On Tue, Nov 22, 2022 at 3:39 PM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
On Tue, Nov 22, 2022 at 4:32 PM Pedro Falcato <pedro.falcato@gmail.com> wrote:
> +  // Testing algorithm inspired by linux's arch/x86/kernel/cpu/rdrand.c:x86_init_rdrand
> +  // as relicensed by the author, Jason Donenfeld, in the EDK2 mailing list.
> +  // As is, the algorithm samples rdrand $RDRAND_TEST_SAMPLES times and expects
> +  // a different result $RDRAND_MIN_CHANGE times for reliable RDRAND usage.

You don't need to pepper my name all over the source. :)

I just wanted to properly credit you :) If you're not okay with it I can remove it in a v3. 

> +  for (TestIteration = 0; TestIteration < RDRAND_TEST_SAMPLES; TestIteration++) {
> +    UINT32  Sample;
> +    //
> +    // Note: We use a retry loop for rdrand. Normal users get this in BaseRng.c
> +    // Any failure to get a random number will assume RDRAND does not work.
> +    //
> +    for (Idx = 0; Idx < RDRAND_RETRIES; Idx++) {
> +      if (AsmRdRand32 (&Sample)) {

The linux code will use a 64bit value on 64bit machines. I suggest you
do the same here -- use native word size. I think EFI calls this a
"UINTN".
 
Hmm, do you reckon it makes a difference? I'm not intimately familiar with HWRNG internals. Unfortunately there's no AsmRdRandUintn
so this would take some per-bitness #define's which... yeah, I'd rather not.

Pedro