]>
Commit | Line | Data |
---|---|---|
2ad9d4cb | 1 | #include "builtin.h" |
9cf6d335 SV |
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" |
2fb3f6db | 11 | #include "exec_cmd.h" |
4614043c | 12 | #include "streaming.h" |
b8a2486f | 13 | #include "thread-utils.h" |
9cf6d335 SV |
14 | |
15 | static const char index_pack_usage[] = | |
e337a04d | 16 | "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])"; |
9cf6d335 | 17 | |
9cba13ca | 18 | struct object_entry { |
aa7e44bf | 19 | struct pack_idx_entry idx; |
2d477051 | 20 | unsigned long size; |
41730576 NTND |
21 | unsigned char hdr_size; |
22 | signed char type; | |
23 | signed char real_type; | |
9cf6d335 SV |
24 | }; |
25 | ||
41730576 | 26 | struct object_stat { |
38a45561 JH |
27 | unsigned delta_depth; |
28 | int base_object_no; | |
53dda6ff NP |
29 | }; |
30 | ||
f41aebd4 | 31 | struct base_data { |
4a438cab SP |
32 | struct base_data *base; |
33 | struct base_data *child; | |
03993e13 | 34 | struct object_entry *obj; |
f41aebd4 SP |
35 | void *data; |
36 | unsigned long size; | |
2baad220 NTND |
37 | int ref_first, ref_last; |
38 | int ofs_first, ofs_last; | |
f41aebd4 SP |
39 | }; |
40 | ||
b8a2486f NTND |
41 | struct thread_local { |
42 | #ifndef NO_PTHREADS | |
43 | pthread_t thread; | |
44 | #endif | |
45 | struct base_data *base_cache; | |
46 | size_t base_cache_used; | |
39539495 | 47 | int pack_fd; |
b8a2486f NTND |
48 | }; |
49 | ||
0153be05 MK |
50 | #define FLAG_LINK (1u<<20) |
51 | #define FLAG_CHECKED (1u<<21) | |
52 | ||
c6458e60 NTND |
53 | struct ofs_delta_entry { |
54 | off_t offset; | |
55 | int obj_no; | |
56 | }; | |
57 | ||
58 | struct ref_delta_entry { | |
59 | unsigned char sha1[20]; | |
636171cb | 60 | int obj_no; |
9cf6d335 SV |
61 | }; |
62 | ||
9cf6d335 | 63 | static struct object_entry *objects; |
41730576 | 64 | static struct object_stat *obj_stat; |
c6458e60 NTND |
65 | static struct ofs_delta_entry *ofs_deltas; |
66 | static struct ref_delta_entry *ref_deltas; | |
b8a2486f | 67 | static struct thread_local nothread_data; |
9cf6d335 | 68 | static int nr_objects; |
c6458e60 NTND |
69 | static int nr_ofs_deltas; |
70 | static int nr_ref_deltas; | |
71 | static int ref_deltas_alloc; | |
636171cb | 72 | static int nr_resolved_deltas; |
b8a2486f | 73 | static int nr_threads; |
9cf6d335 | 74 | |
e42797f5 | 75 | static int from_stdin; |
0153be05 | 76 | static int strict; |
c6807a40 | 77 | static int do_fsck_object; |
22410549 | 78 | static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT; |
3c9af366 | 79 | static int verbose; |
e376f17f | 80 | static int show_resolving_progress; |
3aba2fdd | 81 | static int show_stat; |
c6807a40 | 82 | static int check_self_contained_and_connected; |
3c9af366 | 83 | |
dc6a0757 | 84 | static struct progress *progress; |
e42797f5 | 85 | |
2d477051 NP |
86 | /* We always read in 4kB chunks. */ |
87 | static unsigned char input_buffer[4096]; | |
d7dd0223 NP |
88 | static unsigned int input_offset, input_len; |
89 | static off_t consumed_bytes; | |
411481be | 90 | static off_t max_input_size; |
d1a0ed18 | 91 | static unsigned deepest_delta; |
9126f009 | 92 | static git_SHA_CTX input_ctx; |
ee5743ce | 93 | static uint32_t input_crc32; |
39539495 NTND |
94 | static int input_fd, output_fd; |
95 | static const char *curr_pack; | |
2d477051 | 96 | |
b8a2486f NTND |
97 | #ifndef NO_PTHREADS |
98 | ||
99 | static struct thread_local *thread_data; | |
100 | static int nr_dispatched; | |
101 | static int threads_active; | |
102 | ||
103 | static pthread_mutex_t read_mutex; | |
104 | #define read_lock() lock_mutex(&read_mutex) | |
105 | #define read_unlock() unlock_mutex(&read_mutex) | |
106 | ||
107 | static pthread_mutex_t counter_mutex; | |
108 | #define counter_lock() lock_mutex(&counter_mutex) | |
109 | #define counter_unlock() unlock_mutex(&counter_mutex) | |
110 | ||
111 | static pthread_mutex_t work_mutex; | |
112 | #define work_lock() lock_mutex(&work_mutex) | |
113 | #define work_unlock() unlock_mutex(&work_mutex) | |
114 | ||
3aba2fdd NTND |
115 | static pthread_mutex_t deepest_delta_mutex; |
116 | #define deepest_delta_lock() lock_mutex(&deepest_delta_mutex) | |
117 | #define deepest_delta_unlock() unlock_mutex(&deepest_delta_mutex) | |
118 | ||
ab791dd1 JK |
119 | static pthread_mutex_t type_cas_mutex; |
120 | #define type_cas_lock() lock_mutex(&type_cas_mutex) | |
121 | #define type_cas_unlock() unlock_mutex(&type_cas_mutex) | |
122 | ||
b8a2486f NTND |
123 | static pthread_key_t key; |
124 | ||
125 | static inline void lock_mutex(pthread_mutex_t *mutex) | |
126 | { | |
127 | if (threads_active) | |
128 | pthread_mutex_lock(mutex); | |
129 | } | |
130 | ||
131 | static inline void unlock_mutex(pthread_mutex_t *mutex) | |
132 | { | |
133 | if (threads_active) | |
134 | pthread_mutex_unlock(mutex); | |
135 | } | |
136 | ||
137 | /* | |
138 | * Mutex and conditional variable can't be statically-initialized on Windows. | |
139 | */ | |
140 | static void init_thread(void) | |
141 | { | |
39539495 | 142 | int i; |
b8a2486f NTND |
143 | init_recursive_mutex(&read_mutex); |
144 | pthread_mutex_init(&counter_mutex, NULL); | |
145 | pthread_mutex_init(&work_mutex, NULL); | |
ab791dd1 | 146 | pthread_mutex_init(&type_cas_mutex, NULL); |
3aba2fdd NTND |
147 | if (show_stat) |
148 | pthread_mutex_init(&deepest_delta_mutex, NULL); | |
b8a2486f NTND |
149 | pthread_key_create(&key, NULL); |
150 | thread_data = xcalloc(nr_threads, sizeof(*thread_data)); | |
39539495 NTND |
151 | for (i = 0; i < nr_threads; i++) { |
152 | thread_data[i].pack_fd = open(curr_pack, O_RDONLY); | |
153 | if (thread_data[i].pack_fd == -1) | |
154 | die_errno(_("unable to open %s"), curr_pack); | |
155 | } | |
156 | ||
b8a2486f NTND |
157 | threads_active = 1; |
158 | } | |
159 | ||
160 | static void cleanup_thread(void) | |
161 | { | |
39539495 | 162 | int i; |
b8a2486f NTND |
163 | if (!threads_active) |
164 | return; | |
165 | threads_active = 0; | |
166 | pthread_mutex_destroy(&read_mutex); | |
167 | pthread_mutex_destroy(&counter_mutex); | |
168 | pthread_mutex_destroy(&work_mutex); | |
ab791dd1 | 169 | pthread_mutex_destroy(&type_cas_mutex); |
3aba2fdd NTND |
170 | if (show_stat) |
171 | pthread_mutex_destroy(&deepest_delta_mutex); | |
39539495 NTND |
172 | for (i = 0; i < nr_threads; i++) |
173 | close(thread_data[i].pack_fd); | |
b8a2486f NTND |
174 | pthread_key_delete(key); |
175 | free(thread_data); | |
176 | } | |
177 | ||
178 | #else | |
179 | ||
180 | #define read_lock() | |
181 | #define read_unlock() | |
182 | ||
183 | #define counter_lock() | |
184 | #define counter_unlock() | |
185 | ||
186 | #define work_lock() | |
187 | #define work_unlock() | |
188 | ||
3aba2fdd NTND |
189 | #define deepest_delta_lock() |
190 | #define deepest_delta_unlock() | |
191 | ||
e0e21283 EB |
192 | #define type_cas_lock() |
193 | #define type_cas_unlock() | |
194 | ||
b8a2486f NTND |
195 | #endif |
196 | ||
197 | ||
22410549 | 198 | static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options) |
0153be05 MK |
199 | { |
200 | if (!obj) | |
201 | return -1; | |
202 | ||
203 | if (type != OBJ_ANY && obj->type != type) | |
f2fd0760 | 204 | die(_("object type mismatch at %s"), oid_to_hex(&obj->oid)); |
0153be05 MK |
205 | |
206 | obj->flags |= FLAG_LINK; | |
207 | return 0; | |
208 | } | |
209 | ||
210 | /* The content of each linked object must have been checked | |
211 | or it must be already present in the object database */ | |
c6807a40 | 212 | static unsigned check_object(struct object *obj) |
0153be05 MK |
213 | { |
214 | if (!obj) | |
c6807a40 | 215 | return 0; |
0153be05 MK |
216 | |
217 | if (!(obj->flags & FLAG_LINK)) | |
c6807a40 | 218 | return 0; |
0153be05 MK |
219 | |
220 | if (!(obj->flags & FLAG_CHECKED)) { | |
221 | unsigned long size; | |
ed1c9977 | 222 | int type = sha1_object_info(obj->oid.hash, &size); |
77583e77 JK |
223 | if (type <= 0) |
224 | die(_("did not receive expected object %s"), | |
f2fd0760 | 225 | oid_to_hex(&obj->oid)); |
77583e77 JK |
226 | if (type != obj->type) |
227 | die(_("object %s: expected type %s, found %s"), | |
f2fd0760 | 228 | oid_to_hex(&obj->oid), |
77583e77 | 229 | typename(obj->type), typename(type)); |
0153be05 | 230 | obj->flags |= FLAG_CHECKED; |
c6807a40 | 231 | return 1; |
0153be05 | 232 | } |
c6807a40 NTND |
233 | |
234 | return 0; | |
0153be05 MK |
235 | } |
236 | ||
c6807a40 | 237 | static unsigned check_objects(void) |
0153be05 | 238 | { |
c6807a40 | 239 | unsigned i, max, foreign_nr = 0; |
0153be05 MK |
240 | |
241 | max = get_max_object_index(); | |
242 | for (i = 0; i < max; i++) | |
c6807a40 NTND |
243 | foreign_nr += check_object(get_indexed_object(i)); |
244 | return foreign_nr; | |
0153be05 MK |
245 | } |
246 | ||
247 | ||
636171cb | 248 | /* Discard current buffer used content. */ |
a6e8a767 | 249 | static void flush(void) |
636171cb NP |
