check_null_deref.c

   1 /*
   2  * sparse/check_deference.c
   3  *
   4  * Copyright (C) 2006 Dan Carpenter.
   5  *
   6  * Licensed under the Open Software License version 1.1
   7  *
   8  */
   9
  10 #include "token.h"
  11 #include "smatch.h"
  12 #include "smatch_slist.h"
  13
  14 /*
  15  * TODO:  The return_null list of functions should be determined automatically
  16  */
  17 static const char *return_null[] = {
  18         "kmalloc",
  19 };
  20
  21 struct func_n_param {
  22         struct symbol *func;
  23         int param;
  24         int line;
  25 };
  26 ALLOCATOR(func_n_param, "func parameters");
  27 DECLARE_PTR_LIST(param_list, struct func_n_param);
  28
  29 static struct param_list *funcs;
  30 static struct param_list *do_not_call;
  31 static struct param_list *calls;
  32
  33 static int my_id;
  34
  35 STATE(argument);
  36 STATE(arg_nonnull);
  37 STATE(arg_null);
  38 STATE(assumed_nonnull);
  39 STATE(ignore);
  40 STATE(isnull);
  41 STATE(nonnull);
  42
  43 static struct symbol *func_sym;
  44
  45 /*
  46  * merge_func() is to handle assumed_nonnull.
  47  *
  48  * Assumed_nonnull is a funny state.  It's almost the same as
  49  * no state.  Maybe it would be better to delete it completely...
  50  */
  51 static struct smatch_state *merge_func(const char *name, struct symbol *sym,
  52                                        struct smatch_state *s1,
  53                                        struct smatch_state *s2)
  54 {
  55         if (s1 == &ignore || s2 == &ignore)
  56                 return &ignore;
  57         return &merged;
  58
  59 }
  60
  61 static struct smatch_state *unmatched_state(struct sm_state *sm)
  62 {
  63         if (sm->state == &assumed_nonnull)
  64                 return &assumed_nonnull;
  65         return &undefined;
  66 }
  67
  68 static int is_maybe_null_no_arg(const char *name, struct symbol *sym)
  69 {
  70         struct state_list *slist;
  71         struct sm_state *tmp;
  72         int ret = 0;
  73
  74         slist = get_possible_states(name, my_id, sym);
  75         FOR_EACH_PTR(slist, tmp) {
  76                 if (tmp->state == &ignore)
  77                         return 0;
  78                 if (tmp->state == &isnull) {
  79                         SM_DEBUG("is_maybe_null_no_arg() says %s &isnull\n", name);
  80                         ret = 1;
  81                 }
  82                 if (tmp->state == &undefined) {
  83                         SM_DEBUG("is_maybe_null_no_arg() says %s &undefined\n", name);
  84                         ret = 1;
  85                 }
  86         } END_FOR_EACH_PTR(tmp);
  87         return ret;
  88 }
  89
  90
  91 static int is_maybe_null(const char *name, struct symbol *sym)
  92 {
  93         struct state_list *slist;
  94         struct sm_state *tmp;
  95         int ret = 0;
  96
  97         slist = get_possible_states(name, my_id, sym);
  98         FOR_EACH_PTR(slist, tmp) {
  99                 if (tmp->state == &ignore)
 100                         return 0;
 101                 if (tmp->state == &isnull) {
 102                         SM_DEBUG("is_maybe_null() says %s &isnull\n", name);
 103                         ret = 1;
 104                 }
 105                 if (tmp->state == &undefined) {
 106                         SM_DEBUG("is_maybe_null() says %s &undefined\n", name);
 107                         ret = 1;
 108                 }
 109                 if (tmp->state == &arg_null) {
 110                         SM_DEBUG("is_maybe_null() says %s &arg_null\n", name);
 111                         ret = 1;
 112                 }
 113         } END_FOR_EACH_PTR(tmp);
 114         return ret;
 115 }
 116
 117 static int is_maybe_null_arg(char *name, struct symbol *sym)
 118 {
 119         struct state_list *slist;
 120         struct sm_state *tmp;
 121         int maybe_null = 0;
 122
 123         slist = get_possible_states(name, my_id, sym);
 124         FOR_EACH_PTR(slist, tmp) {
 125                 if (tmp->state != &argument && tmp->state != &arg_null &&
 126                     tmp->state != &arg_nonnull && tmp->state !=  &merged &&
 127                     tmp->state != &assumed_nonnull)
 128                         return 0;
 129                 if (tmp->state == &argument || tmp->state == &arg_null)
 130                         maybe_null = 1;
 131         } END_FOR_EACH_PTR(tmp);
 132
