1 The LTTng Documentation
2 =======================
3 Philippe Proulx <pproulx@efficios.com>
7 include::../common/copyright.txt[]
10 include::../common/welcome.txt[]
13 include::../common/audience.txt[]
17 === What's in this documentation?
19 The LTTng Documentation is divided into the following sections:
21 * **<<nuts-and-bolts,Nuts and bolts>>** explains the
22 rudiments of software tracing and the rationale behind the
25 You can skip this section if you’re familiar with software tracing and
26 with the LTTng project.
28 * **<<installing-lttng,Installation>>** describes the steps to
29 install the LTTng packages on common Linux distributions and from
32 You can skip this section if you already properly installed LTTng on
35 * **<<getting-started,Quick start>>** is a concise guide to
36 getting started quickly with LTTng kernel and user space tracing.
38 We recommend this section if you're new to LTTng or to software tracing
41 You can skip this section if you're not new to LTTng.
43 * **<<core-concepts,Core concepts>>** explains the concepts at
46 It's a good idea to become familiar with the core concepts
47 before attempting to use the toolkit.
49 * **<<plumbing,Components of LTTng>>** describes the various components
50 of the LTTng machinery, like the daemons, the libraries, and the
51 command-line interface.
52 * **<<instrumenting,Instrumentation>>** shows different ways to
53 instrument user applications and the Linux kernel.
55 Instrumenting source code is essential to provide a meaningful
58 You can skip this section if you do not have a programming background.
60 * **<<controlling-tracing,Tracing control>>** is divided into topics
61 which demonstrate how to use the vast array of features that
62 LTTng{nbsp}{revision} offers.
63 * **<<reference,Reference>>** contains reference links and tables.
64 * **<<glossary,Glossary>>** is a specialized dictionary of terms related
65 to LTTng or to the field of software tracing.
68 include::../common/convention.txt[]
71 include::../common/acknowledgements.txt[]
75 == What's new in LTTng {revision}?
77 * **Tracing control**:
78 ** Dynamic filter support for <<event,event rules>> in the Linux kernel
79 <<domain,tracing domain>>. For example:
84 lttng enable-event --kernel irq_handler_entry --filter='irq == 28'
88 ** Wildcard support in the instrumentation point name of an event rule
89 in the Linux kernel tracing domain. For example:
94 lttng enable-event --kernel 'sched_*'
98 ** New `lttng track` and `lttng untrack` commands to make
99 <<pid-tracking,PID tracking>> super-fast for both the Linux kernel
100 and the user space tracing domains.
102 When LTTng _tracks_ one or more PIDs, only the processes having those PIDs
103 can emit events for a given tracing session.
105 ** New `--shm-path` option of the `lttng create` command to specify the
106 path where LTTng creates the shared memory holding the ring buffers.
108 This feature is useful when used with persistent memory file systems to
109 extract the latest recorded trace data in the event of a crash requiring
112 The new man:lttng-crash(1) command-line utility can extract trace data
113 from such a file (see <<persistent-memory-file-systems,Record trace data
114 on persistent memory file systems>>).
116 * **User space tracing**:
117 ** New <<python-application,LTTng-UST Python agent>> which makes it easy
118 to trace existing Python applications that are using the standard
119 https://docs.python.org/3/howto/logging.html[`logging` package].
121 This agent is compatible with both the Python 2 and Python 3 languages.
123 ** New <<tracelog,`tracelog()`>> facility to ease the migration from
126 `tracelog()` is similar to <<tracef,`tracef()`>>,
127 but it accepts an additional log level parameter.
129 ** Plugin support in LTTng-UST to provide a custom clock source and to
130 retrieve the current CPU number.
132 This feature exists for very advanced use cases.
135 https://github.com/lttng/lttng-ust/tree/stable-{revision}/doc/examples/clock-override[clock-override]
137 https://github.com/lttng/lttng-ust/tree/stable-{revision}/doc/examples/getcpu-override[getcpu-override]
138 examples for more details.
140 Moreover, LTTng{nbsp}{revision} boasts great stability, benifiting from
141 piles of bug fixes and more-than-welcome internal refactorings.
143 To learn more about the new features of LTTng{nbsp}{revision}, see
144 https://lttng.org/blog/2015/10/14/lttng-2.7-released/[the release announcement].
150 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
151 generation_ is a modern toolkit for tracing Linux systems and
152 applications. So your first question might be:
159 As the history of software engineering progressed and led to what
160 we now take for granted--complex, numerous and
161 interdependent software applications running in parallel on
162 sophisticated operating systems like Linux--the authors of such
163 components, software developers, began feeling a natural
164 urge to have tools that would ensure the robustness and good performance
165 of their masterpieces.
167 One major achievement in this field is, inarguably, the
168 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
169 an essential tool for developers to find and fix bugs. But even the best
170 debugger won't help make your software run faster, and nowadays, faster
171 software means either more work done by the same hardware, or cheaper
172 hardware for the same work.
174 A _profiler_ is often the tool of choice to identify performance
175 bottlenecks. Profiling is suitable to identify _where_ performance is
176 lost in a given software. The profiler outputs a profile, a statistical
177 summary of observed events, which you may use to discover which
178 functions took the most time to execute. However, a profiler won't
179 report _why_ some identified functions are the bottleneck. Bottlenecks
180 might only occur when specific conditions are met, conditions that are
181 sometimes impossible to capture by a statistical profiler, or impossible
182 to reproduce with an application altered by the overhead of an
183 event-based profiler. For a thorough investigation of software
184 performance issues, a history of execution is essential, with the
185 recorded values of variables and context fields you choose, and
186 with as little influence as possible on the instrumented software. This
187 is where tracing comes in handy.
189 _Tracing_ is a technique used to understand what goes on in a running
190 software system. The software used for tracing is called a _tracer_,
191 which is conceptually similar to a tape recorder. When recording,
192 specific instrumentation points placed in the software source code
193 generate events that are saved on a giant tape: a _trace_ file. You
194 can trace user applications and the operating system at the same time,
195 opening the possibility of resolving a wide range of problems that would
196 otherwise be extremely challenging.
198 Tracing is often compared to _logging_. However, tracers and loggers are
199 two different tools, serving two different purposes. Tracers are
200 designed to record much lower-level events that occur much more
201 frequently than log messages, often in the range of thousands per
202 second, with very little execution overhead. Logging is more appropriate
203 for a very high-level analysis of less frequent events: user accesses,
204 exceptional conditions (errors and warnings, for example), database
205 transactions, instant messaging communications, and such. Simply put,
206 logging is one of the many use cases that can be satisfied with tracing.
208 The list of recorded events inside a trace file can be read manually
209 like a log file for the maximum level of detail, but it is generally
210 much more interesting to perform application-specific analyses to
211 produce reduced statistics and graphs that are useful to resolve a
212 given problem. Trace viewers and analyzers are specialized tools
215 In the end, this is what LTTng is: a powerful, open source set of
216 tools to trace the Linux kernel and user applications at the same time.
217 LTTng is composed of several components actively maintained and
218 developed by its link:/community/#where[community].
221 [[lttng-alternatives]]
222 === Alternatives to noch:{LTTng}
224 Excluding proprietary solutions, a few competing software tracers
227 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
228 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
229 user scripts and is responsible for loading code into the
230 Linux kernel for further execution and collecting the outputted data.
231 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
232 subsystem in the Linux kernel in which a virtual machine can execute
233 programs passed from the user space to the kernel. You can attach
234 such programs to tracepoints and KProbes thanks to a system call, and
235 they can output data to the user space when executed thanks to
236 different mechanisms (pipe, VM register values, and eBPF maps, to name
238 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
239 is the de facto function tracer of the Linux kernel. Its user
240 interface is a set of special files in sysfs.
241 * https://perf.wiki.kernel.org/[perf] is
242 a performance analyzing tool for Linux which supports hardware
243 performance counters, tracepoints, as well as other counters and
244 types of probes. perf's controlling utility is the cmd:perf command
246 * http://linux.die.net/man/1/strace[strace]
247 is a command-line utility which records system calls made by a
248 user process, as well as signal deliveries and changes of process
249 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
250 to fulfill its function.
251 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
252 analyze Linux kernel events. You write scripts, or _chisels_ in
253 sysdig's jargon, in Lua and sysdig executes them while the system is
254 being traced or afterwards. sysdig's interface is the cmd:sysdig
255 command-line tool as well as the curses-based cmd:csysdig tool.
256 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
257 user space tracer which uses custom user scripts to produce plain text
258 traces. SystemTap converts the scripts to the C language, and then
259 compiles them as Linux kernel modules which are loaded to produce
260 trace data. SystemTap's primary user interface is the cmd:stap
263 The main distinctive features of LTTng is that it produces correlated
264 kernel and user space traces, as well as doing so with the lowest
265 overhead amongst other solutions. It produces trace files in the
266 http://diamon.org/ctf[CTF] format, a file format optimized
267 for the production and analyses of multi-gigabyte data.
269 LTTng is the result of more than 10 years of active open source
270 development by a community of passionate developers.
271 LTTng{nbsp}{revision} is currently available on major desktop and server
274 The main interface for tracing control is a single command-line tool
275 named cmd:lttng. The latter can create several tracing sessions, enable
276 and disable events on the fly, filter events efficiently with custom
277 user expressions, start and stop tracing, and much more. LTTng can
278 record the traces on the file system or send them over the network, and
279 keep them totally or partially. You can view the traces once tracing
280 becomes inactive or in real-time.
282 <<installing-lttng,Install LTTng now>> and
283 <<getting-started,start tracing>>!
289 **LTTng** is a set of software <<plumbing,components>> which interact to
290 <<instrumenting,instrument>> the Linux kernel and user applications, and
291 to <<controlling-tracing,control tracing>> (start and stop
292 tracing, enable and disable event rules, and the rest). Those
293 components are bundled into the following packages:
295 * **LTTng-tools**: Libraries and command-line interface to
297 * **LTTng-modules**: Linux kernel modules to instrument and
299 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
300 trace user applications.
302 Most distributions mark the LTTng-modules and LTTng-UST packages as
303 optional when installing LTTng-tools (which is always required). In the
304 following sections, we always provide the steps to install all three,
307 * You only need to install LTTng-modules if you intend to trace the
309 * You only need to install LTTng-UST if you intend to trace user
313 .Availability of LTTng{nbsp}{revision} for major Linux distributions.
315 |Distribution |Available in releases |Alternatives
318 |<<ubuntu,Ubuntu{nbsp}16.04 _Xenial Xerus_>>
319 |LTTng{nbsp}2.8 for Ubuntu{nbsp}16.10 _Yakkety Yak_.
321 LTTng{nbsp}{revision} for Ubuntu{nbsp}12.04 _Precise Pangolin_,
322 Ubuntu{nbsp}14.04 _Trusty Tahr_, and Ubuntu{nbsp}16.04 _Xenial Xerus_:
323 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
325 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
326 other Ubuntu releases.
330 |LTTng{nbsp}{revision} for Fedora{nbsp}25 and Fedora{nbsp}26 (not
333 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
334 other Fedora releases.
338 |LTTng{nbsp}2.8 for Debian "stretch" (testing).
340 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
341 other Debian releases.
344 |<<opensuse,openSUSE Leap{nbsp}42.1>>
345 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
346 other openSUSE releases.
350 |<<building-from-source,Build LTTng{nbsp}{revision} from source>>.
354 |LTTng{nbsp}2.8 for Alpine Linux "edge".
356 LTTng{nbsp}2.8 for Alpine Linux{nbsp}3.5 (not released yet).
358 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
359 other Alpine Linux releases.
362 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
366 |<<"buildroot","Buildroot{nbsp}2016.02, Buildroot{nbsp}2016.05,
367 and Buildroot{nbsp}2016.08">>
368 |LTTng{nbsp}2.8 for Buildroot{nbsp}2016.11 (not released yet).
370 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
371 other Buildroot releases.
373 |OpenEmbedded and Yocto
374 |<<oe-yocto,Yocto Project{nbsp}2.1 _Krogoth_>> (`openembedded-core` layer)
375 |LTTng{nbsp}2.8 for Yocto Project{nbsp}2.2 _Morty_.
377 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
378 other Yocto releases.
383 === [[ubuntu-official-repositories]]Ubuntu
385 LTTng{nbsp}{revision} is available on Ubuntu 16.04 _Xenial Xerus_. For
386 previous releases of Ubuntu, <<ubuntu-ppa,use the LTTng
387 Stable{nbsp}{revision} PPA>>.
389 To install LTTng{nbsp}{revision} on Ubuntu{nbsp}16.04 _Xenial Xerus_:
391 . Install the main LTTng{nbsp}{revision} packages:
396 sudo apt-get install lttng-tools
397 sudo apt-get install lttng-modules-dkms
398 sudo apt-get install liblttng-ust-dev
402 . **If you need to instrument and trace
403 <<java-application,Java applications>>**, install the LTTng-UST
409 sudo apt-get install liblttng-ust-agent-java
413 . **If you need to instrument and trace
414 <<python-application,Python applications>>**, install the
415 LTTng-UST Python agent:
420 sudo apt-get install python3-lttngust
426 ==== noch:{LTTng} Stable {revision} PPA
429 https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng Stable{nbsp}{revision} PPA]
430 offers the latest stable LTTng{nbsp}{revision} packages for:
432 * Ubuntu{nbsp}12.04 _Precise Pangolin_
433 * Ubuntu{nbsp}14.04 _Trusty Tahr_
434 * Ubuntu{nbsp}16.04 _Xenial Xerus_
436 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision}
439 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
445 sudo apt-add-repository ppa:lttng/stable-2.7
450 . Install the main LTTng{nbsp}{revision} packages:
455 sudo apt-get install lttng-tools
456 sudo apt-get install lttng-modules-dkms
457 sudo apt-get install liblttng-ust-dev
461 . **If you need to instrument and trace
462 <<java-application,Java applications>>**, install the LTTng-UST
468 sudo apt-get install liblttng-ust-agent-java
472 . **If you need to instrument and trace
473 <<python-application,Python applications>>**, install the
474 LTTng-UST Python agent:
479 sudo apt-get install python3-lttngust
485 === noch:{openSUSE}/RPM
487 To install LTTng{nbsp}{revision} on openSUSE Leap{nbsp}42.1:
489 * Install the main LTTng{nbsp}{revision} packages:
494 sudo zypper install lttng-tools
495 sudo zypper install lttng-modules
496 sudo zypper install lttng-ust-devel
501 .Java and Python application instrumentation and tracing
503 If you need to instrument and trace <<java-application,Java
504 applications>> on openSUSE, you need to build and install
505 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
506 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
507 `--enable-java-agent-all` options to the `configure` script, depending
508 on which Java logging framework you use.
510 If you need to instrument and trace <<python-application,Python
511 applications>> on openSUSE, you need to build and install
512 LTTng-UST{nbsp}{revision} from source and pass the
513 `--enable-python-agent` option to the `configure` script.
520 To install LTTng{nbsp}{revision} on Buildroot{nbsp}2016.02,
521 Buildroot{nbsp}2016.05, or Buildroot{nbsp}2016.08:
523 . Launch the Buildroot configuration tool:
532 . In **Kernel**, check **Linux kernel**.
533 . In **Toolchain**, check **Enable WCHAR support**.
534 . In **Target packages**{nbsp}→ **Debugging, profiling and benchmark**,
535 check **lttng-modules** and **lttng-tools**.
536 . In **Target packages**{nbsp}→ **Libraries**{nbsp}→
537 **Other**, check **lttng-libust**.
541 === OpenEmbedded and Yocto
543 LTTng{nbsp}{revision} recipes are available in the
544 http://layers.openembedded.org/layerindex/branch/master/layer/openembedded-core/[`openembedded-core`]
545 layer for Yocto Project{nbsp}2.1 _Krogoth_ under the following names:
551 With BitBake, the simplest way to include LTTng recipes in your target
552 image is to add them to `IMAGE_INSTALL_append` in path:{conf/local.conf}:
555 IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"
560 . Select a machine and an image recipe.
561 . Click **Edit image recipe**.
562 . Under the **All recipes** tab, search for **lttng**.
563 . Check the desired LTTng recipes.
566 .Java and Python application instrumentation and tracing
568 If you need to instrument and trace <<java-application,Java
569 applications>> on openSUSE, you need to build and install
570 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
571 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
572 `--enable-java-agent-all` options to the `configure` script, depending
573 on which Java logging framework you use.
575 If you need to instrument and trace <<python-application,Python
576 applications>> on openSUSE, you need to build and install
577 LTTng-UST{nbsp}{revision} from source and pass the
578 `--enable-python-agent` option to the `configure` script.
582 [[enterprise-distributions]]
583 === RHEL, SUSE, and other enterprise distributions
585 To install LTTng on enterprise Linux distributions, such as Red Hat
586 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
587 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
590 [[building-from-source]]
591 === Build from source
593 To build and install LTTng{nbsp}{revision} from source:
595 . Using your distribution's package manager, or from source, install
596 the following dependencies of LTTng-tools and LTTng-UST:
599 * https://sourceforge.net/projects/libuuid/[libuuid]
600 * http://directory.fsf.org/wiki/Popt[popt]
601 * http://liburcu.org/[Userspace RCU]
602 * http://www.xmlsoft.org/[libxml2]
605 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
611 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.7.tar.bz2 &&
612 tar -xf lttng-modules-latest-2.7.tar.bz2 &&
613 cd lttng-modules-2.7.* &&
615 sudo make modules_install &&
620 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
626 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.7.tar.bz2 &&
627 tar -xf lttng-ust-latest-2.7.tar.bz2 &&
628 cd lttng-ust-2.7.* &&
638 .Java and Python application tracing
640 If you need to instrument and trace <<java-application,Java
641 applications>>, pass the `--enable-java-agent-jul`,
642 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
643 `configure` script, depending on which Java logging framework you use.
645 If you need to instrument and trace <<python-application,Python
646 applications>>, pass the `--enable-python-agent` option to the
647 `configure` script. You can set the `PYTHON` environment variable to the
648 path to the Python interpreter for which to install the LTTng-UST Python
656 By default, LTTng-UST libraries are installed to
657 dir:{/usr/local/lib}, which is the de facto directory in which to
658 keep self-compiled and third-party libraries.
660 When <<building-tracepoint-providers-and-user-application,linking an
661 instrumented user application with `liblttng-ust`>>:
663 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
665 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
666 man:gcc(1), man:g++(1), or man:clang(1).
