Messages in this thread | | | From | Tim Hockin <> | Subject | Re: CPU Management for Linux? | Date | Mon, 22 Mar 1999 15:51:21 -0600 (EST) |
| |
> > Adding and removing CPU's from a scheduler point of view isnt too > > horrific. You can force a CPU to not be running a task (eg send it > > an inter processor interrupt). When you add a CPU you need to clone > > it an idle thread and a stack. > > the above applies here -- if we tie a process to a specific processor > or set of processors (there are patches which implement and IRIX > compatible API here), how do we remove the processor?
This is an area where I am interested. I have written the IRIX compatible API (almost done with COMPLETE rewrite. Real Soon Now.) and am making wrappers for the sys-V API, and others, too.
One thing I am looking at as a pretty major thing would be run-queues on a per-processor basis.
For example (without a whole lot of thought put into this) what do people think of this:
* there is a 'general' queue, which is for tasks which can run on any processor(s) * each CPU has a run-queue. Tasks bound to this queue have a higher goodness than tasks in the 'general' queue.
?
Tim
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |