Differences

This shows you the differences between two versions of the page.

--- realtime:documentation:howto:howto_rt_tools_cyclictest [2017/02/08 14:23]
anna-maria delete - move into an own namespace
+++ — (current)
@@ Line 1: / Line 1: @@
-====== Cyclictest ======
-Cyclictest is a high resolution test program, written Thomas Gleixner (tglx), maintained by Clark Williams and John Kacur.
-===== Installation =====
-Get the latest sources from the [[https://git.kernel.org/cgit/utils/rt-tests/rt-tests.git/|git repository]], and clone the repository or fetch a released [[https://www.kernel.org/pub/linux/utils/rt-tests/|tarball from the archive]], untar into a directory of your choice and run make in the source directory. If you want to cross compile, just run make CROSS_COMPILE=<your-compiler-prefix> (for example make CROSS_COMPILE=arm-v4t-linux-gnueabi-).
-You can run the resulting binary from there or install it:
-<code sh>
-#> git clone git://git.kernel.org/pub/scm/utils/rt-tests/rt-tests.git
-#> cd rt-tests
-#> make all
-#> cp ./cyclictest /usr/bin/
-#> cyclictest --help
-</code>
-<WRAP left round tip 100%>
-libnuma is required to build cyclictest. Usually, it's safe to have libnuma installed also in non-numa systems, but if you don't want to install the numa libs (e.g. in embedded environment) then compile with make NUMA=0.
-</WRAP>
-===== Run it =====
-Make sure to be root or use sudo to run cyclictest.
-Without parameters cyclictest creates one thread with a 1ms interval timer.
-cyclictest -h provides help text for the various options
-<code bash>
-cyclictest V 2.00
-Usage:
-cyclictest <options>
--a [CPUSET] --affinity     Run thread #N on processor #N, if possible, or if CPUSET
-                           given, pin threads to that set of processors in round-
-                           robin order.  E.g. -a 2 pins all threads to CPU 2,
-                           but -a 3-5,0 -t 5 will run the first and fifth
-                           threads on CPU (0),thread #2 on CPU 3, thread #3
-                           on CPU 4, and thread #5 on CPU 5.
--A USEC  --aligned=USEC    align thread wakeups to a specific offset
--b USEC  --breaktrace=USEC send break trace command when latency > USEC
--B       --preemptirqs     both preempt and irqsoff tracing (used with -b)
--c CLOCK --clock=CLOCK     select clock
-= CLOCK_MONOTONIC (default)
-= CLOCK_REALTIME
--C       --context         context switch tracing (used with -b)
--d DIST  --distance=DIST   distance of thread intervals in us, default=500
--D       --duration=TIME   specify a length for the test run.
-                           Append 'm', 'h', or 'd' to specify minutes, hours or days.
-         --latency=PM_QOS  write PM_QOS to /dev/cpu_dma_latency
--E       --event           event tracing (used with -b)
--f       --ftrace          function trace (when -b is active)
--F       --fifo=<path>     create a named pipe at path and write stats to it
--h       --histogram=US    dump a latency histogram to stdout after the run
-                           US is the max latency time to be be tracked in microseconds
-                           This option runs all threads at the same priority.
--H       --histofall=US    same as -h except with an additional summary column
-         --histfile=<path> dump the latency histogram to <path> instead of stdout
--i INTV  --interval=INTV   base interval of thread in us default=1000
--I       --irqsoff         Irqsoff tracing (used with -b)
--l LOOPS --loops=LOOPS     number of loops: default=0(endless)
-         --laptop          Save battery when running cyclictest
-                           This will give you poorer realtime results
-                           but will not drain your battery so quickly
--m       --mlockall        lock current and future memory allocations
--M       --refresh_on_max  delay updating the screen until a new max
-                           latency is hit. Userful for low bandwidth.
--n       --nanosleep       use clock_nanosleep
-         --notrace         suppress tracing
--N       --nsecs           print results in ns instead of us (default us)
--o RED   --oscope=RED      oscilloscope mode, reduce verbose output by RED
--O TOPT  --traceopt=TOPT   trace option
--p PRIO  --priority=PRIO   priority of highest prio thread
--P       --preemptoff      Preempt off tracing (used with -b)
-         --policy=NAME     policy of measurement thread, where NAME may be one
-                           of: other, normal, batch, idle, fifo or rr.
-         --priospread      spread priority levels starting at specified value
--q       --quiet           print a summary only on exit
--r       --relative        use relative timer instead of absolute
--R       --resolution      check clock resolution, calling clock_gettime() many
-                           times.  List of clock_gettime() values will be
-                           reported with -X
-         --secaligned [USEC] align thread wakeups to the next full second
-                           and apply the optional offset
--s       --system          use sys_nanosleep and sys_setitimer
--S       --smp             Standard SMP testing: options -a -t -n and
