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 "pack-revindex.h"
  12 #include "csum-file.h"
  13 #include "tree-walk.h"
  14 #include "diff.h"
  15 #include "revision.h"
  16 #include "list-objects.h"
  17 #include "progress.h"
  18 #include "refs.h"
  19
  20 #ifdef THREADED_DELTA_SEARCH
  21 #include "thread-utils.h"
  22 #include <pthread.h>
  23 #endif
  24
  25 static const char pack_usage[] = "\
  26 git-pack-objects [{ -q | --progress | --all-progress }] \n\
  27         [--max-pack-size=N] [--local] [--incremental] \n\
  28         [--window=N] [--window-memory=N] [--depth=N] \n\
  29         [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
  30         [--threads=N] [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
  31         [--stdout | base-name] [--include-tag] [--keep-unreachable] \n\
  32         [<ref-list | <object-list]";
  33
  34 struct object_entry {
  35         struct pack_idx_entry idx;
  36         unsigned long size;     /* uncompressed size */
  37         struct packed_git *in_pack;     /* already in pack */
  38         off_t in_pack_offset;
  39         struct object_entry *delta;     /* delta base object */
  40         struct object_entry *delta_child; /* deltified objects who bases me */
  41         struct object_entry *delta_sibling; /* other deltified objects who
  42                                              * uses the same base as me
  43                                              */
  44         void *delta_data;       /* cached delta (uncompressed) */
  45         unsigned long delta_size;       /* delta data size (uncompressed) */
  46         unsigned int hash;      /* name hint hash */
  47         enum object_type type;
  48         enum object_type in_pack_type;  /* could be delta */
  49         unsigned char in_pack_header_size;
  50         unsigned char preferred_base; /* we do not pack this, but is available
  51                                        * to be used as the base object to delta
  52                                        * objects against.
  53                                        */
  54         unsigned char no_try_delta;
  55 };
  56
  57 /*
  58  * Objects we are going to pack are collected in objects array (dynamically
  59  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  60  * in the order we see -- typically rev-list --objects order that gives us
  61  * nice "minimum seek" order.
  62  */
  63 static struct object_entry *objects;
  64 static struct pack_idx_entry **written_list;
  65 static uint32_t nr_objects, nr_alloc, nr_result, nr_written;
  66
  67 static int non_empty;
  68 static int reuse_delta = 1, reuse_object = 1;
  69 static int keep_unreachable, include_tag;
  70 static int local;
  71 static int incremental;
  72 static int allow_ofs_delta;
  73 static const char *base_name;
  74 static int progress = 1;
  75 static int window = 10;
  76 static uint32_t pack_size_limit, pack_size_limit_cfg;
  77 static int depth = 50;
  78 static int delta_search_threads = 1;
  79 static int pack_to_stdout;
  80 static int num_preferred_base;
  81 static struct progress *progress_state;
  82 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  83 static int pack_compression_seen;
  84
  85 static unsigned long delta_cache_size = 0;
  86 static unsigned long max_delta_cache_size = 0;
  87 static unsigned long cache_max_small_delta_size = 1000;
  88
  89 static unsigned long window_memory_limit = 0;
  90
  91 /*
  92  * The object names in objects array are hashed with this hashtable,
  93  * to help looking up the entry by object name.
  94  * This hashtable is built after all the objects are seen.
  95  */
  96 static int *object_ix;
  97 static int object_ix_hashsz;
  98
  99 /*
 100  * stats
 101  */
 102 static uint32_t written, written_delta;
 103 static uint32_t reused, reused_delta;
 104
 105
 106 static void *get_delta(struct object_entry *entry)
 107 {
 108         unsigned long size, base_size, delta_size;
 109         void *buf, *base_buf, *delta_buf;
 110         enum object_type type;
 111
 112         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 113         if (!buf)
 114                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 115         base_buf = read_sha1_file(entry->delta->idx.sha1, &type, &base_size);
 116         if (!base_buf)
 117                 die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 118         delta_buf = diff_delta(base_buf, base_size,
 119                                buf, size, &delta_size, 0);
 120         if (!delta_buf || delta_size != entry->delta_size)
 121                 die("delta size changed");
 122         free(buf);
 123         free(base_buf);
 124         return delta_buf;
 125 }
 126
 127 /*
 128  * The per-object header is a pretty dense thing, which is
 129  *  - first byte: low four bits are "size", then three bits of "type",
 130  *    and the high bit is "size continues".
 131  *  - each byte afterwards: low seven bits are size continuation,
 132  *    with the high bit being "size continues"
 133  */
 134 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 135 {
 136         int n = 1;
 137         unsigned char c;
 138
 139         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 140                 die("bad type %d", type);
 141
 142         c = (type << 4) | (size & 15);
 143         size >>= 4;
 144         while (size) {
 145                 *hdr++ = c | 0x80;
 146                 c = size & 0x7f;
 147                 size >>= 7;
 148                 n++;
 149         }
 150         *hdr = c;
 151         return n;
 152 }
 153
 154 /*
 155  * we are going to reuse the existing object data as is.  make
 156  * sure it is not corrupt.
 157  */
 158 static int check_pack_inflate(struct packed_git *p,
 159                 struct pack_window **w_curs,
 160                 off_t offset,
 161                 off_t len,
 162                 unsigned long expect)
 163 {
 164         z_stream stream;
 165         unsigned char fakebuf[4096], *in;
 166         int st;
 167
 168         memset(&stream, 0, sizeof(stream));
 169         inflateInit(&stream);
 170         do {
 171                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 172                 stream.next_in = in;
 173                 stream.next_out = fakebuf;
 174                 stream.avail_out = sizeof(fakebuf);
 175                 st = inflate(&stream, Z_FINISH);
 176                 offset += stream.next_in - in;
 177         } while (st == Z_OK || st == Z_BUF_ERROR);
 178         inflateEnd(&stream);
 179         return (st == Z_STREAM_END &&
 180                 stream.total_out == expect &&
 181                 stream.total_in == len) ? 0 : -1;
 182 }
 183
 184 static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
 185                           off_t offset, off_t len, unsigned int nr)
 186 {
 187         const uint32_t *index_crc;
 188         uint32_t data_crc = crc32(0, Z_NULL, 0);
 189
 190         do {
 191                 unsigned int avail;
 192                 void *data = use_pack(p, w_curs, offset, &avail);
 193                 if (avail > len)
 194                         avail = len;
 195                 data_crc = crc32(data_crc, data, avail);
 196                 offset += avail;
 197                 len -= avail;
 198         } while (len);
 199
 200         index_crc = p->index_data;
 201         index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
 202
 203         return data_crc != ntohl(*index_crc);
 204 }
 205
 206 static void copy_pack_data(struct sha1file *f,
 207                 struct packed_git *p,
 208                 struct pack_window **w_curs,
 209                 off_t offset,
 210                 off_t len)
 211 {
 212         unsigned char *in;
 213         unsigned int avail;
 214
 215         while (len) {
 216                 in = use_pack(p, w_curs, offset, &avail);
 217                 if (avail > len)
 218                         avail = (unsigned int)len;
 219                 sha1write(f, in, avail);
 220                 offset += avail;
 221                 len -= avail;
 222         }
 223 }
 224
 225 static unsigned long write_object(struct sha1file *f,
 226                                   struct object_entry *entry,
 227                                   off_t write_offset)
 228 {
 229         unsigned long size, limit;
 230         void *buf;
 231         unsigned char header[10], dheader[10];
 232         unsigned hdrlen;
 233         off_t datalen;
 234         enum object_type type;
 235         int usable_delta, to_reuse;
 236
 237         if (!pack_to_stdout)
 238                 crc32_begin(f);
 239
 240         type = entry->type;
 241
 242         /* write limit if limited packsize and not first object */
 243         limit = pack_size_limit && nr_written ?
 244                         pack_size_limit - write_offset : 0;
 245
 246         if (!entry->delta)
 247                 usable_delta = 0;       /* no delta */
 248         else if (!pack_size_limit)
 249                usable_delta = 1;        /* unlimited packfile */
 250         else if (entry->delta->idx.offset == (off_t)-1)
 251                 usable_delta = 0;       /* base was written to another pack */
 252         else if (entry->delta->idx.offset)
 253                 usable_delta = 1;       /* base already exists in this pack */
 254         else
 255                 usable_delta = 0;       /* base could end up in another pack */
 256
 257         if (!reuse_object)
 258                 to_reuse = 0;   /* explicit */
 259         else if (!entry->in_pack)
 260                 to_reuse = 0;   /* can't reuse what we don't have */
 261         else if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA)
 262                                 /* check_object() decided it for us ... */
 263                 to_reuse = usable_delta;
 264                                 /* ... but pack split may override that */
 265         else if (type != entry->in_pack_type)
 266                 to_reuse = 0;   /* pack has delta which is unusable */
 267         else if (entry->delta)
 268                 to_reuse = 0;   /* we want to pack afresh */
 269         else
 270                 to_reuse = 1;   /* we have it in-pack undeltified,
 271                                  * and we do not need to deltify it.
 272                                  */
 273
 274         if (!to_reuse) {
 275                 z_stream stream;
 276                 unsigned long maxsize;
 277                 void *out;
 278                 if (!usable_delta) {
 279                         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 280                         if (!buf)
 281                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 282                 } else if (entry->delta_data) {
 283                         size = entry->delta_size;
 284                         buf = entry->delta_data;
 285                         entry->delta_data = NULL;
 286                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 287                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 288                 } else {
 289                         buf = get_delta(entry);
 290                         size = entry->delta_size;
 291                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 292                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 293                 }
 294                 /* compress the data to store and put compressed length in datalen */
 295                 memset(&stream, 0, sizeof(stream));
 296                 deflateInit(&stream, pack_compression_level);
 297                 maxsize = deflateBound(&stream, size);
 298                 out = xmalloc(maxsize);
 299                 /* Compress it */
 300                 stream.next_in = buf;
 301                 stream.avail_in = size;
 302                 stream.next_out = out;
 303                 stream.avail_out = maxsize;
 304                 while (deflate(&stream, Z_FINISH) == Z_OK)
 305                         /* nothing */;
 306                 deflateEnd(&stream);
 307                 datalen = stream.total_out;
 308
 309                 /*
 310                  * The object header is a byte of 'type' followed by zero or
 311                  * more bytes of length.
 312                  */
 313                 hdrlen = encode_header(type, size, header);
 314
 315                 if (type == OBJ_OFS_DELTA) {
 316                         /*
 317                          * Deltas with relative base contain an additional
 318                          * encoding of the relative offset for the delta
 319                          * base from this object's position in the pack.
 320                          */
 321                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 322                         unsigned pos = sizeof(dheader) - 1;
 323                         dheader[pos] = ofs & 127;
 324                         while (ofs >>= 7)
 325                                 dheader[--pos] = 128 | (--ofs & 127);
 326                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 327                                 free(out);
 328                                 free(buf);
 329                                 return 0;
 330                         }
 331                         sha1write(f, header, hdrlen);
 332                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 333                         hdrlen += sizeof(dheader) - pos;
 334                 } else if (type == OBJ_REF_DELTA) {
 335                         /*
 336                          * Deltas with a base reference contain
 337                          * an additional 20 bytes for the base sha1.
 338                          */
 339                         if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 340                                 free(out);
 341                                 free(buf);
 342                                 return 0;
 343                         }
 344                         sha1write(f, header, hdrlen);
 345                         sha1write(f, entry->delta->idx.sha1, 20);
 346                         hdrlen += 20;
 347                 } else {
 348                         if (limit && hdrlen + datalen + 20 >= limit) {
 349                                 free(out);
 350                                 free(buf);
 351                                 return 0;
 352                         }
 353                         sha1write(f, header, hdrlen);
 354                 }
 355                 sha1write(f, out, datalen);
 356                 free(out);
 357                 free(buf);
 358         }
 359         else {
 360                 struct packed_git *p = entry->in_pack;
 361                 struct pack_window *w_curs = NULL;
 362                 struct revindex_entry *revidx;
 363                 off_t offset;
 364
 365                 if (entry->delta) {
 366                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 367                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 368                         reused_delta++;
 369                 }
 370                 hdrlen = encode_header(type, entry->size, header);
 371                 offset = entry->in_pack_offset;
 372                 revidx = find_pack_revindex(p, offset);
 373                 datalen = revidx[1].offset - offset;
 374                 if (!pack_to_stdout && p->index_version > 1 &&
 375                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
 376                         die("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 377                 offset += entry->in_pack_header_size;
 378                 datalen -= entry->in_pack_header_size;
 379                 if (type == OBJ_OFS_DELTA) {
 380                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 381                         unsigned pos = sizeof(dheader) - 1;
 382                         dheader[pos] = ofs & 127;
 383                         while (ofs >>= 7)
 384                                 dheader[--pos] = 128 | (--ofs & 127);
 385                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit)
 386                                 return 0;
 387                         sha1write(f, header, hdrlen);
 388                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 389                         hdrlen += sizeof(dheader) - pos;
 390                 } else if (type == OBJ_REF_DELTA) {
 391                         if (limit && hdrlen + 20 + datalen + 20 >= limit)
 392                                 return 0;
 393                         sha1write(f, header, hdrlen);
 394                         sha1write(f, entry->delta->idx.sha1, 20);
 395                         hdrlen += 20;
 396                 } else {
 397                         if (limit && hdrlen + datalen + 20 >= limit)
 398                                 return 0;
 399                         sha1write(f, header, hdrlen);
 400                 }
 401
 402                 if (!pack_to_stdout && p->index_version == 1 &&
 403                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
 404                         die("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 405                 copy_pack_data(f, p, &w_curs, offset, datalen);
 406                 unuse_pack(&w_curs);
 407                 reused++;
 408         }
 409         if (usable_delta)
 410                 written_delta++;
 411         written++;
 412         if (!pack_to_stdout)
 413                 entry->idx.crc32 = crc32_end(f);
 414         return hdrlen + datalen;
 415 }
 416
 417 static off_t write_one(struct sha1file *f,
 418                                struct object_entry *e,
 419                                off_t offset)
 420 {
 421         unsigned long size;
 422
 423         /* offset is non zero if object is written already. */
 424         if (e->idx.offset || e->preferred_base)
 425                 return offset;
 426
 427         /* if we are deltified, write out base object first. */
 428         if (e->delta) {
 429                 offset = write_one(f, e->delta, offset);
 430                 if (!offset)
 431                         return 0;
 432         }
 433
 434         e->idx.offset = offset;
 435         size = write_object(f, e, offset);
 436         if (!size) {
 437                 e->idx.offset = 0;
 438                 return 0;
 439         }
 440         written_list[nr_written++] = &e->idx;
 441
 442         /* make sure off_t is sufficiently large not to wrap */
 443         if (offset > offset + size)
 444                 die("pack too large for current definition of off_t");
 445         return offset + size;
 446 }
 447
 448 /* forward declaration for write_pack_file */
 449 static int adjust_perm(const char *path, mode_t mode);
 450
 451 static void write_pack_file(void)
 452 {
 453         uint32_t i = 0, j;
 454         struct sha1file *f;
 455         off_t offset, offset_one, last_obj_offset = 0;
 456         struct pack_header hdr;
 457         uint32_t nr_remaining = nr_result;
 458         time_t last_mtime = 0;
 459
 460         if (progress > pack_to_stdout)
 461                 progress_state = start_progress("Writing objects", nr_result);
 462         written_list = xmalloc(nr_objects * sizeof(*written_list));
 463
 464         do {
 465                 unsigned char sha1[20];
 466                 char *pack_tmp_name = NULL;
 467
 468                 if (pack_to_stdout) {
 469                         f = sha1fd_throughput(1, "<stdout>", progress_state);
 470                 } else {
 471                         char tmpname[PATH_MAX];
 472                         int fd;
 473                         snprintf(tmpname, sizeof(tmpname),
 474                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 475                         fd = xmkstemp(tmpname);
 476                         pack_tmp_name = xstrdup(tmpname);
 477                         f = sha1fd(fd, pack_tmp_name);
 478                 }
 479
 480                 hdr.hdr_signature = htonl(PACK_SIGNATURE);
 481                 hdr.hdr_version = htonl(PACK_VERSION);
 482                 hdr.hdr_entries = htonl(nr_remaining);
 483                 sha1write(f, &hdr, sizeof(hdr));
 484                 offset = sizeof(hdr);
 485                 nr_written = 0;
 486                 for (; i < nr_objects; i++) {
 487                         last_obj_offset = offset;
 488                         offset_one = write_one(f, objects + i, offset);
 489                         if (!offset_one)
 490                                 break;
 491                         offset = offset_one;
 492                         display_progress(progress_state, written);
 493                 }
 494
 495                 /*
 496                  * Did we write the wrong # entries in the header?
 497                  * If so, rewrite it like in fast-import
 498                  */
 499                 if (pack_to_stdout || nr_written == nr_remaining) {
 500                         sha1close(f, sha1, 1);
 501                 } else {
 502                         int fd = sha1close(f, NULL, 0);
 503                         fixup_pack_header_footer(fd, sha1, pack_tmp_name, nr_written);
 504                         close(fd);
 505                 }
 506
 507                 if (!pack_to_stdout) {
 508                         mode_t mode = umask(0);
 509                         struct stat st;
 510                         char *idx_tmp_name, tmpname[PATH_MAX];
 511
 512                         umask(mode);
 513                         mode = 0444 & ~mode;
 514
 515                         idx_tmp_name = write_idx_file(NULL, written_list,
 516                                                       nr_written, sha1);
 517
 518                         snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
 519                                  base_name, sha1_to_hex(sha1));
 520                         if (adjust_perm(pack_tmp_name, mode))
 521                                 die("unable to make temporary pack file readable: %s",
 522                                     strerror(errno));
 523                         if (rename(pack_tmp_name, tmpname))
 524                                 die("unable to rename temporary pack file: %s",
 525                                     strerror(errno));
 526
 527                         /*
 528                          * Packs are runtime accessed in their mtime
 529                          * order since newer packs are more likely to contain
 530                          * younger objects.  So if we are creating multiple
 531                          * packs then we should modify the mtime of later ones
 532                          * to preserve this property.
