Re: [PATCH] KVM: x86: Set BHI_NO in guest when host is not affected by BHI

[Alexandre Chartre]

On 4/11/24 09:34, Nikolay Borisov wrote:
> On 11.04.24 г. 10:24 ч., Alexandre Chartre wrote:
> > When a system is not affected by the BHI bug then KVM should
> > configure guests with BHI_NO to ensure they won't enable any
> > BHI mitigation.
> >
> > Signed-off-by: Alexandre Chartre <alexandre.chartre@xxxxxxxxxx>
> > ---
> >   arch/x86/kvm/x86.c | 3 +++
> >   1 file changed, 3 insertions(+)
> >
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index 984ea2089efc..f43d3c15a6b7 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -1678,6 +1678,9 @@ static u64 kvm_get_arch_capabilities(void)
> >       if (!boot_cpu_has_bug(X86_BUG_GDS) || gds_ucode_mitigated())
> >           data |= ARCH_CAP_GDS_NO;
> > +    if (!boot_cpu_has_bug(X86_BUG_BHI))
> > +        data |= ARCH_CAP_BHI_NO;
> > +
>
> But this is already handled since ARCH_CAP_BHI_NO is added to
> KVM_SUPPORTED_ARCH_CAP so when the host caps are read that bit is
> going to be set there, if it's set for the physical cpu of course.

Correct, if the host has ARCH_CAP_BHI_NO then it will be propagated to the
guest. But the host might not have ARCH_CAP_BHI_NO set and not be affected
by BHI.

That's the case for example of Skylake servers: they don't have ARCH_CAP_BHI_NO,
but they are not affected by BHI because they don't have eIBRS. However, a guest
will think it is affected because it doesn't know if eIBRS is present on the
system (because virtualization can hide it).

I tested on Skylake:

Without the patch, both host and guest are running 6.9.0-rc3, then BHI mitigations are:

- Host:  BHI: Not affected
- Guest: BHI: SW loop, KVM: SW loop

=> so guest enables BHI SW loop mitigation although host doesn't need mitigation.

With the patch on the host, guest still running 6.9.0-rc3, then BHI mitigations are:

- Host:  BHI: Not affected
- Guest: BHI: Not affected

=> now guest doesn't enable BHI mitigation, like the host.

Thanks,

alex.