smatch_buf_size.c

   1 /*
   2  * smatch/smatch_buf_size.c
   3  *
   4  * Copyright (C) 2010 Dan Carpenter.
   5  *
   6  * Licensed under the Open Software License version 1.1
   7  *
   8  */
   9
  10 #include <stdlib.h>
  11 #include <errno.h>
  12 #include "parse.h"
  13 #include "smatch.h"
  14 #include "smatch_slist.h"
  15 #include "smatch_extra.h"
  16
  17 /*
  18  * This check has two smatch IDs.
  19  * my_size_id - used to store the size of arrays.
  20  * my_strlen_id - track the strlen() of buffers.
  21  */
  22
  23 static int my_size_id;
  24 static int my_strlen_id;
  25
  26 struct limiter {
  27         int buf_arg;
  28         int limit_arg;
  29 };
  30 static struct limiter b0_l2 = {0, 2};
  31
  32 static void set_undefined(struct sm_state *sm)
  33 {
  34         if (sm->state != &undefined)
  35                 set_state(sm->owner, sm->name, sm->sym, &undefined);
  36 }
  37
  38 static struct smatch_state *merge_func(struct smatch_state *s1, struct smatch_state *s2)
  39 {
  40         if (PTR_INT(s1->data) == PTR_INT(s2->data))
  41                 return s1;
  42         return &undefined;
  43 }
  44
  45 void set_param_buf_size(const char *name, struct symbol *sym, char *key, char *value)
  46 {
  47         char fullname[256];
  48         unsigned int size;
  49
  50         if (strncmp(key, "$$", 2))
  51                 return;
  52
  53         snprintf(fullname, 256, "%s%s", name, key + 2);
  54
  55         errno = 0;
  56         size = strtoul(value, NULL, 10);
  57         if (errno)
  58                 return;
  59
  60         set_state(my_size_id, fullname, sym, alloc_state_num(size));
  61 }
  62
  63 static int bytes_per_element(struct expression *expr)
  64 {
  65         struct symbol *type;
  66         int bpe;
  67
  68         if (expr->type == EXPR_STRING)
  69                 return 1;
  70         type = get_type(expr);
  71         if (!type)
  72                 return 0;
  73
  74         if (type->type != SYM_PTR && type->type != SYM_ARRAY)
  75                 return 0;
  76
  77         type = get_base_type(type);
  78         bpe = bits_to_bytes(type->bit_size);
  79
  80         if (bpe == -1) /* void pointer */
  81                 bpe = 1;
  82
  83         return bpe;
  84 }
  85
  86 static int bytes_to_elements(struct expression *expr, int bytes)
  87 {
  88         int bpe;
  89
  90         bpe = bytes_per_element(expr);
  91         if (bpe == 0)
  92                 return 0;
  93         return bytes / bpe;
  94 }
  95
  96 static int elements_to_bytes(struct expression *expr, int elements)
  97 {
  98         int bpe;
  99
 100         bpe = bytes_per_element(expr);
 101         return elements * bpe;
 102 }
 103
 104 static int get_initializer_size(struct expression *expr)
 105 {
 106         switch (expr->type) {
 107         case EXPR_STRING:
 108                 return expr->string->length;
 109         case EXPR_INITIALIZER: {
 110                 struct expression *tmp;
 111                 int i = 0;
 112                 int max = 0;
 113
 114                 FOR_EACH_PTR(expr->expr_list, tmp) {
 115                         if (tmp->type == EXPR_INDEX && tmp->idx_to > max)
 116                                 max = tmp->idx_to;
 117                         i++;
 118                 } END_FOR_EACH_PTR(tmp);
 119                 if (max)
 120                         return max + 1;
 121                 return i;
 122         }
 123         case EXPR_SYMBOL:
 124                 return get_array_size(expr);
 125         }
 126         return 0;
 127 }
 128
 129 static int db_size;
 130 static int db_size_callback(void *unused, int argc, char **argv, char **azColName)
 131 {
 132         if (db_size == 0)
 133                 db_size = atoi(argv[0]);
 134         else
 135                 db_size = -1;
 136         return 0;
 137 }
 138
 139 static int size_from_db(struct expression *expr)
 140 {
 141         int this_file_only = 0;
 142         char *name;
 143
 144         if (!option_spammy)
 145                 return 0;
 146
 147         name = get_member_name(expr);
 148         if (!name && is_static(expr)) {
 149                 name = expr_to_var(expr);
