38ff03a5cdab4eb6107b471daa156dfd57753be7
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"
12static const char index_pack_usage[] =
14"git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
15struct object_entry {
17 struct pack_idx_entry idx;
18 unsigned long size;
19 unsigned int hdr_size;
20 enum object_type type;
21 enum object_type real_type;
22 unsigned delta_depth;
23 int base_object_no;
24};
25union delta_base {
27 unsigned char sha1[20];
28 off_t offset;
29};
30struct 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/*
42 * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
43 * to memcmp() only the first 20 bytes.
44 */
45#define UNION_BASE_SZ 20
46#define FLAG_LINK (1u<<20)
48#define FLAG_CHECKED (1u<<21)
49struct delta_entry {
51 union delta_base base;
52 int obj_no;
53};
54static struct object_entry *objects;
56static struct delta_entry *deltas;
57static struct base_data *base_cache;
58static size_t base_cache_used;
59static int nr_objects;
60static int nr_deltas;
61static int nr_resolved_deltas;
62static int from_stdin;
64static int strict;
65static int verbose;
66static struct progress *progress;
68/* We always read in 4kB chunks. */
70static unsigned char input_buffer[4096];
71static unsigned int input_offset, input_len;
72static off_t consumed_bytes;
73static unsigned deepest_delta;
74static git_SHA_CTX input_ctx;
75static uint32_t input_crc32;
76static int input_fd, output_fd, pack_fd;
77static int mark_link(struct object *obj, int type, void *data)
79{
80 if (!obj)
81 return -1;
82 if (type != OBJ_ANY && obj->type != type)
84 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
85 obj->flags |= FLAG_LINK;
87 return 0;
88}
89/* The content of each linked object must have been checked
91 or it must be already present in the object database */
92static void check_object(struct object *obj)
93{
94 if (!obj)
95 return;
96 if (!(obj->flags & FLAG_LINK))
98 return;
99 if (!(obj->flags & FLAG_CHECKED)) {
101 unsigned long size;
102 int type = sha1_object_info(obj->sha1, &size);
103 if (type != obj->type || type <= 0)
104 die("object of unexpected type");
105 obj->flags |= FLAG_CHECKED;
106 return;
107 }
108}
109static void check_objects(void)
111{
112 unsigned i, max;
113 max = get_max_object_index();
115 for (i = 0; i < max; i++)
116 check_object(get_indexed_object(i));
117}
118/* Discard current buffer used content. */
121static void flush(void)
122{
123 if (input_offset) {
124 if (output_fd >= 0)
125 write_or_die(output_fd, input_buffer, input_offset);
126 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
127 memmove(input_buffer, input_buffer + input_offset, input_len);
128 input_offset = 0;
129 }
130}
131/*
133 * Make sure at least "min" bytes are available in the buffer, and
134 * return the pointer to the buffer.
135 */
136static void *fill(int min)
137{
138 if (min <= input_len)
139 return input_buffer + input_offset;
140 if (min > sizeof(input_buffer))
141 die("cannot fill %d bytes", min);
142 flush();
143 do {
144 ssize_t ret = xread(input_fd, input_buffer + input_len,
145 sizeof(input_buffer) - input_len);
146 if (ret <= 0) {
147 if (!ret)
148 die("early EOF");
149 die_errno("read error on input");
150 }
151 input_len += ret;
152 if (from_stdin)
153 display_throughput(progress, consumed_bytes + input_len);
154 } while (input_len < min);
155 return input_buffer;
156}
157static void use(int bytes)
159{
160 if (bytes > input_len)
161 die("used more bytes than were available");
162 input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
163 input_len -= bytes;
164 input_offset += bytes;
165 /* make sure off_t is sufficiently large not to wrap */
167 if (signed_add_overflows(consumed_bytes, bytes))
168 die("pack too large for current definition of off_t");
169 consumed_bytes += bytes;
170}
171static const char *open_pack_file(const char *pack_name)
173{
174 if (from_stdin) {
175 input_fd = 0;
176 if (!pack_name) {
177 static char tmp_file[PATH_MAX];
178 output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
179 "pack/tmp_pack_XXXXXX");
180 pack_name = xstrdup(tmp_file);
181 } else
182 output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
183 if (output_fd < 0)
184 die_errno("unable to create '%s'", pack_name);
185 pack_fd = output_fd;
186 } else {
187 input_fd = open(pack_name, O_RDONLY);
188 if (input_fd < 0)
189 die_errno("cannot open packfile '%s'", pack_name);
190 output_fd = -1;
191 pack_fd = input_fd;
192 }
193 git_SHA1_Init(&input_ctx);
194 return pack_name;
195}
196static void parse_pack_header(void)
198{
199 struct pack_header *hdr = fill(sizeof(struct pack_header));
200 /* Header consistency check */
202 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
203 die("pack signature mismatch");
204 if (!pack_version_ok(hdr->hdr_version))
205 die("pack version %"PRIu32" unsupported",
206 ntohl(hdr->hdr_version));
207 nr_objects = ntohl(hdr->hdr_entries);
209 use(sizeof(struct pack_header));
210}
211static NORETURN void bad_object(unsigned long offset, const char *format,
213 ...) __attribute__((format (printf, 2, 3)));
214static NORETURN void bad_object(unsigned long offset, const char *format, ...)
