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 hash;      /* name hint hash */
  31         enum object_type type;
  32         enum object_type in_pack_type;  /* could be delta */
  33         unsigned long delta_size;       /* delta data size (uncompressed) */
  34 #define in_pack_header_size delta_size  /* only when reusing pack data */
  35         struct object_entry *delta;     /* delta base object */
  36         struct packed_git *in_pack;     /* already in pack */
  37         off_t in_pack_offset;
  38         struct object_entry *delta_child; /* deltified objects who bases me */
  39         struct object_entry *delta_sibling; /* other deltified objects who
  40                                              * uses the same base as me
  41                                              */
  42         int preferred_base;     /* we do not pack this, but is encouraged to
  43                                  * be used as the base objectto delta huge
  44                                  * objects against.
  45                                  */
  46         uint32_t crc32;         /* crc of raw pack data for this object */
  47 };
  48
  49 /*
  50  * Objects we are going to pack are collected in objects array (dynamically
  51  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  52  * in the order we see -- typically rev-list --objects order that gives us
  53  * nice "minimum seek" order.
  54  *
  55  * sorted-by-sha ans sorted-by-type are arrays of pointers that point at
  56  * elements in the objects array.  The former is used to build the pack
  57  * index (lists object names in the ascending order to help offset lookup),
  58  * and the latter is used to group similar things together by try_delta()
  59  * heuristics.
  60  */
  61
  62 static unsigned char object_list_sha1[20];
  63 static int non_empty;
  64 static int no_reuse_delta;
  65 static int local;
  66 static int incremental;
  67 static int allow_ofs_delta;
  68
  69 static struct object_entry **sorted_by_sha, **sorted_by_type;
  70 static struct object_entry *objects;
  71 static uint32_t nr_objects, nr_alloc, nr_result;
  72 static const char *base_name;
  73 static unsigned char pack_file_sha1[20];
  74 static int progress = 1;
  75 static volatile sig_atomic_t progress_update;
  76 static int window = 10;
  77 static int pack_to_stdout;
  78 static int num_preferred_base;
  79
  80 /*
  81  * The object names in objects array are hashed with this hashtable,
  82  * to help looking up the entry by object name.  Binary search from
  83  * sorted_by_sha is also possible but this was easier to code and faster.
  84  * This hashtable is built after all the objects are seen.
  85  */
  86 static int *object_ix;
  87 static int object_ix_hashsz;
  88
  89 /*
  90  * Pack index for existing packs give us easy access to the offsets into
  91  * corresponding pack file where each object's data starts, but the entries
  92  * do not store the size of the compressed representation (uncompressed
  93  * size is easily available by examining the pack entry header).  It is
  94  * also rather expensive to find the sha1 for an object given its offset.
  95  *
  96  * We build a hashtable of existing packs (pack_revindex), and keep reverse
  97  * index here -- pack index file is sorted by object name mapping to offset;
  98  * this pack_revindex[].revindex array is a list of offset/index_nr pairs
  99  * ordered by offset, so if you know the offset of an object, next offset
 100  * is where its packed representation ends and the index_nr can be used to
 101  * get the object sha1 from the main index.
 102  */
 103 struct revindex_entry {
 104         off_t offset;
 105         unsigned int nr;
 106 };
 107 struct pack_revindex {
 108         struct packed_git *p;
 109         struct revindex_entry *revindex;
 110 };
 111 static struct  pack_revindex *pack_revindex;
 112 static int pack_revindex_hashsz;
 113
 114 /*
 115  * stats
 116  */
 117 static uint32_t written, written_delta;
 118 static uint32_t reused, reused_delta;
 119
 120 static int pack_revindex_ix(struct packed_git *p)
 121 {
 122         unsigned long ui = (unsigned long)p;
 123         int i;
 124
 125         ui = ui ^ (ui >> 16); /* defeat structure alignment */
 126         i = (int)(ui % pack_revindex_hashsz);
 127         while (pack_revindex[i].p) {
 128                 if (pack_revindex[i].p == p)
 129                         return i;
 130                 if (++i == pack_revindex_hashsz)
 131                         i = 0;
 132         }
 133         return -1 - i;
 134 }
 135
 136 static void prepare_pack_ix(void)
 137 {
 138         int num;
 139         struct packed_git *p;
 140         for (num = 0, p = packed_git; p; p = p->next)
 141                 num++;
 142         if (!num)
 143                 return;
 144         pack_revindex_hashsz = num * 11;
 145         pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
 146         for (p = packed_git; p; p = p->next) {
 147                 num = pack_revindex_ix(p);
 148                 num = - 1 - num;
 149                 pack_revindex[num].p = p;
 150         }
 151         /* revindex elements are lazily initialized */
 152 }
 153
 154 static int cmp_offset(const void *a_, const void *b_)
 155 {
 156         const struct revindex_entry *a = a_;
 157         const struct revindex_entry *b = b_;
 158         return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0;
 159 }
 160
 161 /*
 162  * Ordered list of offsets of objects in the pack.
 163  */
 164 static void prepare_pack_revindex(struct pack_revindex *rix)
 165 {
 166         struct packed_git *p = rix->p;
 167         int num_ent = p->num_objects;
 168         int i;
 169         const char *index = p->index_data;
 170
 171         rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
 172         index += 4 * 256;
 173
 174         if (p->index_version > 1) {
 175                 const uint32_t *off_32 =
 176                         (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
 177                 const uint32_t *off_64 = off_32 + p->num_objects;
 178                 for (i = 0; i < num_ent; i++) {
 179                         uint32_t off = ntohl(*off_32++);
 180                         if (!(off & 0x80000000)) {
 181                                 rix->revindex[i].offset = off;
 182                         } else {
 183                                 rix->revindex[i].offset =
 184                                         ((uint64_t)ntohl(*off_64++)) << 32;
 185                                 rix->revindex[i].offset |=
 186                                         ntohl(*off_64++);
 187                         }
 188                         rix->revindex[i].nr = i;
 189                 }
 190         } else {
 191                 for (i = 0; i < num_ent; i++) {
 192                         uint32_t hl = *((uint32_t *)(index + 24 * i));
 193                         rix->revindex[i].offset = ntohl(hl);
 194                         rix->revindex[i].nr = i;
 195                 }
 196         }
 197
 198         /* This knows the pack format -- the 20-byte trailer
 199          * follows immediately after the last object data.
 200          */
 201         rix->revindex[num_ent].offset = p->pack_size - 20;
 202         rix->revindex[num_ent].nr = -1;
 203         qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
 204 }
 205
 206 static struct revindex_entry * find_packed_object(struct packed_git *p,
 207                                                   off_t ofs)
 208 {
 209         int num;
 210         int lo, hi;
 211         struct pack_revindex *rix;
 212         struct revindex_entry *revindex;
 213         num = pack_revindex_ix(p);
 214         if (num < 0)
 215                 die("internal error: pack revindex uninitialized");
 216         rix = &pack_revindex[num];
 217         if (!rix->revindex)
 218                 prepare_pack_revindex(rix);
 219         revindex = rix->revindex;
 220         lo = 0;
 221         hi = p->num_objects + 1;
 222         do {
 223                 int mi = (lo + hi) / 2;
 224                 if (revindex[mi].offset == ofs) {
 225                         return revindex + mi;
 226                 }
 227                 else if (ofs < revindex[mi].offset)
 228                         hi = mi;
 229                 else
 230                         lo = mi + 1;
 231         } while (lo < hi);
 232         die("internal error: pack revindex corrupt");
 233 }
 234
 235 static const unsigned char *find_packed_object_name(struct packed_git *p,
 236                                                     off_t ofs)
 237 {
 238         struct revindex_entry *entry = find_packed_object(p, ofs);
 239         return nth_packed_object_sha1(p, entry->nr);
 240 }
 241
 242 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 243 {
 244         unsigned long othersize, delta_size;
 245         enum object_type type;
 246         void *otherbuf = read_sha1_file(entry->delta->sha1, &type, &othersize);
 247         void *delta_buf;
 248
 249         if (!otherbuf)
 250                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
 251         delta_buf = diff_delta(otherbuf, othersize,
 252                                buf, size, &delta_size, 0);
 253         if (!delta_buf || delta_size != entry->delta_size)
 254                 die("delta size changed");
 255         free(buf);
 256         free(otherbuf);
 257         return delta_buf;
 258 }
 259
 260 /*
 261  * The per-object header is a pretty dense thing, which is
 262  *  - first byte: low four bits are "size", then three bits of "type",
 263  *    and the high bit is "size continues".
