pack-objects.c

   1 #include "cache.h"
   2 #include "object.h"
   3 #include "blob.h"
   4 #include "commit.h"
   5 #include "tag.h"
   6 #include "tree.h"
   7 #include "delta.h"
   8 #include "pack.h"
   9 #include "csum-file.h"
  10 #include "tree-walk.h"
  11 #include <sys/time.h>
  12 #include <signal.h>
  13 #include <stdint.h>
  14
  15 static const char pack_usage[] = "git-pack-objects [-q] [--no-reuse-delta] [--non-empty] [--local] [--incremental] [--window=N] [--depth=N] {--stdout | base-name} < object-list";
  16
  17 struct object_entry {
  18         unsigned char sha1[20];
  19         unsigned long size;     /* uncompressed size */
  20         unsigned long offset;   /* offset into the final pack file;
  21                                  * nonzero if already written.
  22                                  */
  23         unsigned int depth;     /* delta depth */
  24         unsigned int delta_limit;       /* base adjustment for in-pack delta */
  25         unsigned int hash;      /* name hint hash */
  26         enum object_type type;
  27         enum object_type in_pack_type;  /* could be delta */
  28         unsigned long delta_size;       /* delta data size (uncompressed) */
  29         struct object_entry *delta;     /* delta base object */
  30         struct packed_git *in_pack;     /* already in pack */
  31         unsigned int in_pack_offset;
  32         struct object_entry *delta_child; /* delitified objects who bases me */
  33         struct object_entry *delta_sibling; /* other deltified objects who
  34                                              * uses the same base as me
  35                                              */
  36         int preferred_base;     /* we do not pack this, but is encouraged to
  37                                  * be used as the base objectto delta huge
  38                                  * objects against.
  39                                  */
  40 };
  41
  42 /*
  43  * Objects we are going to pack are colected in objects array (dynamically
  44  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  45  * in the order we see -- typically rev-list --objects order that gives us
  46  * nice "minimum seek" order.
  47  *
  48  * sorted-by-sha ans sorted-by-type are arrays of pointers that point at
  49  * elements in the objects array.  The former is used to build the pack
  50  * index (lists object names in the ascending order to help offset lookup),
  51  * and the latter is used to group similar things together by try_delta()
  52  * heuristics.
  53  */
  54
  55 static unsigned char object_list_sha1[20];
  56 static int non_empty = 0;
  57 static int no_reuse_delta = 0;
  58 static int local = 0;
  59 static int incremental = 0;
  60 static struct object_entry **sorted_by_sha, **sorted_by_type;
  61 static struct object_entry *objects = NULL;
  62 static int nr_objects = 0, nr_alloc = 0, nr_result = 0;
  63 static const char *base_name;
  64 static unsigned char pack_file_sha1[20];
  65 static int progress = 1;
  66 static volatile sig_atomic_t progress_update = 0;
  67
  68 /*
  69  * The object names in objects array are hashed with this hashtable,
  70  * to help looking up the entry by object name.  Binary search from
  71  * sorted_by_sha is also possible but this was easier to code and faster.
  72  * This hashtable is built after all the objects are seen.
  73  */
  74 static int *object_ix = NULL;
  75 static int object_ix_hashsz = 0;
  76
  77 /*
  78  * Pack index for existing packs give us easy access to the offsets into
  79  * corresponding pack file where each object's data starts, but the entries
  80  * do not store the size of the compressed representation (uncompressed
  81  * size is easily available by examining the pack entry header).  We build
  82  * a hashtable of existing packs (pack_revindex), and keep reverse index
  83  * here -- pack index file is sorted by object name mapping to offset; this
  84  * pack_revindex[].revindex array is an ordered list of offsets, so if you
  85  * know the offset of an object, next offset is where its packed
  86  * representation ends.
  87  */
  88 struct pack_revindex {
  89         struct packed_git *p;
  90         unsigned long *revindex;
  91 } *pack_revindex = NULL;
  92 static int pack_revindex_hashsz = 0;
  93
  94 /*
  95  * stats
  96  */
  97 static int written = 0;
  98 static int written_delta = 0;
  99 static int reused = 0;
 100 static int reused_delta = 0;
 101
 102 static int pack_revindex_ix(struct packed_git *p)
 103 {
 104         unsigned long ui = (unsigned long)p;
 105         int i;
 106
 107         ui = ui ^ (ui >> 16); /* defeat structure alignment */
 108         i = (int)(ui % pack_revindex_hashsz);
 109         while (pack_revindex[i].p) {
 110                 if (pack_revindex[i].p == p)
 111                         return i;
 112                 if (++i == pack_revindex_hashsz)
 113                         i = 0;
 114         }
 115         return -1 - i;
 116 }
 117
 118 static void prepare_pack_ix(void)
 119 {
 120         int num;
 121         struct packed_git *p;
 122         for (num = 0, p = packed_git; p; p = p->next)
 123                 num++;
 124         if (!num)
 125                 return;
 126         pack_revindex_hashsz = num * 11;
 127         pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
 128         for (p = packed_git; p; p = p->next) {
 129                 num = pack_revindex_ix(p);
 130                 num = - 1 - num;
 131                 pack_revindex[num].p = p;
 132         }
 133         /* revindex elements are lazily initialized */
 134 }
 135
 136 static int cmp_offset(const void *a_, const void *b_)
 137 {
 138         unsigned long a = *(unsigned long *) a_;
 139         unsigned long b = *(unsigned long *) b_;
 140         if (a < b)
 141                 return -1;
 142         else if (a == b)
 143                 return 0;
 144         else
 145                 return 1;
 146 }
 147
 148 /*
 149  * Ordered list of offsets of objects in the pack.
