Messages in this thread | | | Subject | Re: [PATCH 7/7] CPU controller V1 - (temporary) cpuset interface | From | Mike Galbraith <> | Date | Tue, 22 Aug 2006 15:23:29 +0000 |
| |
On Tue, 2006-08-22 at 14:41 +0000, Mike Galbraith wrote: > On Tue, 2006-08-22 at 15:40 +0530, Srivatsa Vaddagiri wrote: > > On Tue, Aug 22, 2006 at 11:10:36AM +0000, Mike Galbraith wrote: > > > Doesn't seem to work here, but maybe I'm doing something wrong. > > > > > > I set up cpuset "all" containing cpu 0-1 (all, 1.something cpus I have;) > > > > You are assigning all the CPUs to the cpuset "all" and then making it an > > exclusive/metered cpuset? > > Yeah. > > > I dont think I am handling that case well (yet), primarily because usage of > > remaining tasks (which are not in cpuset "all", "mikeg" & "root") is not > > accounted/controlled. Note that those remaining tasks will be running on one of > > the CPUs assigned to "all". What needs to happen is those remaining tasks need > > to be moved to a separate group (and a runqueue), being given some left-over > > CPU quota (which is left over from assignment of quota to mikeg and root), > > which is not handled in the patches (yet). One of the reason why I havent > > handled it yet is that there is no easy way to retrieve list of tasks attached > > to a cpuset. > > I try it with everything in either root or mikeg.
That didn't work.
> > Can you try assigning (NUM_CPUS-1) cpus to "all" and give it a shot? > > Essentially you need to ensure that only tasks chosen by you are running in > > cpus given to "all" and other child-cpusets under it.
With only cpu 1 in the cpuset, it worked.
-Mike
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |