builtin-pack-objects.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "object.h"
   4 #include "blob.h"
   5 #include "commit.h"
   6 #include "tag.h"
   7 #include "tree.h"
   8 #include "delta.h"
   9 #include "pack.h"
  10 #include "csum-file.h"
  11 #include "tree-walk.h"
  12 #include "diff.h"
  13 #include "revision.h"
  14 #include "list-objects.h"
  15
  16 static const char pack_usage[] = "\
  17 git-pack-objects [{ -q | --progress | --all-progress }] \n\
  18         [--local] [--incremental] [--window=N] [--depth=N] \n\
  19         [--no-reuse-delta] [--delta-base-offset] [--non-empty] \n\
  20         [--revs [--unpacked | --all]*] [--reflog] [--stdout | base-name] \n\
  21         [<ref-list | <object-list]";
  22
  23 struct object_entry {
  24         unsigned char sha1[20];
  25         unsigned long size;     /* uncompressed size */
  26         off_t offset;   /* offset into the final pack file;
  27                                  * nonzero if already written.
  28                                  */
  29         unsigned int depth;     /* delta depth */
  30         unsigned int delta_limit;       /* base adjustment for in-pack delta */
  31         unsigned int hash;      /* name hint hash */
  32         enum object_type type;
  33         enum object_type in_pack_type;  /* could be delta */
  34         unsigned long delta_size;       /* delta data size (uncompressed) */
  35 #define in_pack_header_size delta_size  /* only when reusing pack data */
  36         struct object_entry *delta;     /* delta base object */
  37         struct packed_git *in_pack;     /* already in pack */
  38         off_t in_pack_offset;
  39         struct object_entry *delta_child; /* deltified objects who bases me */
  40         struct object_entry *delta_sibling; /* other deltified objects who
  41                                              * uses the same base as me
  42                                              */
  43         int preferred_base;     /* we do not pack this, but is encouraged to
  44                                  * be used as the base objectto delta huge
  45                                  * objects against.
  46                                  */
  47         uint32_t crc32;         /* crc of raw pack data for this object */
  48 };
  49
  50 /*
  51  * Objects we are going to pack are collected in objects array (dynamically
  52  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  53  * in the order we see -- typically rev-list --objects order that gives us
  54  * nice "minimum seek" order.
  55  *
  56  * sorted-by-sha ans sorted-by-type are arrays of pointers that point at
  57  * elements in the objects array.  The former is used to build the pack
  58  * index (lists object names in the ascending order to help offset lookup),
  59  * and the latter is used to group similar things together by try_delta()
  60  * heuristics.
  61  */
  62
  63 static unsigned char object_list_sha1[20];
  64 static int non_empty;
  65 static int no_reuse_delta;
  66 static int local;
  67 static int incremental;
  68 static int allow_ofs_delta;
  69
  70 static struct object_entry **sorted_by_sha, **sorted_by_type;
  71 static struct object_entry *objects;
  72 static uint32_t nr_objects, nr_alloc, nr_result;
  73 static const char *base_name;
  74 static unsigned char pack_file_sha1[20];
  75 static int progress = 1;
  76 static volatile sig_atomic_t progress_update;
  77 static int window = 10;
  78 static int pack_to_stdout;
  79 static int num_preferred_base;
  80
  81 /*
  82  * The object names in objects array are hashed with this hashtable,
  83  * to help looking up the entry by object name.  Binary search from
  84  * sorted_by_sha is also possible but this was easier to code and faster.
  85  * This hashtable is built after all the objects are seen.
  86  */
  87 static int *object_ix;
  88 static int object_ix_hashsz;
  89
  90 /*
  91  * Pack index for existing packs give us easy access to the offsets into
  92  * corresponding pack file where each object's data starts, but the entries
  93  * do not store the size of the compressed representation (uncompressed
  94  * size is easily available by examining the pack entry header).  It is
  95  * also rather expensive to find the sha1 for an object given its offset.
  96  *
  97  * We build a hashtable of existing packs (pack_revindex), and keep reverse
  98  * index here -- pack index file is sorted by object name mapping to offset;
  99  * this pack_revindex[].revindex array is a list of offset/index_nr pairs
 100  * ordered by offset, so if you know the offset of an object, next offset
 101  * is where its packed representation ends and the index_nr can be used to
 102  * get the object sha1 from the main index.
 103  */
 104 struct revindex_entry {
 105         off_t offset;
 106         unsigned int nr;
 107 };
 108 struct pack_revindex {
 109         struct packed_git *p;
 110         struct revindex_entry *revindex;
 111 };
 112 static struct  pack_revindex *pack_revindex;
 113 static int pack_revindex_hashsz;
 114
 115 /*
 116  * stats
 117  */
 118 static uint32_t written, written_delta;
 119 static uint32_t reused, reused_delta;
 120
 121 static int pack_revindex_ix(struct packed_git *p)
 122 {
 123         unsigned long ui = (unsigned long)p;
 124         int i;
 125
 126         ui = ui ^ (ui >> 16); /* defeat structure alignment */
 127         i = (int)(ui % pack_revindex_hashsz);
 128         while (pack_revindex[i].p) {
 129                 if (pack_revindex[i].p == p)
 130                         return i;
 131                 if (++i == pack_revindex_hashsz)
 132                         i = 0;
 133         }
 134         return -1 - i;
 135 }
 136
 137 static void prepare_pack_ix(void)
 138 {
 139         int num;
 140         struct packed_git *p;
 141         for (num = 0, p = packed_git; p; p = p->next)
 142                 num++;
 143         if (!num)
 144                 return;
 145         pack_revindex_hashsz = num * 11;
 146         pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
 147         for (p = packed_git; p; p = p->next) {
 148                 num = pack_revindex_ix(p);
 149                 num = - 1 - num;
 150                 pack_revindex[num].p = p;
 151         }
 152         /* revindex elements are lazily initialized */
 153 }
 154
 155 static int cmp_offset(const void *a_, const void *b_)
 156 {
 157         const struct revindex_entry *a = a_;
 158         const struct revindex_entry *b = b_;
 159         return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0;
 160 }
 161
 162 /*
 163  * Ordered list of offsets of objects in the pack.
 164  */
 165 static void prepare_pack_revindex(struct pack_revindex *rix)
 166 {
 167         struct packed_git *p = rix->p;
 168         int num_ent = p->num_objects;
 169         int i;
 170         const char *index = p->index_data;
 171
 172         rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
 173         index += 4 * 256;
 174
 175         if (p->index_version > 1) {
 176                 const uint32_t *off_32 =
 177                         (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
 178                 const uint32_t *off_64 = off_32 + p->num_objects;
 179                 for (i = 0; i < num_ent; i++) {
 180                         uint32_t off = ntohl(*off_32++);
 181                         if (!(off & 0x80000000)) {
 182                                 rix->revindex[i].offset = off;
 183                         } else {
 184                                 rix->revindex[i].offset =
 185                                         ((uint64_t)ntohl(*off_64++)) << 32;
 186                                 rix->revindex[i].offset |=
 187                                         ntohl(*off_64++);
 188                         }
 189                         rix->revindex[i].nr = i;
 190                 }
 191         } else {
 192                 for (i = 0; i < num_ent; i++) {
 193                         uint32_t hl = *((uint32_t *)(index + 24 * i));
 194                         rix->revindex[i].offset = ntohl(hl);
 195                         rix->revindex[i].nr = i;
 196                 }
 197         }
 198
 199         /* This knows the pack format -- the 20-byte trailer
 200          * follows immediately after the last object data.
