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 }] \n\
  20         [--max-pack-size=N] [--local] [--incremental] \n\
  21         [--window=N] [--window-memory=N] [--depth=N] \n\
  22         [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
  23         [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
  24         [--stdout | base-name] [<ref-list | <object-list]";
  25
  26 struct object_entry {
  27         struct pack_idx_entry idx;
  28         unsigned long size;     /* uncompressed size */
  29         struct packed_git *in_pack;     /* already in pack */
  30         off_t in_pack_offset;
  31         struct object_entry *delta;     /* delta base object */
  32         struct object_entry *delta_child; /* deltified objects who bases me */
  33         struct object_entry *delta_sibling; /* other deltified objects who
  34                                              * uses the same base as me
  35                                              */
  36         void *delta_data;       /* cached delta (uncompressed) */
  37         unsigned long delta_size;       /* delta data size (uncompressed) */
  38         unsigned int hash;      /* name hint hash */
  39         enum object_type type;
  40         enum object_type in_pack_type;  /* could be delta */
  41         unsigned char in_pack_header_size;
  42         unsigned char preferred_base; /* we do not pack this, but is available
  43                                        * to be used as the base object to delta
  44                                        * objects against.
  45                                        */
  46         unsigned char no_try_delta;
  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 static struct object_entry *objects;
  56 static struct object_entry **written_list;
  57 static uint32_t nr_objects, nr_alloc, nr_result, nr_written;
  58
  59 static int non_empty;
  60 static int no_reuse_delta, no_reuse_object;
  61 static int local;
  62 static int incremental;
  63 static int allow_ofs_delta;
  64 static const char *pack_tmp_name, *idx_tmp_name;
  65 static char tmpname[PATH_MAX];
  66 static const char *base_name;
  67 static int progress = 1;
  68 static int window = 10;
  69 static uint32_t pack_size_limit;
  70 static int depth = 50;
  71 static int pack_to_stdout;
  72 static int num_preferred_base;
  73 static struct progress progress_state;
  74 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  75 static int pack_compression_seen;
  76
  77 static unsigned long delta_cache_size = 0;
  78 static unsigned long max_delta_cache_size = 0;
  79 static unsigned long cache_max_small_delta_size = 1000;
  80
  81 static unsigned long window_memory_limit = 0;
  82
  83 /*
  84  * The object names in objects array are hashed with this hashtable,
  85  * to help looking up the entry by object name.
  86  * This hashtable is built after all the objects are seen.
  87  */
  88 static int *object_ix;
  89 static int object_ix_hashsz;
  90
  91 /*
  92  * Pack index for existing packs give us easy access to the offsets into
  93  * corresponding pack file where each object's data starts, but the entries
  94  * do not store the size of the compressed representation (uncompressed
  95  * size is easily available by examining the pack entry header).  It is
  96  * also rather expensive to find the sha1 for an object given its offset.
  97  *
  98  * We build a hashtable of existing packs (pack_revindex), and keep reverse
  99  * index here -- pack index file is sorted by object name mapping to offset;
 100  * this pack_revindex[].revindex array is a list of offset/index_nr pairs
 101  * ordered by offset, so if you know the offset of an object, next offset
 102  * is where its packed representation ends and the index_nr can be used to
 103  * get the object sha1 from the main index.
 104  */
 105 struct revindex_entry {
 106         off_t offset;
 107         unsigned int nr;
 108 };
 109 struct pack_revindex {
 110         struct packed_git *p;
 111         struct revindex_entry *revindex;
 112 };
 113 static struct  pack_revindex *pack_revindex;
 114 static int pack_revindex_hashsz;
 115
 116 /*
 117  * stats
 118  */
 119 static uint32_t written, written_delta;
 120 static uint32_t reused, 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 = p->num_objects;
 170         int i;
 171         const char *index = p->index_data;
 172
 173         rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
 174         index += 4 * 256;
 175
 176         if (p->index_version > 1) {
 177                 const uint32_t *off_32 =
 178                         (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
 179                 const uint32_t *off_64 = off_32 + p->num_objects;
 180                 for (i = 0; i < num_ent; i++) {
 181                         uint32_t off = ntohl(*off_32++);
 182                         if (!(off & 0x80000000)) {
 183                                 rix->revindex[i].offset = off;
 184                         } else {
 185                                 rix->revindex[i].offset =
 186                                         ((uint64_t)ntohl(*off_64++)) << 32;
 187                                 rix->revindex[i].offset |=
 188                                         ntohl(*off_64++);
 189                         }
 190                         rix->revindex[i].nr = i;
 191                 }
 192         } else {
 193                 for (i = 0; i < num_ent; i++) {
 194                         uint32_t hl = *((uint32_t *)(index + 24 * i));
 195                         rix->revindex[i].offset = ntohl(hl);
 196                         rix->revindex[i].nr = i;
 197                 }
 198         }
 199
 200         /* This knows the pack format -- the 20-byte trailer
 201          * follows immediately after the last object data.
 202          */
 203         rix->revindex[num_ent].offset = p->pack_size - 20;
 204         rix->revindex[num_ent].nr = -1;
 205         qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
 206 }
 207
 208 static struct revindex_entry * find_packed_object(struct packed_git *p,
 209                                                   off_t ofs)
 210 {
 211         int num;
 212         int lo, hi;
 213         struct pack_revindex *rix;
 214         struct revindex_entry *revindex;
 215         num = pack_revindex_ix(p);
 216         if (num < 0)
 217                 die("internal error: pack revindex uninitialized");
 218         rix = &pack_revindex[num];
 219         if (!rix->revindex)
 220                 prepare_pack_revindex(rix);
 221         revindex = rix->revindex;
 222         lo = 0;
 223         hi = p->num_objects + 1;
 224         do {
 225                 int mi = (lo + hi) / 2;
 226                 if (revindex[mi].offset == ofs) {
 227                         return revindex + mi;
 228                 }
 229                 else if (ofs < revindex[mi].offset)
 230                         hi = mi;
 231                 else
 232                         lo = mi + 1;
 233         } while (lo < hi);
 234         die("internal error: pack revindex corrupt");
 235 }
 236
 237 static const unsigned char *find_packed_object_name(struct packed_git *p,
 238                                                     off_t ofs)
 239 {
 240         struct revindex_entry *entry = find_packed_object(p, ofs);
 241         return nth_packed_object_sha1(p, entry->nr);
 242 }
 243
 244 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 245 {
 246         unsigned long othersize, delta_size;
 247         enum object_type type;
 248         void *otherbuf = read_sha1_file(entry->delta->idx.sha1, &type, &othersize);
 249         void *delta_buf;
 250
 251         if (!otherbuf)
 252                 die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 253         delta_buf = diff_delta(otherbuf, othersize,
 254                                buf, size, &delta_size, 0);
 255         if (!delta_buf || delta_size != entry->delta_size)
 256                 die("delta size changed");
 257         free(buf);
 258         free(otherbuf);
 259         return delta_buf;
 260 }
 261
 262 /*
 263  * The per-object header is a pretty dense thing, which is
 264  *  - first byte: low four bits are "size", then three bits of "type",
 265  *    and the high bit is "size continues".
 266  *  - each byte afterwards: low seven bits are size continuation,
 267  *    with the high bit being "size continues"
 268  */
 269 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 270 {
 271         int n = 1;
 272         unsigned char c;
 273
 274         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 275                 die("bad type %d", type);
 276
 277         c = (type << 4) | (size & 15);
 278         size >>= 4;
 279         while (size) {
 280                 *hdr++ = c | 0x80;
 281                 c = size & 0x7f;
 282                 size >>= 7;
 283                 n++;
 284         }
 285         *hdr = c;
 286         return n;
 287 }
 288
 289 /*
 290  * we are going to reuse the existing object data as is.  make
 291  * sure it is not corrupt.
