Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 4 Jul 2018 20:39:50 +0200 (CEST)
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Paolo Bonzini <pbonzini@redhat.com>
cc:     Jingqi Liu <jingqi.liu@intel.com>, rkrcmar@redhat.com,
        mingo@redhat.com, hpa@zytor.com, x86@kernel.org,
        linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        wei.w.wang@intel.com, Robert Hoo <robert.hu@linux.intel.com>
Subject: Re: [PATCH v2] KVM: Expose the split lock detection feature to guest
 VM
In-Reply-To: <a83b9b41-afbb-016d-528b-029f1b6228ae@redhat.com>
Message-ID: <alpine.DEB.2.21.1807042035340.12092@nanos.tec.linutronix.de>
References: <1530709593-87702-1-git-send-email-jingqi.liu@intel.com> <7813a04b-8058-538b-8c9b-cdd01ce119ac@redhat.com> <alpine.DEB.2.21.1807041636340.1752@nanos.tec.linutronix.de> <a83b9b41-afbb-016d-528b-029f1b6228ae@redhat.com>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, 4 Jul 2018, Paolo Bonzini wrote:
> On 04/07/2018 16:51, Thomas Gleixner wrote:
> > There is no rush for this to be in KVM/QEMU now because all what exists for
> > this new split lock thing is 'silicon' running on an emulator. And w/o
> > support in the kernel proper this is completely useless.
> 
> That's good.  I assumed it was IceLake, in which case the feature would
> block the definition of a standard IceLake CPU model in QEMU.
> 
> > So this needs the following things:
> > 
> >   1) Proper enumeration via CPUID or MISC_FEATURES. The current detection
> >      hack is just broken.
> 
> Yes please.
> 
> >   2) A proper host side implementation, which then automatically makes the
> >      stuff usable in a guest once it is exposed.
> 
> If the CPUID bit or MISC_FEATURES is added, you don't even need the host
> side for the guests to use it.  It's only needed now because of the ugly
> MSR-based detection.
> 
> >   3) A proper way how to expose MSR_TEST_CTL to the guest, but surely not
> >      with extra split_lock_ctrl voodoo. It's an MSR nothing else. KVM/QEMU
> >      have standartized ways to deal with MSRs and the required selective
> >      bitwise access control.
> 
> That part is pretty much standard, I'm not worried about it.  We have
> one variable in struct kvm_vcpu_arch for each MSR (or set of MSRs) that
> we expose, so that's the split_lock_ctrl voodoo. :)
> 
> Once the detection is sorted out, KVM is easy.

That's what I thought and even if it was IceLeak then they still can flip a
CPUID/MISC FEATURE bit in their binary BIOS blob or ucode. We really need
to push back hard on these half baken features which need voodoo
programming to detect.

Thanks,

	tglx