Re: [Xen-devel] [PATCH v5 07/15] VMX: add VMFUNC leaf 0 (EPTP switching) to emulator.

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 14.07.15 at 02:14, <edmund.h.white@xxxxxxxxx> wrote:
> --- a/xen/arch/x86/hvm/emulate.c
> +++ b/xen/arch/x86/hvm/emulate.c
> @@ -1436,6 +1436,19 @@ static int hvmemul_invlpg(
>      return rc;
>  }
>  
> +static int hvmemul_vmfunc(
> +    struct x86_emulate_ctxt *ctxt)
> +{
> +    int rc;
> +
> +    rc = hvm_funcs.altp2m_vcpu_emulate_vmfunc(ctxt->regs);
> +    if ( rc != X86EMUL_OKAY )
> +    {
> +        hvmemul_inject_hw_exception(TRAP_invalid_op, 0, ctxt);
> +    }
> +    return rc;

Pointless braces and missing blank line before final return.

> @@ -1830,6 +1831,19 @@ static void vmx_vcpu_update_vmfunc_ve(struct vcpu *v)
>      vmx_vmcs_exit(v);
>  }
>  
> +static int vmx_vcpu_emulate_vmfunc(struct cpu_user_regs *regs)
> +{
> +    int rc = X86EMUL_EXCEPTION;
> +    struct vcpu *curr = current;
> +
> +    if ( !cpu_has_vmx_vmfunc && altp2m_active(curr->domain) &&
> +         regs->eax == 0 &&
> +         p2m_switch_vcpu_altp2m_by_id(curr, (uint16_t)regs->ecx) )

Documentation suggests that the upper 32 bits of RAX are being
ignored, and that all 32 bits of ECX are being used.

> @@ -3234,6 +3263,15 @@ void vmx_vmexit_handler(struct cpu_user_regs *regs)
>              update_guest_eip();
>          break;
>  
> +    case EXIT_REASON_VMFUNC:
> +        if ( (vmx_vmfunc_intercept(regs) == X86EMUL_EXCEPTION) ||
> +             (vmx_vmfunc_intercept(regs) == X86EMUL_UNHANDLEABLE) ||
> +             (vmx_vmfunc_intercept(regs) == X86EMUL_RETRY) )

Why would you want to invoke the function 3 times? How about
simply != X86EMUL_OKAY?

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel