cppcheck/lib/checkbufferoverrun.cpp

840 lines
34 KiB
C++
Raw Normal View History

/*
* Cppcheck - A tool for static C/C++ code analysis
2019-02-09 07:24:06 +01:00
* Copyright (C) 2007-2019 Cppcheck team.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
//---------------------------------------------------------------------------
// Buffer overrun..
//---------------------------------------------------------------------------
#include "checkbufferoverrun.h"
2017-05-27 04:33:47 +02:00
#include "astutils.h"
#include "library.h"
#include "mathlib.h"
2017-05-27 04:33:47 +02:00
#include "settings.h"
#include "symboldatabase.h"
2017-05-27 04:33:47 +02:00
#include "token.h"
#include "tokenize.h"
#include "tokenlist.h"
2017-05-23 14:58:43 +02:00
#include "utils.h"
2017-05-27 04:33:47 +02:00
#include "valueflow.h"
2017-05-27 04:33:47 +02:00
#include <tinyxml2.h>
#include <algorithm>
#include <cstdlib>
2017-05-27 04:33:47 +02:00
#include <sstream>
#include <stack>
2017-05-27 04:33:47 +02:00
#include <utility>
//---------------------------------------------------------------------------
2009-03-20 18:16:21 +01:00
// Register this check class (by creating a static instance of it)
2011-10-13 20:53:06 +02:00
namespace {
CheckBufferOverrun instance;
2009-03-20 18:16:21 +01:00
}
//---------------------------------------------------------------------------
// CWE ids used:
2016-08-15 18:04:55 +02:00
static const CWE CWE131(131U); // Incorrect Calculation of Buffer Size
static const CWE CWE170(170U); // Improper Null Termination
static const CWE CWE_ARRAY_INDEX_THEN_CHECK(398U); // Indicator of Poor Code Quality
2016-08-15 18:04:55 +02:00
static const CWE CWE682(682U); // Incorrect Calculation
static const CWE CWE758(758U); // Reliance on Undefined, Unspecified, or Implementation-Defined Behavior
static const CWE CWE_POINTER_ARITHMETIC_OVERFLOW(758U); // Reliance on Undefined, Unspecified, or Implementation-Defined Behavior
2019-03-23 08:41:20 +01:00
static const CWE CWE_BUFFER_UNDERRUN(786U); // Access of Memory Location Before Start of Buffer
static const CWE CWE_BUFFER_OVERRUN(788U); // Access of Memory Location After End of Buffer
//---------------------------------------------------------------------------
static const ValueFlow::Value *getBufferSizeValue(const Token *tok)
{
for (const ValueFlow::Value &value : tok->values()) {
if (value.isBufferSizeValue())
return &value;
}
return nullptr;
}
static size_t getMinFormatStringOutputLength(const std::vector<const Token*> &parameters, unsigned int formatStringArgNr)
{
if (formatStringArgNr == 0 || formatStringArgNr > parameters.size())
return 0;
if (parameters[formatStringArgNr - 1]->tokType() != Token::eString)
return 0;
const std::string &formatString = parameters[formatStringArgNr - 1]->str();
bool percentCharFound = false;
std::size_t outputStringSize = 0;
bool handleNextParameter = false;
std::string digits_string;
bool i_d_x_f_found = false;
2010-12-31 09:30:56 +01:00
std::size_t parameterLength = 0;
unsigned int inputArgNr = formatStringArgNr;
for (std::string::size_type i = 1; i + 1 < formatString.length(); ++i) {
if (formatString[i] == '\\') {
if (i < formatString.length() - 1 && formatString[i + 1] == '0')
break;
++outputStringSize;
++i;
continue;
}
2011-10-13 20:53:06 +02:00
if (percentCharFound) {
switch (formatString[i]) {
case 'f':
case 'x':
case 'X':
case 'i':
i_d_x_f_found = true;
2015-06-28 18:07:31 +02:00
handleNextParameter = true;
parameterLength = 1; // TODO
2015-06-28 18:07:31 +02:00
break;
case 'c':
case 'e':
case 'E':
case 'g':
case 'o':
case 'u':
case 'p':
case 'n':
