netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Simon Horman <horms@kernel.org>
To: Markus Theil <theil.markus@gmail.com>
Cc: linux-crypto@vger.kernel.org, davem@davemloft.net,
	netdev@vger.kernel.org, akpm@linux-foundation.org,
	Jason@zx2c4.com
Subject: Re: [PATCH 2/2] prandom/random32: switch to Xoshiro256++
Date: Mon, 17 Feb 2025 11:18:06 +0000	[thread overview]
Message-ID: <20250217111806.GJ1615191@kernel.org> (raw)
In-Reply-To: <20250214081840.47229-3-theil.markus@gmail.com>

On Fri, Feb 14, 2025 at 09:18:40AM +0100, Markus Theil wrote:
> The current Linux PRNG is based on LFSR113, which means:
> - needs some warmup rounds to yield better statistical properties
> - seeds/initial states must be of certain structure
> - does not pass L’Ecuyer's BigCrush in TestU01
> 
> While of course, there is no clear "best" PRNG, replace with
> Xoshiro256++, which seams to be a sensible replacement, from
> todays point of view:
> - only needs one bit set to 1 in the seed, needs no warmup, when
>   seeded with splitmix64.
> - Also has statistical evaluation, like LFSR113.
> - Passes BigCrush in TestU01.
> 
> The code got smaller, because some edge cases are ruled out now.
> I kept the test vectors and adapted them to this RNG.
> 
> Signed-off-by: Markus Theil <theil.markus@gmail.com>

...

> diff --git a/lib/random32.c b/lib/random32.c

...

> +/**
> + * prandom_seed_state - set seed for prandom_u32_state().
> + * @state: pointer to state structure to receive the seed.
> + * @seed: arbitrary 64-bit value to use as a seed.
> + *
> + * splitmix64 init as suggested for xoshiro256++
> + * See: https://prng.di.unimi.it/splitmix64.c
> + */
> +void prandom_seed_state(struct rnd_state *state, u64 seed)
>  {
> -	/* Calling RNG ten times to satisfy recurrence condition */
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> -	prandom_u32_state(state);
> +	int i;
> +
> +	for (i = 0; i < ARRAY_SIZE(state->s); ++i) {
> +		seed += 0x9e3779b97f4a7c15;
> +		u64 z = seed;
> +		z = (z ^ (z >> 30)) * 0xbf58476d1ce4e5b9;
> +		z = (z ^ (z >> 27)) * 0x94d049bb133111eb;
> +        state->s[i] = z ^ (z >> 31);

nit: The indentation seems off here.

> +	}
>  }
> +EXPORT_SYMBOL(prandom_seed_state);

...

  reply	other threads:[~2025-02-17 11:18 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-02-14  8:18 [PATCH 0/2] prandom: add crypto warnings and switch to new PRNG Markus Theil
2025-02-14  8:18 ` [PATCH 1/2] prandom: add usage comments for cryptography Markus Theil
2025-02-14  8:18 ` [PATCH 2/2] prandom/random32: switch to Xoshiro256++ Markus Theil
2025-02-17 11:18   ` Simon Horman [this message]
2025-03-22 18:24     ` Markus Theil
2025-03-24 17:32       ` Simon Horman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250217111806.GJ1615191@kernel.org \
    --to=horms@kernel.org \
    --cc=Jason@zx2c4.com \
    --cc=akpm@linux-foundation.org \
    --cc=davem@davemloft.net \
    --cc=linux-crypto@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=theil.markus@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).