 264  *  - each byte afterwards: low seven bits are size continuation,
 265  *    with the high bit being "size continues"
 266  */
 267 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 268 {
 269         int n = 1;
 270         unsigned char c;
 271
 272         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 273                 die("bad type %d", type);
 274
 275         c = (type << 4) | (size & 15);
 276         size >>= 4;
 277         while (size) {
 278                 *hdr++ = c | 0x80;
 279                 c = size & 0x7f;
 280                 size >>= 7;
 281                 n++;
 282         }
 283         *hdr = c;
 284         return n;
 285 }
 286
 287 /*
 288  * we are going to reuse the existing object data as is.  make
 289  * sure it is not corrupt.
 290  */
 291 static int check_pack_inflate(struct packed_git *p,
 292                 struct pack_window **w_curs,
 293                 off_t offset,
 294                 off_t len,
 295                 unsigned long expect)
 296 {
 297         z_stream stream;
 298         unsigned char fakebuf[4096], *in;
 299         int st;
 300
 301         memset(&stream, 0, sizeof(stream));
 302         inflateInit(&stream);
 303         do {
 304                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 305                 stream.next_in = in;
 306                 stream.next_out = fakebuf;
 307                 stream.avail_out = sizeof(fakebuf);
 308                 st = inflate(&stream, Z_FINISH);
 309                 offset += stream.next_in - in;
 310         } while (st == Z_OK || st == Z_BUF_ERROR);
 311         inflateEnd(&stream);
 312         return (st == Z_STREAM_END &&
 313                 stream.total_out == expect &&
 314                 stream.total_in == len) ? 0 : -1;
 315 }
 316
 317 static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
 318                           off_t offset, off_t len, unsigned int nr)
 319 {
 320         const uint32_t *index_crc;
 321         uint32_t data_crc = crc32(0, Z_NULL, 0);
 322
 323         do {
 324                 unsigned int avail;
 325                 void *data = use_pack(p, w_curs, offset, &avail);
 326                 if (avail > len)
 327                         avail = len;
 328                 data_crc = crc32(data_crc, data, avail);
 329                 offset += avail;
 330                 len -= avail;
 331         } while (len);
 332
 333         index_crc = p->index_data;
 334         index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
 335
 336         return data_crc != ntohl(*index_crc);
 337 }
 338
 339 static void copy_pack_data(struct sha1file *f,
 340                 struct packed_git *p,
 341                 struct pack_window **w_curs,
 342                 off_t offset,
 343                 off_t len)
 344 {
 345         unsigned char *in;
 346         unsigned int avail;
 347
 348         while (len) {
 349                 in = use_pack(p, w_curs, offset, &avail);
 350                 if (avail > len)
 351                         avail = (unsigned int)len;
 352                 sha1write(f, in, avail);
 353                 offset += avail;
 354                 len -= avail;
 355         }
 356 }
 357
 358 static int check_loose_inflate(unsigned char *data, unsigned long len, unsigned long expect)
 359 {
 360         z_stream stream;
 361         unsigned char fakebuf[4096];
 362         int st;
 363
 364         memset(&stream, 0, sizeof(stream));
 365         stream.next_in = data;
 366         stream.avail_in = len;
 367         stream.next_out = fakebuf;
 368         stream.avail_out = sizeof(fakebuf);
 369         inflateInit(&stream);
 370
 371         while (1) {
 372                 st = inflate(&stream, Z_FINISH);
 373                 if (st == Z_STREAM_END || st == Z_OK) {
 374                         st = (stream.total_out == expect &&
 375                               stream.total_in == len) ? 0 : -1;
 376                         break;
 377                 }
 378                 if (st != Z_BUF_ERROR) {
 379                         st = -1;
 380                         break;
 381                 }
 382                 stream.next_out = fakebuf;
 383                 stream.avail_out = sizeof(fakebuf);
 384         }
 385         inflateEnd(&stream);
 386         return st;
 387 }
 388
 389 static int revalidate_loose_object(struct object_entry *entry,
 390                                    unsigned char *map,
 391                                    unsigned long mapsize)
 392 {
 393         /* we already know this is a loose object with new type header. */
 394         enum object_type type;
 395         unsigned long size, used;
 396
 397         if (pack_to_stdout)
 398                 return 0;
 399
 400         used = unpack_object_header_gently(map, mapsize, &type, &size);
 401         if (!used)
 402                 return -1;
 403         map += used;
 404         mapsize -= used;
 405         return check_loose_inflate(map, mapsize, size);
 406 }
 407
 408 static unsigned long write_object(struct sha1file *f,
 409                                   struct object_entry *entry)
 410 {
 411         unsigned long size;
 412         enum object_type type;
 413         void *buf;
 414         unsigned char header[10];
 415         unsigned hdrlen;
 416         off_t datalen;
 417         enum object_type obj_type;
 418         int to_reuse = 0;
 419
 420         if (!pack_to_stdout)
 421                 crc32_begin(f);
 422
 423         obj_type = entry->type;
 424         if (! entry->in_pack)
 425                 to_reuse = 0;   /* can't reuse what we don't have */
 426         else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
 427                 to_reuse = 1;   /* check_object() decided it for us */
 428         else if (obj_type != entry->in_pack_type)
 429                 to_reuse = 0;   /* pack has delta which is unusable */
 430         else if (entry->delta)
 431                 to_reuse = 0;   /* we want to pack afresh */
 432         else
 433                 to_reuse = 1;   /* we have it in-pack undeltified,
 434                                  * and we do not need to deltify it.
 435                                  */
 436
 437         if (!entry->in_pack && !entry->delta) {
 438                 unsigned char *map;
 439                 unsigned long mapsize;
 440                 map = map_sha1_file(entry->sha1, &mapsize);
 441                 if (map && !legacy_loose_object(map)) {
 442                         /* We can copy straight into the pack file */
 443                         if (revalidate_loose_object(entry, map, mapsize))
 444                                 die("corrupt loose object %s",
 445                                     sha1_to_hex(entry->sha1));
 446                         sha1write(f, map, mapsize);
 447                         munmap(map, mapsize);
 448                         written++;
 449                         reused++;
 450                         return mapsize;
 451                 }
 452                 if (map)
 453                         munmap(map, mapsize);
 454         }
 455
 456         if (!to_reuse) {
 457                 buf = read_sha1_file(entry->sha1, &type, &size);
 458                 if (!buf)
 459                         die("unable to read %s", sha1_to_hex(entry->sha1));
 460                 if (size != entry->size)
 461                         die("object %s size inconsistency (%lu vs %lu)",
 462                             sha1_to_hex(entry->sha1), size, entry->size);
 463                 if (entry->delta) {
 464                         buf = delta_against(buf, size, entry);
 465                         size = entry->delta_size;
 466                         obj_type = (allow_ofs_delta && entry->delta->offset) ?
 467                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 468                 }
 469                 /*
 470                  * The object header is a byte of 'type' followed by zero or
 471                  * more bytes of length.
 472                  */
 473                 hdrlen = encode_header(obj_type, size, header);
 474                 sha1write(f, header, hdrlen);
 475
 476                 if (obj_type == OBJ_OFS_DELTA) {
 477                         /*
 478                          * Deltas with relative base contain an additional
 479                          * encoding of the relative offset for the delta
 480                          * base from this object's position in the pack.
 481                          */
 482                         off_t ofs = entry->offset - entry->delta->offset;
 483                         unsigned pos = sizeof(header) - 1;
 484                         header[pos] = ofs & 127;
 485                         while (ofs >>= 7)
 486                                 header[--pos] = 128 | (--ofs & 127);
 487                         sha1write(f, header + pos, sizeof(header) - pos);
 488                         hdrlen += sizeof(header) - pos;
 489                 } else if (obj_type == OBJ_REF_DELTA) {
 490                         /*
 491                          * Deltas with a base reference contain
 492                          * an additional 20 bytes for the base sha1.
 493                          */
 494                         sha1write(f, entry->delta->sha1, 20);
 495                         hdrlen += 20;
 496                 }
 497                 datalen = sha1write_compressed(f, buf, size);
 498                 free(buf);
 499         }
 500         else {
 501                 struct packed_git *p = entry->in_pack;
 502                 struct pack_window *w_curs = NULL;
 503                 struct revindex_entry *revidx;
 504                 off_t offset;
 505
 506                 if (entry->delta) {
 507                         obj_type = (allow_ofs_delta && entry->delta->offset) ?
 508                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 509                         reused_delta++;
 510                 }
 511                 hdrlen = encode_header(obj_type, entry->size, header);
 512                 sha1write(f, header, hdrlen);
 513                 if (obj_type == OBJ_OFS_DELTA) {
 514                         off_t ofs = entry->offset - entry->delta->offset;
 515                         unsigned pos = sizeof(header) - 1;
 516                         header[pos] = ofs & 127;
 517                         while (ofs >>= 7)
 518                                 header[--pos] = 128 | (--ofs & 127);
 519                         sha1write(f, header + pos, sizeof(header) - pos);
 520                         hdrlen += sizeof(header) - pos;
 521                 } else if (obj_type == OBJ_REF_DELTA) {
 522                         sha1write(f, entry->delta->sha1, 20);
 523                         hdrlen += 20;
 524                 }
 525
 526                 offset = entry->in_pack_offset;
 527                 revidx = find_packed_object(p, offset);
 528                 datalen = revidx[1].offset - offset;
 529                 if (!pack_to_stdout && p->index_version > 1 &&
 530                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
 531                         die("bad packed object CRC for %s", sha1_to_hex(entry->sha1));
 532                 offset += entry->in_pack_header_size;
 533                 datalen -= entry->in_pack_header_size;
 534                 if (!pack_to_stdout && p->index_version == 1 &&
 535                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
 536                         die("corrupt packed object for %s", sha1_to_hex(entry->sha1));
 537                 copy_pack_data(f, p, &w_curs, offset, datalen);
 538                 unuse_pack(&w_curs);
 539                 reused++;
 540         }
 541         if (entry->delta)
 542                 written_delta++;
 543         written++;
 544         if (!pack_to_stdout)
 545                 entry->crc32 = crc32_end(f);
 546         return hdrlen + datalen;
 547 }
 548
 549 static off_t write_one(struct sha1file *f,
 550                                struct object_entry *e,
 551                                off_t offset)
 552 {
 553         unsigned long size;
 554
 555         /* offset is non zero if object is written already. */
 556         if (e->offset || e->preferred_base)
 557                 return offset;
 558
 559         /* if we are deltified, write out base object first. */
 560         if (e->delta)
 561                 offset = write_one(f, e->delta, offset);
 562
 563         e->offset = offset;
 564         size = write_object(f, e);
 565
 566         /* make sure off_t is sufficiently large not to wrap */
 567         if (offset > offset + size)
 568                 die("pack too large for current definition of off_t");
 569         return offset + size;
 570 }
 571
 572 static off_t write_pack_file(void)
 573 {
 574         uint32_t i;
 575         struct sha1file *f;
 576         off_t offset, last_obj_offset = 0;
 577         struct pack_header hdr;
 578         unsigned last_percent = 999;
 579         int do_progress = progress;
 580
 581         if (!base_name) {
 582                 f = sha1fd(1, "<stdout>");
 583                 do_progress >>= 1;
 584         }
 585         else
 586                 f = sha1create("%s-%s.%s", base_name,
 587                                sha1_to_hex(object_list_sha1), "pack");
 588         if (do_progress)
 589                 fprintf(stderr, "Writing %u objects.\n", nr_result);
 590
 591         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 592         hdr.hdr_version = htonl(PACK_VERSION);
 593         hdr.hdr_entries = htonl(nr_result);
 594         sha1write(f, &hdr, sizeof(hdr));
 595         offset = sizeof(hdr);
 596         if (!nr_result)
 597                 goto done;
 598         for (i = 0; i < nr_objects; i++) {
 599                 last_obj_offset = offset;
 600                 offset = write_one(f, objects + i, offset);
 601                 if (do_progress) {
 602                         unsigned percent = written * 100 / nr_result;
 603                         if (progress_update || percent != last_percent) {
 604                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
 605                                         percent, written, nr_result);
 606                                 progress_update = 0;
 607                                 last_percent = percent;
 608                         }
 609                 }
 610         }
 611         if (do_progress)
 612                 fputc('\n', stderr);
 613  done:
 614         if (written != nr_result)
 615                 die("wrote %u objects while expecting %u", written, nr_result);
 616         sha1close(f, pack_file_sha1, 1);
 617
 618         return last_obj_offset;
 619 }
 620
 621 static uint32_t index_default_version = 1;
 622 static uint32_t index_off32_limit = 0x7fffffff;
 623
 624 static void write_index_file(off_t last_obj_offset)
 625 {
 626         uint32_t i;
 627         struct sha1file *f = sha1create("%s-%s.%s", base_name,
 628                                         sha1_to_hex(object_list_sha1), "idx");
 629         struct object_entry **list = sorted_by_sha;
 630         struct object_entry **last = list + nr_result;
 631         uint32_t array[256];
 632         uint32_t index_version;
 633
 634         /* if last object's offset is >= 2^31 we should use index V2 */
 635         index_version = (last_obj_offset >> 31) ? 2 : index_default_version;
 636
 637         /* index versions 2 and above need a header */
 638         if (index_version >= 2) {
 639                 struct pack_idx_header hdr;
 640                 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 641                 hdr.idx_version = htonl(index_version);
 642                 sha1write(f, &hdr, sizeof(hdr));
 643         }
 644
 645         /*
 646          * Write the first-level table (the list is sorted,
 647          * but we use a 256-entry lookup to be able to avoid
 648          * having to do eight extra binary search iterations).
