git.oblomov.eu Git - git/blob - builtin-pack-objects.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "attr.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "delta.h"
  10 #include "pack.h"
  11 #include "csum-file.h"
  12 #include "tree-walk.h"
  13 #include "diff.h"
  14 #include "revision.h"
  15 #include "list-objects.h"
  16 #include "progress.h"
  17
  18 static const char pack_usage[] = "\
  19 git-pack-objects [{ -q | --progress | --all-progress }] [--max-pack-size=N] \n\
  20         [--local] [--incremental] [--window=N] [--depth=N] \n\
  21         [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
  22         [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
  23         [--stdout | base-name] [<ref-list | <object-list]";
  24
  25 struct object_entry {
  26         struct pack_idx_entry idx;
  27         unsigned long size;     /* uncompressed size */
  28
  29         unsigned int hash;      /* name hint hash */
  30         unsigned int depth;     /* delta depth */
  31         struct packed_git *in_pack;     /* already in pack */
  32         off_t in_pack_offset;
  33         struct object_entry *delta;     /* delta base object */
  34         struct object_entry *delta_child; /* deltified objects who bases me */
  35         struct object_entry *delta_sibling; /* other deltified objects who
  36                                              * uses the same base as me
  37                                              */
  38         void *delta_data;       /* cached delta (uncompressed) */
  39         unsigned long delta_size;       /* delta data size (uncompressed) */
  40         enum object_type type;
  41         enum object_type in_pack_type;  /* could be delta */
  42         unsigned char in_pack_header_size;
  43         unsigned char preferred_base; /* we do not pack this, but is available
  44                                        * to be used as the base object to delta
  45                                        * objects against.
  46                                        */
  47         unsigned char no_try_delta;
  48 };
  49
  50 /*
  51  * Objects we are going to pack are collected in objects array (dynamically
  52  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  53  * in the order we see -- typically rev-list --objects order that gives us
  54  * nice "minimum seek" order.
  55  */
  56 static struct object_entry *objects;
  57 static struct object_entry **written_list;
  58 static uint32_t nr_objects, nr_alloc, nr_result, nr_written;
  59
  60 static int non_empty;
  61 static int no_reuse_delta, no_reuse_object;
  62 static int local;
  63 static int incremental;
  64 static int allow_ofs_delta;
  65 static const char *pack_tmp_name, *idx_tmp_name;
  66 static char tmpname[PATH_MAX];
  67 static const char *base_name;
  68 static int progress = 1;
  69 static int window = 10;
  70 static uint32_t pack_size_limit;
  71 static int depth = 50;
  72 static int pack_to_stdout;
  73 static int num_preferred_base;
  74 static struct progress progress_state;
  75 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  76 static int pack_compression_seen;
  77
  78 static unsigned long delta_cache_size = 0;
  79 static unsigned long max_delta_cache_size = 0;
  80 static unsigned long cache_max_small_delta_size = 1000;
  81
  82 /*
  83  * The object names in objects array are hashed with this hashtable,
  84  * to help looking up the entry by object name.
  85  * This hashtable is built after all the objects are seen.
  86  */
  87 static int *object_ix;
  88 static int object_ix_hashsz;
  89
  90 /*
  91  * Pack index for existing packs give us easy access to the offsets into
  92  * corresponding pack file where each object's data starts, but the entries
  93  * do not store the size of the compressed representation (uncompressed
  94  * size is easily available by examining the pack entry header).  It is
  95  * also rather expensive to find the sha1 for an object given its offset.
  96  *
  97  * We build a hashtable of existing packs (pack_revindex), and keep reverse
  98  * index here -- pack index file is sorted by object name mapping to offset;
  99  * this pack_revindex[].revindex array is a list of offset/index_nr pairs
 100  * ordered by offset, so if you know the offset of an object, next offset
 101  * is where its packed representation ends and the index_nr can be used to
 102  * get the object sha1 from the main index.
 103  */
 104 struct revindex_entry {
 105         off_t offset;
 106         unsigned int nr;
 107 };
 108 struct pack_revindex {
 109         struct packed_git *p;
 110         struct revindex_entry *revindex;
 111 };
 112 static struct  pack_revindex *pack_revindex;
 113 static int pack_revindex_hashsz;
 114
 115 /*
 116  * stats
 117  */
 118 static uint32_t written, written_delta;
 119 static uint32_t reused, reused_delta;
 120
 121 static int pack_revindex_ix(struct packed_git *p)
 122 {
 123         unsigned long ui = (unsigned long)p;
 124         int i;
 125
 126         ui = ui ^ (ui >> 16); /* defeat structure alignment */
 127         i = (int)(ui % pack_revindex_hashsz);
 128         while (pack_revindex[i].p) {
 129                 if (pack_revindex[i].p == p)
 130                         return i;
 131                 if (++i == pack_revindex_hashsz)
 132                         i = 0;
 133         }
 134         return -1 - i;
 135 }
 136
 137 static void prepare_pack_ix(void)
 138 {
 139         int num;
 140         struct packed_git *p;
 141         for (num = 0, p = packed_git; p; p = p->next)
 142                 num++;
 143         if (!num)
 144                 return;
 145         pack_revindex_hashsz = num * 11;
 146         pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
 147         for (p = packed_git; p; p = p->next) {
 148                 num = pack_revindex_ix(p);
 149                 num = - 1 - num;
 150                 pack_revindex[num].p = p;
 151         }
 152         /* revindex elements are lazily initialized */
 153 }
 154
 155 static int cmp_offset(const void *a_, const void *b_)
 156 {
 157         const struct revindex_entry *a = a_;
 158         const struct revindex_entry *b = b_;
 159         return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0;
 160 }
 161
 162 /*
 163  * Ordered list of offsets of objects in the pack.
 164  */
 165 static void prepare_pack_revindex(struct pack_revindex *rix)
 166 {
 167         struct packed_git *p = rix->p;
 168         int num_ent = p->num_objects;
 169         int i;
 170         const char *index = p->index_data;
 171
 172         rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
 173         index += 4 * 256;
 174
 175         if (p->index_version > 1) {
 176                 const uint32_t *off_32 =
 177                         (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
 178                 const uint32_t *off_64 = off_32 + p->num_objects;
 179                 for (i = 0; i < num_ent; i++) {
 180                         uint32_t off = ntohl(*off_32++);
 181                         if (!(off & 0x80000000)) {
 182                                 rix->revindex[i].offset = off;
 183                         } else {
 184                                 rix->revindex[i].offset =
 185                                         ((uint64_t)ntohl(*off_64++)) << 32;
 186                                 rix->revindex[i].offset |=
 187                                         ntohl(*off_64++);
 188                         }
 189                         rix->revindex[i].nr = i;
 190                 }
 191         } else {
 192                 for (i = 0; i < num_ent; i++) {
 193                         uint32_t hl = *((uint32_t *)(index + 24 * i));
 194                         rix->revindex[i].offset = ntohl(hl);
 195                         rix->revindex[i].nr = i;
 196                 }
 197         }
 198
 199         /* This knows the pack format -- the 20-byte trailer
 200          * follows immediately after the last object data.
