ts::BetterSystemRandomGenerator Class Reference

Improved system-provided PRNG (pseudo-random numbers generator). More...

#include <tsBetterSystemRandomGenerator.h>

Inheritance diagram for ts::BetterSystemRandomGenerator:

Collaboration diagram for ts::BetterSystemRandomGenerator:

Public Member Functions
virtual UString	name () const override
	Algorithm name (informational only).
virtual bool	read (void *, size_t) override
	Get random data.
virtual bool	readByteBlock (ByteBlock &data, size_t size)
	Get random data in a byte block.
template<typename INT_T > requires std::integral<INT_T>
bool	readInt (INT_T &value, INT_T min=std::numeric_limits< INT_T >::min(), INT_T max=std::numeric_limits< INT_T >::max())
	Get a random integer value inside a defined range.
virtual bool	ready () const override
	Check if the PRNG is ready.
virtual bool	seed (const void *, size_t) override
	Seed (add entropy) to the PRNG.

Static Public Member Functions
static BetterSystemRandomGenerator &	Instance ()
	Get the instance of the singleton of this class.

Detailed Description

Improved system-provided PRNG (pseudo-random numbers generator).

Use SystemRandomGenerator as base and add AES-based post-processing. Implemented as a thread-safe singleton.

An entropy state is stored in $HOME/.tsseed. The value of a state is initially loaded from this file. The first time the generator is used (no file), the state is loaded from the system PRNG.

A fixed AES-128 key K is used for the post-processing. The size of the state is 16 bytes, the AES block size.

Description of post-processing, added to the sytem PRNG:

• R1 = read SystemRandomGenerator
• R2 = AES-128[K] (R1)
• R3 = R2 xor state
• R4 = AES-128[K] (R3)
• R4 ==> output of BetterSystemRandomGenerator
• R5 = read SystemRandomGenerator
• state = SHA-256 (R5 xor R4 xor state)

Known limitations:

• The entropy file is rewritten after each block => poor performances.
• Concurrent processes overwrite the same .tsseed file.

Member Function Documentation

Instance()

static BetterSystemRandomGenerator & ts::BetterSystemRandomGenerator::Instance ( )

inlinestatic

Get the instance of the singleton of this class.

Returns

The instance of the singleton of this class.

name()

virtual UString ts::BetterSystemRandomGenerator::name ( ) const

overridevirtual

Algorithm name (informational only).

Returns

The algorithm name.

Reimplemented from ts::SystemRandomGenerator.

ready()

virtual bool ts::BetterSystemRandomGenerator::ready ( ) const

overridevirtual

Check if the PRNG is ready.

If not ready, it must be seeded again.

Returns

True if the PRNG is ready.

Reimplemented from ts::SystemRandomGenerator.

read()

virtual bool ts::BetterSystemRandomGenerator::read ( void *  addr, size_t  size  )

overridevirtual

Get random data.

Parameters

[out]	addr	Address of returned random data.
[in]	size	Requested size in bytes of random data. The method always fill the buffer.

Returns

True on success, false on error.

Reimplemented from ts::SystemRandomGenerator.

seed()

virtual bool ts::SystemRandomGenerator::seed ( const void *  addr, size_t  size  )

overridevirtualinherited

Seed (add entropy) to the PRNG.

Parameters

[in]	addr	Address of entropy data.
[in]	size	Size in bytes of entropy data.

Returns

True on success, false on error.

Implements ts::RandomGenerator.

readByteBlock()

virtual bool ts::RandomGenerator::readByteBlock ( ByteBlock &  data, size_t  size  )

virtualinherited

Get random data in a byte block.

The default implementation uses read().

Parameters

[out]	data	Returned random data.
[in]	size	Requested size in bytes of random data.

Returns

True on success, false on error.

readInt()

template<typename INT_T >
requires std::integral<INT_T>

bool ts::RandomGenerator::readInt ( INT_T &  value, INT_T  min = std::numeric_limits<INT_T>::min(), INT_T  max = std::numeric_limits<INT_T>::max()  )

inherited

Get a random integer value inside a defined range.

Template Parameters

INT_T

An integer type for the result.

Parameters

[out]	value	The return integer value.
[in]	min	Minimum value to return.
[in]	max	Maximum value to return.

Returns

True on success, false on error.

---

The documentation for this class was generated from the following file:

• tsBetterSystemRandomGenerator.h