 150                 this_file_only = 1;
 151         }
 152         if (!name)
 153                 return 0;
 154
 155         db_size = 0;
 156         run_sql(db_size_callback, "select size from type_size where type = '%s' and file = '%s'",
 157                         name, get_filename());
 158         if (db_size == -1)
 159                 return 0;
 160         if (db_size != 0)
 161                 return db_size;
 162         if (this_file_only)
 163                 return 0;
 164
 165         run_sql(db_size_callback, "select size from type_size where type = '%s'",
 166                         name);
 167
 168         if (db_size == -1)
 169                 db_size = 0;
 170
 171         return db_size;
 172 }
 173
 174 static void db_returns_buf_size(struct expression *expr, int param, char *unused, char *math)
 175 {
 176         struct expression *call;
 177         sval_t sval;
 178
 179         if (expr->type != EXPR_ASSIGNMENT)
 180                 return;
 181         call = strip_expr(expr->right);
 182
 183         if (!parse_call_math(call, math, &sval))
 184                 return;
 185         set_state_expr(my_size_id, expr->left, alloc_state_num(sval.value));
 186 }
 187
 188 static int get_real_array_size(struct expression *expr)
 189 {
 190         struct symbol *type;
 191         sval_t sval;
 192
 193         if (expr->type == EXPR_BINOP) /* array elements foo[5] */
 194                 return 0;
 195
 196         type = get_type(expr);
 197         if (!type || type->type != SYM_ARRAY)
 198                 return 0;
 199
 200         if (!get_implied_value(type->array_size, &sval))
 201                 return 0;
 202
 203         /* People put one element arrays on the end of structs */
 204         if (sval.value == 1)
 205                 return 0;
 206
 207         return sval.value;
 208 }
 209
 210 static int get_size_from_initializer(struct expression *expr)
 211 {
 212         if (expr->type != EXPR_SYMBOL || !expr->symbol || !expr->symbol->initializer)
 213                 return 0;
 214         if (expr->symbol->initializer == expr) /* int a = a; */
 215                 return 0;
 216         return get_initializer_size(expr->symbol->initializer);
 217 }
 218
 219 static int get_stored_size_bytes(struct expression *expr)
 220 {
 221         struct smatch_state *state;
 222
 223         state = get_state_expr(my_size_id, expr);
 224         if (!state)
 225                 return 0;
 226         return PTR_INT(state->data);
 227 }
 228
 229 static int get_stored_size_bytes_min(struct expression *expr)
 230 {
 231         struct sm_state *sm, *tmp;
 232         int min = 0;
 233
 234         sm = get_sm_state_expr(my_size_id, expr);
 235         if (!sm)
 236                 return 0;
 237         FOR_EACH_PTR(sm->possible, tmp) {
 238                 if (PTR_INT(tmp->state->data) <= 0)
 239                         continue;
 240                 if (PTR_INT(tmp->state->data) < min)
 241                         min = PTR_INT(tmp->state->data);
 242         } END_FOR_EACH_PTR(tmp);
 243
 244         return min;
 245 }
 246
 247 static int get_bytes_from_address(struct expression *expr)
 248 {
 249         struct symbol *type;
 250         int ret;
 251
 252         if (!option_spammy)
 253                 return 0;
 254         if (expr->type != EXPR_PREOP || expr->op != '&')
 255                 return 0;
 256         type = get_type(expr);
 257         if (!type)
 258                 return 0;
 259
 260         if (type->type == SYM_PTR)
 261                 type = get_base_type(type);
 262
 263         ret = bits_to_bytes(type->bit_size);
 264         if (ret == -1)
 265                 return 0;
 266         if (ret == 1)
 267                 return 0;  /* ignore char pointers */
 268
 269         return ret;
 270 }
 271
 272 static int get_size_from_strlen(struct expression *expr)
 273 {
 274         struct smatch_state *state;
 275         sval_t len;
 276
 277         state = get_state_expr(my_strlen_id, expr);
 278         if (!state || !state->data)
 279                 return 0;
 280         if (!get_implied_max((struct expression *)state->data, &len))
 281                 return 0;
 282         if (sval_is_max(len))
