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 "thread-utils.h"
13static const char index_pack_usage[] =
15"git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
16struct object_entry {
18 struct pack_idx_entry idx;
19 unsigned long size;
20 unsigned int hdr_size;
21 enum object_type type;
22 enum object_type real_type;
23 unsigned delta_depth;
24 int base_object_no;
25};
26union delta_base {
28 unsigned char sha1[20];
29 off_t offset;
30};
31struct base_data {
33 struct base_data *base;
34 struct base_data *child;
35 struct object_entry *obj;
36 void *data;
37 unsigned long size;
38 int ref_first, ref_last;
39 int ofs_first, ofs_last;
40};
41#if !defined(NO_PTHREADS) && defined(NO_PREAD)
43/* NO_PREAD uses compat/pread.c, which is not thread-safe. Disable threading. */
44#define NO_PTHREADS
45#endif
46struct thread_local {
48#ifndef NO_PTHREADS
49 pthread_t thread;
50#endif
51 struct base_data *base_cache;
52 size_t base_cache_used;
53};
54/*
56 * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
57 * to memcmp() only the first 20 bytes.
58 */
59#define UNION_BASE_SZ 20
60#define FLAG_LINK (1u<<20)
62#define FLAG_CHECKED (1u<<21)
63struct delta_entry {
65 union delta_base base;
66 int obj_no;
67};
68static struct object_entry *objects;
70static struct delta_entry *deltas;
71static struct thread_local nothread_data;
72static int nr_objects;
73static int nr_deltas;
74static int nr_resolved_deltas;
75static int nr_threads;
76static int from_stdin;
78static int strict;
79static int verbose;
80static struct progress *progress;
82/* We always read in 4kB chunks. */
84static unsigned char input_buffer[4096];
85static unsigned int input_offset, input_len;
86static off_t consumed_bytes;
87static unsigned deepest_delta;
88static git_SHA_CTX input_ctx;
89static uint32_t input_crc32;
90static int input_fd, output_fd, pack_fd;
91#ifndef NO_PTHREADS
93static struct thread_local *thread_data;
95static int nr_dispatched;
96static int threads_active;
97static pthread_mutex_t read_mutex;
99#define read_lock() lock_mutex(&read_mutex)
100#define read_unlock() unlock_mutex(&read_mutex)
101static pthread_mutex_t counter_mutex;
103#define counter_lock() lock_mutex(&counter_mutex)
104#define counter_unlock() unlock_mutex(&counter_mutex)
105static pthread_mutex_t work_mutex;
107#define work_lock() lock_mutex(&work_mutex)
108#define work_unlock() unlock_mutex(&work_mutex)
109static pthread_key_t key;
111static inline void lock_mutex(pthread_mutex_t *mutex)
113{
114 if (threads_active)
115 pthread_mutex_lock(mutex);
116}
117static inline void unlock_mutex(pthread_mutex_t *mutex)
119{
120 if (threads_active)
121 pthread_mutex_unlock(mutex);
122}
123/*
125 * Mutex and conditional variable can't be statically-initialized on Windows.
126 */
127static void init_thread(void)
128{
129 init_recursive_mutex(&read_mutex);
130 pthread_mutex_init(&counter_mutex, NULL);
131 pthread_mutex_init(&work_mutex, NULL);
132 pthread_key_create(&key, NULL);
133 thread_data = xcalloc(nr_threads, sizeof(*thread_data));
134 threads_active = 1;
135}
136static void cleanup_thread(void)
138{
139 if (!threads_active)
140 return;
141 threads_active = 0;
142 pthread_mutex_destroy(&read_mutex);
143 pthread_mutex_destroy(&counter_mutex);
144 pthread_mutex_destroy(&work_mutex);
145 pthread_key_delete(key);
146 free(thread_data);
147}
148#else
150#define read_lock()
152#define read_unlock()
153#define counter_lock()
155#define counter_unlock()
156#define work_lock()
158#define work_unlock()
159#endif
161static int mark_link(struct object *obj, int type, void *data)
164{
165 if (!obj)
166 return -1;
167 if (type != OBJ_ANY && obj->type != type)
169 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
170 obj->flags |= FLAG_LINK;
172 return 0;
173}
174/* The content of each linked object must have been checked
176 or it must be already present in the object database */
177static void check_object(struct object *obj)
178{
179 if (!obj)
180 return;
181 if (!(obj->flags & FLAG_LINK))
183 return;
184 if (!(obj->flags & FLAG_CHECKED)) {
186 unsigned long size;
187 int type = sha1_object_info(obj->sha1, &size);
188 if (type != obj->type || type <= 0)
189 die(_("object of unexpected type"));
190 obj->flags |= FLAG_CHECKED;
191 return;
192 }
193}
194static void check_objects(void)
196{
197 unsigned i, max;
198 max = get_max_object_index();
200 for (i = 0; i < max; i++)
201 check_object(get_indexed_object(i));
202}
203/* Discard current buffer used content. */
206static void flush(void)
207{
208 if (input_offset) {
209 if (output_fd >= 0)
210 write_or_die(output_fd, input_buffer, input_offset);
211 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
212 memmove(input_buffer, input_buffer + input_offset, input_len);
213 input_offset = 0;
214 }
215}
216/*
218 * Make sure at least "min" bytes are available in the buffer, and
219 * return the pointer to the buffer.
