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
 256         p1_base = p1->pack->index_data;
 257         p2_base = p2->pack->index_data;
 258         p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
 259         p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
 260         p1_step = (p1->pack->index_version < 2) ? 24 : 20;
 261         p2_step = (p2->pack->index_version < 2) ? 24 : 20;
 262
 263         while (p1_off < p1->pack->num_objects * p1_step &&
 264                p2_off < p2->pack->num_objects * p2_step)
 265         {
 266                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 267                 /* cmp ~ p1 - p2 */
 268                 if (cmp == 0) {
 269                         p1_hint = llist_sorted_remove(p1->unique_objects,
 270                                         p1_base + p1_off, p1_hint);
 271                         p2_hint = llist_sorted_remove(p2->unique_objects,
 272                                         p1_base + p1_off, p2_hint);
 273                         p1_off += p1_step;
 274                         p2_off += p2_step;
 275                         continue;
 276                 }
 277                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 278                         p1_off += p1_step;
 279                 } else { /* p2 has the object, p1 doesn't */
 280                         p2_off += p2_step;
 281                 }
 282         }
 283 }
 284
 285 static void pll_free(struct pll *l)
 286 {
 287         struct pll *old;
 288         struct pack_list *opl;
 289
 290         while (l) {
 291                 old = l;
 292                 while (l->pl) {
 293                         opl = l->pl;
 294                         l->pl = opl->next;
 295                         free(opl);
 296                 }
 297                 l = l->next;
 298                 free(old);
 299         }
 300 }
 301
 302 /* all the permutations have to be free()d at the same time,
 303  * since they refer to each other
 304  */
 305 static struct pll * get_permutations(struct pack_list *list, int n)
 306 {
 307         struct pll *subset, *ret = NULL, *new_pll = NULL;
 308
 309         if (list == NULL || pack_list_size(list) < n || n == 0)
 310                 return NULL;
 311
 312         if (n == 1) {
 313                 while (list) {
 314                         new_pll = xmalloc(sizeof(*new_pll));
 315                         new_pll->pl = NULL;
 316                         pack_list_insert(&new_pll->pl, list);
 317                         new_pll->next = ret;
 318                         ret = new_pll;
 319                         list = list->next;
 320                 }
 321                 return ret;
 322         }
 323
 324         while (list->next) {
 325                 subset = get_permutations(list->next, n - 1);
 326                 while (subset) {
 327                         new_pll = xmalloc(sizeof(*new_pll));
 328                         new_pll->pl = subset->pl;
 329                         pack_list_insert(&new_pll->pl, list);
 330                         new_pll->next = ret;
 331                         ret = new_pll;
 332                         subset = subset->next;
 333                 }
 334                 list = list->next;
 335         }
 336         return ret;
 337 }
 338
 339 static int is_superset(struct pack_list *pl, struct llist *list)
 340 {
 341         struct llist *diff;
 342
 343         diff = llist_copy(list);
 344
 345         while (pl) {
 346                 llist_sorted_difference_inplace(diff, pl->all_objects);
 347                 if (diff->size == 0) { /* we're done */
 348                         llist_free(diff);
 349                         return 1;
 350                 }
 351                 pl = pl->next;
 352         }
 353         llist_free(diff);
 354         return 0;
 355 }
 356
 357 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 358 {
 359         size_t ret = 0;
 360         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 361         const unsigned char *p1_base, *p2_base;
 362
 363         p1_base = p1->index_data;
 364         p2_base = p2->index_data;
 365         p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
 366         p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
 367         p1_step = (p1->index_version < 2) ? 24 : 20;
 368         p2_step = (p2->index_version < 2) ? 24 : 20;
 369
 370         while (p1_off < p1->num_objects * p1_step &&
 371                p2_off < p2->num_objects * p2_step)
 372         {
 373                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 374                 /* cmp ~ p1 - p2 */
 375                 if (cmp == 0) {
 376                         ret++;
 377                         p1_off += p1_step;
 378                         p2_off += p2_step;
 379                         continue;
 380                 }
 381                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 382                         p1_off += p1_step;
 383                 } else { /* p2 has the object, p1 doesn't */
 384                         p2_off += p2_step;
 385                 }
 386         }
 387         return ret;
 388 }
 389
 390 /* another O(n^2) function ... */
 391 static size_t get_pack_redundancy(struct pack_list *pl)
 392 {
 393         struct pack_list *subset;
 394         size_t ret = 0;
 395
 396         if (pl == NULL)
 397                 return 0;
 398
 399         while ((subset = pl->next)) {
