Re: [PATCH 1/8] KVM: arm64: Enforce absence of FEAT_FGT on FGT registers

[Oliver Upton <oliver.upton@linux.dev>]

Hey,

On Wed, Sep 17, 2025 at 05:58:33PM +0100, Marc Zyngier wrote:

Did you mean to add changelogs to these patches?

> Signed-off-by: Marc Zyngier <maz@kernel.org>
> ---
>  arch/arm64/kvm/config.c | 201 +++++++++++++++++++++++++++-------------
>  1 file changed, 137 insertions(+), 64 deletions(-)
> 
> diff --git a/arch/arm64/kvm/config.c b/arch/arm64/kvm/config.c
> index da66c4a147752..42b834c82a20d 100644
> --- a/arch/arm64/kvm/config.c
> +++ b/arch/arm64/kvm/config.c
> @@ -8,11 +8,16 @@
>  #include <asm/sysreg.h>
>  
>  struct reg_bits_to_feat_map {
> -	u64		bits;
> +	union {
> +		u64	bits;
> +		u64	*res0p;
> +	};
>  
>  #define	NEVER_FGU	BIT(0)	/* Can trap, but never UNDEF */
>  #define	CALL_FUNC	BIT(1)	/* Needs to evaluate tons of crap */
>  #define	FIXED_VALUE	BIT(2)	/* RAZ/WI or RAO/WI in KVM */
> +#define	RES0_POINTER	BIT(3)	/* Pointer to RES0 value instead of bits */
> +
>  	unsigned long	flags;
>  
>  	union {
> @@ -28,9 +33,16 @@ struct reg_bits_to_feat_map {
>  	};
>  };
>  
> -#define __NEEDS_FEAT_3(m, f, id, fld, lim)		\
> +struct reg_to_feat_map {
> +	const char				*name;
> +	const struct reg_bits_to_feat_map	feat_map;

Some documentation might help with confusion between this and
bit_feat_map. IIUC you're using a single NEEDS_FEAT() expression to RES0
the whole sucker based on whether or not the FEAT_XXX for the EL2
register is present?

> +	const struct reg_bits_to_feat_map	*bit_feat_map;
> +	const unsigned int			bit_feat_map_sz;
> +};

Ok, differentiating "reg_to_feat_map" and "reg_bits_to_feat_map" is a
bit hard on the reader... Could this maybe be called "reg_feat_map_desc"
or similar?

This refactoring could also be done as a separate patch.
 
Thanks,
Oliver