builtin-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, verbose, alt_odb;
  17
  18 struct llist_item {
  19         struct llist_item *next;
  20         const 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;
  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                                                const 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,
 132                                                    const unsigned char *sha1)
 133 {
 134         return llist_insert(list, list->back, sha1);
 135 }
 136
 137 static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
 138                         const unsigned char *sha1, struct llist_item *hint)
 139 {
 140         struct llist_item *prev = NULL, *l;
 141
 142         l = (hint == NULL) ? list->front : hint;
 143         while (l) {
 144                 int cmp = hashcmp(l->sha1, sha1);
 145                 if (cmp > 0) { /* we insert before this entry */
 146                         return llist_insert(list, prev, sha1);
 147                 }
 148                 if(!cmp) { /* already exists */
 149                         return l;
 150                 }
 151                 prev = l;
 152                 l = l->next;
 153         }
 154         /* insert at the end */
 155         return llist_insert_back(list, sha1);
 156 }
 157
 158 /* returns a pointer to an item in front of sha1 */
 159 static inline struct llist_item * llist_sorted_remove(struct llist *list, const unsigned char *sha1, struct llist_item *hint)
 160 {
 161         struct llist_item *prev, *l;
 162
 163 redo_from_start:
 164         l = (hint == NULL) ? list->front : hint;
 165         prev = NULL;
 166         while (l) {
 167                 int cmp = hashcmp(l->sha1, sha1);
 168                 if (cmp > 0) /* not in list, since sorted */
 169                         return prev;
 170                 if(!cmp) { /* found */
 171                         if (prev == NULL) {
 172                                 if (hint != NULL && hint != list->front) {
 173                                         /* we don't know the previous element */
 174                                         hint = NULL;
 175                                         goto redo_from_start;
 176                                 }
 177                                 list->front = l->next;
 178                         } else
 179                                 prev->next = l->next;
 180                         if (l == list->back)
 181                                 list->back = prev;
 182                         llist_item_put(l);
 183                         list->size--;
 184                         return prev;
 185                 }
 186                 prev = l;
 187                 l = l->next;
 188         }
 189         return prev;
 190 }
 191
 192 /* computes A\B */
 193 static void llist_sorted_difference_inplace(struct llist *A,
 194                                      struct llist *B)
 195 {
 196         struct llist_item *hint, *b;
 197
 198         hint = NULL;
 199         b = B->front;
 200
 201         while (b) {
 202                 hint = llist_sorted_remove(A, b->sha1, hint);
 203                 b = b->next;
 204         }
 205 }
 206
 207 static inline struct pack_list * pack_list_insert(struct pack_list **pl,
 208                                            struct pack_list *entry)
 209 {
 210         struct pack_list *p = xmalloc(sizeof(struct pack_list));
 211         memcpy(p, entry, sizeof(struct pack_list));
 212         p->next = *pl;
 213         *pl = p;
 214         return p;
 215 }
 216
 217 static inline size_t pack_list_size(struct pack_list *pl)
 218 {
 219         size_t ret = 0;
 220         while(pl) {
 221                 ret++;
 222                 pl = pl->next;
 223         }
 224         return ret;
 225 }
 226
 227 static struct pack_list * pack_list_difference(const struct pack_list *A,
 228                                                const struct pack_list *B)
 229 {
 230         struct pack_list *ret;
 231         const struct pack_list *pl;
 232
 233         if (A == NULL)
 234                 return NULL;
 235
 236         pl = B;
 237         while (pl != NULL) {
 238                 if (A->pack == pl->pack)
 239                         return pack_list_difference(A->next, B);
 240                 pl = pl->next;
 241         }
 242         ret = xmalloc(sizeof(struct pack_list));
 243         memcpy(ret, A, sizeof(struct pack_list));
 244         ret->next = pack_list_difference(A->next, B);
 245         return ret;
 246 }
 247
 248 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 249 {
 250         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 251         const unsigned char *p1_base, *p2_base;
 252         struct llist_item *p1_hint = NULL, *p2_hint = NULL;
 253
 254         p1_base = p1->pack->index_data;
 255         p2_base = p2->pack->index_data;
 256         p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
 257         p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
 258         p1_step = (p1->pack->index_version < 2) ? 24 : 20;
 259         p2_step = (p2->pack->index_version < 2) ? 24 : 20;
 260
 261         while (p1_off < p1->pack->num_objects * p1_step &&
 262                p2_off < p2->pack->num_objects * p2_step)
 263         {
 264                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 265                 /* cmp ~ p1 - p2 */
 266                 if (cmp == 0) {
 267                         p1_hint = llist_sorted_remove(p1->unique_objects,
 268                                         p1_base + p1_off, p1_hint);
 269                         p2_hint = llist_sorted_remove(p2->unique_objects,
 270                                         p1_base + p1_off, p2_hint);
 271                         p1_off += p1_step;
 272                         p2_off += p2_step;
 273                         continue;
 274                 }
 275                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 276                         p1_off += p1_step;
 277                 } else { /* p2 has the object, p1 doesn't */
 278                         p2_off += p2_step;
 279                 }
 280         }
 281 }
 282
 283 static void pll_free(struct pll *l)
 284 {
 285         struct pll *old;
 286         struct pack_list *opl;
 287
 288         while (l) {
 289                 old = l;
 290                 while (l->pl) {
 291                         opl = l->pl;
 292                         l->pl = opl->next;
 293                         free(opl);
 294                 }
 295                 l = l->next;
 296                 free(old);
 297         }
 298 }
 299
 300 /* all the permutations have to be free()d at the same time,
 301  * since they refer to each other
 302  */
 303 static struct pll * get_permutations(struct pack_list *list, int n)
 304 {
 305         struct pll *subset, *ret = NULL, *new_pll = NULL, *pll;
 306
 307         if (list == NULL || pack_list_size(list) < n || n == 0)
 308                 return NULL;
 309
 310         if (n == 1) {
 311                 while (list) {
 312                         new_pll = xmalloc(sizeof(pll));
 313                         new_pll->pl = NULL;
 314                         pack_list_insert(&new_pll->pl, list);
 315                         new_pll->next = ret;
 316                         ret = new_pll;
 317                         list = list->next;
 318                 }
 319                 return ret;
 320         }
 321
 322         while (list->next) {
 323                 subset = get_permutations(list->next, n - 1);
 324                 while (subset) {
 325                         new_pll = xmalloc(sizeof(pll));
 326                         new_pll->pl = subset->pl;
 327                         pack_list_insert(&new_pll->pl, list);
 328                         new_pll->next = ret;
 329                         ret = new_pll;
 330                         subset = subset->next;
 331                 }
 332                 list = list->next;
 333         }
 334         return ret;
 335 }
 336
 337 static int is_superset(struct pack_list *pl, struct llist *list)
 338 {
 339         struct llist *diff;
 340
 341         diff = llist_copy(list);
 342
 343         while (pl) {
 344                 llist_sorted_difference_inplace(diff, pl->all_objects);
 345                 if (diff->size == 0) { /* we're done */
 346                         llist_free(diff);
 347                         return 1;
 348                 }
 349                 pl = pl->next;
 350         }
 351         llist_free(diff);
 352         return 0;
 353 }
 354
 355 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 356 {
 357         size_t ret = 0;
 358         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 359         const unsigned char *p1_base, *p2_base;
 360
 361         p1_base = p1->index_data;
 362         p2_base = p2->index_data;
 363         p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
 364         p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
 365         p1_step = (p1->index_version < 2) ? 24 : 20;
 366         p2_step = (p2->index_version < 2) ? 24 : 20;
 367
 368         while (p1_off < p1->num_objects * p1_step &&
 369                p2_off < p2->num_objects * p2_step)
 370         {
 371                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 372                 /* cmp ~ p1 - p2 */
 373                 if (cmp == 0) {
 374                         ret++;
 375                         p1_off += p1_step;
 376                         p2_off += p2_step;
 377                         continue;
 378                 }
 379                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 380                         p1_off += p1_step;
 381                 } else { /* p2 has the object, p1 doesn't */
 382                         p2_off += p2_step;
 383                 }
 384         }
 385         return ret;
 386 }
 387
 388 /* another O(n^2) function ... */
 389 static size_t get_pack_redundancy(struct pack_list *pl)
