Re: [PATCH] X86: cpuid faulting feature enable

[Keir Fraser <keir.xen@gmail.com>]

On 04/07/2011 02:06, "Shan, Haitao" <haitao.shan@intel.com> wrote:

> diff -r 177ddb2f4ebf -r 71b58748cfee xen/arch/x86/cpu/cpu.h
> --- a/xen/arch/x86/cpu/cpu.h    Fri Jul 01 20:48:00 2011 +0100
> +++ b/xen/arch/x86/cpu/cpu.h    Fri Jul 01 22:28:53 2011 +0100
> @@ -30,4 +30,4 @@ extern void display_cacheinfo(struct cpu
>  extern void generic_identify(struct cpuinfo_x86 * c);
> 
>  extern void early_intel_workaround(struct cpuinfo_x86 *c);
> -
> +extern void intel_cpuid_faulting_flip(unsigned int enable);
> 
> Should we remove this part at all? Seems it is not used now.

Done. Thanks.

> Shan Haitao
> 
> -----Original Message-----
> From: Keir Fraser [mailto:keir.xen@gmail.com]
> Sent: Saturday, July 02, 2011 3:50 PM
> To: Tian, Kevin; Liu, Jinsong; xen-devel@lists.xensource.com
> Cc: Shan, Haitao; Li, Xin
> Subject: Re: [PATCH] X86: cpuid faulting feature enable
> 
> On 02/07/2011 04:15, "Tian, Kevin" <kevin.tian@intel.com> wrote:
> 
>>> From: Keir Fraser [mailto:keir.xen@gmail.com] On Behalf Of Keir
>>> Fraser
>>> Sent: Saturday, July 02, 2011 5:31 AM
>>> 
>>> On 01/07/2011 15:32, "Liu, Jinsong" <jinsong.liu@intel.com> wrote:
>>> 
>>>> X86: cpuid faulting feature enable
>>>> 
>>>> Latest Intel processor add cpuid faulting feature. This patch is
>>>> used to support cpuid faulting in Xen.
>>>> Like cpuid spoofing, cpuid faulting mainly used to support live migration.
>>>> When cpl>0, cpuid instruction will produce GP, vmm then emulate
>>>> execution
>>> of
>>>> the cpuid instruction. Hence will appear to guest software the value
>>>> chosen
>>> by
>>>> the vmm.
>>> 
>>> I fixed this up quite a bit and applied as c/s 23653. Please take a
>>> look and give it a test. In particular note the changes I made in
>>> intel_init(), to make sure that APs are at least as featureful as the
>>> BSP w.r.t. cpuid faulting.
>>> 
>> 
>> 2.42 + } else if (boot_cpu_has(X86_FEATURE_CPUID_FAULTING)) {
>>     2.43 +  BUG_ON(!probe_intel_cpuid_faulting());
>>     2.44 +  set_bit(X86_FEATURE_CPUID_FAULTING, c->x86_capability);
>>     2.45 + }
>>     2.46 +
>>     2.47 + if (!cpu_has_cpuid_faulting)
>>     2.48 +  set_cpuidmask(c);
>>     2.49 +
>>     2.50 + BUG_ON(cpu_has(c, X86_FEATURE_CPUID_FAULTING) !=
>>     2.51 +        boot_cpu_has(X86_FEATURE_CPUID_FAULTING));
>> 
>> the latter BUG_ON is useless since if AP is inconsistent with BSP we
>> already get bug triggered in earlier BUG_ON.
> 
> Oops, I forgot to remove the redundant second BUG_ON. I'll do that now.
> Thanks!
> 
>  -- Keir
> 
>> Thanks
>> Kevin
> 
>