?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'jk/rebase-i-squash-count-fix' into maint
[git/raj.git] / builtin / index-pack.c
blobf4b87c6c9f901e5834ce9e50db089fb487c6eafc
1 #include "builtin.h"
2 #include "delta.h"
3 #include "pack.h"
4 #include "csum-file.h"
5 #include "blob.h"
6 #include "commit.h"
7 #include "tag.h"
8 #include "tree.h"
9 #include "progress.h"
10 #include "fsck.h"
11 #include "exec_cmd.h"
12 #include "streaming.h"
13 #include "thread-utils.h"
14
15 static const char index_pack_usage[] =
16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
17
18 struct object_entry {
19 struct pack_idx_entry idx;
20 unsigned long size;
21 unsigned char hdr_size;
22 signed char type;
23 signed char real_type;
24 };
25
26 struct object_stat {
27 unsigned delta_depth;
28 int base_object_no;
29 };
30
31 struct base_data {
32 struct base_data *base;
33 struct base_data *child;
34 struct object_entry *obj;
35 void *data;
36 unsigned long size;
37 int ref_first, ref_last;
38 int ofs_first, ofs_last;
39 };
40
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;
47 int pack_fd;
48 };
49
50 #define FLAG_LINK (1u<<20)
51 #define FLAG_CHECKED (1u<<21)
52
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];
60 int obj_no;
61 };
62
63 static struct object_entry *objects;
64 static struct object_stat *obj_stat;
65 static struct ofs_delta_entry *ofs_deltas;
66 static struct ref_delta_entry *ref_deltas;
67 static struct thread_local nothread_data;
68 static int nr_objects;
69 static int nr_ofs_deltas;
70 static int nr_ref_deltas;
71 static int ref_deltas_alloc;
72 static int nr_resolved_deltas;
73 static int nr_threads;
74
75 static int from_stdin;
76 static int strict;
77 static int do_fsck_object;
78 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
79 static int verbose;
80 static int show_resolving_progress;
81 static int show_stat;
82 static int check_self_contained_and_connected;
83
84 static struct progress *progress;
85
86 /* We always read in 4kB chunks. */
87 static unsigned char input_buffer[4096];
88 static unsigned int input_offset, input_len;
89 static off_t consumed_bytes;
90 static off_t max_input_size;
91 static unsigned deepest_delta;
92 static git_SHA_CTX input_ctx;
93 static uint32_t input_crc32;
94 static int input_fd, output_fd;
95 static const char *curr_pack;
96
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
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
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
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 {
142 int i;
143 init_recursive_mutex(&read_mutex);
144 pthread_mutex_init(&counter_mutex, NULL);
145 pthread_mutex_init(&work_mutex, NULL);
146 pthread_mutex_init(&type_cas_mutex, NULL);
147 if (show_stat)
148 pthread_mutex_init(&deepest_delta_mutex, NULL);
149 pthread_key_create(&key, NULL);
150 thread_data = xcalloc(nr_threads, sizeof(*thread_data));
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
157 threads_active = 1;
158 }
159
160 static void cleanup_thread(void)
161 {
162 int i;
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);
169 pthread_mutex_destroy(&type_cas_mutex);
170 if (show_stat)
171 pthread_mutex_destroy(&deepest_delta_mutex);
172 for (i = 0; i < nr_threads; i++)
173 close(thread_data[i].pack_fd);
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
189 #define deepest_delta_lock()
190 #define deepest_delta_unlock()
191
192 #define type_cas_lock()
193 #define type_cas_unlock()
194
195 #endif
196
197
198 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
199 {
200 if (!obj)
201 return -1;
202
203 if (type != OBJ_ANY && obj->type != type)
204 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
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 */
212 static unsigned check_object(struct object *obj)
213 {
214 if (!obj)
215 return 0;
216
217 if (!(obj->flags & FLAG_LINK))
218 return 0;
219
220 if (!(obj->flags & FLAG_CHECKED)) {
221 unsigned long size;
222 int type = sha1_object_info(obj->oid.hash, &size);
223 if (type <= 0)
224 die(_("did not receive expected object %s"),
225 oid_to_hex(&obj->oid));
226 if (type != obj->type)
227 die(_("object %s: expected type %s, found %s"),
228 oid_to_hex(&obj->oid),
229 typename(obj->type), typename(type));
230 obj->flags |= FLAG_CHECKED;
231 return 1;
232 }
233
234 return 0;
235 }
236
237 static unsigned check_objects(void)
238 {
239 unsigned i, max, foreign_nr = 0;
240
241 max = get_max_object_index();
242 for (i = 0; i < max; i++)
243 foreign_nr += check_object(get_indexed_object(i));
244 return foreign_nr;
245 }
246
247
248 /* Discard current buffer used content. */
249 static void flush(void)
250 {
251 if (input_offset) {
252 if (output_fd >= 0)
253 write_or_die(output_fd, input_buffer, input_offset);
254 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
255 memmove(input_buffer, input_buffer + input_offset, input_len);
256 input_offset = 0;
257 }
258 }
259
260 /*
261 * Make sure at least "min" bytes are available in the buffer, and
262 * return the pointer to the buffer.
263 */
264 static void *fill(int min)
265 {
266 if (min <= input_len)
267 return input_buffer + input_offset;
268 if (min > sizeof(input_buffer))
269 die(Q_("cannot fill %d byte",
270 "cannot fill %d bytes",
271 min),
272 min);
273 flush();
274 do {
275 ssize_t ret = xread(input_fd, input_buffer + input_len,
276 sizeof(input_buffer) - input_len);
277 if (ret <= 0) {
278 if (!ret)
279 die(_("early EOF"));
280 die_errno(_("read error on input"));
281 }
282 input_len += ret;
283 if (from_stdin)
284 display_throughput(progress, consumed_bytes + input_len);
285 } while (input_len < min);
286 return input_buffer;
287 }
288
289 static void use(int bytes)
290 {
291 if (bytes > input_len)
292 die(_("used more bytes than were available"));
293 input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
294 input_len -= bytes;
295 input_offset += bytes;
296
297 /* make sure off_t is sufficiently large not to wrap */
298 if (signed_add_overflows(consumed_bytes, bytes))
299 die(_("pack too large for current definition of off_t"));
300 consumed_bytes += bytes;
301 if (max_input_size && consumed_bytes > max_input_size)
302 die(_("pack exceeds maximum allowed size"));
303 }
304
305 static const char *open_pack_file(const char *pack_name)
306 {
307 if (from_stdin) {
308 input_fd = 0;
309 if (!pack_name) {
310 static char tmp_file[PATH_MAX];
311 output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
312 "pack/tmp_pack_XXXXXX");
313 pack_name = xstrdup(tmp_file);
314 } else
315 output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
316 if (output_fd < 0)
317 die_errno(_("unable to create '%s'"), pack_name);
318 nothread_data.pack_fd = output_fd;
319 } else {
320 input_fd = open(pack_name, O_RDONLY);
321 if (input_fd < 0)
322 die_errno(_("cannot open packfile '%s'"), pack_name);
323 output_fd = -1;
324 nothread_data.pack_fd = input_fd;
325 }
326 git_SHA1_Init(&input_ctx);
327 return pack_name;
328 }
329
330 static void parse_pack_header(void)
331 {
332 struct pack_header *hdr = fill(sizeof(struct pack_header));
333
334 /* Header consistency check */
335 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
336 die(_("pack signature mismatch"));
337 if (!pack_version_ok(hdr->hdr_version))
338 die(_("pack version %"PRIu32" unsupported"),
339 ntohl(hdr->hdr_version));
340
341 nr_objects = ntohl(hdr->hdr_entries);
342 use(sizeof(struct pack_header));
343 }
344
345 static NORETURN void bad_object(off_t offset, const char *format,
346 ...) __attribute__((format (printf, 2, 3)));
347
348 static NORETURN void bad_object(off_t offset, const char *format, ...)
