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 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 ** You can attach <<java-application-context,Java application-specific
79 context fields>> to a <<channel,channel>> with the
80 man:lttng-add-context(1) command:
85 lttng add-context --jul --type='$app.retriever:cur_msg_id'
89 Here, `$app` is the prefix of all application-specific context fields,
90 `retriever` names a _context information retriever_ defined at the
91 application level, and `cur_msg_id` names a context field read from this
94 Both the `java.util.logging` and Apache log4j <<domain,tracing domains>>
97 ** You can use Java application-specific <<adding-context,context>>
98 fields in the <<enabling-disabling-events,filter expression>> of an
104 lttng enable-event --log4j my_logger \
105 --filter='$app.retriever:cur_msg_id == 23'
109 ** New `lttng status` command which is the equivalent of +lttng list
110 __CUR__+, where +__CUR__+ is the name of the current
111 <<tracing-session,tracing session>>.
113 See man:lttng-status(1).
115 ** New `lttng metadata regenerate` command to
116 <<metadata-regenerate,regenerate the metadata file of an LTTng
117 trace>> at any moment. This command is meant to be used to resample
118 the wall time following a major
119 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
120 so that a system which boots with an incorrect wall time can be
121 traced before its wall time is NTP-corrected.
123 See man:lttng-metadata(1).
125 ** New command-line interface warnings when <<event,event records>> or
126 whole sub-buffers are
127 <<channel-overwrite-mode-vs-discard-mode,lost>>. The warning messages
128 are printed when a <<tracing-session,tracing session>> is
129 <<basic-tracing-session-control,stopped>> (man:lttng-stop(1)
132 * **User space tracing**:
133 ** Shared object base address dump in order to map <<event,event
134 records>> to original source location (file and line number).
136 If you attach the `ip` and `vpid` <<adding-context,context fields>> to a
137 user space <<channel,channel>> and if you use the
138 <<liblttng-ust-dl,path:{liblttng-ust-dl.so} helper>>, you can retrieve
139 the source location where a given event record was generated.
141 The http://diamon.org/babeltrace/[Babeltrace] trace viewer supports this
142 state dump and those context fields since version 1.4 to print the
143 source location of a given event record. http://tracecompass.org/[Trace
144 Compass] also supports this since version 2.0.
146 ** A <<java-application,Java application>> which uses
147 `java.util.logging` now adds an LTTng-UST log handler to the desired
150 The previous workflow was to initialize the LTTng-UST Java agent
151 by calling `LTTngAgent.getLTTngAgent()`. This had the effect of adding
152 an LTTng-UST log handler to the root loggers.
154 ** A <<java-application,Java application>> which uses Apache log4j now
155 adds an LTTng-UST log appender to the desired log4j loggers.
157 The previous workflow was to initialize the LTTng-UST Java agent
158 by calling `LTTngAgent.getLTTngAgent()`. This had the effect of adding
159 an LTTng-UST appender to the root loggers.
161 ** Any <<java-application,Java application>> can provide
162 <<java-application-context,dynamic context fields>> while running
163 thanks to a new API provided by the <<lttng-ust-agents,LTTng-UST Java
164 agent>>. You can require LTTng to record specific context fields in
165 event records, and you can use them in the filter expression of
166 <<event,event rules>>.
168 * **Linux kernel tracing**:
169 ** The LTTng kernel modules can now be built into a Linux kernel image,
170 that is, not as loadable modules.
173 https://github.com/lttng/lttng-modules/blob/stable-{revision}/README.md#kernel-built-in-support[`README.md`]
176 ** New instrumentation:
177 *** ARM64 architecture support.
179 *** x86 `irq_vectors`.
180 ** New <<adding-context,context fields>>:
183 *** `need_reschedule`
184 *** `migratable` (specific to RT-Preempt)
185 ** Clock source plugin support for advanced cases where a custom source
186 of time is needed to timestamp LTTng event records.
188 See https://github.com/lttng/lttng-modules/blob/stable-{revision}/lttng-clock.h[`lttng-clock.h`]
189 for an overview of the small API.
192 ** The link:/man[man pages] of the man:lttng(1) command-line tool are
193 split into one man page per command (à la Git), for example:
198 man lttng-enable-event
202 You can also use the `--help` option of any man:lttng(1) command to
205 The content and formatting of all the LTTng man pages has improved
212 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
213 generation_ is a modern toolkit for tracing Linux systems and
214 applications. So your first question might be:
221 As the history of software engineering progressed and led to what
222 we now take for granted--complex, numerous and
223 interdependent software applications running in parallel on
224 sophisticated operating systems like Linux--the authors of such
225 components, software developers, began feeling a natural
226 urge to have tools that would ensure the robustness and good performance
227 of their masterpieces.
229 One major achievement in this field is, inarguably, the
230 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
231 an essential tool for developers to find and fix bugs. But even the best
232 debugger won't help make your software run faster, and nowadays, faster
233 software means either more work done by the same hardware, or cheaper
234 hardware for the same work.
236 A _profiler_ is often the tool of choice to identify performance
237 bottlenecks. Profiling is suitable to identify _where_ performance is
238 lost in a given software. The profiler outputs a profile, a statistical
239 summary of observed events, which you may use to discover which
240 functions took the most time to execute. However, a profiler won't
241 report _why_ some identified functions are the bottleneck. Bottlenecks
242 might only occur when specific conditions are met, conditions that are
243 sometimes impossible to capture by a statistical profiler, or impossible
244 to reproduce with an application altered by the overhead of an
245 event-based profiler. For a thorough investigation of software
246 performance issues, a history of execution is essential, with the
247 recorded values of variables and context fields you choose, and
248 with as little influence as possible on the instrumented software. This
249 is where tracing comes in handy.
251 _Tracing_ is a technique used to understand what goes on in a running
252 software system. The software used for tracing is called a _tracer_,
253 which is conceptually similar to a tape recorder. When recording,
254 specific instrumentation points placed in the software source code
255 generate events that are saved on a giant tape: a _trace_ file. You
256 can trace user applications and the operating system at the same time,
257 opening the possibility of resolving a wide range of problems that would
258 otherwise be extremely challenging.
260 Tracing is often compared to _logging_. However, tracers and loggers are
261 two different tools, serving two different purposes. Tracers are
262 designed to record much lower-level events that occur much more
263 frequently than log messages, often in the range of thousands per
264 second, with very little execution overhead. Logging is more appropriate
265 for a very high-level analysis of less frequent events: user accesses,
266 exceptional conditions (errors and warnings, for example), database
267 transactions, instant messaging communications, and such. Simply put,
268 logging is one of the many use cases that can be satisfied with tracing.
270 The list of recorded events inside a trace file can be read manually
271 like a log file for the maximum level of detail, but it is generally
272 much more interesting to perform application-specific analyses to
273 produce reduced statistics and graphs that are useful to resolve a
274 given problem. Trace viewers and analyzers are specialized tools
277 In the end, this is what LTTng is: a powerful, open source set of
278 tools to trace the Linux kernel and user applications at the same time.
279 LTTng is composed of several components actively maintained and
280 developed by its link:/community/#where[community].
283 [[lttng-alternatives]]
284 === Alternatives to noch:{LTTng}
286 Excluding proprietary solutions, a few competing software tracers
289 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
290 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
291 user scripts and is responsible for loading code into the
292 Linux kernel for further execution and collecting the outputted data.
293 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
294 subsystem in the Linux kernel in which a virtual machine can execute
295 programs passed from the user space to the kernel. You can attach
296 such programs to tracepoints and KProbes thanks to a system call, and
297 they can output data to the user space when executed thanks to
298 different mechanisms (pipe, VM register values, and eBPF maps, to name
300 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
301 is the de facto function tracer of the Linux kernel. Its user
302 interface is a set of special files in sysfs.
303 * https://perf.wiki.kernel.org/[perf] is
304 a performance analyzing tool for Linux which supports hardware
305 performance counters, tracepoints, as well as other counters and
306 types of probes. perf's controlling utility is the cmd:perf command
308 * http://linux.die.net/man/1/strace[strace]
309 is a command-line utility which records system calls made by a
310 user process, as well as signal deliveries and changes of process
311 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
312 to fulfill its function.
313 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
314 analyze Linux kernel events. You write scripts, or _chisels_ in
315 sysdig's jargon, in Lua and sysdig executes them while the system is
316 being traced or afterwards. sysdig's interface is the cmd:sysdig
317 command-line tool as well as the curses-based cmd:csysdig tool.
318 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
319 user space tracer which uses custom user scripts to produce plain text
320 traces. SystemTap converts the scripts to the C language, and then
321 compiles them as Linux kernel modules which are loaded to produce
322 trace data. SystemTap's primary user interface is the cmd:stap
325 The main distinctive features of LTTng is that it produces correlated
326 kernel and user space traces, as well as doing so with the lowest
327 overhead amongst other solutions. It produces trace files in the
328 http://diamon.org/ctf[CTF] format, a file format optimized
329 for the production and analyses of multi-gigabyte data.
331 LTTng is the result of more than 10 years of active open source
332 development by a community of passionate developers.
333 LTTng{nbsp}{revision} is currently available on major desktop and server
336 The main interface for tracing control is a single command-line tool
337 named cmd:lttng. The latter can create several tracing sessions, enable
338 and disable events on the fly, filter events efficiently with custom
339 user expressions, start and stop tracing, and much more. LTTng can
340 record the traces on the file system or send them over the network, and
341 keep them totally or partially. You can view the traces once tracing
342 becomes inactive or in real-time.
344 <<installing-lttng,Install LTTng now>> and
345 <<getting-started,start tracing>>!
351 **LTTng** is a set of software <<plumbing,components>> which interact to
352 <<instrumenting,instrument>> the Linux kernel and user applications, and
353 to <<controlling-tracing,control tracing>> (start and stop
354 tracing, enable and disable event rules, and the rest). Those
355 components are bundled into the following packages:
357 * **LTTng-tools**: Libraries and command-line interface to
359 * **LTTng-modules**: Linux kernel modules to instrument and
361 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
362 trace user applications.
364 Most distributions mark the LTTng-modules and LTTng-UST packages as
365 optional when installing LTTng-tools (which is always required). In the
366 following sections, we always provide the steps to install all three,
369 * You only need to install LTTng-modules if you intend to trace the
371 * You only need to install LTTng-UST if you intend to trace user
375 .Availability of LTTng{nbsp}{revision} for major Linux distributions as of 2 December 2016.
377 |Distribution |Available in releases |Alternatives
379 |https://www.ubuntu.com/[Ubuntu]
380 |<<ubuntu,Ubuntu{nbsp}16.10 _Yakkety Yak_>>.
381 |LTTng{nbsp}{revision} for Ubuntu{nbsp}14.04 _Trusty Tahr_
382 and Ubuntu{nbsp}16.04 _Xenial Xerus_:
383 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
385 LTTng{nbsp}2.9 for Ubuntu{nbsp}14.04 _Trusty Tahr_
386 and Ubuntu{nbsp}16.04 _Xenial Xerus_:
387 link:/docs/v2.9#doc-ubuntu-ppa[use the LTTng Stable{nbsp}2.9 PPA].
389 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
390 other Ubuntu releases.
392 |https://getfedora.org/[Fedora]
393 |<<fedora,Fedora{nbsp}25>>.
394 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
395 other Fedora releases.
397 |https://www.debian.org/[Debian]
398 |<<debian,Debian "stretch" (testing)>>.
399 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
400 previous Debian releases.
402 |https://www.opensuse.org/[openSUSE]
404 |<<building-from-source,Build LTTng{nbsp}{revision} from source>>.
406 |https://www.archlinux.org/[Arch Linux]
408 |link:/docs/v2.9#doc-arch-linux[LTTng{nbsp}2.9 from the AUR].
410 |https://alpinelinux.org/[Alpine Linux]
411 |<<alpine-linux,Alpine Linux "edge">>.
412 |LTTng{nbsp}{revision} for Alpine Linux{nbsp}3.5 (not released yet).
414 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
415 other Alpine Linux releases.
417 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
418 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
421 |https://buildroot.org/[Buildroot]
422 |<<buildroot,Buildroot 2016.11>>.
423 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
424 other Buildroot releases.
426 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
427 https://www.yoctoproject.org/[Yocto]
428 |<<oe-yocto,Yocto Project{nbsp}2.2 _Morty_>> (`openembedded-core` layer).
429 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
430 other OpenEmbedded releases.
435 === [[ubuntu-official-repositories]]Ubuntu
437 LTTng{nbsp}{revision} is available on Ubuntu{nbsp}16.10 _Yakkety Yak_.
438 For previous releases of Ubuntu, <<ubuntu-ppa,use the LTTng
439 Stable{nbsp}{revision} PPA>>.
441 To install LTTng{nbsp}{revision} on Ubuntu{nbsp}16.10 _Yakkety Yak_:
443 . Install the main LTTng{nbsp}{revision} packages:
448 sudo apt-get install lttng-tools
449 sudo apt-get install lttng-modules-dkms
450 sudo apt-get install liblttng-ust-dev
454 . **If you need to instrument and trace
455 <<java-application,Java applications>>**, install the LTTng-UST
461 sudo apt-get install liblttng-ust-agent-java
465 . **If you need to instrument and trace
466 <<python-application,Python{nbsp}3 applications>>**, install the
467 LTTng-UST Python agent:
472 sudo apt-get install python3-lttngust
478 ==== noch:{LTTng} Stable {revision} PPA
480 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
481 Stable{nbsp}{revision} PPA] offers the latest stable
482 LTTng{nbsp}{revision} packages for:
484 * Ubuntu{nbsp}14.04 _Trusty Tahr_
485 * Ubuntu{nbsp}16.04 _Xenial Xerus_
487 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
489 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
495 sudo apt-add-repository ppa:lttng/stable-2.8
500 . Install the main LTTng{nbsp}{revision} packages:
505 sudo apt-get install lttng-tools
506 sudo apt-get install lttng-modules-dkms
507 sudo apt-get install liblttng-ust-dev
511 . **If you need to instrument and trace
512 <<java-application,Java applications>>**, install the LTTng-UST
518 sudo apt-get install liblttng-ust-agent-java
522 . **If you need to instrument and trace
523 <<python-application,Python{nbsp}3 applications>>**, install the
524 LTTng-UST Python agent:
529 sudo apt-get install python3-lttngust
537 To install LTTng{nbsp}{revision} on Fedora{nbsp}25:
539 . Install the LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision}
545 sudo yum install lttng-tools
546 sudo yum install lttng-ust
550 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
556 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
557 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
558 cd lttng-modules-2.8.* &&
560 sudo make modules_install &&
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.
585 To install LTTng{nbsp}{revision} on Debian "stretch" (testing):
587 . Install the main LTTng{nbsp}{revision} packages:
592 sudo apt-get install lttng-modules-dkms
593 sudo apt-get install liblttng-ust-dev
594 sudo apt-get install lttng-tools
598 . **If you need to instrument and trace <<java-application,Java
599 applications>>**, install the LTTng-UST Java agent:
604 sudo apt-get install liblttng-ust-agent-java
608 . **If you need to instrument and trace <<python-application,Python
609 applications>>**, install the LTTng-UST Python agent:
614 sudo apt-get install python3-lttngust
622 To install LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision} on
625 . Make sure your system is
626 https://wiki.alpinelinux.org/wiki/Edge[configured for "edge"].
627 . Enable the _testing_ repository by uncommenting the corresponding
628 line in path:{/etc/apk/repositories}.
629 . Add the LTTng packages:
634 sudo apk add lttng-tools
635 sudo apk add lttng-ust-dev
639 To install LTTng-modules{nbsp}{revision} (Linux kernel tracing support)
640 on Alpine Linux "edge":
642 . Add the vanilla Linux kernel:
647 apk add linux-vanilla linux-vanilla-dev
651 . Reboot with the vanilla Linux kernel.
652 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
658 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
659 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
660 cd lttng-modules-2.8.* &&
662 sudo make modules_install &&
668 [[enterprise-distributions]]
669 === RHEL, SUSE, and other enterprise distributions
671 To install LTTng on enterprise Linux distributions, such as Red Hat
672 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
673 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
679 To install LTTng{nbsp}{revision} on Buildroot{nbsp}2016.11:
681 . Launch the Buildroot configuration tool:
690 . In **Kernel**, check **Linux kernel**.
691 . In **Toolchain**, check **Enable WCHAR support**.
692 . In **Target packages**{nbsp}→ **Debugging, profiling and benchmark**,
693 check **lttng-modules** and **lttng-tools**.
694 . In **Target packages**{nbsp}→ **Libraries**{nbsp}→
695 **Other**, check **lttng-libust**.
699 === OpenEmbedded and Yocto
701 LTTng{nbsp}{revision} recipes are available in the
702 http://layers.openembedded.org/layerindex/branch/master/layer/openembedded-core/[`openembedded-core`]
703 layer for Yocto Project{nbsp}2.2 _Morty_ under the following names:
709 With BitBake, the simplest way to include LTTng recipes in your target
710 image is to add them to `IMAGE_INSTALL_append` in path:{conf/local.conf}:
713 IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"
718 . Select a machine and an image recipe.
719 . Click **Edit image recipe**.
720 . Under the **All recipes** tab, search for **lttng**.
721 . Check the desired LTTng recipes.
724 .Java and Python application instrumentation and tracing
726 If you need to instrument and trace <<java-application,Java
727 applications>> on openSUSE, you need to build and install
728 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
729 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
730 `--enable-java-agent-all` options to the `configure` script, depending
731 on which Java logging framework you use.
733 If you need to instrument and trace <<python-application,Python
734 applications>> on openSUSE, you need to build and install
735 LTTng-UST{nbsp}{revision} from source and pass the
736 `--enable-python-agent` option to the `configure` script.
740 [[building-from-source]]
741 === Build from source
743 To build and install LTTng{nbsp}{revision} from source:
745 . Using your distribution's package manager, or from source, install
746 the following dependencies of LTTng-tools and LTTng-UST:
749 * https://sourceforge.net/projects/libuuid/[libuuid]
750 * http://directory.fsf.org/wiki/Popt[popt]
751 * http://liburcu.org/[Userspace RCU]
752 * http://www.xmlsoft.org/[libxml2]
755 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
761 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
762 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
763 cd lttng-modules-2.8.* &&
765 sudo make modules_install &&
770 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
776 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.8.tar.bz2 &&
777 tar -xf lttng-ust-latest-2.8.tar.bz2 &&
778 cd lttng-ust-2.8.* &&
788 .Java and Python application tracing
790 If you need to instrument and trace <<java-application,Java
791 applications>>, pass the `--enable-java-agent-jul`,
792 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
793 `configure` script, depending on which Java logging framework you use.
