Re: [Qemu-devel] [PATCH v1 13/29] target-microblaze: Use TCGv for load/store addresses

[Edgar E. Iglesias]

On Thu, May 03, 2018 at 11:21:54AM -0700, Richard Henderson wrote:
> On 05/03/2018 02:19 AM, Edgar E. Iglesias wrote:
> > From: "Edgar E. Iglesias" <address@hidden>
> > 
> > Use TCGv for load/store addresses, allowing for future
> > computation of 64-bit load/store address.
> > 
> > No functional change.
> > 
> > Signed-off-by: Edgar E. Iglesias <address@hidden>
> 
> Reviewed-by: Richard Henderson <address@hidden>
> 
> 
> >          if (dc->imm == 0) {
> > -            tcg_gen_mov_i32(t, cpu_R[dc->ra]);
> > -            return;
> > +            tcg_gen_mov_i32(t32, cpu_R[dc->ra]);
> > +        } else {
> > +            tcg_gen_movi_i32(t32, (int32_t)((int16_t)dc->imm));
> > +            tcg_gen_add_i32(t32, cpu_R[dc->ra], t32);
> >          }
> 
> For future cleanup, this should be
> 
>   tcg_gen_addi_i32(t32, cpu_R[dc->ra], (int16_t)dc->imm);
> 
> The check for 0 will be done inside of tcg_gen_addi_i32.


Thanks, I've added a follow-up patch that cleans this up in v2 of this series.

Cheers,
Edgar