 390 {
 391         struct pack_list *subset;
 392         size_t ret = 0;
 393
 394         if (pl == NULL)
 395                 return 0;
 396
 397         while ((subset = pl->next)) {
 398                 while(subset) {
 399                         ret += sizeof_union(pl->pack, subset->pack);
 400                         subset = subset->next;
 401                 }
 402                 pl = pl->next;
 403         }
 404         return ret;
 405 }
 406
 407 static inline off_t pack_set_bytecount(struct pack_list *pl)
 408 {
 409         off_t ret = 0;
 410         while (pl) {
 411                 ret += pl->pack->pack_size;
 412                 ret += pl->pack->index_size;
 413                 pl = pl->next;
 414         }
 415         return ret;
 416 }
 417
 418 static void minimize(struct pack_list **min)
 419 {
 420         struct pack_list *pl, *unique = NULL,
 421                 *non_unique = NULL, *min_perm = NULL;
 422         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 423         struct llist *missing;
 424         off_t min_perm_size = 0, perm_size;
 425         int n;
 426
 427         pl = local_packs;
 428         while (pl) {
 429                 if(pl->unique_objects->size)
 430                         pack_list_insert(&unique, pl);
 431                 else
 432                         pack_list_insert(&non_unique, pl);
 433                 pl = pl->next;
 434         }
 435         /* find out which objects are missing from the set of unique packs */
 436         missing = llist_copy(all_objects);
 437         pl = unique;
 438         while (pl) {
 439                 llist_sorted_difference_inplace(missing, pl->all_objects);
 440                 pl = pl->next;
 441         }
 442
 443         /* return if there are no objects missing from the unique set */
 444         if (missing->size == 0) {
 445                 *min = unique;
 446                 return;
 447         }
 448
 449         /* find the permutations which contain all missing objects */
 450         for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
 451                 perm_all = perm = get_permutations(non_unique, n);
 452                 while (perm) {
 453                         if (is_superset(perm->pl, missing)) {
 454                                 new_perm = xmalloc(sizeof(struct pll));
 455                                 memcpy(new_perm, perm, sizeof(struct pll));
 456                                 new_perm->next = perm_ok;
 457                                 perm_ok = new_perm;
 458                         }
 459                         perm = perm->next;
 460                 }
 461                 if (perm_ok)
 462                         break;
 463                 pll_free(perm_all);
 464         }
 465         if (perm_ok == NULL)
 466                 die("Internal error: No complete sets found!\n");
 467
 468         /* find the permutation with the smallest size */
 469         perm = perm_ok;
 470         while (perm) {
 471                 perm_size = pack_set_bytecount(perm->pl);
 472                 if (!min_perm_size || min_perm_size > perm_size) {
 473                         min_perm_size = perm_size;
 474                         min_perm = perm->pl;
 475                 }
 476                 perm = perm->next;
 477         }
 478         *min = min_perm;
 479         /* add the unique packs to the list */
 480         pl = unique;
 481         while(pl) {
 482                 pack_list_insert(min, pl);
 483                 pl = pl->next;
 484         }
 485 }
 486
 487 static void load_all_objects(void)
 488 {
 489         struct pack_list *pl = local_packs;
 490         struct llist_item *hint, *l;
 491
 492         llist_init(&all_objects);
 493
 494         while (pl) {
 495                 hint = NULL;
 496                 l = pl->all_objects->front;
 497                 while (l) {
 498                         hint = llist_insert_sorted_unique(all_objects,
 499                                                           l->sha1, hint);
 500                         l = l->next;
 501                 }
 502                 pl = pl->next;
 503         }
 504         /* remove objects present in remote packs */
 505         pl = altodb_packs;
 506         while (pl) {
 507                 llist_sorted_difference_inplace(all_objects, pl->all_objects);
 508                 pl = pl->next;
 509         }
 510 }
 511
 512 /* this scales like O(n^2) */
 513 static void cmp_local_packs(void)
 514 {
 515         struct pack_list *subset, *pl = local_packs;
 516
 517         while ((subset = pl)) {
 518                 while((subset = subset->next))
 519                         cmp_two_packs(pl, subset);
 520                 pl = pl->next;
 521         }
 522 }
 523
 524 static void scan_alt_odb_packs(void)
 525 {
 526         struct pack_list *local, *alt;
 527
 528         alt = altodb_packs;
 529         while (alt) {
 530                 local = local_packs;
 531                 while (local) {
