1e71b7053SJung-uk Kim=pod 2e71b7053SJung-uk Kim 3e71b7053SJung-uk Kim=head1 NAME 4e71b7053SJung-uk Kim 5e71b7053SJung-uk KimRAND 6e71b7053SJung-uk Kim- the OpenSSL random generator 7e71b7053SJung-uk Kim 8e71b7053SJung-uk Kim=head1 DESCRIPTION 9e71b7053SJung-uk Kim 10e71b7053SJung-uk KimRandom numbers are a vital part of cryptography, they are needed to provide 11e71b7053SJung-uk Kimunpredictability for tasks like key generation, creating salts, and many more. 12e71b7053SJung-uk KimSoftware-based generators must be seeded with external randomness before they 13e71b7053SJung-uk Kimcan be used as a cryptographically-secure pseudo-random number generator 14e71b7053SJung-uk Kim(CSPRNG). 15e71b7053SJung-uk KimThe availability of common hardware with special instructions and 16e71b7053SJung-uk Kimmodern operating systems, which may use items such as interrupt jitter 17e71b7053SJung-uk Kimand network packet timings, can be reasonable sources of seeding material. 18e71b7053SJung-uk Kim 19e71b7053SJung-uk KimOpenSSL comes with a default implementation of the RAND API which is based on 20e71b7053SJung-uk Kimthe deterministic random bit generator (DRBG) model as described in 21e71b7053SJung-uk Kim[NIST SP 800-90A Rev. 1]. The default random generator will initialize 22e71b7053SJung-uk Kimautomatically on first use and will be fully functional without having 23e71b7053SJung-uk Kimto be initialized ('seeded') explicitly. 24e71b7053SJung-uk KimIt seeds and reseeds itself automatically using trusted random sources 25e71b7053SJung-uk Kimprovided by the operating system. 26e71b7053SJung-uk Kim 27e71b7053SJung-uk KimAs a normal application developer, you do not have to worry about any details, 28e71b7053SJung-uk Kimjust use L<RAND_bytes(3)> to obtain random data. 29e71b7053SJung-uk KimHaving said that, there is one important rule to obey: Always check the error 30e71b7053SJung-uk Kimreturn value of L<RAND_bytes(3)> and do not take randomness for granted. 31da327cd2SJung-uk KimAlthough (re-)seeding is automatic, it can fail because no trusted random source 32da327cd2SJung-uk Kimis available or the trusted source(s) temporarily fail to provide sufficient 33da327cd2SJung-uk Kimrandom seed material. 34da327cd2SJung-uk KimIn this case the CSPRNG enters an error state and ceases to provide output, 35da327cd2SJung-uk Kimuntil it is able to recover from the error by reseeding itself. 36*b077aed3SPierre ProncheryFor more details on reseeding and error recovery, see L<EVP_RAND(7)>. 37e71b7053SJung-uk Kim 38e71b7053SJung-uk KimFor values that should remain secret, you can use L<RAND_priv_bytes(3)> 39e71b7053SJung-uk Kiminstead. 40*b077aed3SPierre ProncheryThis method does not provide 'better' randomness, it uses the same type of 41*b077aed3SPierre ProncheryCSPRNG. 42e71b7053SJung-uk KimThe intention behind using a dedicated CSPRNG exclusively for private 43e71b7053SJung-uk Kimvalues is that none of its output should be visible to an attacker (e.g., 44e71b7053SJung-uk Kimused as salt value), in order to reveal as little information as 45e71b7053SJung-uk Kimpossible about its internal state, and that a compromise of the "public" 46e71b7053SJung-uk KimCSPRNG instance will not affect the secrecy of these private values. 47e71b7053SJung-uk Kim 48e71b7053SJung-uk KimIn the rare case where the default implementation does not satisfy your special 49*b077aed3SPierre Proncheryrequirements, the default RAND internals can be replaced by your own 50*b077aed3SPierre ProncheryL<EVP_RAND(3)> objects. 51e71b7053SJung-uk Kim 52*b077aed3SPierre ProncheryChanging the default random generator should be necessary 53*b077aed3SPierre Proncheryonly in exceptional cases and is not recommended, unless you have a profound 54*b077aed3SPierre Proncheryknowledge of cryptographic principles and understand the implications of your 55*b077aed3SPierre Proncherychanges. 56e71b7053SJung-uk Kim 57*b077aed3SPierre Pronchery=head1 DEFAULT SETUP 58e71b7053SJung-uk Kim 59*b077aed3SPierre ProncheryThe default OpenSSL RAND method is based on the EVP_RAND deterministic random 60*b077aed3SPierre Proncherybit generator (DRBG) classes. 61*b077aed3SPierre ProncheryA DRBG is a certain type of cryptographically-secure pseudo-random 62*b077aed3SPierre Proncherynumber generator (CSPRNG), which is described in [NIST SP 800-90A Rev. 1]. 63e71b7053SJung-uk Kim 64e71b7053SJung-uk Kim 65e71b7053SJung-uk Kim=head1 SEE ALSO 66e71b7053SJung-uk Kim 67e71b7053SJung-uk KimL<RAND_bytes(3)>, 68e71b7053SJung-uk KimL<RAND_priv_bytes(3)>, 69*b077aed3SPierre ProncheryL<EVP_RAND(3)>, 70*b077aed3SPierre ProncheryL<RAND_get0_primary(3)>, 71*b077aed3SPierre ProncheryL<EVP_RAND(7)> 72e71b7053SJung-uk Kim 73e71b7053SJung-uk Kim=head1 COPYRIGHT 74e71b7053SJung-uk Kim 75*b077aed3SPierre ProncheryCopyright 2018-2021 The OpenSSL Project Authors. All Rights Reserved. 76e71b7053SJung-uk Kim 77*b077aed3SPierre ProncheryLicensed under the Apache License 2.0 (the "License"). You may not use 78e71b7053SJung-uk Kimthis file except in compliance with the License. You can obtain a copy 79e71b7053SJung-uk Kimin the file LICENSE in the source distribution or at 80e71b7053SJung-uk KimL<https://www.openssl.org/source/license.html>. 81e71b7053SJung-uk Kim 82e71b7053SJung-uk Kim=cut 83