Re: [PATCH v5 12/24] KVM: arm64: Use physical PMSELR for PMXEVTYPER if partitioned

[Oliver Upton <oupton@kernel.org>]

On Tue, Dec 09, 2025 at 08:51:09PM +0000, Colton Lewis wrote:
> Because PMXEVTYPER is trapped and PMSELR is not, it is not appropriate
> to use the virtual PMSELR register when it could be outdated and lead
> to an invalid write. Use the physical register when partitioned.
> 
> Signed-off-by: Colton Lewis <coltonlewis@google.com>
> ---
>  arch/arm64/include/asm/arm_pmuv3.h | 7 ++++++-
>  arch/arm64/kvm/sys_regs.c          | 9 +++++++--
>  2 files changed, 13 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/arm_pmuv3.h b/arch/arm64/include/asm/arm_pmuv3.h
> index 27c4d6d47da31..60600f04b5902 100644
> --- a/arch/arm64/include/asm/arm_pmuv3.h
> +++ b/arch/arm64/include/asm/arm_pmuv3.h
> @@ -70,11 +70,16 @@ static inline u64 read_pmcr(void)
>  	return read_sysreg(pmcr_el0);
>  }
>  
> -static inline void write_pmselr(u32 val)
> +static inline void write_pmselr(u64 val)
>  {
>  	write_sysreg(val, pmselr_el0);
>  }
>  
> +static inline u64 read_pmselr(void)
> +{
> +	return read_sysreg(pmselr_el0);
> +}
> +
>  static inline void write_pmccntr(u64 val)
>  {
>  	write_sysreg(val, pmccntr_el0);
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 0c9596325519b..2e6d907fa8af2 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -1199,14 +1199,19 @@ static bool writethrough_pmevtyper(struct kvm_vcpu *vcpu, struct sys_reg_params
>  static bool access_pmu_evtyper(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
>  			       const struct sys_reg_desc *r)
>  {
> -	u64 idx, reg;
> +	u64 idx, reg, pmselr;
>  
>  	if (pmu_access_el0_disabled(vcpu))
>  		return false;
>  
>  	if (r->CRn == 9 && r->CRm == 13 && r->Op2 == 1) {
>  		/* PMXEVTYPER_EL0 */
> -		idx = SYS_FIELD_GET(PMSELR_EL0, SEL, __vcpu_sys_reg(vcpu, PMSELR_EL0));
> +		if (kvm_vcpu_pmu_is_partitioned(vcpu))
> +			pmselr = read_pmselr();
> +		else
> +			pmselr = __vcpu_sys_reg(vcpu, PMSELR_EL0);

This isn't preemption safe. Nor should the "if (partitioned) do X else do Y" get
open-coded throughout the shop.

I would rather this be handled with a prepatory patch that provides
generic PMU register accessors to the rest of KVM (e.g.
vcpu_read_pmu_reg() / vcpu_write_pmu_reg()). Internally those helpers
can locate the vCPU's PMU registers (emulated, partitioned in-memory,
partitioned in-CPU).

Thanks,
Oliver