aa814c51 - tools/tracing: Update Makefile to build rtla

Wed, 26 Jan 2022 00:22:34 +0000

tools/tracing: Update Makefile to build rtlaUpdate tracing Makefile to build/install/clean rtla tragets.Link: https://lkml.kernel.org/r/20220126002234.79337-1-skhan@linuxfoundation.orgReviewed-by: Daniel Bristot de Oliveira Signed-off-by: Shuah Khan Signed-off-by: Steven Rostedt (Google) List of files: /linux-6.15/tools/tracing/Makefile

e23db805 - tracing/tools: Add the latency-collector to tools directory

Fri, 12 Feb 2021 13:44:21 +0000

tracing/tools: Add the latency-collector to tools directoryThis is a tool that is intended to work around the fact that thepreemptoff, irqsoff, and preemptirqsoff tracers only work inoverwrite mode. The idea is to act randomly in such a way that wedo not systematically lose any latencies, so that if enough testingis done, all latencies will be captured. If the same burst oflatencies is repeated, then sooner or later we will have captured allthe latencies.It also works with the wakeup_dl, wakeup_rt, and wakeup tracers.However, in that case it is probably not useful to use the randomsleep functionality.The reason why it may be desirable to catch all latencies with a longtest campaign is that for some organizations, it's necessary to testthe kernel in the field and not practical for developers to workiteratively with field testers. Because of cost and project schedulesit is not possible to start a new test campaign every time a latencyproblem has been fixed.It uses inotify to detect changes to /sys/kernel/tracing/trace.When a latency is detected, it will either sleep or printimmediately, depending on a function that act as an unfair cointoss.If immediate print is chosen, it means that we open/sys/kernel/tracing/trace and thereby cause a blackout periodthat will hide any subsequent latencies.If sleep is chosen, it means that we wait before opening/sys/kernel/tracing/trace, by default for 1000 ms, to see ifthere is another latency during this period. If there is, then we willlose the previous latency. The coin will be tossed again with adifferent probability, and we will either print the new latency, orpossibly a subsequent one.The probability for the unfair coin toss is chosen so that thereis equal probability to obtain any of the latencies in a burst.However, this assumes that we make an assumption of how manylatencies there can be. By default the program assumes that thereare no more than 2 latencies in a burst, the probability of immediateprintout will be:1/2 and 1Thus, the probability of getting each of the two latencies will be 1/2.If we ever find that there is more than one latency in a series,meaning that we reach the probability of 1, then the table will beexpanded to:1/3, 1/2, and 1Thus, we assume that there are no more than three latencies and eachwith a probability of 1/3 of being captured. If the probability of 1is reached in the new table, that is we see more than two closelyoccurring latencies, then the table will again be extended, and soon.On my systems, it seems like this scheme works fairly well, aslong as the latencies we trace are long enough, 300 us seems to beenough. This userspace program receive the inotify event at the endof a latency, and it has time until the end of the next latencyto react, that is to open /sys/kernel/tracing/trace. Thus,if we trace latencies that are >300 us, then we have at least 300 usto react.The minimum latency will of course not be 300 us on all systems, itwill depend on the hardware, kernel version, workload andconfiguration.Example usage:In one shell, give the following command:sudo latency-collector -rvv -t preemptirqsoff -s 2000 -a 3This will trace latencies > 2000us with the preemptirqsoff tracer,using random sleep with maximum verbosity, with a probabilitytable initialized to a size of 3.In another shell, generate a few bursts of latencies:root@host:~# modprobe preemptirq_delay_test delay=3000 test_mode=alternateburst_size=3root@host:~# echo 1 > /sys/kernel/preemptirq_delay_test/triggerroot@host:~# echo 1 > /sys/kernel/preemptirq_delay_test/triggerroot@host:~# echo 1 > /sys/kernel/preemptirq_delay_test/triggerroot@host:~# echo 1 > /sys/kernel/preemptirq_delay_test/triggerIf all goes well, you should be getting stack traces that showsall the different latencies, i.e. you should see all the threefunctions preemptirqtest_0, preemptirqtest_1, preemptirqtest_2 in thestack traces.Link: https://lkml.kernel.org/r/20210212134421.172750-2-Viktor.Rosendahl@bmw.deSigned-off-by: Viktor Rosendahl Signed-off-by: Steven Rostedt (VMware) List of files: /linux-6.15/tools/tracing/Makefile

Changes in Makefile

aa814c51 - tools/tracing: Update Makefile to build rtla

e23db805 - tracing/tools: Add the latency-collector to tools directory