Re: [PATCH v3 3/3] tests/qtest: Introduce tests for AMD/Xilinx Versal TRNG device

[Peter Maydell]

On Wed, 4 Oct 2023 at 15:27, Tong Ho <tong.ho@amd.com> wrote:
>
> Signed-off-by: Tong Ho <tong.ho@amd.com>
> ---
>  tests/qtest/meson.build             |   2 +-
>  tests/qtest/xlnx-versal-trng-test.c | 490 ++++++++++++++++++++++++++++
>  2 files changed, 491 insertions(+), 1 deletion(-)
>  create mode 100644 tests/qtest/xlnx-versal-trng-test.c
>



> +static void trng_test_autogen(void)
> +{
> +    const size_t cnt = 512 / 32;
> +    uint32_t rng[cnt], prng[cnt];
> +    size_t n;
> +
> +    trng_reset();
> +
> +    /* PRNG run #1 */
> +    trng_reseed(prng_seed);
> +    trng_generate(true);
> +
> +    n = trng_collect(prng, cnt);
> +    if (n != cnt) {
> +        FAILED("PRNG_1 Auto-gen test failed: expected = %u, got = %u",
> +               (unsigned)cnt, (unsigned)n);
> +    }
> +
> +    /* TRNG, should not match PRNG */
> +    trng_reseed(NULL);
> +    trng_generate(true);
> +
> +    n = trng_collect(rng, cnt);
> +    if (n != cnt) {
> +        FAILED("TRNG Auto-gen test failed: expected = %u, got = %u",
> +               (unsigned)cnt, (unsigned)n);
> +    }
> +
> +    if (!memcmp(rng, prng, sizeof(rng))) {
> +        FAILED("TRNG test failed: matching PRNG");
> +    }

Why can't the TRNG output ever match the PRNG output ?
It's random numbers, so you might get (un)lucky.
This sort of check results in flaky CI, because very very
occasionally the RNG will give you output that results in
a test failure.

> +
> +    /* PRNG #2: should matches run #1 */
> +    trng_reseed(prng_seed);
> +    trng_generate(true);
> +
> +    n = trng_collect(rng, cnt);
> +    if (n != cnt) {
> +        FAILED("PRNG_2 Auto-gen test failed: expected = %u, got = %u",
> +               (unsigned)cnt, (unsigned)n);
> +    }
> +
> +    if (memcmp(rng, prng, sizeof(rng))) {
> +        FAILED("PRNG_2 Auto-gen test failed: does not match PRNG_1");
> +    }
> +}
> +

thanks
-- PMM