 201          */
 202         rix->revindex[num_ent].offset = p->pack_size - 20;
 203         rix->revindex[num_ent].nr = -1;
 204         qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
 205 }
 206
 207 static struct revindex_entry * find_packed_object(struct packed_git *p,
 208                                                   off_t ofs)
 209 {
 210         int num;
 211         int lo, hi;
 212         struct pack_revindex *rix;
 213         struct revindex_entry *revindex;
 214         num = pack_revindex_ix(p);
 215         if (num < 0)
 216                 die("internal error: pack revindex uninitialized");
 217         rix = &pack_revindex[num];
 218         if (!rix->revindex)
 219                 prepare_pack_revindex(rix);
 220         revindex = rix->revindex;
 221         lo = 0;
 222         hi = p->num_objects + 1;
 223         do {
 224                 int mi = (lo + hi) / 2;
 225                 if (revindex[mi].offset == ofs) {
 226                         return revindex + mi;
 227                 }
 228                 else if (ofs < revindex[mi].offset)
 229                         hi = mi;
 230                 else
 231                         lo = mi + 1;
 232         } while (lo < hi);
 233         die("internal error: pack revindex corrupt");
 234 }
 235
 236 static const unsigned char *find_packed_object_name(struct packed_git *p,
 237                                                     off_t ofs)
 238 {
 239         struct revindex_entry *entry = find_packed_object(p, ofs);
 240         return nth_packed_object_sha1(p, entry->nr);
 241 }
 242
 243 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 244 {
 245         unsigned long othersize, delta_size;
 246         enum object_type type;
 247         void *otherbuf = read_sha1_file(entry->delta->idx.sha1, &type, &othersize);
 248         void *delta_buf;
 249
 250         if (!otherbuf)
 251                 die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 252         delta_buf = diff_delta(otherbuf, othersize,
 253                                buf, size, &delta_size, 0);
 254         if (!delta_buf || delta_size != entry->delta_size)
 255                 die("delta size changed");
 256         free(buf);
 257         free(otherbuf);
 258         return delta_buf;
 259 }
 260
 261 /*
 262  * The per-object header is a pretty dense thing, which is
 263  *  - first byte: low four bits are "size", then three bits of "type",
 264  *    and the high bit is "size continues".
 265  *  - each byte afterwards: low seven bits are size continuation,
 266  *    with the high bit being "size continues"
 267  */
 268 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 269 {
 270         int n = 1;
 271         unsigned char c;
 272
 273         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 274                 die("bad type %d", type);
 275
 276         c = (type << 4) | (size & 15);
 277         size >>= 4;
 278         while (size) {
 279                 *hdr++ = c | 0x80;
 280                 c = size & 0x7f;
 281                 size >>= 7;
 282                 n++;
 283         }
 284         *hdr = c;
 285         return n;
 286 }
 287
 288 /*
 289  * we are going to reuse the existing object data as is.  make
 290  * sure it is not corrupt.
 291  */
 292 static int check_pack_inflate(struct packed_git *p,
 293                 struct pack_window **w_curs,
 294                 off_t offset,
 295                 off_t len,
 296                 unsigned long expect)
 297 {
 298         z_stream stream;
 299         unsigned char fakebuf[4096], *in;
 300         int st;
 301
 302         memset(&stream, 0, sizeof(stream));
 303         inflateInit(&stream);
 304         do {
 305                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 306                 stream.next_in = in;
 307                 stream.next_out = fakebuf;
 308                 stream.avail_out = sizeof(fakebuf);
 309                 st = inflate(&stream, Z_FINISH);
 310                 offset += stream.next_in - in;
 311         } while (st == Z_OK || st == Z_BUF_ERROR);
 312         inflateEnd(&stream);
 313         return (st == Z_STREAM_END &&
 314                 stream.total_out == expect &&
 315                 stream.total_in == len) ? 0 : -1;
 316 }
 317
 318 static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
 319                           off_t offset, off_t len, unsigned int nr)
 320 {
 321         const uint32_t *index_crc;
 322         uint32_t data_crc = crc32(0, Z_NULL, 0);
 323
 324         do {
 325                 unsigned int avail;
 326                 void *data = use_pack(p, w_curs, offset, &avail);
 327                 if (avail > len)
 328                         avail = len;
 329                 data_crc = crc32(data_crc, data, avail);
 330                 offset += avail;
 331                 len -= avail;
 332         } while (len);
 333
 334         index_crc = p->index_data;
 335         index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
 336
 337         return data_crc != ntohl(*index_crc);
 338 }
 339
 340 static void copy_pack_data(struct sha1file *f,
 341                 struct packed_git *p,
 342                 struct pack_window **w_curs,
 343                 off_t offset,
 344                 off_t len)
 345 {
 346         unsigned char *in;
 347         unsigned int avail;
 348
 349         while (len) {
 350                 in = use_pack(p, w_curs, offset, &avail);
 351                 if (avail > len)
 352                         avail = (unsigned int)len;
 353                 sha1write(f, in, avail);
 354                 offset += avail;
 355                 len -= avail;
 356         }
 357 }
 358
 359 static unsigned long write_object(struct sha1file *f,
 360                                   struct object_entry *entry,
 361                                   off_t write_offset)
 362 {
 363         unsigned long size;
 364         enum object_type type;
 365         void *buf;
 366         unsigned char header[10];
 367         unsigned char dheader[10];
 368         unsigned hdrlen;
 369         off_t datalen;
 370         enum object_type obj_type;
 371         int to_reuse = 0;
 372         /* write limit if limited packsize and not first object */
 373         unsigned long limit = pack_size_limit && nr_written ?
 374                                 pack_size_limit - write_offset : 0;
 375                                 /* no if no delta */
 376         int usable_delta =      !entry->delta ? 0 :
 377                                 /* yes if unlimited packfile */
 378                                 !pack_size_limit ? 1 :
 379                                 /* no if base written to previous pack */
 380                                 entry->delta->idx.offset == (off_t)-1 ? 0 :
 381                                 /* otherwise double-check written to this
 382                                  * pack,  like we do below
 383                                  */
 384                                 entry->delta->idx.offset ? 1 : 0;
 385
 386         if (!pack_to_stdout)
 387                 crc32_begin(f);
 388
 389         obj_type = entry->type;
 390         if (no_reuse_object)
 391                 to_reuse = 0;   /* explicit */
 392         else if (!entry->in_pack)
 393                 to_reuse = 0;   /* can't reuse what we don't have */
 394         else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
 395                                 /* check_object() decided it for us ... */
 396                 to_reuse = usable_delta;
 397                                 /* ... but pack split may override that */
 398         else if (obj_type != entry->in_pack_type)
 399                 to_reuse = 0;   /* pack has delta which is unusable */
 400         else if (entry->delta)
 401                 to_reuse = 0;   /* we want to pack afresh */
 402         else
 403                 to_reuse = 1;   /* we have it in-pack undeltified,
 404                                  * and we do not need to deltify it.
 405                                  */
 406
 407         if (!to_reuse) {
 408                 z_stream stream;
 409                 unsigned long maxsize;
 410                 void *out;
 411                 if (!usable_delta) {
 412                         buf = read_sha1_file(entry->idx.sha1, &obj_type, &size);
 413                         if (!buf)
 414                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 415                 } else if (entry->delta_data) {
 416                         size = entry->delta_size;
 417                         buf = entry->delta_data;
 418                         entry->delta_data = NULL;
 419                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 420                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 421                 } else {
 422                         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 423                         if (!buf)
 424                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 425                         buf = delta_against(buf, size, entry);
 426                         size = entry->delta_size;
 427                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 428                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 429                 }
 430                 /* compress the data to store and put compressed length in datalen */
 431                 memset(&stream, 0, sizeof(stream));
 432                 deflateInit(&stream, pack_compression_level);
 433                 maxsize = deflateBound(&stream, size);
 434                 out = xmalloc(maxsize);
 435                 /* Compress it */
 436                 stream.next_in = buf;
 437                 stream.avail_in = size;
 438                 stream.next_out = out;
 439                 stream.avail_out = maxsize;
 440                 while (deflate(&stream, Z_FINISH) == Z_OK)
 441                         /* nothing */;
 442                 deflateEnd(&stream);
 443                 datalen = stream.total_out;
 444                 deflateEnd(&stream);
 445                 /*
 446                  * The object header is a byte of 'type' followed by zero or
 447                  * more bytes of length.