349 {
350 va_list params;
351 char buf[1024];
352
353 va_start(params, format);
354 vsnprintf(buf, sizeof(buf), format, params);
355 va_end(params);
356 die(_("pack has bad object at offset %"PRIuMAX": %s"),
357 (uintmax_t)offset, buf);
358 }
359
360 static inline struct thread_local *get_thread_data(void)
361 {
362 #ifndef NO_PTHREADS
363 if (threads_active)
364 return pthread_getspecific(key);
365 assert(!threads_active &&
366 "This should only be reached when all threads are gone");
367 #endif
368 return &nothread_data;
369 }
370
371 #ifndef NO_PTHREADS
372 static void set_thread_data(struct thread_local *data)
373 {
374 if (threads_active)
375 pthread_setspecific(key, data);
376 }
377 #endif
378
379 static struct base_data *alloc_base_data(void)
380 {
381 struct base_data *base = xcalloc(1, sizeof(struct base_data));
382 base->ref_last = -1;
383 base->ofs_last = -1;
384 return base;
385 }
386
387 static void free_base_data(struct base_data *c)
388 {
389 if (c->data) {
390 free(c->data);
391 c->data = NULL;
392 get_thread_data()->base_cache_used -= c->size;
393 }
394 }
395
396 static void prune_base_data(struct base_data *retain)
397 {
398 struct base_data *b;
399 struct thread_local *data = get_thread_data();
400 for (b = data->base_cache;
401 data->base_cache_used > delta_base_cache_limit && b;
402 b = b->child) {
403 if (b->data && b != retain)
404 free_base_data(b);
405 }
406 }
407
408 static void link_base_data(struct base_data *base, struct base_data *c)
409 {
410 if (base)
411 base->child = c;
412 else
413 get_thread_data()->base_cache = c;
414
415 c->base = base;
416 c->child = NULL;
417 if (c->data)
418 get_thread_data()->base_cache_used += c->size;
419 prune_base_data(c);
420 }
421
422 static void unlink_base_data(struct base_data *c)
423 {
424 struct base_data *base = c->base;
425 if (base)
426 base->child = NULL;
427 else
428 get_thread_data()->base_cache = NULL;
429 free_base_data(c);
430 }
431
432 static int is_delta_type(enum object_type type)
433 {
434 return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
435 }
436
437 static void *unpack_entry_data(off_t offset, unsigned long size,
438 enum object_type type, unsigned char *sha1)
439 {
440 static char fixed_buf[8192];
441 int status;
442 git_zstream stream;
443 void *buf;
444 git_SHA_CTX c;
445 char hdr[32];
446 int hdrlen;
447
448 if (!is_delta_type(type)) {
449 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), size) + 1;
450 git_SHA1_Init(&c);
451 git_SHA1_Update(&c, hdr, hdrlen);
452 } else
453 sha1 = NULL;
454 if (type == OBJ_BLOB && size > big_file_threshold)
455 buf = fixed_buf;
456 else
457 buf = xmallocz(size);
458
459 memset(&stream, 0, sizeof(stream));
460 git_inflate_init(&stream);
461 stream.next_out = buf;
462 stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
463
464 do {
465 unsigned char *last_out = stream.next_out;
466 stream.next_in = fill(1);
467 stream.avail_in = input_len;
468 status = git_inflate(&stream, 0);
469 use(input_len - stream.avail_in);
470 if (sha1)
471 git_SHA1_Update(&c, last_out, stream.next_out - last_out);
472 if (buf == fixed_buf) {
473 stream.next_out = buf;
474 stream.avail_out = sizeof(fixed_buf);
475 }
476 } while (status == Z_OK);
477 if (stream.total_out != size || status != Z_STREAM_END)
478 bad_object(offset, _("inflate returned %d"), status);
479 git_inflate_end(&stream);
480 if (sha1)
481 git_SHA1_Final(sha1, &c);
482 return buf == fixed_buf ? NULL : buf;
483 }
484
485 static void *unpack_raw_entry(struct object_entry *obj,
486 off_t *ofs_offset,
487 unsigned char *ref_sha1,
488 unsigned char *sha1)
489 {
490 unsigned char *p;
491 unsigned long size, c;
492 off_t base_offset;
493 unsigned shift;
494 void *data;
495
496 obj->idx.offset = consumed_bytes;
497 input_crc32 = crc32(0, NULL, 0);
498
499 p = fill(1);
500 c = *p;
501 use(1);
502 obj->type = (c >> 4) & 7;
503 size = (c & 15);
504 shift = 4;
505 while (c & 0x80) {
506 p = fill(1);
507 c = *p;
508 use(1);
509 size += (c & 0x7f) << shift;
510 shift += 7;
511 }
512 obj->size = size;
513
514 switch (obj->type) {
515 case OBJ_REF_DELTA:
516 hashcpy(ref_sha1, fill(20));
517 use(20);
518 break;
519 case OBJ_OFS_DELTA:
520 p = fill(1);
521 c = *p;
522 use(1);
523 base_offset = c & 127;
524 while (c & 128) {
525 base_offset += 1;
526 if (!base_offset || MSB(base_offset, 7))
527 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
528 p = fill(1);
529 c = *p;
530 use(1);
531 base_offset = (base_offset << 7) + (c & 127);
532 }
533 *ofs_offset = obj->idx.offset - base_offset;
534 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
535 bad_object(obj->idx.offset, _("delta base offset is out of bound"));
536 break;
537 case OBJ_COMMIT:
538 case OBJ_TREE:
539 case OBJ_BLOB:
540 case OBJ_TAG:
541 break;
542 default:
543 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
544 }
545 obj->hdr_size = consumed_bytes - obj->idx.offset;
546
547 data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
548 obj->idx.crc32 = input_crc32;
549 return data;
550 }
551
552 static void *unpack_data(struct object_entry *obj,
553 int (*consume)(const unsigned char *, unsigned long, void *),
554 void *cb_data)
555 {
556 off_t from = obj[0].idx.offset + obj[0].hdr_size;
557 off_t len = obj[1].idx.offset - from;
558 unsigned char *data, *inbuf;
559 git_zstream stream;
560 int status;
561
562 data = xmallocz(consume ? 64*1024 : obj->size);
563 inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
564
565 memset(&stream, 0, sizeof(stream));
566 git_inflate_init(&stream);
567 stream.next_out = data;
568 stream.avail_out = consume ? 64*1024 : obj->size;
569
570 do {
571 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
572 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
573 if (n < 0)
574 die_errno(_("cannot pread pack file"));
575 if (!n)
576 die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
577 "premature end of pack file, %"PRIuMAX" bytes missing",
578 (unsigned int)len),
579 (uintmax_t)len);
580 from += n;
581 len -= n;
582 stream.next_in = inbuf;
583 stream.avail_in = n;
584 if (!consume)
585 status = git_inflate(&stream, 0);
586 else {
587 do {
588 status = git_inflate(&stream, 0);
589 if (consume(data, stream.next_out - data, cb_data)) {
590 free(inbuf);
591 free(data);
592 return NULL;
593 }
594 stream.next_out = data;
595 stream.avail_out = 64*1024;