handleNextParameter = true;
parameterLength = 1; // TODO
break;
case 'd':
i_d_x_f_found = true;
parameterLength = 1;
if (inputArgNr < parameters.size() && parameters[inputArgNr]->hasKnownIntValue())
parameterLength = MathLib::toString(parameters[inputArgNr]->getKnownIntValue()).length();
handleNextParameter = true;
break;
case 's':
parameterLength = 0;
if (inputArgNr < parameters.size() && parameters[inputArgNr]->tokType() == Token::eString)
parameterLength = Token::getStrLength(parameters[inputArgNr]);
2009-10-08 15:27:46 +02:00
handleNextParameter = true;
break;
2009-10-08 15:27:46 +02:00
}
}
2009-10-08 15:27:46 +02:00
if (formatString[i] == '%')
percentCharFound = !percentCharFound;
2011-10-13 20:53:06 +02:00
else if (percentCharFound) {
digits_string.append(1, formatString[i]);
}
if (!percentCharFound)
outputStringSize++;
2011-10-13 20:53:06 +02:00
if (handleNextParameter) {
unsigned int tempDigits = static_cast<unsigned int>(std::abs(std::atoi(digits_string.c_str())));
if (i_d_x_f_found)
tempDigits = std::max(static_cast<unsigned int>(tempDigits), 1U);
2009-10-08 15:27:46 +02:00
2011-10-13 20:53:06 +02:00
if (digits_string.find('.') != std::string::npos) {
const std::string endStr = digits_string.substr(digits_string.find('.') + 1);
const unsigned int maxLen = std::max(static_cast<unsigned int>(std::abs(std::atoi(endStr.c_str()))), 1U);
if (formatString[i] == 's') {
// For strings, the length after the dot "%.2s" will limit
// the length of the string.
if (parameterLength > maxLen)
parameterLength = maxLen;
2011-10-13 20:53:06 +02:00
} else {
// For integers, the length after the dot "%.2d" can
// increase required length
if (tempDigits < maxLen)
tempDigits = maxLen;
}
}
if (tempDigits < parameterLength)
outputStringSize += parameterLength;
2009-10-08 15:27:46 +02:00
else
outputStringSize += tempDigits;
2009-10-08 15:27:46 +02:00
parameterLength = 0;
digits_string.clear();
i_d_x_f_found = false;
percentCharFound = false;
handleNextParameter = false;
++inputArgNr;
}
}
return outputStringSize;
}
//---------------------------------------------------------------------------
static bool getDimensionsEtc(const Token * const arrayToken, const Settings *settings, std::vector<Dimension> * const dimensions, ErrorPath * const errorPath, bool * const mightBeLarger)
{
const Token *array = arrayToken;
while (Token::Match(array, ".|::"))
array = array->astOperand2();
if (!array->variable())
return false;
if (array->variable()->isArray() && !array->variable()->dimensions().empty()) {
*dimensions = array->variable()->dimensions();
if (dimensions->size() >= 1 && ((*dimensions)[0].num <= 1 || !(*dimensions)[0].tok)) {
visitAstNodes(arrayToken,
[&](const Token *child) {
if (child->originalName() == "->") {
*mightBeLarger = true;
return ChildrenToVisit::none;
}
return ChildrenToVisit::op1_and_op2;
});
}
} else if (const Token *stringLiteral = array->getValueTokenMinStrSize()) {
Dimension dim;
dim.tok = nullptr;
dim.num = Token::getStrSize(stringLiteral);
dim.known = array->hasKnownValue();
dimensions->emplace_back(dim);
} else if (array->valueType() && array->valueType()->pointer >= 1 && array->valueType()->isIntegral()) {
const ValueFlow::Value *value = getBufferSizeValue(array);
if (!value)
return false;
*errorPath = value->errorPath;
Dimension dim;
dim.known = value->isKnown();
dim.tok = nullptr;
dim.num = value->intvalue / array->valueType()->typeSize(*settings);
dimensions->emplace_back(dim);
}
return !dimensions->empty();
}
static std::vector<const ValueFlow::Value *> getOverrunIndexValues(const Token *tok, const Token *arrayToken, const std::vector<Dimension> &dimensions, const std::vector<const Token *> &indexTokens)
