builtin / pack-objects.con commit pack-objects: drop packlist index_pos optimization (3a37876)
   1#include "builtin.h"
   2#include "cache.h"
   3#include "repository.h"
   4#include "config.h"
   5#include "attr.h"
   6#include "object.h"
   7#include "blob.h"
   8#include "commit.h"
   9#include "tag.h"
  10#include "tree.h"
  11#include "delta.h"
  12#include "pack.h"
  13#include "pack-revindex.h"
  14#include "csum-file.h"
  15#include "tree-walk.h"
  16#include "diff.h"
  17#include "revision.h"
  18#include "list-objects.h"
  19#include "list-objects-filter.h"
  20#include "list-objects-filter-options.h"
  21#include "pack-objects.h"
  22#include "progress.h"
  23#include "refs.h"
  24#include "streaming.h"
  25#include "thread-utils.h"
  26#include "pack-bitmap.h"
  27#include "delta-islands.h"
  28#include "reachable.h"
  29#include "sha1-array.h"
  30#include "argv-array.h"
  31#include "list.h"
  32#include "packfile.h"
  33#include "object-store.h"
  34#include "dir.h"
  35#include "midx.h"
  36#include "trace2.h"
  37
  38#define IN_PACK(obj) oe_in_pack(&to_pack, obj)
  39#define SIZE(obj) oe_size(&to_pack, obj)
  40#define SET_SIZE(obj,size) oe_set_size(&to_pack, obj, size)
  41#define DELTA_SIZE(obj) oe_delta_size(&to_pack, obj)
  42#define DELTA(obj) oe_delta(&to_pack, obj)
  43#define DELTA_CHILD(obj) oe_delta_child(&to_pack, obj)
  44#define DELTA_SIBLING(obj) oe_delta_sibling(&to_pack, obj)
  45#define SET_DELTA(obj, val) oe_set_delta(&to_pack, obj, val)
  46#define SET_DELTA_EXT(obj, oid) oe_set_delta_ext(&to_pack, obj, oid)
  47#define SET_DELTA_SIZE(obj, val) oe_set_delta_size(&to_pack, obj, val)
  48#define SET_DELTA_CHILD(obj, val) oe_set_delta_child(&to_pack, obj, val)
  49#define SET_DELTA_SIBLING(obj, val) oe_set_delta_sibling(&to_pack, obj, val)
  50
  51static const char *pack_usage[] = {
  52        N_("git pack-objects --stdout [<options>...] [< <ref-list> | < <object-list>]"),
  53        N_("git pack-objects [<options>...] <base-name> [< <ref-list> | < <object-list>]"),
  54        NULL
  55};
  56
  57/*
  58 * Objects we are going to pack are collected in the `to_pack` structure.
  59 * It contains an array (dynamically expanded) of the object data, and a map
  60 * that can resolve SHA1s to their position in the array.
  61 */
  62static struct packing_data to_pack;
  63
  64static struct pack_idx_entry **written_list;
  65static uint32_t nr_result, nr_written, nr_seen;
  66static struct bitmap_index *bitmap_git;
  67static uint32_t write_layer;
  68
  69static int non_empty;
  70static int reuse_delta = 1, reuse_object = 1;
  71static int keep_unreachable, unpack_unreachable, include_tag;
  72static timestamp_t unpack_unreachable_expiration;
  73static int pack_loose_unreachable;
  74static int local;
  75static int have_non_local_packs;
  76static int incremental;
  77static int ignore_packed_keep_on_disk;
  78static int ignore_packed_keep_in_core;
  79static int allow_ofs_delta;
  80static struct pack_idx_option pack_idx_opts;
  81static const char *base_name;
  82static int progress = 1;
  83static int window = 10;
  84static unsigned long pack_size_limit;
  85static int depth = 50;
  86static int delta_search_threads;
  87static int pack_to_stdout;
  88static int sparse;
  89static int thin;
  90static int num_preferred_base;
  91static struct progress *progress_state;
  92
  93static struct packed_git *reuse_packfile;
  94static uint32_t reuse_packfile_objects;
  95static off_t reuse_packfile_offset;
  96
  97static int use_bitmap_index_default = 1;
  98static int use_bitmap_index = -1;
  99static enum {
 100        WRITE_BITMAP_FALSE = 0,
 101        WRITE_BITMAP_QUIET,
 102        WRITE_BITMAP_TRUE,
 103} write_bitmap_index;
 104static uint16_t write_bitmap_options = BITMAP_OPT_HASH_CACHE;
 105
 106static int exclude_promisor_objects;
 107
 108static int use_delta_islands;
 109
 110static unsigned long delta_cache_size = 0;
 111static unsigned long max_delta_cache_size = DEFAULT_DELTA_CACHE_SIZE;
 112static unsigned long cache_max_small_delta_size = 1000;
 113
 114static unsigned long window_memory_limit = 0;
 115
 116static struct list_objects_filter_options filter_options;
 117
 118enum missing_action {
 119        MA_ERROR = 0,      /* fail if any missing objects are encountered */
 120        MA_ALLOW_ANY,      /* silently allow ALL missing objects */
 121        MA_ALLOW_PROMISOR, /* silently allow all missing PROMISOR objects */
 122};
 123static enum missing_action arg_missing_action;
 124static show_object_fn fn_show_object;
 125
 126/*
 127 * stats
 128 */
 129static uint32_t written, written_delta;
 130static uint32_t reused, reused_delta;
 131
 132/*
 133 * Indexed commits
 134 */
 135static struct commit **indexed_commits;
 136static unsigned int indexed_commits_nr;
 137static unsigned int indexed_commits_alloc;
 138
 139static void index_commit_for_bitmap(struct commit *commit)
 140{
 141        if (indexed_commits_nr >= indexed_commits_alloc) {
 142                indexed_commits_alloc = (indexed_commits_alloc + 32) * 2;
 143                REALLOC_ARRAY(indexed_commits, indexed_commits_alloc);
 144        }
 145
 146        indexed_commits[indexed_commits_nr++] = commit;
 147}
 148
 149static void *get_delta(struct object_entry *entry)
 150{
 151        unsigned long size, base_size, delta_size;
 152        void *buf, *base_buf, *delta_buf;
 153        enum object_type type;
 154
 155        buf = read_object_file(&entry->idx.oid, &type, &size);
 156        if (!buf)
 157                die(_("unable to read %s"), oid_to_hex(&entry->idx.oid));
 158        base_buf = read_object_file(&DELTA(entry)->idx.oid, &type,
 159                                    &base_size);
 160        if (!base_buf)
 161                die("unable to read %s",
 162                    oid_to_hex(&DELTA(entry)->idx.oid));
 163        delta_buf = diff_delta(base_buf, base_size,
 164                               buf, size, &delta_size, 0);
 165        /*
 166         * We succesfully computed this delta once but dropped it for
 167         * memory reasons. Something is very wrong if this time we
 168         * recompute and create a different delta.
 169         */
 170        if (!delta_buf || delta_size != DELTA_SIZE(entry))
 171                BUG("delta size changed");
 172        free(buf);
 173        free(base_buf);
 174        return delta_buf;
 175}
 176
 177static unsigned long do_compress(void **pptr, unsigned long size)
 178{
 179        git_zstream stream;
 180        void *in, *out;
 181        unsigned long maxsize;
 182
 183        git_deflate_init(&stream, pack_compression_level);
 184        maxsize = git_deflate_bound(&stream, size);
 185
 186        in = *pptr;
 187        out = xmalloc(maxsize);
 188        *pptr = out;
 189
 190        stream.next_in = in;
 191        stream.avail_in = size;
 192        stream.next_out = out;
 193        stream.avail_out = maxsize;
 194        while (git_deflate(&stream, Z_FINISH) == Z_OK)
 195                ; /* nothing */
 196        git_deflate_end(&stream);
 197
 198        free(in);
 199        return stream.total_out;
 200}
 201
 202static unsigned long write_large_blob_data(struct git_istream *st, struct hashfile *f,
 203                                           const struct object_id *oid)
 204{
 205        git_zstream stream;
 206        unsigned char ibuf[1024 * 16];
 207        unsigned char obuf[1024 * 16];
 208        unsigned long olen = 0;
 209
 210        git_deflate_init(&stream, pack_compression_level);
 211
 212        for (;;) {
 213                ssize_t readlen;
 214                int zret = Z_OK;
 215                readlen = read_istream(st, ibuf, sizeof(ibuf));
 216                if (readlen == -1)
 217                        die(_("unable to read %s"), oid_to_hex(oid));
 218
 219                stream.next_in = ibuf;
 220                stream.avail_in = readlen;
 221                while ((stream.avail_in || readlen == 0) &&
 222                       (zret == Z_OK || zret == Z_BUF_ERROR)) {
 223                        stream.next_out = obuf;
 224                        stream.avail_out = sizeof(obuf);
 225                        zret = git_deflate(&stream, readlen ? 0 : Z_FINISH);
 226                        hashwrite(f, obuf, stream.next_out - obuf);
 227                        olen += stream.next_out - obuf;
 228                }
 229                if (stream.avail_in)
 230                        die(_("deflate error (%d)"), zret);
 231                if (readlen == 0) {
 232                        if (zret != Z_STREAM_END)
 233                                die(_("deflate error (%d)"), zret);
 234                        break;
 235                }
 236        }
 237        git_deflate_end(&stream);
 238        return olen;
 239}
 240
 241/*
 242 * we are going to reuse the existing object data as is.  make
 243 * sure it is not corrupt.
 244 */
 245static int check_pack_inflate(struct packed_git *p,
 246                struct pack_window **w_curs,
 247                off_t offset,
 248                off_t len,
 249                unsigned long expect)
 250{
 251        git_zstream stream;
 252        unsigned char fakebuf[4096], *in;
 253        int st;
 254
 255        memset(&stream, 0, sizeof(stream));
 256        git_inflate_init(&stream);
 257        do {
 258                in = use_pack(p, w_curs, offset, &stream.avail_in);
 259                stream.next_in = in;
 260                stream.next_out = fakebuf;
 261                stream.avail_out = sizeof(fakebuf);
 262                st = git_inflate(&stream, Z_FINISH);
 263                offset += stream.next_in - in;
 264        } while (st == Z_OK || st == Z_BUF_ERROR);
 265        git_inflate_end(&stream);
 266        return (st == Z_STREAM_END &&
 267                stream.total_out == expect &&
 268                stream.total_in == len) ? 0 : -1;
 269}
 270
 271static void copy_pack_data(struct hashfile *f,
 272                struct packed_git *p,
 273                struct pack_window **w_curs,
 274                off_t offset,
 275                off_t len)
 276{
 277        unsigned char *in;
 278        unsigned long avail;
 279
 280        while (len) {
 281                in = use_pack(p, w_curs, offset, &avail);
 282                if (avail > len)
 283                        avail = (unsigned long)len;
 284                hashwrite(f, in, avail);
 285                offset += avail;
 286                len -= avail;
 287        }
 288}
 289
 290/* Return 0 if we will bust the pack-size limit */
 291static unsigned long write_no_reuse_object(struct hashfile *f, struct object_entry *entry,
 292                                           unsigned long limit, int usable_delta)
 293{
 294        unsigned long size, datalen;
 295        unsigned char header[MAX_PACK_OBJECT_HEADER],
 296                      dheader[MAX_PACK_OBJECT_HEADER];
 297        unsigned hdrlen;
 298        enum object_type type;
 299        void *buf;
 300        struct git_istream *st = NULL;
 301        const unsigned hashsz = the_hash_algo->rawsz;
 302
 303        if (!usable_delta) {
 304                if (oe_type(entry) == OBJ_BLOB &&
 305                    oe_size_greater_than(&to_pack, entry, big_file_threshold) &&
 306                    (st = open_istream(&entry->idx.oid, &type, &size, NULL)) != NULL)
 307                        buf = NULL;
 308                else {
 309                        buf = read_object_file(&entry->idx.oid, &type, &size);
 310                        if (!buf)
 311                                die(_("unable to read %s"),
 312                                    oid_to_hex(&entry->idx.oid));
 313                }
 314                /*
 315                 * make sure no cached delta data remains from a
 316                 * previous attempt before a pack split occurred.
 317                 */
 318                FREE_AND_NULL(entry->delta_data);
 319                entry->z_delta_size = 0;
 320        } else if (entry->delta_data) {
 321                size = DELTA_SIZE(entry);
 322                buf = entry->delta_data;
 323                entry->delta_data = NULL;
 324                type = (allow_ofs_delta && DELTA(entry)->idx.offset) ?
 325                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 326        } else {
 327                buf = get_delta(entry);
 328                size = DELTA_SIZE(entry);
 329                type = (allow_ofs_delta && DELTA(entry)->idx.offset) ?
 330                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 331        }
 332
 333        if (st) /* large blob case, just assume we don't compress well */
 334                datalen = size;
 335        else if (entry->z_delta_size)
 336                datalen = entry->z_delta_size;
 337        else
 338                datalen = do_compress(&buf, size);
 339
 340        /*
 341         * The object header is a byte of 'type' followed by zero or
 342         * more bytes of length.
 343         */
 344        hdrlen = encode_in_pack_object_header(header, sizeof(header),
 345                                              type, size);
 346
 347        if (type == OBJ_OFS_DELTA) {
 348                /*
 349                 * Deltas with relative base contain an additional
 350                 * encoding of the relative offset for the delta
 351                 * base from this object's position in the pack.
 352                 */
 353                off_t ofs = entry->idx.offset - DELTA(entry)->idx.offset;
 354                unsigned pos = sizeof(dheader) - 1;
 355                dheader[pos] = ofs & 127;
 356                while (ofs >>= 7)
 357                        dheader[--pos] = 128 | (--ofs & 127);
 358                if (limit && hdrlen + sizeof(dheader) - pos + datalen + hashsz >= limit) {
 359                        if (st)
 360                                close_istream(st);
 361                        free(buf);
 362                        return 0;
 363                }
 364                hashwrite(f, header, hdrlen);
 365                hashwrite(f, dheader + pos, sizeof(dheader) - pos);
 366                hdrlen += sizeof(dheader) - pos;
 367        } else if (type == OBJ_REF_DELTA) {
 368                /*
 369                 * Deltas with a base reference contain
 370                 * additional bytes for the base object ID.
 371                 */
 372                if (limit && hdrlen + hashsz + datalen + hashsz >= limit) {
 373                        if (st)
 374                                close_istream(st);
 375                        free(buf);
 376                        return 0;
 377                }
 378                hashwrite(f, header, hdrlen);
 379                hashwrite(f, DELTA(entry)->idx.oid.hash, hashsz);
 380                hdrlen += hashsz;
 381        } else {
 382                if (limit && hdrlen + datalen + hashsz >= limit) {
 383                        if (st)
 384                                close_istream(st);
 385                        free(buf);
 386                        return 0;
 387                }
 388                hashwrite(f, header, hdrlen);
 389        }
 390        if (st) {
 391                datalen = write_large_blob_data(st, f, &entry->idx.oid);
 392                close_istream(st);
 393        } else {
 394                hashwrite(f, buf, datalen);
 395                free(buf);
 396        }
 397
 398        return hdrlen + datalen;
 399}
 400
 401/* Return 0 if we will bust the pack-size limit */
 402static off_t write_reuse_object(struct hashfile *f, struct object_entry *entry,
 403                                unsigned long limit, int usable_delta)
 404{
 405        struct packed_git *p = IN_PACK(entry);
 406        struct pack_window *w_curs = NULL;
 407        struct revindex_entry *revidx;
 408        off_t offset;
 409        enum object_type type = oe_type(entry);
 410        off_t datalen;
 411        unsigned char header[MAX_PACK_OBJECT_HEADER],
 412                      dheader[MAX_PACK_OBJECT_HEADER];
 413        unsigned hdrlen;
 414        const unsigned hashsz = the_hash_algo->rawsz;
 415        unsigned long entry_size = SIZE(entry);
 416
 417        if (DELTA(entry))
 418                type = (allow_ofs_delta && DELTA(entry)->idx.offset) ?