250 | { |
251 | if (input_offset) { | |
252 | if (output_fd >= 0) | |
253 | write_or_die(output_fd, input_buffer, input_offset); | |
9126f009 | 254 | git_SHA1_Update(&input_ctx, input_buffer, input_offset); |
554a2636 | 255 | memmove(input_buffer, input_buffer + input_offset, input_len); |
636171cb NP |
256 | input_offset = 0; |
257 | } | |
258 | } | |
259 | ||
2d477051 NP |
260 | /* |
261 | * Make sure at least "min" bytes are available in the buffer, and | |
262 | * return the pointer to the buffer. | |
263 | */ | |
b89c4e93 | 264 | static void *fill(int min) |
9cf6d335 | 265 | { |
2d477051 NP |
266 | if (min <= input_len) |
267 | return input_buffer + input_offset; | |
268 | if (min > sizeof(input_buffer)) | |
c2b97ecf NTND |
269 | die(Q_("cannot fill %d byte", |
270 | "cannot fill %d bytes", | |
271 | min), | |
272 | min); | |
636171cb | 273 | flush(); |
2d477051 | 274 | do { |
8a912bcb | 275 | ssize_t ret = xread(input_fd, input_buffer + input_len, |
2d477051 NP |
276 | sizeof(input_buffer) - input_len); |
277 | if (ret <= 0) { | |
278 | if (!ret) | |
c2b97ecf NTND |
279 | die(_("early EOF")); |
280 | die_errno(_("read error on input")); | |
2d477051 NP |
281 | } |
282 | input_len += ret; | |
218558af NP |
283 | if (from_stdin) |
284 | display_throughput(progress, consumed_bytes + input_len); | |
2d477051 NP |
285 | } while (input_len < min); |
286 | return input_buffer; | |
287 | } | |
288 | ||
289 | static void use(int bytes) | |
290 | { | |
291 | if (bytes > input_len) | |
c2b97ecf | 292 | die(_("used more bytes than were available")); |
ee5743ce | 293 | input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes); |
2d477051 NP |
294 | input_len -= bytes; |
295 | input_offset += bytes; | |
d7dd0223 NP |
296 | |
297 | /* make sure off_t is sufficiently large not to wrap */ | |
c03c8315 | 298 | if (signed_add_overflows(consumed_bytes, bytes)) |
c2b97ecf | 299 | die(_("pack too large for current definition of off_t")); |
2d477051 | 300 | consumed_bytes += bytes; |
411481be JK |
301 | if (max_input_size && consumed_bytes > max_input_size) |
302 | die(_("pack exceeds maximum allowed size")); | |
2d477051 | 303 | } |
9cf6d335 | 304 | |
3bb72562 | 305 | static const char *open_pack_file(const char *pack_name) |
2d477051 | 306 | { |
e42797f5 NP |
307 | if (from_stdin) { |
308 | input_fd = 0; | |
309 | if (!pack_name) { | |
594fa999 JK |
310 | struct strbuf tmp_file = STRBUF_INIT; |
311 | output_fd = odb_mkstemp(&tmp_file, | |
6e180cdc | 312 | "pack/tmp_pack_XXXXXX"); |
594fa999 | 313 | pack_name = strbuf_detach(&tmp_file, NULL); |
892e723a | 314 | } else { |
e42797f5 | 315 | output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600); |
892e723a JK |
316 | if (output_fd < 0) |
317 | die_errno(_("unable to create '%s'"), pack_name); | |
318 | } | |
39539495 | 319 | nothread_data.pack_fd = output_fd; |
e42797f5 NP |
320 | } else { |
321 | input_fd = open(pack_name, O_RDONLY); | |
322 | if (input_fd < 0) | |
c2b97ecf | 323 | die_errno(_("cannot open packfile '%s'"), pack_name); |
e42797f5 | 324 | output_fd = -1; |
39539495 | 325 | nothread_data.pack_fd = input_fd; |
e42797f5 | 326 | } |
9126f009 | 327 | git_SHA1_Init(&input_ctx); |
e42797f5 | 328 | return pack_name; |
9cf6d335 SV |
329 | } |
330 | ||
331 | static void parse_pack_header(void) | |
332 | { | |
2d477051 | 333 | struct pack_header *hdr = fill(sizeof(struct pack_header)); |
9cf6d335 SV |
334 | |
335 | /* Header consistency check */ | |
9cf6d335 | 336 | if (hdr->hdr_signature != htonl(PACK_SIGNATURE)) |
c2b97ecf | 337 | die(_("pack signature mismatch")); |
d60fc1c8 | 338 | if (!pack_version_ok(hdr->hdr_version)) |
f350df42 | 339 | die(_("pack version %"PRIu32" unsupported"), |
6e1c2344 | 340 | ntohl(hdr->hdr_version)); |
9cf6d335 SV |
341 | |
342 | nr_objects = ntohl(hdr->hdr_entries); | |
2d477051 | 343 | use(sizeof(struct pack_header)); |
9cf6d335 SV |
344 | } |
345 | ||
fd3e6747 | 346 | static NORETURN void bad_object(off_t offset, const char *format, |
a4f3131c | 347 | ...) __attribute__((format (printf, 2, 3))); |
9cf6d335 | 348 | |
fd3e6747 | 349 | static NORETURN void bad_object(off_t offset, const char *format, ...) |
9cf6d335 SV |
350 | { |
351 | va_list params; | |
352 | char buf[1024]; | |
353 | ||
354 | va_start(params, format); | |
355 | vsnprintf(buf, sizeof(buf), format, params); | |
356 | va_end(params); | |
fd3e6747 NTND |
357 | die(_("pack has bad object at offset %"PRIuMAX": %s"), |
358 | (uintmax_t)offset, buf); | |
9cf6d335 SV |
359 | } |
360 | ||
b8a2486f NTND |
361 | static inline struct thread_local *get_thread_data(void) |
362 | { | |
363 | #ifndef NO_PTHREADS | |
364 | if (threads_active) | |
365 | return pthread_getspecific(key); | |
366 | assert(!threads_active && | |
367 | "This should only be reached when all threads are gone"); | |
368 | #endif | |
369 | return ¬hread_data; | |
370 | } | |
371 | ||
372 | #ifndef NO_PTHREADS | |
373 | static void set_thread_data(struct thread_local *data) | |
374 | { | |
375 | if (threads_active) | |
376 | pthread_setspecific(key, data); | |
377 | } | |
378 | #endif | |
379 | ||
2baad220 NTND |
380 | static struct base_data *alloc_base_data(void) |
381 | { | |
51a60f5b | 382 | struct base_data *base = xcalloc(1, sizeof(struct base_data)); |
2baad220 NTND |
383 | base->ref_last = -1; |
384 | base->ofs_last = -1; | |
385 | return base; | |
386 | } | |
387 | ||
6a87ed97 NP |
388 | static void free_base_data(struct base_data *c) |
389 | { | |
390 | if (c->data) { | |
391 | free(c->data); | |
392 | c->data = NULL; | |
b8a2486f | 393 | get_thread_data()->base_cache_used -= c->size; |
6a87ed97 NP |
394 | } |
395 | } | |
396 | ||
92392b4a SP |
397 | static void prune_base_data(struct base_data *retain) |
398 | { | |
8e24cbae | 399 | struct base_data *b; |
b8a2486f NTND |
400 | struct thread_local *data = get_thread_data(); |
401 | for (b = data->base_cache; | |
402 | data->base_cache_used > delta_base_cache_limit && b; | |
92392b4a | 403 | b = b->child) { |
6a87ed97 NP |
404 | if (b->data && b != retain) |
405 | free_base_data(b); | |
92392b4a SP |
406 | } |
407 | } | |
408 | ||
4a438cab SP |
409 | static void link_base_data(struct base_data *base, struct base_data *c) |
410 | { | |
411 | if (base) | |
412 | base->child = c; | |
413 | else | |
b8a2486f | 414 | get_thread_data()->base_cache = c; |
4a438cab SP |
415 | |
416 | c->base = base; | |
417 | c->child = NULL; | |
9441b61d | 418 | if (c->data) |
b8a2486f | 419 | get_thread_data()->base_cache_used += c->size; |
92392b4a | 420 | prune_base_data(c); |
4a438cab SP |
421 | } |
422 | ||
423 | static void unlink_base_data(struct base_data *c) | |
424 | { | |
425 | struct base_data *base = c->base; | |
426 | if (base) | |
427 | base->child = NULL; | |
428 | else | |
b8a2486f | 429 | get_thread_data()->base_cache = NULL; |
6a87ed97 | 430 | free_base_data(c); |
4a438cab SP |
431 | } |
432 | ||
681b07de NTND |
433 | static int is_delta_type(enum object_type type) |
434 | { | |
435 | return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA); | |
436 | } | |
437 | ||
da49a7da | 438 | static void *unpack_entry_data(off_t offset, unsigned long size, |
681b07de | 439 | enum object_type type, unsigned char *sha1) |
9cf6d335 | 440 | { |
9ec2dde9 | 441 | static char fixed_buf[8192]; |
7ce4721a | 442 | int status; |
ef49a7a0 | 443 | git_zstream stream; |
9ec2dde9 | 444 | void *buf; |
681b07de NTND |
445 | git_SHA_CTX c; |
446 | char hdr[32]; | |
447 | int hdrlen; | |
448 | ||
449 | if (!is_delta_type(type)) { | |
ef1286d3 | 450 | hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), size) + 1; |
681b07de NTND |
451 | git_SHA1_Init(&c); |
452 | git_SHA1_Update(&c, hdr, hdrlen); | |
453 | } else | |
454 | sha1 = NULL; | |
9ec2dde9 NTND |
455 | if (type == OBJ_BLOB && size > big_file_threshold) |
456 | buf = fixed_buf; | |
457 | else | |
a1e920a0 | 458 | buf = xmallocz(size); |
9cf6d335 SV |
459 | |
460 | memset(&stream, 0, sizeof(stream)); | |
7ce4721a | 461 | git_inflate_init(&stream); |
9cf6d335 | 462 | stream.next_out = buf; |
9ec2dde9 | 463 | stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size; |
9cf6d335 | 464 | |
7ce4721a | 465 | do { |
681b07de | 466 | unsigned char *last_out = stream.next_out; |
2d477051 NP |
467 | stream.next_in = fill(1); |
468 | stream.avail_in = input_len; | |
7ce4721a NP |
469 | status = git_inflate(&stream, 0); |
470 | use(input_len - stream.avail_in); | |
681b07de NTND |
471 | if (sha1) |
472 | git_SHA1_Update(&c, last_out, stream.next_out - last_out); | |
9ec2dde9 NTND |
473 | if (buf == fixed_buf) { |
474 | stream.next_out = buf; | |
475 | stream.avail_out = sizeof(fixed_buf); | |
476 | } | |
7ce4721a NP |
477 | } while (status == Z_OK); |
478 | if (stream.total_out != size || status != Z_STREAM_END) | |
c2b97ecf | 479 | bad_object(offset, _("inflate returned %d"), status); |
