]> git.ipfire.org Git - thirdparty/git.git/blame - builtin-pack-objects.c
pack-objects: pass fullname down to add_object_entry()
[thirdparty/git.git] / builtin-pack-objects.c
CommitLineData
5d4a6003 1#include "builtin.h"
c323ac7d
LT
2#include "cache.h"
3#include "object.h"
8e440259
PE
4#include "blob.h"
5#include "commit.h"
6#include "tag.h"
7#include "tree.h"
c323ac7d 8#include "delta.h"
a733cb60 9#include "pack.h"
c38138cd 10#include "csum-file.h"
1b0c7174 11#include "tree-walk.h"
b5d97e6b
JH
12#include "diff.h"
13#include "revision.h"
14#include "list-objects.h"
96a02f8f 15#include "progress.h"
c323ac7d 16
231f240b
NP
17static const char pack_usage[] = "\
18git-pack-objects [{ -q | --progress | --all-progress }] \n\
19 [--local] [--incremental] [--window=N] [--depth=N] \n\
fa736f72
NP
20 [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
21 [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
22 [--stdout | base-name] [<ref-list | <object-list]";
c323ac7d 23
c323ac7d
LT
24struct object_entry {
25 unsigned char sha1[20];
a3fbf4df
NP
26 uint32_t crc32; /* crc of raw pack data for this object */
27 off_t offset; /* offset into the final pack file */
3f9ac8d2 28 unsigned long size; /* uncompressed size */
3f9ac8d2 29 unsigned int hash; /* name hint hash */
a3fbf4df 30 unsigned int depth; /* delta depth */
3f9ac8d2 31 struct packed_git *in_pack; /* already in pack */
6777a59f 32 off_t in_pack_offset;
a3fbf4df 33 struct object_entry *delta; /* delta base object */
82e5a82f 34 struct object_entry *delta_child; /* deltified objects who bases me */
15b4d577
JH
35 struct object_entry *delta_sibling; /* other deltified objects who
36 * uses the same base as me
37 */
a3fbf4df
NP
38 unsigned long delta_size; /* delta data size (uncompressed) */
39 enum object_type type;
40 enum object_type in_pack_type; /* could be delta */
41 unsigned char in_pack_header_size;
42 unsigned char preferred_base; /* we do not pack this, but is available
43 * to be used as the base objectto delta
44 * objects against.
45 */
c323ac7d
LT
46};
47
3f9ac8d2 48/*
82e5a82f 49 * Objects we are going to pack are collected in objects array (dynamically
3f9ac8d2
JH
50 * expanded). nr_objects & nr_alloc controls this array. They are stored
51 * in the order we see -- typically rev-list --objects order that gives us
52 * nice "minimum seek" order.
3f9ac8d2 53 */
e4d58311
NP
54static struct object_entry *objects;
55static uint32_t nr_objects, nr_alloc, nr_result;
3f9ac8d2 56
96f1e58f 57static int non_empty;
fa736f72 58static int no_reuse_delta, no_reuse_object;
96f1e58f
DR
59static int local;
60static int incremental;
be6b1914 61static int allow_ofs_delta;
81a216a5
NP
62static const char *pack_tmp_name, *idx_tmp_name;
63static char tmpname[PATH_MAX];
e1808845 64static unsigned char pack_file_sha1[20];
024701f1 65static int progress = 1;
4812a93a 66static int window = 10;
618e613a 67static int depth = 50;
df6d6101 68static int pack_to_stdout;
8d1d8f83 69static int num_preferred_base;
96a02f8f 70static struct progress progress_state;
960ccca6
DH
71static int pack_compression_level = Z_DEFAULT_COMPRESSION;
72static int pack_compression_seen;
c323ac7d 73
3f9ac8d2
JH
74/*
75 * The object names in objects array are hashed with this hashtable,
e4d58311 76 * to help looking up the entry by object name.
3f9ac8d2
JH
77 * This hashtable is built after all the objects are seen.
78 */
96f1e58f
DR
79static int *object_ix;
80static int object_ix_hashsz;
3f9ac8d2
JH
81
82/*
83 * Pack index for existing packs give us easy access to the offsets into
84 * corresponding pack file where each object's data starts, but the entries
85 * do not store the size of the compressed representation (uncompressed
780e6e73
NP
86 * size is easily available by examining the pack entry header). It is
87 * also rather expensive to find the sha1 for an object given its offset.
88 *
89 * We build a hashtable of existing packs (pack_revindex), and keep reverse
90 * index here -- pack index file is sorted by object name mapping to offset;
91 * this pack_revindex[].revindex array is a list of offset/index_nr pairs
92 * ordered by offset, so if you know the offset of an object, next offset
93 * is where its packed representation ends and the index_nr can be used to
94 * get the object sha1 from the main index.
3f9ac8d2 95 */
780e6e73 96struct revindex_entry {
6777a59f 97 off_t offset;
780e6e73
NP
98 unsigned int nr;
99};
3f9ac8d2
JH
100struct pack_revindex {
101 struct packed_git *p;
780e6e73
NP
102 struct revindex_entry *revindex;
103};
104static struct pack_revindex *pack_revindex;
96f1e58f 105static int pack_revindex_hashsz;
3f9ac8d2
JH
106
107/*
108 * stats
109 */
7cadf491
SP
110static uint32_t written, written_delta;
111static uint32_t reused, reused_delta;
3f9ac8d2
JH
112
113static int pack_revindex_ix(struct packed_git *p)
114{
2b74cffa 115 unsigned long ui = (unsigned long)p;
3f9ac8d2
JH
116 int i;
117
118 ui = ui ^ (ui >> 16); /* defeat structure alignment */
119 i = (int)(ui % pack_revindex_hashsz);
120 while (pack_revindex[i].p) {
121 if (pack_revindex[i].p == p)
122 return i;
123 if (++i == pack_revindex_hashsz)
124 i = 0;
125 }
126 return -1 - i;
127}
128
129static void prepare_pack_ix(void)
130{
131 int num;
132 struct packed_git *p;
133 for (num = 0, p = packed_git; p; p = p->next)
134 num++;
135 if (!num)
136 return;
137 pack_revindex_hashsz = num * 11;
138 pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
139 for (p = packed_git; p; p = p->next) {
140 num = pack_revindex_ix(p);
141 num = - 1 - num;
142 pack_revindex[num].p = p;
143 }
144 /* revindex elements are lazily initialized */
145}
146
147static int cmp_offset(const void *a_, const void *b_)
148{
780e6e73
NP
149 const struct revindex_entry *a = a_;
150 const struct revindex_entry *b = b_;
151 return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0;
3f9ac8d2
JH
152}
153
154/*
155 * Ordered list of offsets of objects in the pack.
156 */
157static void prepare_pack_revindex(struct pack_revindex *rix)
158{
159 struct packed_git *p = rix->p;
57059091 160 int num_ent = p->num_objects;
3f9ac8d2 161 int i;
42873078 162 const char *index = p->index_data;
3f9ac8d2 163
780e6e73 164 rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
c553ca25
NP
165 index += 4 * 256;
166
167 if (p->index_version > 1) {
168 const uint32_t *off_32 =
169 (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
170 const uint32_t *off_64 = off_32 + p->num_objects;
171 for (i = 0; i < num_ent; i++) {
172 uint32_t off = ntohl(*off_32++);
173 if (!(off & 0x80000000)) {
174 rix->revindex[i].offset = off;
175 } else {
176 rix->revindex[i].offset =
177 ((uint64_t)ntohl(*off_64++)) << 32;
178 rix->revindex[i].offset |=
179 ntohl(*off_64++);
180 }
181 rix->revindex[i].nr = i;
182 }
183 } else {
184 for (i = 0; i < num_ent; i++) {
185 uint32_t hl = *((uint32_t *)(index + 24 * i));
186 rix->revindex[i].offset = ntohl(hl);
187 rix->revindex[i].nr = i;
188 }
3f9ac8d2 189 }
c553ca25 190
3f9ac8d2
JH
191 /* This knows the pack format -- the 20-byte trailer
192 * follows immediately after the last object data.