 292  */
 293 static int check_pack_inflate(struct packed_git *p,
 294                 struct pack_window **w_curs,
 295                 off_t offset,
 296                 off_t len,
 297                 unsigned long expect)
 298 {
 299         z_stream stream;
 300         unsigned char fakebuf[4096], *in;
 301         int st;
 302
 303         memset(&stream, 0, sizeof(stream));
 304         inflateInit(&stream);
 305         do {
 306                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 307                 stream.next_in = in;
 308                 stream.next_out = fakebuf;
 309                 stream.avail_out = sizeof(fakebuf);
 310                 st = inflate(&stream, Z_FINISH);
 311                 offset += stream.next_in - in;
 312         } while (st == Z_OK || st == Z_BUF_ERROR);
 313         inflateEnd(&stream);
 314         return (st == Z_STREAM_END &&
 315                 stream.total_out == expect &&
 316                 stream.total_in == len) ? 0 : -1;
 317 }
 318
 319 static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
 320                           off_t offset, off_t len, unsigned int nr)
 321 {
 322         const uint32_t *index_crc;
 323         uint32_t data_crc = crc32(0, Z_NULL, 0);
 324
 325         do {
 326                 unsigned int avail;
 327                 void *data = use_pack(p, w_curs, offset, &avail);
 328                 if (avail > len)
 329                         avail = len;
 330                 data_crc = crc32(data_crc, data, avail);
 331                 offset += avail;
 332                 len -= avail;
 333         } while (len);
 334
 335         index_crc = p->index_data;
 336         index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
 337
 338         return data_crc != ntohl(*index_crc);
 339 }
 340
 341 static void copy_pack_data(struct sha1file *f,
 342                 struct packed_git *p,
 343                 struct pack_window **w_curs,
 344                 off_t offset,
 345                 off_t len)
 346 {
 347         unsigned char *in;
 348         unsigned int avail;
 349
 350         while (len) {
 351                 in = use_pack(p, w_curs, offset, &avail);
 352                 if (avail > len)
 353                         avail = (unsigned int)len;
 354                 sha1write(f, in, avail);
 355                 offset += avail;
 356                 len -= avail;
 357         }
 358 }
 359
 360 static unsigned long write_object(struct sha1file *f,
 361                                   struct object_entry *entry,
 362                                   off_t write_offset)
 363 {
 364         unsigned long size;
 365         enum object_type type;
 366         void *buf;
 367         unsigned char header[10];
 368         unsigned char dheader[10];
 369         unsigned hdrlen;
 370         off_t datalen;
 371         enum object_type obj_type;
 372         int to_reuse = 0;
 373         /* write limit if limited packsize and not first object */
 374         unsigned long limit = pack_size_limit && nr_written ?
 375                                 pack_size_limit - write_offset : 0;
 376                                 /* no if no delta */
 377         int usable_delta =      !entry->delta ? 0 :
 378                                 /* yes if unlimited packfile */
 379                                 !pack_size_limit ? 1 :
 380                                 /* no if base written to previous pack */
 381                                 entry->delta->idx.offset == (off_t)-1 ? 0 :
 382                                 /* otherwise double-check written to this
 383                                  * pack,  like we do below
 384                                  */
 385                                 entry->delta->idx.offset ? 1 : 0;
 386
 387         if (!pack_to_stdout)
 388                 crc32_begin(f);
 389
 390         obj_type = entry->type;
 391         if (no_reuse_object)
 392                 to_reuse = 0;   /* explicit */
 393         else if (!entry->in_pack)
 394                 to_reuse = 0;   /* can't reuse what we don't have */
 395         else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
 396                                 /* check_object() decided it for us ... */
 397                 to_reuse = usable_delta;
 398                                 /* ... but pack split may override that */
 399         else if (obj_type != entry->in_pack_type)
 400                 to_reuse = 0;   /* pack has delta which is unusable */
 401         else if (entry->delta)
 402                 to_reuse = 0;   /* we want to pack afresh */
 403         else
 404                 to_reuse = 1;   /* we have it in-pack undeltified,
 405                                  * and we do not need to deltify it.
 406                                  */
 407
 408         if (!to_reuse) {
 409                 z_stream stream;
 410                 unsigned long maxsize;
 411                 void *out;
 412                 if (!usable_delta) {
 413                         buf = read_sha1_file(entry->idx.sha1, &obj_type, &size);
 414                         if (!buf)
 415                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 416                 } else if (entry->delta_data) {
 417                         size = entry->delta_size;
 418                         buf = entry->delta_data;
 419                         entry->delta_data = NULL;
 420                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 421                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 422                 } else {
 423                         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 424                         if (!buf)
 425                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 426                         buf = delta_against(buf, size, entry);
 427                         size = entry->delta_size;
 428                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 429                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 430                 }
 431                 /* compress the data to store and put compressed length in datalen */
 432                 memset(&stream, 0, sizeof(stream));
 433                 deflateInit(&stream, pack_compression_level);
 434                 maxsize = deflateBound(&stream, size);
 435                 out = xmalloc(maxsize);
 436                 /* Compress it */
 437                 stream.next_in = buf;
 438                 stream.avail_in = size;
 439                 stream.next_out = out;
 440                 stream.avail_out = maxsize;
 441                 while (deflate(&stream, Z_FINISH) == Z_OK)
 442                         /* nothing */;
 443                 deflateEnd(&stream);
 444                 datalen = stream.total_out;
 445                 deflateEnd(&stream);
 446                 /*
 447                  * The object header is a byte of 'type' followed by zero or
 448                  * more bytes of length.
 449                  */
 450                 hdrlen = encode_header(obj_type, size, header);
 451
 452                 if (obj_type == OBJ_OFS_DELTA) {
 453                         /*
 454                          * Deltas with relative base contain an additional
 455                          * encoding of the relative offset for the delta
 456                          * base from this object's position in the pack.
 457                          */
 458                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 459                         unsigned pos = sizeof(dheader) - 1;
 460                         dheader[pos] = ofs & 127;
 461                         while (ofs >>= 7)
 462                                 dheader[--pos] = 128 | (--ofs & 127);
 463                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 464                                 free(out);
 465                                 free(buf);
 466                                 return 0;
 467                         }
 468                         sha1write(f, header, hdrlen);
 469                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 470                         hdrlen += sizeof(dheader) - pos;
 471                 } else if (obj_type == OBJ_REF_DELTA) {
 472                         /*
 473                          * Deltas with a base reference contain
 474                          * an additional 20 bytes for the base sha1.
 475                          */
 476                         if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 477                                 free(out);
 478                                 free(buf);
 479                                 return 0;
 480                         }
 481                         sha1write(f, header, hdrlen);
 482                         sha1write(f, entry->delta->idx.sha1, 20);
 483                         hdrlen += 20;
 484                 } else {
 485                         if (limit && hdrlen + datalen + 20 >= limit) {
 486                                 free(out);
 487                                 free(buf);
 488                                 return 0;
 489                         }
 490                         sha1write(f, header, hdrlen);
 491                 }
 492                 sha1write(f, out, datalen);
 493                 free(out);
 494                 free(buf);
 495         }
 496         else {
 497                 struct packed_git *p = entry->in_pack;
 498                 struct pack_window *w_curs = NULL;
 499                 struct revindex_entry *revidx;
 500                 off_t offset;
 501
 502                 if (entry->delta) {
 503                         obj_type = (allow_ofs_delta && entry->delta->idx.offset) ?