220 */
221static void *fill(int min)
222{
223 if (min <= input_len)
224 return input_buffer + input_offset;
225 if (min > sizeof(input_buffer))
226 die(Q_("cannot fill %d byte",
227 "cannot fill %d bytes",
228 min),
229 min);
230 flush();
231 do {
232 ssize_t ret = xread(input_fd, input_buffer + input_len,
233 sizeof(input_buffer) - input_len);
234 if (ret <= 0) {
235 if (!ret)
236 die(_("early EOF"));
237 die_errno(_("read error on input"));
238 }
239 input_len += ret;
240 if (from_stdin)
241 display_throughput(progress, consumed_bytes + input_len);
242 } while (input_len < min);
243 return input_buffer;
244}
245static void use(int bytes)
247{
248 if (bytes > input_len)
249 die(_("used more bytes than were available"));
250 input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
251 input_len -= bytes;
252 input_offset += bytes;
253 /* make sure off_t is sufficiently large not to wrap */
255 if (signed_add_overflows(consumed_bytes, bytes))
256 die(_("pack too large for current definition of off_t"));
257 consumed_bytes += bytes;
258}
259static const char *open_pack_file(const char *pack_name)
261{
262 if (from_stdin) {
263 input_fd = 0;
264 if (!pack_name) {
265 static char tmp_file[PATH_MAX];
266 output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
267 "pack/tmp_pack_XXXXXX");
268 pack_name = xstrdup(tmp_file);
269 } else
270 output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
271 if (output_fd < 0)
272 die_errno(_("unable to create '%s'"), pack_name);
273 pack_fd = output_fd;
274 } else {
275 input_fd = open(pack_name, O_RDONLY);
276 if (input_fd < 0)
277 die_errno(_("cannot open packfile '%s'"), pack_name);
278 output_fd = -1;
279 pack_fd = input_fd;
280 }
281 git_SHA1_Init(&input_ctx);
282 return pack_name;
283}
284static void parse_pack_header(void)
286{
287 struct pack_header *hdr = fill(sizeof(struct pack_header));
288 /* Header consistency check */
290 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
291 die(_("pack signature mismatch"));
292 if (!pack_version_ok(hdr->hdr_version))
293 die("pack version %"PRIu32" unsupported",
294 ntohl(hdr->hdr_version));
295 nr_objects = ntohl(hdr->hdr_entries);
297 use(sizeof(struct pack_header));
298}
299static NORETURN void bad_object(unsigned long offset, const char *format,
301 ...) __attribute__((format (printf, 2, 3)));
302static NORETURN void bad_object(unsigned long offset, const char *format, ...)
304{
305 va_list params;
306 char buf[1024];
307 va_start(params, format);
309 vsnprintf(buf, sizeof(buf), format, params);
310 va_end(params);
311 die(_("pack has bad object at offset %lu: %s"), offset, buf);
312}
313static inline struct thread_local *get_thread_data(void)
315{
316#ifndef NO_PTHREADS
317 if (threads_active)
318 return pthread_getspecific(key);
319 assert(!threads_active &&
320 "This should only be reached when all threads are gone");
321#endif
322 return ¬hread_data;
323}
324#ifndef NO_PTHREADS
326static void set_thread_data(struct thread_local *data)
327{
328 if (threads_active)
329 pthread_setspecific(key, data);
330}
331#endif
332static struct base_data *alloc_base_data(void)
334{
335 struct base_data *base = xmalloc(sizeof(struct base_data));
336 memset(base, 0, sizeof(*base));
337 base->ref_last = -1;
338 base->ofs_last = -1;
339 return base;
340}
341static void free_base_data(struct base_data *c)
343{
344 if (c->data) {
345 free(c->data);
346 c->data = NULL;
347 get_thread_data()->base_cache_used -= c->size;
348 }
349}
350static void prune_base_data(struct base_data *retain)
352{
353 struct base_data *b;
354 struct thread_local *data = get_thread_data();
355 for (b = data->base_cache;
356 data->base_cache_used > delta_base_cache_limit && b;
357 b = b->child) {
358 if (b->data && b != retain)
359 free_base_data(b);
360 }
361}
362static void link_base_data(struct base_data *base, struct base_data *c)
364{
365 if (base)
366 base->child = c;
367 else
368 get_thread_data()->base_cache = c;
369 c->base = base;
371 c->child = NULL;
372 if (c->data)
373 get_thread_data()->base_cache_used += c->size;
374 prune_base_data(c);
375}
376static void unlink_base_data(struct base_data *c)
378{
379 struct base_data *base = c->base;
380 if (base)
381 base->child = NULL;
382 else
383 get_thread_data()->base_cache = NULL;
384 free_base_data(c);
385}
386static int is_delta_type(enum object_type type)
388{
389 return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
390}
391static void *unpack_entry_data(unsigned long offset, unsigned long size,
393 enum object_type type, unsigned char *sha1)
394{
395 int status;
396 git_zstream stream;
397 void *buf = xmalloc(size);
398 git_SHA_CTX c;
399 char hdr[32];
400 int hdrlen;
401 if (!is_delta_type(type)) {
403 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
404 git_SHA1_Init(&c);
405 git_SHA1_Update(&c, hdr, hdrlen);
406 } else
407 sha1 = NULL;
408 memset(&stream, 0, sizeof(stream));
410 git_inflate_init(&stream);
411 stream.next_out = buf;
412 stream.avail_out = size;
413 do {
415 unsigned char *last_out = stream.next_out;
416 stream.next_in = fill(1);
417 stream.avail_in = input_len;
418 status = git_inflate(&stream, 0);
419 use(input_len - stream.avail_in);
420 if (sha1)
421 git_SHA1_Update(&c, last_out, stream.next_out - last_out);
422 } while (status == Z_OK);
