On Mon, Oct 14, 2013 at 03:36:38PM +0200, Ingo Molnar wrote: > > * Peter Zijlstra <[email protected]> wrote: > > > On Mon, Oct 14, 2013 at 02:38:55PM +0200, Ingo Molnar wrote: > > > > > > [...] the only 'issue' I have is the cgroup abi muck. We clearly > > > > need a bit more discussion on what/how we want things there but > > > > there are no easy answers :/ So I'd say lets try this and see where > > > > we'll find ourselves. > > > > > > I'd suggest we leave out the cgroup ABI muck from the first round of > > > upstream merge - do it in a second round, that will give it more > > > attention. > > > > I'm afraid that'd give rise to some very weird situations for people > > using cgroups :/ > > Why? One solution would be to just not offer bandwidth management > initially - but use some sane default. > > Yes, this doesn't offer "true" deadline scheduling yet, but would allow us > to move most of the code upstream, without any ABI changes initially > (other than adding the SCHED_DEADLINE policy and such).
I suppose we could keep the global sysctl thingies but indeed ignore all the cgroup limits for now. That would allow some admission control but leaves out the hairiest part. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
