check_overflow.c

   1 /*
   2  * smatch/check_overflow.c
   3  *
   4  * Copyright (C) 2010 Dan Carpenter.
   5  *
   6  * Licensed under the Open Software License version 1.1
   7  *
   8  */
   9
  10 #include <stdlib.h>
  11 #include "parse.h"
  12 #include "smatch.h"
  13 #include "smatch_slist.h"
  14 #include "smatch_extra.h"
  15
  16 struct bound {
  17         int param;
  18         int size;
  19 };
  20
  21 /*
  22  * This check has two smatch IDs.
  23  * my_used_id - keeps a record of array offsets that have been used.
  24  *              If the code checks that they are within bounds later on,
  25  *              we complain about using an array offset before checking
  26  *              that it is within bounds.
  27  */
  28 static int my_used_id;
  29
  30 static struct symbol *this_func;
  31
  32 static void match_function_def(struct symbol *sym)
  33 {
  34         this_func = sym;
  35 }
  36
  37 struct limiter {
  38         int buf_arg;
  39         int limit_arg;
  40 };
  41 static struct limiter b0_l2 = {0, 2};
  42 static struct limiter b1_l2 = {1, 2};
  43
  44 static void print_args(struct expression *expr, int size)
  45 {
  46         struct symbol *sym;
  47         char *name;
  48         struct symbol *arg;
  49         const char *arg_name;
  50         int i;
  51
  52         if (!option_info)
  53                 return;
  54
  55         name = get_variable_from_expr(expr, &sym);
  56         if (!name || !sym)
  57                 goto free;
  58
  59         i = 0;
  60         FOR_EACH_PTR(this_func->ctype.base_type->arguments, arg) {
  61                 arg_name = (arg->ident?arg->ident->name:"-");
  62                 if (sym == arg && !strcmp(name, arg_name)) {
  63                         sm_info("param %d array index. size %d", i, size);
  64                         goto free;
  65                 }
  66                 i++;
  67         } END_FOR_EACH_PTR(arg);
  68 free:
  69         free_string(name);
  70 }
  71
  72 static void delete(const char *name, struct symbol *sym, struct expression *expr, void *unused)
  73 {
  74         delete_state(my_used_id, name, sym);
  75 }
  76
  77 static int definitely_just_used_as_limiter(struct expression *array, struct expression *offset)
  78 {
  79         long long val;
  80         struct expression *tmp;
  81         int step = 0;
  82         int dot_ops = 0;
  83
  84         if (!get_implied_value(offset, &val))
  85                 return 0;
  86         if (get_array_size(array) != val)
  87                 return 0;
  88
  89         FOR_EACH_PTR_REVERSE(big_expression_stack, tmp) {
  90                 if (step == 0) {
  91                         step = 1;
  92                         continue;
  93                 }
  94                 if (tmp->type == EXPR_PREOP && tmp->op == '(')
  95                         continue;
  96                 if (tmp->op == '.' && !dot_ops++)
  97                         continue;
  98                 if (step == 1 && tmp->op == '&') {
  99                         step = 2;
 100                         continue;
 101                 }
 102                 if (step == 2 && tmp->type == EXPR_COMPARE)
 103                         return 1;
 104                 return 0;
 105         } END_FOR_EACH_PTR_REVERSE(tmp);
 106         return 0;
 107 }
 108
 109 static void array_check(struct expression *expr)
 110 {
 111         struct expression *array_expr;
 112         int array_size;
 113         struct expression *offset;
 114         long long max;
 115         char *name;
 116
 117         expr = strip_expr(expr);
 118         if (!is_array(expr))
 119                 return;
 120
 121         array_expr = strip_parens(expr->unop->left);
 122         array_size = get_array_size(array_expr);
 123         if (!array_size || array_size == 1)
 124                 return;
 125
 126         offset = get_array_offset(expr);
 127         if (!get_fuzzy_max(offset, &max)) {
 128                 if (getting_address())
 129                         return;
 130                 set_state_expr(my_used_id, offset, alloc_state_num(array_size));
 131                 add_modification_hook_expr(my_used_id, offset, &delete, NULL);
 132                 print_args(offset, array_size);
 133         } else if (array_size <= max) {
 134                 const char *level = "error";
 135
 136                 if (getting_address())
 137                         level = "warn";
 138
 139                 if (definitely_just_used_as_limiter(array_expr, offset))
 140                         return;
 141
 142                 if (!option_spammy) {
 143                         struct smatch_state *state;
 144
 145                         state = get_state_expr(SMATCH_EXTRA, offset);
 146                         if (state && is_whole_range(state))
 147                                 return;
 148                 }
 149
 150                 name = get_variable_from_expr_complex(array_expr, NULL);
 151                 /* Blast.  Smatch can't figure out glibc's strcmp __strcmp_cg()
 152                  * so it prints an error every time you compare to a string
 153                  * literal array with 4 or less chars.