 419                        OBJ_OFS_DELTA : OBJ_REF_DELTA;
 420        hdrlen = encode_in_pack_object_header(header, sizeof(header),
 421                                              type, entry_size);
 422
 423        offset = entry->in_pack_offset;
 424        revidx = find_pack_revindex(p, offset);
 425        datalen = revidx[1].offset - offset;
 426        if (!pack_to_stdout && p->index_version > 1 &&
 427            check_pack_crc(p, &w_curs, offset, datalen, revidx->nr)) {
 428                error(_("bad packed object CRC for %s"),
 429                      oid_to_hex(&entry->idx.oid));
 430                unuse_pack(&w_curs);
 431                return write_no_reuse_object(f, entry, limit, usable_delta);
 432        }
 433
 434        offset += entry->in_pack_header_size;
 435        datalen -= entry->in_pack_header_size;
 436
 437        if (!pack_to_stdout && p->index_version == 1 &&
 438            check_pack_inflate(p, &w_curs, offset, datalen, entry_size)) {
 439                error(_("corrupt packed object for %s"),
 440                      oid_to_hex(&entry->idx.oid));
 441                unuse_pack(&w_curs);
 442                return write_no_reuse_object(f, entry, limit, usable_delta);
 443        }
 444
 445        if (type == OBJ_OFS_DELTA) {
 446                off_t ofs = entry->idx.offset - DELTA(entry)->idx.offset;
 447                unsigned pos = sizeof(dheader) - 1;
 448                dheader[pos] = ofs & 127;
 449                while (ofs >>= 7)
 450                        dheader[--pos] = 128 | (--ofs & 127);
 451                if (limit && hdrlen + sizeof(dheader) - pos + datalen + hashsz >= limit) {
 452                        unuse_pack(&w_curs);
 453                        return 0;
 454                }
 455                hashwrite(f, header, hdrlen);
 456                hashwrite(f, dheader + pos, sizeof(dheader) - pos);
 457                hdrlen += sizeof(dheader) - pos;
 458                reused_delta++;
 459        } else if (type == OBJ_REF_DELTA) {
 460                if (limit && hdrlen + hashsz + datalen + hashsz >= limit) {
 461                        unuse_pack(&w_curs);
 462                        return 0;
 463                }
 464                hashwrite(f, header, hdrlen);
 465                hashwrite(f, DELTA(entry)->idx.oid.hash, hashsz);
 466                hdrlen += hashsz;
 467                reused_delta++;
 468        } else {
 469                if (limit && hdrlen + datalen + hashsz >= limit) {
 470                        unuse_pack(&w_curs);
 471                        return 0;
 472                }
 473                hashwrite(f, header, hdrlen);
 474        }
 475        copy_pack_data(f, p, &w_curs, offset, datalen);
 476        unuse_pack(&w_curs);
 477        reused++;
 478        return hdrlen + datalen;
 479}
 480
 481/* Return 0 if we will bust the pack-size limit */
 482static off_t write_object(struct hashfile *f,
 483                          struct object_entry *entry,
 484                          off_t write_offset)
 485{
 486        unsigned long limit;
 487        off_t len;
 488        int usable_delta, to_reuse;
 489
 490        if (!pack_to_stdout)
 491                crc32_begin(f);
 492
 493        /* apply size limit if limited packsize and not first object */
 494        if (!pack_size_limit || !nr_written)
 495                limit = 0;
 496        else if (pack_size_limit <= write_offset)
 497                /*
 498                 * the earlier object did not fit the limit; avoid
 499                 * mistaking this with unlimited (i.e. limit = 0).
 500                 */
 501                limit = 1;
 502        else
 503                limit = pack_size_limit - write_offset;
 504
 505        if (!DELTA(entry))
 506                usable_delta = 0;       /* no delta */
 507        else if (!pack_size_limit)
 508               usable_delta = 1;        /* unlimited packfile */
 509        else if (DELTA(entry)->idx.offset == (off_t)-1)
 510                usable_delta = 0;       /* base was written to another pack */
 511        else if (DELTA(entry)->idx.offset)
 512                usable_delta = 1;       /* base already exists in this pack */
 513        else
 514                usable_delta = 0;       /* base could end up in another pack */
 515
 516        if (!reuse_object)
 517                to_reuse = 0;   /* explicit */
 518        else if (!IN_PACK(entry))
 519                to_reuse = 0;   /* can't reuse what we don't have */
 520        else if (oe_type(entry) == OBJ_REF_DELTA ||
 521                 oe_type(entry) == OBJ_OFS_DELTA)
 522                                /* check_object() decided it for us ... */
 523                to_reuse = usable_delta;
 524                                /* ... but pack split may override that */
 525        else if (oe_type(entry) != entry->in_pack_type)
 526                to_reuse = 0;   /* pack has delta which is unusable */
 527        else if (DELTA(entry))
 528                to_reuse = 0;   /* we want to pack afresh */
 529        else
 530                to_reuse = 1;   /* we have it in-pack undeltified,
 531                                 * and we do not need to deltify it.
 532                                 */
 533
 534        if (!to_reuse)
 535                len = write_no_reuse_object(f, entry, limit, usable_delta);
 536        else
 537                len = write_reuse_object(f, entry, limit, usable_delta);
 538        if (!len)
 539                return 0;
 540
 541        if (usable_delta)
 542                written_delta++;
 543        written++;
 544        if (!pack_to_stdout)
 545                entry->idx.crc32 = crc32_end(f);
 546        return len;
 547}
 548
 549enum write_one_status {
 550        WRITE_ONE_SKIP = -1, /* already written */
 551        WRITE_ONE_BREAK = 0, /* writing this will bust the limit; not written */
 552        WRITE_ONE_WRITTEN = 1, /* normal */
 553        WRITE_ONE_RECURSIVE = 2 /* already scheduled to be written */
 554};
 555
 556static enum write_one_status write_one(struct hashfile *f,
 557                                       struct object_entry *e,
 558                                       off_t *offset)
 559{
 560        off_t size;
 561        int recursing;
 562
 563        /*
 564         * we set offset to 1 (which is an impossible value) to mark
 565         * the fact that this object is involved in "write its base
 566         * first before writing a deltified object" recursion.
 567         */
 568        recursing = (e->idx.offset == 1);
 569        if (recursing) {
 570                warning(_("recursive delta detected for object %s"),
 571                        oid_to_hex(&e->idx.oid));
 572                return WRITE_ONE_RECURSIVE;
 573        } else if (e->idx.offset || e->preferred_base) {
 574                /* offset is non zero if object is written already. */
 575                return WRITE_ONE_SKIP;
 576        }
 577
 578        /* if we are deltified, write out base object first. */
 579        if (DELTA(e)) {
 580                e->idx.offset = 1; /* now recurse */
 581                switch (write_one(f, DELTA(e), offset)) {
 582                case WRITE_ONE_RECURSIVE:
 583                        /* we cannot depend on this one */
 584                        SET_DELTA(e, NULL);
 585                        break;
 586                default:
 587                        break;
 588                case WRITE_ONE_BREAK:
 589                        e->idx.offset = recursing;
 590                        return WRITE_ONE_BREAK;
 591                }
 592        }
 593
 594        e->idx.offset = *offset;
 595        size = write_object(f, e, *offset);
 596        if (!size) {
 597                e->idx.offset = recursing;
 598                return WRITE_ONE_BREAK;
 599        }
 600        written_list[nr_written++] = &e->idx;
 601
 602        /* make sure off_t is sufficiently large not to wrap */
 603        if (signed_add_overflows(*offset, size))
 604                die(_("pack too large for current definition of off_t"));
 605        *offset += size;
 606        return WRITE_ONE_WRITTEN;
 607}
 608
 609static int mark_tagged(const char *path, const struct object_id *oid, int flag,
 610                       void *cb_data)
 611{
 612        struct object_id peeled;
 613        struct object_entry *entry = packlist_find(&to_pack, oid);
 614
 615        if (entry)
 616                entry->tagged = 1;
 617        if (!peel_ref(path, &peeled)) {
 618                entry = packlist_find(&to_pack, &peeled);
 619                if (entry)
 620                        entry->tagged = 1;
 621        }
 622        return 0;
 623}
 624
 625static inline void add_to_write_order(struct object_entry **wo,
 626                               unsigned int *endp,
 627                               struct object_entry *e)
 628{
 629        if (e->filled || oe_layer(&to_pack, e) != write_layer)
 630                return;
 631        wo[(*endp)++] = e;
 632        e->filled = 1;
 633}
 634
 635static void add_descendants_to_write_order(struct object_entry **wo,
 636                                           unsigned int *endp,
 637                                           struct object_entry *e)
 638{
 639        int add_to_order = 1;
 640        while (e) {
 641                if (add_to_order) {
 642                        struct object_entry *s;
 643                        /* add this node... */
 644                        add_to_write_order(wo, endp, e);
 645                        /* all its siblings... */
 646                        for (s = DELTA_SIBLING(e); s; s = DELTA_SIBLING(s)) {
 647                                add_to_write_order(wo, endp, s);
 648                        }
 649                }
 650                /* drop down a level to add left subtree nodes if possible */
 651                if (DELTA_CHILD(e)) {
 652                        add_to_order = 1;
 653                        e = DELTA_CHILD(e);
 654                } else {
 655                        add_to_order = 0;
 656                        /* our sibling might have some children, it is next */
 657                        if (DELTA_SIBLING(e)) {
 658                                e = DELTA_SIBLING(e);
 659                                continue;
 660                        }
 661                        /* go back to our parent node */
 662                        e = DELTA(e);
 663                        while (e && !DELTA_SIBLING(e)) {
 664                                /* we're on the right side of a subtree, keep
 665                                 * going up until we can go right again */
 666                                e = DELTA(e);
 667                        }
 668                        if (!e) {
 669                                /* done- we hit our original root node */
 670                                return;
 671                        }
 672                        /* pass it off to sibling at this level */
 673                        e = DELTA_SIBLING(e);
 674                }
 675        };
 676}
 677
 678static void add_family_to_write_order(struct object_entry **wo,
 679                                      unsigned int *endp,
 680                                      struct object_entry *e)
 681{
 682        struct object_entry *root;
 683
 684        for (root = e; DELTA(root); root = DELTA(root))
 685                ; /* nothing */
 686        add_descendants_to_write_order(wo, endp, root);
 687}
 688
 689static void compute_layer_order(struct object_entry **wo, unsigned int *wo_end)
 690{
 691        unsigned int i, last_untagged;
 692        struct object_entry *objects = to_pack.objects;
 693
 694        for (i = 0; i < to_pack.nr_objects; i++) {
 695                if (objects[i].tagged)
 696                        break;
 697                add_to_write_order(wo, wo_end, &objects[i]);
 698        }
 699        last_untagged = i;
 700
 701        /*
 702         * Then fill all the tagged tips.
 703         */
 704        for (; i < to_pack.nr_objects; i++) {
 705                if (objects[i].tagged)
 706                        add_to_write_order(wo, wo_end, &objects[i]);
 707        }
 708
 709        /*
 710         * And then all remaining commits and tags.
 711         */
 712        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 713                if (oe_type(&objects[i]) != OBJ_COMMIT &&
 714                    oe_type(&objects[i]) != OBJ_TAG)
 715                        continue;
 716                add_to_write_order(wo, wo_end, &objects[i]);
 717        }
 718
 719        /*
 720         * And then all the trees.
 721         */
 722        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 723                if (oe_type(&objects[i]) != OBJ_TREE)
 724                        continue;
 725                add_to_write_order(wo, wo_end, &objects[i]);
 726        }
 727
 728        /*
 729         * Finally all the rest in really tight order
 730         */
 731        for (i = last_untagged; i < to_pack.nr_objects; i++) {
 732                if (!objects[i].filled && oe_layer(&to_pack, &objects[i]) == write_layer)
 733                        add_family_to_write_order(wo, wo_end, &objects[i]);
 734        }
 735}
 736
 737static struct object_entry **compute_write_order(void)
 738{
 739        uint32_t max_layers = 1;
 740        unsigned int i, wo_end;
 741
 742        struct object_entry **wo;
 743        struct object_entry *objects = to_pack.objects;
 744
 745        for (i = 0; i < to_pack.nr_objects; i++) {
 746                objects[i].tagged = 0;
 747                objects[i].filled = 0;
 748                SET_DELTA_CHILD(&objects[i], NULL);
 749                SET_DELTA_SIBLING(&objects[i], NULL);
 750        }
 751
 752        /*
 753         * Fully connect delta_child/delta_sibling network.
 754         * Make sure delta_sibling is sorted in the original
 755         * recency order.
 756         */
 757        for (i = to_pack.nr_objects; i > 0;) {
 758                struct object_entry *e = &objects[--i];
 759                if (!DELTA(e))
 760                        continue;
 761                /* Mark me as the first child */
 762                e->delta_sibling_idx = DELTA(e)->delta_child_idx;
 763                SET_DELTA_CHILD(DELTA(e), e);
 764        }
 765
 766        /*
 767         * Mark objects that are at the tip of tags.
 768         */
 769        for_each_tag_ref(mark_tagged, NULL);
 770
 771        if (use_delta_islands)
 772                max_layers = compute_pack_layers(&to_pack);
 773
 774        ALLOC_ARRAY(wo, to_pack.nr_objects);
 775        wo_end = 0;
 776
 777        for (; write_layer < max_layers; ++write_layer)
 778                compute_layer_order(wo, &wo_end);
 779
 780        if (wo_end != to_pack.nr_objects)
 781                die(_("ordered %u objects, expected %"PRIu32),
 782                    wo_end, to_pack.nr_objects);
 783
 784        return wo;
 785}
 786
 787static off_t write_reused_pack(struct hashfile *f)
 788{
 789        unsigned char buffer[8192];
 790        off_t to_write, total;
 791        int fd;
 792
 793        if (!is_pack_valid(reuse_packfile))
 794                die(_("packfile is invalid: %s"), reuse_packfile->pack_name);
 795
 796        fd = git_open(reuse_packfile->pack_name);
 797        if (fd < 0)
 798                die_errno(_("unable to open packfile for reuse: %s"),
 799                          reuse_packfile->pack_name);
 800
 801        if (lseek(fd, sizeof(struct pack_header), SEEK_SET) == -1)
 802                die_errno(_("unable to seek in reused packfile"));
 803
 804        if (reuse_packfile_offset < 0)
 805                reuse_packfile_offset = reuse_packfile->pack_size - the_hash_algo->rawsz;
 806
 807        total = to_write = reuse_packfile_offset - sizeof(struct pack_header);
 808
 809        while (to_write) {
 810                int read_pack = xread(fd, buffer, sizeof(buffer));
 811
 812                if (read_pack <= 0)
 813                        die_errno(_("unable to read from reused packfile"));
 814
 815                if (read_pack > to_write)
 816                        read_pack = to_write;
 817
 818                hashwrite(f, buffer, read_pack);
 819                to_write -= read_pack;
 820
 821                /*
 822                 * We don't know the actual number of objects written,
 823                 * only how many bytes written, how many bytes total, and
 824                 * how many objects total. So we can fake it by pretending all
 825                 * objects we are writing are the same size. This gives us a
 826                 * smooth progress meter, and at the end it matches the true
 827                 * answer.
 828                 */
 829                written = reuse_packfile_objects *
 830                                (((double)(total - to_write)) / total);
 831                display_progress(progress_state, written);
 832        }
 833
 834        close(fd);
 835        written = reuse_packfile_objects;
 836        display_progress(progress_state, written);
 837        return reuse_packfile_offset - sizeof(struct pack_header);
 838}
 839
 840static const char no_split_warning[] = N_(
 841"disabling bitmap writing, packs are split due to pack.packSizeLimit"
 842);
 843
 844static void write_pack_file(void)
 845{
 846        uint32_t i = 0, j;
 847        struct hashfile *f;
 848        off_t offset;
 849        uint32_t nr_remaining = nr_result;
 850        time_t last_mtime = 0;
 851        struct object_entry **write_order;
 852
 853        if (progress > pack_to_stdout)
 854                progress_state = start_progress(_("Writing objects"), nr_result);
 855        ALLOC_ARRAY(written_list, to_pack.nr_objects);
 856        write_order = compute_write_order();
 857
 858        do {
 859                struct object_id oid;
 860                char *pack_tmp_name = NULL;
 861
 862                if (pack_to_stdout)
 863                        f = hashfd_throughput(1, "<stdout>", progress_state);
 864                else
 865                        f = create_tmp_packfile(&pack_tmp_name);
 866
 867                offset = write_pack_header(f, nr_remaining);
 868
 869                if (reuse_packfile) {
 870                        off_t packfile_size;
 871                        assert(pack_to_stdout);
 872
 873                        packfile_size = write_reused_pack(f);
 874                        offset += packfile_size;
 875                }
 876
 877                nr_written = 0;
 878                for (; i < to_pack.nr_objects; i++) {
 879                        struct object_entry *e = write_order[i];
 880                        if (write_one(f, e, &offset) == WRITE_ONE_BREAK)
 881                                break;
 882                        display_progress(progress_state, written);
 883                }
 884
 885                /*
 886                 * Did we write the wrong # entries in the header?
