Skip to content
/ gfc Public

Implementation of a Generalized-Feistel Cipher for generating random permutations.

License

Notifications You must be signed in to change notification settings

maxmouchet/gfc

Repository files navigation

🎲 gfc — fast & lazy random permutations

Tests Status PyPI

gfc is a C implementation of a Generalized-Feistel Cipher [1, alg. 3] for generating random permutations.
It uses Speck 64/128 as the random function, and can generate permutations with up to 2^64 elements.
The permutation is computed, and reversed, on-the-fly, without any mutable state and by using very little memory.

Usage

C / C++

API

#include <gfc/gfc.h>
GFC* gfc_init(uint64_t range, uint64_t rounds, uint64_t seed);
void gfc_destroy(GFC* gfc);
uint64_t gfc_decrypt(const GFC* gfc, uint64_t m);
uint64_t gfc_encrypt(const GFC* gfc, uint64_t m);

Example

// main.c
// gcc -Iinclude/ src/gfc.c main.c -o main
#include <assert.h>
#include <gfc/gfc.h>

int main() {
  GFC* gfc = gfc_init(65536, 6, 42);
  
  for (uint64_t i = 0; i < 65536; i++) {
    uint64_t enc = gfc_encrypt(gfc, i);
    uint64_t dec = gfc_decrypt(gfc, enc);
    assert(enc != i);
    assert(dec == i);
  }

  gfc_destroy(gfc);
  return 0;
}

CMake Integration

cmake_minimum_required(VERSION 3.12)
project(example)
add_subdirectory(gfc)
add_executable(main main.c)
target_link_libraries(main PRIVATE gfc)
git submodule add https://github.com/maxmouchet/gfc.git
mkdir build && cd build
cmake .. && cmake --build .
./main

Python

pip install pygfc
from pygfc import Permutation
# Permutation(range, rounds, seed)
perm = Permutation(2 ** 16, 8, 42)
assert set(perm) == set(range(2 ** 16))
assert all(perm.inv(perm[i]) == i for i in range(2 ** 16))

Dependencies

The Speck implementation is from madmo/speck and is licensed under the ISC license (MIT-compatible).

References

[1] Black, John, and Phillip Rogaway. "Ciphers with arbitrary finite domains." Cryptographers’ track at the RSA conference. Springer, Berlin, Heidelberg, 2002. https://web.cs.ucdavis.edu/~rogaway/papers/subset.pdf