Re: [RFC][PATCH 0/4] x86/entry: disallow #DB more

From: Peter Zijlstra
Date: Mon May 25 2020 - 14:08:41 EST

On Mon, May 25, 2020 at 10:19:08AM -0700, Andy Lutomirski wrote:

> How about adding it to cpu_tlbstate? A lot of NMIs are going to read
> that anyway to check CR3.

That might work I suppose; we're really pushing the name of it though.
Also, that's PTI specific IIRC, and we're getting to the point where a
significant number of CPUs no longer need that, right?

> And blaming KVM is a bit misplaced. This isnât KVMâs fault â itâs
> Intelâs. VT-x has two modes: DR access exits and DR access doesnât
> exit. Thereâs no shadow mode.

It's virt, I can't be arsed to care, whoever misdesigned it.
We already have debugreg pvops, they can do shadow there.