{
const Token *array = arrayToken;
while (Token::Match(array, ".|::"))
array = array->astOperand2();
for (int cond = 0; cond < 2; cond++) {
bool equal = false;
bool overflow = false;
bool allKnown = true;
std::vector<const ValueFlow::Value *> indexValues;
for (size_t i = 0; i < dimensions.size() && i < indexTokens.size(); ++i) {
const ValueFlow::Value *value = indexTokens[i]->getMaxValue(cond == 1);
indexValues.push_back(value);
if (!value)
continue;
if (!value->isKnown()) {
if (!allKnown)
continue;
allKnown = false;
}
if (array->variable()->isArray() && dimensions[i].num == 0)
continue;
if (value->intvalue == dimensions[i].num)
equal = true;
else if (value->intvalue > dimensions[i].num)
overflow = true;
}
if (equal && tok->str() != "[")
continue;
if (!overflow && equal) {
const Token *parent = tok;
while (Token::simpleMatch(parent, "["))
parent = parent->astParent();
if (!parent || parent->isUnaryOp("&"))
continue;
}
if (overflow || equal)
return indexValues;
}
return std::vector<const ValueFlow::Value *>();
}
void CheckBufferOverrun::arrayIndex()
{
for (const Token *tok = mTokenizer->tokens(); tok; tok = tok->next()) {
if (tok->str() != "[")
continue;
const Token *array = tok->astOperand1();
while (Token::Match(array, ".|::"))
array = array->astOperand2();
if (!array|| !array->variable() || array->variable()->nameToken() == array)
continue;
if (!array->scope()->isExecutable()) {
// LHS in non-executable scope => This is just a definition
const Token *parent = tok;
while (parent && !Token::simpleMatch(parent->astParent(), "="))
parent = parent->astParent();
if (!parent || parent == parent->astParent()->astOperand1())
continue;
}
std::vector<const Token *> indexTokens;
for (const Token *tok2 = tok; tok2 && tok2->str() == "["; tok2 = tok2->link()->next()) {
if (!tok2->astOperand2()) {
indexTokens.clear();
break;
}
indexTokens.emplace_back(tok2->astOperand2());
}
if (indexTokens.empty())
continue;
std::vector<Dimension> dimensions;
ErrorPath errorPath;
2019-03-19 13:17:27 +01:00
bool mightBeLarger = false;
if (!getDimensionsEtc(tok->astOperand1(), mSettings, &dimensions, &errorPath, &mightBeLarger))
continue;
// Positive index
if (!mightBeLarger) { // TODO check arrays with dim 1 also
const std::vector<const ValueFlow::Value *> &indexValues = getOverrunIndexValues(tok, tok->astOperand1(), dimensions, indexTokens);
if (!indexValues.empty())
arrayIndexError(tok, dimensions, indexValues);
}
// Negative index
bool neg = false;
std::vector<const ValueFlow::Value *> negativeIndexes;
for (size_t i = 0; i < indexTokens.size(); ++i) {
const ValueFlow::Value *negativeValue = indexTokens[i]->getValueLE(-1, mSettings);
negativeIndexes.emplace_back(negativeValue);
if (negativeValue)
neg = true;
}
if (neg) {
negativeIndexError(tok, dimensions, negativeIndexes);
}
}
}
static std::string stringifyIndexes(const std::string &array, const std::vector<const ValueFlow::Value *> &indexValues)
{
if (indexValues.size() == 1)
return MathLib::toString(indexValues[0]->intvalue);
std::ostringstream ret;
ret << array;
for (const ValueFlow::Value *index : indexValues) {
ret << "[";
if (index)
ret << index->intvalue;
else
ret << "*";
ret << "]";
}
return ret.str();
}
static std::string arrayIndexMessage(const Token *tok, const std::vector<Dimension> &dimensions, const std::vector<const ValueFlow::Value *> &indexValues, const Token *condition)
2019-03-11 15:32:30 +01:00
{
std::string array = tok->astOperand1()->expressionString();
for (const Dimension &dim : dimensions)
array += "[" + MathLib::toString(dim.num) + "]";
std::ostringstream errmsg;
if (condition)
errmsg << ValueFlow::eitherTheConditionIsRedundant(condition)
<< " or the array '" + array + "' is accessed at index " << stringifyIndexes(tok->astOperand1()->expressionString(), indexValues) << ", which is out of bounds.";
else
errmsg << "Array '" << array << "' accessed at index " << stringifyIndexes(tok->astOperand1()->expressionString(), indexValues) << ", which is out of bounds.";
return errmsg.str();
}
void CheckBufferOverrun::arrayIndexError(const Token *tok, const std::vector<Dimension> &dimensions, const std::vector<const ValueFlow::Value *> &indexes)
{
if (!tok) {
2019-03-23 08:41:20 +01:00
reportError(tok, Severity::error, "arrayIndexOutOfBounds", "Array 'arr[16]' accessed at index 16, which is out of bounds.", CWE_BUFFER_OVERRUN, false);
reportError(tok, Severity::warning, "arrayIndexOutOfBoundsCond", "Array 'arr[16]' accessed at index 16, which is out of bounds.", CWE_BUFFER_OVERRUN, false);
return;
}
const Token *condition = nullptr;
const ValueFlow::Value *index = nullptr;
for (const ValueFlow::Value *indexValue: indexes) {
if (!indexValue)
continue;
if (!indexValue->errorSeverity() && !mSettings->isEnabled(Settings::WARNING))
return;
if (indexValue->condition)
condition = indexValue->condition;
if (!index || !indexValue->errorPath.empty())
index = indexValue;
}
reportError(getErrorPath(tok, index, "Array index out of bounds"),
index->errorSeverity() ? Severity::error : Severity::warning,
2019-03-11 21:39:39 +01:00
index->condition ? "arrayIndexOutOfBoundsCond" : "arrayIndexOutOfBounds",
arrayIndexMessage(tok, dimensions, indexes, condition),
2019-03-23 08:41:20 +01:00
CWE_BUFFER_OVERRUN,
index->isInconclusive());
2015-11-08 12:39:08 +01:00
}
void CheckBufferOverrun::negativeIndexError(const Token *tok, const std::vector<Dimension> &dimensions, const std::vector<const ValueFlow::Value *> &indexes)
2015-11-08 12:39:08 +01:00
{
if (!tok) {
2019-03-23 08:41:20 +01:00
reportError(tok, Severity::error, "negativeIndex", "Negative array index", CWE_BUFFER_UNDERRUN, false);
return;
2015-11-08 12:39:08 +01:00
}
const Token *condition = nullptr;
const ValueFlow::Value *negativeValue = nullptr;
for (const ValueFlow::Value *indexValue: indexes) {
if (!indexValue)
continue;
if (!indexValue->errorSeverity() && !mSettings->isEnabled(Settings::WARNING))
return;
if (indexValue->condition)
condition = indexValue->condition;
if (!negativeValue || !indexValue->errorPath.empty())
negativeValue = indexValue;
}
reportError(getErrorPath(tok, negativeValue, "Negative array index"),
negativeValue->errorSeverity() ? Severity::error : Severity::warning,
"negativeIndex",
arrayIndexMessage(tok, dimensions, indexes, condition),
2019-03-23 08:41:20 +01:00
CWE_BUFFER_UNDERRUN,
negativeValue->isInconclusive());
}
//---------------------------------------------------------------------------
void CheckBufferOverrun::pointerArithmetic()
{
if (!mSettings->isEnabled(Settings::PORTABILITY))
return;
for (const Token *tok = mTokenizer->tokens(); tok; tok = tok->next()) {
if (!Token::Match(tok, "+|-"))
continue;
if (!tok->valueType() || tok->valueType()->pointer == 0)
continue;
if (!tok->astOperand1() || !tok->astOperand2())
continue;
if (!tok->astOperand1()->valueType() || !tok->astOperand2()->valueType())
continue;
const Token *arrayToken, *indexToken;
if (tok->astOperand1()->valueType()->pointer > 0) {
arrayToken = tok->astOperand1();
indexToken = tok->astOperand2();
} else {
arrayToken = tok->astOperand2();
indexToken = tok->astOperand1();
}
if (!indexToken || !indexToken->valueType() || indexToken->valueType()->pointer > 0 || !indexToken->valueType()->isIntegral())
continue;
std::vector<Dimension> dimensions;
ErrorPath errorPath;
bool mightBeLarger = false;
if (!getDimensionsEtc(arrayToken, mSettings, &dimensions, &errorPath, &mightBeLarger))
continue;
if (tok->str() == "+") {
// Positive index
if (!mightBeLarger) { // TODO check arrays with dim 1 also
const std::vector<const Token *> indexTokens{indexToken};
const std::vector<const ValueFlow::Value *> &indexValues = getOverrunIndexValues(tok, arrayToken, dimensions, indexTokens);
if (!indexValues.empty())
pointerArithmeticError(tok, indexToken, indexValues.front());
}
if (const ValueFlow::Value *neg = indexToken->getValueLE(-1, mSettings))
pointerArithmeticError(tok, indexToken, neg);
} else if (tok->str() == "-") {
// TODO
}
}
}
void CheckBufferOverrun::pointerArithmeticError(const Token *tok, const Token *indexToken, const ValueFlow::Value *indexValue)
{
if (!tok) {
reportError(tok, Severity::portability, "pointerOutOfBounds", "Pointer arithmetic overflow.", CWE_POINTER_ARITHMETIC_OVERFLOW, false);
reportError(tok, Severity::portability, "pointerOutOfBoundsCond", "Pointer arithmetic overflow.", CWE_POINTER_ARITHMETIC_OVERFLOW, false);
return;
}
std::string errmsg;
if (indexValue->condition)
errmsg = "Undefined behaviour, when '" + indexToken->expressionString() + "' is " + MathLib::toString(indexValue->intvalue) + " the pointer arithmetic '" + tok->expressionString() + "' is out of bounds.";
else
errmsg = "Undefined behaviour, pointer arithmetic '" + tok->expressionString() + "' is out of bounds.";
reportError(getErrorPath(tok, indexValue, "Pointer arithmetic overflow"),
Severity::portability,
indexValue->condition ? "pointerOutOfBoundsCond" : "pointerOutOfBounds",
errmsg,
CWE_POINTER_ARITHMETIC_OVERFLOW,
indexValue->isInconclusive());
}
//---------------------------------------------------------------------------
ValueFlow::Value CheckBufferOverrun::getBufferSize(const Token *bufTok) const
2016-10-29 12:18:11 +02:00
{
if (!bufTok->valueType())
return ValueFlow::Value(-1);
const Variable *var = bufTok->variable();
if (!var || var->dimensions().empty()) {
const ValueFlow::Value *value = getBufferSizeValue(bufTok);
if (value)
return *value;
}
if (!var)
return ValueFlow::Value(-1);
MathLib::bigint dim = 1;
for (const Dimension &d : var->dimensions())
dim *= d.num;
ValueFlow::Value v;
v.setKnown();
v.valueType = ValueFlow::Value::ValueType::BUFFER_SIZE;
if (var->isPointerArray())
v.intvalue = dim * mSettings->sizeof_pointer;
2019-03-19 21:06:37 +01:00
else if (var->isPointer())
return ValueFlow::Value(-1);
else {
const MathLib::bigint typeSize = bufTok->valueType()->typeSize(*mSettings);
v.intvalue = dim * typeSize;
}
return v;
2016-10-29 12:18:11 +02:00
}
//---------------------------------------------------------------------------
2016-10-29 12:18:11 +02:00
static bool checkBufferSize(const Token *ftok, const Library::ArgumentChecks::MinSize &minsize, const std::vector<const Token *> &args, const MathLib::bigint bufferSize, const Settings *settings)
{
const Token * const arg = (minsize.arg > 0 && minsize.arg - 1 < args.size()) ? args[minsize.arg - 1] : nullptr;
const Token * const arg2 = (minsize.arg2 > 0 && minsize.arg2 - 1 < args.size()) ? args[minsize.arg2 - 1] : nullptr;
switch (minsize.type) {
case Library::ArgumentChecks::MinSize::Type::STRLEN:
if (settings->library.isargformatstr(ftok, minsize.arg)) {
return getMinFormatStringOutputLength(args, minsize.arg) < bufferSize;
} else if (arg) {
const Token *strtoken = arg->getValueTokenMaxStrLength();
if (strtoken)
return Token::getStrLength(strtoken) < bufferSize;
