]>
Commit | Line | Data |
---|---|---|
9cf6d335 SV |
1 | #include "cache.h" |
2 | #include "delta.h" | |
3 | #include "pack.h" | |
4 | #include "csum-file.h" | |
8e440259 PE |
5 | #include "blob.h" |
6 | #include "commit.h" | |
7 | #include "tag.h" | |
8 | #include "tree.h" | |
96a02f8f | 9 | #include "progress.h" |
0153be05 | 10 | #include "fsck.h" |
9cf6d335 SV |
11 | |
12 | static const char index_pack_usage[] = | |
0153be05 | 13 | "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] [--strict] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }"; |
9cf6d335 SV |
14 | |
15 | struct object_entry | |
16 | { | |
aa7e44bf | 17 | struct pack_idx_entry idx; |
2d477051 NP |
18 | unsigned long size; |
19 | unsigned int hdr_size; | |
9cf6d335 SV |
20 | enum object_type type; |
21 | enum object_type real_type; | |
9cf6d335 SV |
22 | }; |
23 | ||
53dda6ff NP |
24 | union delta_base { |
25 | unsigned char sha1[20]; | |
d7dd0223 | 26 | off_t offset; |
53dda6ff NP |
27 | }; |
28 | ||
3c552873 NP |
29 | /* |
30 | * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want | |
31 | * to memcmp() only the first 20 bytes. | |
32 | */ | |
33 | #define UNION_BASE_SZ 20 | |
34 | ||
0153be05 MK |
35 | #define FLAG_LINK (1u<<20) |
36 | #define FLAG_CHECKED (1u<<21) | |
37 | ||
9cf6d335 SV |
38 | struct delta_entry |
39 | { | |
53dda6ff | 40 | union delta_base base; |
636171cb | 41 | int obj_no; |
9cf6d335 SV |
42 | }; |
43 | ||
9cf6d335 SV |
44 | static struct object_entry *objects; |
45 | static struct delta_entry *deltas; | |
46 | static int nr_objects; | |
47 | static int nr_deltas; | |
636171cb | 48 | static int nr_resolved_deltas; |
9cf6d335 | 49 | |
e42797f5 | 50 | static int from_stdin; |
0153be05 | 51 | static int strict; |
3c9af366 NP |
52 | static int verbose; |
53 | ||
dc6a0757 | 54 | static struct progress *progress; |
e42797f5 | 55 | |
2d477051 NP |
56 | /* We always read in 4kB chunks. */ |
57 | static unsigned char input_buffer[4096]; | |
d7dd0223 NP |
58 | static unsigned int input_offset, input_len; |
59 | static off_t consumed_bytes; | |
2d477051 | 60 | static SHA_CTX input_ctx; |
ee5743ce | 61 | static uint32_t input_crc32; |
6d2fa7f1 | 62 | static int input_fd, output_fd, pack_fd; |
2d477051 | 63 | |
0153be05 MK |
64 | static int mark_link(struct object *obj, int type, void *data) |
65 | { | |
66 | if (!obj) | |
67 | return -1; | |
68 | ||
69 | if (type != OBJ_ANY && obj->type != type) | |
70 | die("object type mismatch at %s", sha1_to_hex(obj->sha1)); | |
71 | ||
72 | obj->flags |= FLAG_LINK; | |
73 | return 0; | |
74 | } | |
75 | ||
76 | /* The content of each linked object must have been checked | |
77 | or it must be already present in the object database */ | |
78 | static void check_object(struct object *obj) | |
79 | { | |
80 | if (!obj) | |
81 | return; | |
82 | ||
83 | if (!(obj->flags & FLAG_LINK)) | |
84 | return; | |
85 | ||
86 | if (!(obj->flags & FLAG_CHECKED)) { | |
87 | unsigned long size; | |
88 | int type = sha1_object_info(obj->sha1, &size); | |
89 | if (type != obj->type || type <= 0) | |
90 | die("object of unexpected type"); | |
91 | obj->flags |= FLAG_CHECKED; | |
92 | return; | |
93 | } | |
94 | } | |
95 | ||
96 | static void check_objects(void) | |
97 | { | |
98 | unsigned i, max; | |
99 | ||
100 | max = get_max_object_index(); | |
101 | for (i = 0; i < max; i++) | |
102 | check_object(get_indexed_object(i)); | |
103 | } | |
104 | ||
105 | ||
636171cb | 106 | /* Discard current buffer used content. */ |
a6e8a767 | 107 | static void flush(void) |
636171cb NP |
108 | { |
109 | if (input_offset) { | |
110 | if (output_fd >= 0) | |
111 | write_or_die(output_fd, input_buffer, input_offset); | |
112 | SHA1_Update(&input_ctx, input_buffer, input_offset); | |
554a2636 | 113 | memmove(input_buffer, input_buffer + input_offset, input_len); |
636171cb NP |
114 | input_offset = 0; |
115 | } | |
116 | } | |
117 | ||
2d477051 NP |
118 | /* |
119 | * Make sure at least "min" bytes are available in the buffer, and | |
120 | * return the pointer to the buffer. | |
121 | */ | |
b89c4e93 | 122 | static void *fill(int min) |
9cf6d335 | 123 | { |
2d477051 NP |
124 | if (min <= input_len) |
125 | return input_buffer + input_offset; | |
126 | if (min > sizeof(input_buffer)) | |
127 | die("cannot fill %d bytes", min); | |
636171cb | 128 | flush(); |
2d477051 | 129 | do { |
8a912bcb | 130 | ssize_t ret = xread(input_fd, input_buffer + input_len, |
2d477051 NP |
