Re: [PATCH] kvm: x86: move architecture-specific code into kvm_arch_vcpu_fault

[Sean Christopherson]

On Wed, Aug 11, 2021, Hou Wenlong wrote:
> The function kvm_arch_vcpu_fault can handle architecture-specific
> case, so move pio-data fault case into it for x86.
> 
> Signed-off-by: Hou Wenlong <houwenlong93@xxxxxxxxxxxxxxxxx>
> ---
>  arch/x86/kvm/x86.c  | 8 ++++++++
>  virt/kvm/kvm_main.c | 4 ----
>  2 files changed, 8 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index e5d5c5ed7dd4..30b0706eced8 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -5348,6 +5348,14 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
>  
>  vm_fault_t kvm_arch_vcpu_fault(struct kvm_vcpu *vcpu, struct vm_fault *vmf)
>  {
> +	if (vmf->pgoff == KVM_PIO_PAGE_OFFSET) {
> +		struct page *page = virt_to_page(vcpu->arch.pio_data);
> +
> +		get_page(page);
> +		vmf->page = page;
> +		return 0;
> +	}
> +
>  	return VM_FAULT_SIGBUS;

What about a prep patch (below) to refactor kvm_arch_vcpu_fault() to return
a struct page pointer instead of vm_fault_t?  That would simplify this patch to:

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 8c1871f0211c..1c5d68ced3be 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5348,6 +5348,8 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
 
 struct page *kvm_arch_vcpu_fault(struct kvm_vcpu *vcpu, struct vm_fault *vmf)
 {
+       if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
+               virt_to_page(vcpu->arch.pio_data);
        return NULL;
 }
 
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index a75848799712..c3b1e8f55251 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -3486,10 +3486,6 @@ static vm_fault_t kvm_vcpu_fault(struct vm_fault *vmf)
 
        if (vmf->pgoff == 0)
                page = virt_to_page(vcpu->run);
-#ifdef CONFIG_X86
-       else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
-               page = virt_to_page(vcpu->arch.pio_data);
-#endif
 #ifdef CONFIG_KVM_MMIO
        else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
                page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);