Re: [PATCH] KVM: x86/mmu: Add capability to zap only sptes for the affected memslot

From: Alex Williamson <alex.williamson@redhat.com>
To: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Vitaly Kuznetsov <vkuznets@redhat.com>,
	Wanpeng Li <wanpengli@tencent.com>,
	Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	Xiong Zhang <xiong.y.zhang@intel.com>,
	Wayne Boyer <wayne.boyer@intel.com>,
	Zhenyu Wang <zhenyuw@linux.intel.com>,
	Jun Nakajima <jun.nakajima@intel.com>,
	Weijiang Yang <weijiang.yang@intel.com>
Subject: Re: [PATCH] KVM: x86/mmu: Add capability to zap only sptes for the affected memslot
Date: Tue, 21 Jul 2020 10:00:36 -0600	[thread overview]
Message-ID: <20200721100036.464d4440@w520.home> (raw)
In-Reply-To: <20200721030319.GD20375@linux.intel.com>

On Mon, 20 Jul 2020 20:03:19 -0700
Sean Christopherson <sean.j.christopherson@intel.com> wrote:

> +Weijiang
> 
> On Mon, Jul 13, 2020 at 12:06:50PM -0700, Sean Christopherson wrote:
> > The only ideas I have going forward are to:
> > 
> >   a) Reproduce the bug outside of your environment and find a resource that
> >      can go through the painful bisection.  
> 
> We're trying to reproduce the original issue in the hopes of biesecting, but
> have not yet discovered the secret sauce.  A few questions:
> 
>   - Are there any known hardware requirements, e.g. specific flavor of GPU?

I'm using an old GeForce GT635, I don't think there's anything special
about this card.

>   - What's the average time to failure when running FurMark/PassMark?  E.g.
>     what's a reasonable time to wait before rebooting to rerun the tests (I
>     assume this is what you meant when you said you sometimes needed to
>     reboot to observe failure).

The failure mode ranges from graphics glitches, ex. vectors drawn
across a window during the test or stray lines when interacting with
the Windows menu button, to graphics driver failures triggering an
error dialog, usually from PassMark.  I usually start FurMark, run the
stress test for ~10s, kill it, then run a PassMark benchmark.  If I
don't observe any glitching during the run, I'll trigger the Windows
menu a few times, then reboot and try again.  The graphics tests within
PassMark are generally when the glitches are triggered, both 2D and 3D,
sometimes it's sufficient to only run those tests rather than the full
system benchmark.  That's largely the trouble with this bisect is that
the test is very interactive and requires observation.  Sometimes when
it fails it snowballs into worse and worse errors and there's high
confidence that it's bad, but other times you'll be suspicious that
something occurred and need to repeat the testing.  Thanks,

Alex