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