 400                 while (subset) {
 401                         ret += sizeof_union(pl->pack, subset->pack);
 402                         subset = subset->next;
 403                 }
 404                 pl = pl->next;
 405         }
 406         return ret;
 407 }
 408
 409 static inline off_t pack_set_bytecount(struct pack_list *pl)
 410 {
 411         off_t ret = 0;
 412         while (pl) {
 413                 ret += pl->pack->pack_size;
 414                 ret += pl->pack->index_size;
 415                 pl = pl->next;
 416         }
 417         return ret;
 418 }
 419
 420 static void minimize(struct pack_list **min)
 421 {
 422         struct pack_list *pl, *unique = NULL,
 423                 *non_unique = NULL, *min_perm = NULL;
 424         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 425         struct llist *missing;
 426         off_t min_perm_size = 0, perm_size;
 427         int n;
 428
 429         pl = local_packs;
 430         while (pl) {
 431                 if (pl->unique_objects->size)
 432                         pack_list_insert(&unique, pl);
 433                 else
 434                         pack_list_insert(&non_unique, pl);
 435                 pl = pl->next;
 436         }
 437         /* find out which objects are missing from the set of unique packs */
 438         missing = llist_copy(all_objects);
 439         pl = unique;
 440         while (pl) {
 441                 llist_sorted_difference_inplace(missing, pl->all_objects);
 442                 pl = pl->next;
 443         }
 444
 445         /* return if there are no objects missing from the unique set */
 446         if (missing->size == 0) {
 447                 *min = unique;
 448                 free(missing);
 449                 return;
 450         }
 451
 452         /* find the permutations which contain all missing objects */
 453         for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
 454                 perm_all = perm = get_permutations(non_unique, n);
 455                 while (perm) {
 456                         if (is_superset(perm->pl, missing)) {
 457                                 new_perm = xmalloc(sizeof(struct pll));
 458                                 memcpy(new_perm, perm, sizeof(struct pll));
 459                                 new_perm->next = perm_ok;
 460                                 perm_ok = new_perm;
 461                         }
 462                         perm = perm->next;
 463                 }
 464                 if (perm_ok)
 465                         break;
 466                 pll_free(perm_all);
 467         }
 468         if (perm_ok == NULL)
 469                 die("Internal error: No complete sets found!");
 470
 471         /* find the permutation with the smallest size */
 472         perm = perm_ok;
 473         while (perm) {
 474                 perm_size = pack_set_bytecount(perm->pl);
 475                 if (!min_perm_size || min_perm_size > perm_size) {
 476                         min_perm_size = perm_size;
 477                         min_perm = perm->pl;
 478                 }
 479                 perm = perm->next;
 480         }
 481         *min = min_perm;
 482         /* add the unique packs to the list */
 483         pl = unique;
 484         while (pl) {
 485                 pack_list_insert(min, pl);
 486                 pl = pl->next;
 487         }
 488 }
 489
 490 static void load_all_objects(void)
 491 {
 492         struct pack_list *pl = local_packs;
 493         struct llist_item *hint, *l;
 494
 495         llist_init(&all_objects);
 496
 497         while (pl) {
 498                 hint = NULL;
 499                 l = pl->all_objects->front;
 500                 while (l) {
 501                         hint = llist_insert_sorted_unique(all_objects,
 502                                                           l->sha1, hint);
 503                         l = l->next;
 504                 }
 505                 pl = pl->next;
 506         }
 507         /* remove objects present in remote packs */
 508         pl = altodb_packs;
 509         while (pl) {
 510                 llist_sorted_difference_inplace(all_objects, pl->all_objects);
 511                 pl = pl->next;
 512         }
 513 }
 514
 515 /* this scales like O(n^2) */
 516 static void cmp_local_packs(void)
 517 {
 518         struct pack_list *subset, *pl = local_packs;
 519
 520         while ((subset = pl)) {
 521                 while ((subset = subset->next))
 522                         cmp_two_packs(pl, subset);
 523                 pl = pl->next;
 524         }
 525 }
 526
 527 static void scan_alt_odb_packs(void)
 528 {
 529         struct pack_list *local, *alt;
 530
 531         alt = altodb_packs;
 532         while (alt) {
 533                 local = local_packs;
 534                 while (local) {
 535                         llist_sorted_difference_inplace(local->unique_objects,
 536                                                         alt->all_objects);
 537                         local = local->next;
 538                 }
 539                 llist_sorted_difference_inplace(all_objects, alt->all_objects);
 540                 alt = alt->next;
 541         }
 542 }
 543
 544 static struct pack_list * add_pack(struct packed_git *p)
 545 {
 546         struct pack_list l;
 547         unsigned long off = 0, step;
 548         const unsigned char *base;
 549
 550         if (!p->pack_local && !(alt_odb || verbose))
 551                 return NULL;