795 If you need to instrument and trace <<python-application,Python
796 applications>>, pass the `--enable-python-agent` option to the
797 `configure` script. You can set the `PYTHON` environment variable to the
798 path to the Python interpreter for which to install the LTTng-UST Python
806 By default, LTTng-UST libraries are installed to
807 dir:{/usr/local/lib}, which is the de facto directory in which to
808 keep self-compiled and third-party libraries.
810 When <<building-tracepoint-providers-and-user-application,linking an
811 instrumented user application with `liblttng-ust`>>:
813 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
815 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
816 man:gcc(1), man:g++(1), or man:clang(1).
820 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
826 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
827 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
828 cd lttng-tools-2.8.* &&
836 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
837 previous steps automatically for a given version of LTTng and confine
838 the installed files in a specific directory. This can be useful to test
839 LTTng without installing it on your system.
845 This is a short guide to get started quickly with LTTng kernel and user
848 Before you follow this guide, make sure to <<installing-lttng,install>>
851 This tutorial walks you through the steps to:
853 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
854 . <<tracing-your-own-user-application,Trace a user application>> written
856 . <<viewing-and-analyzing-your-traces,View and analyze the
860 [[tracing-the-linux-kernel]]
861 === Trace the Linux kernel
863 The following command lines start with cmd:sudo because you need root
864 privileges to trace the Linux kernel. You can avoid using cmd:sudo if
865 your Unix user is a member of the <<tracing-group,tracing group>>.
867 . Create a <<tracing-session,tracing session>> which writes its traces
868 to dir:{/tmp/my-kernel-trace}:
873 sudo lttng create my-kernel-session --output=/tmp/my-kernel-trace
877 . List the available kernel tracepoints and system calls:
883 lttng list --kernel --syscall
887 . Create <<event,event rules>> which match the desired instrumentation
888 point names, for example the `sched_switch` and `sched_process_fork`
889 tracepoints, and the man:open(2) and man:close(2) system calls:
894 sudo lttng enable-event --kernel sched_switch,sched_process_fork
895 sudo lttng enable-event --kernel --syscall open,close
899 You can also create an event rule which matches _all_ the Linux kernel
900 tracepoints (this will generate a lot of data when tracing):
905 sudo lttng enable-event --kernel --all
909 . <<basic-tracing-session-control,Start tracing>>:
918 . Do some operation on your system for a few seconds. For example,
919 load a website, or list the files of a directory.
920 . <<basic-tracing-session-control,Stop tracing>> and destroy the
931 The man:lttng-destroy(1) command does not destroy the trace data; it
932 only destroys the state of the tracing session.
934 . For the sake of this example, make the recorded trace accessible to
940 sudo chown -R $(whoami) /tmp/my-kernel-trace
944 See <<viewing-and-analyzing-your-traces,View and analyze the
945 recorded events>> to view the recorded events.
948 [[tracing-your-own-user-application]]
949 === Trace a user application
951 This section steps you through a simple example to trace a
952 _Hello world_ program written in C.
954 To create the traceable user application:
956 . Create the tracepoint provider header file, which defines the
957 tracepoints and the events they can generate:
963 #undef TRACEPOINT_PROVIDER
964 #define TRACEPOINT_PROVIDER hello_world
966 #undef TRACEPOINT_INCLUDE
967 #define TRACEPOINT_INCLUDE "./hello-tp.h"
969 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
972 #include <lttng/tracepoint.h>
982 ctf_string(my_string_field, my_string_arg)
983 ctf_integer(int, my_integer_field, my_integer_arg)
987 #endif /* _HELLO_TP_H */
989 #include <lttng/tracepoint-event.h>
993 . Create the tracepoint provider package source file:
999 #define TRACEPOINT_CREATE_PROBES
1000 #define TRACEPOINT_DEFINE
1002 #include "hello-tp.h"
1006 . Build the tracepoint provider package:
1011 gcc -c -I. hello-tp.c
1015 . Create the _Hello World_ application source file:
1022 #include "hello-tp.h"
1024 int main(int argc, char *argv[])
1028 puts("Hello, World!\nPress Enter to continue...");
1031 * The following getchar() call is only placed here for the purpose
1032 * of this demonstration, to pause the application in order for
1033 * you to have time to list its tracepoints. It is not
1039 * A tracepoint() call.
1041 * Arguments, as defined in hello-tp.h:
1043 * 1. Tracepoint provider name (required)
1044 * 2. Tracepoint name (required)
1045 * 3. my_integer_arg (first user-defined argument)
1046 * 4. my_string_arg (second user-defined argument)
1048 * Notice the tracepoint provider and tracepoint names are
1049 * NOT strings: they are in fact parts of variables that the
1050 * macros in hello-tp.h create.
1052 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
1054 for (x = 0; x < argc; ++x) {
1055 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
1058 puts("Quitting now!");
1059 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
1066 . Build the application:
1075 . Link the application with the tracepoint provider package,
1076 `liblttng-ust`, and `libdl`:
1081 gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
1085 Here's the whole build process:
1088 .User space tracing tutorial's build steps.
1089 image::ust-flow.png[]
1091 To trace the user application:
1093 . Run the application with a few arguments:
1098 ./hello world and beyond
1107 Press Enter to continue...
1111 . Start an LTTng <<lttng-sessiond,session daemon>>:
1116 lttng-sessiond --daemonize
1120 Note that a session daemon might already be running, for example as
1121 a service that the distribution's service manager started.
1123 . List the available user space tracepoints:
1128 lttng list --userspace
1132 You see the `hello_world:my_first_tracepoint` tracepoint listed
1133 under the `./hello` process.
1135 . Create a <<tracing-session,tracing session>>:
1140 lttng create my-user-space-session
1144 . Create an <<event,event rule>> which matches the
1145 `hello_world:my_first_tracepoint` event name:
1150 lttng enable-event --userspace hello_world:my_first_tracepoint
1154 . <<basic-tracing-session-control,Start tracing>>:
1163 . Go back to the running `hello` application and press Enter. The
1164 program executes all `tracepoint()` instrumentation points and exits.
1165 . <<basic-tracing-session-control,Stop tracing>> and destroy the
1176 The man:lttng-destroy(1) command does not destroy the trace data; it
1177 only destroys the state of the tracing session.
1179 By default, LTTng saves the traces in
1180 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1181 where +__name__+ is the tracing session name. The
1182 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1184 See <<viewing-and-analyzing-your-traces,View and analyze the
1185 recorded events>> to view the recorded events.
1188 [[viewing-and-analyzing-your-traces]]
1189 === View and analyze the recorded events
1191 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1192 kernel>> and <<tracing-your-own-user-application,Trace a user
1193 application>> tutorials, you can inspect the recorded events.
1195 Many tools are available to read LTTng traces:
1197 * **cmd:babeltrace** is a command-line utility which converts trace
1198 formats; it supports the format that LTTng produces, CTF, as well as a
1199 basic text output which can be ++grep++ed. The cmd:babeltrace command
1200 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1201 * Babeltrace also includes
1202 **https://www.python.org/[Python] bindings** so
1203 that you can easily open and read an LTTng trace with your own script,
1204 benefiting from the power of Python.
1205 * http://tracecompass.org/[**Trace Compass**]
1206 is a graphical user interface for viewing and analyzing any type of
1207 logs or traces, including LTTng's.
1208 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1209 project which includes many high-level analyses of LTTng kernel
1210 traces, like scheduling statistics, interrupt frequency distribution,
1211 top CPU usage, and more.
1213 NOTE: This section assumes that the traces recorded during the previous
1214 tutorials were saved to their default location, in the
1215 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1216 environment variable defaults to `$HOME` if not set.
1219 [[viewing-and-analyzing-your-traces-bt]]
1220 ==== Use the cmd:babeltrace command-line tool
1222 The simplest way to list all the recorded events of a trace is to pass
1223 its path to cmd:babeltrace with no options:
1227 babeltrace ~/lttng-traces/my-user-space-session*
1230 cmd:babeltrace finds all traces recursively within the given path and
1231 prints all their events, merging them in chronological order.
1233 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1238 babeltrace /tmp/my-kernel-trace | grep _switch
1241 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1242 count the recorded events:
1246 babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
1250 [[viewing-and-analyzing-your-traces-bt-python]]
1251 ==== Use the Babeltrace Python bindings
1253 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1254 is useful to isolate events by simple matching using man:grep(1) and
1255 similar utilities. However, more elaborate filters, such as keeping only
1256 event records with a field value falling within a specific range, are
1257 not trivial to write using a shell. Moreover, reductions and even the
1258 most basic computations involving multiple event records are virtually
1259 impossible to implement.
1261 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1262 to read the event records of an LTTng trace sequentially and compute the
1263 desired information.
1265 The following script accepts an LTTng Linux kernel trace path as its
1266 first argument and prints the short names of the top 5 running processes
1267 on CPU 0 during the whole trace:
1272 from collections import Counter
1278 if len(sys.argv) != 2:
1279 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1280 print(msg, file=sys.stderr)
1283 # A trace collection contains one or more traces
1284 col = babeltrace.TraceCollection()
1286 # Add the trace provided by the user (LTTng traces always have
1288 if col.add_trace(sys.argv[1], 'ctf') is None:
1289 raise RuntimeError('Cannot add trace')
1291 # This counter dict contains execution times:
1293 # task command name -> total execution time (ns)
1294 exec_times = Counter()
1296 # This contains the last `sched_switch` timestamp
1300 for event in col.events:
1301 # Keep only `sched_switch` events
1302 if event.name != 'sched_switch':
1305 # Keep only events which happened on CPU 0
1306 if event['cpu_id'] != 0:
1310 cur_ts = event.timestamp
1316 # Previous task command (short) name
1317 prev_comm = event['prev_comm']
1319 # Initialize entry in our dict if not yet done
1320 if prev_comm not in exec_times:
1321 exec_times[prev_comm] = 0
1323 # Compute previous command execution time
1324 diff = cur_ts - last_ts
1326 # Update execution time of this command
1327 exec_times[prev_comm] += diff
1329 # Update last timestamp
1333 for name, ns in exec_times.most_common(5):
1335 print('{:20}{} s'.format(name, s))
1340 if __name__ == '__main__':
1341 sys.exit(0 if top5proc() else 1)
1348 python3 top5proc.py /tmp/my-kernel-trace/kernel
1354 swapper/0 48.607245889 s
1355 chromium 7.192738188 s
1356 pavucontrol 0.709894415 s
1357 Compositor 0.660867933 s
1358 Xorg.bin 0.616753786 s
1361 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1362 weren't using the CPU that much when tracing, its first position in the
1367 == [[understanding-lttng]]Core concepts
1369 From a user's perspective, the LTTng system is built on a few concepts,
1370 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1371 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1372 Understanding how those objects relate to eachother is key in mastering
1375 The core concepts are:
1377 * <<tracing-session,Tracing session>>
1378 * <<domain,Tracing domain>>
1379 * <<channel,Channel and ring buffer>>
1380 * <<"event","Instrumentation point, event rule, event, and event record">>
1386 A _tracing session_ is a stateful dialogue between you and
1387 a <<lttng-sessiond,session daemon>>. You can
1388 <<creating-destroying-tracing-sessions,create a new tracing
1389 session>> with the `lttng create` command.
1391 Anything that you do when you control LTTng tracers happens within a
1392 tracing session. In particular, a tracing session:
1395 * Has its own set of trace files.
1396 * Has its own state of activity (started or stopped).
1397 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1399 * Has its own <<channel,channels>> which have their own
1400 <<event,event rules>>.
1403 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1404 image::concepts.png[]
1406 Those attributes and objects are completely isolated between different
1409 A tracing session is analogous to a cash machine session:
1410 the operations you do on the banking system through the cash machine do
1411 not alter the data of other users of the same system. In the case of
1412 the cash machine, a session lasts as long as your bank card is inside.
1413 In the case of LTTng, a tracing session lasts from the `lttng create`
1414 command to the `lttng destroy` command.
1417 .Each Unix user has its own set of tracing sessions.
1418 image::many-sessions.png[]
1421 [[tracing-session-mode]]
1422 ==== Tracing session mode
1424 LTTng can send the generated trace data to different locations. The
1425 _tracing session mode_ dictates where to send it. The following modes
1426 are available in LTTng{nbsp}{revision}:
1429 LTTng writes the traces to the file system of the machine being traced
1432 Network streaming mode::
1433 LTTng sends the traces over the network to a
1434 <<lttng-relayd,relay daemon>> running on a remote system.
1437 LTTng does not write the traces by default. Instead, you can request
1438 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1439 current tracing buffers, and to write it to the target's file system
1440 or to send it over the network to a <<lttng-relayd,relay daemon>>
1441 running on a remote system.
1444 This mode is similar to the network streaming mode, but a live
1445 trace viewer can connect to the distant relay daemon to
1446 <<lttng-live,view event records as LTTng generates them>> by
1453 A _tracing domain_ is a namespace for event sources. A tracing domain
1454 has its own properties and features.
1456 There are currently five available tracing domains:
1460 * `java.util.logging` (JUL)
1464 You must specify a tracing domain when using some commands to avoid
1465 ambiguity. For example, since all the domains support named tracepoints
1466 as event sources (instrumentation points that you manually insert in the
1467 source code), you need to specify a tracing domain when
1468 <<enabling-disabling-events,creating an event rule>> because all the
1469 tracing domains could have tracepoints with the same names.
1471 Some features are reserved to specific tracing domains. Dynamic function
1472 entry and return instrumentation points, for example, are currently only
1473 supported in the Linux kernel tracing domain, but support for other
1474 tracing domains could be added in the future.
1476 You can create <<channel,channels>> in the Linux kernel and user space
1477 tracing domains. The other tracing domains have a single default
1482 === Channel and ring buffer
1484 A _channel_ is an object which is responsible for a set of ring buffers.
1485 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1486 tracer emits an event, it can record it to one or more
1487 sub-buffers. The attributes of a channel determine what to do when
1488 there's no space left for a new event record because all sub-buffers
1489 are full, where to send a full sub-buffer, and other behaviours.
1491 A channel is always associated to a <<domain,tracing domain>>. The
1492 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1493 a default channel which you cannot configure.
1495 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1496 an event, it records it to the sub-buffers of all
1497 the enabled channels with a satisfied event rule, as long as those
1498 channels are part of active <<tracing-session,tracing sessions>>.
1501 [[channel-buffering-schemes]]
1502 ==== Per-user vs. per-process buffering schemes
1504 A channel has at least one ring buffer _per CPU_. LTTng always
1505 records an event to the ring buffer associated to the CPU on which it
1508 Two _buffering schemes_ are available when you
1509 <<enabling-disabling-channels,create a channel>> in the
1510 user space <<domain,tracing domain>>:
1512 Per-user buffering::
1513 Allocate one set of ring buffers--one per CPU--shared by all the
1514 instrumented processes of each Unix user.
1518 .Per-user buffering scheme.
1519 image::per-user-buffering.png[]
1522 Per-process buffering::
1523 Allocate one set of ring buffers--one per CPU--for each
1524 instrumented process.
1528 .Per-process buffering scheme.
1529 image::per-process-buffering.png[]
1532 The per-process buffering scheme tends to consume more memory than the
1533 per-user option because systems generally have more instrumented
1534 processes than Unix users running instrumented processes. However, the
1535 per-process buffering scheme ensures that one process having a high
1536 event throughput won't fill all the shared sub-buffers of the same
1539 The Linux kernel tracing domain has only one available buffering scheme
1540 which is to allocate a single set of ring buffers for the whole system.
1541 This scheme is similar to the per-user option, but with a single, global
1542 user "running" the kernel.
1545 [[channel-overwrite-mode-vs-discard-mode]]
1546 ==== Overwrite vs. discard event loss modes
1548 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1549 arc in the following animation) of a specific channel's ring buffer.
1550 When there's no space left in a sub-buffer, the tracer marks it as
1551 consumable (red) and another, empty sub-buffer starts receiving the
1552 following event records. A <<lttng-consumerd,consumer daemon>>
1553 eventually consumes the marked sub-buffer (returns to white).
1556 [role="docsvg-channel-subbuf-anim"]
1561 In an ideal world, sub-buffers are consumed faster than they are filled,
1562 as is the case in the previous animation. In the real world,
1563 however, all sub-buffers can be full at some point, leaving no space to
1564 record the following events.
1566 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1567 available, it is acceptable to lose event records when the alternative
1568 would be to cause substantial delays in the instrumented application's
1569 execution. LTTng privileges performance over integrity; it aims at
1570 perturbing the traced system as little as possible in order to make
1571 tracing of subtle race conditions and rare interrupt cascades possible.
1573 When it comes to losing event records because no empty sub-buffer is
1574 available, the channel's _event loss mode_ determines what to do. The
1575 available event loss modes are:
1578 Drop the newest event records until a the tracer
1579 releases a sub-buffer.
1582 Clear the sub-buffer containing the oldest event records and start
1583 writing the newest event records there.
1585 This mode is sometimes called _flight recorder mode_ because it's
1587 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1588 always keep a fixed amount of the latest data.
1590 Which mechanism you should choose depends on your context: prioritize
1591 the newest or the oldest event records in the ring buffer?
1593 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1594 as soon as a there's no space left for a new event record, whereas in
1595 discard mode, the tracer only discards the event record that doesn't
1598 In discard mode, LTTng increments a count of lost event records when
1599 an event record is lost and saves this count to the trace. In
1600 overwrite mode, LTTng keeps no information when it overwrites a
1601 sub-buffer before consuming it.
1603 There are a few ways to decrease your probability of losing event
1605 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1606 how you can fine-une the sub-buffer count and size of a channel to
1607 virtually stop losing event records, though at the cost of greater
1611 [[channel-subbuf-size-vs-subbuf-count]]
1612 ==== Sub-buffer count and size
1614 When you <<enabling-disabling-channels,create a channel>>, you can
1615 set its number of sub-buffers and their size.
1617 Note that there is noticeable CPU overhead introduced when
1618 switching sub-buffers (marking a full one as consumable and switching
1619 to an empty one for the following events to be recorded). Knowing this,
1620 the following list presents a few practical situations along with how
1621 to configure the sub-buffer count and size for them:
1623 * **High event throughput**: In general, prefer bigger sub-buffers to
1624 lower the risk of losing event records.
1626 Having bigger sub-buffers also ensures a lower sub-buffer switching
1629 The number of sub-buffers is only meaningful if you create the channel
1630 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1631 other sub-buffers are left unaltered.
1633 * **Low event throughput**: In general, prefer smaller sub-buffers
1634 since the risk of losing event records is low.
1636 Because events occur less frequently, the sub-buffer switching frequency
1637 should remain low and thus the tracer's overhead should not be a
1640 * **Low memory system**: If your target system has a low memory
1641 limit, prefer fewer first, then smaller sub-buffers.
1643 Even if the system is limited in memory, you want to keep the
1644 sub-buffers as big as possible to avoid a high sub-buffer switching
1647 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1648 which means event data is very compact. For example, the average
1649 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1650 sub-buffer size of 1{nbsp}MiB is considered big.
