Commit | Line | Data |
---|---|---|
d03ab969 | 1 | %%% -*-latex-*- |
2 | %%% | |
d03ab969 | 3 | %%% Description of Catacomb's random number generator |
4 | %%% | |
5 | %%% (c) 1999 Straylight/Edgeware | |
6 | %%% | |
7 | ||
8 | %%%----- Licensing notice --------------------------------------------------- | |
9 | %%% | |
10 | %%% This file is part of Catacomb. | |
11 | %%% | |
12 | %%% Catacomb is free software; you can redistribute it and/or modify | |
13 | %%% it under the terms of the GNU Library General Public License as | |
14 | %%% published by the Free Software Foundation; either version 2 of the | |
15 | %%% License, or (at your option) any later version. | |
45c0fd36 | 16 | %%% |
d03ab969 | 17 | %%% Catacomb is distributed in the hope that it will be useful, |
18 | %%% but WITHOUT ANY WARRANTY; without even the implied warranty of | |
19 | %%% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
20 | %%% GNU Library General Public License for more details. | |
45c0fd36 | 21 | %%% |
d03ab969 | 22 | %%% You should have received a copy of the GNU Library General Public |
23 | %%% License along with Catacomb; if not, write to the Free | |
24 | %%% Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, | |
25 | %%% MA 02111-1307, USA. | |
26 | ||
d03ab969 | 27 | %%%----- Header ------------------------------------------------------------- |
28 | ||
29 | \documentclass[a4paper, article, 10pt, notitlepage, numbering]{strayman} | |
30 | \usepackage[palatino, helvetica, courier, maths=cmr]{mdwfonts} | |
31 | \usepackage{mdwtab, mathenv} | |
32 | \usepackage[T1]{fontenc} | |
33 | \usepackage{cmtt, url} | |
34 | \usepackage[tpic, all]{xy} | |
35 | \usepackage{mathbbol} | |
36 | % \usepackage{crypto} | |
37 | ||
38 | \def\mdw{{\normalfont[{\bfseries\itshape mdw}]}} | |
39 | \urlstyle{tt} | |
40 | \def\email{\begingroup\urlstyle{rm}\Url} | |
41 | \urldef\myemail\email{mdw@nsict.org} | |
42 | \def\Z{\mathbb{Z}} | |
43 | \let\assign\leftarrow | |
44 | \let\xor\oplus | |
45 | \let\bigxor\bigoplus | |
25936864 | 46 | \def\cat{\mathbin{\|}} |
d03ab969 | 47 | |
48 | \title{The Catacomb random number generator} | |
49 | \author{Mark Wooding, \myemail} | |
50 | ||
51 | %%%----- The main document -------------------------------------------------- | |
52 | ||
53 | \begin{document} | |
54 | ||
55 | \maketitle | |
56 | ||
57 | \begin{abstract} | |
58 | The author describes the random number generator used in the | |
59 | Straylight/\-Edgeware `Catacomb' library. While the generator is | |
60 | superficially similar to (for example) the Linux and OpenBSD random number | |
61 | generators, it introduces a number of its own innovations which improve | |
62 | both security and performance. | |
45c0fd36 | 63 | |
d03ab969 | 64 | The Catacomb generator uses an optional secret key, which can provide |
65 | additional security against forward state compromise extension. It uses a | |
66 | catastrophic reseeding operation to prevent a compromise yielding | |
67 | information about past generator states. This operation works on | |
68 | arbitrary-sized blocks of data, so the generator's output buffer can be | |
69 | large. This minimizes the effect of the reseeding overhead. | |
70 | \end{abstract} | |
71 | ||
72 | \tableofcontents | |
73 | ||
74 | ||
75 | \section{The one-way transformation} | |
25936864 | 76 | \label{sec:oneway} |
d03ab969 | 77 | |
78 | The most novel part of the generator\footnote{I believe this construction to | |
79 | be novel. If I'm wrong, let me know.} is the one-way transformation which is | |
80 | used to allow pooled input data to affect the output buffer. | |
81 | ||
82 | Let $H$ be some collision-resistant hash function, and let $E_k$ be a | |
83 | symmetric cipher with key $k$. Then I can define the one-way transformation | |
84 | $T$ by | |