 283                 return 0;
 284         return len.value + 1; /* add one because strlen doesn't include the NULL */
 285 }
 286
 287 static struct expression *remove_addr_fluff(struct expression *expr)
 288 {
 289         struct expression *tmp;
 290         sval_t sval;
 291
 292         expr = strip_expr(expr);
 293
 294         /* remove '&' and '*' operations that cancel */
 295         while (expr->type == EXPR_PREOP && expr->op == '&') {
 296                 tmp = strip_expr(expr->unop);
 297                 if (tmp->type != EXPR_PREOP)
 298                         break;
 299                 if (tmp->op != '*')
 300                         break;
 301                 expr = strip_expr(tmp->unop);
 302         }
 303
 304         /* "foo + 0" is just "foo" */
 305         if (expr->type == EXPR_BINOP && expr->op == '+' &&
 306             get_value(expr->right, &sval) && sval.value == 0)
 307                 return expr->left;
 308
 309         return expr;
 310 }
 311
 312 static int is_last_member_of_struct(struct symbol *sym, struct ident *member)
 313 {
 314         struct symbol *tmp;
 315         int i;
 316
 317         i = 0;
 318         FOR_EACH_PTR_REVERSE(sym->symbol_list, tmp) {
 319                 if (i++ || !tmp->ident)
 320                         return 0;
 321                 if (tmp->ident == member)
 322                         return 1;
 323                 return 0;
 324         } END_FOR_EACH_PTR_REVERSE(tmp);
 325
 326         return 0;
 327 }
 328
 329 static int get_stored_size_end_struct_bytes(struct expression *expr)
 330 {
 331         struct symbol *type;
 332         char *name;
 333         struct symbol *sym;
 334         struct smatch_state *state;
 335         sval_t sval;
 336
 337         if (expr->type == EXPR_BINOP) /* array elements foo[5] */
 338                 return 0;
 339
 340         type = get_type(expr);
 341         if (!type || type->type != SYM_ARRAY)
 342                 return 0;
 343
 344         if (!get_implied_value(type->array_size, &sval))
 345                 return 0;
 346
 347         if (sval.value != 0 && sval.value != 1)
 348                 return 0;
 349
 350         name = expr_to_var_sym(expr, &sym);
 351         free_string(name);
 352         if (!sym || !sym->ident || !sym->ident->name)
 353                 return 0;
 354         if (!sym->bit_size)
 355                 return 0;
 356
 357         if (sym->type != SYM_NODE)
 358                 return 0;
 359
 360         state = get_state(my_size_id, sym->ident->name, sym);
 361         if (!state || !state->data)
 362                 return 0;
 363
 364         sym = get_real_base_type(sym);
 365         if (!sym || sym->type != SYM_PTR)
 366                 return 0;
 367         sym = get_real_base_type(sym);
 368         if (!sym || sym->type != SYM_STRUCT)
 369                 return 0;
 370         if (!is_last_member_of_struct(sym, expr->member))
 371                 return 0;
 372
 373         return PTR_INT(state->data) - bits_to_bytes(sym->bit_size) +
 374                 bits_to_bytes(type->bit_size);
 375 }
 376
 377 int get_array_size_bytes(struct expression *expr)
 378 {
 379         int size;
 380
 381         expr = remove_addr_fluff(expr);
 382         if (!expr)
 383                 return 0;
 384
 385         /* strcpy(foo, "BAR"); */
 386         if (expr->type == EXPR_STRING)
 387                 return expr->string->length;
 388
 389         /* buf[4] */
 390         size = get_real_array_size(expr);
 391         if (size)
 392                 return elements_to_bytes(expr, size);
 393
 394         /* buf = malloc(1024); */
 395         size = get_stored_size_bytes(expr);
 396         if (size)
 397                 return size;
 398
 399         size = get_stored_size_end_struct_bytes(expr);
 400         if (size)
 401                 return size;
 402
 403         /* char *foo = "BAR" */
 404         size = get_size_from_initializer(expr);
 405         if (size)
 406                 return elements_to_bytes(expr, size);
 407
 408         size = get_bytes_from_address(expr);
 409         if (size)
 410                 return size;
 411
 412         /* if (strlen(foo) > 4) */
