Re: [PATCH v6 34/43] arm64: RME: Propagate number of breakpoints and watchpoints to userspace

[Gavin Shan <gshan@redhat.com>]

On 2/3/25 9:15 AM, Gavin Shan wrote:
> On 12/13/24 1:55 AM, Steven Price wrote:
>> From: Jean-Philippe Brucker <jean-philippe@linaro.org>
>>
>> The RMM describes the maximum number of BPs/WPs available to the guest
>> in the Feature Register 0. Propagate those numbers into ID_AA64DFR0_EL1,
>> which is visible to userspace. A VMM needs this information in order to
>> set up realm parameters.
>>
>> Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
>> Signed-off-by: Steven Price <steven.price@arm.com>
>> ---
>>   arch/arm64/include/asm/kvm_rme.h |  2 ++
>>   arch/arm64/kvm/rme.c             | 22 ++++++++++++++++++++++
>>   arch/arm64/kvm/sys_regs.c        |  2 +-
>>   3 files changed, 25 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/arm64/include/asm/kvm_rme.h b/arch/arm64/include/asm/kvm_rme.h
>> index 0d89ab1645c1..f8e37907e2d5 100644
>> --- a/arch/arm64/include/asm/kvm_rme.h
>> +++ b/arch/arm64/include/asm/kvm_rme.h
>> @@ -85,6 +85,8 @@ void kvm_init_rme(void);
>>   u32 kvm_realm_ipa_limit(void);
>>   u32 kvm_realm_vgic_nr_lr(void);
>> +u64 kvm_realm_reset_id_aa64dfr0_el1(const struct kvm_vcpu *vcpu, u64 val);
>> +
>>   bool kvm_rme_supports_sve(void);
>>   int kvm_realm_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap);
>> diff --git a/arch/arm64/kvm/rme.c b/arch/arm64/kvm/rme.c
>> index e562e77c1f94..d21042d5ec9a 100644
>> --- a/arch/arm64/kvm/rme.c
>> +++ b/arch/arm64/kvm/rme.c
>> @@ -63,6 +63,28 @@ u32 kvm_realm_vgic_nr_lr(void)
>>       return u64_get_bits(rmm_feat_reg0, RMI_FEATURE_REGISTER_0_GICV3_NUM_LRS);
>>   }
>> +u64 kvm_realm_reset_id_aa64dfr0_el1(const struct kvm_vcpu *vcpu, u64 val)
>> +{
>> +    u32 bps = u64_get_bits(rmm_feat_reg0, RMI_FEATURE_REGISTER_0_NUM_BPS);
>> +    u32 wps = u64_get_bits(rmm_feat_reg0, RMI_FEATURE_REGISTER_0_NUM_WPS);
>> +    u32 ctx_cmps;
>> +
>> +    if (!kvm_is_realm(vcpu->kvm))
>> +        return val;
>> +
>> +    /* Ensure CTX_CMPs is still valid */
>> +    ctx_cmps = FIELD_GET(ID_AA64DFR0_EL1_CTX_CMPs, val);
>> +    ctx_cmps = min(bps, ctx_cmps);
>> +
>> +    val &= ~(ID_AA64DFR0_EL1_BRPs_MASK | ID_AA64DFR0_EL1_WRPs_MASK |
>> +         ID_AA64DFR0_EL1_CTX_CMPs);
>> +    val |= FIELD_PREP(ID_AA64DFR0_EL1_BRPs_MASK, bps) |
>> +           FIELD_PREP(ID_AA64DFR0_EL1_WRPs_MASK, wps) |
>> +           FIELD_PREP(ID_AA64DFR0_EL1_CTX_CMPs, ctx_cmps);
>> +
>> +    return val;
>> +}
>> +
> 
> The the filed ID_AA64DFR0_EL1_WRPs_MASK of the system register ID_AA64DFR0_EL1 is
> writtable, as declared in sys_reg.c. We need to consolidate the field when the
> system register is written.
> 
>          ID_FILTERED(ID_AA64DFR0_EL1, id_aa64dfr0_el1,
>                      ID_AA64DFR0_EL1_DoubleLock_MASK |
>                      ID_AA64DFR0_EL1_WRPs_MASK |
>                      ID_AA64DFR0_EL1_PMUVer_MASK |
>                      ID_AA64DFR0_EL1_DebugVer_MASK),
> 

Please ignore this comment. The consolidation when the system register is
written has been covered by PATCH[35/43].

>>   static int get_start_level(struct realm *realm)
>>   {
>>       return 4 - stage2_pgtable_levels(realm->ia_bits);
>> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
>> index a4713609e230..55cde43b36b9 100644
>> --- a/arch/arm64/kvm/sys_regs.c
>> +++ b/arch/arm64/kvm/sys_regs.c
>> @@ -1806,7 +1806,7 @@ static u64 sanitise_id_aa64dfr0_el1(const struct kvm_vcpu *vcpu, u64 val)
>>       /* Hide SPE from guests */
>>       val &= ~ID_AA64DFR0_EL1_PMSVer_MASK;
>> -    return val;
>> +    return kvm_realm_reset_id_aa64dfr0_el1(vcpu, val);
>>   }
>>   static int set_id_aa64dfr0_el1(struct kvm_vcpu *vcpu,

Thanks,
Gavin