 504                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 505                         reused_delta++;
 506                 }
 507                 hdrlen = encode_header(obj_type, entry->size, header);
 508                 offset = entry->in_pack_offset;
 509                 revidx = find_packed_object(p, offset);
 510                 datalen = revidx[1].offset - offset;
 511                 if (!pack_to_stdout && p->index_version > 1 &&
 512                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
 513                         die("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 514                 offset += entry->in_pack_header_size;
 515                 datalen -= entry->in_pack_header_size;
 516                 if (obj_type == OBJ_OFS_DELTA) {
 517                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 518                         unsigned pos = sizeof(dheader) - 1;
 519                         dheader[pos] = ofs & 127;
 520                         while (ofs >>= 7)
 521                                 dheader[--pos] = 128 | (--ofs & 127);
 522                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit)
 523                                 return 0;
 524                         sha1write(f, header, hdrlen);
 525                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 526                         hdrlen += sizeof(dheader) - pos;
 527                 } else if (obj_type == OBJ_REF_DELTA) {
 528                         if (limit && hdrlen + 20 + datalen + 20 >= limit)
 529                                 return 0;
 530                         sha1write(f, header, hdrlen);
 531                         sha1write(f, entry->delta->idx.sha1, 20);
 532                         hdrlen += 20;
 533                 } else {
 534                         if (limit && hdrlen + datalen + 20 >= limit)
 535                                 return 0;
 536                         sha1write(f, header, hdrlen);
 537                 }
 538
 539                 if (!pack_to_stdout && p->index_version == 1 &&
 540                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
 541                         die("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 542                 copy_pack_data(f, p, &w_curs, offset, datalen);
 543                 unuse_pack(&w_curs);
 544                 reused++;
 545         }
 546         if (usable_delta)
 547                 written_delta++;
 548         written++;
 549         if (!pack_to_stdout)
 550                 entry->idx.crc32 = crc32_end(f);
 551         return hdrlen + datalen;
 552 }
 553
 554 static off_t write_one(struct sha1file *f,
 555                                struct object_entry *e,
 556                                off_t offset)
 557 {
 558         unsigned long size;
 559
 560         /* offset is non zero if object is written already. */
 561         if (e->idx.offset || e->preferred_base)
 562                 return offset;
 563
 564         /* if we are deltified, write out base object first. */
 565         if (e->delta) {
 566                 offset = write_one(f, e->delta, offset);
 567                 if (!offset)
 568                         return 0;
 569         }
 570
 571         e->idx.offset = offset;
 572         size = write_object(f, e, offset);
 573         if (!size) {
 574                 e->idx.offset = 0;
 575                 return 0;
 576         }
 577         written_list[nr_written++] = e;
 578
 579         /* make sure off_t is sufficiently large not to wrap */
 580         if (offset > offset + size)
 581                 die("pack too large for current definition of off_t");
 582         return offset + size;
 583 }
 584
 585 static int open_object_dir_tmp(const char *path)
 586 {
 587     snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
 588     return xmkstemp(tmpname);
 589 }
 590
 591 /* forward declaration for write_pack_file */
 592 static int adjust_perm(const char *path, mode_t mode);
 593
 594 static void write_pack_file(void)
 595 {
 596         uint32_t i = 0, j;
 597         struct sha1file *f;
 598         off_t offset, offset_one, last_obj_offset = 0;
 599         struct pack_header hdr;
 600         int do_progress = progress >> pack_to_stdout;
 601         uint32_t nr_remaining = nr_result;
 602
 603         if (do_progress)
 604                 start_progress(&progress_state, "Writing %u objects...", "", nr_result);
 605         written_list = xmalloc(nr_objects * sizeof(struct object_entry *));
 606
 607         do {
 608                 unsigned char sha1[20];
 609
 610                 if (pack_to_stdout) {
 611                         f = sha1fd(1, "<stdout>");
 612                 } else {
 613                         int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
 614                         pack_tmp_name = xstrdup(tmpname);
 615                         f = sha1fd(fd, pack_tmp_name);
 616                 }
 617
 618                 hdr.hdr_signature = htonl(PACK_SIGNATURE);
 619                 hdr.hdr_version = htonl(PACK_VERSION);
 620                 hdr.hdr_entries = htonl(nr_remaining);
 621                 sha1write(f, &hdr, sizeof(hdr));
 622                 offset = sizeof(hdr);
 623                 nr_written = 0;
 624                 for (; i < nr_objects; i++) {
 625                         last_obj_offset = offset;
 626                         offset_one = write_one(f, objects + i, offset);
 627                         if (!offset_one)
 628                                 break;
 629                         offset = offset_one;
 630                         if (do_progress)
 631                                 display_progress(&progress_state, written);
 632                 }
 633
 634                 /*
 635                  * Did we write the wrong # entries in the header?
 636                  * If so, rewrite it like in fast-import
 637                  */
 638                 if (pack_to_stdout || nr_written == nr_remaining) {
 639                         sha1close(f, sha1, 1);
 640                 } else {
 641                         sha1close(f, sha1, 0);
 642                         fixup_pack_header_footer(f->fd, sha1, pack_tmp_name, nr_written);
 643                         close(f->fd);
 644                 }
 645
 646                 if (!pack_to_stdout) {
 647                         mode_t mode = umask(0);
 648
 649                         umask(mode);
 650                         mode = 0444 & ~mode;
 651
 652                         idx_tmp_name = write_idx_file(NULL,
 653                                 (struct pack_idx_entry **) written_list, nr_written, sha1);
 654                         snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
 655                                  base_name, sha1_to_hex(sha1));
 656                         if (adjust_perm(pack_tmp_name, mode))
 657                                 die("unable to make temporary pack file readable: %s",
 658                                     strerror(errno));
 659                         if (rename(pack_tmp_name, tmpname))
 660                                 die("unable to rename temporary pack file: %s",
 661                                     strerror(errno));
 662                         snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
 663                                  base_name, sha1_to_hex(sha1));
 664                         if (adjust_perm(idx_tmp_name, mode))
 665                                 die("unable to make temporary index file readable: %s",
 666                                     strerror(errno));
 667                         if (rename(idx_tmp_name, tmpname))
 668                                 die("unable to rename temporary index file: %s",
 669                                     strerror(errno));
 670                         puts(sha1_to_hex(sha1));
 671                 }
 672
 673                 /* mark written objects as written to previous pack */
 674                 for (j = 0; j < nr_written; j++) {
 675                         written_list[j]->idx.offset = (off_t)-1;
 676                 }
 677                 nr_remaining -= nr_written;
 678         } while (nr_remaining && i < nr_objects);
 679
 680         free(written_list);
 681         if (do_progress)
 682                 stop_progress(&progress_state);
 683         if (written != nr_result)
 684                 die("wrote %u objects while expecting %u", written, nr_result);
 685         /*
 686          * We have scanned through [0 ... i).  Since we have written
 687          * the correct number of objects,  the remaining [i ... nr_objects)
 688          * items must be either already written (due to out-of-order delta base)
 689          * or a preferred base.  Count those which are neither and complain if any.