216{
217 va_list params;
218 char buf[1024];
219 va_start(params, format);
221 vsnprintf(buf, sizeof(buf), format, params);
222 va_end(params);
223 die("pack has bad object at offset %lu: %s", offset, buf);
224}
225static struct base_data *alloc_base_data(void)
227{
228 struct base_data *base = xmalloc(sizeof(struct base_data));
229 memset(base, 0, sizeof(*base));
230 base->ref_last = -1;
231 base->ofs_last = -1;
232 return base;
233}
234static void free_base_data(struct base_data *c)
236{
237 if (c->data) {
238 free(c->data);
239 c->data = NULL;
240 base_cache_used -= c->size;
241 }
242}
243static void prune_base_data(struct base_data *retain)
245{
246 struct base_data *b;
247 for (b = base_cache;
248 base_cache_used > delta_base_cache_limit && b;
249 b = b->child) {
250 if (b->data && b != retain)
251 free_base_data(b);
252 }
253}
254static void link_base_data(struct base_data *base, struct base_data *c)
256{
257 if (base)
258 base->child = c;
259 else
260 base_cache = c;
261 c->base = base;
263 c->child = NULL;
264 if (c->data)
265 base_cache_used += c->size;
266 prune_base_data(c);
267}
268static void unlink_base_data(struct base_data *c)
270{
271 struct base_data *base = c->base;
272 if (base)
273 base->child = NULL;
274 else
275 base_cache = NULL;
276 free_base_data(c);
277}
278static void *unpack_entry_data(unsigned long offset, unsigned long size)
280{
281 int status;
282 git_zstream stream;
283 void *buf = xmalloc(size);
284 memset(&stream, 0, sizeof(stream));
286 git_inflate_init(&stream);
287 stream.next_out = buf;
288 stream.avail_out = size;
289 do {
291 stream.next_in = fill(1);
292 stream.avail_in = input_len;
293 status = git_inflate(&stream, 0);
294 use(input_len - stream.avail_in);
295 } while (status == Z_OK);
296 if (stream.total_out != size || status != Z_STREAM_END)
297 bad_object(offset, "inflate returned %d", status);
298 git_inflate_end(&stream);
299 return buf;
300}
301static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
303{
304 unsigned char *p;
305 unsigned long size, c;
306 off_t base_offset;
307 unsigned shift;
308 void *data;
309 obj->idx.offset = consumed_bytes;
311 input_crc32 = crc32(0, NULL, 0);
312 p = fill(1);
314 c = *p;
315 use(1);
316 obj->type = (c >> 4) & 7;
317 size = (c & 15);
318 shift = 4;
319 while (c & 0x80) {
320 p = fill(1);
321 c = *p;
322 use(1);
323 size += (c & 0x7f) << shift;
324 shift += 7;
325 }
326 obj->size = size;
327 switch (obj->type) {
329 case OBJ_REF_DELTA:
330 hashcpy(delta_base->sha1, fill(20));
331 use(20);
332 break;
333 case OBJ_OFS_DELTA:
334 memset(delta_base, 0, sizeof(*delta_base));
335 p = fill(1);
336 c = *p;
337 use(1);
338 base_offset = c & 127;
339 while (c & 128) {
340 base_offset += 1;
341 if (!base_offset || MSB(base_offset, 7))
342 bad_object(obj->idx.offset, "offset value overflow for delta base object");
343 p = fill(1);
344 c = *p;
345 use(1);
346 base_offset = (base_offset << 7) + (c & 127);
347 }
348 delta_base->offset = obj->idx.offset - base_offset;
349 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
350 bad_object(obj->idx.offset, "delta base offset is out of bound");
351 break;
352 case OBJ_COMMIT:
353 case OBJ_TREE:
354 case OBJ_BLOB:
355 case OBJ_TAG:
356 break;
357 default:
358 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
359 }
360 obj->hdr_size = consumed_bytes - obj->idx.offset;
361 data = unpack_entry_data(obj->idx.offset, obj->size);