596 } while (status == Z_OK && stream.avail_in);
597 }
598 } while (len && status == Z_OK && !stream.avail_in);
599
600 /* This has been inflated OK when first encountered, so... */
601 if (status != Z_STREAM_END || stream.total_out != obj->size)
602 die(_("serious inflate inconsistency"));
603
604 git_inflate_end(&stream);
605 free(inbuf);
606 if (consume) {
607 free(data);
608 data = NULL;
609 }
610 return data;
611 }
612
613 static void *get_data_from_pack(struct object_entry *obj)
614 {
615 return unpack_data(obj, NULL, NULL);
616 }
617
618 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
619 enum object_type type1,
620 enum object_type type2)
621 {
622 int cmp = type1 - type2;
623 if (cmp)
624 return cmp;
625 return offset1 < offset2 ? -1 :
626 offset1 > offset2 ? 1 :
627 0;
628 }
629
630 static int find_ofs_delta(const off_t offset, enum object_type type)
631 {
632 int first = 0, last = nr_ofs_deltas;
633
634 while (first < last) {
635 int next = (first + last) / 2;
636 struct ofs_delta_entry *delta = &ofs_deltas[next];
637 int cmp;
638
639 cmp = compare_ofs_delta_bases(offset, delta->offset,
640 type, objects[delta->obj_no].type);
641 if (!cmp)
642 return next;
643 if (cmp < 0) {
644 last = next;
645 continue;
646 }
647 first = next+1;
648 }
649 return -first-1;
650 }
651
652 static void find_ofs_delta_children(off_t offset,
653 int *first_index, int *last_index,
654 enum object_type type)
655 {
656 int first = find_ofs_delta(offset, type);
657 int last = first;
658 int end = nr_ofs_deltas - 1;
659
660 if (first < 0) {
661 *first_index = 0;
662 *last_index = -1;
663 return;
664 }
665 while (first > 0 && ofs_deltas[first - 1].offset == offset)
666 --first;
667 while (last < end && ofs_deltas[last + 1].offset == offset)
668 ++last;
669 *first_index = first;
670 *last_index = last;
671 }
672
673 static int compare_ref_delta_bases(const unsigned char *sha1,
674 const unsigned char *sha2,
675 enum object_type type1,
676 enum object_type type2)
677 {
678 int cmp = type1 - type2;
679 if (cmp)
680 return cmp;
681 return hashcmp(sha1, sha2);
682 }
683
684 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
685 {
686 int first = 0, last = nr_ref_deltas;
687
688 while (first < last) {
689 int next = (first + last) / 2;
690 struct ref_delta_entry *delta = &ref_deltas[next];
691 int cmp;
692
693 cmp = compare_ref_delta_bases(sha1, delta->sha1,
694 type, objects[delta->obj_no].type);
695 if (!cmp)
696 return next;
697 if (cmp < 0) {
698 last = next;
699 continue;
700 }
701 first = next+1;
702 }
703 return -first-1;
704 }
705
706 static void find_ref_delta_children(const unsigned char *sha1,
707 int *first_index, int *last_index,
708 enum object_type type)
709 {
710 int first = find_ref_delta(sha1, type);
711 int last = first;
712 int end = nr_ref_deltas - 1;
713
714 if (first < 0) {
715 *first_index = 0;
716 *last_index = -1;
717 return;
718 }
719 while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
720 --first;
721 while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
722 ++last;
723 *first_index = first;
724 *last_index = last;
725 }
726
727 struct compare_data {
728 struct object_entry *entry;
729 struct git_istream *st;
730 unsigned char *buf;
731 unsigned long buf_size;
732 };
733
734 static int compare_objects(const unsigned char *buf, unsigned long size,
735 void *cb_data)
736 {
737 struct compare_data *data = cb_data;
738
739 if (data->buf_size < size) {
740 free(data->buf);
741 data->buf = xmalloc(size);
742 data->buf_size = size;
743 }
744
745 while (size) {
746 ssize_t len = read_istream(data->st, data->buf, size);
747 if (len == 0)
748 die(_("SHA1 COLLISION FOUND WITH %s !"),
749 sha1_to_hex(data->entry->idx.sha1));
750 if (len < 0)
751 die(_("unable to read %s"),
752 sha1_to_hex(data->entry->idx.sha1));
753 if (memcmp(buf, data->buf, len))
754 die(_("SHA1 COLLISION FOUND WITH %s !"),
755 sha1_to_hex(data->entry->idx.sha1));
756 size -= len;
757 buf += len;
758 }
759 return 0;
760 }
761
762 static int check_collison(struct object_entry *entry)
763 {
764 struct compare_data data;
765 enum object_type type;
766 unsigned long size;
767
768 if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
769 return -1;
770
771 memset(&data, 0, sizeof(data));
772 data.entry = entry;
773 data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
774 if (!data.st)
775 return -1;
776 if (size != entry->size || type != entry->type)
777 die(_("SHA1 COLLISION FOUND WITH %s !"),
778 sha1_to_hex(entry->idx.sha1));
779 unpack_data(entry, compare_objects, &data);
780 close_istream(data.st);
781 free(data.buf);
782 return 0;
783 }
784
785 static void sha1_object(const void *data, struct object_entry *obj_entry,
786 unsigned long size, enum object_type type,
787 const unsigned char *sha1)
788 {
789 void *new_data = NULL;
790 int collision_test_needed = 0;
791
792 assert(data || obj_entry);
793
794 if (startup_info->have_repository) {
795 read_lock();
796 collision_test_needed = has_sha1_file_with_flags(sha1, HAS_SHA1_QUICK);
797 read_unlock();
798 }
799
800 if (collision_test_needed && !data) {
801 read_lock();
802 if (!check_collison(obj_entry))
803 collision_test_needed = 0;
804 read_unlock();
805 }
806 if (collision_test_needed) {
807 void *has_data;
808 enum object_type has_type;
809 unsigned long has_size;
810 read_lock();
811 has_type = sha1_object_info(sha1, &has_size);
812 if (has_type != type || has_size != size)
813 die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
814 has_data = read_sha1_file(sha1, &has_type, &has_size);
815 read_unlock();
816 if (!data)
817 data = new_data = get_data_from_pack(obj_entry);
818 if (!has_data)
819 die(_("cannot read existing object %s"), sha1_to_hex(sha1));
820 if (size != has_size || type != has_type ||
821 memcmp(data, has_data, size) != 0)
822 die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
823 free(has_data);
824 }
825
826 if (strict) {
827 read_lock();
828 if (type == OBJ_BLOB) {
829 struct blob *blob = lookup_blob(sha1);
830 if (blob)
831 blob->object.flags |= FLAG_CHECKED;
832 else
833 die(_("invalid blob object %s"), sha1_to_hex(sha1));
834 } else {
835 struct object *obj;
836 int eaten;
837 void *buf = (void *) data;
838
839 assert(data && "data can only be NULL for large _blobs_");
840
841 /*
842 * we do not need to free the memory here, as the
843 * buf is deleted by the caller.
844 */
845 obj = parse_object_buffer(sha1, type, size, buf, &eaten);
846 if (!obj)
847 die(_("invalid %s"), typename(type));
848 if (do_fsck_object &&
849 fsck_object(obj, buf, size, &fsck_options))
850 die(_("Error in object"));
851 if (fsck_walk(obj, NULL, &fsck_options))
852 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
853
854 if (obj->type == OBJ_TREE) {
855 struct tree *item = (struct tree *) obj;
856 item->buffer = NULL;
857 obj->parsed = 0;
858 }
859 if (obj->type == OBJ_COMMIT) {
860 struct commit *commit = (struct commit *) obj;
861 if (detach_commit_buffer(commit, NULL) != data)
862 die("BUG: parse_object_buffer transmogrified our buffer");
863 }
864 obj->flags |= FLAG_CHECKED;
865 }
866 read_unlock();
867 }
868
869 free(new_data);
870 }
871
872 /*
873 * This function is part of find_unresolved_deltas(). There are two
874 * walkers going in the opposite ways.
875 *
876 * The first one in find_unresolved_deltas() traverses down from
877 * parent node to children, deflating nodes along the way. However,
878 * memory for deflated nodes is limited by delta_base_cache_limit, so
879 * at some point parent node's deflated content may be freed.
880 *
881 * The second walker is this function, which goes from current node up
882 * to top parent if necessary to deflate the node. In normal
883 * situation, its parent node would be already deflated, so it just
884 * needs to apply delta.
885 *
886 * In the worst case scenario, parent node is no longer deflated because
887 * we're running out of delta_base_cache_limit; we need to re-deflate
888 * parents, possibly up to the top base.
889 *
890 * All deflated objects here are subject to be freed if we exceed
891 * delta_base_cache_limit, just like in find_unresolved_deltas(), we
892 * just need to make sure the last node is not freed.
893 */
894 static void *get_base_data(struct base_data *c)
895 {
896 if (!c->data) {
897 struct object_entry *obj = c->obj;
898 struct base_data **delta = NULL;
899 int delta_nr = 0, delta_alloc = 0;
900
901 while (is_delta_type(c->obj->type) && !c->data) {
902 ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
903 delta[delta_nr++] = c;
904 c = c->base;
905 }
906 if (!delta_nr) {
907 c->data = get_data_from_pack(obj);
908 c->size = obj->size;
909 get_thread_data()->base_cache_used += c->size;
910 prune_base_data(c);
911 }
912 for (; delta_nr > 0; delta_nr--) {
913 void *base, *raw;
914 c = delta[delta_nr - 1];
915 obj = c->obj;
916 base = get_base_data(c->base);
917 raw = get_data_from_pack(obj);
918 c->data = patch_delta(
919 base, c->base->size,
920 raw, obj->size,
921 &c->size);
922 free(raw);
923 if (!c->data)
924 bad_object(obj->idx.offset, _("failed to apply delta"));
925 get_thread_data()->base_cache_used += c->size;
926 prune_base_data(c);
927 }
928 free(delta);
929 }
930 return c->data;
931 }
932
933 static void resolve_delta(struct object_entry *delta_obj,
934 struct base_data *base, struct base_data *result)
935 {
936 void *base_data, *delta_data;
937
938 if (show_stat) {
939 int i = delta_obj - objects;
940 int j = base->obj - objects;
941 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
942 deepest_delta_lock();
943 if (deepest_delta < obj_stat[i].delta_depth)
944 deepest_delta = obj_stat[i].delta_depth;
945 deepest_delta_unlock();
946 obj_stat[i].base_object_no = j;
947 }
948 delta_data = get_data_from_pack(delta_obj);
949 base_data = get_base_data(base);
950 result->obj = delta_obj;
951 result->data = patch_delta(base_data, base->size,
952 delta_data, delta_obj->size, &result->size);
953 free(delta_data);
954 if (!result->data)
955 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
956 hash_sha1_file(result->data, result->size,
957 typename(delta_obj->real_type), delta_obj->idx.sha1);
958 sha1_object(result->data, NULL, result->size, delta_obj->real_type,
959 delta_obj->idx.sha1);
960 counter_lock();
961 nr_resolved_deltas++;
962 counter_unlock();
963 }
964
965 /*
966 * Standard boolean compare-and-swap: atomically check whether "*type" is
967 * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
968 * and return false.
