From 5fd895f1cf8998bdb620d13f85554f0ef97daf7a Mon Sep 17 00:00:00 2001 From: rusefi Date: Sun, 5 May 2019 09:41:44 -0400 Subject: [PATCH] docs --- firmware/controllers/scheduling/single_timer_executor.cpp | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/firmware/controllers/scheduling/single_timer_executor.cpp b/firmware/controllers/scheduling/single_timer_executor.cpp index c898452816..be2f30a88e 100644 --- a/firmware/controllers/scheduling/single_timer_executor.cpp +++ b/firmware/controllers/scheduling/single_timer_executor.cpp @@ -1,13 +1,16 @@ /** * @file SingleTimerExecutor.cpp * - * This class combines the powers of a 1MHz hardware timer from microsecond_timer.c + * This class combines the powers of a 1MHz hardware timer from microsecond_timer.cpp * and pending events queue event_queue.cpp * * As of version 2.6.x, ChibiOS tick-based kernel is not capable of scheduling events * with the level of precision we need, and realistically it should not. * - * See https://rusefi.com/forum/viewtopic.php?f=5&t=373&sid=e285ebd9a5677a83235116007e8eb65d&start=360#p30895 + * Update: actually newer ChibiOS has tickless mode and what we have here is pretty much the same thing :) + * open question if rusEfi should simply migrate to ChibiOS tickless scheduling (which would increase coupling with ChibiOS) + * + * See https://rusefi.com/forum/viewtopic.php?f=5&t=373&start=360#p30895 * for some performance data: with 'debug' firmware we spend about 5% of CPU in TIM5 handler which seem to be executed * about 1500 times a second *