363 obj->idx.crc32 = input_crc32;
364 return data;
365}
366static void *get_data_from_pack(struct object_entry *obj)
368{
369 off_t from = obj[0].idx.offset + obj[0].hdr_size;
370 unsigned long len = obj[1].idx.offset - from;
371 unsigned char *data, *inbuf;
372 git_zstream stream;
373 int status;
374 data = xmalloc(obj->size);
376 inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
377 memset(&stream, 0, sizeof(stream));
379 git_inflate_init(&stream);
380 stream.next_out = data;
381 stream.avail_out = obj->size;
382 do {
384 ssize_t n = (len < 64*1024) ? len : 64*1024;
385 n = pread(pack_fd, inbuf, n, from);
386 if (n < 0)
387 die_errno("cannot pread pack file");
388 if (!n)
389 die("premature end of pack file, %lu bytes missing", len);
390 from += n;
391 len -= n;
392 stream.next_in = inbuf;
393 stream.avail_in = n;
394 status = git_inflate(&stream, 0);
395 } while (len && status == Z_OK && !stream.avail_in);
396 /* This has been inflated OK when first encountered, so... */
398 if (status != Z_STREAM_END || stream.total_out != obj->size)
399 die("serious inflate inconsistency");
400 git_inflate_end(&stream);
402 free(inbuf);
403 return data;
404}
405static int compare_delta_bases(const union delta_base *base1,
407 const union delta_base *base2,
408 enum object_type type1,
409 enum object_type type2)
410{
411 int cmp = type1 - type2;
412 if (cmp)
413 return cmp;
414 return memcmp(base1, base2, UNION_BASE_SZ);
415}
416static int find_delta(const union delta_base *base, enum object_type type)
418{
419 int first = 0, last = nr_deltas;
420 while (first < last) {
422 int next = (first + last) / 2;
423 struct delta_entry *delta = &deltas[next];
424 int cmp;
425 cmp = compare_delta_bases(base, &delta->base,
427 type, objects[delta->obj_no].type);
428 if (!cmp)
429 return next;
430 if (cmp < 0) {
431 last = next;
432 continue;
433 }
434 first = next+1;
435 }
436 return -first-1;
437}
438static void find_delta_children(const union delta_base *base,
440 int *first_index, int *last_index,
441 enum object_type type)
442{
443 int first = find_delta(base, type);
444 int last = first;
445 int end = nr_deltas - 1;
446 if (first < 0) {
448 *first_index = 0;
449 *last_index = -1;
450 return;
451 }
452 while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
453 --first;
454 while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
455 ++last;
456 *first_index = first;
457 *last_index = last;
458}
459static void sha1_object(const void *data, unsigned long size,
461 enum object_type type, unsigned char *sha1)
462{
463 hash_sha1_file(data, size, typename(type), sha1);
464 if (has_sha1_file(sha1)) {
465 void *has_data;
466 enum object_type has_type;
467 unsigned long has_size;
468 has_data = read_sha1_file(sha1, &has_type, &has_size);
469 if (!has_data)
470 die("cannot read existing object %s", sha1_to_hex(sha1));
471 if (size != has_size || type != has_type ||
472 memcmp(data, has_data, size) != 0)
473 die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
474 free(has_data);
475 }
476 if (strict) {
477 if (type == OBJ_BLOB) {
478 struct blob *blob = lookup_blob(sha1);
479 if (blob)
480 blob->object.flags |= FLAG_CHECKED;
481 else
482 die("invalid blob object %s", sha1_to_hex(sha1));
483 } else {
484 struct object *obj;
485 int eaten;
486 void *buf = (void *) data;
487 /*
489 * we do not need to free the memory here, as the
490 * buf is deleted by the caller.
