On Tue, 2021-04-06 at 22:03 +0200, Viktor Rosendahl wrote: > > > The usefulness lies in the case where you have latencies that systematically > occur very close to each other. > I should have dropped the "very" from the statement above. > > > Admittedly, I have never detected any real world case like the one described > above in the kernel but I don't think that it's completely unthinkable. > I forgot to mention that the kernel has a synthetic example of this situation in the preemptirq_delay_test module, if you use the burst_size parameter. If you want to see all the functions preemptirqtest_[0-9] in your stack traces, then you probably need to use the --random option. Then again this burst feature was implemented by me in order to test the latency-collector. best regards, Viktor
|