423 if (stream.total_out != size || status != Z_STREAM_END)
424 bad_object(offset, _("inflate returned %d"), status);
425 git_inflate_end(&stream);
426 if (sha1)
427 git_SHA1_Final(sha1, &c);
428 return buf;
429}
430static void *unpack_raw_entry(struct object_entry *obj,
432 union delta_base *delta_base,
433 unsigned char *sha1)
434{
435 unsigned char *p;
436 unsigned long size, c;
437 off_t base_offset;
438 unsigned shift;
439 void *data;
440 obj->idx.offset = consumed_bytes;
442 input_crc32 = crc32(0, NULL, 0);
443 p = fill(1);
445 c = *p;
446 use(1);
447 obj->type = (c >> 4) & 7;
448 size = (c & 15);
449 shift = 4;
450 while (c & 0x80) {
451 p = fill(1);
452 c = *p;
453 use(1);
454 size += (c & 0x7f) << shift;
455 shift += 7;
456 }
457 obj->size = size;
458 switch (obj->type) {
460 case OBJ_REF_DELTA:
461 hashcpy(delta_base->sha1, fill(20));
462 use(20);
463 break;
464 case OBJ_OFS_DELTA:
465 memset(delta_base, 0, sizeof(*delta_base));
466 p = fill(1);
467 c = *p;
468 use(1);
469 base_offset = c & 127;
470 while (c & 128) {
471 base_offset += 1;
472 if (!base_offset || MSB(base_offset, 7))
473 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
474 p = fill(1);
475 c = *p;
476 use(1);
477 base_offset = (base_offset << 7) + (c & 127);
478 }
479 delta_base->offset = obj->idx.offset - base_offset;
480 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
481 bad_object(obj->idx.offset, _("delta base offset is out of bound"));
482 break;
483 case OBJ_COMMIT:
484 case OBJ_TREE:
485 case OBJ_BLOB:
486 case OBJ_TAG:
487 break;
488 default:
489 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
490 }
491 obj->hdr_size = consumed_bytes - obj->idx.offset;
492 data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
494 obj->idx.crc32 = input_crc32;
495 return data;
496}
497static void *get_data_from_pack(struct object_entry *obj)
499{
500 off_t from = obj[0].idx.offset + obj[0].hdr_size;
501 unsigned long len = obj[1].idx.offset - from;
502 unsigned char *data, *inbuf;
503 git_zstream stream;
504 int status;
505 data = xmalloc(obj->size);
507 inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
508 memset(&stream, 0, sizeof(stream));
510 git_inflate_init(&stream);
511 stream.next_out = data;
512 stream.avail_out = obj->size;
513 do {
515 ssize_t n = (len < 64*1024) ? len : 64*1024;
516 n = pread(pack_fd, inbuf, n, from);
517 if (n < 0)
518 die_errno(_("cannot pread pack file"));
519 if (!n)
520 die(Q_("premature end of pack file, %lu byte missing",
521 "premature end of pack file, %lu bytes missing",
522 len),
523 len);
524 from += n;
525 len -= n;
526 stream.next_in = inbuf;
527 stream.avail_in = n;
528 status = git_inflate(&stream, 0);
529 } while (len && status == Z_OK && !stream.avail_in);
530 /* This has been inflated OK when first encountered, so... */
532 if (status != Z_STREAM_END || stream.total_out != obj->size)
533 die(_("serious inflate inconsistency"));
534 git_inflate_end(&stream);
536 free(inbuf);
537 return data;
538}
539static int compare_delta_bases(const union delta_base *base1,
541 const union delta_base *base2,
542 enum object_type type1,
543 enum object_type type2)
544{
545 int cmp = type1 - type2;
546 if (cmp)
547 return cmp;
548 return memcmp(base1, base2, UNION_BASE_SZ);
549}
550static int find_delta(const union delta_base *base, enum object_type type)
552{
553 int first = 0, last = nr_deltas;
554 while (first < last) {
556 int next = (first + last) / 2;
557 struct delta_entry *delta = &deltas[next];
558 int cmp;
559 cmp = compare_delta_bases(base, &delta->base,
561 type, objects[delta->obj_no].type);
562 if (!cmp)
563 return next;
564 if (cmp < 0) {
565 last = next;
566 continue;
567 }
568 first = next+1;
569 }
570 return -first-1;
571}
572static void find_delta_children(const union delta_base *base,
574 int *first_index, int *last_index,
575 enum object_type type)
576{
577 int first = find_delta(base, type);
578 int last = first;
579 int end = nr_deltas - 1;
580 if (first < 0) {
582 *first_index = 0;
583 *last_index = -1;
584 return;
585 }
586 while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
587 --first;
588 while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
589 ++last;
590 *first_index = first;
591 *last_index = last;
592}
593static void sha1_object(const void *data, unsigned long size,
595 enum object_type type, const unsigned char *sha1)
596{
597 read_lock();
598 if (has_sha1_file(sha1)) {
599 void *has_data;
600 enum object_type has_type;
601 unsigned long has_size;
602 has_data = read_sha1_file(sha1, &has_type, &has_size);
603 read_unlock();
604 if (!has_data)
605 die(_("cannot read existing object %s"), sha1_to_hex(sha1));
606 if (size != has_size || type != has_type ||
607 memcmp(data, has_data, size) != 0)
608 die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
609 free(has_data);
610 } else
611 read_unlock();
612 if (strict) {
614 read_lock();
615 if (type == OBJ_BLOB) {
616 struct blob *blob = lookup_blob(sha1);
617 if (blob)
618 blob->object.flags |= FLAG_CHECKED;
619 else
620 die(_("invalid blob object %s"), sha1_to_hex(sha1));
621 } else {
622 struct object *obj;
623 int eaten;
624 void *buf = (void *) data;
625 /*
627 * we do not need to free the memory here, as the
628 * buf is deleted by the caller.
