]> git.ipfire.org Git - thirdparty/git.git/blame - builtin/index-pack.c
Git 2.3
[thirdparty/git.git] / builtin / index-pack.c
CommitLineData
2ad9d4cb 1#include "builtin.h"
9cf6d335
SV
2#include "delta.h"
3#include "pack.h"
4#include "csum-file.h"
8e440259
PE
5#include "blob.h"
6#include "commit.h"
7#include "tag.h"
8#include "tree.h"
96a02f8f 9#include "progress.h"
0153be05 10#include "fsck.h"
2fb3f6db 11#include "exec_cmd.h"
4614043c 12#include "streaming.h"
b8a2486f 13#include "thread-utils.h"
9cf6d335
SV
14
15static const char index_pack_usage[] =
e337a04d 16"git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
9cf6d335 17
9cba13ca 18struct object_entry {
aa7e44bf 19 struct pack_idx_entry idx;
2d477051
NP
20 unsigned long size;
21 unsigned int hdr_size;
9cf6d335
SV
22 enum object_type type;
23 enum object_type real_type;
38a45561
JH
24 unsigned delta_depth;
25 int base_object_no;
9cf6d335
SV
26};
27
53dda6ff
NP
28union delta_base {
29 unsigned char sha1[20];
d7dd0223 30 off_t offset;
53dda6ff
NP
31};
32
f41aebd4 33struct base_data {
4a438cab
SP
34 struct base_data *base;
35 struct base_data *child;
03993e13 36 struct object_entry *obj;
f41aebd4
SP
37 void *data;
38 unsigned long size;
2baad220
NTND
39 int ref_first, ref_last;
40 int ofs_first, ofs_last;
f41aebd4
SP
41};
42
b8a2486f
NTND
43struct thread_local {
44#ifndef NO_PTHREADS
45 pthread_t thread;
46#endif
47 struct base_data *base_cache;
48 size_t base_cache_used;
39539495 49 int pack_fd;
b8a2486f
NTND
50};
51
3c552873
NP
52/*
53 * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
54 * to memcmp() only the first 20 bytes.
55 */
56#define UNION_BASE_SZ 20
57
0153be05
MK
58#define FLAG_LINK (1u<<20)
59#define FLAG_CHECKED (1u<<21)
60
9cba13ca 61struct delta_entry {
53dda6ff 62 union delta_base base;
636171cb 63 int obj_no;
9cf6d335
SV
64};
65
9cf6d335
SV
66static struct object_entry *objects;
67static struct delta_entry *deltas;
b8a2486f 68static struct thread_local nothread_data;
9cf6d335
SV
69static int nr_objects;
70static int nr_deltas;
636171cb 71static int nr_resolved_deltas;
b8a2486f 72static int nr_threads;
9cf6d335 73
e42797f5 74static int from_stdin;
0153be05 75static int strict;
c6807a40 76static int do_fsck_object;
3c9af366 77static int verbose;
3aba2fdd 78static int show_stat;
c6807a40 79static int check_self_contained_and_connected;
3c9af366 80
dc6a0757 81static struct progress *progress;
e42797f5 82
2d477051
NP
83/* We always read in 4kB chunks. */
84static unsigned char input_buffer[4096];
d7dd0223
NP
85static unsigned int input_offset, input_len;
86static off_t consumed_bytes;
d1a0ed18 87static unsigned deepest_delta;
9126f009 88static git_SHA_CTX input_ctx;
ee5743ce 89static uint32_t input_crc32;
39539495
NTND
90static int input_fd, output_fd;
91static const char *curr_pack;
2d477051 92
b8a2486f
NTND
93#ifndef NO_PTHREADS
94
95static struct thread_local *thread_data;
96static int nr_dispatched;
97static int threads_active;
98
99static pthread_mutex_t read_mutex;
100#define read_lock() lock_mutex(&read_mutex)
101#define read_unlock() unlock_mutex(&read_mutex)
102
103static pthread_mutex_t counter_mutex;
104#define counter_lock() lock_mutex(&counter_mutex)
105#define counter_unlock() unlock_mutex(&counter_mutex)
106
107static pthread_mutex_t work_mutex;
108#define work_lock() lock_mutex(&work_mutex)
109#define work_unlock() unlock_mutex(&work_mutex)
110
3aba2fdd
NTND
111static pthread_mutex_t deepest_delta_mutex;
112#define deepest_delta_lock() lock_mutex(&deepest_delta_mutex)
113#define deepest_delta_unlock() unlock_mutex(&deepest_delta_mutex)
114
ab791dd1
JK
115static pthread_mutex_t type_cas_mutex;
116#define type_cas_lock() lock_mutex(&type_cas_mutex)
117#define type_cas_unlock() unlock_mutex(&type_cas_mutex)
118
b8a2486f
NTND
119static pthread_key_t key;
120
121static inline void lock_mutex(pthread_mutex_t *mutex)
122{
123 if (threads_active)
124 pthread_mutex_lock(mutex);
125}
126
127static inline void unlock_mutex(pthread_mutex_t *mutex)
128{
129 if (threads_active)
130 pthread_mutex_unlock(mutex);
131}
132
133/*
134 * Mutex and conditional variable can't be statically-initialized on Windows.
135 */
136static void init_thread(void)
137{
39539495 138 int i;
b8a2486f
NTND
139 init_recursive_mutex(&read_mutex);
140 pthread_mutex_init(&counter_mutex, NULL);
141 pthread_mutex_init(&work_mutex, NULL);
ab791dd1 142 pthread_mutex_init(&type_cas_mutex, NULL);
3aba2fdd
NTND
143 if (show_stat)
144 pthread_mutex_init(&deepest_delta_mutex, NULL);
b8a2486f
NTND
145 pthread_key_create(&key, NULL);
146 thread_data = xcalloc(nr_threads, sizeof(*thread_data));
39539495
NTND
147 for (i = 0; i < nr_threads; i++) {
148 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
149 if (thread_data[i].pack_fd == -1)
150 die_errno(_("unable to open %s"), curr_pack);
151 }
152
b8a2486f
NTND
153 threads_active = 1;
154}
155
156static void cleanup_thread(void)
157{
39539495 158 int i;
b8a2486f
NTND
159 if (!threads_active)
160 return;
161 threads_active = 0;
162 pthread_mutex_destroy(&read_mutex);
163 pthread_mutex_destroy(&counter_mutex);
164 pthread_mutex_destroy(&work_mutex);
ab791dd1 165 pthread_mutex_destroy(&type_cas_mutex);
3aba2fdd
NTND
166 if (show_stat)
167 pthread_mutex_destroy(&deepest_delta_mutex);
39539495
NTND
168 for (i = 0; i < nr_threads; i++)
169 close(thread_data[i].pack_fd);
b8a2486f
NTND
170 pthread_key_delete(key);
171 free(thread_data);
172}
173
174#else
175
176#define read_lock()
177#define read_unlock()
178
179#define counter_lock()
180#define counter_unlock()
181
182#define work_lock()
183#define work_unlock()
184
3aba2fdd
NTND
185#define deepest_delta_lock()
186#define deepest_delta_unlock()
187
e0e21283
EB
188#define type_cas_lock()
189#define type_cas_unlock()
190
b8a2486f
NTND
191#endif
192
193
0153be05
MK
194static int mark_link(struct object *obj, int type, void *data)
195{
196 if (!obj)
197 return -1;
198
199 if (type != OBJ_ANY && obj->type != type)
c2b97ecf 200 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
0153be05
MK
201
202 obj->flags |= FLAG_LINK;
203 return 0;
204}
205
206/* The content of each linked object must have been checked
207 or it must be already present in the object database */
c6807a40 208static unsigned check_object(struct object *obj)
0153be05
MK
209{
210 if (!obj)
c6807a40 211 return 0;
0153be05
MK
212
213 if (!(obj->flags & FLAG_LINK))
c6807a40 214 return 0;
0153be05
MK
215
216 if (!(obj->flags & FLAG_CHECKED)) {
217 unsigned long size;
218 int type = sha1_object_info(obj->sha1, &size);
77583e77
JK
219 if (type <= 0)
220 die(_("did not receive expected object %s"),
221 sha1_to_hex(obj->sha1));
222 if (type != obj->type)
223 die(_("object %s: expected type %s, found %s"),
224 sha1_to_hex(obj->sha1),
225 typename(obj->type), typename(type));
0153be05 226 obj->flags |= FLAG_CHECKED;
c6807a40 227 return 1;
0153be05 228 }
c6807a40
NTND
229
230 return 0;
0153be05
MK
231}
232
c6807a40 233static unsigned check_objects(void)
0153be05 234{
c6807a40 235 unsigned i, max, foreign_nr = 0;
0153be05
MK
236
237 max = get_max_object_index();
238 for (i = 0; i < max; i++)
c6807a40
NTND
239 foreign_nr += check_object(get_indexed_object(i));
240 return foreign_nr;
0153be05
MK
241}
242
243
636171cb 244/* Discard current buffer used content. */
a6e8a767 245static void flush(void)
636171cb
NP
246{
247 if (input_offset) {
248 if (output_fd >= 0)
249 write_or_die(output_fd, input_buffer, input_offset);
9126f009 250 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
554a2636 251 memmove(input_buffer, input_buffer + input_offset, input_len);
636171cb
NP
252 input_offset = 0;
253 }
254}
255
2d477051
NP
256/*
257 * Make sure at least "min" bytes are available in the buffer, and
258 * return the pointer to the buffer.
