lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <Zs-_G0FGBbo4tyat@google.com>
Date: Wed, 28 Aug 2024 17:21:47 -0700
From: Sean Christopherson <seanjc@...gle.com>
To: Jim Mattson <jmattson@...gle.com>
Cc: Tom Lendacky <thomas.lendacky@....com>, Thomas Gleixner <tglx@...utronix.de>, 
	Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>, 
	Dave Hansen <dave.hansen@...ux.intel.com>, "H. Peter Anvin" <hpa@...or.com>, 
	Paolo Bonzini <pbonzini@...hat.com>, Pawan Gupta <pawan.kumar.gupta@...ux.intel.com>, 
	Josh Poimboeuf <jpoimboe@...nel.org>, Sandipan Das <sandipan.das@....com>, 
	Kai Huang <kai.huang@...el.com>, x86@...nel.org, linux-kernel@...r.kernel.org, 
	kvm@...r.kernel.org, Venkatesh Srinivas <venkateshs@...omium.org>
Subject: Re: [PATCH v3 4/4] KVM: x86: AMD's IBPB is not equivalent to Intel's IBPB

On Fri, Aug 23, 2024, Jim Mattson wrote:
> On Fri, Aug 23, 2024 at 3:48 PM Jim Mattson <jmattson@...gle.com> wrote:
> >
> > On Fri, Aug 23, 2024 at 3:12 PM Tom Lendacky <thomas.lendacky@....com> wrote:
> > >
> > > On 8/23/24 15:51, Sean Christopherson wrote:
> > > > On Fri, Aug 23, 2024, Tom Lendacky wrote:
> > > >> On 8/23/24 13:53, Jim Mattson wrote:
> > > >>> From Intel's documention [1], "CPUID.(EAX=07H,ECX=0):EDX[26]
> > > >>> enumerates support for indirect branch restricted speculation (IBRS)
> > > >>> and the indirect branch predictor barrier (IBPB)." Further, from [2],
> > > >>> "Software that executed before the IBPB command cannot control the
> > > >>> predicted targets of indirect branches (4) executed after the command
> > > >>> on the same logical processor," where footnote 4 reads, "Note that
> > > >>> indirect branches include near call indirect, near jump indirect and
> > > >>> near return instructions. Because it includes near returns, it follows
> > > >>> that **RSB entries created before an IBPB command cannot control the
> > > >>> predicted targets of returns executed after the command on the same
> > > >>> logical processor.**" [emphasis mine]
> > > >>>
> > > >>> On the other hand, AMD's IBPB "may not prevent return branch
> > > >>> predictions from being specified by pre-IBPB branch targets" [3].
> > > >>>
> > > >>> However, some AMD processors have an "enhanced IBPB" [terminology
> > > >>> mine] which does clear the return address predictor. This feature is
> > > >>> enumerated by CPUID.80000008:EDX.IBPB_RET[bit 30] [4].
> > > >>>
> > > >>> Adjust the cross-vendor features enumerated by KVM_GET_SUPPORTED_CPUID
> > > >>> accordingly.
> > > >>>
> > > >>> [1] https://www.intel.com/content/www/us/en/developer/articles/technical/software-security-guidance/technical-documentation/cpuid-enumeration-and-architectural-msrs.html
> > > >>> [2] https://www.intel.com/content/www/us/en/developer/articles/technical/software-security-guidance/technical-documentation/speculative-execution-side-channel-mitigations.html#Footnotes
> > > >>> [3] https://www.amd.com/en/resources/product-security/bulletin/amd-sb-1040.html
> > > >>> [4] https://www.amd.com/content/dam/amd/en/documents/processor-tech-docs/programmer-references/24594.pdf
> > > >>>
> > > >>> Fixes: 0c54914d0c52 ("KVM: x86: use Intel speculation bugs and features as derived in generic x86 code")
> > > >>> Suggested-by: Venkatesh Srinivas <venkateshs@...omium.org>
> > > >>> Signed-off-by: Jim Mattson <jmattson@...gle.com>
> > > >>> ---
> > > >>>  arch/x86/kvm/cpuid.c | 6 +++++-
> > > >>>  1 file changed, 5 insertions(+), 1 deletion(-)
> > > >>>
> > > >>> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> > > >>> index ec7b2ca3b4d3..c8d7d928ffc7 100644
