[Cryptech Tech] arm

[Bernd Paysan]

Am Samstag, 10. Januar 2015, 09:14:42 schrieb Randy Bush:
> we think we want
>   o no or minimal magic blobs because it's inside the security boundary
>   o support for booting, flash file system, and usb
>   o some speed, but the crypto is done in the fpga
>   o rtos components without rms virus

What I'm using on embedded ARM Cortex Mx right now is Mecrist Stellaris.  It's 
a native code Forth system, and as such, it is both sufficiently fast and 
sufficiently simple to be audited (a lot simpler than most of the 
competition), and it contains some simple RTOS components.

Of course, it's GPLv3, so it doesn't fit the "rms virus-free" requirement, 
which I do object from the wording: this is Microsoft propaganda.  The 
transitive freedom of GPL is not malicious, though it can get in the way with 
certain business models ;-).  Microsoft's business model of "embrace and 
extend", to be precise.

I can understand why hardware description and GPL don't work well together, 
because once the thing is real hardware (instead of an FPGA), nobody can 
modify it any longer, so the transitive freedoms of the GPL make little sense.

However, this part is firmware.  This is where you, as user, absolutely *want* 
to be sure that nobody inserted a backdoor and then locked you out.  That's 
the usual thing what happens with free software in the embedded field, as long 
as it's not GPLv3: Somebody takes that software, adds a bit here or there, and 
then locks out the end user.

For a HSM, I would rather make 100% sure that this can't happen.  Therefore, 
for the firmware, I would rather like to see the transitive freedom, and 
especially *not* a TPM, locked down by the manufacturer, which, for the end 
user, is another problem rather than a solution.

-- 
Bernd Paysan
"If you want it done right, you have to do it yourself"
http://bernd-paysan.de/
net2o ID: kQusJzA;7*?t=uy at X}1GWr!+0qqp_Cn176t4(dQ*