 887                 * If so, rewrite it like in fast-import
 888                 */
 889                if (pack_to_stdout) {
 890                        finalize_hashfile(f, oid.hash, CSUM_HASH_IN_STREAM | CSUM_CLOSE);
 891                } else if (nr_written == nr_remaining) {
 892                        finalize_hashfile(f, oid.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 893                } else {
 894                        int fd = finalize_hashfile(f, oid.hash, 0);
 895                        fixup_pack_header_footer(fd, oid.hash, pack_tmp_name,
 896                                                 nr_written, oid.hash, offset);
 897                        close(fd);
 898                        if (write_bitmap_index) {
 899                                if (write_bitmap_index != WRITE_BITMAP_QUIET)
 900                                        warning(_(no_split_warning));
 901                                write_bitmap_index = 0;
 902                        }
 903                }
 904
 905                if (!pack_to_stdout) {
 906                        struct stat st;
 907                        struct strbuf tmpname = STRBUF_INIT;
 908
 909                        /*
 910                         * Packs are runtime accessed in their mtime
 911                         * order since newer packs are more likely to contain
 912                         * younger objects.  So if we are creating multiple
 913                         * packs then we should modify the mtime of later ones
 914                         * to preserve this property.
 915                         */
 916                        if (stat(pack_tmp_name, &st) < 0) {
 917                                warning_errno(_("failed to stat %s"), pack_tmp_name);
 918                        } else if (!last_mtime) {
 919                                last_mtime = st.st_mtime;
 920                        } else {
 921                                struct utimbuf utb;
 922                                utb.actime = st.st_atime;
 923                                utb.modtime = --last_mtime;
 924                                if (utime(pack_tmp_name, &utb) < 0)
 925                                        warning_errno(_("failed utime() on %s"), pack_tmp_name);
 926                        }
 927
 928                        strbuf_addf(&tmpname, "%s-", base_name);
 929
 930                        if (write_bitmap_index) {
 931                                bitmap_writer_set_checksum(oid.hash);
 932                                bitmap_writer_build_type_index(
 933                                        &to_pack, written_list, nr_written);
 934                        }
 935
 936                        finish_tmp_packfile(&tmpname, pack_tmp_name,
 937                                            written_list, nr_written,
 938                                            &pack_idx_opts, oid.hash);
 939
 940                        if (write_bitmap_index) {
 941                                strbuf_addf(&tmpname, "%s.bitmap", oid_to_hex(&oid));
 942
 943                                stop_progress(&progress_state);
 944
 945                                bitmap_writer_show_progress(progress);
 946                                bitmap_writer_reuse_bitmaps(&to_pack);
 947                                bitmap_writer_select_commits(indexed_commits, indexed_commits_nr, -1);
 948                                bitmap_writer_build(&to_pack);
 949                                bitmap_writer_finish(written_list, nr_written,
 950                                                     tmpname.buf, write_bitmap_options);
 951                                write_bitmap_index = 0;
 952                        }
 953
 954                        strbuf_release(&tmpname);
 955                        free(pack_tmp_name);
 956                        puts(oid_to_hex(&oid));
 957                }
 958
 959                /* mark written objects as written to previous pack */
 960                for (j = 0; j < nr_written; j++) {
 961                        written_list[j]->offset = (off_t)-1;
 962                }
 963                nr_remaining -= nr_written;
 964        } while (nr_remaining && i < to_pack.nr_objects);
 965
 966        free(written_list);
 967        free(write_order);
 968        stop_progress(&progress_state);
 969        if (written != nr_result)
 970                die(_("wrote %"PRIu32" objects while expecting %"PRIu32),
 971                    written, nr_result);
 972        trace2_data_intmax("pack-objects", the_repository,
 973                           "write_pack_file/wrote", nr_result);
 974}
 975
 976static int no_try_delta(const char *path)
 977{
 978        static struct attr_check *check;
 979
 980        if (!check)
 981                check = attr_check_initl("delta", NULL);
 982        git_check_attr(the_repository->index, path, check);
 983        if (ATTR_FALSE(check->items[0].value))
 984                return 1;
 985        return 0;
 986}
 987
 988/*
 989 * When adding an object, check whether we have already added it
 990 * to our packing list. If so, we can skip. However, if we are
 991 * being asked to excludei t, but the previous mention was to include
 992 * it, make sure to adjust its flags and tweak our numbers accordingly.
 993 *
 994 * As an optimization, we pass out the index position where we would have
 995 * found the item, since that saves us from having to look it up again a
 996 * few lines later when we want to add the new entry.
 997 */
 998static int have_duplicate_entry(const struct object_id *oid,
 999                                int exclude)
1000{
1001        struct object_entry *entry;
1002
1003        entry = packlist_find(&to_pack, oid);
1004        if (!entry)
1005                return 0;
1006
1007        if (exclude) {
1008                if (!entry->preferred_base)
1009                        nr_result--;
1010                entry->preferred_base = 1;
1011        }
1012
1013        return 1;
1014}
1015
1016static int want_found_object(int exclude, struct packed_git *p)
1017{
1018        if (exclude)
1019                return 1;
1020        if (incremental)
1021                return 0;
1022
1023        /*
1024         * When asked to do --local (do not include an object that appears in a
1025         * pack we borrow from elsewhere) or --honor-pack-keep (do not include
1026         * an object that appears in a pack marked with .keep), finding a pack
1027         * that matches the criteria is sufficient for us to decide to omit it.
1028         * However, even if this pack does not satisfy the criteria, we need to
1029         * make sure no copy of this object appears in _any_ pack that makes us
1030         * to omit the object, so we need to check all the packs.
1031         *
1032         * We can however first check whether these options can possible matter;
1033         * if they do not matter we know we want the object in generated pack.
1034         * Otherwise, we signal "-1" at the end to tell the caller that we do
1035         * not know either way, and it needs to check more packs.
1036         */
1037        if (!ignore_packed_keep_on_disk &&
1038            !ignore_packed_keep_in_core &&
1039            (!local || !have_non_local_packs))
1040                return 1;
1041
1042        if (local && !p->pack_local)
1043                return 0;
1044        if (p->pack_local &&
1045            ((ignore_packed_keep_on_disk && p->pack_keep) ||
1046             (ignore_packed_keep_in_core && p->pack_keep_in_core)))
1047                return 0;
1048
1049        /* we don't know yet; keep looking for more packs */
1050        return -1;
1051}
1052
1053/*
1054 * Check whether we want the object in the pack (e.g., we do not want
1055 * objects found in non-local stores if the "--local" option was used).
1056 *
1057 * If the caller already knows an existing pack it wants to take the object
1058 * from, that is passed in *found_pack and *found_offset; otherwise this
1059 * function finds if there is any pack that has the object and returns the pack
1060 * and its offset in these variables.
1061 */
1062static int want_object_in_pack(const struct object_id *oid,
1063                               int exclude,
1064                               struct packed_git **found_pack,
1065                               off_t *found_offset)
1066{
1067        int want;
1068        struct list_head *pos;
1069        struct multi_pack_index *m;
1070
1071        if (!exclude && local && has_loose_object_nonlocal(oid))
1072                return 0;
1073
1074        /*
1075         * If we already know the pack object lives in, start checks from that
1076         * pack - in the usual case when neither --local was given nor .keep files
1077         * are present we will determine the answer right now.
1078         */
1079        if (*found_pack) {
1080                want = want_found_object(exclude, *found_pack);
1081                if (want != -1)
1082                        return want;
1083        }
1084
1085        for (m = get_multi_pack_index(the_repository); m; m = m->next) {
1086                struct pack_entry e;
1087                if (fill_midx_entry(the_repository, oid, &e, m)) {
1088                        struct packed_git *p = e.p;
1089                        off_t offset;
1090
1091                        if (p == *found_pack)
1092                                offset = *found_offset;
1093                        else
1094                                offset = find_pack_entry_one(oid->hash, p);
1095
1096                        if (offset) {
1097                                if (!*found_pack) {
1098                                        if (!is_pack_valid(p))
1099                                                continue;
1100                                        *found_offset = offset;
1101                                        *found_pack = p;
1102                                }
1103                                want = want_found_object(exclude, p);
1104                                if (want != -1)
1105                                        return want;
1106                        }
1107                }
1108        }
1109
1110        list_for_each(pos, get_packed_git_mru(the_repository)) {
1111                struct packed_git *p = list_entry(pos, struct packed_git, mru);
1112                off_t offset;
1113
1114                if (p == *found_pack)
1115                        offset = *found_offset;
1116                else
1117                        offset = find_pack_entry_one(oid->hash, p);
1118
1119                if (offset) {
1120                        if (!*found_pack) {
1121                                if (!is_pack_valid(p))
1122                                        continue;
1123                                *found_offset = offset;
1124                                *found_pack = p;
1125                        }
1126                        want = want_found_object(exclude, p);
1127                        if (!exclude && want > 0)
1128                                list_move(&p->mru,
1129                                          get_packed_git_mru(the_repository));
1130                        if (want != -1)
1131                                return want;
1132                }
1133        }
1134
1135        return 1;
1136}
1137
1138static void create_object_entry(const struct object_id *oid,
1139                                enum object_type type,
1140                                uint32_t hash,
1141                                int exclude,
1142                                int no_try_delta,
1143                                struct packed_git *found_pack,
1144                                off_t found_offset)
1145{
1146        struct object_entry *entry;
1147
1148        entry = packlist_alloc(&to_pack, oid);
1149        entry->hash = hash;
1150        oe_set_type(entry, type);
1151        if (exclude)
1152                entry->preferred_base = 1;
1153        else
1154                nr_result++;
1155        if (found_pack) {
1156                oe_set_in_pack(&to_pack, entry, found_pack);
1157                entry->in_pack_offset = found_offset;
1158        }
1159
1160        entry->no_try_delta = no_try_delta;
1161}
1162
1163static const char no_closure_warning[] = N_(
1164"disabling bitmap writing, as some objects are not being packed"
1165);
1166
1167static int add_object_entry(const struct object_id *oid, enum object_type type,
1168                            const char *name, int exclude)
1169{
1170        struct packed_git *found_pack = NULL;
1171        off_t found_offset = 0;
1172
1173        display_progress(progress_state, ++nr_seen);
1174
1175        if (have_duplicate_entry(oid, exclude))
1176                return 0;
1177
1178        if (!want_object_in_pack(oid, exclude, &found_pack, &found_offset)) {
1179                /* The pack is missing an object, so it will not have closure */
1180                if (write_bitmap_index) {
1181                        if (write_bitmap_index != WRITE_BITMAP_QUIET)
1182                                warning(_(no_closure_warning));
1183                        write_bitmap_index = 0;
1184                }
1185                return 0;
1186        }
1187
1188        create_object_entry(oid, type, pack_name_hash(name),
1189                            exclude, name && no_try_delta(name),
1190                            found_pack, found_offset);
1191        return 1;
1192}
1193
1194static int add_object_entry_from_bitmap(const struct object_id *oid,
1195                                        enum object_type type,
1196                                        int flags, uint32_t name_hash,
1197                                        struct packed_git *pack, off_t offset)
1198{
1199        display_progress(progress_state, ++nr_seen);
1200
1201        if (have_duplicate_entry(oid, 0))
1202                return 0;
1203
1204        if (!want_object_in_pack(oid, 0, &pack, &offset))
1205                return 0;
1206
1207        create_object_entry(oid, type, name_hash, 0, 0, pack, offset);
1208        return 1;
1209}
1210
1211struct pbase_tree_cache {
1212        struct object_id oid;
1213        int ref;
1214        int temporary;
1215        void *tree_data;
1216        unsigned long tree_size;
1217};
1218
1219static struct pbase_tree_cache *(pbase_tree_cache[256]);
1220static int pbase_tree_cache_ix(const struct object_id *oid)
1221{
1222        return oid->hash[0] % ARRAY_SIZE(pbase_tree_cache);
1223}
1224static int pbase_tree_cache_ix_incr(int ix)
1225{
1226        return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
1227}
1228
1229static struct pbase_tree {
1230        struct pbase_tree *next;
1231        /* This is a phony "cache" entry; we are not
1232         * going to evict it or find it through _get()
1233         * mechanism -- this is for the toplevel node that
1234         * would almost always change with any commit.
1235         */
1236        struct pbase_tree_cache pcache;
1237} *pbase_tree;
1238
1239static struct pbase_tree_cache *pbase_tree_get(const struct object_id *oid)
1240{
1241        struct pbase_tree_cache *ent, *nent;
1242        void *data;
1243        unsigned long size;
1244        enum object_type type;
1245        int neigh;
1246        int my_ix = pbase_tree_cache_ix(oid);
1247        int available_ix = -1;
1248
1249        /* pbase-tree-cache acts as a limited hashtable.
1250         * your object will be found at your index or within a few
1251         * slots after that slot if it is cached.
1252         */
1253        for (neigh = 0; neigh < 8; neigh++) {
1254                ent = pbase_tree_cache[my_ix];
1255                if (ent && oideq(&ent->oid, oid)) {
1256                        ent->ref++;
1257                        return ent;
1258                }
1259                else if (((available_ix < 0) && (!ent || !ent->ref)) ||
1260                         ((0 <= available_ix) &&
1261                          (!ent && pbase_tree_cache[available_ix])))
1262                        available_ix = my_ix;
1263                if (!ent)
1264                        break;
1265                my_ix = pbase_tree_cache_ix_incr(my_ix);
1266        }
1267
1268        /* Did not find one.  Either we got a bogus request or
1269         * we need to read and perhaps cache.