 150  */
 151 static void prepare_pack_revindex(struct pack_revindex *rix)
 152 {
 153         struct packed_git *p = rix->p;
 154         int num_ent = num_packed_objects(p);
 155         int i;
 156         void *index = p->index_base + 256;
 157
 158         rix->revindex = xmalloc(sizeof(unsigned long) * (num_ent + 1));
 159         for (i = 0; i < num_ent; i++) {
 160                 uint32_t hl = *((uint32_t *)(index + 24 * i));
 161                 rix->revindex[i] = ntohl(hl);
 162         }
 163         /* This knows the pack format -- the 20-byte trailer
 164          * follows immediately after the last object data.
 165          */
 166         rix->revindex[num_ent] = p->pack_size - 20;
 167         qsort(rix->revindex, num_ent, sizeof(unsigned long), cmp_offset);
 168 }
 169
 170 static unsigned long find_packed_object_size(struct packed_git *p,
 171                                              unsigned long ofs)
 172 {
 173         int num;
 174         int lo, hi;
 175         struct pack_revindex *rix;
 176         unsigned long *revindex;
 177         num = pack_revindex_ix(p);
 178         if (num < 0)
 179                 die("internal error: pack revindex uninitialized");
 180         rix = &pack_revindex[num];
 181         if (!rix->revindex)
 182                 prepare_pack_revindex(rix);
 183         revindex = rix->revindex;
 184         lo = 0;
 185         hi = num_packed_objects(p) + 1;
 186         do {
 187                 int mi = (lo + hi) / 2;
 188                 if (revindex[mi] == ofs) {
 189                         return revindex[mi+1] - ofs;
 190                 }
 191                 else if (ofs < revindex[mi])
 192                         hi = mi;
 193                 else
 194                         lo = mi + 1;
 195         } while (lo < hi);
 196         die("internal error: pack revindex corrupt");
 197 }
 198
 199 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 200 {
 201         unsigned long othersize, delta_size;
 202         char type[10];
 203         void *otherbuf = read_sha1_file(entry->delta->sha1, type, &othersize);
 204         void *delta_buf;
 205
 206         if (!otherbuf)
 207                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
 208         delta_buf = diff_delta(otherbuf, othersize,
 209                                buf, size, &delta_size, 0);
 210         if (!delta_buf || delta_size != entry->delta_size)
 211                 die("delta size changed");
 212         free(buf);
 213         free(otherbuf);
 214         return delta_buf;
 215 }
 216
 217 /*
 218  * The per-object header is a pretty dense thing, which is
 219  *  - first byte: low four bits are "size", then three bits of "type",
 220  *    and the high bit is "size continues".
 221  *  - each byte afterwards: low seven bits are size continuation,
 222  *    with the high bit being "size continues"
 223  */
 224 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 225 {
 226         int n = 1;
 227         unsigned char c;
 228
 229         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 230                 die("bad type %d", type);
 231
 232         c = (type << 4) | (size & 15);
 233         size >>= 4;
 234         while (size) {
 235                 *hdr++ = c | 0x80;
 236                 c = size & 0x7f;
 237                 size >>= 7;
 238                 n++;
 239         }
 240         *hdr = c;
 241         return n;
 242 }
 243
 244 static unsigned long write_object(struct sha1file *f,
 245                                   struct object_entry *entry)
 246 {
 247         unsigned long size;
 248         char type[10];
 249         void *buf;
 250         unsigned char header[10];
 251         unsigned hdrlen, datalen;
 252         enum object_type obj_type;
 253         int to_reuse = 0;
 254
 255         if (entry->preferred_base)
 256                 return 0;
 257
 258         obj_type = entry->type;
 259         if (! entry->in_pack)
 260                 to_reuse = 0;   /* can't reuse what we don't have */
 261         else if (obj_type == OBJ_DELTA)
 262                 to_reuse = 1;   /* check_object() decided it for us */
 263         else if (obj_type != entry->in_pack_type)
 264                 to_reuse = 0;   /* pack has delta which is unusable */
 265         else if (entry->delta)
 266                 to_reuse = 0;   /* we want to pack afresh */
 267         else
 268                 to_reuse = 1;   /* we have it in-pack undeltified,
 269                                  * and we do not need to deltify it.
 270                                  */
 271
 272         if (! to_reuse) {
 273                 buf = read_sha1_file(entry->sha1, type, &size);
 274                 if (!buf)
 275                         die("unable to read %s", sha1_to_hex(entry->sha1));
 276                 if (size != entry->size)
 277                         die("object %s size inconsistency (%lu vs %lu)",
 278                             sha1_to_hex(entry->sha1), size, entry->size);
 279                 if (entry->delta) {
 280                         buf = delta_against(buf, size, entry);
 281                         size = entry->delta_size;
 282                         obj_type = OBJ_DELTA;
 283                 }
 284                 /*
 285                  * The object header is a byte of 'type' followed by zero or
 286                  * more bytes of length.  For deltas, the 20 bytes of delta
 287                  * sha1 follows that.
 288                  */
 289                 hdrlen = encode_header(obj_type, size, header);
 290                 sha1write(f, header, hdrlen);
 291
 292                 if (entry->delta) {
 293                         sha1write(f, entry->delta, 20);
 294                         hdrlen += 20;
 295                 }
 296                 datalen = sha1write_compressed(f, buf, size);
 297                 free(buf);
 298         }
 299         else {
 300                 struct packed_git *p = entry->in_pack;
 301                 use_packed_git(p);
 302
 303                 datalen = find_packed_object_size(p, entry->in_pack_offset);
 304                 buf = p->pack_base + entry->in_pack_offset;
 305                 sha1write(f, buf, datalen);
 306                 unuse_packed_git(p);
 307                 hdrlen = 0; /* not really */
 308                 if (obj_type == OBJ_DELTA)
 309                         reused_delta++;
 310                 reused++;
 311         }
 312         if (obj_type == OBJ_DELTA)
 313                 written_delta++;
 314         written++;
 315         return hdrlen + datalen;
 316 }
 317
 318 static unsigned long write_one(struct sha1file *f,
 319                                struct object_entry *e,
 320                                unsigned long offset)
 321 {
 322         if (e->offset)
 323                 /* offset starts from header size and cannot be zero
 324                  * if it is written already.
 325                  */
 326                 return offset;
 327         e->offset = offset;
 328         offset += write_object(f, e);
 329         /* if we are deltified, write out its base object. */
 330         if (e->delta)
 331                 offset = write_one(f, e->delta, offset);
 332         return offset;
 333 }
 334
 335 static void write_pack_file(void)
 336 {
 337         int i;
 338         struct sha1file *f;
 339         unsigned long offset;
 340         struct pack_header hdr;
 341         unsigned last_percent = 999;
 342         int do_progress = 0;
 343
 344         if (!base_name)
 345                 f = sha1fd(1, "<stdout>");
 346         else {
 347                 f = sha1create("%s-%s.%s", base_name,
 348                                sha1_to_hex(object_list_sha1), "pack");
 349                 do_progress = progress;
 350         }
 351         if (do_progress)
 352                 fprintf(stderr, "Writing %d objects.\n", nr_result);
 353
 354         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 355         hdr.hdr_version = htonl(PACK_VERSION);
 356         hdr.hdr_entries = htonl(nr_result);
 357         sha1write(f, &hdr, sizeof(hdr));
 358         offset = sizeof(hdr);
 359         if (!nr_result)
 360                 goto done;
 361         for (i = 0; i < nr_objects; i++) {
 362                 offset = write_one(f, objects + i, offset);
 363                 if (do_progress) {
 364                         unsigned percent = written * 100 / nr_result;
 365                         if (progress_update || percent != last_percent) {
 366                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
 367                                         percent, written, nr_result);
 368                                 progress_update = 0;
 369                                 last_percent = percent;
 370                         }
 371                 }
 372         }
 373         if (do_progress)
 374                 fputc('\n', stderr);
 375  done:
 376         sha1close(f, pack_file_sha1, 1);
 377 }
 378
 379 static void write_index_file(void)
 380 {
 381         int i;
 382         struct sha1file *f = sha1create("%s-%s.%s", base_name,
 383                                         sha1_to_hex(object_list_sha1), "idx");
 384         struct object_entry **list = sorted_by_sha;
 385         struct object_entry **last = list + nr_result;
 386         unsigned int array[256];
 387
 388         /*
 389          * Write the first-level table (the list is sorted,
 390          * but we use a 256-entry lookup to be able to avoid
 391          * having to do eight extra binary search iterations).
 392          */
 393         for (i = 0; i < 256; i++) {
 394                 struct object_entry **next = list;
 395                 while (next < last) {
 396                         struct object_entry *entry = *next;
 397                         if (entry->sha1[0] != i)
 398                                 break;
 399                         next++;
 400                 }
 401                 array[i] = htonl(next - sorted_by_sha);
 402                 list = next;
 403         }
 404         sha1write(f, array, 256 * sizeof(int));
 405
 406         /*
 407          * Write the actual SHA1 entries..
 408          */
 409         list = sorted_by_sha;
 410         for (i = 0; i < nr_result; i++) {
 411                 struct object_entry *entry = *list++;
 412                 unsigned int offset = htonl(entry->offset);
 413                 sha1write(f, &offset, 4);
 414                 sha1write(f, entry->sha1, 20);
 415         }
 416         sha1write(f, pack_file_sha1, 20);
 417         sha1close(f, NULL, 1);
 418 }
 419
 420 static int locate_object_entry_hash(const unsigned char *sha1)
 421 {
 422         int i;
 423         unsigned int ui;
 424         memcpy(&ui, sha1, sizeof(unsigned int));
 425         i = ui % object_ix_hashsz;
 426         while (0 < object_ix[i]) {
 427                 if (!memcmp(sha1, objects[object_ix[i]-1].sha1, 20))
 428                         return i;
 429                 if (++i == object_ix_hashsz)
 430                         i = 0;
 431         }
 432         return -1 - i;
 433 }
 434
 435 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 436 {
 437         int i;
 438
 439         if (!object_ix_hashsz)
 440                 return NULL;
 441
 442         i = locate_object_entry_hash(sha1);
 443         if (0 <= i)
 444                 return &objects[object_ix[i]-1];
 445         return NULL;
 446 }
 447
 448 static void rehash_objects(void)
 449 {
 450         int i;
 451         struct object_entry *oe;
 452
 453         object_ix_hashsz = nr_objects * 3;
 454         if (object_ix_hashsz < 1024)
 455                 object_ix_hashsz = 1024;
 456         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 457         memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 458         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 459                 int ix = locate_object_entry_hash(oe->sha1);
 460                 if (0 <= ix)
 461                         continue;
 462                 ix = -1 - ix;
 463                 object_ix[ix] = i + 1;
 464         }
 465 }
 466
 467 struct name_path {
 468         struct name_path *up;
 469         const char *elem;
 470         int len;
 471 };
 472
 473 #define DIRBITS 12
 474
 475 static unsigned name_hash(struct name_path *path, const char *name)
 476 {
 477         struct name_path *p = path;
 478         const char *n = name + strlen(name);
 479         unsigned hash = 0, name_hash = 0, name_done = 0;
 480
 481         if (n != name && n[-1] == '\n')
 482                 n--;
 483         while (name <= --n) {
 484                 unsigned char c = *n;
 485                 if (c == '/' && !name_done) {
 486                         name_hash = hash;
 487                         name_done = 1;
 488                         hash = 0;
 489                 }
 490                 hash = hash * 11 + c;
 491         }
 492         if (!name_done) {
 493                 name_hash = hash;
 494                 hash = 0;
 495         }
 496         for (p = path; p; p = p->up) {
 497                 hash = hash * 11 + '/';
 498                 n = p->elem + p->len;
 499                 while (p->elem <= --n) {
 500                         unsigned char c = *n;
 501                         hash = hash * 11 + c;
 502                 }
 503         }
 504         /*
 505          * Make sure "Makefile" and "t/Makefile" are hashed separately
 506          * but close enough.