969 */
970 static int compare_and_swap_type(signed char *type,
971 enum object_type want,
972 enum object_type set)
973 {
974 enum object_type old;
975
976 type_cas_lock();
977 old = *type;
978 if (old == want)
979 *type = set;
980 type_cas_unlock();
981
982 return old == want;
983 }
984
985 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
986 struct base_data *prev_base)
987 {
988 if (base->ref_last == -1 && base->ofs_last == -1) {
989 find_ref_delta_children(base->obj->idx.sha1,
990 &base->ref_first, &base->ref_last,
991 OBJ_REF_DELTA);
992
993 find_ofs_delta_children(base->obj->idx.offset,
994 &base->ofs_first, &base->ofs_last,
995 OBJ_OFS_DELTA);
996
997 if (base->ref_last == -1 && base->ofs_last == -1) {
998 free(base->data);
999 return NULL;
1000 }
1001
1002 link_base_data(prev_base, base);
1003 }
1004
1005 if (base->ref_first <= base->ref_last) {
1006 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1007 struct base_data *result = alloc_base_data();
1008
1009 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1010 base->obj->real_type))
1011 die("BUG: child->real_type != OBJ_REF_DELTA");
1012
1013 resolve_delta(child, base, result);
1014 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1015 free_base_data(base);
1016
1017 base->ref_first++;
1018 return result;
1019 }
1020
1021 if (base->ofs_first <= base->ofs_last) {
1022 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1023 struct base_data *result = alloc_base_data();
1024
1025 assert(child->real_type == OBJ_OFS_DELTA);
1026 child->real_type = base->obj->real_type;
1027 resolve_delta(child, base, result);
1028 if (base->ofs_first == base->ofs_last)
1029 free_base_data(base);
1030
1031 base->ofs_first++;
1032 return result;
1033 }
1034
1035 unlink_base_data(base);
1036 return NULL;
1037 }
1038
1039 static void find_unresolved_deltas(struct base_data *base)
1040 {
1041 struct base_data *new_base, *prev_base = NULL;
1042 for (;;) {
1043 new_base = find_unresolved_deltas_1(base, prev_base);
1044
1045 if (new_base) {
1046 prev_base = base;
1047 base = new_base;
1048 } else {
1049 free(base);
1050 base = prev_base;
1051 if (!base)
1052 return;
1053 prev_base = base->base;
1054 }
1055 }
1056 }
1057
1058 static int compare_ofs_delta_entry(const void *a, const void *b)
1059 {
1060 const struct ofs_delta_entry *delta_a = a;
1061 const struct ofs_delta_entry *delta_b = b;
1062
1063 return delta_a->offset < delta_b->offset ? -1 :
1064 delta_a->offset > delta_b->offset ? 1 :
1065 0;
1066 }
1067
1068 static int compare_ref_delta_entry(const void *a, const void *b)
1069 {
1070 const struct ref_delta_entry *delta_a = a;
1071 const struct ref_delta_entry *delta_b = b;
1072
1073 return hashcmp(delta_a->sha1, delta_b->sha1);
1074 }
1075
1076 static void resolve_base(struct object_entry *obj)
1077 {
1078 struct base_data *base_obj = alloc_base_data();
1079 base_obj->obj = obj;
1080 base_obj->data = NULL;
1081 find_unresolved_deltas(base_obj);
1082 }
1083
1084 #ifndef NO_PTHREADS
1085 static void *threaded_second_pass(void *data)
1086 {
1087 set_thread_data(data);
1088 for (;;) {
1089 int i;
1090 counter_lock();
1091 display_progress(progress, nr_resolved_deltas);
1092 counter_unlock();
1093 work_lock();
1094 while (nr_dispatched < nr_objects &&
1095 is_delta_type(objects[nr_dispatched].type))
1096 nr_dispatched++;
1097 if (nr_dispatched >= nr_objects) {
1098 work_unlock();
1099 break;
1100 }
1101 i = nr_dispatched++;
1102 work_unlock();
1103
1104 resolve_base(&objects[i]);
1105 }
1106 return NULL;
1107 }
1108 #endif
1109
1110 /*
1111 * First pass:
1112 * - find locations of all objects;
1113 * - calculate SHA1 of all non-delta objects;
1114 * - remember base (SHA1 or offset) for all deltas.
1115 */
1116 static void parse_pack_objects(unsigned char *sha1)
1117 {
1118 int i, nr_delays = 0;
1119 struct ofs_delta_entry *ofs_delta = ofs_deltas;
1120 unsigned char ref_delta_sha1[20];
1121 struct stat st;
1122
1123 if (verbose)
1124 progress = start_progress(
1125 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1126 nr_objects);
1127 for (i = 0; i < nr_objects; i++) {
1128 struct object_entry *obj = &objects[i];
1129 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1130 ref_delta_sha1, obj->idx.sha1);
1131 obj->real_type = obj->type;
1132 if (obj->type == OBJ_OFS_DELTA) {
1133 nr_ofs_deltas++;
1134 ofs_delta->obj_no = i;
1135 ofs_delta++;
1136 } else if (obj->type == OBJ_REF_DELTA) {
1137 ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1138 hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1139 ref_deltas[nr_ref_deltas].obj_no = i;
1140 nr_ref_deltas++;
1141 } else if (!data) {
1142 /* large blobs, check later */
1143 obj->real_type = OBJ_BAD;
1144 nr_delays++;
1145 } else
1146 sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1147 free(data);
1148 display_progress(progress, i+1);
1149 }
1150 objects[i].idx.offset = consumed_bytes;
1151 stop_progress(&progress);
1152
1153 /* Check pack integrity */
1154 flush();
1155 git_SHA1_Final(sha1, &input_ctx);
1156 if (hashcmp(fill(20), sha1))
1157 die(_("pack is corrupted (SHA1 mismatch)"));
1158 use(20);
1159
1160 /* If input_fd is a file, we should have reached its end now. */
1161 if (fstat(input_fd, &st))
1162 die_errno(_("cannot fstat packfile"));
1163 if (S_ISREG(st.st_mode) &&
1164 lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1165 die(_("pack has junk at the end"));
1166
1167 for (i = 0; i < nr_objects; i++) {
1168 struct object_entry *obj = &objects[i];
1169 if (obj->real_type != OBJ_BAD)
1170 continue;
1171 obj->real_type = obj->type;
1172 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1173 nr_delays--;
1174 }
1175 if (nr_delays)
1176 die(_("confusion beyond insanity in parse_pack_objects()"));
1177 }
1178
1179 /*
1180 * Second pass:
1181 * - for all non-delta objects, look if it is used as a base for
1182 * deltas;
1183 * - if used as a base, uncompress the object and apply all deltas,
1184 * recursively checking if the resulting object is used as a base
1185 * for some more deltas.
1186 */
1187 static void resolve_deltas(void)
1188 {
1189 int i;
1190
1191 if (!nr_ofs_deltas && !nr_ref_deltas)
1192 return;
1193
1194 /* Sort deltas by base SHA1/offset for fast searching */
1195 QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1196 QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1197
1198 if (verbose || show_resolving_progress)
1199 progress = start_progress(_("Resolving deltas"),
1200 nr_ref_deltas + nr_ofs_deltas);
1201
1202 #ifndef NO_PTHREADS
1203 nr_dispatched = 0;
1204 if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1205 init_thread();
1206 for (i = 0; i < nr_threads; i++) {
1207 int ret = pthread_create(&thread_data[i].thread, NULL,
1208 threaded_second_pass, thread_data + i);
1209 if (ret)
1210 die(_("unable to create thread: %s"),
1211 strerror(ret));
1212 }
1213 for (i = 0; i < nr_threads; i++)
1214 pthread_join(thread_data[i].thread, NULL);
1215 cleanup_thread();
1216 return;
1217 }
1218 #endif
1219
1220 for (i = 0; i < nr_objects; i++) {
1221 struct object_entry *obj = &objects[i];
1222
1223 if (is_delta_type(obj->type))
1224 continue;
1225 resolve_base(obj);
1226 display_progress(progress, nr_resolved_deltas);
1227 }
1228 }
1229
1230 /*
1231 * Third pass:
1232 * - append objects to convert thin pack to full pack if required
1233 * - write the final 20-byte SHA-1
1234 */
1235 static void fix_unresolved_deltas(struct sha1file *f);
1236 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1237 {
1238 if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1239 stop_progress(&progress);
1240 /* Flush remaining pack final 20-byte SHA1. */
1241 flush();
1242 return;
1243 }
1244
1245 if (fix_thin_pack) {
1246 struct sha1file *f;
1247 unsigned char read_sha1[20], tail_sha1[20];
1248 struct strbuf msg = STRBUF_INIT;
1249 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1250 int nr_objects_initial = nr_objects;
1251 if (nr_unresolved <= 0)
1252 die(_("confusion beyond insanity"));
1253 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1254 memset(objects + nr_objects + 1, 0,
1255 nr_unresolved * sizeof(*objects));
1256 f = sha1fd(output_fd, curr_pack);
1257 fix_unresolved_deltas(f);
1258 strbuf_addf(&msg, Q_("completed with %d local object",
1259 "completed with %d local objects",
1260 nr_objects - nr_objects_initial),
1261 nr_objects - nr_objects_initial);
1262 stop_progress_msg(&progress, msg.buf);
1263 strbuf_release(&msg);
1264 sha1close(f, tail_sha1, 0);
1265 hashcpy(read_sha1, pack_sha1);
1266 fixup_pack_header_footer(output_fd, pack_sha1,
1267 curr_pack, nr_objects,
1268 read_sha1, consumed_bytes-20);
1269 if (hashcmp(read_sha1, tail_sha1) != 0)
1270 die(_("Unexpected tail checksum for %s "
1271 "(disk corruption?)"), curr_pack);
1272 }
1273 if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1274 die(Q_("pack has %d unresolved delta",
1275 "pack has %d unresolved deltas",
1276 nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1277 nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1278 }
1279
1280 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1281 {
1282 git_zstream stream;
1283 int status;
1284 unsigned char outbuf[4096];
1285
1286 git_deflate_init(&stream, zlib_compression_level);
1287 stream.next_in = in;
1288 stream.avail_in = size;
1289
1290 do {
1291 stream.next_out = outbuf;
1292 stream.avail_out = sizeof(outbuf);
1293 status = git_deflate(&stream, Z_FINISH);
1294 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1295 } while (status == Z_OK);
1296
1297 if (status != Z_STREAM_END)
1298 die(_("unable to deflate appended object (%d)"), status);
1299 size = stream.total_out;
1300 git_deflate_end(&stream);
1301 return size;
1302 }
1303
1304 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1305 const unsigned char *sha1, void *buf,
1306 unsigned long size, enum object_type type)
1307 {
1308 struct object_entry *obj = &objects[nr_objects++];
1309 unsigned char header[10];
1310 unsigned long s = size;
1311 int n = 0;
1312 unsigned char c = (type << 4) | (s & 15);
1313 s >>= 4;
1314 while (s) {
1315 header[n++] = c | 0x80;
1316 c = s & 0x7f;
1317 s >>= 7;
1318 }
1319 header[n++] = c;
1320 crc32_begin(f);
1321 sha1write(f, header, n);
1322 obj[0].size = size;
1323 obj[0].hdr_size = n;
1324 obj[0].type = type;
1325 obj[0].real_type = type;
1326 obj[1].idx.offset = obj[0].idx.offset + n;
1327 obj[1].idx.offset += write_compressed(f, buf, size);
1328 obj[0].idx.crc32 = crc32_end(f);
1329 sha1flush(f);
1330 hashcpy(obj->idx.sha1, sha1);
1331 return obj;
1332 }
1333
1334 static int delta_pos_compare(const void *_a, const void *_b)
1335 {
1336 struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1337 struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1338 return a->obj_no - b->obj_no;
1339 }
1340
1341 static void fix_unresolved_deltas(struct sha1file *f)
1342 {
1343 struct ref_delta_entry **sorted_by_pos;
1344 int i;
1345
1346 /*
1347 * Since many unresolved deltas may well be themselves base objects
1348 * for more unresolved deltas, we really want to include the
1349 * smallest number of base objects that would cover as much delta
1350 * as possible by picking the
1351 * trunc deltas first, allowing for other deltas to resolve without
1352 * additional base objects. Since most base objects are to be found
1353 * before deltas depending on them, a good heuristic is to start
1354 * resolving deltas in the same order as their position in the pack.
1355 */
1356 ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1357 for (i = 0; i < nr_ref_deltas; i++)
1358 sorted_by_pos[i] = &ref_deltas[i];
1359 QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1360
1361 for (i = 0; i < nr_ref_deltas; i++) {
1362 struct ref_delta_entry *d = sorted_by_pos[i];
1363 enum object_type type;
1364 struct base_data *base_obj = alloc_base_data();
1365
1366 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1367 continue;
1368 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1369 if (!base_obj->data)
1370 continue;
1371
1372 if (check_sha1_signature(d->sha1, base_obj->data,
1373 base_obj->size, typename(type)))
1374 die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1375 base_obj->obj = append_obj_to_pack(f, d->sha1,
1376 base_obj->data, base_obj->size, type);
1377 find_unresolved_deltas(base_obj);
1378 display_progress(progress, nr_resolved_deltas);
1379 }
1380 free(sorted_by_pos);
1381 }
1382
1383 static void final(const char *final_pack_name, const char *curr_pack_name,
1384 const char *final_index_name, const char *curr_index_name,
1385 const char *keep_name, const char *keep_msg,
1386 unsigned char *sha1)
1387 {
1388 const char *report = "pack";
1389 char name[PATH_MAX];
1390 int err;
1391
1392 if (!from_stdin) {
1393 close(input_fd);
1394 } else {
1395 fsync_or_die(output_fd, curr_pack_name);
1396 err = close(output_fd);
1397 if (err)
1398 die_errno(_("error while closing pack file"));
1399 }
1400
1401 if (keep_msg) {
1402 int keep_fd, keep_msg_len = strlen(keep_msg);
1403
1404 if (!keep_name)
1405 keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1406 else
1407 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1408
1409 if (keep_fd < 0) {
1410 if (errno != EEXIST)
1411 die_errno(_("cannot write keep file '%s'"),
1412 keep_name ? keep_name : name);
1413 } else {
1414 if (keep_msg_len > 0) {
1415 write_or_die(keep_fd, keep_msg, keep_msg_len);
1416 write_or_die(keep_fd, "\n", 1);
1417 }
1418 if (close(keep_fd) != 0)
1419 die_errno(_("cannot close written keep file '%s'"),
1420 keep_name ? keep_name : name);
1421 report = "keep";
1422 }
1423 }
1424
1425 if (final_pack_name != curr_pack_name) {
1426 if (!final_pack_name) {
1427 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1428 get_object_directory(), sha1_to_hex(sha1));
1429 final_pack_name = name;
1430 }
1431 if (finalize_object_file(curr_pack_name, final_pack_name))
1432 die(_("cannot store pack file"));
1433 } else if (from_stdin)
1434 chmod(final_pack_name, 0444);
1435
1436 if (final_index_name != curr_index_name) {
1437 if (!final_index_name) {
1438 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1439 get_object_directory(), sha1_to_hex(sha1));
1440 final_index_name = name;
1441 }
1442 if (finalize_object_file(curr_index_name, final_index_name))
1443 die(_("cannot store index file"));
1444 } else
1445 chmod(final_index_name, 0444);
1446
1447 if (!from_stdin) {
1448 printf("%s\n", sha1_to_hex(sha1));
1449 } else {
1450 char buf[48];
1451 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1452 report, sha1_to_hex(sha1));
1453 write_or_die(1, buf, len);
1454
1455 /*
1456 * Let's just mimic git-unpack-objects here and write
1457 * the last part of the input buffer to stdout.
