Re: [PATCH v5 09/20] KVM: nVMX: Propagate extended instruction information

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The patch defines KVM_X86_FEATURE_APX but fails to initialize it in KVM's supported capabilities via kvm_cpu_cap_init(), causing the new nested VMCS propagation to be dead code.
--

--- Patch [9]: [PATCH v5 09/20] KVM: nVMX: Propagate extended instruction information ---
commit 2483dfe37a64f0e1963aeef4b3b32d649b5a5dda
Author: Chang S. Bae <chang.seok.bae@intel.com>

KVM: nVMX: Propagate extended instruction information

Define the new extended_instruction_info field in struct vmcs12 and
propagate it to nested VMX. Gate the propagation on the guest APX
enumeration, which aligns with bare metal behavior.

Link: https://lore.kernel.org/aRvOSnaUt1E+%2FpkC@intel.com

> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 8e0523e40477d..1684f5d1dbf4d 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
[ ... ]
> @@ -4766,6 +4766,12 @@ static void prepare_vmcs12(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
>  		vmcs12->vm_exit_intr_info = exit_intr_info;
>  		vmcs12->vm_exit_instruction_len = exit_insn_len;
>  		vmcs12->vmx_instruction_info = vmcs_read32(VMX_INSTRUCTION_INFO);
> +		/*
> +		 * The APX enumeration guarantees the presence of the extended
> +		 * fields. This CPUID bit alone is sufficient to rely on it.
> +		 */
> +		if (guest_cpu_cap_has(vcpu, X86_FEATURE_APX))
> +			vmcs12->extended_instruction_info = vmcs_read64(EXTENDED_INSTRUCTION_INFO);

[Severity: Medium]
Will this condition ever evaluate to true without first initializing the APX
capability in KVM's supported features?

When userspace sets the guest CPUID via KVM_SET_CPUID2, KVM masks off
capabilities that are not explicitly initialized. It appears F(APX) is missing
from the CPUID_7_1_EDX capability mask in kvm_set_cpu_caps():

arch/x86/kvm/cpuid.c:kvm_set_cpu_caps() {
	...
	kvm_cpu_cap_init(CPUID_7_1_EDX,
		F(AVX_VNNI_INT8),
		F(AVX_NE_CONVERT),
		F(AMX_COMPLEX),
		F(AVX_VNNI_INT16),
		F(PREFETCHITI),
		F(AVX10),
	);
	...
}

Does this omission cause the newly added nested VMX propagation for APX to act
as dead code, since the APX bit in vcpu->arch.cpu_caps would always remain 0?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260720171949.498680-1-chang.seok.bae@intel.com?part=9