smatch_kernel_user_data.c

   1 /*
   2  * Copyright (C) 2011 Dan Carpenter.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License
   6  * as published by the Free Software Foundation; either version 2
   7  * of the License, or (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, see http://www.gnu.org/copyleft/gpl.txt
  16  */
  17
  18 /*
  19  * There are a couple checks that try to see if a variable
  20  * comes from the user.  It would be better to unify them
  21  * into one place.  Also it we should follow the data down
  22  * the call paths.  Hence this file.
  23  */
  24
  25 #include "smatch.h"
  26 #include "smatch_slist.h"
  27 #include "smatch_extra.h"
  28
  29 static int my_id;
  30 static int my_call_id;
  31
  32 static struct expression *ignore_clear;
  33
  34 STATE(called);
  35
  36 struct user_fn_info {
  37         const char *name;
  38         int type;
  39         int param;
  40         const char *key;
  41         const sval_t *implies_start, *implies_end;
  42         func_hook *call_back;
  43 };
  44
  45 static struct user_fn_info func_table[] = {
  46         { "iov_iter_count", USER_DATA, -1, "$" },
  47         { "simple_strtol", USER_DATA, -1, "$" },
  48         { "simple_strtoll", USER_DATA, -1, "$" },
  49         { "simple_strtoul", USER_DATA, -1, "$" },
  50         { "simple_strtoull", USER_DATA, -1, "$" },
  51         { "kvm_register_read", USER_DATA, -1, "$" },
  52 };
  53
  54 static struct smatch_state *empty_state(struct sm_state *sm)
  55 {
  56         return alloc_estate_empty();
  57 }
  58
  59 static struct smatch_state *new_state(struct symbol *type)
  60 {
  61         struct smatch_state *state;
  62
  63         if (!type || type_is_ptr(type))
  64                 return NULL;
  65
  66         state = alloc_estate_whole(type);
  67         estate_set_new(state);
  68         return state;
  69 }
  70
  71 static void pre_merge_hook(struct sm_state *cur, struct sm_state *other)
  72 {
  73         struct smatch_state *user = cur->state;
  74         struct smatch_state *extra;
  75         struct smatch_state *state;
  76         struct range_list *rl;
  77
  78         extra = __get_state(SMATCH_EXTRA, cur->name, cur->sym);
  79         if (!extra)
  80                 return;
  81         rl = rl_intersection(estate_rl(user), estate_rl(extra));
  82         state = alloc_estate_rl(clone_rl(rl));
  83         if (estate_capped(user) || is_capped_var_sym(cur->name, cur->sym))
  84                 estate_set_capped(state);
  85         if (estate_treat_untagged(user))
  86                 estate_set_treat_untagged(state);
  87         if (estates_equiv(state, cur->state))
  88                 return;
  89         if (estate_new(cur->state))
  90                 estate_set_new(state);
  91         set_state(my_id, cur->name, cur->sym, state);
  92 }
  93
  94 static void extra_nomod_hook(const char *name, struct symbol *sym, struct expression *expr, struct smatch_state *state)
  95 {
  96         struct smatch_state *user, *new;
  97         struct range_list *rl;
  98
  99         user = __get_state(my_id, name, sym);
 100         if (!user)
 101                 return;
 102         rl = rl_intersection(estate_rl(user), estate_rl(state));
 103         if (rl_equiv(rl, estate_rl(user)))
 104                 return;
 105         new = alloc_estate_rl(rl);
 106         if (estate_capped(user))
 107                 estate_set_capped(new);
 108         if (estate_treat_untagged(user))
 109                 estate_set_treat_untagged(new);
 110         set_state(my_id, name, sym, new);
 111 }
 112
 113 static void store_type_info(struct expression *expr, struct smatch_state *state)
 114 {
 115         struct symbol *type;
 116         char *type_str, *member;
 117
 118         if (__in_fake_assign)
 119                 return;
 120
 121         if (!estate_rl(state))
 122                 return;
 123
 124         expr = strip_expr(expr);
 125         if (!expr || expr->type != EXPR_DEREF || !expr->member)
 126                 return;
 127
 128         type = get_type(expr->deref);
 129         if (!type || !type->ident)
 130                 return;
 131
 132         type_str = type_to_str(type);
 133         if (!type_str)
 134                 return;
 135         member = get_member_name(expr);
 136         if (!member)
 137                 return;
 138
 139         sql_insert_function_type_info(USER_DATA, type_str, member, state->name);
 140 }
 141
 142 static void set_user_data(struct expression *expr, struct smatch_state *state)
 143 {
 144         store_type_info(expr, state);
 145         set_state_expr(my_id, expr, state);
 146 }
 147
 148 static bool user_rl_known(struct expression *expr)
 149 {
 150         struct range_list *rl;
 151         sval_t close_to_max;
 152
 153         if (!get_user_rl(expr, &rl))
 154                 return true;
 155
 156         close_to_max = sval_type_max(rl_type(rl));
 157         close_to_max.value -= 100;
 158
 159         if (sval_cmp(rl_max(rl), close_to_max) >= 0)
 160                 return false;
 161         return true;
 162 }
 163
 164 static bool is_array_index_mask_nospec(struct expression *expr)
 165 {
 166         struct expression *orig;
 167
 168         orig = get_assigned_expr(expr);
 169         if (!orig || orig->type != EXPR_CALL)
 170                 return false;
 171         return sym_name_is("array_index_mask_nospec", orig->fn);
 172 }
 173
 174 static bool binop_capped(struct expression *expr)
 175 {
 176         struct range_list *left_rl;
 177         int comparison;
 178         sval_t sval;
 179
 180         if (expr->op == '-' && get_user_rl(expr->left, &left_rl)) {
 181                 if (user_rl_capped(expr->left))
 182                         return true;
 183                 comparison = get_comparison(expr->left, expr->right);
 184                 if (comparison && show_special(comparison)[0] == '>')
 185                         return true;
 186                 return false;
 187         }
 188
 189         if (expr->op == '&' || expr->op == '%') {
 190                 bool left_user, left_capped, right_user, right_capped;
 191
 192                 if (!get_value(expr->right, &sval) && is_capped(expr->right))
 193                         return true;
 194                 if (is_array_index_mask_nospec(expr->right))
 195                         return true;
 196                 if (is_capped(expr->left))
 197                         return true;
 198                 left_user = is_user_rl(expr->left);
 199                 right_user = is_user_rl(expr->right);
 200                 if (!left_user && !right_user)
 201                         return true;
 202
 203                 left_capped = user_rl_capped(expr->left);
 204                 right_capped = user_rl_capped(expr->right);
 205
 206                 if (left_user && left_capped) {
 207                         if (!right_user)
 208                                 return true;
 209                         if (right_user && right_capped)
 210                                 return true;
 211                         return false;
 212                 }
 213                 if (right_user && right_capped) {
 214                         if (!left_user)
 215                                 return true;
 216                         return false;
 217                 }
 218                 return false;
 219         }
 220
 221         /*
 222          * Generally "capped" means that we capped it to an unknown value.