}
break;
case Library::ArgumentChecks::MinSize::Type::ARGVALUE:
if (arg && arg->hasKnownIntValue())
return arg->getKnownIntValue() <= bufferSize;
break;
case Library::ArgumentChecks::MinSize::Type::SIZEOF:
// TODO
break;
case Library::ArgumentChecks::MinSize::Type::MUL:
if (arg && arg2 && arg->hasKnownIntValue() && arg2->hasKnownIntValue())
return (arg->getKnownIntValue() * arg2->getKnownIntValue()) <= bufferSize;
break;
case Library::ArgumentChecks::MinSize::Type::VALUE:
return minsize.value <= bufferSize;
case Library::ArgumentChecks::MinSize::Type::NONE:
break;
};
return true;
}
void CheckBufferOverrun::bufferOverflow()
{
const SymbolDatabase *symbolDatabase = mTokenizer->getSymbolDatabase();
for (const Scope * scope : symbolDatabase->functionScopes) {
for (const Token *tok = scope->bodyStart; tok != scope->bodyEnd; tok = tok->next()) {
if (!Token::Match(tok, "%name% (") || Token::simpleMatch(tok, ") {"))
continue;
if (!mSettings->library.hasminsize(tok))
continue;
const std::vector<const Token *> args = getArguments(tok);
for (unsigned int argnr = 0; argnr < args.size(); ++argnr) {
if (!args[argnr]->valueType() || args[argnr]->valueType()->pointer == 0)
continue;
const std::vector<Library::ArgumentChecks::MinSize> *minsizes = mSettings->library.argminsizes(tok, argnr + 1);
if (!minsizes || minsizes->empty())
continue;
// Get buffer size..
const Token *argtok = args[argnr];
while (argtok && argtok->isCast())
argtok = argtok->astOperand2() ? argtok->astOperand2() : argtok->astOperand1();
while (Token::Match(argtok, ".|::"))
argtok = argtok->astOperand2();
if (!argtok || !argtok->variable())
continue;
// TODO: strcpy(buf+10, "hello");
const ValueFlow::Value bufferSize = getBufferSize(argtok);
if (bufferSize.intvalue <= 1)
continue;
bool error = true;
for (const Library::ArgumentChecks::MinSize &minsize : *minsizes) {
if (checkBufferSize(tok, minsize, args, bufferSize.intvalue, mSettings)) {
error = false;
break;
}
}
if (error)
bufferOverflowError(args[argnr], &bufferSize);
}
}
}
}
void CheckBufferOverrun::bufferOverflowError(const Token *tok, const ValueFlow::Value *value)
{
2019-03-23 08:41:20 +01:00
reportError(getErrorPath(tok, value, "Buffer overrun"), Severity::error, "bufferAccessOutOfBounds", "Buffer is accessed out of bounds: " + (tok ? tok->expressionString() : "buf"), CWE_BUFFER_OVERRUN, false);
}
//---------------------------------------------------------------------------
void CheckBufferOverrun::arrayIndexThenCheck()
{
if (!mSettings->isEnabled(Settings::PORTABILITY))
return;
const SymbolDatabase *symbolDatabase = mTokenizer->getSymbolDatabase();
for (const Scope * const scope : symbolDatabase->functionScopes) {
for (const Token *tok = scope->bodyStart; tok && tok != scope->bodyEnd; tok = tok->next()) {
if (Token::simpleMatch(tok, "sizeof (")) {
tok = tok->linkAt(1);
continue;
}
if (Token::Match(tok, "%name% [ %var% ]")) {
tok = tok->next();
const unsigned int indexID = tok->next()->varId();
const std::string& indexName(tok->strAt(1));
// Iterate AST upwards
const Token* tok2 = tok;
const Token* tok3 = tok2;
while (tok2->astParent() && tok2->tokType() != Token::eLogicalOp) {
tok3 = tok2;
tok2 = tok2->astParent();
}
// Ensure that we ended at a logical operator and that we came from its left side
if (tok2->tokType() != Token::eLogicalOp || tok2->astOperand1() != tok3)
continue;
// check if array index is ok
// statement can be closed in parentheses, so "(| " is using
if (Token::Match(tok2, "&& (| %varid% <|<=", indexID))
arrayIndexThenCheckError(tok, indexName);
else if (Token::Match(tok2, "&& (| %any% >|>= %varid% !!+", indexID))
