On Wed, Sep 25, 2019 at 7:30 PM Xiaoyao Li <xiaoyao.li@xxxxxxxxx> wrote:
>
> On 9/26/2019 8:04 AM, Jim Mattson wrote:
> > When the guest CPUID information represents an AMD vCPU, return all
> > zeroes for queries of undefined CPUID leaves, whether or not they are
> > in range.
> >
> > Signed-off-by: Jim Mattson <jmattson@xxxxxxxxxx>
> > Fixes: bd22f5cfcfe8f6 ("KVM: move and fix substitue search for missing CPUID entries")
> > Reviewed-by: Marc Orr <marcorr@xxxxxxxxxx>
> > Reviewed-by: Peter Shier <pshier@xxxxxxxxxx>
> > Reviewed-by: Jacob Xu <jacobhxu@xxxxxxxxxx>
> > Cc: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
> > Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> > ---
> > arch/x86/kvm/cpuid.c | 6 ++++--
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> > index 35e2f930a4b79..0377d2820a7aa 100644
> > --- a/arch/x86/kvm/cpuid.c
> > +++ b/arch/x86/kvm/cpuid.c
> > @@ -988,9 +988,11 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
> > /*
> > * Intel CPUID semantics treats any query for an out-of-range
> > * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were
> > - * requested.
> > + * requested. AMD CPUID semantics returns all zeroes for any
> > + * undefined leaf, whether or not the leaf is in range.
> > */
> > - if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
> > + if (!entry && check_limit && !guest_cpuid_is_amd(vcpu) &&
> > + !cpuid_function_in_range(vcpu, function)) {
>
> IIUC, the parameter check_limit is to indicate whether return highest
> basic leaf when out-of-range. Here you just makes check_limit meaningless.
That's right. For AMD CPUID semantics, there is no need for check_limit.
> Maybe we can do like this to use check_limit reasonably：
>
> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index 0377d2820a7a..e6a61f3f6c0c 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -1035,7 +1035,8 @@ int kvm_emulate_cpuid(struct kvm_vcpu *vcpu)
>
> eax = kvm_rax_read(vcpu);
> ecx = kvm_rcx_read(vcpu);
> - kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx, true);
> + kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx,
> + guest_cpuid_is_amd(vcpu) ? false: true);
> kvm_rax_write(vcpu, eax);
> kvm_rbx_write(vcpu, ebx);
> kvm_rcx_write(vcpu, ecx);
>
> > max = kvm_find_cpuid_entry(vcpu, 0, 0);
> > if (max) {
> > function = max->eax;
Since over-limit CPUID queries should be rare, it seems unfortunate to
pay the cost of guest_cpuid_is_amd() for every emulated CPUID
instruction.
