We are using the imx6ul processor in an application where we need to obtain certification for random number generation sources. For this, we capture ~128MB of random data from /dev/random, /dev/urandom and /dev/hwrng interfaces and feed it to NIST statistical tests.
According to the Final Analysis Reports, "Overlapping (Periodic) Template Matching Test" scores below the minimum pass rate in all the above interfaces.
- Why does this particular test fail on all interfaces?
- What are the implications of this result on the certification?
- Will some post-processing such as whitening help fix this?
- Is there any specific configuration for the CAAM RNG that we should use? We are using yocto linux-3.14.38 without any modifications.
PS:
imx6ul SRM says the RNG should be instantiated in test mode to obtain data for certification. Even in test mode the above test fails. Below are extracts from the test reports for /dev/hwrng:
In normal mode:
------------------------------------------------------------------------------
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 P-VALUE PROPORTION STATISTICAL TEST
------------------------------------------------------------------------------
6 2 1 0 0 1 0 0 0 0 0.000199 6/10 * OverlappingTemplate
In test mode:
------------------------------------------------------------------------------
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 P-VALUE PROPORTION STATISTICAL TEST
------------------------------------------------------------------------------
7 1 1 1 0 0 0 0 0 0 0.000003 * 5/10 * OverlappingTemplate
The minimum pass rate for this test is 8 for a sample size of 10 binary sequences.