 448                  */
 449                 hdrlen = encode_header(obj_type, size, header);
 450
 451                 if (obj_type == OBJ_OFS_DELTA) {
 452                         /*
 453                          * Deltas with relative base contain an additional
 454                          * encoding of the relative offset for the delta
 455                          * base from this object's position in the pack.
 456                          */
 457                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 458                         unsigned pos = sizeof(dheader) - 1;
 459                         dheader[pos] = ofs & 127;
 460                         while (ofs >>= 7)
 461                                 dheader[--pos] = 128 | (--ofs & 127);
 462                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 463                                 free(out);
 464                                 free(buf);
 465                                 return 0;
 466                         }
 467                         sha1write(f, header, hdrlen);
 468                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 469                         hdrlen += sizeof(dheader) - pos;
 470                 } else if (obj_type == OBJ_REF_DELTA) {
 471                         /*
 472                          * Deltas with a base reference contain
 473                          * an additional 20 bytes for the base sha1.
 474                          */
 475                         if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 476                                 free(out);
 477                                 free(buf);
 478                                 return 0;
 479                         }
 480                         sha1write(f, header, hdrlen);
 481                         sha1write(f, entry->delta->idx.sha1, 20);
 482                         hdrlen += 20;
 483                 } else {
 484                         if (limit && hdrlen + datalen + 20 >= limit) {
 485                                 free(out);
 486                                 free(buf);
 487                                 return 0;
 488                         }
 489                         sha1write(f, header, hdrlen);
 490                 }
 491                 sha1write(f, out, datalen);
 492                 free(out);
 493                 free(buf);
 494         }
 495         else {
 496                 struct packed_git *p = entry->in_pack;
 497                 struct pack_window *w_curs = NULL;
 498                 struct revindex_entry *revidx;
 499                 off_t offset;
 500
 501                 if (entry->delta) {
 502                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 503                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 504                         reused_delta++;
 505                 }
 506                 hdrlen = encode_header(obj_type, entry->size, header);
 507                 offset = entry->in_pack_offset;
 508                 revidx = find_packed_object(p, offset);
 509                 datalen = revidx[1].offset - offset;
 510                 if (!pack_to_stdout && p->index_version > 1 &&
 511                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
 512                         die("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 513                 offset += entry->in_pack_header_size;
 514                 datalen -= entry->in_pack_header_size;
 515                 if (obj_type == OBJ_OFS_DELTA) {
 516                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 517                         unsigned pos = sizeof(dheader) - 1;
 518                         dheader[pos] = ofs & 127;
 519                         while (ofs >>= 7)
 520                                 dheader[--pos] = 128 | (--ofs & 127);
 521                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit)
 522                                 return 0;
 523                         sha1write(f, header, hdrlen);
 524                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 525                         hdrlen += sizeof(dheader) - pos;
 526                 } else if (obj_type == OBJ_REF_DELTA) {
 527                         if (limit && hdrlen + 20 + datalen + 20 >= limit)
 528                                 return 0;
 529                         sha1write(f, header, hdrlen);
 530                         sha1write(f, entry->delta->idx.sha1, 20);
 531                         hdrlen += 20;
 532                 } else {
 533                         if (limit && hdrlen + datalen + 20 >= limit)
 534                                 return 0;
 535                         sha1write(f, header, hdrlen);
 536                 }
 537
 538                 if (!pack_to_stdout && p->index_version == 1 &&
 539                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
 540                         die("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 541                 copy_pack_data(f, p, &w_curs, offset, datalen);
 542                 unuse_pack(&w_curs);
 543                 reused++;
 544         }
 545         if (usable_delta)
 546                 written_delta++;
 547         written++;
 548         if (!pack_to_stdout)
 549                 entry->idx.crc32 = crc32_end(f);
 550         return hdrlen + datalen;
 551 }
 552
 553 static off_t write_one(struct sha1file *f,
 554                                struct object_entry *e,
 555                                off_t offset)
 556 {
 557         unsigned long size;
 558
 559         /* offset is non zero if object is written already. */
 560         if (e->idx.offset || e->preferred_base)
 561                 return offset;
 562
 563         /* if we are deltified, write out base object first. */
 564         if (e->delta) {
 565                 offset = write_one(f, e->delta, offset);
 566                 if (!offset)
 567                         return 0;
 568         }
 569
 570         e->idx.offset = offset;
 571         size = write_object(f, e, offset);
 572         if (!size) {
 573                 e->idx.offset = 0;
 574                 return 0;
 575         }
 576         written_list[nr_written++] = e;
 577
 578         /* make sure off_t is sufficiently large not to wrap */
 579         if (offset > offset + size)
 580                 die("pack too large for current definition of off_t");
 581         return offset + size;
 582 }
 583
 584 static int open_object_dir_tmp(const char *path)
 585 {
 586     snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
 587     return mkstemp(tmpname);
 588 }
 589
 590 /* forward declaration for write_pack_file */
 591 static int adjust_perm(const char *path, mode_t mode);
 592
 593 static void write_pack_file(void)
 594 {
 595         uint32_t i = 0, j;
 596         struct sha1file *f;
 597         off_t offset, offset_one, last_obj_offset = 0;
 598         struct pack_header hdr;
 599         int do_progress = progress >> pack_to_stdout;
 600         uint32_t nr_remaining = nr_result;
 601
 602         if (do_progress)
 603                 start_progress(&progress_state, "Writing %u objects...", "", nr_result);
 604         written_list = xmalloc(nr_objects * sizeof(struct object_entry *));
 605
 606         do {
 607                 unsigned char sha1[20];
 608
 609                 if (pack_to_stdout) {
 610                         f = sha1fd(1, "<stdout>");
 611                 } else {
 612                         int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
 613                         if (fd < 0)
 614                                 die("unable to create %s: %s\n", tmpname, strerror(errno));
 615                         pack_tmp_name = xstrdup(tmpname);
 616                         f = sha1fd(fd, pack_tmp_name);
 617                 }
 618
 619                 hdr.hdr_signature = htonl(PACK_SIGNATURE);
 620                 hdr.hdr_version = htonl(PACK_VERSION);
 621                 hdr.hdr_entries = htonl(nr_remaining);
 622                 sha1write(f, &hdr, sizeof(hdr));
 623                 offset = sizeof(hdr);
 624                 nr_written = 0;
 625                 for (; i < nr_objects; i++) {
 626                         last_obj_offset = offset;
 627                         offset_one = write_one(f, objects + i, offset);
 628                         if (!offset_one)
 629                                 break;
 630                         offset = offset_one;
 631                         if (do_progress)
 632                                 display_progress(&progress_state, written);
 633                 }
 634
 635                 /*
 636                  * Did we write the wrong # entries in the header?