arrayIndexThenCheckError(tok, indexName);
}
}
}
}
void CheckBufferOverrun::arrayIndexThenCheckError(const Token *tok, const std::string &indexName)
{
reportError(tok, Severity::style, "arrayIndexThenCheck",
"$symbol:" + indexName + "\n"
"Array index '$symbol' is used before limits check.\n"
"Defensive programming: The variable '$symbol' is used as an array index before it "
"is checked that is within limits. This can mean that the array might be accessed out of bounds. "
"Reorder conditions such as '(a[i] && i < 10)' to '(i < 10 && a[i])'. That way the array will "
"not be accessed if the index is out of limits.", CWE_ARRAY_INDEX_THEN_CHECK, false);
}
//---------------------------------------------------------------------------
void CheckBufferOverrun::stringNotZeroTerminated()
{
// this is currently 'inconclusive'. See TestBufferOverrun::terminateStrncpy3
if (!mSettings->isEnabled(Settings::WARNING) || !mSettings->inconclusive)
return;
const SymbolDatabase *symbolDatabase = mTokenizer->getSymbolDatabase();
for (const Scope * const scope : symbolDatabase->functionScopes) {
for (const Token *tok = scope->bodyStart; tok && tok != scope->bodyEnd; tok = tok->next()) {
if (!Token::simpleMatch(tok, "strncpy ("))
continue;
const std::vector<const Token *> args = getArguments(tok);
if (args.size() != 3)
continue;
const Token *sizeToken = args[2];
if (!sizeToken->hasKnownIntValue())
continue;
const ValueFlow::Value &bufferSize = getBufferSize(args[0]);
if (bufferSize.intvalue < 0 || sizeToken->getKnownIntValue() < bufferSize.intvalue)
continue;
const Token *srcValue = args[1]->getValueTokenMaxStrLength();
if (srcValue && Token::getStrLength(srcValue) < sizeToken->getKnownIntValue())
continue;
// Is the buffer zero terminated after the call?
bool isZeroTerminated = false;
for (const Token *tok2 = tok->next()->link(); tok2 != scope->bodyEnd; tok2 = tok2->next()) {
if (!Token::Match(tok2, "] ="))
continue;
const Token *rhs = tok2->next()->astOperand2();
if (!rhs || !rhs->hasKnownIntValue() || rhs->getKnownIntValue() != 0)
continue;
if (isSameExpression(mTokenizer->isCPP(), false, args[0], tok2->link()->astOperand1(), mSettings->library, false, false))
isZeroTerminated = true;
}
if (isZeroTerminated)
continue;
// TODO: Locate unsafe string usage..
terminateStrncpyError(tok, args[0]->expressionString());
}
}
}
void CheckBufferOverrun::terminateStrncpyError(const Token *tok, const std::string &varname)
{
const std::string shortMessage = "The buffer '$symbol' may not be null-terminated after the call to strncpy().";
reportError(tok, Severity::warning, "terminateStrncpy",
"$symbol:" + varname + '\n' +
shortMessage + '\n' +
shortMessage + ' ' +
"If the source string's size fits or exceeds the given size, strncpy() does not add a "
"zero at the end of the buffer. This causes bugs later in the code if the code "
"assumes buffer is null-terminated.", CWE170, true);
}
void CheckBufferOverrun::bufferNotZeroTerminatedError(const Token *tok, const std::string &varname, const std::string &function)
{
const std::string errmsg = "$symbol:" + varname + '\n' +
"$symbol:" + function + '\n' +
"The buffer '" + varname + "' is not null-terminated after the call to " + function + "().\n"
"The buffer '" + varname + "' is not null-terminated after the call to " + function + "(). "
"This will cause bugs later in the code if the code assumes the buffer is null-terminated.";
reportError(tok, Severity::warning, "bufferNotZeroTerminated", errmsg, CWE170, true);
}
2019-03-23 08:36:10 +01:00
//---------------------------------------------------------------------------
// CTU..