131 | sizeof(input_buffer) - input_len); |
132 | if (ret <= 0) { | |
133 | if (!ret) | |
134 | die("early EOF"); | |
135 | die("read error on input: %s", strerror(errno)); | |
136 | } | |
137 | input_len += ret; | |
218558af NP |
138 | if (from_stdin) |
139 | display_throughput(progress, consumed_bytes + input_len); | |
2d477051 NP |
140 | } while (input_len < min); |
141 | return input_buffer; | |
142 | } | |
143 | ||
144 | static void use(int bytes) | |
145 | { | |
146 | if (bytes > input_len) | |
147 | die("used more bytes than were available"); | |
ee5743ce | 148 | input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes); |
2d477051 NP |
149 | input_len -= bytes; |
150 | input_offset += bytes; | |
d7dd0223 NP |
151 | |
152 | /* make sure off_t is sufficiently large not to wrap */ | |
153 | if (consumed_bytes > consumed_bytes + bytes) | |
154 | die("pack too large for current definition of off_t"); | |
2d477051 NP |
155 | consumed_bytes += bytes; |
156 | } | |
9cf6d335 | 157 | |
4049b9cf | 158 | static char *open_pack_file(char *pack_name) |
2d477051 | 159 | { |
e42797f5 NP |
160 | if (from_stdin) { |
161 | input_fd = 0; | |
162 | if (!pack_name) { | |
163 | static char tmpfile[PATH_MAX]; | |
164 | snprintf(tmpfile, sizeof(tmpfile), | |
0e55181f | 165 | "%s/tmp_pack_XXXXXX", get_object_directory()); |
7647b17f | 166 | output_fd = xmkstemp(tmpfile); |
e42797f5 NP |
167 | pack_name = xstrdup(tmpfile); |
168 | } else | |
169 | output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600); | |
170 | if (output_fd < 0) | |
171 | die("unable to create %s: %s\n", pack_name, strerror(errno)); | |
6d2fa7f1 | 172 | pack_fd = output_fd; |
e42797f5 NP |
173 | } else { |
174 | input_fd = open(pack_name, O_RDONLY); | |
175 | if (input_fd < 0) | |
176 | die("cannot open packfile '%s': %s", | |
177 | pack_name, strerror(errno)); | |
178 | output_fd = -1; | |
6d2fa7f1 | 179 | pack_fd = input_fd; |
e42797f5 | 180 | } |
2d477051 | 181 | SHA1_Init(&input_ctx); |
e42797f5 | 182 | return pack_name; |
9cf6d335 SV |
183 | } |
184 | ||
185 | static void parse_pack_header(void) | |
186 | { | |
2d477051 | 187 | struct pack_header *hdr = fill(sizeof(struct pack_header)); |
9cf6d335 SV |
188 | |
189 | /* Header consistency check */ | |
9cf6d335 | 190 | if (hdr->hdr_signature != htonl(PACK_SIGNATURE)) |
e42797f5 | 191 | die("pack signature mismatch"); |
d60fc1c8 | 192 | if (!pack_version_ok(hdr->hdr_version)) |
e42797f5 | 193 | die("pack version %d unsupported", ntohl(hdr->hdr_version)); |
9cf6d335 SV |
194 | |
195 | nr_objects = ntohl(hdr->hdr_entries); | |
2d477051 | 196 | use(sizeof(struct pack_header)); |
9cf6d335 SV |
197 | } |
198 | ||
199 | static void bad_object(unsigned long offset, const char *format, | |
200 | ...) NORETURN __attribute__((format (printf, 2, 3))); | |
201 | ||
202 | static void bad_object(unsigned long offset, const char *format, ...) | |
203 | { | |
204 | va_list params; | |
205 | char buf[1024]; | |
206 | ||
207 | va_start(params, format); | |
208 | vsnprintf(buf, sizeof(buf), format, params); | |
209 | va_end(params); | |
e42797f5 | 210 | die("pack has bad object at offset %lu: %s", offset, buf); |
9cf6d335 SV |
211 | } |
212 | ||
2d477051 | 213 | static void *unpack_entry_data(unsigned long offset, unsigned long size) |
9cf6d335 | 214 | { |
9cf6d335 SV |
215 | z_stream stream; |
216 | void *buf = xmalloc(size); | |
217 | ||
218 | memset(&stream, 0, sizeof(stream)); | |
219 | stream.next_out = buf; | |
220 | stream.avail_out = size; | |
2d477051 NP |
221 | stream.next_in = fill(1); |
222 | stream.avail_in = input_len; | |
9cf6d335 SV |
223 | inflateInit(&stream); |
224 | ||
225 | for (;;) { | |
226 | int ret = inflate(&stream, 0); | |
2d477051 NP |
227 | use(input_len - stream.avail_in); |
228 | if (stream.total_out == size && ret == Z_STREAM_END) | |
9cf6d335 SV |
229 | break; |
230 | if (ret != Z_OK) | |
231 | bad_object(offset, "inflate returned %d", ret); | |
2d477051 NP |
232 | stream.next_in = fill(1); |
233 | stream.avail_in = input_len; | |
9cf6d335 SV |
234 | } |
235 | inflateEnd(&stream); | |
9cf6d335 SV |
236 | return buf; |
237 | } | |
238 | ||
2d477051 | 239 | static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base) |
9cf6d335 | 240 | { |
2d477051 | 241 | unsigned char *p, c; |
d7dd0223 NP |
242 | unsigned long size; |
243 | off_t base_offset; | |
9cf6d335 | 244 | unsigned shift; |
ee5743ce | 245 | void *data; |
9cf6d335 | 246 | |
