On Mon, 21 Mar 2016 15:11:45 +0100 Peter Zijlstra <pet...@infradead.org> wrote:
> So the problem with preempt_enable_trace() is that we just called a > _notrace(), and while I now know these are two different trace thingies, > I might have forgotten that in a few days. > What about preempt_latency_start() for the disable, and preempt_latency_stop() for the enable? Make it more like a stop watch? -- Steve