On Fri, 3 Apr 2026 08:46:00 GMT, Jaromir Hamala <[email protected]> wrote:

> I've been toying in this area on x86, except the problem on x86 is opposite: 
> It does not clone the `MaskAll` node enough:
> 
> 1. there is a missing rule for long - only nodes with int constant are cloned
> 2. `XorVMask` is not treated as commutative, so even the existing rule does 
> not always match
> 
> I have 
> [WIP](https://github.com/openjdk/jdk/compare/master...jerrinot:jdk:jh_knot_kxor),
>  is it worth sending a PR?

I think so~

-------------

PR Comment: https://git.openjdk.org/jdk/pull/30013#issuecomment-4242723774

Reply via email to