aa7e44bf | 247 | obj->idx.offset = consumed_bytes; |
ee5743ce | 248 | input_crc32 = crc32(0, Z_NULL, 0); |
2d477051 NP |
249 | |
250 | p = fill(1); | |
251 | c = *p; | |
252 | use(1); | |
253 | obj->type = (c >> 4) & 7; | |
9cf6d335 SV |
254 | size = (c & 15); |
255 | shift = 4; | |
256 | while (c & 0x80) { | |
2d477051 NP |
257 | p = fill(1); |
258 | c = *p; | |
259 | use(1); | |
9cf6d335 SV |
260 | size += (c & 0x7fUL) << shift; |
261 | shift += 7; | |
262 | } | |
2d477051 | 263 | obj->size = size; |
9cf6d335 | 264 | |
2d477051 | 265 | switch (obj->type) { |
eb32d236 | 266 | case OBJ_REF_DELTA: |
2d477051 NP |
267 | hashcpy(delta_base->sha1, fill(20)); |
268 | use(20); | |
53dda6ff NP |
269 | break; |
270 | case OBJ_OFS_DELTA: | |
271 | memset(delta_base, 0, sizeof(*delta_base)); | |
2d477051 NP |
272 | p = fill(1); |
273 | c = *p; | |
274 | use(1); | |
53dda6ff NP |
275 | base_offset = c & 127; |
276 | while (c & 128) { | |
277 | base_offset += 1; | |
8723f216 | 278 | if (!base_offset || MSB(base_offset, 7)) |
aa7e44bf | 279 | bad_object(obj->idx.offset, "offset value overflow for delta base object"); |
2d477051 NP |
280 | p = fill(1); |
281 | c = *p; | |
282 | use(1); | |
53dda6ff NP |
283 | base_offset = (base_offset << 7) + (c & 127); |
284 | } | |
aa7e44bf GB |
285 | delta_base->offset = obj->idx.offset - base_offset; |
286 | if (delta_base->offset >= obj->idx.offset) | |
287 | bad_object(obj->idx.offset, "delta base offset is out of bound"); | |
53dda6ff | 288 | break; |
9cf6d335 SV |
289 | case OBJ_COMMIT: |
290 | case OBJ_TREE: | |
291 | case OBJ_BLOB: | |
292 | case OBJ_TAG: | |
9cf6d335 SV |
293 | break; |
294 | default: | |
aa7e44bf | 295 | bad_object(obj->idx.offset, "unknown object type %d", obj->type); |
9cf6d335 | 296 | } |
aa7e44bf | 297 | obj->hdr_size = consumed_bytes - obj->idx.offset; |
2d477051 | 298 | |
aa7e44bf GB |
299 | data = unpack_entry_data(obj->idx.offset, obj->size); |
300 | obj->idx.crc32 = input_crc32; | |
ee5743ce | 301 | return data; |
2d477051 NP |
302 | } |
303 | ||
b89c4e93 | 304 | static void *get_data_from_pack(struct object_entry *obj) |
2d477051 | 305 | { |
a91ef6e7 | 306 | off_t from = obj[0].idx.offset + obj[0].hdr_size; |
aa7e44bf | 307 | unsigned long len = obj[1].idx.offset - from; |
a91d49cd | 308 | unsigned long rdy = 0; |
6d2fa7f1 | 309 | unsigned char *src, *data; |
2d477051 NP |
310 | z_stream stream; |
311 | int st; | |
9cf6d335 | 312 | |
6d2fa7f1 | 313 | src = xmalloc(len); |
a91d49cd SP |
314 | data = src; |
315 | do { | |
316 | ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy); | |
317 | if (n <= 0) | |
318 | die("cannot pread pack file: %s", strerror(errno)); | |
319 | rdy += n; | |
320 | } while (rdy < len); | |
2d477051 NP |
321 | data = xmalloc(obj->size); |
322 | memset(&stream, 0, sizeof(stream)); | |
323 | stream.next_out = data; | |
324 | stream.avail_out = obj->size; | |
6d2fa7f1 | 325 | stream.next_in = src; |
2d477051 NP |
326 | stream.avail_in = len; |
327 | inflateInit(&stream); | |
328 | while ((st = inflate(&stream, Z_FINISH)) == Z_OK); | |
329 | inflateEnd(&stream); | |
330 | if (st != Z_STREAM_END || stream.total_out != obj->size) | |
331 | die("serious inflate inconsistency"); | |
6d2fa7f1 | 332 | free(src); |
9cf6d335 SV |
333 | return data; |
334 | } | |
335 | ||
53dda6ff | 336 | static int find_delta(const union delta_base *base) |
9cf6d335 SV |
337 | { |
338 | int first = 0, last = nr_deltas; | |
339 | ||
340 | while (first < last) { | |
341 | int next = (first + last) / 2; | |
342 | struct delta_entry *delta = &deltas[next]; | |
343 | int cmp; | |
344 | ||
3c552873 | 345 | cmp = memcmp(base, &delta->base, UNION_BASE_SZ); |
9cf6d335 SV |
346 | if (!cmp) |
347 | return next; | |
348 | if (cmp < 0) { | |
349 | last = next; | |
350 | continue; | |
351 | } | |
352 | first = next+1; | |
353 | } | |
354 | return -first-1; | |
355 | } | |
356 | ||
b89c4e93 NP |
357 | static int find_delta_children(const union delta_base *base, |
358 | int *first_index, int *last_index) | |
9cf6d335 | 359 | { |
53dda6ff | 360 | int first = find_delta(base); |
9cf6d335 SV |
361 | int last = first; |
362 | int end = nr_deltas - 1; | |
363 | ||
364 | if (first < 0) | |
365 | return -1; | |
3c552873 | 366 | while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ)) |
9cf6d335 | 367 | --first; |
3c552873 | 368 | while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ)) |
9cf6d335 SV |
369 | ++last; |
370 | *first_index = first; | |