 690          */
 691         for (j = 0; i < nr_objects; i++) {
 692                 struct object_entry *e = objects + i;
 693                 j += !e->idx.offset && !e->preferred_base;
 694         }
 695         if (j)
 696                 die("wrote %u objects as expected but %u unwritten", written, j);
 697 }
 698
 699 static int locate_object_entry_hash(const unsigned char *sha1)
 700 {
 701         int i;
 702         unsigned int ui;
 703         memcpy(&ui, sha1, sizeof(unsigned int));
 704         i = ui % object_ix_hashsz;
 705         while (0 < object_ix[i]) {
 706                 if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1))
 707                         return i;
 708                 if (++i == object_ix_hashsz)
 709                         i = 0;
 710         }
 711         return -1 - i;
 712 }
 713
 714 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 715 {
 716         int i;
 717
 718         if (!object_ix_hashsz)
 719                 return NULL;
 720
 721         i = locate_object_entry_hash(sha1);
 722         if (0 <= i)
 723                 return &objects[object_ix[i]-1];
 724         return NULL;
 725 }
 726
 727 static void rehash_objects(void)
 728 {
 729         uint32_t i;
 730         struct object_entry *oe;
 731
 732         object_ix_hashsz = nr_objects * 3;
 733         if (object_ix_hashsz < 1024)
 734                 object_ix_hashsz = 1024;
 735         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 736         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 737         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 738                 int ix = locate_object_entry_hash(oe->idx.sha1);
 739                 if (0 <= ix)
 740                         continue;
 741                 ix = -1 - ix;
 742                 object_ix[ix] = i + 1;
 743         }
 744 }
 745
 746 static unsigned name_hash(const char *name)
 747 {
 748         unsigned char c;
 749         unsigned hash = 0;
 750
 751         if (!name)
 752                 return 0;
 753
 754         /*
 755          * This effectively just creates a sortable number from the
 756          * last sixteen non-whitespace characters. Last characters
 757          * count "most", so things that end in ".c" sort together.
 758          */
 759         while ((c = *name++) != 0) {
 760                 if (isspace(c))
 761                         continue;
 762                 hash = (hash >> 2) + (c << 24);
 763         }
 764         return hash;
 765 }
 766
 767 static void setup_delta_attr_check(struct git_attr_check *check)
 768 {
 769         static struct git_attr *attr_delta;
 770
 771         if (!attr_delta)
 772                 attr_delta = git_attr("delta", 5);
 773
 774         check[0].attr = attr_delta;
 775 }
 776
 777 static int no_try_delta(const char *path)
 778 {
 779         struct git_attr_check check[1];
 780
 781         setup_delta_attr_check(check);
 782         if (git_checkattr(path, ARRAY_SIZE(check), check))
 783                 return 0;
 784         if (ATTR_FALSE(check->value))
 785                 return 1;
 786         return 0;
 787 }
 788
 789 static int add_object_entry(const unsigned char *sha1, enum object_type type,
 790                             const char *name, int exclude)
 791 {
 792         struct object_entry *entry;
 793         struct packed_git *p, *found_pack = NULL;
 794         off_t found_offset = 0;
 795         int ix;
 796         unsigned hash = name_hash(name);
 797
 798         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 799         if (ix >= 0) {
 800                 if (exclude) {
 801                         entry = objects + object_ix[ix] - 1;
 802                         if (!entry->preferred_base)
 803                                 nr_result--;
 804                         entry->preferred_base = 1;
 805                 }
 806                 return 0;
 807         }
 808
 809         for (p = packed_git; p; p = p->next) {
 810                 off_t offset = find_pack_entry_one(sha1, p);
 811                 if (offset) {
 812                         if (!found_pack) {
 813                                 found_offset = offset;
 814                                 found_pack = p;
 815                         }
 816                         if (exclude)
 817                                 break;
 818                         if (incremental)
 819                                 return 0;
 820                         if (local && !p->pack_local)
 821                                 return 0;
 822                 }
 823         }
 824
 825         if (nr_objects >= nr_alloc) {
 826                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 827                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 828         }
 829
 830         entry = objects + nr_objects++;
 831         memset(entry, 0, sizeof(*entry));
 832         hashcpy(entry->idx.sha1, sha1);
 833         entry->hash = hash;
 834         if (type)
 835                 entry->type = type;
 836         if (exclude)
 837                 entry->preferred_base = 1;
 838         else
 839                 nr_result++;
 840         if (found_pack) {
 841                 entry->in_pack = found_pack;
 842                 entry->in_pack_offset = found_offset;
 843         }
 844
 845         if (object_ix_hashsz * 3 <= nr_objects * 4)
 846                 rehash_objects();
 847         else
 848                 object_ix[-1 - ix] = nr_objects;
 849
 850         if (progress)
 851                 display_progress(&progress_state, nr_objects);
 852
 853         if (name && no_try_delta(name))
 854                 entry->no_try_delta = 1;
 855
 856         return 1;
 857 }
 858
 859 struct pbase_tree_cache {
 860         unsigned char sha1[20];
 861         int ref;
 862         int temporary;
 863         void *tree_data;
 864         unsigned long tree_size;
 865 };
 866
 867 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 868 static int pbase_tree_cache_ix(const unsigned char *sha1)
 869 {
 870         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 871 }
 872 static int pbase_tree_cache_ix_incr(int ix)
 873 {
 874         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 875 }
 876
 877 static struct pbase_tree {
 878         struct pbase_tree *next;
 879         /* This is a phony "cache" entry; we are not
 880          * going to evict it nor find it through _get()
 881          * mechanism -- this is for the toplevel node that
 882          * would almost always change with any commit.
 883          */
 884         struct pbase_tree_cache pcache;
 885 } *pbase_tree;
 886
 887 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 888 {
 889         struct pbase_tree_cache *ent, *nent;
 890         void *data;
 891         unsigned long size;
 892         enum object_type type;
 893         int neigh;
 894         int my_ix = pbase_tree_cache_ix(sha1);
 895         int available_ix = -1;
 896
 897         /* pbase-tree-cache acts as a limited hashtable.
 898          * your object will be found at your index or within a few
 899          * slots after that slot if it is cached.
 900          */
 901         for (neigh = 0; neigh < 8; neigh++) {
 902                 ent = pbase_tree_cache[my_ix];
 903                 if (ent && !hashcmp(ent->sha1, sha1)) {
 904                         ent->ref++;
 905                         return ent;
 906                 }
 907                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 908                          ((0 <= available_ix) &&
 909                           (!ent && pbase_tree_cache[available_ix])))
 910                         available_ix = my_ix;
 911                 if (!ent)
 912                         break;
 913                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 914         }
 915
 916         /* Did not find one.  Either we got a bogus request or
 917          * we need to read and perhaps cache.
