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] \n\
  32         [--keep-unreachable | --unpack-unreachable] \n\
  33         [<ref-list | <object-list]";
  34
  35 struct object_entry {
  36         struct pack_idx_entry idx;
  37         unsigned long size;     /* uncompressed size */
  38         struct packed_git *in_pack;     /* already in pack */
  39         off_t in_pack_offset;
  40         struct object_entry *delta;     /* delta base object */
  41         struct object_entry *delta_child; /* deltified objects who bases me */
  42         struct object_entry *delta_sibling; /* other deltified objects who
  43                                              * uses the same base as me
  44                                              */
  45         void *delta_data;       /* cached delta (uncompressed) */
  46         unsigned long delta_size;       /* delta data size (uncompressed) */
  47         unsigned long z_delta_size;     /* delta data size (compressed) */
  48         unsigned int hash;      /* name hint hash */
  49         enum object_type type;
  50         enum object_type in_pack_type;  /* could be delta */
  51         unsigned char in_pack_header_size;
  52         unsigned char preferred_base; /* we do not pack this, but is available
  53                                        * to be used as the base object to delta
  54                                        * objects against.
  55                                        */
  56         unsigned char no_try_delta;
  57 };
  58
  59 /*
  60  * Objects we are going to pack are collected in objects array (dynamically
  61  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  62  * in the order we see -- typically rev-list --objects order that gives us
  63  * nice "minimum seek" order.
  64  */
  65 static struct object_entry *objects;
  66 static struct pack_idx_entry **written_list;
  67 static uint32_t nr_objects, nr_alloc, nr_result, nr_written;
  68
  69 static int non_empty;
  70 static int reuse_delta = 1, reuse_object = 1;
  71 static int keep_unreachable, unpack_unreachable, include_tag;
  72 static int local;
  73 static int incremental;
  74 static int allow_ofs_delta;
  75 static const char *base_name;
  76 static int progress = 1;
  77 static int window = 10;
  78 static uint32_t pack_size_limit, pack_size_limit_cfg;
  79 static int depth = 50;
  80 static int delta_search_threads = 1;
  81 static int pack_to_stdout;
  82 static int num_preferred_base;
  83 static struct progress *progress_state;
  84 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  85 static int pack_compression_seen;
  86
  87 static unsigned long delta_cache_size = 0;
  88 static unsigned long max_delta_cache_size = 0;
  89 static unsigned long cache_max_small_delta_size = 1000;
  90
  91 static unsigned long window_memory_limit = 0;
  92
  93 /*
  94  * The object names in objects array are hashed with this hashtable,
  95  * to help looking up the entry by object name.
  96  * This hashtable is built after all the objects are seen.
  97  */
  98 static int *object_ix;
  99 static int object_ix_hashsz;
 100
 101 /*
 102  * stats
 103  */
 104 static uint32_t written, written_delta;
 105 static uint32_t reused, reused_delta;
 106
 107
 108 static void *get_delta(struct object_entry *entry)
 109 {
 110         unsigned long size, base_size, delta_size;
 111         void *buf, *base_buf, *delta_buf;
 112         enum object_type type;
 113
 114         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 115         if (!buf)
 116                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 117         base_buf = read_sha1_file(entry->delta->idx.sha1, &type, &base_size);
 118         if (!base_buf)
 119                 die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 120         delta_buf = diff_delta(base_buf, base_size,
 121                                buf, size, &delta_size, 0);
 122         if (!delta_buf || delta_size != entry->delta_size)
 123                 die("delta size changed");
 124         free(buf);
 125         free(base_buf);
 126         return delta_buf;
 127 }
 128
 129 static unsigned long do_compress(void **pptr, unsigned long size)
 130 {
 131         z_stream stream;
 132         void *in, *out;
 133         unsigned long maxsize;
 134
 135         memset(&stream, 0, sizeof(stream));
 136         deflateInit(&stream, pack_compression_level);
 137         maxsize = deflateBound(&stream, size);
 138
 139         in = *pptr;
 140         out = xmalloc(maxsize);
 141         *pptr = out;
 142
 143         stream.next_in = in;
 144         stream.avail_in = size;
 145         stream.next_out = out;
 146         stream.avail_out = maxsize;
 147         while (deflate(&stream, Z_FINISH) == Z_OK)
 148                 ; /* nothing */
 149         deflateEnd(&stream);
 150
 151         free(in);
 152         return stream.total_out;
 153 }
 154
 155 /*
 156  * The per-object header is a pretty dense thing, which is
 157  *  - first byte: low four bits are "size", then three bits of "type",
 158  *    and the high bit is "size continues".
 159  *  - each byte afterwards: low seven bits are size continuation,
 160  *    with the high bit being "size continues"
 161  */
 162 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 163 {
 164         int n = 1;
 165         unsigned char c;
 166
 167         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 168                 die("bad type %d", type);
 169
 170         c = (type << 4) | (size & 15);
 171         size >>= 4;
 172         while (size) {
 173                 *hdr++ = c | 0x80;
 174                 c = size & 0x7f;
 175                 size >>= 7;
 176                 n++;
 177         }
 178         *hdr = c;
 179         return n;
 180 }
 181
 182 /*
 183  * we are going to reuse the existing object data as is.  make
 184  * sure it is not corrupt.
 185  */
 186 static int check_pack_inflate(struct packed_git *p,
 187                 struct pack_window **w_curs,
 188                 off_t offset,
 189                 off_t len,
 190                 unsigned long expect)
 191 {
 192         z_stream stream;
 193         unsigned char fakebuf[4096], *in;
 194         int st;
 195
 196         memset(&stream, 0, sizeof(stream));
 197         inflateInit(&stream);
 198         do {
 199                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 200                 stream.next_in = in;
 201                 stream.next_out = fakebuf;
 202                 stream.avail_out = sizeof(fakebuf);
 203                 st = inflate(&stream, Z_FINISH);
 204                 offset += stream.next_in - in;
 205         } while (st == Z_OK || st == Z_BUF_ERROR);
 206         inflateEnd(&stream);
 207         return (st == Z_STREAM_END &&
 208                 stream.total_out == expect &&
 209                 stream.total_in == len) ? 0 : -1;
 210 }
 211
 212 static void copy_pack_data(struct sha1file *f,
 213                 struct packed_git *p,
 214                 struct pack_window **w_curs,
 215                 off_t offset,
 216                 off_t len)
 217 {
 218         unsigned char *in;
 219         unsigned int avail;
 220
 221         while (len) {
 222                 in = use_pack(p, w_curs, offset, &avail);
 223                 if (avail > len)
 224                         avail = (unsigned int)len;
 225                 sha1write(f, in, avail);
 226                 offset += avail;
 227                 len -= avail;
 228         }
 229 }
 230
 231 static unsigned long write_object(struct sha1file *f,
 232                                   struct object_entry *entry,
 233                                   off_t write_offset)
 234 {
 235         unsigned long size, limit, datalen;
 236         void *buf;
 237         unsigned char header[10], dheader[10];
 238         unsigned hdrlen;
 239         enum object_type type;
 240         int usable_delta, to_reuse;
 241
 242         if (!pack_to_stdout)
 243                 crc32_begin(f);
 244
 245         type = entry->type;
 246
 247         /* write limit if limited packsize and not first object */
 248         limit = pack_size_limit && nr_written ?
 249                         pack_size_limit - write_offset : 0;
 250
 251         if (!entry->delta)
 252                 usable_delta = 0;       /* no delta */
 253         else if (!pack_size_limit)
 254                usable_delta = 1;        /* unlimited packfile */
 255         else if (entry->delta->idx.offset == (off_t)-1)
 256                 usable_delta = 0;       /* base was written to another pack */
 257         else if (entry->delta->idx.offset)
 258                 usable_delta = 1;       /* base already exists in this pack */
 259         else
 260                 usable_delta = 0;       /* base could end up in another pack */
 261
 262         if (!reuse_object)
 263                 to_reuse = 0;   /* explicit */
 264         else if (!entry->in_pack)
 265                 to_reuse = 0;   /* can't reuse what we don't have */
 266         else if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA)
 267                                 /* check_object() decided it for us ... */
 268                 to_reuse = usable_delta;
 269                                 /* ... but pack split may override that */
 270         else if (type != entry->in_pack_type)
 271                 to_reuse = 0;   /* pack has delta which is unusable */
 272         else if (entry->delta)
 273                 to_reuse = 0;   /* we want to pack afresh */
 274         else
 275                 to_reuse = 1;   /* we have it in-pack undeltified,
 276                                  * and we do not need to deltify it.
 277                                  */
 278
 279         if (!to_reuse) {
 280                 no_reuse:
 281                 if (!usable_delta) {
 282                         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 283                         if (!buf)
 284                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 285                         /*
 286                          * make sure no cached delta data remains from a
 287                          * previous attempt before a pack split occured.
 288                          */
 289                         free(entry->delta_data);
 290                         entry->delta_data = NULL;
 291                         entry->z_delta_size = 0;
 292                 } else if (entry->delta_data) {
 293                         size = entry->delta_size;
 294                         buf = entry->delta_data;
 295                         entry->delta_data = NULL;
 296                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 297                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 298                 } else {
 299                         buf = get_delta(entry);
 300                         size = entry->delta_size;
 301                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 302                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 303                 }
 304
 305                 if (entry->z_delta_size)
 306                         datalen = entry->z_delta_size;
 307                 else
 308                         datalen = do_compress(&buf, size);
 309
 310                 /*
 311                  * The object header is a byte of 'type' followed by zero or
 312                  * more bytes of length.
