Re: [PATCH 8/8] KVM: x86: add SPEC_CTRL and IBPB_SUPPORT to MSR and CPUID lists

[Eric Wheeler]

On Tue, 9 Jan 2018, Paolo Bonzini wrote:

> Expose them to userspace, now that guests can use them.
> I am not adding cpufeatures here to avoid having a kernel
> that shows spec_ctrl in /proc/cpuinfo and actually has no
> support whatsoever for IBRS/IBPB.  Keep the ugly special-casing
> for now, and clean it up once the generic arch/x86/ code
> learns about them.
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index daa1918031df..4abb37d9f4d8 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -1032,6 +1032,7 @@ unsigned int kvm_get_pt_addr_cnt(void)
>  	MSR_IA32_RTIT_ADDR1_A, MSR_IA32_RTIT_ADDR1_B,
>  	MSR_IA32_RTIT_ADDR2_A, MSR_IA32_RTIT_ADDR2_B,
>  	MSR_IA32_RTIT_ADDR3_A, MSR_IA32_RTIT_ADDR3_B,
> +	MSR_IA32_SPEC_CTRL,
>  };

Hi Paolo,

Thank you for posting this!

I am trying to merge this into 4.14 which does not have 
kvm_get_pt_addr_cnt. The rest of the patch commits, but this gets 
rejected. Is this a necessary part of the commit?

patching file arch/x86/kvm/cpuid.c
Hunk #1 succeeded at 389 (offset -8 lines).
Hunk #2 succeeded at 479 (offset -9 lines).
Hunk #3 succeeded at 636 (offset -27 lines).
patching file arch/x86/kvm/x86.c
Hunk #1 FAILED at 1032.
1 out of 1 hunk FAILED -- saving rejects to file arch/x86/kvm/x86.c.rej

]# cat arch/x86/kvm/x86.c.rej
--- arch/x86/kvm/x86.c
+++ arch/x86/kvm/x86.c
@@ -1032,6 +1032,7 @@
 	MSR_IA32_RTIT_ADDR1_A, MSR_IA32_RTIT_ADDR1_B,
 	MSR_IA32_RTIT_ADDR2_A, MSR_IA32_RTIT_ADDR2_B,
 	MSR_IA32_RTIT_ADDR3_A, MSR_IA32_RTIT_ADDR3_B,
+	MSR_IA32_SPEC_CTRL,
 };
 
 static unsigned num_msrs_to_save;


Thank you for your help!

--
Eric Wheeler


>  
>  static unsigned num_msrs_to_save;
> -- 
> 1.8.3.1
> 
> 
>