 918          */
 919         data = read_sha1_file(sha1, &type, &size);
 920         if (!data)
 921                 return NULL;
 922         if (type != OBJ_TREE) {
 923                 free(data);
 924                 return NULL;
 925         }
 926
 927         /* We need to either cache or return a throwaway copy */
 928
 929         if (available_ix < 0)
 930                 ent = NULL;
 931         else {
 932                 ent = pbase_tree_cache[available_ix];
 933                 my_ix = available_ix;
 934         }
 935
 936         if (!ent) {
 937                 nent = xmalloc(sizeof(*nent));
 938                 nent->temporary = (available_ix < 0);
 939         }
 940         else {
 941                 /* evict and reuse */
 942                 free(ent->tree_data);
 943                 nent = ent;
 944         }
 945         hashcpy(nent->sha1, sha1);
 946         nent->tree_data = data;
 947         nent->tree_size = size;
 948         nent->ref = 1;
 949         if (!nent->temporary)
 950                 pbase_tree_cache[my_ix] = nent;
 951         return nent;
 952 }
 953
 954 static void pbase_tree_put(struct pbase_tree_cache *cache)
 955 {
 956         if (!cache->temporary) {
 957                 cache->ref--;
 958                 return;
 959         }
 960         free(cache->tree_data);
 961         free(cache);
 962 }
 963
 964 static int name_cmp_len(const char *name)
 965 {
 966         int i;
 967         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 968                 ;
 969         return i;
 970 }
 971
 972 static void add_pbase_object(struct tree_desc *tree,
 973                              const char *name,
 974                              int cmplen,
 975                              const char *fullname)
 976 {
 977         struct name_entry entry;
 978         int cmp;
 979
 980         while (tree_entry(tree,&entry)) {
 981                 if (S_ISGITLINK(entry.mode))
 982                         continue;
 983                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 984                       memcmp(name, entry.path, cmplen);
 985                 if (cmp > 0)
 986                         continue;
 987                 if (cmp < 0)
 988                         return;
 989                 if (name[cmplen] != '/') {
 990                         add_object_entry(entry.sha1,
 991                                          S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB,
 992                                          fullname, 1);
 993                         return;
 994                 }
 995                 if (S_ISDIR(entry.mode)) {
 996                         struct tree_desc sub;
 997                         struct pbase_tree_cache *tree;
 998                         const char *down = name+cmplen+1;
 999                         int downlen = name_cmp_len(down);
1000
1001                         tree = pbase_tree_get(entry.sha1);
1002                         if (!tree)
1003                                 return;
1004                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
1005
1006                         add_pbase_object(&sub, down, downlen, fullname);
1007                         pbase_tree_put(tree);
1008                 }
1009         }
1010 }
1011
1012 static unsigned *done_pbase_paths;
1013 static int done_pbase_paths_num;
1014 static int done_pbase_paths_alloc;
1015 static int done_pbase_path_pos(unsigned hash)
1016 {
1017         int lo = 0;
1018         int hi = done_pbase_paths_num;
1019         while (lo < hi) {
1020                 int mi = (hi + lo) / 2;
1021                 if (done_pbase_paths[mi] == hash)
1022                         return mi;
1023                 if (done_pbase_paths[mi] < hash)
1024                         hi = mi;
1025                 else
1026                         lo = mi + 1;
1027         }
1028         return -lo-1;
1029 }
1030
1031 static int check_pbase_path(unsigned hash)
1032 {
1033         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
1034         if (0 <= pos)
1035                 return 1;
1036         pos = -pos - 1;
1037         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
1038                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
1039                 done_pbase_paths = xrealloc(done_pbase_paths,
1040                                             done_pbase_paths_alloc *
1041                                             sizeof(unsigned));
1042         }
1043         done_pbase_paths_num++;
1044         if (pos < done_pbase_paths_num)
1045                 memmove(done_pbase_paths + pos + 1,
1046                         done_pbase_paths + pos,
1047                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1048         done_pbase_paths[pos] = hash;
1049         return 0;
1050 }
1051
1052 static void add_preferred_base_object(const char *name)
1053 {
1054         struct pbase_tree *it;
1055         int cmplen;
1056         unsigned hash = name_hash(name);
1057
1058         if (!num_preferred_base || check_pbase_path(hash))
1059                 return;
1060
1061         cmplen = name_cmp_len(name);
1062         for (it = pbase_tree; it; it = it->next) {
1063                 if (cmplen == 0) {
1064                         add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
1065                 }
1066                 else {
1067                         struct tree_desc tree;
1068                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1069                         add_pbase_object(&tree, name, cmplen, name);
1070                 }
1071         }
1072 }
1073
1074 static void add_preferred_base(unsigned char *sha1)
1075 {
1076         struct pbase_tree *it;
1077         void *data;
1078         unsigned long size;
1079         unsigned char tree_sha1[20];
1080
1081         if (window <= num_preferred_base++)
1082                 return;
1083
1084         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1085         if (!data)
1086                 return;
1087
1088         for (it = pbase_tree; it; it = it->next) {
1089                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1090                         free(data);
1091                         return;
1092                 }
1093         }
1094
1095         it = xcalloc(1, sizeof(*it));
1096         it->next = pbase_tree;
1097         pbase_tree = it;
1098
1099         hashcpy(it->pcache.sha1, tree_sha1);
1100         it->pcache.tree_data = data;
1101         it->pcache.tree_size = size;
1102 }
1103
1104 static void check_object(struct object_entry *entry)
1105 {
1106         if (entry->in_pack) {
1107                 struct packed_git *p = entry->in_pack;
1108                 struct pack_window *w_curs = NULL;
1109                 const unsigned char *base_ref = NULL;
1110                 struct object_entry *base_entry;
1111                 unsigned long used, used_0;
1112                 unsigned int avail;
1113                 off_t ofs;
1114                 unsigned char *buf, c;
1115
1116                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1117
1118                 /*
1119                  * We want in_pack_type even if we do not reuse delta
1120                  * since non-delta representations could still be reused.
1121                  */
1122                 used = unpack_object_header_gently(buf, avail,
1123                                                    &entry->in_pack_type,
1124                                                    &entry->size);
1125
1126                 /*
1127                  * Determine if this is a delta and if so whether we can
1128                  * reuse it or not.  Otherwise let's find out as cheaply as
1129                  * possible what the actual type and size for this object is.
1130                  */
1131                 switch (entry->in_pack_type) {
1132                 default:
1133                         /* Not a delta hence we've already got all we need. */
1134                         entry->type = entry->in_pack_type;
1135                         entry->in_pack_header_size = used;
1136                         unuse_pack(&w_curs);
1137                         return;
1138                 case OBJ_REF_DELTA:
1139                         if (!no_reuse_delta && !entry->preferred_base)
1140                                 base_ref = use_pack(p, &w_curs,
1141                                                 entry->in_pack_offset + used, NULL);
1142                         entry->in_pack_header_size = used + 20;
1143                         break;
1144                 case OBJ_OFS_DELTA:
1145                         buf = use_pack(p, &w_curs,
1146                                        entry->in_pack_offset + used, NULL);
1147                         used_0 = 0;
1148                         c = buf[used_0++];
1149                         ofs = c & 127;
1150                         while (c & 128) {
1151                                 ofs += 1;
1152                                 if (!ofs || MSB(ofs, 7))
1153                                         die("delta base offset overflow in pack for %s",
1154                                             sha1_to_hex(entry->idx.sha1));
1155                                 c = buf[used_0++];
1156                                 ofs = (ofs << 7) + (c & 127);
1157                         }
1158                         if (ofs >= entry->in_pack_offset)
1159                                 die("delta base offset out of bound for %s",
1160                                     sha1_to_hex(entry->idx.sha1));
1161                         ofs = entry->in_pack_offset - ofs;
1162                         if (!no_reuse_delta && !entry->preferred_base)
1163                                 base_ref = find_packed_object_name(p, ofs);
1164                         entry->in_pack_header_size = used + used_0;
1165                         break;
1166                 }
1167
1168                 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1169                         /*
1170                          * If base_ref was set above that means we wish to
1171                          * reuse delta data, and we even found that base
1172                          * in the list of objects we want to pack. Goodie!
1173                          *
1174                          * Depth value does not matter - find_deltas() will
1175                          * never consider reused delta as the base object to
1176                          * deltify other objects against, in order to avoid
1177                          * circular deltas.
1178                          */
1179                         entry->type = entry->in_pack_type;
1180                         entry->delta = base_entry;
1181                         entry->delta_sibling = base_entry->delta_child;
1182                         base_entry->delta_child = entry;
1183                         unuse_pack(&w_curs);
1184                         return;
1185                 }
1186
1187                 if (entry->type) {
1188                         /*
1189                          * This must be a delta and we already know what the
1190                          * final object type is.  Let's extract the actual
1191                          * object size from the delta header.
1192                          */
1193                         entry->size = get_size_from_delta(p, &w_curs,
1194                                         entry->in_pack_offset + entry->in_pack_header_size);
1195                         unuse_pack(&w_curs);
1196                         return;
1197                 }
1198
1199                 /*
1200                  * No choice but to fall back to the recursive delta walk
1201                  * with sha1_object_info() to find about the object type
1202                  * at this point...