 533                          */
 534                         if (stat(tmpname, &st) < 0) {
 535                                 warning("failed to stat %s: %s",
 536                                         tmpname, strerror(errno));
 537                         } else if (!last_mtime) {
 538                                 last_mtime = st.st_mtime;
 539                         } else {
 540                                 struct utimbuf utb;
 541                                 utb.actime = st.st_atime;
 542                                 utb.modtime = --last_mtime;
 543                                 if (utime(tmpname, &utb) < 0)
 544                                         warning("failed utime() on %s: %s",
 545                                                 tmpname, strerror(errno));
 546                         }
 547
 548                         snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
 549                                  base_name, sha1_to_hex(sha1));
 550                         if (adjust_perm(idx_tmp_name, mode))
 551                                 die("unable to make temporary index file readable: %s",
 552                                     strerror(errno));
 553                         if (rename(idx_tmp_name, tmpname))
 554                                 die("unable to rename temporary index file: %s",
 555                                     strerror(errno));
 556
 557                         free(idx_tmp_name);
 558                         free(pack_tmp_name);
 559                         puts(sha1_to_hex(sha1));
 560                 }
 561
 562                 /* mark written objects as written to previous pack */
 563                 for (j = 0; j < nr_written; j++) {
 564                         written_list[j]->offset = (off_t)-1;
 565                 }
 566                 nr_remaining -= nr_written;
 567         } while (nr_remaining && i < nr_objects);
 568
 569         free(written_list);
 570         stop_progress(&progress_state);
 571         if (written != nr_result)
 572                 die("wrote %u objects while expecting %u", written, nr_result);
 573         /*
 574          * We have scanned through [0 ... i).  Since we have written
 575          * the correct number of objects,  the remaining [i ... nr_objects)
 576          * items must be either already written (due to out-of-order delta base)
 577          * or a preferred base.  Count those which are neither and complain if any.
 578          */
 579         for (j = 0; i < nr_objects; i++) {
 580                 struct object_entry *e = objects + i;
 581                 j += !e->idx.offset && !e->preferred_base;
 582         }
 583         if (j)
 584                 die("wrote %u objects as expected but %u unwritten", written, j);
 585 }
 586
 587 static int locate_object_entry_hash(const unsigned char *sha1)
 588 {
 589         int i;
 590         unsigned int ui;
 591         memcpy(&ui, sha1, sizeof(unsigned int));
 592         i = ui % object_ix_hashsz;
 593         while (0 < object_ix[i]) {
 594                 if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1))
 595                         return i;
 596                 if (++i == object_ix_hashsz)
 597                         i = 0;
 598         }
 599         return -1 - i;
 600 }
 601
 602 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 603 {
 604         int i;
 605
 606         if (!object_ix_hashsz)
 607                 return NULL;
 608
 609         i = locate_object_entry_hash(sha1);
 610         if (0 <= i)
 611                 return &objects[object_ix[i]-1];
 612         return NULL;
 613 }
 614
 615 static void rehash_objects(void)
 616 {
 617         uint32_t i;
 618         struct object_entry *oe;
 619
 620         object_ix_hashsz = nr_objects * 3;
 621         if (object_ix_hashsz < 1024)
 622                 object_ix_hashsz = 1024;
 623         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 624         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 625         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 626                 int ix = locate_object_entry_hash(oe->idx.sha1);
 627                 if (0 <= ix)
 628                         continue;
 629                 ix = -1 - ix;
 630                 object_ix[ix] = i + 1;
 631         }
 632 }
 633
 634 static unsigned name_hash(const char *name)
 635 {
 636         unsigned char c;
 637         unsigned hash = 0;
 638
 639         if (!name)
 640                 return 0;
 641
 642         /*
 643          * This effectively just creates a sortable number from the
 644          * last sixteen non-whitespace characters. Last characters
 645          * count "most", so things that end in ".c" sort together.
 646          */
 647         while ((c = *name++) != 0) {
 648                 if (isspace(c))
 649                         continue;
 650                 hash = (hash >> 2) + (c << 24);
 651         }
 652         return hash;
 653 }
 654
 655 static void setup_delta_attr_check(struct git_attr_check *check)
 656 {
 657         static struct git_attr *attr_delta;
 658
 659         if (!attr_delta)
 660                 attr_delta = git_attr("delta", 5);
 661
 662         check[0].attr = attr_delta;
 663 }
 664
 665 static int no_try_delta(const char *path)
 666 {
 667         struct git_attr_check check[1];
 668
 669         setup_delta_attr_check(check);
 670         if (git_checkattr(path, ARRAY_SIZE(check), check))
 671                 return 0;
 672         if (ATTR_FALSE(check->value))
 673                 return 1;
 674         return 0;
 675 }
 676
 677 static int add_object_entry(const unsigned char *sha1, enum object_type type,
 678                             const char *name, int exclude)
 679 {
 680         struct object_entry *entry;
 681         struct packed_git *p, *found_pack = NULL;
 682         off_t found_offset = 0;
 683         int ix;
 684         unsigned hash = name_hash(name);
 685
 686         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 687         if (ix >= 0) {
 688                 if (exclude) {
 689                         entry = objects + object_ix[ix] - 1;
 690                         if (!entry->preferred_base)
 691                                 nr_result--;
 692                         entry->preferred_base = 1;
 693                 }
 694                 return 0;
 695         }
 696
 697         for (p = packed_git; p; p = p->next) {
 698                 off_t offset = find_pack_entry_one(sha1, p);
 699                 if (offset) {
 700                         if (!found_pack) {
 701                                 found_offset = offset;
 702                                 found_pack = p;
 703                         }
 704                         if (exclude)
 705                                 break;
 706                         if (incremental)
 707                                 return 0;
 708                         if (local && !p->pack_local)
 709                                 return 0;
 710                 }
 711         }
 712
 713         if (nr_objects >= nr_alloc) {
 714                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 715                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 716         }
 717
 718         entry = objects + nr_objects++;
 719         memset(entry, 0, sizeof(*entry));
 720         hashcpy(entry->idx.sha1, sha1);
 721         entry->hash = hash;
 722         if (type)
 723                 entry->type = type;
 724         if (exclude)
 725                 entry->preferred_base = 1;
 726         else
 727                 nr_result++;
 728         if (found_pack) {
 729                 entry->in_pack = found_pack;
 730                 entry->in_pack_offset = found_offset;
 731         }
 732
 733         if (object_ix_hashsz * 3 <= nr_objects * 4)
 734                 rehash_objects();
 735         else
 736                 object_ix[-1 - ix] = nr_objects;
 737
 738         display_progress(progress_state, nr_objects);
 739
 740         if (name && no_try_delta(name))
 741                 entry->no_try_delta = 1;
 742
 743         return 1;
 744 }
 745
 746 struct pbase_tree_cache {
 747         unsigned char sha1[20];
 748         int ref;
 749         int temporary;
 750         void *tree_data;
 751         unsigned long tree_size;
 752 };
 753
 754 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 755 static int pbase_tree_cache_ix(const unsigned char *sha1)
 756 {
 757         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 758 }
 759 static int pbase_tree_cache_ix_incr(int ix)
 760 {
 761         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 762 }
 763
 764 static struct pbase_tree {
 765         struct pbase_tree *next;
 766         /* This is a phony "cache" entry; we are not
 767          * going to evict it nor find it through _get()
 768          * mechanism -- this is for the toplevel node that
 769          * would almost always change with any commit.
 770          */
 771         struct pbase_tree_cache pcache;
 772 } *pbase_tree;
 773
 774 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 775 {
 776         struct pbase_tree_cache *ent, *nent;
 777         void *data;
 778         unsigned long size;
 779         enum object_type type;
 780         int neigh;
 781         int my_ix = pbase_tree_cache_ix(sha1);
 782         int available_ix = -1;
 783
 784         /* pbase-tree-cache acts as a limited hashtable.
 785          * your object will be found at your index or within a few
 786          * slots after that slot if it is cached.
 787          */
 788         for (neigh = 0; neigh < 8; neigh++) {
 789                 ent = pbase_tree_cache[my_ix];
 790                 if (ent && !hashcmp(ent->sha1, sha1)) {
 791                         ent->ref++;
 792                         return ent;
 793                 }
 794                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 795                          ((0 <= available_ix) &&
 796                           (!ent && pbase_tree_cache[available_ix])))
 797                         available_ix = my_ix;
 798                 if (!ent)
 799                         break;
 800                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 801         }
 802
 803         /* Did not find one.  Either we got a bogus request or
 804          * we need to read and perhaps cache.
 805          */
 806         data = read_sha1_file(sha1, &type, &size);
 807         if (!data)
 808                 return NULL;
 809         if (type != OBJ_TREE) {
 810                 free(data);
 811                 return NULL;
 812         }
 813
 814         /* We need to either cache or return a throwaway copy */
 815
 816         if (available_ix < 0)
 817                 ent = NULL;
 818         else {
 819                 ent = pbase_tree_cache[available_ix];
 820                 my_ix = available_ix;
 821         }
 822
 823         if (!ent) {
 824                 nent = xmalloc(sizeof(*nent));
 825                 nent->temporary = (available_ix < 0);
 826         }
 827         else {
 828                 /* evict and reuse */
 829                 free(ent->tree_data);
 830                 nent = ent;
 831         }
 832         hashcpy(nent->sha1, sha1);
 833         nent->tree_data = data;
 834         nent->tree_size = size;
 835         nent->ref = 1;
 836         if (!nent->temporary)
 837                 pbase_tree_cache[my_ix] = nent;
 838         return nent;
 839 }
 840
 841 static void pbase_tree_put(struct pbase_tree_cache *cache)
 842 {
 843         if (!cache->temporary) {
 844                 cache->ref--;
 845                 return;
 846         }
 847         free(cache->tree_data);
 848         free(cache);
 849 }
 850
 851 static int name_cmp_len(const char *name)
 852 {
 853         int i;
 854         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 855                 ;
 856         return i;
 857 }
 858
 859 static void add_pbase_object(struct tree_desc *tree,
 860                              const char *name,
 861                              int cmplen,
 862                              const char *fullname)
 863 {
 864         struct name_entry entry;
 865         int cmp;
 866
 867         while (tree_entry(tree,&entry)) {
 868                 if (S_ISGITLINK(entry.mode))
 869                         continue;
 870                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 871                       memcmp(name, entry.path, cmplen);
 872                 if (cmp > 0)
 873                         continue;
 874                 if (cmp < 0)
 875                         return;
 876                 if (name[cmplen] != '/') {
 877                         add_object_entry(entry.sha1,
 878                                          object_type(entry.mode),
 879                                          fullname, 1);
 880                         return;
 881                 }
 882                 if (S_ISDIR(entry.mode)) {
 883                         struct tree_desc sub;
 884                         struct pbase_tree_cache *tree;
 885                         const char *down = name+cmplen+1;
 886                         int downlen = name_cmp_len(down);
 887
 888                         tree = pbase_tree_get(entry.sha1);
 889                         if (!tree)
 890                                 return;
 891                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
 892
 893                         add_pbase_object(&sub, down, downlen, fullname);
 894                         pbase_tree_put(tree);
 895                 }
 896         }
 897 }
 898
 899 static unsigned *done_pbase_paths;
 900 static int done_pbase_paths_num;
 901 static int done_pbase_paths_alloc;
 902 static int done_pbase_path_pos(unsigned hash)
 903 {
 904         int lo = 0;
 905         int hi = done_pbase_paths_num;
 906         while (lo < hi) {
 907                 int mi = (hi + lo) / 2;
 908                 if (done_pbase_paths[mi] == hash)
 909                         return mi;
 910                 if (done_pbase_paths[mi] < hash)
 911                         hi = mi;
 912                 else
 913                         lo = mi + 1;
 914         }
 915         return -lo-1;
 916 }
 917
 918 static int check_pbase_path(unsigned hash)
 919 {
 920         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
 921         if (0 <= pos)
 922                 return 1;
 923         pos = -pos - 1;
 924         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
 925                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
 926                 done_pbase_paths = xrealloc(done_pbase_paths,
 927                                             done_pbase_paths_alloc *
 928                                             sizeof(unsigned));
 929         }
 930         done_pbase_paths_num++;
 931         if (pos < done_pbase_paths_num)
 932                 memmove(done_pbase_paths + pos + 1,
 933                         done_pbase_paths + pos,
 934                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
 935         done_pbase_paths[pos] = hash;
 936         return 0;
 937 }
 938
 939 static void add_preferred_base_object(const char *name)
 940 {
 941         struct pbase_tree *it;
 942         int cmplen;
 943         unsigned hash = name_hash(name);
 944
 945         if (!num_preferred_base || check_pbase_path(hash))
 946                 return;
 947
 948         cmplen = name_cmp_len(name);
 949         for (it = pbase_tree; it; it = it->next) {
 950                 if (cmplen == 0) {
 951                         add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
 952                 }
 953                 else {
 954                         struct tree_desc tree;
 955                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
 956                         add_pbase_object(&tree, name, cmplen, name);
 957                 }
 958         }
 959 }
 960
 961 static void add_preferred_base(unsigned char *sha1)
 962 {
 963         struct pbase_tree *it;
 964         void *data;
 965         unsigned long size;
 966         unsigned char tree_sha1[20];
 967
 968         if (window <= num_preferred_base++)
 969                 return;
 970
 971         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
 972         if (!data)
 973                 return;
 974
 975         for (it = pbase_tree; it; it = it->next) {
 976                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
 977                         free(data);
 978                         return;
 979                 }
 980         }
 981
 982         it = xcalloc(1, sizeof(*it));
 983         it->next = pbase_tree;
 984         pbase_tree = it;
 985
 986         hashcpy(it->pcache.sha1, tree_sha1);
 987         it->pcache.tree_data = data;
 988         it->pcache.tree_size = size;
 989 }
 990
 991 static void check_object(struct object_entry *entry)
 992 {
 993         if (entry->in_pack) {
 994                 struct packed_git *p = entry->in_pack;
 995                 struct pack_window *w_curs = NULL;
 996                 const unsigned char *base_ref = NULL;
 997                 struct object_entry *base_entry;
 998                 unsigned long used, used_0;
 999                 unsigned int avail;
1000                 off_t ofs;
1001                 unsigned char *buf, c;
1002
1003                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1004
1005                 /*
1006                  * We want in_pack_type even if we do not reuse delta
1007                  * since non-delta representations could still be reused.
1008                  */
1009                 used = unpack_object_header_gently(buf, avail,
1010                                                    &entry->in_pack_type,
1011                                                    &entry->size);
1012
1013                 /*
1014                  * Determine if this is a delta and if so whether we can
1015                  * reuse it or not.  Otherwise let's find out as cheaply as
1016                  * possible what the actual type and size for this object is.
1017                  */
1018                 switch (entry->in_pack_type) {
1019                 default:
1020                         /* Not a delta hence we've already got all we need. */
1021                         entry->type = entry->in_pack_type;
1022                         entry->in_pack_header_size = used;
1023                         unuse_pack(&w_curs);
1024                         return;
1025                 case OBJ_REF_DELTA:
1026                         if (reuse_delta && !entry->preferred_base)
1027                                 base_ref = use_pack(p, &w_curs,
1028                                                 entry->in_pack_offset + used, NULL);
1029                         entry->in_pack_header_size = used + 20;
1030                         break;
1031                 case OBJ_OFS_DELTA:
1032                         buf = use_pack(p, &w_curs,
1033                                        entry->in_pack_offset + used, NULL);
1034                         used_0 = 0;
1035                         c = buf[used_0++];
1036                         ofs = c & 127;
1037                         while (c & 128) {
1038                                 ofs += 1;
1039                                 if (!ofs || MSB(ofs, 7))
1040                                         die("delta base offset overflow in pack for %s",
1041                                             sha1_to_hex(entry->idx.sha1));
1042                                 c = buf[used_0++];
1043                                 ofs = (ofs << 7) + (c & 127);
1044                         }
1045                         if (ofs >= entry->in_pack_offset)
1046                                 die("delta base offset out of bound for %s",
1047                                     sha1_to_hex(entry->idx.sha1));
1048                         ofs = entry->in_pack_offset - ofs;
1049                         if (reuse_delta && !entry->preferred_base) {
1050                                 struct revindex_entry *revidx;
1051                                 revidx = find_pack_revindex(p, ofs);
1052                                 base_ref = nth_packed_object_sha1(p, revidx->nr);
1053                         }
1054                         entry->in_pack_header_size = used + used_0;
1055                         break;
1056                 }
1057
1058                 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1059                         /*
1060                          * If base_ref was set above that means we wish to
1061                          * reuse delta data, and we even found that base
1062                          * in the list of objects we want to pack. Goodie!
1063                          *
1064                          * Depth value does not matter - find_deltas() will
1065                          * never consider reused delta as the base object to
1066                          * deltify other objects against, in order to avoid
1067                          * circular deltas.
1068                          */
1069                         entry->type = entry->in_pack_type;
1070                         entry->delta = base_entry;
1071                         entry->delta_sibling = base_entry->delta_child;
1072                         base_entry->delta_child = entry;
1073                         unuse_pack(&w_curs);
1074                         return;
1075                 }
1076
1077                 if (entry->type) {
1078                         /*
1079                          * This must be a delta and we already know what the
1080                          * final object type is.  Let's extract the actual
1081                          * object size from the delta header.
1082                          */
1083                         entry->size = get_size_from_delta(p, &w_curs,
1084                                         entry->in_pack_offset + entry->in_pack_header_size);
1085                         unuse_pack(&w_curs);
1086                         return;
1087                 }
1088
1089                 /*
1090                  * No choice but to fall back to the recursive delta walk
1091                  * with sha1_object_info() to find about the object type
1092                  * at this point...
