Re: [PATCH] target/arm: Handle denormals correctly for FMOPA (widening)

[Peter Maydell]

On Tue, 30 Jul 2024 at 23:36, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 7/31/24 01:58, Peter Maydell wrote:
> > -void HELPER(sme_fmopa_h)(void *vza, void *vzn, void *vzm, void *vpn,
> > -                         void *vpm, void *vst, uint32_t desc)
> > +void HELPER(sme_fmopa_h)(CPUARMState *env,
> > +                         void *vza, void *vzn, void *vzm, void *vpn,
> > +                         void *vpm, uint32_t desc)
>
> I think the env pointer should replace vst in the argument ordering.
> That keeps with the standard gvec signature.

Here I was going with our standard signature for everything
that isn't gvec and putting the env pointer first. I'd have
done that for the gvec helpers in the FEAT_EBF16 series too,
except the gvec machinery doesn't allow it AFAICT (and also
doesn't let you pass it as 'env' and insists on 'ptr' I think).

But I don't feel strongly, so I'll change this.

thanks
-- PMM