 154                  */
 155                 if (name && strcmp(name, "__s1") && strcmp(name, "__s2")) {
 156                         sm_msg("%s: buffer overflow '%s' %d <= %lld",
 157                                 level, name, array_size, max);
 158                 }
 159                 free_string(name);
 160         }
 161 }
 162
 163 static void match_condition(struct expression *expr)
 164 {
 165         int left;
 166         long long val;
 167         struct state_list *slist;
 168         struct sm_state *tmp;
 169         int boundary;
 170
 171         if (!expr || expr->type != EXPR_COMPARE)
 172                 return;
 173         if (get_implied_value(expr->left, &val))
 174                 left = 1;
 175         else if (get_implied_value(expr->right, &val))
 176                 left = 0;
 177         else
 178                 return;
 179
 180         if (left)
 181                 slist = get_possible_states_expr(my_used_id, expr->right);
 182         else
 183                 slist = get_possible_states_expr(my_used_id, expr->left);
 184         if (!slist)
 185                 return;
 186         FOR_EACH_PTR(slist, tmp) {
 187                 if (tmp->state == &merged)
 188                         continue;
 189                 boundary = (int)tmp->state->data;
 190                 boundary -= val;
 191                 if (boundary < 1 && boundary > -1) {
 192                         char *name;
 193
 194                         name = get_variable_from_expr((left ? expr->right : expr->left), NULL);
 195                         sm_msg("error: testing array offset '%s' after use.", name);
 196                         return;
 197                 }
 198         } END_FOR_EACH_PTR(tmp);
 199 }
 200
 201 static void match_strcpy(const char *fn, struct expression *expr, void *unused)
 202 {
 203         struct expression *dest;
 204         struct expression *data;
 205         char *dest_name = NULL;
 206         char *data_name = NULL;
 207         int dest_size;
 208         int data_size;
 209
 210         dest = get_argument_from_call_expr(expr->args, 0);
 211         data = get_argument_from_call_expr(expr->args, 1);
 212         dest_size = get_array_size_bytes(dest);
 213         data_size = get_array_size_bytes(data);
 214
 215         if (!dest_size)
 216                 return;
 217
 218         /* If the size of both arrays is known and the destination
 219          * buffer is larger than the source buffer, we're okay.