 552
 553         l.pack = p;
 554         llist_init(&l.all_objects);
 555
 556         if (open_pack_index(p))
 557                 return NULL;
 558
 559         base = p->index_data;
 560         base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
 561         step = (p->index_version < 2) ? 24 : 20;
 562         while (off < p->num_objects * step) {
 563                 llist_insert_back(l.all_objects, base + off);
 564                 off += step;
 565         }
 566         /* this list will be pruned in cmp_two_packs later */
 567         l.unique_objects = llist_copy(l.all_objects);
 568         if (p->pack_local)
 569                 return pack_list_insert(&local_packs, &l);
 570         else
 571                 return pack_list_insert(&altodb_packs, &l);
 572 }
 573
 574 static struct pack_list * add_pack_file(const char *filename)
 575 {
 576         struct packed_git *p = get_packed_git(the_repository);
 577
 578         if (strlen(filename) < 40)
 579                 die("Bad pack filename: %s", filename);
 580
 581         while (p) {
 582                 if (strstr(p->pack_name, filename))
 583                         return add_pack(p);
 584                 p = p->next;
 585         }
 586         die("Filename %s not found in packed_git", filename);
 587 }
 588
 589 static void load_all(void)
 590 {
 591         struct packed_git *p = get_packed_git(the_repository);
 592
 593         while (p) {
 594                 add_pack(p);
 595                 p = p->next;
 596         }
 597 }
 598
 599 int cmd_pack_redundant(int argc, const char **argv, const char *prefix)
 600 {
 601         int i;
 602         struct pack_list *min, *red, *pl;
 603         struct llist *ignore;
 604         unsigned char *sha1;
 605         char buf[42]; /* 40 byte sha1 + \n + \0 */
 606
 607         if (argc == 2 && !strcmp(argv[1], "-h"))
 608                 usage(pack_redundant_usage);
 609
 610         for (i = 1; i < argc; i++) {
 611                 const char *arg = argv[i];
 612                 if (!strcmp(arg, "--")) {
 613                         i++;
 614                         break;
 615                 }
 616                 if (!strcmp(arg, "--all")) {
 617                         load_all_packs = 1;
 618                         continue;
 619                 }
 620                 if (!strcmp(arg, "--verbose")) {
 621                         verbose = 1;
 622                         continue;
 623                 }
 624                 if (!strcmp(arg, "--alt-odb")) {
 625                         alt_odb = 1;
 626                         continue;
 627                 }
 628                 if (*arg == '-')
 629                         usage(pack_redundant_usage);
 630                 else
 631                         break;
 632         }
 633
 634         if (load_all_packs)
 635                 load_all();
 636         else
 637                 while (*(argv + i) != NULL)
 638                         add_pack_file(*(argv + i++));
 639
 640         if (local_packs == NULL)
 641                 die("Zero packs found!");
 642
 643         load_all_objects();
 644
 645         cmp_local_packs();
 646         if (alt_odb)
 647                 scan_alt_odb_packs();
 648
 649         /* ignore objects given on stdin */
 650         llist_init(&ignore);
 651         if (!isatty(0)) {
 652                 while (fgets(buf, sizeof(buf), stdin)) {
 653                         sha1 = xmalloc(20);
 654                         if (get_sha1_hex(buf, sha1))
 655                                 die("Bad sha1 on stdin: %s", buf);
 656                         llist_insert_sorted_unique(ignore, sha1, NULL);
 657                 }
 658         }
 659         llist_sorted_difference_inplace(all_objects, ignore);
 660         pl = local_packs;
 661         while (pl) {
 662                 llist_sorted_difference_inplace(pl->unique_objects, ignore);
 663                 pl = pl->next;
 664         }
 665
 666         minimize(&min);
 667
 668         if (verbose) {
 669                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 670                         (unsigned long)pack_list_size(altodb_packs));
 671                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 672                 pl = min;
 673                 while (pl) {
 674                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 675                         pl = pl->next;
 676                 }
 677                 fprintf(stderr, "containing %lu duplicate objects "
 678                                 "with a total size of %lukb.\n",
 679                         (unsigned long)get_pack_redundancy(min),
 680                         (unsigned long)pack_set_bytecount(min)/1024);
 681                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 682                         (unsigned long)all_objects->size);
 683                 fprintf(stderr, "Redundant packs (with indexes):\n");
 684         }
 685         pl = red = pack_list_difference(local_packs, min);
 686         while (pl) {
 687                 printf("%s\n%s\n",
 688                        sha1_pack_index_name(pl->pack->sha1),
 689                        pl->pack->pack_name);
 690                 pl = pl->next;
 691         }
 692         if (verbose)
 693                 fprintf(stderr, "%luMB of redundant packs in total.\n",
 694                         (unsigned long)pack_set_bytecount(red)/(1024*1024));
 695
 696         return 0;
 697 }