On Sat, Dec 8, 2018 at 12:10 PM Karol Herbst <kher...@redhat.com> wrote:
>
> opnd() might delete the passed in instruction, but it's used through
> i->srcExists() later in visit
>
> v2: use continue instead return
Can you verify that this change has no effect on compilation results
(at least the high-level stats) for shader-db?
>
> Signed-off-by: Karol Herbst <kher...@redhat.com>
> ---
> .../nouveau/codegen/nv50_ir_peephole.cpp | 71 +++++++++++--------
> 1 file changed, 43 insertions(+), 28 deletions(-)
>
> diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
> b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
> index 202faf0746a..d7980a44ab1 100644
> --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
> +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
> @@ -370,7 +370,8 @@ private:
>
> void expr(Instruction *, ImmediateValue&, ImmediateValue&);
> void expr(Instruction *, ImmediateValue&, ImmediateValue&,
> ImmediateValue&);
> - void opnd(Instruction *, ImmediateValue&, int s);
> + /* true if i was deleted */
> + bool opnd(Instruction *i, ImmediateValue&, int s);
> void opnd3(Instruction *, ImmediateValue&);
>
> void unary(Instruction *, const ImmediateValue&);
> @@ -421,11 +422,13 @@ ConstantFolding::visit(BasicBlock *bb)
> i->src(0).getImmediate(src0) && i->src(1).getImmediate(src1))
> expr(i, src0, src1);
> else
> - if (i->srcExists(0) && i->src(0).getImmediate(src0))
> - opnd(i, src0, 0);
> - else
> + if (i->srcExists(0) && i->src(0).getImmediate(src0)) {
> + if (opnd(i, src0, 0))
> + continue;
> + } else
Please make this whole (outer) if/else sequence have { }. Even though
technically they're not necessary anywhere, by the time it gets to
having nested ifs ... it's a good idea.
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