371 | *last_index = last; | |
372 | return 0; | |
373 | } | |
374 | ||
375 | static void sha1_object(const void *data, unsigned long size, | |
9096c660 | 376 | enum object_type type, unsigned char *sha1) |
9cf6d335 | 377 | { |
ce9fbf16 | 378 | hash_sha1_file(data, size, typename(type), sha1); |
9096c660 | 379 | if (has_sha1_file(sha1)) { |
8685da42 NP |
380 | void *has_data; |
381 | enum object_type has_type; | |
382 | unsigned long has_size; | |
383 | has_data = read_sha1_file(sha1, &has_type, &has_size); | |
384 | if (!has_data) | |
385 | die("cannot read existing object %s", sha1_to_hex(sha1)); | |
386 | if (size != has_size || type != has_type || | |
387 | memcmp(data, has_data, size) != 0) | |
388 | die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1)); | |
bbf4b41b | 389 | free(has_data); |
8685da42 | 390 | } |
0153be05 MK |
391 | if (strict) { |
392 | if (type == OBJ_BLOB) { | |
393 | struct blob *blob = lookup_blob(sha1); | |
394 | if (blob) | |
395 | blob->object.flags |= FLAG_CHECKED; | |
396 | else | |
397 | die("invalid blob object %s", sha1_to_hex(sha1)); | |
398 | } else { | |
399 | struct object *obj; | |
400 | int eaten; | |
401 | void *buf = (void *) data; | |
402 | ||
403 | /* | |
404 | * we do not need to free the memory here, as the | |
405 | * buf is deleted by the caller. | |
406 | */ | |
407 | obj = parse_object_buffer(sha1, type, size, buf, &eaten); | |
408 | if (!obj) | |
409 | die("invalid %s", typename(type)); | |
410 | if (fsck_object(obj, 1, fsck_error_function)) | |
411 | die("Error in object"); | |
412 | if (fsck_walk(obj, mark_link, 0)) | |
413 | die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1)); | |
414 | ||
415 | if (obj->type == OBJ_TREE) { | |
416 | struct tree *item = (struct tree *) obj; | |
417 | item->buffer = NULL; | |
418 | } | |
419 | if (obj->type == OBJ_COMMIT) { | |
420 | struct commit *commit = (struct commit *) obj; | |
421 | commit->buffer = NULL; | |
422 | } | |
423 | obj->flags |= FLAG_CHECKED; | |
424 | } | |
425 | } | |
9cf6d335 SV |
426 | } |
427 | ||
636171cb | 428 | static void resolve_delta(struct object_entry *delta_obj, void *base_data, |
9cf6d335 SV |
429 | unsigned long base_size, enum object_type type) |
430 | { | |
9cf6d335 SV |
431 | void *delta_data; |
432 | unsigned long delta_size; | |
433 | void *result; | |
434 | unsigned long result_size; | |
53dda6ff | 435 | union delta_base delta_base; |
9cf6d335 SV |
436 | int j, first, last; |
437 | ||
636171cb NP |
438 | delta_obj->real_type = type; |
439 | delta_data = get_data_from_pack(delta_obj); | |
440 | delta_size = delta_obj->size; | |
9cf6d335 SV |
441 | result = patch_delta(base_data, base_size, delta_data, delta_size, |
442 | &result_size); | |
443 | free(delta_data); | |
444 | if (!result) | |
aa7e44bf GB |
445 | bad_object(delta_obj->idx.offset, "failed to apply delta"); |
446 | sha1_object(result, result_size, type, delta_obj->idx.sha1); | |
636171cb | 447 | nr_resolved_deltas++; |
53dda6ff | 448 | |
aa7e44bf | 449 | hashcpy(delta_base.sha1, delta_obj->idx.sha1); |
b89c4e93 | 450 | if (!find_delta_children(&delta_base, &first, &last)) { |
636171cb NP |
451 | for (j = first; j <= last; j++) { |
452 | struct object_entry *child = objects + deltas[j].obj_no; | |
453 | if (child->real_type == OBJ_REF_DELTA) | |
454 | resolve_delta(child, result, result_size, type); | |
455 | } | |
53dda6ff NP |
456 | } |
457 | ||
458 | memset(&delta_base, 0, sizeof(delta_base)); | |
aa7e44bf | 459 | delta_base.offset = delta_obj->idx.offset; |
b89c4e93 | 460 | if (!find_delta_children(&delta_base, &first, &last)) { |
636171cb NP |
461 | for (j = first; j <= last; j++) { |
462 | struct object_entry *child = objects + deltas[j].obj_no; | |
463 | if (child->real_type == OBJ_OFS_DELTA) | |
464 | resolve_delta(child, result, result_size, type); | |
465 | } | |
9cf6d335 | 466 | } |
53dda6ff | 467 | |
9cf6d335 SV |
468 | free(result); |
469 | } | |
470 | ||
471 | static int compare_delta_entry(const void *a, const void *b) | |
472 | { | |
473 | const struct delta_entry *delta_a = a; | |
474 | const struct delta_entry *delta_b = b; | |
3c552873 | 475 | return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ); |
9cf6d335 SV |
476 | } |
477 | ||
2d477051 NP |
478 | /* Parse all objects and return the pack content SHA1 hash */ |
479 | static void parse_pack_objects(unsigned char *sha1) | |
9cf6d335 | 480 | { |
96a02f8f | 481 | int i; |