1203                  */
1204                 unuse_pack(&w_curs);
1205         }
1206
1207         entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1208         if (entry->type < 0)
1209                 die("unable to get type of object %s",
1210                     sha1_to_hex(entry->idx.sha1));
1211 }
1212
1213 static int pack_offset_sort(const void *_a, const void *_b)
1214 {
1215         const struct object_entry *a = *(struct object_entry **)_a;
1216         const struct object_entry *b = *(struct object_entry **)_b;
1217
1218         /* avoid filesystem trashing with loose objects */
1219         if (!a->in_pack && !b->in_pack)
1220                 return hashcmp(a->idx.sha1, b->idx.sha1);
1221
1222         if (a->in_pack < b->in_pack)
1223                 return -1;
1224         if (a->in_pack > b->in_pack)
1225                 return 1;
1226         return a->in_pack_offset < b->in_pack_offset ? -1 :
1227                         (a->in_pack_offset > b->in_pack_offset);
1228 }
1229
1230 static void get_object_details(void)
1231 {
1232         uint32_t i;
1233         struct object_entry **sorted_by_offset;
1234
1235         sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1236         for (i = 0; i < nr_objects; i++)
1237                 sorted_by_offset[i] = objects + i;
1238         qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1239
1240         prepare_pack_ix();
1241         for (i = 0; i < nr_objects; i++)
1242                 check_object(sorted_by_offset[i]);
1243         free(sorted_by_offset);
1244 }
1245
1246 static int type_size_sort(const void *_a, const void *_b)
1247 {
1248         const struct object_entry *a = *(struct object_entry **)_a;
1249         const struct object_entry *b = *(struct object_entry **)_b;
1250
1251         if (a->type < b->type)
1252                 return -1;
1253         if (a->type > b->type)
1254                 return 1;
1255         if (a->hash < b->hash)
1256                 return -1;
1257         if (a->hash > b->hash)
1258                 return 1;
1259         if (a->preferred_base < b->preferred_base)
1260                 return -1;
1261         if (a->preferred_base > b->preferred_base)
1262                 return 1;
1263         if (a->size < b->size)
1264                 return -1;
1265         if (a->size > b->size)
1266                 return 1;
1267         return a > b ? -1 : (a < b);  /* newest last */
1268 }
1269
1270 struct unpacked {
1271         struct object_entry *entry;
1272         void *data;
1273         struct delta_index *index;
1274         unsigned depth;
1275 };
1276
1277 static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1278                            unsigned long delta_size)
1279 {
1280         if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1281                 return 0;
1282
1283         if (delta_size < cache_max_small_delta_size)
1284                 return 1;
1285
1286         /* cache delta, if objects are large enough compared to delta size */
1287         if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1288                 return 1;
1289
1290         return 0;
1291 }
1292
1293 /*
1294  * We search for deltas _backwards_ in a list sorted by type and
1295  * by size, so that we see progressively smaller and smaller files.
1296  * That's because we prefer deltas to be from the bigger file
1297  * to the smaller - deletes are potentially cheaper, but perhaps
1298  * more importantly, the bigger file is likely the more recent
1299  * one.
1300  */
1301 static int try_delta(struct unpacked *trg, struct unpacked *src,
1302                      unsigned max_depth, unsigned long *mem_usage)
1303 {
1304         struct object_entry *trg_entry = trg->entry;
1305         struct object_entry *src_entry = src->entry;
1306         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1307         unsigned ref_depth;
1308         enum object_type type;
1309         void *delta_buf;
1310
1311         /* Don't bother doing diffs between different types */
1312         if (trg_entry->type != src_entry->type)
1313                 return -1;
1314
1315         /*
1316          * We do not bother to try a delta that we discarded
1317          * on an earlier try, but only when reusing delta data.
1318          */
1319         if (!no_reuse_delta && trg_entry->in_pack &&
1320             trg_entry->in_pack == src_entry->in_pack &&
1321             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1322             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1323                 return 0;
1324
1325         /* Let's not bust the allowed depth. */
1326         if (src->depth >= max_depth)
1327                 return 0;
1328
1329         /* Now some size filtering heuristics. */
1330         trg_size = trg_entry->size;
1331         if (!trg_entry->delta) {
1332                 max_size = trg_size/2 - 20;
1333                 ref_depth = 1;
1334         } else {
1335                 max_size = trg_entry->delta_size;
1336                 ref_depth = trg->depth;
1337         }
1338         max_size = max_size * (max_depth - src->depth) /
1339                                                 (max_depth - ref_depth + 1);
1340         if (max_size == 0)
1341                 return 0;
1342         src_size = src_entry->size;
1343         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1344         if (sizediff >= max_size)
1345                 return 0;
1346         if (trg_size < src_size / 32)
1347                 return 0;
1348
1349         /* Load data if not already done */
1350         if (!trg->data) {
1351                 trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1352                 if (!trg->data)
1353                         die("object %s cannot be read",
1354                             sha1_to_hex(trg_entry->idx.sha1));
1355                 if (sz != trg_size)
1356                         die("object %s inconsistent object length (%lu vs %lu)",
1357                             sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1358                 *mem_usage += sz;
1359         }
1360         if (!src->data) {
1361                 src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1362                 if (!src->data)
1363                         die("object %s cannot be read",
1364                             sha1_to_hex(src_entry->idx.sha1));
1365                 if (sz != src_size)
1366                         die("object %s inconsistent object length (%lu vs %lu)",
1367                             sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1368                 *mem_usage += sz;
1369         }
1370         if (!src->index) {
1371                 src->index = create_delta_index(src->data, src_size);
1372                 if (!src->index) {
1373                         static int warned = 0;
1374                         if (!warned++)
1375                                 warning("suboptimal pack - out of memory");
1376                         return 0;
1377                 }
1378                 *mem_usage += sizeof_delta_index(src->index);
1379         }
1380
1381         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1382         if (!delta_buf)
1383                 return 0;
1384
1385         if (trg_entry->delta) {
1386                 /* Prefer only shallower same-sized deltas. */
1387                 if (delta_size == trg_entry->delta_size &&
1388                     src->depth + 1 >= trg->depth) {
1389                         free(delta_buf);
1390                         return 0;
1391                 }
1392         }
1393
1394         trg_entry->delta = src_entry;
1395         trg_entry->delta_size = delta_size;
1396         trg->depth = src->depth + 1;
1397
1398         if (trg_entry->delta_data) {
1399                 delta_cache_size -= trg_entry->delta_size;
1400                 free(trg_entry->delta_data);
1401                 trg_entry->delta_data = NULL;
1402         }
1403
1404         if (delta_cacheable(src_size, trg_size, delta_size)) {
1405                 trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1406                 delta_cache_size += trg_entry->delta_size;
1407         } else
1408                 free(delta_buf);
1409         return 1;
1410 }
1411
1412 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1413 {
1414         struct object_entry *child = me->delta_child;
1415         unsigned int m = n;
1416         while (child) {
1417                 unsigned int c = check_delta_limit(child, n + 1);
1418                 if (m < c)
1419                         m = c;
1420                 child = child->delta_sibling;
1421         }
1422         return m;
1423 }
1424
1425 static unsigned long free_unpacked(struct unpacked *n)
1426 {
1427         unsigned long freed_mem = sizeof_delta_index(n->index);
1428         free_delta_index(n->index);
1429         n->index = NULL;
1430         if (n->data) {
1431                 freed_mem += n->entry->size;
1432                 free(n->data);
1433                 n->data = NULL;
1434         }
1435         n->entry = NULL;
1436         n->depth = 0;
1437         return freed_mem;
1438 }
1439
1440 static void find_deltas(struct object_entry **list, unsigned list_size,
1441                         unsigned nr_deltas, int window, int depth)
1442 {
1443         uint32_t i = list_size, idx = 0, count = 0, processed = 0;
1444         unsigned int array_size = window * sizeof(struct unpacked);
1445         struct unpacked *array;
1446         unsigned long mem_usage = 0;
1447
1448         array = xmalloc(array_size);
1449         memset(array, 0, array_size);
1450         if (progress)
1451                 start_progress(&progress_state, "Deltifying %u objects...", "", nr_deltas);
1452
1453         do {
1454                 struct object_entry *entry = list[--i];
1455                 struct unpacked *n = array + idx;
1456                 int j, max_depth, best_base = -1;
1457
1458                 mem_usage -= free_unpacked(n);
1459                 n->entry = entry;
1460
1461                 while (window_memory_limit &&
1462                        mem_usage > window_memory_limit &&
1463                        count > 1) {
1464                         uint32_t tail = (idx + window - count) % window;
1465                         mem_usage -= free_unpacked(array + tail);
1466                         count--;
1467                 }
1468
1469                 /* We do not compute delta to *create* objects we are not
1470                  * going to pack.
1471                  */
1472                 if (entry->preferred_base)
1473                         goto next;
1474
1475                 if (progress)
1476                         display_progress(&progress_state, ++processed);
1477
1478                 /*
1479                  * If the current object is at pack edge, take the depth the
1480                  * objects that depend on the current object into account
1481                  * otherwise they would become too deep.
1482                  */
1483                 max_depth = depth;
1484                 if (entry->delta_child) {
1485                         max_depth -= check_delta_limit(entry, 0);
1486                         if (max_depth <= 0)
1487                                 goto next;
1488                 }
1489
1490                 j = window;
1491                 while (--j > 0) {
1492                         int ret;
1493                         uint32_t other_idx = idx + j;
1494                         struct unpacked *m;
1495                         if (other_idx >= window)
1496                                 other_idx -= window;
1497                         m = array + other_idx;
1498                         if (!m->entry)
1499                                 break;
1500                         ret = try_delta(n, m, max_depth, &mem_usage);
1501                         if (ret < 0)
1502                                 break;
1503                         else if (ret > 0)
1504                                 best_base = other_idx;
1505                 }
1506
1507                 /* if we made n a delta, and if n is already at max
1508                  * depth, leaving it in the window is pointless.  we
1509                  * should evict it first.
1510                  */
1511                 if (entry->delta && depth <= n->depth)
1512                         continue;
1513
1514                 /*
1515                  * Move the best delta base up in the window, after the
1516                  * currently deltified object, to keep it longer.  It will
1517                  * be the first base object to be attempted next.
1518                  */
1519                 if (entry->delta) {
1520                         struct unpacked swap = array[best_base];
1521                         int dist = (window + idx - best_base) % window;
1522                         int dst = best_base;
1523                         while (dist--) {
1524                                 int src = (dst + 1) % window;
1525                                 array[dst] = array[src];
1526                                 dst = src;
1527                         }
1528                         array[dst] = swap;
1529                 }
1530
1531                 next:
1532                 idx++;
1533                 if (count + 1 < window)
1534                         count++;
1535                 if (idx >= window)
1536                         idx = 0;
1537         } while (i > 0);
1538
1539         if (progress)
1540                 stop_progress(&progress_state);
1541
1542         for (i = 0; i < window; ++i) {
1543                 free_delta_index(array[i].index);
1544                 free(array[i].data);
1545         }
1546         free(array);
1547 }
1548
1549 static void prepare_pack(int window, int depth)
1550 {
1551         struct object_entry **delta_list;
1552         uint32_t i, n, nr_deltas;
1553
1554         get_object_details();
1555
1556         if (!nr_objects || !window || !depth)
1557                 return;
1558
1559         delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1560         nr_deltas = n = 0;
1561
1562         for (i = 0; i < nr_objects; i++) {
1563                 struct object_entry *entry = objects + i;
1564
1565                 if (entry->delta)
1566                         /* This happens if we decided to reuse existing
1567                          * delta from a pack.  "!no_reuse_delta &&" is implied.
1568                          */
1569                         continue;
1570
1571                 if (entry->size < 50)
1572                         continue;
1573
1574                 if (entry->no_try_delta)
1575                         continue;
1576
1577                 if (!entry->preferred_base)
1578                         nr_deltas++;
1579
1580                 delta_list[n++] = entry;
1581         }
1582
1583         if (nr_deltas) {
1584                 qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
1585                 find_deltas(delta_list, n, nr_deltas, window+1, depth);
1586         }
1587         free(delta_list);
1588 }
1589
1590 static int git_pack_config(const char *k, const char *v)
1591 {
1592         if(!strcmp(k, "pack.window")) {
1593                 window = git_config_int(k, v);
1594                 return 0;
1595         }
1596         if (!strcmp(k, "pack.windowmemory")) {
1597                 window_memory_limit = git_config_ulong(k, v);
1598                 return 0;
1599         }
1600         if (!strcmp(k, "pack.depth")) {
1601                 depth = git_config_int(k, v);
1602                 return 0;
1603         }
1604         if (!strcmp(k, "pack.compression")) {
1605                 int level = git_config_int(k, v);
1606                 if (level == -1)
1607                         level = Z_DEFAULT_COMPRESSION;
1608                 else if (level < 0 || level > Z_BEST_COMPRESSION)
1609                         die("bad pack compression level %d", level);
1610                 pack_compression_level = level;
1611                 pack_compression_seen = 1;
1612                 return 0;
1613         }
1614         if (!strcmp(k, "pack.deltacachesize")) {
1615                 max_delta_cache_size = git_config_int(k, v);
1616                 return 0;
1617         }
1618         if (!strcmp(k, "pack.deltacachelimit")) {
1619                 cache_max_small_delta_size = git_config_int(k, v);
1620                 return 0;
1621         }
1622         return git_default_config(k, v);
1623 }
1624
1625 static void read_object_list_from_stdin(void)
1626 {
1627         char line[40 + 1 + PATH_MAX + 2];
1628         unsigned char sha1[20];
1629
1630         for (;;) {
1631                 if (!fgets(line, sizeof(line), stdin)) {
1632                         if (feof(stdin))
1633                                 break;
1634                         if (!ferror(stdin))
1635                                 die("fgets returned NULL, not EOF, not error!");
1636                         if (errno != EINTR)
1637                                 die("fgets: %s", strerror(errno));
1638                         clearerr(stdin);
1639                         continue;
1640                 }
1641                 if (line[0] == '-') {
1642                         if (get_sha1_hex(line+1, sha1))
1643                                 die("expected edge sha1, got garbage:\n %s",
1644                                     line);
1645                         add_preferred_base(sha1);
1646                         continue;
1647                 }
1648                 if (get_sha1_hex(line, sha1))
1649                         die("expected sha1, got garbage:\n %s", line);
1650
1651                 add_preferred_base_object(line+41);
1652                 add_object_entry(sha1, 0, line+41, 0);
1653         }
1654 }
1655
1656 static void show_commit(struct commit *commit)
1657 {
1658         add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
1659 }
1660
1661 static void show_object(struct object_array_entry *p)
1662 {
1663         add_preferred_base_object(p->name);
1664         add_object_entry(p->item->sha1, p->item->type, p->name, 0);
1665 }
1666
1667 static void show_edge(struct commit *commit)
1668 {
1669         add_preferred_base(commit->object.sha1);
1670 }
1671
1672 static void get_object_list(int ac, const char **av)
1673 {
1674         struct rev_info revs;
1675         char line[1000];
1676         int flags = 0;
1677
1678         init_revisions(&revs, NULL);
1679         save_commit_buffer = 0;
1680         track_object_refs = 0;
1681         setup_revisions(ac, av, &revs, NULL);
1682
1683         while (fgets(line, sizeof(line), stdin) != NULL) {
1684                 int len = strlen(line);
1685                 if (line[len - 1] == '\n')
1686                         line[--len] = 0;
1687                 if (!len)
1688                         break;
1689                 if (*line == '-') {
1690                         if (!strcmp(line, "--not")) {
1691                                 flags ^= UNINTERESTING;
1692                                 continue;
1693                         }
1694                         die("not a rev '%s'", line);
1695                 }
1696                 if (handle_revision_arg(line, &revs, flags, 1))
1697                         die("bad revision '%s'", line);
1698         }