> > > >>> --- a/arch/x86/kvm/cpuid.c
> > > >>> +++ b/arch/x86/kvm/cpuid.c
> > > >>> @@ -690,7 +690,9 @@ void kvm_set_cpu_caps(void)
> > > >>>     kvm_cpu_cap_set(X86_FEATURE_TSC_ADJUST);
> > > >>>     kvm_cpu_cap_set(X86_FEATURE_ARCH_CAPABILITIES);
> > > >>>
> > > >>> -   if (boot_cpu_has(X86_FEATURE_IBPB) && boot_cpu_has(X86_FEATURE_IBRS))
> > > >>> +   if (boot_cpu_has(X86_FEATURE_AMD_IBPB_RET) &&
> > > >>> +       boot_cpu_has(X86_FEATURE_AMD_IBPB) &&
> > > >>> +       boot_cpu_has(X86_FEATURE_AMD_IBRS))
> > > >>>             kvm_cpu_cap_set(X86_FEATURE_SPEC_CTRL);
> > > >>>     if (boot_cpu_has(X86_FEATURE_STIBP))
> > > >>>             kvm_cpu_cap_set(X86_FEATURE_INTEL_STIBP);
> > > >>> @@ -759,6 +761,8 @@ void kvm_set_cpu_caps(void)
> > > >>>      * arch/x86/kernel/cpu/bugs.c is kind enough to
> > > >>>      * record that in cpufeatures so use them.
> > > >>>      */
> > > >>> +   if (boot_cpu_has(X86_FEATURE_SPEC_CTRL))
> > > >>> +           kvm_cpu_cap_set(X86_FEATURE_AMD_IBPB_RET);
> > > >>
> > > >> If SPEC_CTRL is set, then IBPB is set, so you can't have AMD_IBPB_RET
> > > >> without AMD_IBPB, but it just looks odd seeing them set with separate
> > > >> checks with no relationship dependency for AMD_IBPB_RET on AMD_IBPB.
> > > >> That's just me, though, not worth a v4 unless others feel the same.
> > > >
> > > > You thinking something like this (at the end, after the dust settles)?
> > > >
> > > >       if (WARN_ON_ONCE(kvm_cpu_cap_has(X86_FEATURE_AMD_IBPB_RET) &&
> > > >                        !kvm_cpu_cap_has(X86_FEATURE_AMD_IBPB)))
> > > >               kvm_cpu_cap_clear(X86_FEATURE_AMD_IBPB_RET);
> > >
> > > I was just thinking more along the lines of:
> > >
> > >         if (boot_cpu_has(X86_FEATURE_IBPB)) {
> > >                 kvm_cpu_cap_set(X86_FEATURE_AMD_IBPB);
> > >                 if (boot_cpu_has(X86_FEATURE_SPEC_CTRL))
> > >                         kvm_cpu_cap_set(X86_FEATURE_AMD_IBPB_RET);
> > >         }
> >
> > AFAICT, there are just two reasons that X86_FEATURE_IBPB gets set:
> > 1. The CPU reports CPUID.(EAX=7,ECX=0):EDX[bit 26] (aka X86_FEATURE_SPEC_CTRL)
> > 2. The CPU reports CPUID Fn8000_0008_EBX[IBPB] (aka X86_FEATURE_AMD_IBPB)
> >
> > Clearly, in the second case, the KVM cpu capability for AMD_IBPB will
> > already be set, since it's specified in the mask for
> > CPUID_8000_0008_EBX.
> >
> > If this block of code is just trying to populate CPUID Fn8000_0008_EBX
> > on Intel processors, I'd rather change all of the predicates to test
> > for Intel features, rather than vendor-neutral features, so that the
> > derivation is clear. But maybe this block of code is also trying to
> > populate CPUID Fn8000_0008_EBX on AMD processors that may have some of
> > these features, but don't enumerate them via CPUID?
> 
> There's another argument for just nuking these cross-vendor
> derivations. How do we factor in CVE-2022-26373 (Post-barrier Return
> Stack Buffer Predictions)?
> Intel CPUs without IA32_ARCH_CAPABILITIES.PBRSB_NO[bit 24] have a
> weaker IBPB than AMD CPUs with CPUID Fn8000_0008_EBX[IBPB_RET], and
> probably should not be enumerating that CPUID bit.
> 
> Trying to derive cross-vendor mitigation equivalence is just going to
> end in tears.

Agreed, but I also don't want to break existing setups.  Is there a bare minimum
of sorts that we can advertise to userspace?  E.g. something that might be
imperfect, but has acceptable tradeoffs/risks for the existing code?

And then put a stake in the ground saying no more of these shenanigans.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