https://github.com/MicroControleurMonde/RP2040_ROSC_RNG/blob/main/Lisez_moi_fr.md
A Micro-python library that provides an interface to generate a random number using the recommended method (ROSC reading) by Raspberry.
- Library :
rosc_random_generator.py- Python library for generating random numbers - Test Script:
test_rosc_random_generator.py- Example script demonstrating how to use the library.
Test_ROSC.py- The initial code that I used to test RP2040 ROSC.
- Context
- Project description
- Prerequisites
- Use
- Example of output
- Operation
- Performance test
- Statistical and quality tests
- Acknowledgment
- Disclaimer
- Special tribute
Raspberry foundation recently updated their Pico Datasheet:
Datasheet - Build-date: 2024-10-15 / Build-version: eec2b0c-clean
RP2040 Pico datasheet (Chapter 2.17)
Page 222. Chapter 2.17. Ring Oscillator (ROSC), here is what we can read:
2.17.5. Random Number Generator
If the system clocks are running from the XOSC and/or PLLs the ROSC can be used to generate
random numbers. Simply enable the ROSC and read the RANDOMBIT register to get a 1-bit
random number and read it n times to get an n-bit value. This does not meet the requirements
of randomness for security systems because it can be compromised, but it may be useful in
less critical applications. If the cores are running from the ROSC then the value will not be
random because the timing of the register read will be correlated to the phase of the ROSC.
The random number generator uses the ROSC (Resonance Oscillator) to produce random bits.
This project implements a True Random Number Generator (TRNG) using the RP2040 microcontroller. The generator use the Ring Oscillator as recommended in the datasheet..
- RP2040-based microcontroller.
- MicroPython firmware installed on the board.
machine,time(built-in with MicroPython).
- rosc_random_generator.py – Library for generating random numbers
This Python library provides a class ROSCRandomGenerator that generates random numbers based on the RP2040's RSOC.
- test_rosc_random_generator.py – Example script
This script imports the ROSCRandomGenerator class from rosc_random_generator.py and generates random numbers
- Save thess codes in files named
rosc_random_generator.pyandtest_rosc_random_generator.py. - To use this library in another project, just import it like any Python module.
The RP2040 ROSC RNG generates random integer numbers as 32-bit values.
ROSC is stable. Starting random number generation...
Random value # 1: 2880935910
Random value # 2: 3756622924
Random value # 3: 4041343900
Random value # 4: 41689329
Random value # 5: 1860732466
Random value # 6: 2376253135
Random value # 7: 2057898098
Random value # 8: 3392796777
Random value # 9: 1501861143
Random value # 10: 238555703
- 1 - The program declares the variables and constants required to interact with the ROSC.
- 2 - It checks the state of the ROSC to ensure that it is stable.
- 3 - If the ROSC is stable, it generates 32-bit random values, displaying each value and regularly checking the stability of the ROSC.
- 4 - If instability is detected, the program resets the ROSC and waits for it to stabilize before continuing.
- 5 - Once the values have been generated, it displays an end message.
This way, this process ensures that the ROSC is correctly initialised and stable before random values are generated, and manages the reset in the event of a stability problem during the process.
Caution: no writing on flash !
-
Estimated time to generate 1'425'424 values: 2'143 seconds (approx. 35 minutes)
-
Average time per value: 0.0015 sec
-
Throughput: 2660 Bytes/sec
-
Number of values generated per second: 665 values/sec
Note: Following a kind reminder from Scruss (Link), I changed the output file format from ASCII to Binary.
When converting random integers into ASCII strings, there is a loss of entropy ! A 32-bit
integer represents a value that can be expressed in a range of 4,294,967,296 possible
values (from 0 to 2^32 - 1). This integer range contains 32 bits of entropy.
Due to the way ASCII characters are represented (for example, numbers or letters),
some of the entropy is lost due to the limited number of possible characters. Each
character in ASCII contains approximately 3.45 bits of entropy instead of the
8 possible entropy bits for a complete byte (Log₂ 11 ≈ 3.459).
(www.fourmilab.ch) John Walker
-
Sample size: 5.43 Mo MB
-
Total generated: 1'425'424 values
-
Entropy = 7.973888 bits per byte (measure of randomness - Max=8)
Scatter plot representation [Full size]
Values histogram [Full size]
(https://webhome.phy.duke.edu/~rgb/General/dieharder.php) Robert G. Brown
-
Sample size: 5.43 Mo MB
-
Total generated: 1'425'424 values
(https://simul.iro.umontreal.ca/testu01/tu01.html)
-
C code snippet (only Main):
int main() { // Path to binary file const char *filename= "/home/user_name/rp2040_rosc_rng_1425424.bin"; // Create a generator from the file unif01_Gen *gen = create_gen(filename); if (gen == NULL) { return 1; } // Example with the Rabbit test battery printf("Starting tests from the \n file"); bbattery_Rabbit(gen, 1000); // Test on 1000 bits // Memory release close(global_gen->file); free(global_gen); // Free the generator state free(gen); // Free the generator return 0;
- Some low p-values suggest that the generator may have non-random structures or patterns detected by these tests.
(https://salsa.debian.org/hmh/rng-tools/)
-
Test result: all the FIPS tests failed
⚠️ Which implies a post-generator treatment...
djentA reimplementation of the Fourmilab/John Walker random number test program ent with several improvements.
(https://github.com/dj-on-github/djent)
sudo apt-get install libmpfr-dev libgmp-dev
And to update the Makefile accordingly before compilation!
The project is based on the RP2040 microcontroller, and its MicroPython firmware.
Waveshare RP2040-Plus: MicroPython ef518cbf2-dirty on 2023-03-24;XIAO seed studio: MicroPython v1.22.1 on 2024-01-05;
The code contained in this repository is provided “as is”, without any warranty of performance, accuracy or result. The author shall not be liable for any direct or indirect damages that may result from the use of this code, including, but not limited to, loss of data or interruption of service.
Use of this code is entirely at your own risk. Please ensure that you fully understand the code before using it in a production environment or integrating it into your projects.
We are deeply saddened to hear of the passing of John Walker, one of the founders of Autodesk. He died at home in Switzerland, on 2nd February 2024, aged 74.