Re: [PATCH v2 1/3] target/ppc: Fixed call to deferred exception

[Lucas Mateus Martins Araujo e Castro]

On 19/11/2021 06:18, Richard Henderson wrote:

> On
>       11/18/21 2:25 PM, Lucas Mateus Castro (alqotel) wrote:
>       
> > +    if ((fpscr & FP_VXSOFT)
> >         && (fpscr_ve != 0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXSOFT;
> >         
> >         +    } else if ((fpscr & FP_OX) && (fpscr &
> >         FP_OE)) {
> >         
> >         +        error = POWERPC_EXCP_FP_OX;
> >         
> >         +    } else if ((fpscr & FP_UX) && (fpscr &
> >         FP_UE)) {
> >         
> >         +        error = POWERPC_EXCP_FP_UX;
> >         
> >         +    } else if ((fpscr & FP_XX) && (fpscr &
> >         FP_XE)) {
> >         
> >         +        error = POWERPC_EXCP_FP_XX;
> >         
> >         +    } else if ((fpscr & FP_ZX) && (fpscr &
> >         FP_ZE)) {
> >         
> >         +        error = POWERPC_EXCP_FP_ZX;
> >         
> >         +    } else if ((fpscr & FP_VXSNAN) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXSNAN;
> >         
> >         +    } else if ((fpscr & FP_VXISI) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXISI;
> >         
> >         +    } else if ((fpscr & FP_VXIDI) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXIDI;
> >         
> >         +    } else if ((fpscr & FP_VXZDZ) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXZDZ;
> >         
> >         +    } else if ((fpscr & FP_VXIMZ) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXIMZ;
> >         
> >         +    } else if ((fpscr & FP_VXVC) && (fpscr_ve !=
> >         0)) {
> >         
> >         +        error = POWERPC_EXCP_FP_VXVC;
> >         
> >         +    }
> >         
> >       
>       
>       Is there a defined order for these in the manual?  I couldn't find
>       it quickly if so.  If
>       
>       there is no defined order, I think you should test VE only once.
>       
>     
    I also couldn't find a defined order, so I chose to prioritize
    VXSOFT then go by ascending order of the bit number. In the v3 I'll
    move VXSOFT with the others invalid operation bits to check VE once
    then.
>       
>       Drop the use of fpscr_ve and use fpscr & FP_VE instead. (I
>       think these hidden uses of *env
>       
>       are evil and should be banished, but that's a bit of a job.)
>       
>       
>       You could say
>       
>       
>           } else {
>       
>               return;
>       
>           }
>       
>       
> > +
> >         
> >         +    if (error) {
> >         
> >       
>       
>       and then remove this test.
>       
>     
    Ok, I'll do these in v3.
>       
>       The rest of it looks good.
>       
>       
>       
>       r~
>       
>

--
Lucas Mateus M. Araujo e Castro
Instituto de Pesquisas ELDORADO
Departamento Computação Embarcada
Estagiario
Aviso Legal - Disclaimer