I’m considering the concepts behind multiprocessing, and I’m trying to come up with some reason why a ready list is used that contains all runnable threads that aren’t running, as opposed to a list of all runnable threads with the head of the data structure being the running thread(s)?
Thanks for your opinions.
EDIT: Let me clarify. As far as I know, thread packages use a ready list to identify those processes that are ready to run, while the running process is identified by a separate variable. Why don’t they just include the running processes in the ready list data structure with the running thread at the head of the structure, making the thread package all inclusive. Would multiprocessing cause problems in this design scheme?
Because a thread can only run on one processor (core) at a time. The list (queue, really) of threads that are ready to run is used primarily by the scheduler when it’s looking for what thread it should run; if a thread is already running on one CPU, it can’t be run on another CPU at the same time, so the scheduler does not want to look at it (at that time — sometime later when it’s not running and eligible to run again, it will care about it again…)