linkDice

A cryptographically secure pseudo-random number generator CLI tool which generates byte arrays with entropy from the best random source from your machine* optionally externally seeded by multiple true random number generators and supports various byte-to-text encodings like hex or base64 and for many programming languages. The output may be printed to the command line or to a file. This implementation uses the HMAC Deterministic Random Bit Generator (DRBG) schema as defined in NIST SP800-90Ar1.

^{* depending on the used provider}

Main features:

• Supports all common byte encodings and more (hex, base32, base36, base64, base85, etc.)
• Optional secure seeding of random generator with random.org, Hotbits and ANU Quantum Random Numbers Server
• Generates code for random byte arrays for many programming languages (java, c, c#, kotlin, phyton, swift, go,...)
• NIST SP800-90Ar1 HMAC_DRBG tested with official test vectors
• Output to command line or file with automatic column formatting with upper limit of 10GiB+ of random data (~20MiB/s)
• Entropy warnings if seed is weak
• Additional output configuration like "www-form-urlencoding", padding of output and appended crc32 checksum

Example usage generating randoms with 24 byte-length (not char length) and default encoding:

1java -jar dice.jar 24

More examples:

1java -jar dice.jar 16 --count 100

2java -jar dice.jar 16 --encoding "base64"

3java -jar dice.jar 16 --encoding "java"

4java -jar dice.jar 4096 --encoding "raw" --count 1024 --file "./rnd-4-MiB-outputfile.txt"

5java -jar dice.jar 16 --seed "myBadRandomSeed"

6java -jar dice.jar 16 --offline

7java -jar dice.jar 32 --encoding "base85" --urlencode --padding --crc32

This should run on any Windows, Mac or Linux machine.

linkRequirements

• Java Runtime Environment (JRE) 8

linkDownload

Grab jar from latest Release

Using the *.exe Launcher: Launch4J is used to wrap the .jar into an Windows executable. It should automatically download the needed JRE if required.

linkDemo

linkUse Cases

linkCreating Nonces, Tokens, Identifiers or Passwords

I'll recommend base32, base36 or base58 because these encodings are typically url-safe. 16 byte usually suffice for globally unique, infeasible to brute force number.

1java -jar dice.jar 16 -e "base36"

If you require fixed char-sized output either use hex encoding or other encodings supporting paddings like base32 or base64. For passwords, high-density encodings are recommended like base85.

linkCreating static byte arrays for your application

You can create static salts, or randoms to hardcode, in your code. Just pick your programming language to get the correct syntax (see below). E.g.:

1java -jar dice.jar 16 -e "java"

linkCreating files of entropy

Create a file 4MiB full of raw random bytes with this call:

1java -jar dice.jar 4096 -c 1024 -e "raw" -f "./rnd-outfile.txt"

This will create random chunks of 4Kib (the maximum allowed size per chunk) repeated 1024 times. Currently the tool is capped to 10 GiB of generated random data per call. Successive calls will append data, not overwrite it.

linkCommand Line Interface

1   --anuquantum             Enable external, supposed true random generator ANU Quantum; note this service is known

2                            to be slow (only when online).

3-c,--count          How many randoms should be generated. Automatically chosen if this argument is omitted.

4   --crc32                  If this flag is set, 4 bytes of CRC32 checksum will be appended to every random value.

5                            If you need to check the integrity of the data.

6-d,--debug                  Prints additional info for debugging.

7-e,--encoding       Output byte-to-text encoding. Available encodings include:

8                            binary, octal, dec, base16, BASE16, base26, base32, base36, base58, base64, base64-url,

9                            base85, c, c#, java, go, kotlin, node, js, perl, php, python3, ruby, rust, swift, img,

10                            raw, utf8

11-f,--file             Prints the random data to given file instead of the command line. Will create the file

12                            if it does not exist or append the data if it does.

13-h,--help                   Shows this page.

14-o,--offline                Skips request to external random generators (random.org & hotbits) for seeding (use when

15                            offline).

16-p,--padding                If this flag is set, byte-to-text output will be padded to full byte if needed.

17-r,--robot                  If this flag is set, output will be more friendly for scripting (ie. no verbose text,

18                            only the randoms 1 per line)

19-s,--seed    Uses either the 64-bit integer interpretation or the utf-8 byte representation of given

20                            parameter to seed the internal random generator. Warns if entropy is low.

21-u,--urlencode              Uses 'www-form-urlencoded' encoding scheme, also misleadingly known as URL encoding, on

22                            the output strings

23-v,--version                Prints application version.

linkSupported Encodings

linkByte-to-Text Encodings

linkProgramming Languages

linkOther

linkDigital Signatures

linkSigned Jar

The provided JARs in the Github release page are signed with my private key:

1CN=Patrick Favre-Bulle, OU=Private, O=PF Github Open Source, L=Vienna, ST=Vienna, C=AT

2Validity: Thu Sep 07 16:40:57 SGT 2017 to: Fri Feb 10 16:40:57 SGT 2034

3SHA1: 06:DE:F2:C5:F7:BC:0C:11:ED:35:E2:0F:B1:9F:78:99:0F:BE:43:C4

4SHA256: 2B:65:33:B0:1C:0D:2A:69:4E:2D:53:8F:29:D5:6C:D6:87:AF:06:42:1F:1A:EE:B3:3C:E0:6D:0B:65:A1:AA:88

Use the jarsigner tool (found in your $JAVA_HOME/bin folder) folder to verify.

linkSigned Commits

All tags and commits by me are signed with git with my private key:

1GPG key ID: 4FDF85343912A3AB

2Fingerprint: 2FB392FB05158589B767960C4FDF85343912A3AB

linkDeterministic Random Bit Generation

As cryptographically secure pseudorandom number generator, the NIST SP800-90Ar1 recommendation HMAC-DRBG is used in an implementation derived from the google/rappor project. HMAC-DRBG seems to be a better choice than the also recommended HASH-DRBG approach. Java 9 is expected to have it's own provider for it. There is no known issue with Java's current SHA1-PRNG implementation, but it is less studied thant the NIST recommendation.

This implementation uses HMAC-SHA512 internally and reseeds itself after 1 MiB of random data generation which is well below the maximum NIST recommendation.

References:

• Bruce Schneider: Proof that HMAC-DRBG has No Back Doors
• Formal Verficiation of the HMAC-DRBG Pseudo Random Number Generator
• Security Analysis of DRBG Using HMAC in NIST SP800-90Ar1

linkOutput Test Results

This PRNG was tested with Robert G. Brown's dieharder test suite for random number generators, which among others, includes the NIST Statistical Test Suite.

The reports can be seen in /misc/reports/* in this repo.

linkDRBG Seeding & Input Sources

A DRGB needs to be seeded by strong entropy sources so it can safely be expanded to create unpredictable pseudo random output. SP800-90Ar1 defines different types of input for the DRGB. This implementation uses the following types:

linkEntropy Input

This implementation uses multiple entropy sources to seed it's random bit generator. All these sources are combined and a weak source will not weaken the overall output. This ensures that even if one source fails the output is still cryptographically strong. Below is a detailed description of the used sources:

linkStrong Secure Random Seed

This is the main entropy source. This implementation uses the SecureRandom class with its getStrongInstance() constructor to get the best cryptographic random generator available. Internally SecureRandom chooses among providers available at runtime. The best of those access the OS own entropy pools (e.g. /dev/random in *nix systems) since the OS has better access to various random sources.

Further reading:

• The Right Way to use Secure Random
• Discussion on seeding random generators

linkExternal Random Service Seeding

Per default the tool tries to fetch a seed from an external (supposedly true) random source.

Because there are various opinions what technique delivers truly random data, this tool incorporates 3 different services backed by different hardware RNG. Also to mitigate the fact that if one ore more source is either compromised or produces predictable outcome, the other source will mitigate that flaw.

Using an external random might open a new attack vector if, for example, an attacker might read the seed send over the network. There are 2 measures against this:

• The connections is encrypted with TLS (ie. HTTPS) and the random is signed by the creator which will be verified by a local pinned certificate (only random.org).
• The seed is only a part of the entropy source and the knowledge of it does not make it possible to guess the random bits. Therefore there is no sole trust in an external service. Every generation of random data will see seeding from both local and external sources.

