git.oblomov.eu Git - git/blob - builtin-pack-objects.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "attr.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "delta.h"
  10 #include "pack.h"
  11 #include "pack-revindex.h"
  12 #include "csum-file.h"
  13 #include "tree-walk.h"
  14 #include "diff.h"
  15 #include "revision.h"
  16 #include "list-objects.h"
  17 #include "progress.h"
  18 #include "refs.h"
  19
  20 #ifdef THREADED_DELTA_SEARCH
  21 #include "thread-utils.h"
  22 #include <pthread.h>
  23 #endif
  24
  25 static const char pack_usage[] = "\
  26 git pack-objects [{ -q | --progress | --all-progress }] \n\
  27         [--max-pack-size=N] [--local] [--incremental] \n\
  28         [--window=N] [--window-memory=N] [--depth=N] \n\
  29         [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
  30         [--threads=N] [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
  31         [--stdout | base-name] [--include-tag] \n\
  32         [--keep-unreachable | --unpack-unreachable] \n\
  33         [<ref-list | <object-list]";
  34
  35 struct object_entry {
  36         struct pack_idx_entry idx;
  37         unsigned long size;     /* uncompressed size */
  38         struct packed_git *in_pack;     /* already in pack */
  39         off_t in_pack_offset;
  40         struct object_entry *delta;     /* delta base object */
  41         struct object_entry *delta_child; /* deltified objects who bases me */
  42         struct object_entry *delta_sibling; /* other deltified objects who
  43                                              * uses the same base as me
  44                                              */
  45         void *delta_data;       /* cached delta (uncompressed) */
  46         unsigned long delta_size;       /* delta data size (uncompressed) */
  47         unsigned long z_delta_size;     /* delta data size (compressed) */
  48         unsigned int hash;      /* name hint hash */
  49         enum object_type type;
  50         enum object_type in_pack_type;  /* could be delta */
  51         unsigned char in_pack_header_size;
  52         unsigned char preferred_base; /* we do not pack this, but is available
  53                                        * to be used as the base object to delta
  54                                        * objects against.
  55                                        */
  56         unsigned char no_try_delta;
  57 };
  58
  59 /*
  60  * Objects we are going to pack are collected in objects array (dynamically
  61  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  62  * in the order we see -- typically rev-list --objects order that gives us
  63  * nice "minimum seek" order.
  64  */
  65 static struct object_entry *objects;
  66 static struct pack_idx_entry **written_list;
  67 static uint32_t nr_objects, nr_alloc, nr_result, nr_written;
  68
  69 static int non_empty;
  70 static int reuse_delta = 1, reuse_object = 1;
  71 static int keep_unreachable, unpack_unreachable, include_tag;
  72 static int local;
  73 static int incremental;
  74 static int ignore_packed_keep;
  75 static int allow_ofs_delta;
  76 static const char *base_name;
  77 static int progress = 1;
  78 static int window = 10;
  79 static uint32_t pack_size_limit, pack_size_limit_cfg;
  80 static int depth = 50;
  81 static int delta_search_threads;
  82 static int pack_to_stdout;
  83 static int num_preferred_base;
  84 static struct progress *progress_state;
  85 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  86 static int pack_compression_seen;
  87
  88 static unsigned long delta_cache_size = 0;
  89 static unsigned long max_delta_cache_size = 0;
  90 static unsigned long cache_max_small_delta_size = 1000;
  91
  92 static unsigned long window_memory_limit = 0;
  93
  94 /*
  95  * The object names in objects array are hashed with this hashtable,
  96  * to help looking up the entry by object name.
  97  * This hashtable is built after all the objects are seen.
  98  */
  99 static int *object_ix;
 100 static int object_ix_hashsz;
 101
 102 /*
 103  * stats
 104  */
 105 static uint32_t written, written_delta;
 106 static uint32_t reused, reused_delta;
 107
 108
 109 static void *get_delta(struct object_entry *entry)
 110 {
 111         unsigned long size, base_size, delta_size;
 112         void *buf, *base_buf, *delta_buf;
 113         enum object_type type;
 114
 115         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 116         if (!buf)
 117                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 118         base_buf = read_sha1_file(entry->delta->idx.sha1, &type, &base_size);
 119         if (!base_buf)
 120                 die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1));
 121         delta_buf = diff_delta(base_buf, base_size,
 122                                buf, size, &delta_size, 0);
 123         if (!delta_buf || delta_size != entry->delta_size)
 124                 die("delta size changed");
 125         free(buf);
 126         free(base_buf);
 127         return delta_buf;
 128 }
 129
 130 static unsigned long do_compress(void **pptr, unsigned long size)
 131 {
 132         z_stream stream;
 133         void *in, *out;
 134         unsigned long maxsize;
 135
 136         memset(&stream, 0, sizeof(stream));
 137         deflateInit(&stream, pack_compression_level);
 138         maxsize = deflateBound(&stream, size);
 139
 140         in = *pptr;
 141         out = xmalloc(maxsize);
 142         *pptr = out;
 143
 144         stream.next_in = in;
 145         stream.avail_in = size;
 146         stream.next_out = out;
 147         stream.avail_out = maxsize;
 148         while (deflate(&stream, Z_FINISH) == Z_OK)
 149                 ; /* nothing */
 150         deflateEnd(&stream);
 151
 152         free(in);
 153         return stream.total_out;
 154 }
 155
 156 /*
 157  * The per-object header is a pretty dense thing, which is
 158  *  - first byte: low four bits are "size", then three bits of "type",
 159  *    and the high bit is "size continues".
 160  *  - each byte afterwards: low seven bits are size continuation,
 161  *    with the high bit being "size continues"
 162  */
 163 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 164 {
 165         int n = 1;
 166         unsigned char c;
 167
 168         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 169                 die("bad type %d", type);
 170
 171         c = (type << 4) | (size & 15);
 172         size >>= 4;
 173         while (size) {
 174                 *hdr++ = c | 0x80;
 175                 c = size & 0x7f;
 176                 size >>= 7;
 177                 n++;
 178         }
 179         *hdr = c;
 180         return n;
 181 }
 182
 183 /*
 184  * we are going to reuse the existing object data as is.  make
 185  * sure it is not corrupt.
 186  */
 187 static int check_pack_inflate(struct packed_git *p,
 188                 struct pack_window **w_curs,
 189                 off_t offset,
 190                 off_t len,
 191                 unsigned long expect)
 192 {
 193         z_stream stream;
 194         unsigned char fakebuf[4096], *in;
 195         int st;
 196
 197         memset(&stream, 0, sizeof(stream));
 198         git_inflate_init(&stream);
 199         do {
 200                 in = use_pack(p, w_curs, offset, &stream.avail_in);
 201                 stream.next_in = in;
 202                 stream.next_out = fakebuf;
 203                 stream.avail_out = sizeof(fakebuf);
 204                 st = git_inflate(&stream, Z_FINISH);
 205                 offset += stream.next_in - in;
 206         } while (st == Z_OK || st == Z_BUF_ERROR);
 207         git_inflate_end(&stream);
 208         return (st == Z_STREAM_END &&
 209                 stream.total_out == expect &&
 210                 stream.total_in == len) ? 0 : -1;
 211 }
 212
 213 static void copy_pack_data(struct sha1file *f,
 214                 struct packed_git *p,
 215                 struct pack_window **w_curs,
 216                 off_t offset,
 217                 off_t len)
 218 {
 219         unsigned char *in;
 220         unsigned int avail;
 221
 222         while (len) {
 223                 in = use_pack(p, w_curs, offset, &avail);
 224                 if (avail > len)
 225                         avail = (unsigned int)len;
 226                 sha1write(f, in, avail);
 227                 offset += avail;
 228                 len -= avail;
 229         }
 230 }
 231
 232 static unsigned long write_object(struct sha1file *f,
 233                                   struct object_entry *entry,
 234                                   off_t write_offset)
 235 {
 236         unsigned long size, limit, datalen;
 237         void *buf;
 238         unsigned char header[10], dheader[10];
 239         unsigned hdrlen;
 240         enum object_type type;
 241         int usable_delta, to_reuse;
 242
 243         if (!pack_to_stdout)
 244                 crc32_begin(f);
 245
 246         type = entry->type;
 247
 248         /* write limit if limited packsize and not first object */
 249         if (!pack_size_limit || !nr_written)
 250                 limit = 0;
 251         else if (pack_size_limit <= write_offset)
 252                 /*
 253                  * the earlier object did not fit the limit; avoid
 254                  * mistaking this with unlimited (i.e. limit = 0).
 255                  */
 256                 limit = 1;
 257         else
 258                 limit = pack_size_limit - write_offset;
 259
 260         if (!entry->delta)
 261                 usable_delta = 0;       /* no delta */
 262         else if (!pack_size_limit)
 263                usable_delta = 1;        /* unlimited packfile */
 264         else if (entry->delta->idx.offset == (off_t)-1)
 265                 usable_delta = 0;       /* base was written to another pack */
 266         else if (entry->delta->idx.offset)
 267                 usable_delta = 1;       /* base already exists in this pack */
 268         else
 269                 usable_delta = 0;       /* base could end up in another pack */
 270
 271         if (!reuse_object)
 272                 to_reuse = 0;   /* explicit */
 273         else if (!entry->in_pack)
 274                 to_reuse = 0;   /* can't reuse what we don't have */
 275         else if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA)
 276                                 /* check_object() decided it for us ... */
 277                 to_reuse = usable_delta;
 278                                 /* ... but pack split may override that */
 279         else if (type != entry->in_pack_type)
 280                 to_reuse = 0;   /* pack has delta which is unusable */
 281         else if (entry->delta)
 282                 to_reuse = 0;   /* we want to pack afresh */
 283         else
 284                 to_reuse = 1;   /* we have it in-pack undeltified,
 285                                  * and we do not need to deltify it.
 286                                  */
 287
 288         if (!to_reuse) {
 289                 no_reuse:
 290                 if (!usable_delta) {
 291                         buf = read_sha1_file(entry->idx.sha1, &type, &size);
 292                         if (!buf)
 293                                 die("unable to read %s", sha1_to_hex(entry->idx.sha1));
 294                         /*
 295                          * make sure no cached delta data remains from a
 296                          * previous attempt before a pack split occurred.