1270         */
1271        data = read_object_file(oid, &type, &size);
1272        if (!data)
1273                return NULL;
1274        if (type != OBJ_TREE) {
1275                free(data);
1276                return NULL;
1277        }
1278
1279        /* We need to either cache or return a throwaway copy */
1280
1281        if (available_ix < 0)
1282                ent = NULL;
1283        else {
1284                ent = pbase_tree_cache[available_ix];
1285                my_ix = available_ix;
1286        }
1287
1288        if (!ent) {
1289                nent = xmalloc(sizeof(*nent));
1290                nent->temporary = (available_ix < 0);
1291        }
1292        else {
1293                /* evict and reuse */
1294                free(ent->tree_data);
1295                nent = ent;
1296        }
1297        oidcpy(&nent->oid, oid);
1298        nent->tree_data = data;
1299        nent->tree_size = size;
1300        nent->ref = 1;
1301        if (!nent->temporary)
1302                pbase_tree_cache[my_ix] = nent;
1303        return nent;
1304}
1305
1306static void pbase_tree_put(struct pbase_tree_cache *cache)
1307{
1308        if (!cache->temporary) {
1309                cache->ref--;
1310                return;
1311        }
1312        free(cache->tree_data);
1313        free(cache);
1314}
1315
1316static int name_cmp_len(const char *name)
1317{
1318        int i;
1319        for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
1320                ;
1321        return i;
1322}
1323
1324static void add_pbase_object(struct tree_desc *tree,
1325                             const char *name,
1326                             int cmplen,
1327                             const char *fullname)
1328{
1329        struct name_entry entry;
1330        int cmp;
1331
1332        while (tree_entry(tree,&entry)) {
1333                if (S_ISGITLINK(entry.mode))
1334                        continue;
1335                cmp = tree_entry_len(&entry) != cmplen ? 1 :
1336                      memcmp(name, entry.path, cmplen);
1337                if (cmp > 0)
1338                        continue;
1339                if (cmp < 0)
1340                        return;
1341                if (name[cmplen] != '/') {
1342                        add_object_entry(&entry.oid,
1343                                         object_type(entry.mode),
1344                                         fullname, 1);
1345                        return;
1346                }
1347                if (S_ISDIR(entry.mode)) {
1348                        struct tree_desc sub;
1349                        struct pbase_tree_cache *tree;
1350                        const char *down = name+cmplen+1;
1351                        int downlen = name_cmp_len(down);
1352
1353                        tree = pbase_tree_get(&entry.oid);
1354                        if (!tree)
1355                                return;
1356                        init_tree_desc(&sub, tree->tree_data, tree->tree_size);
1357
1358                        add_pbase_object(&sub, down, downlen, fullname);
1359                        pbase_tree_put(tree);
1360                }
1361        }
1362}
1363
1364static unsigned *done_pbase_paths;
1365static int done_pbase_paths_num;
1366static int done_pbase_paths_alloc;
1367static int done_pbase_path_pos(unsigned hash)
1368{
1369        int lo = 0;
1370        int hi = done_pbase_paths_num;
1371        while (lo < hi) {
1372                int mi = lo + (hi - lo) / 2;
1373                if (done_pbase_paths[mi] == hash)
1374                        return mi;
1375                if (done_pbase_paths[mi] < hash)
1376                        hi = mi;
1377                else
1378                        lo = mi + 1;
1379        }
1380        return -lo-1;
1381}
1382
1383static int check_pbase_path(unsigned hash)
1384{
1385        int pos = done_pbase_path_pos(hash);
1386        if (0 <= pos)
1387                return 1;
1388        pos = -pos - 1;
1389        ALLOC_GROW(done_pbase_paths,
1390                   done_pbase_paths_num + 1,
1391                   done_pbase_paths_alloc);
1392        done_pbase_paths_num++;
1393        if (pos < done_pbase_paths_num)
1394                MOVE_ARRAY(done_pbase_paths + pos + 1, done_pbase_paths + pos,
1395                           done_pbase_paths_num - pos - 1);
1396        done_pbase_paths[pos] = hash;
1397        return 0;
1398}
1399
1400static void add_preferred_base_object(const char *name)
1401{
1402        struct pbase_tree *it;
1403        int cmplen;
1404        unsigned hash = pack_name_hash(name);
1405
1406        if (!num_preferred_base || check_pbase_path(hash))
1407                return;
1408
1409        cmplen = name_cmp_len(name);
1410        for (it = pbase_tree; it; it = it->next) {
1411                if (cmplen == 0) {
1412                        add_object_entry(&it->pcache.oid, OBJ_TREE, NULL, 1);
1413                }
1414                else {
1415                        struct tree_desc tree;
1416                        init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1417                        add_pbase_object(&tree, name, cmplen, name);
1418                }
1419        }
1420}
1421
1422static void add_preferred_base(struct object_id *oid)
1423{
1424        struct pbase_tree *it;
1425        void *data;
1426        unsigned long size;
1427        struct object_id tree_oid;
1428
1429        if (window <= num_preferred_base++)
1430                return;
1431
1432        data = read_object_with_reference(the_repository, oid,
1433                                          tree_type, &size, &tree_oid);
1434        if (!data)
1435                return;
1436
1437        for (it = pbase_tree; it; it = it->next) {
1438                if (oideq(&it->pcache.oid, &tree_oid)) {
1439                        free(data);
1440                        return;
1441                }
1442        }
1443
1444        it = xcalloc(1, sizeof(*it));
1445        it->next = pbase_tree;
1446        pbase_tree = it;
1447
1448        oidcpy(&it->pcache.oid, &tree_oid);
1449        it->pcache.tree_data = data;
1450        it->pcache.tree_size = size;
1451}
1452
1453static void cleanup_preferred_base(void)
1454{
1455        struct pbase_tree *it;
1456        unsigned i;
1457
1458        it = pbase_tree;
1459        pbase_tree = NULL;
1460        while (it) {
1461                struct pbase_tree *tmp = it;
1462                it = tmp->next;
1463                free(tmp->pcache.tree_data);
1464                free(tmp);
1465        }
1466
1467        for (i = 0; i < ARRAY_SIZE(pbase_tree_cache); i++) {
1468                if (!pbase_tree_cache[i])
1469                        continue;
1470                free(pbase_tree_cache[i]->tree_data);
1471                FREE_AND_NULL(pbase_tree_cache[i]);
1472        }
1473
1474        FREE_AND_NULL(done_pbase_paths);
1475        done_pbase_paths_num = done_pbase_paths_alloc = 0;
1476}
1477
1478/*
1479 * Return 1 iff the object specified by "delta" can be sent
1480 * literally as a delta against the base in "base_sha1". If
1481 * so, then *base_out will point to the entry in our packing
1482 * list, or NULL if we must use the external-base list.
1483 *
1484 * Depth value does not matter - find_deltas() will
1485 * never consider reused delta as the base object to
1486 * deltify other objects against, in order to avoid
1487 * circular deltas.
1488 */
1489static int can_reuse_delta(const unsigned char *base_sha1,
1490                           struct object_entry *delta,
1491                           struct object_entry **base_out)
1492{
1493        struct object_entry *base;
1494        struct object_id base_oid;
1495
1496        if (!base_sha1)
1497                return 0;
1498
1499        oidread(&base_oid, base_sha1);
1500
1501        /*
1502         * First see if we're already sending the base (or it's explicitly in
1503         * our "excluded" list).
1504         */
1505        base = packlist_find(&to_pack, &base_oid);
1506        if (base) {
1507                if (!in_same_island(&delta->idx.oid, &base->idx.oid))
1508                        return 0;
1509                *base_out = base;
1510                return 1;
1511        }
1512
1513        /*
1514         * Otherwise, reachability bitmaps may tell us if the receiver has it,
1515         * even if it was buried too deep in history to make it into the
1516         * packing list.
1517         */
1518        if (thin && bitmap_has_oid_in_uninteresting(bitmap_git, &base_oid)) {
1519                if (use_delta_islands) {
1520                        if (!in_same_island(&delta->idx.oid, &base_oid))
1521                                return 0;
1522                }
1523                *base_out = NULL;
1524                return 1;
1525        }
1526
1527        return 0;
1528}
1529
1530static void check_object(struct object_entry *entry)
1531{
1532        unsigned long canonical_size;
1533
1534        if (IN_PACK(entry)) {
1535                struct packed_git *p = IN_PACK(entry);
1536                struct pack_window *w_curs = NULL;
1537                const unsigned char *base_ref = NULL;
1538                struct object_entry *base_entry;
1539                unsigned long used, used_0;
1540                unsigned long avail;
1541                off_t ofs;
1542                unsigned char *buf, c;
1543                enum object_type type;
1544                unsigned long in_pack_size;
1545
1546                buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1547
1548                /*
1549                 * We want in_pack_type even if we do not reuse delta
1550                 * since non-delta representations could still be reused.
1551                 */
1552                used = unpack_object_header_buffer(buf, avail,
1553                                                   &type,
1554                                                   &in_pack_size);
1555                if (used == 0)
1556                        goto give_up;
1557
1558                if (type < 0)
1559                        BUG("invalid type %d", type);
1560                entry->in_pack_type = type;
1561
1562                /*
1563                 * Determine if this is a delta and if so whether we can
1564                 * reuse it or not.  Otherwise let's find out as cheaply as
1565                 * possible what the actual type and size for this object is.
1566                 */
1567                switch (entry->in_pack_type) {
1568                default:
1569                        /* Not a delta hence we've already got all we need. */
1570                        oe_set_type(entry, entry->in_pack_type);
1571                        SET_SIZE(entry, in_pack_size);
1572                        entry->in_pack_header_size = used;
1573                        if (oe_type(entry) < OBJ_COMMIT || oe_type(entry) > OBJ_BLOB)
1574                                goto give_up;
1575                        unuse_pack(&w_curs);
1576                        return;
1577                case OBJ_REF_DELTA:
1578                        if (reuse_delta && !entry->preferred_base)
1579                                base_ref = use_pack(p, &w_curs,
1580                                                entry->in_pack_offset + used, NULL);
1581                        entry->in_pack_header_size = used + the_hash_algo->rawsz;
1582                        break;
1583                case OBJ_OFS_DELTA:
1584                        buf = use_pack(p, &w_curs,
1585                                       entry->in_pack_offset + used, NULL);
1586                        used_0 = 0;
1587                        c = buf[used_0++];
1588                        ofs = c & 127;
1589                        while (c & 128) {
1590                                ofs += 1;
1591                                if (!ofs || MSB(ofs, 7)) {
1592                                        error(_("delta base offset overflow in pack for %s"),
1593                                              oid_to_hex(&entry->idx.oid));
1594                                        goto give_up;
1595                                }
1596                                c = buf[used_0++];
1597                                ofs = (ofs << 7) + (c & 127);
1598                        }
1599                        ofs = entry->in_pack_offset - ofs;
1600                        if (ofs <= 0 || ofs >= entry->in_pack_offset) {
1601                                error(_("delta base offset out of bound for %s"),
1602                                      oid_to_hex(&entry->idx.oid));
1603                                goto give_up;
1604                        }
1605                        if (reuse_delta && !entry->preferred_base) {
1606                                struct revindex_entry *revidx;
1607                                revidx = find_pack_revindex(p, ofs);
1608                                if (!revidx)
1609                                        goto give_up;
1610                                base_ref = nth_packed_object_sha1(p, revidx->nr);
1611                        }
1612                        entry->in_pack_header_size = used + used_0;
1613                        break;
1614                }
1615
1616                if (can_reuse_delta(base_ref, entry, &base_entry)) {
1617                        oe_set_type(entry, entry->in_pack_type);
1618                        SET_SIZE(entry, in_pack_size); /* delta size */
1619                        SET_DELTA_SIZE(entry, in_pack_size);
1620
1621                        if (base_entry) {
1622                                SET_DELTA(entry, base_entry);
1623                                entry->delta_sibling_idx = base_entry->delta_child_idx;
1624                                SET_DELTA_CHILD(base_entry, entry);
1625                        } else {
1626                                SET_DELTA_EXT(entry, base_ref);
1627                        }
1628
1629                        unuse_pack(&w_curs);
1630                        return;
1631                }
1632
1633                if (oe_type(entry)) {
1634                        off_t delta_pos;
1635
1636                        /*
1637                         * This must be a delta and we already know what the
1638                         * final object type is.  Let's extract the actual
1639                         * object size from the delta header.
1640                         */
1641                        delta_pos = entry->in_pack_offset + entry->in_pack_header_size;
1642                        canonical_size = get_size_from_delta(p, &w_curs, delta_pos);
1643                        if (canonical_size == 0)
1644                                goto give_up;
1645                        SET_SIZE(entry, canonical_size);
1646                        unuse_pack(&w_curs);
1647                        return;
1648                }
1649
1650                /*
1651                 * No choice but to fall back to the recursive delta walk
1652                 * with oid_object_info() to find about the object type
1653                 * at this point...
1654                 */
1655                give_up:
1656                unuse_pack(&w_curs);
1657        }
1658
1659        oe_set_type(entry,
1660                    oid_object_info(the_repository, &entry->idx.oid, &canonical_size));
1661        if (entry->type_valid) {
1662                SET_SIZE(entry, canonical_size);
1663        } else {
1664                /*
1665                 * Bad object type is checked in prepare_pack().  This is
1666                 * to permit a missing preferred base object to be ignored
1667                 * as a preferred base.  Doing so can result in a larger
1668                 * pack file, but the transfer will still take place.
1669                 */
1670        }
1671}
1672
1673static int pack_offset_sort(const void *_a, const void *_b)
1674{
1675        const struct object_entry *a = *(struct object_entry **)_a;
1676        const struct object_entry *b = *(struct object_entry **)_b;
1677        const struct packed_git *a_in_pack = IN_PACK(a);
1678        const struct packed_git *b_in_pack = IN_PACK(b);
1679
1680        /* avoid filesystem trashing with loose objects */
1681        if (!a_in_pack && !b_in_pack)
1682                return oidcmp(&a->idx.oid, &b->idx.oid);
1683
1684        if (a_in_pack < b_in_pack)
1685                return -1;
1686        if (a_in_pack > b_in_pack)
1687                return 1;
1688        return a->in_pack_offset < b->in_pack_offset ? -1 :
1689                        (a->in_pack_offset > b->in_pack_offset);
1690}
1691
1692/*
1693 * Drop an on-disk delta we were planning to reuse. Naively, this would
1694 * just involve blanking out the "delta" field, but we have to deal
1695 * with some extra book-keeping:
1696 *
1697 *   1. Removing ourselves from the delta_sibling linked list.
1698 *
1699 *   2. Updating our size/type to the non-delta representation. These were
1700 *      either not recorded initially (size) or overwritten with the delta type
1701 *      (type) when check_object() decided to reuse the delta.
1702 *
1703 *   3. Resetting our delta depth, as we are now a base object.
1704 */
1705static void drop_reused_delta(struct object_entry *entry)
1706{
1707        unsigned *idx = &to_pack.objects[entry->delta_idx - 1].delta_child_idx;
1708        struct object_info oi = OBJECT_INFO_INIT;
1709        enum object_type type;
1710        unsigned long size;
1711
1712        while (*idx) {
1713                struct object_entry *oe = &to_pack.objects[*idx - 1];
1714
1715                if (oe == entry)
1716                        *idx = oe->delta_sibling_idx;
1717                else
1718                        idx = &oe->delta_sibling_idx;
1719        }
1720        SET_DELTA(entry, NULL);
1721        entry->depth = 0;
1722
1723        oi.sizep = &size;
1724        oi.typep = &type;
1725        if (packed_object_info(the_repository, IN_PACK(entry), entry->in_pack_offset, &oi) < 0) {
1726                /*
1727                 * We failed to get the info from this pack for some reason;
1728                 * fall back to oid_object_info, which may find another copy.
1729                 * And if that fails, the error will be recorded in oe_type(entry)
1730                 * and dealt with in prepare_pack().
1731                 */
1732                oe_set_type(entry,
1733                            oid_object_info(the_repository, &entry->idx.oid, &size));
1734        } else {
1735                oe_set_type(entry, type);
1736        }
1737        SET_SIZE(entry, size);
1738}
1739
1740/*
1741 * Follow the chain of deltas from this entry onward, throwing away any links
1742 * that cause us to hit a cycle (as determined by the DFS state flags in
1743 * the entries).
1744 *
1745 * We also detect too-long reused chains that would violate our --depth
1746 * limit.
1747 */
1748static void break_delta_chains(struct object_entry *entry)
1749{
1750        /*
1751         * The actual depth of each object we will write is stored as an int,
1752         * as it cannot exceed our int "depth" limit. But before we break
1753         * changes based no that limit, we may potentially go as deep as the
1754         * number of objects, which is elsewhere bounded to a uint32_t.
1755         */
1756        uint32_t total_depth;
1757        struct object_entry *cur, *next;
1758
1759        for (cur = entry, total_depth = 0;
1760             cur;
1761             cur = DELTA(cur), total_depth++) {
1762                if (cur->dfs_state == DFS_DONE) {
1763                        /*
1764                         * We've already seen this object and know it isn't
1765                         * part of a cycle. We do need to append its depth
1766                         * to our count.
1767                         */
1768                        total_depth += cur->depth;
1769                        break;
1770                }
1771
1772                /*
1773                 * We break cycles before looping, so an ACTIVE state (or any
1774                 * other cruft which made its way into the state variable)
1775                 * is a bug.
1776                 */
1777                if (cur->dfs_state != DFS_NONE)
1778                        BUG("confusing delta dfs state in first pass: %d",
1779                            cur->dfs_state);
1780
1781                /*
1782                 * Now we know this is the first time we've seen the object. If
1783                 * it's not a delta, we're done traversing, but we'll mark it
1784                 * done to save time on future traversals.