53dda6ff | 482 | struct delta_entry *delta = deltas; |
9cf6d335 | 483 | void *data; |
2d477051 | 484 | struct stat st; |
9cf6d335 SV |
485 | |
486 | /* | |
487 | * First pass: | |
488 | * - find locations of all objects; | |
489 | * - calculate SHA1 of all non-delta objects; | |
b89c4e93 | 490 | * - remember base (SHA1 or offset) for all deltas. |
9cf6d335 | 491 | */ |
13aaf148 | 492 | if (verbose) |
29e63ed3 NP |
493 | progress = start_progress( |
494 | from_stdin ? "Receiving objects" : "Indexing objects", | |
495 | nr_objects); | |
9cf6d335 SV |
496 | for (i = 0; i < nr_objects; i++) { |
497 | struct object_entry *obj = &objects[i]; | |
2d477051 | 498 | data = unpack_raw_entry(obj, &delta->base); |
9cf6d335 | 499 | obj->real_type = obj->type; |
53dda6ff NP |
500 | if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) { |
501 | nr_deltas++; | |
636171cb | 502 | delta->obj_no = i; |
53dda6ff | 503 | delta++; |
9cf6d335 | 504 | } else |
aa7e44bf | 505 | sha1_object(data, obj->size, obj->type, obj->idx.sha1); |
9cf6d335 | 506 | free(data); |
4d4fcc54 | 507 | display_progress(progress, i+1); |
9cf6d335 | 508 | } |
aa7e44bf | 509 | objects[i].idx.offset = consumed_bytes; |
4d4fcc54 | 510 | stop_progress(&progress); |
2d477051 NP |
511 | |
512 | /* Check pack integrity */ | |
636171cb | 513 | flush(); |
2d477051 NP |
514 | SHA1_Final(sha1, &input_ctx); |
515 | if (hashcmp(fill(20), sha1)) | |
e42797f5 | 516 | die("pack is corrupted (SHA1 mismatch)"); |
9bee2478 | 517 | use(20); |
2d477051 NP |
518 | |
519 | /* If input_fd is a file, we should have reached its end now. */ | |
520 | if (fstat(input_fd, &st)) | |
e42797f5 | 521 | die("cannot fstat packfile: %s", strerror(errno)); |
fa257b05 JS |
522 | if (S_ISREG(st.st_mode) && |
523 | lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size) | |
e42797f5 | 524 | die("pack has junk at the end"); |
9cf6d335 | 525 | |
3c9af366 NP |
526 | if (!nr_deltas) |
527 | return; | |
528 | ||
53dda6ff | 529 | /* Sort deltas by base SHA1/offset for fast searching */ |
9cf6d335 SV |
530 | qsort(deltas, nr_deltas, sizeof(struct delta_entry), |
531 | compare_delta_entry); | |
532 | ||
533 | /* | |
534 | * Second pass: | |
535 | * - for all non-delta objects, look if it is used as a base for | |
536 | * deltas; | |
537 | * - if used as a base, uncompress the object and apply all deltas, | |
538 | * recursively checking if the resulting object is used as a base | |
539 | * for some more deltas. | |
540 | */ | |
13aaf148 | 541 | if (verbose) |
dc6a0757 | 542 | progress = start_progress("Resolving deltas", nr_deltas); |
9cf6d335 SV |
543 | for (i = 0; i < nr_objects; i++) { |
544 | struct object_entry *obj = &objects[i]; | |
53dda6ff NP |
545 | union delta_base base; |
546 | int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last; | |
9cf6d335 | 547 | |
53dda6ff | 548 | if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) |
9cf6d335 | 549 | continue; |
aa7e44bf | 550 | hashcpy(base.sha1, obj->idx.sha1); |
b89c4e93 | 551 | ref = !find_delta_children(&base, &ref_first, &ref_last); |
53dda6ff | 552 | memset(&base, 0, sizeof(base)); |
aa7e44bf | 553 | base.offset = obj->idx.offset; |
b89c4e93 | 554 | ofs = !find_delta_children(&base, &ofs_first, &ofs_last); |
53dda6ff | 555 | if (!ref && !ofs) |
9cf6d335 | 556 | continue; |
2d477051 | 557 | data = get_data_from_pack(obj); |
53dda6ff | 558 | if (ref) |
636171cb NP |
559 | for (j = ref_first; j <= ref_last; j++) { |
560 | struct object_entry *child = objects + deltas[j].obj_no; | |
561 | if (child->real_type == OBJ_REF_DELTA) | |
562 | resolve_delta(child, data, | |
2d477051 | 563 | obj->size, obj->type); |
636171cb | 564 | } |
53dda6ff | 565 | if (ofs) |
636171cb NP |
566 | for (j = ofs_first; j <= ofs_last; j++) { |
567 | struct object_entry *child = objects + deltas[j].obj_no; | |
568 | if (child->real_type == OBJ_OFS_DELTA) | |
569 | resolve_delta(child, data, | |
2d477051 | 570 | obj->size, obj->type); |
636171cb | 571 | } |
9cf6d335 | 572 | free(data); |
4d4fcc54 | 573 | display_progress(progress, nr_resolved_deltas); |
9cf6d335 | 574 | } |
636171cb NP |
575 | } |
576 | ||
ee5743ce | 577 | static int write_compressed(int fd, void *in, unsigned int size, uint32_t *obj_crc) |
636171cb NP |
578 | { |
579 | z_stream stream; | |
580 | unsigned long maxsize; | |
581 | void *out; | |
582 | ||
583 | memset(&stream, 0, sizeof(stream)); | |
584 | deflateInit(&stream, zlib_compression_level); | |
585 | maxsize = deflateBound(&stream, size); | |
586 | out = xmalloc(maxsize); | |
587 | ||
588 | /* Compress it */ | |
589 | stream.next_in = in; | |
590 | stream.avail_in = size; | |
591 | stream.next_out = out; | |
592 | stream.avail_out = maxsize; | |
593 | while (deflate(&stream, Z_FINISH) == Z_OK); | |
594 | deflateEnd(&stream); | |
595 | ||
596 | size = stream.total_out; | |
597 | write_or_die(fd, out, size); | |
ee5743ce | 598 | *obj_crc = crc32(*obj_crc, out, size); |
636171cb NP |
599 | free(out); |
600 | return size; | |
601 | } | |
602 | ||
9096c660 | 603 | static void append_obj_to_pack(const unsigned char *sha1, void *buf, |
636171cb NP |
604 | unsigned long size, enum object_type type) |
605 | { | |
606 | struct object_entry *obj = &objects[nr_objects++]; | |
607 | unsigned char header[10]; | |
608 | unsigned long s = size; | |
609 | int n = 0; | |
610 | unsigned char c = (type << 4) | (s & 15); | |
611 | s >>= 4; | |
612 | while (s) { | |
613 | header[n++] = c | 0x80; | |
614 | c = s & 0x7f; | |
615 | s >>= 7; | |
616 | } | |
617 | header[n++] = c; | |
618 | write_or_die(output_fd, header, n); | |
aa7e44bf GB |
619 | obj[0].idx.crc32 = crc32(0, Z_NULL, 0); |
620 | obj[0].idx.crc32 = crc32(obj[0].idx.crc32, header, n); | |
621 | obj[1].idx.offset = obj[0].idx.offset + n; | |
622 | obj[1].idx.offset += write_compressed(output_fd, buf, size, &obj[0].idx.crc32); | |
623 | hashcpy(obj->idx.sha1, sha1); | |
636171cb NP |
624 | } |
625 | ||
626 | static int delta_pos_compare(const void *_a, const void *_b) | |
627 | { | |
628 | struct delta_entry *a = *(struct delta_entry **)_a; | |
629 | struct delta_entry *b = *(struct delta_entry **)_b; | |
630 | return a->obj_no - b->obj_no; | |
631 | } | |
9cf6d335 | 632 | |
636171cb NP |
633 | static void fix_unresolved_deltas(int nr_unresolved) |
634 | { | |
635 | struct delta_entry **sorted_by_pos; | |
96a02f8f | 636 | int i, n = 0; |
636171cb NP |
637 | |
638 | /* | |
639 | * Since many unresolved deltas may well be themselves base objects | |
640 | * for more unresolved deltas, we really want to include the | |
641 | * smallest number of base objects that would cover as much delta | |
642 | * as possible by picking the | |
643 | * trunc deltas first, allowing for other deltas to resolve without | |
644 | * additional base objects. Since most base objects are to be found | |
645 | * before deltas depending on them, a good heuristic is to start | |
646 | * resolving deltas in the same order as their position in the pack. | |
647 | */ | |
648 | sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos)); | |
9cf6d335 | 649 | for (i = 0; i < nr_deltas; i++) { |
636171cb NP |
650 | if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA) |
651 | continue; | |
652 | sorted_by_pos[n++] = &deltas[i]; | |
9cf6d335 | 653 | } |
636171cb NP |
654 | qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare); |
655 | ||
656 | for (i = 0; i < n; i++) { | |
657 | struct delta_entry *d = sorted_by_pos[i]; | |
658 | void *data; | |
659 | unsigned long size; | |
21666f1a | 660 | enum object_type type; |
636171cb NP |
661 | int j, first, last; |
662 | ||
663 | if (objects[d->obj_no].real_type != OBJ_REF_DELTA) | |
664 | continue; | |
21666f1a | 665 | data = read_sha1_file(d->base.sha1, &type, &size); |
636171cb NP |
666 | if (!data) |
667 | continue; | |
636171cb | 668 | |
b89c4e93 | 669 | find_delta_children(&d->base, &first, &last); |
636171cb NP |
670 | for (j = first; j <= last; j++) { |
671 | struct object_entry *child = objects + deltas[j].obj_no; | |
672 | if (child->real_type == OBJ_REF_DELTA) | |
21666f1a | 673 | resolve_delta(child, data, size, type); |
636171cb NP |
674 | } |
675 | ||
9096c660 NP |
676 | if (check_sha1_signature(d->base.sha1, data, size, typename(type))) |
677 | die("local object %s is corrupt", sha1_to_hex(d->base.sha1)); | |
678 | append_obj_to_pack(d->base.sha1, data, size, type); | |
636171cb | 679 | free(data); |
4d4fcc54 | 680 | display_progress(progress, nr_resolved_deltas); |
636171cb NP |
681 | } |
682 | free(sorted_by_pos); | |
683 | } | |
684 | ||
e42797f5 NP |
685 | static void final(const char *final_pack_name, const char *curr_pack_name, |
686 | const char *final_index_name, const char *curr_index_name, | |
b8077709 | 687 | const char *keep_name, const char *keep_msg, |
e42797f5 NP |
688 | unsigned char *sha1) |
689 | { | |
3a55602e | 690 | const char *report = "pack"; |
e42797f5 NP |
691 | char name[PATH_MAX]; |
