Re: [PATCH v8 01/21] xen/x86: add bitmap of enabled emulated devices

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 17/11/15 09:58, Juergen Gross wrote:
> On 17/11/15 10:42, Jan Beulich wrote:
>>>>> On 16.11.15 at 19:33, <andrew.cooper3@citrix.com> wrote:
>>> On 16/11/15 12:18, Jan Beulich wrote:
>>>>>>> On 06.11.15 at 17:05, <roger.pau@citrix.com> wrote:
>>>>> --- a/xen/include/public/arch-x86/xen.h
>>>>> +++ b/xen/include/public/arch-x86/xen.h
>>>>> @@ -265,7 +265,31 @@ typedef struct arch_shared_info arch_shared_info_t;
>>>>>   * XEN_DOMCTL_INTERFACE_VERSION.
>>>>>   */
>>>>>  struct xen_arch_domainconfig {
>>>>> -    char dummy;
>>>>> +#define _XEN_X86_EMU_LAPIC          0
>>>>> +#define XEN_X86_EMU_LAPIC           (1U<<_XEN_X86_EMU_LAPIC)
>>>>> +#define _XEN_X86_EMU_HPET           1
>>>>> +#define XEN_X86_EMU_HPET            (1U<<_XEN_X86_EMU_HPET)
>>>>> +#define _XEN_X86_EMU_PM             2
>>>>> +#define XEN_X86_EMU_PM              (1U<<_XEN_X86_EMU_PM)
>>>>> +#define _XEN_X86_EMU_RTC            3
>>>>> +#define XEN_X86_EMU_RTC             (1U<<_XEN_X86_EMU_RTC)
>>>>> +#define _XEN_X86_EMU_IOAPIC         4
>>>>> +#define XEN_X86_EMU_IOAPIC          (1U<<_XEN_X86_EMU_IOAPIC)
>>>>> +#define _XEN_X86_EMU_PIC            5
>>>>> +#define XEN_X86_EMU_PIC             (1U<<_XEN_X86_EMU_PIC)
>>>>> +#define _XEN_X86_EMU_VGA            6
>>>>> +#define XEN_X86_EMU_VGA             (1U<<_XEN_X86_EMU_VGA)
>>>>> +#define _XEN_X86_EMU_IOMMU          7
>>>>> +#define XEN_X86_EMU_IOMMU           (1U<<_XEN_X86_EMU_IOMMU)
>>>>> +#define _XEN_X86_EMU_PIT            8
>>>>> +#define XEN_X86_EMU_PIT             (1U<<_XEN_X86_EMU_PIT)
>>>> While only used for a domctl (so far), I still think we should aim at
>>>> making this a complete set (i.e. preempt future additions to the
>>>> set if at all possible). I say this because - having looked again - I'm
>>>> missing things like MTRR, PAT, and 8254 here.
>>> Use (or not) of MTRR and PAT should be controlled exclusively via the
>>> guests cpuid policy.  Unlike the above bits, they are architectural
>>> components of the CPU itself, rather than external devices on the
>>> motherboard.
>> For MTRR - yes, agreed. But what CPUID bit do I not recall that allows
>> identifying PAT availability?
> INTEL SDM Vol. 3 11.12.1:
>
> An operating system or executive can detect the availability of the PAT
> by executing the CPUID instruction with a value of 1 in the EAX
> register. Support for the PAT is indicated by the PAT flag (bit 16 of
> the values returned to EDX register).

Indeed.  X86_FEATURE_PAT.

~Andrew