629 */
630 obj = parse_object_buffer(sha1, type, size, buf, &eaten);
631 if (!obj)
632 die(_("invalid %s"), typename(type));
633 if (fsck_object(obj, 1, fsck_error_function))
634 die(_("Error in object"));
635 if (fsck_walk(obj, mark_link, NULL))
636 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
637 if (obj->type == OBJ_TREE) {
639 struct tree *item = (struct tree *) obj;
640 item->buffer = NULL;
641 }
642 if (obj->type == OBJ_COMMIT) {
643 struct commit *commit = (struct commit *) obj;
644 commit->buffer = NULL;
645 }
646 obj->flags |= FLAG_CHECKED;
647 }
648 read_unlock();
649 }
650}
651/*
653 * This function is part of find_unresolved_deltas(). There are two
654 * walkers going in the opposite ways.
655 *
656 * The first one in find_unresolved_deltas() traverses down from
657 * parent node to children, deflating nodes along the way. However,
658 * memory for deflated nodes is limited by delta_base_cache_limit, so
659 * at some point parent node's deflated content may be freed.
660 *
661 * The second walker is this function, which goes from current node up
662 * to top parent if necessary to deflate the node. In normal
663 * situation, its parent node would be already deflated, so it just
664 * needs to apply delta.
665 *
666 * In the worst case scenario, parent node is no longer deflated because
667 * we're running out of delta_base_cache_limit; we need to re-deflate
668 * parents, possibly up to the top base.
669 *
670 * All deflated objects here are subject to be freed if we exceed
671 * delta_base_cache_limit, just like in find_unresolved_deltas(), we
672 * just need to make sure the last node is not freed.
673 */
674static void *get_base_data(struct base_data *c)
675{
676 if (!c->data) {
677 struct object_entry *obj = c->obj;
678 struct base_data **delta = NULL;
679 int delta_nr = 0, delta_alloc = 0;
680 while (is_delta_type(c->obj->type) && !c->data) {
682 ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
683 delta[delta_nr++] = c;
684 c = c->base;
685 }
686 if (!delta_nr) {
687 c->data = get_data_from_pack(obj);
688 c->size = obj->size;
689 get_thread_data()->base_cache_used += c->size;
690 prune_base_data(c);
691 }
692 for (; delta_nr > 0; delta_nr--) {
693 void *base, *raw;
694 c = delta[delta_nr - 1];
695 obj = c->obj;
696 base = get_base_data(c->base);
697 raw = get_data_from_pack(obj);
698 c->data = patch_delta(
699 base, c->base->size,
700 raw, obj->size,
701 &c->size);
702 free(raw);
703 if (!c->data)
704 bad_object(obj->idx.offset, _("failed to apply delta"));
705 get_thread_data()->base_cache_used += c->size;
706 prune_base_data(c);
707 }
708 free(delta);
709 }
710 return c->data;
711}
712static void resolve_delta(struct object_entry *delta_obj,
714 struct base_data *base, struct base_data *result)
715{
716 void *base_data, *delta_data;
717 delta_obj->real_type = base->obj->real_type;
719 delta_obj->delta_depth = base->obj->delta_depth + 1;
720 if (deepest_delta < delta_obj->delta_depth)
721 deepest_delta = delta_obj->delta_depth;
722 delta_obj->base_object_no = base->obj - objects;
723 delta_data = get_data_from_pack(delta_obj);
724 base_data = get_base_data(base);
725 result->obj = delta_obj;
726 result->data = patch_delta(base_data, base->size,
727 delta_data, delta_obj->size, &result->size);
728 free(delta_data);
729 if (!result->data)
730 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
731 hash_sha1_file(result->data, result->size,
732 typename(delta_obj->real_type), delta_obj->idx.sha1);
733 sha1_object(result->data, result->size, delta_obj->real_type,
734 delta_obj->idx.sha1);
735 counter_lock();
736 nr_resolved_deltas++;
737 counter_unlock();
738}
739static struct base_data *find_unresolved_deltas_1(struct base_data *base,
741 struct base_data *prev_base)
742{
743 if (base->ref_last == -1 && base->ofs_last == -1) {
744 union delta_base base_spec;
745 hashcpy(base_spec.sha1, base->obj->idx.sha1);
747 find_delta_children(&base_spec,
748 &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
749 memset(&base_spec, 0, sizeof(base_spec));
751 base_spec.offset = base->obj->idx.offset;
752 find_delta_children(&base_spec,
753 &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
754 if (base->ref_last == -1 && base->ofs_last == -1) {
756 free(base->data);
757 return NULL;
758 }
759 link_base_data(prev_base, base);
761 }
762 if (base->ref_first <= base->ref_last) {
764 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
765 struct base_data *result = alloc_base_data();
766 assert(child->real_type == OBJ_REF_DELTA);
768 resolve_delta(child, base, result);
769 if (base->ref_first == base->ref_last && base->ofs_last == -1)
770 free_base_data(base);
771 base->ref_first++;
773 return result;
774 }
775 if (base->ofs_first <= base->ofs_last) {
777 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
778 struct base_data *result = alloc_base_data();
779 assert(child->real_type == OBJ_OFS_DELTA);
781 resolve_delta(child, base, result);
782 if (base->ofs_first == base->ofs_last)
783 free_base_data(base);
784 base->ofs_first++;
786 return result;
787 }
788 unlink_base_data(base);
790 return NULL;
791}
792static void find_unresolved_deltas(struct base_data *base)
794{
795 struct base_data *new_base, *prev_base = NULL;
796 for (;;) {
797 new_base = find_unresolved_deltas_1(base, prev_base);
798 if (new_base) {
800 prev_base = base;
801 base = new_base;
802 } else {
803 free(base);
804 base = prev_base;
805 if (!base)
806 return;
807 prev_base = base->base;
808 }
809 }
810}
811static int compare_delta_entry(const void *a, const void *b)
813{
814 const struct delta_entry *delta_a = a;
815 const struct delta_entry *delta_b = b;
816 /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
818 return compare_delta_bases(&delta_a->base, &delta_b->base,
819 objects[delta_a->obj_no].type,
820 objects[delta_b->obj_no].type);
821}
822static void resolve_base(struct object_entry *obj)
824{
825 struct base_data *base_obj = alloc_base_data();
826 base_obj->obj = obj;
827 base_obj->data = NULL;
828 find_unresolved_deltas(base_obj);
829}
830#ifndef NO_PTHREADS
832static void *threaded_second_pass(void *data)
833{
834 set_thread_data(data);
835 for (;;) {
836 int i;
837 work_lock();
838 display_progress(progress, nr_resolved_deltas);
839 while (nr_dispatched < nr_objects &&
840 is_delta_type(objects[nr_dispatched].type))
841 nr_dispatched++;
842 if (nr_dispatched >= nr_objects) {
843 work_unlock();
844 break;
845 }
846 i = nr_dispatched++;
847 work_unlock();
848 resolve_base(&objects[i]);
850 }
851 return NULL;
852}
853#endif
854/*
856 * First pass:
857 * - find locations of all objects;
858 * - calculate SHA1 of all non-delta objects;
859 * - remember base (SHA1 or offset) for all deltas.
