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