1652 The previous situations highlight the major trade-off between a few big
1653 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1654 frequency vs. how much data is lost in overwrite mode. Assuming a
1655 constant event throughput and using the overwrite mode, the two
1656 following configurations have the same ring buffer total size:
1659 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1664 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1665 switching frequency, but if a sub-buffer overwrite happens, half of
1666 the event records so far (4{nbsp}MiB) are definitely lost.
1667 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1668 overhead as the previous configuration, but if a sub-buffer
1669 overwrite happens, only the eighth of event records so far are
1672 In discard mode, the sub-buffers count parameter is pointless: use two
1673 sub-buffers and set their size according to the requirements of your
1677 [[channel-switch-timer]]
1678 ==== Switch timer period
1680 The _switch timer period_ is an important configurable attribute of
1681 a channel to ensure periodic sub-buffer flushing.
1683 When the _switch timer_ expires, a sub-buffer switch happens. You can
1684 set the switch timer period attribute when you
1685 <<enabling-disabling-channels,create a channel>> to ensure that event
1686 data is consumed and committed to trace files or to a distant relay
1687 daemon periodically in case of a low event throughput.
1690 [role="docsvg-channel-switch-timer"]
1695 This attribute is also convenient when you use big sub-buffers to cope
1696 with a sporadic high event throughput, even if the throughput is
1700 [[channel-read-timer]]
1701 ==== Read timer period
1703 By default, the LTTng tracers use a notification mechanism to signal a
1704 full sub-buffer so that a consumer daemon can consume it. When such
1705 notifications must be avoided, for example in real-time applications,
1706 you can use the channel's _read timer_ instead. When the read timer
1707 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1708 consumable sub-buffers.
1711 [[tracefile-rotation]]
1712 ==== Trace file count and size
1714 By default, trace files can grow as large as needed. You can set the
1715 maximum size of each trace file that a channel writes when you
1716 <<enabling-disabling-channels,create a channel>>. When the size of
1717 a trace file reaches the channel's fixed maximum size, LTTng creates
1718 another file to contain the next event records. LTTng appends a file
1719 count to each trace file name in this case.
1721 If you set the trace file size attribute when you create a channel, the
1722 maximum number of trace files that LTTng creates is _unlimited_ by
1723 default. To limit them, you can also set a maximum number of trace
1724 files. When the number of trace files reaches the channel's fixed
1725 maximum count, the oldest trace file is overwritten. This mechanism is
1726 called _trace file rotation_.
1730 === Instrumentation point, event rule, event, and event record
1732 An _event rule_ is a set of conditions which must be **all** satisfied
1733 for LTTng to record an occuring event.
1735 You set the conditions when you <<enabling-disabling-events,create
1738 You always attach an event rule to <<channel,channel>> when you create
1741 When an event passes the conditions of an event rule, LTTng records it
1742 in one of the attached channel's sub-buffers.
1744 The available conditions, as of LTTng{nbsp}{revision}, are:
1746 * The event rule _is enabled_.
1747 * The instrumentation point's type _is{nbsp}T_.
1748 * The instrumentation point's name (sometimes called _event name_)
1749 _matches{nbsp}N_, but _is not{nbsp}E_.
1750 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1751 _is exactly{nbsp}L_.
1752 * The fields of the event's payload _satisfy_ a filter
1753 expression{nbsp}__F__.
1755 As you can see, all the conditions but the dynamic filter are related to
1756 the event rule's status or to the instrumentation point, not to the
1757 occurring events. This is why, without a filter, checking if an event
1758 passes an event rule is not a dynamic task: when you create or modify an
1759 event rule, all the tracers of its tracing domain enable or disable the
1760 instrumentation points themselves once. This is possible because the
1761 attributes of an instrumentation point (type, name, and log level) are
1762 defined statically. In other words, without a dynamic filter, the tracer
1763 _does not evaluate_ the arguments of an instrumentation point unless it
1764 matches an enabled event rule.
1766 Note that, for LTTng to record an event, the <<channel,channel>> to
1767 which a matching event rule is attached must also be enabled, and the
1768 tracing session owning this channel must be active.
1771 .Logical path from an instrumentation point to an event record.
1772 image::event-rule.png[]
1774 .Event, event record, or event rule?
1776 With so many similar terms, it's easy to get confused.
1778 An **event** is the consequence of the execution of an _instrumentation
1779 point_, like a tracepoint that you manually place in some source code,
1780 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1781 time. Different actions can be taken upon the occurance of an event,
1782 like record the event's payload to a buffer.
1784 An **event record** is the representation of an event in a sub-buffer. A
1785 tracer is responsible for capturing the payload of an event, current
1786 context variables, the event's ID, and the event's timestamp. LTTng
1787 can append this sub-buffer to a trace file.
1789 An **event rule** is a set of conditions which must all be satisfied for
1790 LTTng to record an occuring event. Events still occur without
1791 satisfying event rules, but LTTng does not record them.
1796 == Components of noch:{LTTng}
1798 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1799 to call LTTng a simple _tool_ since it is composed of multiple
1800 interacting components. This section describes those components,
1801 explains their respective roles, and shows how they connect together to
1802 form the LTTng ecosystem.
1804 The following diagram shows how the most important components of LTTng
1805 interact with user applications, the Linux kernel, and you:
1808 .Control and trace data paths between LTTng components.
1809 image::plumbing.png[]
1811 The LTTng project incorporates:
1813 * **LTTng-tools**: Libraries and command-line interface to
1814 control tracing sessions.
1815 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1816 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1817 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1818 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1819 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1820 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1822 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1823 headers to instrument and trace any native user application.
1824 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1825 *** `liblttng-ust-libc-wrapper`
1826 *** `liblttng-ust-pthread-wrapper`
1827 *** `liblttng-ust-cyg-profile`
1828 *** `liblttng-ust-cyg-profile-fast`
1829 *** `liblttng-ust-dl`
1830 ** User space tracepoint provider source files generator command-line
1831 tool (man:lttng-gen-tp(1)).
1832 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1833 Java applications using `java.util.logging` or
1834 Apache log4j 1.2 logging.
1835 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1836 Python applications using the standard `logging` package.
1837 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1839 ** LTTng kernel tracer module.
1840 ** Tracing ring buffer kernel modules.
1841 ** Probe kernel modules.
1842 ** LTTng logger kernel module.
1846 === Tracing control command-line interface
1849 .The tracing control command-line interface.
1850 image::plumbing-lttng-cli.png[]
1852 The _man:lttng(1) command-line tool_ is the standard user interface to
1853 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1854 is part of LTTng-tools.
1856 The cmd:lttng tool is linked with
1857 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1858 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1860 The cmd:lttng tool has a Git-like interface:
1864 lttng <general options> <command> <command options>
1867 The <<controlling-tracing,Tracing control>> section explores the
1868 available features of LTTng using the cmd:lttng tool.
1871 [[liblttng-ctl-lttng]]
1872 === Tracing control library
1875 .The tracing control library.
1876 image::plumbing-liblttng-ctl.png[]
1878 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1879 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1880 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1882 The <<lttng-cli,cmd:lttng command-line tool>>
1883 is linked with `liblttng-ctl`.
1885 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1890 #include <lttng/lttng.h>
1893 Some objects are referenced by name (C string), such as tracing
1894 sessions, but most of them require to create a handle first using
1895 `lttng_create_handle()`.
1897 The best available developer documentation for `liblttng-ctl` is, as of
1898 LTTng{nbsp}{revision}, its installed header files. Every function and
1899 structure is thoroughly documented.
1903 === User space tracing library
1906 .The user space tracing library.
1907 image::plumbing-liblttng-ust.png[]
1909 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1910 is the LTTng user space tracer. It receives commands from a
1911 <<lttng-sessiond,session daemon>>, for example to
1912 enable and disable specific instrumentation points, and writes event
1913 records to ring buffers shared with a
1914 <<lttng-consumerd,consumer daemon>>.
1915 `liblttng-ust` is part of LTTng-UST.
1917 Public C header files are installed beside `liblttng-ust` to
1918 instrument any <<c-application,C or $$C++$$ application>>.
1920 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1921 packages, use their own library providing tracepoints which is
1922 linked with `liblttng-ust`.
1924 An application or library does not have to initialize `liblttng-ust`
1925 manually: its constructor does the necessary tasks to properly register
1926 to a session daemon. The initialization phase also enables the
1927 instrumentation points matching the <<event,event rules>> that you
1931 [[lttng-ust-agents]]
1932 === User space tracing agents
1935 .The user space tracing agents.
1936 image::plumbing-lttng-ust-agents.png[]
1938 The _LTTng-UST Java and Python agents_ are regular Java and Python
1939 packages which add LTTng tracing capabilities to the
1940 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1942 In the case of Java, the
1943 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1944 core logging facilities] and
1945 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1946 Note that Apache Log4{nbsp}2 is not supported.
1948 In the case of Python, the standard
1949 https://docs.python.org/3/library/logging.html[`logging`] package
1950 is supported. Both Python 2 and Python 3 modules can import the
1951 LTTng-UST Python agent package.
1953 The applications using the LTTng-UST agents are in the
1954 `java.util.logging` (JUL),
1955 log4j, and Python <<domain,tracing domains>>.
1957 Both agents use the same mechanism to trace the log statements. When an
1958 agent is initialized, it creates a log handler that attaches to the root
1959 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1960 When the application executes a log statement, it is passed to the
1961 agent's log handler by the root logger. The agent's log handler calls a
1962 native function in a tracepoint provider package shared library linked
1963 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1964 other fields, like its logger name and its log level. This native
1965 function contains a user space instrumentation point, hence tracing the
1968 The log level condition of an
1969 <<event,event rule>> is considered when tracing
1970 a Java or a Python application, and it's compatible with the standard
1971 JUL, log4j, and Python log levels.
1975 === LTTng kernel modules
1978 .The LTTng kernel modules.
1979 image::plumbing-lttng-modules.png[]
1981 The _LTTng kernel modules_ are a set of Linux kernel modules
1982 which implement the kernel tracer of the LTTng project. The LTTng
1983 kernel modules are part of LTTng-modules.
1985 The LTTng kernel modules include:
1987 * A set of _probe_ modules.
1989 Each module attaches to a specific subsystem
1990 of the Linux kernel using its tracepoint instrument points. There are
1991 also modules to attach to the entry and return points of the Linux
1992 system call functions.
1994 * _Ring buffer_ modules.
1996 A ring buffer implementation is provided as kernel modules. The LTTng
1997 kernel tracer writes to the ring buffer; a
1998 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
2000 * The _LTTng kernel tracer_ module.
2001 * The _LTTng logger_ module.
2003 The LTTng logger module implements the special path:{/proc/lttng-logger}
2004 file so that any executable can generate LTTng events by opening and
2005 writing to this file.
2007 See <<proc-lttng-logger-abi,LTTng logger>>.
2009 Generally, you do not have to load the LTTng kernel modules manually
2010 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
2011 daemon>> loads the necessary modules when starting. If you have extra
2012 probe modules, you can specify to load them to the session daemon on
2015 The LTTng kernel modules are installed in
2016 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
2017 the kernel release (see `uname --kernel-release`).
2024 .The session daemon.
2025 image::plumbing-sessiond.png[]
2027 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
2028 managing tracing sessions and for controlling the various components of
2029 LTTng. The session daemon is part of LTTng-tools.
2031 The session daemon sends control requests to and receives control
2034 * The <<lttng-ust,user space tracing library>>.
2036 Any instance of the user space tracing library first registers to
2037 a session daemon. Then, the session daemon can send requests to
2038 this instance, such as:
2041 ** Get the list of tracepoints.
2042 ** Share an <<event,event rule>> so that the user space tracing library
2043 can enable or disable tracepoints. Amongst the possible conditions
2044 of an event rule is a filter expression which `liblttng-ust` evalutes
2045 when an event occurs.
2046 ** Share <<channel,channel>> attributes and ring buffer locations.
2049 The session daemon and the user space tracing library use a Unix
2050 domain socket for their communication.
2052 * The <<lttng-ust-agents,user space tracing agents>>.
2054 Any instance of a user space tracing agent first registers to
2055 a session daemon. Then, the session daemon can send requests to
2056 this instance, such as:
2059 ** Get the list of loggers.
2060 ** Enable or disable a specific logger.
2063 The session daemon and the user space tracing agent use a TCP connection
2064 for their communication.
2066 * The <<lttng-modules,LTTng kernel tracer>>.
2067 * The <<lttng-consumerd,consumer daemon>>.
2069 The session daemon sends requests to the consumer daemon to instruct
2070 it where to send the trace data streams, amongst other information.
2072 * The <<lttng-relayd,relay daemon>>.
2074 The session daemon receives commands from the
2075 <<liblttng-ctl-lttng,tracing control library>>.
2077 The root session daemon loads the appropriate
2078 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
2079 a <<lttng-consumerd,consumer daemon>> as soon as you create
2080 an <<event,event rule>>.
2082 The session daemon does not send and receive trace data: this is the
2083 role of the <<lttng-consumerd,consumer daemon>> and
2084 <<lttng-relayd,relay daemon>>. It does, however, generate the
2085 http://diamon.org/ctf/[CTF] metadata stream.
2087 Each Unix user can have its own session daemon instance. The
2088 tracing sessions managed by different session daemons are completely
2091 The root user's session daemon is the only one which is
2092 allowed to control the LTTng kernel tracer, and its spawned consumer
2093 daemon is the only one which is allowed to consume trace data from the
2094 LTTng kernel tracer. Note, however, that any Unix user which is a member
2095 of the <<tracing-group,tracing group>> is allowed
2096 to create <<channel,channels>> in the
2097 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
2100 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
2101 session daemon when using its `create` command if none is currently
2102 running. You can also start the session daemon manually.
2109 .The consumer daemon.
2110 image::plumbing-consumerd.png[]
2112 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
2113 ring buffers with user applications or with the LTTng kernel modules to
2114 collect trace data and send it to some location (on disk or to a
2115 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
2116 is part of LTTng-tools.
2118 You do not start a consumer daemon manually: a consumer daemon is always
2119 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
2120 <<event,event rule>>, that is, before you start tracing. When you kill
2121 its owner session daemon, the consumer daemon also exits because it is
2122 the session daemon's child process. Command-line options of
2123 man:lttng-sessiond(8) target the consumer daemon process.
2125 There are up to two running consumer daemons per Unix user, whereas only
2126 one session daemon can run per user. This is because each process can be
2127 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
2128 and 64-bit processes, it is more efficient to have separate
2129 corresponding 32-bit and 64-bit consumer daemons. The root user is an
2130 exception: it can have up to _three_ running consumer daemons: 32-bit
2131 and 64-bit instances for its user applications, and one more
2132 reserved for collecting kernel trace data.
2140 image::plumbing-relayd.png[]
2142 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
2143 between remote session and consumer daemons, local trace files, and a
2144 remote live trace viewer. The relay daemon is part of LTTng-tools.
2146 The main purpose of the relay daemon is to implement a receiver of
2147 <<sending-trace-data-over-the-network,trace data over the network>>.
2148 This is useful when the target system does not have much file system
2149 space to record trace files locally.
2151 The relay daemon is also a server to which a
2152 <<lttng-live,live trace viewer>> can
2153 connect. The live trace viewer sends requests to the relay daemon to
2154 receive trace data as the target system emits events. The
2155 communication protocol is named _LTTng live_; it is used over TCP
2158 Note that you can start the relay daemon on the target system directly.
2159 This is the setup of choice when the use case is to view events as
2160 the target system emits them without the need of a remote system.
2164 == [[using-lttng]]Instrumentation
2166 There are many examples of tracing and monitoring in our everyday life:
2168 * You have access to real-time and historical weather reports and
2169 forecasts thanks to weather stations installed around the country.
2170 * You know your heart is safe thanks to an electrocardiogram.
2171 * You make sure not to drive your car too fast and to have enough fuel
2172 to reach your destination thanks to gauges visible on your dashboard.
2174 All the previous examples have something in common: they rely on
2175 **instruments**. Without the electrodes attached to the surface of your
2176 body's skin, cardiac monitoring is futile.
2178 LTTng, as a tracer, is no different from those real life examples. If
2179 you're about to trace a software system or, in other words, record its
2180 history of execution, you better have **instrumentation points** in the
2181 subject you're tracing, that is, the actual software.
2183 Various ways were developed to instrument a piece of software for LTTng
2184 tracing. The most straightforward one is to manually place
2185 instrumentation points, called _tracepoints_, in the software's source
2186 code. It is also possible to add instrumentation points dynamically in
2187 the Linux kernel <<domain,tracing domain>>.
2189 If you're only interested in tracing the Linux kernel, your
2190 instrumentation needs are probably already covered by LTTng's built-in
2191 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2192 user application which is already instrumented for LTTng tracing.
2193 In such cases, you can skip this whole section and read the topics of
2194 the <<controlling-tracing,Tracing control>> section.
2196 Many methods are available to instrument a piece of software for LTTng
2199 * <<c-application,User space instrumentation for C and $$C++$$
2201 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2202 * <<java-application,User space Java agent>>.
2203 * <<python-application,User space Python agent>>.
2204 * <<proc-lttng-logger-abi,LTTng logger>>.
2205 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2209 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2211 The procedure to instrument a C or $$C++$$ user application with
2212 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2214 . <<tracepoint-provider,Create the source files of a tracepoint provider
2216 . <<probing-the-application-source-code,Add tracepoints to
2217 the application's source code>>.
2218 . <<building-tracepoint-providers-and-user-application,Build and link
2219 a tracepoint provider package and the user application>>.
2221 If you need quick, man:printf(3)-like instrumentation, you can skip
2222 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2225 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2226 instrument a user application with `liblttng-ust`.
2229 [[tracepoint-provider]]
2230 ==== Create the source files of a tracepoint provider package
2232 A _tracepoint provider_ is a set of compiled functions which provide
2233 **tracepoints** to an application, the type of instrumentation point
2234 supported by LTTng-UST. Those functions can emit events with
2235 user-defined fields and serialize those events as event records to one
2236 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2237 macro, which you <<probing-the-application-source-code,insert in a user
2238 application's source code>>, calls those functions.
2240 A _tracepoint provider package_ is an object file (`.o`) or a shared
2241 library (`.so`) which contains one or more tracepoint providers.
2242 Its source files are:
2244 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2245 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2247 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2248 the LTTng user space tracer, at run time.
2251 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2252 image::ust-app.png[]
2254 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2255 skip creating and using a tracepoint provider and use
2256 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2260 ===== Create a tracepoint provider header file template
2262 A _tracepoint provider header file_ contains the tracepoint
2263 definitions of a tracepoint provider.
2265 To create a tracepoint provider header file:
2267 . Start from this template:
2271 .Tracepoint provider header file template (`.h` file extension).
