Re: [PATCH v7 03/12] KVM: arm64: PMU: Clear PM{C,I}NTEN{SET,CLR} and PMOVS{SET,CLR} on vCPU reset

From: Eric Auger
Date: Tue Oct 17 2023 - 05:24:45 EST


Hi,
On 10/16/23 23:28, Raghavendra Rao Ananta wrote:
> On Mon, Oct 16, 2023 at 12:45 PM Eric Auger <eauger@xxxxxxxxxx> wrote:
>>
>> Hi Raghavendra,
>>
>> On 10/10/23 01:08, Raghavendra Rao Ananta wrote:
>>> From: Reiji Watanabe <reijiw@xxxxxxxxxx>
>>>
>>> On vCPU reset, PMCNTEN{SET,CLR}_EL0, PMINTEN{SET,CLR}_EL1, and
>>> PMOVS{SET,CLR}_EL1 for a vCPU are reset by reset_pmu_reg().
>> PMOVS{SET,CLR}_EL0?
> Ah, yes. It should be PMOVS{SET,CLR}_EL0.
>
>>> This function clears RAZ bits of those registers corresponding
>>> to unimplemented event counters on the vCPU, and sets bits
>>> corresponding to implemented event counters to a predefined
>>> pseudo UNKNOWN value (some bits are set to 1).
>>>
>>> The function identifies (un)implemented event counters on the
>>> vCPU based on the PMCR_EL0.N value on the host. Using the host
>>> value for this would be problematic when KVM supports letting
>>> userspace set PMCR_EL0.N to a value different from the host value
>>> (some of the RAZ bits of those registers could end up being set to 1).
>>>
>>> Fix this by clearing the registers so that it can ensure
>>> that all the RAZ bits are cleared even when the PMCR_EL0.N value
>>> for the vCPU is different from the host value. Use reset_val() to
>>> do this instead of fixing reset_pmu_reg(), and remove
>>> reset_pmu_reg(), as it is no longer used.
>> do you intend to restore the 'unknown' behavior at some point?
>>
> I believe Reiji's (original author) intention was to keep them
> cleared, which would still imply an 'unknown' behavior. Do you think
> there's an issue with this?
Then why do we bother using reset_unknown in the other places if
clearing the bits is enough here?

Thanks

Eric
>
> Thank you.
> Raghavendra
>> Thanks
>>
>> Eric
>>>
>>> Signed-off-by: Reiji Watanabe <reijiw@xxxxxxxxxx>
>>> Signed-off-by: Raghavendra Rao Ananta <rananta@xxxxxxxxxx>
>>> ---
>>> arch/arm64/kvm/sys_regs.c | 21 +--------------------
>>> 1 file changed, 1 insertion(+), 20 deletions(-)
>>>
>>> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
>>> index 818a52e257ed..3dbb7d276b0e 100644
>>> --- a/arch/arm64/kvm/sys_regs.c
>>> +++ b/arch/arm64/kvm/sys_regs.c
>>> @@ -717,25 +717,6 @@ static unsigned int pmu_visibility(const struct kvm_vcpu *vcpu,
>>> return REG_HIDDEN;
>>> }
>>>
>>> -static u64 reset_pmu_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)
>>> -{
>>> - u64 n, mask = BIT(ARMV8_PMU_CYCLE_IDX);
>>> -
>>> - /* No PMU available, any PMU reg may UNDEF... */
>>> - if (!kvm_arm_support_pmu_v3())
>>> - return 0;
>>> -
>>> - n = read_sysreg(pmcr_el0) >> ARMV8_PMU_PMCR_N_SHIFT;
>>> - n &= ARMV8_PMU_PMCR_N_MASK;
>>> - if (n)
>>> - mask |= GENMASK(n - 1, 0);
>>> -
>>> - reset_unknown(vcpu, r);
>>> - __vcpu_sys_reg(vcpu, r->reg) &= mask;
>>> -
>>> - return __vcpu_sys_reg(vcpu, r->reg);
>>> -}
>>> -
>>> static u64 reset_pmevcntr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)
>>> {
>>> reset_unknown(vcpu, r);
>>> @@ -1115,7 +1096,7 @@ static bool access_pmuserenr(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
>>> trap_wcr, reset_wcr, 0, 0, get_wcr, set_wcr }
>>>
>>> #define PMU_SYS_REG(name) \
>>> - SYS_DESC(SYS_##name), .reset = reset_pmu_reg, \
>>> + SYS_DESC(SYS_##name), .reset = reset_val, \
>>> .visibility = pmu_visibility
>>>
>>> /* Macro to expand the PMEVCNTRn_EL0 register */
>>
>