259 */
b89c4e93 260static void *fill(int min)
9cf6d335 261{
2d477051
NP
262 if (min <= input_len)
263 return input_buffer + input_offset;
264 if (min > sizeof(input_buffer))
c2b97ecf
NTND
265 die(Q_("cannot fill %d byte",
266 "cannot fill %d bytes",
267 min),
268 min);
636171cb 269 flush();
2d477051 270 do {
8a912bcb 271 ssize_t ret = xread(input_fd, input_buffer + input_len,
2d477051
NP
272 sizeof(input_buffer) - input_len);
273 if (ret <= 0) {
274 if (!ret)
c2b97ecf
NTND
275 die(_("early EOF"));
276 die_errno(_("read error on input"));
2d477051
NP
277 }
278 input_len += ret;
218558af
NP
279 if (from_stdin)
280 display_throughput(progress, consumed_bytes + input_len);
2d477051
NP
281 } while (input_len < min);
282 return input_buffer;
283}
284
285static void use(int bytes)
286{
287 if (bytes > input_len)
c2b97ecf 288 die(_("used more bytes than were available"));
ee5743ce 289 input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
2d477051
NP
290 input_len -= bytes;
291 input_offset += bytes;
d7dd0223
NP
292
293 /* make sure off_t is sufficiently large not to wrap */
c03c8315 294 if (signed_add_overflows(consumed_bytes, bytes))
c2b97ecf 295 die(_("pack too large for current definition of off_t"));
2d477051
NP
296 consumed_bytes += bytes;
297}
9cf6d335 298
3bb72562 299static const char *open_pack_file(const char *pack_name)
2d477051 300{
e42797f5
NP
301 if (from_stdin) {
302 input_fd = 0;
303 if (!pack_name) {
ab1900a3
ÆAB
304 static char tmp_file[PATH_MAX];
305 output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
6e180cdc 306 "pack/tmp_pack_XXXXXX");
ab1900a3 307 pack_name = xstrdup(tmp_file);
e42797f5
NP
308 } else
309 output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
310 if (output_fd < 0)
c2b97ecf 311 die_errno(_("unable to create '%s'"), pack_name);
39539495 312 nothread_data.pack_fd = output_fd;
e42797f5
NP
313 } else {
314 input_fd = open(pack_name, O_RDONLY);
315 if (input_fd < 0)
c2b97ecf 316 die_errno(_("cannot open packfile '%s'"), pack_name);
e42797f5 317 output_fd = -1;
39539495 318 nothread_data.pack_fd = input_fd;
e42797f5 319 }
9126f009 320 git_SHA1_Init(&input_ctx);
e42797f5 321 return pack_name;
9cf6d335
SV
322}
323
324static void parse_pack_header(void)
325{
2d477051 326 struct pack_header *hdr = fill(sizeof(struct pack_header));
9cf6d335
SV
327
328 /* Header consistency check */
9cf6d335 329 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
c2b97ecf 330 die(_("pack signature mismatch"));
d60fc1c8 331 if (!pack_version_ok(hdr->hdr_version))
f350df42 332 die(_("pack version %"PRIu32" unsupported"),
6e1c2344 333 ntohl(hdr->hdr_version));
9cf6d335
SV
334
335 nr_objects = ntohl(hdr->hdr_entries);
2d477051 336 use(sizeof(struct pack_header));
9cf6d335
SV
337}
338
a4f3131c
EFL
339static NORETURN void bad_object(unsigned long offset, const char *format,
340 ...) __attribute__((format (printf, 2, 3)));
9cf6d335 341
c2e86add 342static NORETURN void bad_object(unsigned long offset, const char *format, ...)
9cf6d335
SV
343{
344 va_list params;
345 char buf[1024];
346
347 va_start(params, format);
348 vsnprintf(buf, sizeof(buf), format, params);
349 va_end(params);
c2b97ecf 350 die(_("pack has bad object at offset %lu: %s"), offset, buf);
9cf6d335
SV
351}
352
b8a2486f
NTND
353static inline struct thread_local *get_thread_data(void)
354{
355#ifndef NO_PTHREADS
356 if (threads_active)
357 return pthread_getspecific(key);
358 assert(!threads_active &&
359 "This should only be reached when all threads are gone");
360#endif
361 return &nothread_data;
362}
363
364#ifndef NO_PTHREADS
365static void set_thread_data(struct thread_local *data)
366{
367 if (threads_active)
368 pthread_setspecific(key, data);
369}
370#endif
371
2baad220
NTND
372static struct base_data *alloc_base_data(void)
373{
51a60f5b 374 struct base_data *base = xcalloc(1, sizeof(struct base_data));
2baad220
NTND
375 base->ref_last = -1;
376 base->ofs_last = -1;
377 return base;
378}
379
6a87ed97
NP
380static void free_base_data(struct base_data *c)
381{
382 if (c->data) {
383 free(c->data);
384 c->data = NULL;
b8a2486f 385 get_thread_data()->base_cache_used -= c->size;
6a87ed97
NP
386 }
387}
388
92392b4a
SP
389static void prune_base_data(struct base_data *retain)
390{
8e24cbae 391 struct base_data *b;
b8a2486f
NTND
392 struct thread_local *data = get_thread_data();
393 for (b = data->base_cache;
394 data->base_cache_used > delta_base_cache_limit && b;
92392b4a 395 b = b->child) {
6a87ed97
NP
396 if (b->data && b != retain)
397 free_base_data(b);
92392b4a
SP
398 }
399}
400
4a438cab
SP
401static void link_base_data(struct base_data *base, struct base_data *c)
402{
403 if (base)
404 base->child = c;
405 else
b8a2486f 406 get_thread_data()->base_cache = c;
4a438cab
SP
407
408 c->base = base;
409 c->child = NULL;
9441b61d 410 if (c->data)
b8a2486f 411 get_thread_data()->base_cache_used += c->size;
92392b4a 412 prune_base_data(c);
4a438cab
SP
413}
414
415static void unlink_base_data(struct base_data *c)
416{
417 struct base_data *base = c->base;
418 if (base)
419 base->child = NULL;
420 else
b8a2486f 421 get_thread_data()->base_cache = NULL;
6a87ed97 422 free_base_data(c);
4a438cab
SP
423}
424
681b07de
NTND
425static int is_delta_type(enum object_type type)
426{
427 return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
428}
429
430static void *unpack_entry_data(unsigned long offset, unsigned long size,
431 enum object_type type, unsigned char *sha1)
9cf6d335 432{
9ec2dde9 433 static char fixed_buf[8192];
7ce4721a 434 int status;
ef49a7a0 435 git_zstream stream;
9ec2dde9 436 void *buf;
681b07de
NTND
437 git_SHA_CTX c;
438 char hdr[32];
439 int hdrlen;
440
441 if (!is_delta_type(type)) {
442 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
443 git_SHA1_Init(&c);
444 git_SHA1_Update(&c, hdr, hdrlen);
445 } else
446 sha1 = NULL;
9ec2dde9
NTND
447 if (type == OBJ_BLOB && size > big_file_threshold)
448 buf = fixed_buf;
449 else
a1e920a0 450 buf = xmallocz(size);
9cf6d335
SV
451
452 memset(&stream, 0, sizeof(stream));
7ce4721a 453 git_inflate_init(&stream);
9cf6d335 454 stream.next_out = buf;
9ec2dde9 455 stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
9cf6d335 456
7ce4721a 457 do {
681b07de 458 unsigned char *last_out = stream.next_out;
2d477051
NP
459 stream.next_in = fill(1);
460 stream.avail_in = input_len;
7ce4721a
NP
461 status = git_inflate(&stream, 0);
462 use(input_len - stream.avail_in);
681b07de
NTND
463 if (sha1)
464 git_SHA1_Update(&c, last_out, stream.next_out - last_out);
9ec2dde9
NTND
465 if (buf == fixed_buf) {
466 stream.next_out = buf;
467 stream.avail_out = sizeof(fixed_buf);
468 }
7ce4721a
NP
469 } while (status == Z_OK);
470 if (stream.total_out != size || status != Z_STREAM_END)
c2b97ecf 471 bad_object(offset, _("inflate returned %d"), status);
39c68542 472 git_inflate_end(&stream);
681b07de
NTND
473 if (sha1)
