[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 2/2] x86: hv_evmcs CPU flag support
From: |
Vitaly Kuznetsov |
Subject: |
Re: [Qemu-devel] [PATCH 2/2] x86: hv_evmcs CPU flag support |
Date: |
Mon, 22 Oct 2018 17:40:46 +0200 |
Roman Kagan <address@hidden> writes:
> On Fri, Oct 19, 2018 at 01:14:32PM +0200, Vitaly Kuznetsov wrote:
>> --- a/target/i386/kvm.c
>> +++ b/target/i386/kvm.c
>> @@ -798,6 +798,7 @@ int kvm_arch_init_vcpu(CPUState *cs)
>> uint32_t unused;
>> struct kvm_cpuid_entry2 *c;
>> uint32_t signature[3];
>> + uint16_t evmcs_version;
>> int kvm_base = KVM_CPUID_SIGNATURE;
>> int r;
>> Error *local_err = NULL;
>> @@ -841,7 +842,7 @@ int kvm_arch_init_vcpu(CPUState *cs)
>> memset(signature, 0, 12);
>> memcpy(signature, cpu->hyperv_vendor_id, len);
>> }
>> - c->eax = HV_CPUID_MIN;
>> + c->eax = cpu->hyperv_evmcs ? HV_CPUID_MIN_NESTED : HV_CPUID_MIN;
>
>
> I think these two aren't meant to be used on the hypervisor side. My
> understanding is that HV_CPUID_MIN is only there as a reminder that the
> real Hyper-V exposes at least that many hypervisor-specific leaves so
> the guest can rely on that. So I'd rather use directly
> HV_CPUID_IMPLEMENT_LIMITS : HV_CPUID_NESTED_FEATURES, and not introduce
> HV_CPUID_MIN_NESTED.
Makes sense, will do v2.
> Maybe better yet is to update this field with the
> maximum value while populating HV_* leaves:
>
> if (hyperv_enabled(cpu)) {
> uint32_t *cpuid_40000000_eax;
> c = &cpuid_data.entries[cpuid_i++];
> c->function = HV_CPUID_VENDOR_AND_MAX_FUNCTIONS;
> cpuid_40000000_eax = &c->eax;
> *cpuid_40000000_eax = c->function;
>
> ....
>
> c = &cpuid_data.entries[cpuid_i++];
> c->function = HV_CPUID_...;
> *cpuid_40000000_eax = max(*cpuid_40000000_eax, c->function);
>
> but I think it can be done later and doesn't need to hold this patch.
>
> Another question related to this: are the guests OK with leaves
> 0x40000006..0x40000009 missing?
They seem to be, however, after you've asked I'm leaning towards zeroing
them 'just in case'.
--
Vitaly