39c68542 | 480 | git_inflate_end(&stream); |
681b07de NTND |
481 | if (sha1) |
482 | git_SHA1_Final(sha1, &c); | |
9ec2dde9 | 483 | return buf == fixed_buf ? NULL : buf; |
9cf6d335 SV |
484 | } |
485 | ||
681b07de | 486 | static void *unpack_raw_entry(struct object_entry *obj, |
c6458e60 NTND |
487 | off_t *ofs_offset, |
488 | unsigned char *ref_sha1, | |
681b07de | 489 | unsigned char *sha1) |
9cf6d335 | 490 | { |
48fb7deb LT |
491 | unsigned char *p; |
492 | unsigned long size, c; | |
d7dd0223 | 493 | off_t base_offset; |
9cf6d335 | 494 | unsigned shift; |
ee5743ce | 495 | void *data; |
9cf6d335 | 496 | |
aa7e44bf | 497 | obj->idx.offset = consumed_bytes; |
1e4cd68c | 498 | input_crc32 = crc32(0, NULL, 0); |
2d477051 NP |
499 | |
500 | p = fill(1); | |
501 | c = *p; | |
502 | use(1); | |
503 | obj->type = (c >> 4) & 7; | |
9cf6d335 SV |
504 | size = (c & 15); |
505 | shift = 4; | |
506 | while (c & 0x80) { | |
2d477051 NP |
507 | p = fill(1); |
508 | c = *p; | |
509 | use(1); | |
48fb7deb | 510 | size += (c & 0x7f) << shift; |
9cf6d335 SV |
511 | shift += 7; |
512 | } | |
2d477051 | 513 | obj->size = size; |
9cf6d335 | 514 | |
2d477051 | 515 | switch (obj->type) { |
eb32d236 | 516 | case OBJ_REF_DELTA: |
c6458e60 | 517 | hashcpy(ref_sha1, fill(20)); |
2d477051 | 518 | use(20); |
53dda6ff NP |
519 | break; |
520 | case OBJ_OFS_DELTA: | |
2d477051 NP |
521 | p = fill(1); |
522 | c = *p; | |
523 | use(1); | |
53dda6ff NP |
524 | base_offset = c & 127; |
525 | while (c & 128) { | |
526 | base_offset += 1; | |
8723f216 | 527 | if (!base_offset || MSB(base_offset, 7)) |
c2b97ecf | 528 | bad_object(obj->idx.offset, _("offset value overflow for delta base object")); |
2d477051 NP |
529 | p = fill(1); |
530 | c = *p; | |
531 | use(1); | |
53dda6ff NP |
532 | base_offset = (base_offset << 7) + (c & 127); |
533 | } | |
c6458e60 NTND |
534 | *ofs_offset = obj->idx.offset - base_offset; |
535 | if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset) | |
c2b97ecf | 536 | bad_object(obj->idx.offset, _("delta base offset is out of bound")); |
53dda6ff | 537 | break; |
9cf6d335 SV |
538 | case OBJ_COMMIT: |
539 | case OBJ_TREE: | |
540 | case OBJ_BLOB: | |
541 | case OBJ_TAG: | |
9cf6d335 SV |
542 | break; |
543 | default: | |
c2b97ecf | 544 | bad_object(obj->idx.offset, _("unknown object type %d"), obj->type); |
9cf6d335 | 545 | } |
aa7e44bf | 546 | obj->hdr_size = consumed_bytes - obj->idx.offset; |
2d477051 | 547 | |
681b07de | 548 | data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1); |
aa7e44bf | 549 | obj->idx.crc32 = input_crc32; |
ee5743ce | 550 | return data; |
2d477051 NP |
551 | } |
552 | ||
8a2e163c NTND |
553 | static void *unpack_data(struct object_entry *obj, |
554 | int (*consume)(const unsigned char *, unsigned long, void *), | |
555 | void *cb_data) | |
2d477051 | 556 | { |
a91ef6e7 | 557 | off_t from = obj[0].idx.offset + obj[0].hdr_size; |
7171a0b0 | 558 | off_t len = obj[1].idx.offset - from; |
776ea370 | 559 | unsigned char *data, *inbuf; |
ef49a7a0 | 560 | git_zstream stream; |
776ea370 | 561 | int status; |
9cf6d335 | 562 | |
a1e920a0 | 563 | data = xmallocz(consume ? 64*1024 : obj->size); |
7171a0b0 | 564 | inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024); |
776ea370 | 565 | |
2d477051 | 566 | memset(&stream, 0, sizeof(stream)); |
776ea370 | 567 | git_inflate_init(&stream); |
2d477051 | 568 | stream.next_out = data; |
8a2e163c | 569 | stream.avail_out = consume ? 64*1024 : obj->size; |
776ea370 NP |
570 | |
571 | do { | |
7171a0b0 | 572 | ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024; |
53f52cd9 | 573 | n = xpread(get_thread_data()->pack_fd, inbuf, n, from); |
776ea370 | 574 | if (n < 0) |
c2b97ecf | 575 | die_errno(_("cannot pread pack file")); |
776ea370 | 576 | if (!n) |
7171a0b0 NTND |
577 | die(Q_("premature end of pack file, %"PRIuMAX" byte missing", |
578 | "premature end of pack file, %"PRIuMAX" bytes missing", | |
579 | (unsigned int)len), | |
580 | (uintmax_t)len); | |
776ea370 NP |
581 | from += n; |
582 | len -= n; | |
583 | stream.next_in = inbuf; | |
584 | stream.avail_in = n; | |
f8b09038 JK |
585 | if (!consume) |
586 | status = git_inflate(&stream, 0); | |
587 | else { | |
588 | do { | |
589 | status = git_inflate(&stream, 0); | |
590 | if (consume(data, stream.next_out - data, cb_data)) { | |
591 | free(inbuf); | |
592 | free(data); | |
593 | return NULL; | |
594 | } | |
595 | stream.next_out = data; | |
596 | stream.avail_out = 64*1024; | |
597 | } while (status == Z_OK && stream.avail_in); | |
8a2e163c | 598 | } |
776ea370 NP |
599 | } while (len && status == Z_OK && !stream.avail_in); |
600 | ||
601 | /* This has been inflated OK when first encountered, so... */ | |
602 | if (status != Z_STREAM_END || stream.total_out != obj->size) | |
c2b97ecf | 603 | die(_("serious inflate inconsistency")); |
776ea370 NP |
604 | |
605 | git_inflate_end(&stream); | |
606 | free(inbuf); | |
8a2e163c NTND |
607 | if (consume) { |
608 | free(data); | |
609 | data = NULL; | |
610 | } | |
9cf6d335 SV |
611 | return data; |
612 | } | |
613 | ||
8a2e163c NTND |
614 | static void *get_data_from_pack(struct object_entry *obj) |
615 | { | |
616 | return unpack_data(obj, NULL, NULL); | |
617 | } | |
618 | ||
c6458e60 NTND |
619 | static int compare_ofs_delta_bases(off_t offset1, off_t offset2, |
620 | enum object_type type1, | |
621 | enum object_type type2) | |
7218a215 JH |
622 | { |
623 | int cmp = type1 - type2; | |
624 | if (cmp) | |
625 | return cmp; | |
f0e7f11d JK |
626 | return offset1 < offset2 ? -1 : |
627 | offset1 > offset2 ? 1 : | |
628 | 0; | |
7218a215 JH |
629 | } |
630 | ||
c6458e60 | 631 | static int find_ofs_delta(const off_t offset, enum object_type type) |
9cf6d335 | 632 | { |
c6458e60 NTND |
633 | int first = 0, last = nr_ofs_deltas; |
634 | ||
635 | while (first < last) { | |
636 | int next = (first + last) / 2; | |
637 | struct ofs_delta_entry *delta = &ofs_deltas[next]; | |
638 | int cmp; | |
639 | ||
640 | cmp = compare_ofs_delta_bases(offset, delta->offset, | |
641 | type, objects[delta->obj_no].type); | |
642 | if (!cmp) | |
643 | return next; | |
644 | if (cmp < 0) { | |
645 | last = next; | |
646 | continue; | |
647 | } | |
648 | first = next+1; | |
649 | } | |
650 | return -first-1; | |
9cf6d335 SV |
651 | } |
652 | ||
c6458e60 NTND |
653 | static void find_ofs_delta_children(off_t offset, |
654 | int *first_index, int *last_index, | |
655 | enum object_type type) | |
9cf6d335 | 656 | { |
c6458e60 | 657 | int first = find_ofs_delta(offset, type); |
9cf6d335 | 658 | int last = first; |
c6458e60 | 659 | int end = nr_ofs_deltas - 1; |
9cf6d335 | 660 | |
6a87ed97 NP |
661 | if (first < 0) { |
662 | *first_index = 0; | |
663 | *last_index = -1; | |
664 | return; | |
665 | } | |
c6458e60 | 666 | while (first > 0 && ofs_deltas[first - 1].offset == offset) |
9cf6d335 | 667 | --first; |
c6458e60 NTND |
668 | while (last < end && ofs_deltas[last + 1].offset == offset) |
669 | ++last; | |
670 | *first_index = first; | |
671 | *last_index = last; | |
672 | } | |
673 | ||
674 | static int compare_ref_delta_bases(const unsigned char *sha1, | |
675 | const unsigned char *sha2, | |
676 | enum object_type type1, | |
677 | enum object_type type2) | |
678 | { | |
679 | int cmp = type1 - type2; | |
680 | if (cmp) | |
681 | return cmp; | |
682 | return hashcmp(sha1, sha2); | |
683 | } | |
684 | ||
685 | static int find_ref_delta(const unsigned char *sha1, enum object_type type) | |
686 | { | |
687 | int first = 0, last = nr_ref_deltas; | |
688 | ||
689 | while (first < last) { | |
690 | int next = (first + last) / 2; | |
691 | struct ref_delta_entry *delta = &ref_deltas[next]; | |
692 | int cmp; | |
693 | ||
694 | cmp = compare_ref_delta_bases(sha1, delta->sha1, | |
695 | type, objects[delta->obj_no].type); | |
696 | if (!cmp) | |
697 | return next; | |
698 | if (cmp < 0) { | |
699 | last = next; | |
700 | continue; | |
701 | } | |
702 | first = next+1; | |
703 | } | |
704 | return -first-1; | |
705 | } | |
706 | ||
707 | static void find_ref_delta_children(const unsigned char *sha1, | |
708 | int *first_index, int *last_index, | |
709 | enum object_type type) | |
710 | { | |
711 | int first = find_ref_delta(sha1, type); | |
712 | int last = first; | |
713 | int end = nr_ref_deltas - 1; | |
714 | ||
715 | if (first < 0) { | |
716 | *first_index = 0; | |
717 | *last_index = -1; | |
718 | return; | |
719 | } | |
720 | while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1)) | |
721 | --first; | |
722 | while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1)) | |
9cf6d335 SV |
723 | ++last; |
724 | *first_index = first; | |
725 | *last_index = last; | |
9cf6d335 SV |
726 | } |
727 | ||
4614043c NTND |
728 | struct compare_data { |
729 | struct object_entry *entry; | |
730 | struct git_istream *st; | |
731 | unsigned char *buf; | |
732 | unsigned long buf_size; | |
