From: Sean Christopherson <sean.j.christopherson@intel.com>
To: Yang Weijiang <weijiang.yang@intel.com>
Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
pbonzini@redhat.com, jmattson@google.com,
yu.c.zhang@linux.intel.com, yu-cheng.yu@intel.com
Subject: Re: [PATCH v8 3/7] KVM: VMX: Pass through CET related MSRs
Date: Wed, 18 Dec 2019 08:02:28 -0800 [thread overview]
Message-ID: <20191218160228.GB25201@linux.intel.com> (raw)
In-Reply-To: <20191218135513.GB7926@local-michael-cet-test>
On Wed, Dec 18, 2019 at 09:55:13PM +0800, Yang Weijiang wrote:
> On Tue, Dec 17, 2019 at 04:34:55PM -0800, Sean Christopherson wrote:
> > On Mon, Dec 16, 2019 at 10:18:16AM +0800, Yang Weijiang wrote:
> > > On Tue, Dec 10, 2019 at 01:18:21PM -0800, Sean Christopherson wrote:
> > > > On Fri, Nov 01, 2019 at 04:52:18PM +0800, Yang Weijiang wrote:
> > > > > CET MSRs pass through Guest directly to enhance performance.
> > > > > CET runtime control settings are stored in MSR_IA32_{U,S}_CET,
> > > > > Shadow Stack Pointer(SSP) are stored in MSR_IA32_PL{0,1,2,3}_SSP,
> > > > > SSP table base address is stored in MSR_IA32_INT_SSP_TAB,
> > > > > these MSRs are defined in kernel and re-used here.
> > > > >
> > > > +
> > > > > static void vmx_cpuid_update(struct kvm_vcpu *vcpu)
> > > > > {
> > > > > struct vcpu_vmx *vmx = to_vmx(vcpu);
> > > > > @@ -7025,6 +7087,9 @@ static void vmx_cpuid_update(struct kvm_vcpu *vcpu)
> > > > > if (boot_cpu_has(X86_FEATURE_INTEL_PT) &&
> > > > > guest_cpuid_has(vcpu, X86_FEATURE_INTEL_PT))
> > > > > update_intel_pt_cfg(vcpu);
> > > > > +
> > > > > + if (!is_guest_mode(vcpu))
> > > > > + vmx_pass_cet_msrs(vcpu);
> > > >
> > > > Hmm, this looks insufficent, e.g. deliberately toggling CET from on->off
> > > > while in guest mode would put KVM in a weird state as the msr bitmap for
> > > > L1 would still allow L1 to access the CET MSRs.
> > > >
> > > Hi, Sean,
> > > I don't get you, there's guest mode check before access CET msrs, it'll
> > > fail if it's in guest mode.
> >
> > KVM can exit to userspae while L2 is active. If userspace then did a
> > KVM_SET_CPUID2, e.g. instead of KVM_RUN, vmx_cpuid_update() would skip
> > vmx_pass_cet_msrs() and KVM would never update L1's MSR bitmaps.
> >
> Thanks, it makes sense to me. Given current implementation, how about
> removing above check and adding it in CET CPUID
> enumeration for L2 so that no CET msrs passed through to L2 when
> is_guest_mode() is true?
But you still need to update L1's MSR bitmaps. That can obviously be done
all at once, but it's annoying and IMO unnecessarily complex.
> > > > Allowing KVM_SET_CPUID{2} while running a nested guest seems bogus, can we
> > > > kill that path entirely with -EINVAL?
> > > >
> > > Do you mean don't expose CET cpuids to L2 guest?
> >
> > I mean completely disallow KVM_SET_CPUID and KVM_SET_CPUID2 if
> > is_guest_mode() is true. My question is mostly directed at Paolo and
> > anyone else that has an opinion on whether we can massage the ABI to
> > retroactively change KVM_SET_CPUID{2} behavior.
>
> This sounds like something deserving an individual patch after get
> agreement in community. I'll put it aside right now.
Normally I would agree, but I think in this case it would significantly
reduce the complexity and implementation cost of CET support. I'll send a
patch to kickstart the conversation, it's a tiny change in terms of code.
next prev parent reply other threads:[~2019-12-18 16:02 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-11-01 8:52 [PATCH v8 0/7] Introduce support for guest CET feature Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 1/7] KVM: CPUID: Fix IA32_XSS support in CPUID(0xd,i) enumeration Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 2/7] KVM: VMX: Define CET VMCS fields and #CP flag Yang Weijiang
2019-12-10 21:00 ` Sean Christopherson
2019-12-11 1:45 ` Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 3/7] KVM: VMX: Pass through CET related MSRs Yang Weijiang
2019-12-10 21:18 ` Sean Christopherson
2019-12-11 1:32 ` Yang Weijiang
2019-12-11 1:50 ` Sean Christopherson
2019-12-11 2:27 ` Yang Weijiang
2019-12-16 2:18 ` Yang Weijiang
2019-12-18 0:34 ` Sean Christopherson
2019-12-18 13:55 ` Yang Weijiang
2019-12-18 16:02 ` Sean Christopherson [this message]
2019-11-01 8:52 ` [PATCH v8 4/7] KVM: VMX: Load CET states on vmentry/vmexit Yang Weijiang
2019-12-10 21:23 ` Sean Christopherson
2019-12-11 1:54 ` Yang Weijiang
2019-12-11 16:35 ` Sean Christopherson
2019-12-12 1:04 ` Yang Weijiang
2019-12-18 0:30 ` Sean Christopherson
2019-12-18 13:20 ` Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 5/7] KVM: X86: Enable CET bits update in IA32_XSS Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 6/7] KVM: X86: Load guest fpu state when accessing MSRs managed by XSAVES Yang Weijiang
2019-12-10 21:27 ` Sean Christopherson
2019-12-11 2:03 ` Yang Weijiang
2019-11-01 8:52 ` [PATCH v8 7/7] KVM: X86: Add user-space access interface for CET MSRs Yang Weijiang
2019-12-10 21:58 ` Sean Christopherson
2019-12-11 2:19 ` Yang Weijiang
2019-12-11 16:27 ` Sean Christopherson
2019-12-12 0:42 ` Yang Weijiang
2019-12-12 16:03 ` [PATCH v8 0/7] Introduce support for guest CET feature Konrad Rzeszutek Wilk
2019-12-13 0:44 ` Yang Weijiang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20191218160228.GB25201@linux.intel.com \
--to=sean.j.christopherson@intel.com \
--cc=jmattson@google.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=weijiang.yang@intel.com \
--cc=yu-cheng.yu@intel.com \
--cc=yu.c.zhang@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).