netdev - Re: Revising prandom

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:   Tue, 26 Mar 2019 12:23:40 -0700
From:   Stephen Hemminger <stephen@...workplumber.org>
To:     George Spelvin <lkml@....org>
Cc:     daniel@...earbox.net, hannes@...essinduktion.org,
        netdev@...r.kernel.org
Subject: Re: Revising prandom_32 generator

On Tue, 26 Mar 2019 19:07:01 GMT
George Spelvin <lkml@....org> wrote:

> lfsr113 is indeed trivial to predict.  It's a 113-bit LFSR defined
> by a degree-113 polynomial.  (The implementation as four separate
> polynomials of degree 31, 29, 28 and 25 doesn't change this.)  Given
> any 113 bits of its output (not necessarily consecutive), that's
> 113 boolean linear equations in 113 unknowns to find the internal
> state.
> 
> I don't have PoC code, but Gaussian elimination is not exactly
> rocket science.

If some code is using existing lfsr in a manner where prediction
would be a problem, then it is probably using the PRNG incorrectly
and should be using a cryptographic RNG.