 201          */
 202         rix->revindex[num_ent].offset = p->pack_size - 20;
 203         rix->revindex[num_ent].nr = -1;
 204         qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
 205 }
 206
 207 static struct revindex_entry * find_packed_object(struct packed_git *p,
 208                                                   off_t ofs)
 209 {
 210         int num;
 211         int lo, hi;
 212         struct pack_revindex *rix;
 213         struct revindex_entry *revindex;
 214         num = pack_revindex_ix(p);
 215         if (num < 0)
 216                 die("internal error: pack revindex uninitialized");
 217         rix = &pack_revindex[num];
 218         if (!rix->revindex)
 219                 prepare_pack_revindex(rix);
 220         revindex = rix->revindex;
 221         lo = 0;
 222         hi = p->num_objects + 1;
 223         do {
 224                 int mi = (lo + hi) / 2;
 225                 if (revindex[mi].offset == ofs) {
 226                         return revindex + mi;
 227                 }
 228                 else if (ofs < revindex[mi].offset)
 229                         hi = mi;
 230                 else
 231                         lo = mi + 1;
 232         } while (lo < hi);
 233         die("internal error: pack revindex corrupt");
 234 }
 235
 236 static off_t find_packed_object_size(struct packed_git *p, off_t ofs)
 237 {
 238         struct revindex_entry *entry = find_packed_object(p, ofs);
 239         return entry[1].offset - ofs;
 240 }
 241
 242 static const unsigned char *find_packed_object_name(struct packed_git *p,
 243                                                     off_t ofs)
 244 {
 245         struct revindex_entry *entry = find_packed_object(p, ofs);
 246         return nth_packed_object_sha1(p, entry->nr);
 247 }
 248
 249 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 250 {
 251         unsigned long othersize, delta_size;
 252         enum object_type type;
 253         void *otherbuf = read_sha1_file(entry->delta->sha1, &type, &othersize);
 254         void *delta_buf;
 255
 256         if (!otherbuf)
 257                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
 258         delta_buf = diff_delta(otherbuf, othersize,
 259                                buf, size, &delta_size, 0);
 260         if (!delta_buf || delta_size != entry->delta_size)
 261                 die("delta size changed");
 262         free(buf);
 263         free(otherbuf);
 264         return delta_buf;
 265 }
 266
 267 /*
 268  * The per-object header is a pretty dense thing, which is
 269  *  - first byte: low four bits are "size", then three bits of "type",
 270  *    and the high bit is "size continues".
 271  *  - each byte afterwards: low seven bits are size continuation,
 272  *    with the high bit being "size continues"
 273  */
 274 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 275 {
 276         int n = 1;
 277         unsigned char c;
 278
 279         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 280                 die("bad type %d", type);
 281
 282         c = (type << 4) | (size & 15);
 283         size >>= 4;
 284         while (size) {
 285                 *hdr++ = c | 0x80;
 286                 c = size & 0x7f;
 287                 size >>= 7;
 288                 n++;
 289         }
 290         *hdr = c;
 291         return n;
 292 }
 293
 294 /*
 295  * we are going to reuse the existing object data as is.  make
 296  * sure it is not corrupt.
 297  */
 298 static int check_pack_inflate(struct packed_git *p,
 299                 struct pack_window **w_curs,
 300                 off_t offset,
 301                 off_t len,
 302                 unsigned long expect)
 303 {
 304         z_stream stream;
 305         unsigned char fakebuf[4096], *in;
 306         int st;
 307
 308         memset(&stream, 0, sizeof(stream));
 309         inflateInit(&stream);
 310         do {
 311                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 312                 stream.next_in = in;
 313                 stream.next_out = fakebuf;
 314                 stream.avail_out = sizeof(fakebuf);
 315                 st = inflate(&stream, Z_FINISH);
 316                 offset += stream.next_in - in;
 317         } while (st == Z_OK || st == Z_BUF_ERROR);
 318         inflateEnd(&stream);
 319         return (st == Z_STREAM_END &&
 320                 stream.total_out == expect &&
 321                 stream.total_in == len) ? 0 : -1;
 322 }
 323
 324 static void copy_pack_data(struct sha1file *f,
 325                 struct packed_git *p,
 326                 struct pack_window **w_curs,
 327                 off_t offset,
 328                 off_t len)
 329 {
 330         unsigned char *in;
 331         unsigned int avail;
 332
 333         while (len) {
 334                 in = use_pack(p, w_curs, offset, &avail);
 335                 if (avail > len)
 336                         avail = (unsigned int)len;
 337                 sha1write(f, in, avail);
 338                 offset += avail;
 339                 len -= avail;
 340         }
 341 }
 342
 343 static int check_loose_inflate(unsigned char *data, unsigned long len, unsigned long expect)
 344 {
 345         z_stream stream;
 346         unsigned char fakebuf[4096];
 347         int st;
 348
 349         memset(&stream, 0, sizeof(stream));
 350         stream.next_in = data;
 351         stream.avail_in = len;
 352         stream.next_out = fakebuf;
 353         stream.avail_out = sizeof(fakebuf);
 354         inflateInit(&stream);
 355
 356         while (1) {
 357                 st = inflate(&stream, Z_FINISH);
 358                 if (st == Z_STREAM_END || st == Z_OK) {
 359                         st = (stream.total_out == expect &&
 360                               stream.total_in == len) ? 0 : -1;
 361                         break;
 362                 }
 363                 if (st != Z_BUF_ERROR) {
 364                         st = -1;
 365                         break;
 366                 }
 367                 stream.next_out = fakebuf;
 368                 stream.avail_out = sizeof(fakebuf);
 369         }
 370         inflateEnd(&stream);
 371         return st;
 372 }
 373
 374 static int revalidate_loose_object(struct object_entry *entry,
 375                                    unsigned char *map,
 376                                    unsigned long mapsize)
 377 {
 378         /* we already know this is a loose object with new type header. */
 379         enum object_type type;
 380         unsigned long size, used;
 381
 382         if (pack_to_stdout)
 383                 return 0;
 384
 385         used = unpack_object_header_gently(map, mapsize, &type, &size);
 386         if (!used)
 387                 return -1;
 388         map += used;
 389         mapsize -= used;
 390         return check_loose_inflate(map, mapsize, size);
 391 }
 392
 393 static unsigned long write_object(struct sha1file *f,
 394                                   struct object_entry *entry)
 395 {
 396         unsigned long size;
 397         enum object_type type;
 398         void *buf;
 399         unsigned char header[10];
 400         unsigned hdrlen;
 401         off_t datalen;
 402         enum object_type obj_type;
 403         int to_reuse = 0;
 404
 405         if (!pack_to_stdout)
 406                 crc32_begin(f);
 407
 408         obj_type = entry->type;
 409         if (! entry->in_pack)
 410                 to_reuse = 0;   /* can't reuse what we don't have */
 411         else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
 412                 to_reuse = 1;   /* check_object() decided it for us */
 413         else if (obj_type != entry->in_pack_type)
 414                 to_reuse = 0;   /* pack has delta which is unusable */
 415         else if (entry->delta)
 416                 to_reuse = 0;   /* we want to pack afresh */
 417         else
 418                 to_reuse = 1;   /* we have it in-pack undeltified,
 419                                  * and we do not need to deltify it.