491 */
492 obj = parse_object_buffer(sha1, type, size, buf, &eaten);
493 if (!obj)
494 die("invalid %s", typename(type));
495 if (fsck_object(obj, 1, fsck_error_function))
496 die("Error in object");
497 if (fsck_walk(obj, mark_link, NULL))
498 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
499 if (obj->type == OBJ_TREE) {
501 struct tree *item = (struct tree *) obj;
502 item->buffer = NULL;
503 }
504 if (obj->type == OBJ_COMMIT) {
505 struct commit *commit = (struct commit *) obj;
506 commit->buffer = NULL;
507 }
508 obj->flags |= FLAG_CHECKED;
509 }
510 }
511}
512static int is_delta_type(enum object_type type)
514{
515 return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
516}
517static void *get_base_data(struct base_data *c)
519{
520 if (!c->data) {
521 struct object_entry *obj = c->obj;
522 if (is_delta_type(obj->type)) {
524 void *base = get_base_data(c->base);
525 void *raw = get_data_from_pack(obj);
526 c->data = patch_delta(
527 base, c->base->size,
528 raw, obj->size,
529 &c->size);
530 free(raw);
531 if (!c->data)
532 bad_object(obj->idx.offset, "failed to apply delta");
533 } else {
534 c->data = get_data_from_pack(obj);
535 c->size = obj->size;
536 }
537 base_cache_used += c->size;
539 prune_base_data(c);
540 }
541 return c->data;
542}
543static void resolve_delta(struct object_entry *delta_obj,
545 struct base_data *base, struct base_data *result)
546{
547 void *base_data, *delta_data;
548 delta_obj->real_type = base->obj->real_type;
550 delta_obj->delta_depth = base->obj->delta_depth + 1;
551 if (deepest_delta < delta_obj->delta_depth)
552 deepest_delta = delta_obj->delta_depth;
553 delta_obj->base_object_no = base->obj - objects;
554 delta_data = get_data_from_pack(delta_obj);
555 base_data = get_base_data(base);
556 result->obj = delta_obj;
557 result->data = patch_delta(base_data, base->size,
558 delta_data, delta_obj->size, &result->size);
559 free(delta_data);
560 if (!result->data)
561 bad_object(delta_obj->idx.offset, "failed to apply delta");
562 sha1_object(result->data, result->size, delta_obj->real_type,
563 delta_obj->idx.sha1);
564 nr_resolved_deltas++;
565}
566static struct base_data *find_unresolved_deltas_1(struct base_data *base,
568 struct base_data *prev_base)
569{
570 if (base->ref_last == -1 && base->ofs_last == -1) {
571 union delta_base base_spec;
572 hashcpy(base_spec.sha1, base->obj->idx.sha1);
574 find_delta_children(&base_spec,
575 &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
576 memset(&base_spec, 0, sizeof(base_spec));
578 base_spec.offset = base->obj->idx.offset;
579 find_delta_children(&base_spec,
580 &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
581 if (base->ref_last == -1 && base->ofs_last == -1) {
583 free(base->data);
584 return NULL;
585 }
586 link_base_data(prev_base, base);
588 }
589 if (base->ref_first <= base->ref_last) {
591 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
592 struct base_data *result = alloc_base_data();
593 assert(child->real_type == OBJ_REF_DELTA);
595 resolve_delta(child, base, result);
596 if (base->ref_first == base->ref_last && base->ofs_last == -1)
597 free_base_data(base);
598 base->ref_first++;
600 return result;
601 }
602 if (base->ofs_first <= base->ofs_last) {
604 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
605 struct base_data *result = alloc_base_data();
606 assert(child->real_type == OBJ_OFS_DELTA);
608 resolve_delta(child, base, result);
609 if (base->ofs_first == base->ofs_last)
610 free_base_data(base);
611 base->ofs_first++;
613 return result;
614 }
615 unlink_base_data(base);
617 return NULL;
618}
619static void find_unresolved_deltas(struct base_data *base)
621{
622 struct base_data *new_base, *prev_base = NULL;
623 for (;;) {
624 new_base = find_unresolved_deltas_1(base, prev_base);
625 if (new_base) {
627 prev_base = base;
628 base = new_base;
629 } else {
630 free(base);
631 base = prev_base;
632 if (!base)
633 return;
634 prev_base = base->base;
635 }
636 }
637}
638static int compare_delta_entry(const void *a, const void *b)
640{
641 const struct delta_entry *delta_a = a;
642 const struct delta_entry *delta_b = b;
643 /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
645 return compare_delta_bases(&delta_a->base, &delta_b->base,
646 objects[delta_a->obj_no].type,
647 objects[delta_b->obj_no].type);
648}
649/* Parse all objects and return the pack content SHA1 hash */
651static void parse_pack_objects(unsigned char *sha1)
652{
653 int i;
654 struct delta_entry *delta = deltas;
655 struct stat st;
656 /*
658 * First pass:
659 * - find locations of all objects;
660 * - calculate SHA1 of all non-delta objects;
661 * - remember base (SHA1 or offset) for all deltas.