1093                  */
1094                 unuse_pack(&w_curs);
1095         }
1096
1097         entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1098         if (entry->type < 0)
1099                 die("unable to get type of object %s",
1100                     sha1_to_hex(entry->idx.sha1));
1101 }
1102
1103 static int pack_offset_sort(const void *_a, const void *_b)
1104 {
1105         const struct object_entry *a = *(struct object_entry **)_a;
1106         const struct object_entry *b = *(struct object_entry **)_b;
1107
1108         /* avoid filesystem trashing with loose objects */
1109         if (!a->in_pack && !b->in_pack)
1110                 return hashcmp(a->idx.sha1, b->idx.sha1);
1111
1112         if (a->in_pack < b->in_pack)
1113                 return -1;
1114         if (a->in_pack > b->in_pack)
1115                 return 1;
1116         return a->in_pack_offset < b->in_pack_offset ? -1 :
1117                         (a->in_pack_offset > b->in_pack_offset);
1118 }
1119
1120 static void get_object_details(void)
1121 {
1122         uint32_t i;
1123         struct object_entry **sorted_by_offset;
1124
1125         sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1126         for (i = 0; i < nr_objects; i++)
1127                 sorted_by_offset[i] = objects + i;
1128         qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1129
1130         init_pack_revindex();
1131
1132         for (i = 0; i < nr_objects; i++)
1133                 check_object(sorted_by_offset[i]);
1134
1135         free(sorted_by_offset);
1136 }
1137
1138 /*
1139  * We search for deltas in a list sorted by type, by filename hash, and then
1140  * by size, so that we see progressively smaller and smaller files.
1141  * That's because we prefer deltas to be from the bigger file
1142  * to the smaller -- deletes are potentially cheaper, but perhaps
1143  * more importantly, the bigger file is likely the more recent
1144  * one.  The deepest deltas are therefore the oldest objects which are
1145  * less susceptible to be accessed often.
1146  */
1147 static int type_size_sort(const void *_a, const void *_b)
1148 {
1149         const struct object_entry *a = *(struct object_entry **)_a;
1150         const struct object_entry *b = *(struct object_entry **)_b;
1151
1152         if (a->type > b->type)
1153                 return -1;
1154         if (a->type < b->type)
1155                 return 1;
1156         if (a->hash > b->hash)
1157                 return -1;
1158         if (a->hash < b->hash)
1159                 return 1;
1160         if (a->preferred_base > b->preferred_base)
1161                 return -1;
1162         if (a->preferred_base < b->preferred_base)
1163                 return 1;
1164         if (a->size > b->size)
1165                 return -1;
1166         if (a->size < b->size)
1167                 return 1;
1168         return a < b ? -1 : (a > b);  /* newest first */
1169 }
1170
1171 struct unpacked {
1172         struct object_entry *entry;
1173         void *data;
1174         struct delta_index *index;
1175         unsigned depth;
1176 };
1177
1178 static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1179                            unsigned long delta_size)
1180 {
1181         if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1182                 return 0;
1183
1184         if (delta_size < cache_max_small_delta_size)
1185                 return 1;
1186
1187         /* cache delta, if objects are large enough compared to delta size */
1188         if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1189                 return 1;
1190
1191         return 0;
1192 }
1193
1194 #ifdef THREADED_DELTA_SEARCH
1195
1196 static pthread_mutex_t read_mutex = PTHREAD_MUTEX_INITIALIZER;
1197 #define read_lock()             pthread_mutex_lock(&read_mutex)
1198 #define read_unlock()           pthread_mutex_unlock(&read_mutex)
1199
1200 static pthread_mutex_t cache_mutex = PTHREAD_MUTEX_INITIALIZER;
1201 #define cache_lock()            pthread_mutex_lock(&cache_mutex)
1202 #define cache_unlock()          pthread_mutex_unlock(&cache_mutex)
1203
1204 static pthread_mutex_t progress_mutex = PTHREAD_MUTEX_INITIALIZER;
1205 #define progress_lock()         pthread_mutex_lock(&progress_mutex)
1206 #define progress_unlock()       pthread_mutex_unlock(&progress_mutex)
1207
1208 #else
1209
1210 #define read_lock()             (void)0
1211 #define read_unlock()           (void)0
1212 #define cache_lock()            (void)0
1213 #define cache_unlock()          (void)0
1214 #define progress_lock()         (void)0
1215 #define progress_unlock()       (void)0
1216
1217 #endif
1218
1219 static int try_delta(struct unpacked *trg, struct unpacked *src,
1220                      unsigned max_depth, unsigned long *mem_usage)
1221 {
1222         struct object_entry *trg_entry = trg->entry;
1223         struct object_entry *src_entry = src->entry;
1224         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1225         unsigned ref_depth;
1226         enum object_type type;
1227         void *delta_buf;
1228
1229         /* Don't bother doing diffs between different types */
1230         if (trg_entry->type != src_entry->type)
1231                 return -1;
1232
1233         /*
1234          * We do not bother to try a delta that we discarded
1235          * on an earlier try, but only when reusing delta data.
1236          */
1237         if (reuse_delta && trg_entry->in_pack &&
1238             trg_entry->in_pack == src_entry->in_pack &&
1239             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1240             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1241                 return 0;
1242
1243         /* Let's not bust the allowed depth. */
1244         if (src->depth >= max_depth)
1245                 return 0;
1246
1247         /* Now some size filtering heuristics. */
1248         trg_size = trg_entry->size;
1249         if (!trg_entry->delta) {
1250                 max_size = trg_size/2 - 20;
1251                 ref_depth = 1;
1252         } else {
1253                 max_size = trg_entry->delta_size;
1254                 ref_depth = trg->depth;
1255         }
1256         max_size = max_size * (max_depth - src->depth) /
1257                                                 (max_depth - ref_depth + 1);
1258         if (max_size == 0)
1259                 return 0;
1260         src_size = src_entry->size;
1261         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1262         if (sizediff >= max_size)
1263                 return 0;
1264         if (trg_size < src_size / 32)
1265                 return 0;
1266
1267         /* Load data if not already done */
1268         if (!trg->data) {
1269                 read_lock();
1270                 trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1271                 read_unlock();
1272                 if (!trg->data)
1273                         die("object %s cannot be read",
1274                             sha1_to_hex(trg_entry->idx.sha1));
1275                 if (sz != trg_size)
1276                         die("object %s inconsistent object length (%lu vs %lu)",
1277                             sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1278                 *mem_usage += sz;
1279         }
1280         if (!src->data) {
1281                 read_lock();
1282                 src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1283                 read_unlock();
1284                 if (!src->data)
1285                         die("object %s cannot be read",
1286                             sha1_to_hex(src_entry->idx.sha1));
1287                 if (sz != src_size)
1288                         die("object %s inconsistent object length (%lu vs %lu)",
1289                             sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1290                 *mem_usage += sz;
1291         }
1292         if (!src->index) {
1293                 src->index = create_delta_index(src->data, src_size);
1294                 if (!src->index) {
1295                         static int warned = 0;
1296                         if (!warned++)
1297                                 warning("suboptimal pack - out of memory");
1298                         return 0;
1299                 }
1300                 *mem_usage += sizeof_delta_index(src->index);
1301         }
1302
1303         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1304         if (!delta_buf)
1305                 return 0;
1306
1307         if (trg_entry->delta) {
1308                 /* Prefer only shallower same-sized deltas. */
1309                 if (delta_size == trg_entry->delta_size &&
1310                     src->depth + 1 >= trg->depth) {
1311                         free(delta_buf);
1312                         return 0;
1313                 }
1314         }
1315
1316         /*
1317          * Handle memory allocation outside of the cache
1318          * accounting lock.  Compiler will optimize the strangeness
1319          * away when THREADED_DELTA_SEARCH is not defined.
1320          */
1321         free(trg_entry->delta_data);
1322         cache_lock();
1323         if (trg_entry->delta_data) {
1324                 delta_cache_size -= trg_entry->delta_size;
1325                 trg_entry->delta_data = NULL;
1326         }
1327         if (delta_cacheable(src_size, trg_size, delta_size)) {
1328                 delta_cache_size += delta_size;
1329                 cache_unlock();
1330                 trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1331         } else {
1332                 cache_unlock();
1333                 free(delta_buf);
1334         }
1335
1336         trg_entry->delta = src_entry;
1337         trg_entry->delta_size = delta_size;
1338         trg->depth = src->depth + 1;
1339
1340         return 1;
1341 }
1342
1343 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1344 {
1345         struct object_entry *child = me->delta_child;
1346         unsigned int m = n;
1347         while (child) {
1348                 unsigned int c = check_delta_limit(child, n + 1);
1349                 if (m < c)
1350                         m = c;
1351                 child = child->delta_sibling;
1352         }
1353         return m;
1354 }
1355
1356 static unsigned long free_unpacked(struct unpacked *n)
1357 {
1358         unsigned long freed_mem = sizeof_delta_index(n->index);
1359         free_delta_index(n->index);
1360         n->index = NULL;
1361         if (n->data) {
1362                 freed_mem += n->entry->size;
1363                 free(n->data);
1364                 n->data = NULL;
1365         }
1366         n->entry = NULL;
1367         n->depth = 0;
1368         return freed_mem;
1369 }
1370
1371 static void find_deltas(struct object_entry **list, unsigned *list_size,
1372                         int window, int depth, unsigned *processed)
1373 {
1374         uint32_t i, idx = 0, count = 0;
1375         unsigned int array_size = window * sizeof(struct unpacked);
1376         struct unpacked *array;
1377         unsigned long mem_usage = 0;
1378
1379         array = xmalloc(array_size);
1380         memset(array, 0, array_size);
1381
1382         for (;;) {
1383                 struct object_entry *entry = *list++;
1384                 struct unpacked *n = array + idx;
1385                 int j, max_depth, best_base = -1;
1386
1387                 progress_lock();
1388                 if (!*list_size) {
1389                         progress_unlock();
1390                         break;
1391                 }
1392                 (*list_size)--;
1393                 if (!entry->preferred_base) {
1394                         (*processed)++;
1395                         display_progress(progress_state, *processed);
1396                 }
1397                 progress_unlock();
1398
1399                 mem_usage -= free_unpacked(n);
1400                 n->entry = entry;
1401
1402                 while (window_memory_limit &&
1403                        mem_usage > window_memory_limit &&
1404                        count > 1) {
1405                         uint32_t tail = (idx + window - count) % window;
1406                         mem_usage -= free_unpacked(array + tail);
1407                         count--;
1408                 }
1409
1410                 /* We do not compute delta to *create* objects we are not
1411                  * going to pack.
