Intel 64 and IA-32 Architectures Software Developers Manual Volume 1, Basic Architecture

4-32 Vol. 1
DATA TYPES
the destination. Alternately, a denormal-operand or inexact-result exception can
accompany a numeric underflow or overflow exception with both exceptions being
handled.
The precedence for floating-point exceptions is as follows:
1. Invalid-operation exception, subdivided as follows:
a. stack underflow (occurs with x87 FPU only)
b. stack overflow (occurs with x87 FPU only)
c. operand of unsupported format (occurs with x87 FPU only when using the
double extended-precision floating-point format)
d. SNaN operand
2. QNaN operand. Though this is not an exception, the handling of a QNaN operand
has precedence over lower-priority exceptions. For example, a QNaN divided by
zero results in a QNaN, not a zero-divide exception.
3. Any other invalid-operation exception not mentioned above or a divide-by-zero
exception.
4. Denormal-operand exception. If masked, then instruction execution continues
and a lower-priority exception can occur as well.
5. Numeric overflow and underflow exceptions; possibly in conjunction with the
inexact-result exception.
6. Inexact-result exception.
Invalid operation, zero divide, and denormal operand exceptions are detected before
a floating-point operation begins. Overflow, underflow, and precision exceptions are
not detected until a true result has been computed. When an unmasked pre-opera-
tion exception is detected, the destination operand has not yet been updated, and
appears as if the offending instruction has not been executed. When an unmasked
post-operation exception is detected, the destination operand may be updated with
a result, depending on the nature of the exception (except for SSE/SSE2/SSE3
instructions, which do not update their destination operands in such cases).
4.9.3 Typical Actions of a Floating-Point Exception Handler
After the floating-point exception handler is invoked, the processor handles the
exception in the same manner that it handles non-floating-point exceptions. The
floating-point exception handler is normally part of the operating system or execu-
tive software, and it usually invokes a user-registered floating-point exception
handle.
A typical action of the exception handler is to store state information in memory.
Other typical exception handler actions include:
• Examining the stored state information to determine the nature of the error
• Taking actions to correct the condition that caused the error