Re: Unexpected switches to in-band

[Philippe Gerum <rpm@xenomai.org>]

Łukasz Majewski <lukma@nabladev.com> writes:

>
> <idle>-0       [002] *.~1.   128.466469: evl_timer_shot:
> latmus_pulse_handler at 128.467208 (delay: 739 us, 195206 cycles)
>
> Why there is such increase in "delay" (253 vs 739 us) ?

This is the delay until the next timer tick in an absolute timeline,
therefore the larger the value, the lesser time we consumed to handle
the last tick prior to programming the next one (e.g. 128.466214s +
253us = 128.466467s). The time consumed is still huge for this class of
hardware, but ftrace + memory stressing do explain this.

-- 
Philippe.