474 git_SHA1_Final(sha1, &c);
9ec2dde9 475 return buf == fixed_buf ? NULL : buf;
9cf6d335
SV
476}
477
681b07de
NTND
478static void *unpack_raw_entry(struct object_entry *obj,
479 union delta_base *delta_base,
480 unsigned char *sha1)
9cf6d335 481{
48fb7deb
LT
482 unsigned char *p;
483 unsigned long size, c;
d7dd0223 484 off_t base_offset;
9cf6d335 485 unsigned shift;
ee5743ce 486 void *data;
9cf6d335 487
aa7e44bf 488 obj->idx.offset = consumed_bytes;
1e4cd68c 489 input_crc32 = crc32(0, NULL, 0);
2d477051
NP
490
491 p = fill(1);
492 c = *p;
493 use(1);
494 obj->type = (c >> 4) & 7;
9cf6d335
SV
495 size = (c & 15);
496 shift = 4;
497 while (c & 0x80) {
2d477051
NP
498 p = fill(1);
499 c = *p;
500 use(1);
48fb7deb 501 size += (c & 0x7f) << shift;
9cf6d335
SV
502 shift += 7;
503 }
2d477051 504 obj->size = size;
9cf6d335 505
2d477051 506 switch (obj->type) {
eb32d236 507 case OBJ_REF_DELTA:
2d477051
NP
508 hashcpy(delta_base->sha1, fill(20));
509 use(20);
53dda6ff
NP
510 break;
511 case OBJ_OFS_DELTA:
512 memset(delta_base, 0, sizeof(*delta_base));
2d477051
NP
513 p = fill(1);
514 c = *p;
515 use(1);
53dda6ff
NP
516 base_offset = c & 127;
517 while (c & 128) {
518 base_offset += 1;
8723f216 519 if (!base_offset || MSB(base_offset, 7))
c2b97ecf 520 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
2d477051
NP
521 p = fill(1);
522 c = *p;
523 use(1);
53dda6ff
NP
524 base_offset = (base_offset << 7) + (c & 127);
525 }
aa7e44bf 526 delta_base->offset = obj->idx.offset - base_offset;
d8f32556 527 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
c2b97ecf 528 bad_object(obj->idx.offset, _("delta base offset is out of bound"));
53dda6ff 529 break;
9cf6d335
SV
530 case OBJ_COMMIT:
531 case OBJ_TREE:
532 case OBJ_BLOB:
533 case OBJ_TAG:
9cf6d335
SV
534 break;
535 default:
c2b97ecf 536 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
9cf6d335 537 }
aa7e44bf 538 obj->hdr_size = consumed_bytes - obj->idx.offset;
2d477051 539
681b07de 540 data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
aa7e44bf 541 obj->idx.crc32 = input_crc32;
ee5743ce 542 return data;
2d477051
NP
543}
544
8a2e163c
NTND
545static void *unpack_data(struct object_entry *obj,
546 int (*consume)(const unsigned char *, unsigned long, void *),
547 void *cb_data)
2d477051 548{
a91ef6e7 549 off_t from = obj[0].idx.offset + obj[0].hdr_size;
aa7e44bf 550 unsigned long len = obj[1].idx.offset - from;
776ea370 551 unsigned char *data, *inbuf;
ef49a7a0 552 git_zstream stream;
776ea370 553 int status;
9cf6d335 554
a1e920a0 555 data = xmallocz(consume ? 64*1024 : obj->size);
776ea370
NP
556 inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
557
2d477051 558 memset(&stream, 0, sizeof(stream));
776ea370 559 git_inflate_init(&stream);
2d477051 560 stream.next_out = data;
8a2e163c 561 stream.avail_out = consume ? 64*1024 : obj->size;
776ea370
NP
562
563 do {
564 ssize_t n = (len < 64*1024) ? len : 64*1024;
53f52cd9 565 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
776ea370 566 if (n < 0)
c2b97ecf 567 die_errno(_("cannot pread pack file"));
776ea370 568 if (!n)
c2b97ecf
NTND
569 die(Q_("premature end of pack file, %lu byte missing",
570 "premature end of pack file, %lu bytes missing",
571 len),
572 len);
776ea370
NP
573 from += n;
574 len -= n;
575 stream.next_in = inbuf;
576 stream.avail_in = n;
f8b09038
JK
577 if (!consume)
578 status = git_inflate(&stream, 0);
579 else {
580 do {
581 status = git_inflate(&stream, 0);
582 if (consume(data, stream.next_out - data, cb_data)) {
583 free(inbuf);
584 free(data);
585 return NULL;
586 }
587 stream.next_out = data;
588 stream.avail_out = 64*1024;
589 } while (status == Z_OK && stream.avail_in);
8a2e163c 590 }
776ea370
NP
591 } while (len && status == Z_OK && !stream.avail_in);
592
593 /* This has been inflated OK when first encountered, so... */
594 if (status != Z_STREAM_END || stream.total_out != obj->size)
c2b97ecf 595 die(_("serious inflate inconsistency"));
776ea370
NP
596
597 git_inflate_end(&stream);
598 free(inbuf);
8a2e163c
NTND
599 if (consume) {
600 free(data);
601 data = NULL;
602 }
9cf6d335
SV
603 return data;
604}
605
8a2e163c
NTND
606static void *get_data_from_pack(struct object_entry *obj)
607{
608 return unpack_data(obj, NULL, NULL);
609}
610
7218a215
JH
611static int compare_delta_bases(const union delta_base *base1,
612 const union delta_base *base2,
613 enum object_type type1,
614 enum object_type type2)
615{
616 int cmp = type1 - type2;
617 if (cmp)
618 return cmp;
619 return memcmp(base1, base2, UNION_BASE_SZ);
620}
621
622static int find_delta(const union delta_base *base, enum object_type type)
9cf6d335
SV
623{
624 int first = 0, last = nr_deltas;
625
626 while (first < last) {
627 int next = (first + last) / 2;
628 struct delta_entry *delta = &deltas[next];
629 int cmp;
630
7218a215
JH
631 cmp = compare_delta_bases(base, &delta->base,
632 type, objects[delta->obj_no].type);
9cf6d335
SV
633 if (!cmp)
634 return next;
635 if (cmp < 0) {
636 last = next;
637 continue;
638 }
639 first = next+1;
640 }
641 return -first-1;
642}
643
6a87ed97 644static void find_delta_children(const union delta_base *base,
7218a215
JH
645 int *first_index, int *last_index,
646 enum object_type type)
9cf6d335 647{
7218a215 648 int first = find_delta(base, type);
9cf6d335
SV
649 int last = first;
650 int end = nr_deltas - 1;
651
6a87ed97
NP
652 if (first < 0) {
653 *first_index = 0;
654 *last_index = -1;
655 return;
656 }
3c552873 657 while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
9cf6d335 658 --first;
3c552873 659 while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
9cf6d335
SV
660 ++last;
661 *first_index = first;
662 *last_index = last;
9cf6d335
SV
663}
664
4614043c
NTND
665struct compare_data {
666 struct object_entry *entry;
667 struct git_istream *st;
668 unsigned char *buf;
669 unsigned long buf_size;
670};
671
672static int compare_objects(const unsigned char *buf, unsigned long size,
673 void *cb_data)
674{
675 struct compare_data *data = cb_data;
676
677 if (data->buf_size < size) {
678 free(data->buf);
679 data->buf = xmalloc(size);
680 data->buf_size = size;
681 }
682
683 while (size) {
684 ssize_t len = read_istream(data->st, data->buf, size);
685 if (len == 0)
686 die(_("SHA1 COLLISION FOUND WITH %s !"),
687 sha1_to_hex(data->entry->idx.sha1));
688 if (len < 0)
689 die(_("unable to read %s"),
690 sha1_to_hex(data->entry->idx.sha1));
691 if (memcmp(buf, data->buf, len))
692 die(_("SHA1 COLLISION FOUND WITH %s !"),
693 sha1_to_hex(data->entry->idx.sha1));
694 size -= len;
695 buf += len;
696 }
697 return 0;
698}
699
700static int check_collison(struct object_entry *entry)
701{
702 struct compare_data data;
703 enum object_type type;
704 unsigned long size;
705
706 if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
707 return -1;
708
709 memset(&data, 0, sizeof(data));
710 data.entry = entry;
