On systems based on the IA-32 architectures, when an application calls a function that returns a floating-point value, the returned floating-point value is supposed to be on the top of the floating-point stack. If the return value is not used, the compiler must pop the value off of the floating-point stack in order to keep the floating-point stack in the correct state.
On systems based on Intel(R) 64 architectures, floating-point values are usually returned in the xmm0 register. The floating-point stack is used only when the return value is a long double on Linux* and Mac OS* X systems.
If the application calls a function without defining or incorrectly defining the function's prototype, the compiler cannot determine if the function must return a floating-point value. Consequently, the return value is not popped off the floating-point stack if it is not used. This can cause the floating-point stack to overflow.
The overflow of the stack results in two undesirable situations:
A NaN value gets involved in the floating-point calculations
The program results become unpredictable; the point where the program starts making errors can be arbitrarily far away from the point of the actual error.
For systems based on the IA-32 and Intel® 64 architectures, the -fp-stack-check (Linux* and Mac OS* X) or /Qfp-stack-check (Windows*) option checks whether a program makes a correct call to a function that should return a floating-point value. If an incorrect call is detected, the option places a code that marks the incorrect call in the program. The -fp-stack-check (Linux* and Mac OS* X) or /Qfp-stack-check (Windows*) option marks the incorrect call and makes it easy to find the error.
Note
The -fp-stack-check (Linux* and Mac OS* X) and the /Qfp-stack-check (Windows*) option causes significant code generation after every function/subroutine call to ensure that the floating-point stack is maintained in the correct state. Therefore, using this option slows down the program being compiled. Use the option only as a debugging aid to find floating point stack underflow/overflow problems, which can be otherwise hard to find.