 313                  */
 314                 hdrlen = encode_header(type, size, header);
 315
 316                 if (type == OBJ_OFS_DELTA) {
 317                         /*
 318                          * Deltas with relative base contain an additional
 319                          * encoding of the relative offset for the delta
 320                          * base from this object's position in the pack.
 321                          */
 322                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 323                         unsigned pos = sizeof(dheader) - 1;
 324                         dheader[pos] = ofs & 127;
 325                         while (ofs >>= 7)
 326                                 dheader[--pos] = 128 | (--ofs & 127);
 327                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 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(buf);
 341                                 return 0;
 342                         }
 343                         sha1write(f, header, hdrlen);
 344                         sha1write(f, entry->delta->idx.sha1, 20);
 345                         hdrlen += 20;
 346                 } else {
 347                         if (limit && hdrlen + datalen + 20 >= limit) {
 348                                 free(buf);
 349                                 return 0;
 350                         }
 351                         sha1write(f, header, hdrlen);
 352                 }
 353                 sha1write(f, buf, datalen);
 354                 free(buf);
 355         }
 356         else {
 357                 struct packed_git *p = entry->in_pack;
 358                 struct pack_window *w_curs = NULL;
 359                 struct revindex_entry *revidx;
 360                 off_t offset;
 361
 362                 if (entry->delta)
 363                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 364                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 365                 hdrlen = encode_header(type, entry->size, header);
 366
 367                 offset = entry->in_pack_offset;
 368                 revidx = find_pack_revindex(p, offset);
 369                 datalen = revidx[1].offset - offset;
 370                 if (!pack_to_stdout && p->index_version > 1 &&
 371                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr)) {
 372                         error("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 373                         unuse_pack(&w_curs);
 374                         goto no_reuse;
 375                 }
 376
 377                 offset += entry->in_pack_header_size;
 378                 datalen -= entry->in_pack_header_size;
 379                 if (!pack_to_stdout && p->index_version == 1 &&
 380                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size)) {
 381                         error("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 382                         unuse_pack(&w_curs);
 383                         goto no_reuse;
 384                 }
 385
 386                 if (type == OBJ_OFS_DELTA) {
 387                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 388                         unsigned pos = sizeof(dheader) - 1;
 389                         dheader[pos] = ofs & 127;
 390                         while (ofs >>= 7)
 391                                 dheader[--pos] = 128 | (--ofs & 127);
 392                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit)
 393                                 return 0;
 394                         sha1write(f, header, hdrlen);
 395                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 396                         hdrlen += sizeof(dheader) - pos;
 397                         reused_delta++;
 398                 } else if (type == OBJ_REF_DELTA) {
 399                         if (limit && hdrlen + 20 + datalen + 20 >= limit)
 400                                 return 0;
 401                         sha1write(f, header, hdrlen);
 402                         sha1write(f, entry->delta->idx.sha1, 20);
 403                         hdrlen += 20;
 404                         reused_delta++;
 405                 } else {
 406                         if (limit && hdrlen + datalen + 20 >= limit)
 407                                 return 0;
 408                         sha1write(f, header, hdrlen);
 409                 }
 410                 copy_pack_data(f, p, &w_curs, offset, datalen);
 411                 unuse_pack(&w_curs);
 412                 reused++;
 413         }
 414         if (usable_delta)
 415                 written_delta++;
 416         written++;
 417         if (!pack_to_stdout)
 418                 entry->idx.crc32 = crc32_end(f);
 419         return hdrlen + datalen;
 420 }
 421
 422 static int write_one(struct sha1file *f,
 423                                struct object_entry *e,
 424                                off_t *offset)
 425 {
 426         unsigned long size;
 427
 428         /* offset is non zero if object is written already. */
 429         if (e->idx.offset || e->preferred_base)
 430                 return 1;
 431
 432         /* if we are deltified, write out base object first. */
 433         if (e->delta && !write_one(f, e->delta, offset))
 434                 return 0;
 435
 436         e->idx.offset = *offset;
 437         size = write_object(f, e, *offset);
 438         if (!size) {
 439                 e->idx.offset = 0;
 440                 return 0;
 441         }
 442         written_list[nr_written++] = &e->idx;
 443
 444         /* make sure off_t is sufficiently large not to wrap */
 445         if (*offset > *offset + size)
 446                 die("pack too large for current definition of off_t");
 447         *offset += size;
 448         return 1;
 449 }
 450
 451 /* forward declaration for write_pack_file */
 452 static int adjust_perm(const char *path, mode_t mode);
 453
 454 static void write_pack_file(void)
 455 {
 456         uint32_t i = 0, j;
 457         struct sha1file *f;
 458         off_t offset;
 459         struct pack_header hdr;
 460         uint32_t nr_remaining = nr_result;
 461         time_t last_mtime = 0;
 462
 463         if (progress > pack_to_stdout)
 464                 progress_state = start_progress("Writing objects", nr_result);
 465         written_list = xmalloc(nr_objects * sizeof(*written_list));
 466
 467         do {
 468                 unsigned char sha1[20];
 469                 char *pack_tmp_name = NULL;
 470
 471                 if (pack_to_stdout) {
 472                         f = sha1fd_throughput(1, "<stdout>", progress_state);
 473                 } else {
 474                         char tmpname[PATH_MAX];
 475                         int fd;
 476                         snprintf(tmpname, sizeof(tmpname),
 477                                  "%s/pack/tmp_pack_XXXXXX", get_object_directory());
 478                         fd = xmkstemp(tmpname);
 479                         pack_tmp_name = xstrdup(tmpname);
 480                         f = sha1fd(fd, pack_tmp_name);
 481                 }
 482
 483                 hdr.hdr_signature = htonl(PACK_SIGNATURE);
 484                 hdr.hdr_version = htonl(PACK_VERSION);
 485                 hdr.hdr_entries = htonl(nr_remaining);
 486                 sha1write(f, &hdr, sizeof(hdr));
 487                 offset = sizeof(hdr);
 488                 nr_written = 0;
 489                 for (; i < nr_objects; i++) {
 490                         if (!write_one(f, objects + i, &offset))
 491                                 break;
 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) {
 500                         sha1close(f, sha1, CSUM_CLOSE);
 501                 } else if (nr_written == nr_remaining) {
 502                         sha1close(f, sha1, CSUM_FSYNC);
 503                 } else {
 504                         int fd = sha1close(f, sha1, 0);
 505                         fixup_pack_header_footer(fd, sha1, pack_tmp_name,
 506                                                  nr_written, sha1, offset);
 507                         close(fd);
 508                 }
 509
 510                 if (!pack_to_stdout) {
 511                         mode_t mode = umask(0);
 512                         struct stat st;
 513                         char *idx_tmp_name, tmpname[PATH_MAX];
 514
 515                         umask(mode);
 516                         mode = 0444 & ~mode;
 517
 518                         idx_tmp_name = write_idx_file(NULL, written_list,
 519                                                       nr_written, sha1);
 520
 521                         snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
 522                                  base_name, sha1_to_hex(sha1));
 523                         if (adjust_perm(pack_tmp_name, mode))
 524                                 die("unable to make temporary pack file readable: %s",
 525                                     strerror(errno));
 526                         if (rename(pack_tmp_name, tmpname))
 527                                 die("unable to rename temporary pack file: %s",
 528                                     strerror(errno));
 529
 530                         /*
 531                          * Packs are runtime accessed in their mtime
 532                          * order since newer packs are more likely to contain
 533                          * younger objects.  So if we are creating multiple
 534                          * packs then we should modify the mtime of later ones
 535                          * to preserve this property.
 536                          */
 537                         if (stat(tmpname, &st) < 0) {
 538                                 warning("failed to stat %s: %s",
 539                                         tmpname, strerror(errno));
 540                         } else if (!last_mtime) {
 541                                 last_mtime = st.st_mtime;
 542                         } else {
 543                                 struct utimbuf utb;
 544                                 utb.actime = st.st_atime;
 545                                 utb.modtime = --last_mtime;
 546                                 if (utime(tmpname, &utb) < 0)
 547                                         warning("failed utime() on %s: %s",
 548                                                 tmpname, strerror(errno));
 549                         }
 550
 551                         snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
 552                                  base_name, sha1_to_hex(sha1));
 553                         if (adjust_perm(idx_tmp_name, mode))
 554                                 die("unable to make temporary index file readable: %s",
 555                                     strerror(errno));
 556                         if (rename(idx_tmp_name, tmpname))
 557                                 die("unable to rename temporary index file: %s",
 558                                     strerror(errno));
 559
 560                         free(idx_tmp_name);
 561                         free(pack_tmp_name);
 562                         puts(sha1_to_hex(sha1));
 563                 }
 564
 565                 /* mark written objects as written to previous pack */
 566                 for (j = 0; j < nr_written; j++) {
 567                         written_list[j]->offset = (off_t)-1;
 568                 }
 569                 nr_remaining -= nr_written;
 570         } while (nr_remaining && i < nr_objects);
 571
 572         free(written_list);
 573         stop_progress(&progress_state);
 574         if (written != nr_result)
 575                 die("wrote %"PRIu32" objects while expecting %"PRIu32,
 576                         written, nr_result);
 577         /*
 578          * We have scanned through [0 ... i).  Since we have written
 579          * the correct number of objects,  the remaining [i ... nr_objects)
 580          * items must be either already written (due to out-of-order delta base)
 581          * or a preferred base.  Count those which are neither and complain if any.