711 data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
712 if (!data.st)
713 return -1;
714 if (size != entry->size || type != entry->type)
715 die(_("SHA1 COLLISION FOUND WITH %s !"),
716 sha1_to_hex(entry->idx.sha1));
717 unpack_data(entry, compare_objects, &data);
718 close_istream(data.st);
719 free(data.buf);
720 return 0;
721}
722
9ec2dde9
NTND
723static void sha1_object(const void *data, struct object_entry *obj_entry,
724 unsigned long size, enum object_type type,
725 const unsigned char *sha1)
9cf6d335 726{
9ec2dde9 727 void *new_data = NULL;
4614043c 728 int collision_test_needed;
9ec2dde9
NTND
729
730 assert(data || obj_entry);
731
b8a2486f 732 read_lock();
4614043c
NTND
733 collision_test_needed = has_sha1_file(sha1);
734 read_unlock();
735
736 if (collision_test_needed && !data) {
737 read_lock();
738 if (!check_collison(obj_entry))
739 collision_test_needed = 0;
740 read_unlock();
741 }
742 if (collision_test_needed) {
8685da42
NP
743 void *has_data;
744 enum object_type has_type;
745 unsigned long has_size;
4614043c
NTND
746 read_lock();
747 has_type = sha1_object_info(sha1, &has_size);
748 if (has_type != type || has_size != size)
749 die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
8685da42 750 has_data = read_sha1_file(sha1, &has_type, &has_size);
b8a2486f 751 read_unlock();
4614043c
NTND
752 if (!data)
753 data = new_data = get_data_from_pack(obj_entry);
8685da42 754 if (!has_data)
c2b97ecf 755 die(_("cannot read existing object %s"), sha1_to_hex(sha1));
8685da42
NP
756 if (size != has_size || type != has_type ||
757 memcmp(data, has_data, size) != 0)
c2b97ecf 758 die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
bbf4b41b 759 free(has_data);
4614043c 760 }
b8a2486f 761
0153be05 762 if (strict) {
b8a2486f 763 read_lock();
0153be05
MK
764 if (type == OBJ_BLOB) {
765 struct blob *blob = lookup_blob(sha1);
766 if (blob)
767 blob->object.flags |= FLAG_CHECKED;
768 else
c2b97ecf 769 die(_("invalid blob object %s"), sha1_to_hex(sha1));
0153be05
MK
770 } else {
771 struct object *obj;
772 int eaten;
773 void *buf = (void *) data;
774
920734b0 775 assert(data && "data can only be NULL for large _blobs_");
9ec2dde9 776
0153be05
MK
777 /*
778 * we do not need to free the memory here, as the
779 * buf is deleted by the caller.
780 */
781 obj = parse_object_buffer(sha1, type, size, buf, &eaten);
782 if (!obj)
c2b97ecf 783 die(_("invalid %s"), typename(type));
c6807a40 784 if (do_fsck_object &&
90a398bb
JS
785 fsck_object(obj, buf, size, 1,
786 fsck_error_function))
c2b97ecf 787 die(_("Error in object"));
2af202be 788 if (fsck_walk(obj, mark_link, NULL))
c2b97ecf 789 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
0153be05
MK
790
791 if (obj->type == OBJ_TREE) {
792 struct tree *item = (struct tree *) obj;
793 item->buffer = NULL;
6e454b9a 794 obj->parsed = 0;
0153be05
MK
795 }
796 if (obj->type == OBJ_COMMIT) {
797 struct commit *commit = (struct commit *) obj;
8597ea3a 798 if (detach_commit_buffer(commit, NULL) != data)
0fb370da 799 die("BUG: parse_object_buffer transmogrified our buffer");
0153be05
MK
800 }
801 obj->flags |= FLAG_CHECKED;
802 }
b8a2486f 803 read_unlock();
0153be05 804 }
9ec2dde9
NTND
805
806 free(new_data);
9cf6d335
SV
807}
808
20e95d0a
NTND
809/*
810 * This function is part of find_unresolved_deltas(). There are two
811 * walkers going in the opposite ways.
812 *
813 * The first one in find_unresolved_deltas() traverses down from
814 * parent node to children, deflating nodes along the way. However,
815 * memory for deflated nodes is limited by delta_base_cache_limit, so
816 * at some point parent node's deflated content may be freed.
817 *
818 * The second walker is this function, which goes from current node up
819 * to top parent if necessary to deflate the node. In normal
820 * situation, its parent node would be already deflated, so it just
821 * needs to apply delta.
822 *
823 * In the worst case scenario, parent node is no longer deflated because
824 * we're running out of delta_base_cache_limit; we need to re-deflate
825 * parents, possibly up to the top base.
826 *
827 * All deflated objects here are subject to be freed if we exceed
828 * delta_base_cache_limit, just like in find_unresolved_deltas(), we
829 * just need to make sure the last node is not freed.
830 */
92392b4a
SP
831static void *get_base_data(struct base_data *c)
832{
833 if (!c->data) {
834 struct object_entry *obj = c->obj;
20e95d0a
NTND
835 struct base_data **delta = NULL;
836 int delta_nr = 0, delta_alloc = 0;
92392b4a 837
20e95d0a
NTND
838 while (is_delta_type(c->obj->type) && !c->data) {
839 ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
840 delta[delta_nr++] = c;
841 c = c->base;
842 }
843 if (!delta_nr) {
844 c->data = get_data_from_pack(obj);
845 c->size = obj->size;
b8a2486f 846 get_thread_data()->base_cache_used += c->size;
20e95d0a
NTND
847 prune_base_data(c);
848 }
849 for (; delta_nr > 0; delta_nr--) {
850 void *base, *raw;
851 c = delta[delta_nr - 1];
852 obj = c->obj;
853 base = get_base_data(c->base);
854 raw = get_data_from_pack(obj);
92392b4a
SP
855 c->data = patch_delta(
856 base, c->base->size,
857 raw, obj->size,
858 &c->size);
859 free(raw);
860 if (!c->data)
c2b97ecf 861 bad_object(obj->idx.offset, _("failed to apply delta"));
b8a2486f 862 get_thread_data()->base_cache_used += c->size;
20e95d0a 863 prune_base_data(c);
9441b61d 864 }
20e95d0a 865 free(delta);
92392b4a
SP
866 }
867 return c->data;
868}
869
f41aebd4 870static void resolve_delta(struct object_entry *delta_obj,
9441b61d 871 struct base_data *base, struct base_data *result)
9cf6d335 872{
ce3f6dc6 873 void *base_data, *delta_data;
9cf6d335 874
3aba2fdd
NTND
875 if (show_stat) {
876 delta_obj->delta_depth = base->obj->delta_depth + 1;
877 deepest_delta_lock();
878 if (deepest_delta < delta_obj->delta_depth)
879 deepest_delta = delta_obj->delta_depth;
880 deepest_delta_unlock();
881 }
38a45561 882 delta_obj->base_object_no = base->obj - objects;
636171cb 883 delta_data = get_data_from_pack(delta_obj);
ce3f6dc6 884 base_data = get_base_data(base);
9441b61d 885 result->obj = delta_obj;
ce3f6dc6
NP
886 result->data = patch_delta(base_data, base->size,
887 delta_data, delta_obj->size, &result->size);
9cf6d335 888 free(delta_data);
9441b61d 889 if (!result->data)
c2b97ecf 890 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
681b07de
NTND
891 hash_sha1_file(result->data, result->size,
892 typename(delta_obj->real_type), delta_obj->idx.sha1);
9ec2dde9 893 sha1_object(result->data, NULL, result->size, delta_obj->real_type,
9441b61d 894 delta_obj->idx.sha1);
b8a2486f 895 counter_lock();
636171cb 896 nr_resolved_deltas++;
b8a2486f 897 counter_unlock();
9441b61d
NP
898}
899
ab791dd1
JK
900/*
901 * Standard boolean compare-and-swap: atomically check whether "*type" is
902 * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
903 * and return false.
