On Thu, Jun 29, 2017 at 5:10 PM, Wilco Dijkstra <wilco.dijks...@arm.com> wrote:
> Richard Biener wrote:
>
>> int f (int m, int c)
>> {
>> return (m & 1) * c;
>> }
>
> This case (integer[0,1] rather than boolean input) should be transformed into
> c & -(m & 1).
The proposed patch handled both the same. This means the pattern
shouldn't use range-info
but instead match a more complex
(mult (convert (cmp @0 @1)) @2)
? Note that from a gimple perspective c & -(m & 1) is more complex
than (m & 1) * c
and thus non-canonical.
> Wilco
