Strings and Integers

OWASP Code Review Guide Table of Contents

Introduction:

Strings are not a defined Type in C or C++, but simply a contiguous array of characters terminated by a null (\0) character. The length of the string is the amount of characters which precede the null character. C++ does contain template classes which address this feature of the programming language: std::basic_string and std::string These classes address some security issues but not all.

|W|E|L|C|O|M|E|\0|

Common String Errors

Common string errors can be related to mistakes in implementation, which may cause drastic security and availability issues. C/C++ do not have the comfort other programming languages provide, such as Java and C# .NET, relating to buffer overflows and such due to a String Type not being defined.

Common issues include:

1. Input validation errors
2. Unbounded errors
3. Truncation issues
4. Out-of-bounds writes
5. String Termination errors
6. Off-by-one errors

Some of the issues mentioned above have been covered in the Reviewing Code for Buffer Overruns and Overflows section in this guide.

Unbounded Errors

String Copies

Unbounded errors occur when data is copied from a unbounded source to a fixed length character array.

void main(void) {
 char Name[10];
 puts("Enter your name:");
 gets(Name); <-- Here the name input by the user can be of arbitrary length over running the Name array.
...
 }

String Termination Errors

Failure to properly terminate strings with a null can result in system failure.

int main(int argc, char* argv[]) {
 char a[16];
 char b[16];
 char c[32];
 strncpy(a, "0123456789abcdef", sizeof(a));
 strncpy(b, "0123456789abcdef", sizeof(b));
 strncpy(c, a, sizeof(c));
}

Verify that the following are used:

strncpy() instead of strcpy()
snprintf() instead of sprintf()
fgets() instead of gets()

Off by One Error

Looping through arrays should be looped in a n-1 manner, as we must remember arrays and vectors start as 0. This is not specific to C/C++, but Java and C# also.)

Off-by-one errors are common to looping functionality, wherein a looping functionality is performed on an object in order to manipulate the contents of an object such as copy or add information. The off-by-one error is a result of an error on the loop counting functionality.

for (i = 0; i < 5; i++) {
   /* Do Stuff */
}

Here i starts with a value of 0, it then increments to 1, then 2, 3 & 4. When i reaches 5 then the condition i<5 is false and the loop terminates.

If the condition was set such that i<=5 (less than or equal to 5), the loop won’t terminate until i reaches 6, which may not be what is intended.

Also, counting from 1 instead of 0 can cause similar issues, as there would be one less iteration. Both of these issues relate to an off-by-one error where the loop either under or over counts.

Issues with Integers

Integer Overflows

When an integer is increased beyond its maximum range or decreased below its minimum value, overflows occur. Overflows can be signed or unsigned. Signed when the overflow carries over to the sign bit, unsigned when the value being intended to be represented is no longer represented correctly.

int x;
x = INT_MAX; // 2,147,483,647
x++;
Here x would have the value of -2,147,483,648 after the increment

It is important when reviewing the code that some measure should be implemented such that the overflow does not occur. This is not the same as relying on the value "never going to reach this value (2,147,483,647)". This may be done by some supporting logic or a post increment check.

unsigned int y;
y = UINT_MAX; // 4,294,967,295;
y++;
Here y would have a value of 0 after the increment

Also, here we can see the result of an unsigned int being incremented, which loops the integer back to the value 0. As before, this should also be examined to see if there are any compensating controls to prevent this from happening.

Integer Conversion

When converting from a signed to an unsigned integer, care must also be taken to prevent a representation error.

int x = -3;
unsigned short y;
y = x;

Here y would have the value of 65533 due to the loopback effect of the conversion from signed to unsigned.