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 struct object_id *oid;
  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->oid = list->front->oid;
  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->oid = old_item->oid;
 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 struct object_id *oid)
 112 {
 113         struct llist_item *new_item = llist_item_get();
 114         new_item->oid = oid;
 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 struct object_id *oid)
 135 {
 136         return llist_insert(list, list->back, oid);
 137 }
 138
 139 static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
 140                         const struct object_id *oid, 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 = oidcmp(l->oid, oid);
 147                 if (cmp > 0) { /* we insert before this entry */
 148                         return llist_insert(list, prev, oid);
 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, oid);
 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 struct object_id *oid, 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 = oidcmp(l->oid, oid);
 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->oid, 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                                         (const struct object_id *)(p1_base + p1_off),
 272                                         p1_hint);
 273                         p2_hint = llist_sorted_remove(p2->unique_objects,
 274                                         (const struct object_id *)(p1_base + p1_off),
 275                                         p2_hint);
 276                         p1_off += p1_step;
 277                         p2_off += p2_step;
 278                         continue;
 279                 }
 280                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 281                         p1_off += p1_step;
 282                 } else { /* p2 has the object, p1 doesn't */
 283                         p2_off += p2_step;
 284                 }
 285         }
 286 }
 287
 288 static void pll_free(struct pll *l)
 289 {
 290         struct pll *old;
 291         struct pack_list *opl;
 292
 293         while (l) {
 294                 old = l;
 295                 while (l->pl) {
 296                         opl = l->pl;
 297                         l->pl = opl->next;
 298                         free(opl);
 299                 }
 300                 l = l->next;
 301                 free(old);
 302         }
 303 }
 304
 305 /* all the permutations have to be free()d at the same time,
 306  * since they refer to each other
 307  */
 308 static struct pll * get_permutations(struct pack_list *list, int n)
 309 {
 310         struct pll *subset, *ret = NULL, *new_pll = NULL;
 311
 312         if (list == NULL || pack_list_size(list) < n || n == 0)
 313                 return NULL;
 314
 315         if (n == 1) {
 316                 while (list) {
 317                         new_pll = xmalloc(sizeof(*new_pll));
 318                         new_pll->pl = NULL;
 319                         pack_list_insert(&new_pll->pl, list);
 320                         new_pll->next = ret;
 321                         ret = new_pll;
 322                         list = list->next;
 323                 }
 324                 return ret;
 325         }
 326
 327         while (list->next) {
 328                 subset = get_permutations(list->next, n - 1);
 329                 while (subset) {
 330                         new_pll = xmalloc(sizeof(*new_pll));
 331                         new_pll->pl = subset->pl;
 332                         pack_list_insert(&new_pll->pl, list);
 333                         new_pll->next = ret;
 334                         ret = new_pll;
 335                         subset = subset->next;
 336                 }
 337                 list = list->next;
 338         }
 339         return ret;
 340 }
 341
 342 static int is_superset(struct pack_list *pl, struct llist *list)
 343 {
 344         struct llist *diff;
 345
 346         diff = llist_copy(list);
 347
 348         while (pl) {
 349                 llist_sorted_difference_inplace(diff, pl->all_objects);
 350                 if (diff->size == 0) { /* we're done */
 351                         llist_free(diff);
 352                         return 1;
 353                 }
 354                 pl = pl->next;
 355         }
 356         llist_free(diff);
 357         return 0;
 358 }
 359
 360 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 361 {
 362         size_t ret = 0;
 363         unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
 364         const unsigned char *p1_base, *p2_base;
 365         const unsigned int hashsz = the_hash_algo->rawsz;
 366
 367         p1_base = p1->index_data;
 368         p2_base = p2->index_data;
 369         p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
 370         p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
 371         p1_step = hashsz + ((p1->index_version < 2) ? 4 : 0);
 372         p2_step = hashsz + ((p2->index_version < 2) ? 4 : 0);
 373
 374         while (p1_off < p1->num_objects * p1_step &&
 375                p2_off < p2->num_objects * p2_step)
 376         {
 377                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 378                 /* cmp ~ p1 - p2 */
 379                 if (cmp == 0) {
 380                         ret++;
 381                         p1_off += p1_step;
 382                         p2_off += p2_step;
 383                         continue;
 384                 }
 385                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 386                         p1_off += p1_step;
 387                 } else { /* p2 has the object, p1 doesn't */
