builtin-pack-objects.c

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