builtin / pack-objects.con commit Merge branch 'rs/checkout-init-macro' (e683f17)
   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 "pack-objects.h"
  18#include "progress.h"
  19#include "refs.h"
  20#include "streaming.h"
  21#include "thread-utils.h"
  22#include "pack-bitmap.h"
  23#include "reachable.h"
  24#include "sha1-array.h"
  25#include "argv-array.h"
  26
  27static const char *pack_usage[] = {
  28        N_("git pack-objects --stdout [<options>...] [< <ref-list> | < <object-list>]"),
  29        N_("git pack-objects [<options>...] <base-name> [< <ref-list> | < <object-list>]"),
  30        NULL
  31};
  32
  33/*
  34 * Objects we are going to pack are collected in the `to_pack` structure.
  35 * It contains an array (dynamically expanded) of the object data, and a map
  36 * that can resolve SHA1s to their position in the array.
  37 */
  38static struct packing_data to_pack;
  39
  40static struct pack_idx_entry **written_list;
  41static uint32_t nr_result, nr_written;
  42
  43static int non_empty;
  44static int reuse_delta = 1, reuse_object = 1;
  45static int keep_unreachable, unpack_unreachable, include_tag;
  46static unsigned long unpack_unreachable_expiration;
  47static int pack_loose_unreachable;
  48static int local;
  49static int have_non_local_packs;
  50static int incremental;
  51static int ignore_packed_keep;
  52static int allow_ofs_delta;
  53static struct pack_idx_option pack_idx_opts;
  54static const char *base_name;
  55static int progress = 1;
  56static int window = 10;
  57static unsigned long pack_size_limit;
  58static int depth = 50;
  59static int delta_search_threads;
  60static int pack_to_stdout;
  61static int num_preferred_base;
  62static struct progress *progress_state;
  63static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  64static int pack_compression_seen;
  65
  66static struct packed_git *reuse_packfile;
  67static uint32_t reuse_packfile_objects;
  68static off_t reuse_packfile_offset;
  69
  70static int use_bitmap_index_default = 1;
  71static int use_bitmap_index = -1;
  72static int write_bitmap_index;
  73static uint16_t write_bitmap_options;
  74
  75static unsigned long delta_cache_size = 0;
  76static unsigned long max_delta_cache_size = 256 * 1024 * 1024;
  77static unsigned long cache_max_small_delta_size = 1000;
  78
  79static unsigned long window_memory_limit = 0;
  80
  81/*
  82 * stats
  83 */
  84static uint32_t written, written_delta;
  85static uint32_t reused, reused_delta;
  86
  87/*
  88 * Indexed commits
  89 */
  90static struct commit **indexed_commits;
  91static unsigned int indexed_commits_nr;
  92static unsigned int indexed_commits_alloc;
  93
  94static void index_commit_for_bitmap(struct commit *commit)
  95{
  96        if (indexed_commits_nr >= indexed_commits_alloc) {
  97                indexed_commits_alloc = (indexed_commits_alloc + 32) * 2;
  98                REALLOC_ARRAY(indexed_commits, indexed_commits_alloc);
  99        }
 100
 101        indexed_commits[indexed_commits_nr++] = commit;
 102}
 103
 104static void *get_delta(struct object_entry *entry)
 105{
 106        unsigned long size, base_size, delta_size;
 107        void *buf, *base_buf, *delta_buf;
 108        enum object_type type;
 109
 110        buf = read_sha1_file(entry->idx.sha1, &type, &size);
 111        if (!buf)
 112                die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 113        base_buf = read_sha1_file(entry->delta->idx.sha1, &type, &base_size);
 114        if (!base_buf)
 115                die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 116        delta_buf = diff_delta(base_buf, base_size,
 117                               buf, size, &delta_size, 0);
 118        if (!delta_buf || delta_size != entry->delta_size)
 119                die("delta size changed");
 120        free(buf);
 121        free(base_buf);
 122        return delta_buf;
 123}
 124
 125static unsigned long do_compress(void **pptr, unsigned long size)
 126{
 127        git_zstream stream;
 128        void *in, *out;
 129        unsigned long maxsize;
 130
 131        git_deflate_init(&stream, pack_compression_level);
 132        maxsize = git_deflate_bound(&stream, size);
 133
 134        in = *pptr;
 135        out = xmalloc(maxsize);
 136        *pptr = out;
 137
 138        stream.next_in = in;
 139        stream.avail_in = size;
 140        stream.next_out = out;
 141        stream.avail_out = maxsize;
 142        while (git_deflate(&stream, Z_FINISH) == Z_OK)
 143                ; /* nothing */
 144        git_deflate_end(&stream);
 145
 146        free(in);
 147        return stream.total_out;
 148}
 149
 150static unsigned long write_large_blob_data(struct git_istream *st, struct sha1file *f,
 151                                           const unsigned char *sha1)
 152{
 153        git_zstream stream;
 154        unsigned char ibuf[1024 * 16];
 155        unsigned char obuf[1024 * 16];
 156        unsigned long olen = 0;
 157
 158        git_deflate_init(&stream, pack_compression_level);
 159
 160        for (;;) {
 161                ssize_t readlen;
 162                int zret = Z_OK;
 163                readlen = read_istream(st, ibuf, sizeof(ibuf));
 164                if (readlen == -1)
 165                        die(_("unable to read %s"), sha1_to_hex(sha1));
 166
 167                stream.next_in = ibuf;
 168                stream.avail_in = readlen;
 169                while ((stream.avail_in || readlen == 0) &&
 170                       (zret == Z_OK || zret == Z_BUF_ERROR)) {
 171                        stream.next_out = obuf;
 172                        stream.avail_out = sizeof(obuf);
 173                        zret = git_deflate(&stream, readlen ? 0 : Z_FINISH);
 174                        sha1write(f, obuf, stream.next_out - obuf);
 175                        olen += stream.next_out - obuf;
 176                }
 177                if (stream.avail_in)
 178                        die(_("deflate error (%d)"), zret);
 179                if (readlen == 0) {
 180                        if (zret != Z_STREAM_END)
 181                                die(_("deflate error (%d)"), zret);
 182                        break;
 183                }
 184        }
 185        git_deflate_end(&stream);
 186        return olen;
 187}
 188
 189/*
 190 * we are going to reuse the existing object data as is.  make
 191 * sure it is not corrupt.
 192 */
 193static int check_pack_inflate(struct packed_git *p,
 194                struct pack_window **w_curs,
 195                off_t offset,
 196                off_t len,
 197                unsigned long expect)
 198{
 199        git_zstream stream;
 200        unsigned char fakebuf[4096], *in;
 201        int st;
 202
 203        memset(&stream, 0, sizeof(stream));
 204        git_inflate_init(&stream);
 205        do {
 206                in = use_pack(p, w_curs, offset, &stream.avail_in);
 207                stream.next_in = in;
 208                stream.next_out = fakebuf;
 209                stream.avail_out = sizeof(fakebuf);
 210                st = git_inflate(&stream, Z_FINISH);
 211                offset += stream.next_in - in;
 212        } while (st == Z_OK || st == Z_BUF_ERROR);
 213        git_inflate_end(&stream);
 214        return (st == Z_STREAM_END &&
 215                stream.total_out == expect &&
 216                stream.total_in == len) ? 0 : -1;
 217}
 218
 219static void copy_pack_data(struct sha1file *f,
 220                struct packed_git *p,
 221                struct pack_window **w_curs,
 222                off_t offset,
 223                off_t len)
 224{
 225        unsigned char *in;
 226        unsigned long avail;
 227
 228        while (len) {
 229                in = use_pack(p, w_curs, offset, &avail);
 230                if (avail > len)
 231                        avail = (unsigned long)len;
 232                sha1write(f, in, avail);
 233                offset += avail;
 234                len -= avail;
 235        }
 236}
 237
 238/* Return 0 if we will bust the pack-size limit */
 239static unsigned long write_no_reuse_object(struct sha1file *f, struct object_entry *entry,
 240                                           unsigned long limit, int usable_delta)
 241{
 242        unsigned long size, datalen;
 243        unsigned char header[10], dheader[10];
 244        unsigned hdrlen;
 245        enum object_type type;
 246        void *buf;
 247        struct git_istream *st = NULL;
 248
 249        if (!usable_delta) {
 250                if (entry->type == OBJ_BLOB &&
 251                    entry->size > big_file_threshold &&
 252                    (st = open_istream(entry->idx.sha1, &type, &size, NULL)) != NULL)
 253                        buf = NULL;
 254                else {
 255                        buf = read_sha1_file(entry->idx.sha1, &type, &size);
 256                        if (!buf)
 257                                die(_("unable to read %s"), sha1_to_hex(entry->idx.sha1));
 258                }
 259                /*
 260                 * make sure no cached delta data remains from a
 261                 * previous attempt before a pack split occurred.
 262                 */
 263                free(entry->delta_data);
 264                entry->delta_data = NULL;
 265                entry->z_delta_size = 0;
 266        } else if (entry->delta_data) {
 267                size = entry->delta_size;
 268                buf = entry->delta_data;
 269                entry->delta_data = NULL;
 270                type = (allow_ofs_delta && entry->delta->idx.offset) ?
 271                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 272        } else {
 273                buf = get_delta(entry);
 274                size = entry->delta_size;
 275                type = (allow_ofs_delta && entry->delta->idx.offset) ?
 276                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 277        }
 278
 279        if (st) /* large blob case, just assume we don't compress well */
 280                datalen = size;
 281        else if (entry->z_delta_size)
 282                datalen = entry->z_delta_size;
 283        else
 284                datalen = do_compress(&buf, size);
 285
 286        /*
 287         * The object header is a byte of 'type' followed by zero or
 288         * more bytes of length.
 289         */
 290        hdrlen = encode_in_pack_object_header(type, size, header);
 291
 292        if (type == OBJ_OFS_DELTA) {
 293                /*
 294                 * Deltas with relative base contain an additional
 295                 * encoding of the relative offset for the delta
 296                 * base from this object's position in the pack.
 297                 */
 298                off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 299                unsigned pos = sizeof(dheader) - 1;
 300                dheader[pos] = ofs & 127;
 301                while (ofs >>= 7)
 302                        dheader[--pos] = 128 | (--ofs & 127);
 303                if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 304                        if (st)
 305                                close_istream(st);
 306                        free(buf);
 307                        return 0;
 308                }
 309                sha1write(f, header, hdrlen);
 310                sha1write(f, dheader + pos, sizeof(dheader) - pos);
 311                hdrlen += sizeof(dheader) - pos;
 312        } else if (type == OBJ_REF_DELTA) {
 313                /*
 314                 * Deltas with a base reference contain
 315                 * an additional 20 bytes for the base sha1.
 316                 */
 317                if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 318                        if (st)
 319                                close_istream(st);
 320                        free(buf);
 321                        return 0;
 322                }
 323                sha1write(f, header, hdrlen);
 324                sha1write(f, entry->delta->idx.sha1, 20);
 325                hdrlen += 20;
 326        } else {
 327                if (limit && hdrlen + datalen + 20 >= limit) {
 328                        if (st)
 329                                close_istream(st);
 330                        free(buf);
 331                        return 0;
 332                }
 333                sha1write(f, header, hdrlen);
 334        }
 335        if (st) {
 336                datalen = write_large_blob_data(st, f, entry->idx.sha1);
 337                close_istream(st);
 338        } else {
 339                sha1write(f, buf, datalen);
 340                free(buf);
 341        }
 342
 343        return hdrlen + datalen;
 344}
 345
 346/* Return 0 if we will bust the pack-size limit */
 347static off_t write_reuse_object(struct sha1file *f, struct object_entry *entry,
 348                                unsigned long limit, int usable_delta)
 349{
 350        struct packed_git *p = entry->in_pack;
 351        struct pack_window *w_curs = NULL;
 352        struct revindex_entry *revidx;
 353        off_t offset;
 354        enum object_type type = entry->type;
 355        off_t datalen;
 356        unsigned char header[10], dheader[10];
 357        unsigned hdrlen;
 358
 359        if (entry->delta)
 360                type = (allow_ofs_delta && entry->delta->idx.offset) ?
 361                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 362        hdrlen = encode_in_pack_object_header(type, entry->size, header);
 363
 364        offset = entry->in_pack_offset;
 365        revidx = find_pack_revindex(p, offset);
 366        datalen = revidx[1].offset - offset;
 367        if (!pack_to_stdout && p->index_version > 1 &&
 368            check_pack_crc(p, &w_curs, offset, datalen, revidx->nr)) {
 369                error("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 370                unuse_pack(&w_curs);
 371                return write_no_reuse_object(f, entry, limit, usable_delta);
 372        }
 373
 374        offset += entry->in_pack_header_size;
 375        datalen -= entry->in_pack_header_size;
 376
 377        if (!pack_to_stdout && p->index_version == 1 &&
 378            check_pack_inflate(p, &w_curs, offset, datalen, entry->size)) {
 379                error("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 380                unuse_pack(&w_curs);
 381                return write_no_reuse_object(f, entry, limit, usable_delta);
 382        }
 383
 384        if (type == OBJ_OFS_DELTA) {
 385                off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 386                unsigned pos = sizeof(dheader) - 1;
 387                dheader[pos] = ofs & 127;
 388                while (ofs >>= 7)
 389                        dheader[--pos] = 128 | (--ofs & 127);
 390                if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 391                        unuse_pack(&w_curs);
 392                        return 0;
 393                }
 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                        unuse_pack(&w_curs);
 401                        return 0;
 402                }
 403                sha1write(f, header, hdrlen);
 404                sha1write(f, entry->delta->idx.sha1, 20);
 405                hdrlen += 20;
 406                reused_delta++;
 407        } else {
 408                if (limit && hdrlen + datalen + 20 >= limit) {
 409                        unuse_pack(&w_curs);
 410                        return 0;
 411                }
 412                sha1write(f, header, hdrlen);
 413        }
 414        copy_pack_data(f, p, &w_curs, offset, datalen);
 415        unuse_pack(&w_curs);
 416        reused++;
 417        return hdrlen + datalen;
 418}
 419
 420/* Return 0 if we will bust the pack-size limit */
 421static off_t write_object(struct sha1file *f,
 422                          struct object_entry *entry,
 423                          off_t write_offset)
 424{
 425        unsigned long limit;
 426        off_t len;
 427        int usable_delta, to_reuse;
 428
 429        if (!pack_to_stdout)
 430                crc32_begin(f);
 431
 432        /* apply size limit if limited packsize and not first object */
 433        if (!pack_size_limit || !nr_written)
 434                limit = 0;
 435        else if (pack_size_limit <= write_offset)
 436                /*
 437                 * the earlier object did not fit the limit; avoid
 438                 * mistaking this with unlimited (i.e. limit = 0).
 439                 */
 440                limit = 1;
 441        else
 442                limit = pack_size_limit - write_offset;
 443
 444        if (!entry->delta)
 445                usable_delta = 0;       /* no delta */
 446        else if (!pack_size_limit)
 447               usable_delta = 1;        /* unlimited packfile */
 448        else if (entry->delta->idx.offset == (off_t)-1)
 449                usable_delta = 0;       /* base was written to another pack */
 450        else if (entry->delta->idx.offset)
 451                usable_delta = 1;       /* base already exists in this pack */
 452        else
 453                usable_delta = 0;       /* base could end up in another pack */
 454
 455        if (!reuse_object)
 456                to_reuse = 0;   /* explicit */
 457        else if (!entry->in_pack)
 458                to_reuse = 0;   /* can't reuse what we don't have */
 459        else if (entry->type == OBJ_REF_DELTA || entry->type == OBJ_OFS_DELTA)
 460                                /* check_object() decided it for us ... */
 461                to_reuse = usable_delta;
 462                                /* ... but pack split may override that */
 463        else if (entry->type != entry->in_pack_type)
 464                to_reuse = 0;   /* pack has delta which is unusable */
 465        else if (entry->delta)
 466                to_reuse = 0;   /* we want to pack afresh */
 467        else
 468                to_reuse = 1;   /* we have it in-pack undeltified,
 469                                 * and we do not need to deltify it.
 470                                 */
 471
 472        if (!to_reuse)
 473                len = write_no_reuse_object(f, entry, limit, usable_delta);
 474        else
 475                len = write_reuse_object(f, entry, limit, usable_delta);
 476        if (!len)
 477                return 0;
 478
 479        if (usable_delta)
 480                written_delta++;
 481        written++;
 482        if (!pack_to_stdout)
 483                entry->idx.crc32 = crc32_end(f);
 484        return len;
 485}
 486
 487enum write_one_status {
 488        WRITE_ONE_SKIP = -1, /* already written */
 489        WRITE_ONE_BREAK = 0, /* writing this will bust the limit; not written */
 490        WRITE_ONE_WRITTEN = 1, /* normal */
 491        WRITE_ONE_RECURSIVE = 2 /* already scheduled to be written */
 492};
 493
 494static enum write_one_status write_one(struct sha1file *f,
 495                                       struct object_entry *e,
 496                                       off_t *offset)
 497{
 498        off_t size;
 499        int recursing;
 500
 501        /*
 502         * we set offset to 1 (which is an impossible value) to mark
 503         * the fact that this object is involved in "write its base
 504         * first before writing a deltified object" recursion.
 505         */
 506        recursing = (e->idx.offset == 1);
 507        if (recursing) {
 508                warning("recursive delta detected for object %s",
 509                        sha1_to_hex(e->idx.sha1));
 510                return WRITE_ONE_RECURSIVE;
 511        } else if (e->idx.offset || e->preferred_base) {
 512                /* offset is non zero if object is written already. */
 513                return WRITE_ONE_SKIP;
 514        }
 515
 516        /* if we are deltified, write out base object first. */
 517        if (e->delta) {
 518                e->idx.offset = 1; /* now recurse */
 519                switch (write_one(f, e->delta, offset)) {
 520                case WRITE_ONE_RECURSIVE:
 521                        /* we cannot depend on this one */
 522                        e->delta = NULL;
 523                        break;
 524                default:
 525                        break;
 526                case WRITE_ONE_BREAK:
 527                        e->idx.offset = recursing;
 528                        return WRITE_ONE_BREAK;
 529                }
 530        }
 531
 532        e->idx.offset = *offset;
 533        size = write_object(f, e, *offset);
 534        if (!size) {
 535                e->idx.offset = recursing;
 536                return WRITE_ONE_BREAK;
 537        }
 538        written_list[nr_written++] = &e->idx;
 539
 540        /* make sure off_t is sufficiently large not to wrap */
 541        if (signed_add_overflows(*offset, size))
 542                die("pack too large for current definition of off_t");
 543        *offset += size;
 544        return WRITE_ONE_WRITTEN;
 545}
 546
 547static int mark_tagged(const char *path, const struct object_id *oid, int flag,
 548                       void *cb_data)
 549{
 550        unsigned char peeled[20];
 551        struct object_entry *entry = packlist_find(&to_pack, oid->hash, NULL);
 552
 553        if (entry)
 554                entry->tagged = 1;
 555        if (!peel_ref(path, peeled)) {
 556                entry = packlist_find(&to_pack, peeled, NULL);
 557                if (entry)
 558                        entry->tagged = 1;
 559        }
 560        return 0;
 561}
 562
 563static inline void add_to_write_order(struct object_entry **wo,
 564                               unsigned int *endp,
 565                               struct object_entry *e)
 566{
 567        if (e->filled)
 568                return;
 569        wo[(*endp)++] = e;
 570        e->filled = 1;
 571}
 572
 573static void add_descendants_to_write_order(struct object_entry **wo,
 574                                           unsigned int *endp,
 575                                           struct object_entry *e)
 576{
 577        int add_to_order = 1;
 578        while (e) {
 579                if (add_to_order) {
 580                        struct object_entry *s;
 581                        /* add this node... */
 582                        add_to_write_order(wo, endp, e);
 583                        /* all its siblings... */
 584                        for (s = e->delta_sibling; s; s = s->delta_sibling) {
 585                                add_to_write_order(wo, endp, s);
 586                        }
 587                }
 588                /* drop down a level to add left subtree nodes if possible */
 589                if (e->delta_child) {
 590                        add_to_order = 1;
 591                        e = e->delta_child;
 592                } else {
 593                        add_to_order = 0;
 594                        /* our sibling might have some children, it is next */
 595                        if (e->delta_sibling) {
 596                                e = e->delta_sibling;
 597                                continue;
 598                        }
 599                        /* go back to our parent node */
 600                        e = e->delta;
 601                        while (e && !e->delta_sibling) {
 602                                /* we're on the right side of a subtree, keep
 603                                 * going up until we can go right again */
 604                                e = e->delta;
 605                        }
 606                        if (!e) {
 607                                /* done- we hit our original root node */
 608                                return;
 609                        }
 610                        /* pass it off to sibling at this level */
 611                        e = e->delta_sibling;
 612                }
 613        };
 614}
 615
 616static void add_family_to_write_order(struct object_entry **wo,
 617                                      unsigned int *endp,
 618                                      struct object_entry *e)
 619{
 620        struct object_entry *root;
 621
 622        for (root = e; root->delta; root = root->delta)
 623                ; /* nothing */
 624        add_descendants_to_write_order(wo, endp, root);
 625}
 626
 627static struct object_entry **compute_write_order(void)
 628{
 629        unsigned int i, wo_end, last_untagged;
 630
 631        struct object_entry **wo;
 632        struct object_entry *objects = to_pack.objects;
 633
 634        for (i = 0; i < to_pack.nr_objects; i++) {
 635                objects[i].tagged = 0;
 636                objects[i].filled = 0;
 637                objects[i].delta_child = NULL;
 638                objects[i].delta_sibling = NULL;
 639        }
 640
 641        /*
 642         * Fully connect delta_child/delta_sibling network.
 643         * Make sure delta_sibling is sorted in the original
 644         * recency order.
 645         */
 646        for (i = to_pack.nr_objects; i > 0;) {
 647                struct object_entry *e = &objects[--i];
 648                if (!e->delta)
 649                        continue;
 650                /* Mark me as the first child */
 651                e->delta_sibling = e->delta->delta_child;
 652                e->delta->delta_child = e;
 653        }
 654
 655        /*
 656         * Mark objects that are at the tip of tags.
 657         */
 658        for_each_tag_ref(mark_tagged, NULL);
 659
 660        /*
 661         * Give the objects in the original recency order until
 662         * we see a tagged tip.
 663         */
 664        ALLOC_ARRAY(wo, to_pack.nr_objects);
 665        for (i = wo_end = 0; i < to_pack.nr_objects; i++) {
 666                if (objects[i].tagged)
 667                        break;
 668                add_to_write_order(wo, &wo_end, &objects[i]);
 669        }
 670        last_untagged = i;
 671
 672        /*
 673         * Then fill all the tagged tips.
 674         */
 675        for (; i < to_pack.nr_objects; i++) {
 676                if (objects[i].tagged)
 677                        add_to_write_order(wo, &wo_end, &objects[i]);
 678        }
 679
 680        /*
 681         * And then all remaining commits and tags.
 682         */
 683        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 684                if (objects[i].type != OBJ_COMMIT &&
 685                    objects[i].type != OBJ_TAG)
 686                        continue;
 687                add_to_write_order(wo, &wo_end, &objects[i]);
 688        }
 689
 690        /*
 691         * And then all the trees.
 692         */
 693        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 694                if (objects[i].type != OBJ_TREE)
 695                        continue;
 696                add_to_write_order(wo, &wo_end, &objects[i]);
 697        }
 698
 699        /*
 700         * Finally all the rest in really tight order
 701         */
 702        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 703                if (!objects[i].filled)
 704                        add_family_to_write_order(wo, &wo_end, &objects[i]);
 705        }
 706
 707        if (wo_end != to_pack.nr_objects)
 708                die("ordered %u objects, expected %"PRIu32, wo_end, to_pack.nr_objects);
 709
 710        return wo;
 711}
 712
 713static off_t write_reused_pack(struct sha1file *f)
 714{
 715        unsigned char buffer[8192];
 716        off_t to_write, total;
 717        int fd;
 718
 719        if (!is_pack_valid(reuse_packfile))
 720                die("packfile is invalid: %s", reuse_packfile->pack_name);
 721
 722        fd = git_open_noatime(reuse_packfile->pack_name);
 723        if (fd < 0)
 724                die_errno("unable to open packfile for reuse: %s",
 725                          reuse_packfile->pack_name);
 726
 727        if (lseek(fd, sizeof(struct pack_header), SEEK_SET) == -1)
 728                die_errno("unable to seek in reused packfile");
 729
 730        if (reuse_packfile_offset < 0)
 731                reuse_packfile_offset = reuse_packfile->pack_size - 20;
 732
 733        total = to_write = reuse_packfile_offset - sizeof(struct pack_header);
 734
 735        while (to_write) {
 736                int read_pack = xread(fd, buffer, sizeof(buffer));
 737
 738                if (read_pack <= 0)
 739                        die_errno("unable to read from reused packfile");
 740
 741                if (read_pack > to_write)
 742                        read_pack = to_write;
 743
 744                sha1write(f, buffer, read_pack);
 745                to_write -= read_pack;
 746
 747                /*
 748                 * We don't know the actual number of objects written,
 749                 * only how many bytes written, how many bytes total, and
 750                 * how many objects total. So we can fake it by pretending all
 751                 * objects we are writing are the same size. This gives us a
 752                 * smooth progress meter, and at the end it matches the true
 753                 * answer.