 220          */
 221         if (data_size && dest_size >= data_size)
 222                 return;
 223
 224         dest_name = get_variable_from_expr_complex(dest, NULL);
 225         data_name = get_variable_from_expr_complex(data, NULL);
 226
 227         if (data_size)
 228                 sm_msg("error: %s() '%s' too large for '%s' (%d vs %d)",
 229                         fn, data_name, dest_name, data_size, dest_size);
 230         else if (option_spammy)
 231                 sm_msg("warn: %s() '%s' of unknown size might be too large for '%s'",
 232                         fn, data_name, dest_name);
 233
 234         free_string(dest_name);
 235         free_string(data_name);
 236 }
 237
 238 static void match_limited(const char *fn, struct expression *expr, void *_limiter)
 239 {
 240         struct limiter *limiter = (struct limiter *)_limiter;
 241         struct expression *dest;
 242         struct expression *data;
 243         char *dest_name = NULL;
 244         long long needed;
 245         int has;
 246
 247         dest = get_argument_from_call_expr(expr->args, limiter->buf_arg);
 248         data = get_argument_from_call_expr(expr->args, limiter->limit_arg);
 249         if (!get_fuzzy_max(data, &needed))
 250                 return;
 251         has = get_array_size_bytes(dest);
 252         if (!has)
 253                 return;
 254         if (has >= needed)
 255                 return;
 256
 257         dest_name = get_variable_from_expr_complex(dest, NULL);
 258         sm_msg("error: %s() '%s' too small (%d vs %lld)", fn, dest_name, has, needed);
 259         free_string(dest_name);
 260 }
 261
 262 static void match_array_func(const char *fn, struct expression *expr, void *info)
 263 {
 264         struct bound *bound_info = (struct bound *)info;
 265         struct expression *arg;
 266         long long offset;
 267
 268         arg = get_argument_from_call_expr(expr->args, bound_info->param);
 269         if (!get_implied_max(arg, &offset))
 270                 return;
 271         if (offset >= bound_info->size)
 272                 sm_msg("error: buffer overflow calling %s. param %d.  %lld >= %d",
 273                         fn, bound_info->param, offset, bound_info->size);
 274 }
 275
 276 static void register_array_funcs(void)
 277 {
 278         struct token *token;
 279         const char *func;
 280         struct bound *bound_info = NULL;
 281         char name[256];
 282
 283         snprintf(name, 256, "%s.array_bounds", option_project_str);
 284         token = get_tokens_file(name);
 285         if (!token)
 286                 return;
 287         if (token_type(token) != TOKEN_STREAMBEGIN)
 288                 return;
 289         token = token->next;
 290         while (token_type(token) != TOKEN_STREAMEND) {
 291                 bound_info = malloc(sizeof(*bound_info));
 292                 if (token_type(token) != TOKEN_IDENT)
 293                         break;
 294                 func = show_ident(token->ident);
 295                 token = token->next;
 296                 if (token_type(token) != TOKEN_NUMBER)
 297                         break;
 298                 bound_info->param = atoi(token->number);
 299                 token = token->next;
 300                 if (token_type(token) != TOKEN_NUMBER)
 301                         break;
 302                 bound_info->size = atoi(token->number);
 303                 add_function_hook(func, &match_array_func, bound_info);
 304                 token = token->next;
 305         }
 306         if (token_type(token) != TOKEN_STREAMEND) {
 307                 printf("failed to load %s", name);
 308                 free(bound_info);
 309         }
 310         clear_token_alloc();
 311 }
 312
 313 void check_overflow(int id)
 314 {
 315         my_used_id = id;
 316         add_hook(&match_function_def, FUNC_DEF_HOOK);
 317         add_hook(&array_check, OP_HOOK);
 318         add_hook(&match_condition, CONDITION_HOOK);
 319         add_function_hook("strcpy", &match_strcpy, NULL);
 320         add_function_hook("strncpy", &match_limited, &b0_l2);
 321         add_function_hook("memset", &match_limited, &b0_l2);
 322         if (option_project == PROJ_KERNEL) {
 323                 add_function_hook("copy_to_user", &match_limited, &b0_l2);
 324                 add_function_hook("copy_to_user", &match_limited, &b1_l2);
 325                 add_function_hook("_copy_to_user", &match_limited, &b0_l2);
 326                 add_function_hook("_copy_to_user", &match_limited, &b1_l2);
 327                 add_function_hook("__copy_to_user", &match_limited, &b0_l2);
 328                 add_function_hook("__copy_to_user", &match_limited, &b1_l2);
 329                 add_function_hook("copy_from_user", &match_limited, &b0_l2);
 330                 add_function_hook("copy_from_user", &match_limited, &b1_l2);
 331                 add_function_hook("_copy_from_user", &match_limited, &b0_l2);
 332                 add_function_hook("_copy_from_user", &match_limited, &b1_l2);
 333                 add_function_hook("__copy_from_user", &match_limited, &b0_l2);
 334                 add_function_hook("__copy_from_user", &match_limited, &b1_l2);
 335                 add_function_hook("__builtin_memset", &match_limited, &b0_l2);
 336         }
 337         if (option_spammy)
 338                 register_array_funcs();
 339 }