Re: [Xen-devel] [PATCH v2] x86: do not enable global pages when virtualized on AMD hardware

[Jan Beulich <jbeulich@xxxxxxxx>]

On 04.12.2019 12:52, Roger Pau Monné wrote:
> On Wed, Dec 04, 2019 at 12:05:35PM +0100, Jan Beulich wrote:
>> On 04.12.2019 11:44, Roger Pau Monne wrote:
>>> --- a/xen/arch/x86/pv/domain.c
>>> +++ b/xen/arch/x86/pv/domain.c
>>> @@ -118,11 +118,18 @@ unsigned long pv_fixup_guest_cr4(const struct vcpu 
>>> *v, unsigned long cr4)
>>>              (mmu_cr4_features & PV_CR4_GUEST_VISIBLE_MASK));
>>>  }
>>>  
>>> +static int opt_global_pages = -1;
>>
>> int8_t __read_mostly
>>
>>> +boolean_runtime_param("global-pages", opt_global_pages);
>>> +
>>>  unsigned long pv_make_cr4(const struct vcpu *v)
>>>  {
>>>      const struct domain *d = v->domain;
>>>      unsigned long cr4 = mmu_cr4_features &
>>>          ~(X86_CR4_PCIDE | X86_CR4_PGE | X86_CR4_TSD);
>>> +    bool pge = opt_global_pages == -1 ? (!cpu_has_hypervisor ||
>>> +                                         boot_cpu_data.x86_vendor !=
>>> +                                         X86_VENDOR_AMD)
>>> +                                      : !!opt_global_pages;
>>
>> Let's avoid re-doing this evaluation each time we come here.
>> Post boot the value can only change to 0 or 1. Hence in some
>> __init function you can apply the default calculation done
>> here.
> 
> I've assumed that boolean_runtime_param can be changed during runtime
> by the user, and hence the value calculated at boot time would become
> stale if the user changes it after boot, which should be fine for this
> option.

I'm afraid I can't decide whether you agree or disagree with my
comment.

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel