os/doc
3
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

- Update the random(4) 2013Q3 entry

Browse source 
Submitted by:	markm
This commit is contained in:
Gabor Pali 2013-10-15 08:41:46 +00:00
parent 4fdd168f8d
commit b206d08631
Notes: svn2git 2020-12-08 03:00:23 +00:00 
svn path=/head/; revision=42962
1 changed files with 47 additions and 8 deletions

55
en_US.ISO8859-1/htdocs/news/status/report-2013-07-2013-09.xml
View file

@ -949,7 +949,7 @@
</project>
<project cat='kern'>
<title>Updating <tt>random(4)</tt></title>
<title>Reworking <tt>random(4)</tt></title>
<contact>
<person>
@ -978,16 +978,55 @@
</contact>
<body>
<p>Update of <tt>random(4)</tt> to account for recent revelations,
and make more extensible, is in progress.</p>
<p>Random numbers require a lot more thought and preparation that
would naively appear to be the case. For simulations, number
sequences that are repeatable but sufficiently disordered are
often what is required to achieve required experimental
duplication ability, and many programmers are familiar with
these. For cryptography, it is essential that an attacker not
be able to predict or guess the output sequence, thus giving a
source of security-critical secret material for uses such as
passwords or "key material".</p>
<p>&os;'s random number generator, available as the pseudo-file
<tt>/dev/random</tt> produces unpredictable numbers intended for
cryptographic use, and is thus a Cryptograpically-Secured
Pseudo-Random Number Generator, or CSPRNG. The security is
given by careful design of the output generator (based on a
block cipher) and input entropy accumulation queues. The latter
uses hashes to accumulate stochastic information harvested from
various places in the kernel to provide highly unpredictable
input to the generator. The algorithm for doing this, Yarrow,
by Schneier et al, may be found by web search.</p>
<p>&os;'s CSPRNG also allowed for certain stochastic sources,
deemed to be "high-quality", to directly supply the
<tt>random(4)</tt> device without going through Yarrow. With
recent revelations over possible government surveillance and
involvement in the selection of these "high-quality" sources, it
is felt that they can no longer be trusted, and must therefore
also be processed though Yarrow.</p>
<p>The matter was discussed at various levels of formality at the
Cambridge Developer Summit in August, and at EuroBSDcon 2013 in
September.</p>
<p>This work is now done, and the <tt>random(4)</tt> CSPRNG is now
brought to a more paranoid, modern standard of distrust with
regard to its entropy sources. Infrastructure work was also
done to facilitate certain entropy-source choices for the
convenience of the system administrators.</p>
<p>Future work is now going ahead with the implementation of the
Fortuna algorithm by Ferguson and Schneier as an upgrade or
alternative to Yarrow. Initially a choice will be presented,
and decisions on the future of the CSPRNG processing algorithms
in use will be made in the future as needs arise.</p>
</body>
<help>
<task>Fortuna is to be an alternative for Yarrow in FreeBSD&nbsp;11.x.</task>
<task>Yarrow may be deprecated in a couple of years.</task>
<task>FIPS 800-90b support is planned.</task>
<task>A full, in-depth review of entropy is going to be done with
external help.</task>
<task>Implement FIPS 800-90b support.</task>
<task>A full, in-depth review of entropy.</task>
</help>
</project>