860 */
861static void parse_pack_objects(unsigned char *sha1)
862{
863 int i;
864 struct delta_entry *delta = deltas;
865 struct stat st;
866 if (verbose)
868 progress = start_progress(
869 from_stdin ? _("Receiving objects") : _("Indexing objects"),
870 nr_objects);
871 for (i = 0; i < nr_objects; i++) {
872 struct object_entry *obj = &objects[i];
873 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
874 obj->real_type = obj->type;
875 if (is_delta_type(obj->type)) {
876 nr_deltas++;
877 delta->obj_no = i;
878 delta++;
879 } else
880 sha1_object(data, obj->size, obj->type, obj->idx.sha1);
881 free(data);
882 display_progress(progress, i+1);
883 }
884 objects[i].idx.offset = consumed_bytes;
885 stop_progress(&progress);
886 /* Check pack integrity */
888 flush();
889 git_SHA1_Final(sha1, &input_ctx);
890 if (hashcmp(fill(20), sha1))
891 die(_("pack is corrupted (SHA1 mismatch)"));
892 use(20);
893 /* If input_fd is a file, we should have reached its end now. */
895 if (fstat(input_fd, &st))
896 die_errno(_("cannot fstat packfile"));
897 if (S_ISREG(st.st_mode) &&
898 lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
899 die(_("pack has junk at the end"));
900}
901/*
903 * Second pass:
904 * - for all non-delta objects, look if it is used as a base for
905 * deltas;
906 * - if used as a base, uncompress the object and apply all deltas,
907 * recursively checking if the resulting object is used as a base
908 * for some more deltas.
909 */
910static void resolve_deltas(void)
911{
912 int i;
913 if (!nr_deltas)
915 return;
916 /* Sort deltas by base SHA1/offset for fast searching */
918 qsort(deltas, nr_deltas, sizeof(struct delta_entry),
919 compare_delta_entry);
920 if (verbose)
922 progress = start_progress(_("Resolving deltas"), nr_deltas);
923#ifndef NO_PTHREADS
925 nr_dispatched = 0;
926 if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
927 init_thread();
928 for (i = 0; i < nr_threads; i++) {
929 int ret = pthread_create(&thread_data[i].thread, NULL,
930 threaded_second_pass, thread_data + i);
931 if (ret)
932 die("unable to create thread: %s", strerror(ret));
933 }
934 for (i = 0; i < nr_threads; i++)
935 pthread_join(thread_data[i].thread, NULL);
936 cleanup_thread();
937 return;
938 }
939#endif
940 for (i = 0; i < nr_objects; i++) {
942 struct object_entry *obj = &objects[i];
943 if (is_delta_type(obj->type))
945 continue;
946 resolve_base(obj);
947 display_progress(progress, nr_resolved_deltas);
948 }
949}
950/*
952 * Third pass:
953 * - append objects to convert thin pack to full pack if required
954 * - write the final 20-byte SHA-1
955 */
956static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
957static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
958{
959 if (nr_deltas == nr_resolved_deltas) {
960 stop_progress(&progress);
961 /* Flush remaining pack final 20-byte SHA1. */
962 flush();
963 return;
964 }
965 if (fix_thin_pack) {
967 struct sha1file *f;
968 unsigned char read_sha1[20], tail_sha1[20];
969 char msg[48];
970 int nr_unresolved = nr_deltas - nr_resolved_deltas;
971 int nr_objects_initial = nr_objects;
972 if (nr_unresolved <= 0)
973 die(_("confusion beyond insanity"));
974 objects = xrealloc(objects,
975 (nr_objects + nr_unresolved + 1)
976 * sizeof(*objects));
977 f = sha1fd(output_fd, curr_pack);
978 fix_unresolved_deltas(f, nr_unresolved);
979 sprintf(msg, "completed with %d local objects",
980 nr_objects - nr_objects_initial);
981 stop_progress_msg(&progress, msg);
982 sha1close(f, tail_sha1, 0);
983 hashcpy(read_sha1, pack_sha1);
984 fixup_pack_header_footer(output_fd, pack_sha1,
985 curr_pack, nr_objects,
986 read_sha1, consumed_bytes-20);
987 if (hashcmp(read_sha1, tail_sha1) != 0)
988 die("Unexpected tail checksum for %s "
989 "(disk corruption?)", curr_pack);
990 }
991 if (nr_deltas != nr_resolved_deltas)
992 die(Q_("pack has %d unresolved delta",
993 "pack has %d unresolved deltas",
994 nr_deltas - nr_resolved_deltas),
995 nr_deltas - nr_resolved_deltas);
996}
997static int write_compressed(struct sha1file *f, void *in, unsigned int size)
999{
1000 git_zstream stream;
1001 int status;
1002 unsigned char outbuf[4096];
1003 memset(&stream, 0, sizeof(stream));
1005 git_deflate_init(&stream, zlib_compression_level);
1006 stream.next_in = in;
1007 stream.avail_in = size;
1008 do {
1010 stream.next_out = outbuf;
1011 stream.avail_out = sizeof(outbuf);
1012 status = git_deflate(&stream, Z_FINISH);
1013 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1014 } while (status == Z_OK);
1015 if (status != Z_STREAM_END)
1017 die(_("unable to deflate appended object (%d)"), status);
1018 size = stream.total_out;
1019 git_deflate_end(&stream);
1020 return size;
1021}
1022static struct object_entry *append_obj_to_pack(struct sha1file *f,
1024 const unsigned char *sha1, void *buf,