733 | }; | |
734 | ||
735 | static int compare_objects(const unsigned char *buf, unsigned long size, | |
736 | void *cb_data) | |
737 | { | |
738 | struct compare_data *data = cb_data; | |
739 | ||
740 | if (data->buf_size < size) { | |
741 | free(data->buf); | |
742 | data->buf = xmalloc(size); | |
743 | data->buf_size = size; | |
744 | } | |
745 | ||
746 | while (size) { | |
747 | ssize_t len = read_istream(data->st, data->buf, size); | |
748 | if (len == 0) | |
749 | die(_("SHA1 COLLISION FOUND WITH %s !"), | |
750 | sha1_to_hex(data->entry->idx.sha1)); | |
751 | if (len < 0) | |
752 | die(_("unable to read %s"), | |
753 | sha1_to_hex(data->entry->idx.sha1)); | |
754 | if (memcmp(buf, data->buf, len)) | |
755 | die(_("SHA1 COLLISION FOUND WITH %s !"), | |
756 | sha1_to_hex(data->entry->idx.sha1)); | |
757 | size -= len; | |
758 | buf += len; | |
759 | } | |
760 | return 0; | |
761 | } | |
762 | ||
763 | static int check_collison(struct object_entry *entry) | |
764 | { | |
765 | struct compare_data data; | |
766 | enum object_type type; | |
767 | unsigned long size; | |
768 | ||
769 | if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB) | |
770 | return -1; | |
771 | ||
772 | memset(&data, 0, sizeof(data)); | |
773 | data.entry = entry; | |
774 | data.st = open_istream(entry->idx.sha1, &type, &size, NULL); | |
775 | if (!data.st) | |
776 | return -1; | |
777 | if (size != entry->size || type != entry->type) | |
778 | die(_("SHA1 COLLISION FOUND WITH %s !"), | |
779 | sha1_to_hex(entry->idx.sha1)); | |
780 | unpack_data(entry, compare_objects, &data); | |
781 | close_istream(data.st); | |
782 | free(data.buf); | |
783 | return 0; | |
784 | } | |
785 | ||
9ec2dde9 NTND |
786 | static void sha1_object(const void *data, struct object_entry *obj_entry, |
787 | unsigned long size, enum object_type type, | |
788 | const unsigned char *sha1) | |
9cf6d335 | 789 | { |
9ec2dde9 | 790 | void *new_data = NULL; |
29401e15 | 791 | int collision_test_needed = 0; |
9ec2dde9 NTND |
792 | |
793 | assert(data || obj_entry); | |
794 | ||
29401e15 JK |
795 | if (startup_info->have_repository) { |
796 | read_lock(); | |
797 | collision_test_needed = has_sha1_file_with_flags(sha1, HAS_SHA1_QUICK); | |
798 | read_unlock(); | |
799 | } | |
4614043c NTND |
800 | |
801 | if (collision_test_needed && !data) { | |
802 | read_lock(); | |
803 | if (!check_collison(obj_entry)) | |
804 | collision_test_needed = 0; | |
805 | read_unlock(); | |
806 | } | |
807 | if (collision_test_needed) { | |
8685da42 NP |
808 | void *has_data; |
809 | enum object_type has_type; | |
810 | unsigned long has_size; | |
4614043c NTND |
811 | read_lock(); |
812 | has_type = sha1_object_info(sha1, &has_size); | |
51054177 JK |
813 | if (has_type < 0) |
814 | die(_("cannot read existing object info %s"), sha1_to_hex(sha1)); | |
4614043c NTND |
815 | if (has_type != type || has_size != size) |
816 | die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1)); | |
8685da42 | 817 | has_data = read_sha1_file(sha1, &has_type, &has_size); |
b8a2486f | 818 | read_unlock(); |
4614043c NTND |
819 | if (!data) |
820 | data = new_data = get_data_from_pack(obj_entry); | |
8685da42 | 821 | if (!has_data) |
c2b97ecf | 822 | die(_("cannot read existing object %s"), sha1_to_hex(sha1)); |
8685da42 NP |
823 | if (size != has_size || type != has_type || |
824 | memcmp(data, has_data, size) != 0) | |
c2b97ecf | 825 | die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1)); |
bbf4b41b | 826 | free(has_data); |
4614043c | 827 | } |
b8a2486f | 828 | |
0153be05 | 829 | if (strict) { |
b8a2486f | 830 | read_lock(); |
0153be05 MK |
831 | if (type == OBJ_BLOB) { |
832 | struct blob *blob = lookup_blob(sha1); | |
833 | if (blob) | |
834 | blob->object.flags |= FLAG_CHECKED; | |
835 | else | |
c2b97ecf | 836 | die(_("invalid blob object %s"), sha1_to_hex(sha1)); |
0153be05 MK |
837 | } else { |
838 | struct object *obj; | |
839 | int eaten; | |
840 | void *buf = (void *) data; | |
841 | ||
920734b0 | 842 | assert(data && "data can only be NULL for large _blobs_"); |
9ec2dde9 | 843 | |
0153be05 MK |
844 | /* |
845 | * we do not need to free the memory here, as the | |
846 | * buf is deleted by the caller. | |
847 | */ | |
848 | obj = parse_object_buffer(sha1, type, size, buf, &eaten); | |
849 | if (!obj) | |
c2b97ecf | 850 | die(_("invalid %s"), typename(type)); |
c6807a40 | 851 | if (do_fsck_object && |
22410549 | 852 | fsck_object(obj, buf, size, &fsck_options)) |
c2b97ecf | 853 | die(_("Error in object")); |
22410549 | 854 | if (fsck_walk(obj, NULL, &fsck_options)) |
f2fd0760 | 855 | die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid)); |
0153be05 MK |
856 | |
857 | if (obj->type == OBJ_TREE) { | |
858 | struct tree *item = (struct tree *) obj; | |
859 | item->buffer = NULL; | |
6e454b9a | 860 | obj->parsed = 0; |
0153be05 MK |
861 | } |
862 | if (obj->type == OBJ_COMMIT) { | |
863 | struct commit *commit = (struct commit *) obj; | |
8597ea3a | 864 | if (detach_commit_buffer(commit, NULL) != data) |
0fb370da | 865 | die("BUG: parse_object_buffer transmogrified our buffer"); |
0153be05 MK |
866 | } |
867 | obj->flags |= FLAG_CHECKED; | |
868 | } | |
b8a2486f | 869 | read_unlock(); |
0153be05 | 870 | } |
9ec2dde9 NTND |
871 | |
872 | free(new_data); | |
9cf6d335 SV |
873 | } |
874 | ||
20e95d0a NTND |
875 | /* |
876 | * This function is part of find_unresolved_deltas(). There are two | |
877 | * walkers going in the opposite ways. | |
878 | * | |
879 | * The first one in find_unresolved_deltas() traverses down from | |
880 | * parent node to children, deflating nodes along the way. However, | |
881 | * memory for deflated nodes is limited by delta_base_cache_limit, so | |
882 | * at some point parent node's deflated content may be freed. | |
883 | * | |
884 | * The second walker is this function, which goes from current node up | |
885 | * to top parent if necessary to deflate the node. In normal | |
886 | * situation, its parent node would be already deflated, so it just | |
887 | * needs to apply delta. | |
888 | * | |
889 | * In the worst case scenario, parent node is no longer deflated because | |
890 | * we're running out of delta_base_cache_limit; we need to re-deflate | |
891 | * parents, possibly up to the top base. | |
892 | * | |
893 | * All deflated objects here are subject to be freed if we exceed | |
894 | * delta_base_cache_limit, just like in find_unresolved_deltas(), we | |
895 | * just need to make sure the last node is not freed. | |
896 | */ | |
92392b4a SP |
897 | static void *get_base_data(struct base_data *c) |
898 | { | |
899 | if (!c->data) { | |
900 | struct object_entry *obj = c->obj; | |
20e95d0a NTND |
901 | struct base_data **delta = NULL; |
902 | int delta_nr = 0, delta_alloc = 0; | |
92392b4a | 903 | |
20e95d0a NTND |
904 | while (is_delta_type(c->obj->type) && !c->data) { |
905 | ALLOC_GROW(delta, delta_nr + 1, delta_alloc); | |
906 | delta[delta_nr++] = c; | |
907 | c = c->base; | |
908 | } | |
909 | if (!delta_nr) { | |
910 | c->data = get_data_from_pack(obj); | |
911 | c->size = obj->size; | |
b8a2486f | 912 | get_thread_data()->base_cache_used += c->size; |
20e95d0a NTND |
913 | prune_base_data(c); |
914 | } | |
915 | for (; delta_nr > 0; delta_nr--) { | |
916 | void *base, *raw; | |
917 | c = delta[delta_nr - 1]; | |
918 | obj = c->obj; | |
919 | base = get_base_data(c->base); | |
920 | raw = get_data_from_pack(obj); | |
92392b4a SP |
921 | c->data = patch_delta( |
922 | base, c->base->size, | |
923 | raw, obj->size, | |
924 | &c->size); | |
925 | free(raw); | |
926 | if (!c->data) | |
c2b97ecf | 927 | bad_object(obj->idx.offset, _("failed to apply delta")); |
b8a2486f | 928 | get_thread_data()->base_cache_used += c->size; |
20e95d0a | 929 | prune_base_data(c); |
9441b61d | 930 | } |
20e95d0a | 931 | free(delta); |
92392b4a SP |
932 | } |
933 | return c->data; | |
934 | } | |
935 | ||
f41aebd4 | 936 | static void resolve_delta(struct object_entry *delta_obj, |
9441b61d | 937 | struct base_data *base, struct base_data *result) |
9cf6d335 | 938 | { |
ce3f6dc6 | 939 | void *base_data, *delta_data; |
9cf6d335 | 940 | |
3aba2fdd | 941 | if (show_stat) { |
41730576 NTND |
942 | int i = delta_obj - objects; |
943 | int j = base->obj - objects; | |
944 | obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1; | |
3aba2fdd | 945 | deepest_delta_lock(); |
41730576 NTND |
946 | if (deepest_delta < obj_stat[i].delta_depth) |
947 | deepest_delta = obj_stat[i].delta_depth; | |
3aba2fdd | 948 | deepest_delta_unlock(); |
41730576 | 949 | obj_stat[i].base_object_no = j; |
3aba2fdd | 950 | } |
636171cb | 951 | delta_data = get_data_from_pack(delta_obj); |
ce3f6dc6 | 952 | base_data = get_base_data(base); |
9441b61d | 953 | result->obj = delta_obj; |
ce3f6dc6 NP |
954 | result->data = patch_delta(base_data, base->size, |