 413         size = get_size_from_strlen(expr);
 414         if (size)
 415                 return size;
 416
 417         return size_from_db(expr);
 418 }
 419
 420 int get_array_size_bytes_min(struct expression *expr)
 421 {
 422         int size;
 423         int tmp;
 424
 425         size = get_array_size_bytes(expr);
 426
 427         tmp = get_stored_size_bytes_min(expr);
 428         if (size <= 0 || (tmp >= 1 && tmp < size))
 429                 size = tmp;
 430         return size;
 431 }
 432
 433 int get_array_size(struct expression *expr)
 434 {
 435         int bytes;
 436
 437         bytes = get_array_size_bytes(expr);
 438         return bytes_to_elements(expr, bytes);
 439 }
 440
 441 static void match_strlen_condition(struct expression *expr)
 442 {
 443         struct expression *left;
 444         struct expression *right;
 445         struct expression *str = NULL;
 446         int strlen_left = 0;
 447         int strlen_right = 0;
 448         sval_t sval;
 449         struct smatch_state *true_state = NULL;
 450         struct smatch_state *false_state = NULL;
 451
 452         if (expr->type != EXPR_COMPARE)
 453                 return;
 454         left = strip_expr(expr->left);
 455         right = strip_expr(expr->right);
 456
 457         if (left->type == EXPR_CALL && sym_name_is("strlen", left->fn)) {
 458                 str = get_argument_from_call_expr(left->args, 0);
 459                 strlen_left = 1;
 460         }
 461         if (right->type == EXPR_CALL && sym_name_is("strlen", right->fn)) {
 462                 str = get_argument_from_call_expr(right->args, 0);
 463                 strlen_right = 1;
 464         }
 465
 466         if (!strlen_left && !strlen_right)
 467                 return;
 468         if (strlen_left && strlen_right)
 469                 return;
 470
 471         if (strlen_left) {
 472                 if (!get_value(right, &sval))
 473                         return;
 474         }
 475         if (strlen_right) {
 476                 if (!get_value(left, &sval))
 477                         return;
 478         }
 479
 480         if (expr->op == SPECIAL_EQUAL) {
 481                 set_true_false_states_expr(my_size_id, str, alloc_state_num(sval.value + 1), NULL);
 482                 return;
 483         }
 484         if (expr->op == SPECIAL_NOTEQUAL) {
 485                 set_true_false_states_expr(my_size_id, str, NULL, alloc_state_num(sval.value + 1));
 486                 return;
 487         }
 488
 489         switch (expr->op) {
 490         case '<':
 491         case SPECIAL_UNSIGNED_LT:
 492                 if (strlen_left)
 493                         true_state = alloc_state_num(sval.value);
 494                 else
 495                         false_state = alloc_state_num(sval.value + 1);
 496                 break;
 497         case SPECIAL_LTE:
 498         case SPECIAL_UNSIGNED_LTE:
 499                 if (strlen_left)
 500                         true_state = alloc_state_num(sval.value + 1);
 501                 else
 502                         false_state = alloc_state_num(sval.value);
 503                 break;
 504         case SPECIAL_GTE:
 505         case SPECIAL_UNSIGNED_GTE:
 506                 if (strlen_left)
 507                         false_state = alloc_state_num(sval.value);
 508                 else
 509                         true_state = alloc_state_num(sval.value + 1);
 510                 break;
 511         case '>':
 512         case SPECIAL_UNSIGNED_GT:
 513                 if (strlen_left)
 514                         false_state = alloc_state_num(sval.value + 1);
 515                 else
 516                         true_state = alloc_state_num(sval.value);
 517                 break;
 518         }
 519         set_true_false_states_expr(my_size_id, str, true_state, false_state);
 520 }
 521
 522 static struct expression *strip_ampersands(struct expression *expr)
 523 {
 524         struct symbol *type;
 525
 526         if (expr->type != EXPR_PREOP)
 527                 return expr;
 528         if (expr->op != '&')
 529                 return expr;
 530         type = get_type(expr->unop);
 531         if (!type || type->type != SYM_ARRAY)
