pack-redundant.c

   1 /*
   2 *
   3 * Copyright 2005, Lukas Sandstrom <lukass@etek.chalmers.se>
   4 *
   5 * This file is licensed under the GPL v2.
   6 *
   7 */
   8
   9 #include "cache.h"
  10
  11 static const char pack_redundant_usage[] =
  12 "git-pack-redundant [ --verbose ] [ --alt-odb ] < --all | <.pack filename> ...>";
  13
  14 static int load_all_packs = 0, verbose = 0, alt_odb = 0;
  15
  16 struct llist_item {
  17         struct llist_item *next;
  18         char *sha1;
  19 };
  20 static struct llist {
  21         struct llist_item *front;
  22         struct llist_item *back;
  23         size_t size;
  24 } *all_objects; /* all objects which must be present in local packfiles */
  25
  26 static struct pack_list {
  27         struct pack_list *next;
  28         struct packed_git *pack;
  29         struct llist *unique_objects;
  30         struct llist *all_objects;
  31 } *local_packs = NULL, *altodb_packs = NULL;
  32
  33 struct pll {
  34         struct pll *next;
  35         struct pack_list *pl;
  36         size_t pl_size;
  37 };
  38
  39 static struct llist_item *free_nodes = NULL;
  40
  41 static inline struct llist_item *llist_item_get()
  42 {
  43         struct llist_item *new;
  44         if ( free_nodes ) {
  45                 new = free_nodes;
  46                 free_nodes = free_nodes->next;
  47         } else
  48                 new = xmalloc(sizeof(struct llist_item));
  49
  50         return new;
  51 }
  52
  53 static inline void llist_item_put(struct llist_item *item)
  54 {
  55         item->next = free_nodes;
  56         free_nodes = item;
  57 }
  58
  59 static void llist_free(struct llist *list)
  60 {
  61         while((list->back = list->front)) {
  62                 list->front = list->front->next;
  63                 llist_item_put(list->back);
  64         }
  65         free(list);
  66 }
  67
  68 static inline void llist_init(struct llist **list)
  69 {
  70         *list = xmalloc(sizeof(struct llist));
  71         (*list)->front = (*list)->back = NULL;
  72         (*list)->size = 0;
  73 }
  74
  75 static struct llist * llist_copy(struct llist *list)
  76 {
  77         struct llist *ret;
  78         struct llist_item *new, *old, *prev;
  79
  80         llist_init(&ret);
  81
  82         if ((ret->size = list->size) == 0)
  83                 return ret;
  84
  85         new = ret->front = llist_item_get();
  86         new->sha1 = list->front->sha1;
  87
  88         old = list->front->next;
  89         while (old) {
  90                 prev = new;
  91                 new = llist_item_get();
  92                 prev->next = new;
  93                 new->sha1 = old->sha1;
  94                 old = old->next;
  95         }
  96         new->next = NULL;
  97         ret->back = new;
  98
  99         return ret;
 100 }
 101
 102 static inline struct llist_item * llist_insert(struct llist *list,
 103                                         struct llist_item *after, char *sha1)
 104 {
 105         struct llist_item *new = llist_item_get();
 106         new->sha1 = sha1;
 107         new->next = NULL;
 108
 109         if (after != NULL) {
 110                 new->next = after->next;
 111                 after->next = new;
 112                 if (after == list->back)
 113                         list->back = new;
 114         } else {/* insert in front */
 115                 if (list->size == 0)
 116                         list->back = new;
 117                 else
 118                         new->next = list->front;
 119                 list->front = new;
 120         }
 121         list->size++;
 122         return new;
 123 }
 124
 125 static inline struct llist_item * llist_insert_back(struct llist *list, char *sha1)
 126 {
 127         return llist_insert(list, list->back, sha1);
 128 }
 129
 130 static inline struct llist_item * llist_insert_sorted_unique(struct llist *list,
 131                                         char *sha1, struct llist_item *hint)
 132 {
 133         struct llist_item *prev = NULL, *l;
 134
 135         l = (hint == NULL) ? list->front : hint;
 136         while (l) {
 137                 int cmp = memcmp(l->sha1, sha1, 20);
 138                 if (cmp > 0) { /* we insert before this entry */
 139                         return llist_insert(list, prev, sha1);
 140                 }
 141                 if(!cmp) { /* already exists */
 142                         return l;
 143                 }
 144                 prev = l;
 145                 l = l->next;
 146         }
 147         /* insert at the end */
 148         return llist_insert_back(list, sha1);
 149 }
 150
 151 /* returns a pointer to an item in front of sha1 */
 152 static inline struct llist_item * llist_sorted_remove(struct llist *list, char *sha1,
 153                                                struct llist_item *hint)
 154 {
 155         struct llist_item *prev, *l;
 156
 157 redo_from_start:
 158         l = (hint == NULL) ? list->front : hint;
 159         prev = NULL;
 160         while (l) {
 161                 int cmp = memcmp(l->sha1, sha1, 20);
 162                 if (cmp > 0) /* not in list, since sorted */
 163                         return prev;
 164                 if(!cmp) { /* found */
 165                         if (prev == NULL) {
 166                                 if (hint != NULL && hint != list->front) {
 167                                         /* we don't know the previous element */
 168                                         hint = NULL;
 169                                         goto redo_from_start;
 170                                 }
 171                                 list->front = l->next;
 172                         } else
 173                                 prev->next = l->next;
 174                         if (l == list->back)
 175                                 list->back = prev;
 176                         llist_item_put(l);
 177                         list->size--;
 178                         return prev;
 179                 }
 180                 prev = l;
 181                 l = l->next;
 182         }
 183         return prev;
 184 }
 185
 186 /* computes A\B */
 187 static void llist_sorted_difference_inplace(struct llist *A,
 188                                      struct llist *B)
 189 {
 190         struct llist_item *hint, *b;
 191
 192         hint = NULL;
 193         b = B->front;
 194
 195         while (b) {
 196                 hint = llist_sorted_remove(A, b->sha1, hint);
 197                 b = b->next;
 198         }
 199 }
 200
 201 static inline struct pack_list * pack_list_insert(struct pack_list **pl,
 202                                            struct pack_list *entry)
 203 {
 204         struct pack_list *p = xmalloc(sizeof(struct pack_list));
 205         memcpy(p, entry, sizeof(struct pack_list));
 206         p->next = *pl;
 207         *pl = p;
 208         return p;
 209 }
 210
 211 static inline size_t pack_list_size(struct pack_list *pl)
 212 {
 213         size_t ret = 0;
 214         while(pl) {
 215                 ret++;
 216                 pl = pl->next;
 217         }
 218         return ret;
 219 }
 220
 221 static struct pack_list * pack_list_difference(struct pack_list *A,
 222                                         struct pack_list *B)
 223 {
 224         struct pack_list *ret, *pl;
 225
 226         if (A == NULL)
 227                 return NULL;
 228
 229         pl = B;
 230         while (pl != NULL) {
 231                 if (A->pack == pl->pack)
 232                         return pack_list_difference(A->next, B);
 233                 pl = pl->next;
 234         }
 235         ret = xmalloc(sizeof(struct pack_list));
 236         memcpy(ret, A, sizeof(struct pack_list));
 237         ret->next = pack_list_difference(A->next, B);
 238         return ret;
 239 }
 240
 241 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 242 {
 243         int p1_off, p2_off;
 244         void *p1_base, *p2_base;
 245         struct llist_item *p1_hint = NULL, *p2_hint = NULL;
 246
 247         p1_off = p2_off = 256 * 4 + 4;
 248         p1_base = (void *)p1->pack->index_base;
 249         p2_base = (void *)p2->pack->index_base;
 250
 251         while (p1_off <= p1->pack->index_size - 3 * 20 &&
 252                p2_off <= p2->pack->index_size - 3 * 20)
 253         {
 254                 int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
 255                 /* cmp ~ p1 - p2 */
 256                 if (cmp == 0) {
 257                         p1_hint = llist_sorted_remove(p1->unique_objects,
 258                                         p1_base + p1_off, p1_hint);
 259                         p2_hint = llist_sorted_remove(p2->unique_objects,
 260                                         p1_base + p1_off, p2_hint);
 261                         p1_off+=24;
 262                         p2_off+=24;
 263                         continue;
 264                 }
 265                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 266                         p1_off+=24;
 267                 } else { /* p2 has the object, p1 doesn't */
 268                         p2_off+=24;
 269                 }
 270         }
 271 }
 272
 273 static void pll_insert(struct pll **pll, struct pll **hint_table)
 274 {
 275         struct pll *prev;
 276         int i = (*pll)->pl_size - 1;
 277
 278         if (hint_table[i] == NULL) {
 279                 hint_table[i--] = *pll;
 280                 for (; i >= 0; --i) {
 281                         if (hint_table[i] != NULL)
 282                                 break;
 283                 }