//---------------------------------------------------------------------------
std::string CheckBufferOverrun::MyFileInfo::toString() const
2019-03-23 08:36:10 +01:00
{
return CTU::toString(unsafeUsage);
}
bool CheckBufferOverrun::isCtuUnsafeBufferUsage(const Check *check, const Token *argtok, MathLib::bigint *offset)
{
const CheckBufferOverrun *c = dynamic_cast<const CheckBufferOverrun *>(check);
if (!c)
return false;
if (!argtok->valueType())
return false;
if (!Token::Match(argtok, "%name% [") || argtok->astParent() != argtok->next() || Token::simpleMatch(argtok->linkAt(1), "] ["))
return false;
if (!argtok->next()->astOperand2())
return false;
if (!argtok->next()->astOperand2()->hasKnownIntValue())
return false;
if (!offset)
return false;
*offset = argtok->next()->astOperand2()->getKnownIntValue() * argtok->valueType()->typeSize(*c->mSettings);
return true;
}
/** @brief Parse current TU and extract file info */
Check::FileInfo *CheckBufferOverrun::getFileInfo(const Tokenizer *tokenizer, const Settings *settings) const
2019-03-23 08:36:10 +01:00
{
2019-03-23 15:26:13 +01:00
CheckBufferOverrun checkBufferOverrun(tokenizer, settings, nullptr);
const std::list<CTU::FileInfo::UnsafeUsage> &unsafeUsage = CTU::getUnsafeUsage(tokenizer, settings, &checkBufferOverrun, isCtuUnsafeBufferUsage);
2019-03-23 08:36:10 +01:00
if (unsafeUsage.empty())
return nullptr;
MyFileInfo *fileInfo = new MyFileInfo;
fileInfo->unsafeUsage = unsafeUsage;
return fileInfo;
}
Check::FileInfo * CheckBufferOverrun::loadFileInfoFromXml(const tinyxml2::XMLElement *xmlElement) const
2019-03-23 08:36:10 +01:00
{
const std::list<CTU::FileInfo::UnsafeUsage> &unsafeUsage = CTU::loadUnsafeUsageListFromXml(xmlElement);
if (unsafeUsage.empty())
return nullptr;
MyFileInfo *fileInfo = new MyFileInfo;
fileInfo->unsafeUsage = unsafeUsage;
return fileInfo;
}
/** @brief Analyse all file infos for all TU */
2019-03-23 19:03:57 +01:00
bool CheckBufferOverrun::analyseWholeProgram(const CTU::FileInfo *ctu, const std::list<Check::FileInfo*> &fileInfo, const Settings& settings, ErrorLogger &errorLogger)
{
2019-03-23 08:36:10 +01:00
if (!ctu)
return false;
bool foundErrors = false;
(void)settings; // This argument is unused
const std::map<std::string, std::list<const CTU::FileInfo::CallBase *>> callsMap = ctu->getCallsMap();
2019-03-23 19:03:57 +01:00
for (Check::FileInfo *fi1 : fileInfo) {
2019-03-23 08:36:10 +01:00
const MyFileInfo *fi = dynamic_cast<MyFileInfo*>(fi1);
if (!fi)
continue;
for (const CTU::FileInfo::UnsafeUsage &unsafeUsage : fi->unsafeUsage) {
const CTU::FileInfo::FunctionCall *functionCall = nullptr;
const std::list<ErrorLogger::ErrorMessage::FileLocation> &locationList =
2019-03-23 19:03:57 +01:00
ctu->getErrorPath(CTU::FileInfo::InvalidValueType::bufferOverflow,
unsafeUsage,
callsMap,
"Using argument ARG",
&functionCall,
false);
2019-03-23 08:36:10 +01:00
if (locationList.empty())
continue;
if (unsafeUsage.value > 0) {
const ErrorLogger::ErrorMessage errmsg(locationList,
emptyString,
Severity::error,
"Buffer access out of bounds; '" + unsafeUsage.myArgumentName + "' buffer size is " + MathLib::toString(functionCall->callArgValue) + " and it is accessed at offset " + MathLib::toString(unsafeUsage.value) + ".",
"ctubufferoverrun",
2019-03-23 08:41:20 +01:00
CWE_BUFFER_OVERRUN, false);
2019-03-23 08:36:10 +01:00
errorLogger.reportErr(errmsg);
} else {
const ErrorLogger::ErrorMessage errmsg(locationList,
emptyString,
Severity::error,
"Buffer access out of bounds; buffer '" + unsafeUsage.myArgumentName + "' is accessed at offset " + MathLib::toString(unsafeUsage.value) + ".",
"ctubufferunderrun",
2019-03-23 08:41:20 +01:00
CWE_BUFFER_UNDERRUN, false);
2019-03-23 08:36:10 +01:00
errorLogger.reportErr(errmsg);
}
foundErrors = true;
}
}
return foundErrors;
}