2273 #undef TRACEPOINT_PROVIDER
2274 #define TRACEPOINT_PROVIDER provider_name
2276 #undef TRACEPOINT_INCLUDE
2277 #define TRACEPOINT_INCLUDE "./tp.h"
2279 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2282 #include <lttng/tracepoint.h>
2285 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2286 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2291 #include <lttng/tracepoint-event.h>
2297 * `provider_name` with the name of your tracepoint provider.
2298 * `"tp.h"` with the name of your tracepoint provider header file.
2300 . Below the `#include <lttng/tracepoint.h>` line, put your
2301 <<defining-tracepoints,tracepoint definitions>>.
2303 Your tracepoint provider name must be unique amongst all the possible
2304 tracepoint provider names used on the same target system. We
2305 suggest to include the name of your project or company in the name,
2306 for example, `org_lttng_my_project_tpp`.
2308 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2309 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2310 write are the <<defining-tracepoints,tracepoint definitions>>.
2313 [[defining-tracepoints]]
2314 ===== Create a tracepoint definition
2316 A _tracepoint definition_ defines, for a given tracepoint:
2318 * Its **input arguments**. They are the macro parameters that the
2319 `tracepoint()` macro accepts for this particular tracepoint
2320 in the user application's source code.
2321 * Its **output event fields**. They are the sources of event fields
2322 that form the payload of any event that the execution of the
2323 `tracepoint()` macro emits for this particular tracepoint.
2325 You can create a tracepoint definition by using the
2326 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2328 <<tpp-header,tracepoint provider header file template>>.
2330 The syntax of the `TRACEPOINT_EVENT()` macro is:
2333 .`TRACEPOINT_EVENT()` macro syntax.
2336 /* Tracepoint provider name */
2339 /* Tracepoint name */
2342 /* Input arguments */
2347 /* Output event fields */
2356 * `provider_name` with your tracepoint provider name.
2357 * `tracepoint_name` with your tracepoint name.
2358 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2359 * `fields` with the <<tpp-def-output-fields,output event field>>
2362 This tracepoint emits events named `provider_name:tracepoint_name`.
2365 .Event name's length limitation
2367 The concatenation of the tracepoint provider name and the
2368 tracepoint name must not exceed **254 characters**. If it does, the
2369 instrumented application compiles and runs, but LTTng throws multiple
2370 warnings and you could experience serious issues.
2373 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2376 .`TP_ARGS()` macro syntax.
2385 * `type` with the C type of the argument.
2386 * `arg_name` with the argument name.
2388 You can repeat `type` and `arg_name` up to 10 times to have
2389 more than one argument.
2391 .`TP_ARGS()` usage with three arguments.
2403 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2404 tracepoint definition with no input arguments.
2406 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2407 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2408 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2409 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2412 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2413 C expression that the tracer evalutes at the `tracepoint()` macro site
2414 in the application's source code. This expression provides a field's
2415 source of data. The argument expression can include input argument names
2416 listed in the `TP_ARGS()` macro.
2418 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2419 must be unique within a given tracepoint definition.
2421 Here's a complete tracepoint definition example:
2423 .Tracepoint definition.
2425 The following tracepoint definition defines a tracepoint which takes
2426 three input arguments and has four output event fields.
2430 #include "my-custom-structure.h"
2436 const struct my_custom_structure*, my_custom_structure,
2441 ctf_string(query_field, query)
2442 ctf_float(double, ratio_field, ratio)
2443 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2444 ctf_integer(int, send_size, my_custom_structure->send_size)
2449 You can refer to this tracepoint definition with the `tracepoint()`
2450 macro in your application's source code like this:
2454 tracepoint(my_provider, my_tracepoint,
2455 my_structure, some_ratio, the_query);
2459 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2460 if they satisfy an enabled <<event,event rule>>.
2463 [[using-tracepoint-classes]]
2464 ===== Use a tracepoint class
2466 A _tracepoint class_ is a class of tracepoints which share the same
2467 output event field definitions. A _tracepoint instance_ is one
2468 instance of such a defined tracepoint class, with its own tracepoint
2471 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2472 shorthand which defines both a tracepoint class and a tracepoint
2473 instance at the same time.
2475 When you build a tracepoint provider package, the C or $$C++$$ compiler
2476 creates one serialization function for each **tracepoint class**. A
2477 serialization function is responsible for serializing the event fields
2478 of a tracepoint to a sub-buffer when tracing.
2480 For various performance reasons, when your situation requires multiple
2481 tracepoint definitions with different names, but with the same event
2482 fields, we recommend that you manually create a tracepoint class
2483 and instantiate as many tracepoint instances as needed. One positive
2484 effect of such a design, amongst other advantages, is that all
2485 tracepoint instances of the same tracepoint class reuse the same
2486 serialization function, thus reducing
2487 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2489 .Use a tracepoint class and tracepoint instances.
2491 Consider the following three tracepoint definitions:
2503 ctf_integer(int, userid, userid)
2504 ctf_integer(size_t, len, len)
2516 ctf_integer(int, userid, userid)
2517 ctf_integer(size_t, len, len)
2529 ctf_integer(int, userid, userid)
2530 ctf_integer(size_t, len, len)
2535 In this case, we create three tracepoint classes, with one implicit
2536 tracepoint instance for each of them: `get_account`, `get_settings`, and
2537 `get_transaction`. However, they all share the same event field names
2538 and types. Hence three identical, yet independent serialization
2539 functions are created when you build the tracepoint provider package.
2541 A better design choice is to define a single tracepoint class and three
2542 tracepoint instances:
2546 /* The tracepoint class */
2547 TRACEPOINT_EVENT_CLASS(
2548 /* Tracepoint provider name */
2551 /* Tracepoint class name */
2554 /* Input arguments */
2560 /* Output event fields */
2562 ctf_integer(int, userid, userid)
2563 ctf_integer(size_t, len, len)
2567 /* The tracepoint instances */
2568 TRACEPOINT_EVENT_INSTANCE(
2569 /* Tracepoint provider name */
2572 /* Tracepoint class name */
2575 /* Tracepoint name */
2578 /* Input arguments */
2584 TRACEPOINT_EVENT_INSTANCE(
2593 TRACEPOINT_EVENT_INSTANCE(
2606 [[assigning-log-levels]]
2607 ===== Assign a log level to a tracepoint definition
2609 You can assign an optional _log level_ to a
2610 <<defining-tracepoints,tracepoint definition>>.
2612 Assigning different levels of severity to tracepoint definitions can
2613 be useful: when you <<enabling-disabling-events,create an event rule>>,
2614 you can target tracepoints having a log level as severe as a specific
2617 The concept of LTTng-UST log levels is similar to the levels found
2618 in typical logging frameworks:
2620 * In a logging framework, the log level is given by the function
2621 or method name you use at the log statement site: `debug()`,
2622 `info()`, `warn()`, `error()`, and so on.
2623 * In LTTng-UST, you statically assign the log level to a tracepoint
2624 definition; any `tracepoint()` macro invocation which refers to
2625 this definition has this log level.
2627 You can assign a log level to a tracepoint definition with the
2628 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2629 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2630 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2633 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2636 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2638 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2643 * `provider_name` with the tracepoint provider name.
2644 * `tracepoint_name` with the tracepoint name.
2645 * `log_level` with the log level to assign to the tracepoint
2646 definition named `tracepoint_name` in the `provider_name`
2647 tracepoint provider.
2649 See man:lttng-ust(3) for a list of available log level names.
2651 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2655 /* Tracepoint definition */
2664 ctf_integer(int, userid, userid)
2665 ctf_integer(size_t, len, len)
2669 /* Log level assignment */
2670 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2676 ===== Create a tracepoint provider package source file
2678 A _tracepoint provider package source file_ is a C source file which
2679 includes a <<tpp-header,tracepoint provider header file>> to expand its
2680 macros into event serialization and other functions.
2682 You can always use the following tracepoint provider package source
2686 .Tracepoint provider package source file template.
2688 #define TRACEPOINT_CREATE_PROBES
2693 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2694 header file>> name. You may also include more than one tracepoint
2695 provider header file here to create a tracepoint provider package
2696 holding more than one tracepoint providers.
2699 [[probing-the-application-source-code]]
2700 ==== Add tracepoints to an application's source code
2702 Once you <<tpp-header,create a tracepoint provider header file>>, you
2703 can use the `tracepoint()` macro in your application's
2704 source code to insert the tracepoints that this header
2705 <<defining-tracepoints,defines>>.
2707 The `tracepoint()` macro takes at least two parameters: the tracepoint
2708 provider name and the tracepoint name. The corresponding tracepoint
2709 definition defines the other parameters.
2711 .`tracepoint()` usage.
2713 The following <<defining-tracepoints,tracepoint definition>> defines a
2714 tracepoint which takes two input arguments and has two output event
2718 .Tracepoint provider header file.
2720 #include "my-custom-structure.h"
2727 const char*, cmd_name
2730 ctf_string(cmd_name, cmd_name)
2731 ctf_integer(int, number_of_args, argc)
2736 You can refer to this tracepoint definition with the `tracepoint()`
2737 macro in your application's source code like this:
2740 .Application's source file.
2744 int main(int argc, char* argv[])
2746 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2752 Note how the application's source code includes
2753 the tracepoint provider header file containing the tracepoint
2754 definitions to use, path:{tp.h}.
2757 .`tracepoint()` usage with a complex tracepoint definition.
2759 Consider this complex tracepoint definition, where multiple event
2760 fields refer to the same input arguments in their argument expression
2764 .Tracepoint provider header file.
2766 /* For `struct stat` */
2767 #include <sys/types.h>
2768 #include <sys/stat.h>
2780 ctf_integer(int, my_constant_field, 23 + 17)
2781 ctf_integer(int, my_int_arg_field, my_int_arg)
2782 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2783 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2784 my_str_arg[2] + my_str_arg[3])
2785 ctf_string(my_str_arg_field, my_str_arg)
2786 ctf_integer_hex(off_t, size_field, st->st_size)
2787 ctf_float(double, size_dbl_field, (double) st->st_size)
2788 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2789 size_t, strlen(my_str_arg) / 2)
2794 You can refer to this tracepoint definition with the `tracepoint()`
2795 macro in your application's source code like this:
2798 .Application's source file.
2800 #define TRACEPOINT_DEFINE
2807 stat("/etc/fstab", &s);
2808 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2814 If you look at the event record that LTTng writes when tracing this
2815 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2816 it should look like this:
2818 .Event record fields
2820 |Field's name |Field's value
2821 |`my_constant_field` |40
2822 |`my_int_arg_field` |23
2823 |`my_int_arg_field2` |529
2825 |`my_str_arg_field` |`Hello, World!`
2826 |`size_field` |0x12d
2827 |`size_dbl_field` |301.0
2828 |`half_my_str_arg_field` |`Hello,`
2832 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2833 compute--they use the call stack, for example. To avoid this
2834 computation when the tracepoint is disabled, you can use the
2835 `tracepoint_enabled()` and `do_tracepoint()` macros.
2837 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2841 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2843 tracepoint_enabled(provider_name, tracepoint_name)
2844 do_tracepoint(provider_name, tracepoint_name, ...)
2849 * `provider_name` with the tracepoint provider name.
2850 * `tracepoint_name` with the tracepoint name.
2852 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2853 `tracepoint_name` from the provider named `provider_name` is enabled
2856 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2857 if the tracepoint is enabled. Using `tracepoint()` with
2858 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2859 the `tracepoint_enabled()` check, thus a race condition is
2860 possible in this situation:
2863 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2865 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2866 stuff = prepare_stuff();
2869 tracepoint(my_provider, my_tracepoint, stuff);
2872 If the tracepoint is enabled after the condition, then `stuff` is not
2873 prepared: the emitted event will either contain wrong data, or the whole
2874 application could crash (segmentation fault, for example).
2876 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2877 `STAP_PROBEV()` call. If you need it, you must emit
2881 [[building-tracepoint-providers-and-user-application]]
2882 ==== Build and link a tracepoint provider package and an application
2884 Once you have one or more <<tpp-header,tracepoint provider header
2885 files>> and a <<tpp-source,tracepoint provider package source file>>,
2886 you can create the tracepoint provider package by compiling its source
2887 file. From here, multiple build and run scenarios are possible. The
2888 following table shows common application and library configurations
2889 along with the required command lines to achieve them.
2891 In the following diagrams, we use the following file names:
2894 Executable application.
2897 Application's object file.
2900 Tracepoint provider package object file.
2903 Tracepoint provider package archive file.
2906 Tracepoint provider package shared object file.
2909 User library object file.
2912 User library shared object file.
2914 We use the following symbols in the diagrams of table below:
2917 .Symbols used in the build scenario diagrams.
2918 image::ust-sit-symbols.png[]
2920 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2921 variable in the following instructions.
2923 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2924 .Common tracepoint provider package scenarios.
2926 |Scenario |Instructions
2929 The instrumented application is statically linked with
2930 the tracepoint provider package object.
2932 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2935 include::../common/ust-sit-step-tp-o.txt[]
2937 To build the instrumented application:
2939 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2944 #define TRACEPOINT_DEFINE
2948 . Compile the application source file:
2957 . Build the application:
2962 gcc -o app app.o tpp.o -llttng-ust -ldl
2966 To run the instrumented application:
2968 * Start the application:
2978 The instrumented application is statically linked with the
2979 tracepoint provider package archive file.
2981 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2984 To create the tracepoint provider package archive file:
2986 . Compile the <<tpp-source,tracepoint provider package source file>>:
2995 . Create the tracepoint provider package archive file:
3004 To build the instrumented application:
3006 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3011 #define TRACEPOINT_DEFINE
3015 . Compile the application source file:
3024 . Build the application:
3029 gcc -o app app.o tpp.a -llttng-ust -ldl
3033 To run the instrumented application:
3035 * Start the application:
3045 The instrumented application is linked with the tracepoint provider
3046 package shared object.
3048 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
3051 include::../common/ust-sit-step-tp-so.txt[]
3053 To build the instrumented application:
3055 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3060 #define TRACEPOINT_DEFINE
3064 . Compile the application source file:
3073 . Build the application:
3078 gcc -o app app.o -ldl -L. -ltpp
3082 To run the instrumented application:
3084 * Start the application:
3094 The tracepoint provider package shared object is preloaded before the
3095 instrumented application starts.
3097 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
3100 include::../common/ust-sit-step-tp-so.txt[]
3102 To build the instrumented application:
3104 . In path:{app.c}, before including path:{tpp.h}, add the
3110 #define TRACEPOINT_DEFINE
3111 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3115 . Compile the application source file:
3124 . Build the application:
3129 gcc -o app app.o -ldl
3133 To run the instrumented application with tracing support:
3135 * Preload the tracepoint provider package shared object and
3136 start the application:
3141 LD_PRELOAD=./libtpp.so ./app
3145 To run the instrumented application without tracing support:
3147 * Start the application:
3157 The instrumented application dynamically loads the tracepoint provider
3158 package shared object.
3160 See the <<dlclose-warning,warning about `dlclose()`>>.
3162 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3165 include::../common/ust-sit-step-tp-so.txt[]
3167 To build the instrumented application:
3169 . In path:{app.c}, before including path:{tpp.h}, add the
3175 #define TRACEPOINT_DEFINE
3176 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3180 . Compile the application source file:
3189 . Build the application:
3194 gcc -o app app.o -ldl
3198 To run the instrumented application:
3200 * Start the application:
3210 The application is linked with the instrumented user library.
3212 The instrumented user library is statically linked with the tracepoint
3213 provider package object file.
3215 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3218 include::../common/ust-sit-step-tp-o-fpic.txt[]
3220 To build the instrumented user library:
3222 . In path:{emon.c}, before including path:{tpp.h}, add the
3228 #define TRACEPOINT_DEFINE
3232 . Compile the user library source file:
3237 gcc -I. -fpic -c emon.c
3241 . Build the user library shared object:
3246 gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3250 To build the application:
3252 . Compile the application source file:
3261 . Build the application:
3266 gcc -o app app.o -L. -lemon
3270 To run the application:
3272 * Start the application:
3282 The application is linked with the instrumented user library.
3284 The instrumented user library is linked with the tracepoint provider
3285 package shared object.
3287 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3290 include::../common/ust-sit-step-tp-so.txt[]
3292 To build the instrumented user library:
3294 . In path:{emon.c}, before including path:{tpp.h}, add the
3300 #define TRACEPOINT_DEFINE
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 -L. -ltpp
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 tracepoint provider package shared object is preloaded before the
3357 The application is linked with the instrumented user library.
3359 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3362 include::../common/ust-sit-step-tp-so.txt[]
3364 To build the instrumented user library:
3366 . In path:{emon.c}, before including path:{tpp.h}, add the
3372 #define TRACEPOINT_DEFINE
3373 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3377 . Compile the user library source file:
3382 gcc -I. -fpic -c emon.c
3386 . Build the user library shared object:
3391 gcc -shared -o libemon.so emon.o -ldl
3395 To build the application:
3397 . Compile the application source file:
3406 . Build the application:
3411 gcc -o app app.o -L. -lemon
3415 To run the application with tracing support:
3417 * Preload the tracepoint provider package shared object and
3418 start the application:
3423 LD_PRELOAD=./libtpp.so ./app
3427 To run the application without tracing support:
3429 * Start the application:
3439 The application is linked with the instrumented user library.
3441 The instrumented user library dynamically loads the tracepoint provider
3442 package shared object.
3444 See the <<dlclose-warning,warning about `dlclose()`>>.
3446 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3449 include::../common/ust-sit-step-tp-so.txt[]
3451 To build the instrumented user library:
3453 . In path:{emon.c}, before including path:{tpp.h}, add the
3459 #define TRACEPOINT_DEFINE
3460 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3464 . Compile the user library source file:
3469 gcc -I. -fpic -c emon.c
3473 . Build the user library shared object:
3478 gcc -shared -o libemon.so emon.o -ldl
3482 To build the application:
3484 . Compile the application source file:
3493 . Build the application:
3498 gcc -o app app.o -L. -lemon
3502 To run the application:
3504 * Start the application:
3514 The application dynamically loads the instrumented user library.
3516 The instrumented user library is linked with the tracepoint provider
3517 package shared object.
3519 See the <<dlclose-warning,warning about `dlclose()`>>.
3521 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3524 include::../common/ust-sit-step-tp-so.txt[]
3526 To build the instrumented user library:
3528 . In path:{emon.c}, before including path:{tpp.h}, add the
3534 #define TRACEPOINT_DEFINE
3538 . Compile the user library source file:
3543 gcc -I. -fpic -c emon.c
3547 . Build the user library shared object:
3552 gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3556 To build the application:
3558 . Compile the application source file:
3567 . Build the application:
3572 gcc -o app app.o -ldl -L. -lemon
3576 To run the application:
3578 * Start the application:
3588 The application dynamically loads the instrumented user library.
3590 The instrumented user library dynamically loads the tracepoint provider
3591 package shared object.