 649          */
 650         for (i = 0; i < 256; i++) {
 651                 struct object_entry **next = list;
 652                 while (next < last) {
 653                         struct object_entry *entry = *next;
 654                         if (entry->sha1[0] != i)
 655                                 break;
 656                         next++;
 657                 }
 658                 array[i] = htonl(next - sorted_by_sha);
 659                 list = next;
 660         }
 661         sha1write(f, array, 256 * 4);
 662
 663         /*
 664          * Write the actual SHA1 entries..
 665          */
 666         list = sorted_by_sha;
 667         for (i = 0; i < nr_result; i++) {
 668                 struct object_entry *entry = *list++;
 669                 if (index_version < 2) {
 670                         uint32_t offset = htonl(entry->offset);
 671                         sha1write(f, &offset, 4);
 672                 }
 673                 sha1write(f, entry->sha1, 20);
 674         }
 675
 676         if (index_version >= 2) {
 677                 unsigned int nr_large_offset = 0;
 678
 679                 /* write the crc32 table */
 680                 list = sorted_by_sha;
 681                 for (i = 0; i < nr_objects; i++) {
 682                         struct object_entry *entry = *list++;
 683                         uint32_t crc32_val = htonl(entry->crc32);
 684                         sha1write(f, &crc32_val, 4);
 685                 }
 686
 687                 /* write the 32-bit offset table */
 688                 list = sorted_by_sha;
 689                 for (i = 0; i < nr_objects; i++) {
 690                         struct object_entry *entry = *list++;
 691                         uint32_t offset = (entry->offset <= index_off32_limit) ?
 692                                 entry->offset : (0x80000000 | nr_large_offset++);
 693                         offset = htonl(offset);
 694                         sha1write(f, &offset, 4);
 695                 }
 696
 697                 /* write the large offset table */
 698                 list = sorted_by_sha;
 699                 while (nr_large_offset) {
 700                         struct object_entry *entry = *list++;
 701                         uint64_t offset = entry->offset;
 702                         if (offset > index_off32_limit) {
 703                                 uint32_t split[2];
 704                                 split[0]        = htonl(offset >> 32);
 705                                 split[1] = htonl(offset & 0xffffffff);
 706                                 sha1write(f, split, 8);
 707                                 nr_large_offset--;
 708                         }
 709                 }
 710         }
 711
 712         sha1write(f, pack_file_sha1, 20);
 713         sha1close(f, NULL, 1);
 714 }
 715
 716 static int locate_object_entry_hash(const unsigned char *sha1)
 717 {
 718         int i;
 719         unsigned int ui;
 720         memcpy(&ui, sha1, sizeof(unsigned int));
 721         i = ui % object_ix_hashsz;
 722         while (0 < object_ix[i]) {
 723                 if (!hashcmp(sha1, objects[object_ix[i] - 1].sha1))
 724                         return i;
 725                 if (++i == object_ix_hashsz)
 726                         i = 0;
 727         }
 728         return -1 - i;
 729 }
 730
 731 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 732 {
 733         int i;
 734
 735         if (!object_ix_hashsz)
 736                 return NULL;
 737
 738         i = locate_object_entry_hash(sha1);
 739         if (0 <= i)
 740                 return &objects[object_ix[i]-1];
 741         return NULL;
 742 }
 743
 744 static void rehash_objects(void)
 745 {
 746         uint32_t i;
 747         struct object_entry *oe;
 748
 749         object_ix_hashsz = nr_objects * 3;
 750         if (object_ix_hashsz < 1024)
 751                 object_ix_hashsz = 1024;
 752         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 753         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 754         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 755                 int ix = locate_object_entry_hash(oe->sha1);
 756                 if (0 <= ix)
 757                         continue;
 758                 ix = -1 - ix;
 759                 object_ix[ix] = i + 1;
 760         }
 761 }
 762
 763 static unsigned name_hash(const char *name)
 764 {
 765         unsigned char c;
 766         unsigned hash = 0;
 767
 768         /*
 769          * This effectively just creates a sortable number from the
 770          * last sixteen non-whitespace characters. Last characters
 771          * count "most", so things that end in ".c" sort together.
 772          */
 773         while ((c = *name++) != 0) {
 774                 if (isspace(c))
 775                         continue;
 776                 hash = (hash >> 2) + (c << 24);
 777         }
 778         return hash;
 779 }
 780
 781 static int add_object_entry(const unsigned char *sha1, unsigned hash, int exclude)
 782 {
 783         struct object_entry *entry;
 784         struct packed_git *p, *found_pack = NULL;
 785         off_t found_offset = 0;
 786         int ix;
 787
 788         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 789         if (ix >= 0) {
 790                 if (exclude) {
 791                         entry = objects + object_ix[ix] - 1;
 792                         entry->preferred_base = 1;
 793                 }
 794                 return 0;
 795         }
 796
 797         if (!exclude) {
 798                 for (p = packed_git; p; p = p->next) {
 799                         off_t offset = find_pack_entry_one(sha1, p);
 800                         if (offset) {
 801                                 if (incremental)
 802                                         return 0;
 803                                 if (local && !p->pack_local)
 804                                         return 0;
 805                                 if (!found_pack) {
 806                                         found_offset = offset;
 807                                         found_pack = p;
 808                                 }
 809                         }
 810                 }
 811         }
 812
 813         if (nr_objects >= nr_alloc) {
 814                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 815                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 816         }
 817
 818         entry = objects + nr_objects++;
 819         memset(entry, 0, sizeof(*entry));
 820         hashcpy(entry->sha1, sha1);
 821         entry->hash = hash;
 822         if (exclude)
 823                 entry->preferred_base = 1;
 824         if (found_pack) {
 825                 entry->in_pack = found_pack;
 826                 entry->in_pack_offset = found_offset;
 827         }
 828
 829         if (object_ix_hashsz * 3 <= nr_objects * 4)
 830                 rehash_objects();
 831         else
 832                 object_ix[-1 - ix] = nr_objects;
 833
 834         if (progress_update) {
 835                 fprintf(stderr, "Counting objects...%u\r", nr_objects);
 836                 progress_update = 0;
 837         }
 838
 839         return 1;
 840 }
 841
 842 struct pbase_tree_cache {
 843         unsigned char sha1[20];
 844         int ref;
 845         int temporary;
 846         void *tree_data;
 847         unsigned long tree_size;
 848 };
 849
 850 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 851 static int pbase_tree_cache_ix(const unsigned char *sha1)
 852 {
 853         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 854 }
 855 static int pbase_tree_cache_ix_incr(int ix)
 856 {
 857         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 858 }
 859
 860 static struct pbase_tree {
 861         struct pbase_tree *next;
 862         /* This is a phony "cache" entry; we are not
 863          * going to evict it nor find it through _get()
 864          * mechanism -- this is for the toplevel node that
 865          * would almost always change with any commit.
 866          */
 867         struct pbase_tree_cache pcache;
 868 } *pbase_tree;
 869
 870 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 871 {
 872         struct pbase_tree_cache *ent, *nent;
 873         void *data;
 874         unsigned long size;
 875         enum object_type type;
 876         int neigh;
 877         int my_ix = pbase_tree_cache_ix(sha1);
 878         int available_ix = -1;
 879
 880         /* pbase-tree-cache acts as a limited hashtable.
 881          * your object will be found at your index or within a few
 882          * slots after that slot if it is cached.