955 | delta_data, delta_obj->size, &result->size); | |
9cf6d335 | 956 | free(delta_data); |
9441b61d | 957 | if (!result->data) |
c2b97ecf | 958 | bad_object(delta_obj->idx.offset, _("failed to apply delta")); |
681b07de NTND |
959 | hash_sha1_file(result->data, result->size, |
960 | typename(delta_obj->real_type), delta_obj->idx.sha1); | |
9ec2dde9 | 961 | sha1_object(result->data, NULL, result->size, delta_obj->real_type, |
9441b61d | 962 | delta_obj->idx.sha1); |
b8a2486f | 963 | counter_lock(); |
636171cb | 964 | nr_resolved_deltas++; |
b8a2486f | 965 | counter_unlock(); |
9441b61d NP |
966 | } |
967 | ||
ab791dd1 JK |
968 | /* |
969 | * Standard boolean compare-and-swap: atomically check whether "*type" is | |
970 | * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched | |
971 | * and return false. | |
972 | */ | |
41730576 | 973 | static int compare_and_swap_type(signed char *type, |
ab791dd1 JK |
974 | enum object_type want, |
975 | enum object_type set) | |
976 | { | |
977 | enum object_type old; | |
978 | ||
979 | type_cas_lock(); | |
980 | old = *type; | |
981 | if (old == want) | |
982 | *type = set; | |
983 | type_cas_unlock(); | |
984 | ||
985 | return old == want; | |
986 | } | |
987 | ||
2baad220 NTND |
988 | static struct base_data *find_unresolved_deltas_1(struct base_data *base, |
989 | struct base_data *prev_base) | |
9441b61d | 990 | { |
2baad220 | 991 | if (base->ref_last == -1 && base->ofs_last == -1) { |
c6458e60 NTND |
992 | find_ref_delta_children(base->obj->idx.sha1, |
993 | &base->ref_first, &base->ref_last, | |
994 | OBJ_REF_DELTA); | |
9441b61d | 995 | |
c6458e60 NTND |
996 | find_ofs_delta_children(base->obj->idx.offset, |
997 | &base->ofs_first, &base->ofs_last, | |
998 | OBJ_OFS_DELTA); | |
4a438cab | 999 | |
2baad220 NTND |
1000 | if (base->ref_last == -1 && base->ofs_last == -1) { |
1001 | free(base->data); | |
1002 | return NULL; | |
1003 | } | |
53dda6ff | 1004 | |
2baad220 NTND |
1005 | link_base_data(prev_base, base); |
1006 | } | |
4a438cab | 1007 | |
2baad220 | 1008 | if (base->ref_first <= base->ref_last) { |
c6458e60 | 1009 | struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no; |
2baad220 | 1010 | struct base_data *result = alloc_base_data(); |
7218a215 | 1011 | |
ab791dd1 JK |
1012 | if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA, |
1013 | base->obj->real_type)) | |
1014 | die("BUG: child->real_type != OBJ_REF_DELTA"); | |
1015 | ||
2baad220 NTND |
1016 | resolve_delta(child, base, result); |
1017 | if (base->ref_first == base->ref_last && base->ofs_last == -1) | |
7218a215 | 1018 | free_base_data(base); |
2baad220 NTND |
1019 | |
1020 | base->ref_first++; | |
1021 | return result; | |
53dda6ff NP |
1022 | } |
1023 | ||
2baad220 | 1024 | if (base->ofs_first <= base->ofs_last) { |
c6458e60 | 1025 | struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no; |
2baad220 | 1026 | struct base_data *result = alloc_base_data(); |
7218a215 JH |
1027 | |
1028 | assert(child->real_type == OBJ_OFS_DELTA); | |
ab791dd1 | 1029 | child->real_type = base->obj->real_type; |
2baad220 NTND |
1030 | resolve_delta(child, base, result); |
1031 | if (base->ofs_first == base->ofs_last) | |
7218a215 | 1032 | free_base_data(base); |
2baad220 NTND |
1033 | |
1034 | base->ofs_first++; | |
1035 | return result; | |
9cf6d335 | 1036 | } |
53dda6ff | 1037 | |
9441b61d | 1038 | unlink_base_data(base); |
2baad220 NTND |
1039 | return NULL; |
1040 | } | |
1041 | ||
1042 | static void find_unresolved_deltas(struct base_data *base) | |
1043 | { | |
1044 | struct base_data *new_base, *prev_base = NULL; | |
1045 | for (;;) { | |
1046 | new_base = find_unresolved_deltas_1(base, prev_base); | |
1047 | ||
1048 | if (new_base) { | |
1049 | prev_base = base; | |
1050 | base = new_base; | |
1051 | } else { | |
1052 | free(base); | |
1053 | base = prev_base; | |
1054 | if (!base) | |
1055 | return; | |
1056 | prev_base = base->base; | |
1057 | } | |
1058 | } | |
9cf6d335 SV |
1059 | } |
1060 | ||
c6458e60 | 1061 | static int compare_ofs_delta_entry(const void *a, const void *b) |
9cf6d335 | 1062 | { |
c6458e60 NTND |
1063 | const struct ofs_delta_entry *delta_a = a; |
1064 | const struct ofs_delta_entry *delta_b = b; | |
7218a215 | 1065 | |
f0e7f11d JK |
1066 | return delta_a->offset < delta_b->offset ? -1 : |
1067 | delta_a->offset > delta_b->offset ? 1 : | |
1068 | 0; | |
c6458e60 NTND |
1069 | } |
1070 | ||
1071 | static int compare_ref_delta_entry(const void *a, const void *b) | |
9cf6d335 | 1072 | { |
c6458e60 NTND |
1073 | const struct ref_delta_entry *delta_a = a; |
1074 | const struct ref_delta_entry *delta_b = b; | |
7218a215 | 1075 | |
c6458e60 | 1076 | return hashcmp(delta_a->sha1, delta_b->sha1); |
9cf6d335 SV |
1077 | } |
1078 | ||
5272f755 NTND |
1079 | static void resolve_base(struct object_entry *obj) |
1080 | { | |
1081 | struct base_data *base_obj = alloc_base_data(); | |
1082 | base_obj->obj = obj; | |
1083 | base_obj->data = NULL; | |
1084 | find_unresolved_deltas(base_obj); | |
1085 | } | |
1086 | ||
b8a2486f NTND |
1087 | #ifndef NO_PTHREADS |
1088 | static void *threaded_second_pass(void *data) | |
1089 | { | |
1090 | set_thread_data(data); | |
1091 | for (;;) { | |
1092 | int i; | |
8f82aad4 | 1093 | counter_lock(); |
b8a2486f | 1094 | display_progress(progress, nr_resolved_deltas); |
8f82aad4 TR |
1095 | counter_unlock(); |
1096 | work_lock(); | |
b8a2486f NTND |
1097 | while (nr_dispatched < nr_objects && |
1098 | is_delta_type(objects[nr_dispatched].type)) | |
1099 | nr_dispatched++; | |
1100 | if (nr_dispatched >= nr_objects) { | |
1101 | work_unlock(); | |
1102 | break; | |
1103 | } | |
1104 | i = nr_dispatched++; | |
1105 | work_unlock(); | |
1106 | ||
1107 | resolve_base(&objects[i]); | |
1108 | } | |
1109 | return NULL; | |
1110 | } | |
1111 | #endif | |
1112 | ||
5272f755 NTND |
1113 | /* |
1114 | * First pass: | |
1115 | * - find locations of all objects; | |
1116 | * - calculate SHA1 of all non-delta objects; | |
1117 | * - remember base (SHA1 or offset) for all deltas. | |
1118 | */ | |
2d477051 | 1119 | static void parse_pack_objects(unsigned char *sha1) |
9cf6d335 | 1120 | { |
9ec2dde9 | 1121 | int i, nr_delays = 0; |
c6458e60 NTND |
1122 | struct ofs_delta_entry *ofs_delta = ofs_deltas; |
1123 | unsigned char ref_delta_sha1[20]; | |
2d477051 | 1124 | struct stat st; |
9cf6d335 | 1125 | |
13aaf148 | 1126 | if (verbose) |
29e63ed3 | 1127 | progress = start_progress( |
c2b97ecf | 1128 | from_stdin ? _("Receiving objects") : _("Indexing objects"), |
29e63ed3 | 1129 | nr_objects); |
9cf6d335 SV |
1130 | for (i = 0; i < nr_objects; i++) { |
1131 | struct object_entry *obj = &objects[i]; | |
c6458e60 NTND |
1132 | void *data = unpack_raw_entry(obj, &ofs_delta->offset, |
1133 | ref_delta_sha1, obj->idx.sha1); | |
9cf6d335 | 1134 | obj->real_type = obj->type; |
c6458e60 NTND |
1135 | if (obj->type == OBJ_OFS_DELTA) { |
1136 | nr_ofs_deltas++; | |
1137 | ofs_delta->obj_no = i; | |
1138 | ofs_delta++; | |
1139 | } else if (obj->type == OBJ_REF_DELTA) { | |
1140 | ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc); | |
1141 | hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1); | |
1142 | ref_deltas[nr_ref_deltas].obj_no = i; | |
1143 | nr_ref_deltas++; | |
9ec2dde9 NTND |
1144 | } else if (!data) { |
1145 | /* large blobs, check later */ | |
1146 | obj->real_type = OBJ_BAD; | |
1147 | nr_delays++; | |
9cf6d335 | 1148 | } else |
9ec2dde9 | 1149 | sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1); |
9cf6d335 | 1150 | free(data); |
4d4fcc54 | 1151 | display_progress(progress, i+1); |
9cf6d335 | 1152 | } |
aa7e44bf | 1153 | objects[i].idx.offset = consumed_bytes; |
4d4fcc54 | 1154 | stop_progress(&progress); |
2d477051 NP |
1155 | |
1156 | /* Check pack integrity */ | |
636171cb | 1157 | flush(); |
9126f009 | 1158 | git_SHA1_Final(sha1, &input_ctx); |
2d477051 | 1159 | if (hashcmp(fill(20), sha1)) |
c2b97ecf | 1160 | die(_("pack is corrupted (SHA1 mismatch)")); |
9bee2478 | 1161 | use(20); |
2d477051 NP |
1162 | |
1163 | /* If input_fd is a file, we should have reached its end now. */ | |
1164 | if (fstat(input_fd, &st)) | |
c2b97ecf | 1165 | die_errno(_("cannot fstat packfile")); |
fa257b05 JS |
1166 | if (S_ISREG(st.st_mode) && |
1167 | lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size) | |
c2b97ecf | 1168 | die(_("pack has junk at the end")); |
9ec2dde9 NTND |
1169 | |
1170 | for (i = 0; i < nr_objects; i++) { | |
1171 | struct object_entry *obj = &objects[i]; | |
1172 | if (obj->real_type != OBJ_BAD) | |
1173 | continue; | |
1174 | obj->real_type = obj->type; | |
1175 | sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1); | |
1176 | nr_delays--; | |
1177 | } | |
1178 | if (nr_delays) | |
1179 | die(_("confusion beyond insanity in parse_pack_objects()")); | |
5272f755 NTND |
1180 | } |
1181 | ||
1182 | /* | |
1183 | * Second pass: | |