692 | int err; | |
693 | ||
694 | if (!from_stdin) { | |
695 | close(input_fd); | |
696 | } else { | |
697 | err = close(output_fd); | |
698 | if (err) | |
699 | die("error while closing pack file: %s", strerror(errno)); | |
700 | chmod(curr_pack_name, 0444); | |
701 | } | |
702 | ||
b8077709 SP |
703 | if (keep_msg) { |
704 | int keep_fd, keep_msg_len = strlen(keep_msg); | |
705 | if (!keep_name) { | |
706 | snprintf(name, sizeof(name), "%s/pack/pack-%s.keep", | |
707 | get_object_directory(), sha1_to_hex(sha1)); | |
708 | keep_name = name; | |
709 | } | |
9ca4a201 NP |
710 | keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600); |
711 | if (keep_fd < 0) { | |
712 | if (errno != EEXIST) | |
713 | die("cannot write keep file"); | |
714 | } else { | |
715 | if (keep_msg_len > 0) { | |
716 | write_or_die(keep_fd, keep_msg, keep_msg_len); | |
717 | write_or_die(keep_fd, "\n", 1); | |
718 | } | |
91c8d590 JM |
719 | if (close(keep_fd) != 0) |
720 | die("cannot write keep file"); | |
576162a4 | 721 | report = "keep"; |
b8077709 | 722 | } |
b8077709 SP |
723 | } |
724 | ||
e42797f5 NP |
725 | if (final_pack_name != curr_pack_name) { |
726 | if (!final_pack_name) { | |
727 | snprintf(name, sizeof(name), "%s/pack/pack-%s.pack", | |
728 | get_object_directory(), sha1_to_hex(sha1)); | |
729 | final_pack_name = name; | |
730 | } | |
731 | if (move_temp_to_file(curr_pack_name, final_pack_name)) | |
732 | die("cannot store pack file"); | |
733 | } | |
734 | ||
735 | chmod(curr_index_name, 0444); | |
736 | if (final_index_name != curr_index_name) { | |
737 | if (!final_index_name) { | |
738 | snprintf(name, sizeof(name), "%s/pack/pack-%s.idx", | |
739 | get_object_directory(), sha1_to_hex(sha1)); | |
740 | final_index_name = name; | |
741 | } | |
742 | if (move_temp_to_file(curr_index_name, final_index_name)) | |
743 | die("cannot store index file"); | |
744 | } | |
576162a4 NP |
745 | |
746 | if (!from_stdin) { | |
747 | printf("%s\n", sha1_to_hex(sha1)); | |
748 | } else { | |
749 | char buf[48]; | |
750 | int len = snprintf(buf, sizeof(buf), "%s\t%s\n", | |
751 | report, sha1_to_hex(sha1)); | |
d1b2ddc8 | 752 | write_or_die(1, buf, len); |
576162a4 NP |
753 | |
754 | /* | |
755 | * Let's just mimic git-unpack-objects here and write | |
756 | * the last part of the input buffer to stdout. | |
757 | */ | |
758 | while (input_len) { | |
759 | err = xwrite(1, input_buffer + input_offset, input_len); | |
760 | if (err <= 0) | |
761 | break; | |
762 | input_len -= err; | |
763 | input_offset += err; | |
764 | } | |
765 | } | |
9cf6d335 SV |
766 | } |
767 | ||
4d00bda2 NP |
768 | static int git_index_pack_config(const char *k, const char *v) |
769 | { | |
770 | if (!strcmp(k, "pack.indexversion")) { | |
771 | pack_idx_default_version = git_config_int(k, v); | |
772 | if (pack_idx_default_version > 2) | |
773 | die("bad pack.indexversion=%d", pack_idx_default_version); | |
774 | return 0; | |
775 | } | |
776 | return git_default_config(k, v); | |
777 | } | |
778 | ||
9cf6d335 SV |
779 | int main(int argc, char **argv) |
780 | { | |
636171cb | 781 | int i, fix_thin_pack = 0; |
4049b9cf NP |
782 | char *curr_pack, *pack_name = NULL; |
783 | char *curr_index, *index_name = NULL; | |
b8077709 SP |
784 | const char *keep_name = NULL, *keep_msg = NULL; |
785 | char *index_name_buf = NULL, *keep_name_buf = NULL; | |
aa7e44bf | 786 | struct pack_idx_entry **idx_objects; |
84c8d8ae | 787 | unsigned char sha1[20]; |
9cf6d335 | 788 | |
4d00bda2 NP |
789 | git_config(git_index_pack_config); |
790 | ||
9cf6d335 | 791 | for (i = 1; i < argc; i++) { |
4049b9cf | 792 | char *arg = argv[i]; |
9cf6d335 SV |
793 | |
794 | if (*arg == '-') { | |
e42797f5 NP |
795 | if (!strcmp(arg, "--stdin")) { |
796 | from_stdin = 1; | |
636171cb NP |
797 | } else if (!strcmp(arg, "--fix-thin")) { |
798 | fix_thin_pack = 1; | |
0153be05 MK |
799 | } else if (!strcmp(arg, "--strict")) { |
800 | strict = 1; | |
b8077709 SP |
801 | } else if (!strcmp(arg, "--keep")) { |
802 | keep_msg = ""; | |
cc44c765 | 803 | } else if (!prefixcmp(arg, "--keep=")) { |
b8077709 | 804 | keep_msg = arg + 7; |
cc44c765 | 805 | } else if (!prefixcmp(arg, "--pack_header=")) { |
bed006fb NP |
806 | struct pack_header *hdr; |
807 | char *c; | |
808 | ||
809 | hdr = (struct pack_header *)input_buffer; | |
810 | hdr->hdr_signature = htonl(PACK_SIGNATURE); | |
811 | hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10)); | |
812 | if (*c != ',') | |
