builtin-pack-objects.c

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