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 "builtin.h"
  10 #include "repository.h"
  11 #include "packfile.h"
  12 #include "object-store.h"
  13
  14 #define BLKSIZE 512
  15
  16 static const char pack_redundant_usage[] =
  17 "git pack-redundant [--verbose] [--alt-odb] (--all | <filename.pack>...)";
  18
  19 static int load_all_packs, verbose, alt_odb;
  20
  21 struct llist_item {
  22         struct llist_item *next;
  23         const unsigned char *sha1;
  24 };
  25 static struct llist {
  26         struct llist_item *front;
  27         struct llist_item *back;
  28         size_t size;
  29 } *all_objects; /* all objects which must be present in local packfiles */
  30
  31 static struct pack_list {
  32         struct pack_list *next;
  33         struct packed_git *pack;
  34         struct llist *unique_objects;
  35         struct llist *all_objects;
  36 } *local_packs = NULL, *altodb_packs = NULL;
  37
  38 struct pll {
  39         struct pll *next;
  40         struct pack_list *pl;
  41 };
  42
  43 static struct llist_item *free_nodes;
  44
  45 static inline void llist_item_put(struct llist_item *item)
  46 {
  47         item->next = free_nodes;
  48         free_nodes = item;
  49 }
  50
  51 static inline struct llist_item *llist_item_get(void)
  52 {
  53         struct llist_item *new_item;
  54         if ( free_nodes ) {
  55                 new_item = free_nodes;
  56                 free_nodes = free_nodes->next;
  57         } else {
  58                 int i = 1;
  59                 ALLOC_ARRAY(new_item, BLKSIZE);
  60                 for (; i < BLKSIZE; i++)
  61                         llist_item_put(&new_item[i]);
  62         }
  63         return new_item;
  64 }
  65
  66 static void llist_free(struct llist *list)
  67 {
  68         while ((list->back = list->front)) {
  69                 list->front = list->front->next;
  70                 llist_item_put(list->back);
  71         }
  72         free(list);
  73 }
  74
  75 static inline void llist_init(struct llist **list)
  76 {
  77         *list = xmalloc(sizeof(struct llist));
  78         (*list)->front = (*list)->back = NULL;
  79         (*list)->size = 0;
  80 }
  81
  82 static struct llist * llist_copy(struct llist *list)
  83 {
  84         struct llist *ret;
  85         struct llist_item *new_item, *old_item, *prev;
  86
  87         llist_init(&ret);
  88
  89         if ((ret->size = list->size) == 0)
  90                 return ret;
  91
  92         new_item = ret->front = llist_item_get();
  93         new_item->sha1 = list->front->sha1;
  94
  95         old_item = list->front->next;
  96         while (old_item) {
  97                 prev = new_item;
  98                 new_item = llist_item_get();
  99                 prev->next = new_item;
 100                 new_item->sha1 = old_item->sha1;
 101                 old_item = old_item->next;
 102         }
 103         new_item->next = NULL;
 104         ret->back = new_item;
 105
 106         return ret;
 107 }
 108
 109 static inline struct llist_item *llist_insert(struct llist *list,
 110                                               struct llist_item *after,
 111                                                const unsigned char *sha1)
 112 {
 113         struct llist_item *new_item = llist_item_get();
 114         new_item->sha1 = sha1;
 115         new_item->next = NULL;
 116
 117         if (after != NULL) {
 118                 new_item->next = after->next;
 119                 after->next = new_item;
 120                 if (after == list->back)
 121                         list->back = new_item;
 122         } else {/* insert in front */
 123                 if (list->size == 0)
 124                         list->back = new_item;
 125                 else
 126                         new_item->next = list->front;
 127                 list->front = new_item;
 128         }
 129         list->size++;
 130         return new_item;
 131 }
 132
 133 static inline struct llist_item *llist_insert_back(struct llist *list,
 134                                                    const unsigned char *sha1)
 135 {
 136         return llist_insert(list, list->back, sha1);
 137 }
 138
 139 static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
 140                         const unsigned char *sha1, struct llist_item *hint)
 141 {
 142         struct llist_item *prev = NULL, *l;
 143
 144         l = (hint == NULL) ? list->front : hint;
 145         while (l) {