 532                         llist_sorted_difference_inplace(local->unique_objects,
 533                                                         alt->all_objects);
 534                         local = local->next;
 535                 }
 536                 llist_sorted_difference_inplace(all_objects, alt->all_objects);
 537                 alt = alt->next;
 538         }
 539 }
 540
 541 static struct pack_list * add_pack(struct packed_git *p)
 542 {
 543         struct pack_list l;
 544         unsigned long off = 0, step;
 545         const unsigned char *base;
 546
 547         if (!p->pack_local && !(alt_odb || verbose))
 548                 return NULL;
 549
 550         l.pack = p;
 551         llist_init(&l.all_objects);
 552
 553         if (open_pack_index(p))
 554                 return NULL;
 555
 556         base = p->index_data;
 557         base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
 558         step = (p->index_version < 2) ? 24 : 20;
 559         while (off < p->num_objects * step) {
 560                 llist_insert_back(l.all_objects, base + off);
 561                 off += step;
 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 cmd_pack_redundant(int argc, const char **argv, const char *prefix)
 597 {
 598         int i;
 599         struct pack_list *min, *red, *pl;
 600         struct llist *ignore;
 601         unsigned char *sha1;
 602         char buf[42]; /* 40 byte sha1 + \n + \0 */
 603
 604         for (i = 1; i < argc; i++) {
 605                 const char *arg = argv[i];
 606                 if(!strcmp(arg, "--")) {
 607                         i++;
 608                         break;
 609                 }
 610                 if(!strcmp(arg, "--all")) {
 611                         load_all_packs = 1;
 612                         continue;
 613                 }
 614                 if(!strcmp(arg, "--verbose")) {
 615                         verbose = 1;
 616                         continue;
 617                 }
 618                 if(!strcmp(arg, "--alt-odb")) {
 619                         alt_odb = 1;
 620                         continue;
 621                 }
 622                 if(*arg == '-')
 623                         usage(pack_redundant_usage);
 624                 else
 625                         break;
 626         }
 627
 628         prepare_packed_git();
 629
 630         if (load_all_packs)
 631                 load_all();
 632         else
 633                 while (*(argv + i) != NULL)
 634                         add_pack_file(*(argv + i++));
 635
 636         if (local_packs == NULL)
 637                 die("Zero packs found!\n");
 638
 639         load_all_objects();
 640
 641         cmp_local_packs();
 642         if (alt_odb)
 643                 scan_alt_odb_packs();
 644
 645         /* ignore objects given on stdin */
 646         llist_init(&ignore);
 647         if (!isatty(0)) {
 648                 while (fgets(buf, sizeof(buf), stdin)) {
 649                         sha1 = xmalloc(20);
 650                         if (get_sha1_hex(buf, sha1))
 651                                 die("Bad sha1 on stdin: %s", buf);
 652                         llist_insert_sorted_unique(ignore, sha1, NULL);
 653                 }
 654         }
 655         llist_sorted_difference_inplace(all_objects, ignore);
 656         pl = local_packs;
 657         while (pl) {
 658                 llist_sorted_difference_inplace(pl->unique_objects, ignore);
 659                 pl = pl->next;
 660         }
 661
 662         minimize(&min);
 663
 664         if (verbose) {
 665                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 666                         (unsigned long)pack_list_size(altodb_packs));
 667                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 668                 pl = min;
 669                 while (pl) {
 670                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 671                         pl = pl->next;
 672                 }
 673                 fprintf(stderr, "containing %lu duplicate objects "
 674                                 "with a total size of %lukb.\n",
 675                         (unsigned long)get_pack_redundancy(min),
 676                         (unsigned long)pack_set_bytecount(min)/1024);
 677                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 678                         (unsigned long)all_objects->size);
 679                 fprintf(stderr, "Redundant packs (with indexes):\n");
 680         }
 681         pl = red = pack_list_difference(local_packs, min);
 682         while (pl) {
 683                 printf("%s\n%s\n",
 684                        sha1_pack_index_name(pl->pack->sha1),
 685                        pl->pack->pack_name);
 686                 pl = pl->next;
 687         }
 688         if (verbose)
 689                 fprintf(stderr, "%luMB of redundant packs in total.\n",
 690                         (unsigned long)pack_set_bytecount(red)/(1024*1024));
 691
 692         return 0;
 693 }