Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 21 Apr 2021 22:03:17 +0200
From:   Corentin Labbe <clabbe.montjoie@gmail.com>
To:     herbert@gondor.apana.org.au, linux-crypto@vger.kernel.org
Cc:     linus.walleij@linaro.org, linux-kernel@vger.kernel.org
Subject: cortina/gemini: hwrng: what is its quality ?
Message-ID: <YICFBQ8mQRJ4zSh9@Red>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Precedence: bulk

hello

I work on the crypto part of the cortina/gemini SL3516 SoC.
The datasheet mention a HWRNG in its IP but really briefly:
"""
The implementation is a 32-bit Hardware Random Number Generator that has a uniformed
distribution between 0 and 2^32 -1. The hardware randomness is created by sampling data from
different clock domains, and feeding it as input to the 32-bit maximum length LFSR (Linear Feedback
Shift Register)
"""

Piping its output to rngtest give:
dd if=/dev/hwrng count=2000 bs=2048 | rngtest
rngtest 6.11
rngtest: starting FIPS tests...
rngtest: entropy source drained
rngtest: bits received from input: 32768000
rngtest: FIPS 140-2 successes: 1191
rngtest: FIPS 140-2 failures: 447
rngtest: FIPS 140-2(2001-10-10) Monobit: 183
rngtest: FIPS 140-2(2001-10-10) Poker: 116
rngtest: FIPS 140-2(2001-10-10) Runs: 346
2000+0 records in
2000+0 records out
rngtest: FIPS 140-2(2001-10-10) Long run: 0
rngtest: FIPS 140-2(2001-10-10) Continuous run: 0
rngtest: input channel speed: (min=303.606; avg=3143.352; max=9712.208)Kibits/s
rngtest: FIPS tests speed: (min=7.104; avg=10.332; max=10.638)Mibits/s
rngtest: Program run time: 13303224 microseconds

That's a quite number of failure.
Can the hwrng still be used with some "hwrng->quality" setting ?
Or it is just too many failure to be used ?

Regards