On 2015-10-13 20:04:36, Kenneth Graunke wrote: > On Tuesday, October 13, 2015 01:44:55 PM Jordan Justen wrote: > > The commit shown below caused compute shaders to hit the unreachable > > in the default of the switch block. Restore compute shaders to use the > > fragment shader path. > > > > Also, simplify the fragment/compute path to only support scalar mode. > > > > commit 2953c3d76178d7589947e6ea1dbd902b7b02b3d4 > > Author: Kenneth Graunke <kenn...@whitecape.org> > > Date: Fri Aug 14 15:15:11 2015 -0700 > > > > i965/vs: Map scalar VS input locations properly; avoid tons of MOVs. > > > > Signed-off-by: Jordan Justen <jordan.l.jus...@intel.com> > > Cc: Kenneth Graunke <kenn...@whitecape.org> > > --- > > src/mesa/drivers/dri/i965/brw_nir.c | 4 +++- > > 1 file changed, 3 insertions(+), 1 deletion(-) > > > > diff --git a/src/mesa/drivers/dri/i965/brw_nir.c > > b/src/mesa/drivers/dri/i965/brw_nir.c > > index 4f35d81..357ee4f 100644 > > --- a/src/mesa/drivers/dri/i965/brw_nir.c > > +++ b/src/mesa/drivers/dri/i965/brw_nir.c > > @@ -96,8 +96,10 @@ brw_nir_lower_inputs(nir_shader *nir, bool is_scalar) > > } > > break; > > case MESA_SHADER_FRAGMENT: > > + case MESA_SHADER_COMPUTE: > > + assert(is_scalar); > > nir_assign_var_locations(&nir->inputs, &nir->num_inputs, > > - is_scalar ? type_size_scalar : > > type_size_vec4); > > + type_size_scalar); > > break; > > default: > > unreachable("unsupported shader stage"); > > > > I didn't think compute shaders had inputs...so it might make more sense > just to do: > > case MESA_SHADER_COMPUTE: > /* Compute shaders have no inputs. */ > break;
Huh. I looked at nir_assign_var_locations and it seemed to ignore UBO's and SSBO's, but not normal uniforms. So, I thought it might be needed for uniforms. But, I didn't see a regression after skipping calling it. Can you confirm it is not needed for uniforms? Strange, it looks like fs_visitor::nir_setup_uniforms will use var->data.driver_location which nir_assign_var_locations sets. So, I'm wondering why not calling it is not causing trouble. > Either way, sorry for breaking this, and... > Reviewed-by: Kenneth Graunke <kenn...@whitecape.org> No problem. Hitting unreachable is easy to debug. :) -Jordan _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev