* Tony Lindgren tony@atomide.com [180821 08:58]:
- Kevin Hilman khilman@baylibre.com [180820 21:32]:
Any other ideas for where the delay is coming from?
I think v4.16 is OK or at least I did not have any luck reproducing it so far on v4.16. I can reproduce it with v4.17, tried bisecting yesterday but of course it went nowhere as there's no easy way to know if some commit is good without booting it tens of times.
I think I managed to bisect it down to commit 554c8aa8ecad ("sched: idle: Select idle state before stopping the tick"). And looks like there's a fix being discussed for it at:
https://lore.kernel.org/lkml/2161372.IsD4PDzmmY@aspire.rjw.lan/
At least so far I have not been able to reproduce the issue with the patch being discussed, maybe give it a try on your test system too.
I'm guessing that probably the only reason why it was not caught earlier with omap2plus_defconfig is that it has more devices enabled and produces more interrupts compared to multi_v7_defconfig.
Regards,
Tony