 950de2cd 2020-03-18 stsp sizeof(struct got_packidx_object_id) * (nindexed - idx));
470 950de2cd 2020-03-18 stsp memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
471 950de2cd 2020-03-18 stsp sizeof(uint32_t) * (nindexed - idx));
473 950de2cd 2020-03-18 stsp add_indexed_object(packidx, idx, obj);
476 668a20f6 2020-03-18 stsp static const struct got_error *
477 668a20f6 2020-03-18 stsp index_pack(struct got_pack *pack, int idxfd, uint8_t *pack_hash,
478 668a20f6 2020-03-18 stsp struct imsgbuf *ibuf)
480 668a20f6 2020-03-18 stsp const struct got_error *err;
481 668a20f6 2020-03-18 stsp struct got_packfile_hdr hdr;
482 668a20f6 2020-03-18 stsp struct got_packidx packidx;
483 668a20f6 2020-03-18 stsp char buf[8];
484 7bad1537 2020-03-18 stsp int nobj, nvalid, nloose, nresolved = 0, i;
485 668a20f6 2020-03-18 stsp struct got_indexed_object **objects = NULL, *obj;
486 668a20f6 2020-03-18 stsp SHA1_CTX ctx;
487 668a20f6 2020-03-18 stsp uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
488 668a20f6 2020-03-18 stsp ssize_t r, w;
489 0fd91daf 2020-03-18 stsp int pass, have_ref_deltas = 0;
491 668a20f6 2020-03-18 stsp /* Check pack file header. */
492 668a20f6 2020-03-18 stsp r = read(pack->fd, &hdr, sizeof(hdr));
493 668a20f6 2020-03-18 stsp if (r == -1)
494 668a20f6 2020-03-18 stsp return got_error_from_errno("read");
495 668a20f6 2020-03-18 stsp if (r < sizeof(hdr))
496 668a20f6 2020-03-18 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
497 668a20f6 2020-03-18 stsp "short packfile header");
499 668a20f6 2020-03-18 stsp if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
500 668a20f6 2020-03-18 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
501 668a20f6 2020-03-18 stsp "bad packfile signature");
502 668a20f6 2020-03-18 stsp if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
503 668a20f6 2020-03-18 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
504 668a20f6 2020-03-18 stsp "bad packfile version");
505 668a20f6 2020-03-18 stsp nobj = betoh32(hdr.nobjects);
506 668a20f6 2020-03-18 stsp if (nobj == 0)
507 668a20f6 2020-03-18 stsp return got_error_msg(GOT_ERR_BAD_PACKFILE,
508 668a20f6 2020-03-18 stsp "bad packfile with zero objects");
511 668a20f6 2020-03-18 stsp * Create an in-memory pack index which will grow as objects
512 668a20f6 2020-03-18 stsp * IDs in the pack file are discovered. Only fields used to
513 668a20f6 2020-03-18 stsp * read deltified objects will be needed by the pack.c library
514 668a20f6 2020-03-18 stsp * code, so setting up just a pack index header is sufficient.
516 668a20f6 2020-03-18 stsp memset(&packidx, 0, sizeof(packidx));
517 668a20f6 2020-03-18 stsp packidx.hdr.magic = malloc(sizeof(uint32_t));
518 668a20f6 2020-03-18 stsp if (packidx.hdr.magic == NULL)
519 668a20f6 2020-03-18 stsp return got_error_from_errno("calloc");
520 668a20f6 2020-03-18 stsp *packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
521 668a20f6 2020-03-18 stsp packidx.hdr.version = malloc(sizeof(uint32_t));
522 668a20f6 2020-03-18 stsp if (packidx.hdr.version == NULL) {
523 668a20f6 2020-03-18 stsp err = got_error_from_errno("malloc");
526 668a20f6 2020-03-18 stsp *packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
527 668a20f6 2020-03-18 stsp packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
528 668a20f6 2020-03-18 stsp sizeof(uint32_t));
529 668a20f6 2020-03-18 stsp if (packidx.hdr.fanout_table == NULL) {
530 668a20f6 2020-03-18 stsp err = got_error_from_errno("calloc");
533 668a20f6 2020-03-18 stsp packidx.hdr.sorted_ids = calloc(nobj,
534 668a20f6 2020-03-18 stsp sizeof(struct got_packidx_object_id));
535 668a20f6 2020-03-18 stsp if (packidx.hdr.sorted_ids == NULL) {
536 668a20f6 2020-03-18 stsp err = got_error_from_errno("calloc");
539 668a20f6 2020-03-18 stsp packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
540 668a20f6 2020-03-18 stsp if (packidx.hdr.offsets == NULL) {
541 668a20f6 2020-03-18 stsp err = got_error_from_errno("calloc");
544 668a20f6 2020-03-18 stsp /* Large offsets table is empty for pack files < 2 GB. */
545 668a20f6 2020-03-18 stsp if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
546 668a20f6 2020-03-18 stsp packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
547 668a20f6 2020-03-18 stsp if (packidx.hdr.large_offsets == NULL) {
548 668a20f6 2020-03-18 stsp err = got_error_from_errno("calloc");
553 668a20f6 2020-03-18 stsp nvalid = 0;
554 668a20f6 2020-03-18 stsp nloose = 0;
555 668a20f6 2020-03-18 stsp objects = calloc(nobj, sizeof(struct got_indexed_object *));
556 668a20f6 2020-03-18 stsp if (objects == NULL)
557 668a20f6 2020-03-18 stsp return got_error_from_errno("calloc");
560 668a20f6 2020-03-18 stsp * First pass: locate all objects and identify un-deltified objects.
562 668a20f6 2020-03-18 stsp * When this pass has completed we will know offset, type, size, and
563 668a20f6 2020-03-18 stsp * CRC information for all objects in this pack file. We won't know
564 668a20f6 2020-03-18 stsp * any of the actual object IDs of deltified objects yet since we
565 668a20f6 2020-03-18 stsp * will not yet attempt to combine deltas.
568 668a20f6 2020-03-18 stsp for (i = 0; i < nobj; i++) {
569 668a20f6 2020-03-18 stsp err = got_privsep_send_index_pack_progress(ibuf, nobj, i + 1,
570 668a20f6 2020-03-18 stsp nloose, 0);
574 668a20f6 2020-03-18 stsp obj = calloc(1, sizeof(*obj));
575 668a20f6 2020-03-18 stsp if (obj == NULL) {
576 668a20f6 2020-03-18 stsp err = got_error_from_errno("calloc");
579 1e87a3c3 2020-03-18 stsp obj->crc = crc32(0L, NULL, 0);
581 668a20f6 2020-03-18 stsp /* Store offset to type+size information for this object. */
582 668a20f6 2020-03-18 stsp obj->off = lseek(pack->fd, 0, SEEK_CUR);
583 668a20f6 2020-03-18 stsp if (obj->off == -1) {
584 668a20f6 2020-03-18 stsp err = got_error_from_errno("lseek");
588 668a20f6 2020-03-18 stsp err = read_packed_object(pack, obj);
592 668a20f6 2020-03-18 stsp objects[i] = obj;
594 1e87a3c3 2020-03-18 stsp if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
595 1e87a3c3 2020-03-18 stsp SEEK_SET) == -1) {
596 1e87a3c3 2020-03-18 stsp err = got_error_from_errno("lseek");
600 668a20f6 2020-03-18 stsp if (obj->type == GOT_OBJ_TYPE_BLOB ||
601 668a20f6 2020-03-18 stsp obj->type == GOT_OBJ_TYPE_TREE ||
602 668a20f6 2020-03-18 stsp obj->type == GOT_OBJ_TYPE_COMMIT ||
603 668a20f6 2020-03-18 stsp obj->type == GOT_OBJ_TYPE_TAG) {
604 668a20f6 2020-03-18 stsp objects[i]->valid = 1;
606 0fd91daf 2020-03-18 stsp } else if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
607 0fd91daf 2020-03-18 stsp have_ref_deltas = 1;
609 668a20f6 2020-03-18 stsp nvalid = nloose;
611 0fd91daf 2020-03-18 stsp /* In order to resolve ref deltas we need an in-progress pack index. */
612 0fd91daf 2020-03-18 stsp if (have_ref_deltas)
613 0fd91daf 2020-03-18 stsp make_packidx(&packidx, nobj, objects);
616 668a20f6 2020-03-18 stsp * Second pass: We can now resolve deltas to compute the IDs of
617 668a20f6 2020-03-18 stsp * objects which appear in deltified form. Because deltas can be
618 668a20f6 2020-03-18 stsp * chained this pass may require a couple of iterations until all
619 668a20f6 2020-03-18 stsp * IDs of deltified objects have been discovered.
622 668a20f6 2020-03-18 stsp while (nvalid != nobj) {
624 668a20f6 2020-03-18 stsp for (i = 0; i < nobj; i++) {
625 668a20f6 2020-03-18 stsp if (objects[i]->type != GOT_OBJ_TYPE_REF_DELTA &&
626 668a20f6 2020-03-18 stsp objects[i]->type != GOT_OBJ_TYPE_OFFSET_DELTA)
629 668a20f6 2020-03-18 stsp if (objects[i]->valid)
632 668a20f6 2020-03-18 stsp obj = objects[i];
633 cbc66309 2020-03-18 stsp if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
635 668a20f6 2020-03-18 stsp err = got_error_from_errno("lseek");
639 668a20f6 2020-03-18 stsp err = resolve_deltified_object(pack, &packidx, obj);
641 668a20f6 2020-03-18 stsp if (err->code != GOT_ERR_NO_OBJ)
644 668a20f6 2020-03-18 stsp * We cannot resolve this object yet because
645 668a20f6 2020-03-18 stsp * a delta base is unknown. Try again later.