904 */
905static int compare_and_swap_type(enum object_type *type,
906 enum object_type want,
907 enum object_type set)
908{
909 enum object_type old;
910
911 type_cas_lock();
912 old = *type;
913 if (old == want)
914 *type = set;
915 type_cas_unlock();
916
917 return old == want;
918}
919
2baad220
NTND
920static struct base_data *find_unresolved_deltas_1(struct base_data *base,
921 struct base_data *prev_base)
9441b61d 922{
2baad220 923 if (base->ref_last == -1 && base->ofs_last == -1) {
9441b61d
NP
924 union delta_base base_spec;
925
926 hashcpy(base_spec.sha1, base->obj->idx.sha1);
7218a215 927 find_delta_children(&base_spec,
2baad220 928 &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
53dda6ff 929
9441b61d
NP
930 memset(&base_spec, 0, sizeof(base_spec));
931 base_spec.offset = base->obj->idx.offset;
7218a215 932 find_delta_children(&base_spec,
2baad220 933 &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
4a438cab 934
2baad220
NTND
935 if (base->ref_last == -1 && base->ofs_last == -1) {
936 free(base->data);
937 return NULL;
938 }
53dda6ff 939
2baad220
NTND
940 link_base_data(prev_base, base);
941 }
4a438cab 942
2baad220
NTND
943 if (base->ref_first <= base->ref_last) {
944 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
945 struct base_data *result = alloc_base_data();
7218a215 946
ab791dd1
JK
947 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
948 base->obj->real_type))
949 die("BUG: child->real_type != OBJ_REF_DELTA");
950
2baad220
NTND
951 resolve_delta(child, base, result);
952 if (base->ref_first == base->ref_last && base->ofs_last == -1)
7218a215 953 free_base_data(base);
2baad220
NTND
954
955 base->ref_first++;
956 return result;
53dda6ff
NP
957 }
958
2baad220
NTND
959 if (base->ofs_first <= base->ofs_last) {
960 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
961 struct base_data *result = alloc_base_data();
7218a215
JH
962
963 assert(child->real_type == OBJ_OFS_DELTA);
ab791dd1 964 child->real_type = base->obj->real_type;
2baad220
NTND
965 resolve_delta(child, base, result);
966 if (base->ofs_first == base->ofs_last)
7218a215 967 free_base_data(base);
2baad220
NTND
968
969 base->ofs_first++;
970 return result;
9cf6d335 971 }
53dda6ff 972
9441b61d 973 unlink_base_data(base);
2baad220
NTND
974 return NULL;
975}
976
977static void find_unresolved_deltas(struct base_data *base)
978{
979 struct base_data *new_base, *prev_base = NULL;
980 for (;;) {
981 new_base = find_unresolved_deltas_1(base, prev_base);
982
983 if (new_base) {
984 prev_base = base;
985 base = new_base;
986 } else {
987 free(base);
988 base = prev_base;
989 if (!base)
990 return;
991 prev_base = base->base;
992 }
993 }
9cf6d335
SV
994}
995
996static int compare_delta_entry(const void *a, const void *b)
997{
998 const struct delta_entry *delta_a = a;
999 const struct delta_entry *delta_b = b;
7218a215
JH
1000
1001 /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
1002 return compare_delta_bases(&delta_a->base, &delta_b->base,
1003 objects[delta_a->obj_no].type,
1004 objects[delta_b->obj_no].type);
9cf6d335
SV
1005}
1006
5272f755
NTND
1007static void resolve_base(struct object_entry *obj)
1008{
1009 struct base_data *base_obj = alloc_base_data();
1010 base_obj->obj = obj;
1011 base_obj->data = NULL;
1012 find_unresolved_deltas(base_obj);
1013}
1014
b8a2486f
NTND
1015#ifndef NO_PTHREADS
1016static void *threaded_second_pass(void *data)
1017{
1018 set_thread_data(data);
1019 for (;;) {
1020 int i;
8f82aad4 1021 counter_lock();
b8a2486f 1022 display_progress(progress, nr_resolved_deltas);
8f82aad4
TR
1023 counter_unlock();
1024 work_lock();
b8a2486f
NTND
1025 while (nr_dispatched < nr_objects &&
1026 is_delta_type(objects[nr_dispatched].type))
1027 nr_dispatched++;
1028 if (nr_dispatched >= nr_objects) {
1029 work_unlock();
1030 break;
1031 }
1032 i = nr_dispatched++;
1033 work_unlock();
1034
1035 resolve_base(&objects[i]);
1036 }
1037 return NULL;
1038}
1039#endif
1040
5272f755
NTND
1041/*
1042 * First pass:
1043 * - find locations of all objects;
1044 * - calculate SHA1 of all non-delta objects;
1045 * - remember base (SHA1 or offset) for all deltas.
1046 */
2d477051 1047static void parse_pack_objects(unsigned char *sha1)
9cf6d335 1048{
9ec2dde9 1049 int i, nr_delays = 0;
53dda6ff 1050 struct delta_entry *delta = deltas;
2d477051 1051 struct stat st;
9cf6d335 1052
13aaf148 1053 if (verbose)
29e63ed3 1054 progress = start_progress(
c2b97ecf 1055 from_stdin ? _("Receiving objects") : _("Indexing objects"),
29e63ed3 1056 nr_objects);
9cf6d335
SV
1057 for (i = 0; i < nr_objects; i++) {
1058 struct object_entry *obj = &objects[i];
681b07de 1059 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
9cf6d335 1060 obj->real_type = obj->type;
4f8ec74e 1061 if (is_delta_type(obj->type)) {
53dda6ff 1062 nr_deltas++;
636171cb 1063 delta->obj_no = i;
53dda6ff 1064 delta++;
9ec2dde9
NTND
1065 } else if (!data) {
1066 /* large blobs, check later */
1067 obj->real_type = OBJ_BAD;
1068 nr_delays++;
9cf6d335 1069 } else
9ec2dde9 1070 sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
9cf6d335 1071 free(data);
4d4fcc54 1072 display_progress(progress, i+1);
9cf6d335 1073 }
aa7e44bf 1074 objects[i].idx.offset = consumed_bytes;
4d4fcc54 1075 stop_progress(&progress);
2d477051
NP
1076
1077 /* Check pack integrity */
636171cb 1078 flush();
9126f009 1079 git_SHA1_Final(sha1, &input_ctx);
2d477051 1080 if (hashcmp(fill(20), sha1))
c2b97ecf 1081 die(_("pack is corrupted (SHA1 mismatch)"));
9bee2478 1082 use(20);
2d477051
NP
1083
1084 /* If input_fd is a file, we should have reached its end now. */
1085 if (fstat(input_fd, &st))
c2b97ecf 1086 die_errno(_("cannot fstat packfile"));
fa257b05
JS
1087 if (S_ISREG(st.st_mode) &&
1088 lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
c2b97ecf 1089 die(_("pack has junk at the end"));
9ec2dde9
NTND
1090
1091 for (i = 0; i < nr_objects; i++) {
1092 struct object_entry *obj = &objects[i];
1093 if (obj->real_type != OBJ_BAD)
1094 continue;
1095 obj->real_type = obj->type;
1096 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1097 nr_delays--;
1098 }
1099 if (nr_delays)
1100 die(_("confusion beyond insanity in parse_pack_objects()"));
5272f755
NTND
1101}
1102
1103/*
1104 * Second pass:
1105 * - for all non-delta objects, look if it is used as a base for
1106 * deltas;
1107 * - if used as a base, uncompress the object and apply all deltas,
1108 * recursively checking if the resulting object is used as a base
1109 * for some more deltas.
