[Libre-soc-isa] [Bug 817] Big Integer Math (sv.adde, sv.subfe, sv.madded, 128 by 64-bit -> 64-bit div/rem, maybe more...)

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=817

--- Comment #49 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
https://www.felixcloutier.com/x86/idiv

did some investigation, divmod2du is basically identical to idivq

    ELSE IF OperandSize = 64 (* Doublequadword/quadword operation *)
            temp := RDX:RAX / SRC; (* Signed division *)
            IF (temp > 7FFFFFFFFFFFFFFFH) or (temp < 8000000000000000H)
            (* If a positive result is greater than 7FFFFFFFFFFFFFFFH
            or a negative result is less than 8000000000000000H *)
                THEN
                    #DE; (* Divide error *)
                ELSE
                    RAX := temp;
                    RDX := RDE:RAX SignedModulus SRC;
            FI;
        FI;

which is 3-in, 2-out.

* RDX:RAX is the 128 bit divisor (2 64bit regs)
* SRC is the dividend (1 64bit reg)
* RAX:RDX is the twin div/mod result

    dividend[0:(XLEN*2)-1] <- (RC) || (RA)
    divisor[0:(XLEN*2)-1] <- [0]*XLEN || (RB)
    result <- dividend / divisor
    modulo <- dividend % divisor
    RT <- result[XLEN:(XLEN*2)-1]
    RS <- modulo[XLEN:(XLEN*2)-1]

-- 
You are receiving this mail because:
You are on the CC list for the bug.