1412                  */
1413                 if (entry->preferred_base)
1414                         goto next;
1415
1416                 /*
1417                  * If the current object is at pack edge, take the depth the
1418                  * objects that depend on the current object into account
1419                  * otherwise they would become too deep.
1420                  */
1421                 max_depth = depth;
1422                 if (entry->delta_child) {
1423                         max_depth -= check_delta_limit(entry, 0);
1424                         if (max_depth <= 0)
1425                                 goto next;
1426                 }
1427
1428                 j = window;
1429                 while (--j > 0) {
1430                         int ret;
1431                         uint32_t other_idx = idx + j;
1432                         struct unpacked *m;
1433                         if (other_idx >= window)
1434                                 other_idx -= window;
1435                         m = array + other_idx;
1436                         if (!m->entry)
1437                                 break;
1438                         ret = try_delta(n, m, max_depth, &mem_usage);
1439                         if (ret < 0)
1440                                 break;
1441                         else if (ret > 0)
1442                                 best_base = other_idx;
1443                 }
1444
1445                 /* if we made n a delta, and if n is already at max
1446                  * depth, leaving it in the window is pointless.  we
1447                  * should evict it first.
1448                  */
1449                 if (entry->delta && depth <= n->depth)
1450                         continue;
1451
1452                 /*
1453                  * Move the best delta base up in the window, after the
1454                  * currently deltified object, to keep it longer.  It will
1455                  * be the first base object to be attempted next.
1456                  */
1457                 if (entry->delta) {
1458                         struct unpacked swap = array[best_base];
1459                         int dist = (window + idx - best_base) % window;
1460                         int dst = best_base;
1461                         while (dist--) {
1462                                 int src = (dst + 1) % window;
1463                                 array[dst] = array[src];
1464                                 dst = src;
1465                         }
1466                         array[dst] = swap;
1467                 }
1468
1469                 next:
1470                 idx++;
1471                 if (count + 1 < window)
1472                         count++;
1473                 if (idx >= window)
1474                         idx = 0;
1475         }
1476
1477         for (i = 0; i < window; ++i) {
1478                 free_delta_index(array[i].index);
1479                 free(array[i].data);
1480         }
1481         free(array);
1482 }
1483
1484 #ifdef THREADED_DELTA_SEARCH
1485
1486 /*
1487  * The main thread waits on the condition that (at least) one of the workers
1488  * has stopped working (which is indicated in the .working member of
1489  * struct thread_params).
1490  * When a work thread has completed its work, it sets .working to 0 and
1491  * signals the main thread and waits on the condition that .data_ready
1492  * becomes 1.
1493  */
1494
1495 struct thread_params {
1496         pthread_t thread;
1497         struct object_entry **list;
1498         unsigned list_size;
1499         unsigned remaining;
1500         int window;
1501         int depth;
1502         int working;
1503         int data_ready;
1504         pthread_mutex_t mutex;
1505         pthread_cond_t cond;
1506         unsigned *processed;
1507 };
1508
1509 static pthread_cond_t progress_cond = PTHREAD_COND_INITIALIZER;
1510
1511 static void *threaded_find_deltas(void *arg)
1512 {
1513         struct thread_params *me = arg;
1514
1515         while (me->remaining) {
1516                 find_deltas(me->list, &me->remaining,
1517                             me->window, me->depth, me->processed);
1518
1519                 progress_lock();
1520                 me->working = 0;
1521                 pthread_cond_signal(&progress_cond);
1522                 progress_unlock();
1523
1524                 /*
1525                  * We must not set ->data_ready before we wait on the
1526                  * condition because the main thread may have set it to 1
1527                  * before we get here. In order to be sure that new
1528                  * work is available if we see 1 in ->data_ready, it
1529                  * was initialized to 0 before this thread was spawned
1530                  * and we reset it to 0 right away.
1531                  */
1532                 pthread_mutex_lock(&me->mutex);
1533                 while (!me->data_ready)
1534                         pthread_cond_wait(&me->cond, &me->mutex);
1535                 me->data_ready = 0;
1536                 pthread_mutex_unlock(&me->mutex);
1537         }
1538         /* leave ->working 1 so that this doesn't get more work assigned */
1539         return NULL;
1540 }
1541
1542 static void ll_find_deltas(struct object_entry **list, unsigned list_size,
1543                            int window, int depth, unsigned *processed)
1544 {
1545         struct thread_params p[delta_search_threads];
1546         int i, ret, active_threads = 0;
1547
1548         if (delta_search_threads <= 1) {
1549                 find_deltas(list, &list_size, window, depth, processed);
1550                 return;
1551         }
1552
1553         /* Partition the work amongst work threads. */
1554         for (i = 0; i < delta_search_threads; i++) {
1555                 unsigned sub_size = list_size / (delta_search_threads - i);
1556
1557                 p[i].window = window;
1558                 p[i].depth = depth;
1559                 p[i].processed = processed;
1560                 p[i].working = 1;
1561                 p[i].data_ready = 0;
1562
1563                 /* try to split chunks on "path" boundaries */
1564                 while (sub_size && sub_size < list_size &&
1565                        list[sub_size]->hash &&
1566                        list[sub_size]->hash == list[sub_size-1]->hash)
1567                         sub_size++;
1568
1569                 p[i].list = list;
1570                 p[i].list_size = sub_size;
1571                 p[i].remaining = sub_size;
1572
1573                 list += sub_size;
1574                 list_size -= sub_size;
1575         }
1576
1577         /* Start work threads. */
1578         for (i = 0; i < delta_search_threads; i++) {
1579                 if (!p[i].list_size)
1580                         continue;
1581                 pthread_mutex_init(&p[i].mutex, NULL);
1582                 pthread_cond_init(&p[i].cond, NULL);
1583                 ret = pthread_create(&p[i].thread, NULL,
1584                                      threaded_find_deltas, &p[i]);
1585                 if (ret)
1586                         die("unable to create thread: %s", strerror(ret));
1587                 active_threads++;
1588         }
1589
1590         /*
1591          * Now let's wait for work completion.  Each time a thread is done
1592          * with its work, we steal half of the remaining work from the
1593          * thread with the largest number of unprocessed objects and give
1594          * it to that newly idle thread.  This ensure good load balancing
1595          * until the remaining object list segments are simply too short
1596          * to be worth splitting anymore.
1597          */
1598         while (active_threads) {
1599                 struct thread_params *target = NULL;
1600                 struct thread_params *victim = NULL;
1601                 unsigned sub_size = 0;
1602
1603                 progress_lock();
1604                 for (;;) {
1605                         for (i = 0; !target && i < delta_search_threads; i++)
1606                                 if (!p[i].working)
1607                                         target = &p[i];
1608                         if (target)
1609                                 break;
1610                         pthread_cond_wait(&progress_cond, &progress_mutex);
1611                 }
1612
1613                 for (i = 0; i < delta_search_threads; i++)
1614                         if (p[i].remaining > 2*window &&
1615                             (!victim || victim->remaining < p[i].remaining))
1616                                 victim = &p[i];
1617                 if (victim) {
1618                         sub_size = victim->remaining / 2;
1619                         list = victim->list + victim->list_size - sub_size;
1620                         while (sub_size && list[0]->hash &&
1621                                list[0]->hash == list[-1]->hash) {
1622                                 list++;
1623                                 sub_size--;
1624                         }
1625                         if (!sub_size) {
1626                                 /*
1627                                  * It is possible for some "paths" to have
1628                                  * so many objects that no hash boundary
1629                                  * might be found.  Let's just steal the
1630                                  * exact half in that case.