193 */
780e6e73
NP
194 rix->revindex[num_ent].offset = p->pack_size - 20;
195 rix->revindex[num_ent].nr = -1;
196 qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
3f9ac8d2
JH
197}
198
780e6e73 199static struct revindex_entry * find_packed_object(struct packed_git *p,
6777a59f 200 off_t ofs)
3f9ac8d2
JH
201{
202 int num;
203 int lo, hi;
204 struct pack_revindex *rix;
780e6e73 205 struct revindex_entry *revindex;
3f9ac8d2
JH
206 num = pack_revindex_ix(p);
207 if (num < 0)
208 die("internal error: pack revindex uninitialized");
209 rix = &pack_revindex[num];
210 if (!rix->revindex)
211 prepare_pack_revindex(rix);
212 revindex = rix->revindex;
213 lo = 0;
57059091 214 hi = p->num_objects + 1;
3f9ac8d2
JH
215 do {
216 int mi = (lo + hi) / 2;
780e6e73
NP
217 if (revindex[mi].offset == ofs) {
218 return revindex + mi;
3f9ac8d2 219 }
780e6e73 220 else if (ofs < revindex[mi].offset)
3f9ac8d2
JH
221 hi = mi;
222 else
223 lo = mi + 1;
224 } while (lo < hi);
225 die("internal error: pack revindex corrupt");
226}
227
42873078
NP
228static const unsigned char *find_packed_object_name(struct packed_git *p,
229 off_t ofs)
780e6e73
NP
230{
231 struct revindex_entry *entry = find_packed_object(p, ofs);
d72308e0 232 return nth_packed_object_sha1(p, entry->nr);
780e6e73
NP
233}
234
c323ac7d
LT
235static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
236{
237 unsigned long othersize, delta_size;
21666f1a
NP
238 enum object_type type;
239 void *otherbuf = read_sha1_file(entry->delta->sha1, &type, &othersize);
c323ac7d
LT
240 void *delta_buf;
241
242 if (!otherbuf)
243 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
8ee378a0 244 delta_buf = diff_delta(otherbuf, othersize,
dcde55bc 245 buf, size, &delta_size, 0);
75c42d8c 246 if (!delta_buf || delta_size != entry->delta_size)
c323ac7d
LT
247 die("delta size changed");
248 free(buf);
249 free(otherbuf);
250 return delta_buf;
251}
252
a733cb60
LT
253/*
254 * The per-object header is a pretty dense thing, which is
255 * - first byte: low four bits are "size", then three bits of "type",
256 * and the high bit is "size continues".
257 * - each byte afterwards: low seven bits are size continuation,
258 * with the high bit being "size continues"
259 */
260static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
261{
01247d87 262 int n = 1;
a733cb60
LT
263 unsigned char c;
264
eb32d236 265 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
a733cb60
LT
266 die("bad type %d", type);
267
01247d87
LT
268 c = (type << 4) | (size & 15);
269 size >>= 4;
270 while (size) {
a733cb60 271 *hdr++ = c | 0x80;
01247d87
LT
272 c = size & 0x7f;
273 size >>= 7;
274 n++;
a733cb60
LT
275 }
276 *hdr = c;
277 return n;
278}
279
780e6e73
NP
280/*
281 * we are going to reuse the existing object data as is. make
282 * sure it is not corrupt.
283 */
079afb18
SP
284static int check_pack_inflate(struct packed_git *p,
285 struct pack_window **w_curs,
6777a59f
SP
286 off_t offset,
287 off_t len,
079afb18
SP
288 unsigned long expect)
289{
290 z_stream stream;
291 unsigned char fakebuf[4096], *in;
292 int st;
293
294 memset(&stream, 0, sizeof(stream));
295 inflateInit(&stream);
296 do {
297 in = use_pack(p, w_curs, offset, &stream.avail_in);
298 stream.next_in = in;
299 stream.next_out = fakebuf;
300 stream.avail_out = sizeof(fakebuf);
301 st = inflate(&stream, Z_FINISH);
302 offset += stream.next_in - in;
303 } while (st == Z_OK || st == Z_BUF_ERROR);
304 inflateEnd(&stream);
305 return (st == Z_STREAM_END &&
306 stream.total_out == expect &&
307 stream.total_in == len) ? 0 : -1;
308}
309
91ecbeca
NP
310static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
311 off_t offset, off_t len, unsigned int nr)
312{
313 const uint32_t *index_crc;
314 uint32_t data_crc = crc32(0, Z_NULL, 0);
315
316 do {
317 unsigned int avail;
318 void *data = use_pack(p, w_curs, offset, &avail);
319 if (avail > len)
320 avail = len;
321 data_crc = crc32(data_crc, data, avail);
322 offset += avail;
323 len -= avail;
324 } while (len);
325
326 index_crc = p->index_data;
327 index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
328
329 return data_crc != ntohl(*index_crc);
330}
331
079afb18
SP
332static void copy_pack_data(struct sha1file *f,
333 struct packed_git *p,
334 struct pack_window **w_curs,
6777a59f
SP
335 off_t offset,
336 off_t len)
079afb18
SP
337{
338 unsigned char *in;
339 unsigned int avail;
340
341 while (len) {
342 in = use_pack(p, w_curs, offset, &avail);
343 if (avail > len)
6777a59f 344 avail = (unsigned int)len;
079afb18
SP
345 sha1write(f, in, avail);
346 offset += avail;
347 len -= avail;
348 }
349}
350
d7dd0223 351static unsigned long write_object(struct sha1file *f,
7a979d99 352 struct object_entry *entry)
c323ac7d
LT
353{
354 unsigned long size;
21666f1a 355 enum object_type type;
3f9ac8d2 356 void *buf;
a733cb60 357 unsigned char header[10];
6777a59f
SP
358 unsigned hdrlen;
359 off_t datalen;
a733cb60 360 enum object_type obj_type;
ab7cd7bb 361 int to_reuse = 0;
c323ac7d 362
78d1e84f
NP
363 if (!pack_to_stdout)
364 crc32_begin(f);
365
a733cb60 366 obj_type = entry->type;
fa736f72
NP
367 if (no_reuse_object)
368 to_reuse = 0; /* explicit */
369 else if (!entry->in_pack)
ab7cd7bb 370 to_reuse = 0; /* can't reuse what we don't have */
780e6e73 371 else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
ab7cd7bb
JH
372 to_reuse = 1; /* check_object() decided it for us */
373 else if (obj_type != entry->in_pack_type)
374 to_reuse = 0; /* pack has delta which is unusable */
375 else if (entry->delta)
376 to_reuse = 0; /* we want to pack afresh */
377 else
378 to_reuse = 1; /* we have it in-pack undeltified,
379 * and we do not need to deltify it.
380 */
381
df6d6101 382 if (!to_reuse) {
21666f1a 383 buf = read_sha1_file(entry->sha1, &type, &size);
3f9ac8d2
JH
384 if (!buf)
385 die("unable to read %s", sha1_to_hex(entry->sha1));
386 if (size != entry->size)
387 die("object %s size inconsistency (%lu vs %lu)",
388 sha1_to_hex(entry->sha1), size, entry->size);
389 if (entry->delta) {
390 buf = delta_against(buf, size, entry);
391 size = entry->delta_size;
be6b1914
NP
392 obj_type = (allow_ofs_delta && entry->delta->offset) ?
393 OBJ_OFS_DELTA : OBJ_REF_DELTA;
3f9ac8d2
JH
394 }
395 /*
396 * The object header is a byte of 'type' followed by zero or
be6b1914 397 * more bytes of length.
3f9ac8d2
JH
398 */
399 hdrlen = encode_header(obj_type, size, header);
400 sha1write(f, header, hdrlen);
401
be6b1914
NP
402 if (obj_type == OBJ_OFS_DELTA) {
403 /*
404 * Deltas with relative base contain an additional
405 * encoding of the relative offset for the delta
406 * base from this object's position in the pack.
407 */
6777a59f 408 off_t ofs = entry->offset - entry->delta->offset;
be6b1914
NP
409 unsigned pos = sizeof(header) - 1;
410 header[pos] = ofs & 127;
411 while (ofs >>= 7)
412 header[--pos] = 128 | (--ofs & 127);
413 sha1write(f, header + pos, sizeof(header) - pos);
414 hdrlen += sizeof(header) - pos;
415 } else if (obj_type == OBJ_REF_DELTA) {
416 /*
417 * Deltas with a base reference contain
418 * an additional 20 bytes for the base sha1.
419 */
420 sha1write(f, entry->delta->sha1, 20);
3f9ac8d2
JH
421 hdrlen += 20;
422 }
960ccca6 423 datalen = sha1write_compressed(f, buf, size, pack_compression_level);
3f9ac8d2 424 free(buf);
c323ac7d 425 }
3f9ac8d2
JH
426 else {
427 struct packed_git *p = entry->in_pack;
03e79c88 428 struct pack_window *w_curs = NULL;
91ecbeca 429 struct revindex_entry *revidx;
6777a59f 430 off_t offset;
3f9ac8d2 431
780e6e73
NP
432 if (entry->delta) {
433 obj_type = (allow_ofs_delta && entry->delta->offset) ?
434 OBJ_OFS_DELTA : OBJ_REF_DELTA;
435 reused_delta++;
436 }
437 hdrlen = encode_header(obj_type, entry->size, header);
438 sha1write(f, header, hdrlen);
439 if (obj_type == OBJ_OFS_DELTA) {
6777a59f 440 off_t ofs = entry->offset - entry->delta->offset;
780e6e73
NP
441 unsigned pos = sizeof(header) - 1;
442 header[pos] = ofs & 127;
443 while (ofs >>= 7)
444 header[--pos] = 128 | (--ofs & 127);
445 sha1write(f, header + pos, sizeof(header) - pos);
446 hdrlen += sizeof(header) - pos;
447 } else if (obj_type == OBJ_REF_DELTA) {
448 sha1write(f, entry->delta->sha1, 20);
449 hdrlen += 20;
450 }
df6d6101 451
91ecbeca
NP
452 offset = entry->in_pack_offset;
453 revidx = find_packed_object(p, offset);
454 datalen = revidx[1].offset - offset;
455 if (!pack_to_stdout && p->index_version > 1 &&
456 check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
457 die("bad packed object CRC for %s", sha1_to_hex(entry->sha1));
458 offset += entry->in_pack_header_size;
459 datalen -= entry->in_pack_header_size;
460 if (!pack_to_stdout && p->index_version == 1 &&
461 check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
462 die("corrupt packed object for %s", sha1_to_hex(entry->sha1));
079afb18 463 copy_pack_data(f, p, &w_curs, offset, datalen);
03e79c88 464 unuse_pack(&w_curs);
3f9ac8d2 465 reused++;
a733cb60 466 }
be6b1914 467 if (entry->delta)
ab7cd7bb 468 written_delta++;
3f9ac8d2 469 written++;
78d1e84f
NP
470 if (!pack_to_stdout)
471 entry->crc32 = crc32_end(f);
c323ac7d
LT
472 return hdrlen + datalen;
473}
474
6777a59f 475static off_t write_one(struct sha1file *f,
9d5ab962 476 struct object_entry *e,
6777a59f 477 off_t offset)
9d5ab962 478{
d7dd0223
NP
479 unsigned long size;
480
481 /* offset is non zero if object is written already. */
be6b1914 482 if (e->offset || e->preferred_base)
9d5ab962 483 return offset;
d7dd0223
NP
484
485 /* if we are deltified, write out base object first. */
9d5ab962
JH
486 if (e->delta)
487 offset = write_one(f, e->delta, offset);
d7dd0223 488
be6b1914 489 e->offset = offset;
d7dd0223
NP
490 size = write_object(f, e);
491
492 /* make sure off_t is sufficiently large not to wrap */
493 if (offset > offset + size)
494 die("pack too large for current definition of off_t");
495 return offset + size;
9d5ab962
JH
496}
497
3082acfa
AR
498static int open_object_dir_tmp(const char *path)
499{
500 snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
501 return mkstemp(tmpname);
502}
503
c553ca25 504static off_t write_pack_file(void)
c323ac7d 505{
7cadf491 506 uint32_t i;
d22b9290 507 struct sha1file *f;
c553ca25 508 off_t offset, last_obj_offset = 0;
a733cb60 509 struct pack_header hdr;
fa438a2e 510 int do_progress = progress;
c323ac7d 511
81a216a5 512 if (pack_to_stdout) {
d22b9290 513 f = sha1fd(1, "<stdout>");
fa438a2e 514 do_progress >>= 1;
81a216a5 515 } else {
3082acfa 516 int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
81a216a5
NP
517 if (fd < 0)
518 die("unable to create %s: %s\n", tmpname, strerror(errno));
519 pack_tmp_name = xstrdup(tmpname);
520 f = sha1fd(fd, pack_tmp_name);
fa438a2e 521 }
81a216a5 522
13aaf148
NP
523 if (do_progress)
524 start_progress(&progress_state, "Writing %u objects...", "", nr_result);
183bdb2c 525
a733cb60 526 hdr.hdr_signature = htonl(PACK_SIGNATURE);
01247d87 527 hdr.hdr_version = htonl(PACK_VERSION);
7a979d99 528 hdr.hdr_entries = htonl(nr_result);
a733cb60
LT
529 sha1write(f, &hdr, sizeof(hdr));
530 offset = sizeof(hdr);
f0b0af1b
JH
531 if (!nr_result)
532 goto done;
5e8dc750 533 for (i = 0; i < nr_objects; i++) {
c553ca25 534 last_obj_offset = offset;
9d5ab962 535 offset = write_one(f, objects + i, offset);
96a02f8f
NP
536 if (do_progress)
537 display_progress(&progress_state, written);
5e8dc750 538 }
183bdb2c 539 if (do_progress)
96a02f8f 540 stop_progress(&progress_state);
f0b0af1b 541 done:
67c08ce1 542 if (written != nr_result)
7cadf491 543 die("wrote %u objects while expecting %u", written, nr_result);
e1808845 544 sha1close(f, pack_file_sha1, 1);
c553ca25
NP
545
546 return last_obj_offset;
c323ac7d
LT
547}
548
81a216a5
NP
549static int sha1_sort(const void *_a, const void *_b)
550{
551 const struct object_entry *a = *(struct object_entry **)_a;
552 const struct object_entry *b = *(struct object_entry **)_b;
553 return hashcmp(a->sha1, b->sha1);
554}
555
4ba7d711
NP
556static uint32_t index_default_version = 1;
557static uint32_t index_off32_limit = 0x7fffffff;
558
81a216a5 559static void write_index_file(off_t last_obj_offset, unsigned char *sha1)
c323ac7d 560{
81a216a5
NP
561 struct sha1file *f;
562 struct object_entry **sorted_by_sha, **list, **last;
b18b00a6 563 uint32_t array[256];
81a216a5
NP
564 uint32_t i, index_version;
565 SHA_CTX ctx;
81a216a5 566
3082acfa 567 int fd = open_object_dir_tmp("tmp_idx_XXXXXX");
81a216a5
NP
568 if (fd < 0)
569 die("unable to create %s: %s\n", tmpname, strerror(errno));
570 idx_tmp_name = xstrdup(tmpname);
571 f = sha1fd(fd, idx_tmp_name);
572
573 if (nr_result) {
574 uint32_t j = 0;
575 sorted_by_sha =
576 xcalloc(nr_result, sizeof(struct object_entry *));
577 for (i = 0; i < nr_objects; i++)
578 if (!objects[i].preferred_base)
579 sorted_by_sha[j++] = objects + i;
580 if (j != nr_result)
581 die("listed %u objects while expecting %u", j, nr_result);
582 qsort(sorted_by_sha, nr_result, sizeof(*sorted_by_sha), sha1_sort);
583 list = sorted_by_sha;
584 last = sorted_by_sha + nr_result;
585 } else
586 sorted_by_sha = list = last = NULL;
c553ca25
NP
587
588 /* if last object's offset is >= 2^31 we should use index V2 */
4ba7d711 589 index_version = (last_obj_offset >> 31) ? 2 : index_default_version;
c553ca25
NP
590
591 /* index versions 2 and above need a header */
592 if (index_version >= 2) {
593 struct pack_idx_header hdr;
594 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
595 hdr.idx_version = htonl(index_version);
596 sha1write(f, &hdr, sizeof(hdr));
597 }
c323ac7d
LT
598
599 /*
600 * Write the first-level table (the list is sorted,
601 * but we use a 256-entry lookup to be able to avoid
e1808845 602 * having to do eight extra binary search iterations).
c323ac7d
LT
603 */
604 for (i = 0; i < 256; i++) {
605 struct object_entry **next = list;
606 while (next < last) {
607 struct object_entry *entry = *next;
608 if (entry->sha1[0] != i)
609 break;
610 next++;
611 }
612 array[i] = htonl(next - sorted_by_sha);
613 list = next;
614 }
b18b00a6 615 sha1write(f, array, 256 * 4);
c323ac7d 616
81a216a5
NP
617 /* Compute the SHA1 hash of sorted object names. */
618 SHA1_Init(&ctx);
619
620 /* Write the actual SHA1 entries. */
c323ac7d 621 list = sorted_by_sha;
7a979d99 622 for (i = 0; i < nr_result; i++) {
c323ac7d 623 struct object_entry *entry = *list++;
c553ca25
NP
624 if (index_version < 2) {
625 uint32_t offset = htonl(entry->offset);
626 sha1write(f, &offset, 4);
627 }
c38138cd 628 sha1write(f, entry->sha1, 20);
81a216a5 629 SHA1_Update(&ctx, entry->sha1, 20);
c323ac7d 630 }
c553ca25
NP
631
632 if (index_version >= 2) {
633 unsigned int nr_large_offset = 0;
634
635 /* write the crc32 table */
636 list = sorted_by_sha;
637 for (i = 0; i < nr_objects; i++) {
638 struct object_entry *entry = *list++;
639 uint32_t crc32_val = htonl(entry->crc32);
640 sha1write(f, &crc32_val, 4);
641 }
642
643 /* write the 32-bit offset table */
644 list = sorted_by_sha;
645 for (i = 0; i < nr_objects; i++) {
646 struct object_entry *entry = *list++;
4ba7d711 647 uint32_t offset = (entry->offset <= index_off32_limit) ?