1110 */
1111static void resolve_deltas(void)
1112{
1113 int i;
9cf6d335 1114
3c9af366
NP
1115 if (!nr_deltas)
1116 return;
1117
53dda6ff 1118 /* Sort deltas by base SHA1/offset for fast searching */
9cf6d335
SV
1119 qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1120 compare_delta_entry);
1121
13aaf148 1122 if (verbose)
c2b97ecf 1123 progress = start_progress(_("Resolving deltas"), nr_deltas);
b8a2486f
NTND
1124
1125#ifndef NO_PTHREADS
1126 nr_dispatched = 0;
1127 if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1128 init_thread();
1129 for (i = 0; i < nr_threads; i++) {
1130 int ret = pthread_create(&thread_data[i].thread, NULL,
1131 threaded_second_pass, thread_data + i);
1132 if (ret)
f350df42
NTND
1133 die(_("unable to create thread: %s"),
1134 strerror(ret));
b8a2486f
NTND
1135 }
1136 for (i = 0; i < nr_threads; i++)
1137 pthread_join(thread_data[i].thread, NULL);
1138 cleanup_thread();
1139 return;
1140 }
1141#endif
1142
9cf6d335
SV
1143 for (i = 0; i < nr_objects; i++) {
1144 struct object_entry *obj = &objects[i];
9cf6d335 1145
4f8ec74e 1146 if (is_delta_type(obj->type))
9cf6d335 1147 continue;
5272f755 1148 resolve_base(obj);
4d4fcc54 1149 display_progress(progress, nr_resolved_deltas);
9cf6d335 1150 }
636171cb
NP
1151}
1152
5272f755
NTND
1153/*
1154 * Third pass:
1155 * - append objects to convert thin pack to full pack if required
1156 * - write the final 20-byte SHA-1
1157 */
1158static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1159static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1160{
1161 if (nr_deltas == nr_resolved_deltas) {
1162 stop_progress(&progress);
1163 /* Flush remaining pack final 20-byte SHA1. */
1164 flush();
1165 return;
1166 }
1167
1168 if (fix_thin_pack) {
1169 struct sha1file *f;
1170 unsigned char read_sha1[20], tail_sha1[20];
5c3459fc 1171 struct strbuf msg = STRBUF_INIT;
5272f755
NTND
1172 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1173 int nr_objects_initial = nr_objects;
1174 if (nr_unresolved <= 0)
cc13431a 1175 die(_("confusion beyond insanity"));
2756ca43 1176 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
57165db0
JK
1177 memset(objects + nr_objects + 1, 0,
1178 nr_unresolved * sizeof(*objects));
5272f755
NTND
1179 f = sha1fd(output_fd, curr_pack);
1180 fix_unresolved_deltas(f, nr_unresolved);
5c3459fc
NTND
1181 strbuf_addf(&msg, _("completed with %d local objects"),
1182 nr_objects - nr_objects_initial);
1183 stop_progress_msg(&progress, msg.buf);
1184 strbuf_release(&msg);
5272f755
NTND
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)
f350df42
NTND
1191 die(_("Unexpected tail checksum for %s "
1192 "(disk corruption?)"), curr_pack);
5272f755
NTND
1193 }
1194 if (nr_deltas != nr_resolved_deltas)
cc13431a
JH
1195 die(Q_("pack has %d unresolved delta",
1196 "pack has %d unresolved deltas",
1197 nr_deltas - nr_resolved_deltas),
5272f755
NTND
1198 nr_deltas - nr_resolved_deltas);
1199}
1200
8522148f 1201static int write_compressed(struct sha1file *f, void *in, unsigned int size)
636171cb 1202{
ef49a7a0 1203 git_zstream stream;
7734d7f2
NP
1204 int status;
1205 unsigned char outbuf[4096];
636171cb
NP
1206
1207 memset(&stream, 0, sizeof(stream));
55bb5c91 1208 git_deflate_init(&stream, zlib_compression_level);
636171cb
NP
1209 stream.next_in = in;
1210 stream.avail_in = size;
636171cb 1211
7734d7f2
NP
1212 do {
1213 stream.next_out = outbuf;
1214 stream.avail_out = sizeof(outbuf);
55bb5c91 1215 status = git_deflate(&stream, Z_FINISH);
7734d7f2
NP
1216 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1217 } while (status == Z_OK);
1218
1219 if (status != Z_STREAM_END)
c2b97ecf 1220 die(_("unable to deflate appended object (%d)"), status);
636171cb 1221 size = stream.total_out;
55bb5c91 1222 git_deflate_end(&stream);
636171cb
NP
1223 return size;
1224}
1225
8522148f 1226static struct object_entry *append_obj_to_pack(struct sha1file *f,
03993e13 1227 const unsigned char *sha1, void *buf,
636171cb
NP
1228 unsigned long size, enum object_type type)
1229{
1230 struct object_entry *obj = &objects[nr_objects++];
1231 unsigned char header[10];
1232 unsigned long s = size;
1233 int n = 0;
1234 unsigned char c = (type << 4) | (s & 15);
1235 s >>= 4;
1236 while (s) {
1237 header[n++] = c | 0x80;
1238 c = s & 0x7f;
1239 s >>= 7;
1240 }
1241 header[n++] = c;
8522148f
NP
1242 crc32_begin(f);
1243 sha1write(f, header, n);
72de2883
BS
1244 obj[0].size = size;
1245 obj[0].hdr_size = n;
1246 obj[0].type = type;
1247 obj[0].real_type = type;
aa7e44bf 1248 obj[1].idx.offset = obj[0].idx.offset + n;
8522148f
NP
1249 obj[1].idx.offset += write_compressed(f, buf, size);
1250 obj[0].idx.crc32 = crc32_end(f);
838cd346 1251 sha1flush(f);
aa7e44bf 1252 hashcpy(obj->idx.sha1, sha1);
03993e13 1253 return obj;
636171cb
NP
1254}
1255
1256static int delta_pos_compare(const void *_a, const void *_b)
1257{
1258 struct delta_entry *a = *(struct delta_entry **)_a;
1259 struct delta_entry *b = *(struct delta_entry **)_b;
1260 return a->obj_no - b->obj_no;
1261}
9cf6d335 1262
8522148f 1263static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
636171cb
NP
1264{
1265 struct delta_entry **sorted_by_pos;
96a02f8f 1266 int i, n = 0;
636171cb
NP
1267
1268 /*
1269 * Since many unresolved deltas may well be themselves base objects
1270 * for more unresolved deltas, we really want to include the
1271 * smallest number of base objects that would cover as much delta
1272 * as possible by picking the
1273 * trunc deltas first, allowing for other deltas to resolve without
1274 * additional base objects. Since most base objects are to be found
1275 * before deltas depending on them, a good heuristic is to start
1276 * resolving deltas in the same order as their position in the pack.
