On Thu, 23 Mar 2023, Alexey Dobriyan wrote:
> On Wed, Mar 22, 2023 at 05:07:09PM +0100, Miroslav Benes wrote:
> > On Wed, 22 Mar 2023, Alexey Dobriyan wrote:
> >
> > > Hi, Josh.
> > >
> > > I've been profiling how much time livepatching takes and I have a question
> > > regarding these lines:
> > >
> > > void klp_try_complete_transition(void)
> > > {
> > > ...
> > > if (!complete) {
> > > schedule_delayed_work(&klp_transition_work, round_jiffies_relative(HZ));
> > > return;
> > > }
> > >
> > > }
> > >
> > > The problem here is that if the system is idle, then the previous loop
> > > checking idle tasks will reliably sets "complete = false" and then
> > > patching wastes time waiting for next second so that klp_transition_work
> > > will repeat the same code without reentering itself.
> >
> > Only if klp_try_switch_task() cannot switch the idle task right away. We
> > then prod it using wake_up_if_idle() and handle it in the next iteration.
> >
> > So the question might be why klp_try_switch_task() did not succeed in the
> > first place.
>
> Yes, sort of. Transitioning never happens on the first try becase of idle
> tasks (see debugging patch below). But it should happen immediately
> because machine is idle!
Yes. My guess is that swapper task is running and cannot be switched
through klp_try_switch_task() so we prod it and it migrates itself through
our hook in do_idle().
You can enable a dynamic debug and see if it helps to confirm it
# echo "file kernel/livepatch/* +p" > /sys/kernel/debug/dynamic_debug/control
Miroslav