-                           same priority of all threads
-        --spike=<trigger>  record all spikes > trigger
-        --spike-nodes=[num of nodes]
-                           These are the maximum number of spikes we can record.
-                           The default is 1024 if not specified
-         --smi             Enable SMI counting
--t       --threads         one thread per available processor
--t [NUM] --threads=NUM     number of threads:
-                           without NUM, threads = max_cpus
-                           without -t default = 1
-         --tracemark       write a trace mark when -b latency is exceeded
--T TRACE --tracer=TRACER   set tracing function
-    configured tracers: blk mmiotrace function_graph wakeup_dl wakeup_rt wakeup function nop
--u       --unbuffered      force unbuffered output for live processing
--U       --numa            Standard NUMA testing (similar to SMP option)
-                           thread data structures allocated from local node
--v       --verbose         output values on stdout for statistics
-                           format: n:c:v n=tasknum c=count v=value in us
--w       --wakeup          task wakeup tracing (used with -b)
--W       --wakeuprt        rt task wakeup tracing (used with -b)
-         --dbg_cyclictest  print info useful for debugging cyclictest
-</code>
-More information is available by running ''less ./src/cyclictest/cyclictest.8''. The [[http://www.osadl.org/projects-live-cd.0.html|OSADL Realtime LiveCD project]] provides a script to plot the latency distribution.
-===== Expected Results =====
-<WRAP center round todo 60%>
-TODO: Run all the tests run in Expected results section of https://rt.wiki.kernel.org/index.php/Cyclictest and update here. We need to rerun the tests because they were run in 2006 on a Pentium III system running 2.6.16 kernel. Things have probably changed a bit now. :)
-</WRAP>
-===== FAQ =====
-==== ps shows the wrong scheduling class SCHED_OTHER ====
-Each cyclictest-task consist of one or more threads. ps -ce shows only the main-process not the threads of the main-process. ps -eLc | grep cyclic shows the main-process an the containing threads with the correct scheduler class SCHED_FIFO.
-<code bash>
-#>./cyclictest -t5 -p 80 -n -i 10000
-#> ps -cLe | grep cyclic
-  4764 TS   19 pts/1    00:00:01 cyclictest
-  4765 FF  120 pts/1    00:00:00 cyclictest
-  4766 FF  119 pts/1    00:00:00 cyclictest
-  4767 FF  118 pts/1    00:00:00 cyclictest
-  4768 FF  117 pts/1    00:00:00 cyclictest
-  4769 FF  116 pts/1    00:00:00 cyclictest
-</code>
-==== chrt shows the wrong scheduling class SCHED_OTHER ====
-Don't use the PID of the main-process, but the pid of one of the threads from the main-process. The threads are shown with ps -cLe | grep cyclic.
-<code bash>
-#> chrt -p 4766
-pid 4766's current scheduling policy: SCHED_FIFO
-pid 4766's current scheduling priority: 79
-</code>
-==== taskset for CPU affinity ====
-taskset command is Written by Robert M. Love. SMP operating systems have choices when it comes to scheduling processes: a new or newly rescheduled process can run on any available cpu. However, while it shouldn't matter where a new process runs, an existing process should go back to the same cpu it was running on simply because the cpu may still be caching data that belongs to that process. This is particularly apt to be true if the process is a thread: the other threads in the same program are very likely to have cpu cache of interest to their brethren (though obviously this also diminishes the performance gain that might be seen from multithreading) . For these reasons, scheduling algorithms pay attention to cpu affinity and try to keep it constant.
-It is possible to force a process to run only on a certain cpu. There are Linux system calls (sched_setaffinity and sched_getaffinity) and a command line "taskset".
-<code bash>
-#> taskset -c 3 top
-#> taskset -p [pid]
-</code>
-==== Compile failure because numa.h can't be found ====
-<code bash>
-make
-cc -D VERSION_STRING=0.85 -c src/cyclictest/cyclictest.c -Wall -Wno-nonnull -O2 -DNUMA -D_GNU_SOURCE -Isrc/include
-In file included from src/cyclictest/cyclictest.c:37:0:
-src/cyclictest/rt_numa.h:23:18: fatal error: numa.h: No such file or directory
-compilation terminated.
-make: *** [cyclictest.o] Error 1
-</code>
-Simply install your distribution's numa development package. On Fedora this is numactl-devel, so
-<code bash>
-su -c 'yum install numactl-devel'
-</code>
-This is only required for building. This will not affect the way the test runs on non-numa machines
-===== External Links =====
-==== Current repo ====
-Clone one of the following
-  * [[git://git.kernel.org/pub/scm/utils/rt-tests/rt-tests.git]]
-  * [[https://git.kernel.org/pub/scm/utils/rt-tests/rt-tests.git]]
-  * [[https://kernel.googlesource.com/pub/scm/utils/rt-tests/rt-tests.git]]
-==== rt-tests tarballs ====
-  * [[https://www.kernel.org/pub/linux/utils/rt-tests/]]
-==== Mailing List ====
-  * [[http://marc.info/?l=linux-rt-users|linux-rt-users mailing list]]

Wiki

User Tools

Site Tools

Differences

Page Tools