c553ca25
NP
648 entry->offset : (0x80000000 | nr_large_offset++);
649 offset = htonl(offset);
650 sha1write(f, &offset, 4);
651 }
652
653 /* write the large offset table */
654 list = sorted_by_sha;
655 while (nr_large_offset) {
656 struct object_entry *entry = *list++;
657 uint64_t offset = entry->offset;
4ba7d711 658 if (offset > index_off32_limit) {
c553ca25
NP
659 uint32_t split[2];
660 split[0] = htonl(offset >> 32);
661 split[1] = htonl(offset & 0xffffffff);
662 sha1write(f, split, 8);
663 nr_large_offset--;
664 }
665 }
666 }
667
e1808845
LT
668 sha1write(f, pack_file_sha1, 20);
669 sha1close(f, NULL, 1);
81a216a5
NP
670 free(sorted_by_sha);
671 SHA1_Final(sha1, &ctx);
c323ac7d
LT
672}
673
7a979d99
JH
674static int locate_object_entry_hash(const unsigned char *sha1)
675{
676 int i;
677 unsigned int ui;
678 memcpy(&ui, sha1, sizeof(unsigned int));
679 i = ui % object_ix_hashsz;
680 while (0 < object_ix[i]) {
a89fccd2 681 if (!hashcmp(sha1, objects[object_ix[i] - 1].sha1))
7a979d99
JH
682 return i;
683 if (++i == object_ix_hashsz)
684 i = 0;
685 }
686 return -1 - i;
687}
688
689static struct object_entry *locate_object_entry(const unsigned char *sha1)
690{
691 int i;
692
693 if (!object_ix_hashsz)
694 return NULL;
695
696 i = locate_object_entry_hash(sha1);
697 if (0 <= i)
698 return &objects[object_ix[i]-1];
699 return NULL;
700}
701
702static void rehash_objects(void)
c323ac7d 703{
7cadf491 704 uint32_t i;
7a979d99
JH
705 struct object_entry *oe;
706
707 object_ix_hashsz = nr_objects * 3;
708 if (object_ix_hashsz < 1024)
709 object_ix_hashsz = 1024;
710 object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
5379a5c5 711 memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
7a979d99
JH
712 for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
713 int ix = locate_object_entry_hash(oe->sha1);
714 if (0 <= ix)
715 continue;
716 ix = -1 - ix;
717 object_ix[ix] = i + 1;
718 }
719}
720
ce0bd642 721static unsigned name_hash(const char *name)
1d6b38cc 722{
ce0bd642
LT
723 unsigned char c;
724 unsigned hash = 0;
725
bc32fed5
JH
726 if (!name)
727 return 0;
728
eeef7135 729 /*
ce0bd642
LT
730 * This effectively just creates a sortable number from the
731 * last sixteen non-whitespace characters. Last characters
732 * count "most", so things that end in ".c" sort together.
eeef7135 733 */
ce0bd642
LT
734 while ((c = *name++) != 0) {
735 if (isspace(c))
736 continue;
737 hash = (hash >> 2) + (c << 24);
738 }
1d6b38cc
JH
739 return hash;
740}
741
5c49c116 742static int add_object_entry(const unsigned char *sha1, enum object_type type,
bc32fed5 743 const char *name, int exclude)
c323ac7d 744{
c323ac7d 745 struct object_entry *entry;
29b734e4 746 struct packed_git *p, *found_pack = NULL;
6777a59f 747 off_t found_offset = 0;
29b734e4 748 int ix;
bc32fed5 749 unsigned hash = name_hash(name);
29b734e4
NP
750
751 ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
752 if (ix >= 0) {
753 if (exclude) {
754 entry = objects + object_ix[ix] - 1;
81a216a5
NP
755 if (!entry->preferred_base)
756 nr_result--;
29b734e4
NP
757 entry->preferred_base = 1;
758 }
759 return 0;
760 }
c323ac7d 761
5c49c116
NP
762 for (p = packed_git; p; p = p->next) {
763 off_t offset = find_pack_entry_one(sha1, p);
764 if (offset) {
765 if (!found_pack) {
766 found_offset = offset;
767 found_pack = p;
64560374 768 }
5c49c116
NP
769 if (exclude)
770 break;
771 if (incremental)
772 return 0;
773 if (local && !p->pack_local)
774 return 0;
64560374
LT
775 }
776 }
eb019375 777
29b734e4
NP
778 if (nr_objects >= nr_alloc) {
779 nr_alloc = (nr_alloc + 1024) * 3 / 2;
7cadf491 780 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
c323ac7d 781 }
29b734e4
NP
782
783 entry = objects + nr_objects++;
c323ac7d 784 memset(entry, 0, sizeof(*entry));
e702496e 785 hashcpy(entry->sha1, sha1);
27225f2e 786 entry->hash = hash;
5c49c116
NP
787 if (type)
788 entry->type = type;
29b734e4
NP
789 if (exclude)
790 entry->preferred_base = 1;
81a216a5
NP
791 else
792 nr_result++;
29b734e4
NP
793 if (found_pack) {
794 entry->in_pack = found_pack;
795 entry->in_pack_offset = found_offset;
796 }
7a979d99
JH
797
798 if (object_ix_hashsz * 3 <= nr_objects * 4)
799 rehash_objects();
29b734e4
NP
800 else
801 object_ix[-1 - ix] = nr_objects;
7a979d99 802
96a02f8f
NP
803 if (progress)
804 display_progress(&progress_state, nr_objects);
29b734e4
NP
805
806 return 1;
c323ac7d
LT
807}
808
5379a5c5
JH
809struct pbase_tree_cache {
810 unsigned char sha1[20];
811 int ref;
812 int temporary;
813 void *tree_data;
814 unsigned long tree_size;
815};
816
817static struct pbase_tree_cache *(pbase_tree_cache[256]);
818static int pbase_tree_cache_ix(const unsigned char *sha1)
819{
820 return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
821}
822static int pbase_tree_cache_ix_incr(int ix)
823{
824 return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
825}
826
827static struct pbase_tree {
828 struct pbase_tree *next;
829 /* This is a phony "cache" entry; we are not
830 * going to evict it nor find it through _get()
831 * mechanism -- this is for the toplevel node that
832 * would almost always change with any commit.
833 */
834 struct pbase_tree_cache pcache;
835} *pbase_tree;
836
837static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
838{
839 struct pbase_tree_cache *ent, *nent;
840 void *data;
841 unsigned long size;
21666f1a 842 enum object_type type;
5379a5c5
JH
843 int neigh;
844 int my_ix = pbase_tree_cache_ix(sha1);
845 int available_ix = -1;
846
847 /* pbase-tree-cache acts as a limited hashtable.
848 * your object will be found at your index or within a few
849 * slots after that slot if it is cached.
850 */
851 for (neigh = 0; neigh < 8; neigh++) {
852 ent = pbase_tree_cache[my_ix];
a89fccd2 853 if (ent && !hashcmp(ent->sha1, sha1)) {
5379a5c5
JH
854 ent->ref++;
855 return ent;
856 }
857 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
858 ((0 <= available_ix) &&
859 (!ent && pbase_tree_cache[available_ix])))
860 available_ix = my_ix;
861 if (!ent)
862 break;
863 my_ix = pbase_tree_cache_ix_incr(my_ix);
864 }
865
866 /* Did not find one. Either we got a bogus request or
867 * we need to read and perhaps cache.
