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