 284                 if (hint_table[i] == NULL) /* no elements in list */
 285                         die("Why did this happen?");
 286         }
 287
 288         prev = hint_table[i];
 289         while (prev->next && prev->next->pl_size < (*pll)->pl_size)
 290                 prev = prev->next;
 291
 292         (*pll)->next = prev->next;
 293         prev->next = *pll;
 294 }
 295
 296 /* all the permutations have to be free()d at the same time,
 297  * since they refer to each other
 298  */
 299 static struct pll * get_all_permutations(struct pack_list *list)
 300 {
 301         struct pll *subset, *pll, *new_pll = NULL; /*silence warning*/
 302         static struct pll **hint = NULL;
 303         if (hint == NULL)
 304                 hint = xcalloc(pack_list_size(list), sizeof(struct pll *));
 305
 306         if (list == NULL)
 307                 return NULL;
 308
 309         if (list->next == NULL) {
 310                 new_pll = xmalloc(sizeof(struct pll));
 311                 hint[0] = new_pll;
 312                 new_pll->next = NULL;
 313                 new_pll->pl = list;
 314                 new_pll->pl_size = 1;
 315                 return new_pll;
 316         }
 317
 318         pll = subset = get_all_permutations(list->next);
 319         while (pll) {
 320                 if (pll->pl->pack == list->pack) {
 321                         pll = pll->next;
 322                         continue;
 323                 }
 324                 new_pll = xmalloc(sizeof(struct pll));
 325
 326                 new_pll->pl = xmalloc(sizeof(struct pack_list));
 327                 memcpy(new_pll->pl, list, sizeof(struct pack_list));
 328                 new_pll->pl->next = pll->pl;
 329                 new_pll->pl_size = pll->pl_size + 1;
 330
 331                 pll_insert(&new_pll, hint);
 332
 333                 pll = pll->next;
 334         }
 335         /* add ourself */
 336         new_pll = xmalloc(sizeof(struct pll));
 337         new_pll->pl = xmalloc(sizeof(struct pack_list));
 338         memcpy(new_pll->pl, list, sizeof(struct pack_list));
 339         new_pll->pl->next = NULL;
 340         new_pll->pl_size = 1;
 341         pll_insert(&new_pll, hint);
 342
 343         return hint[0];
 344 }
 345
 346 static int is_superset(struct pack_list *pl, struct llist *list)
 347 {
 348         struct llist *diff;
 349
 350         diff = llist_copy(list);
 351
 352         while (pl) {
 353                 llist_sorted_difference_inplace(diff,
 354                                                 pl->all_objects);
 355                 if (diff->size == 0) { /* we're done */
 356                         llist_free(diff);
 357                         return 1;
 358                 }
 359                 pl = pl->next;
 360         }
 361         llist_free(diff);
 362         return 0;
 363 }
 364
 365 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 366 {
 367         size_t ret = 0;
 368         int p1_off, p2_off;
 369         void *p1_base, *p2_base;
 370
 371         p1_off = p2_off = 256 * 4 + 4;
 372         p1_base = (void *)p1->index_base;
 373         p2_base = (void *)p2->index_base;
 374
 375         while (p1_off <= p1->index_size - 3 * 20 &&
 376                p2_off <= p2->index_size - 3 * 20)
 377         {
 378                 int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
 379                 /* cmp ~ p1 - p2 */
 380                 if (cmp == 0) {
 381                         ret++;
 382                         p1_off+=24;
 383                         p2_off+=24;
 384                         continue;
 385                 }
 386                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 387                         p1_off+=24;
 388                 } else { /* p2 has the object, p1 doesn't */
 389                         p2_off+=24;
 390                 }
 391         }
 392         return ret;
 393 }
 394
 395 /* another O(n^2) function ... */
 396 static size_t get_pack_redundancy(struct pack_list *pl)
 397 {
 398         struct pack_list *subset;
 399         size_t ret = 0;
 400
 401         if (pl == NULL)
 402                 return 0;
 403
 404         while ((subset = pl->next)) {
 405                 while(subset) {
 406                         ret += sizeof_union(pl->pack, subset->pack);
 407                         subset = subset->next;
 408                 }
 409                 pl = pl->next;
 410         }
 411         return ret;
 412 }
 413
 414 static inline size_t pack_set_bytecount(struct pack_list *pl)
 415 {
 416         size_t ret = 0;
 417         while (pl) {
 418                 ret += pl->pack->pack_size;
 419                 ret += pl->pack->index_size;
 420                 pl = pl->next;
 421         }
 422         return ret;
 423 }
 424
 425 static void minimize(struct pack_list **min)
 426 {