 754                 */
 755                written = reuse_packfile_objects *
 756                                (((double)(total - to_write)) / total);
 757                display_progress(progress_state, written);
 758        }
 759
 760        close(fd);
 761        written = reuse_packfile_objects;
 762        display_progress(progress_state, written);
 763        return reuse_packfile_offset - sizeof(struct pack_header);
 764}
 765
 766static const char no_split_warning[] = N_(
 767"disabling bitmap writing, packs are split due to pack.packSizeLimit"
 768);
 769
 770static void write_pack_file(void)
 771{
 772        uint32_t i = 0, j;
 773        struct sha1file *f;
 774        off_t offset;
 775        uint32_t nr_remaining = nr_result;
 776        time_t last_mtime = 0;
 777        struct object_entry **write_order;
 778
 779        if (progress > pack_to_stdout)
 780                progress_state = start_progress(_("Writing objects"), nr_result);
 781        ALLOC_ARRAY(written_list, to_pack.nr_objects);
 782        write_order = compute_write_order();
 783
 784        do {
 785                unsigned char sha1[20];
 786                char *pack_tmp_name = NULL;
 787
 788                if (pack_to_stdout)
 789                        f = sha1fd_throughput(1, "<stdout>", progress_state);
 790                else
 791                        f = create_tmp_packfile(&pack_tmp_name);
 792
 793                offset = write_pack_header(f, nr_remaining);
 794
 795                if (reuse_packfile) {
 796                        off_t packfile_size;
 797                        assert(pack_to_stdout);
 798
 799                        packfile_size = write_reused_pack(f);
 800                        offset += packfile_size;
 801                }
 802
 803                nr_written = 0;
 804                for (; i < to_pack.nr_objects; i++) {
 805                        struct object_entry *e = write_order[i];
 806                        if (write_one(f, e, &offset) == WRITE_ONE_BREAK)
 807                                break;
 808                        display_progress(progress_state, written);
 809                }
 810
 811                /*
 812                 * Did we write the wrong # entries in the header?
 813                 * If so, rewrite it like in fast-import
 814                 */
 815                if (pack_to_stdout) {
 816                        sha1close(f, sha1, CSUM_CLOSE);
 817                } else if (nr_written == nr_remaining) {
 818                        sha1close(f, sha1, CSUM_FSYNC);
 819                } else {
 820                        int fd = sha1close(f, sha1, 0);
 821                        fixup_pack_header_footer(fd, sha1, pack_tmp_name,
 822                                                 nr_written, sha1, offset);
 823                        close(fd);
 824                        if (write_bitmap_index) {
 825                                warning(_(no_split_warning));
 826                                write_bitmap_index = 0;
 827                        }
 828                }
 829
 830                if (!pack_to_stdout) {
 831                        struct stat st;
 832                        struct strbuf tmpname = STRBUF_INIT;
 833
 834                        /*
 835                         * Packs are runtime accessed in their mtime
 836                         * order since newer packs are more likely to contain
 837                         * younger objects.  So if we are creating multiple
 838                         * packs then we should modify the mtime of later ones
 839                         * to preserve this property.
 840                         */
 841                        if (stat(pack_tmp_name, &st) < 0) {
 842                                warning_errno("failed to stat %s", pack_tmp_name);
 843                        } else if (!last_mtime) {
 844                                last_mtime = st.st_mtime;
 845                        } else {
 846                                struct utimbuf utb;
 847                                utb.actime = st.st_atime;
 848                                utb.modtime = --last_mtime;
 849                                if (utime(pack_tmp_name, &utb) < 0)
 850                                        warning_errno("failed utime() on %s", pack_tmp_name);
 851                        }
 852
 853                        strbuf_addf(&tmpname, "%s-", base_name);
 854
 855                        if (write_bitmap_index) {
 856                                bitmap_writer_set_checksum(sha1);
 857                                bitmap_writer_build_type_index(written_list, nr_written);
 858                        }
 859
 860                        finish_tmp_packfile(&tmpname, pack_tmp_name,
 861                                            written_list, nr_written,
 862                                            &pack_idx_opts, sha1);
 863
 864                        if (write_bitmap_index) {
 865                                strbuf_addf(&tmpname, "%s.bitmap", sha1_to_hex(sha1));
 866
 867                                stop_progress(&progress_state);
 868
 869                                bitmap_writer_show_progress(progress);
 870                                bitmap_writer_reuse_bitmaps(&to_pack);
 871                                bitmap_writer_select_commits(indexed_commits, indexed_commits_nr, -1);
 872                                bitmap_writer_build(&to_pack);
 873                                bitmap_writer_finish(written_list, nr_written,
 874                                                     tmpname.buf, write_bitmap_options);
 875                                write_bitmap_index = 0;
 876                        }
 877
 878                        strbuf_release(&tmpname);
 879                        free(pack_tmp_name);
 880                        puts(sha1_to_hex(sha1));
 881                }
 882
 883                /* mark written objects as written to previous pack */
 884                for (j = 0; j < nr_written; j++) {
 885                        written_list[j]->offset = (off_t)-1;
 886                }
 887                nr_remaining -= nr_written;
 888        } while (nr_remaining && i < to_pack.nr_objects);
 889
 890        free(written_list);
 891        free(write_order);
 892        stop_progress(&progress_state);
 893        if (written != nr_result)
 894                die("wrote %"PRIu32" objects while expecting %"PRIu32,
 895                        written, nr_result);
 896}
 897
 898static void setup_delta_attr_check(struct git_attr_check *check)
 899{
 900        static struct git_attr *attr_delta;
 901
 902        if (!attr_delta)
 903                attr_delta = git_attr("delta");
 904
 905        check[0].attr = attr_delta;
 906}
 907
 908static int no_try_delta(const char *path)
 909{
 910        struct git_attr_check check[1];
 911
 912        setup_delta_attr_check(check);
 913        if (git_check_attr(path, ARRAY_SIZE(check), check))
 914                return 0;
 915        if (ATTR_FALSE(check->value))
 916                return 1;
 917        return 0;
 918}
 919
 920/*
 921 * When adding an object, check whether we have already added it
 922 * to our packing list. If so, we can skip. However, if we are
 923 * being asked to excludei t, but the previous mention was to include
 924 * it, make sure to adjust its flags and tweak our numbers accordingly.
 925 *
 926 * As an optimization, we pass out the index position where we would have
 927 * found the item, since that saves us from having to look it up again a
 928 * few lines later when we want to add the new entry.
 929 */
 930static int have_duplicate_entry(const unsigned char *sha1,
 931                                int exclude,
 932                                uint32_t *index_pos)
 933{
 934        struct object_entry *entry;
 935
 936        entry = packlist_find(&to_pack, sha1, index_pos);
 937        if (!entry)
 938                return 0;
 939
 940        if (exclude) {
 941                if (!entry->preferred_base)
 942                        nr_result--;
 943                entry->preferred_base = 1;
 944        }
 945
 946        return 1;
 947}
 948
 949static int want_found_object(int exclude, struct packed_git *p)
 950{
 951        if (exclude)
 952                return 1;
 953        if (incremental)
 954                return 0;
 955
 956        /*
 957         * When asked to do --local (do not include an object that appears in a
 958         * pack we borrow from elsewhere) or --honor-pack-keep (do not include
 959         * an object that appears in a pack marked with .keep), finding a pack
 960         * that matches the criteria is sufficient for us to decide to omit it.
 961         * However, even if this pack does not satisfy the criteria, we need to
 962         * make sure no copy of this object appears in _any_ pack that makes us
 963         * to omit the object, so we need to check all the packs.
 964         *
 965         * We can however first check whether these options can possible matter;
 966         * if they do not matter we know we want the object in generated pack.
 967         * Otherwise, we signal "-1" at the end to tell the caller that we do
 968         * not know either way, and it needs to check more packs.
 969         */
 970        if (!ignore_packed_keep &&
 971            (!local || !have_non_local_packs))
 972                return 1;
 973
 974        if (local && !p->pack_local)
 975                return 0;
 976        if (ignore_packed_keep && p->pack_local && p->pack_keep)
 977                return 0;
 978
 979        /* we don't know yet; keep looking for more packs */
 980        return -1;
 981}
 982
 983/*
 984 * Check whether we want the object in the pack (e.g., we do not want
 985 * objects found in non-local stores if the "--local" option was used).
 986 *
 987 * If the caller already knows an existing pack it wants to take the object
 988 * from, that is passed in *found_pack and *found_offset; otherwise this
 989 * function finds if there is any pack that has the object and returns the pack
 990 * and its offset in these variables.
 991 */
 992static int want_object_in_pack(const unsigned char *sha1,
 993                               int exclude,
 994                               struct packed_git **found_pack,
 995                               off_t *found_offset)
 996{
 997        struct packed_git *p;
 998        int want;
 999
1000        if (!exclude && local && has_loose_object_nonlocal(sha1))
1001                return 0;
1002
1003        /*
1004         * If we already know the pack object lives in, start checks from that
1005         * pack - in the usual case when neither --local was given nor .keep files
1006         * are present we will determine the answer right now.
1007         */
1008        if (*found_pack) {
1009                want = want_found_object(exclude, *found_pack);
1010                if (want != -1)
1011                        return want;
1012        }
1013
1014        for (p = packed_git; p; p = p->next) {
1015                off_t offset;
1016
1017                if (p == *found_pack)
1018                        offset = *found_offset;
1019                else
1020                        offset = find_pack_entry_one(sha1, p);
1021
1022                if (offset) {
1023                        if (!*found_pack) {
1024                                if (!is_pack_valid(p))
1025                                        continue;
1026                                *found_offset = offset;
1027                                *found_pack = p;
1028                        }
1029                        want = want_found_object(exclude, p);
1030                        if (want != -1)
1031                                return want;
1032                }
1033        }
1034
1035        return 1;
1036}
1037
1038static void create_object_entry(const unsigned char *sha1,
1039                                enum object_type type,
1040                                uint32_t hash,
1041                                int exclude,
1042                                int no_try_delta,
1043                                uint32_t index_pos,
1044                                struct packed_git *found_pack,
1045                                off_t found_offset)
1046{
1047        struct object_entry *entry;
1048
1049        entry = packlist_alloc(&to_pack, sha1, index_pos);
1050        entry->hash = hash;
1051        if (type)
1052                entry->type = type;
1053        if (exclude)
1054                entry->preferred_base = 1;
1055        else
1056                nr_result++;
1057        if (found_pack) {
1058                entry->in_pack = found_pack;
1059                entry->in_pack_offset = found_offset;
1060        }
1061
1062        entry->no_try_delta = no_try_delta;
1063}
1064
1065static const char no_closure_warning[] = N_(
1066"disabling bitmap writing, as some objects are not being packed"
1067);
1068
1069static int add_object_entry(const unsigned char *sha1, enum object_type type,
1070                            const char *name, int exclude)
1071{
1072        struct packed_git *found_pack = NULL;
1073        off_t found_offset = 0;
1074        uint32_t index_pos;
1075
1076        if (have_duplicate_entry(sha1, exclude, &index_pos))
1077                return 0;
1078
1079        if (!want_object_in_pack(sha1, exclude, &found_pack, &found_offset)) {
1080                /* The pack is missing an object, so it will not have closure */
1081                if (write_bitmap_index) {
1082                        warning(_(no_closure_warning));
1083                        write_bitmap_index = 0;
1084                }
1085                return 0;
1086        }
1087
1088        create_object_entry(sha1, type, pack_name_hash(name),
1089                            exclude, name && no_try_delta(name),
1090                            index_pos, found_pack, found_offset);
1091
1092        display_progress(progress_state, nr_result);
1093        return 1;
1094}
1095
1096static int add_object_entry_from_bitmap(const unsigned char *sha1,
1097                                        enum object_type type,
1098                                        int flags, uint32_t name_hash,
1099                                        struct packed_git *pack, off_t offset)
1100{
1101        uint32_t index_pos;
1102
1103        if (have_duplicate_entry(sha1, 0, &index_pos))
1104                return 0;
1105
1106        if (!want_object_in_pack(sha1, 0, &pack, &offset))
1107                return 0;
1108
1109        create_object_entry(sha1, type, name_hash, 0, 0, index_pos, pack, offset);
1110
1111        display_progress(progress_state, nr_result);
1112        return 1;
1113}
1114
1115struct pbase_tree_cache {
1116        unsigned char sha1[20];
1117        int ref;
1118        int temporary;
1119        void *tree_data;
1120        unsigned long tree_size;
1121};
1122
1123static struct pbase_tree_cache *(pbase_tree_cache[256]);
1124static int pbase_tree_cache_ix(const unsigned char *sha1)
1125{
1126        return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
1127}
1128static int pbase_tree_cache_ix_incr(int ix)
1129{
1130        return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
1131}
1132
1133static struct pbase_tree {
1134        struct pbase_tree *next;
1135        /* This is a phony "cache" entry; we are not
1136         * going to evict it or find it through _get()
1137         * mechanism -- this is for the toplevel node that
1138         * would almost always change with any commit.