 532                 return expr;
 533         return expr->unop;
 534 }
 535
 536 static void match_array_assignment(struct expression *expr)
 537 {
 538         struct expression *left;
 539         struct expression *right;
 540         int array_size;
 541
 542         if (expr->op != '=')
 543                 return;
 544         left = strip_expr(expr->left);
 545         right = strip_expr(expr->right);
 546         right = strip_ampersands(right);
 547         array_size = get_array_size_bytes(right);
 548         if (array_size)
 549                 set_state_expr(my_size_id, left, alloc_state_num(array_size));
 550 }
 551
 552 static void info_record_alloction(struct expression *buffer, struct expression *size)
 553 {
 554         char *name;
 555         sval_t sval;
 556
 557         if (!option_info)
 558                 return;
 559
 560         name = get_member_name(buffer);
 561         if (!name && is_static(buffer))
 562                 name = expr_to_var(buffer);
 563         if (!name)
 564                 return;
 565         if (get_implied_value(size, &sval))
 566                 sql_insert_type_size(name, sval.value);
 567         else
 568                 sql_insert_type_size(name, -1);
 569
 570         free_string(name);
 571 }
 572
 573 static void match_alloc(const char *fn, struct expression *expr, void *_size_arg)
 574 {
 575         int size_arg = PTR_INT(_size_arg);
 576         struct expression *right;
 577         struct expression *arg;
 578         sval_t bytes;
 579
 580         right = strip_expr(expr->right);
 581         arg = get_argument_from_call_expr(right->args, size_arg);
 582
 583         info_record_alloction(expr->left, arg);
 584
 585         if (!get_implied_value(arg, &bytes))
 586                 return;
 587         set_state_expr(my_size_id, expr->left, alloc_state_num(bytes.value));
 588 }
 589
 590 static void match_calloc(const char *fn, struct expression *expr, void *unused)
 591 {
 592         struct expression *right;
 593         struct expression *arg;
 594         sval_t elements;
 595         sval_t size;
 596
 597         right = strip_expr(expr->right);
 598         arg = get_argument_from_call_expr(right->args, 0);
 599         if (!get_implied_value(arg, &elements))
 600                 return;
 601         arg = get_argument_from_call_expr(right->args, 1);
 602         if (!get_implied_value(arg, &size))
 603                 return;
 604         set_state_expr(my_size_id, expr->left, alloc_state_num(elements.value * size.value));
 605 }
 606
 607 static void match_strlen(const char *fn, struct expression *expr, void *unused)
 608 {
 609         struct expression *right;
 610         struct expression *str;
 611         struct expression *len_expr;
 612         char *len_name;
 613         struct smatch_state *state;
 614
 615         right = strip_expr(expr->right);
 616         str = get_argument_from_call_expr(right->args, 0);
 617         len_expr = strip_expr(expr->left);
 618
 619         len_name = expr_to_var(len_expr);
 620         if (!len_name)
 621                 return;
 622
 623         state = __alloc_smatch_state(0);
 624         state->name = len_name;
 625         state->data = len_expr;
 626         set_state_expr(my_strlen_id, str, state);
 627 }
 628
 629 static void match_limited(const char *fn, struct expression *expr, void *_limiter)
 630 {
 631         struct limiter *limiter = (struct limiter *)_limiter;
 632         struct expression *dest;
 633         struct expression *size_expr;
 634         sval_t size;
 635
 636         dest = get_argument_from_call_expr(expr->args, limiter->buf_arg);
 637         size_expr = get_argument_from_call_expr(expr->args, limiter->limit_arg);
 638         if (!get_implied_max(size_expr, &size))
 639                 return;
 640         set_state_expr(my_size_id, dest, alloc_state_num(size.value));
 641 }
 642
 643 static void match_strcpy(const char *fn, struct expression *expr, void *unused)
 644 {
 645         struct expression fake_assign;
 646
 647         fake_assign.op = '=';
 648         fake_assign.left = get_argument_from_call_expr(expr->args, 0);
 649         fake_assign.right = get_argument_from_call_expr(expr->args, 1);