 637                  * If so, rewrite it like in fast-import
 638                  */
 639                 if (pack_to_stdout || nr_written == nr_remaining) {
 640                         sha1close(f, sha1, 1);
 641                 } else {
 642                         sha1close(f, sha1, 0);
 643                         fixup_pack_header_footer(f->fd, sha1, pack_tmp_name, nr_written);
 644                         close(f->fd);
 645                 }
 646
 647                 if (!pack_to_stdout) {
 648                         mode_t mode = umask(0);
 649
 650                         umask(mode);
 651                         mode = 0444 & ~mode;
 652
 653                         idx_tmp_name = write_idx_file(NULL,
 654                                 (struct pack_idx_entry **) written_list, nr_written, sha1);
 655                         snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
 656                                  base_name, sha1_to_hex(sha1));
 657                         if (adjust_perm(pack_tmp_name, mode))
 658                                 die("unable to make temporary pack file readable: %s",
 659                                     strerror(errno));
 660                         if (rename(pack_tmp_name, tmpname))
 661                                 die("unable to rename temporary pack file: %s",
 662                                     strerror(errno));
 663                         snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
 664                                  base_name, sha1_to_hex(sha1));
 665                         if (adjust_perm(idx_tmp_name, mode))
 666                                 die("unable to make temporary index file readable: %s",
 667                                     strerror(errno));
 668                         if (rename(idx_tmp_name, tmpname))
 669                                 die("unable to rename temporary index file: %s",
 670                                     strerror(errno));
 671                         puts(sha1_to_hex(sha1));
 672                 }
 673
 674                 /* mark written objects as written to previous pack */
 675                 for (j = 0; j < nr_written; j++) {
 676                         written_list[j]->idx.offset = (off_t)-1;
 677                 }
 678                 nr_remaining -= nr_written;
 679         } while (nr_remaining && i < nr_objects);
 680
 681         free(written_list);
 682         if (do_progress)
 683                 stop_progress(&progress_state);
 684         if (written != nr_result)
 685                 die("wrote %u objects while expecting %u", written, nr_result);
 686         /*
 687          * We have scanned through [0 ... i).  Since we have written
 688          * the correct number of objects,  the remaining [i ... nr_objects)
 689          * items must be either already written (due to out-of-order delta base)
 690          * or a preferred base.  Count those which are neither and complain if any.
 691          */
 692         for (j = 0; i < nr_objects; i++) {
 693                 struct object_entry *e = objects + i;
 694                 j += !e->idx.offset && !e->preferred_base;
 695         }
 696         if (j)
 697                 die("wrote %u objects as expected but %u unwritten", written, j);
 698 }
 699
 700 static int locate_object_entry_hash(const unsigned char *sha1)
 701 {
 702         int i;
 703         unsigned int ui;
 704         memcpy(&ui, sha1, sizeof(unsigned int));
 705         i = ui % object_ix_hashsz;
 706         while (0 < object_ix[i]) {
 707                 if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1))
 708                         return i;
 709                 if (++i == object_ix_hashsz)
 710                         i = 0;
 711         }
 712         return -1 - i;
 713 }
 714
 715 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 716 {
 717         int i;
 718
 719         if (!object_ix_hashsz)
 720                 return NULL;
 721
 722         i = locate_object_entry_hash(sha1);
 723         if (0 <= i)
 724                 return &objects[object_ix[i]-1];
 725         return NULL;
 726 }
 727
 728 static void rehash_objects(void)
 729 {
 730         uint32_t i;
 731         struct object_entry *oe;
 732
 733         object_ix_hashsz = nr_objects * 3;
 734         if (object_ix_hashsz < 1024)
 735                 object_ix_hashsz = 1024;
 736         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 737         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 738         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 739                 int ix = locate_object_entry_hash(oe->idx.sha1);
 740                 if (0 <= ix)
 741                         continue;
 742                 ix = -1 - ix;
 743                 object_ix[ix] = i + 1;
 744         }
 745 }
 746
 747 static unsigned name_hash(const char *name)
 748 {
 749         unsigned char c;
 750         unsigned hash = 0;
 751
 752         if (!name)
 753                 return 0;
 754
 755         /*
 756          * This effectively just creates a sortable number from the
 757          * last sixteen non-whitespace characters. Last characters
 758          * count "most", so things that end in ".c" sort together.
 759          */
 760         while ((c = *name++) != 0) {
 761                 if (isspace(c))
 762                         continue;
 763                 hash = (hash >> 2) + (c << 24);
 764         }
 765         return hash;
 766 }
 767
 768 static void setup_delta_attr_check(struct git_attr_check *check)
 769 {
 770         static struct git_attr *attr_delta;
 771
 772         if (!attr_delta)
 773                 attr_delta = git_attr("delta", 5);
 774
 775         check[0].attr = attr_delta;
 776 }
 777
 778 static int no_try_delta(const char *path)
 779 {
 780         struct git_attr_check check[1];
 781
 782         setup_delta_attr_check(check);
 783         if (git_checkattr(path, ARRAY_SIZE(check), check))
 784                 return 0;
 785         if (ATTR_FALSE(check->value))
 786                 return 1;
 787         return 0;
 788 }
 789
 790 static int add_object_entry(const unsigned char *sha1, enum object_type type,
 791                             const char *name, int exclude)
 792 {
 793         struct object_entry *entry;
 794         struct packed_git *p, *found_pack = NULL;
 795         off_t found_offset = 0;
 796         int ix;
 797         unsigned hash = name_hash(name);
 798
 799         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 800         if (ix >= 0) {
 801                 if (exclude) {
 802                         entry = objects + object_ix[ix] - 1;
 803                         if (!entry->preferred_base)
 804                                 nr_result--;
 805                         entry->preferred_base = 1;
 806                 }
 807                 return 0;
 808         }
 809
 810         for (p = packed_git; p; p = p->next) {
 811                 off_t offset = find_pack_entry_one(sha1, p);
 812                 if (offset) {
 813                         if (!found_pack) {
 814                                 found_offset = offset;
 815                                 found_pack = p;
 816                         }
 817                         if (exclude)
 818                                 break;
 819                         if (incremental)
 820                                 return 0;
 821                         if (local && !p->pack_local)
 822                                 return 0;
 823                 }
 824         }
 825
 826         if (nr_objects >= nr_alloc) {
 827                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 828                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 829         }
 830
 831         entry = objects + nr_objects++;
 832         memset(entry, 0, sizeof(*entry));
 833         hashcpy(entry->idx.sha1, sha1);
 834         entry->hash = hash;
 835         if (type)
 836                 entry->type = type;
 837         if (exclude)
 838                 entry->preferred_base = 1;
 839         else
 840                 nr_result++;
 841         if (found_pack) {
 842                 entry->in_pack = found_pack;
 843                 entry->in_pack_offset = found_offset;
 844         }
 845
 846         if (object_ix_hashsz * 3 <= nr_objects * 4)
 847                 rehash_objects();
 848         else
 849                 object_ix[-1 - ix] = nr_objects;
 850
 851         if (progress)
 852                 display_progress(&progress_state, nr_objects);
 853
 854         if (name && no_try_delta(name))
 855                 entry->no_try_delta = 1;
 856
 857         return 1;
 858 }
 859
 860 struct pbase_tree_cache {
 861         unsigned char sha1[20];
 862         int ref;
 863         int temporary;
 864         void *tree_data;
 865         unsigned long tree_size;
 866 };
 867
 868 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 869 static int pbase_tree_cache_ix(const unsigned char *sha1)
 870 {
 871         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 872 }
 873 static int pbase_tree_cache_ix_incr(int ix)
 874 {
 875         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 876 }
 877
 878 static struct pbase_tree {
 879         struct pbase_tree *next;
 880         /* This is a phony "cache" entry; we are not
 881          * going to evict it nor find it through _get()
 882          * mechanism -- this is for the toplevel node that
 883          * would almost always change with any commit.
