> On 16. Feb 2026, at 04:44, Richard Henderson <[email protected]> 
> wrote:
> 
> We no longer support kvm for aarch32, therefore the aarch64
> test is trivially true.
> 
Hello,

Probably harmless but aren’t AArch32 guests on an AArch64 KVM host still 
supported?

aarch64_cpu_set_aarch64 in target/arm/cpu.c unsets ARM_FEATURE_AARCH64
for being able to support that case.

Thank you,
> 
> 
> Signed-off-by: Richard Henderson <[email protected]>
> ---
> target/arm/cpu64.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
> index 5d7c6b7fbb..10c6796548 100644
> --- a/target/arm/cpu64.c
> +++ b/target/arm/cpu64.c
> @@ -767,9 +767,7 @@ static void aarch64_host_initfn(Object *obj)
>     ARMCPU *cpu = ARM_CPU(obj);
> #if defined(CONFIG_KVM)
>     kvm_arm_set_cpu_features_from_host(cpu);
> -    if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64)) {
> -        aarch64_add_sve_properties(obj);
> -    }
> +    aarch64_add_sve_properties(obj);
> #elif defined(CONFIG_HVF)
>     hvf_arm_set_cpu_features_from_host(cpu);
> #elif defined(CONFIG_WHPX)
> -- 
> 2.43.0
> 
> 


Reply via email to