Re: [Xen-devel] [PATCH] nested vmx: fix instruction decode segment limit check

[Tim Deegan <tim@xxxxxxx>]

Hi, 

At 16:17 +0100 on 24 Apr (1335284232), Jan Beulich wrote:
> - no limit check in 64-bit mode (is not special in any way)
> - limit check is needed in compatibility mode
> - canonical address check should instead be performed in 64-bit mode
> - the last accessed byte must be within limits, not the first byte past
>   the accessed range
> 
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> 
> --- a/xen/arch/x86/hvm/vmx/vvmx.c
> +++ b/xen/arch/x86/hvm/vmx/vvmx.c
> @@ -319,7 +319,7 @@ static int decode_vmx_inst(struct cpu_us
>  {
>      struct vcpu *v = current;
>      union vmx_inst_info info;
> -    struct segment_register seg;
> +    struct segment_register seg, cs;
>      unsigned long base, index, seg_base, disp, offset;
>      int scale, size;
>  
> @@ -342,6 +342,11 @@ static int decode_vmx_inst(struct cpu_us
>          hvm_get_segment_register(v, sreg_to_index[info.fields.segment], 
> &seg);
>          seg_base = seg.base;
>  
> +        if ( hvm_long_mode_enabled(v) )
> +            hvm_get_segment_register(v, x86_seg_cs, &cs);
> +        else
> +            memset(&cs, 0, sizeof(cs));
> +

I found this a bit confusing - maybe you could extract the attr.fields.l
bit into a bool here instead of zeroing the struct and extracting it later?

Cheers,

Tim.


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