Am 11.02.2013 um 19:55 schrieb Stanislav Meduna: > On 11.02.2013 16:54, Ralf Müller wrote: > >> It basically is a loop that wakes up every milli second, increments a >> generation counter, does some statistics and goes to sleep for >> another millisecond. > > 3-4.5 seconds is heavy so this is not very probable, but seeing > that you have a thread with prio 99 and have timer wakeup > problems, depending on what the other realtime threads do maybe > the following could be (part of) the issue: > > http://www.spinics.net/lists/linux-rt-users/msg08745.html > http://www.spinics.net/lists/linux-rt-users/msg07804.html Thanks for the links. > Shortly said: you aren't going to have low latency timers > if the ksoftirqd can be preempted by your threads. > > Disclaimer: I have never used RT with 2.6, so I don't know > whether it applies at all. It is not called ksoftirqd on that kernel - it's sirq-timer, sirq-hrtimer, sirq-rtc there. But yes - looking for the priorities once again I found there is one thread, which is not completely trusted to never ever run wild, which had a higher priority than the timer interrupt threads. So I will adjust the interrupt threads first and see if something changes. Maybe you already made my day :) So far - thanks a lot. I will keep you informed if this adjustment helps - but it will take at least two days to install and test. Best regards Ralf-- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
