Dynamic Memory Management Research In Real-time Operating System ARTs-OS

With the development of multimedia information technology and internet, and the enhancement of intelligence for electronic product, embedded system was more and more popular. The safety and reliability of embedded system were crucial in key field such as spaceflight, industry control and medicine. In order to improve the safety and reliability of data storage in real-time embedded system, research on memory management for embedded system was particularly significant. The basic task of dynamic memory management is allocation and release of dynamic memory, assuring the reliability and stability of system.ARTs-OS is Preemptive microkernel embedded real time operating system. Based on the study of memory management of the ARTs-OS, research is focused on the dynamic storage allocation algorithm (DSA), which is the key of improving the performance of embedded real-time system. The dynamic storage allocation algorithm of the ARTs-OS include slab algorithm which is used to allocate and release of small memory blocks, and buddy system which is used of big memory blocks. In comparison of common DSA algorithms, TLSF (Two-Level Segregated Fit) algorithm is optimal performance, which meets the need of embedded RTOS and the time complexity of allocation and release is O(1). By TLSF research, propose an improved algorithm for TLSF. On the basic of TLSF, the improving TLSF combines the policies of FIFO and LIFO, trying not to split a whole memory block and choosing to split the combined block.After introducing the improving TLSF algorithm, propose the optimization strategies in the implementation of the algorithm. At the end, design test tasks and test the module of memory management. Comparing of improving TLSF, TLSF and Buddy, submit the results and analysis the response time and fragment. Result show that improving TLSF has an improvement than TLSF, than buddy system better.