1139         */
1140        struct pbase_tree_cache pcache;
1141} *pbase_tree;
1142
1143static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
1144{
1145        struct pbase_tree_cache *ent, *nent;
1146        void *data;
1147        unsigned long size;
1148        enum object_type type;
1149        int neigh;
1150        int my_ix = pbase_tree_cache_ix(sha1);
1151        int available_ix = -1;
1152
1153        /* pbase-tree-cache acts as a limited hashtable.
1154         * your object will be found at your index or within a few
1155         * slots after that slot if it is cached.
1156         */
1157        for (neigh = 0; neigh < 8; neigh++) {
1158                ent = pbase_tree_cache[my_ix];
1159                if (ent && !hashcmp(ent->sha1, sha1)) {
1160                        ent->ref++;
1161                        return ent;
1162                }
1163                else if (((available_ix < 0) && (!ent || !ent->ref)) ||
1164                         ((0 <= available_ix) &&
1165                          (!ent && pbase_tree_cache[available_ix])))
1166                        available_ix = my_ix;
1167                if (!ent)
1168                        break;
1169                my_ix = pbase_tree_cache_ix_incr(my_ix);
1170        }
1171
1172        /* Did not find one.  Either we got a bogus request or
1173         * we need to read and perhaps cache.
1174         */
1175        data = read_sha1_file(sha1, &type, &size);
1176        if (!data)
1177                return NULL;
1178        if (type != OBJ_TREE) {
1179                free(data);
1180                return NULL;
1181        }
1182
1183        /* We need to either cache or return a throwaway copy */
1184
1185        if (available_ix < 0)
1186                ent = NULL;
1187        else {
1188                ent = pbase_tree_cache[available_ix];
1189                my_ix = available_ix;
1190        }
1191
1192        if (!ent) {
1193                nent = xmalloc(sizeof(*nent));
1194                nent->temporary = (available_ix < 0);
1195        }
1196        else {
1197                /* evict and reuse */
1198                free(ent->tree_data);
1199                nent = ent;
1200        }
1201        hashcpy(nent->sha1, sha1);
1202        nent->tree_data = data;
1203        nent->tree_size = size;
1204        nent->ref = 1;
1205        if (!nent->temporary)
1206                pbase_tree_cache[my_ix] = nent;
1207        return nent;
1208}
1209
1210static void pbase_tree_put(struct pbase_tree_cache *cache)
1211{
1212        if (!cache->temporary) {
1213                cache->ref--;
1214                return;
1215        }
1216        free(cache->tree_data);
1217        free(cache);
1218}
1219
1220static int name_cmp_len(const char *name)
1221{
1222        int i;
1223        for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
1224                ;
1225        return i;
1226}
1227
1228static void add_pbase_object(struct tree_desc *tree,
1229                             const char *name,
1230                             int cmplen,
1231                             const char *fullname)
1232{
1233        struct name_entry entry;
1234        int cmp;
1235
1236        while (tree_entry(tree,&entry)) {
1237                if (S_ISGITLINK(entry.mode))
1238                        continue;
1239                cmp = tree_entry_len(&entry) != cmplen ? 1 :
1240                      memcmp(name, entry.path, cmplen);
1241                if (cmp > 0)
1242                        continue;
1243                if (cmp < 0)
1244                        return;
1245                if (name[cmplen] != '/') {
1246                        add_object_entry(entry.oid->hash,
1247                                         object_type(entry.mode),
1248                                         fullname, 1);
1249                        return;
1250                }
1251                if (S_ISDIR(entry.mode)) {
1252                        struct tree_desc sub;
1253                        struct pbase_tree_cache *tree;
1254                        const char *down = name+cmplen+1;
1255                        int downlen = name_cmp_len(down);
1256
1257                        tree = pbase_tree_get(entry.oid->hash);
1258                        if (!tree)
1259                                return;
1260                        init_tree_desc(&sub, tree->tree_data, tree->tree_size);
1261
1262                        add_pbase_object(&sub, down, downlen, fullname);
1263                        pbase_tree_put(tree);
1264                }
1265        }
1266}
1267
1268static unsigned *done_pbase_paths;
1269static int done_pbase_paths_num;
1270static int done_pbase_paths_alloc;
1271static int done_pbase_path_pos(unsigned hash)
1272{
1273        int lo = 0;
1274        int hi = done_pbase_paths_num;
1275        while (lo < hi) {
1276                int mi = (hi + lo) / 2;
1277                if (done_pbase_paths[mi] == hash)
1278                        return mi;
1279                if (done_pbase_paths[mi] < hash)
1280                        hi = mi;
1281                else
1282                        lo = mi + 1;
1283        }
1284        return -lo-1;
1285}
1286
1287static int check_pbase_path(unsigned hash)
1288{
1289        int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
1290        if (0 <= pos)
1291                return 1;
1292        pos = -pos - 1;
1293        ALLOC_GROW(done_pbase_paths,
1294                   done_pbase_paths_num + 1,
1295                   done_pbase_paths_alloc);
1296        done_pbase_paths_num++;
1297        if (pos < done_pbase_paths_num)
1298                memmove(done_pbase_paths + pos + 1,
1299                        done_pbase_paths + pos,
1300                        (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1301        done_pbase_paths[pos] = hash;
1302        return 0;
1303}
1304
1305static void add_preferred_base_object(const char *name)
1306{
1307        struct pbase_tree *it;
1308        int cmplen;
1309        unsigned hash = pack_name_hash(name);
1310
1311        if (!num_preferred_base || check_pbase_path(hash))
1312                return;
1313
1314        cmplen = name_cmp_len(name);
1315        for (it = pbase_tree; it; it = it->next) {
1316                if (cmplen == 0) {
1317                        add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
1318                }
1319                else {
1320                        struct tree_desc tree;
1321                        init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1322                        add_pbase_object(&tree, name, cmplen, name);
1323                }
1324        }
1325}
1326
1327static void add_preferred_base(unsigned char *sha1)
1328{
1329        struct pbase_tree *it;
1330        void *data;
1331        unsigned long size;
1332        unsigned char tree_sha1[20];
1333
1334        if (window <= num_preferred_base++)
1335                return;
1336
1337        data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1338        if (!data)
1339                return;
1340
1341        for (it = pbase_tree; it; it = it->next) {
1342                if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1343                        free(data);
1344                        return;
1345                }
1346        }
1347
1348        it = xcalloc(1, sizeof(*it));
1349        it->next = pbase_tree;
1350        pbase_tree = it;
1351
1352        hashcpy(it->pcache.sha1, tree_sha1);
1353        it->pcache.tree_data = data;
1354        it->pcache.tree_size = size;
1355}
1356
1357static void cleanup_preferred_base(void)
1358{
1359        struct pbase_tree *it;
1360        unsigned i;
1361
1362        it = pbase_tree;
1363        pbase_tree = NULL;
1364        while (it) {
1365                struct pbase_tree *this = it;
1366                it = this->next;
1367                free(this->pcache.tree_data);
1368                free(this);
1369        }
1370
1371        for (i = 0; i < ARRAY_SIZE(pbase_tree_cache); i++) {
1372                if (!pbase_tree_cache[i])
1373                        continue;
1374                free(pbase_tree_cache[i]->tree_data);
1375                free(pbase_tree_cache[i]);
1376                pbase_tree_cache[i] = NULL;
1377        }
1378
1379        free(done_pbase_paths);
1380        done_pbase_paths = NULL;
1381        done_pbase_paths_num = done_pbase_paths_alloc = 0;
1382}
1383
1384static void check_object(struct object_entry *entry)
1385{
1386        if (entry->in_pack) {
1387                struct packed_git *p = entry->in_pack;
1388                struct pack_window *w_curs = NULL;
1389                const unsigned char *base_ref = NULL;
1390                struct object_entry *base_entry;
1391                unsigned long used, used_0;
1392                unsigned long avail;
1393                off_t ofs;
1394                unsigned char *buf, c;
1395
1396                buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1397
1398                /*
1399                 * We want in_pack_type even if we do not reuse delta
1400                 * since non-delta representations could still be reused.
1401                 */
1402                used = unpack_object_header_buffer(buf, avail,
1403                                                   &entry->in_pack_type,
1404                                                   &entry->size);
1405                if (used == 0)
1406                        goto give_up;
1407
1408                /*
1409                 * Determine if this is a delta and if so whether we can
1410                 * reuse it or not.  Otherwise let's find out as cheaply as
1411                 * possible what the actual type and size for this object is.
1412                 */
1413                switch (entry->in_pack_type) {
1414                default:
1415                        /* Not a delta hence we've already got all we need. */
1416                        entry->type = entry->in_pack_type;
1417                        entry->in_pack_header_size = used;
1418                        if (entry->type < OBJ_COMMIT || entry->type > OBJ_BLOB)
1419                                goto give_up;
1420                        unuse_pack(&w_curs);
1421                        return;
1422                case OBJ_REF_DELTA:
1423                        if (reuse_delta && !entry->preferred_base)
1424                                base_ref = use_pack(p, &w_curs,
1425                                                entry->in_pack_offset + used, NULL);
1426                        entry->in_pack_header_size = used + 20;
1427                        break;
1428                case OBJ_OFS_DELTA:
1429                        buf = use_pack(p, &w_curs,
1430                                       entry->in_pack_offset + used, NULL);
1431                        used_0 = 0;
1432                        c = buf[used_0++];
1433                        ofs = c & 127;
1434                        while (c & 128) {
1435                                ofs += 1;
1436                                if (!ofs || MSB(ofs, 7)) {
1437                                        error("delta base offset overflow in pack for %s",
1438                                              sha1_to_hex(entry->idx.sha1));
1439                                        goto give_up;
1440                                }
1441                                c = buf[used_0++];
1442                                ofs = (ofs << 7) + (c & 127);
1443                        }
1444                        ofs = entry->in_pack_offset - ofs;
1445                        if (ofs <= 0 || ofs >= entry->in_pack_offset) {
1446                                error("delta base offset out of bound for %s",
1447                                      sha1_to_hex(entry->idx.sha1));
1448                                goto give_up;
1449                        }
1450                        if (reuse_delta && !entry->preferred_base) {
1451                                struct revindex_entry *revidx;
1452                                revidx = find_pack_revindex(p, ofs);
1453                                if (!revidx)
1454                                        goto give_up;
1455                                base_ref = nth_packed_object_sha1(p, revidx->nr);
1456                        }
1457                        entry->in_pack_header_size = used + used_0;
1458                        break;
1459                }
1460
1461                if (base_ref && (base_entry = packlist_find(&to_pack, base_ref, NULL))) {
1462                        /*
1463                         * If base_ref was set above that means we wish to
1464                         * reuse delta data, and we even found that base
1465                         * in the list of objects we want to pack. Goodie!
1466                         *
1467                         * Depth value does not matter - find_deltas() will
1468                         * never consider reused delta as the base object to
1469                         * deltify other objects against, in order to avoid
1470                         * circular deltas.
1471                         */
1472                        entry->type = entry->in_pack_type;
1473                        entry->delta = base_entry;
1474                        entry->delta_size = entry->size;
1475                        entry->delta_sibling = base_entry->delta_child;
1476                        base_entry->delta_child = entry;
1477                        unuse_pack(&w_curs);
1478                        return;
1479                }
1480
1481                if (entry->type) {
1482                        /*
1483                         * This must be a delta and we already know what the
1484                         * final object type is.  Let's extract the actual
1485                         * object size from the delta header.
1486                         */
1487                        entry->size = get_size_from_delta(p, &w_curs,
1488                                        entry->in_pack_offset + entry->in_pack_header_size);
1489                        if (entry->size == 0)
1490                                goto give_up;
1491                        unuse_pack(&w_curs);
1492                        return;
1493                }
1494
1495                /*
1496                 * No choice but to fall back to the recursive delta walk
1497                 * with sha1_object_info() to find about the object type
1498                 * at this point...
1499                 */
1500                give_up:
1501                unuse_pack(&w_curs);
1502        }
1503
1504        entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1505        /*
1506         * The error condition is checked in prepare_pack().  This is
1507         * to permit a missing preferred base object to be ignored
1508         * as a preferred base.  Doing so can result in a larger
1509         * pack file, but the transfer will still take place.