1025 unsigned long size, enum object_type type)
1026{
1027 struct object_entry *obj = &objects[nr_objects++];
1028 unsigned char header[10];
1029 unsigned long s = size;
1030 int n = 0;
1031 unsigned char c = (type << 4) | (s & 15);
1032 s >>= 4;
1033 while (s) {
1034 header[n++] = c | 0x80;
1035 c = s & 0x7f;
1036 s >>= 7;
1037 }
1038 header[n++] = c;
1039 crc32_begin(f);
1040 sha1write(f, header, n);
1041 obj[0].size = size;
1042 obj[0].hdr_size = n;
1043 obj[0].type = type;
1044 obj[0].real_type = type;
1045 obj[1].idx.offset = obj[0].idx.offset + n;
1046 obj[1].idx.offset += write_compressed(f, buf, size);
1047 obj[0].idx.crc32 = crc32_end(f);
1048 sha1flush(f);
1049 hashcpy(obj->idx.sha1, sha1);
1050 return obj;
1051}
1052static int delta_pos_compare(const void *_a, const void *_b)
1054{
1055 struct delta_entry *a = *(struct delta_entry **)_a;
1056 struct delta_entry *b = *(struct delta_entry **)_b;
1057 return a->obj_no - b->obj_no;
1058}
1059static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1061{
1062 struct delta_entry **sorted_by_pos;
1063 int i, n = 0;
1064 /*
1066 * Since many unresolved deltas may well be themselves base objects
1067 * for more unresolved deltas, we really want to include the
1068 * smallest number of base objects that would cover as much delta
1069 * as possible by picking the
1070 * trunc deltas first, allowing for other deltas to resolve without
1071 * additional base objects. Since most base objects are to be found
1072 * before deltas depending on them, a good heuristic is to start
1073 * resolving deltas in the same order as their position in the pack.
1074 */
1075 sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1076 for (i = 0; i < nr_deltas; i++) {
1077 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1078 continue;
1079 sorted_by_pos[n++] = &deltas[i];
1080 }
1081 qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1082 for (i = 0; i < n; i++) {
1084 struct delta_entry *d = sorted_by_pos[i];
1085 enum object_type type;
1086 struct base_data *base_obj = alloc_base_data();
1087 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1089 continue;
1090 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1091 if (!base_obj->data)
1092 continue;
1093 if (check_sha1_signature(d->base.sha1, base_obj->data,
1095 base_obj->size, typename(type)))
1096 die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1097 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1098 base_obj->data, base_obj->size, type);
1099 find_unresolved_deltas(base_obj);
1100 display_progress(progress, nr_resolved_deltas);
1101 }
1102 free(sorted_by_pos);
1103}
1104static void final(const char *final_pack_name, const char *curr_pack_name,
1106 const char *final_index_name, const char *curr_index_name,
1107 const char *keep_name, const char *keep_msg,
1108 unsigned char *sha1)
1109{
1110 const char *report = "pack";
1111 char name[PATH_MAX];
1112 int err;
1113 if (!from_stdin) {
1115 close(input_fd);
1116 } else {
1117 fsync_or_die(output_fd, curr_pack_name);
1118 err = close(output_fd);
1119 if (err)
1120 die_errno(_("error while closing pack file"));
1121 }
1122 if (keep_msg) {
1124 int keep_fd, keep_msg_len = strlen(keep_msg);
1125 if (!keep_name)
1127 keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1128 else
1129 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1130 if (keep_fd < 0) {
1132 if (errno != EEXIST)
1133 die_errno(_("cannot write keep file '%s'"),
1134 keep_name);
1135 } else {
1136 if (keep_msg_len > 0) {
1137 write_or_die(keep_fd, keep_msg, keep_msg_len);
1138 write_or_die(keep_fd, "\n", 1);
1139 }
1140 if (close(keep_fd) != 0)
1141 die_errno(_("cannot close written keep file '%s'"),
1142 keep_name);
1143 report = "keep";
1144 }
1145 }
1146 if (final_pack_name != curr_pack_name) {
1148 if (!final_pack_name) {
1149 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1150 get_object_directory(), sha1_to_hex(sha1));
1151 final_pack_name = name;
1152 }
1153 if (move_temp_to_file(curr_pack_name, final_pack_name))
1154 die(_("cannot store pack file"));
1155 } else if (from_stdin)
1156 chmod(final_pack_name, 0444);
1157 if (final_index_name != curr_index_name) {
1159 if (!final_index_name) {
1160 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1161 get_object_directory(), sha1_to_hex(sha1));
1162 final_index_name = name;
1163 }
1164 if (move_temp_to_file(curr_index_name, final_index_name))
1165 die(_("cannot store index file"));
1166 } else
1167 chmod(final_index_name, 0444);
1168 if (!from_stdin) {
1170 printf("%s\n", sha1_to_hex(sha1));
1171 } else {
1172 char buf[48];
1173 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1174 report, sha1_to_hex(sha1));
1175 write_or_die(1, buf, len);
1176 /*
1178 * Let's just mimic git-unpack-objects here and write
1179 * the last part of the input buffer to stdout.