 223          * This is useful because if Smatch doesn't know what the value is then
 224          * we have to trust that it is correct.  But if we known cap value is
 225          * 100 then we can check if 100 is correct and complain if it's wrong.
 226          *
 227          * So then the problem is with BINOP when we take a capped variable
 228          * plus a user variable which is clamped to a known range (uncapped)
 229          * the result should be capped.
 230          */
 231         if ((user_rl_capped(expr->left) || user_rl_known(expr->left)) &&
 232             (user_rl_capped(expr->right) || user_rl_known(expr->right)))
 233                 return true;
 234
 235         return false;
 236 }
 237
 238 bool user_rl_capped_var_sym(const char *name, struct symbol *sym)
 239 {
 240         struct smatch_state *state;
 241
 242         state = get_state(my_id, name, sym);
 243         if (state)
 244                 return estate_capped(state);
 245
 246         return true;
 247 }
 248
 249 bool user_rl_capped(struct expression *expr)
 250 {
 251         struct smatch_state *state;
 252         struct range_list *rl;
 253         sval_t sval;
 254
 255         expr = strip_expr(expr);
 256         if (!expr)
 257                 return false;
 258         if (get_value(expr, &sval))
 259                 return true;
 260         if (expr->type == EXPR_BINOP)
 261                 return binop_capped(expr);
 262         if ((expr->type == EXPR_PREOP || expr->type == EXPR_POSTOP) &&
 263             (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT))
 264                 return user_rl_capped(expr->unop);
 265         state = get_state_expr(my_id, expr);
 266         if (state)
 267                 return estate_capped(state);
 268
 269         if (!get_user_rl(expr, &rl)) {
 270                 /*
 271                  * The non user data parts of a binop are capped and
 272                  * also empty user rl states are capped.
 273                  */
 274                 return true;
 275         }
 276
 277         if (rl_to_sval(rl, &sval))
 278                 return true;
 279
 280         return false;  /* uncapped user data */
 281 }
 282
 283 bool user_rl_treat_untagged(struct expression *expr)
 284 {
 285         struct smatch_state *state;
 286         struct range_list *rl;
 287         sval_t sval;
 288
 289         expr = strip_expr(expr);
 290         if (!expr)
 291                 return false;
 292         if (get_value(expr, &sval))
 293                 return true;
 294
 295         state = get_state_expr(my_id, expr);
 296         if (state)
 297                 return estate_treat_untagged(state);
 298
 299         if (get_user_rl(expr, &rl))
 300                 return false;  /* uncapped user data */
 301
 302         return true;  /* not actually user data */
 303 }
 304
 305 static int is_dev_attr_name(struct expression *expr)
 306 {
 307         char *name;
 308         int ret = 0;
 309
 310         name = expr_to_str(expr);
 311         if (!name)
 312                 return 0;
 313         if (strstr(name, "->attr.name"))
 314                 ret = 1;
 315         free_string(name);
 316         return ret;
 317 }
 318
 319 static bool is_percent_n(struct expression *expr, int pos)
 320 {
 321         char *p;
 322         int cnt = 0;
 323
 324         if (!expr)
 325                 return false;
 326         if (expr->type != EXPR_STRING || !expr->string)
 327                 return false;
 328
 329         p = expr->string->data;
 330         while (*p) {
 331                 if (p[0] != '%' ||
 332                     (p[0] == '%' && p[1] == '%')) {
 333                         p++;
 334                         continue;
 335                 }
 336                 if (pos != cnt) {
 337                         cnt++;
 338                         p++;
 339                         continue;
 340                 }
 341                 if (p[1] == 'n')
 342                         return true;
 343                 return false;
 344         }
 345
 346         return false;
 347 }
 348
 349 static void match_sscanf(const char *fn, struct expression *expr, void *unused)
 350 {
 351         struct expression *str, *format, *arg;
 352         int i;
 353
 354         str = get_argument_from_call_expr(expr->args, 0);
 355         if (is_dev_attr_name(str))
 356                 return;
 357
 358         format = get_argument_from_call_expr(expr->args, 1);
 359         if (is_dev_attr_name(format))
 360                 return;
 361
 362         i = -1;
 363         FOR_EACH_PTR(expr->args, arg) {
 364                 i++;
 365                 if (i < 2)
 366                         continue;
 367                 if (is_percent_n(format, i - 2))
 368                         continue;
 369                 mark_as_user_data(deref_expression(arg), true);
 370         } END_FOR_EACH_PTR(arg);
 371 }
 372
 373 static int comes_from_skb_data(struct expression *expr)
 374 {
 375         expr = strip_expr(expr);
 376         if (!expr || expr->type != EXPR_PREOP || expr->op != '*')
 377                 return 0;
 378
 379         expr = strip_expr(expr->unop);
 380         if (!expr)
 381                 return 0;
 382         if (expr->type == EXPR_BINOP && expr->op == '+')
 383                 expr = strip_expr(expr->left);
 384
 385         return is_skb_data(expr);
 386 }
 387
 388 static int handle_get_user(struct expression *expr)
 389 {
 390         char *name;
 391         int ret = 0;
 392
 393         name = get_macro_name(expr->pos);
 394         if (!name || strcmp(name, "get_user") != 0)
 395                 return 0;
 396
 397         name = expr_to_var(expr->right);
 398         if (!name || (strcmp(name, "__val_gu") != 0 && strcmp(name, "__gu_val") != 0))
 399                 goto free;
 400         set_user_data(expr->left, new_state(get_type(expr->left)));
 401         ret = 1;
 402 free:
 403         free_string(name);
 404         return ret;
 405 }
 406
 407 static bool state_is_new(struct expression *expr)
 408 {
 409         struct smatch_state *state;
 410
 411         state = get_state_expr(my_id, expr);
 412         if (estate_new(state))
 413                 return true;
 414