HMAC-based key derivation function (HKDF) described in RFC5869 is used to expand the external seed to the desired length.

linkRandom.org

Random.org is a website that produces "true random numbers" based on atmospheric noise captured by several radios tuned between stations. The service has existed since 1998 and was built by Dr. Mads Haahr of the School of Computer Science and Statistics at Trinity College, Dublin in Ireland. Random.org offers Transport Layer Security (TLS) encrypted access and signed random data with JSON-RPC 2.0

References:

• Statistical Analysis
• Wikipedia Link

linkHotbits

Hotbits is a "genuine random numbers" service generating data by timing successive pairs of radioactive decays detected by a Geiger-Müller tube interfaced to a computer. This service was created by John Walker in 1996. Hotbits offers raw bytes with a simple HTTP GET request over TLS.

References:

• Statistical Analysis
• Wikipedia Reference

linkANU Quantum Random Numbers Server

A quantum random number generator offered by the Australian National University. The random numbers are generated in real-time by measuring the quantum fluctuations of the vacuum. The services provides a TLS encrypted JSON/REST API.

References:

• Statistical Analysis
• Random Generator Paper

linkLocal seeding

The caller may provide a string that additionally seeds the random bit generator. A seed provided by the user is seen as weak seed and will always be combined with the internal state of a strong SecureRandom instance.

linkThreaded Seed Generator

A port of SUN's threaded seed generator by Joshua Bloch. The seed is produced by counting the number of times the VM manages to loop in a given period. This number roughly reflects the machine load at that point in time. The samples are translated using a permutation (s-box) and then XORed together. This process is non linear and should prevent the samples from "averaging out". This implementation uses faster timings to produce values faster.

This generator is mainly used as a fallback if there is no external seed and the internal seeds fail.

linkNonce Input

The nonce is composed of:

• Monotonically increasing sequence number with a starting value of current JVM startup time
• System nano second time (which has arbitrary starting point)
• JVM uptime in milliseconds
• Current elapsed milliseconds since January 1, 1970 UTC.

The four 8 byte values will be hashed with HKDF.

linkPersonalization String

The goal of a personalization string is to gather as much information about e.g. runtime, machine identifiers and static identifiers to make the call as unique as possible for this particular machine/runtime/version/etc.

For this the following data will be gathered:

• MAC address of all network adapters
• Runtime & OS information (e.g. uptime, current cpu usage, processor count, classpath)
• SCM information (e.g. commit hash, committer, etc.) & version name
• Environmental variables and system properties
• Content of the temp directory

The resulting data will be hashed with HKDF.

linkExample Data

Under misc/example a couple of files of example random data can be found (e.g. 1MiB and 10MiB)

linkBuild

linkJar Sign

If you want to jar sign you need to provide a file keystore.jks in the root folder with the correct credentials set in environment variables ( OPENSOURCE_PROJECTS_KS_PW and OPENSOURCE_PROJECTS_KEY_PW); alias is set as pfopensource.

If you want to skip jar signing just change the skip configuration in the pom.xml jar sign plugin to true:

1true

linkBuild with Maven

Use maven (3.1+) to create a jar including all dependencies

1mvn clean install

linkCheckstyle Config File

This project uses my common-parent which centralized a lot of the plugin versions aswell as providing the checkstyle config rules. Specifically they are maintained in checkstyle-config. Locally the files will be copied after you mvnw install into your target folder and is called target/checkstyle-checker.xml. So if you use a plugin for your IDE, use this file as your local configuration.

linkTech Stack

• Java 8
• Maven
• rxjava2, apache-commons-codec, apache-commons-cli, Retrofit 2
• Errorprone, Proguard, Launch4j, Jar Signing

linkCredits

• HMAC_DRBG implementation derived from google/rappor
• Icon based on cube icon made by Smashicons

linkLicense

Copyright 2017 Patrick Favre-Bulle

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

1http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.