650 668a20f6 2020-03-18 stsp objects[i]->valid = 1;
652 0fd91daf 2020-03-18 stsp if (have_ref_deltas)
653 0fd91daf 2020-03-18 stsp update_packidx(&packidx, nobj, obj);
654 cbc66309 2020-03-18 stsp err = got_privsep_send_index_pack_progress(ibuf,
655 cbc66309 2020-03-18 stsp nobj, nobj, nloose, nresolved + n);
660 668a20f6 2020-03-18 stsp if (pass++ > 3 && n == 0) {
661 668a20f6 2020-03-18 stsp static char msg[64];
662 668a20f6 2020-03-18 stsp snprintf(msg, sizeof(msg), "could not resolve "
663 668a20f6 2020-03-18 stsp "any of deltas; packfile could be corrupt");
664 668a20f6 2020-03-18 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
668 668a20f6 2020-03-18 stsp if (nloose + nresolved == nobj) {
669 668a20f6 2020-03-18 stsp static char msg[64];
670 668a20f6 2020-03-18 stsp snprintf(msg, sizeof(msg),
671 cbc66309 2020-03-18 stsp "fix point reached too early: %d/%d/%d",
672 cbc66309 2020-03-18 stsp nvalid, nresolved, nobj);
673 668a20f6 2020-03-18 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
676 668a20f6 2020-03-18 stsp nresolved += n;
677 668a20f6 2020-03-18 stsp nvalid += nresolved;
680 668a20f6 2020-03-18 stsp if (nloose + nresolved != nobj) {
681 668a20f6 2020-03-18 stsp static char msg[64];
682 668a20f6 2020-03-18 stsp snprintf(msg, sizeof(msg),
683 cbc66309 2020-03-18 stsp "discovered only %d of %d objects",
684 cbc66309 2020-03-18 stsp nloose + nresolved, nobj);
685 668a20f6 2020-03-18 stsp err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
689 0fd91daf 2020-03-18 stsp make_packidx(&packidx, nobj, objects);
691 93658fb9 2020-03-18 stsp SHA1Init(&ctx);
692 668a20f6 2020-03-18 stsp err = hwrite(idxfd, "\xfftOc\x00\x00\x00\x02", 8, &ctx);
695 668a20f6 2020-03-18 stsp err = hwrite(idxfd, packidx.hdr.fanout_table,
696 668a20f6 2020-03-18 stsp GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
699 668a20f6 2020-03-18 stsp err = hwrite(idxfd, packidx.hdr.sorted_ids,
700 668a20f6 2020-03-18 stsp nobj * SHA1_DIGEST_LENGTH, &ctx);
703 93658fb9 2020-03-18 stsp for(i = 0; i < nobj; i++){
704 93658fb9 2020-03-18 stsp PUTBE32(buf, objects[i]->crc);
705 668a20f6 2020-03-18 stsp err = hwrite(idxfd, buf, 4, &ctx);
709 cbc66309 2020-03-18 stsp err = hwrite(idxfd, packidx.hdr.offsets, nobj * sizeof(uint32_t),
713 7bad1537 2020-03-18 stsp if (packidx.nlargeobj > 0) {
714 668a20f6 2020-03-18 stsp err = hwrite(idxfd, packidx.hdr.large_offsets,
715 7bad1537 2020-03-18 stsp packidx.nlargeobj * sizeof(uint64_t), &ctx);
719 668a20f6 2020-03-18 stsp err = hwrite(idxfd, pack_hash, SHA1_DIGEST_LENGTH, &ctx);
723 668a20f6 2020-03-18 stsp SHA1Final(packidx_hash, &ctx);
724 668a20f6 2020-03-18 stsp w = write(idxfd, packidx_hash, sizeof(packidx_hash));
725 668a20f6 2020-03-18 stsp if (w == -1) {
726 668a20f6 2020-03-18 stsp err = got_error_from_errno("write");
729 668a20f6 2020-03-18 stsp if (w != sizeof(packidx_hash)) {
730 668a20f6 2020-03-18 stsp err = got_error(GOT_ERR_IO);
734 668a20f6 2020-03-18 stsp free(packidx.hdr.magic);
735 668a20f6 2020-03-18 stsp free(packidx.hdr.version);
736 668a20f6 2020-03-18 stsp free(packidx.hdr.fanout_table);
737 668a20f6 2020-03-18 stsp free(packidx.hdr.sorted_ids);
738 668a20f6 2020-03-18 stsp free(packidx.hdr.offsets);
739 668a20f6 2020-03-18 stsp free(packidx.hdr.large_offsets);
740 668a20f6 2020-03-18 stsp return err;
