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 char *get_function_call(struct expression *expr)
 199 {
 200         if (expr->type != EXPR_CALL)
 201                 return NULL;
 202         if (expr->fn->type == EXPR_SYMBOL && expr->fn->symbol)
 203                 return expr->fn->symbol->ident->name;
 204         return NULL;
 205 }
 206
 207 static void match_function_call_after(struct expression *expr)
 208 {
 209         struct expression *tmp;
 210         struct symbol *sym;
 211         char *name;
 212         struct symbol *func = NULL;
 213         int i;
 214
 215         if (expr->fn->type == EXPR_SYMBOL)
 216                 func = expr->fn->symbol;
 217
 218         i = 0;
 219         FOR_EACH_PTR(expr->args, tmp) {
 220                 tmp = strip_expr(tmp);
 221                 if (tmp->op == '&') {
 222                         name = get_variable_from_expr(tmp->unop, &sym);
 223                         if (name) {
 224                                 set_state(name, my_id, sym, &assumed_nonnull);
 225                         }
 226                         free_string(name);
 227                 } else if (func) {
 228                         name = get_variable_from_expr(tmp, &sym);
 229                         if (name && is_maybe_null_no_arg(name, sym))
 230                                 add_param(&calls, func, i, get_lineno());
 231                         free_string(name);
 232                 }
 233                 i++;
 234         } END_FOR_EACH_PTR(tmp);
 235 }
 236
 237 static int check_null_returns(const char *name, struct symbol *sym,
 238                         struct expression *right)
 239 {
 240         int i;
 241         char *func_name;
 242
 243         func_name = get_function_call(right);
 244         if (!func_name)
 245                 return 0;
 246
 247         for(i = 0; i < sizeof(*return_null)/sizeof(return_null[0]); i++) {
 248                 if (!strcmp(func_name, return_null[i])) {
 249                         set_state(name, my_id, sym, &undefined);
 250                         return 1;
 251                 }
 252         }
 253         return 0;
 254 }
 255
 256 static int assign_seen;
 257 static void match_assign(struct expression *expr)
 258 {
 259         struct expression *left, *right;
 260         struct symbol *sym;
 261         char *name;
 262
 263         if (assign_seen) {
 264                 assign_seen--;
 265                 return;
 266         }
 267         left = strip_expr(expr->left);
 268         name = get_variable_from_expr(left, &sym);
 269         if (!name)
 270                 return;
 271         right = strip_expr(expr->right);
 272         if (is_zero(right)) {
 273                 set_state(name, my_id, sym, &isnull);
 274                 free_string(name);
 275                 return;
 276         }
 277         if (check_null_returns(name, sym, right)) {
 278                 free_string(name);
 279                 return;
 280         }
 281
 282         /* by default we assume it's assigned something nonnull */
 283         set_state(name, my_id, sym, &assumed_nonnull);
 284         free_string(name);
 285 }
 286
 287 /*
 288  * set_new_true_false_paths() is used in the following conditions
 289  * if (a) { ... if (a) { ... } }
 290  * The problem is that after the second blog a is set to undefined
 291  * even though the second condition is meaning less.  (The second test
 292  * could be a macro for example).
 293  *
 294  * Also, stuff passed to the condition hook is processed behind the scenes
 295  * a bit.  Instead of a condition like (foo == 0) which is true when foo is
 296  * null, we get just (foo) and the true and false states are adjusted later.
 297  * Basically the important thing to remember, is that for us, true is always
 298  * non null and false is always null.
 299  */
 300 static void set_new_true_false_paths(const char *name, struct symbol *sym)
 301 {
 302         struct smatch_state *tmp;
 303
 304         tmp = get_state(name, my_id, sym);
 305
 306         SM_DEBUG("set_new_stuff called at for %s on line %d value='%s'\n",
 307                 name, get_lineno(), show_state(tmp));
 308
 309         if (tmp == &argument) {
 310                 set_true_false_states(name, my_id, sym, &arg_nonnull, &arg_null);
 311                 return;
 312         }
 313
 314         if (!tmp || is_maybe_null(name, sym)) {
 315                 set_true_false_states(name, my_id, sym, &nonnull, &isnull);
 316                 return;
 317         }
 318 }
 319
 320
 321 static void match_condition(struct expression *expr)
 322 {
 323         struct symbol *sym;
 324         char *name;
 325
 326         expr = strip_expr(expr);
 327         switch(expr->type) {
 328         case EXPR_PREOP:
 329         case EXPR_SYMBOL:
 330         case EXPR_DEREF:
 331                 name = get_variable_from_expr(expr, &sym);
 332                 if (!name)
 333                         return;
 334                 set_new_true_false_paths(name, sym);
 335                 free_string(name);
 336                 return;
 337         case EXPR_ASSIGNMENT:
 338                 assign_seen++;
 339                 /*
 340                  * There is a kernel macro that does
 341                  *  for ( ... ; ... || x = NULL ; ) ...