 415         if (expr->type == EXPR_BINOP) {
 416                 if (state_is_new(expr->left))
 417                         return true;
 418                 if (state_is_new(expr->right))
 419                         return true;
 420         }
 421         return false;
 422 }
 423
 424 static struct range_list *strip_negatives(struct range_list *rl)
 425 {
 426         sval_t min = rl_min(rl);
 427         sval_t minus_one = { .type = rl_type(rl), .value = -1 };
 428         sval_t over = { .type = rl_type(rl), .value = INT_MAX + 1ULL };
 429         sval_t max = sval_type_max(rl_type(rl));
 430
 431         if (!rl)
 432                 return NULL;
 433
 434         if (type_unsigned(rl_type(rl)) && type_bits(rl_type(rl)) > 31)
 435                 return remove_range(rl, over, max);
 436
 437         return remove_range(rl, min, minus_one);
 438 }
 439
 440 static bool handle_op_assign(struct expression *expr)
 441 {
 442         struct expression *binop_expr;
 443         struct smatch_state *state;
 444         struct range_list *rl;
 445
 446         switch (expr->op) {
 447         case SPECIAL_ADD_ASSIGN:
 448         case SPECIAL_SUB_ASSIGN:
 449         case SPECIAL_AND_ASSIGN:
 450         case SPECIAL_MOD_ASSIGN:
 451         case SPECIAL_SHL_ASSIGN:
 452         case SPECIAL_SHR_ASSIGN:
 453         case SPECIAL_OR_ASSIGN:
 454         case SPECIAL_XOR_ASSIGN:
 455         case SPECIAL_MUL_ASSIGN:
 456         case SPECIAL_DIV_ASSIGN:
 457                 binop_expr = binop_expression(expr->left,
 458                                               op_remove_assign(expr->op),
 459                                               expr->right);
 460                 if (!get_user_rl(binop_expr, &rl))
 461                         return true;
 462
 463                 rl = cast_rl(get_type(expr->left), rl);
 464                 state = alloc_estate_rl(rl);
 465                 if (expr->op == SPECIAL_AND_ASSIGN ||
 466                     expr->op == SPECIAL_MOD_ASSIGN ||
 467                     user_rl_capped(binop_expr))
 468                         estate_set_capped(state);
 469                 if (user_rl_treat_untagged(expr->left))
 470                         estate_set_treat_untagged(state);
 471                 if (state_is_new(binop_expr))
 472                         estate_set_new(state);
 473                 estate_set_assigned(state);
 474                 set_user_data(expr->left, state);
 475                 return true;
 476         }
 477         return false;
 478 }
 479
 480 static void handle_derefed_pointers(struct expression *expr, bool is_new)
 481 {
 482         expr = strip_expr(expr);
 483         if (expr->type != EXPR_PREOP ||
 484             expr->op != '*')
 485                 return;
 486         expr = strip_expr(expr->unop);
 487         set_array_user_ptr(expr, is_new);
 488 }
 489
 490 static void match_assign(struct expression *expr)
 491 {
 492         struct symbol *left_type, *right_type;
 493         struct range_list *rl = NULL;
 494         static struct expression *handled;
 495         struct smatch_state *state;
 496         struct expression *faked;
 497         bool is_capped = false;
 498         bool is_new = false;
 499
 500         left_type = get_type(expr->left);
 501         if (left_type == &void_ctype)
 502                 return;
 503
 504         faked = get_faked_expression();
 505         if (faked && faked == ignore_clear)
 506                 return;
 507
 508         /* FIXME: handle fake array assignments frob(&user_array[x]); */
 509
 510         if (faked &&
 511             faked->type == EXPR_ASSIGNMENT &&
 512             points_to_user_data(faked->right)) {
 513                 if (is_skb_data(faked->right))
 514                         is_new = true;
 515                 rl = alloc_whole_rl(left_type);
 516                 goto set;
 517         }
 518
 519         if (faked && faked == handled)
 520                 return;
 521         if (is_fake_call(expr->right))
 522                 goto clear_old_state;
 523         if (handle_get_user(expr))
 524                 return;
 525         if (points_to_user_data(expr->right) &&
 526             is_struct_ptr(get_type(expr->left))) {
 527                 handled = expr;
 528                 // This should be handled by smatch_points_to_user_data.c
 529                 // set_array_user_ptr(expr->left);
 530         }
 531
 532         if (handle_op_assign(expr))
 533                 return;
 534         if (expr->op != '=')
 535                 goto clear_old_state;
 536
 537         /* Handled by DB code */
 538         if (expr->right->type == EXPR_CALL)
 539                 return;
 540
 541         if (faked)
 542                 disable_type_val_lookups();
 543         get_user_rl(expr->right, &rl);
 544         if (faked)
 545                 enable_type_val_lookups();
 546         if (!rl)
 547                 goto clear_old_state;
 548
 549         is_capped = user_rl_capped(expr->right);
 550         is_new = state_is_new(expr->right);
 551
 552 set:
 553         right_type = get_type(expr->right);
 554         if (type_is_ptr(left_type)) {
 555                 if (right_type && right_type->type == SYM_ARRAY)
 556                         set_array_user_ptr(expr->left, is_new);
 557                 return;
 558         }
 559
 560         rl = cast_rl(left_type, rl);
 561         if (is_capped && type_unsigned(right_type) && type_signed(left_type))
 562                 rl = strip_negatives(rl);
 563         state = alloc_estate_rl(rl);
 564         if (is_new)
 565                 estate_set_new(state);
 566         if (is_capped)
 567                 estate_set_capped(state);
 568         if (user_rl_treat_untagged(expr->right))
 569                 estate_set_treat_untagged(state);
 570         estate_set_assigned(state);
 571
 572         set_user_data(expr->left, state);
 573         handle_derefed_pointers(expr->left, is_new);
 574         return;
 575
 576 clear_old_state:
 577
 578         /*
 579          * HACK ALERT!!!  This should be at the start of the function.  The
 580          * the problem is that handling "pointer = array;" assignments is
 581          * handled in this function instead of in kernel_points_to_user_data.c.
 582          */
 583         if (type_is_ptr(left_type))
 584                 return;
 585
 586         if (get_state_expr(my_id, expr->left))
 587                 set_user_data(expr->left, alloc_estate_empty());
 588 }
 589
 590 static void handle_eq_noteq(struct expression *expr)
 591 {
 592         struct smatch_state *left_orig, *right_orig;
 593
 594         left_orig = get_state_expr(my_id, expr->left);
 595         right_orig = get_state_expr(my_id, expr->right);
 596
 597         if (!left_orig && !right_orig)
 598                 return;
 599         if (left_orig && right_orig)
 600                 return;
 601
 602         if (left_orig) {
 603                 set_true_false_states_expr(my_id, expr->left,
 604                                 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
 605                                 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
 606         } else {
 607                 set_true_false_states_expr(my_id, expr->right,
 608                                 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
 609                                 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
 610         }
 611 }
 612
 613 static void handle_compare(struct expression *expr)
 614 {
 615         struct expression  *left, *right;
 616         struct range_list *left_rl = NULL;
 617         struct range_list *right_rl = NULL;
 618         struct range_list *user_rl;
 619         struct smatch_state *capped_state;
 620         struct smatch_state *left_true = NULL;
 621         struct smatch_state *left_false = NULL;
 622         struct smatch_state *right_true = NULL;
 623         struct smatch_state *right_false = NULL;
 624         struct symbol *type;
 625         sval_t sval;
 626
 627         left = strip_expr(expr->left);
 628         right = strip_expr(expr->right);
 629
 630         while (left->type == EXPR_ASSIGNMENT)
 631                 left = strip_expr(left->left);
 632
 633         /*
 634          * Conditions are mostly handled by smatch_extra.c, but there are some
 635          * times where the exact values are not known so we can't do that.
 636          *
 637          * Normally, we might consider using smatch_capped.c to supliment smatch
 638          * extra but that doesn't work when we merge unknown uncapped kernel
 639          * data with unknown capped user data.  The result is uncapped user
 640          * data.  We need to keep it separate and say that the user data is
 641          * capped.  In the past, I would have marked this as just regular
 642          * kernel data (not user data) but we can't do that these days because
 643          * we need to track user data for Spectre.
 644          *
 645          * The other situation which we have to handle is when we do have an
 646          * int and we compare against an unknown unsigned kernel variable.  In
 647          * that situation we assume that the kernel data is less than INT_MAX.
 648          * Otherwise then we get all sorts of array underflow false positives.
 649          *
 650          */
 651
 652         /* Handled in smatch_extra.c */
 653         if (get_implied_value(left, &sval) ||
 654             get_implied_value(right, &sval))
 655                 return;
 656
 657         get_user_rl(left, &left_rl);
 658         get_user_rl(right, &right_rl);
 659
 660         /* nothing to do */
 661         if (!left_rl && !right_rl)
 662                 return;
 663         /* if both sides are user data that's not a good limit */
 664         if (left_rl && right_rl)
 665                 return;
 666
 667         if (left_rl)
 668                 user_rl = left_rl;
 669         else
 670                 user_rl = right_rl;
 671
 672         type = get_type(expr);
 673         if (type_unsigned(type))
 674                 user_rl = strip_negatives(user_rl);
 675         capped_state = alloc_estate_rl(user_rl);
 676         estate_set_capped(capped_state);
 677
 678         switch (expr->op) {
 679         case '<':
 680         case SPECIAL_UNSIGNED_LT:
 681         case SPECIAL_LTE:
 682         case SPECIAL_UNSIGNED_LTE:
 683                 if (left_rl)
 684                         left_true = capped_state;
 685                 else
 686                         right_false = capped_state;
 687                 break;
 688         case '>':
 689         case SPECIAL_UNSIGNED_GT:
 690         case SPECIAL_GTE:
 691         case SPECIAL_UNSIGNED_GTE:
 692                 if (left_rl)
 693                         left_false = capped_state;
 694                 else
 695                         right_true = capped_state;
 696                 break;
 697         }
 698
 699         set_true_false_states_expr(my_id, left, left_true, left_false);
 700         set_true_false_states_expr(my_id, right, right_true, right_false);
 701 }
 702
 703 static void match_condition(struct expression *expr)
 704 {
 705         if (expr->type != EXPR_COMPARE)
 706                 return;
 707
 708         if (expr->op == SPECIAL_EQUAL ||
 709             expr->op == SPECIAL_NOTEQUAL) {
 710                 handle_eq_noteq(expr);
 711                 return;
 712         }
 713
 714         handle_compare(expr);
 715 }
 716
 717 static int get_user_macro_rl(struct expression *expr, struct range_list **rl)
 718 {
 719         struct expression *parent;
 720         char *macro;
 721
 722         if (!expr)
 723                 return 0;
 724
 725         macro = get_macro_name(expr->pos);
 726         if (!macro)
 727                 return 0;
 728
 729         /* handle ntohl(foo[i]) where "i" is trusted */
 730         parent = expr_get_parent_expr(expr);
 731         while (parent && parent->type != EXPR_BINOP)
 732                 parent = expr_get_parent_expr(parent);
 733         if (parent && parent->type == EXPR_BINOP) {
