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(my_id, name, 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(my_id, name, 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(my_id, name, 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(my_id, arg->ident->name, 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->type == EXPR_PREOP && tmp->op == '&') {
 213                         set_state_expr(my_id, tmp->unop, &assumed_nonnull);
 214                 } else if (func) {
 215                         name = get_variable_from_expr(tmp, &sym);
 216                         if (final_pass && name && is_maybe_null_no_arg(name, sym))
 217                                 add_param(&calls, func, i, get_lineno());
 218                         free_string(name);
 219                 }
 220                 i++;
 221         } END_FOR_EACH_PTR(tmp);
 222 }
 223
 224 static void match_assign_returns_null(const char *fn, struct expression *expr,
 225                                       void *unused)
 226 {
 227         set_state_expr(my_id, expr->left, &undefined);
 228 }
 229
 230 static void match_assign(struct expression *expr)
 231 {
 232         if (is_zero(expr->right)) {
 233                 set_state_expr(my_id, expr->left, &isnull);
 234                 return;
 235         }
 236         /* hack alert.  we set the state to here but it might get
 237            set later to &undefined in match_function_assign().
 238            By default we assume it's assigned something nonnull here. */
 239         set_state_expr(my_id, expr->left, &assumed_nonnull);
 240 }
 241
 242 /*
 243  * set_new_true_false_paths() is used in the following conditions
 244  * if (a) { ... if (a) { ... } }
 245  * The problem is that after the second blog a is set to undefined
 246  * even though the second condition is meaning less.  (The second test
 247  * could be a macro for example).
 248  *
 249  * Also, stuff passed to the condition hook is processed behind the scenes
 250  * a bit.  Instead of a condition like (foo == 0) which is true when foo is
 251  * null, we get just (foo) and the true and false states are adjusted later.
 252  * Basically the important thing to remember, is that for us, true is always
 253  * non null and false is always null.
 254  */
 255 static void set_new_true_false_paths(struct expression *expr, int recently_assigned)
 256 {
 257         struct smatch_state *tmp;
 258         char *name;
 259         struct symbol *sym;
 260
 261
 262         name = get_variable_from_expr(expr, &sym);
 263         if (!name || !sym)
 264                 goto free;
 265
 266         tmp = get_state_expr(my_id, expr);
 267         sm_debug("set_new_stuff called at for %s on line %d value='%s'\n",
 268                 name, get_lineno(), show_state(tmp));
 269
 270         if (tmp == &argument) {
 271                 set_true_false_states_expr(my_id, expr, &arg_nonnull, &arg_null);
 272                 goto free;
 273         }
 274
 275         if (tmp == &ignore) {
 276                 set_true_false_states_expr(my_id, expr, &ignore, &ignore);
 277                 goto free;
 278         }
 279
 280         if (!tmp || recently_assigned || is_maybe_null(name, sym)) {
 281                 set_true_false_states_expr(my_id, expr, &nonnull, &isnull);
 282                 goto free;
 283         }
 284 free:
 285         free_string(name);
 286
 287 }
 288
 289 static void match_condition(struct expression *expr)
 290 {
 291         expr = strip_expr(expr);
 292         switch (expr->type) {
 293         case EXPR_PREOP:
 294         case EXPR_SYMBOL:
 295         case EXPR_DEREF:
 296                 set_new_true_false_paths(expr, 0);
 297                 return;
 298         case EXPR_ASSIGNMENT:
 299                 /*
 300                  * There is a kernel macro that does
 301                  *  for ( ... ; ... || x = NULL ; ) ...
