CFS dispatcher

Abstract

The invention provides a CFS dispatcher and belongs to the field of linux kernel process scheduling algorithm design. The CFS dispatcher comprises a weight conversion module (1), a time slice distributing module (2), a virtual time module (3), a red-black tree maintaining module (4) and a process selecting module (5). When IO operation is conducted, an IO special dispatching function is called each time the operation is over; when the IO dispatching function is called, the virtual operation time of the process is modified; when the function is called the first time, the virtual operation time is set to be an initial state, and a main dispatcher is called; in next two times of calling, the main dispatcher is normally called; back-and-forth switching is conducted; a next program for operating is selected through the main dispatcher. The CPU time distributed for the IO process by the CFS algorithm can be effectively prolonged, and therefore the IO speed is increased.

Description

technical field

[0001] The invention relates to the design field of linux kernel process scheduling algorithm, in particular to a CFS scheduler. Background technique

[0002] The CFS process scheduling algorithm is a product that appeared after linux2.6.23, and its appearance completely replaced the previous O(n) and O(1) scheduling algorithms; in the previous scheduling algorithm, no matter how optimized, there are always some low priorities The process cannot be scheduled by the CPU. In this case, CFS comes into being with a completely fair scheduling strategy.

[0003] The CFS algorithm is the most used process scheduling algorithm in the current Linux system. Because its idea is simple, easy to implement, and effectively solves the problem of process starvation, it effectively takes O(1) and O(n) scheduling algorithm. But the seemingly perfect algorithm also has its shortcomings. Although it perfectly solves the problem of process waiting, it introduces some new probl...

Images