744 93658fb9 2020-03-18 stsp main(int argc, char **argv)
746 668a20f6 2020-03-18 stsp const struct got_error *err = NULL, *close_err;
747 93658fb9 2020-03-18 stsp struct imsgbuf ibuf;
748 93658fb9 2020-03-18 stsp struct imsg imsg;
749 668a20f6 2020-03-18 stsp int idxfd = -1;
750 668a20f6 2020-03-18 stsp struct got_pack pack;
751 668a20f6 2020-03-18 stsp uint8_t pack_hash[SHA1_DIGEST_LENGTH];
752 668a20f6 2020-03-18 stsp off_t packfile_size;
754 668a20f6 2020-03-18 stsp static int attached;
755 668a20f6 2020-03-18 stsp while (!attached)
759 668a20f6 2020-03-18 stsp memset(&pack, 0, sizeof(pack));
760 668a20f6 2020-03-18 stsp pack.fd = -1;
761 668a20f6 2020-03-18 stsp pack.delta_cache = got_delta_cache_alloc(100,
762 668a20f6 2020-03-18 stsp GOT_DELTA_RESULT_SIZE_CACHED_MAX);
763 668a20f6 2020-03-18 stsp if (pack.delta_cache == NULL) {
764 668a20f6 2020-03-18 stsp err = got_error_from_errno("got_delta_cache_alloc");
768 668a20f6 2020-03-18 stsp imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
770 668a20f6 2020-03-18 stsp err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
773 93658fb9 2020-03-18 stsp if (imsg.hdr.type == GOT_IMSG_STOP)
775 93658fb9 2020-03-18 stsp if (imsg.hdr.type != GOT_IMSG_IDXPACK_REQUEST) {
776 93658fb9 2020-03-18 stsp err = got_error(GOT_ERR_PRIVSEP_MSG);
779 668a20f6 2020-03-18 stsp if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(pack_hash)) {
780 93658fb9 2020-03-18 stsp err = got_error(GOT_ERR_PRIVSEP_LEN);
783 668a20f6 2020-03-18 stsp memcpy(pack_hash, imsg.data, sizeof(pack_hash));
784 668a20f6 2020-03-18 stsp pack.fd = imsg.fd;
786 668a20f6 2020-03-18 stsp err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
789 93658fb9 2020-03-18 stsp if (imsg.hdr.type == GOT_IMSG_STOP)
791 93658fb9 2020-03-18 stsp if (imsg.hdr.type != GOT_IMSG_TMPFD) {
792 93658fb9 2020-03-18 stsp err = got_error(GOT_ERR_PRIVSEP_MSG);
795 93658fb9 2020-03-18 stsp if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
796 93658fb9 2020-03-18 stsp err = got_error(GOT_ERR_PRIVSEP_LEN);
799 93658fb9 2020-03-18 stsp idxfd = imsg.fd;
801 668a20f6 2020-03-18 stsp if (lseek(pack.fd, 0, SEEK_END) == -1) {
802 668a20f6 2020-03-18 stsp err = got_error_from_errno("lseek");
805 668a20f6 2020-03-18 stsp packfile_size = lseek(pack.fd, 0, SEEK_CUR);
806 668a20f6 2020-03-18 stsp if (packfile_size == -1) {
807 668a20f6 2020-03-18 stsp err = got_error_from_errno("lseek");
810 668a20f6 2020-03-18 stsp pack.filesize = packfile_size; /* XXX off_t vs size_t */
812 668a20f6 2020-03-18 stsp if (lseek(pack.fd, 0, SEEK_SET) == -1) {
813 668a20f6 2020-03-18 stsp err = got_error_from_errno("lseek");
817 668a20f6 2020-03-18 stsp err = index_pack(&pack, idxfd, pack_hash, &ibuf);
819 668a20f6 2020-03-18 stsp close_err = got_pack_close(&pack);
820 668a20f6 2020-03-18 stsp if (close_err && err == NULL)
821 668a20f6 2020-03-18 stsp err = close_err;
822 668a20f6 2020-03-18 stsp if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
823 668a20f6 2020-03-18 stsp err = got_error_from_errno("close");
825 668a20f6 2020-03-18 stsp if (err == NULL)
826 93658fb9 2020-03-18 stsp err = got_privsep_send_index_pack_done(&ibuf);
828 668a20f6 2020-03-18 stsp got_privsep_send_error(&ibuf, err);
829 93658fb9 2020-03-18 stsp fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
830 93658fb9 2020-03-18 stsp got_privsep_send_error(&ibuf, err);