1785                 */
1786                if (!DELTA(cur)) {
1787                        cur->dfs_state = DFS_DONE;
1788                        break;
1789                }
1790
1791                /*
1792                 * Mark ourselves as active and see if the next step causes
1793                 * us to cycle to another active object. It's important to do
1794                 * this _before_ we loop, because it impacts where we make the
1795                 * cut, and thus how our total_depth counter works.
1796                 * E.g., We may see a partial loop like:
1797                 *
1798                 *   A -> B -> C -> D -> B
1799                 *
1800                 * Cutting B->C breaks the cycle. But now the depth of A is
1801                 * only 1, and our total_depth counter is at 3. The size of the
1802                 * error is always one less than the size of the cycle we
1803                 * broke. Commits C and D were "lost" from A's chain.
1804                 *
1805                 * If we instead cut D->B, then the depth of A is correct at 3.
1806                 * We keep all commits in the chain that we examined.
1807                 */
1808                cur->dfs_state = DFS_ACTIVE;
1809                if (DELTA(cur)->dfs_state == DFS_ACTIVE) {
1810                        drop_reused_delta(cur);
1811                        cur->dfs_state = DFS_DONE;
1812                        break;
1813                }
1814        }
1815
1816        /*
1817         * And now that we've gone all the way to the bottom of the chain, we
1818         * need to clear the active flags and set the depth fields as
1819         * appropriate. Unlike the loop above, which can quit when it drops a
1820         * delta, we need to keep going to look for more depth cuts. So we need
1821         * an extra "next" pointer to keep going after we reset cur->delta.
1822         */
1823        for (cur = entry; cur; cur = next) {
1824                next = DELTA(cur);
1825
1826                /*
1827                 * We should have a chain of zero or more ACTIVE states down to
1828                 * a final DONE. We can quit after the DONE, because either it
1829                 * has no bases, or we've already handled them in a previous
1830                 * call.
1831                 */
1832                if (cur->dfs_state == DFS_DONE)
1833                        break;
1834                else if (cur->dfs_state != DFS_ACTIVE)
1835                        BUG("confusing delta dfs state in second pass: %d",
1836                            cur->dfs_state);
1837
1838                /*
1839                 * If the total_depth is more than depth, then we need to snip
1840                 * the chain into two or more smaller chains that don't exceed
1841                 * the maximum depth. Most of the resulting chains will contain
1842                 * (depth + 1) entries (i.e., depth deltas plus one base), and
1843                 * the last chain (i.e., the one containing entry) will contain
1844                 * whatever entries are left over, namely
1845                 * (total_depth % (depth + 1)) of them.
1846                 *
1847                 * Since we are iterating towards decreasing depth, we need to
1848                 * decrement total_depth as we go, and we need to write to the
1849                 * entry what its final depth will be after all of the
1850                 * snipping. Since we're snipping into chains of length (depth
1851                 * + 1) entries, the final depth of an entry will be its
1852                 * original depth modulo (depth + 1). Any time we encounter an
1853                 * entry whose final depth is supposed to be zero, we snip it
1854                 * from its delta base, thereby making it so.
1855                 */
1856                cur->depth = (total_depth--) % (depth + 1);
1857                if (!cur->depth)
1858                        drop_reused_delta(cur);
1859
1860                cur->dfs_state = DFS_DONE;
1861        }
1862}
1863
1864static void get_object_details(void)
1865{
1866        uint32_t i;
1867        struct object_entry **sorted_by_offset;
1868
1869        if (progress)
1870                progress_state = start_progress(_("Counting objects"),
1871                                                to_pack.nr_objects);
1872
1873        sorted_by_offset = xcalloc(to_pack.nr_objects, sizeof(struct object_entry *));
1874        for (i = 0; i < to_pack.nr_objects; i++)
1875                sorted_by_offset[i] = to_pack.objects + i;
1876        QSORT(sorted_by_offset, to_pack.nr_objects, pack_offset_sort);
1877
1878        for (i = 0; i < to_pack.nr_objects; i++) {
1879                struct object_entry *entry = sorted_by_offset[i];
1880                check_object(entry);
1881                if (entry->type_valid &&
1882                    oe_size_greater_than(&to_pack, entry, big_file_threshold))
1883                        entry->no_try_delta = 1;
1884                display_progress(progress_state, i + 1);
1885        }
1886        stop_progress(&progress_state);
1887
1888        /*
1889         * This must happen in a second pass, since we rely on the delta
1890         * information for the whole list being completed.
1891         */
1892        for (i = 0; i < to_pack.nr_objects; i++)
1893                break_delta_chains(&to_pack.objects[i]);
1894
1895        free(sorted_by_offset);
1896}
1897
1898/*
1899 * We search for deltas in a list sorted by type, by filename hash, and then
1900 * by size, so that we see progressively smaller and smaller files.
1901 * That's because we prefer deltas to be from the bigger file
1902 * to the smaller -- deletes are potentially cheaper, but perhaps
1903 * more importantly, the bigger file is likely the more recent
1904 * one.  The deepest deltas are therefore the oldest objects which are
1905 * less susceptible to be accessed often.
1906 */
1907static int type_size_sort(const void *_a, const void *_b)
1908{
1909        const struct object_entry *a = *(struct object_entry **)_a;
1910        const struct object_entry *b = *(struct object_entry **)_b;
1911        const enum object_type a_type = oe_type(a);
1912        const enum object_type b_type = oe_type(b);
1913        const unsigned long a_size = SIZE(a);
1914        const unsigned long b_size = SIZE(b);
1915
1916        if (a_type > b_type)
1917                return -1;
1918        if (a_type < b_type)
1919                return 1;
1920        if (a->hash > b->hash)
1921                return -1;
1922        if (a->hash < b->hash)
1923                return 1;
1924        if (a->preferred_base > b->preferred_base)
1925                return -1;
1926        if (a->preferred_base < b->preferred_base)
1927                return 1;
1928        if (use_delta_islands) {
1929                const int island_cmp = island_delta_cmp(&a->idx.oid, &b->idx.oid);
1930                if (island_cmp)
1931                        return island_cmp;
1932        }
1933        if (a_size > b_size)
1934                return -1;
1935        if (a_size < b_size)
1936                return 1;
1937        return a < b ? -1 : (a > b);  /* newest first */
1938}
1939
1940struct unpacked {
1941        struct object_entry *entry;
1942        void *data;
1943        struct delta_index *index;
1944        unsigned depth;
1945};
1946
1947static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1948                           unsigned long delta_size)
1949{
1950        if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1951                return 0;
1952
1953        if (delta_size < cache_max_small_delta_size)
1954                return 1;
1955
1956        /* cache delta, if objects are large enough compared to delta size */
1957        if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1958                return 1;
1959
1960        return 0;
1961}
1962
1963/* Protect delta_cache_size */
1964static pthread_mutex_t cache_mutex;
1965#define cache_lock()            pthread_mutex_lock(&cache_mutex)
1966#define cache_unlock()          pthread_mutex_unlock(&cache_mutex)
1967
1968/*
1969 * Protect object list partitioning (e.g. struct thread_param) and
1970 * progress_state
1971 */
1972static pthread_mutex_t progress_mutex;
1973#define progress_lock()         pthread_mutex_lock(&progress_mutex)
1974#define progress_unlock()       pthread_mutex_unlock(&progress_mutex)
1975
1976/*
1977 * Access to struct object_entry is unprotected since each thread owns
1978 * a portion of the main object list. Just don't access object entries
1979 * ahead in the list because they can be stolen and would need
1980 * progress_mutex for protection.
1981 */
1982
1983/*
1984 * Return the size of the object without doing any delta
1985 * reconstruction (so non-deltas are true object sizes, but deltas
1986 * return the size of the delta data).
1987 */
1988unsigned long oe_get_size_slow(struct packing_data *pack,
1989                               const struct object_entry *e)
1990{
1991        struct packed_git *p;
1992        struct pack_window *w_curs;
1993        unsigned char *buf;
1994        enum object_type type;
1995        unsigned long used, avail, size;
1996
1997        if (e->type_ != OBJ_OFS_DELTA && e->type_ != OBJ_REF_DELTA) {
1998                packing_data_lock(&to_pack);
1999                if (oid_object_info(the_repository, &e->idx.oid, &size) < 0)
2000                        die(_("unable to get size of %s"),
2001                            oid_to_hex(&e->idx.oid));
2002                packing_data_unlock(&to_pack);
2003                return size;
2004        }
2005
2006        p = oe_in_pack(pack, e);
2007        if (!p)
2008                BUG("when e->type is a delta, it must belong to a pack");
2009
2010        packing_data_lock(&to_pack);
2011        w_curs = NULL;
2012        buf = use_pack(p, &w_curs, e->in_pack_offset, &avail);
2013        used = unpack_object_header_buffer(buf, avail, &type, &size);
2014        if (used == 0)
2015                die(_("unable to parse object header of %s"),
2016                    oid_to_hex(&e->idx.oid));
2017
2018        unuse_pack(&w_curs);
2019        packing_data_unlock(&to_pack);
2020        return size;
2021}
2022
2023static int try_delta(struct unpacked *trg, struct unpacked *src,
2024                     unsigned max_depth, unsigned long *mem_usage)
2025{
2026        struct object_entry *trg_entry = trg->entry;
2027        struct object_entry *src_entry = src->entry;
2028        unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
2029        unsigned ref_depth;
2030        enum object_type type;
2031        void *delta_buf;
2032
2033        /* Don't bother doing diffs between different types */
2034        if (oe_type(trg_entry) != oe_type(src_entry))
2035                return -1;
2036
2037        /*
2038         * We do not bother to try a delta that we discarded on an
2039         * earlier try, but only when reusing delta data.  Note that
2040         * src_entry that is marked as the preferred_base should always
2041         * be considered, as even if we produce a suboptimal delta against
2042         * it, we will still save the transfer cost, as we already know
2043         * the other side has it and we won't send src_entry at all.
2044         */
2045        if (reuse_delta && IN_PACK(trg_entry) &&
2046            IN_PACK(trg_entry) == IN_PACK(src_entry) &&
2047            !src_entry->preferred_base &&
2048            trg_entry->in_pack_type != OBJ_REF_DELTA &&
2049            trg_entry->in_pack_type != OBJ_OFS_DELTA)
2050                return 0;
2051
2052        /* Let's not bust the allowed depth. */
2053        if (src->depth >= max_depth)
2054                return 0;
2055
2056        /* Now some size filtering heuristics. */
2057        trg_size = SIZE(trg_entry);
2058        if (!DELTA(trg_entry)) {
2059                max_size = trg_size/2 - the_hash_algo->rawsz;
2060                ref_depth = 1;
2061        } else {
2062                max_size = DELTA_SIZE(trg_entry);
2063                ref_depth = trg->depth;
2064        }
2065        max_size = (uint64_t)max_size * (max_depth - src->depth) /
2066                                                (max_depth - ref_depth + 1);
2067        if (max_size == 0)
2068                return 0;
2069        src_size = SIZE(src_entry);
2070        sizediff = src_size < trg_size ? trg_size - src_size : 0;
2071        if (sizediff >= max_size)
2072                return 0;
2073        if (trg_size < src_size / 32)
2074                return 0;
2075
2076        if (!in_same_island(&trg->entry->idx.oid, &src->entry->idx.oid))
2077                return 0;
2078
2079        /* Load data if not already done */
2080        if (!trg->data) {
2081                packing_data_lock(&to_pack);
2082                trg->data = read_object_file(&trg_entry->idx.oid, &type, &sz);
2083                packing_data_unlock(&to_pack);
2084                if (!trg->data)
2085                        die(_("object %s cannot be read"),
2086                            oid_to_hex(&trg_entry->idx.oid));
2087                if (sz != trg_size)
2088                        die(_("object %s inconsistent object length (%"PRIuMAX" vs %"PRIuMAX")"),
2089                            oid_to_hex(&trg_entry->idx.oid), (uintmax_t)sz,
2090                            (uintmax_t)trg_size);
2091                *mem_usage += sz;
2092        }
2093        if (!src->data) {
2094                packing_data_lock(&to_pack);
2095                src->data = read_object_file(&src_entry->idx.oid, &type, &sz);
2096                packing_data_unlock(&to_pack);
2097                if (!src->data) {
2098                        if (src_entry->preferred_base) {
2099                                static int warned = 0;
2100                                if (!warned++)
2101                                        warning(_("object %s cannot be read"),
2102                                                oid_to_hex(&src_entry->idx.oid));
2103                                /*
2104                                 * Those objects are not included in the
2105                                 * resulting pack.  Be resilient and ignore
2106                                 * them if they can't be read, in case the
2107                                 * pack could be created nevertheless.
2108                                 */
2109                                return 0;
2110                        }
2111                        die(_("object %s cannot be read"),
2112                            oid_to_hex(&src_entry->idx.oid));
2113                }
2114                if (sz != src_size)
2115                        die(_("object %s inconsistent object length (%"PRIuMAX" vs %"PRIuMAX")"),
2116                            oid_to_hex(&src_entry->idx.oid), (uintmax_t)sz,
2117                            (uintmax_t)src_size);
2118                *mem_usage += sz;
2119        }
2120        if (!src->index) {
2121                src->index = create_delta_index(src->data, src_size);
2122                if (!src->index) {
2123                        static int warned = 0;
2124                        if (!warned++)
2125                                warning(_("suboptimal pack - out of memory"));
2126                        return 0;
2127                }
2128                *mem_usage += sizeof_delta_index(src->index);
2129        }
2130
2131        delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
2132        if (!delta_buf)
2133                return 0;
2134
2135        if (DELTA(trg_entry)) {
2136                /* Prefer only shallower same-sized deltas. */
2137                if (delta_size == DELTA_SIZE(trg_entry) &&
2138                    src->depth + 1 >= trg->depth) {
2139                        free(delta_buf);
2140                        return 0;
2141                }
2142        }
2143
2144        /*
2145         * Handle memory allocation outside of the cache
2146         * accounting lock.  Compiler will optimize the strangeness
2147         * away when NO_PTHREADS is defined.
2148         */
2149        free(trg_entry->delta_data);
2150        cache_lock();
2151        if (trg_entry->delta_data) {
2152                delta_cache_size -= DELTA_SIZE(trg_entry);
2153                trg_entry->delta_data = NULL;
2154        }
2155        if (delta_cacheable(src_size, trg_size, delta_size)) {
2156                delta_cache_size += delta_size;
2157                cache_unlock();
2158                trg_entry->delta_data = xrealloc(delta_buf, delta_size);
2159        } else {
2160                cache_unlock();
2161                free(delta_buf);
2162        }
2163
2164        SET_DELTA(trg_entry, src_entry);
2165        SET_DELTA_SIZE(trg_entry, delta_size);
2166        trg->depth = src->depth + 1;
2167
2168        return 1;
2169}
2170
2171static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
2172{
2173        struct object_entry *child = DELTA_CHILD(me);
2174        unsigned int m = n;
2175        while (child) {
2176                const unsigned int c = check_delta_limit(child, n + 1);
2177                if (m < c)
2178                        m = c;
2179                child = DELTA_SIBLING(child);
2180        }
2181        return m;
2182}
2183
2184static unsigned long free_unpacked(struct unpacked *n)
2185{
2186        unsigned long freed_mem = sizeof_delta_index(n->index);
2187        free_delta_index(n->index);
2188        n->index = NULL;
2189        if (n->data) {
2190                freed_mem += SIZE(n->entry);
2191                FREE_AND_NULL(n->data);
2192        }
2193        n->entry = NULL;
2194        n->depth = 0;
2195        return freed_mem;
2196}
2197
2198static void find_deltas(struct object_entry **list, unsigned *list_size,
2199                        int window, int depth, unsigned *processed)
2200{
2201        uint32_t i, idx = 0, count = 0;
2202        struct unpacked *array;
2203        unsigned long mem_usage = 0;
2204
2205        array = xcalloc(window, sizeof(struct unpacked));
2206
2207        for (;;) {
2208                struct object_entry *entry;
2209                struct unpacked *n = array + idx;
2210                int j, max_depth, best_base = -1;
2211
2212                progress_lock();
2213                if (!*list_size) {
2214                        progress_unlock();
2215                        break;
2216                }
2217                entry = *list++;
2218                (*list_size)--;
2219                if (!entry->preferred_base) {
2220                        (*processed)++;
2221                        display_progress(progress_state, *processed);
2222                }
2223                progress_unlock();
2224
2225                mem_usage -= free_unpacked(n);
2226                n->entry = entry;
2227
2228                while (window_memory_limit &&
2229                       mem_usage > window_memory_limit &&
2230                       count > 1) {
2231                        const uint32_t tail = (idx + window - count) % window;
2232                        mem_usage -= free_unpacked(array + tail);
2233                        count--;
2234                }
2235
2236                /* We do not compute delta to *create* objects we are not
2237                 * going to pack.