 582          */
 583         for (j = 0; i < nr_objects; i++) {
 584                 struct object_entry *e = objects + i;
 585                 j += !e->idx.offset && !e->preferred_base;
 586         }
 587         if (j)
 588                 die("wrote %"PRIu32" objects as expected but %"PRIu32
 589                         " unwritten", written, j);
 590 }
 591
 592 static int locate_object_entry_hash(const unsigned char *sha1)
 593 {
 594         int i;
 595         unsigned int ui;
 596         memcpy(&ui, sha1, sizeof(unsigned int));
 597         i = ui % object_ix_hashsz;
 598         while (0 < object_ix[i]) {
 599                 if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1))
 600                         return i;
 601                 if (++i == object_ix_hashsz)
 602                         i = 0;
 603         }
 604         return -1 - i;
 605 }
 606
 607 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 608 {
 609         int i;
 610
 611         if (!object_ix_hashsz)
 612                 return NULL;
 613
 614         i = locate_object_entry_hash(sha1);
 615         if (0 <= i)
 616                 return &objects[object_ix[i]-1];
 617         return NULL;
 618 }
 619
 620 static void rehash_objects(void)
 621 {
 622         uint32_t i;
 623         struct object_entry *oe;
 624
 625         object_ix_hashsz = nr_objects * 3;
 626         if (object_ix_hashsz < 1024)
 627                 object_ix_hashsz = 1024;
 628         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 629         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 630         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 631                 int ix = locate_object_entry_hash(oe->idx.sha1);
 632                 if (0 <= ix)
 633                         continue;
 634                 ix = -1 - ix;
 635                 object_ix[ix] = i + 1;
 636         }
 637 }
 638
 639 static unsigned name_hash(const char *name)
 640 {
 641         unsigned char c;
 642         unsigned hash = 0;
 643
 644         if (!name)
 645                 return 0;
 646
 647         /*
 648          * This effectively just creates a sortable number from the
 649          * last sixteen non-whitespace characters. Last characters
 650          * count "most", so things that end in ".c" sort together.
 651          */
 652         while ((c = *name++) != 0) {
 653                 if (isspace(c))
 654                         continue;
 655                 hash = (hash >> 2) + (c << 24);
 656         }
 657         return hash;
 658 }
 659
 660 static void setup_delta_attr_check(struct git_attr_check *check)
 661 {
 662         static struct git_attr *attr_delta;
 663
 664         if (!attr_delta)
 665                 attr_delta = git_attr("delta", 5);
 666
 667         check[0].attr = attr_delta;
 668 }
 669
 670 static int no_try_delta(const char *path)
 671 {
 672         struct git_attr_check check[1];
 673
 674         setup_delta_attr_check(check);
 675         if (git_checkattr(path, ARRAY_SIZE(check), check))
 676                 return 0;
 677         if (ATTR_FALSE(check->value))
 678                 return 1;
 679         return 0;
 680 }
 681
 682 static int add_object_entry(const unsigned char *sha1, enum object_type type,
 683                             const char *name, int exclude)
 684 {
 685         struct object_entry *entry;
 686         struct packed_git *p, *found_pack = NULL;
 687         off_t found_offset = 0;
 688         int ix;
 689         unsigned hash = name_hash(name);
 690
 691         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 692         if (ix >= 0) {
 693                 if (exclude) {
 694                         entry = objects + object_ix[ix] - 1;
 695                         if (!entry->preferred_base)
 696                                 nr_result--;
 697                         entry->preferred_base = 1;
 698                 }
 699                 return 0;
 700         }
 701
 702         for (p = packed_git; p; p = p->next) {
 703                 off_t offset = find_pack_entry_one(sha1, p);
 704                 if (offset) {
 705                         if (!found_pack) {
 706                                 found_offset = offset;
 707                                 found_pack = p;
 708                         }
 709                         if (exclude)
 710                                 break;
 711                         if (incremental)
 712                                 return 0;
 713                         if (local && !p->pack_local)
 714                                 return 0;
 715                 }
 716         }
 717
 718         if (nr_objects >= nr_alloc) {
 719                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 720                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 721         }
 722
 723         entry = objects + nr_objects++;
 724         memset(entry, 0, sizeof(*entry));
 725         hashcpy(entry->idx.sha1, sha1);
 726         entry->hash = hash;
 727         if (type)
 728                 entry->type = type;
 729         if (exclude)
 730                 entry->preferred_base = 1;
 731         else
 732                 nr_result++;
 733         if (found_pack) {
 734                 entry->in_pack = found_pack;
 735                 entry->in_pack_offset = found_offset;
 736         }
 737
 738         if (object_ix_hashsz * 3 <= nr_objects * 4)
 739                 rehash_objects();
 740         else
 741                 object_ix[-1 - ix] = nr_objects;
 742
 743         display_progress(progress_state, nr_objects);
 744
 745         if (name && no_try_delta(name))
 746                 entry->no_try_delta = 1;
 747
 748         return 1;
 749 }
 750
 751 struct pbase_tree_cache {
 752         unsigned char sha1[20];
 753         int ref;
 754         int temporary;
 755         void *tree_data;
 756         unsigned long tree_size;
 757 };
 758
 759 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 760 static int pbase_tree_cache_ix(const unsigned char *sha1)
 761 {
 762         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 763 }
 764 static int pbase_tree_cache_ix_incr(int ix)
 765 {
 766         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 767 }
 768
 769 static struct pbase_tree {
 770         struct pbase_tree *next;
 771         /* This is a phony "cache" entry; we are not
 772          * going to evict it nor find it through _get()
 773          * mechanism -- this is for the toplevel node that
 774          * would almost always change with any commit.
 775          */
 776         struct pbase_tree_cache pcache;
 777 } *pbase_tree;
 778
 779 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 780 {
 781         struct pbase_tree_cache *ent, *nent;
 782         void *data;
 783         unsigned long size;
 784         enum object_type type;
 785         int neigh;
 786         int my_ix = pbase_tree_cache_ix(sha1);
 787         int available_ix = -1;
 788
 789         /* pbase-tree-cache acts as a limited hashtable.
 790          * your object will be found at your index or within a few
 791          * slots after that slot if it is cached.
 792          */
 793         for (neigh = 0; neigh < 8; neigh++) {
 794                 ent = pbase_tree_cache[my_ix];
 795                 if (ent && !hashcmp(ent->sha1, sha1)) {
 796                         ent->ref++;
 797                         return ent;
 798                 }
 799                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 800                          ((0 <= available_ix) &&
 801                           (!ent && pbase_tree_cache[available_ix])))
 802                         available_ix = my_ix;
 803                 if (!ent)
 804                         break;
 805                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 806         }
 807
 808         /* Did not find one.  Either we got a bogus request or
 809          * we need to read and perhaps cache.