 650         match_array_assignment(&fake_assign);
 651 }
 652
 653 static void match_strndup(const char *fn, struct expression *expr, void *unused)
 654 {
 655         struct expression *fn_expr;
 656         struct expression *size_expr;
 657         sval_t size;
 658
 659         fn_expr = strip_expr(expr->right);
 660         size_expr = get_argument_from_call_expr(fn_expr->args, 1);
 661         if (!get_implied_max(size_expr, &size))
 662                 return;
 663
 664         /* It's easy to forget space for the NUL char */
 665         size.value++;
 666         set_state_expr(my_size_id, expr->left, alloc_state_num(size.value));
 667 }
 668
 669 static void match_call(struct expression *expr)
 670 {
 671         struct expression *arg;
 672         int bytes;
 673         int i;
 674
 675         i = 0;
 676         FOR_EACH_PTR(expr->args, arg) {
 677                 bytes = get_array_size_bytes(arg);
 678                 if (bytes > 1) {
 679                         char buf[11];
 680
 681                         snprintf(buf, sizeof(buf), "%d", bytes);
 682                         sql_insert_caller_info(expr, BUF_SIZE, i, "$$", buf);
 683                 }
 684                 i++;
 685         } END_FOR_EACH_PTR(arg);
 686 }
 687
 688 static void struct_member_callback(struct expression *call, int param, char *printed_name, struct smatch_state *state)
 689 {
 690         if (state == &merged)
 691                 return;
 692         sql_insert_caller_info(call, BUF_SIZE, param, printed_name, state->name);
 693 }
 694
 695 void register_buf_size(int id)
 696 {
 697         my_size_id = id;
 698
 699         add_definition_db_callback(set_param_buf_size, BUF_SIZE);
 700         add_db_return_states_callback(BUF_SIZE, &db_returns_buf_size);
 701
 702         add_function_assign_hook("malloc", &match_alloc, INT_PTR(0));
 703         add_function_assign_hook("calloc", &match_calloc, NULL);
 704         add_function_assign_hook("memdup", &match_alloc, INT_PTR(1));
 705         if (option_project == PROJ_KERNEL) {
 706                 add_function_assign_hook("kmalloc", &match_alloc, INT_PTR(0));
 707                 add_function_assign_hook("kzalloc", &match_alloc, INT_PTR(0));
 708                 add_function_assign_hook("vmalloc", &match_alloc, INT_PTR(0));
 709                 add_function_assign_hook("__vmalloc", &match_alloc, INT_PTR(0));
 710                 add_function_assign_hook("kcalloc", &match_calloc, NULL);
 711                 add_function_assign_hook("kmalloc_array", &match_calloc, NULL);
 712                 add_function_assign_hook("drm_malloc_ab", &match_calloc, NULL);
 713                 add_function_assign_hook("drm_calloc_large", &match_calloc, NULL);
 714                 add_function_assign_hook("sock_kmalloc", &match_alloc, INT_PTR(1));
 715                 add_function_assign_hook("kmemdup", &match_alloc, INT_PTR(1));
 716                 add_function_assign_hook("kmemdup_user", &match_alloc, INT_PTR(1));
 717         }
 718         add_hook(&match_array_assignment, ASSIGNMENT_HOOK);
 719         add_hook(&match_strlen_condition, CONDITION_HOOK);
 720         add_function_assign_hook("strlen", &match_strlen, NULL);
 721
 722         add_function_assign_hook("strndup", match_strndup, NULL);
 723         if (option_project == PROJ_KERNEL)
 724                 add_function_assign_hook("kstrndup", match_strndup, NULL);
 725
 726         add_modification_hook(my_size_id, &set_undefined);
 727
 728         add_merge_hook(my_size_id, &merge_func);
 729 }
 730
 731 void register_strlen(int id)
 732 {
 733         my_strlen_id = id;
 734         add_modification_hook(my_strlen_id, &set_undefined);
 735         add_merge_hook(my_strlen_id, &merge_func);
 736 }
 737
 738 void register_buf_size_late(int id)
 739 {
 740         add_function_hook("strlcpy", &match_limited, &b0_l2);
 741         add_function_hook("strlcat", &match_limited, &b0_l2);
 742         add_function_hook("memscan", &match_limited, &b0_l2);
 743
 744         add_function_hook("strcpy", &match_strcpy, NULL);
 745
 746         add_hook(&match_call, FUNCTION_CALL_HOOK);
 747         add_member_info_callback(my_size_id, struct_member_callback);
 748 }