On Fri, 12 Aug 2011 10:38:35 -0700, Paul Berry <stereotype...@gmail.com> wrote:
> This patch fixes a bug when lowering an integer division:
>
> x/y
>
> to a multiplication by a reciprocal:
>
> int(float(x)*reciprocal(float(y)))
>
> If x was a a plain int and y was an ivecN, the lowering pass
> incorrectly assigned the type of the product to be float, when in fact
> it should be vecN. This caused mesa to abort with an IR validation
> error.
>
> Fixes piglit tests {fs,vs}-op-div-int-ivec{2,3,4}.
> ---
> src/glsl/lower_instructions.cpp | 4 ++++
> 1 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/src/glsl/lower_instructions.cpp b/src/glsl/lower_instructions.cpp
> index 806f863..23aa19b 100644
> --- a/src/glsl/lower_instructions.cpp
> +++ b/src/glsl/lower_instructions.cpp
> @@ -166,6 +166,10 @@ lower_instructions_visitor::div_to_mul_rcp(ir_expression
> *ir)
> else
> op0 = new(ir) ir_expression(ir_unop_u2f, vec_type, ir->operands[0],
> NULL);
>
> + vec_type = glsl_type::get_instance(GLSL_TYPE_FLOAT,
> + ir->type->vector_elements,
> + ir->type->matrix_columns);
> +
> op0 = new(ir) ir_expression(ir_binop_mul, vec_type, op0, op1);
>
> if (ir->operands[1]->type->base_type == GLSL_TYPE_INT) {You could just drop the type argument to ir_expression and let it figure out the right answer for you. Oh, and we'll want a test to make sure we're not doing this approximation when we shouldn't be in GLSL 1.30 -- we have actual integer division hardware we should be using.
pgpPp2FhHw0ht.pgp
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