 734                 char *parent_macro = get_macro_name(parent->pos);
 735
 736                 if (parent_macro && strcmp(macro, parent_macro) == 0)
 737                         return 0;
 738         }
 739
 740         if (strcmp(macro, "ntohl") == 0) {
 741                 *rl = alloc_whole_rl(&uint_ctype);
 742                 return 1;
 743         }
 744         if (strcmp(macro, "ntohs") == 0) {
 745                 *rl = alloc_whole_rl(&ushort_ctype);
 746                 return 1;
 747         }
 748         return 0;
 749 }
 750
 751 static int has_user_data(struct symbol *sym)
 752 {
 753         struct sm_state *tmp;
 754
 755         FOR_EACH_MY_SM(my_id, __get_cur_stree(), tmp) {
 756                 if (tmp->sym == sym)
 757                         return 1;
 758         } END_FOR_EACH_SM(tmp);
 759         return 0;
 760 }
 761
 762 bool we_pass_user_data(struct expression *call)
 763 {
 764         struct expression *arg;
 765         struct symbol *sym;
 766
 767         FOR_EACH_PTR(call->args, arg) {
 768                 if (points_to_user_data(arg))
 769                         return true;
 770                 sym = expr_to_sym(arg);
 771                 if (!sym)
 772                         continue;
 773                 if (has_user_data(sym))
 774                         return true;
 775         } END_FOR_EACH_PTR(arg);
 776
 777         return false;
 778 }
 779
 780 // TODO: faked_assign this should already be handled
 781 static int db_returned_user_rl(struct expression *call, struct range_list **rl)
 782 {
 783         struct smatch_state *state;
 784         char buf[48];
 785
 786         if (is_fake_call(call))
 787                 return 0;
 788         snprintf(buf, sizeof(buf), "return %p", call);
 789         state = get_state(my_id, buf, NULL);
 790         if (!state || !estate_rl(state))
 791                 return 0;
 792         *rl = estate_rl(state);
 793         return 1;
 794 }
 795
 796 struct stree *get_user_stree(void)
 797 {
 798         return get_all_states_stree(my_id);
 799 }
 800
 801 static struct int_stack *user_data_flags, *no_user_data_flags;
 802
 803 static void set_flag(struct int_stack **stack)
 804 {
 805         int num;
 806
 807         num = pop_int(stack);
 808         num = 1;
 809         push_int(stack, num);
 810 }
 811
 812 struct range_list *var_user_rl(struct expression *expr)
 813 {
 814         struct smatch_state *state;
 815         struct range_list *rl;
 816         struct range_list *absolute_rl;
 817
 818         if (expr->type == EXPR_PREOP && expr->op == '&') {
 819                 set_flag(&no_user_data_flags);
 820                 return NULL;
 821         }
 822
 823         if (expr->type == EXPR_BINOP && expr->op == '%') {
 824                 struct range_list *left, *right;
 825
 826                 if (!get_user_rl(expr->right, &right))
 827                         return NULL;
 828                 get_absolute_rl(expr->left, &left);
 829                 rl = rl_binop(left, '%', right);
 830                 goto found;
 831         }
 832
 833         if (expr->type == EXPR_BINOP && expr->op == '/') {
 834                 struct range_list *left = NULL;
 835                 struct range_list *right = NULL;
 836                 struct range_list *abs_right;
 837
 838                 /*
 839                  * The specific bug I'm dealing with is:
 840                  *
 841                  * foo = capped_user / unknown;
 842                  *
 843                  * Instead of just saying foo is now entirely user_rl we should
 844                  * probably say instead that it is not at all user data.
 845                  *
 846                  */
 847
 848                 get_user_rl(expr->left, &left);
 849                 get_user_rl(expr->right, &right);
 850                 get_absolute_rl(expr->right, &abs_right);
 851
 852                 if (left && !right) {
 853                         rl = rl_binop(left, '/', abs_right);
 854                         if (sval_cmp(rl_max(left), rl_max(rl)) < 0)
 855                                 set_flag(&no_user_data_flags);
 856                 }
 857
 858                 return NULL;
 859         }
 860
 861         if (get_user_macro_rl(expr, &rl))
 862                 goto found;
 863
 864         if (comes_from_skb_data(expr)) {
 865                 rl = alloc_whole_rl(get_type(expr));
 866                 goto found;
 867         }
 868
 869         state = get_state_expr(my_id, expr);
 870         if (state && estate_rl(state)) {
 871                 rl = estate_rl(state);
 872                 goto found;
 873         }
 874
 875         if (expr->type == EXPR_CALL && db_returned_user_rl(expr, &rl))
 876                 goto found;
 877
 878         if (expr->type == EXPR_PREOP && expr->op == '*' &&
 879             points_to_user_data(expr->unop)) {
 880                 rl = var_to_absolute_rl(expr);
 881                 goto found;
 882         }
 883
 884         if (is_array(expr)) {
 885                 struct expression *array = get_array_base(expr);
 886
 887                 if (!get_state_expr(my_id, array)) {
 888                         set_flag(&no_user_data_flags);
 889                         return NULL;
 890                 }
 891         }
 892
 893         return NULL;
 894 found:
 895         set_flag(&user_data_flags);
 896         absolute_rl = var_to_absolute_rl(expr);
 897         return rl_intersection(rl, absolute_rl);
 898 }
 899
 900 static bool is_ptr_subtract(struct expression *expr)
 901 {
 902         expr = strip_expr(expr);
 903         if (!expr)
 904                 return false;
 905         if (expr->type == EXPR_BINOP && expr->op == '-' &&
 906             type_is_ptr(get_type(expr->left))) {
 907                 return true;
 908         }
 909         return false;
 910 }
 911
 912 int get_user_rl(struct expression *expr, struct range_list **rl)
