On 2021-01-08 10:32:36 [+0100], Peter Zijlstra wrote: > It's a single task wakeup (the caller), doing that from hardirq context > really should not be a problem, we do lots of that in RT already.
I'm not worry about that single wakeup but about an artificial case where you manage to accumulate multiple single wake ups in a short time frame. Sebastian