868 */
21666f1a 869 data = read_sha1_file(sha1, &type, &size);
5379a5c5
JH
870 if (!data)
871 return NULL;
21666f1a 872 if (type != OBJ_TREE) {
5379a5c5
JH
873 free(data);
874 return NULL;
875 }
876
877 /* We need to either cache or return a throwaway copy */
878
879 if (available_ix < 0)
880 ent = NULL;
881 else {
882 ent = pbase_tree_cache[available_ix];
883 my_ix = available_ix;
884 }
885
886 if (!ent) {
887 nent = xmalloc(sizeof(*nent));
888 nent->temporary = (available_ix < 0);
889 }
890 else {
891 /* evict and reuse */
892 free(ent->tree_data);
893 nent = ent;
894 }
e702496e 895 hashcpy(nent->sha1, sha1);
5379a5c5
JH
896 nent->tree_data = data;
897 nent->tree_size = size;
898 nent->ref = 1;
899 if (!nent->temporary)
900 pbase_tree_cache[my_ix] = nent;
901 return nent;
902}
903
904static void pbase_tree_put(struct pbase_tree_cache *cache)
905{
906 if (!cache->temporary) {
907 cache->ref--;
908 return;
909 }
910 free(cache->tree_data);
911 free(cache);
912}
913
914static int name_cmp_len(const char *name)
915{
916 int i;
917 for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
918 ;
919 return i;
920}
921
922static void add_pbase_object(struct tree_desc *tree,
5379a5c5 923 const char *name,
ce0bd642
LT
924 int cmplen,
925 const char *fullname)
3f9ac8d2 926{
4c068a98 927 struct name_entry entry;
8a5a8d6c 928 int cmp;
4c068a98
LT
929
930 while (tree_entry(tree,&entry)) {
8a5a8d6c
NP
931 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
932 memcmp(name, entry.path, cmplen);
933 if (cmp > 0)
7a979d99 934 continue;
8a5a8d6c
NP
935 if (cmp < 0)
936 return;
5379a5c5 937 if (name[cmplen] != '/') {
5c49c116
NP
938 add_object_entry(entry.sha1,
939 S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB,
bc32fed5 940 fullname, 1);
5379a5c5
JH
941 return;
942 }
8a5a8d6c 943 if (S_ISDIR(entry.mode)) {
7a979d99 944 struct tree_desc sub;
5379a5c5
JH
945 struct pbase_tree_cache *tree;
946 const char *down = name+cmplen+1;
947 int downlen = name_cmp_len(down);
948
4c068a98 949 tree = pbase_tree_get(entry.sha1);
5379a5c5
JH
950 if (!tree)
951 return;
6fda5e51 952 init_tree_desc(&sub, tree->tree_data, tree->tree_size);
5379a5c5 953
ce0bd642 954 add_pbase_object(&sub, down, downlen, fullname);
5379a5c5
JH
955 pbase_tree_put(tree);
956 }
957 }
958}
1d6b38cc 959
5379a5c5
JH
960static unsigned *done_pbase_paths;
961static int done_pbase_paths_num;
962static int done_pbase_paths_alloc;
963static int done_pbase_path_pos(unsigned hash)
964{
965 int lo = 0;
966 int hi = done_pbase_paths_num;
967 while (lo < hi) {
968 int mi = (hi + lo) / 2;
969 if (done_pbase_paths[mi] == hash)
970 return mi;
971 if (done_pbase_paths[mi] < hash)
972 hi = mi;
973 else
974 lo = mi + 1;
975 }
976 return -lo-1;
977}
978
979static int check_pbase_path(unsigned hash)
980{
981 int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
982 if (0 <= pos)
983 return 1;
984 pos = -pos - 1;
985 if (done_pbase_paths_alloc <= done_pbase_paths_num) {
986 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
987 done_pbase_paths = xrealloc(done_pbase_paths,
988 done_pbase_paths_alloc *
989 sizeof(unsigned));
990 }
991 done_pbase_paths_num++;
992 if (pos < done_pbase_paths_num)
993 memmove(done_pbase_paths + pos + 1,
994 done_pbase_paths + pos,
995 (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
996 done_pbase_paths[pos] = hash;
997 return 0;
998}
999
bc32fed5 1000static void add_preferred_base_object(const char *name)
5379a5c5
JH
1001{
1002 struct pbase_tree *it;
8a5a8d6c 1003 int cmplen;
bc32fed5 1004 unsigned hash = name_hash(name);
5379a5c5 1005
8a5a8d6c 1006 if (!num_preferred_base || check_pbase_path(hash))
5379a5c5
JH
1007 return;
1008
8a5a8d6c 1009 cmplen = name_cmp_len(name);
5379a5c5
JH
1010 for (it = pbase_tree; it; it = it->next) {
1011 if (cmplen == 0) {
bc32fed5 1012 add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
5379a5c5
JH
1013 }
1014 else {
1015 struct tree_desc tree;
6fda5e51 1016 init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
ce0bd642 1017 add_pbase_object(&tree, name, cmplen, name);
7a979d99 1018 }
3f9ac8d2 1019 }
3f9ac8d2
JH
1020}
1021
7a979d99 1022static void add_preferred_base(unsigned char *sha1)
3f9ac8d2 1023{
5379a5c5
JH
1024 struct pbase_tree *it;
1025 void *data;
1026 unsigned long size;
1027 unsigned char tree_sha1[20];
1d6b38cc 1028
8d1d8f83
JH
1029 if (window <= num_preferred_base++)
1030 return;
1031
5379a5c5
JH
1032 data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1033 if (!data)
7a979d99 1034 return;
5379a5c5
JH
1035
1036 for (it = pbase_tree; it; it = it->next) {
a89fccd2 1037 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
5379a5c5
JH
1038 free(data);
1039 return;
1040 }
1041 }
1042
1043 it = xcalloc(1, sizeof(*it));
1044 it->next = pbase_tree;
1045 pbase_tree = it;
1046
e702496e 1047 hashcpy(it->pcache.sha1, tree_sha1);
5379a5c5
JH
1048 it->pcache.tree_data = data;
1049 it->pcache.tree_size = size;
3f9ac8d2
JH
1050}
1051
c323ac7d
LT
1052static void check_object(struct object_entry *entry)
1053{
5c49c116 1054 if (entry->in_pack) {
780e6e73 1055 struct packed_git *p = entry->in_pack;
03e79c88 1056 struct pack_window *w_curs = NULL;
5c49c116
NP
1057 const unsigned char *base_ref = NULL;
1058 struct object_entry *base_entry;
1059 unsigned long used, used_0;
6777a59f 1060 unsigned int avail;
5c49c116
NP
1061 off_t ofs;
1062 unsigned char *buf, c;
780e6e73 1063
6777a59f 1064 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
ab7cd7bb 1065
5c49c116 1066 /*
fa736f72
NP
1067 * We want in_pack_type even if we do not reuse delta
1068 * since non-delta representations could still be reused.
ab7cd7bb 1069 */
6777a59f 1070 used = unpack_object_header_gently(buf, avail,
5c49c116
NP
1071 &entry->in_pack_type,
1072 &entry->size);
ab7cd7bb 1073
5c49c116
NP
1074 /*
1075 * Determine if this is a delta and if so whether we can
1076 * reuse it or not. Otherwise let's find out as cheaply as
1077 * possible what the actual type and size for this object is.
3f9ac8d2 1078 */
5c49c116
NP
1079 switch (entry->in_pack_type) {
1080 default:
1081 /* Not a delta hence we've already got all we need. */
1082 entry->type = entry->in_pack_type;
1083 entry->in_pack_header_size = used;
1084 unuse_pack(&w_curs);
1085 return;
1086 case OBJ_REF_DELTA:
1087 if (!no_reuse_delta && !entry->preferred_base)
1088 base_ref = use_pack(p, &w_curs,
1089 entry->in_pack_offset + used, NULL);
1090 entry->in_pack_header_size = used + 20;
1091 break;
1092 case OBJ_OFS_DELTA:
1093 buf = use_pack(p, &w_curs,
1094 entry->in_pack_offset + used, NULL);
1095 used_0 = 0;
1096 c = buf[used_0++];
1097 ofs = c & 127;
1098 while (c & 128) {
1099 ofs += 1;
1100 if (!ofs || MSB(ofs, 7))
1101 die("delta base offset overflow in pack for %s",
780e6e73 1102 sha1_to_hex(entry->sha1));
5c49c116
NP
1103 c = buf[used_0++];
1104 ofs = (ofs << 7) + (c & 127);
780e6e73 1105 }
5c49c116
NP
1106 if (ofs >= entry->in_pack_offset)
1107 die("delta base offset out of bound for %s",
1108 sha1_to_hex(entry->sha1));
1109 ofs = entry->in_pack_offset - ofs;
1110 if (!no_reuse_delta && !entry->preferred_base)
1111 base_ref = find_packed_object_name(p, ofs);
1112 entry->in_pack_header_size = used + used_0;
1113 break;
780e6e73 1114 }
780e6e73 1115
5c49c116
NP
1116 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1117 /*
1118 * If base_ref was set above that means we wish to
1119 * reuse delta data, and we even found that base
1120 * in the list of objects we want to pack. Goodie!
