Re: [PATCH v4 4/6] arm64/cpufeature: Add field details for ID_AA64DFR1_EL1 register

[Will Deacon <will@kernel.org>]

On Tue, Apr 07, 2026 at 09:29:46AM -0500, Rob Herring (Arm) wrote:
> From: Anshuman Khandual <anshuman.khandual@arm.com>
> 
> This adds required field details for ID_AA64DFR1_EL1, and also drops dummy
> ftr_raz[] array which is now redundant. These register fields will be used
> to enable increased breakpoint and watchpoint registers via FEAT_Debugv8p9
> later. The register fields have been marked as FTR_STRICT, unless there is
> a known variation in practice.
> 
> Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
> Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
> ---
>  arch/arm64/kernel/cpufeature.c | 21 ++++++++++++++++-----
>  1 file changed, 16 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index c31f8e17732a..24c8e9147e35 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
> @@ -570,6 +570,21 @@ static const struct arm64_ftr_bits ftr_id_aa64dfr0[] = {
>  	ARM64_FTR_END,
>  };
>  
> +static const struct arm64_ftr_bits ftr_id_aa64dfr1[] = {
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_ABL_CMPs_SHIFT, 8, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_DPFZS_SHIFT, 4, 0),

Why does FTR_LOWER_SAFE make sense for DPFZS? From what I can tell, the
new behaviour isn't opt-in, so maybe an FTR_EXACT of 0 would make more
sense if we have to be non-strict (along with a comment like we have for
DFR0.PMUVer)?

> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_EBEP_SHIFT, 4, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_ITE_SHIFT, 4, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_ABLE_SHIFT, 4, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_PMICNTR_SHIFT, 4, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_SPMU_SHIFT, 4, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_CTX_CMPs_SHIFT, 8, 0),

I find it very weird for this to be non-strict when DFR0.CTX_CMPs _is_
strict.

> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_WRPs_SHIFT, 8, 0),
> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_BRPs_SHIFT, 8, 0),

Given that things like hw_breakpoint_reset() rely on the sanitised
register view to determine the number of {break,watch}points, I think we
have to be strict here unless that is changed.

> +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR1_EL1_SYSPMUID_SHIFT, 8, 0),

Again, I'm not sure that FTR_LOWER_SAFE makes a lot of sense here, but
it's hard to tell without an upstream driver for the system PMU.

> +	ARM64_FTR_END,
> +};
> +
>  static const struct arm64_ftr_bits ftr_mvfr0[] = {
>  	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, MVFR0_EL1_FPRound_SHIFT, 4, 0),
>  	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, MVFR0_EL1_FPShVec_SHIFT, 4, 0),
> @@ -756,10 +771,6 @@ static const struct arm64_ftr_bits ftr_single32[] = {
>  	ARM64_FTR_END,
>  };
>  
> -static const struct arm64_ftr_bits ftr_raz[] = {
> -	ARM64_FTR_END,
> -};
> -
>  #define __ARM64_FTR_REG_OVERRIDE(id_str, id, table, ovr) {	\
>  		.sys_id = id,					\
>  		.reg = 	&(struct arm64_ftr_reg){		\
> @@ -832,7 +843,7 @@ static const struct __ftr_reg_entry {
>  
>  	/* Op1 = 0, CRn = 0, CRm = 5 */
>  	ARM64_FTR_REG(SYS_ID_AA64DFR0_EL1, ftr_id_aa64dfr0),
> -	ARM64_FTR_REG(SYS_ID_AA64DFR1_EL1, ftr_raz),
> +	ARM64_FTR_REG(SYS_ID_AA64DFR1_EL1, ftr_id_aa64dfr1),

I'm guessing that KVM will need some updates for this in its sys reg
handling code?

Will