 133         /* We really only care about arguments if they are null */
 134         return maybe_null;
 135 }
 136
 137 static struct func_n_param *alloc_func_n_param(struct symbol *func, int param,
 138                                                int line)
 139 {
 140         struct func_n_param *tmp = __alloc_func_n_param(0);
 141
 142         tmp->func = func;
 143         tmp->param = param;
 144         tmp->line = line;
 145         return tmp;
 146 }
 147
 148 static int get_arg_num(struct symbol *sym)
 149 {
 150         struct symbol *arg;
 151         int i = 0;
 152
 153         FOR_EACH_PTR(func_sym->ctype.base_type->arguments, arg) {
 154                 if (arg == sym) {
 155                         return i;
 156                 }
 157                 i++;
 158         } END_FOR_EACH_PTR(arg);
 159         return -1;
 160 }
 161
 162 static void add_do_not_call(struct symbol *sym, int line)
 163 {
 164         struct func_n_param *tmp;
 165         int num = get_arg_num(sym);
 166
 167         FOR_EACH_PTR(do_not_call, tmp) {
 168                 if (tmp->func == func_sym && tmp->param == num)
 169                         return;
 170         } END_FOR_EACH_PTR(tmp);
 171         tmp = alloc_func_n_param(func_sym, num, line);
 172         add_ptr_list(&do_not_call, tmp);
 173 }
 174
 175 static void add_param(struct param_list **list, struct symbol *func, int param,
 176                       int line)
 177 {
 178         struct func_n_param *tmp;
 179
 180         tmp = alloc_func_n_param(func, param, line);
 181         add_ptr_list(list, tmp);
 182 }
 183
 184 static void match_function_def(struct symbol *sym)
 185 {
 186         struct symbol *arg;
 187
 188         func_sym = sym;
 189         add_param(&funcs, sym, 0, 0);
 190         FOR_EACH_PTR(sym->ctype.base_type->arguments, arg) {
 191                 if (!arg->ident) {
 192                         continue;
 193                 }
 194                 set_state(arg->ident->name, my_id, arg, &argument);
 195         } END_FOR_EACH_PTR(arg);
 196 }
 197
 198 static void match_function_call_after(struct expression *expr)
 199 {
 200         struct expression *tmp;
 201         struct symbol *sym;
 202         char *name;
 203         struct symbol *func = NULL;
 204         int i;
 205
 206         if (expr->fn->type == EXPR_SYMBOL)
 207                 func = expr->fn->symbol;
 208
 209         i = 0;
 210         FOR_EACH_PTR(expr->args, tmp) {
 211                 tmp = strip_expr(tmp);
 212                 if (tmp->op == '&') {
 213                         name = get_variable_from_expr(tmp->unop, &sym);
 214                         if (name) {
 215                                 set_state(name, my_id, sym, &assumed_nonnull);
 216                         }
 217                         free_string(name);
 218                 } else if (func) {
 219                         name = get_variable_from_expr(tmp, &sym);
 220                         if (name && is_maybe_null_no_arg(name, sym))
 221                                 add_param(&calls, func, i, get_lineno());
 222                         free_string(name);
 223                 }
 224                 i++;
 225         } END_FOR_EACH_PTR(tmp);
 226 }
 227
 228 static void match_assign_returns_null(const char *fn, struct expression *expr,
 229                                       void *unused)
 230 {
 231         struct expression *left;
 232         struct symbol *sym;
 233         char *name;
 234
 235         left = strip_expr(expr->left);
 236         name = get_variable_from_expr(left, &sym);
 237         if (!name)
 238                 return;
 239         set_state(name, my_id, sym, &undefined);
 240         free_string(name);
 241 }
 242
 243 static void match_assign(struct expression *expr)
 244 {
 245         struct expression *left, *right;
 246         struct symbol *sym;
 247         char *name;
 248
 249         left = strip_expr(expr->left);
 250         name = get_variable_from_expr(left, &sym);
 251         if (!name)
 252                 return;
 253         right = strip_expr(expr->right);
 254         if (is_zero(right)) {
 255                 set_state(name, my_id, sym, &isnull);
 256                 free_string(name);
 257                 return;
 258         }
 259         /* hack alert.  we set the state to here but it might get
 260            set later to &undefined in match_function_assign().