670 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
676 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.7.tar.bz2 &&
677 tar -xf lttng-tools-latest-2.7.tar.bz2 &&
678 cd lttng-tools-2.7.* &&
686 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
687 previous steps automatically for a given version of LTTng and confine
688 the installed files in a specific directory. This can be useful to test
689 LTTng without installing it on your system.
695 This is a short guide to get started quickly with LTTng kernel and user
698 Before you follow this guide, make sure to <<installing-lttng,install>>
701 This tutorial walks you through the steps to:
703 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
704 . <<tracing-your-own-user-application,Trace a user application>> written
706 . <<viewing-and-analyzing-your-traces,View and analyze the
710 [[tracing-the-linux-kernel]]
711 === Trace the Linux kernel
713 The following command lines start with cmd:sudo because you need root
714 privileges to trace the Linux kernel. You can avoid using cmd:sudo if
715 your Unix user is a member of the <<lttng-sessiond,tracing group>>.
717 . Create a <<tracing-session,tracing session>>:
722 sudo lttng create my-kernel-session
726 . List the available kernel tracepoints and system calls:
735 . Create an <<event,event rule>> which matches the desired event names,
736 for example `sched_switch` and `sched_process_fork`:
741 sudo lttng enable-event --kernel sched_switch,sched_process_fork
745 You can also create an event rule which _matches_ all the Linux kernel
746 tracepoints (this will generate a lot of data when tracing):
751 sudo lttng enable-event --kernel --all
764 . Do some operation on your system for a few seconds. For example,
765 load a website, or list the files of a directory.
766 . Stop tracing and destroy the tracing session:
776 The `destroy` command does not destroy the trace data; it only destroys
777 the state of the tracing session.
779 By default, LTTng saves the traces in
780 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
781 where +__name__+ is the tracing session name. Note that the
782 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
784 See <<viewing-and-analyzing-your-traces,View and analyze the
785 recorded events>> to view the recorded events.
788 [[tracing-your-own-user-application]]
789 === Trace a user application
791 This section steps you through a simple example to trace a
792 _Hello world_ program written in C.
794 To create the traceable user application:
796 . Create the tracepoint provider header file, which defines the
797 tracepoints and the events they can generate:
803 #undef TRACEPOINT_PROVIDER
804 #define TRACEPOINT_PROVIDER hello_world
806 #undef TRACEPOINT_INCLUDE
807 #define TRACEPOINT_INCLUDE "./hello-tp.h"
809 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
812 #include <lttng/tracepoint.h>
822 ctf_string(my_string_field, my_string_arg)
823 ctf_integer(int, my_integer_field, my_integer_arg)
827 #endif /* _HELLO_TP_H */
829 #include <lttng/tracepoint-event.h>
833 . Create the tracepoint provider package source file:
839 #define TRACEPOINT_CREATE_PROBES
840 #define TRACEPOINT_DEFINE
842 #include "hello-tp.h"
846 . Build the tracepoint provider package:
851 gcc -c -I. hello-tp.c
855 . Create the _Hello World_ application source file:
862 #include "hello-tp.h"
864 int main(int argc, char *argv[])
868 puts("Hello, World!\nPress Enter to continue...");
871 * The following getchar() call is only placed here for the purpose
872 * of this demonstration, to pause the application in order for
873 * you to have time to list its tracepoints. It is not
879 * A tracepoint() call.
881 * Arguments, as defined in hello-tp.h:
883 * 1. Tracepoint provider name (required)
884 * 2. Tracepoint name (required)
885 * 3. my_integer_arg (first user-defined argument)
886 * 4. my_string_arg (second user-defined argument)
888 * Notice the tracepoint provider and tracepoint names are
889 * NOT strings: they are in fact parts of variables that the
890 * macros in hello-tp.h create.
892 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
894 for (x = 0; x < argc; ++x) {
895 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
898 puts("Quitting now!");
899 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
906 . Build the application:
915 . Link the application with the tracepoint provider package,
916 `liblttng-ust`, and `libdl`:
921 gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
925 Here's the whole build process:
928 .User space tracing tutorial's build steps.
929 image::ust-flow.png[]
931 To trace the user application:
933 . Run the application with a few arguments:
938 ./hello world and beyond
947 Press Enter to continue...
951 . Start an LTTng <<lttng-sessiond,session daemon>>:
956 lttng-sessiond --daemonize
960 Note that a session daemon might already be running, for example as
961 a service that the distribution's service manager started.
963 . List the available user space tracepoints:
968 lttng list --userspace
972 You see the `hello_world:my_first_tracepoint` tracepoint listed
973 under the `./hello` process.
975 . Create a <<tracing-session,tracing session>>:
980 lttng create my-user-space-session
984 . Create an <<event,event rule>> which matches the
985 `hello_world:my_first_tracepoint` event name:
990 lttng enable-event --userspace hello_world:my_first_tracepoint
1003 . Go back to the running `hello` application and press Enter. The
1004 program executes all `tracepoint()` instrumentation points and exits.
1005 . Stop tracing and destroy the tracing session:
1015 The `destroy` command does not destroy the trace data; it only destroys
1016 the state of the tracing session.
1018 By default, LTTng saves the traces in
1019 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1020 where +__name__+ is the tracing session name. Note that the
1021 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1023 See <<viewing-and-analyzing-your-traces,View and analyze the
1024 recorded events>> to view the recorded events.
1027 [[viewing-and-analyzing-your-traces]]
1028 === View and analyze the recorded events
1030 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1031 kernel>> and <<tracing-your-own-user-application,Trace a user
1032 application>> tutorials, you can inspect the recorded events.
1034 Many tools are available to read LTTng traces:
1036 * **cmd:babeltrace** is a command-line utility which converts trace
1037 formats; it supports the format that LTTng produces, CTF, as well as a
1038 basic text output which can be ++grep++ed. The cmd:babeltrace command
1039 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1040 * Babeltrace also includes
1041 **https://www.python.org/[Python] bindings** so
1042 that you can easily open and read an LTTng trace with your own script,
1043 benefiting from the power of Python.
1044 * http://tracecompass.org/[**Trace Compass**]
1045 is a graphical user interface for viewing and analyzing any type of
1046 logs or traces, including LTTng's.
1047 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1048 project which includes many high-level analyses of LTTng kernel
1049 traces, like scheduling statistics, interrupt frequency distribution,
1050 top CPU usage, and more.
1052 NOTE: This section assumes that the traces recorded during the previous
1053 tutorials were saved to their default location, in the
1054 dir:{$LTTNG_HOME/lttng-traces} directory. Note that the env:LTTNG_HOME
1055 environment variable defaults to `$HOME` if not set.
1058 [[viewing-and-analyzing-your-traces-bt]]
1059 ==== Use the cmd:babeltrace command-line tool
1061 The simplest way to list all the recorded events of a trace is to pass
1062 its path to cmd:babeltrace with no options:
1066 babeltrace ~/lttng-traces/my-user-space-session*
1069 cmd:babeltrace finds all traces recursively within the given path and
1070 prints all their events, merging them in chronological order.
1072 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1077 babeltrace ~/lttng-traces/my-kernel-session* | grep sys_
1080 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1081 count the recorded events:
1085 babeltrace ~/lttng-traces/my-kernel-session* | grep sys_read | wc --lines
1089 [[viewing-and-analyzing-your-traces-bt-python]]
1090 ==== Use the Babeltrace Python bindings
1092 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1093 is useful to isolate events by simple matching using man:grep(1) and
1094 similar utilities. However, more elaborate filters, such as keeping only
1095 event records with a field value falling within a specific range, are
1096 not trivial to write using a shell. Moreover, reductions and even the
1097 most basic computations involving multiple event records are virtually
1098 impossible to implement.
1100 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1101 to read the event records of an LTTng trace sequentially and compute the
1102 desired information.
1104 The following script accepts an LTTng Linux kernel trace path as its
1105 first argument and prints the short names of the top 5 running processes
1106 on CPU 0 during the whole trace:
1111 from collections import Counter
1117 if len(sys.argv) != 2:
1118 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1119 print(msg, file=sys.stderr)
1122 # A trace collection contains one or more traces
1123 col = babeltrace.TraceCollection()
1125 # Add the trace provided by the user (LTTng traces always have
1127 if col.add_trace(sys.argv[1], 'ctf') is None:
1128 raise RuntimeError('Cannot add trace')
1130 # This counter dict contains execution times:
1132 # task command name -> total execution time (ns)
1133 exec_times = Counter()
1135 # This contains the last `sched_switch` timestamp
1139 for event in col.events:
1140 # Keep only `sched_switch` events
1141 if event.name != 'sched_switch':
1144 # Keep only events which happened on CPU 0
1145 if event['cpu_id'] != 0:
1149 cur_ts = event.timestamp
1155 # Previous task command (short) name
1156 prev_comm = event['prev_comm']
1158 # Initialize entry in our dict if not yet done
1159 if prev_comm not in exec_times:
1160 exec_times[prev_comm] = 0
1162 # Compute previous command execution time
1163 diff = cur_ts - last_ts
1165 # Update execution time of this command
1166 exec_times[prev_comm] += diff
1168 # Update last timestamp
1172 for name, ns in exec_times.most_common(5):
1174 print('{:20}{} s'.format(name, s))
1179 if __name__ == '__main__':
1180 sys.exit(0 if top5proc() else 1)
1187 python3 top5proc.py ~/lttng-traces/my-kernel-session*/kernel
1193 swapper/0 48.607245889 s
1194 chromium 7.192738188 s
1195 pavucontrol 0.709894415 s
1196 Compositor 0.660867933 s
1197 Xorg.bin 0.616753786 s
1200 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1201 weren't using the CPU that much when tracing, its first position in the
1206 == [[understanding-lttng]]Core concepts
1208 From a user's perspective, the LTTng system is built on a few concepts,
1209 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1210 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1211 Understanding how those objects relate to eachother is key in mastering
1214 The core concepts are:
1216 * <<tracing-session,Tracing session>>
1217 * <<domain,Tracing domain>>
1218 * <<channel,Channel and ring buffer>>
1219 * <<"event","Instrumentation point, event rule, event, and event record">>
1225 A _tracing session_ is a stateful dialogue between you and
1226 a <<lttng-sessiond,session daemon>>. You can
1227 <<creating-destroying-tracing-sessions,create a new tracing
1228 session>> with the `lttng create` command.
1230 Anything that you do when you control LTTng tracers happens within a
1231 tracing session. In particular, a tracing session:
1234 * Has its own set of trace files.
1235 * Has its own state of activity (started or stopped).
1236 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1238 * Has its own <<channel,channels>> which have their own
1239 <<event,event rules>>.
1242 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1243 image::concepts.png[]
1245 Those attributes and objects are completely isolated between different
1248 A tracing session is analogous to a cash machine session:
1249 the operations you do on the banking system through the cash machine do
1250 not alter the data of other users of the same system. In the case of
1251 the cash machine, a session lasts as long as your bank card is inside.
1252 In the case of LTTng, a tracing session lasts from the `lttng create`
1253 command to the `lttng destroy` command.
1256 .Each Unix user has its own set of tracing sessions.
1257 image::many-sessions.png[]
1260 [[tracing-session-mode]]
1261 ==== Tracing session mode
1263 LTTng can send the generated trace data to different locations. The
1264 _tracing session mode_ dictates where to send it. The following modes
1265 are available in LTTng{nbsp}{revision}:
1268 LTTng writes the traces to the file system of the machine being traced
1271 Network streaming mode::
1272 LTTng sends the traces over the network to a
1273 <<lttng-relayd,relay daemon>> running on a remote system.
1276 LTTng does not write the traces by default. Instead, you can request
1277 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1278 current tracing buffers, and to write it to the target's file system
1279 or to send it over the network to a <<lttng-relayd,relay daemon>>
1280 running on a remote system.
1283 This mode is similar to the network streaming mode, but a live
1284 trace viewer can connect to the distant relay daemon to
1285 <<lttng-live,view event records as LTTng generates them>> by
1292 A _tracing domain_ is a namespace for event sources. A tracing domain
1293 has its own properties and features.
1295 There are currently five available tracing domains:
1299 * `java.util.logging` (JUL)
1303 You must specify a tracing domain when using some commands to avoid
1304 ambiguity. For example, since all the domains support named tracepoints
1305 as event sources (instrumentation points that you manually insert in the
1306 source code), you need to specify a tracing domain when
1307 <<enabling-disabling-events,creating an event rule>> because all the
1308 tracing domains could have tracepoints with the same names.
1310 Some features are reserved to specific tracing domains. Dynamic function
1311 entry and return instrumentation points, for example, are currently only
1312 supported in the Linux kernel tracing domain, but support for other
1313 tracing domains could be added in the future.
1315 You can create <<channel,channels>> in the Linux kernel and user space
1316 tracing domains. The other tracing domains have a single default
1321 === Channel and ring buffer
1323 A _channel_ is an object which is responsible for a set of ring buffers.
1324 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1325 tracer emits an event, it can record it to one or more
1326 sub-buffers. The attributes of a channel determine what to do when
1327 there's no space left for a new event record because all sub-buffers
1328 are full, where to send a full sub-buffer, and other behaviours.
1330 A channel is always associated to a <<domain,tracing domain>>. The
1331 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1332 a default channel which you cannot configure.
1334 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1335 an event, it records it to the sub-buffers of all
1336 the enabled channels with a satisfied event rule, as long as those
1337 channels are part of active <<tracing-session,tracing sessions>>.
1340 [[channel-buffering-schemes]]
1341 ==== Per-user vs. per-process buffering schemes
1343 A channel has at least one ring buffer _per CPU_. LTTng always
1344 records an event to the ring buffer associated to the CPU on which it
1347 Two _buffering schemes_ are available when you
1348 <<enabling-disabling-channels,create a channel>> in the
1349 user space <<domain,tracing domain>>:
1351 Per-user buffering::
1352 Allocate one set of ring buffers--one per CPU--shared by all the
1353 instrumented processes of each Unix user.
1357 .Per-user buffering scheme.
1358 image::per-user-buffering.png[]
1361 Per-process buffering::
1362 Allocate one set of ring buffers--one per CPU--for each
1363 instrumented process.
1367 .Per-process buffering scheme.
1368 image::per-process-buffering.png[]
1371 The per-process buffering scheme tends to consume more memory than the
1372 per-user option because systems generally have more instrumented
1373 processes than Unix users running instrumented processes. However, the
1374 per-process buffering scheme ensures that one process having a high
1375 event throughput won't fill all the shared sub-buffers of the same
1378 The Linux kernel tracing domain has only one available buffering scheme
1379 which is to allocate a single set of ring buffers for the whole system.
1380 This scheme is similar to the per-user option, but with a single, global
1381 user "running" the kernel.
1384 [[channel-overwrite-mode-vs-discard-mode]]
1385 ==== Overwrite vs. discard event loss modes
1387 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1388 arc in the following animation) of a specific channel's ring buffer.
1389 When there's no space left in a sub-buffer, the tracer marks it as
1390 consumable (red) and another, empty sub-buffer starts receiving the
1391 following event records. A <<lttng-consumerd,consumer daemon>>
1392 eventually consumes the marked sub-buffer (returns to white).
1395 [role="docsvg-channel-subbuf-anim"]
1400 In an ideal world, sub-buffers are consumed faster than they are filled,
1401 as is the case in the previous animation. In the real world,
1402 however, all sub-buffers can be full at some point, leaving no space to
1403 record the following events.
1405 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1406 available, it is acceptable to lose event records when the alternative
1407 would be to cause substantial delays in the instrumented application's
1408 execution. LTTng privileges performance over integrity; it aims at
1409 perturbing the traced system as little as possible in order to make
1410 tracing of subtle race conditions and rare interrupt cascades possible.
1412 When it comes to losing event records because no empty sub-buffer is
1413 available, the channel's _event loss mode_ determines what to do. The
1414 available event loss modes are:
1417 Drop the newest event records until a the tracer
1418 releases a sub-buffer.
1421 Clear the sub-buffer containing the oldest event records and start
1422 writing the newest event records there.
1424 This mode is sometimes called _flight recorder mode_ because it's
1426 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1427 always keep a fixed amount of the latest data.
1429 Which mechanism you should choose depends on your context: prioritize
1430 the newest or the oldest event records in the ring buffer?
1432 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1433 as soon as a there's no space left for a new event record, whereas in
1434 discard mode, the tracer only discards the event record that doesn't
1437 In discard mode, LTTng increments a count of lost event records when
1438 an event record is lost and saves this count to the trace. In
1439 overwrite mode, LTTng keeps no information when it overwrites a
1440 sub-buffer before consuming it.
1442 There are a few ways to decrease your probability of losing event
1444 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1445 how you can fine-une the sub-buffer count and size of a channel to
1446 virtually stop losing event records, though at the cost of greater
1450 [[channel-subbuf-size-vs-subbuf-count]]
1451 ==== Sub-buffer count and size
1453 When you <<enabling-disabling-channels,create a channel>>, you can
1454 set its number of sub-buffers and their size.
1456 Note that there is noticeable CPU overhead introduced when
1457 switching sub-buffers (marking a full one as consumable and switching
1458 to an empty one for the following events to be recorded). Knowing this,
1459 the following list presents a few practical situations along with how
1460 to configure the sub-buffer count and size for them:
1462 * **High event throughput**: In general, prefer bigger sub-buffers to
1463 lower the risk of losing event records.
1465 Having bigger sub-buffers also ensures a lower sub-buffer switching
1468 The number of sub-buffers is only meaningful if you create the channel
1469 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1470 other sub-buffers are left unaltered.
1472 * **Low event throughput**: In general, prefer smaller sub-buffers
1473 since the risk of losing event records is low.
1475 Because events occur less frequently, the sub-buffer switching frequency
1476 should remain low and thus the tracer's overhead should not be a
1479 * **Low memory system**: If your target system has a low memory
1480 limit, prefer fewer first, then smaller sub-buffers.
1482 Even if the system is limited in memory, you want to keep the
1483 sub-buffers as big as possible to avoid a high sub-buffer switching
1486 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1487 which means event data is very compact. For example, the average
1488 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1489 sub-buffer size of 1{nbsp}MiB is considered big.
1491 The previous situations highlight the major trade-off between a few big
1492 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1493 frequency vs. how much data is lost in overwrite mode. Assuming a
1494 constant event throughput and using the overwrite mode, the two
1495 following configurations have the same ring buffer total size:
1498 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1503 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1504 switching frequency, but if a sub-buffer overwrite happens, half of
1505 the event records so far (4{nbsp}MiB) are definitely lost.
1506 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1507 overhead as the previous configuration, but if a sub-buffer
1508 overwrite happens, only the eighth of event records so far are
1511 In discard mode, the sub-buffers count parameter is pointless: use two
1512 sub-buffers and set their size according to the requirements of your
1516 [[channel-switch-timer]]
1517 ==== Switch timer period
1519 The _switch timer period_ is an important configurable attribute of
1520 a channel to ensure periodic sub-buffer flushing.