 427         struct pack_list *pl, *unique = NULL,
 428                 *non_unique = NULL, *min_perm = NULL;
 429         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 430         struct llist *missing;
 431         size_t min_perm_size = (size_t)-1, perm_size;
 432
 433         pl = local_packs;
 434         while (pl) {
 435                 if(pl->unique_objects->size)
 436                         pack_list_insert(&unique, pl);
 437                 else
 438                         pack_list_insert(&non_unique, pl);
 439                 pl = pl->next;
 440         }
 441         /* find out which objects are missing from the set of unique packs */
 442         missing = llist_copy(all_objects);
 443         pl = unique;
 444         while (pl) {
 445                 llist_sorted_difference_inplace(missing,
 446                                                 pl->all_objects);
 447                 pl = pl->next;
 448         }
 449
 450         /* return if there are no objects missing from the unique set */
 451         if (missing->size == 0) {
 452                 *min = unique;
 453                 return;
 454         }
 455
 456         /* find the permutations which contain all missing objects */
 457         perm_all = perm = get_all_permutations(non_unique);
 458         while (perm) {
 459                 if (perm_ok && perm->pl_size > perm_ok->pl_size)
 460                         break; /* ignore all larger permutations */
 461                 if (is_superset(perm->pl, missing)) {
 462                         new_perm = xmalloc(sizeof(struct pll));
 463                         memcpy(new_perm, perm, sizeof(struct pll));
 464                         new_perm->next = perm_ok;
 465                         perm_ok = new_perm;
 466                 }
 467                 perm = perm->next;
 468         }
 469
 470         if (perm_ok == NULL)
 471                 die("Internal error: No complete sets found!\n");
 472
 473         /* find the permutation with the smallest size */
 474         perm = perm_ok;
 475         while (perm) {
 476                 perm_size = pack_set_bytecount(perm->pl);
 477                 if (min_perm_size > perm_size) {
 478                         min_perm_size = perm_size;
 479                         min_perm = perm->pl;
 480                 }
 481                 perm = perm->next;
 482         }
 483         *min = min_perm;
 484         /* add the unique packs to the list */
 485         pl = unique;
 486         while(pl) {
 487                 pack_list_insert(min, pl);
 488                 pl = pl->next;
 489         }
 490 }
 491
 492 static void load_all_objects(void)
 493 {
 494         struct pack_list *pl = local_packs;
 495         struct llist_item *hint, *l;
 496
 497         llist_init(&all_objects);
 498
 499         while (pl) {
 500                 hint = NULL;
 501                 l = pl->all_objects->front;
 502                 while (l) {
 503                         hint = llist_insert_sorted_unique(all_objects,
 504                                                           l->sha1, hint);
 505                         l = l->next;
 506                 }
 507                 pl = pl->next;
 508         }
 509         /* remove objects present in remote packs */
 510         pl = altodb_packs;
 511         while (pl) {
 512                 llist_sorted_difference_inplace(all_objects, pl->all_objects);
 513                 pl = pl->next;
 514         }
 515 }
 516
 517 /* this scales like O(n^2) */
 518 static void cmp_local_packs(void)
 519 {
 520         struct pack_list *subset, *pl = local_packs;
 521
 522         while ((subset = pl)) {
 523                 while((subset = subset->next))
 524                         cmp_two_packs(pl, subset);
 525                 pl = pl->next;
 526         }
 527 }
 528
 529 static void scan_alt_odb_packs(void)
 530 {
 531         struct pack_list *local, *alt;
 532
 533         alt = altodb_packs;
 534         while (alt) {
 535                 local = local_packs;
 536                 while (local) {
 537                         llist_sorted_difference_inplace(local->unique_objects,
 538                                                         alt->all_objects);
 539                         local = local->next;
 540                 }
 541                 alt = alt->next;
 542         }
 543 }
 544
 545 static struct pack_list * add_pack(struct packed_git *p)
 546 {
 547         struct pack_list l;
 548         size_t off;
 549         void *base;
 550
 551         if (!p->pack_local && !(alt_odb || verbose))
 552                 return NULL;
 553
 554         l.pack = p;
 555         llist_init(&l.all_objects);
 556
 557         off = 256 * 4 + 4;
 558         base = (void *)p->index_base;
 559         while (off <= p->index_size - 3 * 20) {
 560                 llist_insert_back(l.all_objects, base + off);