 146                 int cmp = hashcmp(l->sha1, sha1);
 147                 if (cmp > 0) { /* we insert before this entry */
 148                         return llist_insert(list, prev, sha1);
 149                 }
 150                 if (!cmp) { /* already exists */
 151                         return l;
 152                 }
 153                 prev = l;
 154                 l = l->next;
 155         }
 156         /* insert at the end */
 157         return llist_insert_back(list, sha1);
 158 }
 159
 160 /* returns a pointer to an item in front of sha1 */
 161 static inline struct llist_item * llist_sorted_remove(struct llist *list, const unsigned char *sha1, struct llist_item *hint)
 162 {
 163         struct llist_item *prev, *l;
 164
 165 redo_from_start:
 166         l = (hint == NULL) ? list->front : hint;
 167         prev = NULL;
 168         while (l) {
 169                 int cmp = hashcmp(l->sha1, sha1);
 170                 if (cmp > 0) /* not in list, since sorted */
 171                         return prev;
 172                 if (!cmp) { /* found */
 173                         if (prev == NULL) {
 174                                 if (hint != NULL && hint != list->front) {
 175                                         /* we don't know the previous element */
 176                                         hint = NULL;
 177                                         goto redo_from_start;
 178                                 }
 179                                 list->front = l->next;
 180                         } else
 181                                 prev->next = l->next;
 182                         if (l == list->back)
 183                                 list->back = prev;
 184                         llist_item_put(l);
 185                         list->size--;
 186                         return prev;
 187                 }
 188                 prev = l;
 189                 l = l->next;
 190         }
 191         return prev;
 192 }
 193
 194 /* computes A\B */
 195 static void llist_sorted_difference_inplace(struct llist *A,
 196                                      struct llist *B)
 197 {
 198         struct llist_item *hint, *b;
 199
 200         hint = NULL;
 201         b = B->front;
 202
 203         while (b) {
 204                 hint = llist_sorted_remove(A, b->sha1, hint);
 205                 b = b->next;
 206         }
 207 }
 208
 209 static inline struct pack_list * pack_list_insert(struct pack_list **pl,
 210                                            struct pack_list *entry)
 211 {
 212         struct pack_list *p = xmalloc(sizeof(struct pack_list));
 213         memcpy(p, entry, sizeof(struct pack_list));
 214         p->next = *pl;
 215         *pl = p;
 216         return p;
 217 }
 218
 219 static inline size_t pack_list_size(struct pack_list *pl)
 220 {
 221         size_t ret = 0;
 222         while (pl) {
 223                 ret++;
 224                 pl = pl->next;
 225         }
 226         return ret;
 227 }
 228
 229 static struct pack_list * pack_list_difference(const struct pack_list *A,
 230                                                const struct pack_list *B)
 231 {
 232         struct pack_list *ret;
 233         const struct pack_list *pl;
 234
 235         if (A == NULL)
 236                 return NULL;
 237
 238         pl = B;
 239         while (pl != NULL) {
 240                 if (A->pack == pl->pack)
 241                         return pack_list_difference(A->next, B);
 242                 pl = pl->next;
 243         }
 244         ret = xmalloc(sizeof(struct pack_list));
 245         memcpy(ret, A, sizeof(struct pack_list));
 246         ret->next = pack_list_difference(A->next, B);
 247         return ret;
 248 }
 249
 250 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 251 {
 252         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 253         const unsigned char *p1_base, *p2_base;
 254         struct llist_item *p1_hint = NULL, *p2_hint = NULL;
 255         const unsigned int hashsz = the_hash_algo->rawsz;
 256
 257         p1_base = p1->pack->index_data;
 258         p2_base = p2->pack->index_data;
 259         p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
 260         p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
 261         p1_step = hashsz + ((p1->pack->index_version < 2) ? 4 : 0);
 262         p2_step = hashsz + ((p2->pack->index_version < 2) ? 4 : 0);
 263
 264         while (p1_off < p1->pack->num_objects * p1_step &&