 810          */
 811         data = read_sha1_file(sha1, &type, &size);
 812         if (!data)
 813                 return NULL;
 814         if (type != OBJ_TREE) {
 815                 free(data);
 816                 return NULL;
 817         }
 818
 819         /* We need to either cache or return a throwaway copy */
 820
 821         if (available_ix < 0)
 822                 ent = NULL;
 823         else {
 824                 ent = pbase_tree_cache[available_ix];
 825                 my_ix = available_ix;
 826         }
 827
 828         if (!ent) {
 829                 nent = xmalloc(sizeof(*nent));
 830                 nent->temporary = (available_ix < 0);
 831         }
 832         else {
 833                 /* evict and reuse */
 834                 free(ent->tree_data);
 835                 nent = ent;
 836         }
 837         hashcpy(nent->sha1, sha1);
 838         nent->tree_data = data;
 839         nent->tree_size = size;
 840         nent->ref = 1;
 841         if (!nent->temporary)
 842                 pbase_tree_cache[my_ix] = nent;
 843         return nent;
 844 }
 845
 846 static void pbase_tree_put(struct pbase_tree_cache *cache)
 847 {
 848         if (!cache->temporary) {
 849                 cache->ref--;
 850                 return;
 851         }
 852         free(cache->tree_data);
 853         free(cache);
 854 }
 855
 856 static int name_cmp_len(const char *name)
 857 {
 858         int i;
 859         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 860                 ;
 861         return i;
 862 }
 863
 864 static void add_pbase_object(struct tree_desc *tree,
 865                              const char *name,
 866                              int cmplen,
 867                              const char *fullname)
 868 {
 869         struct name_entry entry;
 870         int cmp;
 871
 872         while (tree_entry(tree,&entry)) {
 873                 if (S_ISGITLINK(entry.mode))
 874                         continue;
 875                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 876                       memcmp(name, entry.path, cmplen);
 877                 if (cmp > 0)
 878                         continue;
 879                 if (cmp < 0)
 880                         return;
 881                 if (name[cmplen] != '/') {
 882                         add_object_entry(entry.sha1,
 883                                          object_type(entry.mode),
 884                                          fullname, 1);
 885                         return;
 886                 }
 887                 if (S_ISDIR(entry.mode)) {
 888                         struct tree_desc sub;
 889                         struct pbase_tree_cache *tree;
 890                         const char *down = name+cmplen+1;
 891                         int downlen = name_cmp_len(down);
 892
 893                         tree = pbase_tree_get(entry.sha1);
 894                         if (!tree)
 895                                 return;
 896                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
 897
 898                         add_pbase_object(&sub, down, downlen, fullname);
 899                         pbase_tree_put(tree);
 900                 }
 901         }
 902 }
 903
 904 static unsigned *done_pbase_paths;
 905 static int done_pbase_paths_num;
 906 static int done_pbase_paths_alloc;
 907 static int done_pbase_path_pos(unsigned hash)
 908 {
 909         int lo = 0;
 910         int hi = done_pbase_paths_num;
 911         while (lo < hi) {
 912                 int mi = (hi + lo) / 2;
 913                 if (done_pbase_paths[mi] == hash)
 914                         return mi;
 915                 if (done_pbase_paths[mi] < hash)
 916                         hi = mi;
 917                 else
 918                         lo = mi + 1;
 919         }
 920         return -lo-1;
 921 }
 922
 923 static int check_pbase_path(unsigned hash)
 924 {
 925         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
 926         if (0 <= pos)
 927                 return 1;
 928         pos = -pos - 1;
 929         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
 930                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
 931                 done_pbase_paths = xrealloc(done_pbase_paths,
 932                                             done_pbase_paths_alloc *
 933                                             sizeof(unsigned));
 934         }
 935         done_pbase_paths_num++;
 936         if (pos < done_pbase_paths_num)
 937                 memmove(done_pbase_paths + pos + 1,
 938                         done_pbase_paths + pos,
 939                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
 940         done_pbase_paths[pos] = hash;
 941         return 0;
 942 }
 943
 944 static void add_preferred_base_object(const char *name)
 945 {
 946         struct pbase_tree *it;
 947         int cmplen;
 948         unsigned hash = name_hash(name);
 949
 950         if (!num_preferred_base || check_pbase_path(hash))
 951                 return;
 952
 953         cmplen = name_cmp_len(name);
 954         for (it = pbase_tree; it; it = it->next) {
 955                 if (cmplen == 0) {
 956                         add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
 957                 }
 958                 else {
 959                         struct tree_desc tree;
 960                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
 961                         add_pbase_object(&tree, name, cmplen, name);
 962                 }
 963         }
 964 }
 965
 966 static void add_preferred_base(unsigned char *sha1)
 967 {
 968         struct pbase_tree *it;
 969         void *data;
 970         unsigned long size;
 971         unsigned char tree_sha1[20];
 972
 973         if (window <= num_preferred_base++)
 974                 return;
 975
 976         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
 977         if (!data)
 978                 return;
 979
 980         for (it = pbase_tree; it; it = it->next) {
 981                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
 982                         free(data);
 983                         return;
 984                 }
 985         }
 986
 987         it = xcalloc(1, sizeof(*it));
 988         it->next = pbase_tree;
 989         pbase_tree = it;
 990
 991         hashcpy(it->pcache.sha1, tree_sha1);
 992         it->pcache.tree_data = data;
 993         it->pcache.tree_size = size;
 994 }
 995
 996 static void check_object(struct object_entry *entry)
 997 {
 998         if (entry->in_pack) {
 999                 struct packed_git *p = entry->in_pack;
1000                 struct pack_window *w_curs = NULL;
1001                 const unsigned char *base_ref = NULL;
1002                 struct object_entry *base_entry;
1003                 unsigned long used, used_0;
1004                 unsigned int avail;
1005                 off_t ofs;
1006                 unsigned char *buf, c;
1007
1008                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1009
1010                 /*
1011                  * We want in_pack_type even if we do not reuse delta
1012                  * since non-delta representations could still be reused.
1013                  */
1014                 used = unpack_object_header_buffer(buf, avail,
1015                                                    &entry->in_pack_type,
1016                                                    &entry->size);
1017                 if (used == 0)
1018                         goto give_up;
1019
1020                 /*
1021                  * Determine if this is a delta and if so whether we can
1022                  * reuse it or not.  Otherwise let's find out as cheaply as
1023                  * possible what the actual type and size for this object is.
1024                  */
1025                 switch (entry->in_pack_type) {
1026                 default:
1027                         /* Not a delta hence we've already got all we need. */
1028                         entry->type = entry->in_pack_type;
1029                         entry->in_pack_header_size = used;
1030                         if (entry->type < OBJ_COMMIT || entry->type > OBJ_BLOB)
1031                                 goto give_up;
1032                         unuse_pack(&w_curs);
1033                         return;
1034                 case OBJ_REF_DELTA:
1035                         if (reuse_delta && !entry->preferred_base)
1036                                 base_ref = use_pack(p, &w_curs,
1037                                                 entry->in_pack_offset + used, NULL);
1038                         entry->in_pack_header_size = used + 20;
1039                         break;
1040                 case OBJ_OFS_DELTA:
1041                         buf = use_pack(p, &w_curs,
1042                                        entry->in_pack_offset + used, NULL);
1043                         used_0 = 0;
1044                         c = buf[used_0++];
1045                         ofs = c & 127;
1046                         while (c & 128) {
1047                                 ofs += 1;
1048                                 if (!ofs || MSB(ofs, 7)) {
1049                                         error("delta base offset overflow in pack for %s",
1050                                               sha1_to_hex(entry->idx.sha1));
1051                                         goto give_up;
1052                                 }
1053                                 c = buf[used_0++];
1054                                 ofs = (ofs << 7) + (c & 127);
1055                         }
1056                         ofs = entry->in_pack_offset - ofs;
1057                         if (ofs <= 0 || ofs >= entry->in_pack_offset) {
1058                                 error("delta base offset out of bound for %s",
1059                                       sha1_to_hex(entry->idx.sha1));
1060                                 goto give_up;
1061                         }
1062                         if (reuse_delta && !entry->preferred_base) {
1063                                 struct revindex_entry *revidx;
1064                                 revidx = find_pack_revindex(p, ofs);
1065                                 if (!revidx)
1066                                         goto give_up;
1067                                 base_ref = nth_packed_object_sha1(p, revidx->nr);
1068                         }
1069                         entry->in_pack_header_size = used + used_0;
1070                         break;
1071                 }
1072
1073                 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1074                         /*
1075                          * If base_ref was set above that means we wish to
1076                          * reuse delta data, and we even found that base
1077                          * in the list of objects we want to pack. Goodie!
1078                          *
1079                          * Depth value does not matter - find_deltas() will
1080                          * never consider reused delta as the base object to
1081                          * deltify other objects against, in order to avoid
1082                          * circular deltas.
1083                          */
1084                         entry->type = entry->in_pack_type;
1085                         entry->delta = base_entry;
1086                         entry->delta_size = entry->size;
1087                         entry->delta_sibling = base_entry->delta_child;
1088                         base_entry->delta_child = entry;
1089                         unuse_pack(&w_curs);
1090                         return;
1091                 }
1092
1093                 if (entry->type) {
1094                         /*
1095                          * This must be a delta and we already know what the
1096                          * final object type is.  Let's extract the actual
1097                          * object size from the delta header.
1098                          */
1099                         entry->size = get_size_from_delta(p, &w_curs,
1100                                         entry->in_pack_offset + entry->in_pack_header_size);
1101                         if (entry->size == 0)
1102                                 goto give_up;
1103                         unuse_pack(&w_curs);
1104                         return;
1105                 }
1106
1107                 /*
1108                  * No choice but to fall back to the recursive delta walk
1109                  * with sha1_object_info() to find about the object type
1110                  * at this point...
1111                  */
1112                 give_up:
1113                 unuse_pack(&w_curs);
1114         }
1115
1116         entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1117         /*
1118          * The error condition is checked in prepare_pack().  This is
1119          * to permit a missing preferred base object to be ignored
1120          * as a preferred base.  Doing so can result in a larger
1121          * pack file, but the transfer will still take place.
1122          */
1123 }
1124
1125 static int pack_offset_sort(const void *_a, const void *_b)
1126 {
1127         const struct object_entry *a = *(struct object_entry **)_a;
1128         const struct object_entry *b = *(struct object_entry **)_b;
1129
1130         /* avoid filesystem trashing with loose objects */
1131         if (!a->in_pack && !b->in_pack)
1132                 return hashcmp(a->idx.sha1, b->idx.sha1);
1133
1134         if (a->in_pack < b->in_pack)
1135                 return -1;
1136         if (a->in_pack > b->in_pack)
1137                 return 1;
1138         return a->in_pack_offset < b->in_pack_offset ? -1 :
1139                         (a->in_pack_offset > b->in_pack_offset);
1140 }
1141
1142 static void get_object_details(void)
1143 {
1144         uint32_t i;
1145         struct object_entry **sorted_by_offset;
1146
1147         sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1148         for (i = 0; i < nr_objects; i++)
1149                 sorted_by_offset[i] = objects + i;
1150         qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1151
1152         for (i = 0; i < nr_objects; i++)
1153                 check_object(sorted_by_offset[i]);
1154
1155         free(sorted_by_offset);
1156 }
1157
1158 /*
1159  * We search for deltas in a list sorted by type, by filename hash, and then
1160  * by size, so that we see progressively smaller and smaller files.
1161  * That's because we prefer deltas to be from the bigger file
1162  * to the smaller -- deletes are potentially cheaper, but perhaps
1163  * more importantly, the bigger file is likely the more recent
1164  * one.  The deepest deltas are therefore the oldest objects which are
1165  * less susceptible to be accessed often.