1699
1700         prepare_revision_walk(&revs);
1701         mark_edges_uninteresting(revs.commits, &revs, show_edge);
1702         traverse_commit_list(&revs, show_commit, show_object);
1703 }
1704
1705 static int adjust_perm(const char *path, mode_t mode)
1706 {
1707         if (chmod(path, mode))
1708                 return -1;
1709         return adjust_shared_perm(path);
1710 }
1711
1712 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1713 {
1714         int use_internal_rev_list = 0;
1715         int thin = 0;
1716         uint32_t i;
1717         const char **rp_av;
1718         int rp_ac_alloc = 64;
1719         int rp_ac;
1720
1721         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1722
1723         rp_av[0] = "pack-objects";
1724         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1725         rp_ac = 2;
1726
1727         git_config(git_pack_config);
1728         if (!pack_compression_seen && core_compression_seen)
1729                 pack_compression_level = core_compression_level;
1730
1731         progress = isatty(2);
1732         for (i = 1; i < argc; i++) {
1733                 const char *arg = argv[i];
1734
1735                 if (*arg != '-')
1736                         break;
1737
1738                 if (!strcmp("--non-empty", arg)) {
1739                         non_empty = 1;
1740                         continue;
1741                 }
1742                 if (!strcmp("--local", arg)) {
1743                         local = 1;
1744                         continue;
1745                 }
1746                 if (!strcmp("--incremental", arg)) {
1747                         incremental = 1;
1748                         continue;
1749                 }
1750                 if (!prefixcmp(arg, "--compression=")) {
1751                         char *end;
1752                         int level = strtoul(arg+14, &end, 0);
1753                         if (!arg[14] || *end)
1754                                 usage(pack_usage);
1755                         if (level == -1)
1756                                 level = Z_DEFAULT_COMPRESSION;
1757                         else if (level < 0 || level > Z_BEST_COMPRESSION)
1758                                 die("bad pack compression level %d", level);
1759                         pack_compression_level = level;
1760                         continue;
1761                 }
1762                 if (!prefixcmp(arg, "--max-pack-size=")) {
1763                         char *end;
1764                         pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024;
1765                         if (!arg[16] || *end)
1766                                 usage(pack_usage);
1767                         continue;
1768                 }
1769                 if (!prefixcmp(arg, "--window=")) {
1770                         char *end;
1771                         window = strtoul(arg+9, &end, 0);
1772                         if (!arg[9] || *end)
1773                                 usage(pack_usage);
1774                         continue;
1775                 }
1776                 if (!prefixcmp(arg, "--window-memory=")) {
1777                         if (!git_parse_ulong(arg+16, &window_memory_limit))
1778                                 usage(pack_usage);
1779                         continue;
1780                 }
1781                 if (!prefixcmp(arg, "--depth=")) {
1782                         char *end;
1783                         depth = strtoul(arg+8, &end, 0);
1784                         if (!arg[8] || *end)
1785                                 usage(pack_usage);
1786                         continue;
1787                 }
1788                 if (!strcmp("--progress", arg)) {
1789                         progress = 1;
1790                         continue;
1791                 }
1792                 if (!strcmp("--all-progress", arg)) {
1793                         progress = 2;
1794                         continue;
1795                 }
1796                 if (!strcmp("-q", arg)) {
1797                         progress = 0;
1798                         continue;
1799                 }
1800                 if (!strcmp("--no-reuse-delta", arg)) {
1801                         no_reuse_delta = 1;
1802                         continue;
1803                 }
1804                 if (!strcmp("--no-reuse-object", arg)) {
1805                         no_reuse_object = no_reuse_delta = 1;
1806                         continue;
1807                 }
1808                 if (!strcmp("--delta-base-offset", arg)) {
1809                         allow_ofs_delta = 1;
1810                         continue;
1811                 }
1812                 if (!strcmp("--stdout", arg)) {
1813                         pack_to_stdout = 1;
1814                         continue;
1815                 }
1816                 if (!strcmp("--revs", arg)) {
1817                         use_internal_rev_list = 1;
1818                         continue;
1819                 }
1820                 if (!strcmp("--unpacked", arg) ||
1821                     !prefixcmp(arg, "--unpacked=") ||
1822                     !strcmp("--reflog", arg) ||
1823                     !strcmp("--all", arg)) {
1824                         use_internal_rev_list = 1;
1825                         if (rp_ac >= rp_ac_alloc - 1) {
1826                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1827                                 rp_av = xrealloc(rp_av,
1828                                                  rp_ac_alloc * sizeof(*rp_av));
1829                         }
1830                         rp_av[rp_ac++] = arg;
1831                         continue;
1832                 }
1833                 if (!strcmp("--thin", arg)) {
1834                         use_internal_rev_list = 1;
1835                         thin = 1;
1836                         rp_av[1] = "--objects-edge";
1837                         continue;
1838                 }
1839                 if (!prefixcmp(arg, "--index-version=")) {
1840                         char *c;
1841                         pack_idx_default_version = strtoul(arg + 16, &c, 10);
1842                         if (pack_idx_default_version > 2)
1843                                 die("bad %s", arg);
1844                         if (*c == ',')
1845                                 pack_idx_off32_limit = strtoul(c+1, &c, 0);
1846                         if (*c || pack_idx_off32_limit & 0x80000000)
1847                                 die("bad %s", arg);
1848                         continue;
1849                 }
1850                 usage(pack_usage);
1851         }
1852
1853         /* Traditionally "pack-objects [options] base extra" failed;
1854          * we would however want to take refs parameter that would
1855          * have been given to upstream rev-list ourselves, which means
1856          * we somehow want to say what the base name is.  So the
1857          * syntax would be:
1858          *
1859          * pack-objects [options] base <refs...>
1860          *
1861          * in other words, we would treat the first non-option as the
1862          * base_name and send everything else to the internal revision
1863          * walker.
1864          */
1865
1866         if (!pack_to_stdout)
1867                 base_name = argv[i++];
1868
1869         if (pack_to_stdout != !base_name)
1870                 usage(pack_usage);
1871
1872         if (pack_to_stdout && pack_size_limit)
1873                 die("--max-pack-size cannot be used to build a pack for transfer.");
1874
1875         if (!pack_to_stdout && thin)
1876                 die("--thin cannot be used to build an indexable pack.");
1877
1878         prepare_packed_git();
1879
1880         if (progress)
1881                 start_progress(&progress_state, "Generating pack...",
1882                                "Counting objects: ", 0);
1883         if (!use_internal_rev_list)
1884                 read_object_list_from_stdin();
1885         else {
1886                 rp_av[rp_ac] = NULL;
1887                 get_object_list(rp_ac, rp_av);
1888         }
1889         if (progress) {
1890                 stop_progress(&progress_state);
1891                 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
1892         }
1893
1894         if (non_empty && !nr_result)
1895                 return 0;
1896         if (progress && (nr_objects != nr_result))
1897                 fprintf(stderr, "Result has %u objects.\n", nr_result);
1898         if (nr_result)
1899                 prepare_pack(window, depth);
1900         write_pack_file();
1901         if (progress)
1902                 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
1903                         written, written_delta, reused, reused_delta);
1904         return 0;
1905 }