2238                 */
2239                if (entry->preferred_base)
2240                        goto next;
2241
2242                /*
2243                 * If the current object is at pack edge, take the depth the
2244                 * objects that depend on the current object into account
2245                 * otherwise they would become too deep.
2246                 */
2247                max_depth = depth;
2248                if (DELTA_CHILD(entry)) {
2249                        max_depth -= check_delta_limit(entry, 0);
2250                        if (max_depth <= 0)
2251                                goto next;
2252                }
2253
2254                j = window;
2255                while (--j > 0) {
2256                        int ret;
2257                        uint32_t other_idx = idx + j;
2258                        struct unpacked *m;
2259                        if (other_idx >= window)
2260                                other_idx -= window;
2261                        m = array + other_idx;
2262                        if (!m->entry)
2263                                break;
2264                        ret = try_delta(n, m, max_depth, &mem_usage);
2265                        if (ret < 0)
2266                                break;
2267                        else if (ret > 0)
2268                                best_base = other_idx;
2269                }
2270
2271                /*
2272                 * If we decided to cache the delta data, then it is best
2273                 * to compress it right away.  First because we have to do
2274                 * it anyway, and doing it here while we're threaded will
2275                 * save a lot of time in the non threaded write phase,
2276                 * as well as allow for caching more deltas within
2277                 * the same cache size limit.
2278                 * ...
2279                 * But only if not writing to stdout, since in that case
2280                 * the network is most likely throttling writes anyway,
2281                 * and therefore it is best to go to the write phase ASAP
2282                 * instead, as we can afford spending more time compressing
2283                 * between writes at that moment.
2284                 */
2285                if (entry->delta_data && !pack_to_stdout) {
2286                        unsigned long size;
2287
2288                        size = do_compress(&entry->delta_data, DELTA_SIZE(entry));
2289                        if (size < (1U << OE_Z_DELTA_BITS)) {
2290                                entry->z_delta_size = size;
2291                                cache_lock();
2292                                delta_cache_size -= DELTA_SIZE(entry);
2293                                delta_cache_size += entry->z_delta_size;
2294                                cache_unlock();
2295                        } else {
2296                                FREE_AND_NULL(entry->delta_data);
2297                                entry->z_delta_size = 0;
2298                        }
2299                }
2300
2301                /* if we made n a delta, and if n is already at max
2302                 * depth, leaving it in the window is pointless.  we
2303                 * should evict it first.
2304                 */
2305                if (DELTA(entry) && max_depth <= n->depth)
2306                        continue;
2307
2308                /*
2309                 * Move the best delta base up in the window, after the
2310                 * currently deltified object, to keep it longer.  It will
2311                 * be the first base object to be attempted next.
2312                 */
2313                if (DELTA(entry)) {
2314                        struct unpacked swap = array[best_base];
2315                        int dist = (window + idx - best_base) % window;
2316                        int dst = best_base;
2317                        while (dist--) {
2318                                int src = (dst + 1) % window;
2319                                array[dst] = array[src];
2320                                dst = src;
2321                        }
2322                        array[dst] = swap;
2323                }
2324
2325                next:
2326                idx++;
2327                if (count + 1 < window)
2328                        count++;
2329                if (idx >= window)
2330                        idx = 0;
2331        }
2332
2333        for (i = 0; i < window; ++i) {
2334                free_delta_index(array[i].index);
2335                free(array[i].data);
2336        }
2337        free(array);
2338}
2339
2340static void try_to_free_from_threads(size_t size)
2341{
2342        packing_data_lock(&to_pack);
2343        release_pack_memory(size);
2344        packing_data_unlock(&to_pack);
2345}
2346
2347static try_to_free_t old_try_to_free_routine;
2348
2349/*
2350 * The main object list is split into smaller lists, each is handed to
2351 * one worker.
2352 *
2353 * The main thread waits on the condition that (at least) one of the workers
2354 * has stopped working (which is indicated in the .working member of
2355 * struct thread_params).
2356 *
2357 * When a work thread has completed its work, it sets .working to 0 and
2358 * signals the main thread and waits on the condition that .data_ready
2359 * becomes 1.
2360 *
2361 * The main thread steals half of the work from the worker that has
2362 * most work left to hand it to the idle worker.
2363 */
2364
2365struct thread_params {
2366        pthread_t thread;
2367        struct object_entry **list;
2368        unsigned list_size;
2369        unsigned remaining;
2370        int window;
2371        int depth;
2372        int working;
2373        int data_ready;
2374        pthread_mutex_t mutex;
2375        pthread_cond_t cond;
2376        unsigned *processed;
2377};
2378
2379static pthread_cond_t progress_cond;
2380
2381/*
2382 * Mutex and conditional variable can't be statically-initialized on Windows.
2383 */
2384static void init_threaded_search(void)
2385{
2386        pthread_mutex_init(&cache_mutex, NULL);
2387        pthread_mutex_init(&progress_mutex, NULL);
2388        pthread_cond_init(&progress_cond, NULL);
2389        old_try_to_free_routine = set_try_to_free_routine(try_to_free_from_threads);
2390}
2391
2392static void cleanup_threaded_search(void)
2393{
2394        set_try_to_free_routine(old_try_to_free_routine);
2395        pthread_cond_destroy(&progress_cond);
2396        pthread_mutex_destroy(&cache_mutex);
2397        pthread_mutex_destroy(&progress_mutex);
2398}
2399
2400static void *threaded_find_deltas(void *arg)
2401{
2402        struct thread_params *me = arg;
2403
2404        progress_lock();
2405        while (me->remaining) {
2406                progress_unlock();
2407
2408                find_deltas(me->list, &me->remaining,
2409                            me->window, me->depth, me->processed);
2410
2411                progress_lock();
2412                me->working = 0;
2413                pthread_cond_signal(&progress_cond);
2414                progress_unlock();
2415
2416                /*
2417                 * We must not set ->data_ready before we wait on the
2418                 * condition because the main thread may have set it to 1
2419                 * before we get here. In order to be sure that new
2420                 * work is available if we see 1 in ->data_ready, it
2421                 * was initialized to 0 before this thread was spawned
2422                 * and we reset it to 0 right away.
2423                 */
2424                pthread_mutex_lock(&me->mutex);
2425                while (!me->data_ready)
2426                        pthread_cond_wait(&me->cond, &me->mutex);
2427                me->data_ready = 0;
2428                pthread_mutex_unlock(&me->mutex);
2429
2430                progress_lock();
2431        }
2432        progress_unlock();
2433        /* leave ->working 1 so that this doesn't get more work assigned */
2434        return NULL;
2435}
2436
2437static void ll_find_deltas(struct object_entry **list, unsigned list_size,
2438                           int window, int depth, unsigned *processed)
2439{
2440        struct thread_params *p;
2441        int i, ret, active_threads = 0;
2442
2443        init_threaded_search();
2444
2445        if (delta_search_threads <= 1) {
2446                find_deltas(list, &list_size, window, depth, processed);
2447                cleanup_threaded_search();
2448                return;
2449        }
2450        if (progress > pack_to_stdout)
2451                fprintf_ln(stderr, _("Delta compression using up to %d threads"),
2452                           delta_search_threads);
2453        p = xcalloc(delta_search_threads, sizeof(*p));
2454
2455        /* Partition the work amongst work threads. */
2456        for (i = 0; i < delta_search_threads; i++) {
2457                unsigned sub_size = list_size / (delta_search_threads - i);
2458
2459                /* don't use too small segments or no deltas will be found */
2460                if (sub_size < 2*window && i+1 < delta_search_threads)
2461                        sub_size = 0;
2462
2463                p[i].window = window;
2464                p[i].depth = depth;
2465                p[i].processed = processed;
2466                p[i].working = 1;
2467                p[i].data_ready = 0;
2468
2469                /* try to split chunks on "path" boundaries */
2470                while (sub_size && sub_size < list_size &&
2471                       list[sub_size]->hash &&
2472                       list[sub_size]->hash == list[sub_size-1]->hash)
2473                        sub_size++;
2474
2475                p[i].list = list;
2476                p[i].list_size = sub_size;
2477                p[i].remaining = sub_size;
2478
2479                list += sub_size;
2480                list_size -= sub_size;
2481        }
2482
2483        /* Start work threads. */
2484        for (i = 0; i < delta_search_threads; i++) {
2485                if (!p[i].list_size)
2486                        continue;
2487                pthread_mutex_init(&p[i].mutex, NULL);
2488                pthread_cond_init(&p[i].cond, NULL);
2489                ret = pthread_create(&p[i].thread, NULL,
2490                                     threaded_find_deltas, &p[i]);
2491                if (ret)
2492                        die(_("unable to create thread: %s"), strerror(ret));
2493                active_threads++;
2494        }
2495
2496        /*
2497         * Now let's wait for work completion.  Each time a thread is done
2498         * with its work, we steal half of the remaining work from the
2499         * thread with the largest number of unprocessed objects and give
2500         * it to that newly idle thread.  This ensure good load balancing
2501         * until the remaining object list segments are simply too short
2502         * to be worth splitting anymore.
2503         */
2504        while (active_threads) {
2505                struct thread_params *target = NULL;
2506                struct thread_params *victim = NULL;
2507                unsigned sub_size = 0;
2508
2509                progress_lock();
2510                for (;;) {
2511                        for (i = 0; !target && i < delta_search_threads; i++)
2512                                if (!p[i].working)
2513                                        target = &p[i];
2514                        if (target)
2515                                break;
2516                        pthread_cond_wait(&progress_cond, &progress_mutex);
2517                }
2518
2519                for (i = 0; i < delta_search_threads; i++)
2520                        if (p[i].remaining > 2*window &&
2521                            (!victim || victim->remaining < p[i].remaining))
2522                                victim = &p[i];
2523                if (victim) {
2524                        sub_size = victim->remaining / 2;
2525                        list = victim->list + victim->list_size - sub_size;
2526                        while (sub_size && list[0]->hash &&
2527                               list[0]->hash == list[-1]->hash) {
2528                                list++;
2529                                sub_size--;
2530                        }
2531                        if (!sub_size) {
2532                                /*
2533                                 * It is possible for some "paths" to have
2534                                 * so many objects that no hash boundary
2535                                 * might be found.  Let's just steal the
2536                                 * exact half in that case.
2537                                 */
2538                                sub_size = victim->remaining / 2;
2539                                list -= sub_size;
2540                        }
2541                        target->list = list;
2542                        victim->list_size -= sub_size;
2543                        victim->remaining -= sub_size;
2544                }
2545                target->list_size = sub_size;
2546                target->remaining = sub_size;
2547                target->working = 1;
2548                progress_unlock();
2549
2550                pthread_mutex_lock(&target->mutex);
2551                target->data_ready = 1;
2552                pthread_cond_signal(&target->cond);
2553                pthread_mutex_unlock(&target->mutex);
2554
2555                if (!sub_size) {
2556                        pthread_join(target->thread, NULL);
2557                        pthread_cond_destroy(&target->cond);
2558                        pthread_mutex_destroy(&target->mutex);
2559                        active_threads--;
2560                }
2561        }
2562        cleanup_threaded_search();
2563        free(p);
2564}
2565
2566static void add_tag_chain(const struct object_id *oid)
2567{
2568        struct tag *tag;
2569
2570        /*
2571         * We catch duplicates already in add_object_entry(), but we'd
2572         * prefer to do this extra check to avoid having to parse the
2573         * tag at all if we already know that it's being packed (e.g., if
2574         * it was included via bitmaps, we would not have parsed it
2575         * previously).
2576         */
2577        if (packlist_find(&to_pack, oid))
2578                return;
2579
2580        tag = lookup_tag(the_repository, oid);
2581        while (1) {
2582                if (!tag || parse_tag(tag) || !tag->tagged)
2583                        die(_("unable to pack objects reachable from tag %s"),
2584                            oid_to_hex(oid));
2585
2586                add_object_entry(&tag->object.oid, OBJ_TAG, NULL, 0);
2587
2588                if (tag->tagged->type != OBJ_TAG)
2589                        return;
2590
2591                tag = (struct tag *)tag->tagged;
2592        }
2593}
2594
2595static int add_ref_tag(const char *path, const struct object_id *oid, int flag, void *cb_data)
2596{
2597        struct object_id peeled;
2598
2599        if (starts_with(path, "refs/tags/") && /* is a tag? */
2600            !peel_ref(path, &peeled)    && /* peelable? */
2601            packlist_find(&to_pack, &peeled))      /* object packed? */
2602                add_tag_chain(oid);
2603        return 0;
2604}
2605
2606static void prepare_pack(int window, int depth)
2607{
2608        struct object_entry **delta_list;
2609        uint32_t i, nr_deltas;
2610        unsigned n;
2611
2612        if (use_delta_islands)
2613                resolve_tree_islands(the_repository, progress, &to_pack);
2614
2615        get_object_details();
2616
2617        /*
2618         * If we're locally repacking then we need to be doubly careful
2619         * from now on in order to make sure no stealth corruption gets
2620         * propagated to the new pack.  Clients receiving streamed packs
2621         * should validate everything they get anyway so no need to incur
2622         * the additional cost here in that case.
2623         */
2624        if (!pack_to_stdout)
2625                do_check_packed_object_crc = 1;
2626
2627        if (!to_pack.nr_objects || !window || !depth)
2628                return;
2629
2630        ALLOC_ARRAY(delta_list, to_pack.nr_objects);
2631        nr_deltas = n = 0;
2632
2633        for (i = 0; i < to_pack.nr_objects; i++) {
2634                struct object_entry *entry = to_pack.objects + i;
2635
2636                if (DELTA(entry))
2637                        /* This happens if we decided to reuse existing
2638                         * delta from a pack.  "reuse_delta &&" is implied.
2639                         */
2640                        continue;
2641
2642                if (!entry->type_valid ||
2643                    oe_size_less_than(&to_pack, entry, 50))
2644                        continue;
2645
2646                if (entry->no_try_delta)
2647                        continue;
2648
2649                if (!entry->preferred_base) {
2650                        nr_deltas++;
2651                        if (oe_type(entry) < 0)
2652                                die(_("unable to get type of object %s"),
2653                                    oid_to_hex(&entry->idx.oid));
2654                } else {
2655                        if (oe_type(entry) < 0) {
2656                                /*
2657                                 * This object is not found, but we
2658                                 * don't have to include it anyway.
