Messages in this thread | | | Date | Sat, 11 Feb 2012 14:39:29 +0100 | From | Ingo Molnar <> | Subject | Re: [PATCH 0/4] CPU hotplug, cpusets: Fix CPU online handling related to cpusets |
| |
* Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:
> On Thu, 2012-02-09 at 16:11 +0100, Ingo Molnar wrote: > > > > My understanding of the code is that when a CPU is taken > > > offline, it is removed from all the cpusets and then the > > > scan_for_empty_cpusets() function is run to move tasks from > > > empty cpusets to their parent cpusets. > > > > Why is that done that way? offlining a CPU should be an > > invariant as far as cpusets are concerned. > > Can't, tasks need to run someplace. There's two choices, add a > still online cpu to the now empty cpuset or move the tasks to > a parent that still has online cpus. > > Both are destructive.
You aren't thinking hard enough ;-) There's several solutions off the top of my mind:
1) refuse the "impossible" offlining of the CPU, with a clear enough error to make it actionable
2) offer a 'forced' offlinign of a CPU that will SIGTERM all tasks that are on the now offline CPU and can only be there.
3) offer a 'nice' offlining variant that moves all orphan tasks to their or any other well-defined fallback CPU.
4) *allow* 'impossible' cpusets and just run them on CPU#0 or any other natural approximation. Don't touch the cpuset!
All of these would be exception mechanisms with no need to do anything at hot-replug time.
Thanks,
Ingo
| |