[onerng talk] Possible magnetic vulnerability

[Paul Campbell]

On Fri, 23 Jan 2015 09:56:17 Gerd v. Egidy wrote:
> A solution would be to allow the controller to read out the actual voltage 
> after the stepup via a voltage divider over an adc input pin. 
> 
> Probably a bigger output cap for the stepup would help to reduce the
> influence  effects for the short time till you detect them on the adc.

dropping the voltage by saturating the inductor is likely going to stop the 
generator completely (really just reduce the size of the avalanche signal so 
that it always registers as a '1'  which in turn will cause the OneRNG to 
register a 'stuck at' fault and stop generating data (the led will flash an 
error code).

This is another reason to run both RNGs and use the whitener to smear their 
outputs into the entropy pool.

However even if someone manages to reduce the amount of entropy being 
generated by a source that lower entropy will continue to accumulate into the 
onboard pool  (we mix in all new entropy as it arrives). An attack that 
reduces the amount of entropy the system generates is only an issue if it then 
generates entropy at a rate lower than than it is consumed 

	Paul