Analysis of Randomness Levels in the Kernel Entropy Pool After Boot

Purpose of the thesis is to acquire entropy from the Linux random number generator and analyze quality of randomness for obtained entropy. System noise during normal operation passively generates entropy that is collected and exploited by the generator to create random numbers. The creation process treats entropy which leads us towards modifying kernel to allow access to raw entropy. The modification needs to be such as to allow for storage of the entropy in memory for extraction after the boot is finished. We’ve achieved the goal by coping data incoming into internal pool of the LRNG, which is used every time new entropy is being added into the generator. Custom system call was used as means of copying entropy from kernel to user space allowing us to analyze. Analysis was conducted using various methods to determine quality of entropy. It was determined that entropy in kernel after boot is low. Quality random numbers are important for variety of purposes, mainly to generate safe keys for cryptographic use. Majority of the keys are generated during boot, where there is not much system noise therefore it’s crucial to know whether or not enough quality entropy is available to the system.