662 */
663 if (verbose)
664 progress = start_progress(
665 from_stdin ? "Receiving objects" : "Indexing objects",
666 nr_objects);
667 for (i = 0; i < nr_objects; i++) {
668 struct object_entry *obj = &objects[i];
669 void *data = unpack_raw_entry(obj, &delta->base);
670 obj->real_type = obj->type;
671 if (is_delta_type(obj->type)) {
672 nr_deltas++;
673 delta->obj_no = i;
674 delta++;
675 } else
676 sha1_object(data, obj->size, obj->type, obj->idx.sha1);
677 free(data);
678 display_progress(progress, i+1);
679 }
680 objects[i].idx.offset = consumed_bytes;
681 stop_progress(&progress);
682 /* Check pack integrity */
684 flush();
685 git_SHA1_Final(sha1, &input_ctx);
686 if (hashcmp(fill(20), sha1))
687 die("pack is corrupted (SHA1 mismatch)");
688 use(20);
689 /* If input_fd is a file, we should have reached its end now. */
691 if (fstat(input_fd, &st))
692 die_errno("cannot fstat packfile");
693 if (S_ISREG(st.st_mode) &&
694 lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
695 die("pack has junk at the end");
696 if (!nr_deltas)
698 return;
699 /* Sort deltas by base SHA1/offset for fast searching */
701 qsort(deltas, nr_deltas, sizeof(struct delta_entry),
702 compare_delta_entry);
703 /*
705 * Second pass:
706 * - for all non-delta objects, look if it is used as a base for
707 * deltas;
708 * - if used as a base, uncompress the object and apply all deltas,
709 * recursively checking if the resulting object is used as a base
710 * for some more deltas.
711 */
712 if (verbose)
713 progress = start_progress("Resolving deltas", nr_deltas);
714 for (i = 0; i < nr_objects; i++) {
715 struct object_entry *obj = &objects[i];
716 struct base_data *base_obj = alloc_base_data();
717 if (is_delta_type(obj->type))
719 continue;
720 base_obj->obj = obj;
721 base_obj->data = NULL;
722 find_unresolved_deltas(base_obj);
723 display_progress(progress, nr_resolved_deltas);
724 }
725}
726static int write_compressed(struct sha1file *f, void *in, unsigned int size)
728{
729 git_zstream stream;
730 int status;
731 unsigned char outbuf[4096];
732 memset(&stream, 0, sizeof(stream));
734 git_deflate_init(&stream, zlib_compression_level);
735 stream.next_in = in;
736 stream.avail_in = size;
737 do {
739 stream.next_out = outbuf;
740 stream.avail_out = sizeof(outbuf);
741 status = git_deflate(&stream, Z_FINISH);
742 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
743 } while (status == Z_OK);
744 if (status != Z_STREAM_END)
746 die("unable to deflate appended object (%d)", status);
747 size = stream.total_out;
748 git_deflate_end(&stream);
749 return size;
750}
751static struct object_entry *append_obj_to_pack(struct sha1file *f,
753 const unsigned char *sha1, void *buf,
754 unsigned long size, enum object_type type)
755{
756 struct object_entry *obj = &objects[nr_objects++];
757 unsigned char header[10];
758 unsigned long s = size;
759 int n = 0;
760 unsigned char c = (type << 4) | (s & 15);
761 s >>= 4;
762 while (s) {
763 header[n++] = c | 0x80;
764 c = s & 0x7f;
765 s >>= 7;
766 }
767 header[n++] = c;
768 crc32_begin(f);
769 sha1write(f, header, n);
770 obj[0].size = size;
771 obj[0].hdr_size = n;
772 obj[0].type = type;
773 obj[0].real_type = type;
774 obj[1].idx.offset = obj[0].idx.offset + n;
775 obj[1].idx.offset += write_compressed(f, buf, size);
776 obj[0].idx.crc32 = crc32_end(f);
777 sha1flush(f);
778 hashcpy(obj->idx.sha1, sha1);
779 return obj;
780}
781static int delta_pos_compare(const void *_a, const void *_b)
783{
784 struct delta_entry *a = *(struct delta_entry **)_a;
785 struct delta_entry *b = *(struct delta_entry **)_b;
786 return a->obj_no - b->obj_no;
787}
788static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
790{
791 struct delta_entry **sorted_by_pos;
792 int i, n = 0;
793 /*
795 * Since many unresolved deltas may well be themselves base objects
796 * for more unresolved deltas, we really want to include the
797 * smallest number of base objects that would cover as much delta
798 * as possible by picking the
799 * trunc deltas first, allowing for other deltas to resolve without
800 * additional base objects. Since most base objects are to be found
801 * before deltas depending on them, a good heuristic is to start
802 * resolving deltas in the same order as their position in the pack.