 420                                  */
 421
 422         if (!entry->in_pack && !entry->delta) {
 423                 unsigned char *map;
 424                 unsigned long mapsize;
 425                 map = map_sha1_file(entry->sha1, &mapsize);
 426                 if (map && !legacy_loose_object(map)) {
 427                         /* We can copy straight into the pack file */
 428                         if (revalidate_loose_object(entry, map, mapsize))
 429                                 die("corrupt loose object %s",
 430                                     sha1_to_hex(entry->sha1));
 431                         sha1write(f, map, mapsize);
 432                         munmap(map, mapsize);
 433                         written++;
 434                         reused++;
 435                         return mapsize;
 436                 }
 437                 if (map)
 438                         munmap(map, mapsize);
 439         }
 440
 441         if (!to_reuse) {
 442                 buf = read_sha1_file(entry->sha1, &type, &size);
 443                 if (!buf)
 444                         die("unable to read %s", sha1_to_hex(entry->sha1));
 445                 if (size != entry->size)
 446                         die("object %s size inconsistency (%lu vs %lu)",
 447                             sha1_to_hex(entry->sha1), size, entry->size);
 448                 if (entry->delta) {
 449                         buf = delta_against(buf, size, entry);
 450                         size = entry->delta_size;
 451                         obj_type = (allow_ofs_delta && entry->delta->offset) ?
 452                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 453                 }
 454                 /*
 455                  * The object header is a byte of 'type' followed by zero or
 456                  * more bytes of length.
 457                  */
 458                 hdrlen = encode_header(obj_type, size, header);
 459                 sha1write(f, header, hdrlen);
 460
 461                 if (obj_type == OBJ_OFS_DELTA) {
 462                         /*
 463                          * Deltas with relative base contain an additional
 464                          * encoding of the relative offset for the delta
 465                          * base from this object's position in the pack.
 466                          */
 467                         off_t ofs = entry->offset - entry->delta->offset;
 468                         unsigned pos = sizeof(header) - 1;
 469                         header[pos] = ofs & 127;
 470                         while (ofs >>= 7)
 471                                 header[--pos] = 128 | (--ofs & 127);
 472                         sha1write(f, header + pos, sizeof(header) - pos);
 473                         hdrlen += sizeof(header) - pos;
 474                 } else if (obj_type == OBJ_REF_DELTA) {
 475                         /*
 476                          * Deltas with a base reference contain
 477                          * an additional 20 bytes for the base sha1.
 478                          */
 479                         sha1write(f, entry->delta->sha1, 20);
 480                         hdrlen += 20;
 481                 }
 482                 datalen = sha1write_compressed(f, buf, size);
 483                 free(buf);
 484         }
 485         else {
 486                 struct packed_git *p = entry->in_pack;
 487                 struct pack_window *w_curs = NULL;
 488                 off_t offset;
 489
 490                 if (entry->delta) {
 491                         obj_type = (allow_ofs_delta && entry->delta->offset) ?
 492                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 493                         reused_delta++;
 494                 }
 495                 hdrlen = encode_header(obj_type, entry->size, header);
 496                 sha1write(f, header, hdrlen);
 497                 if (obj_type == OBJ_OFS_DELTA) {
 498                         off_t ofs = entry->offset - entry->delta->offset;
 499                         unsigned pos = sizeof(header) - 1;
 500                         header[pos] = ofs & 127;
 501                         while (ofs >>= 7)
 502                                 header[--pos] = 128 | (--ofs & 127);
 503                         sha1write(f, header + pos, sizeof(header) - pos);
 504                         hdrlen += sizeof(header) - pos;
 505                 } else if (obj_type == OBJ_REF_DELTA) {
 506                         sha1write(f, entry->delta->sha1, 20);
 507                         hdrlen += 20;
 508                 }
 509
 510                 offset = entry->in_pack_offset + entry->in_pack_header_size;
 511                 datalen = find_packed_object_size(p, entry->in_pack_offset)
 512                                 - entry->in_pack_header_size;
 513                 if (!pack_to_stdout && check_pack_inflate(p, &w_curs,
 514                                 offset, datalen, entry->size))
 515                         die("corrupt delta in pack %s", sha1_to_hex(entry->sha1));
 516                 copy_pack_data(f, p, &w_curs, offset, datalen);
 517                 unuse_pack(&w_curs);
 518                 reused++;
 519         }
 520         if (entry->delta)
 521                 written_delta++;
 522         written++;
 523         if (!pack_to_stdout)
 524                 entry->crc32 = crc32_end(f);
 525         return hdrlen + datalen;
 526 }
 527
 528 static off_t write_one(struct sha1file *f,
 529                                struct object_entry *e,
 530                                off_t offset)
 531 {
 532         unsigned long size;
 533
 534         /* offset is non zero if object is written already. */
 535         if (e->offset || e->preferred_base)
 536                 return offset;
 537
 538         /* if we are deltified, write out base object first. */
 539         if (e->delta)
 540                 offset = write_one(f, e->delta, offset);
 541
 542         e->offset = offset;
 543         size = write_object(f, e);
 544
 545         /* make sure off_t is sufficiently large not to wrap */
 546         if (offset > offset + size)
 547                 die("pack too large for current definition of off_t");
 548         return offset + size;
 549 }
 550
 551 static off_t write_pack_file(void)
 552 {
 553         uint32_t i;
 554         struct sha1file *f;
 555         off_t offset, last_obj_offset = 0;
 556         struct pack_header hdr;
 557         unsigned last_percent = 999;
 558         int do_progress = progress;
 559
 560         if (!base_name) {
 561                 f = sha1fd(1, "<stdout>");
 562                 do_progress >>= 1;
 563         }
 564         else
 565                 f = sha1create("%s-%s.%s", base_name,
 566                                sha1_to_hex(object_list_sha1), "pack");
 567         if (do_progress)
 568                 fprintf(stderr, "Writing %u objects.\n", nr_result);
 569
 570         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 571         hdr.hdr_version = htonl(PACK_VERSION);
 572         hdr.hdr_entries = htonl(nr_result);
 573         sha1write(f, &hdr, sizeof(hdr));
 574         offset = sizeof(hdr);
 575         if (!nr_result)
 576                 goto done;
 577         for (i = 0; i < nr_objects; i++) {
 578                 last_obj_offset = offset;
 579                 offset = write_one(f, objects + i, offset);
 580                 if (do_progress) {
 581                         unsigned percent = written * 100 / nr_result;
 582                         if (progress_update || percent != last_percent) {
 583                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
 584                                         percent, written, nr_result);
 585                                 progress_update = 0;
 586                                 last_percent = percent;
 587                         }
 588                 }
 589         }
 590         if (do_progress)
 591                 fputc('\n', stderr);
 592  done:
 593         if (written != nr_result)
 594                 die("wrote %u objects while expecting %u", written, nr_result);
 595         sha1close(f, pack_file_sha1, 1);
 596
 597         return last_obj_offset;
 598 }
 599
 600 static uint32_t index_default_version = 1;
 601 static uint32_t index_off32_limit = 0x7fffffff;
 602
 603 static void write_index_file(off_t last_obj_offset)
 604 {
 605         uint32_t i;
 606         struct sha1file *f = sha1create("%s-%s.%s", base_name,
 607                                         sha1_to_hex(object_list_sha1), "idx");
 608         struct object_entry **list = sorted_by_sha;
 609         struct object_entry **last = list + nr_result;
 610         uint32_t array[256];
 611         uint32_t index_version;
 612
 613         /* if last object's offset is >= 2^31 we should use index V2 */
 614         index_version = (last_obj_offset >> 31) ? 2 : index_default_version;
 615
 616         /* index versions 2 and above need a header */
 617         if (index_version >= 2) {
 618                 struct pack_idx_header hdr;
 619                 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 620                 hdr.idx_version = htonl(index_version);
 621                 sha1write(f, &hdr, sizeof(hdr));
 622         }
 623
 624         /*
 625          * Write the first-level table (the list is sorted,
 626          * but we use a 256-entry lookup to be able to avoid
 627          * having to do eight extra binary search iterations).
 628          */
 629         for (i = 0; i < 256; i++) {
 630                 struct object_entry **next = list;
 631                 while (next < last) {
 632                         struct object_entry *entry = *next;
 633                         if (entry->sha1[0] != i)
 634                                 break;
 635                         next++;
 636                 }
 637                 array[i] = htonl(next - sorted_by_sha);
 638                 list = next;
 639         }
 640         sha1write(f, array, 256 * 4);
 641
 642         /*
 643          * Write the actual SHA1 entries..