1166  */
1167 static int type_size_sort(const void *_a, const void *_b)
1168 {
1169         const struct object_entry *a = *(struct object_entry **)_a;
1170         const struct object_entry *b = *(struct object_entry **)_b;
1171
1172         if (a->type > b->type)
1173                 return -1;
1174         if (a->type < b->type)
1175                 return 1;
1176         if (a->hash > b->hash)
1177                 return -1;
1178         if (a->hash < b->hash)
1179                 return 1;
1180         if (a->preferred_base > b->preferred_base)
1181                 return -1;
1182         if (a->preferred_base < b->preferred_base)
1183                 return 1;
1184         if (a->size > b->size)
1185                 return -1;
1186         if (a->size < b->size)
1187                 return 1;
1188         return a < b ? -1 : (a > b);  /* newest first */
1189 }
1190
1191 struct unpacked {
1192         struct object_entry *entry;
1193         void *data;
1194         struct delta_index *index;
1195         unsigned depth;
1196 };
1197
1198 static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1199                            unsigned long delta_size)
1200 {
1201         if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1202                 return 0;
1203
1204         if (delta_size < cache_max_small_delta_size)
1205                 return 1;
1206
1207         /* cache delta, if objects are large enough compared to delta size */
1208         if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1209                 return 1;
1210
1211         return 0;
1212 }
1213
1214 #ifdef THREADED_DELTA_SEARCH
1215
1216 static pthread_mutex_t read_mutex = PTHREAD_MUTEX_INITIALIZER;
1217 #define read_lock()             pthread_mutex_lock(&read_mutex)
1218 #define read_unlock()           pthread_mutex_unlock(&read_mutex)
1219
1220 static pthread_mutex_t cache_mutex = PTHREAD_MUTEX_INITIALIZER;
1221 #define cache_lock()            pthread_mutex_lock(&cache_mutex)
1222 #define cache_unlock()          pthread_mutex_unlock(&cache_mutex)
1223
1224 static pthread_mutex_t progress_mutex = PTHREAD_MUTEX_INITIALIZER;
1225 #define progress_lock()         pthread_mutex_lock(&progress_mutex)
1226 #define progress_unlock()       pthread_mutex_unlock(&progress_mutex)
1227
1228 #else
1229
1230 #define read_lock()             (void)0
1231 #define read_unlock()           (void)0
1232 #define cache_lock()            (void)0
1233 #define cache_unlock()          (void)0
1234 #define progress_lock()         (void)0
1235 #define progress_unlock()       (void)0
1236
1237 #endif
1238
1239 static int try_delta(struct unpacked *trg, struct unpacked *src,
1240                      unsigned max_depth, unsigned long *mem_usage)
1241 {
1242         struct object_entry *trg_entry = trg->entry;
1243         struct object_entry *src_entry = src->entry;
1244         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1245         unsigned ref_depth;
1246         enum object_type type;
1247         void *delta_buf;
1248
1249         /* Don't bother doing diffs between different types */
1250         if (trg_entry->type != src_entry->type)
1251                 return -1;
1252
1253         /*
1254          * We do not bother to try a delta that we discarded
1255          * on an earlier try, but only when reusing delta data.
1256          */
1257         if (reuse_delta && trg_entry->in_pack &&
1258             trg_entry->in_pack == src_entry->in_pack &&
1259             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1260             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1261                 return 0;
1262
1263         /* Let's not bust the allowed depth. */
1264         if (src->depth >= max_depth)
1265                 return 0;
1266
1267         /* Now some size filtering heuristics. */
1268         trg_size = trg_entry->size;
1269         if (!trg_entry->delta) {
1270                 max_size = trg_size/2 - 20;
1271                 ref_depth = 1;
1272         } else {
1273                 max_size = trg_entry->delta_size;
1274                 ref_depth = trg->depth;
1275         }
1276         max_size = max_size * (max_depth - src->depth) /
1277                                                 (max_depth - ref_depth + 1);
1278         if (max_size == 0)
1279                 return 0;
1280         src_size = src_entry->size;
1281         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1282         if (sizediff >= max_size)
1283                 return 0;
1284         if (trg_size < src_size / 32)
1285                 return 0;
1286
1287         /* Load data if not already done */
1288         if (!trg->data) {
1289                 read_lock();
1290                 trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1291                 read_unlock();
1292                 if (!trg->data)
1293                         die("object %s cannot be read",
1294                             sha1_to_hex(trg_entry->idx.sha1));
1295                 if (sz != trg_size)
1296                         die("object %s inconsistent object length (%lu vs %lu)",
1297                             sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1298                 *mem_usage += sz;
1299         }
1300         if (!src->data) {
1301                 read_lock();
1302                 src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1303                 read_unlock();
1304                 if (!src->data)
1305                         die("object %s cannot be read",
1306                             sha1_to_hex(src_entry->idx.sha1));
1307                 if (sz != src_size)
1308                         die("object %s inconsistent object length (%lu vs %lu)",
1309                             sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1310                 *mem_usage += sz;
1311         }
1312         if (!src->index) {
1313                 src->index = create_delta_index(src->data, src_size);
1314                 if (!src->index) {
1315                         static int warned = 0;
1316                         if (!warned++)
1317                                 warning("suboptimal pack - out of memory");
1318                         return 0;
1319                 }
1320                 *mem_usage += sizeof_delta_index(src->index);
1321         }
1322
1323         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1324         if (!delta_buf)
1325                 return 0;
1326
1327         if (trg_entry->delta) {
1328                 /* Prefer only shallower same-sized deltas. */
1329                 if (delta_size == trg_entry->delta_size &&
1330                     src->depth + 1 >= trg->depth) {
1331                         free(delta_buf);
1332                         return 0;
1333                 }
1334         }
1335
1336         /*
1337          * Handle memory allocation outside of the cache
1338          * accounting lock.  Compiler will optimize the strangeness
1339          * away when THREADED_DELTA_SEARCH is not defined.
1340          */
1341         free(trg_entry->delta_data);
1342         cache_lock();
1343         if (trg_entry->delta_data) {
1344                 delta_cache_size -= trg_entry->delta_size;
1345                 trg_entry->delta_data = NULL;
1346         }
1347         if (delta_cacheable(src_size, trg_size, delta_size)) {
1348                 delta_cache_size += delta_size;
1349                 cache_unlock();
1350                 trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1351         } else {
1352                 cache_unlock();
1353                 free(delta_buf);
1354         }
1355
1356         trg_entry->delta = src_entry;
1357         trg_entry->delta_size = delta_size;
1358         trg->depth = src->depth + 1;
1359
1360         return 1;
1361 }
1362
1363 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1364 {
1365         struct object_entry *child = me->delta_child;
1366         unsigned int m = n;
1367         while (child) {
1368                 unsigned int c = check_delta_limit(child, n + 1);
1369                 if (m < c)
1370                         m = c;
1371                 child = child->delta_sibling;
1372         }
1373         return m;
1374 }
1375
1376 static unsigned long free_unpacked(struct unpacked *n)
1377 {
1378         unsigned long freed_mem = sizeof_delta_index(n->index);
1379         free_delta_index(n->index);
1380         n->index = NULL;
1381         if (n->data) {
1382                 freed_mem += n->entry->size;
1383                 free(n->data);
1384                 n->data = NULL;
1385         }
1386         n->entry = NULL;
1387         n->depth = 0;
1388         return freed_mem;
1389 }
1390
1391 static void find_deltas(struct object_entry **list, unsigned *list_size,
1392                         int window, int depth, unsigned *processed)
1393 {
1394         uint32_t i, idx = 0, count = 0;
1395         struct unpacked *array;
1396         unsigned long mem_usage = 0;
1397
1398         array = xcalloc(window, sizeof(struct unpacked));
1399
1400         for (;;) {
1401                 struct object_entry *entry;
1402                 struct unpacked *n = array + idx;
1403                 int j, max_depth, best_base = -1;
1404
1405                 progress_lock();
1406                 if (!*list_size) {
1407                         progress_unlock();
1408                         break;
1409                 }
1410                 entry = *list++;
1411                 (*list_size)--;
1412                 if (!entry->preferred_base) {
1413                         (*processed)++;
1414                         display_progress(progress_state, *processed);
1415                 }
1416                 progress_unlock();
1417
1418                 mem_usage -= free_unpacked(n);
1419                 n->entry = entry;
1420
1421                 while (window_memory_limit &&
1422                        mem_usage > window_memory_limit &&
1423                        count > 1) {
1424                         uint32_t tail = (idx + window - count) % window;
1425                         mem_usage -= free_unpacked(array + tail);
1426                         count--;
1427                 }
1428
1429                 /* We do not compute delta to *create* objects we are not
1430                  * going to pack.
1431                  */
1432                 if (entry->preferred_base)
1433                         goto next;
1434
1435                 /*
1436                  * If the current object is at pack edge, take the depth the
1437                  * objects that depend on the current object into account
1438                  * otherwise they would become too deep.
1439                  */
1440                 max_depth = depth;
1441                 if (entry->delta_child) {
1442                         max_depth -= check_delta_limit(entry, 0);
1443                         if (max_depth <= 0)
1444                                 goto next;
1445                 }
1446
1447                 j = window;
1448                 while (--j > 0) {
1449                         int ret;
1450                         uint32_t other_idx = idx + j;
1451                         struct unpacked *m;
1452                         if (other_idx >= window)
1453                                 other_idx -= window;
1454                         m = array + other_idx;
1455                         if (!m->entry)
1456                                 break;
1457                         ret = try_delta(n, m, max_depth, &mem_usage);
1458                         if (ret < 0)
1459                                 break;
1460                         else if (ret > 0)
1461                                 best_base = other_idx;
1462                 }
1463
1464                 /*
1465                  * If we decided to cache the delta data, then it is best
1466                  * to compress it right away.  First because we have to do
1467                  * it anyway, and doing it here while we're threaded will
1468                  * save a lot of time in the non threaded write phase,
1469                  * as well as allow for caching more deltas within
1470                  * the same cache size limit.