 507          */
 508         hash = (name_hash<<DIRBITS) | (hash & ((1U<<DIRBITS )-1));
 509         return hash;
 510 }
 511
 512 static int add_object_entry(const unsigned char *sha1, unsigned hash, int exclude)
 513 {
 514         unsigned int idx = nr_objects;
 515         struct object_entry *entry;
 516         struct packed_git *p;
 517         unsigned int found_offset = 0;
 518         struct packed_git *found_pack = NULL;
 519         int ix, status = 0;
 520
 521         if (!exclude) {
 522                 for (p = packed_git; p; p = p->next) {
 523                         struct pack_entry e;
 524                         if (find_pack_entry_one(sha1, &e, p)) {
 525                                 if (incremental)
 526                                         return 0;
 527                                 if (local && !p->pack_local)
 528                                         return 0;
 529                                 if (!found_pack) {
 530                                         found_offset = e.offset;
 531                                         found_pack = e.p;
 532                                 }
 533                         }
 534                 }
 535         }
 536         if ((entry = locate_object_entry(sha1)) != NULL)
 537                 goto already_added;
 538
 539         if (idx >= nr_alloc) {
 540                 unsigned int needed = (idx + 1024) * 3 / 2;
 541                 objects = xrealloc(objects, needed * sizeof(*entry));
 542                 nr_alloc = needed;
 543         }
 544         entry = objects + idx;
 545         nr_objects = idx + 1;
 546         memset(entry, 0, sizeof(*entry));
 547         memcpy(entry->sha1, sha1, 20);
 548         entry->hash = hash;
 549
 550         if (object_ix_hashsz * 3 <= nr_objects * 4)
 551                 rehash_objects();
 552         else {
 553                 ix = locate_object_entry_hash(entry->sha1);
 554                 if (0 <= ix)
 555                         die("internal error in object hashing.");
 556                 object_ix[-1 - ix] = idx + 1;
 557         }
 558         status = 1;
 559
 560  already_added:
 561         if (progress_update) {
 562                 fprintf(stderr, "Counting objects...%d\r", nr_objects);
 563                 progress_update = 0;
 564         }
 565         if (exclude)
 566                 entry->preferred_base = 1;
 567         else {
 568                 if (found_pack) {
 569                         entry->in_pack = found_pack;
 570                         entry->in_pack_offset = found_offset;
 571                 }
 572         }
 573         return status;
 574 }
 575
 576 struct pbase_tree_cache {
 577         unsigned char sha1[20];
 578         int ref;
 579         int temporary;
 580         void *tree_data;
 581         unsigned long tree_size;
 582 };
 583
 584 static struct pbase_tree_cache *(pbase_tree_cache[256]);
 585 static int pbase_tree_cache_ix(const unsigned char *sha1)
 586 {
 587         return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
 588 }
 589 static int pbase_tree_cache_ix_incr(int ix)
 590 {
 591         return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
 592 }
 593
 594 static struct pbase_tree {
 595         struct pbase_tree *next;
 596         /* This is a phony "cache" entry; we are not
 597          * going to evict it nor find it through _get()
 598          * mechanism -- this is for the toplevel node that
 599          * would almost always change with any commit.
 600          */
 601         struct pbase_tree_cache pcache;
 602 } *pbase_tree;
 603
 604 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
 605 {
 606         struct pbase_tree_cache *ent, *nent;
 607         void *data;
 608         unsigned long size;
 609         char type[20];
 610         int neigh;
 611         int my_ix = pbase_tree_cache_ix(sha1);
 612         int available_ix = -1;
 613
 614         /* pbase-tree-cache acts as a limited hashtable.
 615          * your object will be found at your index or within a few
 616          * slots after that slot if it is cached.
 617          */
 618         for (neigh = 0; neigh < 8; neigh++) {
 619                 ent = pbase_tree_cache[my_ix];
 620                 if (ent && !memcmp(ent->sha1, sha1, 20)) {
 621                         ent->ref++;
 622                         return ent;
 623                 }
 624                 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
 625                          ((0 <= available_ix) &&
 626                           (!ent && pbase_tree_cache[available_ix])))
 627                         available_ix = my_ix;
 628                 if (!ent)
 629                         break;
 630                 my_ix = pbase_tree_cache_ix_incr(my_ix);
 631         }
 632
 633         /* Did not find one.  Either we got a bogus request or
 634          * we need to read and perhaps cache.