 644          */
 645         list = sorted_by_sha;
 646         for (i = 0; i < nr_result; i++) {
 647                 struct object_entry *entry = *list++;
 648                 if (index_version < 2) {
 649                         uint32_t offset = htonl(entry->offset);
 650                         sha1write(f, &offset, 4);
 651                 }
 652                 sha1write(f, entry->sha1, 20);
 653         }
 654
 655         if (index_version >= 2) {
 656                 unsigned int nr_large_offset = 0;
 657
 658                 /* write the crc32 table */
 659                 list = sorted_by_sha;
 660                 for (i = 0; i < nr_objects; i++) {
 661                         struct object_entry *entry = *list++;
 662                         uint32_t crc32_val = htonl(entry->crc32);
 663                         sha1write(f, &crc32_val, 4);
 664                 }
 665
 666                 /* write the 32-bit offset table */
 667                 list = sorted_by_sha;
 668                 for (i = 0; i < nr_objects; i++) {
 669                         struct object_entry *entry = *list++;
 670                         uint32_t offset = (entry->offset <= index_off32_limit) ?
 671                                 entry->offset : (0x80000000 | nr_large_offset++);
 672                         offset = htonl(offset);
 673                         sha1write(f, &offset, 4);
 674                 }
 675
 676                 /* write the large offset table */
 677                 list = sorted_by_sha;
 678                 while (nr_large_offset) {
 679                         struct object_entry *entry = *list++;
 680                         uint64_t offset = entry->offset;
 681                         if (offset > index_off32_limit) {
 682                                 uint32_t split[2];
 683                                 split[0]        = htonl(offset >> 32);
 684                                 split[1] = htonl(offset & 0xffffffff);
 685                                 sha1write(f, split, 8);
 686                                 nr_large_offset--;
 687                         }
 688                 }
 689         }
 690
 691         sha1write(f, pack_file_sha1, 20);
 692         sha1close(f, NULL, 1);
 693 }
 694
 695 static int locate_object_entry_hash(const unsigned char *sha1)
 696 {
 697         int i;
 698         unsigned int ui;
 699         memcpy(&ui, sha1, sizeof(unsigned int));
 700         i = ui % object_ix_hashsz;
 701         while (0 < object_ix[i]) {
 702                 if (!hashcmp(sha1, objects[object_ix[i] - 1].sha1))
 703                         return i;
 704                 if (++i == object_ix_hashsz)
 705                         i = 0;
 706         }
 707         return -1 - i;
 708 }
 709
 710 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 711 {
 712         int i;
 713
 714         if (!object_ix_hashsz)
 715                 return NULL;
 716
 717         i = locate_object_entry_hash(sha1);
 718         if (0 <= i)
 719                 return &objects[object_ix[i]-1];
 720         return NULL;
 721 }
 722
 723 static void rehash_objects(void)
 724 {
 725         uint32_t i;
 726         struct object_entry *oe;
 727
 728         object_ix_hashsz = nr_objects * 3;
 729         if (object_ix_hashsz < 1024)
 730                 object_ix_hashsz = 1024;
 731         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 732         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 733         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 734                 int ix = locate_object_entry_hash(oe->sha1);
 735                 if (0 <= ix)
 736                         continue;
 737                 ix = -1 - ix;
 738                 object_ix[ix] = i + 1;
 739         }
 740 }
 741
 742 static unsigned name_hash(const char *name)
 743 {
 744         unsigned char c;
 745         unsigned hash = 0;
 746
 747         /*
 748          * This effectively just creates a sortable number from the
 749          * last sixteen non-whitespace characters. Last characters
 750          * count "most", so things that end in ".c" sort together.
 751          */
 752         while ((c = *name++) != 0) {
 753                 if (isspace(c))
 754                         continue;
 755                 hash = (hash >> 2) + (c << 24);
 756         }
 757         return hash;
 758 }
 759
 760 static int add_object_entry(const unsigned char *sha1, unsigned hash, int exclude)
 761 {
 762         uint32_t idx = nr_objects;
 763         struct object_entry *entry;
 764         struct packed_git *p;
 765         off_t found_offset = 0;
 766         struct packed_git *found_pack = NULL;
 767         int ix, status = 0;
 768
 769         if (!exclude) {
 770                 for (p = packed_git; p; p = p->next) {
 771                         off_t offset = find_pack_entry_one(sha1, p);
 772                         if (offset) {
 773                                 if (incremental)
 774                                         return 0;
 775                                 if (local && !p->pack_local)
 776                                         return 0;
 777                                 if (!found_pack) {
 778                                         found_offset = offset;
 779                                         found_pack = p;
 780                                 }
 781                         }
 782                 }
 783         }
 784         if ((entry = locate_object_entry(sha1)) != NULL)
 785                 goto already_added;
 786
 787         if (idx >= nr_alloc) {
 788                 nr_alloc = (idx + 1024) * 3 / 2;
 789                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 790         }
 791         entry = objects + idx;
 792         nr_objects = idx + 1;
 793         memset(entry, 0, sizeof(*entry));
 794         hashcpy(entry->sha1, sha1);
 795         entry->hash = hash;
 796
 797         if (object_ix_hashsz * 3 <= nr_objects * 4)
 798                 rehash_objects();
 799         else {
 800                 ix = locate_object_entry_hash(entry->sha1);
 801                 if (0 <= ix)
 802                         die("internal error in object hashing.");
 803                 object_ix[-1 - ix] = idx + 1;
 804         }
 805         status = 1;
 806
 807  already_added:
 808         if (progress_update) {
 809                 fprintf(stderr, "Counting objects...%u\r", nr_objects);
 810                 progress_update = 0;
 811         }
 812         if (exclude)
 813                 entry->preferred_base = 1;
 814         else {
 815                 if (found_pack) {
 816                         entry->in_pack = found_pack;
 817                         entry->in_pack_offset = found_offset;
 818                 }
 819         }
 820         return status;
 821 }
 822
 823 struct pbase_tree_cache {
 824         unsigned char sha1[20];
 825         int ref;
 826         int temporary;
 827         void *tree_data;
 828         unsigned long tree_size;
 829 };
 830
 831 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 832 static int pbase_tree_cache_ix(const unsigned char *sha1)
 833 {
 834         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 835 }
 836 static int pbase_tree_cache_ix_incr(int ix)
 837 {
 838         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 839 }
 840
 841 static struct pbase_tree {
 842         struct pbase_tree *next;
 843         /* This is a phony "cache" entry; we are not
 844          * going to evict it nor find it through _get()
 845          * mechanism -- this is for the toplevel node that
 846          * would almost always change with any commit.
 847          */
 848         struct pbase_tree_cache pcache;
 849 } *pbase_tree;
 850
 851 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 852 {
 853         struct pbase_tree_cache *ent, *nent;
 854         void *data;
 855         unsigned long size;
 856         enum object_type type;
 857         int neigh;
 858         int my_ix = pbase_tree_cache_ix(sha1);
 859         int available_ix = -1;
 860
 861         /* pbase-tree-cache acts as a limited hashtable.
 862          * your object will be found at your index or within a few
 863          * slots after that slot if it is cached.
 864          */
 865         for (neigh = 0; neigh < 8; neigh++) {
 866                 ent = pbase_tree_cache[my_ix];
 867                 if (ent && !hashcmp(ent->sha1, sha1)) {
 868                         ent->ref++;
 869                         return ent;
 870                 }
 871                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 872                          ((0 <= available_ix) &&
 873                           (!ent && pbase_tree_cache[available_ix])))
 874                         available_ix = my_ix;
 875                 if (!ent)
 876                         break;
 877                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 878         }
 879
 880         /* Did not find one.  Either we got a bogus request or
 881          * we need to read and perhaps cache.