1471                  * ...
1472                  * But only if not writing to stdout, since in that case
1473                  * the network is most likely throttling writes anyway,
1474                  * and therefore it is best to go to the write phase ASAP
1475                  * instead, as we can afford spending more time compressing
1476                  * between writes at that moment.
1477                  */
1478                 if (entry->delta_data && !pack_to_stdout) {
1479                         entry->z_delta_size = do_compress(&entry->delta_data,
1480                                                           entry->delta_size);
1481                         cache_lock();
1482                         delta_cache_size -= entry->delta_size;
1483                         delta_cache_size += entry->z_delta_size;
1484                         cache_unlock();
1485                 }
1486
1487                 /* if we made n a delta, and if n is already at max
1488                  * depth, leaving it in the window is pointless.  we
1489                  * should evict it first.
1490                  */
1491                 if (entry->delta && max_depth <= n->depth)
1492                         continue;
1493
1494                 /*
1495                  * Move the best delta base up in the window, after the
1496                  * currently deltified object, to keep it longer.  It will
1497                  * be the first base object to be attempted next.
1498                  */
1499                 if (entry->delta) {
1500                         struct unpacked swap = array[best_base];
1501                         int dist = (window + idx - best_base) % window;
1502                         int dst = best_base;
1503                         while (dist--) {
1504                                 int src = (dst + 1) % window;
1505                                 array[dst] = array[src];
1506                                 dst = src;
1507                         }
1508                         array[dst] = swap;
1509                 }
1510
1511                 next:
1512                 idx++;
1513                 if (count + 1 < window)
1514                         count++;
1515                 if (idx >= window)
1516                         idx = 0;
1517         }
1518
1519         for (i = 0; i < window; ++i) {
1520                 free_delta_index(array[i].index);
1521                 free(array[i].data);
1522         }
1523         free(array);
1524 }
1525
1526 #ifdef THREADED_DELTA_SEARCH
1527
1528 /*
1529  * The main thread waits on the condition that (at least) one of the workers
1530  * has stopped working (which is indicated in the .working member of
1531  * struct thread_params).
1532  * When a work thread has completed its work, it sets .working to 0 and
1533  * signals the main thread and waits on the condition that .data_ready
1534  * becomes 1.
1535  */
1536
1537 struct thread_params {
1538         pthread_t thread;
1539         struct object_entry **list;
1540         unsigned list_size;
1541         unsigned remaining;
1542         int window;
1543         int depth;
1544         int working;
1545         int data_ready;
1546         pthread_mutex_t mutex;
1547         pthread_cond_t cond;
1548         unsigned *processed;
1549 };
1550
1551 static pthread_cond_t progress_cond = PTHREAD_COND_INITIALIZER;
1552
1553 static void *threaded_find_deltas(void *arg)
1554 {
1555         struct thread_params *me = arg;
1556
1557         while (me->remaining) {
1558                 find_deltas(me->list, &me->remaining,
1559                             me->window, me->depth, me->processed);
1560
1561                 progress_lock();
1562                 me->working = 0;
1563                 pthread_cond_signal(&progress_cond);
1564                 progress_unlock();
1565
1566                 /*
1567                  * We must not set ->data_ready before we wait on the
1568                  * condition because the main thread may have set it to 1
1569                  * before we get here. In order to be sure that new
1570                  * work is available if we see 1 in ->data_ready, it
1571                  * was initialized to 0 before this thread was spawned
1572                  * and we reset it to 0 right away.
1573                  */
1574                 pthread_mutex_lock(&me->mutex);
1575                 while (!me->data_ready)
1576                         pthread_cond_wait(&me->cond, &me->mutex);
1577                 me->data_ready = 0;
1578                 pthread_mutex_unlock(&me->mutex);
1579         }
1580         /* leave ->working 1 so that this doesn't get more work assigned */
1581         return NULL;
1582 }
1583
1584 static void ll_find_deltas(struct object_entry **list, unsigned list_size,
1585                            int window, int depth, unsigned *processed)
1586 {
1587         struct thread_params p[delta_search_threads];
1588         int i, ret, active_threads = 0;
1589
1590         if (delta_search_threads <= 1) {
1591                 find_deltas(list, &list_size, window, depth, processed);
1592                 return;
1593         }
1594
1595         /* Partition the work amongst work threads. */
1596         for (i = 0; i < delta_search_threads; i++) {
1597                 unsigned sub_size = list_size / (delta_search_threads - i);
1598
1599                 p[i].window = window;
1600                 p[i].depth = depth;
1601                 p[i].processed = processed;
1602                 p[i].working = 1;
1603                 p[i].data_ready = 0;
1604
1605                 /* try to split chunks on "path" boundaries */
1606                 while (sub_size && sub_size < list_size &&
1607                        list[sub_size]->hash &&
1608                        list[sub_size]->hash == list[sub_size-1]->hash)
1609                         sub_size++;
1610
1611                 p[i].list = list;
1612                 p[i].list_size = sub_size;
1613                 p[i].remaining = sub_size;
1614
1615                 list += sub_size;
1616                 list_size -= sub_size;
1617         }
1618
1619         /* Start work threads. */
1620         for (i = 0; i < delta_search_threads; i++) {
1621                 if (!p[i].list_size)
1622                         continue;
1623                 pthread_mutex_init(&p[i].mutex, NULL);
1624                 pthread_cond_init(&p[i].cond, NULL);
1625                 ret = pthread_create(&p[i].thread, NULL,
1626                                      threaded_find_deltas, &p[i]);
1627                 if (ret)
1628                         die("unable to create thread: %s", strerror(ret));
1629                 active_threads++;
1630         }
1631
1632         /*
1633          * Now let's wait for work completion.  Each time a thread is done
1634          * with its work, we steal half of the remaining work from the
1635          * thread with the largest number of unprocessed objects and give
1636          * it to that newly idle thread.  This ensure good load balancing
1637          * until the remaining object list segments are simply too short
1638          * to be worth splitting anymore.
1639          */
1640         while (active_threads) {
1641                 struct thread_params *target = NULL;
1642                 struct thread_params *victim = NULL;
1643                 unsigned sub_size = 0;
1644
1645                 progress_lock();
1646                 for (;;) {
1647                         for (i = 0; !target && i < delta_search_threads; i++)
1648                                 if (!p[i].working)
1649                                         target = &p[i];
1650                         if (target)
1651                                 break;
1652                         pthread_cond_wait(&progress_cond, &progress_mutex);
1653                 }
1654
1655                 for (i = 0; i < delta_search_threads; i++)
1656                         if (p[i].remaining > 2*window &&
1657                             (!victim || victim->remaining < p[i].remaining))
1658                                 victim = &p[i];
1659                 if (victim) {
1660                         sub_size = victim->remaining / 2;
1661                         list = victim->list + victim->list_size - sub_size;
1662                         while (sub_size && list[0]->hash &&
1663                                list[0]->hash == list[-1]->hash) {
1664                                 list++;
1665                                 sub_size--;
1666                         }
1667                         if (!sub_size) {
1668                                 /*
1669                                  * It is possible for some "paths" to have
1670                                  * so many objects that no hash boundary
1671                                  * might be found.  Let's just steal the
1672                                  * exact half in that case.
1673                                  */
1674                                 sub_size = victim->remaining / 2;
1675                                 list -= sub_size;
1676                         }
1677                         target->list = list;
1678                         victim->list_size -= sub_size;
1679                         victim->remaining -= sub_size;
1680                 }
1681                 target->list_size = sub_size;
1682                 target->remaining = sub_size;
1683                 target->working = 1;
1684                 progress_unlock();
1685
1686                 pthread_mutex_lock(&target->mutex);
1687                 target->data_ready = 1;
1688                 pthread_cond_signal(&target->cond);
1689                 pthread_mutex_unlock(&target->mutex);
1690
1691                 if (!sub_size) {
1692                         pthread_join(target->thread, NULL);
1693                         pthread_cond_destroy(&target->cond);
1694                         pthread_mutex_destroy(&target->mutex);
1695                         active_threads--;
1696                 }
1697         }
1698 }
1699
1700 #else
1701 #define ll_find_deltas(l, s, w, d, p)   find_deltas(l, &s, w, d, p)
1702 #endif
1703
1704 static int add_ref_tag(const char *path, const unsigned char *sha1, int flag, void *cb_data)
1705 {
1706         unsigned char peeled[20];
1707
1708         if (!prefixcmp(path, "refs/tags/") && /* is a tag? */
1709             !peel_ref(path, peeled)        && /* peelable? */
1710             !is_null_sha1(peeled)          && /* annotated tag? */
1711             locate_object_entry(peeled))      /* object packed? */
1712                 add_object_entry(sha1, OBJ_TAG, NULL, 0);
1713         return 0;
1714 }
1715
1716 static void prepare_pack(int window, int depth)
1717 {
1718         struct object_entry **delta_list;
1719         uint32_t i, nr_deltas;
1720         unsigned n;
1721
1722         get_object_details();
1723
1724         /*
1725          * If we're locally repacking then we need to be doubly careful
1726          * from now on in order to make sure no stealth corruption gets
1727          * propagated to the new pack.  Clients receiving streamed packs
1728          * should validate everything they get anyway so no need to incur
1729          * the additional cost here in that case.
1730          */
1731         if (!pack_to_stdout)
1732                 do_check_packed_object_crc = 1;
1733
1734         if (!nr_objects || !window || !depth)
1735                 return;
1736
1737         delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1738         nr_deltas = n = 0;
1739
1740         for (i = 0; i < nr_objects; i++) {
1741                 struct object_entry *entry = objects + i;
1742
1743                 if (entry->delta)
1744                         /* This happens if we decided to reuse existing
1745                          * delta from a pack.  "reuse_delta &&" is implied.
1746                          */
1747                         continue;
1748
1749                 if (entry->size < 50)
1750                         continue;
1751
1752                 if (entry->no_try_delta)
1753                         continue;
1754
1755                 if (!entry->preferred_base) {
1756                         nr_deltas++;
1757                         if (entry->type < 0)
1758                                 die("unable to get type of object %s",
1759                                     sha1_to_hex(entry->idx.sha1));
1760                 } else {
1761                         if (entry->type < 0) {
1762                                 /*
1763                                  * This object is not found, but we
1764                                  * don't have to include it anyway.
1765                                  */
1766                                 continue;
1767                         }
1768                 }
1769
1770                 delta_list[n++] = entry;
1771         }
1772
1773         if (nr_deltas && n > 1) {
1774                 unsigned nr_done = 0;
1775                 if (progress)
1776                         progress_state = start_progress("Compressing objects",
1777                                                         nr_deltas);
1778                 qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
1779                 ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
1780                 stop_progress(&progress_state);
1781                 if (nr_done != nr_deltas)
1782                         die("inconsistency with delta count");
1783         }
1784         free(delta_list);
1785 }
1786
1787 static int git_pack_config(const char *k, const char *v, void *cb)
1788 {
1789         if(!strcmp(k, "pack.window")) {
1790                 window = git_config_int(k, v);
1791                 return 0;
1792         }
1793         if (!strcmp(k, "pack.windowmemory")) {
1794                 window_memory_limit = git_config_ulong(k, v);
1795                 return 0;
1796         }
1797         if (!strcmp(k, "pack.depth")) {
1798                 depth = git_config_int(k, v);
1799                 return 0;
1800         }
1801         if (!strcmp(k, "pack.compression")) {
1802                 int level = git_config_int(k, v);
1803                 if (level == -1)
1804                         level = Z_DEFAULT_COMPRESSION;
1805                 else if (level < 0 || level > Z_BEST_COMPRESSION)
1806                         die("bad pack compression level %d", level);
1807                 pack_compression_level = level;
1808                 pack_compression_seen = 1;
1809                 return 0;
1810         }
1811         if (!strcmp(k, "pack.deltacachesize")) {
1812                 max_delta_cache_size = git_config_int(k, v);
1813                 return 0;
1814         }
1815         if (!strcmp(k, "pack.deltacachelimit")) {
1816                 cache_max_small_delta_size = git_config_int(k, v);
1817                 return 0;
1818         }
1819         if (!strcmp(k, "pack.threads")) {
1820                 delta_search_threads = git_config_int(k, v);
1821                 if (delta_search_threads < 0)
1822                         die("invalid number of threads specified (%d)",
1823                             delta_search_threads);
1824 #ifndef THREADED_DELTA_SEARCH
1825                 if (delta_search_threads != 1)
1826                         warning("no threads support, ignoring %s", k);
1827 #endif
1828                 return 0;
1829         }
1830         if (!strcmp(k, "pack.indexversion")) {
1831                 pack_idx_default_version = git_config_int(k, v);
1832                 if (pack_idx_default_version > 2)
1833                         die("bad pack.indexversion=%"PRIu32,
1834                                 pack_idx_default_version);
1835                 return 0;
1836         }
1837         if (!strcmp(k, "pack.packsizelimit")) {
1838                 pack_size_limit_cfg = git_config_ulong(k, v);
1839                 return 0;
1840         }
1841         return git_default_config(k, v, cb);
1842 }
1843
1844 static void read_object_list_from_stdin(void)
1845 {
1846         char line[40 + 1 + PATH_MAX + 2];
1847         unsigned char sha1[20];
1848
1849         for (;;) {
1850                 if (!fgets(line, sizeof(line), stdin)) {
1851                         if (feof(stdin))
1852                                 break;
1853                         if (!ferror(stdin))
1854                                 die("fgets returned NULL, not EOF, not error!");
1855                         if (errno != EINTR)
1856                                 die("fgets: %s", strerror(errno));
1857                         clearerr(stdin);
1858                         continue;
1859                 }
1860                 if (line[0] == '-') {
1861                         if (get_sha1_hex(line+1, sha1))
1862                                 die("expected edge sha1, got garbage:\n %s",
1863                                     line);
1864                         add_preferred_base(sha1);
1865                         continue;
1866                 }
1867                 if (get_sha1_hex(line, sha1))
1868                         die("expected sha1, got garbage:\n %s", line);
1869
1870                 add_preferred_base_object(line+41);
1871                 add_object_entry(sha1, 0, line+41, 0);
1872         }
1873 }
1874
1875 #define OBJECT_ADDED (1u<<20)
1876
1877 static void show_commit(struct commit *commit)
1878 {
1879         add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
1880         commit->object.flags |= OBJECT_ADDED;
1881 }
1882
1883 static void show_object(struct object_array_entry *p)
1884 {
1885         add_preferred_base_object(p->name);
1886         add_object_entry(p->item->sha1, p->item->type, p->name, 0);
1887         p->item->flags |= OBJECT_ADDED;
1888 }
1889
1890 static void show_edge(struct commit *commit)
1891 {
1892         add_preferred_base(commit->object.sha1);
1893 }
1894
1895 struct in_pack_object {
1896         off_t offset;
1897         struct object *object;
1898 };
1899
1900 struct in_pack {
1901         int alloc;
1902         int nr;
1903         struct in_pack_object *array;
1904 };
1905
1906 static void mark_in_pack_object(struct object *object, struct packed_git *p, struct in_pack *in_pack)
1907 {
1908         in_pack->array[in_pack->nr].offset = find_pack_entry_one(object->sha1, p);
1909         in_pack->array[in_pack->nr].object = object;
1910         in_pack->nr++;
1911 }
1912
1913 /*
1914  * Compare the objects in the offset order, in order to emulate the
1915  * "git rev-list --objects" output that produced the pack originally.
1916  */
1917 static int ofscmp(const void *a_, const void *b_)
1918 {
1919         struct in_pack_object *a = (struct in_pack_object *)a_;
1920         struct in_pack_object *b = (struct in_pack_object *)b_;
1921
1922         if (a->offset < b->offset)
1923                 return -1;
1924         else if (a->offset > b->offset)
1925                 return 1;
1926         else
1927                 return hashcmp(a->object->sha1, b->object->sha1);
1928 }
1929
1930 static void add_objects_in_unpacked_packs(struct rev_info *revs)
1931 {
1932         struct packed_git *p;
1933         struct in_pack in_pack;
1934         uint32_t i;
1935
1936         memset(&in_pack, 0, sizeof(in_pack));
1937
1938         for (p = packed_git; p; p = p->next) {
1939                 const unsigned char *sha1;
1940                 struct object *o;
1941
1942                 for (i = 0; i < revs->num_ignore_packed; i++) {
1943                         if (matches_pack_name(p, revs->ignore_packed[i]))
1944                                 break;
1945                 }
1946                 if (revs->num_ignore_packed <= i)
1947                         continue;
1948                 if (open_pack_index(p))
1949                         die("cannot open pack index");
1950
1951                 ALLOC_GROW(in_pack.array,
1952                            in_pack.nr + p->num_objects,
1953                            in_pack.alloc);
1954
1955                 for (i = 0; i < p->num_objects; i++) {
1956                         sha1 = nth_packed_object_sha1(p, i);
1957                         o = lookup_unknown_object(sha1);
1958                         if (!(o->flags & OBJECT_ADDED))
1959                                 mark_in_pack_object(o, p, &in_pack);
1960                         o->flags |= OBJECT_ADDED;
1961                 }
1962         }
1963
1964         if (in_pack.nr) {
1965                 qsort(in_pack.array, in_pack.nr, sizeof(in_pack.array[0]),
1966                       ofscmp);
1967                 for (i = 0; i < in_pack.nr; i++) {
1968                         struct object *o = in_pack.array[i].object;
1969                         add_object_entry(o->sha1, o->type, "", 0);
1970                 }
1971         }
1972         free(in_pack.array);
1973 }
1974
1975 static void loosen_unused_packed_objects(struct rev_info *revs)
1976 {
1977         struct packed_git *p;
1978         uint32_t i;
1979         const unsigned char *sha1;
1980
1981         for (p = packed_git; p; p = p->next) {
1982                 for (i = 0; i < revs->num_ignore_packed; i++) {
1983                         if (matches_pack_name(p, revs->ignore_packed[i]))
1984                                 break;
1985                 }
1986                 if (revs->num_ignore_packed <= i)
1987                         continue;
1988
1989                 if (open_pack_index(p))
1990                         die("cannot open pack index");
1991
1992                 for (i = 0; i < p->num_objects; i++) {
1993                         sha1 = nth_packed_object_sha1(p, i);
1994                         if (!locate_object_entry(sha1))
1995                                 if (force_object_loose(sha1, p->mtime))
1996                                         die("unable to force loose object");
1997                 }
1998         }
1999 }
2000
2001 static void get_object_list(int ac, const char **av)
2002 {
2003         struct rev_info revs;
2004         char line[1000];
2005         int flags = 0;
2006
2007         init_revisions(&revs, NULL);
2008         save_commit_buffer = 0;
2009         setup_revisions(ac, av, &revs, NULL);
2010
2011         while (fgets(line, sizeof(line), stdin) != NULL) {
2012                 int len = strlen(line);
2013                 if (len && line[len - 1] == '\n')
2014                         line[--len] = 0;