 297                          */
 298                         free(entry->delta_data);
 299                         entry->delta_data = NULL;
 300                         entry->z_delta_size = 0;
 301                 } else if (entry->delta_data) {
 302                         size = entry->delta_size;
 303                         buf = entry->delta_data;
 304                         entry->delta_data = NULL;
 305                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 306                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 307                 } else {
 308                         buf = get_delta(entry);
 309                         size = entry->delta_size;
 310                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 311                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 312                 }
 313
 314                 if (entry->z_delta_size)
 315                         datalen = entry->z_delta_size;
 316                 else
 317                         datalen = do_compress(&buf, size);
 318
 319                 /*
 320                  * The object header is a byte of 'type' followed by zero or
 321                  * more bytes of length.
 322                  */
 323                 hdrlen = encode_header(type, size, header);
 324
 325                 if (type == OBJ_OFS_DELTA) {
 326                         /*
 327                          * Deltas with relative base contain an additional
 328                          * encoding of the relative offset for the delta
 329                          * base from this object's position in the pack.
 330                          */
 331                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 332                         unsigned pos = sizeof(dheader) - 1;
 333                         dheader[pos] = ofs & 127;
 334                         while (ofs >>= 7)
 335                                 dheader[--pos] = 128 | (--ofs & 127);
 336                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 337                                 free(buf);
 338                                 return 0;
 339                         }
 340                         sha1write(f, header, hdrlen);
 341                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 342                         hdrlen += sizeof(dheader) - pos;
 343                 } else if (type == OBJ_REF_DELTA) {
 344                         /*
 345                          * Deltas with a base reference contain
 346                          * an additional 20 bytes for the base sha1.
 347                          */
 348                         if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 349                                 free(buf);
 350                                 return 0;
 351                         }
 352                         sha1write(f, header, hdrlen);
 353                         sha1write(f, entry->delta->idx.sha1, 20);
 354                         hdrlen += 20;
 355                 } else {
 356                         if (limit && hdrlen + datalen + 20 >= limit) {
 357                                 free(buf);
 358                                 return 0;
 359                         }
 360                         sha1write(f, header, hdrlen);
 361                 }
 362                 sha1write(f, buf, datalen);
 363                 free(buf);
 364         }
 365         else {
 366                 struct packed_git *p = entry->in_pack;
 367                 struct pack_window *w_curs = NULL;
 368                 struct revindex_entry *revidx;
 369                 off_t offset;
 370
 371                 if (entry->delta)
 372                         type = (allow_ofs_delta && entry->delta->idx.offset) ?
 373                                 OBJ_OFS_DELTA : OBJ_REF_DELTA;
 374                 hdrlen = encode_header(type, entry->size, header);
 375
 376                 offset = entry->in_pack_offset;
 377                 revidx = find_pack_revindex(p, offset);
 378                 datalen = revidx[1].offset - offset;
 379                 if (!pack_to_stdout && p->index_version > 1 &&
 380                     check_pack_crc(p, &w_curs, offset, datalen, revidx->nr)) {
 381                         error("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1));
 382                         unuse_pack(&w_curs);
 383                         goto no_reuse;
 384                 }
 385
 386                 offset += entry->in_pack_header_size;
 387                 datalen -= entry->in_pack_header_size;
 388                 if (!pack_to_stdout && p->index_version == 1 &&
 389                     check_pack_inflate(p, &w_curs, offset, datalen, entry->size)) {
 390                         error("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1));
 391                         unuse_pack(&w_curs);
 392                         goto no_reuse;
 393                 }
 394
 395                 if (type == OBJ_OFS_DELTA) {
 396                         off_t ofs = entry->idx.offset - entry->delta->idx.offset;
 397                         unsigned pos = sizeof(dheader) - 1;
 398                         dheader[pos] = ofs & 127;
 399                         while (ofs >>= 7)
 400                                 dheader[--pos] = 128 | (--ofs & 127);
 401                         if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) {
 402                                 unuse_pack(&w_curs);
 403                                 return 0;
 404                         }
 405                         sha1write(f, header, hdrlen);
 406                         sha1write(f, dheader + pos, sizeof(dheader) - pos);
 407                         hdrlen += sizeof(dheader) - pos;
 408                         reused_delta++;
 409                 } else if (type == OBJ_REF_DELTA) {
 410                         if (limit && hdrlen + 20 + datalen + 20 >= limit) {
 411                                 unuse_pack(&w_curs);
 412                                 return 0;
 413                         }
 414                         sha1write(f, header, hdrlen);
 415                         sha1write(f, entry->delta->idx.sha1, 20);
 416                         hdrlen += 20;
 417                         reused_delta++;
 418                 } else {
 419                         if (limit && hdrlen + datalen + 20 >= limit) {
 420                                 unuse_pack(&w_curs);
 421                                 return 0;
 422                         }
 423                         sha1write(f, header, hdrlen);
 424                 }
 425                 copy_pack_data(f, p, &w_curs, offset, datalen);
 426                 unuse_pack(&w_curs);
 427                 reused++;
 428         }
 429         if (usable_delta)
 430                 written_delta++;
 431         written++;
 432         if (!pack_to_stdout)
 433                 entry->idx.crc32 = crc32_end(f);
 434         return hdrlen + datalen;
 435 }
 436
 437 static int write_one(struct sha1file *f,
 438                                struct object_entry *e,
 439                                off_t *offset)
 440 {
 441         unsigned long size;
 442
 443         /* offset is non zero if object is written already. */
 444         if (e->idx.offset || e->preferred_base)
 445                 return 1;
 446
 447         /* if we are deltified, write out base object first. */
 448         if (e->delta && !write_one(f, e->delta, offset))
 449                 return 0;
 450
 451         e->idx.offset = *offset;
 452         size = write_object(f, e, *offset);
 453         if (!size) {
 454                 e->idx.offset = 0;
 455                 return 0;
 456         }
 457         written_list[nr_written++] = &e->idx;
 458
 459         /* make sure off_t is sufficiently large not to wrap */
 460         if (*offset > *offset + size)
 461                 die("pack too large for current definition of off_t");
 462         *offset += size;
 463         return 1;
 464 }
 465
 466 /* forward declaration for write_pack_file */
 467 static int adjust_perm(const char *path, mode_t mode);
 468
 469 static void write_pack_file(void)
 470 {
 471         uint32_t i = 0, j;
 472         struct sha1file *f;
 473         off_t offset;
 474         struct pack_header hdr;
 475         uint32_t nr_remaining = nr_result;
 476         time_t last_mtime = 0;
 477
 478         if (progress > pack_to_stdout)
 479                 progress_state = start_progress("Writing objects", nr_result);
 480         written_list = xmalloc(nr_objects * sizeof(*written_list));
 481
 482         do {
 483                 unsigned char sha1[20];
 484                 char *pack_tmp_name = NULL;
 485
 486                 if (pack_to_stdout) {
 487                         f = sha1fd_throughput(1, "<stdout>", progress_state);
 488                 } else {
 489                         char tmpname[PATH_MAX];
 490                         int fd;
 491                         fd = odb_mkstemp(tmpname, sizeof(tmpname),
 492                                          "pack/tmp_pack_XXXXXX");
 493                         pack_tmp_name = xstrdup(tmpname);
 494                         f = sha1fd(fd, pack_tmp_name);
 495                 }
 496
 497                 hdr.hdr_signature = htonl(PACK_SIGNATURE);
 498                 hdr.hdr_version = htonl(PACK_VERSION);
 499                 hdr.hdr_entries = htonl(nr_remaining);
 500                 sha1write(f, &hdr, sizeof(hdr));
 501                 offset = sizeof(hdr);
 502                 nr_written = 0;
 503                 for (; i < nr_objects; i++) {
 504                         if (!write_one(f, objects + i, &offset))
 505                                 break;
 506                         display_progress(progress_state, written);
 507                 }
 508
 509                 /*
 510                  * Did we write the wrong # entries in the header?
 511                  * If so, rewrite it like in fast-import
 512                  */
 513                 if (pack_to_stdout) {
 514                         sha1close(f, sha1, CSUM_CLOSE);
 515                 } else if (nr_written == nr_remaining) {
 516                         sha1close(f, sha1, CSUM_FSYNC);
 517                 } else {
 518                         int fd = sha1close(f, sha1, 0);
 519                         fixup_pack_header_footer(fd, sha1, pack_tmp_name,
 520                                                  nr_written, sha1, offset);
 521                         close(fd);
 522                 }
 523
 524                 if (!pack_to_stdout) {
 525                         mode_t mode = umask(0);
 526                         struct stat st;
 527                         char *idx_tmp_name, tmpname[PATH_MAX];
 528
 529                         umask(mode);
 530                         mode = 0444 & ~mode;
 531
 532                         idx_tmp_name = write_idx_file(NULL, written_list,
 533                                                       nr_written, sha1);
 534
 535                         snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
 536                                  base_name, sha1_to_hex(sha1));
 537                         free_pack_by_name(tmpname);
 538                         if (adjust_perm(pack_tmp_name, mode))
 539                                 die("unable to make temporary pack file readable: %s",
 540                                     strerror(errno));
 541                         if (rename(pack_tmp_name, tmpname))
 542                                 die("unable to rename temporary pack file: %s",
 543                                     strerror(errno));
 544
 545                         /*
 546                          * Packs are runtime accessed in their mtime
 547                          * order since newer packs are more likely to contain
 548                          * younger objects.  So if we are creating multiple
 549                          * packs then we should modify the mtime of later ones
 550                          * to preserve this property.