 882          */
 883         data = read_sha1_file(sha1, &type, &size);
 884         if (!data)
 885                 return NULL;
 886         if (type != OBJ_TREE) {
 887                 free(data);
 888                 return NULL;
 889         }
 890
 891         /* We need to either cache or return a throwaway copy */
 892
 893         if (available_ix < 0)
 894                 ent = NULL;
 895         else {
 896                 ent = pbase_tree_cache[available_ix];
 897                 my_ix = available_ix;
 898         }
 899
 900         if (!ent) {
 901                 nent = xmalloc(sizeof(*nent));
 902                 nent->temporary = (available_ix < 0);
 903         }
 904         else {
 905                 /* evict and reuse */
 906                 free(ent->tree_data);
 907                 nent = ent;
 908         }
 909         hashcpy(nent->sha1, sha1);
 910         nent->tree_data = data;
 911         nent->tree_size = size;
 912         nent->ref = 1;
 913         if (!nent->temporary)
 914                 pbase_tree_cache[my_ix] = nent;
 915         return nent;
 916 }
 917
 918 static void pbase_tree_put(struct pbase_tree_cache *cache)
 919 {
 920         if (!cache->temporary) {
 921                 cache->ref--;
 922                 return;
 923         }
 924         free(cache->tree_data);
 925         free(cache);
 926 }
 927
 928 static int name_cmp_len(const char *name)
 929 {
 930         int i;
 931         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 932                 ;
 933         return i;
 934 }
 935
 936 static void add_pbase_object(struct tree_desc *tree,
 937                              const char *name,
 938                              int cmplen,
 939                              const char *fullname)
 940 {
 941         struct name_entry entry;
 942
 943         while (tree_entry(tree,&entry)) {
 944                 unsigned long size;
 945                 enum object_type type;
 946
 947                 if (tree_entry_len(entry.path, entry.sha1) != cmplen ||
 948                     memcmp(entry.path, name, cmplen) ||
 949                     !has_sha1_file(entry.sha1) ||
 950                     (type = sha1_object_info(entry.sha1, &size)) < 0)
 951                         continue;
 952                 if (name[cmplen] != '/') {
 953                         unsigned hash = name_hash(fullname);
 954                         add_object_entry(entry.sha1, hash, 1);
 955                         return;
 956                 }
 957                 if (type == OBJ_TREE) {
 958                         struct tree_desc sub;
 959                         struct pbase_tree_cache *tree;
 960                         const char *down = name+cmplen+1;
 961                         int downlen = name_cmp_len(down);
 962
 963                         tree = pbase_tree_get(entry.sha1);
 964                         if (!tree)
 965                                 return;
 966                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
 967
 968                         add_pbase_object(&sub, down, downlen, fullname);
 969                         pbase_tree_put(tree);
 970                 }
 971         }
 972 }
 973
 974 static unsigned *done_pbase_paths;
 975 static int done_pbase_paths_num;
 976 static int done_pbase_paths_alloc;
 977 static int done_pbase_path_pos(unsigned hash)
 978 {
 979         int lo = 0;
 980         int hi = done_pbase_paths_num;
 981         while (lo < hi) {
 982                 int mi = (hi + lo) / 2;
 983                 if (done_pbase_paths[mi] == hash)
 984                         return mi;
 985                 if (done_pbase_paths[mi] < hash)
 986                         hi = mi;
 987                 else
 988                         lo = mi + 1;
 989         }
 990         return -lo-1;
 991 }
 992
 993 static int check_pbase_path(unsigned hash)
 994 {
 995         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
 996         if (0 <= pos)
 997                 return 1;
 998         pos = -pos - 1;
 999         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
1000                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
1001                 done_pbase_paths = xrealloc(done_pbase_paths,
1002                                             done_pbase_paths_alloc *
1003                                             sizeof(unsigned));
1004         }
1005         done_pbase_paths_num++;
1006         if (pos < done_pbase_paths_num)
1007                 memmove(done_pbase_paths + pos + 1,
1008                         done_pbase_paths + pos,
1009                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1010         done_pbase_paths[pos] = hash;
1011         return 0;
1012 }
1013
1014 static void add_preferred_base_object(const char *name, unsigned hash)
1015 {
1016         struct pbase_tree *it;
1017         int cmplen = name_cmp_len(name);
1018
1019         if (check_pbase_path(hash))
1020                 return;
1021
1022         for (it = pbase_tree; it; it = it->next) {
1023                 if (cmplen == 0) {
1024                         hash = name_hash("");
1025                         add_object_entry(it->pcache.sha1, hash, 1);
1026                 }
1027                 else {
1028                         struct tree_desc tree;
1029                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1030                         add_pbase_object(&tree, name, cmplen, name);
1031                 }
1032         }
1033 }
1034
1035 static void add_preferred_base(unsigned char *sha1)
1036 {
1037         struct pbase_tree *it;
1038         void *data;
1039         unsigned long size;
1040         unsigned char tree_sha1[20];
1041
1042         if (window <= num_preferred_base++)
1043                 return;
1044
1045         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1046         if (!data)
1047                 return;
1048
1049         for (it = pbase_tree; it; it = it->next) {
1050                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1051                         free(data);
1052                         return;
1053                 }
1054         }
1055
1056         it = xcalloc(1, sizeof(*it));
1057         it->next = pbase_tree;
1058         pbase_tree = it;
1059
1060         hashcpy(it->pcache.sha1, tree_sha1);
1061         it->pcache.tree_data = data;
1062         it->pcache.tree_size = size;
1063 }
1064
1065 static void check_object(struct object_entry *entry)
1066 {
1067         if (entry->in_pack && !entry->preferred_base) {
1068                 struct packed_git *p = entry->in_pack;
1069                 struct pack_window *w_curs = NULL;
1070                 unsigned long size, used;
1071                 unsigned int avail;
1072                 unsigned char *buf;
1073                 struct object_entry *base_entry = NULL;
1074
1075                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1076
1077                 /* We want in_pack_type even if we do not reuse delta.
1078                  * There is no point not reusing non-delta representations.
1079                  */
1080                 used = unpack_object_header_gently(buf, avail,
1081                                                    &entry->in_pack_type, &size);
1082
1083                 /* Check if it is delta, and the base is also an object
1084                  * we are going to pack.  If so we will reuse the existing
1085                  * delta.
1086                  */
1087                 if (!no_reuse_delta) {
1088                         unsigned char c;
1089                         const unsigned char *base_name;
1090                         off_t ofs;
1091                         unsigned long used_0;
1092                         /* there is at least 20 bytes left in the pack */
1093                         switch (entry->in_pack_type) {
1094                         case OBJ_REF_DELTA:
1095                                 base_name = use_pack(p, &w_curs,
1096                                         entry->in_pack_offset + used, NULL);
1097                                 used += 20;
1098                                 break;
1099                         case OBJ_OFS_DELTA:
1100                                 buf = use_pack(p, &w_curs,
1101                                         entry->in_pack_offset + used, NULL);
1102                                 used_0 = 0;
1103                                 c = buf[used_0++];
1104                                 ofs = c & 127;
1105                                 while (c & 128) {
1106                                         ofs += 1;
1107                                         if (!ofs || MSB(ofs, 7))
1108                                                 die("delta base offset overflow in pack for %s",
1109                                                     sha1_to_hex(entry->sha1));
1110                                         c = buf[used_0++];
1111                                         ofs = (ofs << 7) + (c & 127);
1112                                 }
1113                                 if (ofs >= entry->in_pack_offset)
1114                                         die("delta base offset out of bound for %s",
1115                                             sha1_to_hex(entry->sha1));
1116                                 ofs = entry->in_pack_offset - ofs;
1117                                 base_name = find_packed_object_name(p, ofs);
1118                                 used += used_0;
1119                                 break;
1120                         default:
1121                                 base_name = NULL;
1122                         }
1123                         if (base_name)
1124                                 base_entry = locate_object_entry(base_name);
1125                 }
1126                 unuse_pack(&w_curs);
1127                 entry->in_pack_header_size = used;
1128
1129                 if (base_entry) {
1130
1131                         /* Depth value does not matter - find_deltas()
1132                          * will never consider reused delta as the
1133                          * base object to deltify other objects
1134                          * against, in order to avoid circular deltas.