 883          */
 884         for (neigh = 0; neigh < 8; neigh++) {
 885                 ent = pbase_tree_cache[my_ix];
 886                 if (ent && !hashcmp(ent->sha1, sha1)) {
 887                         ent->ref++;
 888                         return ent;
 889                 }
 890                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 891                          ((0 <= available_ix) &&
 892                           (!ent && pbase_tree_cache[available_ix])))
 893                         available_ix = my_ix;
 894                 if (!ent)
 895                         break;
 896                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 897         }
 898
 899         /* Did not find one.  Either we got a bogus request or
 900          * we need to read and perhaps cache.
 901          */
 902         data = read_sha1_file(sha1, &type, &size);
 903         if (!data)
 904                 return NULL;
 905         if (type != OBJ_TREE) {
 906                 free(data);
 907                 return NULL;
 908         }
 909
 910         /* We need to either cache or return a throwaway copy */
 911
 912         if (available_ix < 0)
 913                 ent = NULL;
 914         else {
 915                 ent = pbase_tree_cache[available_ix];
 916                 my_ix = available_ix;
 917         }
 918
 919         if (!ent) {
 920                 nent = xmalloc(sizeof(*nent));
 921                 nent->temporary = (available_ix < 0);
 922         }
 923         else {
 924                 /* evict and reuse */
 925                 free(ent->tree_data);
 926                 nent = ent;
 927         }
 928         hashcpy(nent->sha1, sha1);
 929         nent->tree_data = data;
 930         nent->tree_size = size;
 931         nent->ref = 1;
 932         if (!nent->temporary)
 933                 pbase_tree_cache[my_ix] = nent;
 934         return nent;
 935 }
 936
 937 static void pbase_tree_put(struct pbase_tree_cache *cache)
 938 {
 939         if (!cache->temporary) {
 940                 cache->ref--;
 941                 return;
 942         }
 943         free(cache->tree_data);
 944         free(cache);
 945 }
 946
 947 static int name_cmp_len(const char *name)
 948 {
 949         int i;
 950         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 951                 ;
 952         return i;
 953 }
 954
 955 static void add_pbase_object(struct tree_desc *tree,
 956                              const char *name,
 957                              int cmplen,
 958                              const char *fullname)
 959 {
 960         struct name_entry entry;
 961         int cmp;
 962
 963         while (tree_entry(tree,&entry)) {
 964                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 965                       memcmp(name, entry.path, cmplen);
 966                 if (cmp > 0)
 967                         continue;
 968                 if (cmp < 0)
 969                         return;
 970                 if (name[cmplen] != '/') {
 971                         unsigned hash = name_hash(fullname);
 972                         add_object_entry(entry.sha1, hash, 1);
 973                         return;
 974                 }
 975                 if (S_ISDIR(entry.mode)) {
 976                         struct tree_desc sub;
 977                         struct pbase_tree_cache *tree;
 978                         const char *down = name+cmplen+1;
 979                         int downlen = name_cmp_len(down);
 980
 981                         tree = pbase_tree_get(entry.sha1);
 982                         if (!tree)
 983                                 return;
 984                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
 985
 986                         add_pbase_object(&sub, down, downlen, fullname);
 987                         pbase_tree_put(tree);
 988                 }
 989         }
 990 }
 991
 992 static unsigned *done_pbase_paths;
 993 static int done_pbase_paths_num;
 994 static int done_pbase_paths_alloc;
 995 static int done_pbase_path_pos(unsigned hash)
 996 {
 997         int lo = 0;
 998         int hi = done_pbase_paths_num;
 999         while (lo < hi) {
1000                 int mi = (hi + lo) / 2;
1001                 if (done_pbase_paths[mi] == hash)
1002                         return mi;
1003                 if (done_pbase_paths[mi] < hash)
1004                         hi = mi;
1005                 else
1006                         lo = mi + 1;
1007         }
1008         return -lo-1;
1009 }
1010
1011 static int check_pbase_path(unsigned hash)
1012 {
1013         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
1014         if (0 <= pos)
1015                 return 1;
1016         pos = -pos - 1;
1017         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
1018                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
1019                 done_pbase_paths = xrealloc(done_pbase_paths,
1020                                             done_pbase_paths_alloc *
1021                                             sizeof(unsigned));
1022         }
1023         done_pbase_paths_num++;
1024         if (pos < done_pbase_paths_num)
1025                 memmove(done_pbase_paths + pos + 1,
1026                         done_pbase_paths + pos,
1027                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1028         done_pbase_paths[pos] = hash;
1029         return 0;
1030 }
1031
1032 static void add_preferred_base_object(const char *name, unsigned hash)
1033 {
1034         struct pbase_tree *it;
1035         int cmplen;
1036
1037         if (!num_preferred_base || check_pbase_path(hash))
1038                 return;
1039
1040         cmplen = name_cmp_len(name);
1041         for (it = pbase_tree; it; it = it->next) {
1042                 if (cmplen == 0) {
1043                         add_object_entry(it->pcache.sha1, 0, 1);
1044                 }
1045                 else {
1046                         struct tree_desc tree;
1047                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1048                         add_pbase_object(&tree, name, cmplen, name);
1049                 }
1050         }
1051 }
1052
1053 static void add_preferred_base(unsigned char *sha1)
1054 {
1055         struct pbase_tree *it;
1056         void *data;
1057         unsigned long size;
1058         unsigned char tree_sha1[20];
1059
1060         if (window <= num_preferred_base++)
1061                 return;
1062
1063         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1064         if (!data)
1065                 return;
1066
1067         for (it = pbase_tree; it; it = it->next) {
1068                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1069                         free(data);
1070                         return;
1071                 }
1072         }
1073
1074         it = xcalloc(1, sizeof(*it));
1075         it->next = pbase_tree;
1076         pbase_tree = it;
1077
1078         hashcpy(it->pcache.sha1, tree_sha1);
1079         it->pcache.tree_data = data;
1080         it->pcache.tree_size = size;
1081 }
1082
1083 static void check_object(struct object_entry *entry)
1084 {
1085         if (entry->in_pack && !entry->preferred_base) {
1086                 struct packed_git *p = entry->in_pack;
1087                 struct pack_window *w_curs = NULL;
1088                 unsigned long size, used;
1089                 unsigned int avail;
1090                 unsigned char *buf;
1091                 struct object_entry *base_entry = NULL;
1092
1093                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1094
1095                 /* We want in_pack_type even if we do not reuse delta.
1096                  * There is no point not reusing non-delta representations.
1097                  */
1098                 used = unpack_object_header_gently(buf, avail,
1099                                                    &entry->in_pack_type, &size);
1100
1101                 /* Check if it is delta, and the base is also an object
1102                  * we are going to pack.  If so we will reuse the existing
1103                  * delta.