 551                          */
 552                         if (stat(tmpname, &st) < 0) {
 553                                 warning("failed to stat %s: %s",
 554                                         tmpname, strerror(errno));
 555                         } else if (!last_mtime) {
 556                                 last_mtime = st.st_mtime;
 557                         } else {
 558                                 struct utimbuf utb;
 559                                 utb.actime = st.st_atime;
 560                                 utb.modtime = --last_mtime;
 561                                 if (utime(tmpname, &utb) < 0)
 562                                         warning("failed utime() on %s: %s",
 563                                                 tmpname, strerror(errno));
 564                         }
 565
 566                         snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
 567                                  base_name, sha1_to_hex(sha1));
 568                         if (adjust_perm(idx_tmp_name, mode))
 569                                 die("unable to make temporary index file readable: %s",
 570                                     strerror(errno));
 571                         if (rename(idx_tmp_name, tmpname))
 572                                 die("unable to rename temporary index file: %s",
 573                                     strerror(errno));
 574
 575                         free(idx_tmp_name);
 576                         free(pack_tmp_name);
 577                         puts(sha1_to_hex(sha1));
 578                 }
 579
 580                 /* mark written objects as written to previous pack */
 581                 for (j = 0; j < nr_written; j++) {
 582                         written_list[j]->offset = (off_t)-1;
 583                 }
 584                 nr_remaining -= nr_written;
 585         } while (nr_remaining && i < nr_objects);
 586
 587         free(written_list);
 588         stop_progress(&progress_state);
 589         if (written != nr_result)
 590                 die("wrote %"PRIu32" objects while expecting %"PRIu32,
 591                         written, nr_result);
 592         /*
 593          * We have scanned through [0 ... i).  Since we have written
 594          * the correct number of objects,  the remaining [i ... nr_objects)
 595          * items must be either already written (due to out-of-order delta base)
 596          * or a preferred base.  Count those which are neither and complain if any.
 597          */
 598         for (j = 0; i < nr_objects; i++) {
 599                 struct object_entry *e = objects + i;
 600                 j += !e->idx.offset && !e->preferred_base;
 601         }
 602         if (j)
 603                 die("wrote %"PRIu32" objects as expected but %"PRIu32
 604                         " unwritten", written, j);
 605 }
 606
 607 static int locate_object_entry_hash(const unsigned char *sha1)
 608 {
 609         int i;
 610         unsigned int ui;
 611         memcpy(&ui, sha1, sizeof(unsigned int));
 612         i = ui % object_ix_hashsz;
 613         while (0 < object_ix[i]) {
 614                 if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1))
 615                         return i;
 616                 if (++i == object_ix_hashsz)
 617                         i = 0;
 618         }
 619         return -1 - i;
 620 }
 621
 622 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 623 {
 624         int i;
 625
 626         if (!object_ix_hashsz)
 627                 return NULL;
 628
 629         i = locate_object_entry_hash(sha1);
 630         if (0 <= i)
 631                 return &objects[object_ix[i]-1];
 632         return NULL;
 633 }
 634
 635 static void rehash_objects(void)
 636 {
 637         uint32_t i;
 638         struct object_entry *oe;
 639
 640         object_ix_hashsz = nr_objects * 3;
 641         if (object_ix_hashsz < 1024)
 642                 object_ix_hashsz = 1024;
 643         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 644         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 645         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 646                 int ix = locate_object_entry_hash(oe->idx.sha1);
 647                 if (0 <= ix)
 648                         continue;
 649                 ix = -1 - ix;
 650                 object_ix[ix] = i + 1;
 651         }
 652 }
 653
 654 static unsigned name_hash(const char *name)
 655 {
 656         unsigned char c;
 657         unsigned hash = 0;
 658
 659         if (!name)
 660                 return 0;
 661
 662         /*
 663          * This effectively just creates a sortable number from the
 664          * last sixteen non-whitespace characters. Last characters
 665          * count "most", so things that end in ".c" sort together.
 666          */
 667         while ((c = *name++) != 0) {
 668                 if (isspace(c))
 669                         continue;
 670                 hash = (hash >> 2) + (c << 24);
 671         }
 672         return hash;
 673 }
 674
 675 static void setup_delta_attr_check(struct git_attr_check *check)
 676 {
 677         static struct git_attr *attr_delta;
 678
 679         if (!attr_delta)
 680                 attr_delta = git_attr("delta", 5);
 681
 682         check[0].attr = attr_delta;
 683 }
 684
 685 static int no_try_delta(const char *path)
 686 {
 687         struct git_attr_check check[1];
 688
 689         setup_delta_attr_check(check);
 690         if (git_checkattr(path, ARRAY_SIZE(check), check))
 691                 return 0;
 692         if (ATTR_FALSE(check->value))
 693                 return 1;
 694         return 0;
 695 }
 696
 697 static int add_object_entry(const unsigned char *sha1, enum object_type type,
 698                             const char *name, int exclude)
 699 {
 700         struct object_entry *entry;
 701         struct packed_git *p, *found_pack = NULL;
 702         off_t found_offset = 0;
 703         int ix;
 704         unsigned hash = name_hash(name);
 705
 706         ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
 707         if (ix >= 0) {
 708                 if (exclude) {
 709                         entry = objects + object_ix[ix] - 1;
 710                         if (!entry->preferred_base)
 711                                 nr_result--;
 712                         entry->preferred_base = 1;
 713                 }
 714                 return 0;
 715         }
 716
 717         if (!exclude && local && has_loose_object_nonlocal(sha1))
 718                 return 0;
 719
 720         for (p = packed_git; p; p = p->next) {
 721                 off_t offset = find_pack_entry_one(sha1, p);
 722                 if (offset) {
 723                         if (!found_pack) {
 724                                 found_offset = offset;
 725                                 found_pack = p;
 726                         }
 727                         if (exclude)
 728                                 break;
 729                         if (incremental)
 730                                 return 0;
 731                         if (local && !p->pack_local)
 732                                 return 0;
 733                         if (ignore_packed_keep && p->pack_local && p->pack_keep)
 734                                 return 0;
 735                 }
 736         }
 737
 738         if (nr_objects >= nr_alloc) {
 739                 nr_alloc = (nr_alloc  + 1024) * 3 / 2;
 740                 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
 741         }
 742
 743         entry = objects + nr_objects++;
 744         memset(entry, 0, sizeof(*entry));
 745         hashcpy(entry->idx.sha1, sha1);
 746         entry->hash = hash;
 747         if (type)
 748                 entry->type = type;
 749         if (exclude)
 750                 entry->preferred_base = 1;
 751         else
 752                 nr_result++;
 753         if (found_pack) {
 754                 entry->in_pack = found_pack;
 755                 entry->in_pack_offset = found_offset;
 756         }
 757
 758         if (object_ix_hashsz * 3 <= nr_objects * 4)
 759                 rehash_objects();
 760         else
 761                 object_ix[-1 - ix] = nr_objects;
 762
 763         display_progress(progress_state, nr_objects);
 764
 765         if (name && no_try_delta(name))
 766                 entry->no_try_delta = 1;
 767
 768         return 1;
 769 }
 770
 771 struct pbase_tree_cache {
 772         unsigned char sha1[20];
 773         int ref;
 774         int temporary;
 775         void *tree_data;
 776         unsigned long tree_size;
 777 };
 778
 779 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 780 static int pbase_tree_cache_ix(const unsigned char *sha1)
 781 {
 782         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 783 }
 784 static int pbase_tree_cache_ix_incr(int ix)
 785 {
 786         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 787 }
 788
 789 static struct pbase_tree {
 790         struct pbase_tree *next;
 791         /* This is a phony "cache" entry; we are not
 792          * going to evict it nor find it through _get()
 793          * mechanism -- this is for the toplevel node that
 794          * would almost always change with any commit.
 795          */
 796         struct pbase_tree_cache pcache;
 797 } *pbase_tree;
 798
 799 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 800 {
 801         struct pbase_tree_cache *ent, *nent;
 802         void *data;
 803         unsigned long size;
 804         enum object_type type;
 805         int neigh;
 806         int my_ix = pbase_tree_cache_ix(sha1);
 807         int available_ix = -1;
 808
 809         /* pbase-tree-cache acts as a limited hashtable.
 810          * your object will be found at your index or within a few
 811          * slots after that slot if it is cached.
 812          */
 813         for (neigh = 0; neigh < 8; neigh++) {
 814                 ent = pbase_tree_cache[my_ix];
 815                 if (ent && !hashcmp(ent->sha1, sha1)) {
 816                         ent->ref++;
 817                         return ent;
 818                 }
 819                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 820                          ((0 <= available_ix) &&
 821                           (!ent && pbase_tree_cache[available_ix])))
 822                         available_ix = my_ix;
 823                 if (!ent)
 824                         break;
 825                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 826         }
 827
 828         /* Did not find one.  Either we got a bogus request or
 829          * we need to read and perhaps cache.