1121 *
1122 * Depth value does not matter - find_deltas() will
1123 * never consider reused delta as the base object to
1124 * deltify other objects against, in order to avoid
1125 * circular deltas.
3f9ac8d2 1126 */
780e6e73 1127 entry->type = entry->in_pack_type;
5c49c116 1128 entry->delta = base_entry;
15b4d577
JH
1129 entry->delta_sibling = base_entry->delta_child;
1130 base_entry->delta_child = entry;
5c49c116
NP
1131 unuse_pack(&w_curs);
1132 return;
1133 }
ab7cd7bb 1134
5c49c116
NP
1135 if (entry->type) {
1136 /*
1137 * This must be a delta and we already know what the
1138 * final object type is. Let's extract the actual
1139 * object size from the delta header.
1140 */
1141 entry->size = get_size_from_delta(p, &w_curs,
1142 entry->in_pack_offset + entry->in_pack_header_size);
1143 unuse_pack(&w_curs);
3f9ac8d2
JH
1144 return;
1145 }
5c49c116
NP
1146
1147 /*
1148 * No choice but to fall back to the recursive delta walk
1149 * with sha1_object_info() to find about the object type
1150 * at this point...
1151 */
1152 unuse_pack(&w_curs);
36e4d74a 1153 }
3f9ac8d2 1154
21666f1a
NP
1155 entry->type = sha1_object_info(entry->sha1, &entry->size);
1156 if (entry->type < 0)
36e4d74a
JH
1157 die("unable to get type of object %s",
1158 sha1_to_hex(entry->sha1));
3f9ac8d2
JH
1159}
1160
5c49c116
NP
1161static int pack_offset_sort(const void *_a, const void *_b)
1162{
1163 const struct object_entry *a = *(struct object_entry **)_a;
1164 const struct object_entry *b = *(struct object_entry **)_b;
1165
1166 /* avoid filesystem trashing with loose objects */
1167 if (!a->in_pack && !b->in_pack)
1168 return hashcmp(a->sha1, b->sha1);
1169
1170 if (a->in_pack < b->in_pack)
1171 return -1;
1172 if (a->in_pack > b->in_pack)
1173 return 1;
1174 return a->in_pack_offset < b->in_pack_offset ? -1 :
1175 (a->in_pack_offset > b->in_pack_offset);
1176}
1177
c323ac7d
LT
1178static void get_object_details(void)
1179{
7cadf491 1180 uint32_t i;
5c49c116
NP
1181 struct object_entry **sorted_by_offset;
1182
1183 sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1184 for (i = 0; i < nr_objects; i++)
1185 sorted_by_offset[i] = objects + i;
1186 qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
c323ac7d 1187
3f9ac8d2 1188 prepare_pack_ix();
5c49c116
NP
1189 for (i = 0; i < nr_objects; i++)
1190 check_object(sorted_by_offset[i]);
1191 free(sorted_by_offset);
c323ac7d
LT
1192}
1193
9668cf59 1194static int type_size_sort(const void *_a, const void *_b)
c323ac7d 1195{
9668cf59
NP
1196 const struct object_entry *a = *(struct object_entry **)_a;
1197 const struct object_entry *b = *(struct object_entry **)_b;
1198
c323ac7d
LT
1199 if (a->type < b->type)
1200 return -1;
1201 if (a->type > b->type)
1202 return 1;
27225f2e
LT
1203 if (a->hash < b->hash)
1204 return -1;
1205 if (a->hash > b->hash)
1206 return 1;
7a979d99
JH
1207 if (a->preferred_base < b->preferred_base)
1208 return -1;
1209 if (a->preferred_base > b->preferred_base)
1210 return 1;
c323ac7d
LT
1211 if (a->size < b->size)
1212 return -1;
1213 if (a->size > b->size)
1214 return 1;
adcc7095 1215 return a > b ? -1 : (a < b); /* newest last */
c323ac7d
LT
1216}
1217
1218struct unpacked {
1219 struct object_entry *entry;
1220 void *data;
f6c7081a 1221 struct delta_index *index;
c323ac7d
LT
1222};
1223
1224/*
521a4f4c
LT
1225 * We search for deltas _backwards_ in a list sorted by type and
1226 * by size, so that we see progressively smaller and smaller files.
1227 * That's because we prefer deltas to be from the bigger file
1228 * to the smaller - deletes are potentially cheaper, but perhaps
1229 * more importantly, the bigger file is likely the more recent
1230 * one.
c323ac7d 1231 */
f6c7081a 1232static int try_delta(struct unpacked *trg, struct unpacked *src,
560b25a8 1233 unsigned max_depth)
c323ac7d 1234{
f6c7081a
NP
1235 struct object_entry *trg_entry = trg->entry;
1236 struct object_entry *src_entry = src->entry;
560b25a8 1237 unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
21666f1a 1238 enum object_type type;
c323ac7d
LT
1239 void *delta_buf;
1240
1241 /* Don't bother doing diffs between different types */
f6c7081a 1242 if (trg_entry->type != src_entry->type)
c323ac7d
LT
1243 return -1;
1244
7a979d99
JH
1245 /* We do not compute delta to *create* objects we are not
1246 * going to pack.
1247 */
f6c7081a 1248 if (trg_entry->preferred_base)
75c42d8c 1249 return -1;
7a979d99 1250
51d1e83f
LT
1251 /*
1252 * We do not bother to try a delta that we discarded
8dbbd14e 1253 * on an earlier try, but only when reusing delta data.
51d1e83f 1254 */
8dbbd14e 1255 if (!no_reuse_delta && trg_entry->in_pack &&
e9195b58
JH
1256 trg_entry->in_pack == src_entry->in_pack &&
1257 trg_entry->in_pack_type != OBJ_REF_DELTA &&
1258 trg_entry->in_pack_type != OBJ_OFS_DELTA)
51d1e83f
LT
1259 return 0;
1260
898b14ce 1261 /* Let's not bust the allowed depth. */
f6c7081a 1262 if (src_entry->depth >= max_depth)
d116a45a 1263 return 0;
c323ac7d 1264
c3b06a69 1265 /* Now some size filtering heuristics. */
560b25a8
NP
1266 trg_size = trg_entry->size;
1267 max_size = trg_size/2 - 20;
c3b06a69
NP
1268 max_size = max_size * (max_depth - src_entry->depth) / max_depth;
1269 if (max_size == 0)
1270 return 0;
4e8da195 1271 if (trg_entry->delta && trg_entry->delta_size <= max_size)
f6c7081a
NP
1272 max_size = trg_entry->delta_size-1;
1273 src_size = src_entry->size;
560b25a8 1274 sizediff = src_size < trg_size ? trg_size - src_size : 0;
27225f2e 1275 if (sizediff >= max_size)
f527cb8c 1276 return 0;
f6c7081a 1277
560b25a8
NP
1278 /* Load data if not already done */
1279 if (!trg->data) {
21666f1a 1280 trg->data = read_sha1_file(trg_entry->sha1, &type, &sz);
560b25a8
NP
1281 if (sz != trg_size)
1282 die("object %s inconsistent object length (%lu vs %lu)",
1283 sha1_to_hex(trg_entry->sha1), sz, trg_size);
1284 }
1285 if (!src->data) {
21666f1a 1286 src->data = read_sha1_file(src_entry->sha1, &type, &sz);
560b25a8
NP
1287 if (sz != src_size)
1288 die("object %s inconsistent object length (%lu vs %lu)",
1289 sha1_to_hex(src_entry->sha1), sz, src_size);
1290 }
1291 if (!src->index) {
1292 src->index = create_delta_index(src->data, src_size);
1293 if (!src->index)
1294 die("out of memory");
1295 }
1296
1297 delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
c323ac7d 1298 if (!delta_buf)
75c42d8c 1299 return 0;
f6c7081a
NP
1300
1301 trg_entry->delta = src_entry;
1302 trg_entry->delta_size = delta_size;
1303 trg_entry->depth = src_entry->depth + 1;
c323ac7d 1304 free(delta_buf);
f6c7081a 1305 return 1;
c323ac7d
LT
1306}
1307
898b14ce 1308static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
b2504a0d 1309{
898b14ce
NP
1310 struct object_entry *child = me->delta_child;
1311 unsigned int m = n;
1312 while (child) {
1313 unsigned int c = check_delta_limit(child, n + 1);
1314 if (m < c)
1315 m = c;
1316 child = child->delta_sibling;
1317 }
1318 return m;
b2504a0d
NP
1319}
1320
d116a45a 1321static void find_deltas(struct object_entry **list, int window, int depth)
c323ac7d 1322{
7cadf491 1323 uint32_t i = nr_objects, idx = 0, processed = 0;
c323ac7d 1324 unsigned int array_size = window * sizeof(struct unpacked);
7cadf491 1325 struct unpacked *array;
898b14ce 1326 int max_depth;
c323ac7d 1327
7cadf491
SP
1328 if (!nr_objects)
1329 return;
1330 array = xmalloc(array_size);
c323ac7d 1331 memset(array, 0, array_size);
13aaf148
NP
1332 if (progress)
1333 start_progress(&progress_state, "Deltifying %u objects...", "", nr_result);
21fcd1bd 1334
7cadf491
SP
1335 do {
1336 struct object_entry *entry = list[--i];
c323ac7d 1337 struct unpacked *n = array + idx;
c323ac7d
LT
1338 int j;
1339
f0b0af1b
JH
1340 if (!entry->preferred_base)
1341 processed++;
1342
96a02f8f
NP
1343 if (progress)
1344 display_progress(&progress_state, processed);
3f9ac8d2
JH
1345
1346 if (entry->delta)
1347 /* This happens if we decided to reuse existing
ab7cd7bb 1348 * delta from a pack. "!no_reuse_delta &&" is implied.