1104                  */
1105                 if (!no_reuse_delta) {
1106                         unsigned char c;
1107                         const unsigned char *base_name;
1108                         off_t ofs;
1109                         unsigned long used_0;
1110                         /* there is at least 20 bytes left in the pack */
1111                         switch (entry->in_pack_type) {
1112                         case OBJ_REF_DELTA:
1113                                 base_name = use_pack(p, &w_curs,
1114                                         entry->in_pack_offset + used, NULL);
1115                                 used += 20;
1116                                 break;
1117                         case OBJ_OFS_DELTA:
1118                                 buf = use_pack(p, &w_curs,
1119                                         entry->in_pack_offset + used, NULL);
1120                                 used_0 = 0;
1121                                 c = buf[used_0++];
1122                                 ofs = c & 127;
1123                                 while (c & 128) {
1124                                         ofs += 1;
1125                                         if (!ofs || MSB(ofs, 7))
1126                                                 die("delta base offset overflow in pack for %s",
1127                                                     sha1_to_hex(entry->sha1));
1128                                         c = buf[used_0++];
1129                                         ofs = (ofs << 7) + (c & 127);
1130                                 }
1131                                 if (ofs >= entry->in_pack_offset)
1132                                         die("delta base offset out of bound for %s",
1133                                             sha1_to_hex(entry->sha1));
1134                                 ofs = entry->in_pack_offset - ofs;
1135                                 base_name = find_packed_object_name(p, ofs);
1136                                 used += used_0;
1137                                 break;
1138                         default:
1139                                 base_name = NULL;
1140                         }
1141                         if (base_name)
1142                                 base_entry = locate_object_entry(base_name);
1143                 }
1144                 unuse_pack(&w_curs);
1145                 entry->in_pack_header_size = used;
1146
1147                 if (base_entry) {
1148
1149                         /* Depth value does not matter - find_deltas()
1150                          * will never consider reused delta as the
1151                          * base object to deltify other objects
1152                          * against, in order to avoid circular deltas.
1153                          */
1154
1155                         /* uncompressed size of the delta data */
1156                         entry->size = size;
1157                         entry->delta = base_entry;
1158                         entry->type = entry->in_pack_type;
1159
1160                         entry->delta_sibling = base_entry->delta_child;
1161                         base_entry->delta_child = entry;
1162
1163                         return;
1164                 }
1165                 /* Otherwise we would do the usual */
1166         }
1167
1168         entry->type = sha1_object_info(entry->sha1, &entry->size);
1169         if (entry->type < 0)
1170                 die("unable to get type of object %s",
1171                     sha1_to_hex(entry->sha1));
1172 }
1173
1174 static void get_object_details(void)
1175 {
1176         uint32_t i;
1177         struct object_entry *entry;
1178
1179         prepare_pack_ix();
1180         for (i = 0, entry = objects; i < nr_objects; i++, entry++)
1181                 check_object(entry);
1182 }
1183
1184 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
1185
1186 static entry_sort_t current_sort;
1187
1188 static int sort_comparator(const void *_a, const void *_b)
1189 {
1190         struct object_entry *a = *(struct object_entry **)_a;
1191         struct object_entry *b = *(struct object_entry **)_b;
1192         return current_sort(a,b);
1193 }
1194
1195 static struct object_entry **create_sorted_list(entry_sort_t sort)
1196 {
1197         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
1198         uint32_t i;
1199
1200         for (i = 0; i < nr_objects; i++)
1201                 list[i] = objects + i;
1202         current_sort = sort;
1203         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
1204         return list;
1205 }
1206
1207 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
1208 {
1209         return hashcmp(a->sha1, b->sha1);
1210 }
1211
1212 static struct object_entry **create_final_object_list(void)
1213 {
1214         struct object_entry **list;
1215         uint32_t i, j;
1216
1217         for (i = nr_result = 0; i < nr_objects; i++)
1218                 if (!objects[i].preferred_base)
1219                         nr_result++;
1220         list = xmalloc(nr_result * sizeof(struct object_entry *));
1221         for (i = j = 0; i < nr_objects; i++) {
1222                 if (!objects[i].preferred_base)
1223                         list[j++] = objects + i;
1224         }
1225         current_sort = sha1_sort;
1226         qsort(list, nr_result, sizeof(struct object_entry *), sort_comparator);
1227         return list;
1228 }
1229
1230 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
1231 {
1232         if (a->type < b->type)
1233                 return -1;
1234         if (a->type > b->type)
1235                 return 1;
1236         if (a->hash < b->hash)
1237                 return -1;
1238         if (a->hash > b->hash)
1239                 return 1;
1240         if (a->preferred_base < b->preferred_base)
1241                 return -1;
1242         if (a->preferred_base > b->preferred_base)
1243                 return 1;
1244         if (a->size < b->size)
1245                 return -1;
1246         if (a->size > b->size)
1247                 return 1;
1248         return a > b ? -1 : (a < b);  /* newest last */
1249 }
1250
1251 struct unpacked {
1252         struct object_entry *entry;
1253         void *data;
1254         struct delta_index *index;
1255 };
1256
1257 /*
1258  * We search for deltas _backwards_ in a list sorted by type and
1259  * by size, so that we see progressively smaller and smaller files.
1260  * That's because we prefer deltas to be from the bigger file
1261  * to the smaller - deletes are potentially cheaper, but perhaps
1262  * more importantly, the bigger file is likely the more recent
1263  * one.
1264  */
1265 static int try_delta(struct unpacked *trg, struct unpacked *src,
1266                      unsigned max_depth)
1267 {
1268         struct object_entry *trg_entry = trg->entry;
1269         struct object_entry *src_entry = src->entry;
1270         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1271         enum object_type type;
1272         void *delta_buf;
1273
1274         /* Don't bother doing diffs between different types */
1275         if (trg_entry->type != src_entry->type)
1276                 return -1;
1277
1278         /* We do not compute delta to *create* objects we are not
1279          * going to pack.
1280          */
1281         if (trg_entry->preferred_base)
1282                 return -1;
1283
1284         /*
1285          * We do not bother to try a delta that we discarded
1286          * on an earlier try, but only when reusing delta data.
1287          */
1288         if (!no_reuse_delta && trg_entry->in_pack &&
1289             trg_entry->in_pack == src_entry->in_pack &&
1290             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1291             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1292                 return 0;
1293
1294         /* Let's not bust the allowed depth. */
1295         if (src_entry->depth >= max_depth)
1296                 return 0;
1297
1298         /* Now some size filtering heuristics. */
1299         trg_size = trg_entry->size;
1300         max_size = trg_size/2 - 20;
1301         max_size = max_size * (max_depth - src_entry->depth) / max_depth;
1302         if (max_size == 0)
1303                 return 0;
1304         if (trg_entry->delta && trg_entry->delta_size <= max_size)
1305                 max_size = trg_entry->delta_size-1;
1306         src_size = src_entry->size;
1307         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1308         if (sizediff >= max_size)
1309                 return 0;
1310
1311         /* Load data if not already done */
1312         if (!trg->data) {
1313                 trg->data = read_sha1_file(trg_entry->sha1, &type, &sz);
1314                 if (sz != trg_size)
1315                         die("object %s inconsistent object length (%lu vs %lu)",
1316                             sha1_to_hex(trg_entry->sha1), sz, trg_size);
1317         }
1318         if (!src->data) {
1319                 src->data = read_sha1_file(src_entry->sha1, &type, &sz);
1320                 if (sz != src_size)
1321                         die("object %s inconsistent object length (%lu vs %lu)",
1322                             sha1_to_hex(src_entry->sha1), sz, src_size);
1323         }
1324         if (!src->index) {
1325                 src->index = create_delta_index(src->data, src_size);
1326                 if (!src->index)
1327                         die("out of memory");
1328         }
1329
1330         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1331         if (!delta_buf)
1332                 return 0;
1333
1334         trg_entry->delta = src_entry;
1335         trg_entry->delta_size = delta_size;
1336         trg_entry->depth = src_entry->depth + 1;
1337         free(delta_buf);
1338         return 1;
1339 }
1340
1341 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1342 {
1343         struct object_entry *child = me->delta_child;
1344         unsigned int m = n;
1345         while (child) {
1346                 unsigned int c = check_delta_limit(child, n + 1);
1347                 if (m < c)
1348                         m = c;
1349                 child = child->delta_sibling;
1350         }
1351         return m;
1352 }
1353
1354 static void find_deltas(struct object_entry **list, int window, int depth)
1355 {
1356         uint32_t i = nr_objects, idx = 0, processed = 0;
1357         unsigned int array_size = window * sizeof(struct unpacked);
1358         struct unpacked *array;
1359         unsigned last_percent = 999;
1360         int max_depth;
1361
1362         if (!nr_objects)
1363                 return;
1364         array = xmalloc(array_size);
1365         memset(array, 0, array_size);
1366         if (progress)
1367                 fprintf(stderr, "Deltifying %u objects.\n", nr_result);
1368
1369         do {
1370                 struct object_entry *entry = list[--i];
1371                 struct unpacked *n = array + idx;
1372                 int j;
1373
1374                 if (!entry->preferred_base)
1375                         processed++;
1376
1377                 if (progress) {
1378                         unsigned percent = processed * 100 / nr_result;
1379                         if (percent != last_percent || progress_update) {
1380                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
1381                                         percent, processed, nr_result);
1382                                 progress_update = 0;
1383                                 last_percent = percent;
1384                         }
1385                 }
1386
1387                 if (entry->delta)
1388                         /* This happens if we decided to reuse existing
1389                          * delta from a pack.  "!no_reuse_delta &&" is implied.
