Maxim Levitsky <mlevitsk@xxxxxxxxxx> writes:
> On Fri, 2022-07-08 at 16:42 +0200, Vitaly Kuznetsov wrote:
>> KVM has to check guest visible HYPERV_CPUID_NESTED_FEATURES.EBX CPUID
>> leaf to know with Enlightened VMCS definition to use (original or 2022
>> update). Cache the leaf along with other Hyper-V CPUID feature leaves
>> to make the check quick.
>>
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
>> ---
>> arch/x86/include/asm/kvm_host.h | 2 ++
>> arch/x86/kvm/hyperv.c | 9 +++++++++
>> 2 files changed, 11 insertions(+)
>>
>> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
>> index de5a149d0971..077ec9cf3169 100644
>> --- a/arch/x86/include/asm/kvm_host.h
>> +++ b/arch/x86/include/asm/kvm_host.h
>> @@ -616,6 +616,8 @@ struct kvm_vcpu_hv {
>> u32 enlightenments_eax; /* HYPERV_CPUID_ENLIGHTMENT_INFO.EAX */
>> u32 enlightenments_ebx; /* HYPERV_CPUID_ENLIGHTMENT_INFO.EBX */
>> u32 syndbg_cap_eax; /* HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES.EAX */
>> + u32 nested_eax; /* HYPERV_CPUID_NESTED_FEATURES.EAX */
>> + u32 nested_ebx; /* HYPERV_CPUID_NESTED_FEATURES.EBX */
>> } cpuid_cache;
>> };
>>
>> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
>> index e08189211d9a..b666902da4d9 100644
>> --- a/arch/x86/kvm/hyperv.c
>> +++ b/arch/x86/kvm/hyperv.c
>> @@ -2030,6 +2030,15 @@ void kvm_hv_set_cpuid(struct kvm_vcpu *vcpu)
>> hv_vcpu->cpuid_cache.syndbg_cap_eax = entry->eax;
>> else
>> hv_vcpu->cpuid_cache.syndbg_cap_eax = 0;
>> +
>> + entry = kvm_find_cpuid_entry(vcpu, HYPERV_CPUID_NESTED_FEATURES, 0);
>> + if (entry) {
>> + hv_vcpu->cpuid_cache.nested_eax = entry->eax;
>> + hv_vcpu->cpuid_cache.nested_ebx = entry->ebx;
>> + } else {
>> + hv_vcpu->cpuid_cache.nested_eax = 0;
>> + hv_vcpu->cpuid_cache.nested_ebx = 0;
>> + }
>> }
>>
>> int kvm_hv_set_enforce_cpuid(struct kvm_vcpu *vcpu, bool enforce)
>
>
> Small Nitpick:
>
> If I understand correctly, the kvm_find_cpuid_entry can fail if the userspace didn't provide the
> cpuid entry.
>
> Since the code that deals with failback is now repeated 3 times, how about some wrapper function that
> will return all zeros for a non present cpuid entry?
I've opted for wiping the whole hv_vcpu->cpuid_cache with memset(), this
way we don't even need a new helper.
--
Vitaly