 830          */
 831         data = read_sha1_file(sha1, &type, &size);
 832         if (!data)
 833                 return NULL;
 834         if (type != OBJ_TREE) {
 835                 free(data);
 836                 return NULL;
 837         }
 838
 839         /* We need to either cache or return a throwaway copy */
 840
 841         if (available_ix < 0)
 842                 ent = NULL;
 843         else {
 844                 ent = pbase_tree_cache[available_ix];
 845                 my_ix = available_ix;
 846         }
 847
 848         if (!ent) {
 849                 nent = xmalloc(sizeof(*nent));
 850                 nent->temporary = (available_ix < 0);
 851         }
 852         else {
 853                 /* evict and reuse */
 854                 free(ent->tree_data);
 855                 nent = ent;
 856         }
 857         hashcpy(nent->sha1, sha1);
 858         nent->tree_data = data;
 859         nent->tree_size = size;
 860         nent->ref = 1;
 861         if (!nent->temporary)
 862                 pbase_tree_cache[my_ix] = nent;
 863         return nent;
 864 }
 865
 866 static void pbase_tree_put(struct pbase_tree_cache *cache)
 867 {
 868         if (!cache->temporary) {
 869                 cache->ref--;
 870                 return;
 871         }
 872         free(cache->tree_data);
 873         free(cache);
 874 }
 875
 876 static int name_cmp_len(const char *name)
 877 {
 878         int i;
 879         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 880                 ;
 881         return i;
 882 }
 883
 884 static void add_pbase_object(struct tree_desc *tree,
 885                              const char *name,
 886                              int cmplen,
 887                              const char *fullname)
 888 {
 889         struct name_entry entry;
 890         int cmp;
 891
 892         while (tree_entry(tree,&entry)) {
 893                 if (S_ISGITLINK(entry.mode))
 894                         continue;
 895                 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
 896                       memcmp(name, entry.path, cmplen);
 897                 if (cmp > 0)
 898                         continue;
 899                 if (cmp < 0)
 900                         return;
 901                 if (name[cmplen] != '/') {
 902                         add_object_entry(entry.sha1,
 903                                          object_type(entry.mode),
 904                                          fullname, 1);
 905                         return;
 906                 }
 907                 if (S_ISDIR(entry.mode)) {
 908                         struct tree_desc sub;
 909                         struct pbase_tree_cache *tree;
 910                         const char *down = name+cmplen+1;
 911                         int downlen = name_cmp_len(down);
 912
 913                         tree = pbase_tree_get(entry.sha1);
 914                         if (!tree)
 915                                 return;
 916                         init_tree_desc(&sub, tree->tree_data, tree->tree_size);
 917
 918                         add_pbase_object(&sub, down, downlen, fullname);
 919                         pbase_tree_put(tree);
 920                 }
 921         }
 922 }
 923
 924 static unsigned *done_pbase_paths;
 925 static int done_pbase_paths_num;
 926 static int done_pbase_paths_alloc;
 927 static int done_pbase_path_pos(unsigned hash)
 928 {
 929         int lo = 0;
 930         int hi = done_pbase_paths_num;
 931         while (lo < hi) {
 932                 int mi = (hi + lo) / 2;
 933                 if (done_pbase_paths[mi] == hash)
 934                         return mi;
 935                 if (done_pbase_paths[mi] < hash)
 936                         hi = mi;
 937                 else
 938                         lo = mi + 1;
 939         }
 940         return -lo-1;
 941 }
 942
 943 static int check_pbase_path(unsigned hash)
 944 {
 945         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
 946         if (0 <= pos)
 947                 return 1;
 948         pos = -pos - 1;
 949         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
 950                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
 951                 done_pbase_paths = xrealloc(done_pbase_paths,
 952                                             done_pbase_paths_alloc *
 953                                             sizeof(unsigned));
 954         }
 955         done_pbase_paths_num++;
 956         if (pos < done_pbase_paths_num)
 957                 memmove(done_pbase_paths + pos + 1,
 958                         done_pbase_paths + pos,
 959                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
 960         done_pbase_paths[pos] = hash;
 961         return 0;
 962 }
 963
 964 static void add_preferred_base_object(const char *name)
 965 {
 966         struct pbase_tree *it;
 967         int cmplen;
 968         unsigned hash = name_hash(name);
 969
 970         if (!num_preferred_base || check_pbase_path(hash))
 971                 return;
 972
 973         cmplen = name_cmp_len(name);
 974         for (it = pbase_tree; it; it = it->next) {
 975                 if (cmplen == 0) {
 976                         add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
 977                 }
 978                 else {
 979                         struct tree_desc tree;
 980                         init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
 981                         add_pbase_object(&tree, name, cmplen, name);
 982                 }
 983         }
 984 }
 985
 986 static void add_preferred_base(unsigned char *sha1)
 987 {
 988         struct pbase_tree *it;
 989         void *data;
 990         unsigned long size;
 991         unsigned char tree_sha1[20];
 992
 993         if (window <= num_preferred_base++)
 994                 return;
 995
 996         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
 997         if (!data)
 998                 return;
 999
1000         for (it = pbase_tree; it; it = it->next) {
1001                 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1002                         free(data);
1003                         return;
1004                 }
1005         }
1006
1007         it = xcalloc(1, sizeof(*it));
1008         it->next = pbase_tree;
1009         pbase_tree = it;
1010
1011         hashcpy(it->pcache.sha1, tree_sha1);
1012         it->pcache.tree_data = data;
1013         it->pcache.tree_size = size;
1014 }
1015
1016 static void check_object(struct object_entry *entry)
1017 {
1018         if (entry->in_pack) {
1019                 struct packed_git *p = entry->in_pack;
1020                 struct pack_window *w_curs = NULL;
1021                 const unsigned char *base_ref = NULL;
1022                 struct object_entry *base_entry;
1023                 unsigned long used, used_0;
1024                 unsigned int avail;
1025                 off_t ofs;
1026                 unsigned char *buf, c;
1027
1028                 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1029
1030                 /*
1031                  * We want in_pack_type even if we do not reuse delta
1032                  * since non-delta representations could still be reused.
1033                  */
1034                 used = unpack_object_header_buffer(buf, avail,
1035                                                    &entry->in_pack_type,
1036                                                    &entry->size);
1037                 if (used == 0)
1038                         goto give_up;
1039
1040                 /*
1041                  * Determine if this is a delta and if so whether we can
1042                  * reuse it or not.  Otherwise let's find out as cheaply as
1043                  * possible what the actual type and size for this object is.
1044                  */
1045                 switch (entry->in_pack_type) {
1046                 default:
1047                         /* Not a delta hence we've already got all we need. */
1048                         entry->type = entry->in_pack_type;
1049                         entry->in_pack_header_size = used;
1050                         if (entry->type < OBJ_COMMIT || entry->type > OBJ_BLOB)
1051                                 goto give_up;
1052                         unuse_pack(&w_curs);
1053                         return;
1054                 case OBJ_REF_DELTA:
1055                         if (reuse_delta && !entry->preferred_base)
1056                                 base_ref = use_pack(p, &w_curs,
1057                                                 entry->in_pack_offset + used, NULL);
1058                         entry->in_pack_header_size = used + 20;
1059                         break;
1060                 case OBJ_OFS_DELTA:
1061                         buf = use_pack(p, &w_curs,
1062                                        entry->in_pack_offset + used, NULL);
1063                         used_0 = 0;
1064                         c = buf[used_0++];
1065                         ofs = c & 127;
1066                         while (c & 128) {
1067                                 ofs += 1;
1068                                 if (!ofs || MSB(ofs, 7)) {
1069                                         error("delta base offset overflow in pack for %s",
1070                                               sha1_to_hex(entry->idx.sha1));
1071                                         goto give_up;
1072                                 }
1073                                 c = buf[used_0++];
1074                                 ofs = (ofs << 7) + (c & 127);
1075                         }
1076                         ofs = entry->in_pack_offset - ofs;
1077                         if (ofs <= 0 || ofs >= entry->in_pack_offset) {
1078                                 error("delta base offset out of bound for %s",
1079                                       sha1_to_hex(entry->idx.sha1));
1080                                 goto give_up;
1081                         }
1082                         if (reuse_delta && !entry->preferred_base) {
1083                                 struct revindex_entry *revidx;
1084                                 revidx = find_pack_revindex(p, ofs);
1085                                 if (!revidx)
1086                                         goto give_up;
1087                                 base_ref = nth_packed_object_sha1(p, revidx->nr);
1088                         }
1089                         entry->in_pack_header_size = used + used_0;
1090                         break;
1091                 }
1092
1093                 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1094                         /*
1095                          * If base_ref was set above that means we wish to
1096                          * reuse delta data, and we even found that base
1097                          * in the list of objects we want to pack. Goodie!
1098                          *
1099                          * Depth value does not matter - find_deltas() will
1100                          * never consider reused delta as the base object to
1101                          * deltify other objects against, in order to avoid
1102                          * circular deltas.
1103                          */
1104                         entry->type = entry->in_pack_type;
1105                         entry->delta = base_entry;
1106                         entry->delta_size = entry->size;
1107                         entry->delta_sibling = base_entry->delta_child;
1108                         base_entry->delta_child = entry;
1109                         unuse_pack(&w_curs);
1110                         return;
1111                 }
1112
1113                 if (entry->type) {
1114                         /*
1115                          * This must be a delta and we already know what the
1116                          * final object type is.  Let's extract the actual
1117                          * object size from the delta header.
1118                          */
1119                         entry->size = get_size_from_delta(p, &w_curs,
1120                                         entry->in_pack_offset + entry->in_pack_header_size);
1121                         if (entry->size == 0)
1122                                 goto give_up;
1123                         unuse_pack(&w_curs);
1124                         return;
1125                 }
1126
1127                 /*
1128                  * No choice but to fall back to the recursive delta walk
1129                  * with sha1_object_info() to find about the object type
1130                  * at this point...
1131                  */
1132                 give_up:
1133                 unuse_pack(&w_curs);
1134         }
1135
1136         entry->type = sha1_object_info(entry->idx.sha1, &entry->size);
1137         /*
1138          * The error condition is checked in prepare_pack().  This is
1139          * to permit a missing preferred base object to be ignored
1140          * as a preferred base.  Doing so can result in a larger
1141          * pack file, but the transfer will still take place.
1142          */
1143 }
1144
1145 static int pack_offset_sort(const void *_a, const void *_b)
1146 {
1147         const struct object_entry *a = *(struct object_entry **)_a;
1148         const struct object_entry *b = *(struct object_entry **)_b;
1149
1150         /* avoid filesystem trashing with loose objects */
1151         if (!a->in_pack && !b->in_pack)
1152                 return hashcmp(a->idx.sha1, b->idx.sha1);
1153
1154         if (a->in_pack < b->in_pack)
1155                 return -1;
1156         if (a->in_pack > b->in_pack)
1157                 return 1;
1158         return a->in_pack_offset < b->in_pack_offset ? -1 :
1159                         (a->in_pack_offset > b->in_pack_offset);
1160 }
1161
1162 static void get_object_details(void)
1163 {
1164         uint32_t i;
1165         struct object_entry **sorted_by_offset;
1166
1167         sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1168         for (i = 0; i < nr_objects; i++)
1169                 sorted_by_offset[i] = objects + i;
1170         qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1171
1172         for (i = 0; i < nr_objects; i++)
1173                 check_object(sorted_by_offset[i]);
1174
1175         free(sorted_by_offset);
1176 }
1177
1178 /*
1179  * We search for deltas in a list sorted by type, by filename hash, and then
1180  * by size, so that we see progressively smaller and smaller files.
1181  * That's because we prefer deltas to be from the bigger file
1182  * to the smaller -- deletes are potentially cheaper, but perhaps
1183  * more importantly, the bigger file is likely the more recent
1184  * one.  The deepest deltas are therefore the oldest objects which are
1185  * less susceptible to be accessed often.
1186  */
1187 static int type_size_sort(const void *_a, const void *_b)
1188 {
1189         const struct object_entry *a = *(struct object_entry **)_a;
1190         const struct object_entry *b = *(struct object_entry **)_b;
1191
1192         if (a->type > b->type)
1193                 return -1;
1194         if (a->type < b->type)
1195                 return 1;
1196         if (a->hash > b->hash)
1197                 return -1;
1198         if (a->hash < b->hash)
1199                 return 1;
1200         if (a->preferred_base > b->preferred_base)
1201                 return -1;
1202         if (a->preferred_base < b->preferred_base)
1203                 return 1;
1204         if (a->size > b->size)
1205                 return -1;
1206         if (a->size < b->size)
1207                 return 1;
1208         return a < b ? -1 : (a > b);  /* newest first */
1209 }
1210
1211 struct unpacked {
1212         struct object_entry *entry;
1213         void *data;
1214         struct delta_index *index;
1215         unsigned depth;
1216 };
1217
1218 static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
1219                            unsigned long delta_size)
1220 {
1221         if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
1222                 return 0;
1223
1224         if (delta_size < cache_max_small_delta_size)
1225                 return 1;
1226
1227         /* cache delta, if objects are large enough compared to delta size */
1228         if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10))
1229                 return 1;
1230
1231         return 0;
1232 }
1233
1234 #ifdef THREADED_DELTA_SEARCH
1235
1236 static pthread_mutex_t read_mutex = PTHREAD_MUTEX_INITIALIZER;
1237 #define read_lock()             pthread_mutex_lock(&read_mutex)
1238 #define read_unlock()           pthread_mutex_unlock(&read_mutex)
1239
1240 static pthread_mutex_t cache_mutex = PTHREAD_MUTEX_INITIALIZER;
1241 #define cache_lock()            pthread_mutex_lock(&cache_mutex)
1242 #define cache_unlock()          pthread_mutex_unlock(&cache_mutex)
1243
1244 static pthread_mutex_t progress_mutex = PTHREAD_MUTEX_INITIALIZER;
1245 #define progress_lock()         pthread_mutex_lock(&progress_mutex)
1246 #define progress_unlock()       pthread_mutex_unlock(&progress_mutex)
1247
1248 #else
1249
1250 #define read_lock()             (void)0
1251 #define read_unlock()           (void)0
1252 #define cache_lock()            (void)0
1253 #define cache_unlock()          (void)0
1254 #define progress_lock()         (void)0
1255 #define progress_unlock()       (void)0
1256
1257 #endif
1258
1259 static int try_delta(struct unpacked *trg, struct unpacked *src,
1260                      unsigned max_depth, unsigned long *mem_usage)
1261 {
1262         struct object_entry *trg_entry = trg->entry;
1263         struct object_entry *src_entry = src->entry;
1264         unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1265         unsigned ref_depth;
1266         enum object_type type;
1267         void *delta_buf;
1268
1269         /* Don't bother doing diffs between different types */
1270         if (trg_entry->type != src_entry->type)
1271                 return -1;
1272
1273         /*
1274          * We do not bother to try a delta that we discarded
1275          * on an earlier try, but only when reusing delta data.
1276          */
1277         if (reuse_delta && trg_entry->in_pack &&
1278             trg_entry->in_pack == src_entry->in_pack &&
1279             trg_entry->in_pack_type != OBJ_REF_DELTA &&
1280             trg_entry->in_pack_type != OBJ_OFS_DELTA)
1281                 return 0;
1282
1283         /* Let's not bust the allowed depth. */
1284         if (src->depth >= max_depth)
1285                 return 0;
1286
1287         /* Now some size filtering heuristics. */
1288         trg_size = trg_entry->size;
1289         if (!trg_entry->delta) {
1290                 max_size = trg_size/2 - 20;
1291                 ref_depth = 1;
1292         } else {
1293                 max_size = trg_entry->delta_size;
1294                 ref_depth = trg->depth;
1295         }
1296         max_size = (uint64_t)max_size * (max_depth - src->depth) /
1297                                                 (max_depth - ref_depth + 1);
1298         if (max_size == 0)
1299                 return 0;
1300         src_size = src_entry->size;
1301         sizediff = src_size < trg_size ? trg_size - src_size : 0;
1302         if (sizediff >= max_size)
1303                 return 0;
1304         if (trg_size < src_size / 32)
1305                 return 0;
1306
1307         /* Load data if not already done */
1308         if (!trg->data) {
1309                 read_lock();
1310                 trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
1311                 read_unlock();
1312                 if (!trg->data)
1313                         die("object %s cannot be read",
1314                             sha1_to_hex(trg_entry->idx.sha1));
1315                 if (sz != trg_size)
1316                         die("object %s inconsistent object length (%lu vs %lu)",
1317                             sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
1318                 *mem_usage += sz;
1319         }
1320         if (!src->data) {
1321                 read_lock();
1322                 src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
1323                 read_unlock();
1324                 if (!src->data)
1325                         die("object %s cannot be read",
1326                             sha1_to_hex(src_entry->idx.sha1));
1327                 if (sz != src_size)
1328                         die("object %s inconsistent object length (%lu vs %lu)",
1329                             sha1_to_hex(src_entry->idx.sha1), sz, src_size);
1330                 *mem_usage += sz;
1331         }
1332         if (!src->index) {
1333                 src->index = create_delta_index(src->data, src_size);
1334                 if (!src->index) {
1335                         static int warned = 0;
1336                         if (!warned++)
1337                                 warning("suboptimal pack - out of memory");
1338                         return 0;
1339                 }
1340                 *mem_usage += sizeof_delta_index(src->index);
1341         }
1342
1343         delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1344         if (!delta_buf)
1345                 return 0;
1346
1347         if (trg_entry->delta) {
1348                 /* Prefer only shallower same-sized deltas. */
1349                 if (delta_size == trg_entry->delta_size &&
1350                     src->depth + 1 >= trg->depth) {
1351                         free(delta_buf);
1352                         return 0;
1353                 }
1354         }
1355
1356         /*
1357          * Handle memory allocation outside of the cache
1358          * accounting lock.  Compiler will optimize the strangeness
1359          * away when THREADED_DELTA_SEARCH is not defined.
1360          */
1361         free(trg_entry->delta_data);
1362         cache_lock();
1363         if (trg_entry->delta_data) {
1364                 delta_cache_size -= trg_entry->delta_size;
1365                 trg_entry->delta_data = NULL;
1366         }
1367         if (delta_cacheable(src_size, trg_size, delta_size)) {
1368                 delta_cache_size += delta_size;
1369                 cache_unlock();
1370                 trg_entry->delta_data = xrealloc(delta_buf, delta_size);
1371         } else {
1372                 cache_unlock();
1373                 free(delta_buf);
1374         }
1375
1376         trg_entry->delta = src_entry;
1377         trg_entry->delta_size = delta_size;
1378         trg->depth = src->depth + 1;
1379
1380         return 1;
1381 }
1382
1383 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1384 {
1385         struct object_entry *child = me->delta_child;
1386         unsigned int m = n;
1387         while (child) {
1388                 unsigned int c = check_delta_limit(child, n + 1);
1389                 if (m < c)
1390                         m = c;
1391                 child = child->delta_sibling;
1392         }
1393         return m;
1394 }
1395
1396 static unsigned long free_unpacked(struct unpacked *n)
1397 {
1398         unsigned long freed_mem = sizeof_delta_index(n->index);
1399         free_delta_index(n->index);
1400         n->index = NULL;
1401         if (n->data) {
1402                 freed_mem += n->entry->size;
1403                 free(n->data);
1404                 n->data = NULL;
1405         }
1406         n->entry = NULL;
1407         n->depth = 0;
1408         return freed_mem;
1409 }
1410
1411 static void find_deltas(struct object_entry **list, unsigned *list_size,
1412                         int window, int depth, unsigned *processed)
1413 {
1414         uint32_t i, idx = 0, count = 0;
1415         struct unpacked *array;
1416         unsigned long mem_usage = 0;
1417
1418         array = xcalloc(window, sizeof(struct unpacked));
1419
1420         for (;;) {
1421                 struct object_entry *entry;
1422                 struct unpacked *n = array + idx;
1423                 int j, max_depth, best_base = -1;
1424
1425                 progress_lock();
1426                 if (!*list_size) {
1427                         progress_unlock();
1428                         break;
1429                 }
1430                 entry = *list++;
1431                 (*list_size)--;
1432                 if (!entry->preferred_base) {
1433                         (*processed)++;
1434                         display_progress(progress_state, *processed);
1435                 }
1436                 progress_unlock();
1437
1438                 mem_usage -= free_unpacked(n);
1439                 n->entry = entry;
1440
1441                 while (window_memory_limit &&
1442                        mem_usage > window_memory_limit &&
1443                        count > 1) {
1444                         uint32_t tail = (idx + window - count) % window;
1445                         mem_usage -= free_unpacked(array + tail);
1446                         count--;
1447                 }
1448
1449                 /* We do not compute delta to *create* objects we are not
1450                  * going to pack.
1451                  */
1452                 if (entry->preferred_base)
1453                         goto next;
1454
1455                 /*
1456                  * If the current object is at pack edge, take the depth the
1457                  * objects that depend on the current object into account
1458                  * otherwise they would become too deep.
1459                  */
1460                 max_depth = depth;
1461                 if (entry->delta_child) {
1462                         max_depth -= check_delta_limit(entry, 0);
1463                         if (max_depth <= 0)
1464                                 goto next;
1465                 }
1466
1467                 j = window;
1468                 while (--j > 0) {
1469                         int ret;
1470                         uint32_t other_idx = idx + j;
1471                         struct unpacked *m;
1472                         if (other_idx >= window)
1473                                 other_idx -= window;
1474                         m = array + other_idx;
1475                         if (!m->entry)
1476                                 break;
1477                         ret = try_delta(n, m, max_depth, &mem_usage);
1478                         if (ret < 0)
1479                                 break;
1480                         else if (ret > 0)
1481                                 best_base = other_idx;
1482                 }
1483
1484                 /*
1485                  * If we decided to cache the delta data, then it is best
1486                  * to compress it right away.  First because we have to do
1487                  * it anyway, and doing it here while we're threaded will
1488                  * save a lot of time in the non threaded write phase,
1489                  * as well as allow for caching more deltas within
1490                  * the same cache size limit.
1491                  * ...
1492                  * But only if not writing to stdout, since in that case
1493                  * the network is most likely throttling writes anyway,
1494                  * and therefore it is best to go to the write phase ASAP
1495                  * instead, as we can afford spending more time compressing
1496                  * between writes at that moment.
1497                  */
1498                 if (entry->delta_data && !pack_to_stdout) {
1499                         entry->z_delta_size = do_compress(&entry->delta_data,
1500                                                           entry->delta_size);
1501                         cache_lock();
1502                         delta_cache_size -= entry->delta_size;
1503                         delta_cache_size += entry->z_delta_size;
1504                         cache_unlock();
1505                 }
1506
1507                 /* if we made n a delta, and if n is already at max
1508                  * depth, leaving it in the window is pointless.  we
1509                  * should evict it first.
1510                  */
1511                 if (entry->delta && max_depth <= n->depth)
1512                         continue;
1513
1514                 /*
1515                  * Move the best delta base up in the window, after the
1516                  * currently deltified object, to keep it longer.  It will
1517                  * be the first base object to be attempted next.
1518                  */
1519                 if (entry->delta) {
1520                         struct unpacked swap = array[best_base];
1521                         int dist = (window + idx - best_base) % window;
1522                         int dst = best_base;
1523                         while (dist--) {
1524                                 int src = (dst + 1) % window;
1525                                 array[dst] = array[src];
1526                                 dst = src;
1527                         }
1528                         array[dst] = swap;
1529                 }
1530
1531                 next:
1532                 idx++;
1533                 if (count + 1 < window)
1534                         count++;
1535                 if (idx >= window)
1536                         idx = 0;
1537         }
1538
1539         for (i = 0; i < window; ++i) {
1540                 free_delta_index(array[i].index);
1541                 free(array[i].data);
1542         }
1543         free(array);
1544 }
1545
1546 #ifdef THREADED_DELTA_SEARCH
1547
1548 /*
1549  * The main thread waits on the condition that (at least) one of the workers
1550  * has stopped working (which is indicated in the .working member of
1551  * struct thread_params).
1552  * When a work thread has completed its work, it sets .working to 0 and
1553  * signals the main thread and waits on the condition that .data_ready
1554  * becomes 1.
1555  */
1556
1557 struct thread_params {
1558         pthread_t thread;
1559         struct object_entry **list;
1560         unsigned list_size;
1561         unsigned remaining;
1562         int window;
1563         int depth;
1564         int working;
1565         int data_ready;
1566         pthread_mutex_t mutex;
1567         pthread_cond_t cond;
1568         unsigned *processed;
1569 };
1570
1571 static pthread_cond_t progress_cond = PTHREAD_COND_INITIALIZER;
1572
1573 static void *threaded_find_deltas(void *arg)
1574 {
1575         struct thread_params *me = arg;
1576
1577         while (me->remaining) {
1578                 find_deltas(me->list, &me->remaining,
1579                             me->window, me->depth, me->processed);
1580
1581                 progress_lock();
1582                 me->working = 0;
1583                 pthread_cond_signal(&progress_cond);
1584                 progress_unlock();
1585
1586                 /*
1587                  * We must not set ->data_ready before we wait on the
1588                  * condition because the main thread may have set it to 1
1589                  * before we get here. In order to be sure that new
1590                  * work is available if we see 1 in ->data_ready, it
1591                  * was initialized to 0 before this thread was spawned
1592                  * and we reset it to 0 right away.
1593                  */
1594                 pthread_mutex_lock(&me->mutex);
1595                 while (!me->data_ready)
1596                         pthread_cond_wait(&me->cond, &me->mutex);
1597                 me->data_ready = 0;
1598                 pthread_mutex_unlock(&me->mutex);
1599         }
1600         /* leave ->working 1 so that this doesn't get more work assigned */
1601         return NULL;
1602 }
1603
1604 static void ll_find_deltas(struct object_entry **list, unsigned list_size,
1605                            int window, int depth, unsigned *processed)
1606 {
1607         struct thread_params p[delta_search_threads];
1608         int i, ret, active_threads = 0;
1609
1610         if (delta_search_threads <= 1) {
1611                 find_deltas(list, &list_size, window, depth, processed);
1612                 return;
1613         }
1614         if (progress > pack_to_stdout)
1615                 fprintf(stderr, "Delta compression using up to %d threads.\n",
1616                                 delta_search_threads);
1617
1618         /* Partition the work amongst work threads. */
1619         for (i = 0; i < delta_search_threads; i++) {
1620                 unsigned sub_size = list_size / (delta_search_threads - i);
1621
1622                 /* don't use too small segments or no deltas will be found */
1623                 if (sub_size < 2*window && i+1 < delta_search_threads)
1624                         sub_size = 0;
1625
1626                 p[i].window = window;
1627                 p[i].depth = depth;
1628                 p[i].processed = processed;
1629                 p[i].working = 1;
1630                 p[i].data_ready = 0;
1631
1632                 /* try to split chunks on "path" boundaries */
1633                 while (sub_size && sub_size < list_size &&
1634                        list[sub_size]->hash &&
1635                        list[sub_size]->hash == list[sub_size-1]->hash)
1636                         sub_size++;
1637
1638                 p[i].list = list;
1639                 p[i].list_size = sub_size;
1640                 p[i].remaining = sub_size;
1641
1642                 list += sub_size;
1643                 list_size -= sub_size;
1644         }
1645
1646         /* Start work threads. */
1647         for (i = 0; i < delta_search_threads; i++) {
1648                 if (!p[i].list_size)
1649                         continue;
1650                 pthread_mutex_init(&p[i].mutex, NULL);
1651                 pthread_cond_init(&p[i].cond, NULL);
1652                 ret = pthread_create(&p[i].thread, NULL,
1653                                      threaded_find_deltas, &p[i]);
1654                 if (ret)
1655                         die("unable to create thread: %s", strerror(ret));
1656                 active_threads++;
1657         }
1658
1659         /*
1660          * Now let's wait for work completion.  Each time a thread is done
1661          * with its work, we steal half of the remaining work from the
1662          * thread with the largest number of unprocessed objects and give
1663          * it to that newly idle thread.  This ensure good load balancing
1664          * until the remaining object list segments are simply too short
1665          * to be worth splitting anymore.
1666          */
1667         while (active_threads) {
1668                 struct thread_params *target = NULL;
1669                 struct thread_params *victim = NULL;
1670                 unsigned sub_size = 0;
1671
1672                 progress_lock();
1673                 for (;;) {
1674                         for (i = 0; !target && i < delta_search_threads; i++)
1675                                 if (!p[i].working)
1676                                         target = &p[i];
1677                         if (target)
1678                                 break;
1679                         pthread_cond_wait(&progress_cond, &progress_mutex);
1680                 }
1681
1682                 for (i = 0; i < delta_search_threads; i++)
1683                         if (p[i].remaining > 2*window &&
1684                             (!victim || victim->remaining < p[i].remaining))
1685                                 victim = &p[i];
1686                 if (victim) {
1687                         sub_size = victim->remaining / 2;
1688                         list = victim->list + victim->list_size - sub_size;
1689                         while (sub_size && list[0]->hash &&
1690                                list[0]->hash == list[-1]->hash) {
1691                                 list++;
1692                                 sub_size--;
1693                         }
1694                         if (!sub_size) {
1695                                 /*
1696                                  * It is possible for some "paths" to have
1697                                  * so many objects that no hash boundary
1698                                  * might be found.  Let's just steal the
1699                                  * exact half in that case.
1700                                  */
1701                                 sub_size = victim->remaining / 2;
1702                                 list -= sub_size;
1703                         }
1704                         target->list = list;
1705                         victim->list_size -= sub_size;
1706                         victim->remaining -= sub_size;
1707                 }
1708                 target->list_size = sub_size;
1709                 target->remaining = sub_size;
1710                 target->working = 1;
1711                 progress_unlock();
1712
1713                 pthread_mutex_lock(&target->mutex);
1714                 target->data_ready = 1;
1715                 pthread_cond_signal(&target->cond);
1716                 pthread_mutex_unlock(&target->mutex);
1717
1718                 if (!sub_size) {
1719                         pthread_join(target->thread, NULL);
1720                         pthread_cond_destroy(&target->cond);
1721                         pthread_mutex_destroy(&target->mutex);
1722                         active_threads--;
1723                 }
1724         }
1725 }
1726
1727 #else
1728 #define ll_find_deltas(l, s, w, d, p)   find_deltas(l, &s, w, d, p)
1729 #endif
1730
1731 static int add_ref_tag(const char *path, const unsigned char *sha1, int flag, void *cb_data)
1732 {
1733         unsigned char peeled[20];
1734
1735         if (!prefixcmp(path, "refs/tags/") && /* is a tag? */
1736             !peel_ref(path, peeled)        && /* peelable? */
1737             !is_null_sha1(peeled)          && /* annotated tag? */
1738             locate_object_entry(peeled))      /* object packed? */
1739                 add_object_entry(sha1, OBJ_TAG, NULL, 0);
1740         return 0;
1741 }
1742
1743 static void prepare_pack(int window, int depth)
1744 {
1745         struct object_entry **delta_list;
1746         uint32_t i, nr_deltas;
1747         unsigned n;
1748
1749         get_object_details();
1750
1751         /*
1752          * If we're locally repacking then we need to be doubly careful
1753          * from now on in order to make sure no stealth corruption gets
1754          * propagated to the new pack.  Clients receiving streamed packs
1755          * should validate everything they get anyway so no need to incur
1756          * the additional cost here in that case.
1757          */
1758         if (!pack_to_stdout)
1759                 do_check_packed_object_crc = 1;
1760
1761         if (!nr_objects || !window || !depth)
1762                 return;
1763
1764         delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1765         nr_deltas = n = 0;
1766
1767         for (i = 0; i < nr_objects; i++) {
1768                 struct object_entry *entry = objects + i;
1769
1770                 if (entry->delta)
1771                         /* This happens if we decided to reuse existing
1772                          * delta from a pack.  "reuse_delta &&" is implied.
1773                          */
1774                         continue;
1775
1776                 if (entry->size < 50)
1777                         continue;
1778
1779                 if (entry->no_try_delta)
1780                         continue;
1781
1782                 if (!entry->preferred_base) {
1783                         nr_deltas++;
1784                         if (entry->type < 0)
1785                                 die("unable to get type of object %s",
1786                                     sha1_to_hex(entry->idx.sha1));
1787                 } else {
1788                         if (entry->type < 0) {
1789                                 /*
1790                                  * This object is not found, but we
1791                                  * don't have to include it anyway.
1792                                  */
1793                                 continue;
1794                         }
1795                 }
1796
1797                 delta_list[n++] = entry;
1798         }
1799
1800         if (nr_deltas && n > 1) {
1801                 unsigned nr_done = 0;
1802                 if (progress)
1803                         progress_state = start_progress("Compressing objects",
1804                                                         nr_deltas);
1805                 qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
1806                 ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
1807                 stop_progress(&progress_state);
1808                 if (nr_done != nr_deltas)
1809                         die("inconsistency with delta count");
1810         }
1811         free(delta_list);
1812 }
1813
1814 static int git_pack_config(const char *k, const char *v, void *cb)
1815 {
1816         if(!strcmp(k, "pack.window")) {
1817                 window = git_config_int(k, v);
1818                 return 0;
1819         }
1820         if (!strcmp(k, "pack.windowmemory")) {
1821                 window_memory_limit = git_config_ulong(k, v);
1822                 return 0;
1823         }
1824         if (!strcmp(k, "pack.depth")) {
1825                 depth = git_config_int(k, v);
1826                 return 0;
1827         }
1828         if (!strcmp(k, "pack.compression")) {
1829                 int level = git_config_int(k, v);
1830                 if (level == -1)
1831                         level = Z_DEFAULT_COMPRESSION;
1832                 else if (level < 0 || level > Z_BEST_COMPRESSION)
1833                         die("bad pack compression level %d", level);
1834                 pack_compression_level = level;
1835                 pack_compression_seen = 1;
1836                 return 0;
1837         }
1838         if (!strcmp(k, "pack.deltacachesize")) {
1839                 max_delta_cache_size = git_config_int(k, v);
1840                 return 0;
1841         }
1842         if (!strcmp(k, "pack.deltacachelimit")) {
1843                 cache_max_small_delta_size = git_config_int(k, v);
1844                 return 0;
1845         }
1846         if (!strcmp(k, "pack.threads")) {
1847                 delta_search_threads = git_config_int(k, v);
1848                 if (delta_search_threads < 0)
1849                         die("invalid number of threads specified (%d)",
1850                             delta_search_threads);
1851 #ifndef THREADED_DELTA_SEARCH
1852                 if (delta_search_threads != 1)
1853                         warning("no threads support, ignoring %s", k);
1854 #endif
1855                 return 0;
1856         }
1857         if (!strcmp(k, "pack.indexversion")) {
1858                 pack_idx_default_version = git_config_int(k, v);
1859                 if (pack_idx_default_version > 2)
1860                         die("bad pack.indexversion=%"PRIu32,
1861                                 pack_idx_default_version);
1862                 return 0;
1863         }
1864         if (!strcmp(k, "pack.packsizelimit")) {
1865                 pack_size_limit_cfg = git_config_ulong(k, v);
1866                 return 0;
1867         }
1868         return git_default_config(k, v, cb);
1869 }
1870
1871 static void read_object_list_from_stdin(void)
1872 {
1873         char line[40 + 1 + PATH_MAX + 2];
1874         unsigned char sha1[20];
1875
1876         for (;;) {
1877                 if (!fgets(line, sizeof(line), stdin)) {
1878                         if (feof(stdin))
1879                                 break;
1880                         if (!ferror(stdin))
1881                                 die("fgets returned NULL, not EOF, not error!");
1882                         if (errno != EINTR)
1883                                 die("fgets: %s", strerror(errno));
1884                         clearerr(stdin);
1885                         continue;
1886                 }
1887                 if (line[0] == '-') {
1888                         if (get_sha1_hex(line+1, sha1))
1889                                 die("expected edge sha1, got garbage:\n %s",
1890                                     line);
1891                         add_preferred_base(sha1);
1892                         continue;
1893                 }
1894                 if (get_sha1_hex(line, sha1))
1895                         die("expected sha1, got garbage:\n %s", line);
1896
1897                 add_preferred_base_object(line+41);
1898                 add_object_entry(sha1, 0, line+41, 0);
1899         }
1900 }
1901
1902 #define OBJECT_ADDED (1u<<20)
1903
1904 static void show_commit(struct commit *commit, void *data)
1905 {
1906         add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
1907         commit->object.flags |= OBJECT_ADDED;
1908 }
1909
1910 static void show_object(struct object *obj, const struct name_path *path, const char *last)
1911 {
1912         char *name = path_name(path, last);
1913
1914         add_preferred_base_object(name);
1915         add_object_entry(obj->sha1, obj->type, name, 0);
1916         obj->flags |= OBJECT_ADDED;
1917
1918         /*
1919          * We will have generated the hash from the name,
1920          * but not saved a pointer to it - we can free it
1921          */
1922         free((char *)name);
1923 }
1924
1925 static void show_edge(struct commit *commit)
1926 {
1927         add_preferred_base(commit->object.sha1);
1928 }
1929
1930 struct in_pack_object {
1931         off_t offset;
1932         struct object *object;
1933 };
1934
1935 struct in_pack {
1936         int alloc;
1937         int nr;
1938         struct in_pack_object *array;
1939 };
1940
1941 static void mark_in_pack_object(struct object *object, struct packed_git *p, struct in_pack *in_pack)
1942 {
1943         in_pack->array[in_pack->nr].offset = find_pack_entry_one(object->sha1, p);
1944         in_pack->array[in_pack->nr].object = object;
1945         in_pack->nr++;
1946 }
1947
1948 /*
1949  * Compare the objects in the offset order, in order to emulate the
1950  * "git rev-list --objects" output that produced the pack originally.
1951  */
1952 static int ofscmp(const void *a_, const void *b_)
1953 {
1954         struct in_pack_object *a = (struct in_pack_object *)a_;
1955         struct in_pack_object *b = (struct in_pack_object *)b_;
1956
1957         if (a->offset < b->offset)
1958                 return -1;
1959         else if (a->offset > b->offset)
1960                 return 1;
1961         else
1962                 return hashcmp(a->object->sha1, b->object->sha1);
1963 }
1964
1965 static void add_objects_in_unpacked_packs(struct rev_info *revs)
1966 {
1967         struct packed_git *p;
1968         struct in_pack in_pack;
1969         uint32_t i;
1970
1971         memset(&in_pack, 0, sizeof(in_pack));
1972
1973         for (p = packed_git; p; p = p->next) {
1974                 const unsigned char *sha1;
1975                 struct object *o;
1976
1977                 if (!p->pack_local || p->pack_keep)
1978                         continue;
1979                 if (open_pack_index(p))
1980                         die("cannot open pack index");
1981
1982                 ALLOC_GROW(in_pack.array,
1983                            in_pack.nr + p->num_objects,
1984                            in_pack.alloc);
1985
1986                 for (i = 0; i < p->num_objects; i++) {
1987                         sha1 = nth_packed_object_sha1(p, i);
1988                         o = lookup_unknown_object(sha1);
1989                         if (!(o->flags & OBJECT_ADDED))
1990                                 mark_in_pack_object(o, p, &in_pack);
1991                         o->flags |= OBJECT_ADDED;
1992                 }
1993         }
1994
1995         if (in_pack.nr) {
1996                 qsort(in_pack.array, in_pack.nr, sizeof(in_pack.array[0]),
1997                       ofscmp);
1998                 for (i = 0; i < in_pack.nr; i++) {
1999                         struct object *o = in_pack.array[i].object;
2000                         add_object_entry(o->sha1, o->type, "", 0);
2001                 }
2002         }
2003         free(in_pack.array);
2004 }
2005
2006 static int has_sha1_pack_kept_or_nonlocal(const unsigned char *sha1)
2007 {
2008         static struct packed_git *last_found = (void *)1;
2009         struct packed_git *p;
2010
2011         p = (last_found != (void *)1) ? last_found : packed_git;
2012
2013         while (p) {
2014                 if ((!p->pack_local || p->pack_keep) &&
2015                         find_pack_entry_one(sha1, p)) {
2016                         last_found = p;
2017                         return 1;
2018                 }
2019                 if (p == last_found)
2020                         p = packed_git;
2021                 else
2022                         p = p->next;
2023                 if (p == last_found)
2024                         p = p->next;
2025         }
2026         return 0;
2027 }
2028
2029 static void loosen_unused_packed_objects(struct rev_info *revs)
2030 {
2031         struct packed_git *p;
2032         uint32_t i;
2033         const unsigned char *sha1;
2034
2035         for (p = packed_git; p; p = p->next) {
2036                 if (!p->pack_local || p->pack_keep)
2037                         continue;
2038
2039                 if (open_pack_index(p))
2040                         die("cannot open pack index");
2041
2042                 for (i = 0; i < p->num_objects; i++) {
2043                         sha1 = nth_packed_object_sha1(p, i);
2044                         if (!locate_object_entry(sha1) &&
2045                                 !has_sha1_pack_kept_or_nonlocal(sha1))
2046                                 if (force_object_loose(sha1, p->mtime))
2047                                         die("unable to force loose object");
2048                 }
2049         }
2050 }
2051
2052 static void get_object_list(int ac, const char **av)
2053 {
2054         struct rev_info revs;
2055         char line[1000];
2056         int flags = 0;
2057
