In the face of all the disagreements over scheduling approaches, it
occurs to me that Linux would be well served by allowing the scheduler
to be modularized.
Laughable you cry !
What it would take, I think, is a careful analysis of the scheduler
"interface", that is, all the functions that allow threads to be
queued and dequeued, as well as schedule() itself. Once identified in
this way (and if necessary, some cleanup done to create such a defined
interface, but I'm not sure this is needed), it should be relatively
simple to allow the module loading process to load a new version of
the scheduler.
Of course, it would be completely caveat emptor: if your new
scheduling module doesn't work right, nothing can save it. But having
such a mechanism in place might help people with strange scheduling
requirements, and would certainly avoid any attempts to expand the
default scheduler to cope with new and diverse situations.
--pbd
-
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/