1458 */
1459 while (input_len) {
1460 err = xwrite(1, input_buffer + input_offset, input_len);
1461 if (err <= 0)
1462 break;
1463 input_len -= err;
1464 input_offset += err;
1465 }
1466 }
1467 }
1468
1469 static int git_index_pack_config(const char *k, const char *v, void *cb)
1470 {
1471 struct pack_idx_option *opts = cb;
1472
1473 if (!strcmp(k, "pack.indexversion")) {
1474 opts->version = git_config_int(k, v);
1475 if (opts->version > 2)
1476 die(_("bad pack.indexversion=%"PRIu32), opts->version);
1477 return 0;
1478 }
1479 if (!strcmp(k, "pack.threads")) {
1480 nr_threads = git_config_int(k, v);
1481 if (nr_threads < 0)
1482 die(_("invalid number of threads specified (%d)"),
1483 nr_threads);
1484 #ifdef NO_PTHREADS
1485 if (nr_threads != 1)
1486 warning(_("no threads support, ignoring %s"), k);
1487 nr_threads = 1;
1488 #endif
1489 return 0;
1490 }
1491 return git_default_config(k, v, cb);
1492 }
1493
1494 static int cmp_uint32(const void *a_, const void *b_)
1495 {
1496 uint32_t a = *((uint32_t *)a_);
1497 uint32_t b = *((uint32_t *)b_);
1498
1499 return (a < b) ? -1 : (a != b);
1500 }
1501
1502 static void read_v2_anomalous_offsets(struct packed_git *p,
1503 struct pack_idx_option *opts)
1504 {
1505 const uint32_t *idx1, *idx2;
1506 uint32_t i;
1507
1508 /* The address of the 4-byte offset table */
1509 idx1 = (((const uint32_t *)p->index_data)
1510 + 2 /* 8-byte header */
1511 + 256 /* fan out */
1512 + 5 * p->num_objects /* 20-byte SHA-1 table */
1513 + p->num_objects /* CRC32 table */
1514 );
1515
1516 /* The address of the 8-byte offset table */
1517 idx2 = idx1 + p->num_objects;
1518
1519 for (i = 0; i < p->num_objects; i++) {
1520 uint32_t off = ntohl(idx1[i]);
1521 if (!(off & 0x80000000))
1522 continue;
1523 off = off & 0x7fffffff;
1524 check_pack_index_ptr(p, &idx2[off * 2]);
1525 if (idx2[off * 2])
1526 continue;
1527 /*
1528 * The real offset is ntohl(idx2[off * 2]) in high 4
1529 * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1530 * octets. But idx2[off * 2] is Zero!!!
1531 */
1532 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1533 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1534 }
1535
1536 QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1537 }
1538
1539 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1540 {
1541 struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1542
1543 if (!p)
1544 die(_("Cannot open existing pack file '%s'"), pack_name);
1545 if (open_pack_index(p))
1546 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1547
1548 /* Read the attributes from the existing idx file */
1549 opts->version = p->index_version;
1550
1551 if (opts->version == 2)
1552 read_v2_anomalous_offsets(p, opts);
1553
1554 /*
1555 * Get rid of the idx file as we do not need it anymore.
1556 * NEEDSWORK: extract this bit from free_pack_by_name() in
1557 * sha1_file.c, perhaps? It shouldn't matter very much as we
1558 * know we haven't installed this pack (hence we never have
1559 * read anything from it).