 913 {
 914         int user_data, no_user_data;
 915
 916         if (!expr)
 917                 return 0;
 918
 919         if (__in_fake_struct_assign &&
 920             !has_states(__get_cur_stree(), my_id))
 921                 return 0;
 922
 923         if (is_ptr_subtract(expr))
 924                 return 0;
 925
 926         push_int(&user_data_flags, 0);
 927         push_int(&no_user_data_flags, 0);
 928
 929         custom_get_absolute_rl(expr, &var_user_rl, rl);
 930
 931         user_data = pop_int(&user_data_flags);
 932         no_user_data = pop_int(&no_user_data_flags);
 933
 934         if (!user_data || no_user_data)
 935                 *rl = NULL;
 936
 937         return !!*rl;
 938 }
 939
 940 int is_user_rl(struct expression *expr)
 941 {
 942         struct range_list *tmp;
 943
 944         return get_user_rl(expr, &tmp) && tmp;
 945 }
 946
 947 int get_user_rl_var_sym(const char *name, struct symbol *sym, struct range_list **rl)
 948 {
 949         struct smatch_state *state, *extra;
 950
 951         state = get_state(my_id, name, sym);
 952         if (!estate_rl(state))
 953                 return 0;
 954         *rl = estate_rl(state);
 955
 956         extra = get_state(SMATCH_EXTRA, name, sym);
 957         if (estate_rl(extra))
 958                 *rl = rl_intersection(estate_rl(state), estate_rl(extra));
 959
 960         return 1;
 961 }
 962
 963 bool is_socket_stuff(struct symbol *sym)
 964 {
 965         struct symbol *type;
 966
 967         /* This is a hack.
 968          * Basically I never want to consider an skb or sk as user pointer.
 969          * The skb->data is already marked as a source of user data, and if
 970          * anything else is marked as user data it's almost certainly wrong.
 971          *
 972          * Ideally, I would figure out where this bogus data is coming from,
 973          * but possibly it just was stuck in the database from previous updates
 974          * and can't get cleared out without deleting all user data.  Things
 975          * like this gets stuck in the DB because of recursion.
 976          *
 977          * I could make this a temporary hack, but I keep wanting to do it so
 978          * I'm just going to make it permanent.  It either doesn't change
 979          * anything or it makes life better.
 980          */
 981
 982         type = get_real_base_type(sym);
 983         if (!type || type->type != SYM_PTR)
 984                 return false;
 985         type = get_real_base_type(type);
 986         if (!type || type->type != SYM_STRUCT || !type->ident)
 987                 return false;
 988
 989         if (strcmp(type->ident->name, "sk_buff") == 0)
 990                 return true;
 991         if (strcmp(type->ident->name, "sock") == 0)
 992                 return true;
 993         if (strcmp(type->ident->name, "socket") == 0)
 994                 return true;
 995
 996         return false;
 997 }
 998
 999 static void return_info_callback(int return_id, char *return_ranges,
1000                                  struct expression *returned_expr,
1001                                  int param,
1002                                  const char *printed_name,
1003                                  struct sm_state *sm)
1004 {
1005         struct smatch_state *extra;
1006         struct range_list *rl;
1007         char buf[64];
1008
1009         if (is_socket_stuff(sm->sym))
1010                 return;
1011         if (is_ignored_kernel_data(printed_name))
1012                 return;
1013
1014         if (param >= 0) {
1015                 if (strcmp(printed_name, "$") == 0)
1016                         return;
1017                 if (!param_was_set_var_sym(sm->name, sm->sym))
1018                         return;
1019                 if (!estate_assigned(sm->state) &&
1020                     !estate_new(sm->state))
1021                         return;
1022         }
1023         rl = estate_rl(sm->state);
1024         if (!rl)
1025                 return;
1026         extra = get_state(SMATCH_EXTRA, sm->name, sm->sym);
1027         if (estate_rl(extra))
1028                 rl = rl_intersection(estate_rl(sm->state), estate_rl(extra));
1029         if (!rl)
1030                 return;
1031
1032         snprintf(buf, sizeof(buf), "%s%s%s",
1033                  show_rl(rl),
1034                  estate_capped(sm->state) ? "[c]" : "",
1035                  estate_treat_untagged(sm->state) ? "[u]" : "");
1036         sql_insert_return_states(return_id, return_ranges,
1037                                  estate_new(sm->state) ? USER_DATA_SET : USER_DATA,
1038                                  param, printed_name, buf);
1039 }
1040
1041 static bool is_ignored_macro(struct position pos)
1042 {
1043         const char *macro;
1044
1045         macro = get_macro_name(pos);
1046         if (!macro)
1047                 return false;
1048         if (strcmp(macro, "v4l2_subdev_call") == 0)
1049                 return true;
1050         return false;
1051 }
1052
1053 static void caller_info_callback(struct expression *call, int param, char *printed_name, struct sm_state *sm)
1054 {
1055         struct smatch_state *state;
1056         struct range_list *rl;
1057         struct symbol *type;
1058         char buf[64];
1059
1060         if (is_ignored_macro(call->pos))
1061                 return;
1062
1063         if (is_socket_stuff(sm->sym))
1064                 return;
1065         if (is_ignored_kernel_data(printed_name))
1066                 return;
1067
1068         /*
1069          * Smatch uses a hack where if we get an unsigned long we say it's
1070          * both user data and it points to user data.  But if we pass it to a
1071          * function which takes an int, then it's just user data.  There's not
1072          * enough bytes for it to be a pointer.
