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 "gettext.h"
  11 #include "hex.h"
  12 #include "repository.h"
  13 #include "packfile.h"
  14 #include "object-store-ll.h"
  15
  16 #define BLKSIZE 512
  17
  18 static const char pack_redundant_usage[] =
  19 "git pack-redundant [--verbose] [--alt-odb] (--all | <pack-filename>...)";
  20
  21 static int load_all_packs, verbose, alt_odb;
  22
  23 struct llist_item {
  24         struct llist_item *next;
  25         struct object_id oid;
  26 };
  27 static struct llist {
  28         struct llist_item *front;
  29         struct llist_item *back;
  30         size_t size;
  31 } *all_objects; /* all objects which must be present in local packfiles */
  32
  33 static struct pack_list {
  34         struct pack_list *next;
  35         struct packed_git *pack;
  36         struct llist *unique_objects;
  37         struct llist *remaining_objects;
  38         size_t all_objects_size;
  39 } *local_packs = NULL, *altodb_packs = NULL;
  40
  41 static struct llist_item *free_nodes;
  42
  43 static inline void llist_item_put(struct llist_item *item)
  44 {
  45         item->next = free_nodes;
  46         free_nodes = item;
  47 }
  48
  49 static inline struct llist_item *llist_item_get(void)
  50 {
  51         struct llist_item *new_item;
  52         if ( free_nodes ) {
  53                 new_item = free_nodes;
  54                 free_nodes = free_nodes->next;
  55         } else {
  56                 int i = 1;
  57                 ALLOC_ARRAY(new_item, BLKSIZE);
  58                 for (; i < BLKSIZE; i++)
  59                         llist_item_put(&new_item[i]);
  60         }
  61         return new_item;
  62 }
  63
  64 static inline void llist_init(struct llist **list)
  65 {
  66         *list = xmalloc(sizeof(struct llist));
  67         (*list)->front = (*list)->back = NULL;
  68         (*list)->size = 0;
  69 }
  70
  71 static struct llist * llist_copy(struct llist *list)
  72 {
  73         struct llist *ret;
  74         struct llist_item *new_item, *old_item, *prev;
  75
  76         llist_init(&ret);
  77
  78         if ((ret->size = list->size) == 0)
  79                 return ret;
  80
  81         new_item = ret->front = llist_item_get();
  82         new_item->oid = list->front->oid;
  83
  84         old_item = list->front->next;
  85         while (old_item) {
  86                 prev = new_item;
  87                 new_item = llist_item_get();
  88                 prev->next = new_item;
  89                 new_item->oid = old_item->oid;
  90                 old_item = old_item->next;
  91         }
  92         new_item->next = NULL;
  93         ret->back = new_item;
  94
  95         return ret;
  96 }
  97
  98 static inline struct llist_item *llist_insert(struct llist *list,
  99                                               struct llist_item *after,
 100                                               const unsigned char *oid)
 101 {
 102         struct llist_item *new_item = llist_item_get();
 103         oidread(&new_item->oid, oid);
 104         new_item->next = NULL;
 105
 106         if (after) {
 107                 new_item->next = after->next;
 108                 after->next = new_item;
 109                 if (after == list->back)
 110                         list->back = new_item;
 111         } else {/* insert in front */
 112                 if (list->size == 0)
 113                         list->back = new_item;
 114                 else
 115                         new_item->next = list->front;
 116                 list->front = new_item;
 117         }
 118         list->size++;
 119         return new_item;
 120 }
 121
 122 static inline struct llist_item *llist_insert_back(struct llist *list,
 123                                                    const unsigned char *oid)
 124 {
 125         return llist_insert(list, list->back, oid);
 126 }
 127
 128 static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
 129                         const struct object_id *oid, struct llist_item *hint)
 130 {
 131         struct llist_item *prev = NULL, *l;
 132
 133         l = (hint == NULL) ? list->front : hint;
 134         while (l) {
 135                 int cmp = oidcmp(&l->oid, oid);
 136                 if (cmp > 0) { /* we insert before this entry */
 137                         return llist_insert(list, prev, oid->hash);
 138                 }
 139                 if (!cmp) { /* already exists */
 140                         return l;
 141                 }
 142                 prev = l;
 143                 l = l->next;
 144         }
 145         /* insert at the end */
 146         return llist_insert_back(list, oid->hash);
 147 }
 148
 149 /* returns a pointer to an item in front of sha1 */
 150 static inline struct llist_item * llist_sorted_remove(struct llist *list, const unsigned char *oid, struct llist_item *hint)
 151 {