3f9ac8d2
JH
1349 */
1350 continue;
1351
9a8b6a0a
JH
1352 if (entry->size < 50)
1353 continue;
ff45715c
NP
1354 free_delta_index(n->index);
1355 n->index = NULL;
c323ac7d 1356 free(n->data);
560b25a8 1357 n->data = NULL;
c323ac7d 1358 n->entry = entry;
ab7cd7bb 1359
898b14ce
NP
1360 /*
1361 * If the current object is at pack edge, take the depth the
1362 * objects that depend on the current object into account
1363 * otherwise they would become too deep.
1364 */
1365 max_depth = depth;
1366 if (entry->delta_child) {
1367 max_depth -= check_delta_limit(entry, 0);
1368 if (max_depth <= 0)
1369 goto next;
1370 }
1371
78817c15
LT
1372 j = window;
1373 while (--j > 0) {
7cadf491 1374 uint32_t other_idx = idx + j;
c323ac7d 1375 struct unpacked *m;
78817c15
LT
1376 if (other_idx >= window)
1377 other_idx -= window;
c323ac7d
LT
1378 m = array + other_idx;
1379 if (!m->entry)
1380 break;
898b14ce 1381 if (try_delta(n, m, max_depth) < 0)
c323ac7d
LT
1382 break;
1383 }
898b14ce 1384
70ca1a3f
JH
1385 /* if we made n a delta, and if n is already at max
1386 * depth, leaving it in the window is pointless. we
1387 * should evict it first.
70ca1a3f
JH
1388 */
1389 if (entry->delta && depth <= entry->depth)
1390 continue;
ff45715c 1391
898b14ce 1392 next:
521a4f4c
LT
1393 idx++;
1394 if (idx >= window)
1395 idx = 0;
7cadf491 1396 } while (i > 0);
adee7bdf 1397
b2504a0d 1398 if (progress)
96a02f8f 1399 stop_progress(&progress_state);
b2504a0d 1400
f6c7081a 1401 for (i = 0; i < window; ++i) {
ff45715c 1402 free_delta_index(array[i].index);
adee7bdf 1403 free(array[i].data);
f6c7081a 1404 }
adee7bdf 1405 free(array);
c323ac7d
LT
1406}
1407
f3123c4a
JH
1408static void prepare_pack(int window, int depth)
1409{
9668cf59
NP
1410 struct object_entry **delta_list;
1411 uint32_t i;
1412
3f9ac8d2 1413 get_object_details();
9668cf59
NP
1414
1415 if (!window || !depth)
1416 return;
1417
1418 delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1419 for (i = 0; i < nr_objects; i++)
1420 delta_list[i] = objects + i;
1421 qsort(delta_list, nr_objects, sizeof(*delta_list), type_size_sort);
1422 find_deltas(delta_list, window+1, depth);
1423 free(delta_list);
f3123c4a
JH
1424}
1425
4812a93a
JK
1426static int git_pack_config(const char *k, const char *v)
1427{
1428 if(!strcmp(k, "pack.window")) {
1429 window = git_config_int(k, v);
1430 return 0;
1431 }
842aaf93
TT
1432 if(!strcmp(k, "pack.depth")) {
1433 depth = git_config_int(k, v);
1434 return 0;
1435 }
960ccca6
DH
1436 if (!strcmp(k, "pack.compression")) {
1437 int level = git_config_int(k, v);
1438 if (level == -1)
1439 level = Z_DEFAULT_COMPRESSION;
1440 else if (level < 0 || level > Z_BEST_COMPRESSION)
1441 die("bad pack compression level %d", level);
1442 pack_compression_level = level;
1443 pack_compression_seen = 1;
1444 return 0;
1445 }
4812a93a
JK
1446 return git_default_config(k, v);
1447}
1448
b5d97e6b 1449static void read_object_list_from_stdin(void)
c323ac7d 1450{
b5d97e6b
JH
1451 char line[40 + 1 + PATH_MAX + 2];
1452 unsigned char sha1[20];
b2504a0d 1453
da93d12b 1454 for (;;) {
da93d12b
LT
1455 if (!fgets(line, sizeof(line), stdin)) {
1456 if (feof(stdin))
1457 break;
1458 if (!ferror(stdin))
1459 die("fgets returned NULL, not EOF, not error!");
687dd75c
JH
1460 if (errno != EINTR)
1461 die("fgets: %s", strerror(errno));
1462 clearerr(stdin);
1463 continue;
da93d12b 1464 }
7a979d99
JH
1465 if (line[0] == '-') {
1466 if (get_sha1_hex(line+1, sha1))
1467 die("expected edge sha1, got garbage:\n %s",
b5d97e6b 1468 line);
8d1d8f83 1469 add_preferred_base(sha1);
7a979d99 1470 continue;
21fcd1bd 1471 }
c323ac7d 1472 if (get_sha1_hex(line, sha1))
ef07618f 1473 die("expected sha1, got garbage:\n %s", line);
b5d97e6b 1474
bc32fed5
JH
1475 add_preferred_base_object(line+41);
1476 add_object_entry(sha1, 0, line+41, 0);
c323ac7d 1477 }
b5d97e6b
JH
1478}
1479
b5d97e6b
JH
1480static void show_commit(struct commit *commit)
1481{
bc32fed5 1482 add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
b5d97e6b
JH
1483}
1484
1485static void show_object(struct object_array_entry *p)
1486{
bc32fed5
JH
1487 add_preferred_base_object(p->name);
1488 add_object_entry(p->item->sha1, p->item->type, p->name, 0);
b5d97e6b
JH
1489}
1490
8d1d8f83
JH
1491static void show_edge(struct commit *commit)
1492{
1493 add_preferred_base(commit->object.sha1);
1494}
1495
1496static void get_object_list(int ac, const char **av)
b5d97e6b
JH
1497{
1498 struct rev_info revs;
1499 char line[1000];
b5d97e6b
JH
1500 int flags = 0;
1501
b5d97e6b
JH
1502 init_revisions(&revs, NULL);
1503 save_commit_buffer = 0;
1504 track_object_refs = 0;
1505 setup_revisions(ac, av, &revs, NULL);
1506
b5d97e6b
JH
1507 while (fgets(line, sizeof(line), stdin) != NULL) {
1508 int len = strlen(line);
1509 if (line[len - 1] == '\n')
1510 line[--len] = 0;
1511 if (!len)
1512 break;
1513 if (*line == '-') {
1514 if (!strcmp(line, "--not")) {
1515 flags ^= UNINTERESTING;
1516 continue;
1517 }
1518 die("not a rev '%s'", line);
1519 }
1520 if (handle_revision_arg(line, &revs, flags, 1))
1521 die("bad revision '%s'", line);
1522 }
1523
1524 prepare_revision_walk(&revs);
8d1d8f83 1525 mark_edges_uninteresting(revs.commits, &revs, show_edge);
b5d97e6b
JH
1526 traverse_commit_list(&revs, show_commit, show_object);
1527}
1528
b6b32ccb
JH
1529static int adjust_perm(const char *path, mode_t mode)
1530{
1531 if (chmod(path, mode))
1532 return -1;
1533 return adjust_shared_perm(path);
1534}
1535
b5d97e6b
JH
1536int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1537{
b5d97e6b 1538 int use_internal_rev_list = 0;
8d1d8f83 1539 int thin = 0;
7cadf491 1540 uint32_t i;
81a216a5
NP
1541 off_t last_obj_offset;
1542 const char *base_name = NULL;
ffa84ffb
RD
1543 const char **rp_av;
1544 int rp_ac_alloc = 64;
8d1d8f83
JH
1545 int rp_ac;
1546
ffa84ffb
RD
1547 rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1548
8d1d8f83
JH
1549 rp_av[0] = "pack-objects";
1550 rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1551 rp_ac = 2;
