Re: [PATCH 3/3] arm64: Unconditionally enable EPAN support

[Marc Zyngier <maz@kernel.org>]

On Thu, 22 Jan 2026 10:15:56 +0000,
Will Deacon <will@kernel.org> wrote:
> 
> On Wed, Jan 07, 2026 at 06:07:01PM +0000, Marc Zyngier wrote:
> > While FEAT_PAN3 is pretty recent, having it permanently enabled costs
> > exactly nothing, and does help with exec-only mappings on these fancy
> > ARMv9.2 machines that are rumoured to exist.
> 
> I'm not sure it's _entirely_ accurate to say this one costs us "exactly
> nothing":
> 
> > diff --git a/arch/arm64/include/asm/cpucaps.h b/arch/arm64/include/asm/cpucaps.h
> > index 177c691914f87..13c0fa54ea19f 100644
> > --- a/arch/arm64/include/asm/cpucaps.h
> > +++ b/arch/arm64/include/asm/cpucaps.h
> > @@ -19,8 +19,6 @@ cpucap_is_possible(const unsigned int cap)
> >  			   "cap must be < ARM64_NCAPS");
> >  
> >  	switch (cap) {
> > -	case ARM64_HAS_EPAN:
> > -		return IS_ENABLED(CONFIG_ARM64_EPAN);
> >  	case ARM64_SVE:
> >  		return IS_ENABLED(CONFIG_ARM64_SVE);
> >  	case ARM64_SME:
> 
> as this means cpus_have_cap(EPAN) always ends up doing a test_bit(). It's
> not exactly expensive, but it feels a little premature when compared to
> PAN and LSE so I'll probably just take those changes for now.

Ah, fair enough. It can probably wait another few years then!

Cheers,

	M.

-- 
Without deviation from the norm, progress is not possible.