Guard Byte - Principle

Principle

When a program is compiled for debugging, all memory allocations are prefixed and postfixed by guard bytes. Special memory allocation routines may then perform additional tasks to determine unwanted read and write attempts outside the allocated memory. These extra bytes help to detect that the program is writing into (or even reading from) inappropriate memory areas, potentially causing buffer overflows. In case of accessing these bytes by the program's algorithm, the programmer is warned with information assisting him/her to locate the problem.

Checking for the inappropriate access to the guard bytes may be done in two ways:

• by setting a memory breakpoint on a condition of write and/or read to those bytes, or
• by pre-initializing the guard bytes with specific values and checking the values upon deallocation.

The first way is possible only with a debugger that handles such breakpoints, but significantly increases the chance of locating the problem. The second way does not require any debuggers or special environments and can be done even on other computers, but the programmer is alerted about the overflow only upon the deallocation, which is sometimes quite late.

Because guard bytes require additional code to be executed and additional memory to be allocated, they are used only when the program is compiled for debugging. When compiled as a release, guard bytes are not used at all, neither the routines working with them.