 635          */
 636         data = read_sha1_file(sha1, type, &size);
 637         if (!data)
 638                 return NULL;
 639         if (strcmp(type, tree_type)) {
 640                 free(data);
 641                 return NULL;
 642         }
 643
 644         /* We need to either cache or return a throwaway copy */
 645
 646         if (available_ix < 0)
 647                 ent = NULL;
 648         else {
 649                 ent = pbase_tree_cache[available_ix];
 650                 my_ix = available_ix;
 651         }
 652
 653         if (!ent) {
 654                 nent = xmalloc(sizeof(*nent));
 655                 nent->temporary = (available_ix < 0);
 656         }
 657         else {
 658                 /* evict and reuse */
 659                 free(ent->tree_data);
 660                 nent = ent;
 661         }
 662         memcpy(nent->sha1, sha1, 20);
 663         nent->tree_data = data;
 664         nent->tree_size = size;
 665         nent->ref = 1;
 666         if (!nent->temporary)
 667                 pbase_tree_cache[my_ix] = nent;
 668         return nent;
 669 }
 670
 671 static void pbase_tree_put(struct pbase_tree_cache *cache)
 672 {
 673         if (!cache->temporary) {
 674                 cache->ref--;
 675                 return;
 676         }
 677         free(cache->tree_data);
 678         free(cache);
 679 }
 680
 681 static int name_cmp_len(const char *name)
 682 {
 683         int i;
 684         for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
 685                 ;
 686         return i;
 687 }
 688
 689 static void add_pbase_object(struct tree_desc *tree,
 690                              struct name_path *up,
 691                              const char *name,
 692                              int cmplen)
 693 {
 694         while (tree->size) {
 695                 const unsigned char *sha1;
 696                 const char *entry_name;
 697                 int entry_len;
 698                 unsigned mode;
 699                 unsigned long size;
 700                 char type[20];
 701
 702                 sha1 = tree_entry_extract(tree, &entry_name, &mode);
 703                 update_tree_entry(tree);
 704                 entry_len = strlen(entry_name);
 705                 if (entry_len != cmplen ||
 706                     memcmp(entry_name, name, cmplen) ||
 707                     !has_sha1_file(sha1) ||
 708                     sha1_object_info(sha1, type, &size))
 709                         continue;
 710                 if (name[cmplen] != '/') {
 711                         unsigned hash = name_hash(up, name);
 712                         add_object_entry(sha1, hash, 1);
 713                         return;
 714                 }
 715                 if (!strcmp(type, tree_type)) {
 716                         struct tree_desc sub;
 717                         struct name_path me;
 718                         struct pbase_tree_cache *tree;
 719                         const char *down = name+cmplen+1;
 720                         int downlen = name_cmp_len(down);
 721
 722                         tree = pbase_tree_get(sha1);
 723                         if (!tree)
 724                                 return;
 725                         sub.buf = tree->tree_data;
 726                         sub.size = tree->tree_size;
 727
 728                         me.up = up;
 729                         me.elem = entry_name;
 730                         me.len = entry_len;
 731                         add_pbase_object(&sub, &me, down, downlen);
 732                         pbase_tree_put(tree);
 733                 }
 734         }
 735 }
 736
 737 static unsigned *done_pbase_paths;
 738 static int done_pbase_paths_num;
 739 static int done_pbase_paths_alloc;
 740 static int done_pbase_path_pos(unsigned hash)
 741 {
 742         int lo = 0;
 743         int hi = done_pbase_paths_num;
 744         while (lo < hi) {
 745                 int mi = (hi + lo) / 2;
 746                 if (done_pbase_paths[mi] == hash)
 747                         return mi;
 748                 if (done_pbase_paths[mi] < hash)
 749                         hi = mi;
 750                 else
 751                         lo = mi + 1;
 752         }
 753         return -lo-1;
 754 }
 755
 756 static int check_pbase_path(unsigned hash)
 757 {
 758         int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
 759         if (0 <= pos)
 760                 return 1;
 761         pos = -pos - 1;
 762         if (done_pbase_paths_alloc <= done_pbase_paths_num) {
 763                 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
 764                 done_pbase_paths = xrealloc(done_pbase_paths,
 765                                             done_pbase_paths_alloc *
 766                                             sizeof(unsigned));
 767         }
 768         done_pbase_paths_num++;
 769         if (pos < done_pbase_paths_num)
 770                 memmove(done_pbase_paths + pos + 1,
 771                         done_pbase_paths + pos,
 772                         (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
 773         done_pbase_paths[pos] = hash;
 774         return 0;
 775 }
 776
 777 static void add_preferred_base_object(char *name, unsigned hash)
 778 {
 779         struct pbase_tree *it;
 780         int cmplen = name_cmp_len(name);
 781
 782         if (check_pbase_path(hash))
 783                 return;
 784
 785         for (it = pbase_tree; it; it = it->next) {
 786                 if (cmplen == 0) {
 787                         hash = name_hash(NULL, "");
 788                         add_object_entry(it->pcache.sha1, hash, 1);
 789                 }
 790                 else {
 791                         struct tree_desc tree;
 792                         tree.buf = it->pcache.tree_data;
 793                         tree.size = it->pcache.tree_size;
 794                         add_pbase_object(&tree, NULL, name, cmplen);
 795                 }
 796         }
 797 }
 798
 799 static void add_preferred_base(unsigned char *sha1)
 800 {
 801         struct pbase_tree *it;
 802         void *data;
 803         unsigned long size;
 804         unsigned char tree_sha1[20];
 805
 806         data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
 807         if (!data)
 808                 return;
 809
 810         for (it = pbase_tree; it; it = it->next) {
 811                 if (!memcmp(it->pcache.sha1, tree_sha1, 20)) {
 812                         free(data);
 813                         return;
 814                 }
 815         }
 816
 817         it = xcalloc(1, sizeof(*it));
 818         it->next = pbase_tree;
 819         pbase_tree = it;
 820
 821         memcpy(it->pcache.sha1, tree_sha1, 20);
 822         it->pcache.tree_data = data;
 823         it->pcache.tree_size = size;
 824 }
 825
 826 static void check_object(struct object_entry *entry)
 827 {
 828         char type[20];
 829
 830         if (entry->in_pack && !entry->preferred_base) {
 831                 unsigned char base[20];
 832                 unsigned long size;
 833                 struct object_entry *base_entry;
 834
 835                 /* We want in_pack_type even if we do not reuse delta.
 836                  * There is no point not reusing non-delta representations.
 837                  */
 838                 check_reuse_pack_delta(entry->in_pack,
 839                                        entry->in_pack_offset,
 840                                        base, &size,
 841                                        &entry->in_pack_type);
 842
 843                 /* Check if it is delta, and the base is also an object
 844                  * we are going to pack.  If so we will reuse the existing
 845                  * delta.
 846                  */
 847                 if (!no_reuse_delta &&
 848                     entry->in_pack_type == OBJ_DELTA &&
 849                     (base_entry = locate_object_entry(base)) &&
 850                     (!base_entry->preferred_base)) {
 851
 852                         /* Depth value does not matter - find_deltas()
 853                          * will never consider reused delta as the
 854                          * base object to deltify other objects
 855                          * against, in order to avoid circular deltas.