2015                 if (!len)
2016                         break;
2017                 if (*line == '-') {
2018                         if (!strcmp(line, "--not")) {
2019                                 flags ^= UNINTERESTING;
2020                                 continue;
2021                         }
2022                         die("not a rev '%s'", line);
2023                 }
2024                 if (handle_revision_arg(line, &revs, flags, 1))
2025                         die("bad revision '%s'", line);
2026         }
2027
2028         if (prepare_revision_walk(&revs))
2029                 die("revision walk setup failed");
2030         mark_edges_uninteresting(revs.commits, &revs, show_edge);
2031         traverse_commit_list(&revs, show_commit, show_object);
2032
2033         if (keep_unreachable)
2034                 add_objects_in_unpacked_packs(&revs);
2035         if (unpack_unreachable)
2036                 loosen_unused_packed_objects(&revs);
2037 }
2038
2039 static int adjust_perm(const char *path, mode_t mode)
2040 {
2041         if (chmod(path, mode))
2042                 return -1;
2043         return adjust_shared_perm(path);
2044 }
2045
2046 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
2047 {
2048         int use_internal_rev_list = 0;
2049         int thin = 0;
2050         uint32_t i;
2051         const char **rp_av;
2052         int rp_ac_alloc = 64;
2053         int rp_ac;
2054
2055         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
2056
2057         rp_av[0] = "pack-objects";
2058         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
2059         rp_ac = 2;
2060
2061         git_config(git_pack_config, NULL);
2062         if (!pack_compression_seen && core_compression_seen)
2063                 pack_compression_level = core_compression_level;
2064
2065         progress = isatty(2);
2066         for (i = 1; i < argc; i++) {
2067                 const char *arg = argv[i];
2068
2069                 if (*arg != '-')
2070                         break;
2071
2072                 if (!strcmp("--non-empty", arg)) {
2073                         non_empty = 1;
2074                         continue;
2075                 }
2076                 if (!strcmp("--local", arg)) {
2077                         local = 1;
2078                         continue;
2079                 }
2080                 if (!strcmp("--incremental", arg)) {
2081                         incremental = 1;
2082                         continue;
2083                 }
2084                 if (!prefixcmp(arg, "--compression=")) {
2085                         char *end;
2086                         int level = strtoul(arg+14, &end, 0);
2087                         if (!arg[14] || *end)
2088                                 usage(pack_usage);
2089                         if (level == -1)
2090                                 level = Z_DEFAULT_COMPRESSION;
2091                         else if (level < 0 || level > Z_BEST_COMPRESSION)
2092                                 die("bad pack compression level %d", level);
2093                         pack_compression_level = level;
2094                         continue;
2095                 }
2096                 if (!prefixcmp(arg, "--max-pack-size=")) {
2097                         char *end;
2098                         pack_size_limit_cfg = 0;
2099                         pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024;
2100                         if (!arg[16] || *end)
2101                                 usage(pack_usage);
2102                         continue;
2103                 }
2104                 if (!prefixcmp(arg, "--window=")) {
2105                         char *end;
2106                         window = strtoul(arg+9, &end, 0);
2107                         if (!arg[9] || *end)
2108                                 usage(pack_usage);
2109                         continue;
2110                 }
2111                 if (!prefixcmp(arg, "--window-memory=")) {
2112                         if (!git_parse_ulong(arg+16, &window_memory_limit))
2113                                 usage(pack_usage);
2114                         continue;
2115                 }
2116                 if (!prefixcmp(arg, "--threads=")) {
2117                         char *end;
2118                         delta_search_threads = strtoul(arg+10, &end, 0);
2119                         if (!arg[10] || *end || delta_search_threads < 0)
2120                                 usage(pack_usage);
2121 #ifndef THREADED_DELTA_SEARCH
2122                         if (delta_search_threads != 1)
2123                                 warning("no threads support, "
2124                                         "ignoring %s", arg);
2125 #endif
2126                         continue;
2127                 }
2128                 if (!prefixcmp(arg, "--depth=")) {
2129                         char *end;
2130                         depth = strtoul(arg+8, &end, 0);
2131                         if (!arg[8] || *end)
2132                                 usage(pack_usage);
2133                         continue;
2134                 }
2135                 if (!strcmp("--progress", arg)) {
2136                         progress = 1;
2137                         continue;
2138                 }
2139                 if (!strcmp("--all-progress", arg)) {
2140                         progress = 2;
2141                         continue;
2142                 }
2143                 if (!strcmp("-q", arg)) {
2144                         progress = 0;
2145                         continue;
2146                 }
2147                 if (!strcmp("--no-reuse-delta", arg)) {
2148                         reuse_delta = 0;
2149                         continue;
2150                 }
2151                 if (!strcmp("--no-reuse-object", arg)) {
2152                         reuse_object = reuse_delta = 0;
2153                         continue;
2154                 }
2155                 if (!strcmp("--delta-base-offset", arg)) {
2156                         allow_ofs_delta = 1;
2157                         continue;
2158                 }
2159                 if (!strcmp("--stdout", arg)) {
2160                         pack_to_stdout = 1;
2161                         continue;
2162                 }
2163                 if (!strcmp("--revs", arg)) {
2164                         use_internal_rev_list = 1;
2165                         continue;
2166                 }
2167                 if (!strcmp("--keep-unreachable", arg)) {
2168                         keep_unreachable = 1;
2169                         continue;
2170                 }
2171                 if (!strcmp("--unpack-unreachable", arg)) {
2172                         unpack_unreachable = 1;
2173                         continue;
2174                 }
2175                 if (!strcmp("--include-tag", arg)) {
2176                         include_tag = 1;
2177                         continue;
2178                 }
2179                 if (!strcmp("--unpacked", arg) ||
2180                     !prefixcmp(arg, "--unpacked=") ||
2181                     !strcmp("--reflog", arg) ||
2182                     !strcmp("--all", arg)) {
2183                         use_internal_rev_list = 1;
2184                         if (rp_ac >= rp_ac_alloc - 1) {
2185                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
2186                                 rp_av = xrealloc(rp_av,
2187                                                  rp_ac_alloc * sizeof(*rp_av));
2188                         }
2189                         rp_av[rp_ac++] = arg;
2190                         continue;
2191                 }
2192                 if (!strcmp("--thin", arg)) {
2193                         use_internal_rev_list = 1;
2194                         thin = 1;
2195                         rp_av[1] = "--objects-edge";
2196                         continue;
2197                 }
2198                 if (!prefixcmp(arg, "--index-version=")) {
2199                         char *c;
2200                         pack_idx_default_version = strtoul(arg + 16, &c, 10);
2201                         if (pack_idx_default_version > 2)
2202                                 die("bad %s", arg);
2203                         if (*c == ',')
2204                                 pack_idx_off32_limit = strtoul(c+1, &c, 0);
2205                         if (*c || pack_idx_off32_limit & 0x80000000)
2206                                 die("bad %s", arg);
2207                         continue;
2208                 }
2209                 usage(pack_usage);
2210         }
2211
2212         /* Traditionally "pack-objects [options] base extra" failed;
2213          * we would however want to take refs parameter that would
2214          * have been given to upstream rev-list ourselves, which means
2215          * we somehow want to say what the base name is.  So the
2216          * syntax would be:
2217          *
2218          * pack-objects [options] base <refs...>
2219          *
2220          * in other words, we would treat the first non-option as the
2221          * base_name and send everything else to the internal revision
2222          * walker.
2223          */
2224
2225         if (!pack_to_stdout)
2226                 base_name = argv[i++];
2227
2228         if (pack_to_stdout != !base_name)
2229                 usage(pack_usage);
2230
2231         if (!pack_to_stdout && !pack_size_limit)
2232                 pack_size_limit = pack_size_limit_cfg;
2233
2234         if (pack_to_stdout && pack_size_limit)
2235                 die("--max-pack-size cannot be used to build a pack for transfer.");
2236
2237         if (!pack_to_stdout && thin)
2238                 die("--thin cannot be used to build an indexable pack.");
2239
2240         if (keep_unreachable && unpack_unreachable)
2241                 die("--keep-unreachable and --unpack-unreachable are incompatible.");
2242
2243 #ifdef THREADED_DELTA_SEARCH
2244         if (!delta_search_threads)      /* --threads=0 means autodetect */
2245                 delta_search_threads = online_cpus();
2246 #endif
2247
2248         prepare_packed_git();
2249
2250         if (progress)
2251                 progress_state = start_progress("Counting objects", 0);
2252         if (!use_internal_rev_list)
2253                 read_object_list_from_stdin();
2254         else {
2255                 rp_av[rp_ac] = NULL;
2256                 get_object_list(rp_ac, rp_av);
2257         }
2258         if (include_tag && nr_result)
2259                 for_each_ref(add_ref_tag, NULL);
2260         stop_progress(&progress_state);
2261
2262         if (non_empty && !nr_result)
2263                 return 0;
2264         if (nr_result)
2265                 prepare_pack(window, depth);
2266         write_pack_file();
2267         if (progress)
2268                 fprintf(stderr, "Total %"PRIu32" (delta %"PRIu32"),"
2269                         " reused %"PRIu32" (delta %"PRIu32")\n",
2270                         written, written_delta, reused, reused_delta);
2271         return 0;
2272 }