Re: [PATCH] rs6000: Teach rs6000_opaque_type_invalid_use_p about gcall [PR108348]

[Segher Boessenkool <segher@kernel.crashing.org>]

On Mon, Jan 16, 2023 at 09:05:38PM +0800, Kewen.Lin wrote:
> > The *_ok things should only be used for features that can be disabled
> > during configuration, or features that we *want* users to be able to
> > turn off (like FP, VMX, VSX, or HMT or QP float, that kind of thing).
> > That gives quite enough permutations to test already, we do not need to
> > create a whole bunch extra for no reason :-)
> 
> Thanks for the explanation!!  I meant to use powerpc_p9modulo_ok to
> exclude those cases where we can't use -mcpu=power9, as you explained we
> should not worry about it!?

But that selector says whether modulo insns are enabled.  This is not
correct to use here.  I know we have abused these things before, but it
needs to be untangled, not made worse :-)

> Since the test point requires altivec support
> (which is implied when specifying -mcpu=power9, I didn't explicitly specify
> it before), I think I could use powerpc_altivec_ok to replace
> powerpc_p9modulo_ok here, does it sound good to you?

VMX can be turned off even with -mcpu=power9.  So yes, it does need
powerpc_altivec_ok.  Does it need VSX even?


Segher