 856                          */
 857
 858                         /* uncompressed size of the delta data */
 859                         entry->size = entry->delta_size = size;
 860                         entry->delta = base_entry;
 861                         entry->type = OBJ_DELTA;
 862
 863                         entry->delta_sibling = base_entry->delta_child;
 864                         base_entry->delta_child = entry;
 865
 866                         return;
 867                 }
 868                 /* Otherwise we would do the usual */
 869         }
 870
 871         if (sha1_object_info(entry->sha1, type, &entry->size))
 872                 die("unable to get type of object %s",
 873                     sha1_to_hex(entry->sha1));
 874
 875         if (!strcmp(type, commit_type)) {
 876                 entry->type = OBJ_COMMIT;
 877         } else if (!strcmp(type, tree_type)) {
 878                 entry->type = OBJ_TREE;
 879         } else if (!strcmp(type, blob_type)) {
 880                 entry->type = OBJ_BLOB;
 881         } else if (!strcmp(type, tag_type)) {
 882                 entry->type = OBJ_TAG;
 883         } else
 884                 die("unable to pack object %s of type %s",
 885                     sha1_to_hex(entry->sha1), type);
 886 }
 887
 888 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
 889 {
 890         struct object_entry *child = me->delta_child;
 891         unsigned int m = n;
 892         while (child) {
 893                 unsigned int c = check_delta_limit(child, n + 1);
 894                 if (m < c)
 895                         m = c;
 896                 child = child->delta_sibling;
 897         }
 898         return m;
 899 }
 900
 901 static void get_object_details(void)
 902 {
 903         int i;
 904         struct object_entry *entry;
 905
 906         prepare_pack_ix();
 907         for (i = 0, entry = objects; i < nr_objects; i++, entry++)
 908                 check_object(entry);
 909
 910         if (nr_objects == nr_result) {
 911                 /*
 912                  * Depth of objects that depend on the entry -- this
 913                  * is subtracted from depth-max to break too deep
 914                  * delta chain because of delta data reusing.
 915                  * However, we loosen this restriction when we know we
 916                  * are creating a thin pack -- it will have to be
 917                  * expanded on the other end anyway, so do not
 918                  * artificially cut the delta chain and let it go as
 919                  * deep as it wants.
 920                  */
 921                 for (i = 0, entry = objects; i < nr_objects; i++, entry++)
 922                         if (!entry->delta && entry->delta_child)
 923                                 entry->delta_limit =
 924                                         check_delta_limit(entry, 1);
 925         }
 926 }
 927
 928 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
 929
 930 static entry_sort_t current_sort;
 931
 932 static int sort_comparator(const void *_a, const void *_b)
 933 {
 934         struct object_entry *a = *(struct object_entry **)_a;
 935         struct object_entry *b = *(struct object_entry **)_b;
 936         return current_sort(a,b);
 937 }
 938
 939 static struct object_entry **create_sorted_list(entry_sort_t sort)
 940 {
 941         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
 942         int i;
 943
 944         for (i = 0; i < nr_objects; i++)
 945                 list[i] = objects + i;
 946         current_sort = sort;
 947         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
 948         return list;
 949 }
 950
 951 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
 952 {
 953         return memcmp(a->sha1, b->sha1, 20);
 954 }
 955
 956 static struct object_entry **create_final_object_list(void)
 957 {
 958         struct object_entry **list;
 959         int i, j;
 960
 961         for (i = nr_result = 0; i < nr_objects; i++)
 962                 if (!objects[i].preferred_base)
 963                         nr_result++;
 964         list = xmalloc(nr_result * sizeof(struct object_entry *));
 965         for (i = j = 0; i < nr_objects; i++) {
 966                 if (!objects[i].preferred_base)
 967                         list[j++] = objects + i;
 968         }
 969         current_sort = sha1_sort;
 970         qsort(list, nr_result, sizeof(struct object_entry *), sort_comparator);
 971         return list;
 972 }
 973
 974 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
 975 {
 976         if (a->type < b->type)
 977                 return -1;
 978         if (a->type > b->type)
 979                 return 1;
 980         if (a->hash < b->hash)
 981                 return -1;
 982         if (a->hash > b->hash)
 983                 return 1;
 984         if (a->preferred_base < b->preferred_base)
 985                 return -1;
 986         if (a->preferred_base > b->preferred_base)
 987                 return 1;
 988         if (a->size < b->size)
 989                 return -1;
 990         if (a->size > b->size)
 991                 return 1;
 992         return a < b ? -1 : (a > b);
 993 }
 994
 995 struct unpacked {
 996         struct object_entry *entry;
 997         void *data;
 998 };
 999
1000 /*
1001  * We search for deltas _backwards_ in a list sorted by type and
1002  * by size, so that we see progressively smaller and smaller files.
1003  * That's because we prefer deltas to be from the bigger file
1004  * to the smaller - deletes are potentially cheaper, but perhaps
1005  * more importantly, the bigger file is likely the more recent
1006  * one.
1007  */
1008 static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_depth)
1009 {
1010         struct object_entry *cur_entry = cur->entry;
1011         struct object_entry *old_entry = old->entry;
1012         unsigned long size, oldsize, delta_size, sizediff;
1013         long max_size;
1014         void *delta_buf;
1015
1016         /* Don't bother doing diffs between different types */
1017         if (cur_entry->type != old_entry->type)
1018                 return -1;
1019
1020         /* We do not compute delta to *create* objects we are not
1021          * going to pack.
1022          */
1023         if (cur_entry->preferred_base)
1024                 return -1;
1025
1026         /* If the current object is at pack edge, take the depth the
1027          * objects that depend on the current object into account --
1028          * otherwise they would become too deep.
