[Mesa-dev] [PATCH 2/2] ac/nir: fix translation of nir_op_frsq for doubles

[Samuel Pitoiset]

On 01/11/2018 07:06 AM, Timothy Arceri wrote:
> Without this we en up with the llvm error message:

typo: end up

Series is:

Reviewed-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>

> 
> "Both operands to a binary operator are not of the same type!"
> ---
>   src/amd/common/ac_nir_to_llvm.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
> index 8d759b30fb..f1472677e1 100644
> --- a/src/amd/common/ac_nir_to_llvm.c
> +++ b/src/amd/common/ac_nir_to_llvm.c
> @@ -1842,7 +1842,8 @@ static void visit_alu(struct ac_nir_context *ctx, const nir_alu_instr *instr)
>   	case nir_op_frsq:
>   		result = emit_intrin_1f_param(&ctx->ac, "llvm.sqrt",
>   		                              ac_to_float_type(&ctx->ac, def_type), src[0]);
> -		result = ac_build_fdiv(&ctx->ac, ctx->ac.f32_1, result);
> +		result = ac_build_fdiv(&ctx->ac, instr->dest.dest.ssa.bit_size == 32 ? ctx->ac.f32_1 : ctx->ac.f64_1,
> +				       result);
>   		break;
>   	case nir_op_fpow:
>   		result = emit_intrin_2f_param(&ctx->ac, "llvm.pow",
>