 561                 off += 24;
 562         }
 563         /* this list will be pruned in cmp_two_packs later */
 564         l.unique_objects = llist_copy(l.all_objects);
 565         if (p->pack_local)
 566                 return pack_list_insert(&local_packs, &l);
 567         else
 568                 return pack_list_insert(&altodb_packs, &l);
 569 }
 570
 571 static struct pack_list * add_pack_file(char *filename)
 572 {
 573         struct packed_git *p = packed_git;
 574
 575         if (strlen(filename) < 40)
 576                 die("Bad pack filename: %s\n", filename);
 577
 578         while (p) {
 579                 if (strstr(p->pack_name, filename))
 580                         return add_pack(p);
 581                 p = p->next;
 582         }
 583         die("Filename %s not found in packed_git\n", filename);
 584 }
 585
 586 static void load_all(void)
 587 {
 588         struct packed_git *p = packed_git;
 589
 590         while (p) {
 591                 add_pack(p);
 592                 p = p->next;
 593         }
 594 }
 595
 596 int main(int argc, char **argv)
 597 {
 598         int i;
 599         struct pack_list *min, *red, *pl;
 600         struct llist *ignore;
 601         char *sha1, buf[42]; /* 40 byte sha1 + \n + \0 */
 602
 603         for (i = 1; i < argc; i++) {
 604                 const char *arg = argv[i];
 605                 if(!strcmp(arg, "--")) {
 606                         i++;
 607                         break;
 608                 }
 609                 if(!strcmp(arg, "--all")) {
 610                         load_all_packs = 1;
 611                         continue;
 612                 }
 613                 if(!strcmp(arg, "--verbose")) {
 614                         verbose = 1;
 615                         continue;
 616                 }
 617                 if(!strcmp(arg, "--alt-odb")) {
 618                         alt_odb = 1;
 619                         continue;
 620                 }
 621                 if(*arg == '-')
 622                         usage(pack_redundant_usage);
 623                 else
 624                         break;
 625         }
 626
 627         prepare_packed_git();
 628
 629         if (load_all_packs)
 630                 load_all();
 631         else
 632                 while (*(argv + i) != NULL)
 633                         add_pack_file(*(argv + i++));
 634
 635         if (local_packs == NULL)
 636                 die("Zero packs found!\n");
 637
 638         load_all_objects();
 639
 640         cmp_local_packs();
 641         if (alt_odb)
 642                 scan_alt_odb_packs();
 643
 644         /* ignore objects given on stdin */
 645         llist_init(&ignore);
 646         if (!isatty(0)) {
 647                 while (fgets(buf, sizeof(buf), stdin)) {
 648                         sha1 = xmalloc(20);
 649                         if (get_sha1_hex(buf, sha1))
 650                                 die("Bad sha1 on stdin: %s", buf);
 651                         llist_insert_sorted_unique(ignore, sha1, NULL);
 652                 }
 653         }
 654         llist_sorted_difference_inplace(all_objects, ignore);
 655         pl = local_packs;
 656         while (pl) {
 657                 llist_sorted_difference_inplace(pl->unique_objects, ignore);
 658                 pl = pl->next;
 659         }
 660
 661         minimize(&min);
 662
 663         if (verbose) {
 664                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 665                         (unsigned long)pack_list_size(altodb_packs));
 666                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 667                 pl = min;
 668                 while (pl) {
 669                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 670                         pl = pl->next;
 671                 }
 672                 fprintf(stderr, "containing %lu duplicate objects "
 673                                 "with a total size of %lukb.\n",
 674                         (unsigned long)get_pack_redundancy(min),
 675                         (unsigned long)pack_set_bytecount(min)/1024);
 676                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 677                         (unsigned long)all_objects->size);
 678                 fprintf(stderr, "Redundant packs (with indexes):\n");
 679         }
 680         pl = red = pack_list_difference(local_packs, min);
 681         while (pl) {
 682                 printf("%s\n%s\n",
 683                        sha1_pack_index_name(pl->pack->sha1),
 684                        pl->pack->pack_name);
 685                 pl = pl->next;
 686         }
 687         if (verbose)
 688                 fprintf(stderr, "%luMB of redundant packs in total.\n", pack_set_bytecount(red)/(1024*1024));
 689
 690         return 0;
 691 }