3593 See the <<dlclose-warning,warning about `dlclose()`>>.
3595 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3598 include::../common/ust-sit-step-tp-so.txt[]
3600 To build the instrumented user library:
3602 . In path:{emon.c}, before including path:{tpp.h}, add the
3608 #define TRACEPOINT_DEFINE
3609 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3613 . Compile the user library source file:
3618 gcc -I. -fpic -c emon.c
3622 . Build the user library shared object:
3627 gcc -shared -o libemon.so emon.o -ldl
3631 To build the application:
3633 . Compile the application source file:
3642 . Build the application:
3647 gcc -o app app.o -ldl -L. -lemon
3651 To run the application:
3653 * Start the application:
3663 The tracepoint provider package shared object is preloaded before the
3666 The application dynamically loads the instrumented user library.
3668 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3671 include::../common/ust-sit-step-tp-so.txt[]
3673 To build the instrumented user library:
3675 . In path:{emon.c}, before including path:{tpp.h}, add the
3681 #define TRACEPOINT_DEFINE
3682 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3686 . Compile the user library source file:
3691 gcc -I. -fpic -c emon.c
3695 . Build the user library shared object:
3700 gcc -shared -o libemon.so emon.o -ldl
3704 To build the application:
3706 . Compile the application source file:
3715 . Build the application:
3720 gcc -o app app.o -L. -lemon
3724 To run the application with tracing support:
3726 * Preload the tracepoint provider package shared object and
3727 start the application:
3732 LD_PRELOAD=./libtpp.so ./app
3736 To run the application without tracing support:
3738 * Start the application:
3748 The application is statically linked with the tracepoint provider
3749 package object file.
3751 The application is linked with the instrumented user library.
3753 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3756 include::../common/ust-sit-step-tp-o.txt[]
3758 To build the instrumented user library:
3760 . In path:{emon.c}, before including path:{tpp.h}, add the
3766 #define TRACEPOINT_DEFINE
3770 . Compile the user library source file:
3775 gcc -I. -fpic -c emon.c
3779 . Build the user library shared object:
3784 gcc -shared -o libemon.so emon.o
3788 To build the application:
3790 . Compile the application source file:
3799 . Build the application:
3804 gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3808 To run the instrumented application:
3810 * Start the application:
3820 The application is statically linked with the tracepoint provider
3821 package object file.
3823 The application dynamically loads the instrumented user library.
3825 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3828 include::../common/ust-sit-step-tp-o.txt[]
3830 To build the application:
3832 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3837 #define TRACEPOINT_DEFINE
3841 . Compile the application source file:
3850 . Build the application:
3855 gcc -Wl,--export-dynamic -o app app.o tpp.o \
3860 The `--export-dynamic` option passed to the linker is necessary for the
3861 dynamically loaded library to ``see'' the tracepoint symbols defined in
3864 To build the instrumented user library:
3866 . Compile the user library source file:
3871 gcc -I. -fpic -c emon.c
3875 . Build the user library shared object:
3880 gcc -shared -o libemon.so emon.o
3884 To run the application:
3886 * Start the application:
3898 .Do not use man:dlclose(3) on a tracepoint provider package
3900 Never use man:dlclose(3) on any shared object which:
3902 * Is linked with, statically or dynamically, a tracepoint provider
3904 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3905 package shared object.
3907 This is currently considered **unsafe** due to a lack of reference
3908 counting from LTTng-UST to the shared object.
3910 A known workaround (available since glibc 2.2) is to use the
3911 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3912 effect of not unloading the loaded shared object, even if man:dlclose(3)
3915 You can also preload the tracepoint provider package shared object with
3916 the env:LD_PRELOAD environment variable to overcome this limitation.
3920 [[using-lttng-ust-with-daemons]]
3921 ===== Use noch:{LTTng-UST} with daemons
3923 If your instrumented application calls man:fork(2), man:clone(2),
3924 or BSD's man:rfork(2), without a following man:exec(3)-family
3925 system call, you must preload the path:{liblttng-ust-fork.so} shared
3926 object when starting the application.
3930 LD_PRELOAD=liblttng-ust-fork.so ./my-app
3933 If your tracepoint provider package is
3934 a shared library which you also preload, you must put both
3935 shared objects in env:LD_PRELOAD:
3939 LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3943 [[lttng-ust-pkg-config]]
3944 ===== Use noch:{pkg-config}
3946 On some distributions, LTTng-UST ships with a
3947 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3948 metadata file. If this is your case, then you can use cmd:pkg-config to
3949 build an application on the command line:
3953 gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3957 [[instrumenting-32-bit-app-on-64-bit-system]]
3958 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3960 In order to trace a 32-bit application running on a 64-bit system,
3961 LTTng must use a dedicated 32-bit
3962 <<lttng-consumerd,consumer daemon>>.
3964 The following steps show how to build and install a 32-bit consumer
3965 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3966 build and install the 32-bit LTTng-UST libraries, and how to build and
3967 link an instrumented 32-bit application in that context.
3969 To build a 32-bit instrumented application for a 64-bit target system,
3970 assuming you have a fresh target system with no installed Userspace RCU
3973 . Download, build, and install a 32-bit version of Userspace RCU:
3979 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3980 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3981 cd userspace-rcu-0.9.* &&
3982 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3984 sudo make install &&
3989 . Using your distribution's package manager, or from source, install
3990 the following 32-bit versions of the following dependencies of
3991 LTTng-tools and LTTng-UST:
3994 * https://sourceforge.net/projects/libuuid/[libuuid]
3995 * http://directory.fsf.org/wiki/Popt[popt]
3996 * http://www.xmlsoft.org/[libxml2]
3999 . Download, build, and install a 32-bit version of the latest
4000 LTTng-UST{nbsp}{revision}:
4006 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.8.tar.bz2 &&
4007 tar -xf lttng-ust-latest-2.8.tar.bz2 &&
4008 cd lttng-ust-2.8.* &&
4009 ./configure --libdir=/usr/local/lib32 \
4010 CFLAGS=-m32 CXXFLAGS=-m32 \
4011 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
4013 sudo make install &&
4020 Depending on your distribution,
4021 32-bit libraries could be installed at a different location than
4022 `/usr/lib32`. For example, Debian is known to install
4023 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
4025 In this case, make sure to set `LDFLAGS` to all the
4026 relevant 32-bit library paths, for example:
4030 LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
4034 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
4035 the 32-bit consumer daemon:
4041 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
4042 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
4043 cd lttng-tools-2.8.* &&
4044 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
4045 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
4047 cd src/bin/lttng-consumerd &&
4048 sudo make install &&
4053 . From your distribution or from source,
4054 <<installing-lttng,install>> the 64-bit versions of
4055 LTTng-UST and Userspace RCU.
4056 . Download, build, and install the 64-bit version of the
4057 latest LTTng-tools{nbsp}{revision}:
4063 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
4064 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
4065 cd lttng-tools-2.8.* &&
4066 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
4067 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
4069 sudo make install &&
4074 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
4075 when linking your 32-bit application:
4078 -m32 -L/usr/lib32 -L/usr/local/lib32 \
4079 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
4082 For example, let's rebuild the quick start example in
4083 <<tracing-your-own-user-application,Trace a user application>> as an
4084 instrumented 32-bit application:
4089 gcc -m32 -c -I. hello-tp.c
4091 gcc -m32 -o hello hello.o hello-tp.o \
4092 -L/usr/lib32 -L/usr/local/lib32 \
4093 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
4098 No special action is required to execute the 32-bit application and
4099 to trace it: use the command-line man:lttng(1) tool as usual.
4106 man:tracef(3) is a small LTTng-UST API designed for quick,
4107 man:printf(3)-like instrumentation without the burden of
4108 <<tracepoint-provider,creating>> and
4109 <<building-tracepoint-providers-and-user-application,building>>
4110 a tracepoint provider package.
4112 To use `tracef()` in your application:
4114 . In the C or C++ source files where you need to use `tracef()`,
4115 include `<lttng/tracef.h>`:
4120 #include <lttng/tracef.h>
4124 . In the application's source code, use `tracef()` like you would use
4132 tracef("my message: %d (%s)", my_integer, my_string);
4138 . Link your application with `liblttng-ust`:
4143 gcc -o app app.c -llttng-ust
4147 To trace the events that `tracef()` calls emit:
4149 * <<enabling-disabling-events,Create an event rule>> which matches the
4150 `lttng_ust_tracef:*` event name:
4155 lttng enable-event --userspace 'lttng_ust_tracef:*'
4160 .Limitations of `tracef()`
4162 The `tracef()` utility function was developed to make user space tracing
4163 super simple, albeit with notable disadvantages compared to
4164 <<defining-tracepoints,user-defined tracepoints>>:
4166 * All the emitted events have the same tracepoint provider and
4167 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4168 * There is no static type checking.
4169 * The only event record field you actually get, named `msg`, is a string
4170 potentially containing the values you passed to `tracef()`
4171 using your own format string. This also means that you cannot filter
4172 events with a custom expression at run time because there are no
4174 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4175 function behind the scenes to format the strings at run time, its
4176 expected performance is lower than with user-defined tracepoints,
4177 which do not require a conversion to a string.
4179 Taking this into consideration, `tracef()` is useful for some quick
4180 prototyping and debugging, but you should not consider it for any
4181 permanent and serious applicative instrumentation.
4187 ==== Use `tracelog()`
4189 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4190 the difference that it accepts an additional log level parameter.
4192 The goal of `tracelog()` is to ease the migration from logging to
4195 To use `tracelog()` in your application:
4197 . In the C or C++ source files where you need to use `tracelog()`,
4198 include `<lttng/tracelog.h>`:
4203 #include <lttng/tracelog.h>
4207 . In the application's source code, use `tracelog()` like you would use
4208 man:printf(3), except for the first parameter which is the log
4216 tracelog(TRACE_WARNING, "my message: %d (%s)",
4217 my_integer, my_string);
4223 See man:lttng-ust(3) for a list of available log level names.
4225 . Link your application with `liblttng-ust`:
4230 gcc -o app app.c -llttng-ust
4234 To trace the events that `tracelog()` calls emit with a log level
4235 _as severe as_ a specific log level:
4237 * <<enabling-disabling-events,Create an event rule>> which matches the
4238 `lttng_ust_tracelog:*` event name and a minimum level
4244 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4245 --loglevel=TRACE_WARNING
4249 To trace the events that `tracelog()` calls emit with a
4250 _specific log level_:
4252 * Create an event rule which matches the `lttng_ust_tracelog:*`
4253 event name and a specific log level:
4258 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4259 --loglevel-only=TRACE_INFO
4264 [[prebuilt-ust-helpers]]
4265 === Prebuilt user space tracing helpers
4267 The LTTng-UST package provides a few helpers in the form or preloadable
4268 shared objects which automatically instrument system functions and
4271 The helper shared objects are normally found in dir:{/usr/lib}. If you
4272 built LTTng-UST <<building-from-source,from source>>, they are probably
4273 located in dir:{/usr/local/lib}.
4275 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4278 path:{liblttng-ust-libc-wrapper.so}::
4279 path:{liblttng-ust-pthread-wrapper.so}::
4280 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4281 memory and POSIX threads function tracing>>.
4283 path:{liblttng-ust-cyg-profile.so}::
4284 path:{liblttng-ust-cyg-profile-fast.so}::
4285 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4287 path:{liblttng-ust-dl.so}::
4288 <<liblttng-ust-dl,Dynamic linker tracing>>.
4290 To use a user space tracing helper with any user application:
4292 * Preload the helper shared object when you start the application:
4297 LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4301 You can preload more than one helper:
4306 LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4312 [[liblttng-ust-libc-pthread-wrapper]]
4313 ==== Instrument C standard library memory and POSIX threads functions
4315 The path:{liblttng-ust-libc-wrapper.so} and
4316 path:{liblttng-ust-pthread-wrapper.so} helpers
4317 add instrumentation to some C standard library and POSIX
4321 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4323 |TP provider name |TP name |Instrumented function
4325 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4326 |`calloc` |man:calloc(3)
4327 |`realloc` |man:realloc(3)
4328 |`free` |man:free(3)
4329 |`memalign` |man:memalign(3)
4330 |`posix_memalign` |man:posix_memalign(3)
4334 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4336 |TP provider name |TP name |Instrumented function
4338 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4339 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4340 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4341 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4344 When you preload the shared object, it replaces the functions listed
4345 in the previous tables by wrappers which contain tracepoints and call
4346 the replaced functions.
4349 [[liblttng-ust-cyg-profile]]
4350 ==== Instrument function entry and exit
4352 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4353 to the entry and exit points of functions.
4355 man:gcc(1) and man:clang(1) have an option named
4356 https://gcc.gnu.org/onlinedocs/gcc/Code-Gen-Options.html[`-finstrument-functions`]
4357 which generates instrumentation calls for entry and exit to functions.
4358 The LTTng-UST function tracing helpers,
4359 path:{liblttng-ust-cyg-profile.so} and
4360 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4361 to add tracepoints to the two generated functions (which contain
4362 `cyg_profile` in their names, hence the helper's name).
4364 To use the LTTng-UST function tracing helper, the source files to
4365 instrument must be built using the `-finstrument-functions` compiler
4368 There are two versions of the LTTng-UST function tracing helper:
4370 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4371 that you should only use when it can be _guaranteed_ that the
4372 complete event stream is recorded without any lost event record.
4373 Any kind of duplicate information is left out.
4375 Assuming no event record is lost, having only the function addresses on
4376 entry is enough to create a call graph, since an event record always
4377 contains the ID of the CPU that generated it.
4379 You can use a tool like man:addr2line(1) to convert function addresses
4380 back to source file names and line numbers.
4382 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4383 which also works in use cases where event records might get discarded or
4384 not recorded from application startup.
4385 In these cases, the trace analyzer needs more information to be
4386 able to reconstruct the program flow.
4388 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4389 points of this helper.
4391 All the tracepoints that this helper provides have the
4392 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4394 TIP: It's sometimes a good idea to limit the number of source files that
4395 you compile with the `-finstrument-functions` option to prevent LTTng
4396 from writing an excessive amount of trace data at run time. When using
4397 man:gcc(1), you can use the
4398 `-finstrument-functions-exclude-function-list` option to avoid
4399 instrument entries and exits of specific function names.
4404 ==== Instrument the dynamic linker
4406 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4407 man:dlopen(3) and man:dlclose(3) function calls.
4409 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4414 [[java-application]]
4415 === User space Java agent
4417 You can instrument any Java application which uses one of the following
4420 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4421 (JUL) core logging facilities.
4422 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4423 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4426 .LTTng-UST Java agent imported by a Java application.
4427 image::java-app.png[]
4429 Note that the methods described below are new in LTTng{nbsp}{revision}.
4430 Previous LTTng versions use another technique.
4432 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4433 and https://ci.lttng.org/[continuous integration], thus this version is
4434 directly supported. However, the LTTng-UST Java agent is also tested
4435 with OpenJDK{nbsp}7.
4440 ==== Use the LTTng-UST Java agent for `java.util.logging`
4442 To use the LTTng-UST Java agent in a Java application which uses
4443 `java.util.logging` (JUL):
4445 . In the Java application's source code, import the LTTng-UST
4446 log handler package for `java.util.logging`:
4451 import org.lttng.ust.agent.jul.LttngLogHandler;
4455 . Create an LTTng-UST JUL log handler:
4460 Handler lttngUstLogHandler = new LttngLogHandler();
4464 . Add this handler to the JUL loggers which should emit LTTng events:
4469 Logger myLogger = Logger.getLogger("some-logger");
4471 myLogger.addHandler(lttngUstLogHandler);
4475 . Use `java.util.logging` log statements and configuration as usual.
4476 The loggers with an attached LTTng-UST log handler can emit
4479 . Before exiting the application, remove the LTTng-UST log handler from
4480 the loggers attached to it and call its `close()` method:
4485 myLogger.removeHandler(lttngUstLogHandler);
4486 lttngUstLogHandler.close();
4490 This is not strictly necessary, but it is recommended for a clean
4491 disposal of the handler's resources.
4493 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4494 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4496 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4497 path] when you build the Java application.
4499 The JAR files are typically located in dir:{/usr/share/java}.
4501 IMPORTANT: The LTTng-UST Java agent must be
4502 <<installing-lttng,installed>> for the logging framework your
4505 .Use the LTTng-UST Java agent for `java.util.logging`.
4510 import java.io.IOException;
4511 import java.util.logging.Handler;
4512 import java.util.logging.Logger;
4513 import org.lttng.ust.agent.jul.LttngLogHandler;
4517 private static final int answer = 42;
4519 public static void main(String[] argv) throws Exception
4522 Logger logger = Logger.getLogger("jello");
4524 // Create an LTTng-UST log handler
4525 Handler lttngUstLogHandler = new LttngLogHandler();
4527 // Add the LTTng-UST log handler to our logger
4528 logger.addHandler(lttngUstLogHandler);
4531 logger.info("some info");
4532 logger.warning("some warning");
4534 logger.finer("finer information; the answer is " + answer);
4536 logger.severe("error!");
4538 // Not mandatory, but cleaner
4539 logger.removeHandler(lttngUstLogHandler);
4540 lttngUstLogHandler.close();
4549 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4552 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4553 <<enabling-disabling-events,create an event rule>> matching the
4554 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4559 lttng enable-event --jul jello
4563 Run the compiled class:
4567 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4570 <<basic-tracing-session-control,Stop tracing>> and inspect the
4580 You can use the opt:lttng-enable-event(1):--loglevel or
4581 opt:lttng-enable-event(1):--loglevel-only option of the
4582 man:lttng-enable-event(1) command to target a range of JUL log levels
4583 or a specific JUL log level.
4588 ==== Use the LTTng-UST Java agent for Apache log4j
4590 To use the LTTng-UST Java agent in a Java application which uses
4593 . In the Java application's source code, import the LTTng-UST
4594 log appender package for Apache log4j:
4599 import org.lttng.ust.agent.log4j.LttngLogAppender;
4603 . Create an LTTng-UST log4j log appender:
4608 Appender lttngUstLogAppender = new LttngLogAppender();
4612 . Add this appender to the log4j loggers which should emit LTTng events:
4617 Logger myLogger = Logger.getLogger("some-logger");
4619 myLogger.addAppender(lttngUstLogAppender);
4623 . Use Apache log4j log statements and configuration as usual. The
4624 loggers with an attached LTTng-UST log appender can emit LTTng events.
4626 . Before exiting the application, remove the LTTng-UST log appender from
4627 the loggers attached to it and call its `close()` method:
4632 myLogger.removeAppender(lttngUstLogAppender);
4633 lttngUstLogAppender.close();
4637 This is not strictly necessary, but it is recommended for a clean
4638 disposal of the appender's resources.
