[Arm-netbook] bunnie about riscv

[Neil Jansen]

On Sun, Jun 11, 2017 at 4:38 PM, <ronwirring at safe-mail.net> wrote:
> Thank you for the information.
> I have watched a rutkowska video on how complicated
> intel's management features are.
> Difficult.

That's why I'm here, lol.  The Intel stuff is getting bad enough that it
has me wondering what I can do for the open source hardware world.  Moving
to ARM via EOMA68 is a good near-term solution, but even that's not going
to be 100% trustable at lest by bunnie's standards in the lecture.
Something like RISC-V has the potential to get there, but as he pointed
out, even that's not completely open.  I think right now the important
thing is to just be an early adopter of this stuff to show that the
market's there.  bunnie broke the demographics down pretty well, there's
definitely money to be made.  Back to Intel though.  It makes me want to
jump on eBay and pick up some older vintage Intel CPU's that didn't have
the management features, but obviously there's no way to know if those
aren't blown wide open by other means.  Man, very interesting times we live
in.


> Remarkable that you cannot do a verification using a microscope.

You can do exactly this, and it'll get you to maybe 99% of the way there.
Companies like ChipWorks do exactly this for money.  Others do it for hobby
(see: http://www.visual6502.org/, http://siliconpr0n.org/,
https://zeptobars.com/en/, http://www.righto.com/).  It can often get great
results.  bunnie was playing devils advocate by saying even if you did
this, there are still things that can be present but in an obfuscated
manner, that could be malicious or careless.  This doesn't really mean to
throw the baby out with the bathwater.  Having a reverse engineered CPU
with a small possibility of shenanigans is still better than having a 100%
proprietary CPU or a 50% proprietary CPU.  Security through obscurity and
all that.


> We should have libre software hdds and ram.

Can you elaborate on that a bit?  I don't understand what you mean.