1510         */
1511}
1512
1513static int pack_offset_sort(const void *_a, const void *_b)
1514{
1515        const struct object_entry *a = *(struct object_entry **)_a;
1516        const struct object_entry *b = *(struct object_entry **)_b;
1517
1518        /* avoid filesystem trashing with loose objects */
1519        if (!a->in_pack && !b->in_pack)
1520                return hashcmp(a->idx.sha1, b->idx.sha1);
1521
1522        if (a->in_pack < b->in_pack)
1523                return -1;
1524        if (a->in_pack > b->in_pack)
1525                return 1;
1526        return a->in_pack_offset < b->in_pack_offset ? -1 :
1527                        (a->in_pack_offset > b->in_pack_offset);
1528}
1529
1530static void get_object_details(void)
1531{
1532        uint32_t i;
1533        struct object_entry **sorted_by_offset;
1534
1535        sorted_by_offset = xcalloc(to_pack.nr_objects, sizeof(struct object_entry *));
1536        for (i = 0; i < to_pack.nr_objects; i++)
1537                sorted_by_offset[i] = to_pack.objects + i;
1538        qsort(sorted_by_offset, to_pack.nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1539
1540        for (i = 0; i < to_pack.nr_objects; i++) {
1541                struct object_entry *entry = sorted_by_offset[i];
1542                check_object(entry);
1543                if (big_file_threshold < entry->size)
1544                        entry->no_try_delta = 1;
1545        }
1546
1547        free(sorted_by_offset);
1548}
1549
1550/*
1551 * We search for deltas in a list sorted by type, by filename hash, and then
1552 * by size, so that we see progressively smaller and smaller files.
1553 * That's because we prefer deltas to be from the bigger file
1554 * to the smaller -- deletes are potentially cheaper, but perhaps
1555 * more importantly, the bigger file is likely the more recent
1556 * one.  The deepest deltas are therefore the oldest objects which are
1557 * less susceptible to be accessed often.
1558 */
1559static int type_size_sort(const void *_a, const void *_b)
1560{
1561        const struct object_entry *a = *(struct object_entry **)_a;
1562        const struct object_entry *b = *(struct object_entry **)_b;
1563
1564        if (a->type > b->type)
1565                return -1;
1566        if (a->type < b->type)
1567                return 1;
1568        if (a->hash > b->hash)
1569                return -1;
1570        if (a->hash < b->hash)
1571                return 1;
1572        if (a->preferred_base > b->preferred_base)
1573                return -1;
1574        if (a->preferred_base < b->preferred_base)
1575                return 1;
1576        if (a->size > b->size)
1577                return -1;
1578        if (a->size < b->size)
1579                return 1;
1580        return a < b ? -1 : (a > b);  /* newest first */
1581}
1582
1583struct unpacked {
1584        struct object_entry *entry;
1585        void *data;
1586        struct delta_index *index;
1587        unsigned depth;
1588};
1589
1590static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1591                           unsigned long delta_size)
1592{
1593        if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1594                return 0;
1595
1596        if (delta_size < cache_max_small_delta_size)
1597                return 1;
1598
1599        /* cache delta, if objects are large enough compared to delta size */
1600        if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1601                return 1;
1602
1603        return 0;
1604}
1605
1606#ifndef NO_PTHREADS
1607
1608static pthread_mutex_t read_mutex;
1609#define read_lock()             pthread_mutex_lock(&read_mutex)
1610#define read_unlock()           pthread_mutex_unlock(&read_mutex)
1611
1612static pthread_mutex_t cache_mutex;
1613#define cache_lock()            pthread_mutex_lock(&cache_mutex)
1614#define cache_unlock()          pthread_mutex_unlock(&cache_mutex)
1615
1616static pthread_mutex_t progress_mutex;
1617#define progress_lock()         pthread_mutex_lock(&progress_mutex)
1618#define progress_unlock()       pthread_mutex_unlock(&progress_mutex)
1619
1620#else
1621
1622#define read_lock()             (void)0
1623#define read_unlock()           (void)0
1624#define cache_lock()            (void)0
1625#define cache_unlock()          (void)0
1626#define progress_lock()         (void)0
1627#define progress_unlock()       (void)0
1628
1629#endif
1630
1631static int try_delta(struct unpacked *trg, struct unpacked *src,
1632                     unsigned max_depth, unsigned long *mem_usage)
1633{
1634        struct object_entry *trg_entry = trg->entry;
1635        struct object_entry *src_entry = src->entry;
1636        unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1637        unsigned ref_depth;
1638        enum object_type type;
1639        void *delta_buf;
1640
1641        /* Don't bother doing diffs between different types */
1642        if (trg_entry->type != src_entry->type)
1643                return -1;
1644
1645        /*
1646         * We do not bother to try a delta that we discarded on an
1647         * earlier try, but only when reusing delta data.  Note that
1648         * src_entry that is marked as the preferred_base should always
1649         * be considered, as even if we produce a suboptimal delta against
1650         * it, we will still save the transfer cost, as we already know
1651         * the other side has it and we won't send src_entry at all.
1652         */
1653        if (reuse_delta && trg_entry->in_pack &&
1654            trg_entry->in_pack == src_entry->in_pack &&
1655            !src_entry->preferred_base &&
1656            trg_entry->in_pack_type != OBJ_REF_DELTA &&
1657            trg_entry->in_pack_type != OBJ_OFS_DELTA)
1658                return 0;
1659
1660        /* Let's not bust the allowed depth. */
1661        if (src->depth >= max_depth)
1662                return 0;
1663
1664        /* Now some size filtering heuristics. */
1665        trg_size = trg_entry->size;
1666        if (!trg_entry->delta) {
1667                max_size = trg_size/2 - 20;
1668                ref_depth = 1;
1669        } else {
1670                max_size = trg_entry->delta_size;
1671                ref_depth = trg->depth;
1672        }
1673        max_size = (uint64_t)max_size * (max_depth - src->depth) /
1674                                                (max_depth - ref_depth + 1);
1675        if (max_size == 0)
1676                return 0;
1677        src_size = src_entry->size;
1678        sizediff = src_size < trg_size ? trg_size - src_size : 0;
1679        if (sizediff >= max_size)
1680                return 0;
1681        if (trg_size < src_size / 32)
1682                return 0;
1683
1684        /* Load data if not already done */
1685        if (!trg->data) {
1686                read_lock();
1687                trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1688                read_unlock();
1689                if (!trg->data)
1690                        die("object %s cannot be read",
1691                            sha1_to_hex(trg_entry->idx.sha1));
1692                if (sz != trg_size)
1693                        die("object %s inconsistent object length (%lu vs %lu)",
1694                            sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1695                *mem_usage += sz;
1696        }
1697        if (!src->data) {
1698                read_lock();
1699                src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1700                read_unlock();
1701                if (!src->data) {
1702                        if (src_entry->preferred_base) {
1703                                static int warned = 0;
1704                                if (!warned++)
1705                                        warning("object %s cannot be read",
1706                                                sha1_to_hex(src_entry->idx.sha1));
1707                                /*
1708                                 * Those objects are not included in the
1709                                 * resulting pack.  Be resilient and ignore
1710                                 * them if they can't be read, in case the
1711                                 * pack could be created nevertheless.
1712                                 */
1713                                return 0;
1714                        }
1715                        die("object %s cannot be read",
1716                            sha1_to_hex(src_entry->idx.sha1));
1717                }
1718                if (sz != src_size)
1719                        die("object %s inconsistent object length (%lu vs %lu)",
1720                            sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1721                *mem_usage += sz;
1722        }
1723        if (!src->index) {
1724                src->index = create_delta_index(src->data, src_size);
1725                if (!src->index) {
1726                        static int warned = 0;
1727                        if (!warned++)
1728                                warning("suboptimal pack - out of memory");
1729                        return 0;
1730                }
1731                *mem_usage += sizeof_delta_index(src->index);
1732        }
1733
1734        delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1735        if (!delta_buf)
1736                return 0;
1737
1738        if (trg_entry->delta) {
1739                /* Prefer only shallower same-sized deltas. */
1740                if (delta_size == trg_entry->delta_size &&
1741                    src->depth + 1 >= trg->depth) {
1742                        free(delta_buf);
1743                        return 0;
1744                }
1745        }
1746
1747        /*
1748         * Handle memory allocation outside of the cache
1749         * accounting lock.  Compiler will optimize the strangeness
1750         * away when NO_PTHREADS is defined.
1751         */
1752        free(trg_entry->delta_data);
1753        cache_lock();
1754        if (trg_entry->delta_data) {
1755                delta_cache_size -= trg_entry->delta_size;
1756                trg_entry->delta_data = NULL;
1757        }
1758        if (delta_cacheable(src_size, trg_size, delta_size)) {
1759                delta_cache_size += delta_size;
1760                cache_unlock();
1761                trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1762        } else {
1763                cache_unlock();
1764                free(delta_buf);
1765        }
1766
1767        trg_entry->delta = src_entry;
1768        trg_entry->delta_size = delta_size;
1769        trg->depth = src->depth + 1;
1770
1771        return 1;
1772}
1773
1774static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1775{
1776        struct object_entry *child = me->delta_child;
1777        unsigned int m = n;
1778        while (child) {
1779                unsigned int c = check_delta_limit(child, n + 1);
1780                if (m < c)
1781                        m = c;
1782                child = child->delta_sibling;
1783        }
1784        return m;
1785}
1786
1787static unsigned long free_unpacked(struct unpacked *n)
1788{
1789        unsigned long freed_mem = sizeof_delta_index(n->index);
1790        free_delta_index(n->index);
1791        n->index = NULL;
1792        if (n->data) {
1793                freed_mem += n->entry->size;
1794                free(n->data);
1795                n->data = NULL;
1796        }
1797        n->entry = NULL;
1798        n->depth = 0;
1799        return freed_mem;
1800}
1801
1802static void find_deltas(struct object_entry **list, unsigned *list_size,
1803                        int window, int depth, unsigned *processed)
1804{
1805        uint32_t i, idx = 0, count = 0;
1806        struct unpacked *array;
1807        unsigned long mem_usage = 0;
1808
1809        array = xcalloc(window, sizeof(struct unpacked));
1810
1811        for (;;) {
1812                struct object_entry *entry;
1813                struct unpacked *n = array + idx;
1814                int j, max_depth, best_base = -1;
1815
1816                progress_lock();
1817                if (!*list_size) {
1818                        progress_unlock();
1819                        break;
1820                }
1821                entry = *list++;
1822                (*list_size)--;
1823                if (!entry->preferred_base) {
1824                        (*processed)++;
1825                        display_progress(progress_state, *processed);
1826                }
1827                progress_unlock();
1828
1829                mem_usage -= free_unpacked(n);
1830                n->entry = entry;
1831
1832                while (window_memory_limit &&
1833                       mem_usage > window_memory_limit &&
1834                       count > 1) {
1835                        uint32_t tail = (idx + window - count) % window;
1836                        mem_usage -= free_unpacked(array + tail);
1837                        count--;
1838                }
1839
1840                /* We do not compute delta to *create* objects we are not
1841                 * going to pack.
1842                 */
1843                if (entry->preferred_base)
1844                        goto next;
1845
1846                /*
1847                 * If the current object is at pack edge, take the depth the
1848                 * objects that depend on the current object into account
1849                 * otherwise they would become too deep.
1850                 */
1851                max_depth = depth;
1852                if (entry->delta_child) {
1853                        max_depth -= check_delta_limit(entry, 0);
1854                        if (max_depth <= 0)
1855                                goto next;
1856                }
1857
1858                j = window;
1859                while (--j > 0) {
1860                        int ret;
1861                        uint32_t other_idx = idx + j;
1862                        struct unpacked *m;
1863                        if (other_idx >= window)
1864                                other_idx -= window;
1865                        m = array + other_idx;
1866                        if (!m->entry)
1867                                break;
1868                        ret = try_delta(n, m, max_depth, &mem_usage);
1869                        if (ret < 0)
1870                                break;
1871                        else if (ret > 0)
1872                                best_base = other_idx;
1873                }
1874
1875                /*
1876                 * If we decided to cache the delta data, then it is best
1877                 * to compress it right away.  First because we have to do
1878                 * it anyway, and doing it here while we're threaded will
1879                 * save a lot of time in the non threaded write phase,
1880                 * as well as allow for caching more deltas within
1881                 * the same cache size limit.
1882                 * ...
1883                 * But only if not writing to stdout, since in that case
1884                 * the network is most likely throttling writes anyway,
1885                 * and therefore it is best to go to the write phase ASAP
1886                 * instead, as we can afford spending more time compressing
1887                 * between writes at that moment.