 342                  */
 343                 if (is_zero(expr->right)) {
 344                         name = get_variable_from_expr(expr->left, &sym);
 345                         if (!name)
 346                                 return;
 347                         set_true_false_states(name, my_id, sym, NULL, &isnull);
 348                         free_string(name);
 349                         return;
 350                 }
 351                  /* You have to deal with stuff like if (a = b = c) */
 352                 match_condition(expr->right);
 353                 match_condition(expr->left);
 354                 return;
 355         default:
 356                 return;
 357         }
 358 }
 359
 360 static void match_declarations(struct symbol *sym)
 361 {
 362         const char *name;
 363
 364         if ((get_base_type(sym))->type == SYM_ARRAY) {
 365                 return;
 366         }
 367
 368         name = sym->ident->name;
 369
 370         if (sym->initializer) {
 371                 if (is_zero(sym->initializer)) {
 372                         set_state(name, my_id, sym, &isnull);
 373                         return;
 374                 }
 375                 if (check_null_returns(name, sym, strip_expr(sym->initializer)))
 376                         return;
 377                 set_state(name, my_id, sym, &assumed_nonnull);
 378         } else {
 379                 set_state(name, my_id, sym, &undefined);
 380         }
 381 }
 382
 383 static void match_dereferences(struct expression *expr)
 384 {
 385         char *deref = NULL;
 386         struct symbol *sym = NULL;
 387
 388         if (strcmp(show_special(expr->deref->op), "*"))
 389                 return;
 390
 391         deref = get_variable_from_expr(expr->deref->unop, &sym);
 392         if (!deref)
 393                 return;
 394
 395         if (is_maybe_null_arg(deref, sym)) {
 396                 add_do_not_call(sym, get_lineno());
 397                 set_state(deref, my_id, sym, &assumed_nonnull);
 398         } else if  (is_maybe_null(deref, sym)) {
 399                 smatch_msg("error: dereferencing undefined:  '%s'", deref);
 400                 set_state(deref, my_id, sym, &ignore);
 401         }
 402         free_string(deref);
 403 }
 404
 405 static void end_file_processing(void)
 406 {
 407         struct func_n_param *param1, *param2;
 408
 409         // if a function is not static print it out...
 410         FOR_EACH_PTR(do_not_call, param1) {
 411                 if (!(param1->func->ctype.modifiers & MOD_STATIC))
 412                         printf("%s +%d info: unchecked param %s %d\n",
 413                                get_filename(), param1->line,
 414                                param1->func->ident->name, param1->param);
 415         } END_FOR_EACH_PTR(param1);
 416
 417         // if there is an error print it out...
 418         FOR_EACH_PTR(calls, param1) {
 419                 FOR_EACH_PTR(do_not_call, param2) {
 420                         if (param1->func == param2->func &&
 421                             param1->param == param2->param)
 422                                 printf("%s +%d error: cross_func deref %s %d\n",
 423                                        get_filename(), param1->line,
 424                                        param1->func->ident->name,
 425                                        param1->param);
 426                 } END_FOR_EACH_PTR(param2);
 427         } END_FOR_EACH_PTR(param1);
 428
 429         // if someone calls a non-static function print that..
 430         FOR_EACH_PTR(calls, param1) {
 431                 int defined = 0;
 432
 433                 FOR_EACH_PTR(funcs, param2) {
 434                         if (param1->func == param2->func)
 435                                 defined = 1;
 436                 } END_FOR_EACH_PTR(param2);
 437                 if (!defined)
 438                         printf("%s +%d info: undefined param %s %d\n",
 439                                get_filename(), param1->line,
 440                                param1->func->ident->name, param1->param);
 441         } END_FOR_EACH_PTR(param1);
 442 }
 443
 444 void check_null_deref(int id)
 445 {
 446         my_id = id;
 447         add_merge_hook(my_id, &merge_func);
 448         add_unmatched_state_hook(my_id, &unmatched_state);
 449         add_hook(&match_function_def, FUNC_DEF_HOOK);
 450         add_hook(&match_function_call_after, FUNCTION_CALL_AFTER_HOOK);
 451         add_hook(&match_assign, ASSIGNMENT_AFTER_HOOK);
 452         add_hook(&match_condition, CONDITION_HOOK);
 453         add_hook(&match_dereferences, DEREF_HOOK);
 454         add_hook(&match_declarations, DECLARATION_HOOK);
 455         add_hook(&end_file_processing, END_FILE_HOOK);
 456 }