 152         struct llist_item *prev, *l;
 153
 154 redo_from_start:
 155         l = (hint == NULL) ? list->front : hint;
 156         prev = NULL;
 157         while (l) {
 158                 const int cmp = hashcmp(l->oid.hash, oid);
 159                 if (cmp > 0) /* not in list, since sorted */
 160                         return prev;
 161                 if (!cmp) { /* found */
 162                         if (!prev) {
 163                                 if (hint != NULL && hint != list->front) {
 164                                         /* we don't know the previous element */
 165                                         hint = NULL;
 166                                         goto redo_from_start;
 167                                 }
 168                                 list->front = l->next;
 169                         } else
 170                                 prev->next = l->next;
 171                         if (l == list->back)
 172                                 list->back = prev;
 173                         llist_item_put(l);
 174                         list->size--;
 175                         return prev;
 176                 }
 177                 prev = l;
 178                 l = l->next;
 179         }
 180         return prev;
 181 }
 182
 183 /* computes A\B */
 184 static void llist_sorted_difference_inplace(struct llist *A,
 185                                      struct llist *B)
 186 {
 187         struct llist_item *hint, *b;
 188
 189         hint = NULL;
 190         b = B->front;
 191
 192         while (b) {
 193                 hint = llist_sorted_remove(A, b->oid.hash, hint);
 194                 b = b->next;
 195         }
 196 }
 197
 198 static inline struct pack_list * pack_list_insert(struct pack_list **pl,
 199                                            struct pack_list *entry)
 200 {
 201         struct pack_list *p = xmalloc(sizeof(struct pack_list));
 202         memcpy(p, entry, sizeof(struct pack_list));
 203         p->next = *pl;
 204         *pl = p;
 205         return p;
 206 }
 207
 208 static inline size_t pack_list_size(struct pack_list *pl)
 209 {
 210         size_t ret = 0;
 211         while (pl) {
 212                 ret++;
 213                 pl = pl->next;
 214         }
 215         return ret;
 216 }
 217
 218 static struct pack_list * pack_list_difference(const struct pack_list *A,
 219                                                const struct pack_list *B)
 220 {
 221         struct pack_list *ret;
 222         const struct pack_list *pl;
 223
 224         if (!A)
 225                 return NULL;
 226
 227         pl = B;
 228         while (pl != NULL) {
 229                 if (A->pack == pl->pack)
 230                         return pack_list_difference(A->next, B);
 231                 pl = pl->next;
 232         }
 233         ret = xmalloc(sizeof(struct pack_list));
 234         memcpy(ret, A, sizeof(struct pack_list));
 235         ret->next = pack_list_difference(A->next, B);
 236         return ret;
 237 }
 238
 239 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 240 {
 241         size_t p1_off = 0, p2_off = 0, p1_step, p2_step;
 242         const unsigned char *p1_base, *p2_base;
 243         struct llist_item *p1_hint = NULL, *p2_hint = NULL;
 244         const unsigned int hashsz = the_hash_algo->rawsz;
 245
 246         if (!p1->unique_objects)
 247                 p1->unique_objects = llist_copy(p1->remaining_objects);
 248         if (!p2->unique_objects)
 249                 p2->unique_objects = llist_copy(p2->remaining_objects);
 250
 251         p1_base = p1->pack->index_data;
 252         p2_base = p2->pack->index_data;
 253         p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
 254         p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
 255         p1_step = hashsz + ((p1->pack->index_version < 2) ? 4 : 0);
 256         p2_step = hashsz + ((p2->pack->index_version < 2) ? 4 : 0);
 257
 258         while (p1_off < p1->pack->num_objects * p1_step &&
 259                p2_off < p2->pack->num_objects * p2_step)
 260         {
 261                 const int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 262                 /* cmp ~ p1 - p2 */
 263                 if (cmp == 0) {
 264                         p1_hint = llist_sorted_remove(p1->unique_objects,
 265                                         p1_base + p1_off,
 266                                         p1_hint);
 267                         p2_hint = llist_sorted_remove(p2->unique_objects,
 268                                         p1_base + p1_off,
 269                                         p2_hint);
 270                         p1_off += p1_step;
 271                         p2_off += p2_step;
 272                         continue;
 273                 }
 274                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 275                         p1_off += p1_step;