4640 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4641 files, path:{lttng-ust-agent-common.jar} and
4642 path:{lttng-ust-agent-log4j.jar}, in the
4643 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4644 path] when you build the Java application.
4646 The JAR files are typically located in dir:{/usr/share/java}.
4648 IMPORTANT: The LTTng-UST Java agent must be
4649 <<installing-lttng,installed>> for the logging framework your
4652 .Use the LTTng-UST Java agent for Apache log4j.
4657 import org.apache.log4j.Appender;
4658 import org.apache.log4j.Logger;
4659 import org.lttng.ust.agent.log4j.LttngLogAppender;
4663 private static final int answer = 42;
4665 public static void main(String[] argv) throws Exception
4668 Logger logger = Logger.getLogger("jello");
4670 // Create an LTTng-UST log appender
4671 Appender lttngUstLogAppender = new LttngLogAppender();
4673 // Add the LTTng-UST log appender to our logger
4674 logger.addAppender(lttngUstLogAppender);
4677 logger.info("some info");
4678 logger.warn("some warning");
4680 logger.debug("debug information; the answer is " + answer);
4682 logger.fatal("error!");
4684 // Not mandatory, but cleaner
4685 logger.removeAppender(lttngUstLogAppender);
4686 lttngUstLogAppender.close();
4692 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4697 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4700 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4701 <<enabling-disabling-events,create an event rule>> matching the
4702 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4707 lttng enable-event --log4j jello
4711 Run the compiled class:
4715 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4718 <<basic-tracing-session-control,Stop tracing>> and inspect the
4728 You can use the opt:lttng-enable-event(1):--loglevel or
4729 opt:lttng-enable-event(1):--loglevel-only option of the
4730 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4731 or a specific log4j log level.
4735 [[java-application-context]]
4736 ==== Provide application-specific context fields in a Java application
4738 A Java application-specific context field is a piece of state provided
4739 by the application which <<adding-context,you can add>>, using the
4740 man:lttng-add-context(1) command, to each <<event,event record>>
4741 produced by the log statements of this application.
4743 For example, a given object might have a current request ID variable.
4744 You can create a context information retriever for this object and
4745 assign a name to this current request ID. You can then, using the
4746 man:lttng-add-context(1) command, add this context field by name to
4747 the JUL or log4j <<channel,channel>>.
4749 To provide application-specific context fields in a Java application:
4751 . In the Java application's source code, import the LTTng-UST
4752 Java agent context classes and interfaces:
4757 import org.lttng.ust.agent.context.ContextInfoManager;
4758 import org.lttng.ust.agent.context.IContextInfoRetriever;
4762 . Create a context information retriever class, that is, a class which
4763 implements the `IContextInfoRetriever` interface:
4768 class MyContextInfoRetriever implements IContextInfoRetriever
4771 public Object retrieveContextInfo(String key)
4773 if (key.equals("intCtx")) {
4775 } else if (key.equals("strContext")) {
4776 return "context value!";
4785 This `retrieveContextInfo()` method is the only member of the
4786 `IContextInfoRetriever` interface. Its role is to return the current
4787 value of a state by name to create a context field. The names of the
4788 context fields and which state variables they return depends on your
4791 All primitive types and objects are supported as context fields.
4792 When `retrieveContextInfo()` returns an object, the context field
4793 serializer calls its `toString()` method to add a string field to
4794 event records. The method can also return `null`, which means that
4795 no context field is available for the required name.
4797 . Register an instance of your context information retriever class to
4798 the context information manager singleton:
4803 IContextInfoRetriever cir = new MyContextInfoRetriever();
4804 ContextInfoManager cim = ContextInfoManager.getInstance();
4805 cim.registerContextInfoRetriever("retrieverName", cir);
4809 . Before exiting the application, remove your context information
4810 retriever from the context information manager singleton:
4815 ContextInfoManager cim = ContextInfoManager.getInstance();
4816 cim.unregisterContextInfoRetriever("retrieverName");
4820 This is not strictly necessary, but it is recommended for a clean
4821 disposal of some manager's resources.
4823 . Build your Java application with LTTng-UST Java agent support as
4824 usual, following the procedure for either the <<jul,JUL>> or
4825 <<log4j,Apache log4j>> framework.
4828 .Provide application-specific context fields in a Java application.
4833 import java.util.logging.Handler;
4834 import java.util.logging.Logger;
4835 import org.lttng.ust.agent.jul.LttngLogHandler;
4836 import org.lttng.ust.agent.context.ContextInfoManager;
4837 import org.lttng.ust.agent.context.IContextInfoRetriever;
4841 // Our context information retriever class
4842 private static class MyContextInfoRetriever
4843 implements IContextInfoRetriever
4846 public Object retrieveContextInfo(String key) {
4847 if (key.equals("intCtx")) {
4849 } else if (key.equals("strContext")) {
4850 return "context value!";
4857 private static final int answer = 42;
4859 public static void main(String args[]) throws Exception
4861 // Get the context information manager instance
4862 ContextInfoManager cim = ContextInfoManager.getInstance();
4864 // Create and register our context information retriever
4865 IContextInfoRetriever cir = new MyContextInfoRetriever();
4866 cim.registerContextInfoRetriever("myRetriever", cir);
4869 Logger logger = Logger.getLogger("jello");
4871 // Create an LTTng-UST log handler
4872 Handler lttngUstLogHandler = new LttngLogHandler();
4874 // Add the LTTng-UST log handler to our logger
4875 logger.addHandler(lttngUstLogHandler);
4878 logger.info("some info");
4879 logger.warning("some warning");
4881 logger.finer("finer information; the answer is " + answer);
4883 logger.severe("error!");
4885 // Not mandatory, but cleaner
4886 logger.removeHandler(lttngUstLogHandler);
4887 lttngUstLogHandler.close();
4888 cim.unregisterContextInfoRetriever("myRetriever");
4897 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4900 <<creating-destroying-tracing-sessions,Create a tracing session>>
4901 and <<enabling-disabling-events,create an event rule>> matching the
4907 lttng enable-event --jul jello
4910 <<adding-context,Add the application-specific context fields>> to the
4915 lttng add-context --jul --type='$app.myRetriever:intCtx'
4916 lttng add-context --jul --type='$app.myRetriever:strContext'
4919 <<basic-tracing-session-control,Start tracing>>:
4926 Run the compiled class:
4930 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4933 <<basic-tracing-session-control,Stop tracing>> and inspect the
4945 [[python-application]]
4946 === User space Python agent
4948 You can instrument a Python 2 or Python 3 application which uses the
4949 standard https://docs.python.org/3/library/logging.html[`logging`]
4952 Each log statement emits an LTTng event once the
4953 application module imports the
4954 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4957 .A Python application importing the LTTng-UST Python agent.
4958 image::python-app.png[]
4960 To use the LTTng-UST Python agent:
4962 . In the Python application's source code, import the LTTng-UST Python
4972 The LTTng-UST Python agent automatically adds its logging handler to the
4973 root logger at import time.
4975 Any log statement that the application executes before this import does
4976 not emit an LTTng event.
4978 IMPORTANT: The LTTng-UST Python agent must be
4979 <<installing-lttng,installed>>.
4981 . Use log statements and logging configuration as usual.
4982 Since the LTTng-UST Python agent adds a handler to the _root_
4983 logger, you can trace any log statement from any logger.
4985 .Use the LTTng-UST Python agent.
4996 logging.basicConfig()
4997 logger = logging.getLogger('my-logger')
5000 logger.debug('debug message')
5001 logger.info('info message')
5002 logger.warn('warn message')
5003 logger.error('error message')
5004 logger.critical('critical message')
5008 if __name__ == '__main__':
5012 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
5013 logging handler which prints to the standard error stream, is not
5014 strictly required for LTTng-UST tracing to work, but in versions of
5015 Python preceding 3.2, you could see a warning message which indicates
5016 that no handler exists for the logger `my-logger`.
5018 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5019 <<enabling-disabling-events,create an event rule>> matching the
5020 `my-logger` Python logger, and <<basic-tracing-session-control,start
5026 lttng enable-event --python my-logger
5030 Run the Python script:
5037 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5047 You can use the opt:lttng-enable-event(1):--loglevel or
5048 opt:lttng-enable-event(1):--loglevel-only option of the
5049 man:lttng-enable-event(1) command to target a range of Python log levels
5050 or a specific Python log level.
5052 When an application imports the LTTng-UST Python agent, the agent tries
5053 to register to a <<lttng-sessiond,session daemon>>. Note that you must
5054 <<start-sessiond,start the session daemon>> _before_ you run the Python
5055 application. If a session daemon is found, the agent tries to register
5056 to it during 5{nbsp}seconds, after which the application continues
5057 without LTTng tracing support. You can override this timeout value with
5058 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5061 If the session daemon stops while a Python application with an imported
5062 LTTng-UST Python agent runs, the agent retries to connect and to
5063 register to a session daemon every 3{nbsp}seconds. You can override this
5064 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5069 [[proc-lttng-logger-abi]]
5072 The `lttng-tracer` Linux kernel module, part of
5073 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
5074 path:{/proc/lttng-logger} when it's loaded. Any application can write
5075 text data to this file to emit an LTTng event.
5078 .An application writes to the LTTng logger file to emit an LTTng event.
5079 image::lttng-logger.png[]
5081 The LTTng logger is the quickest method--not the most efficient,
5082 however--to add instrumentation to an application. It is designed
5083 mostly to instrument shell scripts:
5087 echo "Some message, some $variable" > /proc/lttng-logger
5090 Any event that the LTTng logger emits is named `lttng_logger` and
5091 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5092 other instrumentation points in the kernel tracing domain, **any Unix
5093 user** can <<enabling-disabling-events,create an event rule>> which
5094 matches its event name, not only the root user or users in the
5095 <<tracing-group,tracing group>>.
5097 To use the LTTng logger:
5099 * From any application, write text data to the path:{/proc/lttng-logger}
5102 The `msg` field of `lttng_logger` event records contains the
5105 NOTE: The maximum message length of an LTTng logger event is
5106 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5107 than one event to contain the remaining data.
5109 You should not use the LTTng logger to trace a user application which
5110 can be instrumented in a more efficient way, namely:
5112 * <<c-application,C and $$C++$$ applications>>.
5113 * <<java-application,Java applications>>.
5114 * <<python-application,Python applications>>.
5116 .Use the LTTng logger.
5121 echo 'Hello, World!' > /proc/lttng-logger
5123 df --human-readable --print-type / > /proc/lttng-logger
5126 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5127 <<enabling-disabling-events,create an event rule>> matching the
5128 `lttng_logger` Linux kernel tracepoint, and
5129 <<basic-tracing-session-control,start tracing>>:
5134 lttng enable-event --kernel lttng_logger
5138 Run the Bash script:
5145 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5156 [[instrumenting-linux-kernel]]
5157 === LTTng kernel tracepoints
5159 NOTE: This section shows how to _add_ instrumentation points to the
5160 Linux kernel. The kernel's subsystems are already thoroughly
5161 instrumented at strategic places for LTTng when you
5162 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5166 There are two methods to instrument the Linux kernel:
5168 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5169 tracepoint which uses the `TRACE_EVENT()` API.
5171 Choose this if you want to instrumentation a Linux kernel tree with an
5172 instrumentation point compatible with ftrace, perf, and SystemTap.
5174 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5175 instrument an out-of-tree kernel module.
5177 Choose this if you don't need ftrace, perf, or SystemTap support.
5181 [[linux-add-lttng-layer]]
5182 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5184 This section shows how to add an LTTng layer to existing ftrace
5185 instrumentation using the `TRACE_EVENT()` API.
5187 This section does not document the `TRACE_EVENT()` macro. You can
5188 read the following articles to learn more about this API:
5190 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
5191 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
5192 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
5194 The following procedure assumes that your ftrace tracepoints are
5195 correctly defined in their own header and that they are created in
5196 one source file using the `CREATE_TRACE_POINTS` definition.
5198 To add an LTTng layer over an existing ftrace tracepoint:
5200 . Make sure the following kernel configuration options are
5206 * `CONFIG_HIGH_RES_TIMERS`
5207 * `CONFIG_TRACEPOINTS`
5210 . Build the Linux source tree with your custom ftrace tracepoints.
5211 . Boot the resulting Linux image on your target system.
5213 Confirm that the tracepoints exist by looking for their names in the
5214 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5215 is your subsystem's name.
5217 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5223 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
5224 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
5225 cd lttng-modules-2.8.*
5229 . In dir:{instrumentation/events/lttng-module}, relative to the root
5230 of the LTTng-modules source tree, create a header file named
5231 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5232 LTTng-modules tracepoint definitions using the LTTng-modules
5235 Start with this template:
5239 .path:{instrumentation/events/lttng-module/my_subsys.h}
5242 #define TRACE_SYSTEM my_subsys
5244 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5245 #define _LTTNG_MY_SUBSYS_H
5247 #include "../../../probes/lttng-tracepoint-event.h"
5248 #include <linux/tracepoint.h>
5250 LTTNG_TRACEPOINT_EVENT(
5252 * Format is identical to TRACE_EVENT()'s version for the three
5253 * following macro parameters:
5256 TP_PROTO(int my_int, const char *my_string),
5257 TP_ARGS(my_int, my_string),
5259 /* LTTng-modules specific macros */
5261 ctf_integer(int, my_int_field, my_int)
5262 ctf_string(my_bar_field, my_bar)
5266 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5268 #include "../../../probes/define_trace.h"
5272 The entries in the `TP_FIELDS()` section are the list of fields for the
5273 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5274 ftrace's `TRACE_EVENT()` macro.
5276 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5277 complete description of the available `ctf_*()` macros.
5279 . Create the LTTng-modules probe's kernel module C source file,
5280 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5285 .path:{probes/lttng-probe-my-subsys.c}
5287 #include <linux/module.h>
5288 #include "../lttng-tracer.h"
5291 * Build-time verification of mismatch between mainline
5292 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5293 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5295 #include <trace/events/my_subsys.h>
5297 /* Create LTTng tracepoint probes */
5298 #define LTTNG_PACKAGE_BUILD
5299 #define CREATE_TRACE_POINTS
5300 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5302 #include "../instrumentation/events/lttng-module/my_subsys.h"
5304 MODULE_LICENSE("GPL and additional rights");
5305 MODULE_AUTHOR("Your name <your-email>");
5306 MODULE_DESCRIPTION("LTTng my_subsys probes");
5307 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5308 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5309 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5310 LTTNG_MODULES_EXTRAVERSION);
5314 . Edit path:{probes/Makefile} and add your new kernel module object
5315 next to the existing ones:
5319 .path:{probes/Makefile}
5323 obj-m += lttng-probe-module.o
5324 obj-m += lttng-probe-power.o
5326 obj-m += lttng-probe-my-subsys.o
5332 . Build and install the LTTng kernel modules:
5337 make KERNELDIR=/path/to/linux
5338 sudo make modules_install
5342 Replace `/path/to/linux` with the path to the Linux source tree where
5343 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5345 Note that you can also use the
5346 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5347 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5348 C code that need to be executed before the event fields are recorded.
5350 The best way to learn how to use the previous LTTng-modules macros is to
5351 inspect the existing LTTng-modules tracepoint definitions in the
5352 dir:{instrumentation/events/lttng-module} header files. Compare them
5353 with the Linux kernel mainline versions in the
5354 dir:{include/trace/events} directory of the Linux source tree.
5358 [[lttng-tracepoint-event-code]]
5359 ===== Use custom C code to access the data for tracepoint fields
5361 Although we recommended to always use the
5362 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5363 the arguments and fields of an LTTng-modules tracepoint when possible,
5364 sometimes you need a more complex process to access the data that the
5365 tracer records as event record fields. In other words, you need local
5366 variables and multiple C{nbsp}statements instead of simple
5367 argument-based expressions that you pass to the
5368 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5370 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5371 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5372 a block of C{nbsp}code to be executed before LTTng records the fields.
5373 The structure of this macro is:
5376 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5378 LTTNG_TRACEPOINT_EVENT_CODE(
5380 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5381 * version for the following three macro parameters:
5384 TP_PROTO(int my_int, const char *my_string),
5385 TP_ARGS(my_int, my_string),
5387 /* Declarations of custom local variables */
5390 unsigned long b = 0;
5391 const char *name = "(undefined)";
5392 struct my_struct *my_struct;
5396 * Custom code which uses both tracepoint arguments
5397 * (in TP_ARGS()) and local variables (in TP_locvar()).
5399 * Local variables are actually members of a structure pointed
5400 * to by the special variable tp_locvar.
5404 tp_locvar->a = my_int + 17;
5405 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5406 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5407 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5408 put_my_struct(tp_locvar->my_struct);
5417 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5418 * version for this, except that tp_locvar members can be
5419 * used in the argument expression parameters of
5420 * the ctf_*() macros.
5423 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5424 ctf_integer(int, my_struct_a, tp_locvar->a)
5425 ctf_string(my_string_field, my_string)
5426 ctf_string(my_struct_name, tp_locvar->name)
5431 IMPORTANT: The C code defined in `TP_code()` must not have any side
5432 effects when executed. In particular, the code must not allocate
5433 memory or get resources without deallocating this memory or putting
5434 those resources afterwards.
5437 [[instrumenting-linux-kernel-tracing]]
5438 ==== Load and unload a custom probe kernel module
5440 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5441 kernel module>> in the kernel before it can emit LTTng events.
5443 To load the default probe kernel modules and a custom probe kernel
5446 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5447 probe modules to load when starting a root <<lttng-sessiond,session
5451 .Load the `my_subsys`, `usb`, and the default probe modules.
5455 sudo lttng-sessiond --extra-kmod-probes=my_subsys,usb
5460 You only need to pass the subsystem name, not the whole kernel module
5463 To load _only_ a given custom probe kernel module:
5465 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5466 modules to load when starting a root session daemon:
5469 .Load only the `my_subsys` and `usb` probe modules.
5473 sudo lttng-sessiond --kmod-probes=my_subsys,usb
5478 To confirm that a probe module is loaded:
5485 lsmod | grep lttng_probe_usb
5489 To unload the loaded probe modules:
5491 * Kill the session daemon with `SIGTERM`:
5496 sudo pkill lttng-sessiond
5500 You can also use man:modprobe(8)'s `--remove` option if the session
5501 daemon terminates abnormally.
5504 [[controlling-tracing]]
5507 Once an application or a Linux kernel is
5508 <<instrumenting,instrumented>> for LTTng tracing,
5511 This section is divided in topics on how to use the various
5512 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5513 command-line tool>>, to _control_ the LTTng daemons and tracers.
5515 NOTE: In the following subsections, we refer to an man:lttng(1) command
5516 using its man page name. For example, instead of _Run the `create`
5517 command to..._, we use _Run the man:lttng-create(1) command to..._.
