Re: [PATCH 0/3] Apple M1 DART IOMMU driver

[Mark Kettenis]

> From: Arnd Bergmann <arnd@xxxxxxxxxx>
> Date: Fri, 26 Mar 2021 21:03:32 +0100
> 
> On Fri, Mar 26, 2021 at 6:28 PM Mark Kettenis <mark.kettenis@xxxxxxxxx> wrote:
> 
> > I haven't figured out how the bypass stuff really works.  Corellium
> > added support for it in their codebase when they added support for
> > Thunderbolt, and some of the DARTs that seem to be related to
> > Thunderbolt do indeed have a "bypass" property.  But it is unclear to
> > me how the different puzzle pieces fit together for Thunderbolt.
> 
> As a general observation, bypass mode for Thunderbolt is what enabled
> the http://thunderclap.io/ attack. This is extremely useful for debugging
> a running kernel from another machine, but it's also something that
> should never be done in a production kernel.

No kidding!  I was surprised to see the bypass support on the
Thunderbolt-related nodes.