qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [RFC PATCH v4 18/20] target/i386: add cpuid Fn8000_001f


From: Eduardo Habkost
Subject: Re: [Qemu-devel] [RFC PATCH v4 18/20] target/i386: add cpuid Fn8000_001f
Date: Wed, 8 Mar 2017 18:29:38 -0300
User-agent: Mutt/1.7.1 (2016-10-04)

On Wed, Mar 08, 2017 at 03:54:20PM -0500, Brijesh Singh wrote:
> Fn8000_001f cpuid provides the memory encryption (aka C-bit) location
> in a page table for the SEV-enabled guest.
> 
> Signed-off-by: Brijesh Singh <address@hidden>
> ---
>  target/i386/cpu.c |    6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index fba9212..44662eb 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -40,6 +40,7 @@
>  #include "qapi/visitor.h"
>  #include "qom/qom-qobject.h"
>  #include "sysemu/arch_init.h"
> +#include "sysemu/sev.h"
>  
>  #if defined(CONFIG_KVM)
>  #include <linux/kvm_para.h>
> @@ -2966,6 +2967,11 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, 
> uint32_t count,
>              *edx = 0;
>          }
>          break;
> +    case 0x8000001F:
> +        if (sev_enabled()) {
> +            host_cpuid(index, 0, eax, ebx, ecx, edx);

This doesn't look future-proof. The architecture may be extended
to include CPUID bits that we can't expose to the guest unless we
implement host-side support for it.

I suggest including only the specific bits we know we can really
expose to the guest, instead of exposing raw host CPUID directly.

Also, exposing only explicitly configured CPUID bits will make it
easier to implement migration support in the future.

> +        }
> +        break;
>      case 0xC0000000:
>          *eax = env->cpuid_xlevel2;
>          *ebx = 0;
> 

-- 
Eduardo



reply via email to

[Prev in Thread] Current Thread [Next in Thread]