1180 */
1181 while (input_len) {
1182 err = xwrite(1, input_buffer + input_offset, input_len);
1183 if (err <= 0)
1184 break;
1185 input_len -= err;
1186 input_offset += err;
1187 }
1188 }
1189}
1190static int git_index_pack_config(const char *k, const char *v, void *cb)
1192{
1193 struct pack_idx_option *opts = cb;
1194 if (!strcmp(k, "pack.indexversion")) {
1196 opts->version = git_config_int(k, v);
1197 if (opts->version > 2)
1198 die("bad pack.indexversion=%"PRIu32, opts->version);
1199 return 0;
1200 }
1201 if (!strcmp(k, "pack.threads")) {
1202 nr_threads = git_config_int(k, v);
1203 if (nr_threads < 0)
1204 die("invalid number of threads specified (%d)",
1205 nr_threads);
1206#ifdef NO_PTHREADS
1207 if (nr_threads != 1)
1208 warning("no threads support, ignoring %s", k);
1209 nr_threads = 1;
1210#endif
1211 return 0;
1212 }
1213 return git_default_config(k, v, cb);
1214}
1215static int cmp_uint32(const void *a_, const void *b_)
1217{
1218 uint32_t a = *((uint32_t *)a_);
1219 uint32_t b = *((uint32_t *)b_);
1220 return (a < b) ? -1 : (a != b);
1222}
1223static void read_v2_anomalous_offsets(struct packed_git *p,
1225 struct pack_idx_option *opts)
1226{
1227 const uint32_t *idx1, *idx2;
1228 uint32_t i;
1229 /* The address of the 4-byte offset table */
1231 idx1 = (((const uint32_t *)p->index_data)
1232 + 2 /* 8-byte header */
1233 + 256 /* fan out */
1234 + 5 * p->num_objects /* 20-byte SHA-1 table */
1235 + p->num_objects /* CRC32 table */
1236 );
1237 /* The address of the 8-byte offset table */
1239 idx2 = idx1 + p->num_objects;
1240 for (i = 0; i < p->num_objects; i++) {
1242 uint32_t off = ntohl(idx1[i]);
1243 if (!(off & 0x80000000))
1244 continue;
1245 off = off & 0x7fffffff;
1246 if (idx2[off * 2])
1247 continue;
1248 /*
1249 * The real offset is ntohl(idx2[off * 2]) in high 4
1250 * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1251 * octets. But idx2[off * 2] is Zero!!!
1252 */
1253 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1254 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1255 }
1256 if (1 < opts->anomaly_nr)
1258 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1259}
1260static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1262{
1263 struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1264 if (!p)
1266 die(_("Cannot open existing pack file '%s'"), pack_name);
1267 if (open_pack_index(p))
1268 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1269 /* Read the attributes from the existing idx file */
1271 opts->version = p->index_version;
1272 if (opts->version == 2)
1274 read_v2_anomalous_offsets(p, opts);
1275 /*
1277 * Get rid of the idx file as we do not need it anymore.
1278 * NEEDSWORK: extract this bit from free_pack_by_name() in
1279 * sha1_file.c, perhaps? It shouldn't matter very much as we
1280 * know we haven't installed this pack (hence we never have
1281 * read anything from it).
1282 */
1283 close_pack_index(p);
1284 free(p);
1285}
1286static void show_pack_info(int stat_only)
1288{
1289 int i, baseobjects = nr_objects - nr_deltas;
1290 unsigned long *chain_histogram = NULL;
1291 if (deepest_delta)
1293 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1294 for (i = 0; i < nr_objects; i++) {
1296 struct object_entry *obj = &objects[i];
1297 if (is_delta_type(obj->type))
1299 chain_histogram[obj->delta_depth - 1]++;
1300 if (stat_only)
1301 continue;
1302 printf("%s %-6s %lu %lu %"PRIuMAX,
1303 sha1_to_hex(obj->idx.sha1),
1304 typename(obj->real_type), obj->size,
1305 (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1306 (uintmax_t)obj->idx.offset);
1307 if (is_delta_type(obj->type)) {
1308 struct object_entry *bobj = &objects[obj->base_object_no];
1309 printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1310 }
1311 putchar('\n');
1312 }
1313 if (baseobjects)
1315 printf_ln(Q_("non delta: %d object",
1316 "non delta: %d objects",
1317 baseobjects),
1318 baseobjects);
1319 for (i = 0; i < deepest_delta; i++) {
1320 if (!chain_histogram[i])
1321 continue;
1322 printf_ln(Q_("chain length = %d: %lu object",
1323 "chain length = %d: %lu objects",
1324 chain_histogram[i]),
1325 i + 1,
1326 chain_histogram[i]);
1327 }
1328}
1329int cmd_index_pack(int argc, const char **argv, const char *prefix)
1331{
1332 int i, fix_thin_pack = 0, verify = 0, stat_only = 0, stat = 0;
1333 const char *curr_pack, *curr_index;
1334 const char *index_name = NULL, *pack_name = NULL;
1335 const char *keep_name = NULL, *keep_msg = NULL;
1336 char *index_name_buf = NULL, *keep_name_buf = NULL;
1337 struct pack_idx_entry **idx_objects;
1338 struct pack_idx_option opts;
1339 unsigned char pack_sha1[20];
1340 if (argc == 2 && !strcmp(argv[1], "-h"))
