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,
 153                                                       const char *sha1,
 154                                                       struct llist_item *hint)
 155 {
 156         struct llist_item *prev, *l;
 157
 158 redo_from_start:
 159         l = (hint == NULL) ? list->front : hint;
 160         prev = NULL;
 161         while (l) {
 162                 int cmp = memcmp(l->sha1, sha1, 20);
 163                 if (cmp > 0) /* not in list, since sorted */
 164                         return prev;
 165                 if(!cmp) { /* found */
 166                         if (prev == NULL) {
 167                                 if (hint != NULL && hint != list->front) {
 168                                         /* we don't know the previous element */
 169                                         hint = NULL;
 170                                         goto redo_from_start;
 171                                 }
 172                                 list->front = l->next;
 173                         } else
 174                                 prev->next = l->next;
 175                         if (l == list->back)
 176                                 list->back = prev;
 177                         llist_item_put(l);
 178                         list->size--;
 179                         return prev;
 180                 }
 181                 prev = l;
 182                 l = l->next;
 183         }
 184         return prev;
 185 }
 186
 187 /* computes A\B */
 188 static void llist_sorted_difference_inplace(struct llist *A,
 189                                      struct llist *B)
 190 {
 191         struct llist_item *hint, *b;
 192
 193         hint = NULL;
 194         b = B->front;
 195
 196         while (b) {
 197                 hint = llist_sorted_remove(A, b->sha1, hint);
 198                 b = b->next;
 199         }
 200 }
 201
 202 static inline struct pack_list * pack_list_insert(struct pack_list **pl,
 203                                            struct pack_list *entry)
 204 {
 205         struct pack_list *p = xmalloc(sizeof(struct pack_list));
 206         memcpy(p, entry, sizeof(struct pack_list));
 207         p->next = *pl;
 208         *pl = p;
 209         return p;
 210 }
 211
 212 static inline size_t pack_list_size(struct pack_list *pl)
 213 {
 214         size_t ret = 0;
 215         while(pl) {
 216                 ret++;
 217                 pl = pl->next;
 218         }
 219         return ret;
 220 }
 221
 222 static struct pack_list * pack_list_difference(const struct pack_list *A,
 223                                                const struct pack_list *B)
 224 {
 225         struct pack_list *ret;
 226         const struct pack_list *pl;
 227
 228         if (A == NULL)
 229                 return NULL;
 230
 231         pl = B;
 232         while (pl != NULL) {
 233                 if (A->pack == pl->pack)
 234                         return pack_list_difference(A->next, B);
 235                 pl = pl->next;
 236         }
 237         ret = xmalloc(sizeof(struct pack_list));
 238         memcpy(ret, A, sizeof(struct pack_list));
 239         ret->next = pack_list_difference(A->next, B);
 240         return ret;
 241 }
 242
 243 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 244 {
 245         int p1_off, p2_off;
 246         void *p1_base, *p2_base;
 247         struct llist_item *p1_hint = NULL, *p2_hint = NULL;
 248
 249         p1_off = p2_off = 256 * 4 + 4;
 250         p1_base = (void *)p1->pack->index_base;
 251         p2_base = (void *)p2->pack->index_base;
 252
 253         while (p1_off <= p1->pack->index_size - 3 * 20 &&
 254                p2_off <= p2->pack->index_size - 3 * 20)
 255         {
 256                 int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
 257                 /* cmp ~ p1 - p2 */
 258                 if (cmp == 0) {
 259                         p1_hint = llist_sorted_remove(p1->unique_objects,
 260                                         p1_base + p1_off, p1_hint);
 261                         p2_hint = llist_sorted_remove(p2->unique_objects,
 262                                         p1_base + p1_off, p2_hint);
 263                         p1_off+=24;
 264                         p2_off+=24;
 265                         continue;
 266                 }
 267                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 268                         p1_off+=24;
 269                 } else { /* p2 has the object, p1 doesn't */
 270                         p2_off+=24;
 271                 }
 272         }
 273 }
 274
 275 static void pll_insert(struct pll **pll, struct pll **hint_table)
 276 {
 277         struct pll *prev;
 278         int i = (*pll)->pl_size - 1;
 279
 280         if (hint_table[i] == NULL) {
 281                 hint_table[i--] = *pll;
 282                 for (; i >= 0; --i) {
 283                         if (hint_table[i] != NULL)
 284                                 break;
 285                 }
 286                 if (hint_table[i] == NULL) /* no elements in list */
 287                         die("Why did this happen?");
 288         }
 289
 290         prev = hint_table[i];
 291         while (prev->next && prev->next->pl_size < (*pll)->pl_size)
 292                 prev = prev->next;
 293