1184 | * - for all non-delta objects, look if it is used as a base for | |
1185 | * deltas; | |
1186 | * - if used as a base, uncompress the object and apply all deltas, | |
1187 | * recursively checking if the resulting object is used as a base | |
1188 | * for some more deltas. | |
1189 | */ | |
1190 | static void resolve_deltas(void) | |
1191 | { | |
1192 | int i; | |
9cf6d335 | 1193 | |
c6458e60 | 1194 | if (!nr_ofs_deltas && !nr_ref_deltas) |
3c9af366 NP |
1195 | return; |
1196 | ||
53dda6ff | 1197 | /* Sort deltas by base SHA1/offset for fast searching */ |
9ed0d8d6 RS |
1198 | QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry); |
1199 | QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry); | |
9cf6d335 | 1200 | |
e376f17f | 1201 | if (verbose || show_resolving_progress) |
c6458e60 NTND |
1202 | progress = start_progress(_("Resolving deltas"), |
1203 | nr_ref_deltas + nr_ofs_deltas); | |
b8a2486f NTND |
1204 | |
1205 | #ifndef NO_PTHREADS | |
1206 | nr_dispatched = 0; | |
1207 | if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) { | |
1208 | init_thread(); | |
1209 | for (i = 0; i < nr_threads; i++) { | |
1210 | int ret = pthread_create(&thread_data[i].thread, NULL, | |
1211 | threaded_second_pass, thread_data + i); | |
1212 | if (ret) | |
f350df42 NTND |
1213 | die(_("unable to create thread: %s"), |
1214 | strerror(ret)); | |
b8a2486f NTND |
1215 | } |
1216 | for (i = 0; i < nr_threads; i++) | |
1217 | pthread_join(thread_data[i].thread, NULL); | |
1218 | cleanup_thread(); | |
1219 | return; | |
1220 | } | |
1221 | #endif | |
1222 | ||
9cf6d335 SV |
1223 | for (i = 0; i < nr_objects; i++) { |
1224 | struct object_entry *obj = &objects[i]; | |
9cf6d335 | 1225 | |
4f8ec74e | 1226 | if (is_delta_type(obj->type)) |
9cf6d335 | 1227 | continue; |
5272f755 | 1228 | resolve_base(obj); |
4d4fcc54 | 1229 | display_progress(progress, nr_resolved_deltas); |
9cf6d335 | 1230 | } |
636171cb NP |
1231 | } |
1232 | ||
5272f755 NTND |
1233 | /* |
1234 | * Third pass: | |
1235 | * - append objects to convert thin pack to full pack if required | |
1236 | * - write the final 20-byte SHA-1 | |
1237 | */ | |
781d9306 | 1238 | static void fix_unresolved_deltas(struct sha1file *f); |
5272f755 NTND |
1239 | static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1) |
1240 | { | |
c6458e60 | 1241 | if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) { |
5272f755 NTND |
1242 | stop_progress(&progress); |
1243 | /* Flush remaining pack final 20-byte SHA1. */ | |
1244 | flush(); | |
1245 | return; | |
1246 | } | |
1247 | ||
1248 | if (fix_thin_pack) { | |
1249 | struct sha1file *f; | |
1250 | unsigned char read_sha1[20], tail_sha1[20]; | |
5c3459fc | 1251 | struct strbuf msg = STRBUF_INIT; |
c6458e60 | 1252 | int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas; |
5272f755 NTND |
1253 | int nr_objects_initial = nr_objects; |
1254 | if (nr_unresolved <= 0) | |
cc13431a | 1255 | die(_("confusion beyond insanity")); |
2756ca43 | 1256 | REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1); |
57165db0 JK |
1257 | memset(objects + nr_objects + 1, 0, |
1258 | nr_unresolved * sizeof(*objects)); | |
5272f755 | 1259 | f = sha1fd(output_fd, curr_pack); |
781d9306 | 1260 | fix_unresolved_deltas(f); |
71d99b81 VA |
1261 | strbuf_addf(&msg, Q_("completed with %d local object", |
1262 | "completed with %d local objects", | |
1263 | nr_objects - nr_objects_initial), | |
5c3459fc NTND |
1264 | nr_objects - nr_objects_initial); |
1265 | stop_progress_msg(&progress, msg.buf); | |
1266 | strbuf_release(&msg); | |
5272f755 NTND |
1267 | sha1close(f, tail_sha1, 0); |
1268 | hashcpy(read_sha1, pack_sha1); | |
1269 | fixup_pack_header_footer(output_fd, pack_sha1, | |
1270 | curr_pack, nr_objects, | |
1271 | read_sha1, consumed_bytes-20); | |
1272 | if (hashcmp(read_sha1, tail_sha1) != 0) | |
f350df42 NTND |
1273 | die(_("Unexpected tail checksum for %s " |
1274 | "(disk corruption?)"), curr_pack); | |
5272f755 | 1275 | } |
c6458e60 | 1276 | if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas) |
cc13431a JH |
1277 | die(Q_("pack has %d unresolved delta", |
1278 | "pack has %d unresolved deltas", | |
c6458e60 NTND |
1279 | nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas), |
1280 | nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas); | |
5272f755 NTND |
1281 | } |
1282 | ||
8522148f | 1283 | static int write_compressed(struct sha1file *f, void *in, unsigned int size) |
636171cb | 1284 | { |
ef49a7a0 | 1285 | git_zstream stream; |
7734d7f2 NP |
1286 | int status; |
1287 | unsigned char outbuf[4096]; | |
636171cb | 1288 | |
55bb5c91 | 1289 | git_deflate_init(&stream, zlib_compression_level); |
636171cb NP |
1290 | stream.next_in = in; |
1291 | stream.avail_in = size; | |
636171cb | 1292 | |
7734d7f2 NP |
1293 | do { |
1294 | stream.next_out = outbuf; | |
1295 | stream.avail_out = sizeof(outbuf); | |
55bb5c91 | 1296 | status = git_deflate(&stream, Z_FINISH); |
7734d7f2 NP |
1297 | sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out); |
1298 | } while (status == Z_OK); | |
1299 | ||
1300 | if (status != Z_STREAM_END) | |
c2b97ecf | 1301 | die(_("unable to deflate appended object (%d)"), status); |
636171cb | 1302 | size = stream.total_out; |
55bb5c91 | 1303 | git_deflate_end(&stream); |
636171cb NP |
1304 | return size; |
1305 | } | |
1306 | ||
8522148f | 1307 | static struct object_entry *append_obj_to_pack(struct sha1file *f, |
03993e13 | 1308 | const unsigned char *sha1, void *buf, |
636171cb NP |
1309 | unsigned long size, enum object_type type) |
1310 | { | |
1311 | struct object_entry *obj = &objects[nr_objects++]; | |
1312 | unsigned char header[10]; | |
1313 | unsigned long s = size; | |
1314 | int n = 0; | |
1315 | unsigned char c = (type << 4) | (s & 15); | |
1316 | s >>= 4; | |
1317 | while (s) { | |
1318 | header[n++] = c | 0x80; | |
1319 | c = s & 0x7f; | |
1320 | s >>= 7; | |
1321 | } | |
1322 | header[n++] = c; | |
8522148f NP |
1323 | crc32_begin(f); |
1324 | sha1write(f, header, n); | |
72de2883 BS |
1325 | obj[0].size = size; |
1326 | obj[0].hdr_size = n; | |
1327 | obj[0].type = type; | |
1328 | obj[0].real_type = type; | |
aa7e44bf | 1329 | obj[1].idx.offset = obj[0].idx.offset + n; |
8522148f NP |
1330 | obj[1].idx.offset += write_compressed(f, buf, size); |
1331 | obj[0].idx.crc32 = crc32_end(f); | |
838cd346 | 1332 | sha1flush(f); |
aa7e44bf | 1333 | hashcpy(obj->idx.sha1, sha1); |
03993e13 | 1334 | return obj; |
636171cb NP |
1335 | } |
1336 | ||
1337 | static int delta_pos_compare(const void *_a, const void *_b) | |
1338 | { | |
c6458e60 NTND |
1339 | struct ref_delta_entry *a = *(struct ref_delta_entry **)_a; |
1340 | struct ref_delta_entry *b = *(struct ref_delta_entry **)_b; | |
636171cb NP |
1341 | return a->obj_no - b->obj_no; |
1342 | } | |
9cf6d335 | 1343 | |
781d9306 | 1344 | static void fix_unresolved_deltas(struct sha1file *f) |
636171cb | 1345 | { |
c6458e60 | 1346 | struct ref_delta_entry **sorted_by_pos; |
781d9306 | 1347 | int i; |
636171cb NP |
1348 | |
1349 | /* | |
1350 | * Since many unresolved deltas may well be themselves base objects | |
1351 | * for more unresolved deltas, we really want to include the | |
1352 | * smallest number of base objects that would cover as much delta | |
1353 | * as possible by picking the | |
1354 | * trunc deltas first, allowing for other deltas to resolve without | |
1355 | * additional base objects. Since most base objects are to be found | |
1356 | * before deltas depending on them, a good heuristic is to start | |
1357 | * resolving deltas in the same order as their position in the pack. | |
1358 | */ | |
b32fa95f | 1359 | ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas); |
c6458e60 | 1360 | for (i = 0; i < nr_ref_deltas; i++) |
781d9306 | 1361 | sorted_by_pos[i] = &ref_deltas[i]; |
9ed0d8d6 | 1362 | QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare); |
636171cb | 1363 | |
781d9306 | 1364 | for (i = 0; i < nr_ref_deltas; i++) { |
c6458e60 | 1365 | struct ref_delta_entry *d = sorted_by_pos[i]; |
21666f1a | 1366 | enum object_type type; |
2baad220 | 1367 | struct base_data *base_obj = alloc_base_data(); |
636171cb NP |
1368 | |
1369 | if (objects[d->obj_no].real_type != OBJ_REF_DELTA) | |
1370 | continue; | |
c6458e60 | 1371 | base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size); |
2baad220 | 1372 | if (!base_obj->data) |
636171cb | 1373 | continue; |
636171cb | 1374 | |
c6458e60 | 1375 | if (check_sha1_signature(d->sha1, base_obj->data, |
2baad220 | 1376 | base_obj->size, typename(type))) |
c6458e60 NTND |
1377 | die(_("local object %s is corrupt"), sha1_to_hex(d->sha1)); |
1378 | base_obj->obj = append_obj_to_pack(f, d->sha1, | |
2baad220 NTND |
1379 | base_obj->data, base_obj->size, type); |
1380 | find_unresolved_deltas(base_obj); | |
