[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <8736ari0x7.fsf@vitty.brq.redhat.com>
Date: Mon, 02 Mar 2020 10:03:16 +0100
From: Vitaly Kuznetsov <vkuznets@...hat.com>
To: Sean Christopherson <sean.j.christopherson@...el.com>
Cc: Paolo Bonzini <pbonzini@...hat.com>,
Wanpeng Li <wanpengli@...cent.com>,
Jim Mattson <jmattson@...gle.com>,
Joerg Roedel <joro@...tes.org>, kvm@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 00/61] KVM: x86: Introduce KVM cpu caps
Sean Christopherson <sean.j.christopherson@...el.com> writes:
> On Tue, Feb 25, 2020 at 04:18:38PM +0100, Vitaly Kuznetsov wrote:
>> Sean Christopherson <sean.j.christopherson@...el.com> writes:
>>
>> >
>> > 7. Profit!
>>
>> Would it be better or worse if we eliminate set_supported_cpuid() hook
>> completely by doing an ugly hack like (completely untested):
>>
>> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
>> index a2a091d328c6..5ad291d48e1b 100644
>> --- a/arch/x86/include/asm/kvm_host.h
>> +++ b/arch/x86/include/asm/kvm_host.h
>> @@ -1145,8 +1145,6 @@ struct kvm_x86_ops {
>>
>> void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
>>
>> - void (*set_supported_cpuid)(struct kvm_cpuid_entry2 *entry);
>> -
>> bool (*has_wbinvd_exit)(void);
>>
>> u64 (*read_l1_tsc_offset)(struct kvm_vcpu *vcpu);
>> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
>> index e8beb1e542a8..88431fc02797 100644
>> --- a/arch/x86/kvm/cpuid.c
>> +++ b/arch/x86/kvm/cpuid.c
>> @@ -749,6 +749,16 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
>> cpuid_entry_override(entry, CPUID_8000_0008_EBX);
>> break;
>> }
>> + case 0x8000000A:
>> + if (boot_cpu_has(X86_FEATURE_SVM)) {
>> + entry->eax = 1; /* SVM revision 1 */
>> + entry->ebx = 8; /* Lets support 8 ASIDs in case we add proper
>> + ASID emulation to nested SVM */
>> + entry->ecx = 0; /* Reserved */
>> + entry->edx = 0; /* Per default do not support any
>> + additional features */
>
> Lucky thing that you suggested this change, patch ("KVM: SVM: Convert
> feature updates from CPUID to KVM cpu caps") was buggy in that clearing
> entry->edx here would wipe out all X86_FEATURE_NRIPS and X86_FEATURE_NPT.
> Only noticed it when moving this code.
>
I plan to give your v2 a spin on AMD Epyc, just in case)
--
Vitaly
Powered by blists - more mailing lists