813 | die("bad %s", arg); | |
814 | hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10)); | |
815 | if (*c) | |
816 | die("bad %s", arg); | |
817 | input_len = sizeof(*hdr); | |
3c9af366 NP |
818 | } else if (!strcmp(arg, "-v")) { |
819 | verbose = 1; | |
e42797f5 | 820 | } else if (!strcmp(arg, "-o")) { |
9cf6d335 SV |
821 | if (index_name || (i+1) >= argc) |
822 | usage(index_pack_usage); | |
823 | index_name = argv[++i]; | |
4ba7d711 NP |
824 | } else if (!prefixcmp(arg, "--index-version=")) { |
825 | char *c; | |
aa7e44bf GB |
826 | pack_idx_default_version = strtoul(arg + 16, &c, 10); |
827 | if (pack_idx_default_version > 2) | |
4ba7d711 NP |
828 | die("bad %s", arg); |
829 | if (*c == ',') | |
aa7e44bf GB |
830 | pack_idx_off32_limit = strtoul(c+1, &c, 0); |
831 | if (*c || pack_idx_off32_limit & 0x80000000) | |
4ba7d711 | 832 | die("bad %s", arg); |
9cf6d335 SV |
833 | } else |
834 | usage(index_pack_usage); | |
835 | continue; | |
836 | } | |
837 | ||
838 | if (pack_name) | |
839 | usage(index_pack_usage); | |
840 | pack_name = arg; | |
841 | } | |
842 | ||
e42797f5 | 843 | if (!pack_name && !from_stdin) |
9cf6d335 | 844 | usage(index_pack_usage); |
636171cb NP |
845 | if (fix_thin_pack && !from_stdin) |
846 | die("--fix-thin cannot be used without --stdin"); | |
e42797f5 | 847 | if (!index_name && pack_name) { |
9cf6d335 | 848 | int len = strlen(pack_name); |
5bb1cda5 | 849 | if (!has_extension(pack_name, ".pack")) |
9cf6d335 SV |
850 | die("packfile name '%s' does not end with '.pack'", |
851 | pack_name); | |
6689f087 | 852 | index_name_buf = xmalloc(len); |
9cf6d335 SV |
853 | memcpy(index_name_buf, pack_name, len - 5); |
854 | strcpy(index_name_buf + len - 5, ".idx"); | |
855 | index_name = index_name_buf; | |
856 | } | |
b8077709 SP |
857 | if (keep_msg && !keep_name && pack_name) { |
858 | int len = strlen(pack_name); | |
859 | if (!has_extension(pack_name, ".pack")) | |
860 | die("packfile name '%s' does not end with '.pack'", | |
861 | pack_name); | |
862 | keep_name_buf = xmalloc(len); | |
863 | memcpy(keep_name_buf, pack_name, len - 5); | |
864 | strcpy(keep_name_buf + len - 5, ".keep"); | |
865 | keep_name = keep_name_buf; | |
866 | } | |
9cf6d335 | 867 | |
e42797f5 | 868 | curr_pack = open_pack_file(pack_name); |
9cf6d335 | 869 | parse_pack_header(); |
636171cb NP |
870 | objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry)); |
871 | deltas = xmalloc(nr_objects * sizeof(struct delta_entry)); | |
2d477051 | 872 | parse_pack_objects(sha1); |
96a02f8f | 873 | if (nr_deltas == nr_resolved_deltas) { |
4d4fcc54 | 874 | stop_progress(&progress); |
96a02f8f NP |
875 | /* Flush remaining pack final 20-byte SHA1. */ |
876 | flush(); | |
877 | } else { | |
636171cb | 878 | if (fix_thin_pack) { |
a984a06a | 879 | char msg[48]; |
636171cb | 880 | int nr_unresolved = nr_deltas - nr_resolved_deltas; |
3c9af366 | 881 | int nr_objects_initial = nr_objects; |
636171cb NP |
882 | if (nr_unresolved <= 0) |
883 | die("confusion beyond insanity"); | |
884 | objects = xrealloc(objects, | |
885 | (nr_objects + nr_unresolved + 1) | |
886 | * sizeof(*objects)); | |
887 | fix_unresolved_deltas(nr_unresolved); | |
a984a06a NP |
888 | sprintf(msg, "completed with %d local objects", |
889 | nr_objects - nr_objects_initial); | |
890 | stop_progress_msg(&progress, msg); | |
cbc6bdab | 891 | fixup_pack_header_footer(output_fd, sha1, |
a984a06a | 892 | curr_pack, nr_objects); |
636171cb NP |
893 | } |
894 | if (nr_deltas != nr_resolved_deltas) | |
895 | die("pack has %d unresolved deltas", | |
896 | nr_deltas - nr_resolved_deltas); | |
636171cb | 897 | } |
9cf6d335 | 898 | free(deltas); |
0153be05 MK |
899 | if (strict) |
900 | check_objects(); | |
aa7e44bf GB |
901 | |
902 | idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *)); | |
903 | for (i = 0; i < nr_objects; i++) | |
904 | idx_objects[i] = &objects[i].idx; | |
905 | curr_index = write_idx_file(index_name, idx_objects, nr_objects, sha1); | |
906 | free(idx_objects); | |
907 | ||
b8077709 SP |
908 | final(pack_name, curr_pack, |
909 | index_name, curr_index, | |
910 | keep_name, keep_msg, | |
911 | sha1); | |
9cf6d335 SV |
912 | free(objects); |
913 | free(index_name_buf); | |
b8077709 | 914 | free(keep_name_buf); |
c85228ed NP |
915 | if (pack_name == NULL) |
916 | free(curr_pack); | |
917 | if (index_name == NULL) | |
918 | free(curr_index); | |
9cf6d335 SV |
919 | |
920 | return 0; | |
921 | } |