The Security Study Of Embedded Linux Operation System

With the development of computer and network technology, information technology brings enormous benefits to the society, however, the computer system security is faced with serious challenges, and the same embedded computing systems security has become one of the urgent problems at present. Operating system is the core of the embedded system. If operating system is weak in security, the security of network communication and application software will be affected, and the security of the whole embedded system will not be guaranteed. Embedded Linux operating system is widely used in embedded application, so it is very important to study its safety.Firstly, an in-depth analysis is made of the key technology of secure operating system. The three kinds of international mainstream computer security evaluation criteria is discussed in this thesis. The structure, target, design principles and development methods of the secure operating system is analyzed. A discussion about how to enhance the operating system security and the development process of secure operating system is made.Secondly, an analysis of Linux operating system security mechanism and the main safety defects of Linux system is made. According to information confidentiality, integrity and availability of information protection, the advantages and disadvantages of mainstream access control model are studied. In view of the deficiencies in terms of flexibility and practicality existed in the two models, by introducing memory component into the system main body, expanding the safety level(including confidentiality and completeness) into safety level interval, two respective improved models are put forward based on the idea of information flow model, namely the MBLP model and the MBiba model. In the MBLP model only if the confidentiality level of the object is in the interval of the subject's, the subject could read the object, and if the confidentiality level of writing is higher than reading, the subject could read the object, which not only guarantees the information integrity, but also makes the model more flexible and practical. However in the Mbiba model, only if the completeness level of the object is in the interval of the subject's, the subject could read the object, and if the completeness level of all the writings are higher than all the readings, the subject could read the object, which increases flexibility and practicability of the model on the premise of guaranteeing the information integrity. Then in this thesis, formal descriptions and detailed security proofs of the two models are given, the analysis of the safety performance of the two models is made, and a contrast of improved model with original model is made.Finally, the LSM security module architecture is introduced. The MBLP model and MBiba model are loaded into the LSM module, then the MBLP model is initialized, so that to make it run in the LSM module.