 388                         p2_off += p2_step;
 389                 }
 390         }
 391         return ret;
 392 }
 393
 394 /* another O(n^2) function ... */
 395 static size_t get_pack_redundancy(struct pack_list *pl)
 396 {
 397         struct pack_list *subset;
 398         size_t ret = 0;
 399
 400         if (pl == NULL)
 401                 return 0;
 402
 403         while ((subset = pl->next)) {
 404                 while (subset) {
 405                         ret += sizeof_union(pl->pack, subset->pack);
 406                         subset = subset->next;
 407                 }
 408                 pl = pl->next;
 409         }
 410         return ret;
 411 }
 412
 413 static inline off_t pack_set_bytecount(struct pack_list *pl)
 414 {
 415         off_t ret = 0;
 416         while (pl) {
 417                 ret += pl->pack->pack_size;
 418                 ret += pl->pack->index_size;
 419                 pl = pl->next;
 420         }
 421         return ret;
 422 }
 423
 424 static void minimize(struct pack_list **min)
 425 {
 426         struct pack_list *pl, *unique = NULL,
 427                 *non_unique = NULL, *min_perm = NULL;
 428         struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
 429         struct llist *missing;
 430         off_t min_perm_size = 0, perm_size;
 431         int n;
 432
 433         pl = local_packs;
 434         while (pl) {
 435                 if (pl->unique_objects->size)
 436                         pack_list_insert(&unique, pl);
 437                 else
 438                         pack_list_insert(&non_unique, pl);
 439                 pl = pl->next;
 440         }
 441         /* find out which objects are missing from the set of unique packs */
 442         missing = llist_copy(all_objects);
 443         pl = unique;
 444         while (pl) {
 445                 llist_sorted_difference_inplace(missing, pl->all_objects);
 446                 pl = pl->next;
 447         }
 448
 449         /* return if there are no objects missing from the unique set */
 450         if (missing->size == 0) {
 451                 *min = unique;
 452                 free(missing);
 453                 return;
 454         }
 455
 456         /* find the permutations which contain all missing objects */
 457         for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
 458                 perm_all = perm = get_permutations(non_unique, n);
 459                 while (perm) {
 460                         if (is_superset(perm->pl, missing)) {
 461                                 new_perm = xmalloc(sizeof(struct pll));
 462                                 memcpy(new_perm, perm, sizeof(struct pll));
 463                                 new_perm->next = perm_ok;
 464                                 perm_ok = new_perm;
 465                         }
 466                         perm = perm->next;
 467                 }
 468                 if (perm_ok)
 469                         break;
 470                 pll_free(perm_all);
 471         }
 472         if (perm_ok == NULL)
 473                 die("Internal error: No complete sets found!");
 474
 475         /* find the permutation with the smallest size */
 476         perm = perm_ok;
 477         while (perm) {
 478                 perm_size = pack_set_bytecount(perm->pl);
 479                 if (!min_perm_size || min_perm_size > perm_size) {
 480                         min_perm_size = perm_size;
 481                         min_perm = perm->pl;
 482                 }
 483                 perm = perm->next;
 484         }
 485         *min = min_perm;
 486         /* add the unique packs to the list */
 487         pl = unique;
 488         while (pl) {
 489                 pack_list_insert(min, pl);
 490                 pl = pl->next;
 491         }
 492 }
 493
 494 static void load_all_objects(void)
 495 {
 496         struct pack_list *pl = local_packs;
 497         struct llist_item *hint, *l;
 498
 499         llist_init(&all_objects);
 500
 501         while (pl) {
 502                 hint = NULL;
 503                 l = pl->all_objects->front;
 504                 while (l) {
 505                         hint = llist_insert_sorted_unique(all_objects,
 506                                                           l->oid, hint);
 507                         l = l->next;
 508                 }
 509                 pl = pl->next;
 510         }
 511         /* remove objects present in remote packs */
 512         pl = altodb_packs;
 513         while (pl) {
 514                 llist_sorted_difference_inplace(all_objects, pl->all_objects);
 515                 pl = pl->next;
 516         }
 517 }
 518
 519 /* this scales like O(n^2) */
 520 static void cmp_local_packs(void)
 521 {
 522         struct pack_list *subset, *pl = local_packs;
 523
 524         while ((subset = pl)) {
 525                 while ((subset = subset->next))
 526                         cmp_two_packs(pl, subset);
 527                 pl = pl->next;
 528         }
 529 }
 530
 531 static void scan_alt_odb_packs(void)
 532 {
 533         struct pack_list *local, *alt;
 534
 535         alt = altodb_packs;
 536         while (alt) {
 537                 local = local_packs;
 538                 while (local) {
 539                         llist_sorted_difference_inplace(local->unique_objects,
 540                                                         alt->all_objects);
 541                         local = local->next;
 542                 }
 543                 llist_sorted_difference_inplace(all_objects, alt->all_objects);
 544                 alt = alt->next;
 545         }
 546 }
 547
 548 static struct pack_list * add_pack(struct packed_git *p)
 549 {
 550         struct pack_list l;
