[v5] Distinguish between variants of IBPB

[PATCH v5 0/4] Distinguish between variants of IBPB

Posted by Jim Mattson 1 month, 2 weeks ago

Prior to Zen4, AMD's IBPB did not flush the RAS (or, in Intel
terminology, the RSB). Hence, the older version of AMD's IBPB was not
equivalent to Intel's IBPB. However, KVM has been treating them as
equivalent, synthesizing Intel's CPUID.(EAX=7,ECX=0):EDX[bit 26] on any
platform that supports the synthetic features X86_FEATURE_IBPB and
X86_FEATURE_IBRS.

Equivalence also requires a previously ignored feature on the AMD side,
CPUID Fn8000_0008_EBX[IBPB_RET], which is enumerated on Zen4.

v5: Restored the first commit, which was unintentionally dropped in v4.
    Added Tom Lendacky's and Thomas Gleixner's Reviewed-by to the two
    commits that have not changed since v3.

v4: Added "guaranteed" to X86_FEATURE_IBPB comment [Pawan]
    Changed logic for deducing AMD IBPB features from Intel IBPB features
    in kvm_set_cpu_caps [Tom]
    Intel CPUs that suffer from PBRSB can't claim AMD_IBPB_RET [myself]

v3: Pass through IBPB_RET from hardware to userspace. [Tom]
    Derive AMD_IBPB from X86_FEATURE_SPEC_CTRL rather than
    X86_FEATURE_IBPB. [Tom]
    Clarify semantics of X86_FEATURE_IBPB.

v2: Use IBPB_RET to identify semantic equality. [Venkatesh]


Jim Mattson (4):
  x86/cpufeatures: Clarify semantics of X86_FEATURE_IBPB
  x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET
  KVM: x86: Advertise AMD_IBPB_RET to userspace
  KVM: x86: AMD's IBPB is not equivalent to Intel's IBPB

 arch/x86/include/asm/cpufeatures.h |  3 ++-
 arch/x86/kvm/cpuid.c               | 12 +++++++++---
 2 files changed, 11 insertions(+), 4 deletions(-)

-- 
2.47.0.rc1.288.g06298d1525-goog

Re: [PATCH v5 0/4] Distinguish between variants of IBPB

Posted by Sean Christopherson 3 weeks, 5 days ago

On Fri, 11 Oct 2024 14:43:49 -0700, Jim Mattson wrote:
> Prior to Zen4, AMD's IBPB did not flush the RAS (or, in Intel
> terminology, the RSB). Hence, the older version of AMD's IBPB was not
> equivalent to Intel's IBPB. However, KVM has been treating them as
> equivalent, synthesizing Intel's CPUID.(EAX=7,ECX=0):EDX[bit 26] on any
> platform that supports the synthetic features X86_FEATURE_IBPB and
> X86_FEATURE_IBRS.
> 
> [...]

Applied to kvm-x86 misc, thanks!

[1/4] x86/cpufeatures: Clarify semantics of X86_FEATURE_IBPB
      https://github.com/kvm-x86/linux/commit/43801a0dbb38
[2/4] x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET
      https://github.com/kvm-x86/linux/commit/99d252e3ae3e
[3/4] KVM: x86: Advertise AMD_IBPB_RET to userspace
      https://github.com/kvm-x86/linux/commit/df9328b0ef66
[4/4] KVM: x86: AMD's IBPB is not equivalent to Intel's IBPB
      https://github.com/kvm-x86/linux/commit/d66e266427e4

--
https://github.com/kvm-x86/linux/tree/next

Re: [PATCH v5 0/4] Distinguish between variants of IBPB

Posted by Borislav Petkov 3 weeks, 4 days ago

On Thu, Oct 31, 2024 at 12:51:33PM -0700, Sean Christopherson wrote:
> [1/4] x86/cpufeatures: Clarify semantics of X86_FEATURE_IBPB
>       https://github.com/kvm-x86/linux/commit/43801a0dbb38
> [2/4] x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET
>       https://github.com/kvm-x86/linux/commit/99d252e3ae3e

ff898623af2e ("x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET")

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH v5 0/4] Distinguish between variants of IBPB

Posted by Sean Christopherson 3 weeks, 4 days ago

On Fri, Nov 01, 2024, Borislav Petkov wrote:
> On Thu, Oct 31, 2024 at 12:51:33PM -0700, Sean Christopherson wrote:
> > [1/4] x86/cpufeatures: Clarify semantics of X86_FEATURE_IBPB
> >       https://github.com/kvm-x86/linux/commit/43801a0dbb38
> > [2/4] x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET
> >       https://github.com/kvm-x86/linux/commit/99d252e3ae3e
> 
> ff898623af2e ("x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET")

Doh.  I think I'll rebase the entire misc branch onto rc5, realistically the only
downside is having to send updates for all of the hashes.

Thanks for catching this!

Re: [PATCH v5 0/4] Distinguish between variants of IBPB

Posted by Sean Christopherson 3 weeks, 4 days ago

On Fri, Nov 01, 2024, Sean Christopherson wrote:
> On Fri, Nov 01, 2024, Borislav Petkov wrote:
> > On Thu, Oct 31, 2024 at 12:51:33PM -0700, Sean Christopherson wrote:
> > > [1/4] x86/cpufeatures: Clarify semantics of X86_FEATURE_IBPB
> > >       https://github.com/kvm-x86/linux/commit/43801a0dbb38
> > > [2/4] x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET
> > >       https://github.com/kvm-x86/linux/commit/99d252e3ae3e
> > 
> > ff898623af2e ("x86/cpufeatures: Define X86_FEATURE_AMD_IBPB_RET")
> 
> Doh.  I think I'll rebase the entire misc branch onto rc5, realistically the only
> downside is having to send updates for all of the hashes.
> 
> Thanks for catching this!

New hashes after the rebase:

[3/4] KVM: x86: Advertise AMD_IBPB_RET to userspace
      https://github.com/kvm-x86/linux/commit/71dd5d5300d2
[4/4] KVM: x86: AMD's IBPB is not equivalent to Intel's IBPB
      https://github.com/kvm-x86/linux/commit/de572491a975