 294         (*pll)->next = prev->next;
 295         prev->next = *pll;
 296 }
 297
 298 /* all the permutations have to be free()d at the same time,
 299  * since they refer to each other
 300  */
 301 static struct pll * get_all_permutations(struct pack_list *list)
 302 {
 303         struct pll *subset, *pll, *new_pll = NULL; /*silence warning*/
 304         static struct pll **hint = NULL;
 305         if (hint == NULL)
 306                 hint = xcalloc(pack_list_size(list), sizeof(struct pll *));
 307
 308         if (list == NULL)
 309                 return NULL;
 310
 311         if (list->next == NULL) {
 312                 new_pll = xmalloc(sizeof(struct pll));
 313                 hint[0] = new_pll;
 314                 new_pll->next = NULL;
 315                 new_pll->pl = list;
 316                 new_pll->pl_size = 1;
 317                 return new_pll;
 318         }
 319
 320         pll = subset = get_all_permutations(list->next);
 321         while (pll) {
 322                 if (pll->pl->pack == list->pack) {
 323                         pll = pll->next;
 324                         continue;
 325                 }
 326                 new_pll = xmalloc(sizeof(struct pll));
 327
 328                 new_pll->pl = xmalloc(sizeof(struct pack_list));
 329                 memcpy(new_pll->pl, list, sizeof(struct pack_list));
 330                 new_pll->pl->next = pll->pl;
 331                 new_pll->pl_size = pll->pl_size + 1;
 332
 333                 pll_insert(&new_pll, hint);
 334
 335                 pll = pll->next;
 336         }
 337         /* add ourself */
 338         new_pll = xmalloc(sizeof(struct pll));
 339         new_pll->pl = xmalloc(sizeof(struct pack_list));
 340         memcpy(new_pll->pl, list, sizeof(struct pack_list));
 341         new_pll->pl->next = NULL;
 342         new_pll->pl_size = 1;
 343         pll_insert(&new_pll, hint);
 344
 345         return hint[0];
 346 }
 347
 348 static int is_superset(struct pack_list *pl, struct llist *list)
 349 {
 350         struct llist *diff;
 351
 352         diff = llist_copy(list);
 353
 354         while (pl) {
 355                 llist_sorted_difference_inplace(diff, pl->all_objects);
 356                 if (diff->size == 0) { /* we're done */
 357                         llist_free(diff);
 358                         return 1;
 359                 }
 360                 pl = pl->next;
 361         }
 362         llist_free(diff);
 363         return 0;
 364 }
 365
 366 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 367 {
 368         size_t ret = 0;
 369         int p1_off, p2_off;
 370         void *p1_base, *p2_base;
 371
 372         p1_off = p2_off = 256 * 4 + 4;
 373         p1_base = (void *)p1->index_base;
 374         p2_base = (void *)p2->index_base;
 375
 376         while (p1_off <= p1->index_size - 3 * 20 &&
 377                p2_off <= p2->index_size - 3 * 20)
 378         {
 379                 int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
 380                 /* cmp ~ p1 - p2 */
 381                 if (cmp == 0) {
 382                         ret++;
 383                         p1_off+=24;
 384                         p2_off+=24;
 385                         continue;
 386                 }
 387                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 388                         p1_off+=24;
 389                 } else { /* p2 has the object, p1 doesn't */
 390                         p2_off+=24;
 391                 }
 392         }
 393         return ret;
 394 }
 395
 396 /* another O(n^2) function ... */
 397 static size_t get_pack_redundancy(struct pack_list *pl)
 398 {
 399         struct pack_list *subset;
 400         size_t ret = 0;
 401
 402         if (pl == NULL)
 403                 return 0;
 404
 405         while ((subset = pl->next)) {
 406                 while(subset) {
 407                         ret += sizeof_union(pl->pack, subset->pack);
 408                         subset = subset->next;
 409                 }
 410                 pl = pl->next;
 411         }
 412         return ret;
 413 }
 414
 415 static inline size_t pack_set_bytecount(struct pack_list *pl)
 416 {
 417         size_t ret = 0;
 418         while (pl) {
 419                 ret += pl->pack->pack_size;
 420                 ret += pl->pack->index_size;
 421                 pl = pl->next;
 422         }
 423         return ret;
 424 }
 425
 426 static void minimize(struct pack_list **min)
 427 {
 428         struct pack_list *pl, *unique = NULL,
 429                 *non_unique = NULL, *min_perm = NULL;
 430         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 431         struct llist *missing;
 432         size_t min_perm_size = (size_t)-1, perm_size;
 433
 434         pl = local_packs;
 435         while (pl) {
 436                 if(pl->unique_objects->size)
 437                         pack_list_insert(&unique, pl);
 438                 else
 439                         pack_list_insert(&non_unique, pl);
 440                 pl = pl->next;
 441         }
 442         /* find out which objects are missing from the set of unique packs */
 443         missing = llist_copy(all_objects);
 444         pl = unique;
 445         while (pl) {
 446                 llist_sorted_difference_inplace(missing,
 447                                                 pl->all_objects);