1277 */
1278 sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
9cf6d335 1279 for (i = 0; i < nr_deltas; i++) {
636171cb
NP
1280 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1281 continue;
1282 sorted_by_pos[n++] = &deltas[i];
9cf6d335 1283 }
636171cb
NP
1284 qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1285
1286 for (i = 0; i < n; i++) {
1287 struct delta_entry *d = sorted_by_pos[i];
21666f1a 1288 enum object_type type;
2baad220 1289 struct base_data *base_obj = alloc_base_data();
636171cb
NP
1290
1291 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1292 continue;
2baad220
NTND
1293 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1294 if (!base_obj->data)
636171cb 1295 continue;
636171cb 1296
2baad220
NTND
1297 if (check_sha1_signature(d->base.sha1, base_obj->data,
1298 base_obj->size, typename(type)))
c2b97ecf 1299 die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
2baad220
NTND
1300 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1301 base_obj->data, base_obj->size, type);
1302 find_unresolved_deltas(base_obj);
4d4fcc54 1303 display_progress(progress, nr_resolved_deltas);
636171cb
NP
1304 }
1305 free(sorted_by_pos);
1306}
1307
e42797f5
NP
1308static void final(const char *final_pack_name, const char *curr_pack_name,
1309 const char *final_index_name, const char *curr_index_name,
b8077709 1310 const char *keep_name, const char *keep_msg,
e42797f5
NP
1311 unsigned char *sha1)
1312{
3a55602e 1313 const char *report = "pack";
e42797f5
NP
1314 char name[PATH_MAX];
1315 int err;
1316
1317 if (!from_stdin) {
1318 close(input_fd);
1319 } else {
4c81b03e 1320 fsync_or_die(output_fd, curr_pack_name);
e42797f5
NP
1321 err = close(output_fd);
1322 if (err)
c2b97ecf 1323 die_errno(_("error while closing pack file"));
e42797f5
NP
1324 }
1325
b8077709
SP
1326 if (keep_msg) {
1327 int keep_fd, keep_msg_len = strlen(keep_msg);
6e180cdc
JH
1328
1329 if (!keep_name)
1330 keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1331 else
1332 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1333
9ca4a201
NP
1334 if (keep_fd < 0) {
1335 if (errno != EEXIST)
c2b97ecf 1336 die_errno(_("cannot write keep file '%s'"),
de983a0a 1337 keep_name ? keep_name : name);
9ca4a201
NP
1338 } else {
1339 if (keep_msg_len > 0) {
1340 write_or_die(keep_fd, keep_msg, keep_msg_len);
1341 write_or_die(keep_fd, "\n", 1);
1342 }
91c8d590 1343 if (close(keep_fd) != 0)
c2b97ecf 1344 die_errno(_("cannot close written keep file '%s'"),
de983a0a 1345 keep_name ? keep_name : name);
576162a4 1346 report = "keep";
b8077709 1347 }
b8077709
SP
1348 }
1349
e42797f5
NP
1350 if (final_pack_name != curr_pack_name) {
1351 if (!final_pack_name) {
1352 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1353 get_object_directory(), sha1_to_hex(sha1));
1354 final_pack_name = name;
1355 }
1356 if (move_temp_to_file(curr_pack_name, final_pack_name))
c2b97ecf 1357 die(_("cannot store pack file"));
fb8b1936 1358 } else if (from_stdin)
33b65030 1359 chmod(final_pack_name, 0444);
e42797f5 1360
e42797f5
NP
1361 if (final_index_name != curr_index_name) {
1362 if (!final_index_name) {
1363 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1364 get_object_directory(), sha1_to_hex(sha1));
1365 final_index_name = name;
1366 }
1367 if (move_temp_to_file(curr_index_name, final_index_name))
c2b97ecf 1368 die(_("cannot store index file"));
fb8b1936
JH
1369 } else
1370 chmod(final_index_name, 0444);
576162a4
NP
1371
1372 if (!from_stdin) {
1373 printf("%s\n", sha1_to_hex(sha1));
1374 } else {
1375 char buf[48];
1376 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1377 report, sha1_to_hex(sha1));
d1b2ddc8 1378 write_or_die(1, buf, len);
576162a4
NP
1379
1380 /*
1381 * Let's just mimic git-unpack-objects here and write
1382 * the last part of the input buffer to stdout.
1383 */
1384 while (input_len) {
1385 err = xwrite(1, input_buffer + input_offset, input_len);
1386 if (err <= 0)
1387 break;
1388 input_len -= err;
1389 input_offset += err;
1390 }
1391 }
9cf6d335
SV
1392}
1393
ef90d6d4 1394static int git_index_pack_config(const char *k, const char *v, void *cb)
4d00bda2 1395{
ebcfb379
JH
1396 struct pack_idx_option *opts = cb;
1397
4d00bda2 1398 if (!strcmp(k, "pack.indexversion")) {
ebcfb379
JH
1399 opts->version = git_config_int(k, v);
1400 if (opts->version > 2)
f350df42 1401 die(_("bad pack.indexversion=%"PRIu32), opts->version);
4d00bda2
NP
1402 return 0;
1403 }
b8a2486f
NTND
1404 if (!strcmp(k, "pack.threads")) {
1405 nr_threads = git_config_int(k, v);
1406 if (nr_threads < 0)
f350df42 1407 die(_("invalid number of threads specified (%d)"),
b8a2486f
NTND
1408 nr_threads);
1409#ifdef NO_PTHREADS
1410 if (nr_threads != 1)
f350df42 1411 warning(_("no threads support, ignoring %s"), k);
b8a2486f
NTND
1412 nr_threads = 1;
1413#endif
1414 return 0;
1415 }
ef90d6d4 1416 return git_default_config(k, v, cb);
4d00bda2
NP
1417}
1418
3c9fc074
JH
1419static int cmp_uint32(const void *a_, const void *b_)
1420{
1421 uint32_t a = *((uint32_t *)a_);
1422 uint32_t b = *((uint32_t *)b_);
1423
1424 return (a < b) ? -1 : (a != b);
1425}
1426
1427static void read_v2_anomalous_offsets(struct packed_git *p,
1428 struct pack_idx_option *opts)
1429{
1430 const uint32_t *idx1, *idx2;
1431 uint32_t i;
1432
1433 /* The address of the 4-byte offset table */
1434 idx1 = (((const uint32_t *)p->index_data)
1435 + 2 /* 8-byte header */
1436 + 256 /* fan out */
1437 + 5 * p->num_objects /* 20-byte SHA-1 table */
1438 + p->num_objects /* CRC32 table */
1439 );
1440
1441 /* The address of the 8-byte offset table */
1442 idx2 = idx1 + p->num_objects;
1443
1444 for (i = 0; i < p->num_objects; i++) {
1445 uint32_t off = ntohl(idx1[i]);
1446 if (!(off & 0x80000000))
1447 continue;
1448 off = off & 0x7fffffff;
1449 if (idx2[off * 2])
1450 continue;
1451 /*
1452 * The real offset is ntohl(idx2[off * 2]) in high 4
1453 * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1454 * octets. But idx2[off * 2] is Zero!!!
1455 */
1456 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1457 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1458 }
1459
1460 if (1 < opts->anomaly_nr)
1461 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1462}
1463
e337a04d
JH
1464static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1465{
1466 struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1467
1468 if (!p)
c2b97ecf 1469 die(_("Cannot open existing pack file '%s'"), pack_name);
e337a04d 1470 if (open_pack_index(p))
c2b97ecf 1471 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
e337a04d
JH
1472
1473 /* Read the attributes from the existing idx file */
1474 opts->version = p->index_version;
1475
3c9fc074
JH
1476 if (opts->version == 2)
1477 read_v2_anomalous_offsets(p, opts);
1478
e337a04d
JH
1479 /*
1480 * Get rid of the idx file as we do not need it anymore.
1481 * NEEDSWORK: extract this bit from free_pack_by_name() in
1482 * sha1_file.c, perhaps? It shouldn't matter very much as we
1483 * know we haven't installed this pack (hence we never have
1484 * read anything from it).
