Re: [PATCH] kvm:x86:Fix incorrect error paths in the function process_smi

[Paolo Bonzini <pbonzini@redhat.com>]

On 18/09/2015 16:46, Nicholas Krause wrote:
> This fixes the incorrect assumation that the function
> kvm_vcpu_write_guest always runs successfully and
> therefore checks if it fails by returning a error code
> and if so return immediately to the caller of the function
> process_smi as we cannot this work's intended work after
> this failed call to kvm_vcpu_write_guest.
> 
> Signed-off-by: Nicholas Krause <xerofoify@gmail.com>
> ---
>  arch/x86/kvm/x86.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index a60bdbc..4743a4b 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -6088,7 +6088,8 @@ static void process_smi(struct kvm_vcpu *vcpu)
>  	else
>  		process_smi_save_state_32(vcpu, buf);
>  
> -	kvm_vcpu_write_guest(vcpu, vcpu->arch.smbase + 0xfe00, buf, sizeof(buf));
> +	if (kvm_vcpu_write_guest(vcpu, vcpu->arch.smbase + 0xfe00, buf, sizeof(buf)))
> +		return;
>  
>  	if (kvm_x86_ops->get_nmi_mask(vcpu))
>  		vcpu->arch.hflags |= HF_SMM_INSIDE_NMI_MASK;
> 

The SMI was requested by the guest.  If the guest failed to set up SMM
correctly, it's the guest's problem.  Aborting the SMI does not match
what a real processor does.

Thanks,

Paolo