1390                          */
1391                         continue;
1392
1393                 if (entry->size < 50)
1394                         continue;
1395                 free_delta_index(n->index);
1396                 n->index = NULL;
1397                 free(n->data);
1398                 n->data = NULL;
1399                 n->entry = entry;
1400
1401                 /*
1402                  * If the current object is at pack edge, take the depth the
1403                  * objects that depend on the current object into account
1404                  * otherwise they would become too deep.
1405                  */
1406                 max_depth = depth;
1407                 if (entry->delta_child) {
1408                         max_depth -= check_delta_limit(entry, 0);
1409                         if (max_depth <= 0)
1410                                 goto next;
1411                 }
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, max_depth) < 0)
1423                                 break;
1424                 }
1425
1426                 /* if we made n a delta, and if n is already at max
1427                  * depth, leaving it in the window is pointless.  we
1428                  * should evict it first.
1429                  */
1430                 if (entry->delta && depth <= entry->depth)
1431                         continue;
1432
1433                 next:
1434                 idx++;
1435                 if (idx >= window)
1436                         idx = 0;
1437         } while (i > 0);
1438
1439         if (progress)
1440                 fputc('\n', stderr);
1441
1442         for (i = 0; i < window; ++i) {
1443                 free_delta_index(array[i].index);
1444                 free(array[i].data);
1445         }
1446         free(array);
1447 }
1448
1449 static void prepare_pack(int window, int depth)
1450 {
1451         get_object_details();
1452         sorted_by_type = create_sorted_list(type_size_sort);
1453         if (window && depth)
1454                 find_deltas(sorted_by_type, window+1, depth);
1455 }
1456
1457 static int reuse_cached_pack(unsigned char *sha1)
1458 {
1459         static const char cache[] = "pack-cache/pack-%s.%s";
1460         char *cached_pack, *cached_idx;
1461         int ifd, ofd, ifd_ix = -1;
1462
1463         cached_pack = git_path(cache, sha1_to_hex(sha1), "pack");
1464         ifd = open(cached_pack, O_RDONLY);
1465         if (ifd < 0)
1466                 return 0;
1467
1468         if (!pack_to_stdout) {
1469                 cached_idx = git_path(cache, sha1_to_hex(sha1), "idx");
1470                 ifd_ix = open(cached_idx, O_RDONLY);
1471                 if (ifd_ix < 0) {
1472                         close(ifd);
1473                         return 0;
1474                 }
1475         }
1476
1477         if (progress)
1478                 fprintf(stderr, "Reusing %u objects pack %s\n", nr_objects,
1479                         sha1_to_hex(sha1));
1480
1481         if (pack_to_stdout) {
1482                 if (copy_fd(ifd, 1))
1483                         exit(1);
1484                 close(ifd);
1485         }
1486         else {
1487                 char name[PATH_MAX];
1488                 snprintf(name, sizeof(name),
1489                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "pack");
1490                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1491                 if (ofd < 0)
1492                         die("unable to open %s (%s)", name, strerror(errno));
1493                 if (copy_fd(ifd, ofd))
1494                         exit(1);
1495                 close(ifd);
1496
1497                 snprintf(name, sizeof(name),
1498                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "idx");
1499                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1500                 if (ofd < 0)
1501                         die("unable to open %s (%s)", name, strerror(errno));
1502                 if (copy_fd(ifd_ix, ofd))
1503                         exit(1);
1504                 close(ifd_ix);
1505                 puts(sha1_to_hex(sha1));
1506         }
1507
1508         return 1;
1509 }
1510
1511 static void progress_interval(int signum)
1512 {
1513         progress_update = 1;
1514 }
1515
1516 static void setup_progress_signal(void)
1517 {
1518         struct sigaction sa;
1519         struct itimerval v;
1520
1521         memset(&sa, 0, sizeof(sa));
1522         sa.sa_handler = progress_interval;
1523         sigemptyset(&sa.sa_mask);
1524         sa.sa_flags = SA_RESTART;
1525         sigaction(SIGALRM, &sa, NULL);
1526
1527         v.it_interval.tv_sec = 1;
1528         v.it_interval.tv_usec = 0;
1529         v.it_value = v.it_interval;
1530         setitimer(ITIMER_REAL, &v, NULL);
1531 }
1532
1533 static int git_pack_config(const char *k, const char *v)
1534 {
1535         if(!strcmp(k, "pack.window")) {
1536                 window = git_config_int(k, v);
1537                 return 0;
1538         }
1539         return git_default_config(k, v);
1540 }
1541
1542 static void read_object_list_from_stdin(void)
1543 {
1544         char line[40 + 1 + PATH_MAX + 2];
1545         unsigned char sha1[20];
1546         unsigned hash;
1547
1548         for (;;) {
1549                 if (!fgets(line, sizeof(line), stdin)) {
1550                         if (feof(stdin))
1551                                 break;
1552                         if (!ferror(stdin))
1553                                 die("fgets returned NULL, not EOF, not error!");
1554                         if (errno != EINTR)
1555                                 die("fgets: %s", strerror(errno));
1556                         clearerr(stdin);
1557                         continue;
1558                 }
1559                 if (line[0] == '-') {
1560                         if (get_sha1_hex(line+1, sha1))
1561                                 die("expected edge sha1, got garbage:\n %s",
1562                                     line);
1563                         add_preferred_base(sha1);
1564                         continue;
1565                 }
1566                 if (get_sha1_hex(line, sha1))
1567                         die("expected sha1, got garbage:\n %s", line);
1568
1569                 hash = name_hash(line+41);
1570                 add_preferred_base_object(line+41, hash);
1571                 add_object_entry(sha1, hash, 0);
1572         }
1573 }
1574
1575 static void show_commit(struct commit *commit)
1576 {
1577         add_object_entry(commit->object.sha1, 0, 0);
1578 }
1579
1580 static void show_object(struct object_array_entry *p)
1581 {
1582         unsigned hash = name_hash(p->name);
1583         add_preferred_base_object(p->name, hash);
1584         add_object_entry(p->item->sha1, hash, 0);
1585 }
1586
1587 static void show_edge(struct commit *commit)
1588 {
1589         add_preferred_base(commit->object.sha1);
1590 }
1591
1592 static void get_object_list(int ac, const char **av)
1593 {
1594         struct rev_info revs;
1595         char line[1000];
1596         int flags = 0;
1597
1598         init_revisions(&revs, NULL);
1599         save_commit_buffer = 0;
1600         track_object_refs = 0;
1601         setup_revisions(ac, av, &revs, NULL);
1602
1603         while (fgets(line, sizeof(line), stdin) != NULL) {
1604                 int len = strlen(line);
1605                 if (line[len - 1] == '\n')
1606                         line[--len] = 0;
1607                 if (!len)
1608                         break;
1609                 if (*line == '-') {
1610                         if (!strcmp(line, "--not")) {
1611                                 flags ^= UNINTERESTING;
1612                                 continue;
1613                         }
1614                         die("not a rev '%s'", line);
1615                 }
1616                 if (handle_revision_arg(line, &revs, flags, 1))
1617                         die("bad revision '%s'", line);
1618         }
1619
