Re: [PATCH v3 1/3] KVM: x86/mmu: Use KVM's max TDP level to determine need for 32-bit TDP root

[Sean Christopherson <seanjc@google.com>]

On Mon, Sep 14, 2026, Yan Zhao wrote:
> On a separate note, though it shouldn't be a real issue, is the below change
> necessary?

I would say this is technically correct, but unnecessary, because I can't imagine
any 32-bit guest firmware would try to relocate BARs above 4G.

> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index bb68a021cfaf..abbba26dbdc6 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -1810,7 +1810,7 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
>                         phys_as = entry->eax & 0xff;
>                         g_phys_as = phys_as;
>                         if (kvm_mmu_get_max_tdp_level() < 5)
> -                               g_phys_as = min(g_phys_as, 48U);
> +                               g_phys_as = min(g_phys_as, kvm_mmu_get_max_tdp_level() == 3 ? 32U : 48U);

FWIW, if we wanted to do this, my vote would be for:

			if (kvm_mmu_get_max_tdp_level() < 4)
				g_phys_as = min(g_phys_as, 32U);
			else if (kvm_mmu_get_max_tdp_level() < 5)
				g_phys_as = min(g_phys_as, 48U);

but I think I'd prefer to just leave this be.  Rr better, drop 32-bit kVM support :-D