5521 === Start a session daemon
5523 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5524 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5527 You will see the following error when you run a command while no session
5531 Error: No session daemon is available
5534 The only command that automatically runs a session daemon is
5535 man:lttng-create(1), which you use to
5536 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5537 this is most of the time the first operation that you do, sometimes it's
5538 not. Some examples are:
5540 * <<list-instrumentation-points,List the available instrumentation points>>.
5541 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5543 [[tracing-group]] Each Unix user must have its own running session
5544 daemon to trace user applications. The session daemon that the root user
5545 starts is the only one allowed to control the LTTng kernel tracer. Users
5546 that are part of the _tracing group_ can control the root session
5547 daemon. The default tracing group name is `tracing`; you can set it to
5548 something else with the opt:lttng-sessiond(8):--group option when you
5549 start the root session daemon.
5551 To start a user session daemon:
5553 * Run man:lttng-sessiond(8):
5558 lttng-sessiond --daemonize
5562 To start the root session daemon:
5564 * Run man:lttng-sessiond(8) as the root user:
5569 sudo lttng-sessiond --daemonize
5573 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5574 start the session daemon in foreground.
5576 To stop a session daemon, use man:kill(1) on its process ID (standard
5579 Note that some Linux distributions could manage the LTTng session daemon
5580 as a service. In this case, you should use the service manager to
5581 start, restart, and stop session daemons.
5584 [[creating-destroying-tracing-sessions]]
5585 === Create and destroy a tracing session
5587 Almost all the LTTng control operations happen in the scope of
5588 a <<tracing-session,tracing session>>, which is the dialogue between the
5589 <<lttng-sessiond,session daemon>> and you.
5591 To create a tracing session with a generated name:
5593 * Use the man:lttng-create(1) command:
5602 The created tracing session's name is `auto` followed by the
5605 To create a tracing session with a specific name:
5607 * Use the optional argument of the man:lttng-create(1) command:
5612 lttng create my-session
5616 Replace `my-session` with the specific tracing session name.
5618 LTTng appends the creation date to the created tracing session's name.
5620 LTTng writes the traces of a tracing session in
5621 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5622 name of the tracing session. Note that the env:LTTNG_HOME environment
5623 variable defaults to `$HOME` if not set.
5625 To output LTTng traces to a non-default location:
5627 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5632 lttng create --output=/tmp/some-directory my-session
5636 You may create as many tracing sessions as you wish.
5638 To list all the existing tracing sessions for your Unix user:
5640 * Use the man:lttng-list(1) command:
5649 When you create a tracing session, it is set as the _current tracing
5650 session_. The following man:lttng(1) commands operate on the current
5651 tracing session when you don't specify one:
5653 [role="list-3-cols"]
5669 To change the current tracing session:
5671 * Use the man:lttng-set-session(1) command:
5676 lttng set-session new-session
5680 Replace `new-session` by the name of the new current tracing session.
5682 When you are done tracing in a given tracing session, you can destroy
5683 it. This operation frees the resources taken by the tracing session
5684 to destroy; it does not destroy the trace data that LTTng wrote for
5685 this tracing session.
5687 To destroy the current tracing session:
5689 * Use the man:lttng-destroy(1) command:
5699 [[list-instrumentation-points]]
5700 === List the available instrumentation points
5702 The <<lttng-sessiond,session daemon>> can query the running instrumented
5703 user applications and the Linux kernel to get a list of available
5704 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5705 they are tracepoints and system calls. For the user space tracing
5706 domain, they are tracepoints. For the other tracing domains, they are
5709 To list the available instrumentation points:
5711 * Use the man:lttng-list(1) command with the requested tracing domain's
5715 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5716 must be a root user, or it must be a member of the
5717 <<tracing-group,tracing group>>).
5718 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5719 kernel system calls (your Unix user must be a root user, or it must be
5720 a member of the tracing group).
5721 * opt:lttng-list(1):--userspace: user space tracepoints.
5722 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5723 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5724 * opt:lttng-list(1):--python: Python loggers.
5727 .List the available user space tracepoints.
5731 lttng list --userspace
5735 .List the available Linux kernel system call tracepoints.
5739 lttng list --kernel --syscall
5744 [[enabling-disabling-events]]
5745 === Create and enable an event rule
5747 Once you <<creating-destroying-tracing-sessions,create a tracing
5748 session>>, you can create <<event,event rules>> with the
5749 man:lttng-enable-event(1) command.
5751 You specify each condition with a command-line option. The available
5752 condition options are shown in the following table.
5754 [role="growable",cols="asciidoc,asciidoc,default"]
5755 .Condition command-line options for the man:lttng-enable-event(1) command.
5757 |Option |Description |Applicable tracing domains
5763 . +--probe=__ADDR__+
5764 . +--function=__ADDR__+
5767 Instead of using the default _tracepoint_ instrumentation type, use:
5769 . A Linux system call.
5770 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5771 . The entry and return points of a Linux function (symbol or address).
5775 |First positional argument.
5778 Tracepoint or system call name. In the case of a Linux KProbe or
5779 function, this is a custom name given to the event rule. With the
5780 JUL, log4j, and Python domains, this is a logger name.
5782 With a tracepoint, logger, or system call name, the last character
5783 can be `*` to match anything that remains.
5790 . +--loglevel=__LEVEL__+
5791 . +--loglevel-only=__LEVEL__+
5794 . Match only tracepoints or log statements with a logging level at
5795 least as severe as +__LEVEL__+.
5796 . Match only tracepoints or log statements with a logging level
5797 equal to +__LEVEL__+.
5799 See man:lttng-enable-event(1) for the list of available logging level
5802 |User space, JUL, log4j, and Python.
5804 |+--exclude=__EXCLUSIONS__+
5807 When you use a `*` character at the end of the tracepoint or logger
5808 name (first positional argument), exclude the specific names in the
5809 comma-delimited list +__EXCLUSIONS__+.
5812 User space, JUL, log4j, and Python.
5814 |+--filter=__EXPR__+
5817 Match only events which satisfy the expression +__EXPR__+.
5819 See man:lttng-enable-event(1) to learn more about the syntax of a
5826 You attach an event rule to a <<channel,channel>> on creation. If you do
5827 not specify the channel with the opt:lttng-enable-event(1):--channel
5828 option, and if the event rule to create is the first in its
5829 <<domain,tracing domain>> for a given tracing session, then LTTng
5830 creates a _default channel_ for you. This default channel is reused in
5831 subsequent invocations of the man:lttng-enable-event(1) command for the
5832 same tracing domain.
5834 An event rule is always enabled at creation time.
5836 The following examples show how you can combine the previous
5837 command-line options to create simple to more complex event rules.
5839 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5843 lttng enable-event --kernel sched_switch
5847 .Create an event rule matching four Linux kernel system calls (default channel).
5851 lttng enable-event --kernel --syscall open,write,read,close
5855 .Create event rules matching tracepoints with filter expressions (default channel).
5859 lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5864 lttng enable-event --kernel --all \
5865 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5870 lttng enable-event --jul my_logger \
5871 --filter='$app.retriever:cur_msg_id > 3'
5874 IMPORTANT: Make sure to always quote the filter string when you
5875 use man:lttng(1) from a shell.
5878 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5882 lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5885 IMPORTANT: Make sure to always quote the wildcard character when you
5886 use man:lttng(1) from a shell.
5889 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5893 lttng enable-event --python my-app.'*' \
5894 --exclude='my-app.module,my-app.hello'
5898 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5902 lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5906 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5910 lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5914 The event rules of a given channel form a whitelist: as soon as an
5915 emitted event passes one of them, LTTng can record the event. For
5916 example, an event named `my_app:my_tracepoint` emitted from a user space
5917 tracepoint with a `TRACE_ERROR` log level passes both of the following
5922 lttng enable-event --userspace my_app:my_tracepoint
5923 lttng enable-event --userspace my_app:my_tracepoint \
5924 --loglevel=TRACE_INFO
5927 The second event rule is redundant: the first one includes
5931 [[disable-event-rule]]
5932 === Disable an event rule
5934 To disable an event rule that you <<enabling-disabling-events,created>>
5935 previously, use the man:lttng-disable-event(1) command. This command
5936 disables _all_ the event rules (of a given tracing domain and channel)
5937 which match an instrumentation point. The other conditions are not
5938 supported as of LTTng{nbsp}{revision}.
5940 The LTTng tracer does not record an emitted event which passes
5941 a _disabled_ event rule.
5943 .Disable an event rule matching a Python logger (default channel).
5947 lttng disable-event --python my-logger
5951 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5955 lttng disable-event --jul '*'
5959 .Disable _all_ the event rules of the default channel.
5961 The opt:lttng-disable-event(1):--all-events option is not, like the
5962 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5963 equivalent of the event name `*` (wildcard): it disables _all_ the event
5964 rules of a given channel.
5968 lttng disable-event --jul --all-events
5972 NOTE: You cannot delete an event rule once you create it.
5976 === Get the status of a tracing session
5978 To get the status of the current tracing session, that is, its
5979 parameters, its channels, event rules, and their attributes:
5981 * Use the man:lttng-status(1) command:
5991 To get the status of any tracing session:
5993 * Use the man:lttng-list(1) command with the tracing session's name:
5998 lttng list my-session
6002 Replace `my-session` with the desired tracing session's name.
6005 [[basic-tracing-session-control]]
6006 === Start and stop a tracing session
6008 Once you <<creating-destroying-tracing-sessions,create a tracing
6010 <<enabling-disabling-events,create one or more event rules>>,
6011 you can start and stop the tracers for this tracing session.
6013 To start tracing in the current tracing session:
6015 * Use the man:lttng-start(1) command:
6024 LTTng is very flexible: you can launch user applications before
6025 or after the you start the tracers. The tracers only record the events
6026 if they pass enabled event rules and if they occur while the tracers are
6029 To stop tracing in the current tracing session:
6031 * Use the man:lttng-stop(1) command:
6040 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
6041 records>> or lost sub-buffers since the last time you ran
6042 man:lttng-start(1), warnings are printed when you run the
6043 man:lttng-stop(1) command.
6046 [[enabling-disabling-channels]]
6047 === Create a channel
6049 Once you create a tracing session, you can create a <<channel,channel>>
6050 with the man:lttng-enable-channel(1) command.
6052 Note that LTTng automatically creates a default channel when, for a
6053 given <<domain,tracing domain>>, no channels exist and you
6054 <<enabling-disabling-events,create>> the first event rule. This default
6055 channel is named `channel0` and its attributes are set to reasonable
6056 values. Therefore, you only need to create a channel when you need
6057 non-default attributes.
6059 You specify each non-default channel attribute with a command-line
6060 option when you use the man:lttng-enable-channel(1) command. The
6061 available command-line options are:
6063 [role="growable",cols="asciidoc,asciidoc"]
6064 .Command-line options for the man:lttng-enable-channel(1) command.
6066 |Option |Description
6072 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
6073 the default _discard_ mode.
6075 |`--buffers-pid` (user space tracing domain only)
6078 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6079 instead of the default per-user buffering scheme.
6081 |+--subbuf-size=__SIZE__+
6084 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6085 either for each Unix user (default), or for each instrumented process.
6087 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6089 |+--num-subbuf=__COUNT__+
6092 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6093 for each Unix user (default), or for each instrumented process.
6095 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6097 |+--tracefile-size=__SIZE__+
6100 Set the maximum size of each trace file that this channel writes within
6101 a stream to +__SIZE__+ bytes instead of no maximum.
6103 See <<tracefile-rotation,Trace file count and size>>.
6105 |+--tracefile-count=__COUNT__+
6108 Limit the number of trace files that this channel creates to
6109 +__COUNT__+ channels instead of no limit.
6111 See <<tracefile-rotation,Trace file count and size>>.
6113 |+--switch-timer=__PERIODUS__+
6116 Set the <<channel-switch-timer,switch timer period>>
6117 to +__PERIODUS__+{nbsp}µs.
6119 |+--read-timer=__PERIODUS__+
6122 Set the <<channel-read-timer,read timer period>>
6123 to +__PERIODUS__+{nbsp}µs.
6125 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6128 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6132 You can only create a channel in the Linux kernel and user space
6133 <<domain,tracing domains>>: other tracing domains have their own channel
6134 created on the fly when <<enabling-disabling-events,creating event
6139 Because of a current LTTng limitation, you must create all channels
6140 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6141 tracing session, that is, before the first time you run
6144 Since LTTng automatically creates a default channel when you use the
6145 man:lttng-enable-event(1) command with a specific tracing domain, you
6146 cannot, for example, create a Linux kernel event rule, start tracing,
6147 and then create a user space event rule, because no user space channel
6148 exists yet and it's too late to create one.
6150 For this reason, make sure to configure your channels properly
6151 before starting the tracers for the first time!
6154 The following examples show how you can combine the previous
6155 command-line options to create simple to more complex channels.
6157 .Create a Linux kernel channel with default attributes.
6161 lttng enable-channel --kernel my-channel
6165 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6169 lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6170 --buffers-pid my-channel
6174 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
6178 lttng enable-channel --kernel --tracefile-count=8 \
6179 --tracefile-size=4194304 my-channel
6183 .Create a user space channel in overwrite (or _flight recorder_) mode.
6187 lttng enable-channel --userspace --overwrite my-channel
6191 You can <<enabling-disabling-events,create>> the same event rule in
6192 two different channels:
6196 lttng enable-event --userspace --channel=my-channel app:tp
6197 lttng enable-event --userspace --channel=other-channel app:tp
6200 If both channels are enabled, when a tracepoint named `app:tp` is
6201 reached, LTTng records two events, one for each channel.
6205 === Disable a channel
6207 To disable a specific channel that you <<enabling-disabling-channels,created>>
6208 previously, use the man:lttng-disable-channel(1) command.
6210 .Disable a specific Linux kernel channel.
6214 lttng disable-channel --kernel my-channel
6218 The state of a channel precedes the individual states of event rules
6219 attached to it: event rules which belong to a disabled channel, even if
6220 they are enabled, are also considered disabled.
6224 === Add context fields to a channel
6226 Event record fields in trace files provide important information about
6227 events that occured previously, but sometimes some external context may
6228 help you solve a problem faster. Examples of context fields are:
6230 * The **process ID**, **thread ID**, **process name**, and
6231 **process priority** of the thread in which the event occurs.
6232 * The **hostname** of the system on which the event occurs.
6233 * The current values of many possible **performance counters** using
6235 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6237 ** Branch instructions, misses, and loads.
6239 * Any context defined at the application level (supported for the
6240 JUL and log4j <<domain,tracing domains>>).
6242 To get the full list of available context fields, see
6243 `lttng add-context --list`. Some context fields are reserved for a
6244 specific <<domain,tracing domain>> (Linux kernel or user space).
6246 You add context fields to <<channel,channels>>. All the events
6247 that a channel with added context fields records contain those fields.
6249 To add context fields to one or all the channels of a given tracing
6252 * Use the man:lttng-add-context(1) command.
6254 .Add context fields to all the channels of the current tracing session.
6256 The following command line adds the virtual process identifier and
6257 the per-thread CPU cycles count fields to all the user space channels
6258 of the current tracing session.
6262 lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6266 .Add a context field to a specific channel.
6268 The following command line adds the thread identifier context field
6269 to the Linux kernel channel named `my-channel` in the current
6274 lttng add-context --kernel --channel=my-channel --type=tid
6278 .Add an application-specific context field to a specific channel.
6280 The following command line adds the `cur_msg_id` context field of the
6281 `retriever` context retriever for all the instrumented
6282 <<java-application,Java applications>> recording <<event,event records>>
6283 in the channel named `my-channel`:
6287 lttng add-context --kernel --channel=my-channel \
6288 --type='$app:retriever:cur_msg_id'
6291 IMPORTANT: Make sure to always quote the `$` character when you
6292 use man:lttng-add-context(1) from a shell.
6295 NOTE: You cannot remove context fields from a channel once you add it.
6300 === Track process IDs
6302 It's often useful to allow only specific process IDs (PIDs) to emit
6303 events. For example, you may wish to record all the system calls made by
6304 a given process (à la http://linux.die.net/man/1/strace[strace]).
6306 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6307 purpose. Both commands operate on a whitelist of process IDs. You _add_
6308 entries to this whitelist with the man:lttng-track(1) command and remove
6309 entries with the man:lttng-untrack(1) command. Any process which has one
6310 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6311 an enabled <<event,event rule>>.
6313 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6314 process with a given tracked ID exit and another process be given this
6315 ID, then the latter would also be allowed to emit events.
6317 .Track and untrack process IDs.
6319 For the sake of the following example, assume the target system has 16
6323 <<creating-destroying-tracing-sessions,create a tracing session>>,
6324 the whitelist contains all the possible PIDs:
6327 .All PIDs are tracked.
6328 image::track-all.png[]
6330 When the whitelist is full and you use the man:lttng-track(1) command to
6331 specify some PIDs to track, LTTng first clears the whitelist, then it
6332 tracks the specific PIDs. After:
6336 lttng track --pid=3,4,7,10,13
6342 .PIDs 3, 4, 7, 10, and 13 are tracked.
6343 image::track-3-4-7-10-13.png[]
6345 You can add more PIDs to the whitelist afterwards:
6349 lttng track --pid=1,15,16
6355 .PIDs 1, 15, and 16 are added to the whitelist.
6356 image::track-1-3-4-7-10-13-15-16.png[]
6358 The man:lttng-untrack(1) command removes entries from the PID tracker's
6359 whitelist. Given the previous example, the following command:
6363 lttng untrack --pid=3,7,10,13
6366 leads to this whitelist:
6369 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6370 image::track-1-4-15-16.png[]
6372 LTTng can track all possible PIDs again using the opt:track(1):--all
6377 lttng track --pid --all
6380 The result is, again:
6383 .All PIDs are tracked.
6384 image::track-all.png[]
6387 .Track only specific PIDs
6389 A very typical use case with PID tracking is to start with an empty
6390 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6391 then add PIDs manually while tracers are active. You can accomplish this
6392 by using the opt:lttng-untrack(1):--all option of the
6393 man:lttng-untrack(1) command to clear the whitelist after you
6394 <<creating-destroying-tracing-sessions,create a tracing session>>:
6398 lttng untrack --pid --all
6404 .No PIDs are tracked.
6405 image::untrack-all.png[]
6407 If you trace with this whitelist configuration, the tracer records no
6408 events for this <<domain,tracing domain>> because no processes are
6409 tracked. You can use the man:lttng-track(1) command as usual to track
6410 specific PIDs, for example:
6414 lttng track --pid=6,11
6420 .PIDs 6 and 11 are tracked.
6421 image::track-6-11.png[]
6426 [[saving-loading-tracing-session]]
6427 === Save and load tracing session configurations
6429 Configuring a <<tracing-session,tracing session>> can be long. Some of
6430 the tasks involved are:
6432 * <<enabling-disabling-channels,Create channels>> with
6433 specific attributes.