4d4fcc54 | 1381 | display_progress(progress, nr_resolved_deltas); |
636171cb NP |
1382 | } |
1383 | free(sorted_by_pos); | |
1384 | } | |
1385 | ||
e42797f5 NP |
1386 | static void final(const char *final_pack_name, const char *curr_pack_name, |
1387 | const char *final_index_name, const char *curr_index_name, | |
b8077709 | 1388 | const char *keep_name, const char *keep_msg, |
e42797f5 NP |
1389 | unsigned char *sha1) |
1390 | { | |
3a55602e | 1391 | const char *report = "pack"; |
f2075480 JK |
1392 | struct strbuf pack_name = STRBUF_INIT; |
1393 | struct strbuf index_name = STRBUF_INIT; | |
1394 | struct strbuf keep_name_buf = STRBUF_INIT; | |
e42797f5 NP |
1395 | int err; |
1396 | ||
1397 | if (!from_stdin) { | |
1398 | close(input_fd); | |
1399 | } else { | |
4c81b03e | 1400 | fsync_or_die(output_fd, curr_pack_name); |
e42797f5 NP |
1401 | err = close(output_fd); |
1402 | if (err) | |
c2b97ecf | 1403 | die_errno(_("error while closing pack file")); |
e42797f5 NP |
1404 | } |
1405 | ||
b8077709 SP |
1406 | if (keep_msg) { |
1407 | int keep_fd, keep_msg_len = strlen(keep_msg); | |
6e180cdc JH |
1408 | |
1409 | if (!keep_name) | |
f2075480 | 1410 | keep_name = odb_pack_name(&keep_name_buf, sha1, "keep"); |
6e180cdc | 1411 | |
f2075480 | 1412 | keep_fd = odb_pack_keep(keep_name); |
9ca4a201 NP |
1413 | if (keep_fd < 0) { |
1414 | if (errno != EEXIST) | |
c2b97ecf | 1415 | die_errno(_("cannot write keep file '%s'"), |
f2075480 | 1416 | keep_name); |
9ca4a201 NP |
1417 | } else { |
1418 | if (keep_msg_len > 0) { | |
1419 | write_or_die(keep_fd, keep_msg, keep_msg_len); | |
1420 | write_or_die(keep_fd, "\n", 1); | |
1421 | } | |
91c8d590 | 1422 | if (close(keep_fd) != 0) |
c2b97ecf | 1423 | die_errno(_("cannot close written keep file '%s'"), |
f2075480 | 1424 | keep_name); |
576162a4 | 1425 | report = "keep"; |
b8077709 | 1426 | } |
b8077709 SP |
1427 | } |
1428 | ||
e42797f5 | 1429 | if (final_pack_name != curr_pack_name) { |
ba47a308 | 1430 | if (!final_pack_name) |
f2075480 | 1431 | final_pack_name = odb_pack_name(&pack_name, sha1, "pack"); |
cb5add58 | 1432 | if (finalize_object_file(curr_pack_name, final_pack_name)) |
c2b97ecf | 1433 | die(_("cannot store pack file")); |
fb8b1936 | 1434 | } else if (from_stdin) |
33b65030 | 1435 | chmod(final_pack_name, 0444); |
e42797f5 | 1436 | |
e42797f5 | 1437 | if (final_index_name != curr_index_name) { |
ba47a308 | 1438 | if (!final_index_name) |
f2075480 | 1439 | final_index_name = odb_pack_name(&index_name, sha1, "idx"); |
cb5add58 | 1440 | if (finalize_object_file(curr_index_name, final_index_name)) |
c2b97ecf | 1441 | die(_("cannot store index file")); |
fb8b1936 JH |
1442 | } else |
1443 | chmod(final_index_name, 0444); | |
576162a4 NP |
1444 | |
1445 | if (!from_stdin) { | |
1446 | printf("%s\n", sha1_to_hex(sha1)); | |
1447 | } else { | |
5b1ef2ce JK |
1448 | struct strbuf buf = STRBUF_INIT; |
1449 | ||
1450 | strbuf_addf(&buf, "%s\t%s\n", report, sha1_to_hex(sha1)); | |
1451 | write_or_die(1, buf.buf, buf.len); | |
1452 | strbuf_release(&buf); | |
576162a4 NP |
1453 | |
1454 | /* | |
1455 | * Let's just mimic git-unpack-objects here and write | |
1456 | * the last part of the input buffer to stdout. | |
1457 | */ | |
1458 | while (input_len) { | |
1459 | err = xwrite(1, input_buffer + input_offset, input_len); | |
1460 | if (err <= 0) | |
1461 | break; | |
1462 | input_len -= err; | |
1463 | input_offset += err; | |
1464 | } | |
1465 | } | |
ba47a308 | 1466 | |
f2075480 JK |
1467 | strbuf_release(&index_name); |
1468 | strbuf_release(&pack_name); | |
1469 | strbuf_release(&keep_name_buf); | |
9cf6d335 SV |
1470 | } |
1471 | ||
ef90d6d4 | 1472 | static int git_index_pack_config(const char *k, const char *v, void *cb) |
4d00bda2 | 1473 | { |
ebcfb379 JH |
1474 | struct pack_idx_option *opts = cb; |
1475 | ||
4d00bda2 | 1476 | if (!strcmp(k, "pack.indexversion")) { |
ebcfb379 JH |
1477 | opts->version = git_config_int(k, v); |
1478 | if (opts->version > 2) | |
f350df42 | 1479 | die(_("bad pack.indexversion=%"PRIu32), opts->version); |
4d00bda2 NP |
1480 | return 0; |
1481 | } | |
b8a2486f NTND |
1482 | if (!strcmp(k, "pack.threads")) { |
1483 | nr_threads = git_config_int(k, v); | |
1484 | if (nr_threads < 0) | |
f350df42 | 1485 | die(_("invalid number of threads specified (%d)"), |
b8a2486f NTND |
1486 | nr_threads); |
1487 | #ifdef NO_PTHREADS | |
1488 | if (nr_threads != 1) | |
f350df42 | 1489 | warning(_("no threads support, ignoring %s"), k); |
b8a2486f NTND |
1490 | nr_threads = 1; |
1491 | #endif | |
1492 | return 0; | |
1493 | } | |
ef90d6d4 | 1494 | return git_default_config(k, v, cb); |
4d00bda2 NP |
1495 | } |
1496 | ||
3c9fc074 JH |
1497 | static int cmp_uint32(const void *a_, const void *b_) |
1498 | { | |
1499 | uint32_t a = *((uint32_t *)a_); | |
1500 | uint32_t b = *((uint32_t *)b_); | |
1501 | ||
1502 | return (a < b) ? -1 : (a != b); | |
1503 | } | |
1504 | ||
1505 | static void read_v2_anomalous_offsets(struct packed_git *p, | |
1506 | struct pack_idx_option *opts) | |
1507 | { | |
1508 | const uint32_t *idx1, *idx2; | |
1509 | uint32_t i; | |
1510 | ||
1511 | /* The address of the 4-byte offset table */ | |
1512 | idx1 = (((const uint32_t *)p->index_data) | |
1513 | + 2 /* 8-byte header */ | |
1514 | + 256 /* fan out */ | |
1515 | + 5 * p->num_objects /* 20-byte SHA-1 table */ | |
1516 | + p->num_objects /* CRC32 table */ | |
1517 | ); | |
1518 | ||
1519 | /* The address of the 8-byte offset table */ | |
1520 | idx2 = idx1 + p->num_objects; | |
1521 | ||
1522 | for (i = 0; i < p->num_objects; i++) { | |
1523 | uint32_t off = ntohl(idx1[i]); | |
1524 | if (!(off & 0x80000000)) | |
1525 | continue; | |
1526 | off = off & 0x7fffffff; | |
47fe3f6e | 1527 | check_pack_index_ptr(p, &idx2[off * 2]); |
3c9fc074 JH |
1528 | if (idx2[off * 2]) |
1529 | continue; | |
1530 | /* | |
1531 | * The real offset is ntohl(idx2[off * 2]) in high 4 | |
1532 | * octets, and ntohl(idx2[off * 2 + 1]) in low 4 | |
1533 | * octets. But idx2[off * 2] is Zero!!! | |
1534 | */ | |
1535 | ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc); | |
1536 | opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]); | |
1537 | } | |
1538 | ||
1b5294de | 1539 | QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32); |
3c9fc074 JH |
1540 | } |
1541 | ||
e337a04d JH |
1542 | static void read_idx_option(struct pack_idx_option *opts, const char *pack_name) |
1543 | { | |
1544 | struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1); | |
1545 | ||
1546 | if (!p) | |
c2b97ecf | 1547 | die(_("Cannot open existing pack file '%s'"), pack_name); |
e337a04d | 1548 | if (open_pack_index(p)) |
c2b97ecf | 1549 | die(_("Cannot open existing pack idx file for '%s'"), pack_name); |
e337a04d JH |
1550 | |
1551 | /* Read the attributes from the existing idx file */ | |
1552 | opts->version = p->index_version; | |
1553 | ||
3c9fc074 JH |
1554 | if (opts->version == 2) |
1555 | read_v2_anomalous_offsets(p, opts); | |
1556 | ||
e337a04d JH |
1557 | /* |
1558 | * Get rid of the idx file as we do not need it anymore. | |
1559 | * NEEDSWORK: extract this bit from free_pack_by_name() in | |
1560 | * sha1_file.c, perhaps? It shouldn't matter very much as we | |
1561 | * know we haven't installed this pack (hence we never have | |
1562 | * read anything from it). | |
1563 | */ | |
1564 | close_pack_index(p); | |
1565 | free(p); | |
1566 | } | |
1567 | ||
38a45561 JH |
1568 | static void show_pack_info(int stat_only) |
1569 | { | |
c6458e60 | 1570 | int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas; |
d1a0ed18 JH |
1571 | unsigned long *chain_histogram = NULL; |
1572 | ||
1573 | if (deepest_delta) | |
1574 | chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long)); | |
1575 | ||
38a45561 JH |
1576 | for (i = 0; i < nr_objects; i++) { |
1577 | struct object_entry *obj = &objects[i]; | |
1578 | ||
d1a0ed18 | 1579 | if (is_delta_type(obj->type)) |
41730576 | 1580 | chain_histogram[obj_stat[i].delta_depth - 1]++; |
38a45561 JH |
1581 | if (stat_only) |
1582 | continue; | |
1583 | printf("%s %-6s %lu %lu %"PRIuMAX, | |
1584 | sha1_to_hex(obj->idx.sha1), | |
1585 | typename(obj->real_type), obj->size, | |
1586 | (unsigned long)(obj[1].idx.offset - obj->idx.offset), | |
1587 | (uintmax_t)obj->idx.offset); | |
1588 | if (is_delta_type(obj->type)) { | |
41730576 NTND |
1589 | struct object_entry *bobj = &objects[obj_stat[i].base_object_no]; |
1590 | printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1)); | |
38a45561 JH |
1591 | } |
1592 | putchar('\n'); | |
1593 | } | |
d1a0ed18 JH |
1594 | |
1595 | if (baseobjects) | |
c2b97ecf NTND |
1596 | printf_ln(Q_("non delta: %d object", |
1597 | "non delta: %d objects", | |
1598 | baseobjects), | |
1599 | baseobjects); | |
d1a0ed18 JH |
1600 | for (i = 0; i < deepest_delta; i++) { |
1601 | if (!chain_histogram[i]) | |
1602 | continue; | |
c2b97ecf NTND |