b5d97e6b
JH
1552
1553 git_config(git_pack_config);
960ccca6
DH
1554 if (!pack_compression_seen && core_compression_seen)
1555 pack_compression_level = core_compression_level;
b5d97e6b
JH
1556
1557 progress = isatty(2);
1558 for (i = 1; i < argc; i++) {
1559 const char *arg = argv[i];
1560
1561 if (*arg != '-')
1562 break;
1563
1564 if (!strcmp("--non-empty", arg)) {
1565 non_empty = 1;
1566 continue;
1567 }
1568 if (!strcmp("--local", arg)) {
1569 local = 1;
1570 continue;
1571 }
b5d97e6b
JH
1572 if (!strcmp("--incremental", arg)) {
1573 incremental = 1;
1574 continue;
1575 }
960ccca6
DH
1576 if (!prefixcmp(arg, "--compression=")) {
1577 char *end;
1578 int level = strtoul(arg+14, &end, 0);
1579 if (!arg[14] || *end)
1580 usage(pack_usage);
1581 if (level == -1)
1582 level = Z_DEFAULT_COMPRESSION;
1583 else if (level < 0 || level > Z_BEST_COMPRESSION)
1584 die("bad pack compression level %d", level);
1585 pack_compression_level = level;
1586 continue;
1587 }
599065a3 1588 if (!prefixcmp(arg, "--window=")) {
b5d97e6b
JH
1589 char *end;
1590 window = strtoul(arg+9, &end, 0);
1591 if (!arg[9] || *end)
1592 usage(pack_usage);
1593 continue;
1594 }
599065a3 1595 if (!prefixcmp(arg, "--depth=")) {
b5d97e6b
JH
1596 char *end;
1597 depth = strtoul(arg+8, &end, 0);
1598 if (!arg[8] || *end)
1599 usage(pack_usage);
1600 continue;
1601 }
1602 if (!strcmp("--progress", arg)) {
1603 progress = 1;
1604 continue;
1605 }
231f240b
NP
1606 if (!strcmp("--all-progress", arg)) {
1607 progress = 2;
1608 continue;
1609 }
b5d97e6b
JH
1610 if (!strcmp("-q", arg)) {
1611 progress = 0;
1612 continue;
1613 }
1614 if (!strcmp("--no-reuse-delta", arg)) {
1615 no_reuse_delta = 1;
1616 continue;
1617 }
fa736f72
NP
1618 if (!strcmp("--no-reuse-object", arg)) {
1619 no_reuse_object = no_reuse_delta = 1;
1620 continue;
1621 }
be6b1914 1622 if (!strcmp("--delta-base-offset", arg)) {
780e6e73 1623 allow_ofs_delta = 1;
be6b1914
NP
1624 continue;
1625 }
b5d97e6b
JH
1626 if (!strcmp("--stdout", arg)) {
1627 pack_to_stdout = 1;
1628 continue;
1629 }
1630 if (!strcmp("--revs", arg)) {
1631 use_internal_rev_list = 1;
1632 continue;
1633 }
8d1d8f83 1634 if (!strcmp("--unpacked", arg) ||
599065a3 1635 !prefixcmp(arg, "--unpacked=") ||
63049292 1636 !strcmp("--reflog", arg) ||
8d1d8f83
JH
1637 !strcmp("--all", arg)) {
1638 use_internal_rev_list = 1;
ffa84ffb
RD
1639 if (rp_ac >= rp_ac_alloc - 1) {
1640 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1641 rp_av = xrealloc(rp_av,
1642 rp_ac_alloc * sizeof(*rp_av));
1643 }
8d1d8f83 1644 rp_av[rp_ac++] = arg;
b5d97e6b
JH
1645 continue;
1646 }
8d1d8f83
JH
1647 if (!strcmp("--thin", arg)) {
1648 use_internal_rev_list = 1;
1649 thin = 1;
1650 rp_av[1] = "--objects-edge";
b5d97e6b 1651 continue;
4ba7d711
NP
1652 }
1653 if (!prefixcmp(arg, "--index-version=")) {
1654 char *c;
1655 index_default_version = strtoul(arg + 16, &c, 10);
1656 if (index_default_version > 2)
1657 die("bad %s", arg);
1658 if (*c == ',')
1659 index_off32_limit = strtoul(c+1, &c, 0);
1660 if (*c || index_off32_limit & 0x80000000)
1661 die("bad %s", arg);
1662 continue;
b5d97e6b
JH
1663 }
1664 usage(pack_usage);
1665 }
1666
1667 /* Traditionally "pack-objects [options] base extra" failed;
1668 * we would however want to take refs parameter that would
1669 * have been given to upstream rev-list ourselves, which means
1670 * we somehow want to say what the base name is. So the
1671 * syntax would be:
1672 *
1673 * pack-objects [options] base <refs...>
1674 *
1675 * in other words, we would treat the first non-option as the
1676 * base_name and send everything else to the internal revision
1677 * walker.
1678 */
1679
1680 if (!pack_to_stdout)
1681 base_name = argv[i++];
1682
1683 if (pack_to_stdout != !base_name)
1684 usage(pack_usage);
1685
8d1d8f83
JH
1686 if (!pack_to_stdout && thin)
1687 die("--thin cannot be used to build an indexable pack.");
b5d97e6b
JH
1688
1689 prepare_packed_git();
1690
13aaf148
NP
1691 if (progress)
1692 start_progress(&progress_state, "Generating pack...",
1693 "Counting objects: ", 0);
b5d97e6b
JH
1694 if (!use_internal_rev_list)
1695 read_object_list_from_stdin();
8d1d8f83
JH
1696 else {
1697 rp_av[rp_ac] = NULL;
1698 get_object_list(rp_ac, rp_av);
1699 }
96a02f8f
NP
1700 if (progress) {
1701 stop_progress(&progress_state);
7cadf491 1702 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
96a02f8f
NP
1703 }
1704
f0b0af1b 1705 if (non_empty && !nr_result)
1c4a2912 1706 return 0;
7a979d99 1707 if (progress && (nr_objects != nr_result))
7cadf491 1708 fprintf(stderr, "Result has %u objects.\n", nr_result);
f7ae6a93
NP
1709 if (nr_result)
1710 prepare_pack(window, depth);
f7ae6a93
NP
1711 last_obj_offset = write_pack_file();
1712 if (!pack_to_stdout) {
81a216a5 1713 unsigned char object_list_sha1[20];
b6b32ccb
JH
1714 mode_t mode = umask(0);
1715
1716 umask(mode);
d83c9af5 1717 mode = 0444 & ~mode;
b6b32ccb 1718
81a216a5
NP
1719 write_index_file(last_obj_offset, object_list_sha1);
1720 snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
1721 base_name, sha1_to_hex(object_list_sha1));
b6b32ccb 1722 if (adjust_perm(pack_tmp_name, mode))
aef5aedd
JH
1723 die("unable to make temporary pack file readable: %s",
1724 strerror(errno));
81a216a5
NP
1725 if (rename(pack_tmp_name, tmpname))
1726 die("unable to rename temporary pack file: %s",
1727 strerror(errno));
1728 snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
1729 base_name, sha1_to_hex(object_list_sha1));
b6b32ccb 1730 if (adjust_perm(idx_tmp_name, mode))
aef5aedd
JH
1731 die("unable to make temporary index file readable: %s",
1732 strerror(errno));
81a216a5
NP
1733 if (rename(idx_tmp_name, tmpname))
1734 die("unable to rename temporary index file: %s",
1735 strerror(errno));
f7ae6a93 1736 puts(sha1_to_hex(object_list_sha1));
5f3de58f 1737 }
ab7cd7bb 1738 if (progress)
7cadf491 1739 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
67c08ce1 1740 written, written_delta, reused, reused_delta);
c323ac7d
LT
1741 return 0;
1742}