6434 * <<adding-context,Add context fields>> to specific channels.
6435 * <<enabling-disabling-events,Create event rules>> with specific log
6436 level and filter conditions.
6438 If you use LTTng to solve real world problems, chances are you have to
6439 record events using the same tracing session setup over and over,
6440 modifying a few variables each time in your instrumented program
6441 or environment. To avoid constant tracing session reconfiguration,
6442 the man:lttng(1) command-line tool can save and load tracing session
6443 configurations to/from XML files.
6445 To save a given tracing session configuration:
6447 * Use the man:lttng-save(1) command:
6452 lttng save my-session
6456 Replace `my-session` with the name of the tracing session to save.
6458 LTTng saves tracing session configurations to
6459 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6460 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6461 the opt:lttng-save(1):--output-path option to change this destination
6464 LTTng saves all configuration parameters, for example:
6466 * The tracing session name.
6467 * The trace data output path.
6468 * The channels with their state and all their attributes.
6469 * The context fields you added to channels.
6470 * The event rules with their state, log level and filter conditions.
6472 To load a tracing session:
6474 * Use the man:lttng-load(1) command:
6479 lttng load my-session
6483 Replace `my-session` with the name of the tracing session to load.
6485 When LTTng loads a configuration, it restores your saved tracing session
6486 as if you just configured it manually.
6488 See man:lttng(1) for the complete list of command-line options. You
6489 can also save and load all many sessions at a time, and decide in which
6490 directory to output the XML files.
6493 [[sending-trace-data-over-the-network]]
6494 === Send trace data over the network
6496 LTTng can send the recorded trace data to a remote system over the
6497 network instead of writing it to the local file system.
6499 To send the trace data over the network:
6501 . On the _remote_ system (which can also be the target system),
6502 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6511 . On the _target_ system, create a tracing session configured to
6512 send trace data over the network:
6517 lttng create my-session --set-url=net://remote-system
6521 Replace `remote-system` by the host name or IP address of the
6522 remote system. See man:lttng-create(1) for the exact URL format.
6524 . On the target system, use the man:lttng(1) command-line tool as usual.
6525 When tracing is active, the target's consumer daemon sends sub-buffers
6526 to the relay daemon running on the remote system intead of flushing
6527 them to the local file system. The relay daemon writes the received
6528 packets to the local file system.
6530 The relay daemon writes trace files to
6531 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6532 +__hostname__+ is the host name of the target system and +__session__+
6533 is the tracing session name. Note that the env:LTTNG_HOME environment
6534 variable defaults to `$HOME` if not set. Use the
6535 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6536 trace files to another base directory.
6541 === View events as LTTng emits them (noch:{LTTng} live)
6543 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6544 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6545 display events as LTTng emits them on the target system while tracing is
6548 The relay daemon creates a _tee_: it forwards the trace data to both
6549 the local file system and to connected live viewers:
6552 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6557 . On the _target system_, create a <<tracing-session,tracing session>>
6563 lttng create --live my-session
6567 This spawns a local relay daemon.
6569 . Start the live viewer and configure it to connect to the relay
6570 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6575 babeltrace --input-format=lttng-live net://localhost/host/hostname/my-session
6582 * `hostname` with the host name of the target system.
6583 * `my-session` with the name of the tracing session to view.
6586 . Configure the tracing session as usual with the man:lttng(1)
6587 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6589 You can list the available live tracing sessions with Babeltrace:
6593 babeltrace --input-format=lttng-live net://localhost
6596 You can start the relay daemon on another system. In this case, you need
6597 to specify the relay daemon's URL when you create the tracing session
6598 with the opt:lttng-create(1):--set-url option. You also need to replace
6599 `localhost` in the procedure above with the host name of the system on
6600 which the relay daemon is running.
6602 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6603 command-line options.
6607 [[taking-a-snapshot]]
6608 === Take a snapshot of the current sub-buffers of a tracing session
6610 The normal behavior of LTTng is to append full sub-buffers to growing
6611 trace data files. This is ideal to keep a full history of the events
6612 that occurred on the target system, but it can
6613 represent too much data in some situations. For example, you may wish
6614 to trace your application continuously until some critical situation
6615 happens, in which case you only need the latest few recorded
6616 events to perform the desired analysis, not multi-gigabyte trace files.
6618 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6619 current sub-buffers of a given <<tracing-session,tracing session>>.
6620 LTTng can write the snapshot to the local file system or send it over
6625 . Create a tracing session in _snapshot mode_:
6630 lttng create --snapshot my-session
6634 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6635 <<channel,channels>> created in this mode is automatically set to
6636 _overwrite_ (flight recorder mode).
6638 . Configure the tracing session as usual with the man:lttng(1)
6639 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6641 . **Optional**: When you need to take a snapshot,
6642 <<basic-tracing-session-control,stop tracing>>.
6644 You can take a snapshot when the tracers are active, but if you stop
6645 them first, you are sure that the data in the sub-buffers does not
6646 change before you actually take the snapshot.
6653 lttng snapshot record --name=my-first-snapshot
6657 LTTng writes the current sub-buffers of all the current tracing
6658 session's channels to trace files on the local file system. Those trace
6659 files have `my-first-snapshot` in their name.
6661 There is no difference between the format of a normal trace file and the
6662 format of a snapshot: viewers of LTTng traces also support LTTng
6665 By default, LTTng writes snapshot files to the path shown by
6666 `lttng snapshot list-output`. You can change this path or decide to send
6667 snapshots over the network using either:
6669 . An output path or URL that you specify when you create the
6671 . An snapshot output path or URL that you add using
6672 `lttng snapshot add-output`
6673 . An output path or URL that you provide directly to the
6674 `lttng snapshot record` command.
6676 Method 3 overrides method 2, which overrides method 1. When you
6677 specify a URL, a relay daemon must listen on a remote system (see
6678 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6683 === Use the machine interface
6685 With any command of the man:lttng(1) command-line tool, you can set the
6686 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6687 XML machine interface output, for example:
6691 lttng --mi=xml enable-event --kernel --syscall open
6694 A schema definition (XSD) is
6695 https://github.com/lttng/lttng-tools/blob/stable-2.8/src/common/mi-lttng-3.0.xsd[available]
6696 to ease the integration with external tools as much as possible.
6700 [[metadata-regenerate]]
6701 === Regenerate the metadata of an LTTng trace
6703 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6704 data stream files and a metadata file. This metadata file contains,
6705 amongst other things, information about the offset of the clock sources
6706 used to timestamp <<event,event records>> when tracing.
6708 If, once a <<tracing-session,tracing session>> is
6709 <<basic-tracing-session-control,started>>, a major
6710 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6711 happens, the trace's clock offset also needs to be updated. You
6712 can use the man:lttng-metadata(1) command to do so.
6714 The main use case of this command is to allow a system to boot with
6715 an incorrect wall time and trace it with LTTng before its wall time
6716 is corrected. Once the system is known to be in a state where its
6717 wall time is correct, it can run `lttng metadata regenerate`.
6719 To regenerate the metadata of an LTTng trace:
6721 * Use the `regenerate` action of the man:lttng-metadata(1) command:
6726 lttng metadata regenerate
6732 `lttng metadata regenerate` has the following limitations:
6734 * Tracing session <<creating-destroying-tracing-sessions,created>>
6736 * User space <<channel,channels>>, if any, using
6737 <<channel-buffering-schemes,per-user buffering>>.
6742 [[persistent-memory-file-systems]]
6743 === Record trace data on persistent memory file systems
6745 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6746 (NVRAM) is random-access memory that retains its information when power
6747 is turned off (non-volatile). Systems with such memory can store data
6748 structures in RAM and retrieve them after a reboot, without flushing
6749 to typical _storage_.
6751 Linux supports NVRAM file systems thanks to either
6752 http://pramfs.sourceforge.net/[PRAMFS] or
6753 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6754 (requires Linux 4.1+).
6756 This section does not describe how to operate such file systems;
6757 we assume that you have a working persistent memory file system.
6759 When you create a <<tracing-session,tracing session>>, you can specify
6760 the path of the shared memory holding the sub-buffers. If you specify a
6761 location on an NVRAM file system, then you can retrieve the latest
6762 recorded trace data when the system reboots after a crash.
6764 To record trace data on a persistent memory file system and retrieve the
6765 trace data after a system crash:
6767 . Create a tracing session with a sub-buffer shared memory path located
6768 on an NVRAM file system:
6773 lttng create --shm-path=/path/to/shm
6777 . Configure the tracing session as usual with the man:lttng(1)
6778 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6780 . After a system crash, use the man:lttng-crash(1) command-line tool to
6781 view the trace data recorded on the NVRAM file system:
6786 lttng-crash /path/to/shm
6790 The binary layout of the ring buffer files is not exactly the same as
6791 the trace files layout. This is why you need to use man:lttng-crash(1)
6792 instead of your preferred trace viewer directly.
6794 To convert the ring buffer files to LTTng trace files:
6796 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6801 lttng-crash --extract=/path/to/trace /path/to/shm
6809 [[lttng-modules-ref]]
6810 === noch:{LTTng-modules}
6813 [[lttng-modules-tp-fields]]
6814 ==== Tracepoint fields macros (for `TP_FIELDS()`)
6816 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
6817 tracepoint fields, which must be listed within `TP_FIELDS()` in
6818 `LTTNG_TRACEPOINT_EVENT()`, are:
6820 [role="func-desc growable",cols="asciidoc,asciidoc"]
6821 .Available macros to define LTTng-modules tracepoint fields
6823 |Macro |Description and parameters
6826 +ctf_integer(__t__, __n__, __e__)+
6828 +ctf_integer_nowrite(__t__, __n__, __e__)+
6830 +ctf_user_integer(__t__, __n__, __e__)+
6832 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
6834 Standard integer, displayed in base 10.
6837 Integer C type (`int`, `long`, `size_t`, ...).
6843 Argument expression.
6846 +ctf_integer_hex(__t__, __n__, __e__)+
6848 +ctf_user_integer_hex(__t__, __n__, __e__)+
6850 Standard integer, displayed in base 16.
6859 Argument expression.
6861 |+ctf_integer_oct(__t__, __n__, __e__)+
6863 Standard integer, displayed in base 8.
6872 Argument expression.
6875 +ctf_integer_network(__t__, __n__, __e__)+
6877 +ctf_user_integer_network(__t__, __n__, __e__)+
6879 Integer in network byte order (big-endian), displayed in base 10.
6888 Argument expression.
6891 +ctf_integer_network_hex(__t__, __n__, __e__)+
6893 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
6895 Integer in network byte order, displayed in base 16.
6904 Argument expression.
6907 +ctf_string(__n__, __e__)+
6909 +ctf_string_nowrite(__n__, __e__)+
6911 +ctf_user_string(__n__, __e__)+
6913 +ctf_user_string_nowrite(__n__, __e__)+
6915 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
6921 Argument expression.
6924 +ctf_array(__t__, __n__, __e__, __s__)+
6926 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
6928 +ctf_user_array(__t__, __n__, __e__, __s__)+
6930 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
6932 Statically-sized array of integers.
6935 Array element C type.
6941 Argument expression.
6947 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
6949 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6951 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
6953 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6955 Statically-sized array of bits.
6957 The type of +__e__+ must be an integer type. +__s__+ is the number
6958 of elements of such type in +__e__+, not the number of bits.
6961 Array element C type.
6967 Argument expression.
6973 +ctf_array_text(__t__, __n__, __e__, __s__)+
6975 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
6977 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
6979 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
6981 Statically-sized array, printed as text.
6983 The string does not need to be null-terminated.
6986 Array element C type (always `char`).
6992 Argument expression.
6998 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7000 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7002 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7004 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7006 Dynamically-sized array of integers.
7008 The type of +__E__+ must be unsigned.
7011 Array element C type.
7017 Argument expression.
7020 Length expression C type.
7026 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7028 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7030 Dynamically-sized array of integers, displayed in base 16.
7032 The type of +__E__+ must be unsigned.
7035 Array element C type.
7041 Argument expression.
7044 Length expression C type.
7049 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7051 Dynamically-sized array of integers in network byte order (big-endian),
7052 displayed in base 10.
7054 The type of +__E__+ must be unsigned.
7057 Array element C type.
7063 Argument expression.
7066 Length expression C type.
7072 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7074 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7076 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7078 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7080 Dynamically-sized array of bits.
7082 The type of +__e__+ must be an integer type. +__s__+ is the number
7083 of elements of such type in +__e__+, not the number of bits.
7085 The type of +__E__+ must be unsigned.
7088 Array element C type.
7094 Argument expression.
7097 Length expression C type.
7103 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7105 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7107 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7109 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7111 Dynamically-sized array, displayed as text.
7113 The string does not need to be null-terminated.
7115 The type of +__E__+ must be unsigned.
7117 The behaviour is undefined if +__e__+ is `NULL`.
7120 Sequence element C type (always `char`).
7126 Argument expression.
7129 Length expression C type.
7135 Use the `_user` versions when the argument expression, `e`, is
7136 a user space address. In the cases of `ctf_user_integer*()` and
7137 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7140 The `_nowrite` versions omit themselves from the session trace, but are
7141 otherwise identical. This means the `_nowrite` fields won't be written
7142 in the recorded trace. Their primary purpose is to make some
7143 of the event context available to the
7144 <<enabling-disabling-events,event filters>> without having to
7145 commit the data to sub-buffers.
7151 Terms related to LTTng and to tracing in general:
7154 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7155 the cmd:babeltrace command, some libraries, and Python bindings.
7157 <<channel-buffering-schemes,buffering scheme>>::
7158 A layout of sub-buffers applied to a given channel.
7160 <<channel,channel>>::
7161 An entity which is responsible for a set of ring buffers.
7163 <<event,Event rules>> are always attached to a specific channel.
7166 A reference of time for a tracer.
7168 <<lttng-consumerd,consumer daemon>>::
7169 A process which is responsible for consuming the full sub-buffers
7170 and write them to a file system or send them over the network.
7172 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7173 mode in which the tracer _discards_ new event records when there's no
7174 sub-buffer space left to store them.
7177 The consequence of the execution of an instrumentation
7178 point, like a tracepoint that you manually place in some source code,
7179 or a Linux kernel KProbe.
7181 An event is said to _occur_ at a specific time. Different actions can
7182 be taken upon the occurance of an event, like record the event's payload
7185 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7186 The mechanism by which event records of a given channel are lost
7187 (not recorded) when there is no sub-buffer space left to store them.
7189 [[def-event-name]]event name::
7190 The name of an event, which is also the name of the event record.
7191 This is also called the _instrumentation point name_.
7194 A record, in a trace, of the payload of an event which occured.
7196 <<event,event rule>>::
7197 Set of conditions which must be satisfied for one or more occuring
7198 events to be recorded.
7200 `java.util.logging`::
7202 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7204 <<instrumenting,instrumentation>>::
7205 The use of LTTng probes to make a piece of software traceable.
7207 instrumentation point::
7208 A point in the execution path of a piece of software that, when
7209 reached by this execution, can emit an event.
7211 instrumentation point name::
7212 See _<<def-event-name,event name>>_.
7215 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7216 developed by the Apache Software Foundation.
7219 Level of severity of a log statement or user space
7220 instrumentation point.
7223 The _Linux Trace Toolkit: next generation_ project.
7225 <<lttng-cli,cmd:lttng>>::
7226 A command-line tool provided by the LTTng-tools project which you
7227 can use to send and receive control messages to and from a
7231 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7232 which is a set of analyzing programs that are used to obtain a
7233 higher level view of an LTTng trace.
7235 cmd:lttng-consumerd::
7236 The name of the consumer daemon program.
7239 A utility provided by the LTTng-tools project which can convert
7240 ring buffer files (usually
7241 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7244 LTTng Documentation::
7247 <<lttng-live,LTTng live>>::
7248 A communication protocol between the relay daemon and live viewers
7249 which makes it possible to see events "live", as they are received by
7252 <<lttng-modules,LTTng-modules>>::
7253 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7254 which contains the Linux kernel modules to make the Linux kernel
7255 instrumentation points available for LTTng tracing.
7258 The name of the relay daemon program.
7260 cmd:lttng-sessiond::
7261 The name of the session daemon program.
7264 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7265 contains the various programs and libraries used to
7266 <<controlling-tracing,control tracing>>.
7268 <<lttng-ust,LTTng-UST>>::
7269 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7270 contains libraries to instrument user applications.
7272 <<lttng-ust-agents,LTTng-UST Java agent>>::
7273 A Java package provided by the LTTng-UST project to allow the
7274 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7277 <<lttng-ust-agents,LTTng-UST Python agent>>::
7278 A Python package provided by the LTTng-UST project to allow the
7279 LTTng instrumentation of Python logging statements.
7281 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7282 The event loss mode in which new event records overwrite older
7283 event records when there's no sub-buffer space left to store them.
7285 <<channel-buffering-schemes,per-process buffering>>::
7286 A buffering scheme in which each instrumented process has its own
7287 sub-buffers for a given user space channel.
7289 <<channel-buffering-schemes,per-user buffering>>::
7290 A buffering scheme in which all the processes of a Unix user share the
7291 same sub-buffer for a given user space channel.
7293 <<lttng-relayd,relay daemon>>::
7294 A process which is responsible for receiving the trace data sent by
7295 a distant consumer daemon.
7298 A set of sub-buffers.
7300 <<lttng-sessiond,session daemon>>::
7301 A process which receives control commands from you and orchestrates
7302 the tracers and various LTTng daemons.
7304 <<taking-a-snapshot,snapshot>>::
7305 A copy of the current data of all the sub-buffers of a given tracing
7306 session, saved as trace files.
7309 One part of an LTTng ring buffer which contains event records.
7312 The time information attached to an event when it is emitted.
7315 A set of files which are the concatenations of one or more
7316 flushed sub-buffers.
7319 The action of recording the events emitted by an application
7320 or by a system, or to initiate such recording by controlling
7324 The http://tracecompass.org[Trace Compass] project and application.
7327 An instrumentation point using the tracepoint mechanism of the Linux
7328 kernel or of LTTng-UST.
7330 tracepoint definition::
7331 The definition of a single tracepoint.
7334 The name of a tracepoint.
7336 tracepoint provider::
7337 A set of functions providing tracepoints to an instrumented user
7340 Not to be confused with a _tracepoint provider package_: many tracepoint
7341 providers can exist within a tracepoint provider package.
7343 tracepoint provider package::
7344 One or more tracepoint providers compiled as an object file or as
7348 A software which records emitted events.
7350 <<domain,tracing domain>>::
7351 A namespace for event sources.
7353 <<tracing-group,tracing group>>::
7354 The Unix group in which a Unix user can be to be allowed to trace the
7357 <<tracing-session,tracing session>>::
7358 A stateful dialogue between you and a <<lttng-sessiond,session
7362 An application running in user space, as opposed to a Linux kernel
7363 module, for example.