1620         prepare_revision_walk(&revs);
1621         mark_edges_uninteresting(revs.commits, &revs, show_edge);
1622         traverse_commit_list(&revs, show_commit, show_object);
1623 }
1624
1625 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1626 {
1627         SHA_CTX ctx;
1628         int depth = 10;
1629         struct object_entry **list;
1630         int use_internal_rev_list = 0;
1631         int thin = 0;
1632         uint32_t i;
1633         const char **rp_av;
1634         int rp_ac_alloc = 64;
1635         int rp_ac;
1636
1637         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1638
1639         rp_av[0] = "pack-objects";
1640         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1641         rp_ac = 2;
1642
1643         git_config(git_pack_config);
1644
1645         progress = isatty(2);
1646         for (i = 1; i < argc; i++) {
1647                 const char *arg = argv[i];
1648
1649                 if (*arg != '-')
1650                         break;
1651
1652                 if (!strcmp("--non-empty", arg)) {
1653                         non_empty = 1;
1654                         continue;
1655                 }
1656                 if (!strcmp("--local", arg)) {
1657                         local = 1;
1658                         continue;
1659                 }
1660                 if (!strcmp("--incremental", arg)) {
1661                         incremental = 1;
1662                         continue;
1663                 }
1664                 if (!prefixcmp(arg, "--window=")) {
1665                         char *end;
1666                         window = strtoul(arg+9, &end, 0);
1667                         if (!arg[9] || *end)
1668                                 usage(pack_usage);
1669                         continue;
1670                 }
1671                 if (!prefixcmp(arg, "--depth=")) {
1672                         char *end;
1673                         depth = strtoul(arg+8, &end, 0);
1674                         if (!arg[8] || *end)
1675                                 usage(pack_usage);
1676                         continue;
1677                 }
1678                 if (!strcmp("--progress", arg)) {
1679                         progress = 1;
1680                         continue;
1681                 }
1682                 if (!strcmp("--all-progress", arg)) {
1683                         progress = 2;
1684                         continue;
1685                 }
1686                 if (!strcmp("-q", arg)) {
1687                         progress = 0;
1688                         continue;
1689                 }
1690                 if (!strcmp("--no-reuse-delta", arg)) {
1691                         no_reuse_delta = 1;
1692                         continue;
1693                 }
1694                 if (!strcmp("--delta-base-offset", arg)) {
1695                         allow_ofs_delta = 1;
1696                         continue;
1697                 }
1698                 if (!strcmp("--stdout", arg)) {
1699                         pack_to_stdout = 1;
1700                         continue;
1701                 }
1702                 if (!strcmp("--revs", arg)) {
1703                         use_internal_rev_list = 1;
1704                         continue;
1705                 }
1706                 if (!strcmp("--unpacked", arg) ||
1707                     !prefixcmp(arg, "--unpacked=") ||
1708                     !strcmp("--reflog", arg) ||
1709                     !strcmp("--all", arg)) {
1710                         use_internal_rev_list = 1;
1711                         if (rp_ac >= rp_ac_alloc - 1) {
1712                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1713                                 rp_av = xrealloc(rp_av,
1714                                                  rp_ac_alloc * sizeof(*rp_av));
1715                         }
1716                         rp_av[rp_ac++] = arg;
1717                         continue;
1718                 }
1719                 if (!strcmp("--thin", arg)) {
1720                         use_internal_rev_list = 1;
1721                         thin = 1;
1722                         rp_av[1] = "--objects-edge";
1723                         continue;
1724                 }
1725                 if (!prefixcmp(arg, "--index-version=")) {
1726                         char *c;
1727                         index_default_version = strtoul(arg + 16, &c, 10);
1728                         if (index_default_version > 2)
1729                                 die("bad %s", arg);
1730                         if (*c == ',')
1731                                 index_off32_limit = strtoul(c+1, &c, 0);
1732                         if (*c || index_off32_limit & 0x80000000)
1733                                 die("bad %s", arg);
1734                         continue;
1735                 }
1736                 usage(pack_usage);
1737         }
1738
1739         /* Traditionally "pack-objects [options] base extra" failed;
1740          * we would however want to take refs parameter that would
1741          * have been given to upstream rev-list ourselves, which means
1742          * we somehow want to say what the base name is.  So the
1743          * syntax would be:
1744          *
1745          * pack-objects [options] base <refs...>
1746          *
1747          * in other words, we would treat the first non-option as the
1748          * base_name and send everything else to the internal revision
1749          * walker.
1750          */
1751
1752         if (!pack_to_stdout)
1753                 base_name = argv[i++];
1754
1755         if (pack_to_stdout != !base_name)
1756                 usage(pack_usage);
1757
1758         if (!pack_to_stdout && thin)
1759                 die("--thin cannot be used to build an indexable pack.");
1760
1761         prepare_packed_git();
1762
1763         if (progress) {
1764                 fprintf(stderr, "Generating pack...\n");
1765                 setup_progress_signal();
1766         }
1767
1768         if (!use_internal_rev_list)
1769                 read_object_list_from_stdin();
1770         else {
1771                 rp_av[rp_ac] = NULL;
1772                 get_object_list(rp_ac, rp_av);
1773         }
1774
1775         if (progress)
1776                 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
1777         sorted_by_sha = create_final_object_list();
1778         if (non_empty && !nr_result)
1779                 return 0;
1780
1781         SHA1_Init(&ctx);
1782         list = sorted_by_sha;
1783         for (i = 0; i < nr_result; i++) {
1784                 struct object_entry *entry = *list++;
1785                 SHA1_Update(&ctx, entry->sha1, 20);
1786         }
1787         SHA1_Final(object_list_sha1, &ctx);
1788         if (progress && (nr_objects != nr_result))
1789                 fprintf(stderr, "Result has %u objects.\n", nr_result);
1790
1791         if (reuse_cached_pack(object_list_sha1))
1792                 ;
1793         else {
1794                 off_t last_obj_offset;
1795                 if (nr_result)
1796                         prepare_pack(window, depth);
1797                 if (progress == 1 && pack_to_stdout) {
1798                         /* the other end usually displays progress itself */
1799                         struct itimerval v = {{0,},};
1800                         setitimer(ITIMER_REAL, &v, NULL);
1801                         signal(SIGALRM, SIG_IGN );
1802                         progress_update = 0;
1803                 }
1804                 last_obj_offset = write_pack_file();
1805                 if (!pack_to_stdout) {
1806                         write_index_file(last_obj_offset);
1807                         puts(sha1_to_hex(object_list_sha1));
1808                 }
1809         }
1810         if (progress)
1811                 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
1812                         written, written_delta, reused, reused_delta);
1813         return 0;
1814 }