1888                 */
1889                if (entry->delta_data && !pack_to_stdout) {
1890                        entry->z_delta_size = do_compress(&entry->delta_data,
1891                                                          entry->delta_size);
1892                        cache_lock();
1893                        delta_cache_size -= entry->delta_size;
1894                        delta_cache_size += entry->z_delta_size;
1895                        cache_unlock();
1896                }
1897
1898                /* if we made n a delta, and if n is already at max
1899                 * depth, leaving it in the window is pointless.  we
1900                 * should evict it first.
1901                 */
1902                if (entry->delta && max_depth <= n->depth)
1903                        continue;
1904
1905                /*
1906                 * Move the best delta base up in the window, after the
1907                 * currently deltified object, to keep it longer.  It will
1908                 * be the first base object to be attempted next.
1909                 */
1910                if (entry->delta) {
1911                        struct unpacked swap = array[best_base];
1912                        int dist = (window + idx - best_base) % window;
1913                        int dst = best_base;
1914                        while (dist--) {
1915                                int src = (dst + 1) % window;
1916                                array[dst] = array[src];
1917                                dst = src;
1918                        }
1919                        array[dst] = swap;
1920                }
1921
1922                next:
1923                idx++;
1924                if (count + 1 < window)
1925                        count++;
1926                if (idx >= window)
1927                        idx = 0;
1928        }
1929
1930        for (i = 0; i < window; ++i) {
1931                free_delta_index(array[i].index);
1932                free(array[i].data);
1933        }
1934        free(array);
1935}
1936
1937#ifndef NO_PTHREADS
1938
1939static void try_to_free_from_threads(size_t size)
1940{
1941        read_lock();
1942        release_pack_memory(size);
1943        read_unlock();
1944}
1945
1946static try_to_free_t old_try_to_free_routine;
1947
1948/*
1949 * The main thread waits on the condition that (at least) one of the workers
1950 * has stopped working (which is indicated in the .working member of
1951 * struct thread_params).
1952 * When a work thread has completed its work, it sets .working to 0 and
1953 * signals the main thread and waits on the condition that .data_ready
1954 * becomes 1.
1955 */
1956
1957struct thread_params {
1958        pthread_t thread;
1959        struct object_entry **list;
1960        unsigned list_size;
1961        unsigned remaining;
1962        int window;
1963        int depth;
1964        int working;
1965        int data_ready;
1966        pthread_mutex_t mutex;
1967        pthread_cond_t cond;
1968        unsigned *processed;
1969};
1970
1971static pthread_cond_t progress_cond;
1972
1973/*
1974 * Mutex and conditional variable can't be statically-initialized on Windows.
1975 */
1976static void init_threaded_search(void)
1977{
1978        init_recursive_mutex(&read_mutex);
1979        pthread_mutex_init(&cache_mutex, NULL);
1980        pthread_mutex_init(&progress_mutex, NULL);
1981        pthread_cond_init(&progress_cond, NULL);
1982        old_try_to_free_routine = set_try_to_free_routine(try_to_free_from_threads);
1983}
1984
1985static void cleanup_threaded_search(void)
1986{
1987        set_try_to_free_routine(old_try_to_free_routine);
1988        pthread_cond_destroy(&progress_cond);
1989        pthread_mutex_destroy(&read_mutex);
1990        pthread_mutex_destroy(&cache_mutex);
1991        pthread_mutex_destroy(&progress_mutex);
1992}
1993
1994static void *threaded_find_deltas(void *arg)
1995{
1996        struct thread_params *me = arg;
1997
1998        while (me->remaining) {
1999                find_deltas(me->list, &me->remaining,
2000                            me->window, me->depth, me->processed);
2001
2002                progress_lock();
2003                me->working = 0;
2004                pthread_cond_signal(&progress_cond);
2005                progress_unlock();
2006
2007                /*
2008                 * We must not set ->data_ready before we wait on the
2009                 * condition because the main thread may have set it to 1
2010                 * before we get here. In order to be sure that new
2011                 * work is available if we see 1 in ->data_ready, it
2012                 * was initialized to 0 before this thread was spawned
2013                 * and we reset it to 0 right away.
2014                 */
2015                pthread_mutex_lock(&me->mutex);
2016                while (!me->data_ready)
2017                        pthread_cond_wait(&me->cond, &me->mutex);
2018                me->data_ready = 0;
2019                pthread_mutex_unlock(&me->mutex);
2020        }
2021        /* leave ->working 1 so that this doesn't get more work assigned */
2022        return NULL;
2023}
2024
2025static void ll_find_deltas(struct object_entry **list, unsigned list_size,
2026                           int window, int depth, unsigned *processed)
2027{
2028        struct thread_params *p;
2029        int i, ret, active_threads = 0;
2030
2031        init_threaded_search();
2032
2033        if (delta_search_threads <= 1) {
2034                find_deltas(list, &list_size, window, depth, processed);
2035                cleanup_threaded_search();
2036                return;
2037        }
2038        if (progress > pack_to_stdout)
2039                fprintf(stderr, "Delta compression using up to %d threads.\n",
2040                                delta_search_threads);
2041        p = xcalloc(delta_search_threads, sizeof(*p));
2042
2043        /* Partition the work amongst work threads. */
2044        for (i = 0; i < delta_search_threads; i++) {
2045                unsigned sub_size = list_size / (delta_search_threads - i);
2046
2047                /* don't use too small segments or no deltas will be found */
2048                if (sub_size < 2*window && i+1 < delta_search_threads)
2049                        sub_size = 0;
2050
2051                p[i].window = window;
2052                p[i].depth = depth;
2053                p[i].processed = processed;
2054                p[i].working = 1;
2055                p[i].data_ready = 0;
2056
2057                /* try to split chunks on "path" boundaries */
2058                while (sub_size && sub_size < list_size &&
2059                       list[sub_size]->hash &&
2060                       list[sub_size]->hash == list[sub_size-1]->hash)
2061                        sub_size++;
2062
2063                p[i].list = list;
2064                p[i].list_size = sub_size;
2065                p[i].remaining = sub_size;
2066
2067                list += sub_size;
2068                list_size -= sub_size;
2069        }
2070
2071        /* Start work threads. */
2072        for (i = 0; i < delta_search_threads; i++) {
2073                if (!p[i].list_size)
2074                        continue;
2075                pthread_mutex_init(&p[i].mutex, NULL);
2076                pthread_cond_init(&p[i].cond, NULL);
2077                ret = pthread_create(&p[i].thread, NULL,
2078                                     threaded_find_deltas, &p[i]);
2079                if (ret)
2080                        die("unable to create thread: %s", strerror(ret));
2081                active_threads++;
2082        }
2083
2084        /*
2085         * Now let's wait for work completion.  Each time a thread is done
2086         * with its work, we steal half of the remaining work from the
2087         * thread with the largest number of unprocessed objects and give
2088         * it to that newly idle thread.  This ensure good load balancing
2089         * until the remaining object list segments are simply too short
2090         * to be worth splitting anymore.
2091         */
2092        while (active_threads) {
2093                struct thread_params *target = NULL;
2094                struct thread_params *victim = NULL;
2095                unsigned sub_size = 0;
2096
2097                progress_lock();
2098                for (;;) {
2099                        for (i = 0; !target && i < delta_search_threads; i++)
2100                                if (!p[i].working)
2101                                        target = &p[i];
2102                        if (target)
2103                                break;
2104                        pthread_cond_wait(&progress_cond, &progress_mutex);
2105                }
2106
2107                for (i = 0; i < delta_search_threads; i++)
2108                        if (p[i].remaining > 2*window &&
2109                            (!victim || victim->remaining < p[i].remaining))
2110                                victim = &p[i];
2111                if (victim) {
2112                        sub_size = victim->remaining / 2;
2113                        list = victim->list + victim->list_size - sub_size;
2114                        while (sub_size && list[0]->hash &&
2115                               list[0]->hash == list[-1]->hash) {
2116                                list++;
2117                                sub_size--;
2118                        }
2119                        if (!sub_size) {
2120                                /*
2121                                 * It is possible for some "paths" to have
2122                                 * so many objects that no hash boundary
2123                                 * might be found.  Let's just steal the
2124                                 * exact half in that case.
2125                                 */
2126                                sub_size = victim->remaining / 2;
2127                                list -= sub_size;
2128                        }
2129                        target->list = list;
2130                        victim->list_size -= sub_size;
2131                        victim->remaining -= sub_size;
2132                }
2133                target->list_size = sub_size;
2134                target->remaining = sub_size;
2135                target->working = 1;
2136                progress_unlock();
2137
2138                pthread_mutex_lock(&target->mutex);
2139                target->data_ready = 1;
2140                pthread_cond_signal(&target->cond);
2141                pthread_mutex_unlock(&target->mutex);
2142
2143                if (!sub_size) {
2144                        pthread_join(target->thread, NULL);
2145                        pthread_cond_destroy(&target->cond);
2146                        pthread_mutex_destroy(&target->mutex);
2147                        active_threads--;
2148                }
2149        }
2150        cleanup_threaded_search();
2151        free(p);
2152}
2153
2154#else
2155#define ll_find_deltas(l, s, w, d, p)   find_deltas(l, &s, w, d, p)
2156#endif
2157
2158static void add_tag_chain(const struct object_id *oid)
2159{
2160        struct tag *tag;
2161
2162        /*
2163         * We catch duplicates already in add_object_entry(), but we'd
2164         * prefer to do this extra check to avoid having to parse the
2165         * tag at all if we already know that it's being packed (e.g., if
2166         * it was included via bitmaps, we would not have parsed it
2167         * previously).
2168         */
2169        if (packlist_find(&to_pack, oid->hash, NULL))
2170                return;
2171
2172        tag = lookup_tag(oid->hash);
2173        while (1) {
2174                if (!tag || parse_tag(tag) || !tag->tagged)
2175                        die("unable to pack objects reachable from tag %s",
2176                            oid_to_hex(oid));
2177
2178                add_object_entry(tag->object.oid.hash, OBJ_TAG, NULL, 0);
2179
2180                if (tag->tagged->type != OBJ_TAG)
2181                        return;
2182
2183                tag = (struct tag *)tag->tagged;
2184        }
2185}
2186
2187static int add_ref_tag(const char *path, const struct object_id *oid, int flag, void *cb_data)
2188{
2189        struct object_id peeled;
2190
2191        if (starts_with(path, "refs/tags/") && /* is a tag? */
2192            !peel_ref(path, peeled.hash)    && /* peelable? */
2193            packlist_find(&to_pack, peeled.hash, NULL))      /* object packed? */
2194                add_tag_chain(oid);
2195        return 0;
2196}
2197
2198static void prepare_pack(int window, int depth)
2199{
2200        struct object_entry **delta_list;
2201        uint32_t i, nr_deltas;
2202        unsigned n;
2203
2204        get_object_details();
2205
2206        /*
2207         * If we're locally repacking then we need to be doubly careful
2208         * from now on in order to make sure no stealth corruption gets
2209         * propagated to the new pack.  Clients receiving streamed packs
2210         * should validate everything they get anyway so no need to incur
2211         * the additional cost here in that case.
2212         */
2213        if (!pack_to_stdout)
2214                do_check_packed_object_crc = 1;
2215
2216        if (!to_pack.nr_objects || !window || !depth)
2217                return;
2218
2219        ALLOC_ARRAY(delta_list, to_pack.nr_objects);
2220        nr_deltas = n = 0;
2221
2222        for (i = 0; i < to_pack.nr_objects; i++) {
2223                struct object_entry *entry = to_pack.objects + i;
2224
2225                if (entry->delta)
2226                        /* This happens if we decided to reuse existing
2227                         * delta from a pack.  "reuse_delta &&" is implied.
2228                         */
2229                        continue;
2230
2231                if (entry->size < 50)
2232                        continue;
2233
2234                if (entry->no_try_delta)
2235                        continue;
2236
2237                if (!entry->preferred_base) {
2238                        nr_deltas++;
2239                        if (entry->type < 0)
2240                                die("unable to get type of object %s",
2241                                    sha1_to_hex(entry->idx.sha1));
2242                } else {
2243                        if (entry->type < 0) {
2244                                /*
2245                                 * This object is not found, but we
2246                                 * don't have to include it anyway.