 302                  */
 303                 if (is_zero(expr->right)) {
 304                         set_true_false_states_expr(my_id, expr->left, NULL, &isnull);
 305                         return;
 306                 }
 307                 set_new_true_false_paths(expr->left, 1);
 308                 return;
 309         }
 310 }
 311
 312 static void match_declarations(struct symbol *sym)
 313 {
 314         const char *name;
 315
 316         if ((get_base_type(sym))->type == SYM_ARRAY) {
 317                 return;
 318         }
 319
 320         name = sym->ident->name;
 321
 322         if (!sym->initializer) {
 323                 set_state(my_id, name, sym, &undefined);
 324                 scoped_state(my_id, name, sym);
 325         }
 326 }
 327
 328 static void match_dereferences(struct expression *expr)
 329 {
 330         char *deref = NULL;
 331         struct symbol *sym = NULL;
 332
 333         if (expr->type != EXPR_PREOP)
 334                 return;
 335         deref = get_variable_from_expr(expr->unop, &sym);
 336         if (!deref)
 337                 return;
 338
 339         if (is_maybe_null_arg(deref, sym)) {
 340                 add_do_not_call(sym, get_lineno());
 341                 set_state(my_id, deref, sym, &assumed_nonnull);
 342         } else if (is_maybe_null(deref, sym)) {
 343                 sm_msg("error: dereferencing undefined:  '%s'", deref);
 344                 set_state(my_id, deref, sym, &ignore);
 345         }
 346         free_string(deref);
 347 }
 348
 349 static void end_file_processing(void)
 350 {
 351         struct func_n_param *param1, *param2;
 352
 353         // if there is an error print it out...
 354         FOR_EACH_PTR(calls, param1) {
 355                 FOR_EACH_PTR(do_not_call, param2) {
 356                         if (param1->func == param2->func &&
 357                             param1->param == param2->param)
 358                                 sm_printf("%s +%d error: cross_func deref %s %d\n",
 359                                        get_filename(), param1->line,
 360                                        param1->func->ident->name,
 361                                        param1->param);
 362                 } END_FOR_EACH_PTR(param2);
 363         } END_FOR_EACH_PTR(param1);
 364
 365         if (!option_info)
 366                 return;
 367
 368         // if a function is not static print it out...
 369         FOR_EACH_PTR(do_not_call, param1) {
 370                 if (!(param1->func->ctype.modifiers & MOD_STATIC))
 371                         sm_printf("%s +%d info: unchecked param %s %d\n",
 372                                get_filename(), param1->line,
 373                                param1->func->ident->name, param1->param);
 374         } END_FOR_EACH_PTR(param1);
 375
 376         // if someone calls a non-static function print that..
 377         FOR_EACH_PTR(calls, param1) {
 378                 int defined = 0;
 379
 380                 FOR_EACH_PTR(funcs, param2) {
 381                         if (param1->func == param2->func)
 382                                 defined = 1;
 383                 } END_FOR_EACH_PTR(param2);
 384                 if (!defined)
 385                         sm_printf("%s +%d info: undefined param %s %d\n",
 386                                get_filename(), param1->line,
 387                                param1->func->ident->name, param1->param);
 388         } END_FOR_EACH_PTR(param1);
 389 }
 390
 391 static void register_allocation_funcs(void)
 392 {
 393         struct token *token;
 394         const char *func;
 395
 396         token = get_tokens_file("kernel.allocation_funcs");
 397         if (!token)
 398                 return;
 399         if (token_type(token) != TOKEN_STREAMBEGIN)
 400                 return;
 401         token = token->next;
 402         while (token_type(token) != TOKEN_STREAMEND) {
 403                 if (token_type(token) != TOKEN_IDENT)
 404                         return;
 405                 func = show_ident(token->ident);
 406                 add_function_assign_hook(func, &match_assign_returns_null,
 407                                          NULL);
 408                 token = token->next;
 409         }
 410         clear_token_alloc();
 411 }
 412
 413 void check_null_deref(int id)
 414 {
 415         int i;
 416
 417         my_id = id;
 418         set_default_state(my_id, &assumed_nonnull);
 419         add_merge_hook(my_id, &merge_func);
 420         add_unmatched_state_hook(my_id, &unmatched_state);
 421         add_hook(&match_function_def, FUNC_DEF_HOOK);
 422         add_hook(&match_function_call_after, FUNCTION_CALL_HOOK);
 423         add_hook(&match_assign, ASSIGNMENT_HOOK);
 424         add_hook(&match_condition, CONDITION_HOOK);
 425         add_hook(&match_dereferences, DEREF_HOOK);
 426         add_hook(&match_declarations, DECLARATION_HOOK);
 427         add_hook(&end_file_processing, END_FILE_HOOK);
 428
 429         for(i = 0; i < ARRAY_SIZE(return_null); i++) {
 430                 add_function_assign_hook(return_null[i],
 431                                          &match_assign_returns_null, NULL);
 432         }
 433         register_allocation_funcs();
 434 }