 276                 } else { /* p2 has the object, p1 doesn't */
 277                         p2_off += p2_step;
 278                 }
 279         }
 280 }
 281
 282 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 283 {
 284         size_t ret = 0;
 285         size_t p1_off = 0, p2_off = 0, p1_step, p2_step;
 286         const unsigned char *p1_base, *p2_base;
 287         const unsigned int hashsz = the_hash_algo->rawsz;
 288
 289         p1_base = p1->index_data;
 290         p2_base = p2->index_data;
 291         p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
 292         p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
 293         p1_step = hashsz + ((p1->index_version < 2) ? 4 : 0);
 294         p2_step = hashsz + ((p2->index_version < 2) ? 4 : 0);
 295
 296         while (p1_off < p1->num_objects * p1_step &&
 297                p2_off < p2->num_objects * p2_step)
 298         {
 299                 int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
 300                 /* cmp ~ p1 - p2 */
 301                 if (cmp == 0) {
 302                         ret++;
 303                         p1_off += p1_step;
 304                         p2_off += p2_step;
 305                         continue;
 306                 }
 307                 if (cmp < 0) { /* p1 has the object, p2 doesn't */
 308                         p1_off += p1_step;
 309                 } else { /* p2 has the object, p1 doesn't */
 310                         p2_off += p2_step;
 311                 }
 312         }
 313         return ret;
 314 }
 315
 316 /* another O(n^2) function ... */
 317 static size_t get_pack_redundancy(struct pack_list *pl)
 318 {
 319         struct pack_list *subset;
 320         size_t ret = 0;
 321
 322         if (!pl)
 323                 return 0;
 324
 325         while ((subset = pl->next)) {
 326                 while (subset) {
 327                         ret += sizeof_union(pl->pack, subset->pack);
 328                         subset = subset->next;
 329                 }
 330                 pl = pl->next;
 331         }
 332         return ret;
 333 }
 334
 335 static inline off_t pack_set_bytecount(struct pack_list *pl)
 336 {
 337         off_t ret = 0;
 338         while (pl) {
 339                 ret += pl->pack->pack_size;
 340                 ret += pl->pack->index_size;
 341                 pl = pl->next;
 342         }
 343         return ret;
 344 }
 345
 346 static int cmp_remaining_objects(const void *a, const void *b)
 347 {
 348         struct pack_list *pl_a = *((struct pack_list **)a);
 349         struct pack_list *pl_b = *((struct pack_list **)b);
 350
 351         if (pl_a->remaining_objects->size == pl_b->remaining_objects->size) {
 352                 /* have the same remaining_objects, big pack first */
 353                 if (pl_a->all_objects_size == pl_b->all_objects_size)
 354                         return 0;
 355                 else if (pl_a->all_objects_size < pl_b->all_objects_size)
 356                         return 1;
 357                 else
 358                         return -1;
 359         } else if (pl_a->remaining_objects->size < pl_b->remaining_objects->size) {
 360                 /* sort by remaining objects, more objects first */
 361                 return 1;
 362         } else {
 363                 return -1;
 364         }
 365 }
 366
 367 /* Sort pack_list, greater size of remaining_objects first */
 368 static void sort_pack_list(struct pack_list **pl)
 369 {
 370         struct pack_list **ary, *p;
 371         int i;
 372         size_t n = pack_list_size(*pl);
 373
 374         if (n < 2)
 375                 return;
 376
 377         /* prepare an array of packed_list for easier sorting */
 378         CALLOC_ARRAY(ary, n);
 379         for (n = 0, p = *pl; p; p = p->next)
 380                 ary[n++] = p;
 381
 382         QSORT(ary, n, cmp_remaining_objects);
 383
 384         /* link them back again */
 385         for (i = 0; i < n - 1; i++)
 386                 ary[i]->next = ary[i + 1];
 387         ary[n - 1]->next = NULL;
 388         *pl = ary[0];
 389
 390         free(ary);
 391 }
 392
 393
 394 static void minimize(struct pack_list **min)
 395 {
 396         struct pack_list *pl, *unique = NULL, *non_unique = NULL;
 397         struct llist *missing, *unique_pack_objects;
 398
 399         pl = local_packs;
 400         while (pl) {
 401                 if (pl->unique_objects->size)
 402                         pack_list_insert(&unique, pl);
 403                 else
 404                         pack_list_insert(&non_unique, pl);
 405                 pl = pl->next;
 406         }
 407         /* find out which objects are missing from the set of unique packs */
 408         missing = llist_copy(all_objects);
 409         pl = unique;
 410         while (pl) {
 411                 llist_sorted_difference_inplace(missing, pl->remaining_objects);