2247                                 */
2248                                continue;
2249                        }
2250                }
2251
2252                delta_list[n++] = entry;
2253        }
2254
2255        if (nr_deltas && n > 1) {
2256                unsigned nr_done = 0;
2257                if (progress)
2258                        progress_state = start_progress(_("Compressing objects"),
2259                                                        nr_deltas);
2260                qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
2261                ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
2262                stop_progress(&progress_state);
2263                if (nr_done != nr_deltas)
2264                        die("inconsistency with delta count");
2265        }
2266        free(delta_list);
2267}
2268
2269static int git_pack_config(const char *k, const char *v, void *cb)
2270{
2271        if (!strcmp(k, "pack.window")) {
2272                window = git_config_int(k, v);
2273                return 0;
2274        }
2275        if (!strcmp(k, "pack.windowmemory")) {
2276                window_memory_limit = git_config_ulong(k, v);
2277                return 0;
2278        }
2279        if (!strcmp(k, "pack.depth")) {
2280                depth = git_config_int(k, v);
2281                return 0;
2282        }
2283        if (!strcmp(k, "pack.compression")) {
2284                int level = git_config_int(k, v);
2285                if (level == -1)
2286                        level = Z_DEFAULT_COMPRESSION;
2287                else if (level < 0 || level > Z_BEST_COMPRESSION)
2288                        die("bad pack compression level %d", level);
2289                pack_compression_level = level;
2290                pack_compression_seen = 1;
2291                return 0;
2292        }
2293        if (!strcmp(k, "pack.deltacachesize")) {
2294                max_delta_cache_size = git_config_int(k, v);
2295                return 0;
2296        }
2297        if (!strcmp(k, "pack.deltacachelimit")) {
2298                cache_max_small_delta_size = git_config_int(k, v);
2299                return 0;
2300        }
2301        if (!strcmp(k, "pack.writebitmaphashcache")) {
2302                if (git_config_bool(k, v))
2303                        write_bitmap_options |= BITMAP_OPT_HASH_CACHE;
2304                else
2305                        write_bitmap_options &= ~BITMAP_OPT_HASH_CACHE;
2306        }
2307        if (!strcmp(k, "pack.usebitmaps")) {
2308                use_bitmap_index_default = git_config_bool(k, v);
2309                return 0;
2310        }
2311        if (!strcmp(k, "pack.threads")) {
2312                delta_search_threads = git_config_int(k, v);
2313                if (delta_search_threads < 0)
2314                        die("invalid number of threads specified (%d)",
2315                            delta_search_threads);
2316#ifdef NO_PTHREADS
2317                if (delta_search_threads != 1)
2318                        warning("no threads support, ignoring %s", k);
2319#endif
2320                return 0;
2321        }
2322        if (!strcmp(k, "pack.indexversion")) {
2323                pack_idx_opts.version = git_config_int(k, v);
2324                if (pack_idx_opts.version > 2)
2325                        die("bad pack.indexversion=%"PRIu32,
2326                            pack_idx_opts.version);
2327                return 0;
2328        }
2329        return git_default_config(k, v, cb);
2330}
2331
2332static void read_object_list_from_stdin(void)
2333{
2334        char line[40 + 1 + PATH_MAX + 2];
2335        unsigned char sha1[20];
2336
2337        for (;;) {
2338                if (!fgets(line, sizeof(line), stdin)) {
2339                        if (feof(stdin))
2340                                break;
2341                        if (!ferror(stdin))
2342                                die("fgets returned NULL, not EOF, not error!");
2343                        if (errno != EINTR)
2344                                die_errno("fgets");
2345                        clearerr(stdin);
2346                        continue;
2347                }
2348                if (line[0] == '-') {
2349                        if (get_sha1_hex(line+1, sha1))
2350                                die("expected edge sha1, got garbage:\n %s",
2351                                    line);
2352                        add_preferred_base(sha1);
2353                        continue;
2354                }
2355                if (get_sha1_hex(line, sha1))
2356                        die("expected sha1, got garbage:\n %s", line);
2357
2358                add_preferred_base_object(line+41);
2359                add_object_entry(sha1, 0, line+41, 0);
2360        }
2361}
2362
2363#define OBJECT_ADDED (1u<<20)
2364
2365static void show_commit(struct commit *commit, void *data)
2366{
2367        add_object_entry(commit->object.oid.hash, OBJ_COMMIT, NULL, 0);
2368        commit->object.flags |= OBJECT_ADDED;
2369
2370        if (write_bitmap_index)
2371                index_commit_for_bitmap(commit);
2372}
2373
2374static void show_object(struct object *obj, const char *name, void *data)
2375{
2376        add_preferred_base_object(name);
2377        add_object_entry(obj->oid.hash, obj->type, name, 0);
2378        obj->flags |= OBJECT_ADDED;
2379}
2380
2381static void show_edge(struct commit *commit)
2382{
2383        add_preferred_base(commit->object.oid.hash);
2384}
2385
2386struct in_pack_object {
2387        off_t offset;
2388        struct object *object;
2389};
2390
2391struct in_pack {
2392        int alloc;
2393        int nr;
2394        struct in_pack_object *array;
2395};
2396
2397static void mark_in_pack_object(struct object *object, struct packed_git *p, struct in_pack *in_pack)
2398{
2399        in_pack->array[in_pack->nr].offset = find_pack_entry_one(object->oid.hash, p);
2400        in_pack->array[in_pack->nr].object = object;
2401        in_pack->nr++;
2402}
2403
2404/*
2405 * Compare the objects in the offset order, in order to emulate the
2406 * "git rev-list --objects" output that produced the pack originally.
2407 */
2408static int ofscmp(const void *a_, const void *b_)
2409{
2410        struct in_pack_object *a = (struct in_pack_object *)a_;
2411        struct in_pack_object *b = (struct in_pack_object *)b_;
2412
2413        if (a->offset < b->offset)
2414                return -1;
2415        else if (a->offset > b->offset)
2416                return 1;
2417        else
2418                return oidcmp(&a->object->oid, &b->object->oid);
2419}
2420
2421static void add_objects_in_unpacked_packs(struct rev_info *revs)
2422{
2423        struct packed_git *p;
2424        struct in_pack in_pack;
2425        uint32_t i;
2426
2427        memset(&in_pack, 0, sizeof(in_pack));
2428
2429        for (p = packed_git; p; p = p->next) {
2430                const unsigned char *sha1;
2431                struct object *o;
2432
2433                if (!p->pack_local || p->pack_keep)
2434                        continue;
2435                if (open_pack_index(p))
2436                        die("cannot open pack index");
2437
2438                ALLOC_GROW(in_pack.array,
2439                           in_pack.nr + p->num_objects,
2440                           in_pack.alloc);
2441
2442                for (i = 0; i < p->num_objects; i++) {
2443                        sha1 = nth_packed_object_sha1(p, i);
2444                        o = lookup_unknown_object(sha1);
2445                        if (!(o->flags & OBJECT_ADDED))
2446                                mark_in_pack_object(o, p, &in_pack);
2447                        o->flags |= OBJECT_ADDED;
2448                }
2449        }
2450
2451        if (in_pack.nr) {
2452                qsort(in_pack.array, in_pack.nr, sizeof(in_pack.array[0]),
2453                      ofscmp);
2454                for (i = 0; i < in_pack.nr; i++) {
2455                        struct object *o = in_pack.array[i].object;
2456                        add_object_entry(o->oid.hash, o->type, "", 0);
2457                }
2458        }
2459        free(in_pack.array);
2460}
2461
2462static int add_loose_object(const unsigned char *sha1, const char *path,
2463                            void *data)
2464{
2465        enum object_type type = sha1_object_info(sha1, NULL);
2466
2467        if (type < 0) {
2468                warning("loose object at %s could not be examined", path);
2469                return 0;
2470        }
2471
2472        add_object_entry(sha1, type, "", 0);
2473        return 0;
2474}
2475
2476/*
2477 * We actually don't even have to worry about reachability here.
2478 * add_object_entry will weed out duplicates, so we just add every
2479 * loose object we find.
2480 */
2481static void add_unreachable_loose_objects(void)
2482{
2483        for_each_loose_file_in_objdir(get_object_directory(),
2484                                      add_loose_object,
2485                                      NULL, NULL, NULL);
2486}
2487
2488static int has_sha1_pack_kept_or_nonlocal(const unsigned char *sha1)
2489{
2490        static struct packed_git *last_found = (void *)1;
2491        struct packed_git *p;
2492
2493        p = (last_found != (void *)1) ? last_found : packed_git;
2494
2495        while (p) {
2496                if ((!p->pack_local || p->pack_keep) &&
2497                        find_pack_entry_one(sha1, p)) {
2498                        last_found = p;
2499                        return 1;
2500                }
2501                if (p == last_found)
2502                        p = packed_git;
2503                else
2504                        p = p->next;
2505                if (p == last_found)
2506                        p = p->next;
2507        }
2508        return 0;
2509}
2510
2511/*
2512 * Store a list of sha1s that are should not be discarded
2513 * because they are either written too recently, or are
2514 * reachable from another object that was.
2515 *
2516 * This is filled by get_object_list.
2517 */
2518static struct sha1_array recent_objects;
2519
2520static int loosened_object_can_be_discarded(const unsigned char *sha1,
2521                                            unsigned long mtime)
2522{
2523        if (!unpack_unreachable_expiration)
2524                return 0;
2525        if (mtime > unpack_unreachable_expiration)
2526                return 0;
2527        if (sha1_array_lookup(&recent_objects, sha1) >= 0)
2528                return 0;
2529        return 1;
2530}
2531
2532static void loosen_unused_packed_objects(struct rev_info *revs)
2533{
2534        struct packed_git *p;
2535        uint32_t i;
2536        const unsigned char *sha1;
2537
2538        for (p = packed_git; p; p = p->next) {
2539                if (!p->pack_local || p->pack_keep)
2540                        continue;
2541
2542                if (open_pack_index(p))
2543                        die("cannot open pack index");
2544
2545                for (i = 0; i < p->num_objects; i++) {
2546                        sha1 = nth_packed_object_sha1(p, i);
2547                        if (!packlist_find(&to_pack, sha1, NULL) &&
2548                            !has_sha1_pack_kept_or_nonlocal(sha1) &&
2549                            !loosened_object_can_be_discarded(sha1, p->mtime))
2550                                if (force_object_loose(sha1, p->mtime))
2551                                        die("unable to force loose object");
2552                }
2553        }
2554}
2555
2556/*
2557 * This tracks any options which pack-reuse code expects to be on, or which a
2558 * reader of the pack might not understand, and which would therefore prevent
2559 * blind reuse of what we have on disk.