1631                                  */
1632                                 sub_size = victim->remaining / 2;
1633                                 list -= sub_size;
1634                         }
1635                         target->list = list;
1636                         victim->list_size -= sub_size;
1637                         victim->remaining -= sub_size;
1638                 }
1639                 target->list_size = sub_size;
1640                 target->remaining = sub_size;
1641                 target->working = 1;
1642                 progress_unlock();
1643
1644                 pthread_mutex_lock(&target->mutex);
1645                 target->data_ready = 1;
1646                 pthread_cond_signal(&target->cond);
1647                 pthread_mutex_unlock(&target->mutex);
1648
1649                 if (!sub_size) {
1650                         pthread_join(target->thread, NULL);
1651                         pthread_cond_destroy(&target->cond);
1652                         pthread_mutex_destroy(&target->mutex);
1653                         active_threads--;
1654                 }
1655         }
1656 }
1657
1658 #else
1659 #define ll_find_deltas(l, s, w, d, p)   find_deltas(l, &s, w, d, p)
1660 #endif
1661
1662 static int add_ref_tag(const char *path, const unsigned char *sha1, int flag, void *cb_data)
1663 {
1664         unsigned char peeled[20];
1665
1666         if (!prefixcmp(path, "refs/tags/") && /* is a tag? */
1667             !peel_ref(path, peeled)        && /* peelable? */
1668             !is_null_sha1(peeled)          && /* annotated tag? */
1669             locate_object_entry(peeled))      /* object packed? */
1670                 add_object_entry(sha1, OBJ_TAG, NULL, 0);
1671         return 0;
1672 }
1673
1674 static void prepare_pack(int window, int depth)
1675 {
1676         struct object_entry **delta_list;
1677         uint32_t i, n, nr_deltas;
1678
1679         get_object_details();
1680
1681         if (!nr_objects || !window || !depth)
1682                 return;
1683
1684         delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1685         nr_deltas = n = 0;
1686
1687         for (i = 0; i < nr_objects; i++) {
1688                 struct object_entry *entry = objects + i;
1689
1690                 if (entry->delta)
1691                         /* This happens if we decided to reuse existing
1692                          * delta from a pack.  "reuse_delta &&" is implied.
1693                          */
1694                         continue;
1695
1696                 if (entry->size < 50)
1697                         continue;
1698
1699                 if (entry->no_try_delta)
1700                         continue;
1701
1702                 if (!entry->preferred_base)
1703                         nr_deltas++;
1704
1705                 delta_list[n++] = entry;
1706         }
1707
1708         if (nr_deltas && n > 1) {
1709                 unsigned nr_done = 0;
1710                 if (progress)
1711                         progress_state = start_progress("Compressing objects",
1712                                                         nr_deltas);
1713                 qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
1714                 ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
1715                 stop_progress(&progress_state);
1716                 if (nr_done != nr_deltas)
1717                         die("inconsistency with delta count");
1718         }
1719         free(delta_list);
1720 }
1721
1722 static int git_pack_config(const char *k, const char *v)
1723 {
1724         if(!strcmp(k, "pack.window")) {
1725                 window = git_config_int(k, v);
1726                 return 0;
1727         }
1728         if (!strcmp(k, "pack.windowmemory")) {
1729                 window_memory_limit = git_config_ulong(k, v);
1730                 return 0;
1731         }
1732         if (!strcmp(k, "pack.depth")) {
1733                 depth = git_config_int(k, v);
1734                 return 0;
1735         }
1736         if (!strcmp(k, "pack.compression")) {
1737                 int level = git_config_int(k, v);
1738                 if (level == -1)
1739                         level = Z_DEFAULT_COMPRESSION;
1740                 else if (level < 0 || level > Z_BEST_COMPRESSION)
1741                         die("bad pack compression level %d", level);
1742                 pack_compression_level = level;
1743                 pack_compression_seen = 1;
1744                 return 0;
1745         }
1746         if (!strcmp(k, "pack.deltacachesize")) {
1747                 max_delta_cache_size = git_config_int(k, v);
1748                 return 0;
1749         }
1750         if (!strcmp(k, "pack.deltacachelimit")) {
1751                 cache_max_small_delta_size = git_config_int(k, v);
1752                 return 0;
1753         }
1754         if (!strcmp(k, "pack.threads")) {
1755                 delta_search_threads = git_config_int(k, v);
1756                 if (delta_search_threads < 0)
1757                         die("invalid number of threads specified (%d)",
1758                             delta_search_threads);
1759 #ifndef THREADED_DELTA_SEARCH
1760                 if (delta_search_threads != 1)
1761                         warning("no threads support, ignoring %s", k);
1762 #endif
1763                 return 0;
1764         }
1765         if (!strcmp(k, "pack.indexversion")) {
1766                 pack_idx_default_version = git_config_int(k, v);
1767                 if (pack_idx_default_version > 2)
1768                         die("bad pack.indexversion=%d", pack_idx_default_version);
1769                 return 0;
1770         }
1771         if (!strcmp(k, "pack.packsizelimit")) {
1772                 pack_size_limit_cfg = git_config_ulong(k, v);
1773                 return 0;
1774         }
1775         return git_default_config(k, v);
1776 }
1777
1778 static void read_object_list_from_stdin(void)
1779 {
1780         char line[40 + 1 + PATH_MAX + 2];
1781         unsigned char sha1[20];
1782
1783         for (;;) {
1784                 if (!fgets(line, sizeof(line), stdin)) {
1785                         if (feof(stdin))
1786                                 break;
1787                         if (!ferror(stdin))
1788                                 die("fgets returned NULL, not EOF, not error!");
1789                         if (errno != EINTR)
1790                                 die("fgets: %s", strerror(errno));
1791                         clearerr(stdin);
1792                         continue;
1793                 }
1794                 if (line[0] == '-') {
1795                         if (get_sha1_hex(line+1, sha1))
1796                                 die("expected edge sha1, got garbage:\n %s",
1797                                     line);
1798                         add_preferred_base(sha1);
1799                         continue;
1800                 }
1801                 if (get_sha1_hex(line, sha1))
1802                         die("expected sha1, got garbage:\n %s", line);
1803
1804                 add_preferred_base_object(line+41);
1805                 add_object_entry(sha1, 0, line+41, 0);
1806         }
1807 }
1808
1809 #define OBJECT_ADDED (1u<<20)
1810
1811 static void show_commit(struct commit *commit)
1812 {
1813         add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
1814         commit->object.flags |= OBJECT_ADDED;
1815 }
1816
1817 static void show_object(struct object_array_entry *p)
1818 {
1819         add_preferred_base_object(p->name);
1820         add_object_entry(p->item->sha1, p->item->type, p->name, 0);
1821         p->item->flags |= OBJECT_ADDED;
1822 }
1823
1824 static void show_edge(struct commit *commit)
1825 {
1826         add_preferred_base(commit->object.sha1);
1827 }
1828
1829 struct in_pack_object {
1830         off_t offset;
1831         struct object *object;
1832 };
1833
1834 struct in_pack {
1835         int alloc;
1836         int nr;
1837         struct in_pack_object *array;
1838 };
1839
1840 static void mark_in_pack_object(struct object *object, struct packed_git *p, struct in_pack *in_pack)
1841 {
1842         in_pack->array[in_pack->nr].offset = find_pack_entry_one(object->sha1, p);
1843         in_pack->array[in_pack->nr].object = object;
1844         in_pack->nr++;
1845 }
1846
1847 /*
1848  * Compare the objects in the offset order, in order to emulate the
1849  * "git-rev-list --objects" output that produced the pack originally.