1073          *
1074          */
1075         type = get_arg_type(call->fn, param);
1076         if (strcmp(printed_name, "$") != 0 && type && type_bits(type) < type_bits(&ptr_ctype))
1077                 return;
1078
1079         if (strcmp(sm->state->name, "") == 0)
1080                 return;
1081
1082         state = __get_state(SMATCH_EXTRA, sm->name, sm->sym);
1083         if (!state || !estate_rl(state))
1084                 rl = estate_rl(sm->state);
1085         else
1086                 rl = rl_intersection(estate_rl(sm->state), estate_rl(state));
1087
1088         if (!rl)
1089                 return;
1090
1091         snprintf(buf, sizeof(buf), "%s%s%s", show_rl(rl),
1092                  estate_capped(sm->state) ? "[c]" : "",
1093                  estate_treat_untagged(sm->state) ? "[u]" : "");
1094         sql_insert_caller_info(call, USER_DATA, param, printed_name, buf);
1095 }
1096
1097 static void db_param_set(struct expression *expr, int param, char *key, char *value)
1098 {
1099         struct expression *arg;
1100         char *name;
1101         struct symbol *sym;
1102         struct smatch_state *state;
1103
1104         while (expr->type == EXPR_ASSIGNMENT)
1105                 expr = strip_expr(expr->right);
1106         if (expr->type != EXPR_CALL)
1107                 return;
1108         if (expr == ignore_clear)
1109                 return;
1110
1111         arg = get_argument_from_call_expr(expr->args, param);
1112         if (!arg)
1113                 return;
1114         name = get_variable_from_key(arg, key, &sym);
1115         if (!name || !sym)
1116                 goto free;
1117
1118         state = get_state(my_id, name, sym);
1119         if (!state)
1120                 goto free;
1121
1122         set_state(my_id, name, sym, alloc_estate_empty());
1123 free:
1124         free_string(name);
1125 }
1126
1127 static bool param_data_capped(const char *value)
1128 {
1129         if (strstr(value, ",c") || strstr(value, "[c"))
1130                 return true;
1131         return false;
1132 }
1133
1134 static bool param_data_treat_untagged(const char *value)
1135 {
1136         if (strstr(value, ",u") || strstr(value, "[u"))
1137                 return true;
1138         return false;
1139 }
1140
1141 static void set_param_user_data(const char *name, struct symbol *sym, char *key, char *value)
1142 {
1143         struct expression *expr;
1144         struct range_list *rl = NULL;
1145         struct smatch_state *state;
1146         struct symbol *type;
1147         char *fullname;
1148
1149         expr = symbol_expression(sym);
1150         fullname = get_variable_from_key(expr, key, NULL);
1151         if (!fullname)
1152                 return;
1153
1154         type = get_member_type_from_key(expr, key);
1155         if (type && type->type == SYM_STRUCT)
1156                 return;
1157
1158         if (!type)
1159                 return;
1160
1161         str_to_rl(type, value, &rl);
1162         rl = swap_mtag_seed(expr, rl);
1163         state = alloc_estate_rl(rl);
1164         if (param_data_capped(value) || is_capped(expr))
1165                 estate_set_capped(state);
1166         if (param_data_treat_untagged(value) || sym->ctype.as == 5)
1167                 estate_set_treat_untagged(state);
1168         set_state(my_id, fullname, sym, state);
1169 }
1170
1171 static void set_called(const char *name, struct symbol *sym, char *key, char *value)
1172 {
1173         set_state(my_call_id, "this_function", NULL, &called);
1174 }
1175
1176 static void match_syscall_definition(struct symbol *sym)
1177 {
1178         struct symbol *arg;
1179         char *macro;
1180         char *name;
1181         int is_syscall = 0;
1182
1183         macro = get_macro_name(sym->pos);
1184         if (macro &&
1185             (strncmp("SYSCALL_DEFINE", macro, strlen("SYSCALL_DEFINE")) == 0 ||
1186              strncmp("COMPAT_SYSCALL_DEFINE", macro, strlen("COMPAT_SYSCALL_DEFINE")) == 0))
1187                 is_syscall = 1;
1188
1189         name = get_function();
1190         if (!option_no_db && get_state(my_call_id, "this_function", NULL) != &called) {
1191                 if (name && strncmp(name, "sys_", 4) == 0)
1192                         is_syscall = 1;
1193         }
1194
1195         if (name && strncmp(name, "compat_sys_", 11) == 0)
1196                 is_syscall = 1;
1197
1198         if (!is_syscall)
1199                 return;
1200
1201         FOR_EACH_PTR(sym->ctype.base_type->arguments, arg) {
1202                 set_state(my_id, arg->ident->name, arg, alloc_estate_whole(get_real_base_type(arg)));
1203         } END_FOR_EACH_PTR(arg);
1204 }
1205
1206 #define OLD 0
1207 #define NEW 1
1208
1209 static void store_user_data_return(struct expression *expr, char *key, char *value, bool is_new)
1210 {
1211         struct smatch_state *state;
1212         struct range_list *rl;
1213         struct symbol *type;
1214         char buf[48];
1215
1216         if (key[0] != '$')
1217                 return;
1218
1219         type = get_type(expr);
1220         snprintf(buf, sizeof(buf), "return %p%s", expr, key + 1);
1221         call_results_to_rl(expr, type, value, &rl);
1222
1223         state = alloc_estate_rl(rl);
1224         if (is_new)
1225                 estate_set_new(state);
1226
1227         set_state(my_id, buf, NULL, state);
1228 }
1229
1230 // FIXME: not a fan of this name, would prefer set_to_user_data() but that's
1231 //        already used.
1232 void mark_as_user_data(struct expression *expr, bool isnew)
1233 {
1234         struct smatch_state *state;
1235
1236         state = alloc_estate_whole(get_type(expr));
1237         if (isnew)
1238                 estate_set_new(state);
1239         set_state_expr(my_id, expr, state);
1240 }
1241
1242 static void set_to_user_data(struct expression *expr, char *key, char *value, bool is_new)
1243 {
1244         struct smatch_state *state;
1245         char *name;
1246         struct symbol *sym;
1247         struct symbol *type;
1248         struct range_list *rl = NULL;
1249
1250         type = get_member_type_from_key(expr, key);
1251         name = get_variable_from_key(expr, key, &sym);
1252         if (!name || !sym)
1253                 goto free;
1254
1255         call_results_to_rl(expr, type, value, &rl);
1256
1257         state = alloc_estate_rl(rl);
1258         if (param_data_capped(value))
1259                 estate_set_capped(state);
1260         if (param_data_treat_untagged(value))
1261                 estate_set_treat_untagged(state);
1262         if (is_new)
1263                 estate_set_new(state);
1264         estate_set_assigned(state);
1265         set_state(my_id, name, sym, state);
1266 free:
1267         free_string(name);
1268 }
1269
1270 static void returns_param_user_data(struct expression *expr, int param, char *key, char *value)
1271 {
1272         struct expression *arg;
1273         struct expression *call;
1274
1275         call = expr;
1276         while (call->type == EXPR_ASSIGNMENT)
1277                 call = strip_expr(call->right);
1278         if (call->type != EXPR_CALL)
1279                 return;
1280
1281         if (!we_pass_user_data(call))
1282                 return;
1283
1284         if (param == -1) {
1285                 if (expr->type != EXPR_ASSIGNMENT) {
1286                         // TODO: faked_assign this should all be handled as a fake assignment
1287                         store_user_data_return(expr, key, value, OLD);
1288                         return;
1289                 }
1290                 set_to_user_data(expr->left, key, value, OLD);
1291                 return;
1292         }
1293
1294         arg = get_argument_from_call_expr(call->args, param);
1295         if (!arg)
1296                 return;
1297         set_to_user_data(arg, key, value, OLD);
1298 }
1299
1300 static void returns_param_user_data_set(struct expression *expr, int param, char *key, char *value)
1301 {
1302         struct expression *arg;
1303
1304         if (param == -1) {
1305                 if (expr->type != EXPR_ASSIGNMENT) {
1306                         store_user_data_return(expr, key, value, NEW);
1307                         return;
1308                 }
1309                 set_to_user_data(expr->left, key, value, NEW);
1310                 return;
1311         }
1312
1313         while (expr->type == EXPR_ASSIGNMENT)
1314                 expr = strip_expr(expr->right);
1315         if (expr->type != EXPR_CALL)
1316                 return;
1317
1318         arg = get_argument_from_call_expr(expr->args, param);
1319         if (!arg)
1320                 return;
1321         set_to_user_data(arg, key, value, NEW);
1322 }
1323
1324 static void set_param_key_user_data(struct expression *expr, const char *name,
1325                                     struct symbol *sym, void *data)
1326 {
1327         struct expression *arg;
1328
1329         arg = gen_expression_from_name_sym(name, sym);
1330         set_state_expr(my_id, arg, new_state(get_type(arg)));
1331 }
1332
1333 static void match_capped(struct expression *expr, const char *name, struct symbol *sym, void *info)
1334 {
1335         struct smatch_state *state, *new;
1336
1337         state = get_state(my_id, name, sym);
1338         if (!state || estate_capped(state))
1339                 return;
1340
1341         new = clone_estate(state);
1342         estate_set_capped(new);
1343
1344         set_state(my_id, name, sym, new);
1345 }
1346
1347 void register_kernel_user_data(int id)
1348 {
1349         struct user_fn_info *info;
1350         int i;
1351
1352         my_id = id;
1353
1354         if (option_project != PROJ_KERNEL)
1355                 return;
1356
1357         set_dynamic_states(my_id);
1358
1359         add_unmatched_state_hook(my_id, &empty_state);
1360         add_extra_nomod_hook(&extra_nomod_hook);
1361         add_pre_merge_hook(my_id, &pre_merge_hook);
1362         add_merge_hook(my_id, &merge_estates);
1363
1364         add_function_hook("sscanf", &match_sscanf, NULL);
1365
1366         add_hook(&match_syscall_definition, AFTER_DEF_HOOK);
1367
1368         add_hook(&match_assign, ASSIGNMENT_HOOK);
1369         select_return_states_hook(PARAM_SET, &db_param_set);
1370         add_hook(&match_condition, CONDITION_HOOK);
1371
1372         add_caller_info_callback(my_id, caller_info_callback);
1373         add_return_info_callback(my_id, return_info_callback);
1374         select_caller_info_hook(set_param_user_data, USER_DATA);
1375         select_return_states_hook(USER_DATA, &returns_param_user_data);
1376         select_return_states_hook(USER_DATA_SET, &returns_param_user_data_set);
1377
1378         select_return_param_key(CAPPED_DATA, &match_capped);
1379         add_function_param_key_hook_late("memcpy", &match_capped, 2, "$", NULL);
1380         add_function_param_key_hook_late("_memcpy", &match_capped, 2, "$", NULL);
1381         add_function_param_key_hook_late("__memcpy", &match_capped, 2, "$", NULL);
1382         add_function_param_key_hook_late("memset", &match_capped, 2, "$", NULL);
1383         add_function_param_key_hook_late("_memset", &match_capped, 2, "$", NULL);
1384         add_function_param_key_hook_late("__memset", &match_capped, 2, "$", NULL);
1385
1386         for (i = 0; i < ARRAY_SIZE(func_table); i++) {
1387                 info = &func_table[i];
1388                 add_function_param_key_hook_late(info->name, &set_param_key_user_data,
1389                                                  info->param, info->key, info);
1390         }
1391 }
1392
1393 void register_kernel_user_data2(int id)
1394 {
1395         my_call_id = id;
1396
1397         if (option_project != PROJ_KERNEL)
1398                 return;
1399         select_caller_info_hook(set_called, INTERNAL);
1400 }