On Mon, 11 Apr 2022 15:23:17 +0200
Nico Boehr <nrb@xxxxxxxxxxxxx> wrote:
> On Wed, 2022-04-06 at 15:31 +0200, Claudio Imbrenda wrote:
> > > diff --git a/s390x/adtl-status.c b/s390x/adtl-status.c
> > > new file mode 100644
> > > index 000000000000..c3ecdbc35a9d
> > >
> [...]
> > > +static void restart_write_vector(void)
> > > +{
> > > + uint8_t *vec_reg;
> > > + /* vlm handles at most 16 registers at a time */
> > > + uint8_t *vec_reg_16_31 = &expected_vec_contents[16][0];
> > > + uint64_t cr0, cr0_mask = ~(1ULL << CTL0_VECTOR);
> >
> > cr0_mask can be const, and you can use ~BIT_ULL(CTL0_VECTOR)
>
> I don't think so, since ng in the inline ASM will store its result
> there.
oops, right
>
> BIT_ULL is much better, thanks. I tend to forget about that.
>
> >
> > > + int one = 1;
> >
> > one can also be const, although I wonder if this can just become an
> > constant in the asm statement
>
> Yes, right, thanks. Should work with mvhi.
>