 265                p2_off < p2->pack->num_objects * p2_step)
 266         {
 267                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 268                 /* cmp ~ p1 - p2 */
 269                 if (cmp == 0) {
 270                         p1_hint = llist_sorted_remove(p1->unique_objects,
 271                                         p1_base + p1_off, p1_hint);
 272                         p2_hint = llist_sorted_remove(p2->unique_objects,
 273                                         p1_base + p1_off, p2_hint);
 274                         p1_off += p1_step;
 275                         p2_off += p2_step;
 276                         continue;
 277                 }
 278                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 279                         p1_off += p1_step;
 280                 } else { /* p2 has the object, p1 doesn't */
 281                         p2_off += p2_step;
 282                 }
 283         }
 284 }
 285
 286 static void pll_free(struct pll *l)
 287 {
 288         struct pll *old;
 289         struct pack_list *opl;
 290
 291         while (l) {
 292                 old = l;
 293                 while (l->pl) {
 294                         opl = l->pl;
 295                         l->pl = opl->next;
 296                         free(opl);
 297                 }
 298                 l = l->next;
 299                 free(old);
 300         }
 301 }
 302
 303 /* all the permutations have to be free()d at the same time,
 304  * since they refer to each other
 305  */
 306 static struct pll * get_permutations(struct pack_list *list, int n)
 307 {
 308         struct pll *subset, *ret = NULL, *new_pll = NULL;
 309
 310         if (list == NULL || pack_list_size(list) < n || n == 0)
 311                 return NULL;
 312
 313         if (n == 1) {
 314                 while (list) {
 315                         new_pll = xmalloc(sizeof(*new_pll));
 316                         new_pll->pl = NULL;
 317                         pack_list_insert(&new_pll->pl, list);
 318                         new_pll->next = ret;
 319                         ret = new_pll;
 320                         list = list->next;
 321                 }
 322                 return ret;
 323         }
 324
 325         while (list->next) {
 326                 subset = get_permutations(list->next, n - 1);
 327                 while (subset) {
 328                         new_pll = xmalloc(sizeof(*new_pll));
 329                         new_pll->pl = subset->pl;
 330                         pack_list_insert(&new_pll->pl, list);
 331                         new_pll->next = ret;
 332                         ret = new_pll;
 333                         subset = subset->next;
 334                 }
 335                 list = list->next;
 336         }
 337         return ret;
 338 }
 339
 340 static int is_superset(struct pack_list *pl, struct llist *list)
 341 {
 342         struct llist *diff;
 343
 344         diff = llist_copy(list);
 345
 346         while (pl) {
 347                 llist_sorted_difference_inplace(diff, pl->all_objects);
 348                 if (diff->size == 0) { /* we're done */
 349                         llist_free(diff);
 350                         return 1;
 351                 }
 352                 pl = pl->next;
 353         }
 354         llist_free(diff);
 355         return 0;
 356 }
 357
 358 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 359 {
 360         size_t ret = 0;
 361         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 362         const unsigned char *p1_base, *p2_base;
 363         const unsigned int hashsz = the_hash_algo->rawsz;
 364
 365         p1_base = p1->index_data;
 366         p2_base = p2->index_data;
 367         p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
 368         p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
 369         p1_step = hashsz + ((p1->index_version < 2) ? 4 : 0);
 370         p2_step = hashsz + ((p2->index_version < 2) ? 4 : 0);
 371
 372         while (p1_off < p1->num_objects * p1_step &&
 373                p2_off < p2->num_objects * p2_step)
 374         {
 375                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 376                 /* cmp ~ p1 - p2 */
 377                 if (cmp == 0) {
 378                         ret++;
 379                         p1_off += p1_step;
 380                         p2_off += p2_step;
 381                         continue;
 382                 }
