Menu
▾
▴
Tree [4a904a] master / History
Read Me
Readme - spass
Secure password/passphrase generator.
Authors
- Author: Guy Rutenberg
- Word list by Arnold G. Reinhold
Usage
By default spass will generate passwords. You can control the length using the -l modifier. Examples:
$ spass mqGP0GEZ $ ./spass -l 10 @-9Me6VNnT
If you pass the -p modifier, spass will output a passphrase instead of a password:
$ ./spass -p -l 4 pecan suey faith signor
For more options see spass --help.
Installation
Dependencies
- CMake-2.8
- PortAudio
- libasound2-dev
- libboost-program-options-dev
- qtbase5-dev
Linux
Just run:
mkdir build cd build cmake .. make sudo make install
Or use one of the provided binary packages.
Windows
Compiling Boost and PortAudio from source is easy. PortAudio uses CMake (just generate and compile), and compiling boost is a matter of: bootstrap and .\b2.
If you use Visual Studio Express 2012, you may need to set the Boost_COMPILER variable in CMake to -vc110, else it won't be detected properly.
In order to select the right Qt installation, set QT_QMAKE_EXECUTABLE to point to the correct path to qmake.exe. CMake will use it to infer the rest of the paths.
Due to a bug in CMake (actually a regression) setting the SubSystem doesn't work as it should. It should be set manually in the Project Properties->Linker->System to "Windows" (at least for release).
Configuration
cmake can receive the following options:
WITH_PORTAUDIO use PortAudio as the audio backend instead of ALSA
[default=OFF]
WITH_OSS use Open Sound System (OSS) as the audio backend
instead of ALSA [default=OFF]
WITH_ALSA use Advanced Linux Sound Architecture (ALSA) as the
audio backend [default=ON]
The options are exclusive, e.g. only one of them can be used each time. If no option is provided, the ALSA is used.
Security
The security of a password generator like spass, is determined by the quality of its underlying random number generator. spass employs a true random number generator based on noise obtained via the microphone. The noise goes through an unbiasing phase and then every 512 bits are compressed into 128 bits using a hash function. After those two phases the output bits should have full entropy.
See also Audio Based True Random Number Generator POC.
The words for the passphrases are taken from a list containing 8192 words. Hence, each word provides 13 bits of entropy.
Caveats
The current implementation can't open the audio device for capture in an exclusive mode (due to ALSA limitation). This means that theoratically, if you use spass on a system with a malicious user, he could record the same noise you're using and guess the output of spass. This means that you've to be in full control of the system your using spass on. While this may seem problematic, in practice it's a requirement for every other cryptography related application as well.