2058         init_revisions(&revs, NULL);
2059         save_commit_buffer = 0;
2060         setup_revisions(ac, av, &revs, NULL);
2061
2062         while (fgets(line, sizeof(line), stdin) != NULL) {
2063                 int len = strlen(line);
2064                 if (len && line[len - 1] == '\n')
2065                         line[--len] = 0;
2066                 if (!len)
2067                         break;
2068                 if (*line == '-') {
2069                         if (!strcmp(line, "--not")) {
2070                                 flags ^= UNINTERESTING;
2071                                 continue;
2072                         }
2073                         die("not a rev '%s'", line);
2074                 }
2075                 if (handle_revision_arg(line, &revs, flags, 1))
2076                         die("bad revision '%s'", line);
2077         }
2078
2079         if (prepare_revision_walk(&revs))
2080                 die("revision walk setup failed");
2081         mark_edges_uninteresting(revs.commits, &revs, show_edge);
2082         traverse_commit_list(&revs, show_commit, show_object, NULL);
2083
2084         if (keep_unreachable)
2085                 add_objects_in_unpacked_packs(&revs);
2086         if (unpack_unreachable)
2087                 loosen_unused_packed_objects(&revs);
2088 }
2089
2090 static int adjust_perm(const char *path, mode_t mode)
2091 {
2092         if (chmod(path, mode))
2093                 return -1;
2094         return adjust_shared_perm(path);
2095 }
2096
2097 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
2098 {
2099         int use_internal_rev_list = 0;
2100         int thin = 0;
2101         uint32_t i;
2102         const char **rp_av;
2103         int rp_ac_alloc = 64;
2104         int rp_ac;
2105
2106         rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
2107
2108         rp_av[0] = "pack-objects";
2109         rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
2110         rp_ac = 2;
2111
2112         git_config(git_pack_config, NULL);
2113         if (!pack_compression_seen && core_compression_seen)
2114                 pack_compression_level = core_compression_level;
2115
2116         progress = isatty(2);
2117         for (i = 1; i < argc; i++) {
2118                 const char *arg = argv[i];
2119
2120                 if (*arg != '-')
2121                         break;
2122
2123                 if (!strcmp("--non-empty", arg)) {
2124                         non_empty = 1;
2125                         continue;
2126                 }
2127                 if (!strcmp("--local", arg)) {
2128                         local = 1;
2129                         continue;
2130                 }
2131                 if (!strcmp("--incremental", arg)) {
2132                         incremental = 1;
2133                         continue;
2134                 }
2135                 if (!strcmp("--honor-pack-keep", arg)) {
2136                         ignore_packed_keep = 1;
2137                         continue;
2138                 }
2139                 if (!prefixcmp(arg, "--compression=")) {
2140                         char *end;
2141                         int level = strtoul(arg+14, &end, 0);
2142                         if (!arg[14] || *end)
2143                                 usage(pack_usage);
2144                         if (level == -1)
2145                                 level = Z_DEFAULT_COMPRESSION;
2146                         else if (level < 0 || level > Z_BEST_COMPRESSION)
2147                                 die("bad pack compression level %d", level);
2148                         pack_compression_level = level;
2149                         continue;
2150                 }
2151                 if (!prefixcmp(arg, "--max-pack-size=")) {
2152                         char *end;
2153                         pack_size_limit_cfg = 0;
2154                         pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024;
2155                         if (!arg[16] || *end)
2156                                 usage(pack_usage);
2157                         continue;
2158                 }
2159                 if (!prefixcmp(arg, "--window=")) {
2160                         char *end;
2161                         window = strtoul(arg+9, &end, 0);
2162                         if (!arg[9] || *end)
2163                                 usage(pack_usage);
2164                         continue;
2165                 }
2166                 if (!prefixcmp(arg, "--window-memory=")) {
2167                         if (!git_parse_ulong(arg+16, &window_memory_limit))
2168                                 usage(pack_usage);
2169                         continue;
2170                 }
2171                 if (!prefixcmp(arg, "--threads=")) {
2172                         char *end;
2173                         delta_search_threads = strtoul(arg+10, &end, 0);
2174                         if (!arg[10] || *end || delta_search_threads < 0)
2175                                 usage(pack_usage);
2176 #ifndef THREADED_DELTA_SEARCH
2177                         if (delta_search_threads != 1)
2178                                 warning("no threads support, "
2179                                         "ignoring %s", arg);
2180 #endif
2181                         continue;
2182                 }
2183                 if (!prefixcmp(arg, "--depth=")) {
2184                         char *end;
2185                         depth = strtoul(arg+8, &end, 0);
2186                         if (!arg[8] || *end)
2187                                 usage(pack_usage);
2188                         continue;
2189                 }
2190                 if (!strcmp("--progress", arg)) {
2191                         progress = 1;
2192                         continue;
2193                 }
2194                 if (!strcmp("--all-progress", arg)) {
2195                         progress = 2;
2196                         continue;
2197                 }
2198                 if (!strcmp("-q", arg)) {
2199                         progress = 0;
2200                         continue;
2201                 }
2202                 if (!strcmp("--no-reuse-delta", arg)) {
2203                         reuse_delta = 0;
2204                         continue;
2205                 }
2206                 if (!strcmp("--no-reuse-object", arg)) {
2207                         reuse_object = reuse_delta = 0;
2208                         continue;
2209                 }
2210                 if (!strcmp("--delta-base-offset", arg)) {
2211                         allow_ofs_delta = 1;
2212                         continue;
2213                 }
2214                 if (!strcmp("--stdout", arg)) {
2215                         pack_to_stdout = 1;
2216                         continue;
2217                 }
2218                 if (!strcmp("--revs", arg)) {
2219                         use_internal_rev_list = 1;
2220                         continue;
2221                 }
2222                 if (!strcmp("--keep-unreachable", arg)) {
2223                         keep_unreachable = 1;
2224                         continue;
2225                 }
2226                 if (!strcmp("--unpack-unreachable", arg)) {
2227                         unpack_unreachable = 1;
2228                         continue;
2229                 }
2230                 if (!strcmp("--include-tag", arg)) {
2231                         include_tag = 1;
2232                         continue;
2233                 }
2234                 if (!strcmp("--unpacked", arg) ||
2235                     !strcmp("--reflog", arg) ||
2236                     !strcmp("--all", arg)) {
2237                         use_internal_rev_list = 1;
2238                         if (rp_ac >= rp_ac_alloc - 1) {
2239                                 rp_ac_alloc = alloc_nr(rp_ac_alloc);
2240                                 rp_av = xrealloc(rp_av,
2241                                                  rp_ac_alloc * sizeof(*rp_av));
2242                         }
2243                         rp_av[rp_ac++] = arg;
2244                         continue;
2245                 }
2246                 if (!strcmp("--thin", arg)) {
2247                         use_internal_rev_list = 1;
2248                         thin = 1;
2249                         rp_av[1] = "--objects-edge";
2250                         continue;
2251                 }
2252                 if (!prefixcmp(arg, "--index-version=")) {
2253                         char *c;
2254                         pack_idx_default_version = strtoul(arg + 16, &c, 10);
2255                         if (pack_idx_default_version > 2)
2256                                 die("bad %s", arg);
2257                         if (*c == ',')
2258                                 pack_idx_off32_limit = strtoul(c+1, &c, 0);
2259                         if (*c || pack_idx_off32_limit & 0x80000000)
2260                                 die("bad %s", arg);
2261                         continue;
2262                 }
2263                 usage(pack_usage);
2264         }
2265
2266         /* Traditionally "pack-objects [options] base extra" failed;
2267          * we would however want to take refs parameter that would
2268          * have been given to upstream rev-list ourselves, which means
2269          * we somehow want to say what the base name is.  So the
2270          * syntax would be:
2271          *
2272          * pack-objects [options] base <refs...>
2273          *
2274          * in other words, we would treat the first non-option as the
2275          * base_name and send everything else to the internal revision
2276          * walker.
2277          */
2278
2279         if (!pack_to_stdout)
2280                 base_name = argv[i++];
2281
2282         if (pack_to_stdout != !base_name)
2283                 usage(pack_usage);
2284
2285         if (!pack_to_stdout && !pack_size_limit)
2286                 pack_size_limit = pack_size_limit_cfg;
2287
2288         if (pack_to_stdout && pack_size_limit)
2289                 die("--max-pack-size cannot be used to build a pack for transfer.");
2290
2291         if (!pack_to_stdout && thin)
2292                 die("--thin cannot be used to build an indexable pack.");
2293
2294         if (keep_unreachable && unpack_unreachable)
2295                 die("--keep-unreachable and --unpack-unreachable are incompatible.");
2296
2297 #ifdef THREADED_DELTA_SEARCH
2298         if (!delta_search_threads)      /* --threads=0 means autodetect */
2299                 delta_search_threads = online_cpus();
2300 #endif
2301
2302         prepare_packed_git();
2303
2304         if (progress)
2305                 progress_state = start_progress("Counting objects", 0);
2306         if (!use_internal_rev_list)
2307                 read_object_list_from_stdin();
2308         else {
2309                 rp_av[rp_ac] = NULL;
2310                 get_object_list(rp_ac, rp_av);
2311         }
2312         if (include_tag && nr_result)
2313                 for_each_ref(add_ref_tag, NULL);
2314         stop_progress(&progress_state);
2315
2316         if (non_empty && !nr_result)
2317                 return 0;
2318         if (nr_result)
2319                 prepare_pack(window, depth);
2320         write_pack_file();
2321         if (progress)
2322                 fprintf(stderr, "Total %"PRIu32" (delta %"PRIu32"),"
2323                         " reused %"PRIu32" (delta %"PRIu32")\n",
2324                         written, written_delta, reused, reused_delta);
2325         return 0;
2326 }