1135                          */
1136
1137                         /* uncompressed size of the delta data */
1138                         entry->size = size;
1139                         entry->delta = base_entry;
1140                         entry->type = entry->in_pack_type;
1141
1142                         entry->delta_sibling = base_entry->delta_child;
1143                         base_entry->delta_child = entry;
1144
1145                         return;
1146                 }
1147                 /* Otherwise we would do the usual */
1148         }
1149
1150         entry->type = sha1_object_info(entry->sha1, &entry->size);
1151         if (entry->type < 0)
1152                 die("unable to get type of object %s",
1153                     sha1_to_hex(entry->sha1));
1154 }
1155
1156 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1157 {
1158         struct object_entry *child = me->delta_child;
1159         unsigned int m = n;
1160         while (child) {
1161                 unsigned int c = check_delta_limit(child, n + 1);
1162                 if (m < c)
1163                         m = c;
1164                 child = child->delta_sibling;
1165         }
1166         return m;
1167 }
1168
1169 static void get_object_details(void)
1170 {
1171         uint32_t i;
1172         struct object_entry *entry;
1173
1174         prepare_pack_ix();
1175         for (i = 0, entry = objects; i < nr_objects; i++, entry++)
1176                 check_object(entry);
1177
1178         if (nr_objects == nr_result) {
1179                 /*
1180                  * Depth of objects that depend on the entry -- this
1181                  * is subtracted from depth-max to break too deep
1182                  * delta chain because of delta data reusing.
1183                  * However, we loosen this restriction when we know we
1184                  * are creating a thin pack -- it will have to be
1185                  * expanded on the other end anyway, so do not
1186                  * artificially cut the delta chain and let it go as
1187                  * deep as it wants.
1188                  */
1189                 for (i = 0, entry = objects; i < nr_objects; i++, entry++)
1190                         if (!entry->delta && entry->delta_child)
1191                                 entry->delta_limit =
1192                                         check_delta_limit(entry, 1);
1193         }
1194 }
1195
1196 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
1197
1198 static entry_sort_t current_sort;
1199
1200 static int sort_comparator(const void *_a, const void *_b)
1201 {
1202         struct object_entry *a = *(struct object_entry **)_a;
1203         struct object_entry *b = *(struct object_entry **)_b;
1204         return current_sort(a,b);
1205 }
1206
1207 static struct object_entry **create_sorted_list(entry_sort_t sort)
1208 {
1209         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
1210         uint32_t i;
1211
1212         for (i = 0; i < nr_objects; i++)
1213                 list[i] = objects + i;
1214         current_sort = sort;
1215         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
1216         return list;
1217 }
1218
1219 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
1220 {
1221         return hashcmp(a->sha1, b->sha1);
1222 }
1223
1224 static struct object_entry **create_final_object_list(void)
1225 {
1226         struct object_entry **list;
1227         uint32_t i, j;
1228
1229         for (i = nr_result = 0; i < nr_objects; i++)
1230                 if (!objects[i].preferred_base)
1231                         nr_result++;
1232         list = xmalloc(nr_result * sizeof(struct object_entry *));
1233         for (i = j = 0; i < nr_objects; i++) {
1234                 if (!objects[i].preferred_base)
1235                         list[j++] = objects + i;
1236         }
1237         current_sort = sha1_sort;
1238         qsort(list, nr_result, sizeof(struct object_entry *), sort_comparator);
1239         return list;
1240 }
1241
1242 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
1243 {
1244         if (a->type < b->type)
1245                 return -1;
1246         if (a->type > b->type)
1247                 return 1;
1248         if (a->hash < b->hash)
1249                 return -1;
1250         if (a->hash > b->hash)
1251                 return 1;
1252         if (a->preferred_base < b->preferred_base)
1253                 return -1;
1254         if (a->preferred_base > b->preferred_base)
1255                 return 1;
1256         if (a->size < b->size)
1257                 return -1;
1258         if (a->size > b->size)
1259                 return 1;
1260         return a < b ? -1 : (a > b);
1261 }
1262
1263 struct unpacked {
1264         struct object_entry *entry;
1265         void *data;
1266         struct delta_index *index;
1267 };
1268
1269 /*
1270  * We search for deltas _backwards_ in a list sorted by type and
1271  * by size, so that we see progressively smaller and smaller files.
1272  * That's because we prefer deltas to be from the bigger file
1273  * to the smaller - deletes are potentially cheaper, but perhaps
1274  * more importantly, the bigger file is likely the more recent
1275  * one.
1276  */
1277 static int try_delta(struct unpacked *trg, struct unpacked *src,
1278                      unsigned max_depth)
1279 {
1280         struct object_entry *trg_entry = trg->entry;
1281         struct object_entry *src_entry = src->entry;
1282         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1283         enum object_type type;
1284         void *delta_buf;
1285
1286         /* Don't bother doing diffs between different types */
1287         if (trg_entry->type != src_entry->type)
1288                 return -1;
1289
1290         /* We do not compute delta to *create* objects we are not
1291          * going to pack.
1292          */
1293         if (trg_entry->preferred_base)
1294                 return -1;
1295
1296         /*
1297          * We do not bother to try a delta that we discarded
1298          * on an earlier try, but only when reusing delta data.
1299          */
1300         if (!no_reuse_delta && trg_entry->in_pack &&
1301             trg_entry->in_pack == src_entry->in_pack &&
1302             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1303             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1304                 return 0;
1305
1306         /*
1307          * If the current object is at pack edge, take the depth the
1308          * objects that depend on the current object into account --
1309          * otherwise they would become too deep.