2560 */
2561static int pack_options_allow_reuse(void)
2562{
2563        return pack_to_stdout && allow_ofs_delta;
2564}
2565
2566static int get_object_list_from_bitmap(struct rev_info *revs)
2567{
2568        if (prepare_bitmap_walk(revs) < 0)
2569                return -1;
2570
2571        if (pack_options_allow_reuse() &&
2572            !reuse_partial_packfile_from_bitmap(
2573                        &reuse_packfile,
2574                        &reuse_packfile_objects,
2575                        &reuse_packfile_offset)) {
2576                assert(reuse_packfile_objects);
2577                nr_result += reuse_packfile_objects;
2578                display_progress(progress_state, nr_result);
2579        }
2580
2581        traverse_bitmap_commit_list(&add_object_entry_from_bitmap);
2582        return 0;
2583}
2584
2585static void record_recent_object(struct object *obj,
2586                                 const char *name,
2587                                 void *data)
2588{
2589        sha1_array_append(&recent_objects, obj->oid.hash);
2590}
2591
2592static void record_recent_commit(struct commit *commit, void *data)
2593{
2594        sha1_array_append(&recent_objects, commit->object.oid.hash);
2595}
2596
2597static void get_object_list(int ac, const char **av)
2598{
2599        struct rev_info revs;
2600        char line[1000];
2601        int flags = 0;
2602
2603        init_revisions(&revs, NULL);
2604        save_commit_buffer = 0;
2605        setup_revisions(ac, av, &revs, NULL);
2606
2607        /* make sure shallows are read */
2608        is_repository_shallow();
2609
2610        while (fgets(line, sizeof(line), stdin) != NULL) {
2611                int len = strlen(line);
2612                if (len && line[len - 1] == '\n')
2613                        line[--len] = 0;
2614                if (!len)
2615                        break;
2616                if (*line == '-') {
2617                        if (!strcmp(line, "--not")) {
2618                                flags ^= UNINTERESTING;
2619                                write_bitmap_index = 0;
2620                                continue;
2621                        }
2622                        if (starts_with(line, "--shallow ")) {
2623                                unsigned char sha1[20];
2624                                if (get_sha1_hex(line + 10, sha1))
2625                                        die("not an SHA-1 '%s'", line + 10);
2626                                register_shallow(sha1);
2627                                use_bitmap_index = 0;
2628                                continue;
2629                        }
2630                        die("not a rev '%s'", line);
2631                }
2632                if (handle_revision_arg(line, &revs, flags, REVARG_CANNOT_BE_FILENAME))
2633                        die("bad revision '%s'", line);
2634        }
2635
2636        if (use_bitmap_index && !get_object_list_from_bitmap(&revs))
2637                return;
2638
2639        if (prepare_revision_walk(&revs))
2640                die("revision walk setup failed");
2641        mark_edges_uninteresting(&revs, show_edge);
2642        traverse_commit_list(&revs, show_commit, show_object, NULL);
2643
2644        if (unpack_unreachable_expiration) {
2645                revs.ignore_missing_links = 1;
2646                if (add_unseen_recent_objects_to_traversal(&revs,
2647                                unpack_unreachable_expiration))
2648                        die("unable to add recent objects");
2649                if (prepare_revision_walk(&revs))
2650                        die("revision walk setup failed");
2651                traverse_commit_list(&revs, record_recent_commit,
2652                                     record_recent_object, NULL);
2653        }
2654
2655        if (keep_unreachable)
2656                add_objects_in_unpacked_packs(&revs);
2657        if (pack_loose_unreachable)
2658                add_unreachable_loose_objects();
2659        if (unpack_unreachable)
2660                loosen_unused_packed_objects(&revs);
2661
2662        sha1_array_clear(&recent_objects);
2663}
2664
2665static int option_parse_index_version(const struct option *opt,
2666                                      const char *arg, int unset)
2667{
2668        char *c;
2669        const char *val = arg;
2670        pack_idx_opts.version = strtoul(val, &c, 10);
2671        if (pack_idx_opts.version > 2)
2672                die(_("unsupported index version %s"), val);
2673        if (*c == ',' && c[1])
2674                pack_idx_opts.off32_limit = strtoul(c+1, &c, 0);
2675        if (*c || pack_idx_opts.off32_limit & 0x80000000)
2676                die(_("bad index version '%s'"), val);
2677        return 0;
2678}
2679
2680static int option_parse_unpack_unreachable(const struct option *opt,
2681                                           const char *arg, int unset)
2682{
2683        if (unset) {
2684                unpack_unreachable = 0;
2685                unpack_unreachable_expiration = 0;
2686        }
2687        else {
2688                unpack_unreachable = 1;
2689                if (arg)
2690                        unpack_unreachable_expiration = approxidate(arg);
2691        }
2692        return 0;
2693}
2694
2695int cmd_pack_objects(int argc, const char **argv, const char *prefix)
2696{
2697        int use_internal_rev_list = 0;
2698        int thin = 0;
2699        int shallow = 0;
2700        int all_progress_implied = 0;
2701        struct argv_array rp = ARGV_ARRAY_INIT;
2702        int rev_list_unpacked = 0, rev_list_all = 0, rev_list_reflog = 0;
2703        int rev_list_index = 0;
2704        struct option pack_objects_options[] = {
2705                OPT_SET_INT('q', "quiet", &progress,
2706                            N_("do not show progress meter"), 0),
2707                OPT_SET_INT(0, "progress", &progress,
2708                            N_("show progress meter"), 1),
2709                OPT_SET_INT(0, "all-progress", &progress,
2710                            N_("show progress meter during object writing phase"), 2),
2711                OPT_BOOL(0, "all-progress-implied",
2712                         &all_progress_implied,
2713                         N_("similar to --all-progress when progress meter is shown")),
2714                { OPTION_CALLBACK, 0, "index-version", NULL, N_("version[,offset]"),
2715                  N_("write the pack index file in the specified idx format version"),
2716                  0, option_parse_index_version },
2717                OPT_MAGNITUDE(0, "max-pack-size", &pack_size_limit,
2718                              N_("maximum size of each output pack file")),
2719                OPT_BOOL(0, "local", &local,
2720                         N_("ignore borrowed objects from alternate object store")),
2721                OPT_BOOL(0, "incremental", &incremental,
2722                         N_("ignore packed objects")),
2723                OPT_INTEGER(0, "window", &window,
2724                            N_("limit pack window by objects")),
2725                OPT_MAGNITUDE(0, "window-memory", &window_memory_limit,
2726                              N_("limit pack window by memory in addition to object limit")),
2727                OPT_INTEGER(0, "depth", &depth,
2728                            N_("maximum length of delta chain allowed in the resulting pack")),
2729                OPT_BOOL(0, "reuse-delta", &reuse_delta,
2730                         N_("reuse existing deltas")),
2731                OPT_BOOL(0, "reuse-object", &reuse_object,
2732                         N_("reuse existing objects")),
2733                OPT_BOOL(0, "delta-base-offset", &allow_ofs_delta,
2734                         N_("use OFS_DELTA objects")),
2735                OPT_INTEGER(0, "threads", &delta_search_threads,
2736                            N_("use threads when searching for best delta matches")),
2737                OPT_BOOL(0, "non-empty", &non_empty,
2738                         N_("do not create an empty pack output")),
2739                OPT_BOOL(0, "revs", &use_internal_rev_list,
2740                         N_("read revision arguments from standard input")),
2741                { OPTION_SET_INT, 0, "unpacked", &rev_list_unpacked, NULL,
2742                  N_("limit the objects to those that are not yet packed"),
2743                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
2744                { OPTION_SET_INT, 0, "all", &rev_list_all, NULL,
2745                  N_("include objects reachable from any reference"),
2746                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
2747                { OPTION_SET_INT, 0, "reflog", &rev_list_reflog, NULL,
2748                  N_("include objects referred by reflog entries"),
2749                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
2750                { OPTION_SET_INT, 0, "indexed-objects", &rev_list_index, NULL,
2751                  N_("include objects referred to by the index"),
2752                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
2753                OPT_BOOL(0, "stdout", &pack_to_stdout,
2754                         N_("output pack to stdout")),
2755                OPT_BOOL(0, "include-tag", &include_tag,
2756                         N_("include tag objects that refer to objects to be packed")),
2757                OPT_BOOL(0, "keep-unreachable", &keep_unreachable,
2758                         N_("keep unreachable objects")),
2759                OPT_BOOL(0, "pack-loose-unreachable", &pack_loose_unreachable,
2760                         N_("pack loose unreachable objects")),
2761                { OPTION_CALLBACK, 0, "unpack-unreachable", NULL, N_("time"),
2762                  N_("unpack unreachable objects newer than <time>"),
2763                  PARSE_OPT_OPTARG, option_parse_unpack_unreachable },
2764                OPT_BOOL(0, "thin", &thin,
2765                         N_("create thin packs")),
2766                OPT_BOOL(0, "shallow", &shallow,
2767                         N_("create packs suitable for shallow fetches")),
2768                OPT_BOOL(0, "honor-pack-keep", &ignore_packed_keep,
2769                         N_("ignore packs that have companion .keep file")),
2770                OPT_INTEGER(0, "compression", &pack_compression_level,
2771                            N_("pack compression level")),
2772                OPT_SET_INT(0, "keep-true-parents", &grafts_replace_parents,
2773                            N_("do not hide commits by grafts"), 0),
2774                OPT_BOOL(0, "use-bitmap-index", &use_bitmap_index,
2775                         N_("use a bitmap index if available to speed up counting objects")),
2776                OPT_BOOL(0, "write-bitmap-index", &write_bitmap_index,
2777                         N_("write a bitmap index together with the pack index")),
2778                OPT_END(),
2779        };
2780
2781        check_replace_refs = 0;
2782
2783        reset_pack_idx_option(&pack_idx_opts);
2784        git_config(git_pack_config, NULL);
2785        if (!pack_compression_seen && core_compression_seen)
2786                pack_compression_level = core_compression_level;
2787
2788        progress = isatty(2);
2789        argc = parse_options(argc, argv, prefix, pack_objects_options,
2790                             pack_usage, 0);
2791
2792        if (argc) {
2793                base_name = argv[0];
2794                argc--;
2795        }
2796        if (pack_to_stdout != !base_name || argc)
2797                usage_with_options(pack_usage, pack_objects_options);
2798
2799        argv_array_push(&rp, "pack-objects");
2800        if (thin) {
2801                use_internal_rev_list = 1;
2802                argv_array_push(&rp, shallow
2803                                ? "--objects-edge-aggressive"
2804                                : "--objects-edge");
2805        } else
2806                argv_array_push(&rp, "--objects");
2807
2808        if (rev_list_all) {
2809                use_internal_rev_list = 1;
2810                argv_array_push(&rp, "--all");
2811        }
2812        if (rev_list_reflog) {
2813                use_internal_rev_list = 1;
2814                argv_array_push(&rp, "--reflog");
2815        }
2816        if (rev_list_index) {
2817                use_internal_rev_list = 1;
2818                argv_array_push(&rp, "--indexed-objects");
2819        }
2820        if (rev_list_unpacked) {
2821                use_internal_rev_list = 1;
2822                argv_array_push(&rp, "--unpacked");
2823        }
2824
2825        if (!reuse_object)
2826                reuse_delta = 0;
2827        if (pack_compression_level == -1)
2828                pack_compression_level = Z_DEFAULT_COMPRESSION;
2829        else if (pack_compression_level < 0 || pack_compression_level > Z_BEST_COMPRESSION)
2830                die("bad pack compression level %d", pack_compression_level);
2831
2832        if (!delta_search_threads)      /* --threads=0 means autodetect */
2833                delta_search_threads = online_cpus();
2834
2835#ifdef NO_PTHREADS
2836        if (delta_search_threads != 1)
2837                warning("no threads support, ignoring --threads");
2838#endif
2839        if (!pack_to_stdout && !pack_size_limit)
2840                pack_size_limit = pack_size_limit_cfg;
2841        if (pack_to_stdout && pack_size_limit)
2842                die("--max-pack-size cannot be used to build a pack for transfer.");
2843        if (pack_size_limit && pack_size_limit < 1024*1024) {
2844                warning("minimum pack size limit is 1 MiB");
2845                pack_size_limit = 1024*1024;
2846        }
2847
2848        if (!pack_to_stdout && thin)
2849                die("--thin cannot be used to build an indexable pack.");
2850
2851        if (keep_unreachable && unpack_unreachable)
2852                die("--keep-unreachable and --unpack-unreachable are incompatible.");
2853        if (!rev_list_all || !rev_list_reflog || !rev_list_index)
2854                unpack_unreachable_expiration = 0;
2855
2856        /*
2857         * "soft" reasons not to use bitmaps - for on-disk repack by default we want
2858         *
2859         * - to produce good pack (with bitmap index not-yet-packed objects are
2860         *   packed in suboptimal order).
2861         *
2862         * - to use more robust pack-generation codepath (avoiding possible
2863         *   bugs in bitmap code and possible bitmap index corruption).
2864         */
2865        if (!pack_to_stdout)
2866                use_bitmap_index_default = 0;
2867
2868        if (use_bitmap_index < 0)
2869                use_bitmap_index = use_bitmap_index_default;
2870
2871        /* "hard" reasons not to use bitmaps; these just won't work at all */
2872        if (!use_internal_rev_list || (!pack_to_stdout && write_bitmap_index) || is_repository_shallow())
2873                use_bitmap_index = 0;
2874
2875        if (pack_to_stdout || !rev_list_all)
2876                write_bitmap_index = 0;
2877
2878        if (progress && all_progress_implied)
2879                progress = 2;
2880
2881        prepare_packed_git();
2882        if (ignore_packed_keep) {
2883                struct packed_git *p;
2884                for (p = packed_git; p; p = p->next)
2885                        if (p->pack_local && p->pack_keep)
2886                                break;
2887                if (!p) /* no keep-able packs found */
2888                        ignore_packed_keep = 0;
2889        }
2890        if (local) {
2891                /*
2892                 * unlike ignore_packed_keep above, we do not want to
2893                 * unset "local" based on looking at packs, as it
2894                 * also covers non-local objects
2895                 */
2896                struct packed_git *p;
2897                for (p = packed_git; p; p = p->next) {
2898                        if (!p->pack_local) {
2899                                have_non_local_packs = 1;
2900                                break;
2901                        }
2902                }
2903        }
2904
2905        if (progress)
2906                progress_state = start_progress(_("Counting objects"), 0);
2907        if (!use_internal_rev_list)
2908                read_object_list_from_stdin();
2909        else {
2910                get_object_list(rp.argc, rp.argv);
2911                argv_array_clear(&rp);
2912        }
2913        cleanup_preferred_base();
2914        if (include_tag && nr_result)
2915                for_each_ref(add_ref_tag, NULL);
2916        stop_progress(&progress_state);
2917
2918        if (non_empty && !nr_result)
2919                return 0;
2920        if (nr_result)
2921                prepare_pack(window, depth);
2922        write_pack_file();
2923        if (progress)
2924                fprintf(stderr, "Total %"PRIu32" (delta %"PRIu32"),"
2925                        " reused %"PRIu32" (delta %"PRIu32")\n",
2926                        written, written_delta, reused, reused_delta);
2927        return 0;
2928}