1850  */
1851 static int ofscmp(const void *a_, const void *b_)
1852 {
1853         struct in_pack_object *a = (struct in_pack_object *)a_;
1854         struct in_pack_object *b = (struct in_pack_object *)b_;
1855
1856         if (a->offset < b->offset)
1857                 return -1;
1858         else if (a->offset > b->offset)
1859                 return 1;
1860         else
1861                 return hashcmp(a->object->sha1, b->object->sha1);
1862 }
1863
1864 static void add_objects_in_unpacked_packs(struct rev_info *revs)
1865 {
1866         struct packed_git *p;
1867         struct in_pack in_pack;
1868         uint32_t i;
1869
1870         memset(&in_pack, 0, sizeof(in_pack));
1871
1872         for (p = packed_git; p; p = p->next) {
1873                 const unsigned char *sha1;
1874                 struct object *o;
1875
1876                 for (i = 0; i < revs->num_ignore_packed; i++) {
1877                         if (matches_pack_name(p, revs->ignore_packed[i]))
1878                                 break;
1879                 }
1880                 if (revs->num_ignore_packed <= i)
1881                         continue;
1882                 if (open_pack_index(p))
1883                         die("cannot open pack index");
1884
1885                 ALLOC_GROW(in_pack.array,
1886                            in_pack.nr + p->num_objects,
1887                            in_pack.alloc);
1888
1889                 for (i = 0; i < p->num_objects; i++) {
1890                         sha1 = nth_packed_object_sha1(p, i);
1891                         o = lookup_unknown_object(sha1);
1892                         if (!(o->flags & OBJECT_ADDED))
1893                                 mark_in_pack_object(o, p, &in_pack);
1894                         o->flags |= OBJECT_ADDED;
1895                 }
1896         }
1897
1898         if (in_pack.nr) {
1899                 qsort(in_pack.array, in_pack.nr, sizeof(in_pack.array[0]),
1900                       ofscmp);
1901                 for (i = 0; i < in_pack.nr; i++) {
1902                         struct object *o = in_pack.array[i].object;
1903                         add_object_entry(o->sha1, o->type, "", 0);
1904                 }
1905         }
1906         free(in_pack.array);
1907 }
1908
1909 static void get_object_list(int ac, const char **av)
1910 {
1911         struct rev_info revs;
1912         char line[1000];
1913         int flags = 0;
1914
1915         init_revisions(&revs, NULL);
1916         save_commit_buffer = 0;
1917         setup_revisions(ac, av, &revs, NULL);
1918
1919         while (fgets(line, sizeof(line), stdin) != NULL) {
1920                 int len = strlen(line);
1921                 if (len && line[len - 1] == '\n')
1922                         line[--len] = 0;
1923                 if (!len)
1924                         break;
1925                 if (*line == '-') {
1926                         if (!strcmp(line, "--not")) {
1927                                 flags ^= UNINTERESTING;
1928                                 continue;
1929                         }
1930                         die("not a rev '%s'", line);
1931                 }
1932                 if (handle_revision_arg(line, &revs, flags, 1))
1933                         die("bad revision '%s'", line);
1934         }
1935
1936         if (prepare_revision_walk(&revs))
1937                 die("revision walk setup failed");
1938         mark_edges_uninteresting(revs.commits, &revs, show_edge);
1939         traverse_commit_list(&revs, show_commit, show_object);
1940
1941         if (keep_unreachable)
1942                 add_objects_in_unpacked_packs(&revs);
1943 }
1944
1945 static int adjust_perm(const char *path, mode_t mode)
1946 {
1947         if (chmod(path, mode))
1948                 return -1;
1949         return adjust_shared_perm(path);
1950 }
1951
1952 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1953 {
1954         int use_internal_rev_list = 0;
1955         int thin = 0;
1956         uint32_t i;
1957         const char **rp_av;
1958         int rp_ac_alloc = 64;
1959         int rp_ac;
1960
1961         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1962
1963         rp_av[0] = "pack-objects";
1964         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1965         rp_ac = 2;
1966
1967         git_config(git_pack_config);
1968         if (!pack_compression_seen && core_compression_seen)
1969                 pack_compression_level = core_compression_level;
1970
1971         progress = isatty(2);
1972         for (i = 1; i < argc; i++) {
1973                 const char *arg = argv[i];
1974
1975                 if (*arg != '-')
1976                         break;
1977
1978                 if (!strcmp("--non-empty", arg)) {
1979                         non_empty = 1;
1980                         continue;
1981                 }
1982                 if (!strcmp("--local", arg)) {
1983                         local = 1;
1984                         continue;
1985                 }
1986                 if (!strcmp("--incremental", arg)) {
1987                         incremental = 1;
1988                         continue;
1989                 }
1990                 if (!prefixcmp(arg, "--compression=")) {
1991                         char *end;
1992                         int level = strtoul(arg+14, &end, 0);
1993                         if (!arg[14] || *end)
1994                                 usage(pack_usage);
1995                         if (level == -1)
1996                                 level = Z_DEFAULT_COMPRESSION;
1997                         else if (level < 0 || level > Z_BEST_COMPRESSION)
1998                                 die("bad pack compression level %d", level);
1999                         pack_compression_level = level;
2000                         continue;
2001                 }
2002                 if (!prefixcmp(arg, "--max-pack-size=")) {
2003                         char *end;
2004                         pack_size_limit_cfg = 0;
2005                         pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024;
2006                         if (!arg[16] || *end)
2007                                 usage(pack_usage);
2008                         continue;
2009                 }
2010                 if (!prefixcmp(arg, "--window=")) {
2011                         char *end;
2012                         window = strtoul(arg+9, &end, 0);
2013                         if (!arg[9] || *end)
2014                                 usage(pack_usage);
2015                         continue;
2016                 }
2017                 if (!prefixcmp(arg, "--window-memory=")) {
2018                         if (!git_parse_ulong(arg+16, &window_memory_limit))
2019                                 usage(pack_usage);
2020                         continue;
2021                 }
2022                 if (!prefixcmp(arg, "--threads=")) {
2023                         char *end;
2024                         delta_search_threads = strtoul(arg+10, &end, 0);
2025                         if (!arg[10] || *end || delta_search_threads < 0)
2026                                 usage(pack_usage);
2027 #ifndef THREADED_DELTA_SEARCH
2028                         if (delta_search_threads != 1)
2029                                 warning("no threads support, "
2030                                         "ignoring %s", arg);
2031 #endif
2032                         continue;
2033                 }
2034                 if (!prefixcmp(arg, "--depth=")) {
2035                         char *end;
2036                         depth = strtoul(arg+8, &end, 0);
2037                         if (!arg[8] || *end)
2038                                 usage(pack_usage);
2039                         continue;
2040                 }
2041                 if (!strcmp("--progress", arg)) {
2042                         progress = 1;
2043                         continue;
2044                 }
2045                 if (!strcmp("--all-progress", arg)) {
2046                         progress = 2;
2047                         continue;
2048                 }
2049                 if (!strcmp("-q", arg)) {
2050                         progress = 0;
2051                         continue;
2052                 }
2053                 if (!strcmp("--no-reuse-delta", arg)) {
2054                         reuse_delta = 0;
2055                         continue;
2056                 }
2057                 if (!strcmp("--no-reuse-object", arg)) {
2058                         reuse_object = reuse_delta = 0;
2059                         continue;
2060                 }
2061                 if (!strcmp("--delta-base-offset", arg)) {
2062                         allow_ofs_delta = 1;
2063                         continue;
2064                 }
2065                 if (!strcmp("--stdout", arg)) {
2066                         pack_to_stdout = 1;
2067                         continue;
2068                 }
2069                 if (!strcmp("--revs", arg)) {
2070                         use_internal_rev_list = 1;
2071                         continue;
2072                 }
2073                 if (!strcmp("--keep-unreachable", arg)) {
2074                         keep_unreachable = 1;
2075                         continue;
2076                 }
2077                 if (!strcmp("--include-tag", arg)) {
2078                         include_tag = 1;
2079                         continue;
2080                 }
2081                 if (!strcmp("--unpacked", arg) ||
2082                     !prefixcmp(arg, "--unpacked=") ||
2083                     !strcmp("--reflog", arg) ||
2084                     !strcmp("--all", arg)) {
2085                         use_internal_rev_list = 1;
2086                         if (rp_ac >= rp_ac_alloc - 1) {
2087                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
2088                                 rp_av = xrealloc(rp_av,
2089                                                  rp_ac_alloc * sizeof(*rp_av));
2090                         }
2091                         rp_av[rp_ac++] = arg;
2092                         continue;
2093                 }
2094                 if (!strcmp("--thin", arg)) {
2095                         use_internal_rev_list = 1;
2096                         thin = 1;
2097                         rp_av[1] = "--objects-edge";
2098                         continue;
2099                 }
2100                 if (!prefixcmp(arg, "--index-version=")) {
2101                         char *c;
2102                         pack_idx_default_version = strtoul(arg + 16, &c, 10);
2103                         if (pack_idx_default_version > 2)
2104                                 die("bad %s", arg);
2105                         if (*c == ',')
2106                                 pack_idx_off32_limit = strtoul(c+1, &c, 0);
2107                         if (*c || pack_idx_off32_limit & 0x80000000)
2108                                 die("bad %s", arg);
2109                         continue;
2110                 }
2111                 usage(pack_usage);
2112         }
2113
2114         /* Traditionally "pack-objects [options] base extra" failed;
2115          * we would however want to take refs parameter that would
2116          * have been given to upstream rev-list ourselves, which means
2117          * we somehow want to say what the base name is.  So the
2118          * syntax would be:
2119          *
2120          * pack-objects [options] base <refs...>
2121          *
2122          * in other words, we would treat the first non-option as the
2123          * base_name and send everything else to the internal revision
2124          * walker.
2125          */
2126
2127         if (!pack_to_stdout)
2128                 base_name = argv[i++];
2129
2130         if (pack_to_stdout != !base_name)
2131                 usage(pack_usage);
2132
2133         if (!pack_to_stdout && !pack_size_limit)
2134                 pack_size_limit = pack_size_limit_cfg;
2135
2136         if (pack_to_stdout && pack_size_limit)
2137                 die("--max-pack-size cannot be used to build a pack for transfer.");
2138
2139         if (!pack_to_stdout && thin)
2140                 die("--thin cannot be used to build an indexable pack.");
2141
2142 #ifdef THREADED_DELTA_SEARCH
2143         if (!delta_search_threads)      /* --threads=0 means autodetect */
2144                 delta_search_threads = online_cpus();
2145 #endif
2146
2147         prepare_packed_git();
2148
2149         if (progress)
2150                 progress_state = start_progress("Counting objects", 0);
2151         if (!use_internal_rev_list)
2152                 read_object_list_from_stdin();
2153         else {
2154                 rp_av[rp_ac] = NULL;
2155                 get_object_list(rp_ac, rp_av);
2156         }
2157         if (include_tag && nr_result)
2158                 for_each_ref(add_ref_tag, NULL);
2159         stop_progress(&progress_state);
2160
2161         if (non_empty && !nr_result)
2162                 return 0;
2163         if (nr_result)
2164                 prepare_pack(window, depth);
2165         write_pack_file();
2166         if (progress)
2167                 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
2168                         written, written_delta, reused, reused_delta);
2169         return 0;
2170 }