Re: [Qemu-devel] [PATCH 4/5] softfloat: Implement fused multiply-add

[Richard Henderson]

On 09/28/2011 10:27 AM, Peter Maydell wrote:
>  
> /*----------------------------------------------------------------------------
> +| Select which NaN to propagate for a three-input operation.
> +| For the moment we assume that no CPU needs the 'larger significand'
> +| information.
> +| Return values : 0 : a; 1 : b; 2 : c; 3 : default-NaN
> +*----------------------------------------------------------------------------*/
> +#if defined(TARGET_ARM)
> +static int pickNaNMulAdd(flag aIsQNaN, flag aIsSNaN, flag bIsQNaN, flag 
> bIsSNaN,
> +                         flag cIsQNaN, flag cIsSNaN, flag infzero 
> STATUS_PARAM)
...
> +#elif defined(TARGET_PPC)
> +static int pickNaNMulAdd(flag aIsQNaN, flag aIsSNaN, flag bIsQNaN, flag 
> bIsSNaN,
> +                         flag cIsQNaN, flag cIsSNaN, flag infzero 
> STATUS_PARAM)
> +{

The function declaration should be outside the #if, so that the interface is
forcibly consistent across the platforms.

> +    cSig64 = (uint64_t)cSig << 39;

This might be more readable as << (62 - 23), since you've just mentioned the
explicit bit at bit 62 above.

> +    if (pSign == cSign) {
> +        /* Addition */
...
> +        shift64RightJamming(zSig64, 32, &zSig64);
> +        return roundAndPackFloat32(zSign, zExp, zSig64 STATUS_VAR);
> +    } else {
> +        /* Subtraction */
...
> +        shift64RightJamming(zSig64, 32, &zSig64);
> +        return roundAndPackFloat32(zSign, zExp, zSig64 STATUS_VAR);
> +    }
> +}

Push those two calls down after the IF?

Similar comments wrt float64_muladd.  But I don't see any actual logic errors
wrt the handling of any of the edge cases.


r~