[Mesa-dev] [PATCH 06/13] nir: don't count removal of lcssa_phi as progress
Timothy Arceri
timothy.arceri at collabora.com
Thu Dec 22 21:03:21 UTC 2016
On Thu, 2016-12-22 at 11:58 -0800, Jason Ekstrand wrote:
> I'm not actually sure that we need this anymore. We never run
> convert_to_lcssa unless we are about to make actual CF modifications
> and, therefore, report progress. I've got a patch running through
> Jenkins now to find out. I'm not strongly opposed to keeping it but
> I'd like to know if it's really needed.
I'm pretty sure it can go. I should have realised that earlier thanks
for spotting.
>
> On Wed, Dec 21, 2016 at 6:26 PM, Timothy Arceri <timothy.arceri at colla
> bora.com> wrote:
> > V2:
> > - make the is_lcssa_phi bool const, suggested by Topi.
> > ---
> > src/compiler/nir/nir_opt_remove_phis.c | 4 +++-
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/src/compiler/nir/nir_opt_remove_phis.c
> > b/src/compiler/nir/nir_opt_remove_phis.c
> > index acaa6e1..99d5d35 100644
> > --- a/src/compiler/nir/nir_opt_remove_phis.c
> > +++ b/src/compiler/nir/nir_opt_remove_phis.c
> > @@ -73,6 +73,7 @@ remove_phis_block(nir_block *block, nir_builder
> > *b)
> > break;
> >
> > nir_phi_instr *phi = nir_instr_as_phi(instr);
> > + const bool is_lcssa_phi = phi->is_lcssa_phi;
> >
> > nir_ssa_def *def = NULL;
> > nir_alu_instr *mov = NULL;
> > @@ -133,7 +134,8 @@ remove_phis_block(nir_block *block, nir_builder
> > *b)
> > nir_ssa_def_rewrite_uses(&phi->dest.ssa,
> > nir_src_for_ssa(def));
> > nir_instr_remove(instr);
> >
> > - progress = true;
> > + if (!is_lcssa_phi)
> > + progress = true;
> > }
> >
> > return progress;
> > --
> > 2.9.3
> >
> > _______________________________________________
> > mesa-dev mailing list
> > mesa-dev at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
> >
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
More information about the mesa-dev
mailing list