 383                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 384                         p1_off += p1_step;
 385                 } else { /* p2 has the object, p1 doesn't */
 386                         p2_off += p2_step;
 387                 }
 388         }
 389         return ret;
 390 }
 391
 392 /* another O(n^2) function ... */
 393 static size_t get_pack_redundancy(struct pack_list *pl)
 394 {
 395         struct pack_list *subset;
 396         size_t ret = 0;
 397
 398         if (pl == NULL)
 399                 return 0;
 400
 401         while ((subset = pl->next)) {
 402                 while (subset) {
 403                         ret += sizeof_union(pl->pack, subset->pack);
 404                         subset = subset->next;
 405                 }
 406                 pl = pl->next;
 407         }
 408         return ret;
 409 }
 410
 411 static inline off_t pack_set_bytecount(struct pack_list *pl)
 412 {
 413         off_t ret = 0;
 414         while (pl) {
 415                 ret += pl->pack->pack_size;
 416                 ret += pl->pack->index_size;
 417                 pl = pl->next;
 418         }
 419         return ret;
 420 }
 421
 422 static void minimize(struct pack_list **min)
 423 {
 424         struct pack_list *pl, *unique = NULL,
 425                 *non_unique = NULL, *min_perm = NULL;
 426         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 427         struct llist *missing;
 428         off_t min_perm_size = 0, perm_size;
 429         int n;
 430
 431         pl = local_packs;
 432         while (pl) {
 433                 if (pl->unique_objects->size)
 434                         pack_list_insert(&unique, pl);
 435                 else
 436                         pack_list_insert(&non_unique, pl);
 437                 pl = pl->next;
 438         }
 439         /* find out which objects are missing from the set of unique packs */
 440         missing = llist_copy(all_objects);
 441         pl = unique;
 442         while (pl) {
 443                 llist_sorted_difference_inplace(missing, pl->all_objects);
 444                 pl = pl->next;
 445         }
 446
 447         /* return if there are no objects missing from the unique set */
 448         if (missing->size == 0) {
 449                 *min = unique;
 450                 free(missing);
 451                 return;
 452         }
 453
 454         /* find the permutations which contain all missing objects */
 455         for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
 456                 perm_all = perm = get_permutations(non_unique, n);
 457                 while (perm) {
 458                         if (is_superset(perm->pl, missing)) {
 459                                 new_perm = xmalloc(sizeof(struct pll));
 460                                 memcpy(new_perm, perm, sizeof(struct pll));
 461                                 new_perm->next = perm_ok;
 462                                 perm_ok = new_perm;
 463                         }
 464                         perm = perm->next;
 465                 }
 466                 if (perm_ok)
 467                         break;
 468                 pll_free(perm_all);
 469         }
 470         if (perm_ok == NULL)
 471                 die("Internal error: No complete sets found!");
 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 || 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                 llist_sorted_difference_inplace(all_objects, alt->all_objects);
 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         unsigned long off = 0, step;
 550         const unsigned char *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         if (open_pack_index(p))
 559                 return NULL;
 560
 561         base = p->index_data;
 562         base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
 563         step = the_hash_algo->rawsz + ((p->index_version < 2) ? 4 : 0);
 564         while (off < p->num_objects * step) {
 565                 llist_insert_back(l.all_objects, base + off);
 566                 off += step;
 567         }
 568         /* this list will be pruned in cmp_two_packs later */
 569         l.unique_objects = llist_copy(l.all_objects);
 570         if (p->pack_local)
 571                 return pack_list_insert(&local_packs, &l);
 572         else
 573                 return pack_list_insert(&altodb_packs, &l);
 574 }
 575
 576 static struct pack_list * add_pack_file(const char *filename)
 577 {
 578         struct packed_git *p = get_packed_git(the_repository);
 579