1342 usage(index_pack_usage);
1343 read_replace_refs = 0;
1345 reset_pack_idx_option(&opts);
1347 git_config(git_index_pack_config, &opts);
1348 if (prefix && chdir(prefix))
1349 die(_("Cannot come back to cwd"));
1350 for (i = 1; i < argc; i++) {
1352 const char *arg = argv[i];
1353 if (*arg == '-') {
1355 if (!strcmp(arg, "--stdin")) {
1356 from_stdin = 1;
1357 } else if (!strcmp(arg, "--fix-thin")) {
1358 fix_thin_pack = 1;
1359 } else if (!strcmp(arg, "--strict")) {
1360 strict = 1;
1361 } else if (!strcmp(arg, "--verify")) {
1362 verify = 1;
1363 } else if (!strcmp(arg, "--verify-stat")) {
1364 verify = 1;
1365 stat = 1;
1366 } else if (!strcmp(arg, "--verify-stat-only")) {
1367 verify = 1;
1368 stat = 1;
1369 stat_only = 1;
1370 } else if (!strcmp(arg, "--keep")) {
1371 keep_msg = "";
1372 } else if (!prefixcmp(arg, "--keep=")) {
1373 keep_msg = arg + 7;
1374 } else if (!prefixcmp(arg, "--threads=")) {
1375 char *end;
1376 nr_threads = strtoul(arg+10, &end, 0);
1377 if (!arg[10] || *end || nr_threads < 0)
1378 usage(index_pack_usage);
1379#ifdef NO_PTHREADS
1380 if (nr_threads != 1)
1381 warning("no threads support, "
1382 "ignoring %s", arg);
1383 nr_threads = 1;
1384#endif
1385 } else if (!prefixcmp(arg, "--pack_header=")) {
1386 struct pack_header *hdr;
1387 char *c;
1388 hdr = (struct pack_header *)input_buffer;
1390 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1391 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1392 if (*c != ',')
1393 die(_("bad %s"), arg);
1394 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1395 if (*c)
1396 die(_("bad %s"), arg);
1397 input_len = sizeof(*hdr);
1398 } else if (!strcmp(arg, "-v")) {
1399 verbose = 1;
1400 } else if (!strcmp(arg, "-o")) {
1401 if (index_name || (i+1) >= argc)
1402 usage(index_pack_usage);
1403 index_name = argv[++i];
1404 } else if (!prefixcmp(arg, "--index-version=")) {
1405 char *c;
1406 opts.version = strtoul(arg + 16, &c, 10);
1407 if (opts.version > 2)
1408 die(_("bad %s"), arg);
1409 if (*c == ',')
1410 opts.off32_limit = strtoul(c+1, &c, 0);
1411 if (*c || opts.off32_limit & 0x80000000)
1412 die(_("bad %s"), arg);
1413 } else
1414 usage(index_pack_usage);
1415 continue;
1416 }
1417 if (pack_name)
1419 usage(index_pack_usage);
1420 pack_name = arg;
1421 }
1422 if (!pack_name && !from_stdin)
1424 usage(index_pack_usage);
1425 if (fix_thin_pack && !from_stdin)
1426 die(_("--fix-thin cannot be used without --stdin"));
1427 if (!index_name && pack_name) {
1428 int len = strlen(pack_name);
1429 if (!has_extension(pack_name, ".pack"))
1430 die(_("packfile name '%s' does not end with '.pack'"),
1431 pack_name);
1432 index_name_buf = xmalloc(len);
1433 memcpy(index_name_buf, pack_name, len - 5);
1434 strcpy(index_name_buf + len - 5, ".idx");
1435 index_name = index_name_buf;
1436 }
1437 if (keep_msg && !keep_name && pack_name) {
1438 int len = strlen(pack_name);
1439 if (!has_extension(pack_name, ".pack"))
1440 die(_("packfile name '%s' does not end with '.pack'"),
1441 pack_name);
1442 keep_name_buf = xmalloc(len);
1443 memcpy(keep_name_buf, pack_name, len - 5);
1444 strcpy(keep_name_buf + len - 5, ".keep");
1445 keep_name = keep_name_buf;
1446 }
1447 if (verify) {
1448 if (!index_name)
1449 die(_("--verify with no packfile name given"));
1450 read_idx_option(&opts, index_name);
1451 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1452 }
1453 if (strict)
1454 opts.flags |= WRITE_IDX_STRICT;
1455#ifndef NO_PTHREADS
1457 if (!nr_threads) {
1458 nr_threads = online_cpus();
1459 /* An experiment showed that more threads does not mean faster */
1460 if (nr_threads > 3)
1461 nr_threads = 3;
1462 }
1463#endif
1464 curr_pack = open_pack_file(pack_name);
1466 parse_pack_header();
1467 objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1468 deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1469 parse_pack_objects(pack_sha1);
1470 resolve_deltas();
1471 conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1472 free(deltas);
1473 if (strict)
1474 check_objects();
1475 if (stat)
1477 show_pack_info(stat_only);
1478 idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1480 for (i = 0; i < nr_objects; i++)
1481 idx_objects[i] = &objects[i].idx;
1482 curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1483 free(idx_objects);
1484 if (!verify)
1486 final(pack_name, curr_pack,
1487 index_name, curr_index,
1488 keep_name, keep_msg,
1489 pack_sha1);
1490 else
1491 close(input_fd);
1492 free(objects);
1493 free(index_name_buf);
1494 free(keep_name_buf);
1495 if (pack_name == NULL)
1496 free((void *) curr_pack);
1497 if (index_name == NULL)
1498 free((void *) curr_index);
1499 return 0;
1501}