1560 */
1561 close_pack_index(p);
1562 free(p);
1563 }
1564
1565 static void show_pack_info(int stat_only)
1566 {
1567 int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1568 unsigned long *chain_histogram = NULL;
1569
1570 if (deepest_delta)
1571 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1572
1573 for (i = 0; i < nr_objects; i++) {
1574 struct object_entry *obj = &objects[i];
1575
1576 if (is_delta_type(obj->type))
1577 chain_histogram[obj_stat[i].delta_depth - 1]++;
1578 if (stat_only)
1579 continue;
1580 printf("%s %-6s %lu %lu %"PRIuMAX,
1581 sha1_to_hex(obj->idx.sha1),
1582 typename(obj->real_type), obj->size,
1583 (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1584 (uintmax_t)obj->idx.offset);
1585 if (is_delta_type(obj->type)) {
1586 struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1587 printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1));
1588 }
1589 putchar('\n');
1590 }
1591
1592 if (baseobjects)
1593 printf_ln(Q_("non delta: %d object",
1594 "non delta: %d objects",
1595 baseobjects),
1596 baseobjects);
1597 for (i = 0; i < deepest_delta; i++) {
1598 if (!chain_histogram[i])
1599 continue;
1600 printf_ln(Q_("chain length = %d: %lu object",
1601 "chain length = %d: %lu objects",
1602 chain_histogram[i]),
1603 i + 1,
1604 chain_histogram[i]);
1605 }
1606 }
1607
1608 static const char *derive_filename(const char *pack_name, const char *suffix,
1609 struct strbuf *buf)
1610 {
1611 size_t len;
1612 if (!strip_suffix(pack_name, ".pack", &len))
1613 die(_("packfile name '%s' does not end with '.pack'"),
1614 pack_name);
1615 strbuf_add(buf, pack_name, len);
1616 strbuf_addstr(buf, suffix);
1617 return buf->buf;
1618 }
1619
1620 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1621 {
1622 int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1623 const char *curr_index;
1624 const char *index_name = NULL, *pack_name = NULL;
1625 const char *keep_name = NULL, *keep_msg = NULL;
1626 struct strbuf index_name_buf = STRBUF_INIT,
1627 keep_name_buf = STRBUF_INIT;
1628 struct pack_idx_entry **idx_objects;
1629 struct pack_idx_option opts;
1630 unsigned char pack_sha1[20];
1631 unsigned foreign_nr = 1; /* zero is a "good" value, assume bad */
1632 int report_end_of_input = 0;
1633
1634 if (argc == 2 && !strcmp(argv[1], "-h"))
1635 usage(index_pack_usage);
1636
1637 check_replace_refs = 0;
1638 fsck_options.walk = mark_link;
1639
1640 reset_pack_idx_option(&opts);
1641 git_config(git_index_pack_config, &opts);
1642 if (prefix && chdir(prefix))
1643 die(_("Cannot come back to cwd"));
1644
1645 for (i = 1; i < argc; i++) {
1646 const char *arg = argv[i];
1647
1648 if (*arg == '-') {
1649 if (!strcmp(arg, "--stdin")) {
1650 from_stdin = 1;
1651 } else if (!strcmp(arg, "--fix-thin")) {
1652 fix_thin_pack = 1;
1653 } else if (!strcmp(arg, "--strict")) {
1654 strict = 1;
1655 do_fsck_object = 1;
1656 } else if (skip_prefix(arg, "--strict=", &arg)) {
1657 strict = 1;
1658 do_fsck_object = 1;
1659 fsck_set_msg_types(&fsck_options, arg);
1660 } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1661 strict = 1;
1662 check_self_contained_and_connected = 1;
1663 } else if (!strcmp(arg, "--verify")) {
1664 verify = 1;
1665 } else if (!strcmp(arg, "--verify-stat")) {
1666 verify = 1;
1667 show_stat = 1;
1668 } else if (!strcmp(arg, "--verify-stat-only")) {
1669 verify = 1;
1670 show_stat = 1;
1671 stat_only = 1;
1672 } else if (!strcmp(arg, "--keep")) {
1673 keep_msg = "";
1674 } else if (starts_with(arg, "--keep=")) {
1675 keep_msg = arg + 7;
1676 } else if (starts_with(arg, "--threads=")) {
1677 char *end;
1678 nr_threads = strtoul(arg+10, &end, 0);
1679 if (!arg[10] || *end || nr_threads < 0)
1680 usage(index_pack_usage);
1681 #ifdef NO_PTHREADS
1682 if (nr_threads != 1)
1683 warning(_("no threads support, "
1684 "ignoring %s"), arg);
1685 nr_threads = 1;
1686 #endif
1687 } else if (starts_with(arg, "--pack_header=")) {
1688 struct pack_header *hdr;
1689 char *c;
1690
1691 hdr = (struct pack_header *)input_buffer;
1692 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1693 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1694 if (*c != ',')
1695 die(_("bad %s"), arg);
1696 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1697 if (*c)
1698 die(_("bad %s"), arg);
1699 input_len = sizeof(*hdr);
1700 } else if (!strcmp(arg, "-v")) {
1701 verbose = 1;
1702 } else if (!strcmp(arg, "--show-resolving-progress")) {
1703 show_resolving_progress = 1;
1704 } else if (!strcmp(arg, "--report-end-of-input")) {
1705 report_end_of_input = 1;
1706 } else if (!strcmp(arg, "-o")) {
1707 if (index_name || (i+1) >= argc)
1708 usage(index_pack_usage);
1709 index_name = argv[++i];
1710 } else if (starts_with(arg, "--index-version=")) {
1711 char *c;
1712 opts.version = strtoul(arg + 16, &c, 10);
1713 if (opts.version > 2)
1714 die(_("bad %s"), arg);
1715 if (*c == ',')
1716 opts.off32_limit = strtoul(c+1, &c, 0);
1717 if (*c || opts.off32_limit & 0x80000000)
1718 die(_("bad %s"), arg);
1719 } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1720 max_input_size = strtoumax(arg, NULL, 10);
1721 } else
1722 usage(index_pack_usage);
1723 continue;
1724 }
1725
1726 if (pack_name)
1727 usage(index_pack_usage);
1728 pack_name = arg;
1729 }
1730
1731 if (!pack_name && !from_stdin)
1732 usage(index_pack_usage);
1733 if (fix_thin_pack && !from_stdin)
1734 die(_("--fix-thin cannot be used without --stdin"));
1735 if (from_stdin && !startup_info->have_repository)
1736 die(_("--stdin requires a git repository"));
1737 if (!index_name && pack_name)
1738 index_name = derive_filename(pack_name, ".idx", &index_name_buf);
1739 if (keep_msg && !keep_name && pack_name)
1740 keep_name = derive_filename(pack_name, ".keep", &keep_name_buf);
1741
1742 if (verify) {
1743 if (!index_name)
1744 die(_("--verify with no packfile name given"));
1745 read_idx_option(&opts, index_name);
1746 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1747 }
1748 if (strict)
1749 opts.flags |= WRITE_IDX_STRICT;
1750
1751 #ifndef NO_PTHREADS
1752 if (!nr_threads) {
1753 nr_threads = online_cpus();
1754 /* An experiment showed that more threads does not mean faster */
1755 if (nr_threads > 3)
1756 nr_threads = 3;
1757 }
1758 #endif
1759
1760 curr_pack = open_pack_file(pack_name);
1761 parse_pack_header();
1762 objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1763 if (show_stat)
1764 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1765 ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1766 parse_pack_objects(pack_sha1);
1767 if (report_end_of_input)
1768 write_in_full(2, "\0", 1);
1769 resolve_deltas();
1770 conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1771 free(ofs_deltas);
1772 free(ref_deltas);
1773 if (strict)
1774 foreign_nr = check_objects();
1775
1776 if (show_stat)
1777 show_pack_info(stat_only);
1778
1779 ALLOC_ARRAY(idx_objects, nr_objects);
1780 for (i = 0; i < nr_objects; i++)
1781 idx_objects[i] = &objects[i].idx;
1782 curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1783 free(idx_objects);
1784
1785 if (!verify)
1786 final(pack_name, curr_pack,
1787 index_name, curr_index,
1788 keep_name, keep_msg,
1789 pack_sha1);
1790 else
1791 close(input_fd);
1792 free(objects);
1793 strbuf_release(&index_name_buf);
1794 strbuf_release(&keep_name_buf);
1795 if (pack_name == NULL)
1796 free((void *) curr_pack);
1797 if (index_name == NULL)
1798 free((void *) curr_index);
1799
1800 /*
1801 * Let the caller know this pack is not self contained
1802 */
1803 if (check_self_contained_and_connected && foreign_nr)
1804 return 1;
1805
1806 return 0;
1807 }
ramsay's fork of git