Specifications

Numeric Data Processor (8087)
Error Handling and Interrupts
A numeric error occurs if an operation is attempted with invalid
operands or if the result of a computation cannot be accurately
represented. Six defined error conditions can occur during the
execution of a numeric instruction:
• Invalid operation
• Overflow
• Zero divisor
• Underflow
• Denormalized operand
• Inexact result
The NDP can be programmed to interrupt the V40 during any or all of
these errors. For an interrupt to be recognized, the NDP must be
programmed to enable interrupts and remove the mask of the selected
error condition. If the NDP interrupts are disabled when an error
occurs, the V40 is not interrupted and NDP operation continues. If the
NDP interrupts are enabled and an error occurs that is masked, the
NDP will flag the error in a status register and automatically execute a
default handling procedure. If the NDP interrupts are enabled and an
error occurs that is not masked, an interrupt will be generated to the
V40. This interrupt must be supported with an interrupt service
routine in the application program.
15-6