If I got it right current scheduler queue missed dead lines. That's a little bit of a problem and may be better idea to ignore execution if dead line is already missed. Using cortex-Mx mirocontroller with nested prioritized interrupt this is what happen if interupt is triggered by a timer.
Read here https://man7.org/linux/man-pages/man7/sched.7.html SCHED_DEADLINE use Earliest Deadline First policy while SCHED_FIFO use fixed priority. Intterupt I mention above is equal to SCHED_FIFO. Both are optimal with some konditions for periodic tasks. SCHED_DEADLINE is a better choice since it allow a higher limit on the available execution to guarantee a dead line is never is missed. In you link it is stated Worst Case Execution Time WCET is easy to calculate but can't agree on this especially for a CPU on an ordinary computer. Software for WCET should be on the whish list for every real time programmer, done these kind of software for 15-20 years but never had the opportunity to use it. Run until done or interrupted by higher priority as done in the example above with the Micro controller may degrade gracefully if dead line is missed. It is of course not good if for example a value for control loop is not updated within period so that old value is used instead but effect is usually small. Queue starve lower priority tasks while abort execution result is fewer or none fully executed so that in example with control loop fewer values are successfully updated. I would rule for option to not queue or abort if dead line is missed and flag a warning. It should hopefully degrade gracefully. Nicklas Karlsson lör 2026-03-21 klockan 06:10 +0000 skrev Alec Ari via Emc-developers: > Hi all, > > Since SCHED_DEADLINE has precedence over SCHED_FIFO shouldn't we be using > that instead? > > https://unix.stackexchange.com/questions/342725/can-sched-fifo-be-preempted-by-sched-deadline#:~:text=The%20deadline%20scheduler%20has%20the,to%20get%20the%20work%20done > . > > Alec > > > _______________________________________________ > Emc-developers mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-developers _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