 884          */
 885         struct pbase_tree_cache pcache;
 886 } *pbase_tree;
 887
 888 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 889 {
 890         struct pbase_tree_cache *ent, *nent;
 891         void *data;
 892         unsigned long size;
 893         enum object_type type;
 894         int neigh;
 895         int my_ix = pbase_tree_cache_ix(sha1);
 896         int available_ix = -1;
 897
 898         /* pbase-tree-cache acts as a limited hashtable.
 899          * your object will be found at your index or within a few
 900          * slots after that slot if it is cached.
 901          */
 902         for (neigh = 0; neigh < 8; neigh++) {
 903                 ent = pbase_tree_cache[my_ix];
 904                 if (ent && !hashcmp(ent->sha1, sha1)) {
 905                         ent->ref++;
 906                         return ent;
 907                 }
 908                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 909                          ((0 <= available_ix) &&
 910                           (!ent && pbase_tree_cache[available_ix])))
 911                         available_ix = my_ix;
 912                 if (!ent)
 913                         break;
 914                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 915         }
 916
 917         /* Did not find one.  Either we got a bogus request or
 918          * we need to read and perhaps cache.
 919          */
 920         data = read_sha1_file(sha1, &type, &size);
 921         if (!data)
 922                 return NULL;
 923         if (type != OBJ_TREE) {
 924                 free(data);
 925                 return NULL;
 926         }
 927
 928         /* We need to either cache or return a throwaway copy */
 929
 930         if (available_ix < 0)
 931                 ent = NULL;
 932         else {
 933                 ent = pbase_tree_cache[available_ix];
 934                 my_ix = available_ix;
 935         }
 936
 937         if (!ent) {
 938                 nent = xmalloc(sizeof(*nent));
 939                 nent->temporary = (available_ix < 0);
 940         }
 941         else {
 942                 /* evict and reuse */
 943                 free(ent->tree_data);
 944                 nent = ent;
 945         }
 946         hashcpy(nent->sha1, sha1);
 947         nent->tree_data = data;
 948         nent->tree_size = size;
 949         nent->ref = 1;
 950         if (!nent->temporary)
 951                 pbase_tree_cache[my_ix] = nent;
 952         return nent;
 953 }
 954
 955 static void pbase_tree_put(struct pbase_tree_cache *cache)
 956 {
 957         if (!cache->temporary) {
 958                 cache->ref--;
 959                 return;
 960         }
 961         free(cache->tree_data);
 962         free(cache);
 963 }
 964
 965 static int name_cmp_len(const char *name)
 966 {
 967         int i;
 968         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 969                 ;
 970         return i;
 971 }
 972
 973 static void add_pbase_object(struct tree_desc *tree,
 974                              const char *name,
 975                              int cmplen,
 976                              const char *fullname)
 977 {
 978         struct name_entry entry;
 979         int cmp;
 980
 981         while (tree_entry(tree,&entry)) {
 982                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 983                       memcmp(name, entry.path, cmplen);
 984                 if (cmp > 0)
 985                         continue;
 986                 if (cmp < 0)
 987                         return;
 988                 if (name[cmplen] != '/') {
 989                         add_object_entry(entry.sha1,
 990                                          S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB,
 991                                          fullname, 1);
 992                         return;
 993                 }
 994                 if (S_ISDIR(entry.mode)) {
 995                         struct tree_desc sub;
 996                         struct pbase_tree_cache *tree;
 997                         const char *down = name+cmplen+1;
 998                         int downlen = name_cmp_len(down);
 999
1000                         tree = pbase_tree_get(entry.sha1);
1001                         if (!tree)
1002                                 return;
1003                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
1004
1005                         add_pbase_object(&sub, down, downlen, fullname);
1006                         pbase_tree_put(tree);
1007                 }
1008         }
1009 }
1010
1011 static unsigned *done_pbase_paths;
1012 static int done_pbase_paths_num;
1013 static int done_pbase_paths_alloc;
1014 static int done_pbase_path_pos(unsigned hash)
1015 {
1016         int lo = 0;
1017         int hi = done_pbase_paths_num;
1018         while (lo < hi) {
1019                 int mi = (hi + lo) / 2;
1020                 if (done_pbase_paths[mi] == hash)
1021                         return mi;
1022                 if (done_pbase_paths[mi] < hash)
1023                         hi = mi;
1024                 else
1025                         lo = mi + 1;
1026         }
1027         return -lo-1;
1028 }
1029
1030 static int check_pbase_path(unsigned hash)
1031 {
1032         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
1033         if (0 <= pos)
1034                 return 1;
1035         pos = -pos - 1;
1036         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
1037                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
1038                 done_pbase_paths = xrealloc(done_pbase_paths,
1039                                             done_pbase_paths_alloc *
1040                                             sizeof(unsigned));
1041         }
1042         done_pbase_paths_num++;
1043         if (pos < done_pbase_paths_num)
1044                 memmove(done_pbase_paths + pos + 1,
1045                         done_pbase_paths + pos,
1046                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1047         done_pbase_paths[pos] = hash;
1048         return 0;
1049 }
1050
1051 static void add_preferred_base_object(const char *name)
1052 {
1053         struct pbase_tree *it;
1054         int cmplen;
1055         unsigned hash = name_hash(name);
1056
1057         if (!num_preferred_base || check_pbase_path(hash))
1058                 return;
1059
1060         cmplen = name_cmp_len(name);
1061         for (it = pbase_tree; it; it = it->next) {
1062                 if (cmplen == 0) {
1063                         add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
1064                 }
1065                 else {
1066                         struct tree_desc tree;
1067                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1068                         add_pbase_object(&tree, name, cmplen, name);
1069                 }
1070         }
1071 }
1072
1073 static void add_preferred_base(unsigned char *sha1)
1074 {
1075         struct pbase_tree *it;
1076         void *data;
1077         unsigned long size;
1078         unsigned char tree_sha1[20];
1079
1080         if (window <= num_preferred_base++)
1081                 return;
1082
1083         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1084         if (!data)
1085                 return;
1086
1087         for (it = pbase_tree; it; it = it->next) {
1088                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1089                         free(data);
1090                         return;
1091                 }
1092         }
1093
1094         it = xcalloc(1, sizeof(*it));
1095         it->next = pbase_tree;
1096         pbase_tree = it;
1097
1098         hashcpy(it->pcache.sha1, tree_sha1);
1099         it->pcache.tree_data = data;
1100         it->pcache.tree_size = size;
1101 }
1102
1103 static void check_object(struct object_entry *entry)
1104 {
1105         if (entry->in_pack) {
1106                 struct packed_git *p = entry->in_pack;
1107                 struct pack_window *w_curs = NULL;
1108                 const unsigned char *base_ref = NULL;
1109                 struct object_entry *base_entry;
1110                 unsigned long used, used_0;
1111                 unsigned int avail;
1112                 off_t ofs;
1113                 unsigned char *buf, c;
1114
1115                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1116
1117                 /*
1118                  * We want in_pack_type even if we do not reuse delta
1119                  * since non-delta representations could still be reused.
1120                  */
1121                 used = unpack_object_header_gently(buf, avail,
1122                                                    &entry->in_pack_type,
1123                                                    &entry->size);
1124
1125                 /*
1126                  * Determine if this is a delta and if so whether we can
1127                  * reuse it or not.  Otherwise let's find out as cheaply as
1128                  * possible what the actual type and size for this object is.