1603 | printf_ln(Q_("chain length = %d: %lu object", |
1604 | "chain length = %d: %lu objects", | |
1605 | chain_histogram[i]), | |
1606 | i + 1, | |
1607 | chain_histogram[i]); | |
d1a0ed18 | 1608 | } |
38a45561 JH |
1609 | } |
1610 | ||
bfee614a JH |
1611 | static const char *derive_filename(const char *pack_name, const char *suffix, |
1612 | struct strbuf *buf) | |
1613 | { | |
1614 | size_t len; | |
1615 | if (!strip_suffix(pack_name, ".pack", &len)) | |
1616 | die(_("packfile name '%s' does not end with '.pack'"), | |
1617 | pack_name); | |
1618 | strbuf_add(buf, pack_name, len); | |
1619 | strbuf_addstr(buf, suffix); | |
1620 | return buf->buf; | |
1621 | } | |
1622 | ||
3bb72562 | 1623 | int cmd_index_pack(int argc, const char **argv, const char *prefix) |
9cf6d335 | 1624 | { |
3aba2fdd | 1625 | int i, fix_thin_pack = 0, verify = 0, stat_only = 0; |
39539495 | 1626 | const char *curr_index; |
3bb72562 | 1627 | const char *index_name = NULL, *pack_name = NULL; |
b8077709 | 1628 | const char *keep_name = NULL, *keep_msg = NULL; |
592ce208 JK |
1629 | struct strbuf index_name_buf = STRBUF_INIT, |
1630 | keep_name_buf = STRBUF_INIT; | |
aa7e44bf | 1631 | struct pack_idx_entry **idx_objects; |
ebcfb379 | 1632 | struct pack_idx_option opts; |
8522148f | 1633 | unsigned char pack_sha1[20]; |
c6807a40 | 1634 | unsigned foreign_nr = 1; /* zero is a "good" value, assume bad */ |
83558686 | 1635 | int report_end_of_input = 0; |
9cf6d335 | 1636 | |
99caeed0 JN |
1637 | if (argc == 2 && !strcmp(argv[1], "-h")) |
1638 | usage(index_pack_usage); | |
1639 | ||
afc711b8 | 1640 | check_replace_refs = 0; |
22410549 | 1641 | fsck_options.walk = mark_link; |
6e2a09d2 | 1642 | |
ebcfb379 JH |
1643 | reset_pack_idx_option(&opts); |
1644 | git_config(git_index_pack_config, &opts); | |
3f8099fc | 1645 | if (prefix && chdir(prefix)) |
c2b97ecf | 1646 | die(_("Cannot come back to cwd")); |
4d00bda2 | 1647 | |
9cf6d335 | 1648 | for (i = 1; i < argc; i++) { |
3bb72562 | 1649 | const char *arg = argv[i]; |
9cf6d335 SV |
1650 | |
1651 | if (*arg == '-') { | |
e42797f5 NP |
1652 | if (!strcmp(arg, "--stdin")) { |
1653 | from_stdin = 1; | |
636171cb NP |
1654 | } else if (!strcmp(arg, "--fix-thin")) { |
1655 | fix_thin_pack = 1; | |
0153be05 MK |
1656 | } else if (!strcmp(arg, "--strict")) { |
1657 | strict = 1; | |
c6807a40 | 1658 | do_fsck_object = 1; |
5d477a33 JS |
1659 | } else if (skip_prefix(arg, "--strict=", &arg)) { |
1660 | strict = 1; | |
1661 | do_fsck_object = 1; | |
1662 | fsck_set_msg_types(&fsck_options, arg); | |
c6807a40 NTND |
1663 | } else if (!strcmp(arg, "--check-self-contained-and-connected")) { |
1664 | strict = 1; | |
1665 | check_self_contained_and_connected = 1; | |
e337a04d JH |
1666 | } else if (!strcmp(arg, "--verify")) { |
1667 | verify = 1; | |
38a45561 JH |
1668 | } else if (!strcmp(arg, "--verify-stat")) { |
1669 | verify = 1; | |
3aba2fdd | 1670 | show_stat = 1; |
38a45561 JH |
1671 | } else if (!strcmp(arg, "--verify-stat-only")) { |
1672 | verify = 1; | |
3aba2fdd | 1673 | show_stat = 1; |
38a45561 | 1674 | stat_only = 1; |
b8077709 SP |
1675 | } else if (!strcmp(arg, "--keep")) { |
1676 | keep_msg = ""; | |
59556548 | 1677 | } else if (starts_with(arg, "--keep=")) { |
b8077709 | 1678 | keep_msg = arg + 7; |
59556548 | 1679 | } else if (starts_with(arg, "--threads=")) { |
b8a2486f NTND |
1680 | char *end; |
1681 | nr_threads = strtoul(arg+10, &end, 0); | |
1682 | if (!arg[10] || *end || nr_threads < 0) | |
1683 | usage(index_pack_usage); | |
1684 | #ifdef NO_PTHREADS | |
1685 | if (nr_threads != 1) | |
f350df42 NTND |
1686 | warning(_("no threads support, " |
1687 | "ignoring %s"), arg); | |
b8a2486f NTND |
1688 | nr_threads = 1; |
1689 | #endif | |
59556548 | 1690 | } else if (starts_with(arg, "--pack_header=")) { |
bed006fb NP |
1691 | struct pack_header *hdr; |
1692 | char *c; | |
1693 | ||
1694 | hdr = (struct pack_header *)input_buffer; | |
1695 | hdr->hdr_signature = htonl(PACK_SIGNATURE); | |
1696 | hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10)); | |
1697 | if (*c != ',') | |
c2b97ecf | 1698 | die(_("bad %s"), arg); |
bed006fb NP |
1699 | hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10)); |
1700 | if (*c) | |
c2b97ecf | 1701 | die(_("bad %s"), arg); |
bed006fb | 1702 | input_len = sizeof(*hdr); |
3c9af366 NP |
1703 | } else if (!strcmp(arg, "-v")) { |
1704 | verbose = 1; | |
e376f17f JK |
1705 | } else if (!strcmp(arg, "--show-resolving-progress")) { |
1706 | show_resolving_progress = 1; | |
83558686 JK |
1707 | } else if (!strcmp(arg, "--report-end-of-input")) { |
1708 | report_end_of_input = 1; | |
e42797f5 | 1709 | } else if (!strcmp(arg, "-o")) { |
9cf6d335 SV |
1710 | if (index_name || (i+1) >= argc) |
1711 | usage(index_pack_usage); | |
1712 | index_name = argv[++i]; | |
59556548 | 1713 | } else if (starts_with(arg, "--index-version=")) { |
4ba7d711 | 1714 | char *c; |
ebcfb379 JH |
1715 | opts.version = strtoul(arg + 16, &c, 10); |
1716 | if (opts.version > 2) | |
c2b97ecf | 1717 | die(_("bad %s"), arg); |
4ba7d711 | 1718 | if (*c == ',') |
ebcfb379 JH |
1719 | opts.off32_limit = strtoul(c+1, &c, 0); |
1720 | if (*c || opts.off32_limit & 0x80000000) | |
c2b97ecf | 1721 | die(_("bad %s"), arg); |
411481be JK |
1722 | } else if (skip_prefix(arg, "--max-input-size=", &arg)) { |
1723 | max_input_size = strtoumax(arg, NULL, 10); | |
9cf6d335 SV |
1724 | } else |
1725 | usage(index_pack_usage); | |
1726 | continue; | |
1727 | } | |
1728 | ||
1729 | if (pack_name) | |
1730 | usage(index_pack_usage); | |
1731 | pack_name = arg; | |
1732 | } | |
1733 | ||
e42797f5 | 1734 | if (!pack_name && !from_stdin) |
9cf6d335 | 1735 | usage(index_pack_usage); |
636171cb | 1736 | if (fix_thin_pack && !from_stdin) |
c2b97ecf | 1737 | die(_("--fix-thin cannot be used without --stdin")); |
7176a314 JK |
1738 | if (from_stdin && !startup_info->have_repository) |
1739 | die(_("--stdin requires a git repository")); | |
bfee614a JH |
1740 | if (!index_name && pack_name) |
1741 | index_name = derive_filename(pack_name, ".idx", &index_name_buf); | |
1742 | if (keep_msg && !keep_name && pack_name) | |
1743 | keep_name = derive_filename(pack_name, ".keep", &keep_name_buf); | |
1744 | ||
e337a04d JH |
1745 | if (verify) { |
1746 | if (!index_name) | |
c2b97ecf | 1747 | die(_("--verify with no packfile name given")); |
e337a04d | 1748 | read_idx_option(&opts, index_name); |
68be2fea | 1749 | opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT; |
e337a04d | 1750 | } |
68be2fea JH |
1751 | if (strict) |
1752 | opts.flags |= WRITE_IDX_STRICT; | |
9cf6d335 | 1753 | |
b8a2486f NTND |
1754 | #ifndef NO_PTHREADS |
1755 | if (!nr_threads) { | |
1756 | nr_threads = online_cpus(); | |
1757 | /* An experiment showed that more threads does not mean faster */ | |
1758 | if (nr_threads > 3) | |
1759 | nr_threads = 3; | |
1760 | } | |
1761 | #endif | |
1762 | ||
e42797f5 | 1763 | curr_pack = open_pack_file(pack_name); |
9cf6d335 | 1764 | parse_pack_header(); |
50a6c8ef | 1765 | objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry)); |
41730576 | 1766 | if (show_stat) |
50a6c8ef | 1767 | obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat)); |
c6458e60 | 1768 | ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry)); |
8522148f | 1769 | parse_pack_objects(pack_sha1); |
83558686 JK |
1770 | if (report_end_of_input) |
1771 | write_in_full(2, "\0", 1); | |
5272f755 NTND |
1772 | resolve_deltas(); |
1773 | conclude_pack(fix_thin_pack, curr_pack, pack_sha1); | |
c6458e60 NTND |
1774 | free(ofs_deltas); |
1775 | free(ref_deltas); | |
0153be05 | 1776 | if (strict) |
c6807a40 | 1777 | foreign_nr = check_objects(); |
aa7e44bf | 1778 | |
3aba2fdd | 1779 | if (show_stat) |
38a45561 JH |
1780 | show_pack_info(stat_only); |
1781 | ||
b32fa95f | 1782 | ALLOC_ARRAY(idx_objects, nr_objects); |
aa7e44bf GB |
1783 | for (i = 0; i < nr_objects; i++) |
1784 | idx_objects[i] = &objects[i].idx; | |
ebcfb379 | 1785 | curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1); |
aa7e44bf GB |
1786 | free(idx_objects); |
1787 | ||
e337a04d JH |
1788 | if (!verify) |
1789 | final(pack_name, curr_pack, | |
1790 | index_name, curr_index, | |
1791 | keep_name, keep_msg, | |
1792 | pack_sha1); | |
1793 | else | |
1794 | close(input_fd); | |
9cf6d335 | 1795 | free(objects); |
592ce208 JK |
1796 | strbuf_release(&index_name_buf); |
1797 | strbuf_release(&keep_name_buf); | |
c85228ed | 1798 | if (pack_name == NULL) |
3bb72562 | 1799 | free((void *) curr_pack); |
c85228ed | 1800 | if (index_name == NULL) |
3bb72562 | 1801 | free((void *) curr_index); |
9cf6d335 | 1802 | |
c6807a40 NTND |
1803 | /* |
1804 | * Let the caller know this pack is not self contained | |
1805 | */ | |
1806 | if (check_self_contained_and_connected && foreign_nr) | |
1807 | return 1; | |
1808 | ||
9cf6d335 SV |
1809 | return 0; |
1810 | } |