1522 When the _switch timer_ expires, a sub-buffer switch happens. You can
1523 set the switch timer period attribute when you
1524 <<enabling-disabling-channels,create a channel>> to ensure that event
1525 data is consumed and committed to trace files or to a distant relay
1526 daemon periodically in case of a low event throughput.
1529 [role="docsvg-channel-switch-timer"]
1534 This attribute is also convenient when you use big sub-buffers to cope
1535 with a sporadic high event throughput, even if the throughput is
1539 [[channel-read-timer]]
1540 ==== Read timer period
1542 By default, the LTTng tracers use a notification mechanism to signal a
1543 full sub-buffer so that a consumer daemon can consume it. When such
1544 notifications must be avoided, for example in real-time applications,
1545 you can use the channel's _read timer_ instead. When the read timer
1546 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1547 consumable sub-buffers.
1550 [[tracefile-rotation]]
1551 ==== Trace file count and size
1553 By default, trace files can grow as large as needed. You can set the
1554 maximum size of each trace file that a channel writes when you
1555 <<enabling-disabling-channels,create a channel>>. When the size of
1556 a trace file reaches the channel's fixed maximum size, LTTng creates
1557 another file to contain the next event records. LTTng appends a file
1558 count to each trace file name in this case.
1560 If you set the trace file size attribute when you create a channel, the
1561 maximum number of trace files that LTTng creates is _unlimited_ by
1562 default. To limit them, you can also set a maximum number of trace
1563 files. When the number of trace files reaches the channel's fixed
1564 maximum count, the oldest trace file is overwritten. This mechanism is
1565 called _trace file rotation_.
1569 === Instrumentation point, event rule, event, and event record
1571 An _event rule_ is a set of conditions which must be **all** satisfied
1572 for LTTng to record an occuring event.
1574 You set the conditions when you <<enabling-disabling-events,create
1577 You always attach an event rule to <<channel,channel>> when you create
1580 When an event passes the conditions of an event rule, LTTng records it
1581 in one of the attached channel's sub-buffers.
1583 The available conditions, as of LTTng{nbsp}{revision}, are:
1585 * The event rule _is enabled_.
1586 * The instrumentation point's type _is{nbsp}T_.
1587 * The instrumentation point's name (sometimes called _event name_)
1588 _matches{nbsp}N_, but _is not{nbsp}E_.
1589 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1590 _is exactly{nbsp}L_.
1591 * The fields of the event's payload _satisfy_ a filter
1592 expression{nbsp}__F__.
1594 As you can see, all the conditions but the dynamic filter are related to
1595 the event rule's status or to the instrumentation point, not to the
1596 occurring events. This is why, without a filter, checking if an event
1597 passes an event rule is not a dynamic task: when you create or modify an
1598 event rule, all the tracers of its tracing domain enable or disable the
1599 instrumentation points themselves once. This is possible because the
1600 attributes of an instrumentation point (type, name, and log level) are
1601 defined statically. In other words, without a dynamic filter, the tracer
1602 _does not evaluate_ the arguments of an instrumentation point unless it
1603 matches an enabled event rule.
1605 Note that, for LTTng to record an event, the <<channel,channel>> to
1606 which a matching event rule is attached must also be enabled, and the
1607 tracing session owning this channel must be active.
1610 .Logical path from an instrumentation point to an event record.
1611 image::event-rule.png[]
1613 .Event, event record, or event rule?
1615 With so many similar terms, it's easy to get confused.
1617 An **event** is the consequence of the execution of an _instrumentation
1618 point_, like a tracepoint that you manually place in some source code,
1619 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1620 time. Different actions can be taken upon the occurance of an event,
1621 like record the event's payload to a buffer.
1623 An **event record** is the representation of an event in a sub-buffer. A
1624 tracer is responsible for capturing the payload of an event, current
1625 context variables, the event's ID, and the event's timestamp. LTTng
1626 can append this sub-buffer to a trace file.
1628 An **event rule** is a set of conditions which must all be satisfied for
1629 LTTng to record an occuring event. Events still occur without
1630 satisfying event rules, but LTTng does not record them.
1635 == Components of noch:{LTTng}
1637 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1638 to call LTTng a simple _tool_ since it is composed of multiple
1639 interacting components. This section describes those components,
1640 explains their respective roles, and shows how they connect together to
1641 form the LTTng ecosystem.
1643 The following diagram shows how the most important components of LTTng
1644 interact with user applications, the Linux kernel, and you:
1647 .Control and trace data paths between LTTng components.
1648 image::plumbing.png[]
1650 The LTTng project incorporates:
1652 * **LTTng-tools**: Libraries and command-line interface to
1653 control tracing sessions.
1654 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1655 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1656 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1657 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1658 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1659 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1661 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1662 headers to instrument and trace any native user application.
1663 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1664 *** `liblttng-ust-libc-wrapper`
1665 *** `liblttng-ust-pthread-wrapper`
1666 *** `liblttng-ust-cyg-profile`
1667 *** `liblttng-ust-cyg-profile-fast`
1668 *** `liblttng-ust-dl`
1669 ** User space tracepoint provider source files generator command-line
1670 tool (man:lttng-gen-tp(1)).
1671 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1672 Java applications using `java.util.logging` or
1673 Apache log4j 1.2 logging.
1674 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1675 Python applications using the standard `logging` package.
1676 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1678 ** LTTng kernel tracer module.
1679 ** Tracing ring buffer kernel modules.
1680 ** Probe kernel modules.
1681 ** LTTng logger kernel module.
1685 === Tracing control command-line interface
1688 .The tracing control command-line interface.
1689 image::plumbing-lttng-cli.png[]
1691 The _man:lttng(1) command-line tool_ is the standard user interface to
1692 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1693 is part of LTTng-tools.
1695 The cmd:lttng tool is linked with
1696 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1697 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1699 The cmd:lttng tool has a Git-like interface:
1703 lttng <general options> <command> <command options>
1706 The <<controlling-tracing,Tracing control>> section explores the
1707 available features of LTTng using the cmd:lttng tool.
1710 [[liblttng-ctl-lttng]]
1711 === Tracing control library
1714 .The tracing control library.
1715 image::plumbing-liblttng-ctl.png[]
1717 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1718 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1719 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1721 The <<lttng-cli,cmd:lttng command-line tool>>
1722 is linked with `liblttng-ctl`.
1724 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1729 #include <lttng/lttng.h>
1732 Some objects are referenced by name (C string), such as tracing
1733 sessions, but most of them require to create a handle first using
1734 `lttng_create_handle()`.
1736 The best available developer documentation for `liblttng-ctl` is, as of
1737 LTTng{nbsp}{revision}, its installed header files. Every function and
1738 structure is thoroughly documented.
1742 === User space tracing library
1745 .The user space tracing library.
1746 image::plumbing-liblttng-ust.png[]
1748 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1749 is the LTTng user space tracer. It receives commands from a
1750 <<lttng-sessiond,session daemon>>, for example to
1751 enable and disable specific instrumentation points, and writes event
1752 records to ring buffers shared with a
1753 <<lttng-consumerd,consumer daemon>>.
1754 `liblttng-ust` is part of LTTng-UST.
1756 Public C header files are installed beside `liblttng-ust` to
1757 instrument any <<c-application,C or $$C++$$ application>>.
1759 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1760 packages, use their own library providing tracepoints which is
1761 linked with `liblttng-ust`.
1763 An application or library does not have to initialize `liblttng-ust`
1764 manually: its constructor does the necessary tasks to properly register
1765 to a session daemon. The initialization phase also enables the
1766 instrumentation points matching the <<event,event rules>> that you
1770 [[lttng-ust-agents]]
1771 === User space tracing agents
1774 .The user space tracing agents.
1775 image::plumbing-lttng-ust-agents.png[]
1777 The _LTTng-UST Java and Python agents_ are regular Java and Python
1778 packages which add LTTng tracing capabilities to the
1779 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1781 In the case of Java, the
1782 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1783 core logging facilities] and
1784 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1785 Note that Apache Log4{nbsp}2 is not supported.
1787 In the case of Python, the standard
1788 https://docs.python.org/3/library/logging.html[`logging`] package
1789 is supported. Both Python 2 and Python 3 modules can import the
1790 LTTng-UST Python agent package.
1792 The applications using the LTTng-UST agents are in the
1793 `java.util.logging` (JUL),
1794 log4j, and Python <<domain,tracing domains>>.
1796 Both agents use the same mechanism to trace the log statements. When an
1797 agent is initialized, it creates a log handler that attaches to the root
1798 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1799 When the application executes a log statement, it is passed to the
1800 agent's log handler by the root logger. The agent's log handler calls a
1801 native function in a tracepoint provider package shared library linked
1802 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1803 other fields, like its logger name and its log level. This native
1804 function contains a user space instrumentation point, hence tracing the
1807 The log level condition of an
1808 <<event,event rule>> is considered when tracing
1809 a Java or a Python application, and it's compatible with the standard
1810 JUL, log4j, and Python log levels.
1814 === LTTng kernel modules
1817 .The LTTng kernel modules.
1818 image::plumbing-lttng-modules.png[]
1820 The _LTTng kernel modules_ are a set of Linux kernel modules
1821 which implement the kernel tracer of the LTTng project. The LTTng
1822 kernel modules are part of LTTng-modules.
1824 The LTTng kernel modules include:
1826 * A set of _probe_ modules.
1828 Each module attaches to a specific subsystem
1829 of the Linux kernel using its tracepoint instrument points. There are
1830 also modules to attach to the entry and return points of the Linux
1831 system call functions.
1833 * _Ring buffer_ modules.
1835 A ring buffer implementation is provided as kernel modules. The LTTng
1836 kernel tracer writes to the ring buffer; a
1837 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1839 * The _LTTng kernel tracer_ module.
1840 * The _LTTng logger_ module.
1842 The LTTng logger module implements the special path:{/proc/lttng-logger}
1843 file so that any executable can generate LTTng events by opening and
1844 writing to this file.
1846 See <<proc-lttng-logger-abi,LTTng logger>>.
1848 Generally, you do not have to load the LTTng kernel modules manually
1849 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1850 daemon>> loads the necessary modules when starting. If you have extra
1851 probe modules, you can specify to load them to the session daemon on
1854 The LTTng kernel modules are installed in
1855 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1856 the kernel release (see `uname --kernel-release`).
1863 .The session daemon.
1864 image::plumbing-sessiond.png[]
1866 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1867 managing tracing sessions and for controlling the various components of
1868 LTTng. The session daemon is part of LTTng-tools.
1870 The session daemon sends control requests to and receives control
1873 * The <<lttng-ust,user space tracing library>>.
1875 Any instance of the user space tracing library first registers to
1876 a session daemon. Then, the session daemon can send requests to
1877 this instance, such as:
1880 ** Get the list of tracepoints.
1881 ** Share an <<event,event rule>> so that the user space tracing library
1882 can enable or disable tracepoints. Amongst the possible conditions
1883 of an event rule is a filter expression which `liblttng-ust` evalutes
1884 when an event occurs.
1885 ** Share <<channel,channel>> attributes and ring buffer locations.
1888 The session daemon and the user space tracing library use a Unix
1889 domain socket for their communication.
1891 * The <<lttng-ust-agents,user space tracing agents>>.
1893 Any instance of a user space tracing agent first registers to
1894 a session daemon. Then, the session daemon can send requests to
1895 this instance, such as:
1898 ** Get the list of loggers.
1899 ** Enable or disable a specific logger.
1902 The session daemon and the user space tracing agent use a TCP connection
1903 for their communication.
1905 * The <<lttng-modules,LTTng kernel tracer>>.
1906 * The <<lttng-consumerd,consumer daemon>>.
1908 The session daemon sends requests to the consumer daemon to instruct
1909 it where to send the trace data streams, amongst other information.
1911 * The <<lttng-relayd,relay daemon>>.
1913 The session daemon receives commands from the
1914 <<liblttng-ctl-lttng,tracing control library>>.
1916 The root session daemon loads the appropriate
1917 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
1918 a <<lttng-consumerd,consumer daemon>> as soon as you create
1919 an <<event,event rule>>.
1921 The session daemon does not send and receive trace data: this is the
1922 role of the <<lttng-consumerd,consumer daemon>> and
1923 <<lttng-relayd,relay daemon>>. It does, however, generate the
1924 http://diamon.org/ctf/[CTF] metadata stream.
1926 Each Unix user can have its own session daemon instance. The
1927 tracing sessions managed by different session daemons are completely
1930 The root user's session daemon is the only one which is
1931 allowed to control the LTTng kernel tracer, and its spawned consumer
1932 daemon is the only one which is allowed to consume trace data from the
1933 LTTng kernel tracer. Note, however, that any Unix user which is a member
1934 of the <<tracing-group,tracing group>> is allowed
1935 to create <<channel,channels>> in the
1936 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
1939 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
1940 session daemon when using its `create` command if none is currently
1941 running. You can also start the session daemon manually.
1948 .The consumer daemon.
1949 image::plumbing-consumerd.png[]
1951 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
1952 ring buffers with user applications or with the LTTng kernel modules to
1953 collect trace data and send it to some location (on disk or to a
1954 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
1955 is part of LTTng-tools.
1957 You do not start a consumer daemon manually: a consumer daemon is always
1958 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
1959 <<event,event rule>>, that is, before you start tracing. When you kill
1960 its owner session daemon, the consumer daemon also exits because it is
1961 the session daemon's child process. Command-line options of
1962 man:lttng-sessiond(8) target the consumer daemon process.
1964 There are up to two running consumer daemons per Unix user, whereas only
1965 one session daemon can run per user. This is because each process can be
1966 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
1967 and 64-bit processes, it is more efficient to have separate
1968 corresponding 32-bit and 64-bit consumer daemons. The root user is an
1969 exception: it can have up to _three_ running consumer daemons: 32-bit
1970 and 64-bit instances for its user applications, and one more
1971 reserved for collecting kernel trace data.
1979 image::plumbing-relayd.png[]
1981 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
1982 between remote session and consumer daemons, local trace files, and a
1983 remote live trace viewer. The relay daemon is part of LTTng-tools.
1985 The main purpose of the relay daemon is to implement a receiver of
1986 <<sending-trace-data-over-the-network,trace data over the network>>.
1987 This is useful when the target system does not have much file system
1988 space to record trace files locally.
1990 The relay daemon is also a server to which a
1991 <<lttng-live,live trace viewer>> can
1992 connect. The live trace viewer sends requests to the relay daemon to
1993 receive trace data as the target system emits events. The
1994 communication protocol is named _LTTng live_; it is used over TCP
1997 Note that you can start the relay daemon on the target system directly.
1998 This is the setup of choice when the use case is to view events as
1999 the target system emits them without the need of a remote system.
2003 == [[using-lttng]]Instrumentation
2005 There are many examples of tracing and monitoring in our everyday life:
2007 * You have access to real-time and historical weather reports and
2008 forecasts thanks to weather stations installed around the country.
2009 * You know your heart is safe thanks to an electrocardiogram.
2010 * You make sure not to drive your car too fast and to have enough fuel
2011 to reach your destination thanks to gauges visible on your dashboard.
2013 All the previous examples have something in common: they rely on
2014 **instruments**. Without the electrodes attached to the surface of your
2015 body's skin, cardiac monitoring is futile.
2017 LTTng, as a tracer, is no different from those real life examples. If
2018 you're about to trace a software system or, in other words, record its
2019 history of execution, you better have **instrumentation points** in the
2020 subject you're tracing, that is, the actual software.
2022 Various ways were developed to instrument a piece of software for LTTng
2023 tracing. The most straightforward one is to manually place
2024 instrumentation points, called _tracepoints_, in the software's source
2025 code. It is also possible to add instrumentation points dynamically in
2026 the Linux kernel <<domain,tracing domain>>.
2028 If you're only interested in tracing the Linux kernel, your
2029 instrumentation needs are probably already covered by LTTng's built-in
2030 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2031 user application which is already instrumented for LTTng tracing.
2032 In such cases, you can skip this whole section and read the topics of
2033 the <<controlling-tracing,Tracing control>> section.
2035 Many methods are available to instrument a piece of software for LTTng
2038 * <<c-application,User space instrumentation for C and $$C++$$
2040 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2041 * <<java-application,User space Java agent>>.
2042 * <<python-application,User space Python agent>>.
2043 * <<proc-lttng-logger-abi,LTTng logger>>.
2044 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2048 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2050 The procedure to instrument a C or $$C++$$ user application with
2051 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2053 . <<tracepoint-provider,Create the source files of a tracepoint provider
2055 . <<probing-the-application-source-code,Add tracepoints to
2056 the application's source code>>.
2057 . <<building-tracepoint-providers-and-user-application,Build and link
2058 a tracepoint provider package and the user application>>.
2060 If you need quick, man:printf(3)-like instrumentation, you can skip
2061 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2064 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2065 instrument a user application with `liblttng-ust`.
2068 [[tracepoint-provider]]
2069 ==== Create the source files of a tracepoint provider package
2071 A _tracepoint provider_ is a set of compiled functions which provide
2072 **tracepoints** to an application, the type of instrumentation point
2073 supported by LTTng-UST. Those functions can emit events with
2074 user-defined fields and serialize those events as event records to one
2075 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2076 macro, which you <<probing-the-application-source-code,insert in a user
2077 application's source code>>, calls those functions.
2079 A _tracepoint provider package_ is an object file (`.o`) or a shared
2080 library (`.so`) which contains one or more tracepoint providers.
2081 Its source files are:
2083 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2084 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2086 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2087 the LTTng user space tracer, at run time.
2090 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2091 image::ust-app.png[]
2093 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2094 skip creating and using a tracepoint provider and use
2095 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2099 ===== Create a tracepoint provider header file template
2101 A _tracepoint provider header file_ contains the tracepoint
2102 definitions of a tracepoint provider.
2104 To create a tracepoint provider header file:
2106 . Start from this template:
2110 .Tracepoint provider header file template (`.h` file extension).
2112 #undef TRACEPOINT_PROVIDER
2113 #define TRACEPOINT_PROVIDER provider_name
2115 #undef TRACEPOINT_INCLUDE
2116 #define TRACEPOINT_INCLUDE "./tp.h"
2118 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2121 #include <lttng/tracepoint.h>
2124 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2125 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2130 #include <lttng/tracepoint-event.h>
2136 * `provider_name` with the name of your tracepoint provider.
2137 * `"tp.h"` with the name of your tracepoint provider header file.