1310          */
1311         if (trg_entry->delta_child) {
1312                 if (max_depth <= trg_entry->delta_limit)
1313                         return 0;
1314                 max_depth -= trg_entry->delta_limit;
1315         }
1316         if (src_entry->depth >= max_depth)
1317                 return 0;
1318
1319         /* Now some size filtering heuristics. */
1320         trg_size = trg_entry->size;
1321         max_size = trg_size/2 - 20;
1322         max_size = max_size * (max_depth - src_entry->depth) / max_depth;
1323         if (max_size == 0)
1324                 return 0;
1325         if (trg_entry->delta && trg_entry->delta_size <= max_size)
1326                 max_size = trg_entry->delta_size-1;
1327         src_size = src_entry->size;
1328         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1329         if (sizediff >= max_size)
1330                 return 0;
1331
1332         /* Load data if not already done */
1333         if (!trg->data) {
1334                 trg->data = read_sha1_file(trg_entry->sha1, &type, &sz);
1335                 if (sz != trg_size)
1336                         die("object %s inconsistent object length (%lu vs %lu)",
1337                             sha1_to_hex(trg_entry->sha1), sz, trg_size);
1338         }
1339         if (!src->data) {
1340                 src->data = read_sha1_file(src_entry->sha1, &type, &sz);
1341                 if (sz != src_size)
1342                         die("object %s inconsistent object length (%lu vs %lu)",
1343                             sha1_to_hex(src_entry->sha1), sz, src_size);
1344         }
1345         if (!src->index) {
1346                 src->index = create_delta_index(src->data, src_size);
1347                 if (!src->index)
1348                         die("out of memory");
1349         }
1350
1351         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1352         if (!delta_buf)
1353                 return 0;
1354
1355         trg_entry->delta = src_entry;
1356         trg_entry->delta_size = delta_size;
1357         trg_entry->depth = src_entry->depth + 1;
1358         free(delta_buf);
1359         return 1;
1360 }
1361
1362 static void progress_interval(int signum)
1363 {
1364         progress_update = 1;
1365 }
1366
1367 static void find_deltas(struct object_entry **list, int window, int depth)
1368 {
1369         uint32_t i = nr_objects, idx = 0, processed = 0;
1370         unsigned int array_size = window * sizeof(struct unpacked);
1371         struct unpacked *array;
1372         unsigned last_percent = 999;
1373
1374         if (!nr_objects)
1375                 return;
1376         array = xmalloc(array_size);
1377         memset(array, 0, array_size);
1378         if (progress)
1379                 fprintf(stderr, "Deltifying %u objects.\n", nr_result);
1380
1381         do {
1382                 struct object_entry *entry = list[--i];
1383                 struct unpacked *n = array + idx;
1384                 int j;
1385
1386                 if (!entry->preferred_base)
1387                         processed++;
1388
1389                 if (progress) {
1390                         unsigned percent = processed * 100 / nr_result;
1391                         if (percent != last_percent || progress_update) {
1392                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
1393                                         percent, processed, nr_result);
1394                                 progress_update = 0;
1395                                 last_percent = percent;
1396                         }
1397                 }
1398
1399                 if (entry->delta)
1400                         /* This happens if we decided to reuse existing
1401                          * delta from a pack.  "!no_reuse_delta &&" is implied.
1402                          */
1403                         continue;
1404
1405                 if (entry->size < 50)
1406                         continue;
1407                 free_delta_index(n->index);
1408                 n->index = NULL;
1409                 free(n->data);
1410                 n->data = NULL;
1411                 n->entry = entry;
1412
1413                 j = window;
1414                 while (--j > 0) {
1415                         uint32_t other_idx = idx + j;
1416                         struct unpacked *m;
1417                         if (other_idx >= window)
1418                                 other_idx -= window;
1419                         m = array + other_idx;
1420                         if (!m->entry)
1421                                 break;
1422                         if (try_delta(n, m, depth) < 0)
1423                                 break;
1424                 }
1425                 /* if we made n a delta, and if n is already at max
1426                  * depth, leaving it in the window is pointless.  we
1427                  * should evict it first.
1428                  */
1429                 if (entry->delta && depth <= entry->depth)
1430                         continue;
1431
1432                 idx++;
1433                 if (idx >= window)
1434                         idx = 0;
1435         } while (i > 0);
1436
1437         if (progress)
1438                 fputc('\n', stderr);
1439
1440         for (i = 0; i < window; ++i) {
1441                 free_delta_index(array[i].index);
1442                 free(array[i].data);
1443         }
1444         free(array);
1445 }
1446
1447 static void prepare_pack(int window, int depth)
1448 {
1449         get_object_details();
1450         sorted_by_type = create_sorted_list(type_size_sort);
1451         if (window && depth)
1452                 find_deltas(sorted_by_type, window+1, depth);
1453 }
1454
1455 static int reuse_cached_pack(unsigned char *sha1)
1456 {
1457         static const char cache[] = "pack-cache/pack-%s.%s";
1458         char *cached_pack, *cached_idx;
1459         int ifd, ofd, ifd_ix = -1;
1460
1461         cached_pack = git_path(cache, sha1_to_hex(sha1), "pack");
1462         ifd = open(cached_pack, O_RDONLY);
1463         if (ifd < 0)
1464                 return 0;
1465
1466         if (!pack_to_stdout) {
1467                 cached_idx = git_path(cache, sha1_to_hex(sha1), "idx");
1468                 ifd_ix = open(cached_idx, O_RDONLY);
1469                 if (ifd_ix < 0) {
1470                         close(ifd);
1471                         return 0;
1472                 }
1473         }
1474
1475         if (progress)
1476                 fprintf(stderr, "Reusing %u objects pack %s\n", nr_objects,
1477                         sha1_to_hex(sha1));
1478
1479         if (pack_to_stdout) {
1480                 if (copy_fd(ifd, 1))
1481                         exit(1);
1482                 close(ifd);
1483         }
1484         else {
1485                 char name[PATH_MAX];
1486                 snprintf(name, sizeof(name),
1487                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "pack");
1488                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1489                 if (ofd < 0)
1490                         die("unable to open %s (%s)", name, strerror(errno));
1491                 if (copy_fd(ifd, ofd))
1492                         exit(1);
1493                 close(ifd);
1494
1495                 snprintf(name, sizeof(name),
1496                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "idx");
1497                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1498                 if (ofd < 0)
1499                         die("unable to open %s (%s)", name, strerror(errno));
1500                 if (copy_fd(ifd_ix, ofd))
1501                         exit(1);
1502                 close(ifd_ix);
1503                 puts(sha1_to_hex(sha1));
1504         }
1505
1506         return 1;
1507 }
1508
1509 static void setup_progress_signal(void)
1510 {
1511         struct sigaction sa;
1512         struct itimerval v;
1513
1514         memset(&sa, 0, sizeof(sa));
1515         sa.sa_handler = progress_interval;
1516         sigemptyset(&sa.sa_mask);
1517         sa.sa_flags = SA_RESTART;
1518         sigaction(SIGALRM, &sa, NULL);
1519
1520         v.it_interval.tv_sec = 1;
1521         v.it_interval.tv_usec = 0;
1522         v.it_value = v.it_interval;
1523         setitimer(ITIMER_REAL, &v, NULL);
1524 }
1525
1526 static int git_pack_config(const char *k, const char *v)
1527 {
1528         if(!strcmp(k, "pack.window")) {
1529                 window = git_config_int(k, v);
1530                 return 0;
1531         }
1532         return git_default_config(k, v);
1533 }
1534
1535 static void read_object_list_from_stdin(void)
1536 {
1537         char line[40 + 1 + PATH_MAX + 2];
1538         unsigned char sha1[20];
1539         unsigned hash;
1540
1541         for (;;) {
1542                 if (!fgets(line, sizeof(line), stdin)) {
1543                         if (feof(stdin))
1544                                 break;
1545                         if (!ferror(stdin))
1546                                 die("fgets returned NULL, not EOF, not error!");
1547                         if (errno != EINTR)
1548                                 die("fgets: %s", strerror(errno));
1549                         clearerr(stdin);
1550                         continue;
1551                 }
1552                 if (line[0] == '-') {
1553                         if (get_sha1_hex(line+1, sha1))
1554                                 die("expected edge sha1, got garbage:\n %s",
1555                                     line);
1556                         add_preferred_base(sha1);
1557                         continue;
1558                 }
1559                 if (get_sha1_hex(line, sha1))
1560                         die("expected sha1, got garbage:\n %s", line);
1561
1562                 hash = name_hash(line+41);
