linux-kernel - Re: [PATCH] random: vary jitter iterations based on cycle counter speed

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 13 Jul 2022 16:53:36 +0200
From:   "Jason A. Donenfeld" <Jason@...c4.com>
To:     Vladimir Murzin <vladimir.murzin@....com>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Linux Crypto Mailing List <linux-crypto@...r.kernel.org>,
        Eric Biggers <ebiggers@...gle.com>,
        "Theodore Ts'o" <tytso@....edu>,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH] random: vary jitter iterations based on cycle counter speed

Hi Vladimir,

On Wed, Jul 13, 2022 at 4:52 PM Vladimir Murzin <vladimir.murzin@....com> wrote:
>
> Hi Jason,
>
> On 7/13/22 15:40, Jason A. Donenfeld wrote:
> > Hi Vladimir,
> >
> > On Wed, Jul 13, 2022 at 03:31:05PM +0100, Vladimir Murzin wrote:
> >> I've just seen on the platform with slow(ish) timer that it is now considered
> >> as a source of entropy with samples_per_bit set to 27 (5.19-rc6 has MAX_SAMPLES_PER_BIT
> >> set to 32). Because of that I see significant delays and I'm trying to understand what
> >> could be wrong with my setup.
> >>
> >> I observe one credit_init_bits(1) call (via entropy_timer()) per ~970 schedule() calls.
> >> Is that somewhat expected? Does it make sense at all?
> >
> > How slow are we talking? Seconds? Minutes? Is it too slow? It's possible
> > that MAX_SAMPLES_PER_BIT=32 is a bit high as a threshold and I should
> > reduce that a bit.
> >
>
> TBH, I run out of patience and never seen it completes, more then seconds. I just was
> curious how much it is should take to get crng_ready() return true.

Ooof. Yea, running this in a VM with various settings I can see that
the current maximum is problematic. I'll fix that up and send a patch.

Jason