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