1029          */
1030         if (cur_entry->delta_child) {
1031                 if (max_depth <= cur_entry->delta_limit)
1032                         return 0;
1033                 max_depth -= cur_entry->delta_limit;
1034         }
1035
1036         size = cur_entry->size;
1037         oldsize = old_entry->size;
1038         sizediff = oldsize > size ? oldsize - size : size - oldsize;
1039
1040         if (size < 50)
1041                 return -1;
1042         if (old_entry->depth >= max_depth)
1043                 return 0;
1044
1045         /*
1046          * NOTE!
1047          *
1048          * We always delta from the bigger to the smaller, since that's
1049          * more space-efficient (deletes don't have to say _what_ they
1050          * delete).
1051          */
1052         max_size = size / 2 - 20;
1053         if (cur_entry->delta)
1054                 max_size = cur_entry->delta_size-1;
1055         if (sizediff >= max_size)
1056                 return 0;
1057         delta_buf = diff_delta(old->data, oldsize,
1058                                cur->data, size, &delta_size, max_size);
1059         if (!delta_buf)
1060                 return 0;
1061         cur_entry->delta = old_entry;
1062         cur_entry->delta_size = delta_size;
1063         cur_entry->depth = old_entry->depth + 1;
1064         free(delta_buf);
1065         return 0;
1066 }
1067
1068 static void progress_interval(int signum)
1069 {
1070         progress_update = 1;
1071 }
1072
1073 static void find_deltas(struct object_entry **list, int window, int depth)
1074 {
1075         int i, idx;
1076         unsigned int array_size = window * sizeof(struct unpacked);
1077         struct unpacked *array = xmalloc(array_size);
1078         unsigned processed = 0;
1079         unsigned last_percent = 999;
1080
1081         memset(array, 0, array_size);
1082         i = nr_objects;
1083         idx = 0;
1084         if (progress)
1085                 fprintf(stderr, "Deltifying %d objects.\n", nr_result);
1086
1087         while (--i >= 0) {
1088                 struct object_entry *entry = list[i];
1089                 struct unpacked *n = array + idx;
1090                 unsigned long size;
1091                 char type[10];
1092                 int j;
1093
1094                 if (!entry->preferred_base)
1095                         processed++;
1096
1097                 if (progress) {
1098                         unsigned percent = processed * 100 / nr_result;
1099                         if (percent != last_percent || progress_update) {
1100                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
1101                                         percent, processed, nr_result);
1102                                 progress_update = 0;
1103                                 last_percent = percent;
1104                         }
1105                 }
1106
1107                 if (entry->delta)
1108                         /* This happens if we decided to reuse existing
1109                          * delta from a pack.  "!no_reuse_delta &&" is implied.
1110                          */
1111                         continue;
1112
1113                 free(n->data);
1114                 n->entry = entry;
1115                 n->data = read_sha1_file(entry->sha1, type, &size);
1116                 if (size != entry->size)
1117                         die("object %s inconsistent object length (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
1118
1119                 j = window;
1120                 while (--j > 0) {
1121                         unsigned int other_idx = idx + j;
1122                         struct unpacked *m;
1123                         if (other_idx >= window)
1124                                 other_idx -= window;
1125                         m = array + other_idx;
1126                         if (!m->entry)
1127                                 break;
1128                         if (try_delta(n, m, depth) < 0)
1129                                 break;
1130                 }
1131 #if 0
1132                 /* if we made n a delta, and if n is already at max
1133                  * depth, leaving it in the window is pointless.  we
1134                  * should evict it first.
1135                  * ... in theory only; somehow this makes things worse.
1136                  */
1137                 if (entry->delta && depth <= entry->depth)
1138                         continue;
1139 #endif
1140                 idx++;
1141                 if (idx >= window)
1142                         idx = 0;
1143         }
1144
1145         if (progress)
1146                 fputc('\n', stderr);
1147
1148         for (i = 0; i < window; ++i)
1149                 free(array[i].data);
1150         free(array);
1151 }
1152
1153 static void prepare_pack(int window, int depth)
1154 {
1155         get_object_details();
1156         sorted_by_type = create_sorted_list(type_size_sort);
1157         if (window && depth)
1158                 find_deltas(sorted_by_type, window+1, depth);
1159 }
1160
1161 static int reuse_cached_pack(unsigned char *sha1, int pack_to_stdout)
1162 {
1163         static const char cache[] = "pack-cache/pack-%s.%s";
1164         char *cached_pack, *cached_idx;
1165         int ifd, ofd, ifd_ix = -1;
1166
1167         cached_pack = git_path(cache, sha1_to_hex(sha1), "pack");
1168         ifd = open(cached_pack, O_RDONLY);
1169         if (ifd < 0)
1170                 return 0;
1171
1172         if (!pack_to_stdout) {
1173                 cached_idx = git_path(cache, sha1_to_hex(sha1), "idx");
1174                 ifd_ix = open(cached_idx, O_RDONLY);
1175                 if (ifd_ix < 0) {
1176                         close(ifd);
1177                         return 0;
1178                 }
1179         }
1180
1181         if (progress)
1182                 fprintf(stderr, "Reusing %d objects pack %s\n", nr_objects,
1183                         sha1_to_hex(sha1));
1184
1185         if (pack_to_stdout) {
1186                 if (copy_fd(ifd, 1))
1187                         exit(1);
1188                 close(ifd);
1189         }
1190         else {
1191                 char name[PATH_MAX];
1192                 snprintf(name, sizeof(name),
1193                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "pack");
1194                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1195                 if (ofd < 0)
1196                         die("unable to open %s (%s)", name, strerror(errno));
1197                 if (copy_fd(ifd, ofd))
1198                         exit(1);
1199                 close(ifd);
1200
1201                 snprintf(name, sizeof(name),
1202                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "idx");
1203                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1204                 if (ofd < 0)