803 */
804 sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
805 for (i = 0; i < nr_deltas; i++) {
806 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
807 continue;
808 sorted_by_pos[n++] = &deltas[i];
809 }
810 qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
811 for (i = 0; i < n; i++) {
813 struct delta_entry *d = sorted_by_pos[i];
814 enum object_type type;
815 struct base_data *base_obj = alloc_base_data();
816 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
818 continue;
819 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
820 if (!base_obj->data)
821 continue;
822 if (check_sha1_signature(d->base.sha1, base_obj->data,
824 base_obj->size, typename(type)))
825 die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
826 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
827 base_obj->data, base_obj->size, type);
828 find_unresolved_deltas(base_obj);
829 display_progress(progress, nr_resolved_deltas);
830 }
831 free(sorted_by_pos);
832}
833static void final(const char *final_pack_name, const char *curr_pack_name,
835 const char *final_index_name, const char *curr_index_name,
836 const char *keep_name, const char *keep_msg,
837 unsigned char *sha1)
838{
839 const char *report = "pack";
840 char name[PATH_MAX];
841 int err;
842 if (!from_stdin) {
844 close(input_fd);
845 } else {
846 fsync_or_die(output_fd, curr_pack_name);
847 err = close(output_fd);
848 if (err)
849 die_errno("error while closing pack file");
850 }
851 if (keep_msg) {
853 int keep_fd, keep_msg_len = strlen(keep_msg);
854 if (!keep_name)
856 keep_fd = odb_pack_keep(name, sizeof(name), sha1);
857 else
858 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
859 if (keep_fd < 0) {
861 if (errno != EEXIST)
862 die_errno("cannot write keep file '%s'",
863 keep_name);
864 } else {
865 if (keep_msg_len > 0) {
866 write_or_die(keep_fd, keep_msg, keep_msg_len);
867 write_or_die(keep_fd, "\n", 1);
868 }
869 if (close(keep_fd) != 0)
870 die_errno("cannot close written keep file '%s'",
871 keep_name);
872 report = "keep";
873 }
874 }
875 if (final_pack_name != curr_pack_name) {
877 if (!final_pack_name) {
878 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
879 get_object_directory(), sha1_to_hex(sha1));
880 final_pack_name = name;
881 }
882 if (move_temp_to_file(curr_pack_name, final_pack_name))
883 die("cannot store pack file");
884 } else if (from_stdin)
885 chmod(final_pack_name, 0444);
886 if (final_index_name != curr_index_name) {
888 if (!final_index_name) {
889 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
890 get_object_directory(), sha1_to_hex(sha1));
891 final_index_name = name;
892 }
893 if (move_temp_to_file(curr_index_name, final_index_name))
894 die("cannot store index file");
895 } else
896 chmod(final_index_name, 0444);
897 if (!from_stdin) {
899 printf("%s\n", sha1_to_hex(sha1));
900 } else {
901 char buf[48];
902 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
903 report, sha1_to_hex(sha1));
904 write_or_die(1, buf, len);
905 /*
907 * Let's just mimic git-unpack-objects here and write
908 * the last part of the input buffer to stdout.
909 */
910 while (input_len) {
911 err = xwrite(1, input_buffer + input_offset, input_len);
912 if (err <= 0)
913 break;
914 input_len -= err;
915 input_offset += err;
916 }
917 }
918}
919static int git_index_pack_config(const char *k, const char *v, void *cb)
921{
922 struct pack_idx_option *opts = cb;
923 if (!strcmp(k, "pack.indexversion")) {
925 opts->version = git_config_int(k, v);
926 if (opts->version > 2)
927 die("bad pack.indexversion=%"PRIu32, opts->version);
928 return 0;
929 }
930 return git_default_config(k, v, cb);
931}
932static int cmp_uint32(const void *a_, const void *b_)
934{
935 uint32_t a = *((uint32_t *)a_);
936 uint32_t b = *((uint32_t *)b_);
937 return (a < b) ? -1 : (a != b);
939}
940static void read_v2_anomalous_offsets(struct packed_git *p,
942 struct pack_idx_option *opts)
943{
944 const uint32_t *idx1, *idx2;
945 uint32_t i;
946 /* The address of the 4-byte offset table */
948 idx1 = (((const uint32_t *)p->index_data)
949 + 2 /* 8-byte header */
950 + 256 /* fan out */
951 + 5 * p->num_objects /* 20-byte SHA-1 table */
952 + p->num_objects /* CRC32 table */
953 );
954 /* The address of the 8-byte offset table */
956 idx2 = idx1 + p->num_objects;
957 for (i = 0; i < p->num_objects; i++) {
959 uint32_t off = ntohl(idx1[i]);
960 if (!(off & 0x80000000))
961 continue;
962 off = off & 0x7fffffff;
963 if (idx2[off * 2])
964 continue;
965 /*
966 * The real offset is ntohl(idx2[off * 2]) in high 4
967 * octets, and ntohl(idx2[off * 2 + 1]) in low 4
968 * octets. But idx2[off * 2] is Zero!!!
