MC68060 Software Package
C-20
M68060 USER’S MANUAL
MOTOROLA
C.3.2.4 EXCEPTIONS DURING EMULATION. Unimplemented data type, unimplemented
effective address, and unimplemented floating-point instruction exception software emula-
tion by the M68060FPSP may determine that the instruction being emulated should take a
BSUN, SNAN, OPERR, OVFL, UNFL, DZ, or INEX exception. These exceptions may either
C.3.2.4.1 Trap-Disabled Operation. If a newly found exception is disabled by the user,
then the default result for that exception is returned as the result of emulation by the
M68060FPSP. The handler then returns the processor to normal processing.
Figure C-9. Disabled vs. Enabled Exception Actions
fsin.x
fp0
<non-fp>
<fp instruction>
TAKES FLOATING-POINT UNIMPLEMENTED
EXCEPTION IMMEDIATELY
(a)
(b)
(1) "fsin" software emulation determines that the sine operation should cause an underflow.
(2) If UNFL is:
DISABLED: the default result is calculated and returned at point "a"; the "exception present"
bit in the FPU is clear.
ENABLED: an fsave frame with the underflow exception set is restored into the FPU at point "a"
with the "exception present" bit set.
The actual underflow will occur as a pre-instruction exception at point "b".
(a)
fdiv.x
fp0, fp1
<non-fp>
<fp instruction>
TAKES FLOATING-POINT UNIMPLEMENTED
DATA TYPE EXCEPTION HERE
(1) fp0 contains a denormalized number and fp1 contains an SNAN; the exceptionis taken as a pre-
instruction exception at point "a".
(2) "fdiv" software emulation determines that the divide should cause a signalling non exception.
(2) If SNAN is:
DISABLED: The default result is calculated and returned at point "a"; the exception present"
bit in the FPU is clear.
ENABLED: an fsave frame with the SNAN exception set is restored into the FPU at point "a" with
the "exception present" bit set. The actual SNAN exception will then occur immediately as a pre-
instruction exception when the unimplemented floating point data type handler
(b)