1205                         die("unable to open %s (%s)", name, strerror(errno));
1206                 if (copy_fd(ifd_ix, ofd))
1207                         exit(1);
1208                 close(ifd_ix);
1209                 puts(sha1_to_hex(sha1));
1210         }
1211
1212         return 1;
1213 }
1214
1215 static void setup_progress_signal(void)
1216 {
1217         struct sigaction sa;
1218         struct itimerval v;
1219
1220         memset(&sa, 0, sizeof(sa));
1221         sa.sa_handler = progress_interval;
1222         sigemptyset(&sa.sa_mask);
1223         sa.sa_flags = SA_RESTART;
1224         sigaction(SIGALRM, &sa, NULL);
1225
1226         v.it_interval.tv_sec = 1;
1227         v.it_interval.tv_usec = 0;
1228         v.it_value = v.it_interval;
1229         setitimer(ITIMER_REAL, &v, NULL);
1230 }
1231
1232 int main(int argc, char **argv)
1233 {
1234         SHA_CTX ctx;
1235         char line[40 + 1 + PATH_MAX + 2];
1236         int window = 10, depth = 10, pack_to_stdout = 0;
1237         struct object_entry **list;
1238         int num_preferred_base = 0;
1239         int i;
1240
1241         setup_git_directory();
1242
1243         for (i = 1; i < argc; i++) {
1244                 const char *arg = argv[i];
1245
1246                 if (*arg == '-') {
1247                         if (!strcmp("--non-empty", arg)) {
1248                                 non_empty = 1;
1249                                 continue;
1250                         }
1251                         if (!strcmp("--local", arg)) {
1252                                 local = 1;
1253                                 continue;
1254                         }
1255                         if (!strcmp("--incremental", arg)) {
1256                                 incremental = 1;
1257                                 continue;
1258                         }
1259                         if (!strncmp("--window=", arg, 9)) {
1260                                 char *end;
1261                                 window = strtoul(arg+9, &end, 0);
1262                                 if (!arg[9] || *end)
1263                                         usage(pack_usage);
1264                                 continue;
1265                         }
1266                         if (!strncmp("--depth=", arg, 8)) {
1267                                 char *end;
1268                                 depth = strtoul(arg+8, &end, 0);
1269                                 if (!arg[8] || *end)
1270                                         usage(pack_usage);
1271                                 continue;
1272                         }
1273                         if (!strcmp("-q", arg)) {
1274                                 progress = 0;
1275                                 continue;
1276                         }
1277                         if (!strcmp("--no-reuse-delta", arg)) {
1278                                 no_reuse_delta = 1;
1279                                 continue;
1280                         }
1281                         if (!strcmp("--stdout", arg)) {
1282                                 pack_to_stdout = 1;
1283                                 continue;
1284                         }
1285                         usage(pack_usage);
1286                 }
1287                 if (base_name)
1288                         usage(pack_usage);
1289                 base_name = arg;
1290         }
1291
1292         if (pack_to_stdout != !base_name)
1293                 usage(pack_usage);
1294
1295         prepare_packed_git();
1296
1297         if (progress) {
1298                 fprintf(stderr, "Generating pack...\n");
1299                 setup_progress_signal();
1300         }
1301
1302         for (;;) {
1303                 unsigned char sha1[20];
1304                 unsigned hash;
1305
1306                 if (!fgets(line, sizeof(line), stdin)) {
1307                         if (feof(stdin))
1308                                 break;
1309                         if (!ferror(stdin))
1310                                 die("fgets returned NULL, not EOF, not error!");
1311                         if (errno != EINTR)
1312                                 die("fgets: %s", strerror(errno));
1313                         clearerr(stdin);
1314                         continue;
1315                 }
1316
1317                 if (line[0] == '-') {
1318                         if (get_sha1_hex(line+1, sha1))
1319                                 die("expected edge sha1, got garbage:\n %s",
1320                                     line+1);
1321                         if (num_preferred_base++ < window)
1322                                 add_preferred_base(sha1);
1323                         continue;
1324                 }
1325                 if (get_sha1_hex(line, sha1))
1326                         die("expected sha1, got garbage:\n %s", line);
1327                 hash = name_hash(NULL, line+41);
1328                 add_preferred_base_object(line+41, hash);
1329                 add_object_entry(sha1, hash, 0);
1330         }
1331         if (progress)
1332                 fprintf(stderr, "Done counting %d objects.\n", nr_objects);
1333         sorted_by_sha = create_final_object_list();
1334         if (non_empty && !nr_result)
1335                 return 0;
1336
1337         SHA1_Init(&ctx);
1338         list = sorted_by_sha;
1339         for (i = 0; i < nr_result; i++) {
1340                 struct object_entry *entry = *list++;
1341                 SHA1_Update(&ctx, entry->sha1, 20);
1342         }
1343         SHA1_Final(object_list_sha1, &ctx);
1344         if (progress && (nr_objects != nr_result))
1345                 fprintf(stderr, "Result has %d objects.\n", nr_result);
1346
1347         if (reuse_cached_pack(object_list_sha1, pack_to_stdout))
1348                 ;
1349         else {
1350                 if (nr_result)
1351                         prepare_pack(window, depth);
1352                 if (progress && pack_to_stdout) {
1353                         /* the other end usually displays progress itself */
1354                         struct itimerval v = {{0,},};
1355                         setitimer(ITIMER_REAL, &v, NULL);
1356                         signal(SIGALRM, SIG_IGN );
1357                         progress_update = 0;
1358                 }
1359                 write_pack_file();
1360                 if (!pack_to_stdout) {
1361                         write_index_file();
1362                         puts(sha1_to_hex(object_list_sha1));
1363                 }
1364         }
1365         if (progress)
1366                 fprintf(stderr, "Total %d, written %d (delta %d), reused %d (delta %d)\n",
1367                         nr_result, written, written_delta, reused, reused_delta);
1368         return 0;
1369 }