85 | \[ T(x) = E_{H(x)}(x) \] | |
86 | ||
87 | I believe, although formal proof seems difficult, that an adversary in | |
88 | posession of $T(x)$ and a portion of the original $x$ cannot reconstruct the | |
89 | remainder of $x$ without breaking one of the cryptographic primitives (which | |
90 | I assume is `difficult') or performing an exhaustive search of one of: the | |
91 | space of the unknown portion of $x$, the range of the hash function $H$, or | |
92 | the keyspace of the cipher $E$. | |
93 | ||
94 | A similar feat of cryptanalysis or exhaustive search seems necessary to work | |
95 | in a forwards direction: given partial knowledge of both $x$ and $T(x)$, the | |
96 | adversary cannot work out the remainder of either without trying every | |
97 | possibility for one or the other unknown portions, or working through the | |
98 | hash- or keyspace. | |
99 | ||
100 | A keyed version of $T$ may be defined, given a keyed hash (or MAC) $H_k$: | |
101 | \[ T_k(x) = E_{H_k(x)}(x) \] | |
102 | If this is done, the adversary cannot work forwards even with \emph{complete} | |
103 | knowledge of $B$, or performing one of the obvious exhaustive searches. | |
104 | ||
105 | ||
25936864 | 106 | \section{Abstract description of the generator} |
d03ab969 | 107 | |
108 | The generator is divided into two parts: an \emph{input pool} which | |
109 | accumulates random input data from the environment, and an \emph{output | |
110 | buffer} which contains data to be passed to clients of the generator on | |
111 | request. | |
112 | ||
25936864 | 113 | \subsection{The input pool and mixing function} |
114 | ||
d03ab969 | 115 | New information is contributed to the generator by mixing it with the input |
116 | pool, using a mixing function derived from the Linux random number generator | |
117 | \cite{linux:devrandom}. The mixing function views the input pool as eight | |
118 | parallel shift registers. Input data is added one octet at a time. Each bit | |
119 | of an input octet is mixed with a different shift register. | |
120 | ||
25936864 | 121 | Formally, let $I$ be the input pool, with size $N_I$ bytes; let $P(x) = a_0 + |
122 | a_1 x + a_2 x^2 + \cdots + a_{N_I} x^{N_I}$ be a primitive polynomial in | |
123 | $\mathrm{GF}(2^{N_I})$ with degree $N_I$; let $i$ be an integer such that $0 | |
124 | \le i < N_I$, and $r$ be an integer such that $0 \le r < 8$; and let $x$ be | |
d03ab969 | 125 | an input byte. The result of mixing $x$ with the pool $I$ is calculated as |
126 | follows: | |
127 | \begin{eqlines*} | |
128 | \begin{spliteqn*} | |
129 | I'[8j + b] = | |
130 | \begin{cases} | |
131 | x\bigl[(r + b) \bmod 8\bigr] \xor | |
45c0fd36 MW |
132 | \bigxor_{0 \le k < N_I} |
133 | a_k I\bigl[8\bigl((j + k) \bmod N_I\bigr) + b\bigr] & if $i = j$ \\ | |
d03ab969 | 134 | I[j + b] & otherwise |
135 | \end{cases} \\ | |
25936864 | 136 | \textrm{for all integers $j$ and $b$ where $0 \le j < N_I$ and |
d03ab969 | 137 | $0 \le b < 8$} |
138 | \end{spliteqn*} | |
139 | \\ | |
140 | I \assign I' \qquad | |
25936864 | 141 | i \assign (i + 1) \bmod N_I \qquad |
d03ab969 | 142 | r \assign (r + 5) \bmod 8 |
143 | \end{eqlines*} | |
144 | Initially, $i$ and $r$ are both zero. The use of 8-bit bytes above is | |
25936864 | 145 | arbitrary but convenient for modern computers. |
146 | ||
147 | The mixing function isn't intended to be cryptographically strong. Its | |
148 | purpose is just to hold data without letting too much of the randomness get | |
149 | away. | |
150 | ||
151 | \subsection{The output buffer} | |
d03ab969 | 152 | |
153 | Newly added data doesn't affect the output buffer until a `gating' operation | |
154 | is performed. This uses the one-way transformation described earlier over | |
155 | the entire generator state. | |
156 | ||
