Re: [PATCH v2 25/25] KVM: x86/mmu: extract initialization of the page walking data

[Sean Christopherson]

On Mon, Feb 21, 2022, Paolo Bonzini wrote:
> +static void kvm_vcpu_init_walker(struct kvm_vcpu *vcpu,
> +				 struct kvm_mmu *mmu,
> +				 union kvm_mmu_paging_mode new_mode)

kvm_vcpu_init_walker() is a rather odd and almost maliciously obtuse.  We're not
short on space for this one, so how about?

static void kvm_mmu_init_guest_walker(struct kvm_vcpu *vcpu,
				      struct kvm_mmu *mmu,
				      union kvm_mmu_paging_mode new_mode)
>  void kvm_init_mmu(struct kvm_vcpu *vcpu)
>  {
>  	struct kvm_mmu_role_regs regs = vcpu_to_role_regs(vcpu);
>  	union kvm_mmu_paging_mode cpu_mode = kvm_calc_cpu_mode(vcpu, &regs);
>  
> +	kvm_vcpu_init_walker(vcpu, vcpu->arch.walk_mmu, cpu_mode);
>  	if (mmu_is_nested(vcpu))
> -		init_kvm_nested_mmu(vcpu, cpu_mode);
> -	else if (tdp_enabled)
> +		return;

Nice!  I really like that this highlights that the nested_mmu crud is just for
the walker.  Can you also add a comment here explaining that part?

> +
> +	if (tdp_enabled)
>  		init_kvm_tdp_mmu(vcpu, cpu_mode);
>  	else
>  		init_kvm_softmmu(vcpu, cpu_mode);
> -- 
> 2.31.1
>