[PATCH v4 3/5] arm64: Add support for FEAT_HAFT

[Catalin Marinas]

On Tue, Nov 05, 2024 at 06:38:51PM +0800, Yicong Yang wrote:
> On 2024/11/5 10:47, Yicong Yang wrote:
> > On 2024/11/5 1:28, Catalin Marinas wrote:
> >> On Sat, Nov 02, 2024 at 06:42:33PM +0800, Yicong Yang wrote:
> >>> diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h
> >>> index 3d261cc123c1..ed8c784ca082 100644
> >>> --- a/arch/arm64/include/asm/cpufeature.h
> >>> +++ b/arch/arm64/include/asm/cpufeature.h
> >>> @@ -838,6 +838,12 @@ static inline bool system_supports_poe(void)
> >>>  		alternative_has_cap_unlikely(ARM64_HAS_S1POE);
> >>>  }
> >>>  
> >>> +static inline bool system_supports_haft(void)
> >>> +{
> >>> +	return IS_ENABLED(CONFIG_ARM64_HAFT) &&
> >>> +		cpus_have_final_cap(ARM64_HAFT);
> >>> +}
> >>
> >> I'm fine with this approach. If we ever get hardware with mismatched
> >> FEAT_HAFT and some secondary CPUs don't come up, we can revisit.
> >>
> >>> diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S
> >>> index ccbae4525891..0bc88df7cb35 100644
> >>> --- a/arch/arm64/mm/proc.S
> >>> +++ b/arch/arm64/mm/proc.S
> >>> @@ -498,6 +498,10 @@ alternative_else_nop_endif
> >>>  	and	x9, x9, ID_AA64MMFR1_EL1_HAFDBS_MASK
> >>>  	cbz	x9, 1f
> >>>  	orr	tcr, tcr, #TCR_HA		// hardware Access flag update
> >>> +
> >>> +#ifdef CONFIG_ARM64_HAFT
> >>> +	orr	tcr2, tcr2, TCR2_EL1x_HAFT
> >>> +#endif /* CONFIG_ARM64_HAFT */
> >>>  1:
> >>>  #endif	/* CONFIG_ARM64_HW_AFDBM */
> >>>  	msr	mair_el1, mair
> >>
> >> If you still want #ifdefs, I'd have left it outside the HW_AFDBM. We
> >> already have a dependency in the Kconfig. Anyway, I can fix this up.
> > 
> > yes it has already depend on the HW_AFDBM. And one asm won't cause much to the
> > Image size if user want CONFIG_ARM64_HAFT=n. I'll drop the #ifdef here.
> > 
> 
> I rethink it and maybe we still need the #ifdef here considering one case: the hardware
> supports FEAT_HAFT while user make CONFIG_ARM64_HAFT=n, in such case the HAFT will be
> enabled unexpectedly if no CONFIG_ARM64_HAFT protection here.

Yes, still keeping the #ifdef but outside of HW_AFDBM. I can fix it up
myself when applying the patches.

-- 
Catalin