 412                 pl = pl->next;
 413         }
 414
 415         *min = unique;
 416
 417         /* return if there are no objects missing from the unique set */
 418         if (missing->size == 0) {
 419                 free(missing);
 420                 return;
 421         }
 422
 423         unique_pack_objects = llist_copy(all_objects);
 424         llist_sorted_difference_inplace(unique_pack_objects, missing);
 425
 426         /* remove unique pack objects from the non_unique packs */
 427         pl = non_unique;
 428         while (pl) {
 429                 llist_sorted_difference_inplace(pl->remaining_objects, unique_pack_objects);
 430                 pl = pl->next;
 431         }
 432
 433         while (non_unique) {
 434                 /* sort the non_unique packs, greater size of remaining_objects first */
 435                 sort_pack_list(&non_unique);
 436                 if (non_unique->remaining_objects->size == 0)
 437                         break;
 438
 439                 pack_list_insert(min, non_unique);
 440
 441                 for (pl = non_unique->next; pl && pl->remaining_objects->size > 0;  pl = pl->next)
 442                         llist_sorted_difference_inplace(pl->remaining_objects, non_unique->remaining_objects);
 443
 444                 non_unique = non_unique->next;
 445         }
 446 }
 447
 448 static void load_all_objects(void)
 449 {
 450         struct pack_list *pl = local_packs;
 451         struct llist_item *hint, *l;
 452
 453         llist_init(&all_objects);
 454
 455         while (pl) {
 456                 hint = NULL;
 457                 l = pl->remaining_objects->front;
 458                 while (l) {
 459                         hint = llist_insert_sorted_unique(all_objects,
 460                                                           &l->oid, hint);
 461                         l = l->next;
 462                 }
 463                 pl = pl->next;
 464         }
 465         /* remove objects present in remote packs */
 466         pl = altodb_packs;
 467         while (pl) {
 468                 llist_sorted_difference_inplace(all_objects, pl->remaining_objects);
 469                 pl = pl->next;
 470         }
 471 }
 472
 473 /* this scales like O(n^2) */
 474 static void cmp_local_packs(void)
 475 {
 476         struct pack_list *subset, *pl = local_packs;
 477
 478         /* only one packfile */
 479         if (!pl->next) {
 480                 llist_init(&pl->unique_objects);
 481                 return;
 482         }
 483
 484         while ((subset = pl)) {
 485                 while ((subset = subset->next))
 486                         cmp_two_packs(pl, subset);
 487                 pl = pl->next;
 488         }
 489 }
 490
 491 static void scan_alt_odb_packs(void)
 492 {
 493         struct pack_list *local, *alt;
 494
 495         alt = altodb_packs;
 496         while (alt) {
 497                 local = local_packs;
 498                 while (local) {
 499                         llist_sorted_difference_inplace(local->remaining_objects,
 500                                                         alt->remaining_objects);
 501                         local = local->next;
 502                 }
 503                 alt = alt->next;
 504         }
 505 }
 506
 507 static struct pack_list * add_pack(struct packed_git *p)
 508 {
 509         struct pack_list l;
 510         size_t off = 0, step;
 511         const unsigned char *base;
 512
 513         if (!p->pack_local && !(alt_odb || verbose))
 514                 return NULL;
 515
 516         l.pack = p;
 517         llist_init(&l.remaining_objects);
 518
 519         if (open_pack_index(p))
 520                 return NULL;
 521
 522         base = p->index_data;
 523         base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
 524         step = the_hash_algo->rawsz + ((p->index_version < 2) ? 4 : 0);
 525         while (off < p->num_objects * step) {
 526                 llist_insert_back(l.remaining_objects, base + off);
 527                 off += step;
 528         }
 529         l.all_objects_size = l.remaining_objects->size;
 530         l.unique_objects = NULL;
 531         if (p->pack_local)
 532                 return pack_list_insert(&local_packs, &l);
 533         else
 534                 return pack_list_insert(&altodb_packs, &l);
 535 }
 536
 537 static struct pack_list * add_pack_file(const char *filename)
 538 {
 539         struct packed_git *p = get_all_packs(the_repository);
 540
 541         if (strlen(filename) < 40)
 542                 die("Bad pack filename: %s", filename);
 543
 544         while (p) {
 545                 if (strstr(p->pack_name, filename))
 546                         return add_pack(p);
 547                 p = p->next;
 548         }
 549         die("Filename %s not found in packed_git", filename);
 550 }
 551
 552 static void load_all(void)
 553 {
 554         struct packed_git *p = get_all_packs(the_repository);
 555
 556         while (p) {
 557                 add_pack(p);
 558                 p = p->next;
 559         }
 560 }
 561
 562 int cmd_pack_redundant(int argc, const char **argv, const char *prefix UNUSED)
 563 {
 564         int i;
 565         int i_still_use_this = 0;
 566         struct pack_list *min = NULL, *red, *pl;
 567         struct llist *ignore;
 568         struct object_id *oid;
 569         char buf[GIT_MAX_HEXSZ + 2]; /* hex hash + \n + \0 */
 570
 571         if (argc == 2 && !strcmp(argv[1], "-h"))
 572                 usage(pack_redundant_usage);
 573
 574         for (i = 1; i < argc; i++) {
 575                 const char *arg = argv[i];
 576                 if (!strcmp(arg, "--")) {
 577                         i++;
 578                         break;
 579                 }
 580                 if (!strcmp(arg, "--all")) {
 581                         load_all_packs = 1;
 582                         continue;
 583                 }
 584                 if (!strcmp(arg, "--verbose")) {
 585                         verbose = 1;
 586                         continue;
 587                 }
 588                 if (!strcmp(arg, "--alt-odb")) {
 589                         alt_odb = 1;
 590                         continue;
 591                 }
 592                 if (!strcmp(arg, "--i-still-use-this")) {
 593                         i_still_use_this = 1;
 594                         continue;
 595                 }
 596                 if (*arg == '-')
 597                         usage(pack_redundant_usage);
 598                 else
 599                         break;
 600         }
 601
 602         if (!i_still_use_this) {
 603                 fputs(_("'git pack-redundant' is nominated for removal.\n"
 604                         "If you still use this command, please add an extra\n"
 605                         "option, '--i-still-use-this', on the command line\n"
 606                         "and let us know you still use it by sending an e-mail\n"
 607                         "to <git@vger.kernel.org>.  Thanks.\n"), stderr);
 608                 die(_("refusing to run without --i-still-use-this"));
 609         }
 610
 611         if (load_all_packs)
 612                 load_all();
 613         else
 614                 while (*(argv + i) != NULL)
 615                         add_pack_file(*(argv + i++));
 616
 617         if (!local_packs)
 618                 die("Zero packs found!");
 619
 620         load_all_objects();
 621
 622         if (alt_odb)
 623                 scan_alt_odb_packs();
 624
 625         /* ignore objects given on stdin */
 626         llist_init(&ignore);
 627         if (!isatty(0)) {
 628                 while (fgets(buf, sizeof(buf), stdin)) {
 629                         oid = xmalloc(sizeof(*oid));
 630                         if (get_oid_hex(buf, oid))
 631                                 die("Bad object ID on stdin: %s", buf);
 632                         llist_insert_sorted_unique(ignore, oid, NULL);
 633                 }
 634         }
 635         llist_sorted_difference_inplace(all_objects, ignore);
 636         pl = local_packs;
 637         while (pl) {
 638                 llist_sorted_difference_inplace(pl->remaining_objects, ignore);
 639                 pl = pl->next;
 640         }
 641
 642         cmp_local_packs();
 643
 644         minimize(&min);
 645
 646         if (verbose) {
 647                 fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
 648                         (unsigned long)pack_list_size(altodb_packs));
 649                 fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
 650                 pl = min;
 651                 while (pl) {
 652                         fprintf(stderr, "\t%s\n", pl->pack->pack_name);
 653                         pl = pl->next;
 654                 }
 655                 fprintf(stderr, "containing %lu duplicate objects "
 656                                 "with a total size of %lukb.\n",
 657                         (unsigned long)get_pack_redundancy(min),
 658                         (unsigned long)pack_set_bytecount(min)/1024);
 659                 fprintf(stderr, "A total of %lu unique objects were considered.\n",
 660                         (unsigned long)all_objects->size);
 661                 fprintf(stderr, "Redundant packs (with indexes):\n");
 662         }
 663         pl = red = pack_list_difference(local_packs, min);
 664         while (pl) {
 665                 printf("%s\n%s\n",
 666                        sha1_pack_index_name(pl->pack->hash),
 667                        pl->pack->pack_name);
 668                 pl = pl->next;
 669         }
 670         if (verbose)
 671                 fprintf(stderr, "%luMB of redundant packs in total.\n",
 672                         (unsigned long)pack_set_bytecount(red)/(1024*1024));
 673
 674         return 0;
 675 }