969 */
970 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
971 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
972 }
973 if (1 < opts->anomaly_nr)
975 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
976}
977static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
979{
980 struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
981 if (!p)
983 die("Cannot open existing pack file '%s'", pack_name);
984 if (open_pack_index(p))
985 die("Cannot open existing pack idx file for '%s'", pack_name);
986 /* Read the attributes from the existing idx file */
988 opts->version = p->index_version;
989 if (opts->version == 2)
991 read_v2_anomalous_offsets(p, opts);
992 /*
994 * Get rid of the idx file as we do not need it anymore.
995 * NEEDSWORK: extract this bit from free_pack_by_name() in
996 * sha1_file.c, perhaps? It shouldn't matter very much as we
997 * know we haven't installed this pack (hence we never have
998 * read anything from it).
999 */
1000 close_pack_index(p);
1001 free(p);
1002}
1003static void show_pack_info(int stat_only)
1005{
1006 int i, baseobjects = nr_objects - nr_deltas;
1007 unsigned long *chain_histogram = NULL;
1008 if (deepest_delta)
1010 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1011 for (i = 0; i < nr_objects; i++) {
1013 struct object_entry *obj = &objects[i];
1014 if (is_delta_type(obj->type))
1016 chain_histogram[obj->delta_depth - 1]++;
1017 if (stat_only)
1018 continue;
1019 printf("%s %-6s %lu %lu %"PRIuMAX,
1020 sha1_to_hex(obj->idx.sha1),
1021 typename(obj->real_type), obj->size,
1022 (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1023 (uintmax_t)obj->idx.offset);
1024 if (is_delta_type(obj->type)) {
1025 struct object_entry *bobj = &objects[obj->base_object_no];
1026 printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1027 }
1028 putchar('\n');
1029 }
1030 if (baseobjects)
1032 printf("non delta: %d object%s\n",
1033 baseobjects, baseobjects > 1 ? "s" : "");
1034 for (i = 0; i < deepest_delta; i++) {
1035 if (!chain_histogram[i])
1036 continue;
1037 printf("chain length = %d: %lu object%s\n",
1038 i + 1,
1039 chain_histogram[i],
1040 chain_histogram[i] > 1 ? "s" : "");
1041 }
1042}
1043int cmd_index_pack(int argc, const char **argv, const char *prefix)
1045{
1046 int i, fix_thin_pack = 0, verify = 0, stat_only = 0, stat = 0;
1047 const char *curr_pack, *curr_index;
1048 const char *index_name = NULL, *pack_name = NULL;
1049 const char *keep_name = NULL, *keep_msg = NULL;
1050 char *index_name_buf = NULL, *keep_name_buf = NULL;
1051 struct pack_idx_entry **idx_objects;
1052 struct pack_idx_option opts;
1053 unsigned char pack_sha1[20];
1054 if (argc == 2 && !strcmp(argv[1], "-h"))
1056 usage(index_pack_usage);
1057 read_replace_refs = 0;
1059 reset_pack_idx_option(&opts);
1061 git_config(git_index_pack_config, &opts);
1062 if (prefix && chdir(prefix))
1063 die("Cannot come back to cwd");
1064 for (i = 1; i < argc; i++) {
1066 const char *arg = argv[i];
1067 if (*arg == '-') {
1069 if (!strcmp(arg, "--stdin")) {
1070 from_stdin = 1;
1071 } else if (!strcmp(arg, "--fix-thin")) {
1072 fix_thin_pack = 1;
1073 } else if (!strcmp(arg, "--strict")) {
1074 strict = 1;
1075 } else if (!strcmp(arg, "--verify")) {
1076 verify = 1;
1077 } else if (!strcmp(arg, "--verify-stat")) {
1078 verify = 1;
1079 stat = 1;
1080 } else if (!strcmp(arg, "--verify-stat-only")) {
1081 verify = 1;
1082 stat = 1;
1083 stat_only = 1;
1084 } else if (!strcmp(arg, "--keep")) {
1085 keep_msg = "";
1086 } else if (!prefixcmp(arg, "--keep=")) {
1087 keep_msg = arg + 7;
1088 } else if (!prefixcmp(arg, "--pack_header=")) {
1089 struct pack_header *hdr;
1090 char *c;
1091 hdr = (struct pack_header *)input_buffer;
1093 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1094 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1095 if (*c != ',')
1096 die("bad %s", arg);
1097 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1098 if (*c)