2139 . Below the `#include <lttng/tracepoint.h>` line, put your
2140 <<defining-tracepoints,tracepoint definitions>>.
2142 Your tracepoint provider name must be unique amongst all the possible
2143 tracepoint provider names used on the same target system. We
2144 suggest to include the name of your project or company in the name,
2145 for example, `org_lttng_my_project_tpp`.
2147 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2148 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2149 write are the <<defining-tracepoints,tracepoint definitions>>.
2152 [[defining-tracepoints]]
2153 ===== Create a tracepoint definition
2155 A _tracepoint definition_ defines, for a given tracepoint:
2157 * Its **input arguments**. They are the macro parameters that the
2158 `tracepoint()` macro accepts for this particular tracepoint
2159 in the user application's source code.
2160 * Its **output event fields**. They are the sources of event fields
2161 that form the payload of any event that the execution of the
2162 `tracepoint()` macro emits for this particular tracepoint.
2164 You can create a tracepoint definition by using the
2165 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2167 <<tpp-header,tracepoint provider header file template>>.
2169 The syntax of the `TRACEPOINT_EVENT()` macro is:
2172 .`TRACEPOINT_EVENT()` macro syntax.
2175 /* Tracepoint provider name */
2178 /* Tracepoint name */
2181 /* Input arguments */
2186 /* Output event fields */
2195 * `provider_name` with your tracepoint provider name.
2196 * `tracepoint_name` with your tracepoint name.
2197 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2198 * `fields` with the <<tpp-def-output-fields,output event field>>
2201 This tracepoint emits events named `provider_name:tracepoint_name`.
2204 .Event name's length limitation
2206 The concatenation of the tracepoint provider name and the
2207 tracepoint name must not exceed **254 characters**. If it does, the
2208 instrumented application compiles and runs, but LTTng throws multiple
2209 warnings and you could experience serious issues.
2212 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2215 .`TP_ARGS()` macro syntax.
2224 * `type` with the C type of the argument.
2225 * `arg_name` with the argument name.
2227 You can repeat `type` and `arg_name` up to 10 times to have
2228 more than one argument.
2230 .`TP_ARGS()` usage with three arguments.
2242 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2243 tracepoint definition with no input arguments.
2245 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2246 `ctf_*()` macros. Each `ctf_*()` macro defines one event field.
2247 See <<liblttng-ust-tp-fields,Tracepoint fields macros>> for a
2248 complete description of the available `ctf_*()` macros.
2249 A `ctf_*()` macro specifies the type, size, and byte order of
2252 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2253 C expression that the tracer evalutes at the `tracepoint()` macro site
2254 in the application's source code. This expression provides a field's
2255 source of data. The argument expression can include input argument names
2256 listed in the `TP_ARGS()` macro.
2258 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2259 must be unique within a given tracepoint definition.
2261 Here's a complete tracepoint definition example:
2263 .Tracepoint definition.
2265 The following tracepoint definition defines a tracepoint which takes
2266 three input arguments and has four output event fields.
2270 #include "my-custom-structure.h"
2276 const struct my_custom_structure*, my_custom_structure,
2281 ctf_string(query_field, query)
2282 ctf_float(double, ratio_field, ratio)
2283 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2284 ctf_integer(int, send_size, my_custom_structure->send_size)
2289 You can refer to this tracepoint definition with the `tracepoint()`
2290 macro in your application's source code like this:
2294 tracepoint(my_provider, my_tracepoint,
2295 my_structure, some_ratio, the_query);
2299 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2300 if they satisfy an enabled <<event,event rule>>.
2303 [[using-tracepoint-classes]]
2304 ===== Use a tracepoint class
2306 A _tracepoint class_ is a class of tracepoints which share the same
2307 output event field definitions. A _tracepoint instance_ is one
2308 instance of such a defined tracepoint class, with its own tracepoint
2311 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2312 shorthand which defines both a tracepoint class and a tracepoint
2313 instance at the same time.
2315 When you build a tracepoint provider package, the C or $$C++$$ compiler
2316 creates one serialization function for each **tracepoint class**. A
2317 serialization function is responsible for serializing the event fields
2318 of a tracepoint to a sub-buffer when tracing.
2320 For various performance reasons, when your situation requires multiple
2321 tracepoint definitions with different names, but with the same event
2322 fields, we recommend that you manually create a tracepoint class
2323 and instantiate as many tracepoint instances as needed. One positive
2324 effect of such a design, amongst other advantages, is that all
2325 tracepoint instances of the same tracepoint class reuse the same
2326 serialization function, thus reducing
2327 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2329 .Use a tracepoint class and tracepoint instances.
2331 Consider the following three tracepoint definitions:
2343 ctf_integer(int, userid, userid)
2344 ctf_integer(size_t, len, len)
2356 ctf_integer(int, userid, userid)
2357 ctf_integer(size_t, len, len)
2369 ctf_integer(int, userid, userid)
2370 ctf_integer(size_t, len, len)
2375 In this case, we create three tracepoint classes, with one implicit
2376 tracepoint instance for each of them: `get_account`, `get_settings`, and
2377 `get_transaction`. However, they all share the same event field names
2378 and types. Hence three identical, yet independent serialization
2379 functions are created when you build the tracepoint provider package.
2381 A better design choice is to define a single tracepoint class and three
2382 tracepoint instances:
2386 /* The tracepoint class */
2387 TRACEPOINT_EVENT_CLASS(
2388 /* Tracepoint provider name */
2391 /* Tracepoint class name */
2394 /* Input arguments */
2400 /* Output event fields */
2402 ctf_integer(int, userid, userid)
2403 ctf_integer(size_t, len, len)
2407 /* The tracepoint instances */
2408 TRACEPOINT_EVENT_INSTANCE(
2409 /* Tracepoint provider name */
2412 /* Tracepoint class name */
2415 /* Tracepoint name */
2418 /* Input arguments */
2424 TRACEPOINT_EVENT_INSTANCE(
2433 TRACEPOINT_EVENT_INSTANCE(
2446 [[assigning-log-levels]]
2447 ===== Assign a log level to a tracepoint definition
2449 You can assign an optional _log level_ to a
2450 <<defining-tracepoints,tracepoint definition>>.
2452 Assigning different levels of severity to tracepoint definitions can
2453 be useful: when you <<enabling-disabling-events,create an event rule>>,
2454 you can target tracepoints having a log level as severe as a specific
2457 The concept of LTTng-UST log levels is similar to the levels found
2458 in typical logging frameworks:
2460 * In a logging framework, the log level is given by the function
2461 or method name you use at the log statement site: `debug()`,
2462 `info()`, `warn()`, `error()`, and so on.
2463 * In LTTng-UST, you statically assign the log level to a tracepoint
2464 definition; any `tracepoint()` macro invocation which refers to
2465 this definition has this log level.
2467 You can assign a log level to a tracepoint definition with the
2468 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2469 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2470 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2473 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2476 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2478 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2483 * `provider_name` with the tracepoint provider name.
2484 * `tracepoint_name` with the tracepoint name.
2485 * `log_level` with the log level to assign to the tracepoint
2486 definition named `tracepoint_name` in the `provider_name`
2487 tracepoint provider.
2489 See <<liblttng-ust-tracepoint-loglevel,Tracepoint log levels>> for
2490 a list of available log level names.
2492 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2496 /* Tracepoint definition */
2505 ctf_integer(int, userid, userid)
2506 ctf_integer(size_t, len, len)
2510 /* Log level assignment */
2511 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2517 ===== Create a tracepoint provider package source file
2519 A _tracepoint provider package source file_ is a C source file which
2520 includes a <<tpp-header,tracepoint provider header file>> to expand its
2521 macros into event serialization and other functions.
2523 You can always use the following tracepoint provider package source
2527 .Tracepoint provider package source file template.
2529 #define TRACEPOINT_CREATE_PROBES
2534 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2535 header file>> name. You may also include more than one tracepoint
2536 provider header file here to create a tracepoint provider package
2537 holding more than one tracepoint providers.
2540 [[probing-the-application-source-code]]
2541 ==== Add tracepoints to an application's source code
2543 Once you <<tpp-header,create a tracepoint provider header file>>, you
2544 can use the `tracepoint()` macro in your application's
2545 source code to insert the tracepoints that this header
2546 <<defining-tracepoints,defined>> defines.
2548 The `tracepoint()` macro takes at least two parameters: the tracepoint
2549 provider name and the tracepoint name. The corresponding tracepoint
2550 definition defines the other parameters.
2552 .`tracepoint()` usage.
2554 The following <<defining-tracepoints,tracepoint definition>> defines a
2555 tracepoint which takes two input arguments and has two output event
2559 .Tracepoint provider header file.
2561 #include "my-custom-structure.h"
2568 const char*, cmd_name
2571 ctf_string(cmd_name, cmd_name)
2572 ctf_integer(int, number_of_args, argc)
2577 You can refer to this tracepoint definition with the `tracepoint()`
2578 macro in your application's source code like this:
2581 .Application's source file.
2585 int main(int argc, char* argv[])
2587 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2593 Note how the application's source code includes
2594 the tracepoint provider header file containing the tracepoint
2595 definitions to use, path:{tp.h}.
2598 .`tracepoint()` usage with a complex tracepoint definition.
2600 Consider this complex tracepoint definition, where multiple event
2601 fields refer to the same input arguments in their argument expression
2605 .Tracepoint provider header file.
2607 /* For `struct stat` */
2608 #include <sys/types.h>
2609 #include <sys/stat.h>
2621 ctf_integer(int, my_constant_field, 23 + 17)
2622 ctf_integer(int, my_int_arg_field, my_int_arg)
2623 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2624 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2625 my_str_arg[2] + my_str_arg[3])
2626 ctf_string(my_str_arg_field, my_str_arg)
2627 ctf_integer_hex(off_t, size_field, st->st_size)
2628 ctf_float(double, size_dbl_field, (double) st->st_size)
2629 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2630 size_t, strlen(my_str_arg) / 2)
2635 You can refer to this tracepoint definition with the `tracepoint()`
2636 macro in your application's source code like this:
2639 .Application's source file.
2641 #define TRACEPOINT_DEFINE
2648 stat("/etc/fstab", &s);
2649 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2655 If you look at the event record that LTTng writes when tracing this
2656 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2657 it should look like this:
2659 .Event record fields
2661 |Field's name |Field's value
2662 |`my_constant_field` |40
2663 |`my_int_arg_field` |23
2664 |`my_int_arg_field2` |529
2666 |`my_str_arg_field` |`Hello, World!`
2667 |`size_field` |0x12d
2668 |`size_dbl_field` |301.0
2669 |`half_my_str_arg_field` |`Hello,`
2673 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2674 compute--they use the call stack, for example. To avoid this
2675 computation when the tracepoint is disabled, you can use the
2676 `tracepoint_enabled()` and `do_tracepoint()` macros.
2678 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2682 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2684 tracepoint_enabled(provider_name, tracepoint_name)
2685 do_tracepoint(provider_name, tracepoint_name, ...)
2690 * `provider_name` with the tracepoint provider name.
2691 * `tracepoint_name` with the tracepoint name.
2693 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2694 `tracepoint_name` from the provider named `provider_name` is enabled
2697 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2698 if the tracepoint is enabled. Using `tracepoint()` with
2699 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2700 the `tracepoint_enabled()` check, thus a race condition is
2701 possible in this situation:
2704 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2706 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2707 stuff = prepare_stuff();
2710 tracepoint(my_provider, my_tracepoint, stuff);
2713 If the tracepoint is enabled after the condition, then `stuff` is not
2714 prepared: the emitted event will either contain wrong data, or the whole
2715 application could crash (segmentation fault, for example).
2717 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2718 `STAP_PROBEV()` call. If you need it, you must emit
2722 [[building-tracepoint-providers-and-user-application]]
2723 ==== Build and link a tracepoint provider package and an application
2725 Once you have one or more <<tpp-header,tracepoint provider header
2726 files>> and a <<tpp-source,tracepoint provider package source file>>,
2727 you can create the tracepoint provider package by compiling its source
2728 file. From here, multiple build and run scenarios are possible. The
2729 following table shows common application and library configurations
2730 along with the required command lines to achieve them.
2732 In the following diagrams, we use the following file names:
2735 Executable application.
2738 Application's object file.
2741 Tracepoint provider package object file.
2744 Tracepoint provider package archive file.
2747 Tracepoint provider package shared object file.
2750 User library object file.
2753 User library shared object file.
2755 The red star indicates that this object file is instrumented
2756 (contains code which uses the `tracepoint()` macro). The spring
2757 symbol between the application and a library means the application is
2758 linked with the library at build time.
2760 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2761 variable in the following instructions.
2763 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2764 .Common tracepoint provider package scenarios.
2766 |Scenario |Instructions
2769 The instrumented application is statically linked with
2770 the tracepoint provider package object.
2772 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2775 include::../common/ust-sit-step-tp-o.txt[]
2777 To build the instrumented application:
2779 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2784 #define TRACEPOINT_DEFINE
2788 . Compile the application source file:
2797 . Build the application:
2802 gcc -o app app.o tpp.o -llttng-ust -ldl
2806 To run the instrumented application:
2808 * Start the application:
2818 The instrumented application is statically linked with the
2819 tracepoint provider package archive file.
2821 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2824 To create the tracepoint provider package archive file:
2826 . Compile the <<tpp-source,tracepoint provider package source file>>:
2835 . Create the tracepoint provider package archive file:
2844 To build the instrumented application:
2846 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2851 #define TRACEPOINT_DEFINE
2855 . Compile the application source file:
2864 . Build the application:
2869 gcc -o app app.o tpp.a -llttng-ust -ldl
2873 To run the instrumented application:
2875 * Start the application:
2885 The instrumented application is linked with the tracepoint provider
2886 package shared object.
2888 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2891 include::../common/ust-sit-step-tp-so.txt[]
2893 To build the instrumented application:
2895 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2900 #define TRACEPOINT_DEFINE
2904 . Compile the application source file:
2913 . Build the application:
2918 gcc -o app app.o -ldl -L. -ltpp
2922 To run the instrumented application:
2924 * Start the application:
2934 The tracepoint provider package shared object is preloaded before the
2935 instrumented application starts.
2937 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
2940 include::../common/ust-sit-step-tp-so.txt[]
2942 To build the instrumented application:
2944 . In path:{app.c}, before including path:{tpp.h}, add the
2950 #define TRACEPOINT_DEFINE
2951 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2955 . Compile the application source file:
2964 . Build the application:
2969 gcc -o app app.o -ldl
2973 To run the instrumented application with tracing support:
2975 * Preload the tracepoint provider package shared object and
2976 start the application:
2981 LD_PRELOAD=./libtpp.so ./app
2985 To run the instrumented application without tracing support:
2987 * Start the application:
2997 The instrumented application dynamically loads the tracepoint provider
2998 package shared object.
3000 See the <<dlclose-warning,warning about `dlclose()`>>.
3002 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3005 include::../common/ust-sit-step-tp-so.txt[]
3007 To build the instrumented application:
3009 . In path:{app.c}, before including path:{tpp.h}, add the
3015 #define TRACEPOINT_DEFINE
3016 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3020 . Compile the application source file:
3029 . Build the application:
3034 gcc -o app app.o -ldl
3038 To run the instrumented application:
3040 * Start the application:
3050 The application is linked with the instrumented user library.
3052 The instrumented user library is statically linked with the tracepoint
3053 provider package object file.
3055 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3058 include::../common/ust-sit-step-tp-o-fpic.txt[]
3060 To build the instrumented user library:
3062 . In path:{emon.c}, before including path:{tpp.h}, add the
3068 #define TRACEPOINT_DEFINE
3072 . Compile the user library source file:
3077 gcc -I. -fpic -c emon.c
3081 . Build the user library shared object:
3086 gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3090 To build the application:
3092 . Compile the application source file:
3101 . Build the application:
3106 gcc -o app app.o -L. -lemon
3110 To run the application:
3112 * Start the application:
3122 The application is linked with the instrumented user library.
3124 The instrumented user library is linked with the tracepoint provider
3125 package shared object.
3127 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3130 include::../common/ust-sit-step-tp-so.txt[]
3132 To build the instrumented user library:
3134 . In path:{emon.c}, before including path:{tpp.h}, add the
3140 #define TRACEPOINT_DEFINE
3144 . Compile the user library source file:
3149 gcc -I. -fpic -c emon.c
3153 . Build the user library shared object:
3158 gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3162 To build the application:
3164 . Compile the application source file:
3173 . Build the application:
3178 gcc -o app app.o -L. -lemon
3182 To run the application:
3184 * Start the application:
3194 The tracepoint provider package shared object is preloaded before the
3197 The application is linked with the instrumented user library.
3199 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3202 include::../common/ust-sit-step-tp-so.txt[]
3204 To build the instrumented user library:
3206 . In path:{emon.c}, before including path:{tpp.h}, add the
3212 #define TRACEPOINT_DEFINE
3213 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3217 . Compile the user library source file:
3222 gcc -I. -fpic -c emon.c
3226 . Build the user library shared object:
3231 gcc -shared -o libemon.so emon.o -ldl
3235 To build the application:
3237 . Compile the application source file:
3246 . Build the application:
3251 gcc -o app app.o -L. -lemon
3255 To run the application with tracing support:
3257 * Preload the tracepoint provider package shared object and
3258 start the application:
3263 LD_PRELOAD=./libtpp.so ./app
3267 To run the application without tracing support:
3269 * Start the application:
3279 The application is linked with the instrumented user library.
3281 The instrumented user library dynamically loads the tracepoint provider
3282 package shared object.
3284 See the <<dlclose-warning,warning about `dlclose()`>>.
3286 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3289 include::../common/ust-sit-step-tp-so.txt[]
3291 To build the instrumented user library:
3293 . In path:{emon.c}, before including path:{tpp.h}, add the
3299 #define TRACEPOINT_DEFINE
3300 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3304 . Compile the user library source file:
3309 gcc -I. -fpic -c emon.c
3313 . Build the user library shared object:
3318 gcc -shared -o libemon.so emon.o -ldl
3322 To build the application:
3324 . Compile the application source file:
3333 . Build the application:
3338 gcc -o app app.o -L. -lemon
3342 To run the application:
3344 * Start the application:
3354 The application dynamically loads the instrumented user library.
3356 The instrumented user library is linked with the tracepoint provider
3357 package shared object.
