You need to add these packages to the target dependencies, for example, in conf/local.conf:

IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"

If you want to run Babeltrace on the target, also append the package babeltrace.

You need to enable the following:

For user space trace tracing, enable this:

There is a package called lttng-babletrace for the target. Buildroot builds the host babeltrace automatically and places in output/host/usr/bin/babeltrace.

LTTng can use the set of ftrace events described above as potential trace points. Initially, they are disabled.

The control interface for LTTng is the lttng command. You can list the kernel probes using the following:

# lttng list --kernel
Kernel events:
-------------
      writeback_nothread (loglevel: TRACE_EMERG (0)) (type: tracepoint)
      writeback_queue (loglevel: TRACE_EMERG (0)) (type: tracepoint)
      writeback_exec (loglevel: TRACE_EMERG (0)) (type: tracepoint)
[...]

Traces are captured in the context of a session which, in this example, is called test:

# lttng create test
Session test created.
Traces will be written in /home/root/lttng-traces/test-20150824-140942
# lttng list
Available tracing sessions:
  1) test (/home/root/lttng-traces/test-20150824-140942) [inactive]

Now enable a few events in the current session. You can enable all kernel tracepoints using the --all option but remember the warning about generating too much trace data. Let's start with a couple of scheduler-related trace events:

# lttng enable-event --kernel sched_switch,sched_process_fork

Check that everything is set up:

# lttng list test
Tracing session test: [inactive]
    Trace path: /home/root/lttng-traces/test-20150824-140942
    Live timer interval (usec): 0

=== Domain: Kernel ===

Channels:
-------------
- channel0: [enabled]

    Attributes:
      overwrite mode: 0
      subbufers size: 26214
      number of subbufers: 4
      switch timer interval: 0
      read timer interval: 200000
      trace file count: 0
      trace file size (bytes): 0
      output: splice()

    Events:
      sched_process_fork (loglevel: TRACE_EMERG (0)) (type: tracepoint) [enabled]
      sched_switch (loglevel: TRACE_EMERG (0)) (type: tracepoint) [enabled]

Now start tracing:

# lttng start

Run the test load and then stop tracing:

# lttng stop

Traces for the session are written to the session directory, lttng-traces/<session>/kernel.

You can use the Babeltrace viewer to dump the raw trace data in text format, in this case, I ran it on the host computer:

$ babeltrace  lttng-traces/test-20150824-140942/kernel

The output is too verbose to fit on this page, so I will leave it as an exercise for you, the reader, to capture and display a trace in this way. The text output from eBabeltrace does have the advantage that it is easy to search for strings using grep and similar commands.

A good choice for a graphical trace viewer is the Trace Compass plug-in for Eclipse, which is now part of the Eclipse IDE for C/C++ Developers bundle. Importing the trace data into Eclipse is characteristically fiddly. Briefly, you need to follow these steps:

In the preceding screenshot, I have zoomed in on the control flow view to show state transitions between dropbear and a shell, and also some activity of the lttng daemon.