1563                 add_preferred_base_object(line+41, hash);
1564                 add_object_entry(sha1, hash, 0);
1565         }
1566 }
1567
1568 static void show_commit(struct commit *commit)
1569 {
1570         unsigned hash = name_hash("");
1571         add_preferred_base_object("", hash);
1572         add_object_entry(commit->object.sha1, hash, 0);
1573 }
1574
1575 static void show_object(struct object_array_entry *p)
1576 {
1577         unsigned hash = name_hash(p->name);
1578         add_preferred_base_object(p->name, hash);
1579         add_object_entry(p->item->sha1, hash, 0);
1580 }
1581
1582 static void show_edge(struct commit *commit)
1583 {
1584         add_preferred_base(commit->object.sha1);
1585 }
1586
1587 static void get_object_list(int ac, const char **av)
1588 {
1589         struct rev_info revs;
1590         char line[1000];
1591         int flags = 0;
1592
1593         init_revisions(&revs, NULL);
1594         save_commit_buffer = 0;
1595         track_object_refs = 0;
1596         setup_revisions(ac, av, &revs, NULL);
1597
1598         while (fgets(line, sizeof(line), stdin) != NULL) {
1599                 int len = strlen(line);
1600                 if (line[len - 1] == '\n')
1601                         line[--len] = 0;
1602                 if (!len)
1603                         break;
1604                 if (*line == '-') {
1605                         if (!strcmp(line, "--not")) {
1606                                 flags ^= UNINTERESTING;
1607                                 continue;
1608                         }
1609                         die("not a rev '%s'", line);
1610                 }
1611                 if (handle_revision_arg(line, &revs, flags, 1))
1612                         die("bad revision '%s'", line);
1613         }
1614
1615         prepare_revision_walk(&revs);
1616         mark_edges_uninteresting(revs.commits, &revs, show_edge);
1617         traverse_commit_list(&revs, show_commit, show_object);
1618 }
1619
1620 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1621 {
1622         SHA_CTX ctx;
1623         int depth = 10;
1624         struct object_entry **list;
1625         int use_internal_rev_list = 0;
1626         int thin = 0;
1627         uint32_t i;
1628         const char **rp_av;
1629         int rp_ac_alloc = 64;
1630         int rp_ac;
1631
1632         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1633
1634         rp_av[0] = "pack-objects";
1635         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1636         rp_ac = 2;
1637
1638         git_config(git_pack_config);
1639
1640         progress = isatty(2);
1641         for (i = 1; i < argc; i++) {
1642                 const char *arg = argv[i];
1643
1644                 if (*arg != '-')
1645                         break;
1646
1647                 if (!strcmp("--non-empty", arg)) {
1648                         non_empty = 1;
1649                         continue;
1650                 }
1651                 if (!strcmp("--local", arg)) {
1652                         local = 1;
1653                         continue;
1654                 }
1655                 if (!strcmp("--incremental", arg)) {
1656                         incremental = 1;
1657                         continue;
1658                 }
1659                 if (!prefixcmp(arg, "--window=")) {
1660                         char *end;
1661                         window = strtoul(arg+9, &end, 0);
1662                         if (!arg[9] || *end)
1663                                 usage(pack_usage);
1664                         continue;
1665                 }
1666                 if (!prefixcmp(arg, "--depth=")) {
1667                         char *end;
1668                         depth = strtoul(arg+8, &end, 0);
1669                         if (!arg[8] || *end)
1670                                 usage(pack_usage);
1671                         continue;
1672                 }
1673                 if (!strcmp("--progress", arg)) {
1674                         progress = 1;
1675                         continue;
1676                 }
1677                 if (!strcmp("--all-progress", arg)) {
1678                         progress = 2;
1679                         continue;
1680                 }
1681                 if (!strcmp("-q", arg)) {
1682                         progress = 0;
1683                         continue;
1684                 }
1685                 if (!strcmp("--no-reuse-delta", arg)) {
1686                         no_reuse_delta = 1;
1687                         continue;
1688                 }
1689                 if (!strcmp("--delta-base-offset", arg)) {
1690                         allow_ofs_delta = 1;
1691                         continue;
1692                 }
1693                 if (!strcmp("--stdout", arg)) {
1694                         pack_to_stdout = 1;
1695                         continue;
1696                 }
1697                 if (!strcmp("--revs", arg)) {
1698                         use_internal_rev_list = 1;
1699                         continue;
1700                 }
1701                 if (!strcmp("--unpacked", arg) ||
1702                     !prefixcmp(arg, "--unpacked=") ||
1703                     !strcmp("--reflog", arg) ||
1704                     !strcmp("--all", arg)) {
1705                         use_internal_rev_list = 1;
1706                         if (rp_ac >= rp_ac_alloc - 1) {
1707                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1708                                 rp_av = xrealloc(rp_av,
1709                                                  rp_ac_alloc * sizeof(*rp_av));
1710                         }
1711                         rp_av[rp_ac++] = arg;
1712                         continue;
1713                 }
1714                 if (!strcmp("--thin", arg)) {
1715                         use_internal_rev_list = 1;
1716                         thin = 1;
1717                         rp_av[1] = "--objects-edge";
1718                         continue;
1719                 }
1720                 if (!prefixcmp(arg, "--index-version=")) {
1721                         char *c;
1722                         index_default_version = strtoul(arg + 16, &c, 10);
1723                         if (index_default_version > 2)
1724                                 die("bad %s", arg);
1725                         if (*c == ',')
1726                                 index_off32_limit = strtoul(c+1, &c, 0);
1727                         if (*c || index_off32_limit & 0x80000000)
1728                                 die("bad %s", arg);
1729                         continue;
1730                 }
1731                 usage(pack_usage);
1732         }
1733
1734         /* Traditionally "pack-objects [options] base extra" failed;
1735          * we would however want to take refs parameter that would
1736          * have been given to upstream rev-list ourselves, which means
1737          * we somehow want to say what the base name is.  So the
1738          * syntax would be:
1739          *
1740          * pack-objects [options] base <refs...>
1741          *
1742          * in other words, we would treat the first non-option as the
1743          * base_name and send everything else to the internal revision
1744          * walker.
1745          */
1746
1747         if (!pack_to_stdout)
1748                 base_name = argv[i++];
1749
1750         if (pack_to_stdout != !base_name)
1751                 usage(pack_usage);
1752
1753         if (!pack_to_stdout && thin)
1754                 die("--thin cannot be used to build an indexable pack.");
1755
1756         prepare_packed_git();
1757
1758         if (progress) {
1759                 fprintf(stderr, "Generating pack...\n");
1760                 setup_progress_signal();
1761         }
1762
1763         if (!use_internal_rev_list)
1764                 read_object_list_from_stdin();
1765         else {
1766                 rp_av[rp_ac] = NULL;
1767                 get_object_list(rp_ac, rp_av);
1768         }
1769
1770         if (progress)
1771                 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
1772         sorted_by_sha = create_final_object_list();
1773         if (non_empty && !nr_result)
1774                 return 0;
1775
1776         SHA1_Init(&ctx);
1777         list = sorted_by_sha;
1778         for (i = 0; i < nr_result; i++) {
1779                 struct object_entry *entry = *list++;
1780                 SHA1_Update(&ctx, entry->sha1, 20);
1781         }
1782         SHA1_Final(object_list_sha1, &ctx);
1783         if (progress && (nr_objects != nr_result))
1784                 fprintf(stderr, "Result has %u objects.\n", nr_result);
1785
1786         if (reuse_cached_pack(object_list_sha1))
1787                 ;
1788         else {
1789                 off_t last_obj_offset;
1790                 if (nr_result)
1791                         prepare_pack(window, depth);
1792                 if (progress == 1 && pack_to_stdout) {
1793                         /* the other end usually displays progress itself */
1794                         struct itimerval v = {{0,},};
1795                         setitimer(ITIMER_REAL, &v, NULL);
1796                         signal(SIGALRM, SIG_IGN );
1797                         progress_update = 0;
1798                 }
1799                 last_obj_offset = write_pack_file();
1800                 if (!pack_to_stdout) {
1801                         write_index_file(last_obj_offset);
1802                         puts(sha1_to_hex(object_list_sha1));
1803                 }
1804         }
1805         if (progress)
1806                 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
1807                         written, written_delta, reused, reused_delta);
1808         return 0;
1809 }