1485 */
1486 close_pack_index(p);
1487 free(p);
1488}
1489
38a45561
JH
1490static void show_pack_info(int stat_only)
1491{
d1a0ed18
JH
1492 int i, baseobjects = nr_objects - nr_deltas;
1493 unsigned long *chain_histogram = NULL;
1494
1495 if (deepest_delta)
1496 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1497
38a45561
JH
1498 for (i = 0; i < nr_objects; i++) {
1499 struct object_entry *obj = &objects[i];
1500
d1a0ed18
JH
1501 if (is_delta_type(obj->type))
1502 chain_histogram[obj->delta_depth - 1]++;
38a45561
JH
1503 if (stat_only)
1504 continue;
1505 printf("%s %-6s %lu %lu %"PRIuMAX,
1506 sha1_to_hex(obj->idx.sha1),
1507 typename(obj->real_type), obj->size,
1508 (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1509 (uintmax_t)obj->idx.offset);
1510 if (is_delta_type(obj->type)) {
1511 struct object_entry *bobj = &objects[obj->base_object_no];
1512 printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1513 }
1514 putchar('\n');
1515 }
d1a0ed18
JH
1516
1517 if (baseobjects)
c2b97ecf
NTND
1518 printf_ln(Q_("non delta: %d object",
1519 "non delta: %d objects",
1520 baseobjects),
1521 baseobjects);
d1a0ed18
JH
1522 for (i = 0; i < deepest_delta; i++) {
1523 if (!chain_histogram[i])
1524 continue;
c2b97ecf
NTND
1525 printf_ln(Q_("chain length = %d: %lu object",
1526 "chain length = %d: %lu objects",
1527 chain_histogram[i]),
1528 i + 1,
1529 chain_histogram[i]);
d1a0ed18 1530 }
38a45561
JH
1531}
1532
3bb72562 1533int cmd_index_pack(int argc, const char **argv, const char *prefix)
9cf6d335 1534{
3aba2fdd 1535 int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
39539495 1536 const char *curr_index;
3bb72562 1537 const char *index_name = NULL, *pack_name = NULL;
b8077709 1538 const char *keep_name = NULL, *keep_msg = NULL;
592ce208
JK
1539 struct strbuf index_name_buf = STRBUF_INIT,
1540 keep_name_buf = STRBUF_INIT;
aa7e44bf 1541 struct pack_idx_entry **idx_objects;
ebcfb379 1542 struct pack_idx_option opts;
8522148f 1543 unsigned char pack_sha1[20];
c6807a40 1544 unsigned foreign_nr = 1; /* zero is a "good" value, assume bad */
9cf6d335 1545
99caeed0
JN
1546 if (argc == 2 && !strcmp(argv[1], "-h"))
1547 usage(index_pack_usage);
1548
afc711b8 1549 check_replace_refs = 0;
6e2a09d2 1550
ebcfb379
JH
1551 reset_pack_idx_option(&opts);
1552 git_config(git_index_pack_config, &opts);
3f8099fc 1553 if (prefix && chdir(prefix))
c2b97ecf 1554 die(_("Cannot come back to cwd"));
4d00bda2 1555
9cf6d335 1556 for (i = 1; i < argc; i++) {
3bb72562 1557 const char *arg = argv[i];
9cf6d335
SV
1558
1559 if (*arg == '-') {
e42797f5
NP
1560 if (!strcmp(arg, "--stdin")) {
1561 from_stdin = 1;
636171cb
NP
1562 } else if (!strcmp(arg, "--fix-thin")) {
1563 fix_thin_pack = 1;
0153be05
MK
1564 } else if (!strcmp(arg, "--strict")) {
1565 strict = 1;
c6807a40
NTND
1566 do_fsck_object = 1;
1567 } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1568 strict = 1;
1569 check_self_contained_and_connected = 1;
e337a04d
JH
1570 } else if (!strcmp(arg, "--verify")) {
1571 verify = 1;
38a45561
JH
1572 } else if (!strcmp(arg, "--verify-stat")) {
1573 verify = 1;
3aba2fdd 1574 show_stat = 1;
38a45561
JH
1575 } else if (!strcmp(arg, "--verify-stat-only")) {
1576 verify = 1;
3aba2fdd 1577 show_stat = 1;
38a45561 1578 stat_only = 1;
b8077709
SP
1579 } else if (!strcmp(arg, "--keep")) {
1580 keep_msg = "";
59556548 1581 } else if (starts_with(arg, "--keep=")) {
b8077709 1582 keep_msg = arg + 7;
59556548 1583 } else if (starts_with(arg, "--threads=")) {
b8a2486f
NTND
1584 char *end;
1585 nr_threads = strtoul(arg+10, &end, 0);
1586 if (!arg[10] || *end || nr_threads < 0)
1587 usage(index_pack_usage);
1588#ifdef NO_PTHREADS
1589 if (nr_threads != 1)
f350df42
NTND
1590 warning(_("no threads support, "
1591 "ignoring %s"), arg);
b8a2486f
NTND
1592 nr_threads = 1;
1593#endif
59556548 1594 } else if (starts_with(arg, "--pack_header=")) {
bed006fb
NP
1595 struct pack_header *hdr;
1596 char *c;
1597
1598 hdr = (struct pack_header *)input_buffer;
1599 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1600 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1601 if (*c != ',')
c2b97ecf 1602 die(_("bad %s"), arg);
bed006fb
NP
1603 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1604 if (*c)
c2b97ecf 1605 die(_("bad %s"), arg);
bed006fb 1606 input_len = sizeof(*hdr);
3c9af366
NP
1607 } else if (!strcmp(arg, "-v")) {
1608 verbose = 1;
e42797f5 1609 } else if (!strcmp(arg, "-o")) {
9cf6d335
SV
1610 if (index_name || (i+1) >= argc)
1611 usage(index_pack_usage);
1612 index_name = argv[++i];
59556548 1613 } else if (starts_with(arg, "--index-version=")) {
4ba7d711 1614 char *c;
ebcfb379
JH
1615 opts.version = strtoul(arg + 16, &c, 10);
1616 if (opts.version > 2)
c2b97ecf 1617 die(_("bad %s"), arg);
4ba7d711 1618 if (*c == ',')
ebcfb379
JH
1619 opts.off32_limit = strtoul(c+1, &c, 0);
1620 if (*c || opts.off32_limit & 0x80000000)
c2b97ecf 1621 die(_("bad %s"), arg);
9cf6d335
SV
1622 } else
1623 usage(index_pack_usage);
1624 continue;
1625 }
1626
1627 if (pack_name)
1628 usage(index_pack_usage);
1629 pack_name = arg;
1630 }
1631
e42797f5 1632 if (!pack_name && !from_stdin)
9cf6d335 1633 usage(index_pack_usage);
636171cb 1634 if (fix_thin_pack && !from_stdin)
c2b97ecf 1635 die(_("--fix-thin cannot be used without --stdin"));
e42797f5 1636 if (!index_name && pack_name) {
592ce208
JK
1637 size_t len;
1638 if (!strip_suffix(pack_name, ".pack", &len))
c2b97ecf 1639 die(_("packfile name '%s' does not end with '.pack'"),
9cf6d335 1640 pack_name);
592ce208
JK
1641 strbuf_add(&index_name_buf, pack_name, len);
1642 strbuf_addstr(&index_name_buf, ".idx");
1643 index_name = index_name_buf.buf;
9cf6d335 1644 }
b8077709 1645 if (keep_msg && !keep_name && pack_name) {
592ce208
JK
1646 size_t len;
1647 if (!strip_suffix(pack_name, ".pack", &len))
c2b97ecf 1648 die(_("packfile name '%s' does not end with '.pack'"),
b8077709 1649 pack_name);
592ce208
JK
1650 strbuf_add(&keep_name_buf, pack_name, len);
1651 strbuf_addstr(&keep_name_buf, ".idx");
1652 keep_name = keep_name_buf.buf;
b8077709 1653 }
e337a04d
JH
1654 if (verify) {
1655 if (!index_name)
c2b97ecf 1656 die(_("--verify with no packfile name given"));
e337a04d 1657 read_idx_option(&opts, index_name);
68be2fea 1658 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
e337a04d 1659 }
68be2fea
JH
1660 if (strict)
1661 opts.flags |= WRITE_IDX_STRICT;
9cf6d335 1662
b8a2486f
NTND
1663#ifndef NO_PTHREADS
1664 if (!nr_threads) {
1665 nr_threads = online_cpus();
1666 /* An experiment showed that more threads does not mean faster */
1667 if (nr_threads > 3)
1668 nr_threads = 3;
1669 }
1670#endif
1671
e42797f5 1672 curr_pack = open_pack_file(pack_name);
9cf6d335 1673 parse_pack_header();
38a45561
JH
1674 objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1675 deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
8522148f 1676 parse_pack_objects(pack_sha1);
5272f755
NTND
1677 resolve_deltas();
1678 conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
9cf6d335 1679 free(deltas);
0153be05 1680 if (strict)
c6807a40 1681 foreign_nr = check_objects();
aa7e44bf 1682
3aba2fdd 1683 if (show_stat)
38a45561
JH
1684 show_pack_info(stat_only);
1685
aa7e44bf
GB
1686 idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1687 for (i = 0; i < nr_objects; i++)
1688 idx_objects[i] = &objects[i].idx;
ebcfb379 1689 curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
aa7e44bf
GB
1690 free(idx_objects);
1691
e337a04d
JH
1692 if (!verify)
1693 final(pack_name, curr_pack,
1694 index_name, curr_index,
1695 keep_name, keep_msg,
1696 pack_sha1);
1697 else
1698 close(input_fd);
9cf6d335 1699 free(objects);
592ce208
JK
1700 strbuf_release(&index_name_buf);
1701 strbuf_release(&keep_name_buf);
c85228ed 1702 if (pack_name == NULL)
3bb72562 1703 free((void *) curr_pack);
c85228ed 1704 if (index_name == NULL)
3bb72562 1705 free((void *) curr_index);
9cf6d335 1706
c6807a40
NTND
1707 /*
1708 * Let the caller know this pack is not self contained
1709 */
1710 if (check_self_contained_and_connected && foreign_nr)
1711 return 1;
1712
9cf6d335
SV
1713 return 0;
1714}