Indication for support for SVE will not depend on whether we
perform the query on the main kvm_state or the temp vcpu.
Mirror kvm_arm_pauth_supported.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/kvm64.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
index d16d4ea250..36ff20c9e3 100644
--- a/target/arm/kvm64.c
+++ b/target/arm/kvm64.c
@@ -496,6 +496,11 @@ static bool kvm_arm_pauth_supported(void)
             kvm_check_extension(kvm_state, KVM_CAP_ARM_PTRAUTH_GENERIC));
 }
 
+static bool kvm_arm_svm_supported(void)
+{
+    return kvm_check_extension(kvm_state, KVM_CAP_ARM_SVE);
+}
+
 bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
 {
     /* Identify the feature bits corresponding to the host CPU, and
@@ -675,7 +680,7 @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
         }
     }
 
-    sve_supported = ioctl(fdarray[0], KVM_CHECK_EXTENSION, KVM_CAP_ARM_SVE) > 0;
+    sve_supported = kvm_arm_svm_supported();
 
     /* Add feature bits that can't appear until after VCPU init. */
     if (sve_supported) {
-- 
2.34.1

Hi Richard,

On 2022/7/26 2:14, Richard Henderson wrote:
> Indication for support for SVE will not depend on whether we
> perform the query on the main kvm_state or the temp vcpu.
> Mirror kvm_arm_pauth_supported.
> 
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/kvm64.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
> index d16d4ea250..36ff20c9e3 100644
> --- a/target/arm/kvm64.c
> +++ b/target/arm/kvm64.c
> @@ -496,6 +496,11 @@ static bool kvm_arm_pauth_supported(void)
>              kvm_check_extension(kvm_state, KVM_CAP_ARM_PTRAUTH_GENERIC));
>  }
>  
> +static bool kvm_arm_svm_supported(void)
> +{
> +    return kvm_check_extension(kvm_state, KVM_CAP_ARM_SVE);
> +}
> +

Not sure if it's a typo. Maybe we should instead use
kvm_arm_sve_supported() which was introduced in commit 14e99e0fbbc6.

Zenghui

On 7/25/22 19:02, Zenghui Yu wrote:
> Hi Richard,
> 
> On 2022/7/26 2:14, Richard Henderson wrote:
>> Indication for support for SVE will not depend on whether we
>> perform the query on the main kvm_state or the temp vcpu.
>> Mirror kvm_arm_pauth_supported.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>  target/arm/kvm64.c | 7 ++++++-
>>  1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
>> index d16d4ea250..36ff20c9e3 100644
>> --- a/target/arm/kvm64.c
>> +++ b/target/arm/kvm64.c
>> @@ -496,6 +496,11 @@ static bool kvm_arm_pauth_supported(void)
>>              kvm_check_extension(kvm_state, KVM_CAP_ARM_PTRAUTH_GENERIC));
>>  }
>>
>> +static bool kvm_arm_svm_supported(void)
>> +{
>> +    return kvm_check_extension(kvm_state, KVM_CAP_ARM_SVE);
>> +}
>> +
> 
> Not sure if it's a typo. Maybe we should instead use
> kvm_arm_sve_supported() which was introduced in commit 14e99e0fbbc6.

Oof, it certainly is.

r~