3359 See the <<dlclose-warning,warning about `dlclose()`>>.
3361 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3364 include::../common/ust-sit-step-tp-so.txt[]
3366 To build the instrumented user library:
3368 . In path:{emon.c}, before including path:{tpp.h}, add the
3374 #define TRACEPOINT_DEFINE
3378 . Compile the user library source file:
3383 gcc -I. -fpic -c emon.c
3387 . Build the user library shared object:
3392 gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3396 To build the application:
3398 . Compile the application source file:
3407 . Build the application:
3412 gcc -o app app.o -ldl -L. -lemon
3416 To run the application:
3418 * Start the application:
3428 The application dynamically loads the instrumented user library.
3430 The instrumented user library dynamically loads the tracepoint provider
3431 package shared object.
3433 See the <<dlclose-warning,warning about `dlclose()`>>.
3435 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3438 include::../common/ust-sit-step-tp-so.txt[]
3440 To build the instrumented user library:
3442 . In path:{emon.c}, before including path:{tpp.h}, add the
3448 #define TRACEPOINT_DEFINE
3449 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3453 . Compile the user library source file:
3458 gcc -I. -fpic -c emon.c
3462 . Build the user library shared object:
3467 gcc -shared -o libemon.so emon.o -ldl
3471 To build the application:
3473 . Compile the application source file:
3482 . Build the application:
3487 gcc -o app app.o -ldl -L. -lemon
3491 To run the application:
3493 * Start the application:
3503 The tracepoint provider package shared object is preloaded before the
3506 The application dynamically loads the instrumented user library.
3508 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3511 include::../common/ust-sit-step-tp-so.txt[]
3513 To build the instrumented user library:
3515 . In path:{emon.c}, before including path:{tpp.h}, add the
3521 #define TRACEPOINT_DEFINE
3522 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3526 . Compile the user library source file:
3531 gcc -I. -fpic -c emon.c
3535 . Build the user library shared object:
3540 gcc -shared -o libemon.so emon.o -ldl
3544 To build the application:
3546 . Compile the application source file:
3555 . Build the application:
3560 gcc -o app app.o -L. -lemon
3564 To run the application with tracing support:
3566 * Preload the tracepoint provider package shared object and
3567 start the application:
3572 LD_PRELOAD=./libtpp.so ./app
3576 To run the application without tracing support:
3578 * Start the application:
3588 The application is statically linked with the tracepoint provider
3589 package object file.
3591 The application is linked with the instrumented user library.
3593 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3596 include::../common/ust-sit-step-tp-o.txt[]
3598 To build the instrumented user library:
3600 . In path:{emon.c}, before including path:{tpp.h}, add the
3606 #define TRACEPOINT_DEFINE
3610 . Compile the user library source file:
3615 gcc -I. -fpic -c emon.c
3619 . Build the user library shared object:
3624 gcc -shared -o libemon.so emon.o
3628 To build the application:
3630 . Compile the application source file:
3639 . Build the application:
3644 gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3648 To run the instrumented application:
3650 * Start the application:
3660 The application is statically linked with the tracepoint provider
3661 package object file.
3663 The application dynamically loads the instrumented user library.
3665 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3668 include::../common/ust-sit-step-tp-o.txt[]
3670 To build the application:
3672 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3677 #define TRACEPOINT_DEFINE
3681 . Compile the application source file:
3690 . Build the application:
3695 gcc -Wl,--export-dynamic -o app app.o tpp.o \
3700 The `--export-dynamic` option passed to the linker is necessary for the
3701 dynamically loaded library to ``see'' the tracepoint symbols defined in
3704 To build the instrumented user library:
3706 . Compile the user library source file:
3711 gcc -I. -fpic -c emon.c
3715 . Build the user library shared object:
3720 gcc -shared -o libemon.so emon.o
3724 To run the application:
3726 * Start the application:
3738 .Do not use man:dlclose(3) on a tracepoint provider package
3740 Never use man:dlclose(3) on any shared object which:
3742 * Is linked with, statically or dynamically, a tracepoint provider
3744 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3745 package shared object.
3747 This is currently considered **unsafe** due to a lack of reference
3748 counting from LTTng-UST to the shared object.
3750 A known workaround (available since glibc 2.2) is to use the
3751 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3752 effect of not unloading the loaded shared object, even if man:dlclose(3)
3755 You can also preload the tracepoint provider package shared object with
3756 the env:LD_PRELOAD environment variable to overcome this limitation.
3760 [[using-lttng-ust-with-daemons]]
3761 ===== Use noch:{LTTng-UST} with daemons
3763 If your instrumented application calls man:fork(2), man:clone(2),
3764 or BSD's man:rfork(2), without a following man:exec(3)-family
3765 system call, you must preload the path:{liblttng-ust-fork.so} shared
3766 object when starting the application.
3770 LD_PRELOAD=liblttng-ust-fork.so ./my-app
3773 If your tracepoint provider package is
3774 a shared library which you also preload, you must put both
3775 shared objects in env:LD_PRELOAD:
3779 LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3783 [[lttng-ust-pkg-config]]
3784 ===== Use noch:{pkg-config}
3786 On some distributions, LTTng-UST ships with a
3787 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3788 metadata file. If this is your case, then you can use cmd:pkg-config to
3789 build an application on the command line:
3793 gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3797 [[instrumenting-32-bit-app-on-64-bit-system]]
3798 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3800 In order to trace a 32-bit application running on a 64-bit system,
3801 LTTng must use a dedicated 32-bit
3802 <<lttng-consumerd,consumer daemon>>.
3804 The following steps show how to build and install a 32-bit consumer
3805 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3806 build and install the 32-bit LTTng-UST libraries, and how to build and
3807 link an instrumented 32-bit application in that context.
3809 To build a 32-bit instrumented application for a 64-bit target system,
3810 assuming you have a fresh target system with no installed Userspace RCU
3813 . Download, build, and install a 32-bit version of Userspace RCU:
3819 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3820 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3821 cd userspace-rcu-0.9.* &&
3822 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3824 sudo make install &&
3829 . Using your distribution's package manager, or from source, install
3830 the following 32-bit versions of the following dependencies of
3831 LTTng-tools and LTTng-UST:
3834 * https://sourceforge.net/projects/libuuid/[libuuid]
3835 * http://directory.fsf.org/wiki/Popt[popt]
3836 * http://www.xmlsoft.org/[libxml2]
3839 . Download, build, and install a 32-bit version of the latest
3840 LTTng-UST{nbsp}{revision}:
3846 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.7.tar.bz2 &&
3847 tar -xf lttng-ust-latest-2.7.tar.bz2 &&
3848 cd lttng-ust-2.7.* &&
3849 ./configure --libdir=/usr/local/lib32 \
3850 CFLAGS=-m32 CXXFLAGS=-m32 \
3851 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3853 sudo make install &&
3860 Depending on your distribution,
3861 32-bit libraries could be installed at a different location than
3862 `/usr/lib32`. For example, Debian is known to install
3863 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3865 In this case, make sure to set `LDFLAGS` to all the
3866 relevant 32-bit library paths, for example:
3870 LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3874 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3875 the 32-bit consumer daemon:
3881 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.7.tar.bz2 &&
3882 tar -xf lttng-tools-latest-2.7.tar.bz2 &&
3883 cd lttng-tools-2.7.* &&
3884 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3885 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3887 cd src/bin/lttng-consumerd &&
3888 sudo make install &&
3893 . From your distribution or from source,
3894 <<installing-lttng,install>> the 64-bit versions of
3895 LTTng-UST and Userspace RCU.
3896 . Download, build, and install the 64-bit version of the
3897 latest LTTng-tools{nbsp}{revision}:
3903 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.7.tar.bz2 &&
3904 tar -xf lttng-tools-latest-2.7.tar.bz2 &&
3905 cd lttng-tools-2.7.* &&
3906 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
3907 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
3909 sudo make install &&
3914 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
3915 when linking your 32-bit application:
3918 -m32 -L/usr/lib32 -L/usr/local/lib32 \
3919 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
3922 For example, let's rebuild the quick start example in
3923 <<tracing-your-own-user-application,Trace a user application>> as an
3924 instrumented 32-bit application:
3929 gcc -m32 -c -I. hello-tp.c
3931 gcc -m32 -o hello hello.o hello-tp.o \
3932 -L/usr/lib32 -L/usr/local/lib32 \
3933 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
3938 No special action is required to execute the 32-bit application and
3939 to trace it: use the command-line man:lttng(1) tool as usual.
3946 `tracef()` is a small LTTng-UST API designed for quick,
3947 man:printf(3)-like instrumentation without the burden of
3948 <<tracepoint-provider,creating>> and
3949 <<building-tracepoint-providers-and-user-application,building>>
3950 a tracepoint provider package.
3952 To use `tracef()` in your application:
3954 . In the C or C++ source files where you need to use `tracef()`,
3955 include `<lttng/tracef.h>`:
3960 #include <lttng/tracef.h>
3964 . In the application's source code, use `tracef()` like you would use
3972 tracef("my message: %d (%s)", my_integer, my_string);
3978 . Link your application with `liblttng-ust`:
3983 gcc -o app app.c -llttng-ust
3987 To trace the events that `tracef()` calls emit:
3989 * <<enabling-disabling-events,Create an event rule>> which matches the
3990 `lttng_ust_tracef:*` event name:
3995 lttng enable-event --userspace 'lttng_ust_tracef:*'
4000 .Limitations of `tracef()`
4002 The `tracef()` utility function was developed to make user space tracing
4003 super simple, albeit with notable disadvantages compared to
4004 <<defining-tracepoints,user-defined tracepoints>>:
4006 * All the emitted events have the same tracepoint provider and
4007 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4008 * There is no static type checking.
4009 * The only event record field you actually get, named `msg`, is a string
4010 potentially containing the values you passed to `tracef()`
4011 using your own format string. This also means that you cannot filter
4012 events with a custom expression at run time because there are no
4014 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4015 function behind the scenes to format the strings at run time, its
4016 expected performance is lower than with user-defined tracepoints,
4017 which do not require a conversion to a string.
4019 Taking this into consideration, `tracef()` is useful for some quick
4020 prototyping and debugging, but you should not consider it for any
4021 permanent and serious applicative instrumentation.
4027 ==== Use `tracelog()`
4029 The `tracelog()` API is very similar to <<tracef,`tracef()`>>, with
4030 the difference that it accepts an additional log level parameter.
4032 The goal of `tracelog()` is to ease the migration from logging to
4035 To use `tracelog()` in your application:
4037 . In the C or C++ source files where you need to use `tracelog()`,
4038 include `<lttng/tracelog.h>`:
4043 #include <lttng/tracelog.h>
4047 . In the application's source code, use `tracelog()` like you would use
4048 man:printf(3), except for the first parameter which is the log
4056 tracelog(TRACE_WARNING, "my message: %d (%s)",
4057 my_integer, my_string);
4063 See <<liblttng-ust-tracepoint-loglevel,Tracepoint log levels>> for
4064 a list of available log level names.
4066 . Link your application with `liblttng-ust`:
4071 gcc -o app app.c -llttng-ust
4075 To trace the events that `tracelog()` calls emit with a log level
4076 _as severe as_ a specific log level:
4078 * <<enabling-disabling-events,Create an event rule>> which matches the
4079 `lttng_ust_tracelog:*` event name and a minimum level
4085 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4086 --loglevel=TRACE_WARNING
4090 To trace the events that `tracelog()` calls emit with a
4091 _specific log level_:
4093 * Create an event rule which matches the `lttng_ust_tracelog:*`
4094 event name and a specific log level:
4099 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4100 --loglevel-only=TRACE_INFO
4105 [[prebuilt-ust-helpers]]
4106 === Prebuilt user space tracing helpers
4108 The LTTng-UST package provides a few helpers in the form or preloadable
4109 shared objects which automatically instrument system functions and
4112 The helper shared objects are normally found in dir:{/usr/lib}. If you
4113 built LTTng-UST <<building-from-source,from source>>, they are probably
4114 located in dir:{/usr/local/lib}.
4116 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4119 path:{liblttng-ust-libc-wrapper.so}::
4120 path:{liblttng-ust-pthread-wrapper.so}::
4121 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4122 memory and POSIX threads function tracing>>.
4124 path:{liblttng-ust-cyg-profile.so}::
4125 path:{liblttng-ust-cyg-profile-fast.so}::
4126 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4128 path:{liblttng-ust-dl.so}::
4129 <<liblttng-ust-dl,Dynamic linker tracing>>.
4131 To use a user space tracing helper with any user application:
4133 * Preload the helper shared object when you start the application:
4138 LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4142 You can preload more than one helper:
4147 LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4153 [[liblttng-ust-libc-pthread-wrapper]]
4154 ==== Instrument C standard library memory and POSIX threads functions
4156 The path:{liblttng-ust-libc-wrapper.so} and
4157 path:{liblttng-ust-pthread-wrapper.so} helpers
4158 add instrumentation to some C standard library and POSIX
4162 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4164 |TP provider name |TP name |Instrumented function
4166 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4167 |`calloc` |man:calloc(3)
4168 |`realloc` |man:realloc(3)
4169 |`free` |man:free(3)
4170 |`memalign` |man:memalign(3)
4171 |`posix_memalign` |man:posix_memalign(3)
4175 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4177 |TP provider name |TP name |Instrumented function
4179 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4180 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4181 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4182 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4185 When you preload the shared object, it replaces the functions listed
4186 in the previous tables by wrappers which contain tracepoints and call
4187 the replaced functions.
4190 [[liblttng-ust-cyg-profile]]
4191 ==== Instrument function entry and exit
4193 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4194 to the entry and exit points of functions.
4196 man:gcc(1) and man:clang(1) have an option named
4197 https://gcc.gnu.org/onlinedocs/gcc/Code-Gen-Options.html[`-finstrument-functions`]
4198 which generates instrumentation calls for entry and exit to functions.
4199 The LTTng-UST function tracing helpers,
4200 path:{liblttng-ust-cyg-profile.so} and
4201 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4202 to add tracepoints to the two generated functions (which contain
4203 `cyg_profile` in their names, hence the helper's name).
4205 To use the LTTng-UST function tracing helper, the source files to
4206 instrument must be built using the `-finstrument-functions` compiler
4209 There are two versions of the LTTng-UST function tracing helper:
4211 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4212 that you should only use when it can be _guaranteed_ that the
4213 complete event stream is recorded without any lost event record.
4214 Any kind of duplicate information is left out.
4216 This version contains the following tracepoints:
4220 .Points instrumented by preloading path:{liblttng-ust-cyg-profile-fast.so}.
4222 |TP provider name |TP name |Instrumented points
4224 .2+|`lttng_ust_cyg_profile_fast`
4230 Address of called function.
4237 Assuming no event record is lost, having only the function addresses on
4238 entry is enough to create a call graph, since an event record always
4239 contains the ID of the CPU that generated it.
4241 You can use a tool like
4242 https://sourceware.org/binutils/docs/binutils/addr2line.html[cmd:addr2line]
4243 to convert function addresses back to source file names and
4246 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4247 which also works in use cases where event records might get discarded or
4248 not recorded from application startup.
4249 In these cases, the trace analyzer needs more information to be
4250 able to reconstruct the program flow.
4252 This version contains the following tracepoints:
4256 .Points instrumented by preloading path:{liblttng-ust-cyg-profile.so}.
4258 |TP provider name |TP name |Instrumented point
4260 .2+|`lttng_ust_cyg_profile`
4266 Address of called function.
4275 Address of called function.
4282 TIP: It's sometimes a good idea to limit the number of source files that
4283 you compile with the `-finstrument-functions` option to prevent LTTng
4284 from writing an excessive amount of trace data at run time. When using
4285 man:gcc(1), you can use the
4286 `-finstrument-functions-exclude-function-list` option to avoid
4287 instrument entries and exits of specific function names.
4289 All the tracepoints that this helper contains have the
4290 <<liblttng-ust-tracepoint-loglevel,log level>> `TRACE_DEBUG_FUNCTION`.
4295 ==== Instrument the dynamic linker
4297 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4298 man:dlopen(3) and man:dlclose(3) function calls.
4301 .Functions instrumented by preloading path:{liblttng-ust-dl.so}.
4303 |TP provider name |TP name |Instrumented function
4311 Memory base address (where the dynamic linker placed the shared
4315 File system path to the loaded shared object.
4318 File size of the the loaded shared object.
4321 Last modification time (seconds since Epoch time) of the loaded shared
4328 Memory base address (where the dynamic linker placed the shared
4334 [[java-application]]
4335 === User space Java agent
4337 You can instrument a Java application which uses one of the following
4340 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4341 (JUL) core logging facilities.
4342 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4343 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4345 Each log statement emits an LTTng event once the
4346 application initializes the <<lttng-ust-agents,LTTng-UST Java agent>>
4350 .LTTng-UST Java agent imported by a Java application.
4351 image::java-app.png[]
4353 NOTE: We use http://openjdk.java.net/[OpenJDK] 7 for development and
4354 https://ci.lttng.org/[continuous integration], thus this version is
4355 directly supported. However, the LTTng-UST Java agent is also
4356 tested with OpenJDK 6.
4358 To use the LTTng-UST Java agent:
4360 . In the Java application's source code, import the LTTng-UST Java
4366 import org.lttng.ust.agent.LTTngAgent;
4370 . As soon as possible after the entry point of the application,
4371 initialize the LTTng-UST Java agent:
4376 LTTngAgent lttngAgent = LTTngAgent.getLTTngAgent();
4380 Any log statement that the application executes before this
4381 initialization does not emit an LTTng event.
4383 . Use `java.util.logging` and/or log4j log statements and configuration
4384 as usual. Since the LTTng-UST Java agent adds a handler to the _root_
4385 loggers, you can trace any log statement from any logger.
4387 . Before exiting the application, dispose the LTTng-UST Java agent:
4392 lttngAgent.dispose();
4396 This is not strictly necessary, but it is recommended for a clean
4397 disposal of the agent's resources.
4399 Any log statement that the application executes after this disposal does
4400 not emit an LTTng event.
4402 . Include the LTTng-UST Java agent's JAR file, path:{liblttng-ust-agent.jar},
4404 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class path]
4405 when building the Java application.
4407 path:{liblttng-ust-agent.jar} is typically located in
4408 dir:{/usr/share/java}.
4410 IMPORTANT: The LTTng-UST Java agent must be
4411 <<installing-lttng,installed>> for the logging framework your
4414 .[[jul]]Use the LTTng-UST Java agent with `java.util.logging`.
