Re: [RFC PATCH 01/10] arm64: cpufeature: allow for version discrepancy in PMU implementations

[Mark Rutland]

On Tue, Jan 03, 2017 at 06:10:18PM +0000, Will Deacon wrote:
> Perf already supports multiple PMU instances for heterogeneous systems,
> so there's no need to be strict in the cpufeature checking, particularly
> as the PMU extension is optional in the architecture.
> 
> Reviewed-by: Suzuki K Poulose <suzuki.poulose@xxxxxxx>
> Signed-off-by: Will Deacon <will.deacon@xxxxxxx>

There are remaining issues with PMU support exposed to KVM guests in
hetereogeneous systems, but I think that's a larger issue with KVM and
heterogeneous CPUs (and we're already aware of it), so FWIW:

Acked-by: Mark Rutland <mark.rutland@xxxxxxx>

Thanks,
Mark.

> ---
>  arch/arm64/kernel/cpufeature.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index fdf8f045929f..47d0226620e8 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
> @@ -184,7 +184,11 @@ static const struct arm64_ftr_bits ftr_id_aa64dfr0[] = {
>  	ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_CTX_CMPS_SHIFT, 4, 0),
>  	ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_WRPS_SHIFT, 4, 0),
>  	ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_BRPS_SHIFT, 4, 0),
> -	S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_PMUVER_SHIFT, 4, 0),
> +	/*
> +	 * We can instantiate multiple PMU instances with different levels
> +	 * of support.
> +	 * */
> +	S_ARM64_FTR_BITS(FTR_NONSTRICT, FTR_EXACT, ID_AA64DFR0_PMUVER_SHIFT, 4, 0),
>  	ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_TRACEVER_SHIFT, 4, 0),
>  	ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_DEBUGVER_SHIFT, 4, 0x6),
>  	ARM64_FTR_END,
> -- 
> 2.1.4
>