 580         if (strlen(filename) < 40)
 581                 die("Bad pack filename: %s", filename);
 582
 583         while (p) {
 584                 if (strstr(p->pack_name, filename))
 585                         return add_pack(p);
 586                 p = p->next;
 587         }
 588         die("Filename %s not found in packed_git", filename);
 589 }
 590
 591 static void load_all(void)
 592 {
 593         struct packed_git *p = get_packed_git(the_repository);
 594
 595         while (p) {
 596                 add_pack(p);
 597                 p = p->next;
 598         }
 599 }
 600
 601 int cmd_pack_redundant(int argc, const char **argv, const char *prefix)
 602 {
 603         int i;
 604         struct pack_list *min, *red, *pl;
 605         struct llist *ignore;
 606         unsigned char *sha1;
 607         char buf[42]; /* 40 byte sha1 + \n + \0 */
 608
 609         if (argc == 2 && !strcmp(argv[1], "-h"))
 610                 usage(pack_redundant_usage);
 611
 612         for (i = 1; i < argc; i++) {
 613                 const char *arg = argv[i];
 614                 if (!strcmp(arg, "--")) {
 615                         i++;
 616                         break;
 617                 }
 618                 if (!strcmp(arg, "--all")) {
 619                         load_all_packs = 1;
 620                         continue;
 621                 }
 622                 if (!strcmp(arg, "--verbose")) {
 623                         verbose = 1;
 624                         continue;
 625                 }
 626                 if (!strcmp(arg, "--alt-odb")) {
 627                         alt_odb = 1;
 628                         continue;
 629                 }
 630                 if (*arg == '-')
 631                         usage(pack_redundant_usage);
 632                 else
 633                         break;
 634         }
 635
 636         if (load_all_packs)
 637                 load_all();
 638         else
 639                 while (*(argv + i) != NULL)
 640                         add_pack_file(*(argv + i++));
 641
 642         if (local_packs == NULL)
 643                 die("Zero packs found!");
 644
 645         load_all_objects();
 646
 647         cmp_local_packs();
 648         if (alt_odb)
 649                 scan_alt_odb_packs();
 650
 651         /* ignore objects given on stdin */
 652         llist_init(&ignore);
 653         if (!isatty(0)) {
 654                 while (fgets(buf, sizeof(buf), stdin)) {
 655                         sha1 = xmalloc(20);
 656                         if (get_sha1_hex(buf, sha1))
 657                                 die("Bad sha1 on stdin: %s", buf);
 658                         llist_insert_sorted_unique(ignore, sha1, NULL);
 659                 }
 660         }
 661         llist_sorted_difference_inplace(all_objects, ignore);
 662         pl = local_packs;
 663         while (pl) {
 664                 llist_sorted_difference_inplace(pl->unique_objects, ignore);
 665                 pl = pl->next;
 666         }
 667
 668         minimize(&min);
 669
 670         if (verbose) {
 671                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 672                         (unsigned long)pack_list_size(altodb_packs));
 673                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 674                 pl = min;
 675                 while (pl) {
 676                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 677                         pl = pl->next;
 678                 }
 679                 fprintf(stderr, "containing %lu duplicate objects "
 680                                 "with a total size of %lukb.\n",
 681                         (unsigned long)get_pack_redundancy(min),
 682                         (unsigned long)pack_set_bytecount(min)/1024);
 683                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 684                         (unsigned long)all_objects->size);
 685                 fprintf(stderr, "Redundant packs (with indexes):\n");
 686         }
 687         pl = red = pack_list_difference(local_packs, min);
 688         while (pl) {
 689                 printf("%s\n%s\n",
 690                        sha1_pack_index_name(pl->pack->sha1),
 691                        pl->pack->pack_name);
 692                 pl = pl->next;
 693         }
 694         if (verbose)
 695                 fprintf(stderr, "%luMB of redundant packs in total.\n",
 696                         (unsigned long)pack_set_bytecount(red)/(1024*1024));
 697
 698         return 0;
 699 }