1099 die("bad %s", arg);
1100 input_len = sizeof(*hdr);
1101 } else if (!strcmp(arg, "-v")) {
1102 verbose = 1;
1103 } else if (!strcmp(arg, "-o")) {
1104 if (index_name || (i+1) >= argc)
1105 usage(index_pack_usage);
1106 index_name = argv[++i];
1107 } else if (!prefixcmp(arg, "--index-version=")) {
1108 char *c;
1109 opts.version = strtoul(arg + 16, &c, 10);
1110 if (opts.version > 2)
1111 die("bad %s", arg);
1112 if (*c == ',')
1113 opts.off32_limit = strtoul(c+1, &c, 0);
1114 if (*c || opts.off32_limit & 0x80000000)
1115 die("bad %s", arg);
1116 } else
1117 usage(index_pack_usage);
1118 continue;
1119 }
1120 if (pack_name)
1122 usage(index_pack_usage);
1123 pack_name = arg;
1124 }
1125 if (!pack_name && !from_stdin)
1127 usage(index_pack_usage);
1128 if (fix_thin_pack && !from_stdin)
1129 die("--fix-thin cannot be used without --stdin");
1130 if (!index_name && pack_name) {
1131 int len = strlen(pack_name);
1132 if (!has_extension(pack_name, ".pack"))
1133 die("packfile name '%s' does not end with '.pack'",
1134 pack_name);
1135 index_name_buf = xmalloc(len);
1136 memcpy(index_name_buf, pack_name, len - 5);
1137 strcpy(index_name_buf + len - 5, ".idx");
1138 index_name = index_name_buf;
1139 }
1140 if (keep_msg && !keep_name && pack_name) {
1141 int len = strlen(pack_name);
1142 if (!has_extension(pack_name, ".pack"))
1143 die("packfile name '%s' does not end with '.pack'",
1144 pack_name);
1145 keep_name_buf = xmalloc(len);
1146 memcpy(keep_name_buf, pack_name, len - 5);
1147 strcpy(keep_name_buf + len - 5, ".keep");
1148 keep_name = keep_name_buf;
1149 }
1150 if (verify) {
1151 if (!index_name)
1152 die("--verify with no packfile name given");
1153 read_idx_option(&opts, index_name);
1154 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1155 }
1156 if (strict)
1157 opts.flags |= WRITE_IDX_STRICT;
1158 curr_pack = open_pack_file(pack_name);
1160 parse_pack_header();
1161 objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1162 deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1163 parse_pack_objects(pack_sha1);
1164 if (nr_deltas == nr_resolved_deltas) {
1165 stop_progress(&progress);
1166 /* Flush remaining pack final 20-byte SHA1. */
1167 flush();
1168 } else {
1169 if (fix_thin_pack) {
1170 struct sha1file *f;
1171 unsigned char read_sha1[20], tail_sha1[20];
1172 char msg[48];
1173 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1174 int nr_objects_initial = nr_objects;
1175 if (nr_unresolved <= 0)
1176 die("confusion beyond insanity");
1177 objects = xrealloc(objects,
1178 (nr_objects + nr_unresolved + 1)
1179 * sizeof(*objects));
1180 f = sha1fd(output_fd, curr_pack);
1181 fix_unresolved_deltas(f, nr_unresolved);
1182 sprintf(msg, "completed with %d local objects",
1183 nr_objects - nr_objects_initial);
1184 stop_progress_msg(&progress, msg);
1185 sha1close(f, tail_sha1, 0);
1186 hashcpy(read_sha1, pack_sha1);
1187 fixup_pack_header_footer(output_fd, pack_sha1,
1188 curr_pack, nr_objects,
1189 read_sha1, consumed_bytes-20);
1190 if (hashcmp(read_sha1, tail_sha1) != 0)
1191 die("Unexpected tail checksum for %s "
1192 "(disk corruption?)", curr_pack);
1193 }
1194 if (nr_deltas != nr_resolved_deltas)
1195 die("pack has %d unresolved deltas",
1196 nr_deltas - nr_resolved_deltas);
1197 }
1198 free(deltas);
1199 if (strict)
1200 check_objects();
1201 if (stat)
1203 show_pack_info(stat_only);
1204 idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1206 for (i = 0; i < nr_objects; i++)
1207 idx_objects[i] = &objects[i].idx;
1208 curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1209 free(idx_objects);
1210 if (!verify)
1212 final(pack_name, curr_pack,
1213 index_name, curr_index,
1214 keep_name, keep_msg,
1215 pack_sha1);
1216 else
1217 close(input_fd);
1218 free(objects);
1219 free(index_name_buf);
1220 free(keep_name_buf);
1221 if (pack_name == NULL)
1222 free((void *) curr_pack);
1223 if (index_name == NULL)
1224 free((void *) curr_index);
1225 return 0;
1227}