ktp.next() for keys

[m-temp]

ktp.next() for keys

[m-temp]

backlog: With ctx.obj.ktp.fixed_key = False the internal lfsr is iterated twice per ktp.next(). This must be taken into account when using the batch-mode and thus, the lfsr on device side.

Possible solutions:

• perform a dummy iteration on the device
• use ktp.next_key() and ktp.next_text() instead of ktp.next()[0] and ktp.next()[1]

[m-temp]

For otbn we are using the second solution. See lowRISC/opentitan#17416 and #128 It's running smoothly, while the dummy iteration on the device creates a significant overhead.

[johannheyszl]

The behaviour/use of the host-side PRNG is hard to understand for reasons described here.

Latest host-side batch capture implementation: #197
Does not use ktp.next().

PRNG likely replaced in future, hence deprioritizing this issue.