157 | Data requested by clients of the generator is read from the output buffer | |
25936864 | 158 | $O$. Initially the buffer contains zeroes. The output buffer is large |
159 | enough for $N_O$ bits. | |
160 | ||
161 | When the generator estimates that there's enough entropy in the input pool, a | |
162 | \emph{gating} operation is performed, using the one-way function described in | |
163 | section~\ref{sec:oneway}. Hash both the input pool and output buffer, and | |
164 | then encrypt both using the hash as the key: | |
165 | \begin{eqlines*} | |
166 | h = H(I \cat O) \\ | |
167 | I \assign E_h(I) \qquad O \assign E_h(O) | |
168 | \end{eqlines*} | |
169 | If the output buffer is exhausted before the next gating operation, it is | |
170 | \emph{stretched} using the one-way function: $O \assign E_{H(O)}(O)$. | |
171 | ||
172 | \subsection{Other tweaks} | |
173 | ||
174 | The first $N_S$ bits of this buffer take part in the output transformation | |
175 | but are never actually output. They're there to make predicting further | |
176 | output from the generator difficult. | |
177 | ||
178 | Also, optionally, the one-way functions can be keyed. This does, of course, | |
179 | beg the question as to where the key comes from. This might be one of those | |
180 | things best done the old-fashioned way with a bunch of coins or dice or | |
181 | something. | |
182 | ||
183 | ||
184 | \section{The actual implementation} | |
185 | ||
186 | The Catacomb implementation of the generator uses the following parameters: | |
187 | \begin{itemize} | |
188 | \item The hash function used in the one-way transformation is RIPEMD-160 | |
189 | \cite{rmd160}; the block cipher is Blowfish, using a 160-bit key. | |
190 | \item The input pool size $N_I$ is 128 bytes. The output buffer size $N_O$ | |
61383268 | 191 | is 512 bytes. The size $N_S$ of the secret part of the output buffer |
25936864 | 192 | is 160 bits (20 bytes). |
193 | \item The polynomial $P(x)$ used for mixing in new input is | |
194 | $1 + x + x^2 + x^7 + x^{128}$. | |
195 | \end{itemize} | |
196 | The hash and block cipher are well-known and respected cryptographic | |
61383268 | 197 | primitives. |
198 | ||
199 | The input pool is rater larger than it strictly needs to be to contain | |
200 | `enough' entropy to bring the generator up to the strength of its | |
201 | cryptographic primitives. The pool is large to reduce the effect of | |
202 | asymptotic behaviour in the amount of entropy in the pool. | |
203 | ||
204 | The output buffer is large simply to improve performance: Blowfish has a | |
205 | heavy key schedule, so it pays to perform fewer rekeyings per byte of data. | |
206 | The precise size of 512 bytes was chosen empirically as being about where the | |
207 | performance improvement stops being linear with the buffer size on my | |
208 | machine. | |
d03ab969 | 209 | |
210 | \begin{thebibliography}{99} | |
45c0fd36 | 211 | |
d03ab969 | 212 | \bibitem{cp:rand} |
213 | J.~Kelsey, B.~Schneier, D.~Wagner, and C.~Hall, ``Cryptographic Attacks on | |
214 | Pseudorandom Number Generators'', \emph{Fast Software Encryption, Fifth | |
215 | International Workshop Proceedings (March 1998)}, Springer-Verlag, 1998, | |
216 | pp. 168--188, \url{http://www.counterpane.com/pseudorandom_number.html} | |
217 | ||
218 | \bibitem{linux:devrandom} | |
219 | T.~Ts'o, ``A string random number generator'', Linux sources, | |
220 | \path{drivers/char/random.c}. | |
221 | ||
222 | \bibitem{mdw:devrandom} | |
223 | M.~Wooding, ``Linux \path{/dev/random} generator security'', Usenet article | |
224 | posted to \mtt{sci.crypt}, July 1998. | |
225 | ||
226 | \end{thebibliography} | |
227 | ||
228 | %%%----- That's all, folks -------------------------------------------------- | |
229 | ||
45c0fd36 | 230 | \end{document} |