4419 import java.util.logging.Logger;
4420 import org.lttng.ust.agent.LTTngAgent;
4424 private static final int answer = 42;
4426 public static void main(String[] argv) throws Exception
4429 Logger logger = Logger.getLogger("jello");
4431 // Call this as soon as possible (before logging)
4432 LTTngAgent lttngAgent = LTTngAgent.getLTTngAgent();
4435 logger.info("some info");
4436 logger.warning("some warning");
4438 logger.finer("finer information; the answer is " + answer);
4440 logger.severe("error!");
4442 // Not mandatory, but cleaner
4443 lttngAgent.dispose();
4448 You can build this example like this:
4452 javac -cp /usr/share/java/liblttng-ust-agent.jar Test.java
4455 You can run the compiled class like this:
4459 java -cp /usr/share/java/liblttng-ust-agent.jar:. Test
4463 .[[log4j]]Use the LTTng-UST Java agent with Apache log4j 1.2.
4468 import org.apache.log4j.Logger;
4469 import org.apache.log4j.BasicConfigurator;
4470 import org.lttng.ust.agent.LTTngAgent;
4474 private static final int answer = 42;
4476 public static void main(String[] argv) throws Exception
4478 // Create and configure a logger
4479 Logger logger = Logger.getLogger(Test.class);
4480 BasicConfigurator.configure();
4482 // Call this as soon as possible (before logging)
4483 LTTngAgent lttngAgent = LTTngAgent.getLTTngAgent();
4486 logger.info("some info");
4487 logger.warn("some warning");
4489 logger.debug("debug information; the answer is " + answer);
4491 logger.error("error!");
4492 logger.fatal("fatal error!");
4494 // Not mandatory, but cleaner
4495 lttngAgent.dispose();
4500 You can build this example like this:
4504 javac -cp /usr/share/java/liblttng-ust-agent.jar:$LOG4JCP Test.java
4507 where `$LOG4JCP` is the path to log4j's JAR file.
4509 You can run the compiled class like this:
4513 java -cp /usr/share/java/liblttng-ust-agent.jar:$LOG4JCP:. Test
4517 When you <<enabling-disabling-events,create an event rule>>, use the
4518 `--jul` (`java.util.logging`) or `--log4j` (log4j) option to target
4520 <<domain,tracing domain>>. You can also use the `--loglevel` or
4521 `--loglevel-only` option to target a range of JUL/log4j log levels or a
4522 specific JUL/log4j log level.
4526 [[python-application]]
4527 === User space Python agent
4529 You can instrument a Python 2 or Python 3 application which uses the
4530 standard https://docs.python.org/3/library/logging.html[`logging`]
4533 Each log statement emits an LTTng event once the
4534 application module imports the
4535 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4538 .A Python application importing the LTTng-UST Python agent.
4539 image::python-app.png[]
4541 To use the LTTng-UST Python agent:
4543 . In the Python application's source code, import the LTTng-UST Python
4553 The LTTng-UST Python agent automatically adds its logging handler to the
4554 root logger at import time.
4556 Any log statement that the application executes before this import does
4557 not emit an LTTng event.
4559 IMPORTANT: The LTTng-UST Python agent must be
4560 <<installing-lttng,installed>>.
4562 . Use log statements and logging configuration as usual.
4563 Since the LTTng-UST Python agent adds a handler to the _root_
4564 logger, you can trace any log statement from any logger.
4566 .Use the LTTng-UST Python agent.
4576 logging.basicConfig()
4577 logger = logging.getLogger('my-logger')
4580 logger.debug('debug message')
4581 logger.info('info message')
4582 logger.warn('warn message')
4583 logger.error('error message')
4584 logger.critical('critical message')
4588 if __name__ == '__main__':
4592 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
4593 logging handler which prints to the standard error stream, is not
4594 strictly required for LTTng-UST tracing to work, but in versions of
4595 Python preceding 3.2, you could see a warning message which indicates
4596 that no handler exists for the logger `my-logger`.
4599 When you <<enabling-disabling-events,create an event rule>>, use the
4600 `--python` option to target the Python
4601 <<domain,tracing domain>>. You can also use
4602 the `--loglevel` or `--loglevel-only` option to target a range of
4603 Python log levels or a specific Python log level.
4605 When an application imports the LTTng-UST Python agent, the agent tries
4606 to register to a <<lttng-sessiond,session daemon>>. Note that you must
4607 start the session daemon _before_ you start the Python application.
4608 If a session daemon is found, the agent tries to register to it
4609 during 5{nbsp}seconds, after which the application continues without
4610 LTTng tracing support. You can override this timeout value with the
4611 env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
4614 If the session daemon stops while a Python application with an imported
4615 LTTng-UST Python agent runs, the agent retries to connect and to
4616 register to a session daemon every 3{nbsp}seconds. You can override this
4617 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
4622 [[proc-lttng-logger-abi]]
4625 The `lttng-tracer` Linux kernel module, part of
4626 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
4627 path:{/proc/lttng-logger} when it's loaded. Any application can write
4628 text data to this file to emit an LTTng event.
4631 .An application writes to the LTTng logger file to emit an LTTng event.
4632 image::lttng-logger.png[]
4634 The LTTng logger is the quickest method--not the most efficient,
4635 however--to add instrumentation to an application. It is designed
4636 mostly to instrument shell scripts:
4640 echo "Some message, some $variable" > /proc/lttng-logger
4643 Any event that the LTTng logger emits is named `lttng_logger` and
4644 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
4645 other instrumentation points in the kernel tracing domain, **any Unix
4646 user** can <<enabling-disabling-events,create an event rule>> which
4647 matches its event name, not only the root user or users in the tracing
4650 To use the LTTng logger:
4652 * From any application, write text data to the path:{/proc/lttng-logger}
4655 The `msg` field of `lttng_logger` event records contains the
4658 NOTE: The maximum message length of an LTTng logger event is
4659 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
4660 than one event to contain the remaining data.
4662 You should not use the LTTng logger to trace a user application which
4663 can be instrumented in a more efficient way, namely:
4665 * <<c-application,C and $$C++$$ applications>>.
4666 * <<java-application,Java applications>>.
4667 * <<python-application,Python applications>>.
4670 [[instrumenting-linux-kernel]]
4671 === LTTng kernel tracepoints
4673 NOTE: This section shows how to _add_ instrumentation points to the
4674 Linux kernel. The kernel's subsystems are already thoroughly
4675 instrumented at strategic places for LTTng when you
4676 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
4680 There are two methods to instrument the Linux kernel:
4682 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
4683 tracepoint which uses the `TRACE_EVENT()` API.
4685 Choose this if you want to instrumentation a Linux kernel tree with an
4686 instrumentation point compatible with ftrace, perf, and SystemTap.
4688 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
4689 instrument an out-of-tree kernel module.
4691 Choose this if you don't need ftrace, perf, or SystemTap support.
4695 [[linux-add-lttng-layer]]
4696 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
4698 This section shows how to add an LTTng layer to existing ftrace
4699 instrumentation using the `TRACE_EVENT()` API.
4701 This section does not document the `TRACE_EVENT()` macro. You can
4702 read the following articles to learn more about this API:
4704 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
4705 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
4706 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
4708 The following procedure assumes that your ftrace tracepoints are
4709 correctly defined in their own header and that they are created in
4710 one source file using the `CREATE_TRACE_POINTS` definition.
4712 To add an LTTng layer over an existing ftrace tracepoint:
4714 . Make sure the following kernel configuration options are
4720 * `CONFIG_HIGH_RES_TIMERS`
4721 * `CONFIG_TRACEPOINTS`
4724 . Build the Linux source tree with your custom ftrace tracepoints.
4725 . Boot the resulting Linux image on your target system.
4727 Confirm that the tracepoints exist by looking for their names in the
4728 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
4729 is your subsystem's name.
4731 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
4737 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
4738 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
4739 cd lttng-modules-2.8.*
4743 . In dir:{instrumentation/events/lttng-module}, relative to the root
4744 of the LTTng-modules source tree, create a header file named
4745 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
4746 LTTng-modules tracepoint definitions using the LTTng-modules
4749 Start with this template:
4753 .path:{instrumentation/events/lttng-module/my_subsys.h}
4756 #define TRACE_SYSTEM my_subsys
4758 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
4759 #define _LTTNG_MY_SUBSYS_H
4761 #include "../../../probes/lttng-tracepoint-event.h"
4762 #include <linux/tracepoint.h>
4764 LTTNG_TRACEPOINT_EVENT(
4766 * Format is identical to TRACE_EVENT()'s version for the three
4767 * following macro parameters:
4770 TP_PROTO(int my_int, const char *my_string),
4771 TP_ARGS(my_int, my_string),
4773 /* LTTng-modules specific macros */
4775 ctf_integer(int, my_int_field, my_int)
4776 ctf_string(my_bar_field, my_bar)
4780 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
4782 #include "../../../probes/define_trace.h"
4786 The entries in the `TP_FIELDS()` section are the list of fields for the
4787 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
4788 ftrace's `TRACE_EVENT()` macro.
4790 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
4791 complete description of the available `ctf_*()` macros.
4793 . Create the LTTng-modules probe's kernel module C source file,
4794 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
4799 .path:{probes/lttng-probe-my-subsys.c}
4801 #include <linux/module.h>
4802 #include "../lttng-tracer.h"
4805 * Build-time verification of mismatch between mainline
4806 * TRACE_EVENT() arguments and the LTTng-modules adaptation
4807 * layer LTTNG_TRACEPOINT_EVENT() arguments.
4809 #include <trace/events/my_subsys.h>
4811 /* Create LTTng tracepoint probes */
4812 #define LTTNG_PACKAGE_BUILD
4813 #define CREATE_TRACE_POINTS
4814 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
4816 #include "../instrumentation/events/lttng-module/my_subsys.h"
4818 MODULE_LICENSE("GPL and additional rights");
4819 MODULE_AUTHOR("Your name <your-email>");
4820 MODULE_DESCRIPTION("LTTng my_subsys probes");
4821 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
4822 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
4823 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
4824 LTTNG_MODULES_EXTRAVERSION);
4828 . Edit path:{probes/Makefile} and add your new kernel module object
4829 next to the existing ones:
4833 .path:{probes/Makefile}
4837 obj-m += lttng-probe-module.o
4838 obj-m += lttng-probe-power.o
4840 obj-m += lttng-probe-my-subsys.o
4846 . Build and install the LTTng kernel modules:
4851 make KERNELDIR=/path/to/linux
4852 sudo make modules_install
4856 Replace `/path/to/linux` with the path to the Linux source tree where
4857 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
4859 Note that you can also use the
4860 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
4861 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
4862 C code that need to be executed before the event fields are recorded.
4864 The best way to learn how to use the previous LTTng-modules macros is to
4865 inspect the existing LTTng-modules tracepoint definitions in the
4866 dir:{instrumentation/events/lttng-module} header files. Compare them
4867 with the Linux kernel mainline versions in the
4868 dir:{include/trace/events} directory of the Linux source tree.
4872 [[lttng-tracepoint-event-code]]
4873 ===== Use custom C code to access the data for tracepoint fields
4875 Although we recommended to always use the
4876 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
4877 the arguments and fields of an LTTng-modules tracepoint when possible,
4878 sometimes you need a more complex process to access the data that the
4879 tracer records as event record fields. In other words, you need local
4880 variables and multiple C{nbsp}statements instead of simple
4881 argument-based expressions that you pass to the
4882 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
4884 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
4885 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
4886 a block of C{nbsp}code to be executed before LTTng records the fields.
4887 The structure of this macro is:
4890 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
4892 LTTNG_TRACEPOINT_EVENT_CODE(
4894 * Format identical to the LTTNG_TRACEPOINT_EVENT()
4895 * version for the following three macro parameters:
4898 TP_PROTO(int my_int, const char *my_string),
4899 TP_ARGS(my_int, my_string),
4901 /* Declarations of custom local variables */
4904 unsigned long b = 0;
4905 const char *name = "(undefined)";
4906 struct my_struct *my_struct;
4910 * Custom code which uses both tracepoint arguments
4911 * (in TP_ARGS()) and local variables (in TP_locvar()).
4913 * Local variables are actually members of a structure pointed
4914 * to by the special variable tp_locvar.
4918 tp_locvar->a = my_int + 17;
4919 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
4920 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
4921 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
4922 put_my_struct(tp_locvar->my_struct);
4931 * Format identical to the LTTNG_TRACEPOINT_EVENT()
4932 * version for this, except that tp_locvar members can be
4933 * used in the argument expression parameters of
4934 * the ctf_*() macros.
4937 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
4938 ctf_integer(int, my_struct_a, tp_locvar->a)
4939 ctf_string(my_string_field, my_string)
4940 ctf_string(my_struct_name, tp_locvar->name)
4945 IMPORTANT: The C code defined in `TP_code()` must not have any side
4946 effects when executed. In particular, the code must not allocate
4947 memory or get resources without deallocating this memory or putting
4948 those resources afterwards.
4951 [[instrumenting-linux-kernel-tracing]]
4952 ==== Load and unload a custom probe kernel module
4954 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
4955 kernel module>> in the kernel before it can emit LTTng events.
4957 To load the default probe kernel modules and a custom probe kernel
4960 * Use the `--extra-kmod-probes` option to give extra probe modules
4961 to load when starting a root <<lttng-sessiond,session daemon>>:
4964 .Load the `my_subsys`, `usb`, and the default probe modules.
4968 sudo lttng-sessiond --extra-kmod-probes=my_subsys,usb
4973 You only need to pass the subsystem name, not the whole kernel module
4976 To load _only_ a given custom probe kernel module:
4978 * Use the `--kmod-probes` option to give the probe modules
4979 to load when starting a root session daemon:
4982 .Load only the `my_subsys` and `usb` probe modules.
4986 sudo lttng-sessiond --kmod-probes=my_subsys,usb
4991 To confirm that a probe module is loaded:
4998 lsmod | grep lttng_probe_usb
5002 To unload the loaded probe modules:
5004 * Kill the session daemon with `SIGTERM`:
5009 sudo pkill lttng-sessiond
5013 You can also use man:modprobe(8)'s `--remove` option if the session
5014 daemon terminates abnormally.
5017 [[controlling-tracing]]
5020 Once an application or a Linux kernel is
5021 <<instrumenting,instrumented>> for LTTng tracing,
5024 This section is divided in topics on how to use the various
5025 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5026 command-line tool>>, to _control_ the LTTng daemons and tracers.
5028 Note that the <<online-lttng-manpages,Online LTTng man pages>> are
5029 more comprehensive than the guides of this section. Refer to them if
5030 your use case is not included in this section.
5034 === Start a session daemon
5036 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5037 _before_ you can use the cmd:lttng command-line tool.
5039 You will see the following error when you run a command while no session
5043 Error: No session daemon is available
5046 The only command that automatically runs a session daemon is `create`,
5047 which you use to <<creating-destroying-tracing-sessions,create a tracing
5048 session>>. While this is most of the time the first operation that you
5049 do, sometimes it's not. Some examples are:
5051 * <<list-instrumentation-points,List the available instrumentation points>>.
5052 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5054 [[tracing-group]] Each Unix user must have its own running session
5055 daemon to trace user applications. The session daemon that the root user
5056 starts is the only one allowed to control the LTTng kernel tracer. Users
5057 that are part of the _tracing group_ can control the root session
5058 daemon. The default tracing group name is `tracing`; you can set it to
5059 something else with the `--group` option when you start the root session
5062 To start a user session daemon:
5064 * Run cmd:lttng-sessiond:
5069 lttng-sessiond --daemonize
5073 To start the root session daemon:
5075 * Run cmd:lttng-sessiond as the root user:
5080 sudo lttng-sessiond --daemonize
5084 In both cases, remove the `--daemonize` option to start the session
5085 daemon in foreground.
5087 To stop a session daemon, use cmd:kill on its process ID (standard
5090 Note that some Linux distributions could manage the LTTng session daemon
5091 as a service. In this case, you should use the service manager to
5092 start, restart, and stop session daemons.
5095 [[creating-destroying-tracing-sessions]]
5096 === Create and destroy a tracing session
5098 Almost all the LTTng control operations happen in the scope of
5099 a <<tracing-session,tracing session>>, which is the dialogue between the
5100 <<lttng-sessiond,session daemon>> and you.
5102 To create a tracing session with a generated name:
5104 * Use the `create` command:
5113 The created tracing session's name is `auto` followed by the
5116 To create a tracing session with a specific name:
5118 * Use the optional argument of the `create` command:
5123 lttng create my-session
5127 Replace `my-session` with the specific tracing session name.
5129 LTTng appends the creation date to the created tracing session's name.
5131 LTTng writes the traces of a tracing session in
5132 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5133 name of the tracing session. Note that the env:LTTNG_HOME environment
5134 variable defaults to `$HOME` if not set.
5136 To output LTTng traces to a non-default location:
5138 * Use the `--output` option of the `create` command:
5143 lttng create --output=/tmp/some-directory my-session
5147 You may create as many tracing sessions as you wish.
5149 To list all the existing tracing sessions for your Unix user:
5151 * Use the `list` command:
5160 When you create a tracing session, it is set as the _current tracing
5161 session_. The following man:lttng(1) commands operate on the current
5162 tracing session when you don't specify one:
5164 [role="list-3-cols"]
5180 To change the current tracing session:
5182 * Use the `set-session` command:
5187 lttng set-session new-session
5191 Replace `new-session` by the name of the new current tracing session.
5193 When you are done tracing in a given tracing session, you can destroy
5194 it. This operation frees the resources taken by the tracing session
5195 to destroy; it does not destroy the trace data that LTTng wrote for
5196 this tracing session.
5198 To destroy the current tracing session:
5200 * Use the `destroy` command:
5210 [[list-instrumentation-points]]
5211 === List the available instrumentation points
5213 The <<lttng-sessiond,session daemon>> can query the running instrumented
5214 user applications and the Linux kernel to get a list of available
5215 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5216 they are tracepoints and system calls. For the user space tracing
5217 domain, they are tracepoints. For the other tracing domains, they are
5220 To list the available instrumentation points:
5222 * Use the `list` command with the requested tracing domain's option
5226 * `--kernel`: Linux kernel tracepoints (your Unix user must be a root
5227 user, or it must be a member of the tracing group).
5228 * `--kernel --syscall`: Linux kernel system calls (your Unix user must
5229 be a root user, or it must be a member of the tracing group).
5230 * `--userspace`: user space tracepoints.
5231 * `--jul`: `java.util.logging` loggers.
5232 * `--log4j`: Apache log4j loggers.
5233 * `--python`: Python loggers.
5236 .List the available user space tracepoints.
5240 lttng list --userspace
5244 .List the available Linux kernel system call tracepoints.
5248 lttng list --kernel --syscall
5253 [[enabling-disabling-events]]
5254 === Create and enable an event rule
5256 Once you <<creating-destroying-tracing-sessions,create a tracing
5257 session>>, you can create <<event,event rules>> with the
5258 `enable-event` command.
5260 You specify each condition with a command-line option. The available
5261 condition options are shown in the following table.
5263 [role="growable",cols="asciidoc,asciidoc,default"]
5264 .Condition command-line options for the `enable-event` command.
5266 |Option |Description |Applicable tracing domains
5272 . +--probe=__ADDR__+
5273 . +--function=__ADDR__+
5276 Instead of using the default _tracepoint_ instrumentation type, use:
5278 . A Linux system call.
5279 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5280 . The entry and return points of a Linux function (symbol or address).
5284 |First positional argument.
5287 Tracepoint or system call name. In the case of a Linux KProbe or
5288 function, this is a custom name given to the event rule. With the
5289 JUL, log4j, and Python domains, this is a logger name.
5291 With a tracepoint, logger, or system call name, the last character
5292 can be `*` to match anything that remains.
5299 . +--loglevel=__LEVEL__+
5300 . +--loglevel-only=__LEVEL__+
5303 . Match only tracepoints or log statements with a logging level at
5304 least as severe as +__LEVEL__+.
5305 . Match only tracepoints or log statements with a logging level
5306 equal to +__LEVEL__+.
5308 You can get the list of available logging level names with
5309 `lttng enable-event --help`.
5311 |User space, JUL, log4j, and Python.
5313 |+--exclude=__EXCLUSIONS__+
5316 When you use a `*` character at the end of the tracepoint or logger
5317 name (first positional argument), exclude the specific names in the
5318 comma-delimited list +__EXCLUSIONS__+.
5321 User space, JUL, log4j, and Python.
5323 |+--filter=__EXPR__+
5326 Match only events which satisfy the expression +__EXPR__+.
5328 +__EXPR__+ is a C-like logical expression where identifiers are event
5329 fields (preceded with `$ctx.` for context fields). Nested expressions
5330 with `(` and `)`, and all the logical and comparison operators of the C
5331 language are supported. The precedence rules of those operators are the
5332 same as in the C language.
5334 When a comparison includes a non-existent event field, the whole filter
5335 expression evaluates to false.
5337 C integer and floating point number constants are supported, as well as
5338 literal strings between double quotes (`"`). Literal strings can
5339 contain a wildcard character (`*`) at the end to match anything that
5340 remains. This wildcard can be escaped using `\*`.
5342 Note that, although it is possible to use this option with the JUL,
5343 log4j, and Python tracing domains, the tracer evalutes the expression
5344 against the equivalent user space event.
5351 for more details about those command-line options.
5353 You attach an event rule to a <<channel,channel>> on creation. If you
5354 do not specify the channel with the `--channel` option, and if the event
5355 rule to create is the first in its <<domain,tracing domain>> for a given
5356 tracing session, then LTTng creates a _default channel_ for you. This
5357 default channel is reused in subsequent invocations of the
5358 `enable-event` command for the same tracing domain.
5360 An event rule is always enabled at creation time.
5362 The following examples show how you can combine the previous
5363 command-line options to create simple to more complex event rules.
5365 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5369 lttng enable-event --kernel sched_switch
5373 .Create an event rule matching four Linux kernel system calls (default channel).
5377 lttng enable-event --kernel --syscall open,write,read,close
5381 .Create an event rule matching a Linux kernel tracepoint with a filter expression (default channel).
5385 lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5388 IMPORTANT: Make sure to always quote the filter string when you
5389 use man:lttng(1) from a shell.
5392 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5396 lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5399 IMPORTANT: Make sure to always quote the wildcard character when you
5400 use man:lttng(1) from a shell.
5403 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5407 lttng enable-event --python my-app.'*' \
5408 --exclude='my-app.module,my-app.hello'
5412 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5416 lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5420 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5424 lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5428 The event rules of a given channel form a whitelist: as soon as an
5429 emitted event passes one of them, LTTng can record the event. For
5430 example, an event named `my_app:my_tracepoint` emitted from a user space
5431 tracepoint with a `TRACE_ERROR` log level passes both of the following
5436 lttng enable-event --userspace my_app:my_tracepoint
5437 lttng enable-event --userspace my_app:my_tracepoint \
5438 --loglevel=TRACE_INFO
5441 The second event rule is redundant: the first one includes
5445 [[disable-event-rule]]
5446 === Disable an event rule
5448 To disable an event rule that you <<enabling-disabling-events,created>>
5449 previously, use the `disable-event` command. This command disables _all_
5450 the event rules (of a given tracing domain and channel) which match an
5451 instrumentation point. The other conditions are not supported as of
5452 LTTng{nbsp}{revision}.
5454 The LTTng tracer does not record an emitted event which passes
5455 a _disabled_ event rule.
5457 .Disable an event rule matching a Python logger (default channel).
5461 lttng disable-event --python my-logger
5465 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5469 lttng disable-event --jul '*'
5473 .Disable _all_ the event rules of the default channel.
5475 The `--all-events` option is not, like the `--all` option of
5476 `enable-event`, the equivalent of the event name `*` (wildcard): it
5477 disables _all_ the event rules of a given channel.
5481 lttng disable-event --jul --all-events
5485 NOTE: You cannot delete an event rule once you create it.
5489 === Get the status of a tracing session
5491 To get the status of a tracing session, that is, its channels, event
5492 rules, and their attributes:
5494 * Use the `list` command with the tracing session's name:
5499 lttng list my-session
5503 Replace `my-session` with your tracing session's name.
5506 [[basic-tracing-session-control]]
5507 === Start and stop a tracing session
5509 Once you <<creating-destroying-tracing-sessions,create a tracing
5511 <<enabling-disabling-events,create one or more event rules>>,
5512 you can start and stop the tracers for this tracing session.
5514 To start tracing in the current tracing session:
5516 * Use the `start` command:
5525 To stop tracing in the current tracing session:
5527 * Use the `stop` command:
5536 LTTng is very flexible: you can launch user applications before
5537 or after the you start the tracers. The tracers only record the events
5538 if they pass enabled event rules and if they occur while the tracers are
5542 [[enabling-disabling-channels]]
5543 === Create a channel
5545 Once you create a tracing session, you can create a <<channel,channel>>
5546 with the `enable-channel` command.
5548 Note that LTTng automatically creates a default channel when, for a
5549 given <<domain,tracing domain>>, no channels exist and you
5550 <<enabling-disabling-events,create>> the first event rule. This default
5551 channel is named `channel0` and its attributes are set to reasonable
5552 values. Therefore, you only need to create a channel when you need
5553 non-default attributes.
5555 You specify each non-default channel attribute with a command-line
5556 option when you use the `enable-channel` command. The available
5557 command-line options are:
5559 [role="growable",cols="asciidoc,asciidoc"]
5560 .Command-line options for the `enable-channel` command.
5562 |Option |Description
5568 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
5569 the default _discard_ mode.
5571 |`--buffers-pid` (user space tracing domain only)
5574 Use the per-process <<channel-buffering-schemes,buffering scheme>>
5575 instead of the default per-user buffering scheme.
5577 |+--subbuf-size=__SIZE__+
5580 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
5581 either for each Unix user (default), or for each instrumented process.
5583 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5585 |+--num-subbuf=__COUNT__+
5588 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
5589 for each Unix user (default), or for each instrumented process.
5591 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5593 |+--tracefile-size=__SIZE__+
5596 Set the maximum size of each trace file that this channel writes within
5597 a stream to +__SIZE__+ bytes instead of no maximum.
5599 See <<tracefile-rotation,Trace file count and size>>.
5601 |+--tracefile-count=__COUNT__+
5604 Limit the number of trace files that this channel creates to
5605 +__COUNT__+ channels instead of no limit.
5607 See <<tracefile-rotation,Trace file count and size>>.
5609 |+--switch-timer=__PERIODUS__+
5612 Set the <<channel-switch-timer,switch timer period>>
5613 to +__PERIODUS__+{nbsp}µs.
5615 |+--read-timer=__PERIODUS__+
5618 Set the <<channel-read-timer,read timer period>>
5619 to +__PERIODUS__+{nbsp}µs.
5621 |+--output=__TYPE__+ (Linux kernel tracing domain only)
5624 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
5629 for more details about those command-line options.
5631 You can only create a channel in the Linux kernel and user space
5632 <<domain,tracing domains>>: other tracing domains have their own
5633 channel created on the fly when
5634 <<enabling-disabling-events,creating event rules>>.
5638 Because of a current LTTng limitation, you must create all channels
5639 _before_ you <<basic-tracing-session-control,start tracing>> in a given
5640 tracing session, that is, before the first time you run `lttng start`.
5642 Since LTTng automatically creates a default channel when you use the
5643 `enable-event` command with a specific tracing domain, you cannot, for
5644 example, create a Linux kernel event rule, start tracing, and then
5645 create a user space event rule, because no user space channel exists yet
5646 and it's too late to create one.
5648 For this reason, make sure to configure your channels properly
5649 before starting the tracers for the first time!
5652 The following examples show how you can combine the previous
5653 command-line options to create simple to more complex channels.
5655 .Create a Linux kernel channel with default attributes.
5659 lttng enable-channel --kernel my-channel
5663 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
5667 lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
5668 --buffers-pid my-channel
5672 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
5676 lttng enable-channel --kernel --tracefile-count=8 \
5677 --tracefile-size=4194304 my-channel
5681 .Create a user space channel in overwrite (or _flight recorder_) mode.
5685 lttng enable-channel --userspace --overwrite my-channel
5689 You can <<enabling-disabling-events,create>> the same event rule in
5690 two different channels:
5694 lttng enable-event --userspace --channel=my-channel app:tp
5695 lttng enable-event --userspace --channel=other-channel app:tp
5698 If both channels are enabled, when a tracepoint named `app:tp` is
5699 reached, LTTng records two events, one for each channel.
5703 === Disable a channel
5705 To disable a specific channel that you <<enabling-disabling-channels,created>>
5706 previously, use the `disable-channel` command.
5708 .Disable a specific Linux kernel channel.
5712 lttng disable-channel --kernel my-channel
5716 The state of a channel precedes the individual states of event rules
5717 attached to it: event rules which belong to a disabled channel, even if
5718 they are enabled, are also considered disabled.
5722 === Add context fields to a channel
5724 Event record fields in trace files provide important information about
5725 events that occured previously, but sometimes some external context may
5726 help you solve a problem faster. Examples of context fields are:
5728 * The **process ID**, **thread ID**, **process name**, and
5729 **process priority** of the thread in which the event occurs.
5730 * The **hostname** of the system on which the event occurs.
5731 * The current values of many possible **performance counters** using
5733 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
5735 ** Branch instructions, misses, and loads.
5738 To get the full list of available context fields, see
5739 `lttng add-context --help`. Some context fields are reserved for a
5740 specific <<domain,tracing domain>> (Linux kernel or user space).
5742 You add context fields to <<channel,channels>>. All the events
5743 that a channel with added context fields records contain those fields.
5745 To add context fields to one or all the channels of a given tracing
5746 session, use the `add-context` command.
5748 .Add context fields to all the channels of the current tracing session.
5750 The following command line adds the virtual process identifier and
5751 the per-thread CPU cycles count fields to all the user space channels
5752 of the current tracing session.
5756 lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
5760 .Add a context field to a specific channel.
5762 The following command line adds the thread identifier context field
5763 to the Linux kernel channel named `my-channel` in the current
5768 lttng add-context --kernel --channel=my-channel --type=tid
5772 NOTE: You cannot remove context fields from a channel once you add it.
5777 === Track process IDs
5779 It's often useful to allow only specific process IDs (PIDs) to emit
5780 events. For example, you may wish to record all the system calls made by
5781 a given process (Ă la http://linux.die.net/man/1/strace[strace]).
5783 The `track` and `untrack` commands serve this purpose. Both commands
5784 operate on a whitelist of process IDs. You _add_ entries to this
5785 whitelist with the `track` command and remove entries with the `untrack`
5786 command. Any process which has one of the PIDs in the whitelist is
5787 allowed to emit LTTng events which pass an enabled <<event,event rule>>.
5789 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
5790 process with a given tracked ID exit and another process be given this
5791 ID, then the latter would also be allowed to emit events.
5793 .Track and untrack process IDs.
5795 For the sake of the following example, assume the target system has 16
5799 <<creating-destroying-tracing-sessions,create a tracing session>>,
5800 the whitelist contains all the possible PIDs:
5803 .All PIDs are tracked.
5804 image::track-all.png[]
5806 When the whitelist is full and you use the `track` command to specify
5807 some PIDs to track, LTTng first clears the whitelist, then it tracks
5808 the specific PIDs. After:
5812 lttng track --pid=3,4,7,10,13
5818 .PIDs 3, 4, 7, 10, and 13 are tracked.
5819 image::track-3-4-7-10-13.png[]
5821 You can add more PIDs to the whitelist afterwards:
5825 lttng track --pid=1,15,16
5831 .PIDs 1, 15, and 16 are added to the whitelist.
5832 image::track-1-3-4-7-10-13-15-16.png[]
5834 The `untrack` command removes entries from the PID tracker's whitelist.
5835 Given the previous example, the following command:
5839 lttng untrack --pid=3,7,10,13
5842 leads to this whitelist:
5845 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
5846 image::track-1-4-15-16.png[]
5848 LTTng can track all possible PIDs again using the `--all` option:
5852 lttng track --pid --all
5855 The result is, again:
5858 .All PIDs are tracked.
5859 image::track-all.png[]
5862 .Track only specific PIDs
5864 A very typical use case with PID tracking is to start with an empty
5865 whitelist, then <<basic-tracing-session-control,start the tracers>>,
5866 and then add PIDs manually while tracers are active. You can accomplish
5867 this by using the `--all` option of the `untrack` command to clear the
5868 whitelist after you create a tracing session:
5872 lttng untrack --pid --all
5878 .No PIDs are tracked.
5879 image::untrack-all.png[]
5881 If you trace with this whitelist configuration, the tracer records no
5882 events for this <<domain,tracing domain>> because no processes are
5883 tracked. You can use the `track` command as usual to track specific
5888 lttng track --pid=6,11
5894 .PIDs 6 and 11 are tracked.
5895 image::track-6-11.png[]
5900 [[saving-loading-tracing-session]]
5901 === Save and load tracing session configurations
5903 Configuring a <<tracing-session,tracing session>> can be long. Some of
5904 the tasks involved are:
5906 * <<enabling-disabling-channels,Create channels>> with
5907 specific attributes.
5908 * <<adding-context,Add context fields>> to specific channels.
5909 * <<enabling-disabling-events,Create event rules>> with specific log
5910 level and filter conditions.
5912 If you use LTTng to solve real world problems, chances are you have to
5913 record events using the same tracing session setup over and over,
5914 modifying a few variables each time in your instrumented program
5915 or environment. To avoid constant tracing session reconfiguration,
5916 the cmd:lttng command-line tool can save and load tracing session
5917 configurations to/from XML files.
5919 To save a given tracing session configuration:
5921 * Use the `save` command:
5926 lttng save my-session
5930 Replace `my-session` with the name of the tracing session to save.
5932 LTTng saves tracing session configurations to
5933 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
5934 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
5935 the `--output-path` option to change this destination directory.
5937 LTTng saves all configuration parameters, for example:
5939 * The tracing session name.
5940 * The trace data output path.
5941 * The channels with their state and all their attributes.
5942 * The context fields you added to channels.
5943 * The event rules with their state, log level and filter conditions.
5945 To load a tracing session:
5947 * Use the `load` command:
5952 lttng load my-session
5956 Replace `my-session` with the name of the tracing session to load.
5958 When LTTng loads a configuration, it restores your saved tracing session
5959 as if you just configured it manually.
5961 See man:lttng(1) for the complete list of command-line options. You
5962 can also save and load all many sessions at a time, and decide in which
5963 directory to output the XML files.
5966 [[sending-trace-data-over-the-network]]
5967 === Send trace data over the network
5969 LTTng can send the recorded trace data to a remote system over the
5970 network instead of writing it to the local file system.
5972 To send the trace data over the network:
5974 . On the _remote_ system (which can also be the target system),
5975 start an LTTng <<lttng-relayd,relay daemon>>:
5984 . On the _target_ system, create a tracing session configured to
5985 send trace data over the network:
5990 lttng create my-session --set-url=net://remote-system
5994 Replace `remote-system` by the host name or IP address of the
5995 remote system. See `lttng create --help` for the exact URL format.
5997 . On the target system, use the cmd:lttng command-line tool as usual.
5998 When tracing is active, the target's consumer daemon sends sub-buffers
5999 to the relay daemon running on the remote system intead of flushing
6000 them to the local file system. The relay daemon writes the received
6001 packets to the local file system.
6003 The relay daemon writes trace files to
6004 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6005 +__hostname__+ is the host name of the target system and +__session__+
6006 is the tracing session name. Note that the env:LTTNG_HOME environment
6007 variable defaults to `$HOME` if not set. Use the `--output` option of
6008 cmd:lttng-relayd to write trace files to another base directory.
6013 === View events as LTTng emits them (noch:{LTTng} live)
6015 LTTng live is a network protocol implemented by the
6016 <<lttng-relayd,relay daemon>> to allow compatible trace viewers to
6017 display events as LTTng emits them on the target system while tracing
6020 The relay daemon creates a _tee_: it forwards the trace data to both
6021 the local file system and to connected live viewers:
6024 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6029 . On the _target system_, create a <<tracing-session,tracing session>>
6035 lttng create --live my-session
6039 This spawns a local relay daemon.
6041 . Start the live viewer and configure it to connect to the relay
6042 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6047 babeltrace --input-format=lttng-live net://localhost/host/hostname/my-session
6054 * `hostname` with the host name of the target system.
6055 * `my-session` with the name of the tracing session to view.
6058 . Configure the tracing session as usual with the cmd:lttng
6059 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6061 You can list the available live tracing sessions with Babeltrace:
6065 babeltrace --input-format=lttng-live net://localhost
6068 You can start the relay daemon on another system. In this case, you need
6069 to specify the relay daemon's URL when you create the tracing session
6070 with the `--set-url` option. You also need to replace `localhost`
6071 in the procedure above with the host name of the system on which the
6072 relay daemon is running.
6074 See man:lttng(1) and man:lttng-relayd(8) for the complete list of
6075 command-line options.
6079 [[taking-a-snapshot]]
6080 === Take a snapshot of the current sub-buffers of a tracing session
6082 The normal behavior of LTTng is to append full sub-buffers to growing
6083 trace data files. This is ideal to keep a full history of the events
6084 that occurred on the target system, but it can
6085 represent too much data in some situations. For example, you may wish
6086 to trace your application continuously until some critical situation
6087 happens, in which case you only need the latest few recorded
6088 events to perform the desired analysis, not multi-gigabyte trace files.