2659                                 */
2660                                continue;
2661                        }
2662                }
2663
2664                delta_list[n++] = entry;
2665        }
2666
2667        if (nr_deltas && n > 1) {
2668                unsigned nr_done = 0;
2669                if (progress)
2670                        progress_state = start_progress(_("Compressing objects"),
2671                                                        nr_deltas);
2672                QSORT(delta_list, n, type_size_sort);
2673                ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
2674                stop_progress(&progress_state);
2675                if (nr_done != nr_deltas)
2676                        die(_("inconsistency with delta count"));
2677        }
2678        free(delta_list);
2679}
2680
2681static int git_pack_config(const char *k, const char *v, void *cb)
2682{
2683        if (!strcmp(k, "pack.window")) {
2684                window = git_config_int(k, v);
2685                return 0;
2686        }
2687        if (!strcmp(k, "pack.windowmemory")) {
2688                window_memory_limit = git_config_ulong(k, v);
2689                return 0;
2690        }
2691        if (!strcmp(k, "pack.depth")) {
2692                depth = git_config_int(k, v);
2693                return 0;
2694        }
2695        if (!strcmp(k, "pack.deltacachesize")) {
2696                max_delta_cache_size = git_config_int(k, v);
2697                return 0;
2698        }
2699        if (!strcmp(k, "pack.deltacachelimit")) {
2700                cache_max_small_delta_size = git_config_int(k, v);
2701                return 0;
2702        }
2703        if (!strcmp(k, "pack.writebitmaphashcache")) {
2704                if (git_config_bool(k, v))
2705                        write_bitmap_options |= BITMAP_OPT_HASH_CACHE;
2706                else
2707                        write_bitmap_options &= ~BITMAP_OPT_HASH_CACHE;
2708        }
2709        if (!strcmp(k, "pack.usebitmaps")) {
2710                use_bitmap_index_default = git_config_bool(k, v);
2711                return 0;
2712        }
2713        if (!strcmp(k, "pack.usesparse")) {
2714                sparse = git_config_bool(k, v);
2715                return 0;
2716        }
2717        if (!strcmp(k, "pack.threads")) {
2718                delta_search_threads = git_config_int(k, v);
2719                if (delta_search_threads < 0)
2720                        die(_("invalid number of threads specified (%d)"),
2721                            delta_search_threads);
2722                if (!HAVE_THREADS && delta_search_threads != 1) {
2723                        warning(_("no threads support, ignoring %s"), k);
2724                        delta_search_threads = 0;
2725                }
2726                return 0;
2727        }
2728        if (!strcmp(k, "pack.indexversion")) {
2729                pack_idx_opts.version = git_config_int(k, v);
2730                if (pack_idx_opts.version > 2)
2731                        die(_("bad pack.indexversion=%"PRIu32),
2732                            pack_idx_opts.version);
2733                return 0;
2734        }
2735        return git_default_config(k, v, cb);
2736}
2737
2738static void read_object_list_from_stdin(void)
2739{
2740        char line[GIT_MAX_HEXSZ + 1 + PATH_MAX + 2];
2741        struct object_id oid;
2742        const char *p;
2743
2744        for (;;) {
2745                if (!fgets(line, sizeof(line), stdin)) {
2746                        if (feof(stdin))
2747                                break;
2748                        if (!ferror(stdin))
2749                                die("BUG: fgets returned NULL, not EOF, not error!");
2750                        if (errno != EINTR)
2751                                die_errno("fgets");
2752                        clearerr(stdin);
2753                        continue;
2754                }
2755                if (line[0] == '-') {
2756                        if (get_oid_hex(line+1, &oid))
2757                                die(_("expected edge object ID, got garbage:\n %s"),
2758                                    line);
2759                        add_preferred_base(&oid);
2760                        continue;
2761                }
2762                if (parse_oid_hex(line, &oid, &p))
2763                        die(_("expected object ID, got garbage:\n %s"), line);
2764
2765                add_preferred_base_object(p + 1);
2766                add_object_entry(&oid, OBJ_NONE, p + 1, 0);
2767        }
2768}
2769
2770/* Remember to update object flag allocation in object.h */
2771#define OBJECT_ADDED (1u<<20)
2772
2773static void show_commit(struct commit *commit, void *data)
2774{
2775        add_object_entry(&commit->object.oid, OBJ_COMMIT, NULL, 0);
2776        commit->object.flags |= OBJECT_ADDED;
2777
2778        if (write_bitmap_index)
2779                index_commit_for_bitmap(commit);
2780
2781        if (use_delta_islands)
2782                propagate_island_marks(commit);
2783}
2784
2785static void show_object(struct object *obj, const char *name, void *data)
2786{
2787        add_preferred_base_object(name);
2788        add_object_entry(&obj->oid, obj->type, name, 0);
2789        obj->flags |= OBJECT_ADDED;
2790
2791        if (use_delta_islands) {
2792                const char *p;
2793                unsigned depth;
2794                struct object_entry *ent;
2795
2796                /* the empty string is a root tree, which is depth 0 */
2797                depth = *name ? 1 : 0;
2798                for (p = strchr(name, '/'); p; p = strchr(p + 1, '/'))
2799                        depth++;
2800
2801                ent = packlist_find(&to_pack, &obj->oid);
2802                if (ent && depth > oe_tree_depth(&to_pack, ent))
2803                        oe_set_tree_depth(&to_pack, ent, depth);
2804        }
2805}
2806
2807static void show_object__ma_allow_any(struct object *obj, const char *name, void *data)
2808{
2809        assert(arg_missing_action == MA_ALLOW_ANY);
2810
2811        /*
2812         * Quietly ignore ALL missing objects.  This avoids problems with
2813         * staging them now and getting an odd error later.
2814         */
2815        if (!has_object_file(&obj->oid))
2816                return;
2817
2818        show_object(obj, name, data);
2819}
2820
2821static void show_object__ma_allow_promisor(struct object *obj, const char *name, void *data)
2822{
2823        assert(arg_missing_action == MA_ALLOW_PROMISOR);
2824
2825        /*
2826         * Quietly ignore EXPECTED missing objects.  This avoids problems with
2827         * staging them now and getting an odd error later.
2828         */
2829        if (!has_object_file(&obj->oid) && is_promisor_object(&obj->oid))
2830                return;
2831
2832        show_object(obj, name, data);
2833}
2834
2835static int option_parse_missing_action(const struct option *opt,
2836                                       const char *arg, int unset)
2837{
2838        assert(arg);
2839        assert(!unset);
2840
2841        if (!strcmp(arg, "error")) {
2842                arg_missing_action = MA_ERROR;
2843                fn_show_object = show_object;
2844                return 0;
2845        }
2846
2847        if (!strcmp(arg, "allow-any")) {
2848                arg_missing_action = MA_ALLOW_ANY;
2849                fetch_if_missing = 0;
2850                fn_show_object = show_object__ma_allow_any;
2851                return 0;
2852        }
2853
2854        if (!strcmp(arg, "allow-promisor")) {
2855                arg_missing_action = MA_ALLOW_PROMISOR;
2856                fetch_if_missing = 0;
2857                fn_show_object = show_object__ma_allow_promisor;
2858                return 0;
2859        }
2860
2861        die(_("invalid value for --missing"));
2862        return 0;
2863}
2864
2865static void show_edge(struct commit *commit)
2866{
2867        add_preferred_base(&commit->object.oid);
2868}
2869
2870struct in_pack_object {
2871        off_t offset;
2872        struct object *object;
2873};
2874
2875struct in_pack {
2876        unsigned int alloc;
2877        unsigned int nr;
2878        struct in_pack_object *array;
2879};
2880
2881static void mark_in_pack_object(struct object *object, struct packed_git *p, struct in_pack *in_pack)
2882{
2883        in_pack->array[in_pack->nr].offset = find_pack_entry_one(object->oid.hash, p);
2884        in_pack->array[in_pack->nr].object = object;
2885        in_pack->nr++;
2886}
2887
2888/*
2889 * Compare the objects in the offset order, in order to emulate the
2890 * "git rev-list --objects" output that produced the pack originally.
2891 */
2892static int ofscmp(const void *a_, const void *b_)
2893{
2894        struct in_pack_object *a = (struct in_pack_object *)a_;
2895        struct in_pack_object *b = (struct in_pack_object *)b_;
2896
2897        if (a->offset < b->offset)
2898                return -1;
2899        else if (a->offset > b->offset)
2900                return 1;
2901        else
2902                return oidcmp(&a->object->oid, &b->object->oid);
2903}
2904
2905static void add_objects_in_unpacked_packs(void)
2906{
2907        struct packed_git *p;
2908        struct in_pack in_pack;
2909        uint32_t i;
2910
2911        memset(&in_pack, 0, sizeof(in_pack));
2912
2913        for (p = get_all_packs(the_repository); p; p = p->next) {
2914                struct object_id oid;
2915                struct object *o;
2916
2917                if (!p->pack_local || p->pack_keep || p->pack_keep_in_core)
2918                        continue;
2919                if (open_pack_index(p))
2920                        die(_("cannot open pack index"));
2921
2922                ALLOC_GROW(in_pack.array,
2923                           in_pack.nr + p->num_objects,
2924                           in_pack.alloc);
2925
2926                for (i = 0; i < p->num_objects; i++) {
2927                        nth_packed_object_oid(&oid, p, i);
2928                        o = lookup_unknown_object(&oid);
2929                        if (!(o->flags & OBJECT_ADDED))
2930                                mark_in_pack_object(o, p, &in_pack);
2931                        o->flags |= OBJECT_ADDED;
2932                }
2933        }
2934
2935        if (in_pack.nr) {
2936                QSORT(in_pack.array, in_pack.nr, ofscmp);
2937                for (i = 0; i < in_pack.nr; i++) {
2938                        struct object *o = in_pack.array[i].object;
2939                        add_object_entry(&o->oid, o->type, "", 0);
2940                }
2941        }
2942        free(in_pack.array);
2943}
2944
2945static int add_loose_object(const struct object_id *oid, const char *path,
2946                            void *data)
2947{
2948        enum object_type type = oid_object_info(the_repository, oid, NULL);
2949
2950        if (type < 0) {
2951                warning(_("loose object at %s could not be examined"), path);
2952                return 0;
2953        }
2954
2955        add_object_entry(oid, type, "", 0);
2956        return 0;
2957}
2958
2959/*
2960 * We actually don't even have to worry about reachability here.
2961 * add_object_entry will weed out duplicates, so we just add every
2962 * loose object we find.
2963 */
2964static void add_unreachable_loose_objects(void)
2965{
2966        for_each_loose_file_in_objdir(get_object_directory(),
2967                                      add_loose_object,
2968                                      NULL, NULL, NULL);
2969}
2970
2971static int has_sha1_pack_kept_or_nonlocal(const struct object_id *oid)
2972{
2973        static struct packed_git *last_found = (void *)1;
2974        struct packed_git *p;
2975
2976        p = (last_found != (void *)1) ? last_found :
2977                                        get_all_packs(the_repository);
2978
2979        while (p) {
2980                if ((!p->pack_local || p->pack_keep ||
2981                                p->pack_keep_in_core) &&
2982                        find_pack_entry_one(oid->hash, p)) {
2983                        last_found = p;
2984                        return 1;
2985                }
2986                if (p == last_found)
2987                        p = get_all_packs(the_repository);
2988                else
2989                        p = p->next;
2990                if (p == last_found)
2991                        p = p->next;
2992        }
2993        return 0;
2994}
2995
2996/*
2997 * Store a list of sha1s that are should not be discarded
2998 * because they are either written too recently, or are
2999 * reachable from another object that was.
3000 *
3001 * This is filled by get_object_list.
3002 */
3003static struct oid_array recent_objects;
3004
3005static int loosened_object_can_be_discarded(const struct object_id *oid,
3006                                            timestamp_t mtime)
3007{
3008        if (!unpack_unreachable_expiration)
3009                return 0;
3010        if (mtime > unpack_unreachable_expiration)
3011                return 0;
3012        if (oid_array_lookup(&recent_objects, oid) >= 0)
3013                return 0;
3014        return 1;
3015}
3016
3017static void loosen_unused_packed_objects(void)
3018{
3019        struct packed_git *p;
3020        uint32_t i;
3021        struct object_id oid;
3022
3023        for (p = get_all_packs(the_repository); p; p = p->next) {
3024                if (!p->pack_local || p->pack_keep || p->pack_keep_in_core)
3025                        continue;
3026
3027                if (open_pack_index(p))
3028                        die(_("cannot open pack index"));
3029
3030                for (i = 0; i < p->num_objects; i++) {
3031                        nth_packed_object_oid(&oid, p, i);
3032                        if (!packlist_find(&to_pack, &oid) &&
3033                            !has_sha1_pack_kept_or_nonlocal(&oid) &&
3034                            !loosened_object_can_be_discarded(&oid, p->mtime))
3035                                if (force_object_loose(&oid, p->mtime))
3036                                        die(_("unable to force loose object"));
3037                }
3038        }
3039}
3040
3041/*
3042 * This tracks any options which pack-reuse code expects to be on, or which a
3043 * reader of the pack might not understand, and which would therefore prevent
3044 * blind reuse of what we have on disk.
