Hey TDK, thanks for the reply. So there's a chance I may have misunderstood but let me put the full quote from the reference manual just to provide more context for why they made me think it was possible to obtain the raw data (perhaps with a special firmware).
From RM0433:
"""
RM0433 Page 1272
34.6 RNG entropy source validation
34.6.1 Introduction
In order to assess the amount of entropy available from the RNG, STMicroelectronics has
tested the peripheral using German BSI AIS-31 statistical tests (T0 to T8). The results can
be provided on demand or the customer can reproduce the tests.
34.6.2
Validation conditions
STMicroelectronics has tested the RNG true random number generator in the following
conditions:
RNG clock rng_clk= 48 MHz (CED bit = ’0’ in RNG_CR register) and rng_clk = 400 kHz
(CED bit = ‘1’ in RNG_CR register).
RM0433 Rev 7RM0433
34.6.3
True random number generator (RNG)
Data collection
In order to run statistical tests it is required to collect samples from the entropy source at raw
data level as well as at the output of the entropy source.
Contact STMicroelectronics if above samples need to be retrieved for your product.
"""
You will notice they say "the customer can reproduce the tests" and "In order to run statistical tests it is required to collect samples from the entropy source at raw
data level as well as at the output of the entropy source. Contact STMicroelectronics if above samples need to be retrieved for your product."
Regarding the note about STM being able to supply the firmware in AN4320, it was unclear to me whether this firmware would supply the raw entropy data, however upon further reading of AN4320 it does appear they are running the tests on the data after the conditioning engine has processed it (which is what one would want to do), so let's exclude that from further discussion, I apologize for misinterpreting it.
I took a brief look at
https://www.bsi.bund.de/SharedDocs/Downloads/DE/BSI/Zertifizierung/Interpretationen/AIS_31_Functionality_classes_for_random_number_generators_e.pdf
Source code available here
https://github.com/mjosaarinen/ais31-testsuite-v1.0
I don't fully understand the T0 through T8 tests but based on the document it does seem like one would want to feed preconditioned data into the tests to verify the characteristics of the raw entropy source are close to ideal. This seems to match with the statement from STM "from the entropy source at rawdata level as well as at the output of the entropy source".
I guess the only remaining comment I'de have is they are not clear whether after one "Contact STMicroelectronics if above samples need to be retrieved for your product." if they are going to provide a tool for one to extract the samples ones-self, or if one is going to need STM to physically extract them for the customer. I can't imagine it's the later, but perhaps it is.