1129                  */
1130                 switch (entry->in_pack_type) {
1131                 default:
1132                         /* Not a delta hence we've already got all we need. */
1133                         entry->type = entry->in_pack_type;
1134                         entry->in_pack_header_size = used;
1135                         unuse_pack(&w_curs);
1136                         return;
1137                 case OBJ_REF_DELTA:
1138                         if (!no_reuse_delta && !entry->preferred_base)
1139                                 base_ref = use_pack(p, &w_curs,
1140                                                 entry->in_pack_offset + used, NULL);
1141                         entry->in_pack_header_size = used + 20;
1142                         break;
1143                 case OBJ_OFS_DELTA:
1144                         buf = use_pack(p, &w_curs,
1145                                        entry->in_pack_offset + used, NULL);
1146                         used_0 = 0;
1147                         c = buf[used_0++];
1148                         ofs = c & 127;
1149                         while (c & 128) {
1150                                 ofs += 1;
1151                                 if (!ofs || MSB(ofs, 7))
1152                                         die("delta base offset overflow in pack for %s",
1153                                             sha1_to_hex(entry->idx.sha1));
1154                                 c = buf[used_0++];
1155                                 ofs = (ofs << 7) + (c & 127);
1156                         }
1157                         if (ofs >= entry->in_pack_offset)
1158                                 die("delta base offset out of bound for %s",
1159                                     sha1_to_hex(entry->idx.sha1));
1160                         ofs = entry->in_pack_offset - ofs;
1161                         if (!no_reuse_delta && !entry->preferred_base)
1162                                 base_ref = find_packed_object_name(p, ofs);
1163                         entry->in_pack_header_size = used + used_0;
1164                         break;
1165                 }
1166
1167                 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1168                         /*
1169                          * If base_ref was set above that means we wish to
1170                          * reuse delta data, and we even found that base
1171                          * in the list of objects we want to pack. Goodie!
1172                          *
1173                          * Depth value does not matter - find_deltas() will
1174                          * never consider reused delta as the base object to
1175                          * deltify other objects against, in order to avoid
1176                          * circular deltas.
1177                          */
1178                         entry->type = entry->in_pack_type;
1179                         entry->delta = base_entry;
1180                         entry->delta_sibling = base_entry->delta_child;
1181                         base_entry->delta_child = entry;
1182                         unuse_pack(&w_curs);
1183                         return;
1184                 }
1185
1186                 if (entry->type) {
1187                         /*
1188                          * This must be a delta and we already know what the
1189                          * final object type is.  Let's extract the actual
1190                          * object size from the delta header.
1191                          */
1192                         entry->size = get_size_from_delta(p, &w_curs,
1193                                         entry->in_pack_offset + entry->in_pack_header_size);
1194                         unuse_pack(&w_curs);
1195                         return;
1196                 }
1197
1198                 /*
1199                  * No choice but to fall back to the recursive delta walk
1200                  * with sha1_object_info() to find about the object type
1201                  * at this point...
1202                  */
1203                 unuse_pack(&w_curs);
1204         }
1205
1206         entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1207         if (entry->type < 0)
1208                 die("unable to get type of object %s",
1209                     sha1_to_hex(entry->idx.sha1));
1210 }
1211
1212 static int pack_offset_sort(const void *_a, const void *_b)
1213 {
1214         const struct object_entry *a = *(struct object_entry **)_a;
1215         const struct object_entry *b = *(struct object_entry **)_b;
1216
1217         /* avoid filesystem trashing with loose objects */
1218         if (!a->in_pack && !b->in_pack)
1219                 return hashcmp(a->idx.sha1, b->idx.sha1);
1220
1221         if (a->in_pack < b->in_pack)
1222                 return -1;
1223         if (a->in_pack > b->in_pack)
1224                 return 1;
1225         return a->in_pack_offset < b->in_pack_offset ? -1 :
1226                         (a->in_pack_offset > b->in_pack_offset);
1227 }
1228
1229 static void get_object_details(void)
1230 {
1231         uint32_t i;
1232         struct object_entry **sorted_by_offset;
1233
1234         sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1235         for (i = 0; i < nr_objects; i++)
1236                 sorted_by_offset[i] = objects + i;
1237         qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1238
1239         prepare_pack_ix();
1240         for (i = 0; i < nr_objects; i++)
1241                 check_object(sorted_by_offset[i]);
1242         free(sorted_by_offset);
1243 }
1244
1245 static int type_size_sort(const void *_a, const void *_b)
1246 {
1247         const struct object_entry *a = *(struct object_entry **)_a;
1248         const struct object_entry *b = *(struct object_entry **)_b;
1249
1250         if (a->type < b->type)
1251                 return -1;
1252         if (a->type > b->type)
1253                 return 1;
1254         if (a->hash < b->hash)
1255                 return -1;
1256         if (a->hash > b->hash)
1257                 return 1;
1258         if (a->preferred_base < b->preferred_base)
1259                 return -1;
1260         if (a->preferred_base > b->preferred_base)
1261                 return 1;
1262         if (a->size < b->size)
1263                 return -1;
1264         if (a->size > b->size)
1265                 return 1;
1266         return a > b ? -1 : (a < b);  /* newest last */
1267 }
1268
1269 struct unpacked {
1270         struct object_entry *entry;
1271         void *data;
1272         struct delta_index *index;
1273 };
1274
1275 static int delta_cacheable(struct unpacked *trg, struct unpacked *src,
1276                             unsigned long src_size, unsigned long trg_size,
1277                             unsigned long delta_size)
1278 {
1279         if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1280                 return 0;
1281
1282         if (delta_size < cache_max_small_delta_size)
1283                 return 1;
1284
1285         /* cache delta, if objects are large enough compared to delta size */
1286         if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1287                 return 1;
1288
1289         return 0;
1290 }
1291
1292 /*
1293  * We search for deltas _backwards_ in a list sorted by type and
1294  * by size, so that we see progressively smaller and smaller files.
1295  * That's because we prefer deltas to be from the bigger file
1296  * to the smaller - deletes are potentially cheaper, but perhaps
1297  * more importantly, the bigger file is likely the more recent
1298  * one.
1299  */
1300 static int try_delta(struct unpacked *trg, struct unpacked *src,
1301                      unsigned max_depth)
1302 {
1303         struct object_entry *trg_entry = trg->entry;
1304         struct object_entry *src_entry = src->entry;
1305         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1306         enum object_type type;
1307         void *delta_buf;
1308
1309         /* Don't bother doing diffs between different types */
1310         if (trg_entry->type != src_entry->type)
1311                 return -1;
1312
1313         /* We do not compute delta to *create* objects we are not
1314          * going to pack.
1315          */
1316         if (trg_entry->preferred_base)
1317                 return -1;
1318
1319         /*
1320          * We do not bother to try a delta that we discarded
1321          * on an earlier try, but only when reusing delta data.