3045 */
3046static int pack_options_allow_reuse(void)
3047{
3048        return pack_to_stdout &&
3049               allow_ofs_delta &&
3050               !ignore_packed_keep_on_disk &&
3051               !ignore_packed_keep_in_core &&
3052               (!local || !have_non_local_packs) &&
3053               !incremental;
3054}
3055
3056static int get_object_list_from_bitmap(struct rev_info *revs)
3057{
3058        if (!(bitmap_git = prepare_bitmap_walk(revs)))
3059                return -1;
3060
3061        if (pack_options_allow_reuse() &&
3062            !reuse_partial_packfile_from_bitmap(
3063                        bitmap_git,
3064                        &reuse_packfile,
3065                        &reuse_packfile_objects,
3066                        &reuse_packfile_offset)) {
3067                assert(reuse_packfile_objects);
3068                nr_result += reuse_packfile_objects;
3069                display_progress(progress_state, nr_result);
3070        }
3071
3072        traverse_bitmap_commit_list(bitmap_git, &add_object_entry_from_bitmap);
3073        return 0;
3074}
3075
3076static void record_recent_object(struct object *obj,
3077                                 const char *name,
3078                                 void *data)
3079{
3080        oid_array_append(&recent_objects, &obj->oid);
3081}
3082
3083static void record_recent_commit(struct commit *commit, void *data)
3084{
3085        oid_array_append(&recent_objects, &commit->object.oid);
3086}
3087
3088static void get_object_list(int ac, const char **av)
3089{
3090        struct rev_info revs;
3091        struct setup_revision_opt s_r_opt = {
3092                .allow_exclude_promisor_objects = 1,
3093        };
3094        char line[1000];
3095        int flags = 0;
3096        int save_warning;
3097
3098        repo_init_revisions(the_repository, &revs, NULL);
3099        save_commit_buffer = 0;
3100        setup_revisions(ac, av, &revs, &s_r_opt);
3101
3102        /* make sure shallows are read */
3103        is_repository_shallow(the_repository);
3104
3105        save_warning = warn_on_object_refname_ambiguity;
3106        warn_on_object_refname_ambiguity = 0;
3107
3108        while (fgets(line, sizeof(line), stdin) != NULL) {
3109                int len = strlen(line);
3110                if (len && line[len - 1] == '\n')
3111                        line[--len] = 0;
3112                if (!len)
3113                        break;
3114                if (*line == '-') {
3115                        if (!strcmp(line, "--not")) {
3116                                flags ^= UNINTERESTING;
3117                                write_bitmap_index = 0;
3118                                continue;
3119                        }
3120                        if (starts_with(line, "--shallow ")) {
3121                                struct object_id oid;
3122                                if (get_oid_hex(line + 10, &oid))
3123                                        die("not an SHA-1 '%s'", line + 10);
3124                                register_shallow(the_repository, &oid);
3125                                use_bitmap_index = 0;
3126                                continue;
3127                        }
3128                        die(_("not a rev '%s'"), line);
3129                }
3130                if (handle_revision_arg(line, &revs, flags, REVARG_CANNOT_BE_FILENAME))
3131                        die(_("bad revision '%s'"), line);
3132        }
3133
3134        warn_on_object_refname_ambiguity = save_warning;
3135
3136        if (use_bitmap_index && !get_object_list_from_bitmap(&revs))
3137                return;
3138
3139        if (use_delta_islands)
3140                load_delta_islands(the_repository, progress);
3141
3142        if (prepare_revision_walk(&revs))
3143                die(_("revision walk setup failed"));
3144        mark_edges_uninteresting(&revs, show_edge, sparse);
3145
3146        if (!fn_show_object)
3147                fn_show_object = show_object;
3148        traverse_commit_list_filtered(&filter_options, &revs,
3149                                      show_commit, fn_show_object, NULL,
3150                                      NULL);
3151
3152        if (unpack_unreachable_expiration) {
3153                revs.ignore_missing_links = 1;
3154                if (add_unseen_recent_objects_to_traversal(&revs,
3155                                unpack_unreachable_expiration))
3156                        die(_("unable to add recent objects"));
3157                if (prepare_revision_walk(&revs))
3158                        die(_("revision walk setup failed"));
3159                traverse_commit_list(&revs, record_recent_commit,
3160                                     record_recent_object, NULL);
3161        }
3162
3163        if (keep_unreachable)
3164                add_objects_in_unpacked_packs();
3165        if (pack_loose_unreachable)
3166                add_unreachable_loose_objects();
3167        if (unpack_unreachable)
3168                loosen_unused_packed_objects();
3169
3170        oid_array_clear(&recent_objects);
3171}
3172
3173static void add_extra_kept_packs(const struct string_list *names)
3174{
3175        struct packed_git *p;
3176
3177        if (!names->nr)
3178                return;
3179
3180        for (p = get_all_packs(the_repository); p; p = p->next) {
3181                const char *name = basename(p->pack_name);
3182                int i;
3183
3184                if (!p->pack_local)
3185                        continue;
3186
3187                for (i = 0; i < names->nr; i++)
3188                        if (!fspathcmp(name, names->items[i].string))
3189                                break;
3190
3191                if (i < names->nr) {
3192                        p->pack_keep_in_core = 1;
3193                        ignore_packed_keep_in_core = 1;
3194                        continue;
3195                }
3196        }
3197}
3198
3199static int option_parse_index_version(const struct option *opt,
3200                                      const char *arg, int unset)
3201{
3202        char *c;
3203        const char *val = arg;
3204
3205        BUG_ON_OPT_NEG(unset);
3206
3207        pack_idx_opts.version = strtoul(val, &c, 10);
3208        if (pack_idx_opts.version > 2)
3209                die(_("unsupported index version %s"), val);
3210        if (*c == ',' && c[1])
3211                pack_idx_opts.off32_limit = strtoul(c+1, &c, 0);
3212        if (*c || pack_idx_opts.off32_limit & 0x80000000)
3213                die(_("bad index version '%s'"), val);
3214        return 0;
3215}
3216
3217static int option_parse_unpack_unreachable(const struct option *opt,
3218                                           const char *arg, int unset)
3219{
3220        if (unset) {
3221                unpack_unreachable = 0;
3222                unpack_unreachable_expiration = 0;
3223        }
3224        else {
3225                unpack_unreachable = 1;
3226                if (arg)
3227                        unpack_unreachable_expiration = approxidate(arg);
3228        }
3229        return 0;
3230}
3231
3232int cmd_pack_objects(int argc, const char **argv, const char *prefix)
3233{
3234        int use_internal_rev_list = 0;
3235        int shallow = 0;
3236        int all_progress_implied = 0;
3237        struct argv_array rp = ARGV_ARRAY_INIT;
3238        int rev_list_unpacked = 0, rev_list_all = 0, rev_list_reflog = 0;
3239        int rev_list_index = 0;
3240        struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
3241        struct option pack_objects_options[] = {
3242                OPT_SET_INT('q', "quiet", &progress,
3243                            N_("do not show progress meter"), 0),
3244                OPT_SET_INT(0, "progress", &progress,
3245                            N_("show progress meter"), 1),
3246                OPT_SET_INT(0, "all-progress", &progress,
3247                            N_("show progress meter during object writing phase"), 2),
3248                OPT_BOOL(0, "all-progress-implied",
3249                         &all_progress_implied,
3250                         N_("similar to --all-progress when progress meter is shown")),
3251                { OPTION_CALLBACK, 0, "index-version", NULL, N_("<version>[,<offset>]"),
3252                  N_("write the pack index file in the specified idx format version"),
3253                  PARSE_OPT_NONEG, option_parse_index_version },
3254                OPT_MAGNITUDE(0, "max-pack-size", &pack_size_limit,
3255                              N_("maximum size of each output pack file")),
3256                OPT_BOOL(0, "local", &local,
3257                         N_("ignore borrowed objects from alternate object store")),
3258                OPT_BOOL(0, "incremental", &incremental,
3259                         N_("ignore packed objects")),
3260                OPT_INTEGER(0, "window", &window,
3261                            N_("limit pack window by objects")),
3262                OPT_MAGNITUDE(0, "window-memory", &window_memory_limit,
3263                              N_("limit pack window by memory in addition to object limit")),
3264                OPT_INTEGER(0, "depth", &depth,
3265                            N_("maximum length of delta chain allowed in the resulting pack")),
3266                OPT_BOOL(0, "reuse-delta", &reuse_delta,
3267                         N_("reuse existing deltas")),
3268                OPT_BOOL(0, "reuse-object", &reuse_object,
3269                         N_("reuse existing objects")),
3270                OPT_BOOL(0, "delta-base-offset", &allow_ofs_delta,
3271                         N_("use OFS_DELTA objects")),
3272                OPT_INTEGER(0, "threads", &delta_search_threads,
3273                            N_("use threads when searching for best delta matches")),
3274                OPT_BOOL(0, "non-empty", &non_empty,
3275                         N_("do not create an empty pack output")),
3276                OPT_BOOL(0, "revs", &use_internal_rev_list,
3277                         N_("read revision arguments from standard input")),
3278                OPT_SET_INT_F(0, "unpacked", &rev_list_unpacked,
3279                              N_("limit the objects to those that are not yet packed"),
3280                              1, PARSE_OPT_NONEG),
3281                OPT_SET_INT_F(0, "all", &rev_list_all,
3282                              N_("include objects reachable from any reference"),
3283                              1, PARSE_OPT_NONEG),
3284                OPT_SET_INT_F(0, "reflog", &rev_list_reflog,
3285                              N_("include objects referred by reflog entries"),
3286                              1, PARSE_OPT_NONEG),
3287                OPT_SET_INT_F(0, "indexed-objects", &rev_list_index,
3288                              N_("include objects referred to by the index"),
3289                              1, PARSE_OPT_NONEG),
3290                OPT_BOOL(0, "stdout", &pack_to_stdout,
3291                         N_("output pack to stdout")),
3292                OPT_BOOL(0, "include-tag", &include_tag,
3293                         N_("include tag objects that refer to objects to be packed")),
3294                OPT_BOOL(0, "keep-unreachable", &keep_unreachable,
3295                         N_("keep unreachable objects")),
3296                OPT_BOOL(0, "pack-loose-unreachable", &pack_loose_unreachable,
3297                         N_("pack loose unreachable objects")),
3298                { OPTION_CALLBACK, 0, "unpack-unreachable", NULL, N_("time"),
3299                  N_("unpack unreachable objects newer than <time>"),
3300                  PARSE_OPT_OPTARG, option_parse_unpack_unreachable },
3301                OPT_BOOL(0, "sparse", &sparse,
3302                         N_("use the sparse reachability algorithm")),
3303                OPT_BOOL(0, "thin", &thin,
3304                         N_("create thin packs")),
3305                OPT_BOOL(0, "shallow", &shallow,
3306                         N_("create packs suitable for shallow fetches")),
3307                OPT_BOOL(0, "honor-pack-keep", &ignore_packed_keep_on_disk,
3308                         N_("ignore packs that have companion .keep file")),
3309                OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
3310                                N_("ignore this pack")),
3311                OPT_INTEGER(0, "compression", &pack_compression_level,
3312                            N_("pack compression level")),
3313                OPT_SET_INT(0, "keep-true-parents", &grafts_replace_parents,
3314                            N_("do not hide commits by grafts"), 0),
3315                OPT_BOOL(0, "use-bitmap-index", &use_bitmap_index,
3316                         N_("use a bitmap index if available to speed up counting objects")),
3317                OPT_SET_INT(0, "write-bitmap-index", &write_bitmap_index,
3318                            N_("write a bitmap index together with the pack index"),
3319                            WRITE_BITMAP_TRUE),
3320                OPT_SET_INT_F(0, "write-bitmap-index-quiet",
3321                              &write_bitmap_index,
3322                              N_("write a bitmap index if possible"),
3323                              WRITE_BITMAP_QUIET, PARSE_OPT_HIDDEN),
3324                OPT_PARSE_LIST_OBJECTS_FILTER(&filter_options),
3325                { OPTION_CALLBACK, 0, "missing", NULL, N_("action"),
3326                  N_("handling for missing objects"), PARSE_OPT_NONEG,
3327                  option_parse_missing_action },
3328                OPT_BOOL(0, "exclude-promisor-objects", &exclude_promisor_objects,
3329                         N_("do not pack objects in promisor packfiles")),
3330                OPT_BOOL(0, "delta-islands", &use_delta_islands,
3331                         N_("respect islands during delta compression")),
3332                OPT_END(),
3333        };
3334
3335        if (DFS_NUM_STATES > (1 << OE_DFS_STATE_BITS))
3336                BUG("too many dfs states, increase OE_DFS_STATE_BITS");
3337
3338        read_replace_refs = 0;
3339
3340        sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
3341        reset_pack_idx_option(&pack_idx_opts);
3342        git_config(git_pack_config, NULL);
3343
3344        progress = isatty(2);
3345        argc = parse_options(argc, argv, prefix, pack_objects_options,
3346                             pack_usage, 0);
3347
3348        if (argc) {
3349                base_name = argv[0];
3350                argc--;
3351        }
3352        if (pack_to_stdout != !base_name || argc)
3353                usage_with_options(pack_usage, pack_objects_options);
3354
3355        if (depth >= (1 << OE_DEPTH_BITS)) {
3356                warning(_("delta chain depth %d is too deep, forcing %d"),
3357                        depth, (1 << OE_DEPTH_BITS) - 1);
3358                depth = (1 << OE_DEPTH_BITS) - 1;
3359        }
3360        if (cache_max_small_delta_size >= (1U << OE_Z_DELTA_BITS)) {
3361                warning(_("pack.deltaCacheLimit is too high, forcing %d"),
3362                        (1U << OE_Z_DELTA_BITS) - 1);
3363                cache_max_small_delta_size = (1U << OE_Z_DELTA_BITS) - 1;
3364        }
3365
3366        argv_array_push(&rp, "pack-objects");
3367        if (thin) {
3368                use_internal_rev_list = 1;
3369                argv_array_push(&rp, shallow
3370                                ? "--objects-edge-aggressive"
3371                                : "--objects-edge");
3372        } else
3373                argv_array_push(&rp, "--objects");
3374
3375        if (rev_list_all) {
3376                use_internal_rev_list = 1;
3377                argv_array_push(&rp, "--all");
3378        }
3379        if (rev_list_reflog) {
3380                use_internal_rev_list = 1;
3381                argv_array_push(&rp, "--reflog");
3382        }
3383        if (rev_list_index) {
3384                use_internal_rev_list = 1;
3385                argv_array_push(&rp, "--indexed-objects");
3386        }
3387        if (rev_list_unpacked) {
3388                use_internal_rev_list = 1;
3389                argv_array_push(&rp, "--unpacked");
3390        }
3391
3392        if (exclude_promisor_objects) {
3393                use_internal_rev_list = 1;
3394                fetch_if_missing = 0;
3395                argv_array_push(&rp, "--exclude-promisor-objects");
3396        }
3397        if (unpack_unreachable || keep_unreachable || pack_loose_unreachable)
3398                use_internal_rev_list = 1;
3399
3400        if (!reuse_object)
3401                reuse_delta = 0;
3402        if (pack_compression_level == -1)
3403                pack_compression_level = Z_DEFAULT_COMPRESSION;
3404        else if (pack_compression_level < 0 || pack_compression_level > Z_BEST_COMPRESSION)
3405                die(_("bad pack compression level %d"), pack_compression_level);
3406
3407        if (!delta_search_threads)      /* --threads=0 means autodetect */
3408                delta_search_threads = online_cpus();
3409
3410        if (!HAVE_THREADS && delta_search_threads != 1)
3411                warning(_("no threads support, ignoring --threads"));
3412        if (!pack_to_stdout && !pack_size_limit)
3413                pack_size_limit = pack_size_limit_cfg;
3414        if (pack_to_stdout && pack_size_limit)
3415                die(_("--max-pack-size cannot be used to build a pack for transfer"));
3416        if (pack_size_limit && pack_size_limit < 1024*1024) {
3417                warning(_("minimum pack size limit is 1 MiB"));
3418                pack_size_limit = 1024*1024;
3419        }
3420
3421        if (!pack_to_stdout && thin)
3422                die(_("--thin cannot be used to build an indexable pack"));
3423
3424        if (keep_unreachable && unpack_unreachable)
3425                die(_("--keep-unreachable and --unpack-unreachable are incompatible"));
3426        if (!rev_list_all || !rev_list_reflog || !rev_list_index)
3427                unpack_unreachable_expiration = 0;
3428
3429        if (filter_options.choice) {
3430                if (!pack_to_stdout)
3431                        die(_("cannot use --filter without --stdout"));
3432                use_bitmap_index = 0;
3433        }
3434
3435        /*
3436         * "soft" reasons not to use bitmaps - for on-disk repack by default we want
3437         *
3438         * - to produce good pack (with bitmap index not-yet-packed objects are
3439         *   packed in suboptimal order).
3440         *
3441         * - to use more robust pack-generation codepath (avoiding possible
3442         *   bugs in bitmap code and possible bitmap index corruption).
3443         */
3444        if (!pack_to_stdout)
3445                use_bitmap_index_default = 0;
3446
3447        if (use_bitmap_index < 0)
3448                use_bitmap_index = use_bitmap_index_default;
3449
3450        /* "hard" reasons not to use bitmaps; these just won't work at all */
3451        if (!use_internal_rev_list || (!pack_to_stdout && write_bitmap_index) || is_repository_shallow(the_repository))
3452                use_bitmap_index = 0;
3453
3454        if (pack_to_stdout || !rev_list_all)
3455                write_bitmap_index = 0;
3456
3457        if (use_delta_islands)
3458                argv_array_push(&rp, "--topo-order");
3459
3460        if (progress && all_progress_implied)
3461                progress = 2;
3462
3463        add_extra_kept_packs(&keep_pack_list);
3464        if (ignore_packed_keep_on_disk) {
3465                struct packed_git *p;
3466                for (p = get_all_packs(the_repository); p; p = p->next)
3467                        if (p->pack_local && p->pack_keep)
3468                                break;
3469                if (!p) /* no keep-able packs found */
3470                        ignore_packed_keep_on_disk = 0;
3471        }
3472        if (local) {
3473                /*
3474                 * unlike ignore_packed_keep_on_disk above, we do not
3475                 * want to unset "local" based on looking at packs, as
3476                 * it also covers non-local objects
3477                 */
3478                struct packed_git *p;
3479                for (p = get_all_packs(the_repository); p; p = p->next) {
3480                        if (!p->pack_local) {
3481                                have_non_local_packs = 1;
3482                                break;
3483                        }
3484                }
3485        }
3486
3487        trace2_region_enter("pack-objects", "enumerate-objects",
3488                            the_repository);
3489        prepare_packing_data(the_repository, &to_pack);
3490
3491        if (progress)
3492                progress_state = start_progress(_("Enumerating objects"), 0);
3493        if (!use_internal_rev_list)
3494                read_object_list_from_stdin();
3495        else {
3496                get_object_list(rp.argc, rp.argv);
3497                argv_array_clear(&rp);
3498        }
3499        cleanup_preferred_base();
3500        if (include_tag && nr_result)
3501                for_each_ref(add_ref_tag, NULL);
3502        stop_progress(&progress_state);
3503        trace2_region_leave("pack-objects", "enumerate-objects",
3504                            the_repository);
3505
3506        if (non_empty && !nr_result)
3507                return 0;
3508        if (nr_result) {
3509                trace2_region_enter("pack-objects", "prepare-pack",
3510                                    the_repository);
3511                prepare_pack(window, depth);
3512                trace2_region_leave("pack-objects", "prepare-pack",
3513                                    the_repository);
3514        }
3515
3516        trace2_region_enter("pack-objects", "write-pack-file", the_repository);
3517        write_pack_file();
3518        trace2_region_leave("pack-objects", "write-pack-file", the_repository);
3519
3520        if (progress)
3521                fprintf_ln(stderr,
3522                           _("Total %"PRIu32" (delta %"PRIu32"),"
3523                             " reused %"PRIu32" (delta %"PRIu32")"),
3524                           written, written_delta, reused, reused_delta);
3525        return 0;
3526}