Function:

FDIV divides destination by source and returns the result in destination.
FDIVP does the same but pops the FPU stack afterwards.

The bug occurs when the instruction operates on an FPU register which is
tagged as empty, but holds a nonzero value and the next FPU instruction
occurs within 35 FPU clock counts. In that case, the current instruction
will use the invalid number in the empty location, producing an invalid
result and causing the following instruction to generate an invalid
result as well. There is no workaround.




FDIVR / FDIVRP  Floating point division reversed / divide & POP
──────────────────────────────────────────────────────────────────────────────