1322          */
1323         if (!no_reuse_delta && trg_entry->in_pack &&
1324             trg_entry->in_pack == src_entry->in_pack &&
1325             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1326             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1327                 return 0;
1328
1329         /* Let's not bust the allowed depth. */
1330         if (src_entry->depth >= max_depth)
1331                 return 0;
1332
1333         /* Now some size filtering heuristics. */
1334         trg_size = trg_entry->size;
1335         max_size = trg_size/2 - 20;
1336         max_size = max_size * (max_depth - src_entry->depth) / max_depth;
1337         if (max_size == 0)
1338                 return 0;
1339         if (trg_entry->delta && trg_entry->delta_size <= max_size)
1340                 max_size = trg_entry->delta_size-1;
1341         src_size = src_entry->size;
1342         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1343         if (sizediff >= max_size)
1344                 return 0;
1345
1346         /* Load data if not already done */
1347         if (!trg->data) {
1348                 trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1349                 if (sz != trg_size)
1350                         die("object %s inconsistent object length (%lu vs %lu)",
1351                             sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1352         }
1353         if (!src->data) {
1354                 src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1355                 if (sz != src_size)
1356                         die("object %s inconsistent object length (%lu vs %lu)",
1357                             sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1358         }
1359         if (!src->index) {
1360                 src->index = create_delta_index(src->data, src_size);
1361                 if (!src->index) {
1362                         static int warned = 0;
1363                         if (!warned++)
1364                                 warning("suboptimal pack - out of memory");
1365                         return 0;
1366                 }
1367         }
1368
1369         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1370         if (!delta_buf)
1371                 return 0;
1372
1373         if (trg_entry->delta_data) {
1374                 delta_cache_size -= trg_entry->delta_size;
1375                 free(trg_entry->delta_data);
1376         }
1377         trg_entry->delta_data = 0;
1378         trg_entry->delta = src_entry;
1379         trg_entry->delta_size = delta_size;
1380         trg_entry->depth = src_entry->depth + 1;
1381
1382         if (delta_cacheable(src, trg, src_size, trg_size, delta_size)) {
1383                 trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1384                 delta_cache_size += trg_entry->delta_size;
1385         } else
1386                 free(delta_buf);
1387         return 1;
1388 }
1389
1390 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1391 {
1392         struct object_entry *child = me->delta_child;
1393         unsigned int m = n;
1394         while (child) {
1395                 unsigned int c = check_delta_limit(child, n + 1);
1396                 if (m < c)
1397                         m = c;
1398                 child = child->delta_sibling;
1399         }
1400         return m;
1401 }
1402
1403 static void find_deltas(struct object_entry **list, int window, int depth)
1404 {
1405         uint32_t i = nr_objects, idx = 0, processed = 0;
1406         unsigned int array_size = window * sizeof(struct unpacked);
1407         struct unpacked *array;
1408         int max_depth;
1409
1410         if (!nr_objects)
1411                 return;
1412         array = xmalloc(array_size);
1413         memset(array, 0, array_size);
1414         if (progress)
1415                 start_progress(&progress_state, "Deltifying %u objects...", "", nr_result);
1416
1417         do {
1418                 struct object_entry *entry = list[--i];
1419                 struct unpacked *n = array + idx;
1420                 int j;
1421
1422                 if (!entry->preferred_base)
1423                         processed++;
1424
1425                 if (progress)
1426                         display_progress(&progress_state, processed);
1427
1428                 if (entry->delta)
1429                         /* This happens if we decided to reuse existing
1430                          * delta from a pack.  "!no_reuse_delta &&" is implied.
1431                          */
1432                         continue;
1433
1434                 if (entry->size < 50)
1435                         continue;
1436
1437                 if (entry->no_try_delta)
1438                         continue;
1439
1440                 free_delta_index(n->index);
1441                 n->index = NULL;
1442                 free(n->data);
1443                 n->data = NULL;
1444                 n->entry = entry;
1445
1446                 /*
1447                  * If the current object is at pack edge, take the depth the
1448                  * objects that depend on the current object into account
1449                  * otherwise they would become too deep.
1450                  */
1451                 max_depth = depth;
1452                 if (entry->delta_child) {
1453                         max_depth -= check_delta_limit(entry, 0);
1454                         if (max_depth <= 0)
1455                                 goto next;
1456                 }
1457
1458                 j = window;
1459                 while (--j > 0) {
1460                         uint32_t other_idx = idx + j;
1461                         struct unpacked *m;
1462                         if (other_idx >= window)
1463                                 other_idx -= window;
1464                         m = array + other_idx;
1465                         if (!m->entry)
1466                                 break;
1467                         if (try_delta(n, m, max_depth) < 0)
1468                                 break;
1469                 }
1470
1471                 /* if we made n a delta, and if n is already at max
1472                  * depth, leaving it in the window is pointless.  we
1473                  * should evict it first.
1474                  */
1475                 if (entry->delta && depth <= entry->depth)
1476                         continue;
1477
1478                 next:
1479                 idx++;
1480                 if (idx >= window)
1481                         idx = 0;
1482         } while (i > 0);
1483
1484         if (progress)
1485                 stop_progress(&progress_state);
1486
1487         for (i = 0; i < window; ++i) {
1488                 free_delta_index(array[i].index);
1489                 free(array[i].data);
1490         }
1491         free(array);
1492 }
1493
1494 static void prepare_pack(int window, int depth)
1495 {
1496         struct object_entry **delta_list;
1497         uint32_t i;
1498
1499         get_object_details();
1500
1501         if (!window || !depth)
1502                 return;
1503
1504         delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1505         for (i = 0; i < nr_objects; i++)
1506                 delta_list[i] = objects + i;
1507         qsort(delta_list, nr_objects, sizeof(*delta_list), type_size_sort);
1508         find_deltas(delta_list, window+1, depth);
1509         free(delta_list);
1510 }
1511
1512 static int git_pack_config(const char *k, const char *v)
1513 {
1514         if(!strcmp(k, "pack.window")) {
1515                 window = git_config_int(k, v);
1516                 return 0;
1517         }
1518         if(!strcmp(k, "pack.depth")) {
1519                 depth = git_config_int(k, v);
1520                 return 0;
1521         }
1522         if (!strcmp(k, "pack.compression")) {
1523                 int level = git_config_int(k, v);
1524                 if (level == -1)
1525                         level = Z_DEFAULT_COMPRESSION;
1526                 else if (level < 0 || level > Z_BEST_COMPRESSION)
1527                         die("bad pack compression level %d", level);
1528                 pack_compression_level = level;
1529                 pack_compression_seen = 1;
1530                 return 0;
1531         }
1532         if (!strcmp(k, "pack.deltacachesize")) {
1533                 max_delta_cache_size = git_config_int(k, v);
1534                 return 0;
1535         }
1536         if (!strcmp(k, "pack.deltacachelimit")) {
1537                 cache_max_small_delta_size = git_config_int(k, v);
1538                 return 0;
1539         }
1540         return git_default_config(k, v);
1541 }
1542
1543 static void read_object_list_from_stdin(void)
1544 {
1545         char line[40 + 1 + PATH_MAX + 2];
1546         unsigned char sha1[20];
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                 add_preferred_base_object(line+41);
1570                 add_object_entry(sha1, 0, line+41, 0);
1571         }
1572 }
1573
1574 static void show_commit(struct commit *commit)
1575 {
1576         add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
1577 }
1578
1579 static void show_object(struct object_array_entry *p)
1580 {
1581         add_preferred_base_object(p->name);
1582         add_object_entry(p->item->sha1, p->item->type, p->name, 0);
1583 }
1584
1585 static void show_edge(struct commit *commit)
1586 {
1587         add_preferred_base(commit->object.sha1);
1588 }
1589
1590 static void get_object_list(int ac, const char **av)
1591 {