6090 With the `snapshot` command, you can take a snapshot of the current
6091 sub-buffers of a given <<tracing-session,tracing session>>. LTTng can
6092 write the snapshot to the local file system or send it over the network.
6096 . Create a tracing session in _snapshot mode_:
6101 lttng create --snapshot my-session
6105 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6106 <<channel,channels>> created in this mode is automatically set to
6107 _overwrite_ (flight recorder mode).
6109 . Configure the tracing session as usual with the cmd:lttng
6110 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6112 . **Optional**: When you need to take a snapshot, stop tracing.
6114 You can take a snapshot when the tracers are active, but if you stop
6115 them first, you are sure that the data in the sub-buffers does not
6116 change before you actually take the snapshot.
6123 lttng snapshot record --name=my-first-snapshot
6127 LTTng writes the current sub-buffers of all the current tracing
6128 session's channels to trace files on the local file system. Those trace
6129 files have `my-first-snapshot` in their name.
6131 There is no difference between the format of a normal trace file and the
6132 format of a snapshot: viewers of LTTng traces also support LTTng
6135 By default, LTTng writes snapshot files to the path shown by
6136 `lttng snapshot list-output`. You can change this path or decide to send
6137 snapshots over the network using either:
6139 . An output path or URL that you specify when you create the
6141 . An snapshot output path or URL that you add using
6142 `lttng snapshot add-output`
6143 . An output path or URL that you provide directly to the
6144 `lttng snapshot record` command.
6146 Method 3 overrides method 2, which overrides method 1. When you
6147 specify a URL, a relay daemon must listen on a remote system (see
6148 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6153 === Use the machine interface
6155 With any command of the cmd:lttng command-line tool, you can use the
6156 `--mi=xml` argument (before the command name) to get an XML machine
6157 interface output, for example:
6161 lttng --mi=xml enable-event --kernel --syscall open
6164 A schema definition (XSD) is
6165 https://github.com/lttng/lttng-tools/blob/stable-{revision}/src/common/mi_lttng.xsd[available]
6166 to ease the integration with external tools as much as possible.
6170 [[persistent-memory-file-systems]]
6171 === Record trace data on persistent memory file systems
6173 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6174 (NVRAM) is random-access memory that retains its information when power
6175 is turned off (non-volatile). Systems with such memory can store data
6176 structures in RAM and retrieve them after a reboot, without flushing
6177 to typical _storage_.
6179 Linux supports NVRAM file systems thanks to either
6180 http://pramfs.sourceforge.net/[PRAMFS] or
6181 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6182 (requires Linux 4.1+).
6184 This section does not describe how to operate such file systems;
6185 we assume that you have a working persistent memory file system.
6187 When you create a <<tracing-session,tracing session>>, you can specify
6188 the path of the shared memory holding the sub-buffers. If you specify a
6189 location on an NVRAM file system, then you can retrieve the latest
6190 recorded trace data when the system reboots after a crash.
6192 To record trace data on a persistent memory file system and retrieve the
6193 trace data after a system crash:
6195 . Create a tracing session with a sub-buffer shared memory path located
6196 on an NVRAM file system:
6201 lttng create --shm-path=/path/to/shm
6205 . Configure the tracing session as usual with the cmd:lttng
6206 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6208 . After a system crash, use the cmd:lttng-crash command-line tool to
6209 view the trace data recorded on the NVRAM file system:
6214 lttng-crash /path/to/shm
6218 The binary layout of the ring buffer files is not exactly the same as
6219 the trace files layout. This is why you need to use the cmd:lttng-crash
6220 utility instead of your preferred trace viewer directly.
6222 To convert the ring buffer files to LTTng trace files:
6224 * Use the `--extract` option of cmd:lttng-crash:
6229 lttng-crash --extract=/path/to/trace /path/to/shm
6233 See man:lttng-crash(1) for the complete list of command-line options.
6239 This section presents various references for LTTng packages such as
6240 links to online manpages, tables that the rest of the text needs,
6241 descriptions of library functions, and more.
6244 [[online-lttng-manpages]]
6245 === Online noch:{LTTng} manpages
6247 LTTng packages currently install the following link:/man[man pages],
6248 available online using the links below:
6252 ** man:lttng-crash(1)
6253 ** man:lttng-sessiond(8)
6254 ** man:lttng-relayd(8)
6256 ** man:lttng-gen-tp(1)
6258 ** man:lttng-ust-cyg-profile(3)
6259 ** man:lttng-ust-dl(3)
6263 === noch:{LTTng-UST}
6265 This section presents references of the LTTng-UST package.
6269 ==== noch:{LTTng-UST} library (+liblttng‑ust+)
6271 The LTTng-UST library, or `liblttng-ust`, is the main shared object
6272 against which user applications are linked to make LTTng user space
6275 The <<c-application,C application>> guide shows the complete
6276 process to instrument, build and run a C/$$C++$$ application using
6277 LTTng-UST, while this section contains a few important tables.
6280 [[liblttng-ust-tp-fields]]
6281 ===== Tracepoint fields macros (for `TP_FIELDS()`)
6283 The available macros to define tracepoint fields, which you must use
6284 within `TP_FIELDS()` in `TRACEPOINT_EVENT()`, are:
6286 [role="func-desc growable",cols="asciidoc,asciidoc"]
6287 .Available macros to define LTTng-UST tracepoint fields
6289 |Macro |Description and parameters
6292 +ctf_integer(__t__, __n__, __e__)+
6294 +ctf_integer_nowrite(__t__, __n__, __e__)+
6296 Standard integer, displayed in base 10.
6299 Integer C type (`int`, `long`, `size_t`, ...).
6305 Argument expression.
6307 |+ctf_integer_hex(__t__, __n__, __e__)+
6309 Standard integer, displayed in base 16.
6318 Argument expression.
6320 |+ctf_integer_network(__t__, __n__, __e__)+
6322 Integer in network byte order (big-endian), displayed in base 10.
6331 Argument expression.
6333 |+ctf_integer_network_hex(__t__, __n__, __e__)+
6335 Integer in network byte order, displayed in base 16.
6344 Argument expression.
6347 +ctf_float(__t__, __n__, __e__)+
6349 +ctf_float_nowrite(__t__, __n__, __e__)+
6351 Floating point number.
6354 Floating point number C type (`float` or `double`).
6360 Argument expression.
6363 +ctf_string(__n__, __e__)+
6365 +ctf_string_nowrite(__n__, __e__)+
6367 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
6373 Argument expression.
6376 +ctf_array(__t__, __n__, __e__, __s__)+
6378 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
6380 Statically-sized array of integers
6383 Array element C type.
6389 Argument expression.
6395 +ctf_array_text(__t__, __n__, __e__, __s__)+
6397 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
6399 Statically-sized array, printed as text.
6401 The string does not need to be null-terminated.
6404 Array element C type (always `char`).
6410 Argument expression.
6416 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
6418 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
6420 Dynamically-sized array of integers.
6422 The type of +__E__+ must be unsigned.
6425 Array element C type.
6431 Argument expression.
6434 Length expression C type.
6440 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
6442 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
6444 Dynamically-sized array, displayed as text.
6446 The string does not need to be null-terminated.
6448 The type of +__E__+ must be unsigned.
6450 The behaviour is undefined if +__e__+ is `NULL`.
6453 Sequence element C type (always `char`).
6459 Argument expression.
6462 Length expression C type.
6468 The `_nowrite` versions omit themselves from the session trace, but are
6469 otherwise identical. This means the tracer does not write the `_nowrite`
6470 fields to the trace. Their primary purpose is to make some of the event
6471 context available to the <<enabling-disabling-events,event filters>>
6472 without having to commit the data to sub-buffers.
6475 [[liblttng-ust-tracepoint-loglevel]]
6476 ===== Tracepoint log levels (for `TRACEPOINT_LOGLEVEL()`)
6478 The following table shows the available log level values for the
6479 `TRACEPOINT_LOGLEVEL()` macro:
6485 Action must be taken immediately.
6488 Critical conditions.
6497 Normal, but significant, condition.
6500 Informational message.
6502 `TRACE_DEBUG_SYSTEM`::
6503 Debug information with system-level scope (set of programs).
6505 `TRACE_DEBUG_PROGRAM`::
6506 Debug information with program-level scope (set of processes).
6508 `TRACE_DEBUG_PROCESS`::
6509 Debug information with process-level scope (set of modules).
6511 `TRACE_DEBUG_MODULE`::
6512 Debug information with module (executable/library) scope (set of units).
6514 `TRACE_DEBUG_UNIT`::
6515 Debug information with compilation unit scope (set of functions).
6517 `TRACE_DEBUG_FUNCTION`::
6518 Debug information with function-level scope.
6520 `TRACE_DEBUG_LINE`::
6521 Debug information with line-level scope (TRACEPOINT_EVENT default).
6524 Debug-level message.
6526 Log levels `TRACE_EMERG` through `TRACE_INFO` and `TRACE_DEBUG` match
6527 http://man7.org/linux/man-pages/man3/syslog.3.html[syslog]
6528 level semantics. Log levels `TRACE_DEBUG_SYSTEM` through `TRACE_DEBUG`
6529 offer more fine-grained selection of debug information.
6532 [[lttng-modules-ref]]
6533 === noch:{LTTng-modules}
6535 This section presents references of the LTTng-modules package.
6539 [[lttng-modules-tp-fields]]
6540 ==== Tracepoint fields macros (for `TP_FIELDS()`)
6542 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
6543 tracepoint fields, which must be listed within `TP_FIELDS()` in
6544 `LTTNG_TRACEPOINT_EVENT()`, are:
6546 [role="func-desc growable",cols="asciidoc,asciidoc"]
6547 .Available macros to define LTTng-modules tracepoint fields
6549 |Macro |Description and parameters
6552 +ctf_integer(__t__, __n__, __e__)+
6554 +ctf_integer_nowrite(__t__, __n__, __e__)+
6556 +ctf_user_integer(__t__, __n__, __e__)+
6558 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
6560 Standard integer, displayed in base 10.
6563 Integer C type (`int`, `long`, `size_t`, ...).
6569 Argument expression.
6572 +ctf_integer_hex(__t__, __n__, __e__)+
6574 +ctf_user_integer_hex(__t__, __n__, __e__)+
6576 Standard integer, displayed in base 16.
6585 Argument expression.
6587 |+ctf_integer_oct(__t__, __n__, __e__)+
6589 Standard integer, displayed in base 8.
6598 Argument expression.
6601 +ctf_integer_network(__t__, __n__, __e__)+
6603 +ctf_user_integer_network(__t__, __n__, __e__)+
6605 Integer in network byte order (big-endian), displayed in base 10.
6614 Argument expression.
6617 +ctf_integer_network_hex(__t__, __n__, __e__)+
6619 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
6621 Integer in network byte order, displayed in base 16.
6630 Argument expression.
6633 +ctf_string(__n__, __e__)+
6635 +ctf_string_nowrite(__n__, __e__)+
6637 +ctf_user_string(__n__, __e__)+
6639 +ctf_user_string_nowrite(__n__, __e__)+
6641 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
6647 Argument expression.
6650 +ctf_array(__t__, __n__, __e__, __s__)+
6652 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
6654 +ctf_user_array(__t__, __n__, __e__, __s__)+
6656 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
6658 Statically-sized array of integers
6661 Array element C type.
6667 Argument expression.
6673 +ctf_array_text(__t__, __n__, __e__, __s__)+
6675 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
6677 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
6679 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
6681 Statically-sized array, printed as text.
6683 The string does not need to be null-terminated.
6686 Array element C type (always `char`).
6692 Argument expression.
6698 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
6700 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
6702 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
6704 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
6706 Dynamically-sized array of integers.
6708 The type of +__E__+ must be unsigned.
6711 Array element C type.
6717 Argument expression.
6720 Length expression C type.
6725 |+ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
6727 Dynamically-sized array of integers, displayed in base 16.
6729 The type of +__E__+ must be unsigned.
6732 Array element C type.
6738 Argument expression.
6741 Length expression C type.
6746 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
6748 Dynamically-sized array of integers in network byte order (big-endian),
6749 displayed in base 10.
6751 The type of +__E__+ must be unsigned.
6754 Array element C type.
6760 Argument expression.
6763 Length expression C type.
6769 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
6771 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
6773 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
6775 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
6777 Dynamically-sized array, displayed as text.
6779 The string does not need to be null-terminated.
6781 The type of +__E__+ must be unsigned.
6783 The behaviour is undefined if +__e__+ is `NULL`.
6786 Sequence element C type (always `char`).
6792 Argument expression.
6795 Length expression C type.
6801 Use the `_user` versions when the argument expression, `e`, is
6802 a user space address. In the cases of `ctf_user_integer*()` and
6803 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
6806 The `_nowrite` versions omit themselves from the session trace, but are
6807 otherwise identical. This means the `_nowrite` fields won't be written
6808 in the recorded trace. Their primary purpose is to make some
6809 of the event context available to the
6810 <<enabling-disabling-events,event filters>> without having to
6811 commit the data to sub-buffers.
6817 Terms related to LTTng and to tracing in general:
6820 The http://diamon.org/babeltrace[Babeltrace] project, which includes
6821 the cmd:babeltrace command, some libraries, and Python bindings.
6823 <<channel-buffering-schemes,buffering scheme>>::
6824 A layout of sub-buffers applied to a given channel.
6826 <<channel,channel>>::
6827 An entity which is responsible for a set of ring buffers.
6829 <<event,Event rules>> are always attached to a specific channel.
6832 A reference of time for a tracer.
6834 <<lttng-consumerd,consumer daemon>>::
6835 A process which is responsible for consuming the full sub-buffers
6836 and write them to a file system or send them over the network.
6838 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
6839 mode in which the tracer _discards_ new event records when there's no
6840 sub-buffer space left to store them.
6843 The consequence of the execution of an instrumentation
6844 point, like a tracepoint that you manually place in some source code,
6845 or a Linux kernel KProbe.
6847 An event is said to _occur_ at a specific time. Different actions can
6848 be taken upon the occurance of an event, like record the event's payload
6851 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
6852 The mechanism by which event records of a given channel are lost
6853 (not recorded) when there is no sub-buffer space left to store them.
6855 [[def-event-name]]event name::
6856 The name of an event, which is also the name of the event record.
6857 This is also called the _instrumentation point name_.
6860 A record, in a trace, of the payload of an event which occured.
6862 <<event,event rule>>::
6863 Set of conditions which must be satisfied for one or more occuring
6864 events to be recorded.
6866 `java.util.logging`::
6868 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
6870 <<instrumenting,instrumentation>>::
6871 The use of LTTng probes to make a piece of software traceable.
6873 instrumentation point::
6874 A point in the execution path of a piece of software that, when
6875 reached by this execution, can emit an event.
6877 instrumentation point name::
6878 See _<<def-event-name,event name>>_.
6881 A http://logging.apache.org/log4j/1.2/[logging library] for Java
6882 developed by the Apache Software Foundation.
6885 Level of severity of a log statement or user space
6886 instrumentation point.
6889 The _Linux Trace Toolkit: next generation_ project.
6891 <<lttng-cli,cmd:lttng>>::
6892 A command-line tool provided by the LTTng-tools project which you
6893 can use to send and receive control messages to and from a
6897 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
6898 which is a set of analyzing programs that are used to obtain a
6899 higher level view of an LTTng trace.
6901 cmd:lttng-consumerd::
6902 The name of the consumer daemon program.
6905 A utility provided by the LTTng-tools project which can convert
6906 ring buffer files (usually
6907 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
6910 LTTng Documentation::
6913 <<lttng-live,LTTng live>>::
6914 A communication protocol between the relay daemon and live viewers
6915 which makes it possible to see events "live", as they are received by
6918 <<lttng-modules,LTTng-modules>>::
6919 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
6920 which contains the Linux kernel modules to make the Linux kernel
6921 instrumentation points available for LTTng tracing.
6924 The name of the relay daemon program.
6926 cmd:lttng-sessiond::
6927 The name of the session daemon program.
6930 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
6931 contains the various programs and libraries used to
6932 <<controlling-tracing,control tracing>>.
6934 <<lttng-ust,LTTng-UST>>::
6935 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
6936 contains libraries to instrument user applications.
6938 <<lttng-ust-agents,LTTng-UST Java agent>>::
6939 A Java package provided by the LTTng-UST project to allow the
6940 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
6943 <<lttng-ust-agents,LTTng-UST Python agent>>::
6944 A Python package provided by the LTTng-UST project to allow the
6945 LTTng instrumentation of Python logging statements.
6947 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
6948 The event loss mode in which new event records overwrite older
6949 event records when there's no sub-buffer space left to store them.
6951 <<channel-buffering-schemes,per-process buffering>>::
6952 A buffering scheme in which each instrumented process has its own
6953 sub-buffers for a given user space channel.
6955 <<channel-buffering-schemes,per-user buffering>>::
6956 A buffering scheme in which all the processes of a Unix user share the
6957 same sub-buffer for a given user space channel.
6959 <<lttng-relayd,relay daemon>>::
6960 A process which is responsible for receiving the trace data sent by
6961 a distant consumer daemon.
6964 A set of sub-buffers.
6966 <<lttng-sessiond,session daemon>>::
6967 A process which receives control commands from you and orchestrates
6968 the tracers and various LTTng daemons.
6970 <<taking-a-snapshot,snapshot>>::
6971 A copy of the current data of all the sub-buffers of a given tracing
6972 session, saved as trace files.
6975 One part of an LTTng ring buffer which contains event records.
6978 The time information attached to an event when it is emitted.
6981 A set of files which are the concatenations of one or more
6982 flushed sub-buffers.
6985 The action of recording the events emitted by an application
6986 or by a system, or to initiate such recording by controlling
6990 The http://tracecompass.org[Trace Compass] project and application.
6993 An instrumentation point using the tracepoint mechanism of the Linux
6994 kernel or of LTTng-UST.
6996 tracepoint definition::
6997 The definition of a single tracepoint.
7000 The name of a tracepoint.
7002 tracepoint provider::
7003 A set of functions providing tracepoints to an instrumented user
7006 Not to be confused with a _tracepoint provider package_: many tracepoint
7007 providers can exist within a tracepoint provider package.
7009 tracepoint provider package::
7010 One or more tracepoint providers compiled as an object file or as
7014 A software which records emitted events.
7016 <<domain,tracing domain>>::
7017 A namespace for event sources.
7020 The Unix group in which a Unix user can be to be allowed to trace the
7023 <<tracing-session,tracing session>>::
7024 A stateful dialogue between you and a <<lttng-sessiond,session
7028 An application running in user space, as opposed to a Linux kernel
7029 module, for example.