1 My Random number library.
3 These routines can be used to generate pseudo random numbers and can be
4 used to 'seed' the pseudo random number generator (RNG). The RNG make no
5 effort to reproduce the same random number stream with each execution.
6 Various other routines in the SSLeay library 'seed' the RNG when suitable
7 'random' input data is available. Read the section at the end for details
8 on the design of the RNG.
13 This routine puts 'num' random bytes into 'buf'. One should make
14 sure RAND_seed() has been called before using this routine.
19 This routine adds more 'seed' data the RNG state. 'num' bytes
20 are added to the RNG state, they are taken from 'buf'. This
21 routine can be called with sensitive data such as user entered
22 passwords. This sensitive data is in no way recoverable from
23 the RAND library routines or state. Try to pass as much data
24 from 'random' sources as possible into the RNG via this function.
25 Also strongly consider using the RAND_load_file() and
26 RAND_write_file() routines.
29 When a program has finished with the RAND library, if it so
30 desires, it can 'zero' all RNG state.
32 The following 3 routines are convenience routines that can be used to
33 'save' and 'restore' data from/to the RNG and it's state.
34 Since the more 'random' data that is feed as seed data the better, why not
35 keep it around between executions of the program? Of course the
36 application should pass more 'random' data in via RAND_seed() and
37 make sure no-one can read the 'random' data file.
42 This routine returns a 'default' name for the location of a 'rand'
43 file. The 'rand' file should keep a sequence of random bytes used
44 to initialise the RNG. The filename is put in 'buf'. Buf is 'size'
45 bytes long. Buf is returned if things go well, if they do not,
46 NULL is returned. The 'rand' file name is generated in the
47 following way. First, if there is a 'RANDFILE' environment
48 variable, it is returned. Second, if there is a 'HOME' environment
49 variable, $HOME/.rand is returned. Third, NULL is returned. NULL
50 is also returned if a buf would overflow.
55 This function 'adds' the 'file' into the RNG state. It does this by
56 doing a RAND_seed() on the value returned from a stat() system call
57 on the file and if 'number' is non-zero, upto 'number' bytes read
58 from the file. The number of bytes passed to RAND_seed() is returned.
62 RAND_write_file() writes N random bytes to the file 'file', where
63 N is the size of the internal RND state (currently 1k).
64 This is a suitable method of saving RNG state for reloading via
67 What follows is a description of this RNG and a description of the rational
70 It should be noted that this RNG is intended to be used to generate
71 'random' keys for various ciphers including generation of DH and RSA keys.
73 It should also be noted that I have just created a system that I am happy with.
74 It may be overkill but that does not worry me. I have not spent that much
75 time on this algorithm so if there are glaring errors, please let me know.
76 Speed has not been a consideration in the design of these routines.
78 First up I will state the things I believe I need for a good RNG.
79 1) A good hashing algorithm to mix things up and to convert the RNG 'state'
81 2) An initial source of random 'state'.
82 3) The state should be very large. If the RNG is being used to generate
83 4096 bit RSA keys, 2 2048 bit random strings are required (at a minimum).
84 If your RNG state only has 128 bits, you are obviously limiting the
85 search space to 128 bits, not 2048. I'm probably getting a little
86 carried away on this last point but it does indicate that it may not be
87 a bad idea to keep quite a lot of RNG state. It should be easier to
88 break a cipher than guess the RNG seed data.
89 4) Any RNG seed data should influence all subsequent random numbers
90 generated. This implies that any random seed data entered will have
91 an influence on all subsequent random numbers generated.
92 5) When using data to seed the RNG state, the data used should not be
93 extractable from the RNG state. I believe this should be a
94 requirement because one possible source of 'secret' semi random
95 data would be a private key or a password. This data must
96 not be disclosed by either subsequent random numbers or a
97 'core' dump left by a program crash.
98 6) Given the same initial 'state', 2 systems should deviate in their RNG state
99 (and hence the random numbers generated) over time if at all possible.
100 7) Given the random number output stream, it should not be possible to determine
101 the RNG state or the next random number.
104 The algorithm is as follows.
106 There is global state made up of a 1023 byte buffer (the 'state'), a
107 working message digest ('md') and a counter ('count').
109 Whenever seed data is added, it is inserted into the 'state' as
111 The input is chopped up into units of 16 bytes (or less for
112 the last block). Each of these blocks is run through the MD5
113 message digest. The data passed to the MD5 digest is the
114 current 'md', the same number of bytes from the 'state'
115 (the location determined by in incremented looping index) as
116 the current 'block' and the new key data 'block'. The result
117 of this is kept in 'md' and also xored into the 'state' at the
118 same locations that were used as input into the MD5.
119 I believe this system addresses points 1 (MD5), 3 (the 'state'),
120 4 (via the 'md'), 5 (by the use of MD5 and xor).
122 When bytes are extracted from the RNG, the following process is used.
123 For each group of 8 bytes (or less), we do the following,
124 Input into MD5, the top 8 bytes from 'md', the byte that are
125 to be overwritten by the random bytes and bytes from the
126 'state' (incrementing looping index). From this digest output
127 (which is kept in 'md'), the top (upto) 8 bytes are
128 returned to the caller and the bottom (upto) 8 bytes are xored
130 Finally, after we have finished 'generation' random bytes for the
131 called, 'count' (which is incremented) and 'md' are fed into MD5 and
132 the results are kept in 'md'.
133 I believe the above addressed points 1 (use of MD5), 6 (by
134 hashing into the 'state' the 'old' data from the caller that
135 is about to be overwritten) and 7 (by not using the 8 bytes
136 given to the caller to update the 'state', but they are used
139 So of the points raised, only 2 is not addressed, but sources of
140 random data will always be a problem.