 261            By default we assume it's assigned something nonnull here. */
 262         set_state(name, my_id, sym, &assumed_nonnull);
 263         free_string(name);
 264 }
 265
 266 /*
 267  * set_new_true_false_paths() is used in the following conditions
 268  * if (a) { ... if (a) { ... } }
 269  * The problem is that after the second blog a is set to undefined
 270  * even though the second condition is meaning less.  (The second test
 271  * could be a macro for example).
 272  *
 273  * Also, stuff passed to the condition hook is processed behind the scenes
 274  * a bit.  Instead of a condition like (foo == 0) which is true when foo is
 275  * null, we get just (foo) and the true and false states are adjusted later.
 276  * Basically the important thing to remember, is that for us, true is always
 277  * non null and false is always null.
 278  */
 279 static void set_new_true_false_paths(const char *name, struct symbol *sym,
 280         int recently_assigned)
 281 {
 282         struct smatch_state *tmp;
 283
 284         tmp = get_state(name, my_id, sym);
 285
 286         SM_DEBUG("set_new_stuff called at for %s on line %d value='%s'\n",
 287                 name, get_lineno(), show_state(tmp));
 288
 289         if (tmp == &argument) {
 290                 set_true_false_states(name, my_id, sym, &arg_nonnull, &arg_null);
 291                 return;
 292         }
 293
 294         if (tmp == &ignore) {
 295                 set_true_false_states(name, my_id, sym, &ignore, &ignore);
 296                 return;
 297         }
 298
 299         if (!tmp || recently_assigned || is_maybe_null(name, sym)) {
 300                 set_true_false_states(name, my_id, sym, &nonnull, &isnull);
 301                 return;
 302         }
 303 }
 304
 305 static void match_var_null_nonnull(struct expression *expr,
 306                                    int recently_assigned)
 307 {
 308         struct symbol *sym;
 309         char *name;
 310
 311         name = get_variable_from_expr(expr, &sym);
 312         if (!name)
 313                 return;
 314         set_new_true_false_paths(name, sym, recently_assigned);
 315         free_string(name);
 316 }
 317
 318 static void match_condition(struct expression *expr)
 319 {
 320         struct symbol *sym;
 321         char *name;
 322         expr = strip_expr(expr);
 323         switch(expr->type) {
 324         case EXPR_PREOP:
 325         case EXPR_SYMBOL:
 326         case EXPR_DEREF:
 327                 match_var_null_nonnull(expr, 0);
 328                 return;
 329         case EXPR_ASSIGNMENT:
 330                 /*
 331                  * There is a kernel macro that does
 332                  *  for ( ... ; ... || x = NULL ; ) ...
 333                  */
 334                 if (is_zero(expr->right)) {
 335                         name = get_variable_from_expr(expr->left, &sym);
 336                         if (!name)
 337                                 return;
 338                         set_true_false_states(name, my_id, sym, NULL, &isnull);
 339                         free_string(name);
 340                         return;
 341                 }
 342                 match_var_null_nonnull(expr->left, 1);
 343                 return;
 344         }
 345 }
 346
 347 static void match_declarations(struct symbol *sym)
 348 {
 349         const char *name;
 350
 351         if ((get_base_type(sym))->type == SYM_ARRAY) {
 352                 return;
 353         }
 354
 355         name = sym->ident->name;
 356
 357         if (sym->initializer) {
 358                 if (is_zero(sym->initializer)) {
 359                         set_state(name, my_id, sym, &isnull);
 360                         scoped_state(name, my_id, sym);
 361                         return;
 362                 }
 363                 set_state(name, my_id, sym, &assumed_nonnull);
 364                 scoped_state(name, my_id, sym);
 365         } else {
 366                 set_state(name, my_id, sym, &undefined);
 367                 scoped_state(name, my_id, sym);
 368         }
 369 }
 370
 371 static void match_dereferences(struct expression *expr)
 372 {
 373         char *deref = NULL;
 374         struct symbol *sym = NULL;
 375
 376         if (strcmp(show_special(expr->deref->op), "*"))
 377                 return;
 378
 379         deref = get_variable_from_expr(expr->deref->unop, &sym);
 380         if (!deref)
 381                 return;
 382
 383         if (is_maybe_null_arg(deref, sym)) {
 384                 add_do_not_call(sym, get_lineno());
 385                 set_state(deref, my_id, sym, &assumed_nonnull);
 386         } else if (is_maybe_null(deref, sym)) {
 387                 smatch_msg("error: dereferencing undefined:  '%s'", deref);
 388                 set_state(deref, my_id, sym, &ignore);
 389         }
 390         free_string(deref);
 391 }
 392
 393 static void end_file_processing(void)
 394 {
 395         struct func_n_param *param1, *param2;
 396
 397         // if a function is not static print it out...