 551         unsigned long off = 0, step;
 552         const unsigned char *base;
 553
 554         if (!p->pack_local && !(alt_odb || verbose))
 555                 return NULL;
 556
 557         l.pack = p;
 558         llist_init(&l.all_objects);
 559
 560         if (open_pack_index(p))
 561                 return NULL;
 562
 563         base = p->index_data;
 564         base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
 565         step = the_hash_algo->rawsz + ((p->index_version < 2) ? 4 : 0);
 566         while (off < p->num_objects * step) {
 567                 llist_insert_back(l.all_objects, (const struct object_id *)(base + off));
 568                 off += step;
 569         }
 570         /* this list will be pruned in cmp_two_packs later */
 571         l.unique_objects = llist_copy(l.all_objects);
 572         if (p->pack_local)
 573                 return pack_list_insert(&local_packs, &l);
 574         else
 575                 return pack_list_insert(&altodb_packs, &l);
 576 }
 577
 578 static struct pack_list * add_pack_file(const char *filename)
 579 {
 580         struct packed_git *p = get_packed_git(the_repository);
 581
 582         if (strlen(filename) < 40)
 583                 die("Bad pack filename: %s", filename);
 584
 585         while (p) {
 586                 if (strstr(p->pack_name, filename))
 587                         return add_pack(p);
 588                 p = p->next;
 589         }
 590         die("Filename %s not found in packed_git", filename);
 591 }
 592
 593 static void load_all(void)
 594 {
 595         struct packed_git *p = get_packed_git(the_repository);
 596
 597         while (p) {
 598                 add_pack(p);
 599                 p = p->next;
 600         }
 601 }
 602
 603 int cmd_pack_redundant(int argc, const char **argv, const char *prefix)
 604 {
 605         int i;
 606         struct pack_list *min, *red, *pl;
 607         struct llist *ignore;
 608         struct object_id *oid;
 609         char buf[GIT_MAX_HEXSZ + 2]; /* hex hash + \n + \0 */
 610
 611         if (argc == 2 && !strcmp(argv[1], "-h"))
 612                 usage(pack_redundant_usage);
 613
 614         for (i = 1; i < argc; i++) {
 615                 const char *arg = argv[i];
 616                 if (!strcmp(arg, "--")) {
 617                         i++;
 618                         break;
 619                 }
 620                 if (!strcmp(arg, "--all")) {
 621                         load_all_packs = 1;
 622                         continue;
 623                 }
 624                 if (!strcmp(arg, "--verbose")) {
 625                         verbose = 1;
 626                         continue;
 627                 }
 628                 if (!strcmp(arg, "--alt-odb")) {
 629                         alt_odb = 1;
 630                         continue;
 631                 }
 632                 if (*arg == '-')
 633                         usage(pack_redundant_usage);
 634                 else
 635                         break;
 636         }
 637
 638         if (load_all_packs)
 639                 load_all();
 640         else
 641                 while (*(argv + i) != NULL)
 642                         add_pack_file(*(argv + i++));
 643
 644         if (local_packs == NULL)
 645                 die("Zero packs found!");
 646
 647         load_all_objects();
 648
 649         cmp_local_packs();
 650         if (alt_odb)
 651                 scan_alt_odb_packs();
 652
 653         /* ignore objects given on stdin */
 654         llist_init(&ignore);
 655         if (!isatty(0)) {
 656                 while (fgets(buf, sizeof(buf), stdin)) {
 657                         oid = xmalloc(sizeof(*oid));
 658                         if (get_oid_hex(buf, oid))
 659                                 die("Bad object ID on stdin: %s", buf);
 660                         llist_insert_sorted_unique(ignore, oid, NULL);
 661                 }
 662         }
 663         llist_sorted_difference_inplace(all_objects, ignore);
 664         pl = local_packs;
 665         while (pl) {
 666                 llist_sorted_difference_inplace(pl->unique_objects, ignore);
 667                 pl = pl->next;
 668         }
 669
 670         minimize(&min);
 671
 672         if (verbose) {
 673                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 674                         (unsigned long)pack_list_size(altodb_packs));
 675                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 676                 pl = min;
 677                 while (pl) {
 678                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 679                         pl = pl->next;
 680                 }
 681                 fprintf(stderr, "containing %lu duplicate objects "
 682                                 "with a total size of %lukb.\n",
 683                         (unsigned long)get_pack_redundancy(min),
 684                         (unsigned long)pack_set_bytecount(min)/1024);
 685                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 686                         (unsigned long)all_objects->size);
 687                 fprintf(stderr, "Redundant packs (with indexes):\n");
 688         }
 689         pl = red = pack_list_difference(local_packs, min);
 690         while (pl) {
 691                 printf("%s\n%s\n",
 692                        sha1_pack_index_name(pl->pack->sha1),
 693                        pl->pack->pack_name);
 694                 pl = pl->next;
 695         }
 696         if (verbose)
 697                 fprintf(stderr, "%luMB of redundant packs in total.\n",
 698                         (unsigned long)pack_set_bytecount(red)/(1024*1024));
 699
 700         return 0;
 701 }