 448                 pl = pl->next;
 449         }
 450
 451         /* return if there are no objects missing from the unique set */
 452         if (missing->size == 0) {
 453                 *min = unique;
 454                 return;
 455         }
 456
 457         /* find the permutations which contain all missing objects */
 458         perm_all = perm = get_all_permutations(non_unique);
 459         while (perm) {
 460                 if (perm_ok && perm->pl_size > perm_ok->pl_size)
 461                         break; /* ignore all larger permutations */
 462                 if (is_superset(perm->pl, missing)) {
 463                         new_perm = xmalloc(sizeof(struct pll));
 464                         memcpy(new_perm, perm, sizeof(struct pll));
 465                         new_perm->next = perm_ok;
 466                         perm_ok = new_perm;
 467                 }
 468                 perm = perm->next;
 469         }
 470
 471         if (perm_ok == NULL)
 472                 die("Internal error: No complete sets found!\n");
 473
 474         /* find the permutation with the smallest size */
 475         perm = perm_ok;
 476         while (perm) {
 477                 perm_size = pack_set_bytecount(perm->pl);
 478                 if (min_perm_size > perm_size) {
 479                         min_perm_size = perm_size;
 480                         min_perm = perm->pl;
 481                 }
 482                 perm = perm->next;
 483         }
 484         *min = min_perm;
 485         /* add the unique packs to the list */
 486         pl = unique;
 487         while(pl) {
 488                 pack_list_insert(min, pl);
 489                 pl = pl->next;
 490         }
 491 }
 492
 493 static void load_all_objects(void)
 494 {
 495         struct pack_list *pl = local_packs;
 496         struct llist_item *hint, *l;
 497
 498         llist_init(&all_objects);
 499
 500         while (pl) {
 501                 hint = NULL;
 502                 l = pl->all_objects->front;
 503                 while (l) {
 504                         hint = llist_insert_sorted_unique(all_objects,
 505                                                           l->sha1, hint);
 506                         l = l->next;
 507                 }
 508                 pl = pl->next;
 509         }
 510         /* remove objects present in remote packs */
 511         pl = altodb_packs;
 512         while (pl) {
 513                 llist_sorted_difference_inplace(all_objects, pl->all_objects);
 514                 pl = pl->next;
 515         }
 516 }
 517
 518 /* this scales like O(n^2) */
 519 static void cmp_local_packs(void)
 520 {
 521         struct pack_list *subset, *pl = local_packs;
 522
 523         while ((subset = pl)) {
 524                 while((subset = subset->next))
 525                         cmp_two_packs(pl, subset);
 526                 pl = pl->next;
 527         }
 528 }
 529
 530 static void scan_alt_odb_packs(void)
 531 {
 532         struct pack_list *local, *alt;
 533
 534         alt = altodb_packs;
 535         while (alt) {
 536                 local = local_packs;
 537                 while (local) {
 538                         llist_sorted_difference_inplace(local->unique_objects,
 539                                                         alt->all_objects);
 540                         local = local->next;
 541                 }
 542                 alt = alt->next;
 543         }
 544 }
 545
 546 static struct pack_list * add_pack(struct packed_git *p)
 547 {
 548         struct pack_list l;
 549         size_t off;
 550         void *base;
 551
 552         if (!p->pack_local && !(alt_odb || verbose))
 553                 return NULL;
 554
 555         l.pack = p;
 556         llist_init(&l.all_objects);
 557
 558         off = 256 * 4 + 4;
 559         base = (void *)p->index_base;
 560         while (off <= p->index_size - 3 * 20) {
 561                 llist_insert_back(l.all_objects, base + off);
 562                 off += 24;
 563         }
 564         /* this list will be pruned in cmp_two_packs later */
 565         l.unique_objects = llist_copy(l.all_objects);
 566         if (p->pack_local)
 567                 return pack_list_insert(&local_packs, &l);
 568         else
 569                 return pack_list_insert(&altodb_packs, &l);
 570 }
 571
 572 static struct pack_list * add_pack_file(char *filename)
 573 {
 574         struct packed_git *p = packed_git;
 575
 576         if (strlen(filename) < 40)
 577                 die("Bad pack filename: %s\n", filename);
 578
 579         while (p) {
 580                 if (strstr(p->pack_name, filename))
 581                         return add_pack(p);
 582                 p = p->next;
 583         }
 584         die("Filename %s not found in packed_git\n", filename);
 585 }
 586
 587 static void load_all(void)
 588 {
 589         struct packed_git *p = packed_git;
 590
 591         while (p) {
 592                 add_pack(p);
 593                 p = p->next;
 594         }
 595 }
 596
 597 int main(int argc, char **argv)
 598 {
 599         int i;
 600         struct pack_list *min, *red, *pl;
 601         struct llist *ignore;
 602         char *sha1, 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", pack_set_bytecount(red)/(1024*1024));
 690
 691         return 0;
 692 }