1592         struct rev_info revs;
1593         char line[1000];
1594         int flags = 0;
1595
1596         init_revisions(&revs, NULL);
1597         save_commit_buffer = 0;
1598         track_object_refs = 0;
1599         setup_revisions(ac, av, &revs, NULL);
1600
1601         while (fgets(line, sizeof(line), stdin) != NULL) {
1602                 int len = strlen(line);
1603                 if (line[len - 1] == '\n')
1604                         line[--len] = 0;
1605                 if (!len)
1606                         break;
1607                 if (*line == '-') {
1608                         if (!strcmp(line, "--not")) {
1609                                 flags ^= UNINTERESTING;
1610                                 continue;
1611                         }
1612                         die("not a rev '%s'", line);
1613                 }
1614                 if (handle_revision_arg(line, &revs, flags, 1))
1615                         die("bad revision '%s'", line);
1616         }
1617
1618         prepare_revision_walk(&revs);
1619         mark_edges_uninteresting(revs.commits, &revs, show_edge);
1620         traverse_commit_list(&revs, show_commit, show_object);
1621 }
1622
1623 static int adjust_perm(const char *path, mode_t mode)
1624 {
1625         if (chmod(path, mode))
1626                 return -1;
1627         return adjust_shared_perm(path);
1628 }
1629
1630 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1631 {
1632         int use_internal_rev_list = 0;
1633         int thin = 0;
1634         uint32_t i;
1635         const char **rp_av;
1636         int rp_ac_alloc = 64;
1637         int rp_ac;
1638
1639         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1640
1641         rp_av[0] = "pack-objects";
1642         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1643         rp_ac = 2;
1644
1645         git_config(git_pack_config);
1646         if (!pack_compression_seen && core_compression_seen)
1647                 pack_compression_level = core_compression_level;
1648
1649         progress = isatty(2);
1650         for (i = 1; i < argc; i++) {
1651                 const char *arg = argv[i];
1652
1653                 if (*arg != '-')
1654                         break;
1655
1656                 if (!strcmp("--non-empty", arg)) {
1657                         non_empty = 1;
1658                         continue;
1659                 }
1660                 if (!strcmp("--local", arg)) {
1661                         local = 1;
1662                         continue;
1663                 }
1664                 if (!strcmp("--incremental", arg)) {
1665                         incremental = 1;
1666                         continue;
1667                 }
1668                 if (!prefixcmp(arg, "--compression=")) {
1669                         char *end;
1670                         int level = strtoul(arg+14, &end, 0);
1671                         if (!arg[14] || *end)
1672                                 usage(pack_usage);
1673                         if (level == -1)
1674                                 level = Z_DEFAULT_COMPRESSION;
1675                         else if (level < 0 || level > Z_BEST_COMPRESSION)
1676                                 die("bad pack compression level %d", level);
1677                         pack_compression_level = level;
1678                         continue;
1679                 }
1680                 if (!prefixcmp(arg, "--max-pack-size=")) {
1681                         char *end;
1682                         pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024;
1683                         if (!arg[16] || *end)
1684                                 usage(pack_usage);
1685                         continue;
1686                 }
1687                 if (!prefixcmp(arg, "--window=")) {
1688                         char *end;
1689                         window = strtoul(arg+9, &end, 0);
1690                         if (!arg[9] || *end)
1691                                 usage(pack_usage);
1692                         continue;
1693                 }
1694                 if (!prefixcmp(arg, "--depth=")) {
1695                         char *end;
1696                         depth = strtoul(arg+8, &end, 0);
1697                         if (!arg[8] || *end)
1698                                 usage(pack_usage);
1699                         continue;
1700                 }
1701                 if (!strcmp("--progress", arg)) {
1702                         progress = 1;
1703                         continue;
1704                 }
1705                 if (!strcmp("--all-progress", arg)) {
1706                         progress = 2;
1707                         continue;
1708                 }
1709                 if (!strcmp("-q", arg)) {
1710                         progress = 0;
1711                         continue;
1712                 }
1713                 if (!strcmp("--no-reuse-delta", arg)) {
1714                         no_reuse_delta = 1;
1715                         continue;
1716                 }
1717                 if (!strcmp("--no-reuse-object", arg)) {
1718                         no_reuse_object = no_reuse_delta = 1;
1719                         continue;
1720                 }
1721                 if (!strcmp("--delta-base-offset", arg)) {
1722                         allow_ofs_delta = 1;
1723                         continue;
1724                 }
1725                 if (!strcmp("--stdout", arg)) {
1726                         pack_to_stdout = 1;
1727                         continue;
1728                 }
1729                 if (!strcmp("--revs", arg)) {
1730                         use_internal_rev_list = 1;
1731                         continue;
1732                 }
1733                 if (!strcmp("--unpacked", arg) ||
1734                     !prefixcmp(arg, "--unpacked=") ||
1735                     !strcmp("--reflog", arg) ||
1736                     !strcmp("--all", arg)) {
1737                         use_internal_rev_list = 1;
1738                         if (rp_ac >= rp_ac_alloc - 1) {
1739                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1740                                 rp_av = xrealloc(rp_av,
1741                                                  rp_ac_alloc * sizeof(*rp_av));
1742                         }
1743                         rp_av[rp_ac++] = arg;
1744                         continue;
1745                 }
1746                 if (!strcmp("--thin", arg)) {
1747                         use_internal_rev_list = 1;
1748                         thin = 1;
1749                         rp_av[1] = "--objects-edge";
1750                         continue;
1751                 }
1752                 if (!prefixcmp(arg, "--index-version=")) {
1753                         char *c;
1754                         pack_idx_default_version = strtoul(arg + 16, &c, 10);
1755                         if (pack_idx_default_version > 2)
1756                                 die("bad %s", arg);
1757                         if (*c == ',')
1758                                 pack_idx_off32_limit = strtoul(c+1, &c, 0);
1759                         if (*c || pack_idx_off32_limit & 0x80000000)
1760                                 die("bad %s", arg);
1761                         continue;
1762                 }
1763                 usage(pack_usage);
1764         }
1765
1766         /* Traditionally "pack-objects [options] base extra" failed;
1767          * we would however want to take refs parameter that would
1768          * have been given to upstream rev-list ourselves, which means
1769          * we somehow want to say what the base name is.  So the
1770          * syntax would be:
1771          *
1772          * pack-objects [options] base <refs...>
1773          *
1774          * in other words, we would treat the first non-option as the
1775          * base_name and send everything else to the internal revision
1776          * walker.
1777          */
1778
1779         if (!pack_to_stdout)
1780                 base_name = argv[i++];
1781
1782         if (pack_to_stdout != !base_name)
1783                 usage(pack_usage);
1784
1785         if (pack_to_stdout && pack_size_limit)
1786                 die("--max-pack-size cannot be used to build a pack for transfer.");
1787
1788         if (!pack_to_stdout && thin)
1789                 die("--thin cannot be used to build an indexable pack.");
1790
1791         prepare_packed_git();
1792
1793         if (progress)
1794                 start_progress(&progress_state, "Generating pack...",
1795                                "Counting objects: ", 0);
1796         if (!use_internal_rev_list)
1797                 read_object_list_from_stdin();
1798         else {
1799                 rp_av[rp_ac] = NULL;
1800                 get_object_list(rp_ac, rp_av);
1801         }
1802         if (progress) {
1803                 stop_progress(&progress_state);
1804                 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
1805         }
1806
1807         if (non_empty && !nr_result)
1808                 return 0;
1809         if (progress && (nr_objects != nr_result))
1810                 fprintf(stderr, "Result has %u objects.\n", nr_result);
1811         if (nr_result)
1812                 prepare_pack(window, depth);
1813         write_pack_file();
1814         if (progress)
1815                 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
1816                         written, written_delta, reused, reused_delta);
1817         return 0;
1818 }