jitterentropy-hashtime in 3.0.2 about jitterentropy-library

[joshuaehill]

The jitterentropy-hashtime function is outputs raw data (i.e., prior to hashing), so the setting of osr (which establishes how many raw samples are fed into the hash per hash output) doesn't directly affect the data being output from jitterentropy-hashtime.

When JENT_CONF_DISABLE_LOOP_SHUFFLE is defined, the first and second column of outputs from the jitterentropy-hashtime have basically the same behavior (other than the second column is sampled later).

Which pre-3.0.2 behavior are you trying to replicate?

[dtudor-dt]

I just noticed other issues that were discussing this change, so my mistake.

So, the pre-3.0.2 behavior was that one column had raw data that was imitating the real behavior of the rngd process (the var case) and one column that had only the lower bound (the single case). Now that difference is removed, right? The rngd behavior is the minimum, lower bound case.

[joshuaehill]

By default, the pseudorandom variation produced by the jent_loop_shuffle function is no longer used in any mode (either testing, or while operating normally). As a consequence, the first and second columns produced by jitterentropy-hashtime should mostly show the same behavior.

The pseudorandom variation induced by jent_loop_shuffle can be turned back on by removing the JENT_CONF_DISABLE_LOOP_SHUFFLE macro prior to compiling the library.

[dtudor-dt]

That cleared things up, thanks a lot.

Would you recommend keeping the JENT_CONF_DISABLE_LOOP_SHUFFLE macro is the end goal is to pass an SP800-90B validation?

[joshuaehill]

Yes. Here is a presentation that I gave on this topic.

[dtudor-dt]

Great, thank you.