 398         FOR_EACH_PTR(do_not_call, param1) {
 399                 if (!(param1->func->ctype.modifiers & MOD_STATIC))
 400                         printf("%s +%d info: unchecked param %s %d\n",
 401                                get_filename(), param1->line,
 402                                param1->func->ident->name, param1->param);
 403         } END_FOR_EACH_PTR(param1);
 404
 405         // if there is an error print it out...
 406         FOR_EACH_PTR(calls, param1) {
 407                 FOR_EACH_PTR(do_not_call, param2) {
 408                         if (param1->func == param2->func &&
 409                             param1->param == param2->param)
 410                                 printf("%s +%d error: cross_func deref %s %d\n",
 411                                        get_filename(), param1->line,
 412                                        param1->func->ident->name,
 413                                        param1->param);
 414                 } END_FOR_EACH_PTR(param2);
 415         } END_FOR_EACH_PTR(param1);
 416
 417         // if someone calls a non-static function print that..
 418         FOR_EACH_PTR(calls, param1) {
 419                 int defined = 0;
 420
 421                 FOR_EACH_PTR(funcs, param2) {
 422                         if (param1->func == param2->func)
 423                                 defined = 1;
 424                 } END_FOR_EACH_PTR(param2);
 425                 if (!defined)
 426                         printf("%s +%d info: undefined param %s %d\n",
 427                                get_filename(), param1->line,
 428                                param1->func->ident->name, param1->param);
 429         } END_FOR_EACH_PTR(param1);
 430 }
 431
 432 static void register_allocation_funcs(void)
 433 {
 434         struct token *token;
 435         const char *func;
 436
 437         token = get_tokens_file("kernel.allocation_funcs");
 438         if (!token)
 439                 return;
 440         if (token_type(token) != TOKEN_STREAMBEGIN)
 441                 return;
 442         token = token->next;
 443         while (token_type(token) != TOKEN_STREAMEND) {
 444                 if (token_type(token) != TOKEN_IDENT)
 445                         return;
 446                 func = show_ident(token->ident);
 447                 add_function_assign_hook(func, &match_assign_returns_null,
 448                                          NULL);
 449                 token = token->next;
 450         }
 451         clear_token_alloc();
 452 }
 453
 454 void check_null_deref(int id)
 455 {
 456         int i;
 457
 458         my_id = id;
 459         add_merge_hook(my_id, &merge_func);
 460         add_unmatched_state_hook(my_id, &unmatched_state);
 461         add_hook(&match_function_def, FUNC_DEF_HOOK);
 462         add_hook(&match_function_call_after, FUNCTION_CALL_HOOK);
 463         add_hook(&match_assign, ASSIGNMENT_HOOK);
 464         add_hook(&match_condition, CONDITION_HOOK);
 465         add_hook(&match_dereferences, DEREF_HOOK);
 466         add_hook(&match_declarations, DECLARATION_HOOK);
 467         add_hook(&end_file_processing, END_FILE_HOOK);
 468
 469         for(i = 0; i < sizeof(*return_null)/sizeof(return_null[0]); i++) {
 470                 add_function_assign_hook(return_null[i],
 471                                          &match_assign_returns_null, NULL);
 472         }
 473         register_allocation_funcs();
 474 }